2D Materials TOPICAL REVIEW • OPEN ACCESS Recent citations Production and processing of graphene and - Review of fabrication methods of large- related materials area transparent graphene electrodes for industry Petri Mustonen et al To cite this article: Claudia Backes et al 2020 2D Mater. 7 022001 - Multifunctional graphene oxide/biopolymer composite aerogels for microcontaminants removal from drinking water Alessandro Kovtun et al - Pei He et al View the article online for updates and enhancements. This content was downloaded from IP address 213.148.215.17 on 16/07/2020 at 10:42 2D Mater. 7 (2020) 022001 https://doi.org/10.1088/2053-1583/ab1e0a OPEN ACCESS TOPICAL REVIEW Production and processing of graphene and related materials RECEIVED Claudia Backes1,2, Amr M Abdelkader3, Concepción Alonso4, Amandine Andrieux-Ledier5, 17 May 2018 Raul Arenal6,7,72, Jon Azpeitia8, Nilanthy Balakrishnan9, Luca Banszerus10, Julien Barjon11, Ruben Bartali12, RE VISED 13 March 2019 Sebastiano Bellani13, Claire Berger14,15, Reinhard Berger16, M M Bernal Ortega17, Carlo Bernard18, ACCEP TED FOR PUBLICATION Peter H Beton9, André Beyer19, Alberto Bianco20, Peter Bøggild65, Francesco Bonaccorso13,68, 30 April 2019 Gabriela Borin Barin21, Cristina Botas22, Rebeca A Bueno8, Daniel Carriazo22,23, Andres Castellanos- PUBLISHED Gomez8, Meganne Christian24, Artur Ciesielski25, Tymoteusz Ciuk26, Matthew T Cole27, Jonathan Coleman2, 29 January 2020 Camilla Coletti13,28, Luigi Crema12, Huanyao Cun18, Daniela Dasler29, Domenico De Fazio3, Noel Díez22, Simon Drieschner30, Georg S Duesberg31, Roman Fasel21,33, Xinliang Feng16, Alberto Fina17, Stiven Forti28, Original content from this work may be used Costas Galiotis34,35, Giovanni Garberoglio36, Jorge M García64, Jose Antonio Garrido38, Marco Gibertini39, under the terms of the Armin Gölzhäuser19, Julio Gómez40, Thomas Greber18, Frank Hauke29, Adrian Hemmi18, Irene Creative Commons Attribution 3.0 licence. Hernandez-Rodriguez8, Andreas Hirsch29, Stephen A Hodge3, Yves Huttel8, Peter U Jepsen65, Ignacio Jimenez8, Any further distribution Ute Kaiser66, Tommi Kaplas32, HoKwon Kim41, Andras Kis41, Konstantinos Papagelis35,42, Kostas Kostarelos43, of this work must Aleksandra Krajewska26,74, Kangho Lee31, Changfeng Li44, Harri Lipsanen44, Andrea Liscio70, Martin R Lohe16, maintain attribution to the author(s) and the Annick Loiseau46, Lucia Lombardi3, Maria Francisca López8, Oliver Martin29, Cristina Martín47, title of the work, journal Lidia Martínez8, Jose Angel Martin-Gago8, José Ignacio Martínez8, Nicola Marzari39, Álvaro Mayoral7,71, citation and DOI. John McManus2, Manuela Melucci45, Javier Méndez8, Cesar Merino48, Pablo Merino8,69, Andreas P Meyer29, Elisa Miniussi18, Vaidotas Miseikis28, Neeraj Mishra28, Vittorio Morandi24, Carmen Munuera8, Roberto Muñoz8, Hugo Nolan2, Luca Ortolani24, Anna K Ott3,73, Irene Palacio8, Vincenzo Palermo45,50, John Parthenios35, Iwona Pasternak26,51, Amalia Patane9, Maurizio Prato23,52,53, Henri Prevost46, Vladimir Prudkovskiy14, Nicola Pugno54,55,56, Teófilo Rojo22,57, Antonio Rossi28, Pascal Ruffieux21, Paolo Samorì25, Léonard Schué46, Eki Setijadi12, Thomas Seyller58, Giorgio Speranza12, Christoph Stampfer10, Ingrid Stenger11, Wlodek Strupinski26,51, Yuri Svirko32, Simone Taioli36,37, Kenneth B K Teo59, Matteo Testi12, Flavia Tomarchio3, Mauro Tortello60, Emanuele Treossi45, Andrey Turchanin61, Ester Vazquez47, Elvira Villaro62, Patrick R Whelan65, Zhenyuan Xia45,50, Rositza Yakimova63, Sheng Yang16, G Reza Yazdi63, Chanyoung Yim31, Duhee Yoon3, Xianghui Zhang19, Xiaodong Zhuang16, Luigi Colombo67, Andrea C Ferrari3 and Mar Garcia-Hernandez8,75 1 Physikalisch-Chemisches Institut, University of Heidelberg, Im Neuenheimer Feld 253, 69120 Heidelberg, Germany 2 School of Physics, Centre for Research on Adaptive Nanostructures and Nanodevices (CRANN) and Advanced Materials and BioEngineering Research (AMBER), Trinity College Dublin, Dublin 2, Ireland 3 Cambridge Graphene Centre, 9 JJ Thomson Avenue, University of Cambridge, CB3 0FA, United Kingdom 4 Department Applied Physical Chemistry, Autonomous University of Madrid, 28049 Madrid, Spain 5 ONERA, Département Physique, Instrumentation, Environnement et Espace—Université Paris Saclay—29 Avenue de la Division Leclerc, 92320 Chatillon, France 6 ARAID Fundation, Calle Mariano de Luna, 50018 Zaragoza, Spain 7 Laboratorio de Microscopias Avanzadas (LMA) Instituto de Nanociencia de Aragon (INA), Universidad de Zaragoza, Calle Mariano Esquillor, 50018 Zaragoza, Spain 8 Materials Science Factory, Instituto de Ciencia de Materiales de Madrid, Consejo Superior de Investigaciones Científicas, C/Sor Juana Ines de la Cruz 3, Madrid, E-28049, Spain 9 School of Physics and Astronomy, University of Nottingham, University Park, Nottingham, NG7 2RD, United Kingdom 10 JARA-FIT and 2nd Institute of Physics, RWTH Aachen University, Aachen, Germany 11 Groupe d’Etude de la Matière Condensée (GEMaC), CNRS—Université Versailles ST Quentin—Université Paris Saclay, 45 Avenue des Etats Unis, Versailles, France 12 Fondazione Bruno Kessler, via Sommarive 18, 38123 Trento, Italy 13 Graphene Labs, Istituto Italiano di Tecnologia, Via Morego 30, 16163 Genova, Italy 14 Institut Néel, CNRS—University Grenoble-Alpes, 25 rue des Martyrs, 38042 Grenoble, France 15 Georgia Institute of Technology, School of physics, Atlanta, GA, United States of America 16 Center for Advancing Electronics Dresden (cfaed) and Department of Chemistry and Food Chemistry, Technische Universität Dresden, Mommsenstr. 4, 01062 Dresden, Germany 17 Dipartimento di Scienza Applicata e Tecnologia, Politecnico di Torino, Alessandria Campus, Viale Teresa Michel 5, 15121 Alessandria, Italy 18 Physik-Institut, University of Zürich, Winterthurerstrasse 190, CH-8057 Zürich, Switzerland 19 Physics of Supramolecular Systems, Bielefeld University, Universitätsstr. 25, 33615 Bielefeld, Germany 20 University of Strasbourg, CNRS, Immunology, Immunopathology and Therapeutic Chemistry, UPR 3572, 67000 Strasbourg, France 21 Empa, Swiss Federal Laboratories for Materials Science and Technology, Überlandstr. 129, 8600 Dübendorf, Switzerland 22 CIC EnergiGUNE. Parque Tecnológico de Álava, 01510-Miñano, Álava, Spain 23 IKERBASQUE, Basque Foundation for Science, 48013-Bilbao, Spain 24 Institute for Microelectronic and Microsystems (IMM) Section of Bologna - CNR, via Gobetti 101, 40129 Bologna, Italy © 2020 The Author(s). Published by IOP Publishing Ltd 2D Mater. 7 (2020) 022001 C Backes et al 25 Université de Strasbourg, CNRS, ISIS, 8 alleé Gaspard Monge, 67000 Strasbourg, France 26 Instytut Technologii Materiałów Elektronicznych, Wólczyńska 133, 01-919 Warszawa, Poland 27 Department of Electronic and Electrical Engineering, BA2 7AY, Bath, United Kingdom 28 Center for Nanotechnology Innovation @NEST, Istituto Italiano di Tecnologia, P.zza S. Silvestro 12, 56127, Pisa, Italy 29 Chair of Organic Chemistry II and Joint Institute of Advanced Materials and Processes, Nikolaus-Fiebiger-Str. 10, Friedrich-Alexander Universität Erlangen-Nürnberg, 91058 Erlangen, Germany 30 Walter Schottky Institut Technische Universität München, Am Coulombwall 4, D-85748 Garching, Germany 31 Institute of Physics, Universität der Bundeswehr München, 85577 Neubiberg, Germany 32 Institute of Photonics, University of Eastern Finland, Yliopistokatu 7, 80100 Joensuu, Finland 33 Department of Chemistry and Biochemistry, University of Bern, 3012 Bern, Switzerland 34 Department of Chemical Engineering, University of Patras, Patras 26504, Greece 35 Institute of Chemical Engineering Sciences, Foundation for Research and Technology- Hellas (FORTH/ICE-HT), Patras 26504, Greece 36 European Centre for Theoretical Studies in Nuclear Physics and Related Areas (ECT*-FBK), 38123, Trento, Italy and Trento Institute for Fundamental Physics and Applications (TIFPA-INFN), 38123 Trento, Italy 37 Faculty of Mathematics and Physics, Charles University, Praha 8, 180 00 Prague, Czech Republic 38 Instituto Catalán de Nanotecnologia ICN2, UAB Campus, Bellaterra, (Barcelona) 08193, Spain 39 Theory and Simulation of Materials (THEOS), and National Centre for Computational Design and Discovery of Novel Materials (MARVEL), École Polytechnique Fédérale de Lausanne, CH-1015 Lausanne, Switzerland 40 Avanzare Innovacion Tecnologica S.L., Avda Lentiscares 4-6 Navarrete, Spain 41 EPFL, Electrical Engineering Institute and Institute of Materials Science and Engineering, École Polytechnique Fédérale de Lausanne (EPFL), CH-1015 Lausanne, Switzerland 42 School of Physics, Department of Solid State Physics, Aristotle University of Thessaloniki, Thessaloniki 54124, Greece 43 University of Manchester, Manchester, United Kingdom 44 Department of Electronics and Nanoengineering, Aalto University, Tietotie 3, 02150 Espoo, Finland 45 Institute of Organic Synthesis and Photoreactivity (ISOF), CNR, via Gobetti 101, 40129, Bologna, Italy 46 Laboratoire d’Etude des Microstructures (LEM), CNRS–ONERA, Université Paris Saclay, 29 Avenue de la Division Leclerc, 92320 Chatillon, France 47 Instituto Regional de Investigación Científica Aplicada (IRICA), Universidad de Castilla-la Mancha, Avda Camilo Jose Cela, 13071, Ciudad Real, Spain 48 ACM Advanced Carbon Materials, Grupo Antolin Ingeniería, SA, Ctra. Madrid—Irún, km. 244.8, E09007 Burgos, Spain 49 Consorzio Nazionale Interuniversitario per le Telecomunicazioni, Via Moruzzi 1, 56124, Pisa, Italy 50 Department of Industrial and Materials Science, Chalmers University of Technology, Horsalsvagen 7A, SE-412 96 Gothenburg, Sweden 51 Faculty of Physics, Warsaw University of Technology, Koszykowa, 75 00-662 Warsaw, Poland 52 Carbon Bionanotechnology Laboratory, CIC biomaGUNE, Paseo de Miramón 182, 20009 Donostia-San Sebastián, Spain 53 Dipartimento di Scienze Chimiche e Farmaceutiche, INSTM UdR Trieste, Università degli Studi di Trieste, Via Licio Giorgieri 1, 34127 Trieste, Italy 54 Laboratory of Bio-Inspired and Graphene Nanomechanics, Department of Civil, Environmental and Mechanical Engineering, University of Trento, Via Mesiano 77, 38123 Trento, Italy 55 Ket Lab, Edoardo Amaldi Foudation, Via del Politecnico snc, 00133 Rome, Italy 56 School of Engineering and Materials Science, Queen Mary University of London, Mile End Road, London E1 4NS, United Kingdom 57 Universidad del País Vasco UPV/EHU, 48080-Bilbao, Spain 58 Technische Universität Chemnitz, Institut für Physik, Reichenhainer Str. 70, 09126 Chemnitz, Germany 59 Buckingway Business Park, Anderson Road, Swavesey, Cambridge CB24 4FQ, United Kingdom 60 Politecnico di Torino, C.so Duca degli Abruzzi 24, 10129 Torino, Italy 61 Institute of Physical Chemistry, Friedrich Schiller University Jena, Lessingstr. 10, 07743 Jena, Germany 62 Interquimica, C/Antonio Nebrija, 8- Bajo. CP 26006, Logroño, Spain 63 University of Linköping, IFM, 58183, Linkoping, Sweden 64 Instituto de Micro y Nanotecnología, IMN-CNM, CSIC (CEI UAM  +  CSIC) Isaac Newton, 8, E-28760, Tres Cantos, Madrid, Spain 65 Center for Nanostructured Graphene (CNG), Technical University of Denmark, Kongens Lyngby, 2800, Denmark 66 Materialwissenschaftliche Elektronenmikroskopie, Universität Ulm, Albert-Einstein-Allee 11, 89081 Ulm, Germany 67 Department of Materials Science and Engineering, University of Texas at Dallas, Richardson, TX 75080, United States of America 68 BeDimensional Spa, Via Albisola 121, 16163 Genova, Italy 69 Instituto de Física Fundamental, CSIC, Serrano 121, 28006, Madrid, Spain 70 Institute for Microelectronic and Microsystems (IMM) Section of Roma—CNR, via del fosso del cavaliere 100, 00133 Roma, Italy 71 Center for High-resolution Electron Microscopy (CћEM), School of Physical Science and Technology, ShanghaiTech University, 393 Middle Huaxia Road, Pudong, Shanghai 72 Instituto de Ciencia de Materiales de Aragon, CSIC-Univ. Zaragoza, Pedro Cerbuna 12, 50009 Zaragoza, Spain 73 College of Engineering, Mathematics and Physical Sciences, University of Exeter, Exeter, EX4 4QF, United Kingdom 74 International Research Centre CENTERA, Institute of High Pressure Physics, Polish Academy of Sciences, Warsaw, 01-142, Poland 75 Author to whom any correspondence should be addressed. E-mail: marmar@icmm.csic.es Keywords: processing of layered materials, inks of layered materials, characterization of layered materials, functionalization of layered materials, synthesis of graphene and related materials, growth of layered materials Abstract We present an overview of the main techniques for production and processing of graphene and related materials (GRMs), as well as the key characterization procedures. We adopt a ‘hands-on’ approach, providing practical details and procedures as derived from literature as well as from the authors’ experience, in order to enable the reader to reproduce the results. 2 2D Mater. 7 (2020) 022001 C Backes et al Section I is devoted to ‘bottom up’ approaches, whereby individual constituents are pieced together into more complex structures. We consider graphene nanoribbons (GNRs) produced either by solution processing or by on-surface synthesis in ultra high vacuum (UHV), as well carbon nanomembranes (CNM). Production of a variety of GNRs with tailored band gaps and edge shapes is now possible. CNMs can be tuned in terms of porosity, crystallinity and electronic behaviour. Section II covers ‘top down’ techniques. These rely on breaking down of a layered precursor, in the graphene case usually natural crystals like graphite or artificially synthesized materials, such as highly oriented pyrolythic graphite, monolayers or few layers (FL) flakes. The main focus of this section is on various exfoliation techniques in a liquid media, either intercalation or liquid phase exfoliation (LPE). The choice of precursor, exfoliation method, medium as well as the control of parameters such as time or temperature are crucial. A definite choice of parameters and conditions yields a particular material with specific properties that makes it more suitable for a targeted application. We cover protocols for the graphitic precursors to graphene oxide (GO). This is an important material for a range of applications in biomedicine, energy storage, nanocomposites, etc. Hummers’ and modified Hummers’ methods are used to make GO that subsequently can be reduced to obtain reduced graphene oxide (RGO) with a variety of strategies. GO flakes are also employed to prepare three-dimensional (3d) low density structures, such as sponges, foams, hydro- or aerogels. The assembly of flakes into 3d structures can provide improved mechanical properties. Aerogels with a highly open structure, with interconnected hierarchical pores, can enhance the accessibility to the whole surface area, as relevant for a number of applications, such as energy storage. The main recipes to yield graphite intercalation compounds (GICs) are also discussed. GICs are suitable precursors for covalent functionalization of graphene, but can also be used for the synthesis of uncharged graphene in solution. Degradation of the molecules intercalated in GICs can be triggered by high temperature treatment or microwave irradiation, creating a gas pressure surge in graphite and exfoliation. Electrochemical exfoliation by applying a voltage in an electrolyte to a graphite electrode can be tuned by varying precursors, electrolytes and potential. Graphite electrodes can be either negatively or positively intercalated to obtain GICs that are subsequently exfoliated. We also discuss the materials that can be amenable to exfoliation, by employing a theoretical data-mining approach. The exfoliation of LMs usually results in a heterogeneous dispersion of flakes with different lateral size and thickness. This is a critical bottleneck for applications, and hinders the full exploitation of GRMs produced by solution processing. The establishment of procedures to control the morphological properties of exfoliated GRMs, which also need to be industrially scalable, is one of the key needs. Section III deals with the processing of flakes. (Ultra)centrifugation techniques have thus far been the most investigated to sort GRMs following ultrasonication, shear mixing, ball milling, microfluidization, and wet-jet milling. It allows sorting by size and thickness. Inks formulated from GRM dispersions can be printed using a number of processes, from inkjet to screen printing. Each technique has specific rheological requirements, as well as geometrical constraints. The solvent choice is critical, not only for the GRM stability, but also in terms of optimizing printing on different substrates, such as glass, Si, plastic, paper, etc, all with different surface energies. Chemical modifications of such substrates is also a key step. Sections IV–VII are devoted to the growth of GRMs on various substrates and their processing after growth to place them on the surface of choice for specific applications. The substrate for graphene growth is a key determinant of the nature and quality of the resultant film. The lattice mismatch between graphene and substrate influences the resulting crystallinity. Growth on insulators, such as SiO2, typically results in films with small crystallites, whereas growth on the close-packed surfaces of metals yields highly crystalline films. Section IV outlines the growth of graphene on SiC substrates. This satisfies the requirements for electronic applications, with well-defined graphene-substrate interface, low trapped impurities and no need for transfer. It also allows graphene structures and devices to be measured directly on the growth substrate. The flatness of the substrate results in graphene with minimal strain and ripples on large areas, allowing spectroscopies and surface science to be performed. We also discuss the surface engineering by intercalation of the resulting graphene, its integration with Si-wafers and the production of nanostructures with the desired shape, with no need for patterning. Section V deals with chemical vapour deposition (CVD) onto various transition metals and on insulators. Growth on Ni results in graphitized polycrystalline films. While the thickness of these films can be optimized by controlling the deposition parameters, such as the type of hydrocarbon 3 2D Mater. 7 (2020) 022001 C Backes et al precursor and temperature, it is difficult to attain single layer graphene (SLG) across large areas, owing to the simultaneous nucleation/growth and solution/precipitation mechanisms. The differing characteristics of polycrystalline Ni films facilitate the growth of graphitic layers at different rates, resulting in regions with differing numbers of graphitic layers. High-quality films can be grown on Cu. Cu is available in a variety of shapes and forms, such as foils, bulks, foams, thin films on other materials and powders, making it attractive for industrial production of large area graphene films. The push to use CVD graphene in applications has also triggered a research line for the direct growth on insulators. The quality of the resulting films is lower than possible to date on metals, but enough, in terms of transmittance and resistivity, for many applications as described in section V. Transfer technologies are the focus of section VI. CVD synthesis of graphene on metals and bottom up molecular approaches require SLG to be transferred to the final target substrates. To have technological impact, the advances in production of high-quality large-area CVD graphene must be commensurate with those on transfer and placement on the final substrates. This is a prerequisite for most applications, such as touch panels, anticorrosion coatings, transparent electrodes and gas sensors etc. New strategies have improved the transferred graphene quality, making CVD graphene a feasible option for CMOS foundries. Methods based on complete etching of the metal substrate in suitable etchants, typically iron chloride, ammonium persulfate, or hydrogen chloride although reliable, are time- and resource- consuming, with damage to graphene and production of metal and etchant residues. Electrochemical delamination in a low-concentration aqueous solution is an alternative. In this case metallic substrates can be reused. Dry transfer is less detrimental for the SLG quality, enabling a deterministic transfer. There is a large range of layered materials (LMs) beyond graphite. Only few of them have been already exfoliated and fully characterized. Section VII deals with the growth of some of these materials. Amongst them, h-BN, transition metal tri- and di-chalcogenides are of paramount importance. The growth of h-BN is at present considered essential for the development of graphene in (opto) electronic applications, as h-BN is ideal as capping layer or substrate. The interesting optical and electronic properties of TMDs also require the development of scalable methods for their production. Large scale growth using chemical/physical vapour deposition or thermal assisted conversion has been thus far limited to a small set, such as h-BN or some TMDs. Heterostructures could also be directly grown. Section VIII discusses advances in GRM functionalization. A broad range of organic molecules can be anchored to the sp2 basal plane by reductive functionalization. Negatively charged graphene can be prepared in liquid phase (e.g. via intercalation chemistry or electrochemically) and can react with electrophiles. This can be achieved both in dispersion or on substrate. The functional groups of GO can be further derivatized. Graphene can also be noncovalently functionalized, in particular with polycyclic aromatic hydrocarbons that assemble on the sp2 carbon network by π–π stacking. In the liquid phase, this can enhance the colloidal stability of SLG/FLG. Approaches to achieve noncovalent on-substrate functionalization are also discussed, which can chemically dope graphene. Research efforts to derivatize CNMs are also summarized, as well as novel routes to selectively address defect sites. In dispersion, edges are the most dominant defects and can be covalently modified. This enhances colloidal stability without modifying the graphene basal plane. Basal plane point defects can also be modified, passivated and healed in ultra-high vacuum. The decoration of graphene with metal nanoparticles (NPs) has also received considerable attention, as it allows to exploit synergistic effects between NPs and graphene. Decoration can be either achieved chemically or in the gas phase. All LMs, can be functionalized and we summarize emerging approaches to covalently and noncovalently functionalize MoS2 both in the liquid and on substrate. Section IX describes some of the most popular characterization techniques, ranging from optical detection to the measurement of the electronic structure. Microscopies play an important role, although macroscopic techniques are also used for the measurement of the properties of these materials and their devices. Raman spectroscopy is paramount for GRMs, while PL is more adequate for non-graphene LMs (see section IX.2). Liquid based methods result in flakes with different thicknesses and dimensions. The qualification of size and thickness can be achieved using imaging techniques, like scanning probe microscopy (SPM) or transmission electron microscopy (TEM) or spectroscopic techniques. Optical microscopy enables the detection of flakes on suitable surfaces as well as the measurement of optical properties. Characterization of exfoliated materials is essential to improve the GRM metrology for applications and quality control. For grown GRMs, SPM can 4 2D Mater. 7 (2020) 022001 C Backes et al be used to probe morphological properties, as well as to study growth mechanisms and quality of transfer. More generally, SPM combined with smart measurement protocols in various modes allows one to get obtain information on mechanical properties, surface potential, work functions, electrical properties, or effectiveness of functionalization. Some of the techniques described are suitable for ‘in situ’ characterization, and can be hosted within the growth chambers. If the diagnosis is made ‘ex situ’, consideration should be given to the preparation of the samples to avoid contamination. Occasionally cleaning methods have to be used prior to measurement. Contents I. Bottom-up 9 I.1. Graphene nanoribbons 9 I.2. Graphene and carbon nanomembranes 15 I.3. Heterostructures from CNMs 21 II. Top-down 23 II.1. Precursors 23 II.2. Liquid phase exfoliation 24 II.3. GO and RGO 34 II.4. Chemical intercalation and reductive exfoliation 41 II.5. Electrochemical exfoliation 42 II.6. Sonication-assisted versus chemical versus electrochemical exfoliation 45 II.7. Computational modelling of exfoliation of LMs 46 III. Processing of dispersions 48 III.1. Size selection 48 III.2. Inks formulation 48 III.3. Printing and deposition of inks 52 III.4. Applications 60 IV. Graphene growth on SiC 62 IV.1. Sublimation 62 IV.2. CVD growth on SiC 81 V. CVD, PVD & MBE 84 V.1. Growth on metals 84 V.2. Graphene growth on semiconductors 108 V.3. Graphene growth on insulators 108 V.4. Brief comparison of all methods 119 VI. Graphene transfer, placement and decoupling from substrate 120 VI.1. Wet transfer 120 VI.2. Semi-dry transfer: hot press lamination and UV assisted transfer 128 VI.3. Dry transfer using h-BN 132 VI.4. Graphene/PMMA sandwich structures 132 VII. Growth and transfer of other layered materials 134 VII.1. Hexagonal BN 134 VII.2. Layered semiconductors 140 VII.3. Layered materials heterostructures 156 VIII. Functionalization of GRMs 160 VIII.1. Covalent functionalization of graphene 160 VIII.2. Functionalization of GO 165 VIII.3. Noncovalent functionalization of graphene 167 VIII.4. Defect functionalization of GRM 174 VIII.5. Decoration with nanoparticles 176 VIII.6. Functionalization of other LMs 182 IX. Characterization methods 185 IX.1. Microscopies 185 IX.2. Spectroscopies 210 IX.3. Electrical characterization 226 IX.4. Mechanical characterization 234 Conclusions241 Acknowledgment241 References242 5 2D Mater. 7 (2020) 022001 C Backes et al List of acronyms DFT Density functional theory DGM Density gradient medium 0d Zero dimensional DGU Density gradient ultracentrifugation 1d One dimensional Dh Decahedral 1L One layer DI Deionized water 1LG 1 layer graphene DIC Differential interference contrast 2D 2D Raman peak of graphene DMAc N,N′-dimethylacetamide 2d Two dimensional DME 1,2-dimethoxyethane 2LG 2 layer graphene DMF Dimethylformamide 3d Three dimensional DMSO Dimethylsulfoxide 3LG 3 layer graphene DOS Density of states 4LG 4 layer graphene DOTA 1,4,7,10-tetraazacyclododecane- AES Auger electron spectroscopy 1,4,7,10-tetraacetic acid AFM Atomic force microscopy DSSC Dye-sensitized solar cell AGNR GNRs with armchair edges EA Elemental analysis ALD Atomic layer deposition EBSD Electron back scattering diffraction APCVD Atmospheric pressure CVD EC Ethyl cellulose APS Ammonium persulfate EDA Ethylene-diamine ARGO Reduced graphene oxide-based aerogels EE Electrochemical exfoliation ARPES  Angle resolved photoemission EEG Electrochemically exfoliated spectroscopy graphene BHJ Bulk hetro junction EG Epitaxial graphene Bipy 2,2′-Bipyridine EVA Ethylene vinyl acetate BL Buffer layer FCC Face-centered cubic BLG Bilayer graphene FESEM  Field emission scanning electron BODIPY 2,6-diiodo-1,3,5,7-tetramethyl-8- microscope phenyl-4,4-difluoroboradiazaindacene FET Field effect transistor BP Black phosphorus FIB Focused ion beam BP3 3-(Biphenyl-4-yl)propane-1-thiol FL Few layer BPT 1,1′-biphenyl-4-thiol FLaT Functional layer transfer BZ Brillouin zone FLG Few layer graphene Bz Benzidine FoM Figure of merit CA Cellulose acetate FS Fermi surface CCS Confinement controlled sublimation FTIR  Fourier transform infrared c-CVD Catalytical-chemical vapor deposition spectroscopy CE Counter electrode FWHM Full width at half maximum CES Constant energy surface G G Raman peak CHP N-cyclohexyl-2-pyrrolidone G3DCN Graphene-based covalent networks CL Cathode luminiscence GANF Grupo Antolin nanofibres CMC Critical micelle concentration GFactor Gauge factor CMOS Complementary metal-oxide GF Graphene foam semiconductor GIC Graphite intercalation compound CMP Conjugated microporous polymers GnP Graphene nanoplatelets CNF Carbon nanofibre GNR  Graphene nanoribbons, nanometer CNM Carbon nanomembrane wide graphene strips exhibiting a CNT Carbon nanotube bandgap COD 1,5-cyclooctadiene GO Graphene oxide CPD Critical point drying GOS Graphene on silicon CTAB Cetyl trimethyl ammonium bromide GRM Graphene and related material CVD Chemical vapour deposition GRMs Graphene and related materials D D Raman peak HAADF High angle annular dark field DA Diels-Alder HATU  O - ( 7 - A z a b e n z o t r i a z o l e - 1 - y l ) - DAP Donor–acceptor pairs N , N , N , N ′ - t e t r a m e t hy l u ro n i u m db Dangling bonds hexafluorophosphate DBT 4-docosyloxy-benzenediazonium HBC Hexabenzocoronene tetrafluoroborate HBC-Br  2 - B r o m o - 1 1 - ( 1 ′ - [ 4 ′ - ( S - DCM Dichloromethane Acetylthiomethyl)phenyl]acetyl)- DEA Dissociative electron attachment 5,8,14,17-tetra(3′,7′-dimethyloctyl)- DES Diethyl sulphide hexa-peri-hexabenzocoronene 6 2D Mater. 7 (2020) 022001 C Backes et al HBC-CN  2 - C y a n o - 1 1 - ( 1 ′ - [ 4 ′ - ( S - NDI Naphthalene diimide Acetylthiomethyl)phenyl]acetyl)- NMP N-methyl-2-pyrrolidone 5,8,14,17-tetra(3′,7′-dimethyloctyl)- NMR Nuclear magnetic resonance hexa-peri-hexabenzocoronene NP Nanoparticle HIM Helium ion microscope NPTH Naphtalene-2-thiol hMDM Human monocyte macrophages ODCB 1,2-dichlorobenzene HMDS Hexamethyldisilazane OPV Organic photovoltaics HOPG Highly oriented pyrolytic graphite ORR Oxygen reduction reaction HPB  S,S′-(3′,4′,5′,6′-tetraphenyl- Otf Trifluoromethanesulfonate [1,1′:2′,1″-terphenyl]-4,4″-diyl) OTFT Organic thin film transistor diethanethioate P3HT Poly(3-hexylthiophene) HPL Hot-press lamination PAH Polycyclic aromatic hydrocarbon HREM High resolution electron microscopy PANi Polyaniline HRP Horseradish peroxidase PCBM  Phenyl-C61-butyric acid methyl HRTEM High resolution transmission electron ester microscopy PDI Perylene diimide ICS Ion cluster source PDMS Polydimethylsiloxane IFM Inverted floating method PE Polyethylene Ih Icosahedral PECVD  Plasma enhanced chemical vapour IPA Isopropyl alcohol deposition JCNM Janus carbon nanomembrane PEDOT Poly(3,4 ethylenedioxythiophene) KPM Kelvin probe microscopy PET Polyethylene terephthalate LCC Liquid cascade centrifugation PhCN Benzonitrile LEED Low energy electron diffraction PI Polyimide LEEM Low energy electron microscopy PL Photoluminescence LIB Lithium ion battery PMMA Poly(methyl methacrylate) LM Layered material PPF Pyrolyzed photoresist film LMH Layered material heterostructure PPP Polu(p -phenylene) LMs Layered materials PS Polystyrene LPCVD Low pressure CVD PTCVD Photo thermal CVD LPE Liquid phase exfoliation PTFE Polytetrafluoroethylene LRI Liquid resin infusion PVA Polyvinyl acetate µ Carrier mobilities PVC Polyvinyl chloride µ-ARPE Micro spot ARPES PVDF Polyvinylidene difluoride MALDI-TOF  Matrix-assisted laser desorption/ PVP Polyvinylpyrrolidone ionization-time of flight PVT Physical vapor transport MBE Molecular beam epitaxy QFBLG Quasi free-standing bilayer graphene MC Mechanically-cleaved QFSLG  Quasi free-standing single layer MEG Multilayer epitaxial graphene graphene MEP Minimum energy path r-(ECR-CVD) Remote electron cyclotron resonance MICS Multiple ion cluster source plasma assisted chemical vapor mIPM Murine intraperitoneal macrophages deposition ML Monolayer RBLM Radial-breathing-like mode MLG Multilayer graphene RCA1, RCA2 Standard sets of wafer cleaning steps MNP Metal nanoparticle which need to be performed before MOCVD  Metalorganic chemical vapour high-temperature processing steps deposition RE Reference electrode MP Moire pattern RF Radiofrequency MS Mass spectrometry RF-ID Radio-frequency identification MSCs Micro super capacitors RGO Reduced graphene oxide MW Microwave RHEED  Reflection high energy electron MW Molecular weight diffraction N Number of layers RIE Reactive ion etching NaCMC Sodium carboxymethylcellulose Rpm Revolutions per minute NBE Near band edge RT Room temperature n-BuLi n-butyllithium RTM Resin transfer moulding NC Nitrocellulose RTP Rapid thermal process Nc-AFM Non-contact atomic force microscopy RZC Rate zonal centrifugation 7 2D Mater. 7 (2020) 022001 C Backes et al SAED Selected area electron diffraction TCB 1,2,4-trichlorobenzene SAM Self-assembled monolayer TCF Transparent conducting film SARPES Spin resolved ARPES TCNQ Tetracyanoquinodimethane SAV Single atom vacancies TEM Transmission electron microscopy SBS Sedimentation based separation TEMPO 2,2,6,6-tetramethyl-1-piperidinyloxy SC Sodium cholate TGA Thermogravimetric analysis SDBS Sodium dodecyl benzene sulfonate THF Tetrahydrofuran SDC Sodium deoxycholate THz-TDS THz-time domain spectroscopy SDS Sodium dodecyl sulfate TLG Trilayer graphene SEC Size exclusion chromatography TMC Transition metal chalcogenide SEM Scanning electron microscopy TMD Transition metal dichalcogenide SET Single electron transfer TOA Tetraoctylammonium SL Single-layer TPT [1″,4′,1′,1]-Terphenyl-4-thiol SLG Single layer graphene Tr-ARPES Time-resolved ARPES Slm Standard liter per minute TRT Thermal release tape SOI Silicon on insulator UHV Ultra high vacuum STEM Scanning transmission electron ULF Ultra low frequency microscopy UV Ultraviolet SThM Scanning thermal microscopy UVA Ultraviolet adhesive STM Scanning tunnelling microscopy Uv–Vis Ultraviolet–visible STS Scanning tunnelling spectroscopy vdW van der Waals SuMBE Supersonic molecular beam epitaxy VPE Vapour phase epitaxy SWCNT Single wall carbon nanotube WE Working electrode T Temperature XPS X-ray photoelectron spectroscopy TAC Thermally assisted conversion Z Reciprocal of the Ohnesorge number Tb Temperature of the bubbler ZGNR GNRs with zigzag edges 8 2D Mater. 7 (2020) 022001 C Backes et al I. Bottom-up nanographenes [22, 23] The reaction is based on the intramolecular oxidative cyclodehydrogenation of I.1. Graphene nanoribbons corresponding oligophenylene precursors and was GNRs are an interesting family of materials extended from defined molecules to polymers, i.e. combining aspect ratios allowing to bridge the range from PAHs to GNRs [23]. Since then, the synthesis of of (sub-) nanometer dimensions with ultimate GNRs through intramolecular cyclodehydrogenation structure-properties relationship (GNR width, below of polyphenylene polymers was achieved, through 5 nm) and mesoscopic dimensions (GNR length up to AA-, AB- and A2B2-type polymerizations as 500 nm). This makes GNRs accessible to established summarized in Refs. [24–27]. top-down contacting strategies and thus allows their The most critical issue in the solution-synthesis device integration. For GNRs with armchair edges is to achieve high (>600 000 g mol−1 on average by (AGNRs), theory predicts the opening of sizable Diels Alder (DA) polymerization) molecular weight bandgaps as soon as their widths falls below ~2 nm of the precursor polymer. While the width of GNRs is [1–5]. This is due to quantum confinement and edge determined by the monomer dimensions, the molec- effects and can qualitatively be understood by slicing ular weight is directly proportional to the number of the graphene Dirac cone along k-lines in the reciprocal repeating units, therefore proportional to the length space that are compatible with the hard-wall boundary of the resulting GNR after the cyclodehydrogenation conditions set by the finite AGNR width. The further -graphitization- step. E.g. DA polymerization pro- these cuts of allowed electronic states are away from vides a molecular weight  >600 000, on average corre­ the K point of the Brillouin zone of graphene, the sponding to a length of 600 nm [23]. larger is their bandgap [6]. AGRNs were predicted to Figure I.1 provides an overview of the polymeriza- show metallic to semiconducting behavior, depending tion routes explained in what follows along with the on their width [2, 7–9]. Generally, AGRNs exhibiting relevant references. widths smaller than 10 nm behave as semiconductors with non-zero bandgaps that increase as the they I.1.2. Preparation of polymer precursors by A2B2- become narrower [5, 7–11]. E.g., AGNRs as narrow poly­merization as 2–3 nm are expected to havea bandgap ~0.7 eV, A2B2-Polymerization requires two monomers which is comparable to that of Ge [5]. In contrast, with complementary functional groups A and early theoretical studies indicated that zigzag GNRs, B (see figure I.1). These can be A  =  Cl, Br, I, Otf ZGNRs, have metallic properties with zero bandgap (Trifluoromethanesulfonate) in combination with B irrespective of the width, showing strongly localized being a boronic acid or boronic acid ester. In this case, edge states at the zigzag sites [7], with ferromagnetic the underlying carbon-carbon bond formation is based coupling along and antiferromagnetic coupling across on the Suzuki-reaction. In contrast, if A is a diene and B the edges [12]. Spin-polarized density functional a dienophile, it belongs to the reaction class of a Diels- theory DFT calculations have found that AGNRs Alder reaction. The most prominent combination for a are always semiconductors and that the ground state Diels-Alder reaction to form PAHs is the combination of ZGNRs has an antiferromagnetic configuration, of a cyclopentadienone and a substituted acetylene where electronic states with opposite spins are highly [28, 29]. The benefit of this inverse electron demand- localized at the two ZGNR edges and are responsible Diels-Alder reaction is the tandem cycloaddition and for the opening of a gap [1712, 1713]. Thus, small carbon-monoxide-extrusion reaction. Therefore, both differences in width and edge configuration lead to reaction classes require different protocols. In all cases, large variations in GNR properties [5, 10, 11], making the polymerization follows a step-growth mechanism it imperative to control the GNR structure on the and is defined by Carother’s equation [30]. In this case, atomic level to achieve the desired (opto)electronic the functionalized monomers first react into monomers, and magnetic properties with high accuracy and dimers, trimer, oligomers and finally high molecular reproducibility. While this is beyond the level of what weight polymers. The stoichiometry of both monomers can be currently controlled by top-down structuring is of fundamental importance to achieve a high methods, such as lithographic patterning [13–16] degree of polymerization and thus molecular weight. or cutting of CNTs[17–20], advances in bottom-up From our experience, a small deviation imbalance in fabrication have shown that GNRs with specific edge stoichiometry or impurity of at least one monomer of structure and width are accessible [21]. Not only even 2% will not allow high molecular weight polymer purely A or ZGNRs can be synthesized. Other types formation. To ensure exact stoichiometry, the purity of in between named chevron- or necklace-type can be monomers as well as dryness is a critical issue. A balance designed and prepared as well [1724]. used to weight the monomers must require an accuracy of 0.1 mg. In a theoretical example, an impurity of 2% at I.1.1. Solution synthesis of GNRs a degree of polymerization of 98%, will cut molecular The concept of solution based bottom-up synthesis weight to half [30]. relies on the preparation of large polycyclic A Suzuki reaction is a palladium catalyzed reaction. aromatic hydrocarbons (PAHs), often referred to as The active catalyst is a Pd(0) species which is oxygen 9 2D Mater. 7 (2020) 022001 C Backes et al Type Polymerization Method Catalyst Polymer Scholl Reaction GNR AA Yamamoto (J. Am. Chem. Soc. 2016, 138, 10136–10139) R R R R R R R R R R , FeCl 3 Cl Cl Ni(COD)2 DCM/MeNO2 R R R R R R R R R R AB Diels-Alder (Nat. Chem. 2014, 6, 126–132) R R R R R FeCl3, O heat DCM/MeNO2 R R R R R Suzuki (Chem. Eur.J. 2016, 22, 9116–9120) R R R R O Br B O DDQ, Pd(PtBu3)2 TfOH/DCM R R R R R R A2B2 Diels-Alder (J. Am. Chem. Soc. 2008, 130, 4216–4217) R R R R R R FeCl3, O O n DCM/MeNO2 R R R R R R Suzuki (J. Am. Chem. Soc. 2008, 130, 4216–4217) R R R R R RR R R RR R O O FeCl3, I I B B Pd(PPh3)4 O O DCM/MeNO2 R R R R R RR R R RR R Figure I.1. Overview individual types (AA, AB and A2B2) of polymerization the reactive parts are colored according to the protocol of reactions. References to examples are given on top of the monomer or monomer combination of main polymerization methods employed in solution synthesis of GNRs. sensitive. This protocol requires the preparation of the With prolonged reaction time in the order of reaction under inert condition. To exclude oxygen from hours to days, the molecular weight of the resulting the reaction, both monomers and a base (potassium polymer will increase. However, the solubility of the carbonate) are usually evacuated in a Schlenk-type formed polymer will decrease. Before the polymer glassware. Afterwards, the solvents (a combination of precipitates out of the solution, the residual terminal toluene, ethanol and water, typically 3:1:1), is bubbled functional groups (halogen or boronic acid) must be with Ar for at least 20 min when a total volume of sol- ‘end-capped’, to avoid undesired atoms at the termi- vent is in the range of 100–200 ml. It is recommended nal positions of the GNR. The ‘end-capping’ must to apply high (~1200 rpm) stirring during bubbling be performed before precipitation of the polymer to to ensure a complete saturation of the solvent mixture ensure conversion of unreacted functional groups. with Ar. After the reaction apparatus is in contact with This is achieved by adding a suitable end-capper, e.g. the preheated oil bath, the catalyst (Pd(PPh3)4) is added bromo-benzene followed by adding an excess of phe- under Ar. During the reaction, no oxygen should enter nylboronic acid in the respective solvent. The reaction the reaction chamber. In addition, it is recommended is continued for several hours (at least one) to ensure to cover the reaction chamber with Al foil to protect it the full conversion of the terminal functional groups. from light. To track the reaction, samples can be taken Afterwards, the reaction is quenched by the addition in different intervals of 15 min to several hours, always of water. After extraction and precipitation into typi- under Ar protection. These (0.1 ml) can be quenched cally methanol, the crude polymer can be character- by adding a drop of water and extracted with an organic ized by mass spectrometry (MS) and analytical site solvent such as dichloromethane or chloroform and exclusion chromatography (SEC) using polystrene or are required for tracking the molecular weight increase Poly(p-phenylene) as internal standard, although the by mass spectrometry [31]. Matrix-assisted laser des- molecular weight values derived from SEC analyses orption/ionization-time of flight (MALDI-TOF) [31] are only rough estimations and the absolute molecu- using tetracyanoquinodimethane (TCNQ) as matrix is lar weights may be obtained by laser light scattering suitable for GNRs. experiments [32]. 10 2D Mater. 7 (2020) 022001 C Backes et al Nevertheless, the SEC data are useful for qualita- In a typical procedure, the precursor polymer is tive comparison of the molecular weights of different dissolved in unstabilized dichloromethane (DCM), polymer samples and a crucial indicator for the result- which is saturated with Ar by bubbling for 15 min. It is ing GNR’s length. At this stage, it is recommended to recommended to apply a continuous DCM saturated narrow the broad molecular weight distribution by gel Ar stream through the reaction chamber. As a starting permeation chromatography or centrifugation into point for novel systems, usually six FeCl3 per hydrogen fractions of a lower polydispersity index (<1.5). UV– to be removed are recommended as oxidant. The FeCl3 vis absorption spectroscopy can also provide a qualita- oxidant is added as suspension (~100 mg per ml) in tive analysis of the polymer length since the wavelength nitromethane. of maximum absorption will red-shift with extension Samples can be taken in sequential time frames of of conjugation length. 15 min to days and analyzed after quenching of metha- In contrast to the Suzuki and Yamamoto reaction, nol. Due to the decreased solubility of the planarized the Diels Alder reaction does not require a metal cata- GNR compared to precursor polymers, it is recom- lyst and can be performed only by the thermal treat- mended to use MALDI-TOF MS, as well as absorption ment of both monomers [32].This is typically con- and Fourier transform infrared spectroscopy (FTIR) ducted in in either in diphenylether as solvent (reflux, for both qualitative and quantitative verification of 20–28) or in the pure melt of monomer at T ~ 260 °C– the degree of cyclodehydrogenation. One of the most 270 °C during 5 h. However, the constant solubility of dominant side reactions is the formation of chlorin- the propagating chain must be ensured similar to the ated species. The amount of chlorination can be con- Suzuki polymerization. trolled by the amount of FeCl3 equivalents (6–12), as well as the reaction time, from minutes to days. We I.1.3. Preparation of precursor polymers by AA-type conduct the reaction generally at RT. polymerization AA-type Yamamoto polymerization (see figure I.1) I.1.6. On-surface synthesis of GRNs is unrestricted by the stoichiometry problem and is On-surface bottom-up use of specifically designed thus easier to handle than A2B2-type polymerization molecular precursor monomers that carry the full methods [33, 34]. Furthermore, it is a highly efficient structural information of the final GNR together with even in sterically demanding systems [35, 36] leaving groups that can be activated on the surface, so which can improve the molecular weights (MW  =   that the target structure is built by establishing covalent 52 000 g mol−1, Mn  =  44 000 g mol−1) of the resulting bonds between activated sites of adjacent precursor polyphenylene precursors over the ones obtained by monomers. By this approach, selective growth of the Suzuki reaction. a single type of GNR is possible [21], and depends The catalytic Ni(0) is not as stable as the Pd(0) solely on the choice of the precursor monomer and an derivative. Therefore, for the preparation of the activation protocol that triggers the surface-assisted ­reaction mixture (Ni(COD)2, COD, bipy in THF), reaction steps under optimized conditions. Advances precaution in avoiding both oxygen and light must in GNR fabrication and characterization are described be taken. As a general rule the active catalyst system is in Ref. [37]. deep purple. We observe that it will quickly turn dark in contact with traces of oxygen. I.1.7. Synthesis The bottom-up synthesis of GNRs on surfaces I.1.4. Preparation of precursor polymers by AB-type critically depends on the atomic perfection of the polymerization used precursor monomers as well as control over the Another effective way to overcome the issues of surface-assisted synthesis steps [21]. In the case of 1d A2B2-type polymerisations and the labile Ni(0) target structures (such as for the case of GNRs) this catalysts is to take advantage of AB-type reactions. AB- is even more pronounced, since any defect changes polymerizations for the solution synthesis of GNRs are the electronic properties or may act as a growth established for Diels-Alder and Suzuki protocols (see stopper. It is therefore crucial to start with ultrapure figure I.1). Theyovercome the issue of stoichiometry precursor monomer samples so that undesired and result in precursor polymers with high molecular coupling configurations arising from contaminations weight. are minimized. We find that purity judged from nuclear magnetic resonance (NMR) spectroscopy is I.1.5. Cyclodehydrogenation of precursor polymer into not sufficient in order to guarantee the lowest defect GNRs density and maximum GNR length, so that precursor The cyclodehydrogenation of precursor polymers monomer samples need to be further purified with up usually follows a similar protocol. The Scholl-reaction, to eight recrystallization steps. Similarly important is an oxidative cyclodehydrogenation using Iron(III) the preparation of ‘clean’ growth substrates: Au(1 1 1) chloride as both oxidant and Lewis acid is the most single crystals were used or Au thin films on mica used. The handling of the reaction is similar for a broad [1725]. The subsequent surface-assisted synthesis variety of GNRs (see figure I.1). steps are schematically depicted in figure I.2. 11 2D Mater. 7 (2020) 022001 C Backes et al Figure I.2. GNR bottom-up synthesis concept. GNR synthesis is achieved by deposition of halogen-substituted precursor monomers at a substrate temperature T0, followed by their activation (halogen cleavage) at T1, polymerization at T2, and cyclodehydrogenation at T3. All steps are accomplished at a pressure below found [39], as well as the formation of GNRs of mul- 2 · 10−9 mbar while heating the sample to a specific T. tiple widths related to partial edge dehydrogenation of In the first step, precursor monomers are deposited on GNRs, which triggers GNR fusion (cross-dehydrogen- a clean substrate held at T0. Quartz crucibles that are ative coupling) to form seamless higher-order GNRs resistively heated up to the T needed are used for main- [39]. After cyclodehydrogenation, the sample is cooled taining a precursor monomer flux of 0.1 nm per min- to RT and either transferred to a connected scanning ute at the sample position, as determined with a quartz tunneling microscope for in situ characterization (see microbalance. The sample temperature is then raised section IX.1) or directly taken out of the UHV cham- to T1 for the halogen cleavage (activation) and to T2 for ber for characterization and/or further processing the polymerization of the activated precursors. Finally, under ambient conditions. GNRs are formed by triggering cyclodehydrogenation The above mentioned parameters are valid for the of the polymers by heating the growth substrate to T3 growth of GNRs on Au(1 1 1), for which the highest [37]. While each of these steps is crucial for GNR syn- quality is achieved for all reported GNR types [1726]. thesis, not all of the intermediate products are easily GNR synthesis under less stringent conditions was accessible for structural characterization. E.g. mono- also reported [40]. Using a CVD setup, the synthesis of mer activation at T1 ideally leads to doubly activated Chevron GNRs [41] and 9-AGNRs [42] were achieved. precursor monomers (biradicals) that coexist with the An overview on published GNR structures is given cleaved halogens at the surface. Practically, however, in figure I.3. The most frequently used halogen atom this phase is often not accessible because the activated is Br. The two main reasons for using Br is its better species frequently undergo polymerization directly at synthetic accessibility for most of the precursor mono- these T [21]. mers (as compared to I) and lower reactivity with the This implies that the activation barrier related to growth substrates (as compared to Cl [43]). diffusion and covalent bond formation between the Depending on the choice of the precursor mono- biradical species is smaller or equal to the energy bar- mer, the selective synthesis of AGNRs with differ- rier for halogen bond cleavage [38]. Deposition, activa- ent width and bandgap was achieved [21, 44, 48, 49]. tion, and polymerization steps can be combined into a On-surface synthesis of ZGNRs was demonstrated single step by depositing precursor monomers directly for 6-ZGNRs using a methyl-based monomer design at the polymerization temperature T2  =  450 K. The [57]. Chiral GNRs with a controlled sequence of arm- time for this combined step is the deposition time chair and zigzag segments along the GNR edge were (1–10 min, depending on target GNR coverage) plus made based on halogen-substitution of bianthryl [58]. 15 min holding time [37]. This step is followed by GNRs with a controlled width-modulation along the the cyclodehydrogenation step, which is triggered by axis, the so-called Chevron GNRs, were synthesized increasing the temperature to T3  =  630 K and holding based on dibromo-substituted 1,2,3,4-tetraphenyltri- it for 15 min. It is crucial not to exceed this T in order phenylene [21]. This monomer design in particular, is to avoid further activation of the formed GNRs. For well accessible for substitutional doping [54, 55, 1727]. higher T3  ≈  660 K covalent crosslinking of GNRs was Since chemical doping is achieved at the level of mon- 12 2D Mater. 7 (2020) 022001 C Backes et al Figure I.3. Overview of bottom-up synthesized GNRs. X marks the leaving group, which is typically a halogen atom (X  =  Br, I, Cl). References are: 5-AGNR [44–47], 7-AGNR: [21], 9-AGNR: [48], 13-AGNR: [49], S-13-AGNR: [50], B2-7-AGNR: [51, 52], Chevron: [21], N2-Chevron: [53, 54], N4-Chevron: [55], NH-Chevron: [56], 6-ZGNR: [57], 6-ZGNR  +  : [57], (3,1)-ChGNR: [58], (4,1)-ChGNR. omer synthesis, a fully controlled level of doping and (figure I.2) after a specific step and, subsequently, periodic arrangement of the doping sites is reached in transferring the substrate to a connected STM the final GNR. Chevron GNRs were fabricated with a chamber. Beside coverage determination, STM was number of nitrogen-substitution patterns [54–56]. For used for the determination of polymer length after the 7-AGNRs, substitutional boron-doping was achieved polymerization step as well as for the determination allowing to controllably modify the electronic proper- of possible undesired coupling motifs that can occur ties [51, 52]. The combination of different precursors by either not entirely purified precursor monomer during the polymerization step gives access to different batches or a not fully selective monomer design, which GNR heterostructures with additional opportunities, potentially allows for covalent coupling configurations making specific electronic properties at the interfaces that are not compatible with the envisaged final of dissimilar GNR segments available [48, 59–62]. GNR structure. With the exception of 5-AGNRs, Width-modulated GNRs were prepared, allowing to polymerization of the activated precursor monomers achieve distinct topological quantum phases [63, 64]. yields structures where not all molecular subunits are planar with respect to the substrate surface [21, 48]. I.1.8. GNR characterization The related apparent height imaged by STM is for all The main method applied for developing new bottom- monomers above 0.25 nm, higher than the apparent up synthesized GNR structures is in situ scanning height of the final GNR structure (~0.19 nm). tunneling microscopy (STM, see section IX.1). It Using this sensitivity, STM allows for a direct access allows accessing the growth at the surface-related to the onset T of the cyclodehydrogenation step synthesis steps by interrupting the growth protocol by identifying polymer segments, where lowered 13 2D Mater. 7 (2020) 022001 C Backes et al Figure I.4. Characterization of 7-AGNRs on their growth substrate and after substrate transfer. (a) Large-scale STM topography image (bias:  −0.5 V; current: 5 pA; T: 4.5 K; scale bar: 30 nm) showing high coverage 7-AGNRs on Au(1 1 1). Inset: small-scale STM image (bias: 0.1 V; current: 30 pA; scale bar: 3 nm). The apparent height of individual 7-AGNRs is 0.19 nm. (b) Constant-height nc-AFM frequency shift image (bias: 2 mV; oscillation amplitude: 0.3 Å; T: 4.5 K; scale bar: 1 nm) taken with a CO-functionalized tip. (c) Raman spectra of 7-AGNR on Au(1 1 1) recorded under ambient conditions after synthesis (green curve) with indicated peak positions of the main modes, and after transfer to Si/SiO2 (black curve; laser 532 nm, power of 2 mW, three scans of 20 s). (d) RBLM peak position for 5-AGNRs, 7-AGNRs and 9-AGNRs (red markers), together with predicted width-dependent RBLM wavenumbers for AGNRs [65]. apparent height indicates the related planarization of precursor monomers and the polymer intermediates, the polymer to the final GNR. For the investigation are hardly accessible by nc-AFM due to the related of individual GNRs, T below ~25 K are needed to ‘constant height’ imaging mode [66]. suppress their mobility on the surface (figure I.4). The main ex situ characterization tool applied An even higher resolution is achieved by using to GNRs is Raman spectroscopy (see section IX.2/ non-contact atomic force microscopy (nc-AFM) Raman) [1714]. Owing to their atomically defined where the tip apex can be decorated with specific structure, GNRs present well defined Raman peaks molecules or atoms to yield unprecedented insight (figure I.4(c)). The main peaks are the G and D into the chemical structure of the synthesized carbon modes with overtones and the width-dependent nanostructures [66]. Tungsten tips attached to a tun- radial-breathing-like mode (RBLM) (~396 cm−1 for ing fork sensor [67] have been used in a low-T STM 7-AGNRs) [40]. The D peak is intrinsic for GNRs (ScientaOmicron) functionalized with CO molecules due to the presence of edges [1715]. For laser energies by dosing CO onto the surface and a controlled pick- ranging from 457 to 633 nm, between 1–2 mW power up procedure [68]. By recording the frequency shift were used (higher power can result in thermal effects image at constant height (figure I.4(b)), the chemical at ambient conditions [1725]). For laser energies in structure of GNRs can be visualized with a resolution the infrared range (785 nm) a max of 10 mW is used that is going down to individual chemical bonds. The [1725]. It is important to select the photon energy of sensitivity is high enough to resolve, e.g. addition- the laser source close to the GNR bandgap to be in ally attached hydrogens at ZGNR edges (H2- instead resonance [48]. For 7-AGNRs with an optical band of H-termination) [57]. With this unique structural gap of 1.9 eV [69], Raman spectra were recorded sensitivity, nc-AFM is complementary to STM (see using a green laser (532 nm, 2.33 eV). For 9-AGNRs section IX.1), which is only indirectly sensitive to the with an optical band gap of 1.1 eV [70], Raman spec- atomic composition by recording an apparent struc- tra were recorded with an infrared laser line (785 nm, ture defined by the local density of states near the 1.55 eV). Exchanging the two laser lines for both Fermi level. The constant-height imaging mode, how- examples leads to a loss of resonance conditions ever, can only be applied to the flat final GNR struc- and the width-characteristic RBLM mode cannot be tures [48]. Non-planar structures, such as the used resolved anymore [1725]. 14 2D Mater. 7 (2020) 022001 C Backes et al I.1.9. GNR transfer Al2O3, glass slides and TEM grids (lacey carbon sup- The on-surface synthesis of GNRs relies on the catalytic ported graphene, TedPella.com). The main charac- action of the metallic growth substrate, and therefore terization tool used to prove that the structure of the requires their transfer onto dielectric substrates for GNRs remains intact upon transfer was Raman spectr­ further characterization of their electronic or optical oscopy. An example of Raman spectra taken before properties. Similarly, in order to explore GNRs as and after transfer of 7-AGNRs is in figure I.4(c)). All active materials in electronic or optoelectronic devices, the characteristic 7-AGNR modes are present in the their transfer to dielectric substrates is required. The spectrum taken after transfer. Most importantly, the potential of GNR-based devices was demonstrated by RBLM at 397 cm−1 remains equally intense and does the realization of GNR-based field effect transistors not show significant broadening. (FETs) with an on current of 1 µA and on/off ratio of 105 [71, 72]. Furthermore, devices using CVD-GNRs I.2. Graphene and carbon nanomembranes as active material were reported with high on/off ratio CNMs with tunable properties can be produced by and high photoresponsivity of 5  ×  105 A W−1 [41, 42]. conversion of aromatic self-assembled monolayers Transfer was achieved using a sacrificial PMMA layer (SAMs), CNMs can be then subsequently converted deposited onto the as-grown GNRs [72], stripped off into GRM and graphene [73–80]. The process is after GNR transfer to the target substrate. A transfer schematically shown in figure I.5(a). First, a SAM method was developed allowing for the transfer is formed on a solid substrate, then the monolayer is of GNRs without PMMA, which avoids possible converted into a CNM [81] via electron irradiation contamination problems related to PMMA residues [82], and finally the CNM is transformed into GRM [40, 55]. This relies on is delamination of Mica from via annealing in vacuum (pyrolysis) or in an inert a Mica/Au/GNR stack in an hydrochloric acid, from atmosphere. By tuning the structure of molecular which the remaining Au/GNR stack is picked up precursors (figure I.5(b)), parameters of the self- with the target substrate. In a last step, the Au layer is assembly, substrate materials, electron irradiation then dissolved leaving a clean GNR film (without Au and annealing conditions, GRM with adjustable or iodine residues) on the target substrate. To do so, crystallinity, thickness, porosity and electronic the Mica/GNR/Au stack was placed on concentrated properties can be produced. (38%) hydrochloric acid (HCl) with the Au/GNR film facing up. After 15–20 min the mica detaches from I.2.1. Molecular self-assembly Au. Once the mica is detached, the acid is removed Molecular self-assembly of aromatic molecules can be from the container. In order to prevent sticking of the performed from solvents [73, 83] or by vapor phase Au film at the container walls, it is important to not deposition [84, 85]. For the self-assembly on coinage remove the acid completely in one step. After initially metal substrates such as Cu, Ag or Au, typically thiol removing ¾ of the acid, water is added and the liquid functional groups are employed providing a covalent removed in five iterative steps. In the last dilution step, binding of the molecules to the substrate [86]. The the container is kept full of water for the pickup of the solvent-based self-assembly of 1,1′-biphenyl-4-thiol Au film with the target substrate. It is important that (BPT, (1a), in figure I.5(b)), an aromatic precursor used the target substrate is free of any impurities. for the production of SLG [74, 79] on Au substrates, With the help of tweezers, the target substrate is can be conducted by immersing the substrate into a approached (facing down) and pressed against the solution of BPT in dimethylformamide (DMF) [85]. floating Au/GNR film until they stick together. The Alternatively, a BPT SAM can be prepared by vapor merged Au/GNR/substrate stack is then pulled out deposition, i.e. exposure of sputter-clean Au substrates of the water. At this stage, the Au film is usually not to BPT vapor [85]. Both methods allow the formation completely flat on the substrate. In order to increase of BPT SAMs with comparable structural quality, the contact between Au film and substrate, 1–2 drops although the solvent-based preparation typically of pure ethanol are added and dried under ambient results in a slightly higher (~5%) packing density of conditions. Once dried, the Au/GNR/substrate stack the formed SAMs. In case of the self-assembly from is further heated on a hot plate (100 °C, 10 min). This solvents the solvent/molecule interactions play an two-step process results in a flattening of the Au film important role [86], and the packing density of the on top of the target substrate. The Au film was then formed SAMs can be tuned by adjusting the solvent removed by adding 1–2 drops of Au etchant (KI/I2, no polarity and concentration of the precursor molecules dilution) on top of the Au film and waiting until the [73]. In comparison to solvent-based self-assembly, Au film was completely etched away (around 5 min). self-assembly by vapor deposition requires high Finally, to clean the GNRs/substrate, this was soaked vacuum equipment, but SAM formation is achieved in water (5–10 min), rinsed with 20–30 ml of acetone/ in a significantly shorter time (~1 h compared to 3 ethanol and water and finally dried under N2 flux days [74, 84, 85]. For practical reasons, these different [1725]. aspects have to be considered when designing the Using this method, 5-AGNRs, 7-AGNRs, 9-AGNRs experiment. Furthermore, vacuum vapor deposition and Chevron GNRs were transferred to SiO2, CaF2, is preferred over solution deposition in the case of the 15 2D Mater. 7 (2020) 022001 C Backes et al Figure I.5. Formation of CNMs and GRM from aromatic self-assembled monolayers. (a) Schematic illustration of the fabrication route for CNMs and graphene: Self-assembled monolayers are (i) prepared on a substrate then (ii) crosslinked by electron irradiation to form CNMs of monomolecular thickness. The CNMs are (iii) released from the underlying substrate, and further annealing at 900 °C transforms them into graphene. (b) Chemical structures of the different precursor molecules used for synthesis of CNMs and graphene. Figure adapted from Ref. [73], copyright American Chemical Society. formation of thiol-based SAMs on oxidative metal with partial desorption of carbon in purely aromatic substrates (e.g. Cu or Ni), as the metal oxidation which SAMs as observed by XPS. The XPS data also show hinders the self-assembly is avoided [84, 86]. that the irradiation and the subsequent molecular reorganization also affects the S-Au bonds. These I.2.2. Electron irradiation induced crosslinking of aro- structural changes at the molecule/substrate interface matic SAMs and formation of CNMs are in agreement with the LEED and STM results Electron irradiation of aromatic SAMs results in showing a loss of the long range (>5 nm) order in the lateral crosslinking of the constituting molecules and SAMs upon electron irradiation [74]. formation of CNMs [73, 82, 87]. The mechanisms Complete crosslinking of aromatic SAMs can be of this process are discussed in Refs. [82] and [88]. also achieved via He+ ion irradiation [94] with expo- Here, we summarize essential features of this process sure dose ~1 mC cm−2, ~60 times smaller than the [89–91]. In aliphatic SAMs, electron irradiation corre­sponding electron irradiation dose [82]. Most results in significant (up to 80%–90%) molecular likely, this effect is due to the energy distribution of decomposition and desorption [92, 93]. In contrast, secondary electrons that have a maximum at energies in aromatic SAMs the same treatment a thin carbon below 50 eV, which results in a more efficient dissocia- layer is formed. To induce the crosslinking in aromatic tive electron attachment (DEA) process. Crosslinking SAMs with low-energy electrons (50–100 eV), typically can also be achieved employing UV/EUV [83] and doses of ~50 mC cm−2, corresponding to ~750 higher energy electrons (few keV) [95]. electrons per molecule, are used. Because of electron After electron irradiation of the aromatic SAMs, irradiation, C–H cleavage takes place [90], which the CNMs can be separated from the original sub- is the predominant process leading to crosslinking strates and transferred using PMMA assisted transfer between adjacent aromatic rings. As suggested by onto new solid or perforated substrates (e.g. grids, see UV photoelectron spectroscopy, quantum chemical section VI) [78], where they form large free-standing and molecular dynamics calculations of BPT SAMs areas (up to 0.3 mm2). Figure I.6 shows helium ion on gold, the formation of single- and double-links microscope (HIM) images of free-standing CNMs (C–C bonds) between phenyl rings of the molecules from different types of aromatic molecular precur- is expected during crosslinking [82]. This is supported sors [73]. These images show unbroken membranes, by UV–vis spectroscopy of the formed CNMs [76]. which demonstrate mechanically stable CNMs. Molecular dynamic calculations suggest [91] that CNMs with large free-standing areas up to ~0.3 mm2 a partial dissociation of the aromatic rings can take can be obtained in this way. Since the thickness of place and play a role in the formation of a carbon CNMs is determined by the size of the precursor mol- network. Such a mechanism would be in agreement ecules and their packing in the SAMs, it can be tailored 16 2D Mater. 7 (2020) 022001 C Backes et al Figure I.6. Helium ion microscope (HIM) images of free-standing CNMs. After crosslinking the nanomembranes aretransferred onto TEM grids and images were taken at different magnifications (see scale bar). CNMs are prepared from: (a) Naphtalene-2- thiol (NPTH, (2a)); (b) 3-(biphenyl-4-yl)propane-1-thiol (BP3, (2b)). (c) 2-Cyano-11-(1′-[4′-(S-Acetylthiomethyl)phenyl] acetyl)-5,8,14,17-tetra(3′,7′-dimethyloctyl)-hexa-peri-hexabenzocoronene (HBC-CN, (3c)); (d) [1″,4′,1′,1]-Terphenyl-4-thiol (TPT, (1c)); (e) 2-Bromo-11-(1′-[4′-(S-Acetylthiomethyl)phenyl]acetyl)-5,8,14,17-tetra(3′,7′-dimethyloctyl)-hexa-peri- hexabenzocoronene (HBC-Br, (3b)); (f) S,S′-(3′,4′,5′,6′-Tetraphenyl-[1,1′:2′,1″-terphenyl]-4,4″-diyl) diethanethioate (HPB, (3a)). The upper left insets show the precursor molecules. The CNMs in (a)–(c) are suspended over Cu grids, CNMs in (d)–(f) over Cu grids with thin holey carbon film. The numbers in the lower left corners in (a)–(d) indicate the CNM thicknesses, as determined from XPS before the transfer. HIM images e and f show CNMs with nanopores, the lower insets show the respective distributions (in%) of pore diameters (in nm). Figure adapted from [73], copyright American Chemical Society. by varying these parameters enabling nanomem- large lateral dimensions of HBC and HPB molecules brane engineering. Figures I.6(a)–(c) show exam- in comparison to smaller molecular precursors (see ples of CNMs where the thinnest nanomembrane figure I.6(b)) and, in the case of HBCs, to the ten- has a thickness ~0.6 nm, while the thickest CNM is dency of the disk like molecules to intermolecular ~2.2 nm. Different CNMs have been fabricated and stacking, which reduces the ordering in the respec- were investigated by HIM [73]. A relation between tive SAMs. The average pore diameter correlates with size and shape of precursor molecules, degree of the SAM thickness and decreases from 6.4 to 3.0 nm order in its SAMs and the appearance of the CNM has when the thickness increases from 1 to 2 nm [73]. been established [73]. If the molecule form a densely packed SAM (1(a)–(c), 2(a)–(c) and (e) in fig- I.2.3. Conversion of CNMs into graphene and GRMs ure I.6(b)), the corresponding CNM is homogeneous via pyrolysis and free of holes above 1.5 nm diameter. Figure I.6(d) I.2.3.1. Formation of nanocrystalline graphene/GRMs shows a HIM image of such a CNM made from ter- CNMs are stable up to to 800 K [83], which enables phenylthiol (1c). Conversely, CNMs made from large their conversion into graphene/GRMs via pyrolysis molecules, i.e. hexabenzocoronenes (HBC, 3(b) and in vacuum or inert atmosphere [73, 78]. The (c) in figure I.6(b)) or S,S′-(3′,4′,5′,6′-tetraphenyl- crystallinity of the produced GRM can be tuned by [1,1′:2′,1″-terphenyl]-4,4″-diyl) diethanethioate the annealing conditions, i.e. T and substrate material. (HPB, 3a in figure I.6(b)), form less ordered SAMs The formation of nanocrystalline sheets by annealing and exhibit pores, as presented in figures I.6(e) and of free-standing CNMs [78] on TEM grids (figure (f). Here the dark spots are pores with diameters of I.7(a)) was reported on Au, on [79] or SiO2[76, 77]. 2–10 nm. Note that these pores have a narrow size dis- Although S is initially present in the CNMs both, XPS tribution (see histogram insets). In case of the HBC (of supported sheets) and scanning Auger microscopy precursor, the mean size of the nanopores is ~6 nm (of suspended sheets), indicate that after annealing with surface density ~9.1  ×  1014 pores m−2 The more above 800 K all sheets consist only of carbon [78, 82]. compact HPB precursor shows a size ~2.4 nm with a At this T, the structural transformation of CNMs sets surface density `1.3  ×  1015 pores m−2. The formation in, evident from appearance of the characteristic D-, of nanopores in these CNMs can be attributed to the G- and 2D peaks in the Raman spectra [96] [77–79]. 17 2D Mater. 7 (2020) 022001 C Backes et al Figure I.7. Conversion of 1,1′-biphenyl-4-thiols (BPT) CNMs into nanocrystalline GRM upon annealing. (a) HIM image of BPT CNM annealed to 1000 K on a AuTEM grid. (b) Atomic structure of a similar sample obtained by aberration-corrected high- resolution TEM (AC-HRTEM, 80 kV). The inset shows a magnified grain boundary where arrangements of carbon atoms into pentagons and heptagons are highlighted. (c) RT sheet resistivity of the samples as a function of annealing T. (d) T dependencies of the normalized sheet conductivity demonstrate the change in electrical transport mechanism, i.e. insulator to metal transition. Figure (b) adapted from [73], figures (c) and (d) adapted from [79], copyright American Chemical Society. This can also be visualized by HRTEM [73, 79]. As electric field effect were studied in Hall-bar devices shown in figure I.7(b), after annealing of a BPT CNM [79]. Samples with lower annealing T (900, 950 and (chemical formula (1a) in figure I.5(b)), most of the 1050 K), (1–3) in figure I.7(d), have insulating behav- sheet area (~70%) is SLG recognized by the hexagonal ior with a positive curvature in T. Their T depend­   1  arrangement of carbon atoms. Randomly oriented ence can be described by σ (T) ∝ exp − TT0 3 , nanocrystalline graphene domains are connected with each other via heptagon-pentagon grain boundaries representing the Mott law [79], characteristic of ther- [97] (see inset to figure I.7(b)); a small fraction mally activated variable range hopping in a 2d system (~20%) of the sheet consists of bilayer graphene (BLG) with weak Coulomb interactions [79]. For the sam- as indicated by the Moiré pattern [98] and some of the ple with the highest degree of transformation, (5) in area (~10%) shows an amorphous carbon phase [99]. figure I.7(d), σ(T) shows a negative curvature with The conversion of CNMs influences the electrical σ(T) ∝ T1/2, i.e. a semi-metallic state. Since the con- and optical properties [76, 78, 79]. Figure I.7(c) shows ductivity of nanocrystalline carbon in the insulat- the sheet resistivity of BPT CNMs as a function of ing regime strongly depends on the density of states, annealing T. The measurements were conducted at RT a large ambipolar electric field effect is observed. The by different methods after respective annealing steps electron mobility in the resulting nanocrystalline car- on both supported and suspended GRM [78, 79]. The bon film~50 cm2 V−1 s−1 at RT [79]. The evolution non-annealed CNMs do not show lateral conductivity. of the transport characteristics upon conversion of Electrical conductivity is first detected after annealing CNMs into nanocrystalline carbon is reminiscent of at ~800 K. After annealing a ~1200 K the conductiv- that observed upon the thermal reduction of graphene ity increases by six orders of magnitude approaching oxide (GO) into reduced graphene oxide (RGO) [100, ~10 kOhm sq−1. To characterize the influence of this 101]. In both cases the final material presents an inter- transformation on electrical transport, the T depend- connected network of graphen nanocrystallites. How- encies of the electrical conductivity, σ(T), and the ever, in case of the RGO, a residual amount of oxygen 18 2D Mater. 7 (2020) 022001 C Backes et al Figure I.8. Conversion of BPT CNMs into graphene on Cu foils. (a) Raman spectra (λexc  =  532 nm) of the conversion of BPT CNMs as a function of T. The sheets after annealing were transferred from Cu onto Si wafers with 300 nm SiO2. (b) HRTEM micrograph of the sheet resolves the honeycomb lattice of graphene. The single layer nature can be determined from the HRTEM image contrast. It is further verified by the SAEDin (c). (d) RT resistivity in vacuum as a function of back-gate voltage using Hall bar devices schematically depicted in (e). (f) Quantum Hall effect at 0.3 K and 15 T. The upper plot shows Shubnikov–de Haas oscillations with the corresponding filling factors and the lower plot shows the Hall resistance as a function of back gate voltage, i.e. varied charge carrier density. Figure adapted from [74], © Wiley-VCH Verlag GmbH & Co. KGaA. containing groups is present [102], whereas nanocrys- as in this case, GRM sheets doped by these elements talline GRM sheets obtained by the conversion BPT can be expected, which are of interest for applications CNMs consist only of carbon [73, 78, 79]. in catalysis [103] or energy storage [104]. The thickness of the formed nanocrystalline sheets depends on the structure of precursor molecules, I.2.3.2. Formation of polycrystalline graphene their ability to form SAMs and to be crosslinked into Employing the conversion of CNMs into GRM by CNMs. Thus, by varying precursors (see figure I.5(b)), performing pyrolysis on catalytically active substrates the thickness of the formed nanocrystalline sheets can like copper, graphene layers with high crystallinity be tuned by a factor ~3 [73]. The resistivity correlates and, therefore, high mobility above 2000 cm2 with the thickness of the sheets, with lower resistivity V−1 s−1 can be attained [74]. As a model system, the for thicker sheets [73, 76]. An interesting opportunity conversion of BPT SAMs is presented (see chemical is opened by using N- or boron-containing precursors, formula (1a) in figure I.5(b)) into graphene on 19 2D Mater. 7 (2020) 022001 C Backes et al Figure I.9. Direct growth of patterned graphene. (a) Schematic of direct growth of patterned graphene. (b) SEM image of graphene microstructures directly written on top of Cu by irradiating a BPT SAM with a focused electron beam (3 keV, 75 mC cm−2 (left)—125 mC cm−2 (right)) and subsequent annealing (800 °C). (c) Same structure after transfer to a SiO2/Si wafer. Figure adapted from [80], © Wiley-VCH Verlag GmbH & Co. KGaA. copper foils. The Raman spectroscopy data (see cm−2, is ~1600 cm2 V−1 s−1. µ values at lower charge figure I.8(a)) show an evolution of the characteristic carrier concentrations are ~2300 cm2 V−1 s−1. Further D, G and 2D Raman peaks as a function of T (see characterization of the transport properties (see sec- section IX.2/Raman for an introduction to the tion IX.3) at low T (T  =  0.3 K) in a magnetic field of 15 technique). The conversion of a CNM into graphene T demonstrates that by varying the charge density with with T is clearly observed. For the highest annealing the back-gate voltage, Shubnikov–de Haas oscillations T (830 °C), the same features as known for single- and resistivity plateaus of the quantum Hall effect spe- layer graphene prepared by mechanical exfoliation cific for SLG are observed [107]. These results confirm (G peak at 1587 cm−1 and a narrow Lorentzian 2D the electronic quality of the grown sheets, l making peak at 2680 cm−1 (FWHM  =  24 cm−1) [105] are them attractive for electronic applications. observed after the conversion. The grown sheets were transferred onto grids and onto oxidized highly doped I.2.3.3. Direct growth of graphenemicropatterns Si-wafers and were characterized by HRTEM and Since only the electron-beam irradiated areas of SAMs by electric transport [74]. The HRTEM and selected undergo conversion into graphen, both large-area area diffraction (SEAD) data confirm formation up to 10 cm2 and more graphene sheets and carbon of SLG [106], figures I.8(b) and (c). The dark-field films of various architectures (e.g. nano-ribbon, dot, TEM imaging shows that the formed sheets consist of anti-dot patterns) can be fabricated from SAMs by graphene single crystals with lateral dimensions up to employing either defocused electron flood exposures 1–2 µm [74]. [74, 78] or exposures by focused electron beams [75]. Electric transport properties of the synthesized Because only electron-irradiated (crosslinked) regions graphene films were studied by four-point meas- of aromatic SAMs are converted into grapheneupon urements in the Hall bar geometry (see inset in annealing, whereas non-irradiated (pristine) SAMs figures I.8(d) and (e)) [74]. Figure I.8(d) presents desorb from the surface [83], the suggested approach the observed ambipolar nature as measured in a FET provides an opportunity to directly grow graphene device. The RT charge carrier mobility, µ, extracted patterns by area-selective electron irradiation [80], from the data at a hole-concentration of 1  ×  1012 figure I.9(a). Therewith, at least four technological 20 2D Mater. 7 (2020) 022001 C Backes et al Figure I.10. Hybrid LMH of 0d and 2d carbons. (a)–(e) Schematic representation of the LMH assembly. (a) Formation of a NBPT SAM on Au. (b) Electron irradiation induced crosslinking and reduction of the terminal nitro groups into amino groups. (c) Formation of a free-standing JNM with the terminal N- and S-faces. (d) Functionalization of the N- and S-faces with C60 and AuNPs. (e) Assembly of a (C60–JNM)n (here n  =  3) hybrid heterostructure by mechanical stacking. Color code for atoms: black—C, grey—H, blue—N, green—S, and red—O. (f) HIM images in the scanning transmission ion mode shows the immobilization of 16 nm Au NPs on a JNM which are uniformly distributed with an coverage ~50%. (g) HIM image of an JNM- (C60–JNM)3 heterostructure spanning a Si window. (h) Young’s moduli of JNM, C60–JNM and JNM-(C60–JNM)n (n  =  1, 2, and 3). Figure adapted from [113], with permission of The Royal Society of Chemistry. steps (spin-coating of photoresist, developing non-planar surfaces, thus it is also feasible to apply the of photoresist, reactive ion etching, stripping developed methodology to create graphene structures of photoresist) employed in the conventional on any 3d shape. microfabrication of graphene electronic devices are omitted. Figure I.9(b) shows a graphene pattern grown I.3. Heterostructures from CNMs on Cu at 800 °C after a BPT SAM is locally crosslinked Stacking of 2d sheets including SLG, hBN, or by a primary electron beam of 3 keV and doses ranging TMDsinto layered materials heterostructures (LMHs) from 75 mC cm−2 to 125 mC cm−2. The transfer have led to novel materials with a potential for of this structure onto a 300 nm SiO2/Si substrate is applications and in fundamental research [111, 112]. presented in figure I.9(c). By comparing the distances The integration of other low dimensional materials into between the single structure elements in figure I.9(b) these LMHs can extend these borders even further. Here (on Cu) and 9(c) (on SiO2/Si), the structural integrity a modular and broadly applicable route is presented is conserved during transfer. Only a few elements to create hybrid LMHs made of individual ~1 nm thick show folding defects or missing parts. In this case, Janus CNMs (JCNM) [113] functionalized with other the local adhesion to the substrate is too weak to low-dimensional materials (see figure I.10). withstand the forces occurring during the removal of JCNMs are produced via electron irradiation the transfer medium in solvent [80]. More advanced induced crosslinking of 4′-nitro-1,1′-biphenyl-4- transfer techniques, like electrochemical delamination thiol SAMs and have different chemical groups on [108] or a clean-lifting transfer [109] may be applied their opposite faces, e.g. amino groups on the top side to avoid or minimize the creation of defects (see (N-side) and sulfur species on the lower side (S-side) section VI). The minimum size of the features in the [114]. They can be independently chemically func- grown pattern correspond to ~1 µm. The lateral tionalized with desired building blocks and assembled resolution is defined in principle by the resolution of into hybrid LMHs via stacking, figures I.10(a)–(e). electron-beam lithography, ~7 nm for SAMs [110]. C60was covalently bound to the amino groups on the Another interesting opportunity is given by the fact N-side of a JCNM [113]. To demonstrate that in the that molecular self-assembly can be conducted on assembly of hybrid heterostructures also the S-side of 21 2D Mater. 7 (2020) 022001 C Backes et al JCNMs can be used, it was functionalized with Au NPs, characterize their mechanical properties. To this end, figure I.10(f). The possibility of bifacial chemical func- the sheets were transferred onto a Si substrate with an tionalization of JCNMs paves the way to hybrid LMH- array of square shaped orifices. Figure I.10(g) shows a swith a variety of other 0d and 1d materials. HIM image of a homogeneous free-standing hybrid In Ref. [113], cm2-sized heterostructure stacks structure of three layers C60/JCNM spanning over an of hybrids JNCM with coupled C60 were fabricated. orifice with dimensions of 40  ×  44 µm2. The LMH can They were characterized with respect to their struc- support its own weight and preserves its mechanical tural, chemical and mechanical properties. The char- integrity. The Young’s moduli of C60/JCNM multilay- acterization by XPS (see section IX.2) shows that the ers were measured by mechanical bulge tests and are chemical composition and effective thickness of the presented in figure I.10(h). Within the accuracy of the individual C60/JCNM layers remains unaffected [113]. measurement, the Young’s moduli have similar values, Individual C60/JCNMs and their heterostructures were demonstrating that the mechanical properties are not further studied by mechanical bulge tests [133] to degraded upon the assembly of the hybrid. 22 2D Mater. 7 (2020) 022001 C Backes et al II. Top-down of oxygen functional groups introduced in the oxida- tion reaction resulting in GO [132], influencing also II.1. Precursors the sonication time required to overcome the vdW A wide variety of GRMs with different characteristics interactions [133], as well as the chemical structure can be obtained through liquid phase techniques of the RGO obtained after thermal [135] or chemical [111, 115]. For a given approach, the selection of reduction [135]. The microstructure of graphite can an appropriate precursor allows to tune the final be distinguished by polarized light microscopy [120], features and properties of the GRM products in order where the different crystalline domains are defined by to optimize their performance for each application. the interference colours (figure II.2(a). Figure II.2(b) Graphite is the reference starting material to produce shows anSEM image of graphite in which these features graphene, but it presents different characteristics like cannot be observed. The distribution of crystal size particle size, crystal size or purity, which have to be also influences the polydispersion in lateral size of the considered. E.g., graphite with small crystal size will obtained GO/RGO material, limiting the maximum limit the maximum lateral size of the final graphene lateral size. The final average size of graphene is usu- flakes. Other precursors, such as nanocarbons (i.e. ally much smaller than the initial crystal size in graph- carbon materials with nanoscale size, such as CNTs) ite, and is mainly dictated by the mechanical process of [17, 116, 117] or pre-graphitic carbons (short ranged exfoliation at the meso- and nano-scale [136, 137]. ordered amorphous carbons) [118, 119] have been GO/RGO has been prepared from different natu- also investigated as potential candidates for graphene ral graphites [138]: flaky, (natural graphite of finely production, as discussed below. powdered crystalline graphite with grain size smaller than 250 µm), lumpy (crystalline graphite with a wal- II.1.1. Graphite nut to a pea size), and amorphous (exhibiting only Graphite consists of stacked graphene layers bonded short range order) [139, 140]. Lower crystallinity of by van der Waals (vdW) forces [120]. Carbon atoms the parent material (lumpy  >  natural  >  amorphous) are hexagonally arranged and SLGs are parallel to leads to higher defects, smaller GO sheets’ size and each other. Graphite has two main allotropic forms, higher surface areas (figure II.3). Natural graphites can hexagonal [121] (figure II.1(a)) and rhombohedral contain silicates and other particles that can contami- [122] (figure II.1(b)). In both cases the carbon nate the final GRM, as detected by thermogravimetric hybridization is sp2, the C–C distance in the basal (TGA) measurements [141]. plane is 0.1417 nm, and the distance between the layers Smaller particle size will result in higher content is 0.3320 nm. The hexagonal form is the most stable, of oxygen functional groups in GO, because of easier with layers stacked in an ABAB sequence (unit cell diffusion of reagents during oxidation. This influences constants: a  =  0.2456 nm, c  =  0.6708 nm) [121]. In the final electrical and thermal properties [142, 143]. Rhombohedral graphite the sequence of the layers Graphite with particle sizes  >200 µm are adequate for is ABCABC (unit cell constants: a  =  0.2566 nm, the production of GRM to be used in thermoset com- c  =  0.100 62 nm) [122]. posites by infusion [144]. Particle size  >600 µm are Graphite can be natural or synthetic.The latter can be ideal for conductive thermoset composites [145]. obtained by subjecting nongraphitic carbons as pitches The discrete particle size of graphite, as observed and cokes to high temperatures (1700 °C–2700 °C) in SEM, is different than the particle size determined in inert atmosphere or vacuum [123–125].This is the by other techniques such as laser diffraction, opti- case of graphitizable carbons, i.e. non-graphitic car- cal microscopy or granulometry. Laser diffraction or bons which, upon thermal treatment, convert into granulometry increase the measured particle size of graphitic carbon, [126]. The degree of graphitization, graphite due to the agglomeration of discrete crys- the amount of disordered phase converted in their tals. SEM can distinguish between different or discrete graphitic counterpart, can be further increased by crystals and allows the measurement of particle size performing thermal treatment under high pressure with more accuracy. In the case of optical microscopy, (100–1000 MPa) [127]. Graphitic materials can also it is more difficult to separate particle sizes. be obtained by CVD of hydrocarbons, such as meth- Expanded graphite is a graphite-derived material ane [128] or ethane [129] at T ~ 1200 °C or by catalytic formed by a two-step oxidation-reduction process (c-CVD) and these synthetic methods are reviewed in that retains the long-range-ordered layered structure section V. of graphite, yielding a large interlayer distance. It can Graphene can be obtained by exfoliation of graph- be used for the preparation of GRM by liquid methods ite [130]. The presence of defects in the layers leads to [147]. Typical expanded graphites used for the prep­ graphene with lower conductivity [120, 131–134]. aration of graphene flakes have a medium diameter Graphite with large crystal size allows one to obtain D50, i.e. the value of the particle diameter at 50% in bigger flakes [132]. The crystal boundaries of the start- the cumulative distribution of particle size diameters, ing graphite have an impact on the amount and type from 15 to 50 µm. 23 2D Mater. 7 (2020) 022001 C Backes et al Figure II.1. (a) Hexagonal and (b) rhombohedral structures of graphite. II.1.2. Non-graphitic carbons GNRs in liquid media [17, 116, 117]. Twisted ribbon- Non-graphitic carbons were also proposed [148, 149] shaped carbon nanofibers are helical carbon nanofibers as precursors for the synthesis of GRM by top-down comprising 5–6 stacked graphene layers [117, techniques. e.g. petroleum and coal-tar pitches pass 153–155]. Two approaches have been proposed for the through a liquid or liquid-crystalline phase when production of graphene flakes and sheets from these subjected to heat treatment in inert atmosphere [153, 154]. The first is based on a mechano-chemical [150]. These materials are graphitizing and develop method, through melamine intercalation during ball a graphitic microstructure upon heating to 3000 °C milling in different solvents such as isopropanol, THF [151]. When these precursors are carbonized at lower or DMF. Ref. [153] produced graphene suspensions temperatures (1000 °C), they become cokes. A coke is in low boiling point solvents such as 2-propanol a solid carbonaceous material derived from distillation (~83 °C) and THF (66 °C) by a scalable method of low-ash, low-sulphur bituminous coal, or derived [153]. Ref. [154] obtained GO/RGO flakes® using from oil refinery coke units or other cracking processes the Hummers’ method [156] and different reduction [152]. It has a partially organized lamellar structure techniques [154, 157, 158]. resulting from the parallel stacking of layers of sp2- GNRs have been produced from CNTs as the start- hybridized carbon atoms. Therefore, Refs. [148, 149] ing material by solution-based oxidative processes suggested that these could be used for the preparation of [17, 117]. CNTs suspended in concentrated H2SO4 graphene flakes. Since high T (up to 2500 °C–3000 °C) were oxidized by adding KMnO4 (500 wt%with are required to obtain synthetic graphites, the direct respect to CNTs) at T  <  70 °C [17].The resulting use of cokes as precursors reduces the environmental GNRs have straight edges and are dispersible in water impact of the whole process. Moreover, the broad and other polar solvents due to the presence of oxygen range of microstructures and chemical compositions moieties. Unzipping single-wall carbon nanotubes of cokes (as determined by the carbon source and the (SWNTs) [17] gives GNRs, but these are difficult to dis- carbonization conditions) makes them a suitable entangle. Multi-wall carbon nanotubes (MWNTs) led starting material for the synthesis of graphene or GO to less entangled GNRs [17], that can be separated to with targeted structures and compositions. single layers after sonication. Carbonyls, carboxyls and Due to the different microstructure and chemical hydroxyl groups present in the GNRs can be reduced composition of cokes and graphites, the conditions for by treatment with aqueous hydrazine (N2H4·H2O) the synthesis of GRM should also be adjusted. Thus, after dispersing them in aqueous sodium dodecyl sul- compared to graphites, cokes require larger amounts phate in order to prevent their re-aggregation. CNT of reactants during their oxidation by the Hummers’ unzipping by intercalation of lithium and ammonia method. Otherwise, its exfoliation into GO would have a and subsequent exfoliation was also reported [116]. much lower yield [148]. When cokes and graphites were However, this procedure yields a mixture of multi- exfoliated in NMP, flakes from coke had a larger number layer GNRs, partially open MWNTs, and flakes, con- of defects and smaller lateral size (~400 nm) than those trary to oxidation with KMnO4, which allows a nearly from graphite (~1 µm) [148]. Similarly, as the crystal- 100% yield of GNRs with no presence of partially open linity of cokes increases, its subsequent exfoliation yields MWNTs [117]. flakes with larger lateral size and in higher yield [149]. II.2. Liquid phase exfoliation II.1.3. Nanocarbons Various strategies have been developed to produce CNTs or nanofibres (CNFs), have also been explored as colloidally stable GRM dispersions. They can be alternative precursors to produce graphene sheets and classified into intercalation-based methods [159–161], 24 2D Mater. 7 (2020) 022001 C Backes et al and methods relying on mechanical exfoliation and bilization and size selection is quite robust and repro- further stabilisation by suitable solvents/surfactants ducible giving access to long-term (weeks-months) (LPE) [161–164]. Oxidative intercalation of graphite stable, dispersions. In order to retain ‘pristine-like’ and subsequent delamination to yield GO are amongst optical properties, similar to those of micromechani- the oldest techniques [156]. cally cleaved (MC) materials, aqueous surfactant LPE is one of the simplest and most versatile strat- solutions are of advantage. E.g., photoluminescence egies to obtain colloidally stable dispersions from a (PL) of MoS2 and WS2 can be observed in single layer broad range of LMs. Energy in the form of sonication (1L)-enriched solutions in sodium cholate (SC) with or shear is used to overcome the van der Waals attrac- emission dominated by excitons and at similar spectral tion in bulk LM. If this is done in a suitable liquid (sol- positions as MC TMDs [185, 186]. Surfactants such as vents with matching solubility parameters, or additives SC are water soluble and do not require the same level that act as electrostatic and/or steric stabilizers [165]), of health precautions as organic solvents [187]. SC is reaggregation is prevented. The resulting flakes can compatible with many processing techniques such as have low defect densities, and can be further solution- vacuum filtration and spraying. However, printing processed by a range of techniques [166]. LPE of MoS2 from aqueous surfactant still remains challenging, and WSe2 was reported in 1989 [167]. as it is difficult to control the rheological properties While LPE is a versatile method for producing large [166]. In addition, there are two other downsides. First, quantities of flakes from bulk powders, it has some some materials (such as GaS [188] or black phospho- disadvantages. The flakes are highly polydisperse with rus [189, 190]), can degrade in the presence of water/ a range of sizes and thicknesses (lateral sizes ranging oxygen [189, 191] so that aqueous surfactant solu- from tens of nanometres to tens of microns, number tions cannot be used. Second, it is difficult to com- of layers (N) from 1 to 100 s of layers). The dispersions pletely remove the surfactant from the resulting flake can also have low monolayer content (YM  <  1%). A surfaces after processing, as it can be trapped between precise, statistical analysis of size distribution due to restacked flakes, potentially deteriorating the result- exfoliation by ultrasound has been reported, for GO ing network properties. A few systematic studies (see sheets [136]. For application areas where LPE flakes are [161–164, 166, 168–173, 192]) exist on the impact of well suited, such as printed optoelectronic and electro- the chemical structure of the surfactant or its concen- chemical devices, optimized performance is unlikely tration on the degree of exfoliation, i.e. N or yield of to arise through direct use of the pristine dispersions. 1L and few layer (FL) flakes as well as their lateral size, This is especially true for those applications requir- with detailed data available for SC or sodium dodecyl ing direct bandgap luminescence of transition metal benzene sulfonate (SDBS). However, a comprehensive dichalcogenides (TMDs). Thus, size selection must be picture is still lacking due to the vast parameter space an integral part of the production process [168]. (materials, surfactants, concentrations, exfoliation Many strategies exist for both exfoliation and size techniques, centrifugation procedure). Systematic selection, depending on the starting material and studies on the effect of less common surfactants, such the desired outcome. A number of review articles as polycyclic aromatic hydrocarbons (PAH), have also on exfoliation in liquid media have been published been reported, correlating the adsorption energy of [115, 161–164, 166, 168–170, 172, 173]. Here, questions the single molecule on graphene to the overall exfo- such as which medium (e.g. solvent or surfactant) or liation performance [193, 194]. Flakes produced in exfoliation method (e.g. sonication or shear) is more aqueous surfactants are smaller, but also thinner than suitable and how to size-select flakes using benchtop in solvents such as N-methyl-2-pyrrolidone (NMP) centrifuges are addressed. and N-Cyclohexyl-2-pyrrolidone (CHP) [174]. This is most likely related to the higher viscosity of these sol- II.2.1. Choice of medium vents compared to aqueous surfactant. This can on the II.2.1.1. Solvents and detergents one hand hinder intercalation of the solvent between In a typical LPE experiment, the bulk crystal is immersed the layers of the crystals during the exfoliation process into a liquid and subjected to sonication [131, 162, and in addition lead to a slower sedimentation in the 173–175], shear mixing [176, 177], microfluidization centrifugal field [195] and hence a larger population [178], ball milling [179] or similar [180, 181]. Prior to of larger/thicker nanosheets. Details on differences designing an experiment, thought needs to be given to in sizes/thicknesses and yields of 1L and FLstrongly the choice of medium. This can be crucial and will often depend on material, exfoliation conditions and cen- depend on the starting material in combination with trifugation conditions making it difficult to quantify the exfoliation method and the purpose of production. these effects. Irrespective of the exfoliation method and/or medium, Solvent stabilisation can be described in the frame- a purification step of the starting materials (e.g. graphite work of solubility parameters [196]. Solvents or sol- flakes) is useful, since these may contain impurities that vent blends can stabilize liquid-exfoliated flakes and can destabilize the dispersion. prevent them from reaggregating, when solubility Aqueous surfactant solutions are widely used as sta- parameters, such as surface tension or Hansen param­ bilizers in LPE [172, 182–184]. Usually, xfoliation, sta- eters, of solvent and solute match [131, 174, 197–200]. 25 2D Mater. 7 (2020) 022001 C Backes et al Chemically unstable materials, such as black phos- dispersants or graphite exfoliators, i.e. they do not trig- phorus (BP), can be protected against degradation ger the exfoliation. These molecules/polymers mainly through a stabilising solvation shell [190]. In addition, act as dispersion stabilizing agent via the non-covalent it is easier to tune rheological properties for process- functionalization, i.e. through the adsorption of their ing techniques such as inkjet printing [201] and sol- hydrophobic moieties on the graphene surface during vent removal is typically easier than for surfactants. the process of exfoliation [165]. Therefore, they can However, certain solvents, such as NMP or ODCB, prevent re-aggregation and increase the stability of can degrade and polymerize [202, 203], changing the both aqueous and organic dispersions [165]. properties of the exfoliated material, while hampering Using water as a LPE media is a natural choice subsequent spectroscopic and microscopic characteri- because of its non-toxicity and offers potential for the zation [204]. In such cases, the solvent can also remain formation of biocompatible graphene-based mat­ on the flakes surfaces. In general, good solvents show erials for biomedical applications [220]. Yet, LPE of a strong, yet noncovalent interaction with the flakes, graphene in water is particularly challenging due to the altering optical properties as, e.g. manifested by PL hydrophobic nature of the sheets. This challenge can shifts in TMDs [186]. Another consideration is that be overcome by using surfactants such as SC, which many suitable solvents, such as NMP, suffer from high allow exfoliated flakes to remain suspended. Among boiling points (>200 °C) and are often toxic [205]. various molecular stabilizers, PAHs [221, 222], substi- Some solvent blends have been identified as promising tuted with numerous side groups, are the most studied (e.g. alcohols and water) [206, 207] to potentially over- compounds. Adsorption of PAHs onto the graphene come the limitation of the high boiling point solvents. surface occurs through π–π interactions between the GRM dispersions in water can also be stabilized by planar π-conjugated surfaces. In these noncovalent using a wide range of polymers [182, 208–212], such interactions, both PAHs and flakes surfaces share the as ethyl cellulose (EC) [209], cellulose acetate (CA) π-orbital electrons, ultimately resulting in the reduc- [208], lignin [211], polyvinylpyrrolidone (PVP) [210], tion of the surface free energy of the dispersion [194]. and even more complex systems [182, 212]. The use Various derivatives have been used to stabilize CNTs of polymers as stabilizing agents in the LPE process dispersions [223], and as in the case of NMP, they have works better than the use of small organic molecules been adopted for LPE of graphite [194, 224–226]. Nev- in terms of the concentration of graphene dispersions. ertheless, not all stabilizers suitable for CNTs, having a Yet, because of the strong polymer/graphene non- curved surface, would be ideal for dispersing graphene covalent interactions, flakes produced by exploiting with a flat surface. Because of the improvement of the this approach cannot be completely separated from exfoliation yield (the amount of solubilized material the polymer/graphene composites [213]. during LPE) and, in particular, increase in the number of SL in LPE in the presence of small PAHs such as pyr- II.2.1.2. Molecule-assisted LPE enes, NDIs, [227] and PDIs [228], and other PAHs are Stable dispersions of graphene in water have also also expected to be suitable to stabilize graphene pro- been obtained using small polyaromatic dyes as duced by LPE. surfactants [194, 214, 215]. Thanks to their aromatic Regardless of the exfoliation yields (% of SL, con- core, these molecules can adsorb strongly on the centration) and the stability of graphene in aqueous graphene surface (EADS ~ 15 Kcal mol−1), forming also dispersions, the use of water as an exfoliation medium ordered layers [193]. They have also a strong (molar is not recommended for electronic devices such as absorption coefficient ε  >  15 000 l g−1 m−1) and FETs. The presence of water at the interface with the unique absorption and emission spectra in the visible dielectric substrate can enhance charge-trapping range, which allows to monitor their interaction with [229]. Therefore, the use of stabilizers in organic sol- graphene in solution and in solid [216, 217]. Many vents has also been explored [230–235]. For more of these molecules are low cost (<100 $ Kg−1) dyes, details on the underlying specific noncovalent interac- widely used in large-scale compounding of polymers, tion, see section VIII.3. e.g. as industrial additives and colorants dyes [218]. While most of the work has been done with graphene- II.2.2. Exfoliation methods where a strong interaction of the aromatic core of the II.2.2.1. General considerations dye with the extended π-electron network of the flakes The purity with respect to contaminations of occurs, these molecules can also be used to exfoliate a the commercially available materials is often wide range of LMs, such as BN [219], WS2 [219] and unpredictable, creating problems for the MoS2 [219], selenides and tellurides [219]. reproducibility in the exfoliated product. Accordingly, The use of properly selected small PAH organic the general recommendation is to perform a two-step molecules or polymers can enhance the exfoliation exfoliation process [185, 186, 236]. In the first one, of graphite, in particular when the molecules/poly- the material in the medium of choice is subjected mers possess a high (EADS ~ 15 Kcal mol−1) energy of to a short (~20% of the time of the total planned adsorption on the basal plane of graphene. However, duration) exfoliation. After this initial step, the these molecules or polymers do not act as graphene dispersion should be subjected to centrifugation at 26 2D Mater. 7 (2020) 022001 C Backes et al intermediate centrifugal acceleration (~5000 g) after size-selected dispersions) for sonication at an ampli- which the supernatant is decanted and discarded and tude of 50% compared to 60% [246]. From experi- the sediment collected in fresh solvent/surfactant. This ence, higher amplitudes can damage the processor can then be followed by a second, longer exfoliation. after prolonged (100 s of hours) use. 500 and 750 W do This approach allows most of the impurities (and very not give noticeable differences [246]. Since the sonic small flakes) to be removed. This is particularly needed energy falls off exponentially from the probe [243], it when working with ionic surfactants, as the impurities is important to match the container to the probe size as are often ionic and therefore, can destabilize the recommended by manufacturers. Even though the dis- surfactant dispersion via charge screening effects. persed concentration of FL flakes increases with soni- A higher initial concentration of the precursor cation time [174, 176, 177, 188, 190, 238–242], 5–7 h of bulk material will also give a higher concentration of sonication provides a compromise between dispersed exfoliated flakes, up to when the dispersed concentra- concentration and time. tion saturates. This initial concentration is typically It is crucial to prevent heating of the sample, espe- >30 g l−1 of precursor concentration [174, 176, 177, cially in tip sonication, but also in bath sonication, 188, 190, 237–240]. Similarly, longer processing times shear exfoliation, etc. This is not only because heat- yield higher dispersed concentrations [174, 176, 177, ing can deteriorate the properties of the material, but 188, 190, 238–242]. While this can also change the also because dispersed concentrations are lower. While length and thickness distributions in the as-obtained cooling in tip sonication can be achieved by position- dispersion [241], this effect can be balanced by appro- ing the reaction vessel, i.e. a metal cup in an ice bath, priate size-selection techniques [168]. it is far more ideal to install a chiller, as the ice must be The choice of the exfoliation method is material replenished every 2 h or so [245]. In addition, keeping dependent. E.g. while graphite is readily exfoliated for the dispersion cool is facilitated by pulsing the sonica- various methods [176, 177, 237] (sonication, shear, tion through the device controller. For a given com- microfluidization, ball milling), this may not be the bination of sonicator, operating conditions, cooling, case for other materials. In particular for TMDs, the sonication time and material, the outcome of tip soni- quality of exfoliation (as measured by the concentra- cation appears reproducible both in terms of dispersed tion/yield of 1L and 2–5 L flakes is better when using concentration and size distribution [185]. tip sonication compared shear mixing [176, 238]. For In general, bath sonication may offer a lower-cost GaS, bath sonication is most effective [188]. When alternative to tip sonication, and additionally, in some attempting to exfoliate a new material, it is important cases, such as for GaS [188], this is preferred because to not only consider the stabilizer, but also to test vari- tip sonication can damage the exfoliated flakes. The ous available exfoliation methods. In particular the energy input is lower in bath sonication [243], as it is role of the crystallite size of the starting material in less. Therefore, less material fragmentation is expected. combination with the exfoliation method has not been However, longer processing times are required investigated systematically. (depending on material) to achieve an equivalent con- centration of dispersed material compared to tip soni- II.2.2.2. Sonication cation. For TMDs, sonication times are longer by a fac- Sonication can be performed in tip or bath sonicators. tor ~20 in bath compared to tip sonication. In general, In tip sonication, energy is imparted to the dispersion bath sonication is less reproducible than tip sonication, media directly [243], whereas in the sonic bath, the and the outcome varies depending on specific bath energy must travel through the tank and dispersion used, filling level of the bath, and positioning of the vial vial before reaching the flakes [243]. Although the within the bath. While more reproducible results can scalability of sonication to produce few layer flakes be obtained by rotating the sample in the bath during has not been demonstrated (compared with shear sonication, this will result in a lower dispersed concen- exfoliation and ball milling), tip sonication remains tration for an equivalent sonication time compared to useful as lab-scale technique, as the material quantity placing vials in ‘hot spots’ (thermolytic centers, where (hundreds of mg) is generally sufficient to test size- the sonication induced cavitation bubbles of the water dependent properties in a range of application areas. in the tank preferentially collapse). Again, it is impor- For LMs, tip sonication is preferred over bath soni- tant to prevent heating of the sample (at least for non cation due to the higher production rates, (i.e. more graphene LMs) by either installing a cooling system or exfoliated material is produced in shorter times, with exchanging the water every 30 min. concentrations in the range of ~1 g l−1 in  <24 h for ini- tial bulk material concentrations of 30–50 g l−1) [174, II.2.2.3. Shear exfoliation 186, 244]. In this case, various sonic tip configurations Alternatively to sonication, one can use shear are available with different probe sizes and shapes and exfoliation in rotor stator mixers [176] or in processors. In typical protocols [185, 186, 245, 246], blenders with rotating blades [177, 238]. Even a solid flathead tips are used at sonication amplitudes kitchen blender can be used, offering a very low ~60%–75%. At lower amplitudes, exfoliation can be cost alternative to sonication [177]. However, the poorer: the mean N WS2 increased by ~1 (across eight household kitchen blenders are not designed to 27 2D Mater. 7 (2020) 022001 C Backes et al Figure II.2. (a) Optical micrograph acquired with an oil-immersion objective (20×) and an one-wave retarder to generate interference colours and (b) SEM picture of the same graphite. Graphite Oxygen functional groups in the basal plane (GO) Oxygen functional groups in the edges (GO) Sheet size, Conductivity Structural defects Crystal size Figure II.3. Schematic model of partially reduced GO and their parent graphites. operate in organic solvents and can be destroyed to a fluid forcing it to pass through a microchannel, when doing so. In addition, continuous operation figure II.5. The key advantage over other LPE methods heats up both sample and mixer, so that the run has is that high shear rates (γ̇   >  106s−1 [248, 249]) can to be paused frequently (after 15 min) and mixer be applied to the whole fluid volume [249], not just and sample cooled in an ice bath. The dispersed locally. Microfluidization was used for the production concentration of FL flakes is not strongly dependent of polymer nanosuspensions [247], in pharmaceutical on the processed volume, so that shear exfoliation applications to produce liposome nanoparticles with is a scalable process with production rates of 0.15 g diameters smaller than 80 nm to be used in eye drops h−1 for graphene and 1.3 mg min−1 for MoS2 for drug delivery to the posterior segment tissues of [176, 177, 238]. The quality of the material in terms of the eye [250], or to produce aspirin nanoemulsions dispersed concentration and mean N is comparable [251], as well as in food applications for oil-in-water to sonication in the case of graphene [177]. nanoemulsions [252]. Microfluidization was also used for the de-agglomeration and dispersion of II.2.2.4. Microfluidization CNTs [253]. Microfluidization is a homogenization method The main step of this process involves placing that applies high pressure (up to 207 MPa) [247] the graphite/solvent mixture into an inlet reservoir. 28 2D Mater. 7 (2020) 022001 C Backes et al Figure II.4. SEM images graphite 0–200 µm (>0.2 mm; 10% max) and a graphite 0–600 µm. Adapted from [146]. Figure II.5. Schematic of microfluidization process. Graphite flakes in SDC/water are added in the inlet reservoir. An intensifier pump applies high pressure (207 MPa) and forces the suspension to pass through the microchannel of the interaction chamber where intense γ̇ ~ 9.2  ×  107 s−1 is generated. The processed material is cooled down and collected from the outlet reservoir. The process can be cycled several times [178]. Adapted from [178]. An intensification pump, driven by electro-hydrau- Below are two typical recipes for the production lics, provides two motions, suction and compres- of graphitic flakes without centrifugation (Method 1), sion, drawing a portion of sample into the processor or FLGs following centrifugation (Method 2). Due to through a one-way valve, and then pushing this sample the dimensions of the interaction chamber, microflu- fraction past the pressure gauge and through sub-mil- idic processing of large (>100 µm) graphite flakes may limetre sized channels within a diamond interaction block the microfluidic channel. Instead, smaller pow- chamber, where graphite exfoliation takes place [178]. der sizes are advisable (typically, the particle diameter Interaction chambers are typically based on two corresponding to 90% cumulative (from 0 to 100%) of geometries, Y- and Z type. The Z-type chamber is the undersize particle size distribution (D90), less than preferred for graphite exfoliation, since the geom- 30 µm). etry consists of sharp turns and a narrow rectangular cross-section for optimal shear. The chamber G10Z is II.2.2.4.1. Method 1: production of multi-layered flakes ~87 µm in diameter. The motion of the intensifica- This method [178] refers to the production of high tion pump creates mixing within these micro chan- yield (100 wt.%), high concentration (up to 100 g l−1) nels, generating pressures  >  30 000 psi (~2100 bar). graphene-based inks that are not centrifuged. Graphite The fluid dynamics of the process can be turbulent or powders (natural or synthetic, particle size  <30 µm) laminar as defined by the Reynolds number, depend- are mixed at a concentration up to 100 g l−1 with SDC ent on the kinematic viscosity and density of the sol- at a concentration 9 g l−1 in water in typical batch vent used [178]. sizes of 250–500 ml. The material is collected in a 29 2D Mater. 7 (2020) 022001 C Backes et al Figure II.6. SEM images taken from coatings comprising (a) starting graphite, (b) after 5 cycles and (c) after 100 cycles. Reproduced from [178]. Figure II.7. (a) Rs and film thickness as a function of processing cycles for a formulation with 73 wt% flakes, (b) bulk and critical thickness as a function of loading (70 cycles). All samples are dried for 10 min at 100 °C. Reproduced from [178]. beaker and placed back in to the inlet reservoir to cycle (300 °C–40 min), Rs reaches 2 Ω □−1 at 25 µm (con- automatically for the desired length of time depending ductivity, σ  =  2  ×  104 S m−1), suitable for electrodes on the number of process cycles, n (where n is in devices such as OPVs [254, 255], organic thin-film typically  <70). The final product is collected. Sodium transistors (OTFTs) [256] or radio-frequency identifi- carboxymethylcellulose (NaCMC) (10 g l−1) is added cation (RF-IDs) [257]. The inks can then be deposited whilst stirring to adjust the viscosity to the required on glass and poly(ethylene terephthalate) PET flex- value. NaCMC is slowly added until fully dissolved in ible substrates using blade coating and screen print- order to avoid the formation of clumps. Clumps that ing to demonstrate the viability for these applications are formed should be ground with a spatula and the (OPVs, OTFTs, RF-IDs). Screen printing is discussed ink left to stir overnight. in section III. Figure II.6 shows SEM images of the coatings comprising the starting graphite (figure II.6(a)), II.2.2.4.2. Method 2: production of FLGs after 5 (figure II.6(b)) and 100 cycles (figure II.6(c)). Below we show that FLG can be obtained by Flake size reduction and platelet-like morphology is adding an additional centrifugation step. Graphite observed after microfluidic processing. powders (natural or synthetic) with D90  <  30 µm Since these inks are not further processed by cen- are mixed at a concentration 12 g l−1 with SDC at a trifugation, YW (yield by weight) is 100 wt.%. 4 wt% of concentration ~0.5 g l−1 in water and microfluidized the exfoliated material consists of FLGs (<4 nm thick) for n typically  <100). Following processing, samples that can be isolated using Method 2 below, and 96 wt% are centrifuged typically at 10 000 rpm (17 000 g) for are flakes in the 4 to 70 nm thickness range. The sta- one hour. Films obtained by vacuum filtration of the bilized dispersion can be used for blade coating and FLG ink have Rs ~ 4.5 kΩ □−1 at 90 nm thickness, screen printing. Figure II.7 plots the sheet resistance significantly better than the equivalent sonication (Rs) and thickness of blade coated films as a function produced graphene films that have Rs ~ 8.9 kΩ □−1 of process cycles at a flake loading of 80 g l−1 (73 wt.%), at 160 nm. The effects of processing cycles, pressure, highlighting that exfoliated materials produce thin- surfactant concentration, centrifugation speed and ner films with lower Rs. The bulk conductivity of films graphite loading on graphene concentration are increases with loading with a plateau above 80 wt.%, shown in figures II.8(a)–(f) AFM indicates that the and the critical thickness at which bulk conductivity is exfoliated material consists of FLG, with ~18% SLG reached is less when loading is higher. After annealing (figure II.9). 30 2D Mater. 7 (2020) 022001 C Backes et al 0.35 0.30 0.30 0.30 Sonicated 9 hours Microfluidized 20 min (a) 0.5 mg/mL SDC 9 mg/mL SDC (b) 20k psi 30k psi (c) 0.25 0.25 0.25 0.20 0.20 0.20 C (mg/mL) C (mg/mL) C (mg/mL) 0.15 0.15 0.15 0.10 0.10 0.10 0.05 0.05 C = 0.237(1-e-0.026 cyc) 0.05 C = 0.215(1-e -0.021 cyc ) C = 0.243(1-e-0.036 cyc) C = 0.237(1-e -0.026 cyc ) 0.00 0.00 0.00 0 2 4 6 8 10 12 0 20 40 60 80 100 120 140 0 20 40 60 80 100 120 140 Surfactant concentration (mg/mL) Cycles (cyc) Cycles (cyc) 0.30 0.15 2.00 0.25 (d) (e) 9 mg/mL SDC, 100 cycles, 30 kpsi, 10 krpm 1h (f) 1.60 0.20 0.10 1.20 C (mg/mL) C (mg/mL) C (mg/mL) 0.15 9 mg/mL SDC, 60 cycles, 30 kpsi Centrifugation speed: 0.80 Gravity, 72 h 0.10 1 krpm, 1 h 0.05 2 krpm, 1 h 4 krpm, 1 h 0.40 0.05 6 krpm, 1 h 10 krpm, 1 h C = 0.0191.Ci 0.00 0.00 0.00 0 20 40 60 80 100 120 140 0 20 40 60 80 100 120 140 0 20 40 60 80 Cycles (cyc) Cycles (cyc) C i (mg/mL) Figure II.8. (a) Effect of SDC concentration at fixed processing cycles (~60 cycles, 20 min, red) compared to sonication processing (9 h, black). (b) Effect of surfactant concentration 0.5 g l−1 (red) and 9 g l−1 (black) and process cycles. (c) Effect of process pressure 20 k psi (black) 30 k psi (red). (d) Effect of centrifugation parameters; (e) effect of using NaCMC as a stabilizer. Dashed red line represents sonication processing using the same NaCMC loading. (f) Effect of increased graphite loading. II.2.2.5. Ball milling interact with the graphene π-system [259], but can Ball milling can produce FLG at low-cost and under also form an extended network on its surface, owing environmentally friendly conditions. It does not to the presence of hydrogen bonds (figure II.10). Mela- require solvents, avoiding the use of toxic organics, mine can then be washed away using hot water, leav- while increasing efficiency in terms of time and energy. ing high quality graphene suspensions, with very small Two effects take place in ball milling [163]: the Raman D peak. shear force, that promotes the synthesis of large-sized In a typical experiment, 7.5 mg of graphite and flakes, of the order of the starting graphite planes, and 22.5 mg of melamine are ball-milled at 100 rpm for the vertical impacts applied by the balls during the roll- 30 min, under air atmosphere, in a stainless steel grind- ing actions, which tend to deteriorate the flakes and ing jar containing ten stainless steel balls of 1 cm of must then be minimized in order to reduce the num- diameter. The resulting solid mixture is dispersed in ber of defects. Ref. [258] reported changes of graphite, 20 ml of solvent, obtaining black suspensions. Filtra- during milling processes, through a nanocrystalline tion can remove melamine from the suspensions, phase prior to amorphization. To avoid this and favour giving rise to FLG dispersions in different solvents. If the shear force process, the use of solid exfoliating DMF is used as a solvent, colloidally stable dispersions agents is recommended. are obtained with no precipitate observed. If water is Exfoliating agents, including triazine derivatives, used as a solvent, a concentration gradient appears, can be used to form multipoint interactions with and some precipitation takes place after stabilization the exfoliated flakes. Amounts around few grams of of the dispersion, at room temperature, for 5 days. graphite and exfoliating agents, such as melamine When preparing dispersions in water, melamine (2,4,6-triamine-1,3,5-triazine), can be used to prepare can also be eliminated by dialysis, which is highly rec- FLGs, which can subsequently be dispersed in organic ommended, mainly because this technique allows solvents, such as DMF or in aqueous media [179]. The keeping the flakes always in dispersion [213]. It is lateral size of the resulting flakes can be modulated advisable to hand-shake the dialysis sack and to apply a depending on the ball-milling conditions (atmos­ pulse of mild (bath sonication for no more than 1 min) phere, revolutions, and time of treatment). Different if any precipitate appears in the black dispersion dur- triazine derivatives were tested, in relation to their abil- ing dialysis. Once melamine is removed, the black dis- ity to exfoliate graphite, not only experimentally but persion is kept in the stabilization container at room also by computational studies, which evidenced that temperature for 5 days, in order to obtain a concentra- melamine is a good option to favour exfoliation [259]. tion gradient. The liquid fraction with stable sheets in Melamine not only has an aromatic nucleus that can suspension needs to be carefully extracted, avoiding 31 2D Mater. 7 (2020) 022001 C Backes et al (a) 20 cycles 60 cycles 100 cycles 2.5 µm 2.5 µm 2.5 µm 1 µm 1 µm 1 µm 60 (b) 40 100 cycles (c) 60 100 cycles Count Count 40 20 20 0 0 0 100 200 300 400 500 600 700 800 0 2 4 6 8 10 12 14 16 18 20 60 60 cycles 60 60 cycles 40 Count Count 40 20 20 0 0 0 200 400 600 800 0 2 4 6 8 10 12 14 16 18 20 60 20 cycles 20 cycles 60 40 Count Count 40 20 20 0 0 0 200 400 600 800 0 2 4 6 8 10 12 14 16 18 20 Lateral size (nm) Apparent height (nm) Figure II.9. (a) AFM height scans of flakes microfluidized for 20, 60 and 100 cycles. (b) Histogram showing the lateral size distribution of flakes with increasing process cycles. (c) Histogram showing the AFM height distribution for increasing process cycles. the meniscus and the precipitate, in which non-exfo- amounts of melamine molecules tune the adsorption liated graphite remains. The obtained aqueous disper- of few water layers on SLG. These reveal that melamine sions, with concentration ~0.1 g l−1, are stable at RT for acts as a non-covalent anchor, keeping some water a few weeks. According to elemental analysis, these dis- molecules near SLG, providing a quantitative estima- persions contain an amount of melamine lower than tion of the non-covalent interactions (dispersion and 1 ppm, which is considered non toxic [260]. hydrogen-bonding), which provide the required driv- Characterization of the FLG shows a low number of ing force to stabilize the SLG-water systems. defects, with Raman ~0.2–0.6, confirming that no oxi- A further advantage is that the aqueous suspen- dation occurs during the preparation [179]. In [261], sions can be lyophilized after having been frozen [262]. ab initio calculations on graphene-melamine-water A very soft and low-density black powder, consisting systems have been performed focusing on how small of FLGs, is produced this way. This solid can be safely 32 2D Mater. 7 (2020) 022001 C Backes et al Figure II.10. Supramolecular assembly of melamine between SLG. The addition of small amounts of solvent during milling was reported to improve exfoliation [153]. Sol- vent drop grinding was shown to be an effective tech- nique to accelerate mechanochemical reactions [265]. The addition of catalytic amounts of a liquid phase facilitates molecular mobility, inducing reactivity in normally inactive systems. Therefore, wet milling con- ditions have been probed, and ball milling exfoliation of carbon nanofibers was reported [153]. Melamine was used as the exfoliating agent: 0.5 ml of solvent was added to 30 mg of nanofibres/melamine mixture. The Hansen solubility parameters of each material were studied in order to discriminate SLG from the rest of Figure II.11. FLG powder (left) and its dispersion in cell materials (FLG, poorly exfoliated carbon fibres and culture medium (right). melamine) after the milling treatment. This allows one to select a suitable solvent to disperse mainly SLGs [153]. The quality of exfoliation was studied by stored and shipped, being easily dispersed in water, Raman spectroscopy. Although there is a compromise culture media or other solvents, using mild sonication between exfoliation quality and graphene concentra- combined with shaking with no change in its struc- tion, these wet conditions follow the same trend estab- ture (figure II.11). The mass yield of the entire process, lished by the Hansen solubility parameters, permitting starting from graphite, is ~30%. SLG dispersion [153]. In [179], powders were characterized with elemen- To achieve larger particle sizes (in the range of tal analysis (EA), TGA, XPS, TEM and Raman spectr­ microns), mechanical exfoliation by ball milling or oscopy. EA resulted in average values ~94.3 wt%C, 0.4 high-shear mixing can be combined with chemical wt%H, 0.4 wt%N and 4.9 wt%O [1728]. TGA showed intercalation. Intercalating Li and ammonium-based a weight loss [1728], in agreement with EA, confirm- ions (such as tetraethylammonium-Et4N+- ions) ing the non-oxidative nature of the milling treatment. between layers weakens the van der Waal forces [266]. This feature was also corroborated by XPS [179], as Ball milling supplies shear forces high enough to exfo- deconvoluted plots confirmed the presence of small liate the intercalated compounds. Using ionic liquid amounts of oxygenated groups, from the C 1s and O 1s (IL) or ammonium based deep eutectic solvents as the core level spectra and traces of melamine. XPS spectra milling liquid accelerates the kinetics of the intercala- (figure II.12) indicate that 15 min ultrasonication does tion because it increases the activity of ammonium not add a significant number of defects. TEM was used ions [266]. Deep eutectic solvents are green solvents to determine the lateral flake sizes [1729], concluding having the advantages of non-flammability [267], that FLG have a wide size distribution ~200–2000 nm. high thermal stability (>150 °C) [267], wide liquid There is an increasing interest in the use of GRMs phase range (usually over a range of 200 °C–500 °C) in biological applications. Dispersions in water and [268], negligible vapour pressure (<10 Pa at RT) [269] culture media allow the preparation of hydrogels for and easy recycling [270]. These properties eliminate drug delivery [259], the study of synaptic functions in the safety problems associated with typical organic sol- culture brain cells [263, 264] as well as environmental vents. An eutectic mixture of urea and choline chloride toxicity studies [262]. (2:1) can be produced by stirring the two components 33 2D Mater. 7 (2020) 022001 C Backes et al until clear liquid is formed. The liquid can be then tiple portions during the reaction, rather than in a sin- charged into the grinding chamber with 2 g graphite, gle portion. Such modification had the same oxidation 4 g battery grade Li chips, and yttria stabilized zirconia efficiency of reiterative Brodie oxidations (achieving milling media. The ball milling runs under a flow of N2 high oxygen contents, up to C:O  =  2:1) but in a single gas inside a glovebox. It is better to pause the grinding synthetic step. However, the Brodie and Staudenmaier process for 30 min every 2 h to allow the system to cool. method was not widely used because of the poten- To recover the exfoliation product, the samples are tial risk of explosions [297]. In 1958 Hummers and then washed with DMSO, and centrifuged at 1500 rpm Offeman proposed a safer approach, currently widely to remove thick flakes. The milling speed should be adopted and known as Hummer’s method [156], con- controlled to below 500 rpm to avoid extensive heating sisting of the replacement of KClO3 by KMnO4 and and to minimize defects in the produced flakes. NaNO3 in concentrated H2SO4. Ref. [298] reported that excluding NaNO3, increasing KMnO4 and a mix- II.3. GO and RGO ture of H2SO4/H3PO4 can improve the oxidation pro- II.3.1. GO cess. In these conditions the reaction is not exothermic GO has a number of properties that are advantageous and no toxic gas is produced. for applications. Due to its oxygenated surface, it can The method proposed in Refs. [272, 298] allows undergo complete exfoliation in water [271–275], one to prepare highly soluble (>4 g l−1) water-soluble yielding colloidal suspensions of individual sheets GO for studying the biological and toxicology effects [273, 276, 277], that can be further functionalized, on living cells [235, 286, 287]. In this way, GO sheets deoxygenated or dispersed in polymeric matrixes to could be prepared with constant surface chemistry, but give new multifunctional materials and composites varying lateral size, spanning from 100 µm down to [276]. This, combined with chemical tailoring less than 100 nm [235]. The lateral size of the GO flakes [216, 217, 278] and biofunctionalization [279–283] after a typical oxidation can be tuned by sonicating the makes GO a promising candidate for biomedical starting solutions for different times, from 0 to 100 h, applications where water compatibility is crucial, e.g. in in milliQ water (for a systematic study of GO size ver- tissue engineering, drug delivery, cancer treatment and sus sonication time see Ref. [136]). GO can also be pre- biosensing [281–289]. Unlike liquid-exfoliated LMs, pared by the Hummer’s method directly on SiC wafers GO is typically one layer thick [276]. This high surface from graphene on SiC, with the property that the GO to volume ratio and the mechanical strength lends itself layers remain on the substrate for further surface stud- to using GO as electrode material in energy storage and ies and electronic devices production [299–301]. conversion [290, 291] or as filtration membrane for It is still not clear which are the most relevant fac- gas and ion separation [292, 293]. In addition, GO has tors regulating the interactions between GO and bio- some unique features. For example, unlike other LMs, logical molecules, tissue structures and organisms. it is amphiphilic, due to the non-uniform distribution Numerous studies [235, 286, 287, 302] have shown of the various oxygen functional groups and can self- that GO chemical purity, chemo-physical parameters assemble into a range of nanostructures with different and morphological properties can play a crucial role morphology at interfaces [294]. When increasing the on cells viability and functionalities [302]. concentration in dispersion, GO undergoes a phase Taking advantage of the size tunability and high transition from colloidal isotropic to nematic liquid solubility of GO, the effect of the lateral size of GO crystalline phase, where the GO sheets are oriented in different biological systems was investigated for in parallel [295]. The liquid crystalline phase leads human and murine phagocytic cells [235], human to an enhanced alignment of GO sheets in those intestinal cells [286], human lung and colon carci- assembled structures, offering improved mechanical noma cells [303], different enzymes of human [287] or and electrical/thermal properties, compared to the bacterial [304] origin, gelatine membranes and fibres random-oriented configuration of GOs which is of animal origin [305]. important for electrochemical applications and opens In Ref. [235], three GO solutions, sonicated for 0, 2 up possible applications of GO in optical switches and and 26 h (with average lateral size ~1.3, 0.27 and 0.13 photonic crystals [295]. µm) were mixed with human monocyte macrophages GO can be prepared by oxidation of graphite using (hMDM) and murine intraperitoneal macrophages highly oxidant reactants such as H2SO4, KMnO4, H2O2 (mIPM), revealing that the flake size has significant [156, 157]. The preparation method strongly influ- impact on different cellular parameters (i.e. cells via- ences the surface chemistry and related chemo-phys- bility, reactive oxygen species (ROS) generation and ical properties, ultimately affecting its functionalities. cellular activation, figure II.13). Ref. [235] reported a In 1859, Brodie first reported the synthesis of GO by strong interaction (>90% internalization) with the adding potassium chlorate (KClO3) to a slurry of cellular membrane, leading to the sliding of flakes graphite in fuming nitric acid (HNO3) in a single por- under the membrane layer. GO samples could be inter- tion [296]. In 1898, Staudenmaier improved Brodie’s nalized by both human and murine macrophages in a method by replacing ~two thirds of fuming HNO3 size-dependent manner. The more the lateral dimen- with concentrated H2SO4 and by adding KClO3 in mul- sions of GO were reduced, the higher were the cellular 34 2D Mater. 7 (2020) 022001 C Backes et al Figure II.12. XPS spectrum of powders produced by ball-milling of graphite. Figure II.13. (a)–(c) SEM image of GO solutions sonicated for different times, then spin coated on 300 nm SiO2/Si(1 0 0): 0 h-GO (a), 2 h-GO (b), and 26 h-GO (c). (d) Flake size distribution of the 0 h-GO, 2 h-GO and 26 h-GO samples from SEM statistics. (e) Optical map of mIPM cells incubated with GO. Red dots indicate where Raman spectra was taken to detect internalization of GO into the cell. Adapted from [238], with permission of The Royal Society of Chemistry. internalization and the effects on cellular functional- GO. The electrochemical GO production can be ity [235]. Flakes with the smallest lateral dimensions achieved in two sequential steps starting using flex- (<300 nm) were better internalized by primary mac- ible graphite paper as anode and Pt as a cathode. In the rophages. A ‘mask effect’ was observed due to the 2d first step (typically 20 min), a stage-I GIC is formed shape of GO, with a preferential parallel orientation of in concentrated H2SO4. In the second step, diluted the GO sheets onto to the cellular surface. (50 wt%) H2SO4 is used as electrolyte and the GIC as An efficient an green electrochemical oxidation of anode. Within seconds, the intercalation compound graphite to GO was reported [306]. Compared to tra- is oxidized and exfoliated. The dispersion produced is ditional GO synthesis, the process bears the advantage then filtered and washed to remove sulfuric acid and that it is scalable, fast (<30 min opposed to hundreds then redispersed by bath sonication if desired. Due to of hours) and the risk of explosions is minimized. the lower acid concentration in the exfoliation step While electrochemical exfoliation has been developed compared to the traditional Hummer’s method, the to synthesize GRM with minimal oxidation (see sec- dispersion is less viscous facilitating the filtration and tion II.5), it has previously not been used to prepare washing step. Importantly, the process can be designed 35 2D Mater. 7 (2020) 022001 C Backes et al in such a way that a continuous oxidation and exfolia- [143], however, in most of the applications in compos- tion is achieved and the degree of oxidation is readily ites, SLG is not needed. tuneable by the concentration of sulfuric acid. The The C/O ratio and the quantification of other most critical step is the choice if starting material, as heteroatoms are important factors to determine the neither graphite powders, not flakes or rods are suit- range of applications suitable for RGO. RGO with high able. (10%) to medium or low (1%) content of O is not suit- able for thermal conductive properties, and a decrease II.3.2. Reduction of GO to the oxygen content to lower than 0.5%w and dis- There is a growing and active investigation into the tances between defects  >15 nm are needed [141]. Also, preparation and use of RGO due to its good electrical the sp2/sp3 ratio is relevant for energy applications conductivity properties (from 10−5 S cm−1 to 1000 [309], catalysis [316] and gas absorption [317]. In S cm−1, [307]), and increase in mechanical reinforce­ many applications, such as mechanical reinforcement ment of polymeric matrixes [308]. E.g., 0.2% RGO can of polymers, increase dispersion in composites or fire improve from 25% to 50% the mechanical properties retardancy [318] and for further covalent functionali- of epoxy matrix composites[309]. Whereas thermal, zation, GO with 15%–25% O concentration is needed chemical and thermochemical reduction of graphene and ~8%–4% O are recomended. and GO are customizable and versatile methods for Surface area is another of the key characteris- preparing different types of RGO, they show several tics. The Brunauer–Emmett–Teller (BET) theory problems that hamper the synthesis of defect–free serves as the basis for an important analysis tech- SLG. Among them, there are difficulties in complete nique for the measurement of the specific surface removal of functional groups and the restitution of area of materials. We will refer to those experimental morphological characteristics of the material prior methods as BET specific surface area, SBET, meas- to the oxidation and sonication processes. RGO is urements. SBET is related to N and is a key factor highly polydisperse [307] in the lateral size, from enabling a material for energy applications. N can be less than 1 µm to more than 100 µm, due to the estimated by dividing the maximum surface area of distribution of crystal size in the starting natural or SLG, 2630 m2 g−1 by the specific surface determined synthetic graphites, and also due to the scissor effect experimentally by BET 2630/SBET, [319]. According that breaks the graphenic planes during the oxidation to Ref. [320], RGO with SBET ~600–700 m2 g−1 are and/or the sonication processes [310]. However, the mostly SLG. RGO thickness is controlled by the exfoliation of GO The density of RGOis another key parameter. [311] and, depending on the preparation method and GRMs with high SBET values produced by thermo- conditions, 1L can be obtained [312]. One of the first chemical reduction exhibit low densities, even lower isolation of graphene was realized by reduction of GO than 4 g l−1 [311]. For some applications a compaction by hydrazine in 1962 [273, 274]. of the RGO for further processing is needed [321]. There are several key characteristics of RGO that Many strategies exist for the chemical, thermal should be considered and evaluated to obtain opti- and thermochemical reduction of GO, which strongly mum performance for a particular application: aver- depend on the process and conditions selected for the age lateral size, thickness, C/O ratio, the quantifica- reduction process, and the final performance required. tion of other heteroatoms, and surface area. Lateral Toxic hydrazine, sodium borohydride, hydroquinone sizes  >5–10 µm are not suitable for fiber reinforced have been used [322], as well as ‘green reducers’ such composites by vacuum assisted liquid resin infusion, as organic acids [323], alcohols [324] and local high T LRI, or resin transfer moulding (RTM), due to the dif- reduction [300]. Biochemical molecules [325], amino ficult filtration of the RGO flakes by the fabrics during acids [325], natural extracts [325] or metals [326], and the processing [313]. Lateral sizes  >30 µm produce several approaches other have been proposed [327]. low percolation threshold composites, as compared Hydroiodic acid, often in combination with other with medium (<20 µm) and small (<5 µm) ones, due acids, has been suggested as potent and versatile reduc- to size of overlapped areas and the higher number of ing agent [328]. GO reduction can be carried out at low contacts between particles [314]. T either in solution or for GO films and papers, in the Thickness is also a key factor for the performance gas phase [328]. GO self-assembled on the air/water of RGO in applications. In photocatalytic applications interface can also be efficiently reduced by hydroiodic of RGO-TiO2 the highest photocatalytic activity was acid [329]. When using hydroiodic acid on GO films, observed with the SLG composites, and it decreased it was found that the strength and ductility of the film with N; this effect is attributed to higher charge car- was improved [157]. Compared to other methods of rier mobility and improved prevention of electron– thermal reduction, such as reduction with hydrazine hole pairs recombination [315]. Better results were or vitamin C, it was found that hydroiodic acid reduces obtained by co-exfoliation of graphite and TiO2 the GO without causing additional damage to the lat- [1716]. In some composites, this effect is also observed tice [328]. 36 2D Mater. 7 (2020) 022001 C Backes et al Based on reproducibility, post-processability and done in vacuum [334] or inert atmosphere [311] in ease up-scaling, alcohols are ideally suited for chemi- a reducing H2 environment [335] and using carbon cal reduction [324]. GO is first dispersed in the selected donor molecules. Alcohols are very effective in the alcohol. Concentrations of GO ~5–20 g l−1 should be reduction of GO and the partial restoration of the used because, before the reflux of GO in alcohol, ultra- graphitic structure due to their carbon donor charac- sonication (of a batch or in a continuous system) is teristics. Other carbon donor molecules or a combi- needed for the exfoliation from graphite oxide to and nation of hydrocarbons [336] are alternatives for the an increase of viscosity should be avoided. The higher production RGO with O content  <2% O at moderate the GO lateral size, the lower GO concentration should T  <  800 °C. to be used, due to the lower stability of the GO in the In thermochemical reduction, the working atmos­ suspension. The main disadvantage of this process is phere comprises a gas carrier, which in most cases is the high volume of alcohols needed. E.g., to process an inert gas such as N2 or Ar, and the carbon donor/ 100 g of GO, a 20 l vessel is needed. However, alcohols source molecule or combination of molecules. Ar is can be recycled by column distillation with recycling preferred as gas carrier over N2, due to the potential yields  >95%. Critical parameters are the selection of doping of with N RGO during the reduction process. the alcohol and the time under reflux. By using x-ray For the restoration atmosphere, it is crucial to control diffraction, the reduction can be monitored based in the carbon donor and also the kinetics of the pyrolytic the decrease of the intensity of the (0 0 1) peak of GO decomposition of the active gas. Carbon molecules and the presence of (0 0 2) peak of RGO [330]. result in the decomposition and formation of other The electrical conductivity of RGO is also depend- non-graphene species and non-desired aromatic mol- ent on the chemical reduction process [307]. A Longer ecules [337]. Thus a low concentration, usually  <5% is reflux time in isopropyl alcohol (IPA) produces recommended. Significant improvement of the reduc- higher conductivity. However, after 15 h no significant tion process when compared to the material obtained improvements are observed, see figure II.14. in H2 was reported [338].The use of carbon donor After this reduction step, filtration of RGO at RT is atmospheres is the most cost-efficient to obtain highly required. This is easy to perform, due to the powdery reduced GO, with O concentration from 0.25% to 1%. aspect of the material. This can be air-dried or oven The reduction T determines the kinetics of the dried at less than 100 °C for 12 h. It is important to thermal decomposition (from 300 to 2000 °C), but control the drying process to prevent contaminations, a low cost- efficient reduction must be pursued in an avoid thermal expansion and decrease density for industrial enviroment and the time and T of the pro- further processing. High pressure and hydrothermal cess should be the lowest possible. Efficient processes methods are more complex [331, 332], but efficient at low T have been reported in [339], obtaining gra- reduction of GO can be achieved, obtaining lower O phene for Li batteries at a low T ~ 300 °C in 5 min. concentration compared with traditional chemical reductors, such as hydrazine [332]. II.3.3. Aerogels based on RGO The products obtained from chemical reduction of Graphene flakes can be used as building blocks for exfoliated GO [307] have usually poor, <100 S cm−1, the preparation of 3d low density structures, such electrical properties due to the high O concentration as sponges, foams, hydro- or aerogels [340–346]. (>10%) and the presence of structural defects [157, These materials exhibit a highly open structure, with 158, 333]. interconnected hierarchical pores that enhance the RGO can be prepared by thermochemical meth- accessibility to the whole surface of the material and ods as an alternative route for reduction of GO [311], improve ion diffusion. The assembly of the graphene where restoring the sp2 SLG structure is addition- sheets into macrostructures can also improve ally achieved by thermal annealing [276]. Besides O the mechanical properties, while retaining other removal by reduction, enhancing the graphitization properties associated to individual graphene layers by repairing the C–C sp2 π-bond network at defect [340, 341]. GO can be used for the preparation of regions during the reduction process [311] can further composites with tailored macroporous structure improve the electrical properties of exfoliated RGO, through the assembly of GO sheets with different for electrical, thermal, and self-sensing applications. templates. After template removal, complex GO- For the preparation of RGO with  <1% O, several based materials with hierarchical porosities can be strategies can be used, such as  >1800 °C annealing T, prepared [347]. The combination of processing at RT long processing times (>60 min at 1500 °C), reducing and a subsequent treatment at moderate T allows the atmospheres, such as H2, alcohols, hydrocarbons, etc. preparation of highly conductive bare carbonaceous Thermochemical reduction can be done at medium electrodes [348] and composites, which may serve as 800 °C or high T (>1200 °C), depending on the final self–standing electrodes for Li ion batteries [347]. requirements. The choice of the working atmosphere Ice–templating [349] is a useful technique for the has an impact on the final RGO characteristic, par­ preparation of macroporous reduced graphene oxide- ticularly on its O content. Thermal reduction can be based aerogels (ARGO). Ice–templating involves the 37 2D Mater. 7 (2020) 022001 C Backes et al Figure II.14. Conductivity of RGO reduced in IPA reflux versus reduction time. Figure II.15. Schematic representation of the route followed for the preparation of RGO composite aerogels. freezing of a GO suspension, then subsequently freeze- waters, but also for immiscible substances in biphasic dried to sublime the ice crystals formed within the systems [1730]. In energy storage applications, ARGO structure. This process is useful for the preparation of was proposed as electrodes for supercapacitors, Li-ion, a variety of a homogeneous nanostructured compos- Na-ion or metal-air batteries [341]. ites, with application in energy storage, catalysis, sens- Ice-templated ARGO have been investigated as ing or separation [350], it offers a better control on the self-standing electrodes for energy storage devices porosity of the resultant materials. The modification [351]. These structures can be assembled into the elec- of parameters such as T or immersion rate, impacts trochemical cells without the need of binders, which the textural properties of the resultant material. do not contribute to the capacity of the electrode, but Ice-templating also allows gives structures with hier- can decrease its conductivity and block the access of archical porosity, which can favor the transport and the electrolyte ions to the surface of the active materi- diffusion of molecules to the whole surface of the als. As anodes for Li ion batteries (LIBs), sp2 graphene material, highly desired in adsorption processes for not only improves the electrodes electronic conductiv- separation and purification purposes and catalytic ity, thus enhancing the power of the device, but it can applications, not only of soluble pollutants in waste also buffer the volume changes undergone by some 38 2D Mater. 7 (2020) 022001 C Backes et al a) b) c) d) 1 cm Figure II.16. Photographs of GO aerogels prepared form different suspensions: (a) GO (2 mg ml−1), (b) GO (6 mg ml−1), (c) Sn-GO and (b) RGO composite aerogel, obtained after thermal reduction of GO aerogels at 800 C. active species used as high energy anodes for LIBs, atmosphere to reduce GO and to form the desired during alloying with Li [352]. crystalline phase. The processing of these materials as self-standing For the preparation of the GO suspension in etha- composites provides additional value to the electrodes nol ~5 g l−1, an aqueous suspension of GO is washed since they can be directly mounted into flexible port- several times with dry ethanol. The suspension is then able electronics [353]. sonicated and centrifuged to remove aggregates in the sediment. When mixing Si NPs and GO, the use of II.3.4. Si-RGO and SnO2-RGO aerogels ethanol is justified, since Si-RGO composites generally ARGO composites can be prepared following the route need pre-oxidation of the Si NPs or ultrasonication, in schematized in figure II.15. First, an homogeneous order to improve their dispersibility in polar media, to suspension of GO, alone or in the presence of a certain be homogeneously mixed with GO [356, 357]. A highly precursor of the desired particles, is prepared. Second, homogeneous, see figure II.17, and stable dispersion of the suspension is rapidly frozen by its immersion non-oxidized Si NPs was achieved in this way [355]. in liquid nitrogen. Then it is freeze-dried to remove Si-GO composites were prepared by mixing Si the ice-crystals formed upon freezing, leaving the nanoparticles with the GO suspension in ethanol macroporous structured in the GO-based monoliths. under inert atmosphere, then, solvent was reduced to Finally, samples are partially reduced by a thermal 10% to be subsequently frozen and freeze-dried, to be treatment conducted under inert atmosphere, yielding then thermally reduced at 1000 ºC under dynamic Ar/ a macroporous carbonaceous network in which RGO H2 (95:5) atmosphere. The resulting material has large sheets are decorated with crystalline particles. macropores in a highly open porous structure formed The microstructure and porosity of the resultant by graphitic flakes homogeneously decorated with Si aerogels can be tuned by changing rate and T of freez- NPs (~50 nm diameter), figure II.17(d). ing. Fast freezing does not allow graphene layers to This procedure also allows the in situ deposition of self-arrange with the ice crystals front, giving rise to the particles on the surface of the graphene flakes. Sn materials containing macropores randomly distrib- and SnO2-ArGO, were prepared by dissolving SnSO4 uted within the structure [354]. in the GO suspension and the the pH was increased to Figure II.16 shows photographs of different aero- 9 using NH3 solution.Then, the suspension was freeze gels obtained by ice templating. In these cases, GO sus- dried to obtain macroporous Sn(OH)4–GO compos- pensions obtained by the sonication of graphite oxide ites [352]. by the Hummer’s method were used [120]. For the GO-based aerogels were thermally reduced under synthesis of Sn-based ARGO composites an aqueous a dynamic argon flow to increase their electrical con- suspension was used, but in the case of Si-ARGO, the ductivity. Heating the samples under a dynamic argon GO and the SiNPs were initially dispersed in ethanol. flow, Sn-ARGO (figure II.17(b)) at 800 °C and SnO2- Homogeneous GO suspensions in the presence of the ARGO (figure II.17(c)) at 650 °C were obtained [352]. precursor are in this way obtained [352, 355]. Then, Figure II.17 shows a highly porous material formed the suspension is frozen in liquid nitrogen, and subse- by RGO flakes which contain large macroporous quently freeze-dried to sublime the ice-crystals, leav- randomly distributed along the structure. In both ing the macroporous structured monoliths. Finally, samples, the Sn sub-micrometer particles showed a the samples undergo thermal treatment under inert homogeneous distribution on the graphene sheets and 39 2D Mater. 7 (2020) 022001 C Backes et al a) b) 10 µm 5 µm c) d) 5 µm 5 µm Figure II.17. Representative SEM images of (a) A-RGO, (b) A-Sn-RGO, (c) A-SnO2-RGO and (d) A-Si-RGO. Figure II.18. (a) Reaction scheme for the quantitative electron transfer from various GICs to PhCN leading to dissolved K+ ions and the red coloured radical anion PhCN.-. (b) Photograph of a sealed vial containing KC8 in a concentration of 5.0  ×  10−1 M in PhCN.-, adapted from [375]. a narrow particle size distribution, higher in the case within 2–0.05 V. These electrodes also showed a good of the metallic Sn particles (~750 nm diameter) than stability during 100 charge–discharge cycles [355]. SnO2 particles (~250 nm diameter), ascribed to the This indicates that ARGO can accommodate volume carbothermal reduction, melting and re-crystalliza- changes of Si particles and, at the same time, improve tion of Sn particles [358] over 650 °C. the conductivity of the electrode without additives. These aerogels have been processed as self-standing Due to their high electronic conductivities and light electrodes, without any binder nor metallic support, weight, graphene aerogels are good candidates as showing promising results as anodes in LIBs[347]. ­support in the preparation of electrodes for next-gen- In the case of the SnO2-ARGO composites, reversible eration Li batteries such as Metal-air batteries or Li-S capacity of 1010 mAh g−1 electrode was measured at 0.05 batteries [359] or Li ion capacitors [360]. A g−1 and 470 mAh g−1 −1 electrode at 2 A g , with reason- able Coulombic efficiencies (>98%) and stability after II.3.5. GO membranes 150 charge–discharge cycles [355]. Si-ARGO elec- Membranes can be fabricated, e.g. by vacuum trodes showed reversible specific capacities ~700 mAh filtration, even though other deposition techniques g−1 can be used [1731]. GO membranes have gas and ion electrode with associated Coulombic efficiencies  >99% 40 2D Mater. 7 (2020) 022001 C Backes et al a) 100 b) 70 = 180 nm ML: 72 % 60 95 50 TG (%) Count 40 250 nm 90 30 85 MoS2 reference 20 LixMoS2 excess MoS2 10 LixMoS2 excess BuLi 80 0 100 200 300 400 500 600 0 1 2 3 4 5 6 7 8 100 LixMoS2 excess BuLi 50 = 350 nm 40 ML: 70 % Ion current (10-11 A) 96 40 m/z 18 30 TG (%) Count 92 30 20 200 nm 88 20 m/z 17 84 10 10 m/z 16 m/z 15 80 0 0 100 200 300 400 500 600 0 1 2 3 4 5 6 7 8 Temperature (°C) t (nm) Figure II.19. Characterization of chemically exfoliated MoS2 with different intercalation conditions. (A) TGA weight loss of MoS2 and MoS2 chemically intercalated with excess of BuLi and MoS2. Bottom: mass profiles for m/z  =  15–18 which constitute the main weight loss in chemically exfoliated MoS2 using excess BuLi. (B) AFM thickness histograms of MoS2 intercalated using excess MoS2 (top) and BuLi (bottom). Inset: representative images. Adapted from [379], copyright American Chemical Society. filtration properties [293] making GO an interesting tuned by adjusting the GO sheet size, the thickness of material for water purification and desalination. the GO membrane, water pH, and the GO membrane Ref. [361] reported that GO based membranes are structure as summarized in Ref. [365]. permeable to water, but hold back most other liquid The key aspect of these type of membranes is the and gases. Research in this area has been expanding self-assembly of the flakes into well-ordered macro- with a range of review papers [293, 362–365]. A brief scopic laminates unlike in GRM-polymer compos- overview will be given here. ites, where the flakes are less ordered and less densely GO-based membranes are both selective and packed [363]. The flakes arrangement in the mem- highly permeable to water, chemically and physically brane is thus an important parameter influencing stable. However, fouling and biocompatibility still separation performance. This not only depends on the need to be addressed [362]. GO/FLG hybrid mem- characteristics of the flakes themselves (size/thickness, branes were reported improved chemical resistance degree of oxidation), but also on the deposition tech- towards chlorine, and enhanced NaCl rejection (up to nique [367]. Highly ordered, interlocked GO flakes 85%) [366]. were obtained by building membranes by spin-coating These GO material membranes can be described a GO dispersion in a layer by layer fashion. These inter- as an interlayer nanocapillary network, due to the con- locked membranes showed better CO2/nitrogen selec- nected interlayer spaces [365]. This network, together tivity than less ordered ones [363, 364]. For a more with the gaps between edges of non-interlocked neigh- detailed discussion on transport mechanisms see Refs. boring GO sheets, and the cracks or holes of the GO [363, 364]. sheet, provide passage for molecules or ions to perme- ate through the GO membrane in aqueous solutions II.4. Chemical intercalation and reductive so that effective separation of different molecules exfoliation and ions can be achieved depending on the network II.4.1. Graphene from GICs morph­ ology, degree of GO oxidation, lateral size, The intercalation of graphite with sulfuric acid can be stacking in combination with size and chemical nature exploited for the production of GO. Graphite can also of molecules and ions. Molecules or ions with smaller be intercalated by a number of other species to yield GIC hydrated radius, less charge and weaker interaction [160]. The fabrication of donor-GIC after intercalation with GO sheets permeate through the GO membrane with, e.g. alkali metals, has gained increasing importance. more easily than larger ones, driven by the high cap- Refs. [368–374] reported that such acceptor-GICs are illary-like pressure [365, 1731]. The separation per- very suitable precursors for covalent functionalization formance at high water flux and rejection rates can be of graphene (see section VIII.1). 41 2D Mater. 7 (2020) 022001 C Backes et al In typical covalent functionalization sequences, 380]. Another approach, based on chlorosulfonic acid the negatively charged SLG first act as reductants assisted exfoliation was developed, allowing to keep for electrophiles (for example alkylhalide) [368], the original semiconducting properties of 2H-TMD which are subsequently attacked by the intermedi- (as MoS2 and WS2) [381, 1412, 1413, 1732, 1733]. ately generated organic radicals or H-atoms, yield- This intercalation chemistry was explored in the ing covalently modified graphene. This wet-chemical 80s [159], but the interest was not in producing exfo- functionalization is facilitated by the fact that, due to liated flakes in a liquid dispersion. Ref. [379] studied Coulomb repulsion, the negatively charged graphe- the impact of intercalation conditions on the final nide [373] layers within the solid GICs can be dis- product of chemically exfoliated MoS2. In a typical persed in suitable organic solvents (see Ref [373] for exfoliation of MoS2 with n-BuLi, the n-BuLi is used in more details). ~10-fold excess. These reaction conditions can lead to One fundamental question is if whether all nega- the introduction of defects, as exemplified by a signifi- tive charges of the graphenide intermediates can cant mass loss in TGA under inert conditions (figure be controlled or even completely removed in such II.19(a)). No specific mass fragments can be assigned redox reactions [376]. Only complete oxidation to this weight loss, suggesting it is not the result of a is expected to avoid reactions with moisture and well-defined surface derivatization [379]. The major O during workup [375], leading to side products mass fragment observed has a mass over charge (m/z) with undesired and additional O and H function- ~18 and was assigned to -OH groups, figure II.19(a). alities. More importantly, the controlled removal This disruption of the structure is not observed when of all negative charges from the solvent exfoliated MoS2 is used in excess over n-BuLi. The work-up is graphenide intermediates with a suitable oxidation more tedious, with less material obtained as individu- reagent allows for the bulk production of defect-free alized flakes.After centrifugation, flakes with similar graphene. Ref. [375] reported that the treatment of lateral dimensions and thickness distribution are pro- K intercalated graphite with benzonitrile (PhCN) duced (figure II.19(b)) suggesting that there is scope leads to discharging, i.e. taking away of the elec- to optimize chemical exfoliation with respect to yield trons from charged layers, of the individual graphe- of 1L MoS2. nide sheets upon the formation of the colored radi- cal anion PhCN.-, figure II.18, which can be used to II.5. Electrochemical exfoliation monitor the accompanying exhaustive and Coulomb Graphite can be exfoliated to SLG and FLG by applying force-driven migration of K counterions from GICs a bias in ionic aqueous (or organic) electrolytes into the surrounding benzonitrile phase. The sup- [137, 382]. Electrochemical exfoliation (EE) requires pression of reactions of dispersed graphenides with one graphite electrode and a metallic counter moisture and air takes place when no treatment with electrode in conductive media, which is not equipment- benzonitrile is provided, resulting in graphene. This intensive. The exfoliation completes in minutes to hours, represents a mild, scalable, and inexpensive method producing gram-scale quantities of flakes with high Yw for wet-chemical graphene production. This reduc- up to 80%. The graphene quality is tunable: sheet size, tive graphite exfoliation approach can be extended to C/O ratio, solubility and electrical conductivity can be water as solvent [377]. modified depending on the types of graphite precursors, electrolytes and operating potentials [383]. II.4.2. Chemical exfoliation of transition metal dichal- The selection of the graphite electrode is an impor- cogenides tant parameter. Various graphite precursors (including LMs, such as TMDs, can be reductively exfoliated via natural graphite and synthetic graphite), in the form intercalation compounds [159]. Typically, MoS2 is of powder, foil, paper, rod, flake and plate, have been intercalated with n-Butyllithium (n-BuLi) in an inert studied for electrochemical exfoliation [397, 1734]. solvent such as hexane [378]. In contrast to graphite, From our perspective, graphite foil is considered as intercalated MoS2 does not react with water [378] so an ideal raw material due to the ease of operation, as that this can be used for subsequent exfoliation. well as good tolerance to the volume expansion during Water is added to the reaction mixture to destroy exfoliation. excess BuLi. At the same time, gas formation occurs, which acts as driving force to expand MoS2 and indi- II.5.1. Aqueous media vidualize the layers. After further washing and centrif- H2SO4 can be used to prepare GICs as the ionic ugation-based purification, a colloidally stable MoS2 diameter of sulfate ion (0.46 nm) is close to the dispersion in water can be obtained with negatively interlayer spacing of graphite (0.33 nm), which is the charged, predominantly 1L flakes (70%–90%). As a prerequisite for efficient intercalation. Efficiency here result of the negative charges, a phase transforma- relates to the degree of exfoliation, a higher ratio of tion from the semiconducting 2H-MoS2 polytype to MLG is considered more efficient than thicker flakes. the metallic 1T polytype is observed. Such negatively In this way efficiency is related to the average thickness charged MoS2 can be used as precursor for subsequent of the exfoliated platelets: the lower the thickness, the covalent functionalisation using electrophiles [379, higher the efficiency. Dilute H2SO4 aqueous solutions 42 2D Mater. 7 (2020) 022001 C Backes et al Figure II.20. Illustrations of graphite after intercalation, expansion, electrochemical exfoliation and reduction steps (from left to right).Adapted from [348], © Wiley-VCH Verlag GmbH & Co. KGaA. were used [384, 385] to encourage graphite exfoliation II.5.2. Non-aqueous media and molten salts because it is less corrosive. Ref. [385] immersed a Another strategy to minimize O content is to use graphite anode together with a counter electrode (Pt cathodic intercalation and O-free electrolytes [389]. foil) into 0.1 M H2SO4The exfoliation started when The intercalation of small cations with graphite from a potential of 10 V was applied. This gave 80% 1-3LG organic solvents is the principle of many energy with high exfoliation yield (60%) and high C/O ratio storage devices [389–393]. Graphene can be prepared (12.3). Under electric current, the oxidation effect on by electrochemical intercalation of Li+ within the graphene is strong, owing to the interplay between graphite interlayer space, followed by dissociation of proton (H+) and sulfate ion at low pH. The oxidation the resultant intercalating compound using prolonged effect is defined as the oxygen content in the exfoliated sonication. Ref. [394] tried to reduce the sonication materials, which usually results from the attack of time by intercalating graphite with Li, followed by a oxygen-containing radicals by water splitting [386]. second intercalation of large tetra-n-butylammonium Water electrolysis generates oxidative radicals (HO· ions. In both cases, the strong decomposition reaction and O·), which contains a collection of sulfate salts of the solvent cations hindered the formation of LiC6 (e.g. sodium sulfate, ammonium sulfate, etc) instead of and/or ammonia GICs. Therefore, another sonication H2SO4 [387]. By doing so, Ref. [387] reported  >85% of step was need to completely detach the graphene flakes. 1-3LG with dimensions up to 40 µm and C/O ~ 17.2, Graphite can be fully exfoliated via an electro- implying a low level of O functionalization. Another chemical process in organic solvents without the need strategy to suppress oxidation is to remove radicals from of sonication or an inert atmosphere, by using DMSO water splitting [388]. A series of scavengers have been saturated with Li and small alkylammonium ions investigated. Ref. [388] reported 2,2,6,6-Tetramethyl- (triethylammonium, Et3NH+) [395]. An electrochem- 1-piperidinyloxy (TEMPO) to be the most effective, ical program is used to apply a controlled cathodic reaching C/O ~ 25.3, average sizes ~5–10 µm and potential on the graphite electrode, enabling the for- hole mobility up to 405 cm2 V−1 s−1. Electrochemical mation of exfoliated powder [266]. In a three-electrode exfoliation can produce not only graphene but also system, a chronoamperometric step of  −1.7 V on highly oxidized GO [306]. E.g., a two-step anodic Ag/AgCl is applied for 5 min, followed by linear sweep oxidation in concentrated (98 wt%) and diluted (50 voltammetry at a rate of 10 mV s−1. The potential is wt%) H sulfuric acid enables the synthesis of GO with then kept at  −5 V for 5 min, to allow intercalation of C/O  <  2 and high yield (95%) of 1L similar to those the electrolyte cations, finally it is swept linearly back to achieved by traditional Hummer’s methods. the open-circuit potential to decompose the resultant There are still several challenges: (1) the exfoliation complex compound. During the first chronoampero- only occurs at a single electrode (anode or cathode), metric and the second linear sweep voltammetry steps, limiting the production rate; (2) insufficient inter- solvated carbanion complexes are formed according calation results in polydisperse flakes (with varying to reactions II.1-5. The second chronoamperometric N), which requires additional separate procedures; step at  −5 V is important to complete the intercalation (3) flakes trend to restack in solution, due to strong process. This intercalated compound could decom- in-plane interaction when the stabilizers are absent, pose cathodically if the potential is held at negative which needs careful selection of suitable electrolytes. values according to reaction (6) or on the oxidation 43 2D Mater. 7 (2020) 022001 C Backes et al Figure II.21. AFM images of HOPG basal surface after treatment by (a) sonication-assisted, (b) chemical exfoliation, (c) electrochemical exfoliation. Z-range: (a) 400 nm, (b) 600 nm, (c) 1 µm. (d)–(f) gradient-filtered version of each image. (g)–(j) Schematic representation of the structure of the HOPG substrate after exfoliation procedures. (j)–(l) Height profiles taken along the lines shown in (a)–(c). From [137], © Wiley-VCH Verlag GmbH & Co. KGaA. cycle according to 7. In both cases, dissociation of the The intercalation from RT electrolytes is a slow intercalated compound is associated with Et3N gas for- process (diffusion coefficient  <1.2  ×  10−10 cm2 s−1) mation, more likely taking place between the graphene [396] and limited to few µm from the edge of the layers, which applies more stress on SLGs and separate graphite grain [397]. Attempts to increase chemical them further apart. activities of Li ions or other cations by using pure ILs or deep eutectic solvents did not overcome the kinetic 2Et3 NH+ + 2e− = H2 + 2Et3 N (II.1) barriers [398]. Hence, it is important to repeat the nEt3 N + e− + C = (Et3 N)n C− (II.2) exfoliation process for more than one cycle to obtain a yield  >70% the big flakes [389]. Another alternative Li+ + e− = Li (II.3) is to use molten salts at T  >  500 °C as the electrolyte. [399–401] developed for the exfoliation of graphite mLi + e− + C = Lim C− (II.4) based on Li intercalation from molten LiCl electrolyte containing water. In a typical experiment, anhydrous nEt3 N + mLi + e− + C = Lim (Et3 N)n C (II.5) LiCl powder, which serves as anode, is charged in a graphite crucible. The cathode is graphite. The cell is Lim (Et3 N)n C− + e− + 2Et3 NH+ heated to ~800 °C, above the melting point of LiCl, = C + mLi + (n + 2) Et3 N + H2 and a stream of water is bubbled into the molten bath. (II.6) Electrolysis is achieved by applying a constant direct Lim (Et3 N)n C− = C + mLi + nEt3 N + e− . (II.7) current between cathode and anode. After allowing 44 2D Mater. 7 (2020) 022001 C Backes et al the cell to cool down to RT (Ar gas flow), graphene trile), which was present in high excess, as a solvent. powder can be recovered from the solidified salt by Graphite was intercalated and partially expanded by washing with hot distilled water and vacuum filtering. electrochemical insertion of ClO4- in acetonitrile Finally, the black powder is heated at 1300 °C under by applying a  +  5V for 0.5 h. In this way, negatively inert gas atmosphere containing hydrogen. The pro- charged ClO4  −   ions intercalated through grain cess demonstrates commercial viability and a molten boundaries or defect sites and favoured the penetra- salt volume ~10 L can produce ~4.5 kg flakes in a day tion of the smaller, uncharged acetonitrile molecules. [400, 402]. The fundamental role of ClO4  −  was demonstrated in a comparison experiment by using acetonitrile alone, II.5.3. Combination of electrochemical and microwave for which no intercalation was observed [348]. The expansion amount of molecules intercalated into graphite was Microwave (MW) irradiation is currently employed high (70% w/w), as estimated from TGA of the interca- in many fields of organic synthesis to shorten lated graphite electrode [348]. reaction times, to enhance both reaction yields and After electrochemical intercalation, MW irra- product purity, to provide eco-sustainable synthetic diation expanded the graphite interlayers through methodologies, by replacing or reducing the use of decomposition and gas evolution of acetonitrile to polluting reagents [217]. Graphite can interact with yield a foam-like, multilayered powder [348]. The the oscillating electrical field of MW radiation, giving expansion was fast (~10 s) and yielded an increase in high T gradients and increased reaction rates (e.g. volume of the initial graphite ~600% [348]. The role for fast functionalization of GO) [217] as compared of acetonitrile in the expansion was demonstrated in to conventional, procedures such as conventional a comparison experiment, in which no expansion was chemical functionalization at RT in solution. observed for graphite treated only in aqueous solu- Therefore, the T attainable using MW (>80 °C) can be tions of HClO4 or NaClO4 without acetonitrile [348]. applied to exfoliate graphite [217]. Figure II.20 shows a schematic of the process, and pho- Electrochemistry can be used for a fast (<30 min) tographs of the material at the different processing and massive (>70%w/w intercalated molecules/ stages. graphite) intercalation of suitable molecules, such as perchlorate ions, into graphite [217]. Degradation of II.6. Sonication-assisted versus chemical versus these molecules in GICs is then triggered by MW irra- electrochemical exfoliation diation, creating a gas pressure surge in graphite and Graphite can be exfoliated by different chemical yielding exfoliation [217]. methods in the liquid phase. One of the simplest Charged perchlorate ions could act as ‘Trojan approaches is ultrasonic treatment in organic horses’, to favour the intercalation of uncharged ace- solvents (see II.2). Some dipolar aprotic solvents (e.g. tonitrile molecules [348]. These work as nanoscopic DMF, NMP) or surfactants in aqueous solutions are foaming agents, and decompose with MW irradiation effective to stabilize graphene in solution without re- to generate a pressure surge within graphite, resulting aggregation [131]. in exfoliation. The process yields soluble, monoatomic A more effective, but disruptive, approach to solu- (52% SLG), large (72% of the sheets extending beyond bilize graphene is covalent modification, in particular 1 µm) GRM sheets characterized and then tested as by the formation of GO (see Section II.3). In presence transparent electrodes [348] and capacitors [348]. of strong oxidants, the aromatic carbon network is oxi- These electrodes can also be functionalized with nano- dized with the creation of hydroxyl, carboxyl and epoxy porous layers of inorganic oxides, after the pre-interca- moieties. The hydrophilic nature of these moieties on lation of electrolyte species (e.g., FeCl3–nitromethane GO facilitates solution processing of concentrated 1L electrolytes), to obtain, e.g. composite electrodes for on different substrates substrates (>60% 1L[277]). LIB [1735]. The GO flakes can be then reduced to give electrically Electrochemical treatment can be performed conductive RGO. However, numerous defects on GO ­starting from HOPG as a working electrode and a Pt are not fully restored after reduction and the electronic wire as a reference electrode. Ions such as sulfates and properties of RGO are poorer compared to pristine perchlorates can be efficiently intercalated into graph- graphene. ite, whereas intercalating uncharged molecules is more Another more controllable, slightly less disrup- difficult [131], and can usually be achieved only by tive approach to exfoliate graphite takes advantage several hours sonication, with a lower yield (few%) of of electrochemistry (see Section II.4). By adjusting both exfoliated material and 1L. The exfoliation pro- the applied potential in suitable electrolytes, graphite cess works only if the surface energy of the intercalated electrodes can be either negatively or positively inter- organic liquid is similar to that of graphene [131]. The calated to obtain GICs, and then exfoliated by solvent process is limited to high-boiling solvents (T  >  100 decomposition directly during electrochemical treat- °C), such as NMP or DMF. To overcome this challenge, ment or by further thermal treatment. Ref. [348] used perchlorate ions to promote the inter- The most common LPE methods are compared: calation of an organic, uncharged molecule (acetoni- sonication in solvent, chemical oxidation, and electro- 45 2D Mater. 7 (2020) 022001 C Backes et al Figure II.22. (a) Binding energy versus relative difference in interlayer distance predicted by vdW (DF2-C09) or non-vdW (revPBE) functionals. Materials classified as easily exfoliable, potentially exfoliable, or with high binding energy are colored differently. Well- known LMs are highlighted in the plot. Adapted from [407], © Springer Nature. (b) Calculated heat of formation for 216 1L-TMDs in the 2H and 1T phases. Oxides have the highest stability, followed by sulfides, selenides, and tellurides, in that order. The stability decreases as the metal ion goes through the transition-metal series. Adapted from [412], copyright American Chemical Society. (c) Examples of patterns predicted for free-standing 1L B polymorphs, from close-packed triangular, honeycomb. The numbers at the bottom right of each model mark the hole density, S1 and S2 are the observed phases. Adapted from [423], © Springer Nature. chemical oxidation. By using a graphite crystal as start- II.7. Computational modelling of exfoliation ing material, it was possible to study the structure of of LMs both the exfoliated and the non-exfoliated fractions on The rush towards innovative applications has the nanometric and mesoscopic scale [137]. stimulated the synthesis and exfoliation of novel Figure II.21 shows AFM images of graphite basal LMs. Graphene, BN, TMDs, and BP are nothing surfaces that are left behind after different exfoliation but the tip of the iceberg of a rapidly increasing LM processes. Exfoliation by sonication in DMF proceeds family. To keep up with this experimental thrust, it layer-by-layer. Only the upper part of the HOPG is is of uttermost importance to expand the portfolio interested, and the process is slow, requiring several of potential LMs. In this respect, first-principles hours to alter the substrate roughness [137]. CE is the simulations offer unprecedented opportunities to most damaging treatment for HOPG, destroying sig- predict and design novel LMs [403, 404] providing nificantly the crystalline layer through deformation new directions for experimental explorations. To and intercalation.The obtained sheets are mostly 1L achieve this computational synthesis of novel LMs, two soluble, tending to re-stack in a uniform LM. EEpro- main strategies have been put forward, that mirror the ceeds by mechanical expansion due to gas formation. bottom-up or top-down techniques of experiments. Whole areas of the substrate rise up due to gas forma- Top-down approaches consist in systematically tion, tearing apart the superficial layers and removing exploring databases of known experimental com- large amounts of material. EE works on a larger scale, pounds, looking for LMs that can be exfoliated into 1L. oxidizing the upper HOPG layers, with more crack- This requires first to recognize if a parent bulk 3d crys- ing and swelling on the macroscopic scale. Damage to tal structure results from the stacking of chemically dis- HOPG is proportional to the applied bias, with disrup- connected components. Geometrical algorithms based tion of crystalline stacking. Exfoliation is fast and more on the comparison between interatomic distances and efficient than in solvent, yielding 1-2L GO  >  50%, van der Waals or covalent radii can be adopted to assess with tunable oxidation. the chemical connectivity between units. The dimen- The differences between the three methods high- sionality of each unit can then be estimated, leading in light the trade-off between speed and efficiency of particular to the identification of 1L. Ref. [405]data- exfoliation on one side, and preservation of material mined subset of structures with low packing ratio from quality on the other. the Inorganic Crystal Structure Database (ICSD) and 46 2D Mater. 7 (2020) 022001 C Backes et al identified 92 LMs, nearly half of which had never been prediction of germanene and silicene [411] by looking discussed before. More recently, this portfolio has been for stable Ge- and Si-based analogues of graphene. A significantly enriched by Refs. [406, 407], who not only list of 171 TMDs and oxides with favorable formation extended the investigation to other databases (includ- energies has been obtained by decorating the so-called ing the Materials Project database and the Crystallo- H and T structural phases with 27 different transition graphic Open Database), but also validated their results metals [412, 413]. Ref. [414] extended this approach to against accurate van-der-Waals density-functional- other classes of materials including halides, semimetal theory calculations of the binding energy. This led to monochalcogenides and atomically thin 1Ls, identi- the discovery of 681 [406] and 1825 [407] exfoliable fying 146 hypothetical LMs for which the electronic LMs that will provide a rich portfolio of candidates and magnetic properties have been computed [1736]. to explore for optimal electronic, optical, catalytic, Another promising way to theoretically synthesize topological, and magnetic properties. E.g. instance, a new LMs bottom up relies on global optim­ization new magnetic LM, Fe3GeTe2, has been already found techniques. Starting from a given set of elements, the [408] among these new exfoliable structures [406], but ground-state structure is obtained by searching for many more promising candidates await discovery. This the most stable atomic configuration and (possibly) approach has been adopted not only to collate these stoichiometry using either evolutionary [415–417] or extensive databases [1737–1740] but also as a one-shot particle swarm optimization [418, 419] algorithms. effort to find novel LMs from bulk layered compounds; These methods have been adopted for instance to pre- e.g. for arsenene and antimonene [409] or 1L Hittorf’s dict 1L materials with no bulk layered counterpart, phosphorus [410]. like SiS [420] novel phases of InP [417] and several 2d Novel LMs materials have also been ‘synthesized’ allotropes of boron [416, 421], known as borophenes. computationally through bottom-up approaches, In the latter case, statistical methods based on cluster starting directly with some chemical intuition. In expansion have been put forward and predicted the this regard, a first possibility is to consider structural structure of borophene on metallic substrates in agree- proto­types of known LMs and create hypothetical new ment with experiments [422]. Figure II.22 provides a structures by substituting elements with chemically classification of LMs as easily exfoliable, potentially similar species [403]. This led, e.g. to the theoretical exfoliable, or with high binding energy. 47 2D Mater. 7 (2020) 022001 C Backes et al III. Processing of dispersions physical parameters of the dispersed flakes, i.e. lateral size and thickness, and physico-chemical properties of III.1. Size selection the solvent they are dispersed in, i.e. the viscosity (η). III.1.1. Sedimentation-based separation Both parameters have a strong influence on the value The exfoliation of LMs usually results in a of f . heterogeneous dispersion of flakes having different Overall, the sedimentation of GRM flakes depends morphology, i.e. lateral size and thickness [166]. There on their mass and frictional coefficient, which is shape are many methods for the selection and processing dependent [115, 166, 424]. In first approximation, of flakes [115, 131, 183–185, 424, 425] by means thick and large GRM flakes, having larger mass with of ultrasonication [115, 131, 174, 183–185, 204, respect to small and thin flakes, sediment faster with 424–427], shear mixing [176, 177, 238, 428–430], respect to the latter, since these have smaller mass. This ball milling [179, 431–436], microfluidization [178], allows to separate GRM flakes with different morph­ and wet-jet milling [437–440]. Centrifugation ology [166, 424], see figure III.1. However, it is chal- based size selection methods can be classified into lenging to isolate large and thin flakes. The definition sedimentation based-separation (SBS) and density of large/thick and small/thin depends on the material gradient ultracentrifugation (DGU) [195]. Materials that is used in combination with the centrifugal accel- subjected to SBS are sorted on the basis of different eration and the medium. This will be detailed further sedimentation rates in response to a centrifugal down below for a few examples. force. Owing to its ease in handling, SBS is the most Thus, taking into account the dependence of the commonly applied size selection technique for GRMs. physical dimensions of the flakes on the centrifugal Typically, small and thin flakes are separated from acceleration (expressed as relative centrifugal field, larger, thicker counterparts. However, achievable RCF, in units of the earth’s gravitational field, g), it is size distributions are still rather broad except for the possible to exploit SBS to prepare GRM flakes of differ- smallest and thinnest flakes. To tackle this, a number ent morphology. In a typical experiment, a dispersion of variations to traditional single step SBS have been is centrifuged at a fixed g-for a given time. Supernatant developed, such as band sedimentation [185] and and sediment are then separated. The sediment con- liquid cascade centrifugation [186]. Different to SBS, tains larger and thicker flakes than the supernatant. By DGU exploits the movement of the dispersed object changing the centrifugal acceleration and centrifuga- to the point in the centrifuge tube where the buoyant tion time, the flake sizes can be adjusted. E.g. the flakes density of the material matches that of the surrounding in the supernatant will be smaller after centrifugation liquid. The addition of density gradient media to the at higher accelerations or longer times. Any centrifuge mixture is typically required to adjust the density of (benchtop or ultracentrifuge) with either fixed angle the liquid to the higher density material [168]. As such, rotor or swinging bucket rotors can be used. DGU has the potential to separate flakes by thickness, Even though it is not required that the centrifu- but is more challenging to perform than SBS. In the gation is run for several hours until an equilibrium following sections we will discuss these different is reached, i.e. all flakes that will sediment at a given strategies, highlighting pros and cons. centrifugal acceleration have reached the bottom of the vial, it is recommended not to use centrifugation III.1.2. Single step sedimentation times shorter than 30 min. When centrifugation times The SBS process is widely used to separate materials are short, a steeper size/thickness gradient is formed in of different nature and morphology ranging from the vial. As a result, the size and thickness distributions metallic nanoparticles (MNPs) [441], CNTs [442–445], in supernatant and sediment will depend on how the to GRM flakes [115, 131, 174, 177, 183–185, 204, sample is decanted, leading to poor reproducibility. 238, 424–428, 430]. A theoretical description of the Centrifugation times ~1–3 h are recommended and sedimentation process can be found in Ref. [195]. lead to reproducible results. This also depends on the In brief, three forces act on the objects during the rotor and vial size, hence the distance the flakes need to centrifugation, which determine their sedimentation travel to reach the bottom [246]. Often sediments are rate (see figure III.1(left)): (I) the centrifugal force then discarded and supernatants collected for analy- Fc  =  m2Dω2r, which is proportional to the mass of flake sis and/or further processing. However this makes it a itself (mGRM), to the square of the angular velocity wasteful process. (ω), and to the distance from the rotational axes (r), With SBS, it has been possible to prepare samples (II) the buoyant force Fb  =  −mS 2r, which is linked having lateral sizes ranging from few nm to a few µm, with the Archimedes’ principle, being proportional both for FLG [131, 176, 184, 424–427, 446, 447] and to the mass of the displaced solvent (mS) times the other LMs [174, 185, 244, 425, 448, 449]. In terms of centrifugal acceleration, and (III) the frictional force thickness, dispersions with SLG content up to ~60% Ff  =  −fv, i.e. the force acting on the GRM flakes were demonstrated in an SDC based aqueous disper- moving with a sedimentation velocity (ν) in the sion [446], while ~33% SLG was reported for disper- solvent. Ff is proportional to the friction coefficient (f ) sion in NMP [447]. Such 1L-rich dispersions from between solvent and the GRM flake. f depends on both SBS also contain predominantly laterally very small 48 2D Mater. 7 (2020) 022001 C Backes et al Figure III.1. Sorting of GRM flakes by ultracentrifugation. The SBS approach allows separating thin and small lateral sized flakes (supernatant) from thicker and larger ones, which precipitate at the bottom of the ultracentrifuge tube as pellet. The final flake Dimensions (lateral size, final flake N) depend on the centrifugal acceleration, material and medium. (<50 nm) flakes [446, 447]. Apart from the difference narrower, i.e. the standard distribution was reduced in percentage of SLG, the flakes processed in water- by ~2 [185]. surfactant dispersions are, on average, also smaller Even though this approach can be advantageous (~200 nm) [184, 427, 446] with respect to NMP when various fractions with narrower thickness dis- (~1 µm) [131, 447]. This difference is related to the tributions than SBS are required, it suffers from a few difference in the viscosity of the solvents. When com- disadvantages. First, the material quantity that can be paring the most widely used solvents (NMP and water processed in a single step is lower because it is benefi- where, with the addition of surfactant, the viscosity cial to keep the sample layer thin, so that most of the decreases) the viscosity of NMP (1.7 mPa s) [450] is volume in the cuvette is taken up by the race layer. Sec- significantly higher than water (~1 mPa s) [451]. A ond, the final size-selected sample is diluted during higher solvent viscosity increases the frictional force the process, making high concentrations inaccessible. [195], ­reducing the sedimentation velocity [195]. Third, a swinging-bucket rotor is required, as a fixed- Therefore, when using similar centrifugation con- angle rotor will not lead to such a defined movement of ditions (centrifugation time and centrifugal accel- flakes through the race layer. Therefore, liquid cascade eration), the flakes retained in the supernatant will be centrifugation (LC) has been suggested as versatile larger/thicker in higher viscosity solvents. alternative to the above mentioned single step SBS. SBS also suffers from the further disadvantage that all flakes(also small  <  50 nm) remain in the III.1.3. Liquid cascade centrifugation supernatant when operating at the low centrifugal LCC is versatile and can be carried out using benchtop acceleration (<1000 g) used to isolate large (µm) centrifuges [186]. This is a multi-step procedure, flakes. Hence, dispersions containing larger flakes whereby various cascades can be designed according on average are significantly more polydisperse and to the desired outcome. To demonstrate this process, a show broad size (20-few µm) and thickness (1–20 general cascade is portrayed in figure III.2. It involves layers) distributions. This can be partially overcome multiple centrifugation steps, each with a higher by a band sedimentation approach, where the disper- centrifugal acceleration [186]. After each step the sion is placed on top of a solvent race layer (without sediment is retained and the supernatant is then used flakes) prior to centrifugation [185]. E.g. a disper- in the proceeding stage. As a result, each sediment sion containing flakes in an aqueous surfactant can contains flakes in a given size range, ‘trapped’ between be onto a race layer of deuterated water containing two centrifugation stages with different speeds. the same surfactant [185]. During centrifugation, the Similar to band sedimentation, small/thin flakes are flakes spread throughout the vial according to their removed from dispersions with predominantly larger/ sedimentation rate, related to their size. The centrifu- thicker flakes. As with traditional SBS, the lateral gation is stopped before the dispersion constituents dimensions and N obtained depend on the design reach the bottom of the vial, allowing for a collec- of the cascade (centrifugal accelerations), starting tion of various fractions in one run. Smaller/thin- material and solvent. A few examples of lateral sizes ner flakes remain closer to the top and are thus effi- and thicknesses are summarised in table III.1. ciently isolated from larger/thicker ones, closer to the Critical to LCC, the resulting sediment can be bottom. This was demonstrated for MoS2 [185]. In redispersed by mild agitation (shaking, or  <5 min this case, centrifugation at 1500 g for 10 min yielded bath sonication) in the respective medium, ena- various fractions ranging from mean lateral sizes of bling one to reach any desired flake concentration, 350 nm and arithmetic mean N~15 in fractions close as well as modification of the concentration of addi- to the bottom, to 40 nm and 2L in fractions extracted tives (such as polymers or surfactants). Virtually no from the top of the vial. Compared to traditional material is wasted in LCC, resulting in the collection ­homogeneous SBS, the distribution histograms were of larger masses of size-selected flakes from a single 49 2D Mater. 7 (2020) 022001 C Backes et al ω 4>ω 3 ω1 ω 2>ω 1 ω 3>ω 2 ω 5>ω 4 ω 6>ω 5 ω 7>ω 6 Discard Collect decreasing sizes in sediments Figure III.2. Schematic representation of liquid cascade centrifugation. Adapted from [186], copyright American Chemical Society. Table III.1. Overview of mean lateral sizes using LCC. For each material, the ‘large/thick’ and ‘small/thin’ fractions are defined by the midpoint of the centrifugation boundaries (given as central g). Data includes centrifugation in solvents (NMP, IPA) and aqueous SC solution. FLG SC FLG NMP WS2 SC MoS2 SC BP NMP GaS IPA [236] [176] [245] [245] [190] [188] Central g 550 21 000 80 1800 0.33 7500 0.33 9900 690 17 500 170 850 (nm) 360 90 1000 160 170 35 270 55 615 150 450 100 8.1 3.3 Not determined 7.5 2.1 7.4 2.4 Not determined 29.5 11 dispersion compared to homogeneous SBS, where cific sizes are targeted, some centrifugation steps can the sediments are discarded. The procedure was thus be skipped. E.g. if medium-sized flakes are desired, the found to be ideal to study size effects in applications sample can be centrifuged at only two different cen- [452]. It was applied to WS2 and MoS2 [186, 245, 452, trifugal accelerations and the sediment redispersed. In 453] as Ni(OH)2 [240], GaS [188], BP [190] and FLG first approximation, flake size selection occurs by mass [176, 236] in solvents [176, 188, 190] as well as aque- in such a standard cascade, still making it difficult to ous surfactant [177, 186, 236, 240, 452] or polymer select large (µm), thin (1–3 L) flakes like in homoge- systems [453]. Due to its versatility and the large acces- neous SBS. However, a design of secondary cascades sible quantities, other reports adopted the procedure [186] involving a combination of long (14 h), low- [454, 455]. Typical mean lateral sizes and N are given in speed centrifugations (~1/5 of the lower g-boundary table III.1 for a range of materials. of the initial trapping) to remove thicker material and From this table it is clear, that the outcome of the size short (60 min), high-speed centrifugations (above the selection depends on the centrifugal accelerations, the higher g-boundary of the initial trapping) to remove density of the material and medium. A deeper under- very small (10–20 nm) flakes has shown potential to standing and systematic analysis will enable the predic- overcome this limitation yielding a ~75% 1L (mean N tion of the outcome of the size selection. In any case, to ~1.3) with an average lateral size ~40 nm in the case of achieve efficient size selection, it is critical to remove the WS2 [186]. supernatant from the sediment as completely as possible. This also means that for this procedure to work, the cen- III.1.3.1. Density gradient centrifugation trifugation time has to be long enough to allow the major- The key disadvantage of the SBS process, not allowing ity of the flakes to sediment to the bottom of the vial. This one to isolate a high percentage (>10%) of 1L flakes does not require centrifugation to equilibrium (unlike of reasonable lateral size (>100 nm), can be overcome DGU). One can obtain a good separation of supernatant with the exploitation of DGU. This enables control on and pellet-like sediment after 2 h of centrifugation, when N, when flakes are subjected to ultracentrifugation in a the filling height of the dispersion in the vial is  <10 cm. preformed DGM. Centrifugation times should be extended if greater filling Two approaches can be exploited for the separa- heights are used [246]. tion of GRM flakes: isopycnic [456] and rate zonal The chosen size-selection cascade can be modified separation [456]. Isopycnic separation allows spatial readily to suit a desired outcome. While this specific separation inside the cuvette of GRM flakes depend- procedure yields flakes sizes and thicknesses over a ing only on their buoyant density. This is defined as the broad size range in the different fractions, if only spe- density of the medium at the corresponding isopycnic 50 2D Mater. 7 (2020) 022001 C Backes et al point and is measured in g cm−3. GRM flakes, during except for the surfactant shell, making this an impor- isopycnic separation, move along the cuvette, until tant aspect of DGU. This fact makes the sorting of they reach the corresponding isopycnic point, i.e. the inorganic GRM flakes challenging in common DGM, point where their buoyant density matches that of the such as iodixanol, with ρ  =  1.32 g cm−1 [115, 469]. surrounding DGM [457]. Hence, in principle, flakes Therefore, in order to separate flakes by N, it is impor- should be sorted according to N. However, it is none- tant to ensure that they have uniform surfactant/poly- theless difficult to obtain dispersions of exclusively mer coverage [183, 446]. This requirement is crucial, large (micron-sized) 1L-GRMs. This is likely related to because the surfactant/polymer layer also contributes the fact that micron-sized SLG is a minority fraction in to the buoyant density of the entire system, thus result- LPE samples. Hence, even if all SLG flakes are isolated ing in slight variations of the buoyant densities of 1L- from the dispersion, the average lateral size will still be and FL- flakes, which can lead to isopycnic separation. in the range of hundreds nm. This results in a spatial separation inside the cuvette, For isopycnic separation, the choice of DGM is overcoming the limitations of conventional ultracen- fundamental. Salts (e.g. sodium chloride, cesium chlo- trifugation in a density constant medium. ride, and lithium chloride), sucrose, and iodixanol are A successful GRM flakes sorting by isopycnic sepa- the most widely used DGMs [458, 459]. The use of salts ration mainly depends on the following: (1) a large has some issues. First, they induce aggregation on the as possible amount of 1L-and FL- flakes in the start- hydrophobic solutes [460, 461] which may negatively ing dispersions. Differences in buoyant density can affect the sorting process. Second, density gradients be correlated to specific thickness (1L, 2L, 3L) only if produced with salts are less stable with respect to those individualized GRM flakes are encapsulated by the using sucrose and iodixanol, due to their lower viscos- surfactant micelles or polymer molecules. This also ity [462]. Sucrose suffers the opposite problem: it has implies the folding and wrinkling of the flakes will be high viscosity, which further increases exponentially at problematic. (2) Uniform flakes coverage. This is ruled high concentrations [463]. Iodixanol seems to be bet- by the adsorption of surfactant or polymer molecules, ter suited, due to density tuneability, with respect to and their aggregation, which can lead to clusters for- sucrose, and higher viscosity compared to salts. mation (i.e. aggregates of several molecules around the The effectiveness of the isopycnic separation is also flakes basal plane). The effectiveness of the isopycnic strongly determined by the type of density profile of separation is demonstrated by the yield of SLG to date, the DGM and its variation. During centrifugation the ~80% [183] with SC as surfactant or N separation of density profile redistributes as the DGM responds to GO [471]. the centrifugal force, resulting in a steeper gradient Isopycnic separation was also used to sort inor- over time [464]. The density profile of the DGM can ganic LMs [168]. One of the main issues is their high be: linear, nonlinear, or step [465, 466]. Linear gra- density. Thus, surfactant molecules, such as SC [469] dients are used when the materials to be sorted have cannot reduce their buoyant density sufficiently to small (<0.05 g cm−3) buoyant density differences. match the density of the gradient. By using Pluronic For this reason, linear gradients can sort surfactant- F68, a greater hydration shell than SC is formed. This micelle encapsulated CNTs with very small buoyant procedure allows to reduce the buoyant density of density difference, such as (6,5) (1.06 g cm−3) and (7,5) the encapsulated LMs [472, 473]. Pluronic F68 allows (1.08 g cm−3) dispersed in SC [467–469]. In nonlinear to reduce the buoyant density of BN [473] as well as gradients, the DGM density changes nonlinearly along TMDs such as WS2, WSe2, MoS2, and MoSe2 [472], the cuvette and the density gradient is established dur- within the limit of the DGM allowing for isopycnic N ing the ultracentrifugation process [465]. Nonlinear separation. gradients are particularly suited to sediment particles Another approach to sort inorganic LMs via isop- over the entire length of the cuvette [465]. By exploit- ycnic separation relies on the mixing of two DGMs ing nonlinear gradients, a variety of depth-density [474]. The addition of cesium chloride to iodixanol, profiles can be produced according to the density vari- determines an increase of the maximum buoyant ation. This allows the trapping of particles of different density supported by the DGM (i.e. 1.56 g cm−3) to densities along the cuvette’s length. Step gradients, the point where even high-density rhenium disulfide formed by stacking layers of different density, are usu- can be sorted layer-by-layer [474]. The addition of ally employed when the materials to be separated have salts as DGM is not compatible with dispersions based larger difference in density (>0.1 g cm−3) [470]. The on ionic surfactants, as these will crash due to charge sharp variation in density at the interface of two differ- screening. ent layers stops materials with a density smaller than Isopycnic DGU-based separation allows one to sort the denser layer, letting the larger density ones pass GRM flakes by thickness and lateral size, rather than through [470]. mass, unlike SBS. However, DGU separation, although For separating GRM flakes, the problem is that widely used in biology [195] and successfully applied these materials have high densities, with no intrinsi- to GRMs [168], has the drawback of being a process cally different density with N (e.g. MoS2, ρ  =  5.06 involving multiple iterations and, as such, it is time- g cm−3, WS2, ρ  =  7.5 g cm−3, WSe2, ρ  =  9.32 g cm−3), consuming. Another issue for the further processing 51 2D Mater. 7 (2020) 022001 C Backes et al Table III.2. Specifications to consider for different printing methods. Viscosity Layer thickness Feature size Registration Throughput Printing method [Pa s] [µm] [µm] [µm] [m2 s−1] Gravure printing 0.05–0.2 [477] 0.8–8 [477] 75 [481] >10 [481] 60 [481] <0.1 [478–480] 3–30 [482] Flexography printing 0.05–0.5 [477] 0.8–2.5 [477] 80 [481] <200 [481] 10 [481] 3–30 [482] Offset printing 30–100 [477] 0.5–2 [477] 10–50 [481] >10 [481] 5–30 [481] 5–50 [483] 3–30 [482] Screen printing 0.5–50 [477] 3–15 [477] 20–100 [481] >25 [481] 2–3 [481] 0.015 [484] 30–100 [481] of the sorted flakes, i.e. ink formulation, is linked with slot-die and ink-jet printing, the formulation of the the presence of the DGM in the dispersion, that may be ink is more complex, requiring the addition of many problematic for processing and/or its removal. components in pre-defined steps. In the following, a Another methodology to sort GRMs in a centrifu- few examples are given. gal field is rate zonal separation (RZS) [457, 475]. In RZS, the ultracentrifugation is stopped during the III.2.1. Tuning of rheological properties transient centrifugal regime, i.e. before the mat­erials The control of the ink rheology is key to ensure under ultracentrifugation reach their isopycnic points consistency and reproducibility. This is usually [475]. RZS exploits differences in the sedimentation achieved by studying flow and deformation of the coefficient of particles under ultracentrifugation. materials under external perturbation [485]. The Therefore, particles with different sedimentation complex structure of the ink determines a rheological coefficients will move along the cuvette at different behavior which is typical of non-Newtonian flow sedimentation velocities [457, 475], determining their [486]. Thus, there is a non-linear relation between spatial separation [441, 475]. RZS was used to separate shear rate γ̇ shear stress (τ), and apparent viscosity η GO flakes with different lateral size [476], exploiting (ηapp  =  σγ̇ ) which depend on γ̇ (or τ). This behavior the fact that the larger is the flake lateral size, the larger is different from Newtonian liquids, where η  =  σγ̇ is the sedimentation coefficient. RZS was also used for [487]. The shear rate determines the η behavior of inks the lateral size selection of BP flakes [425]. Similar to and, thus, their range of application. Depending on DGU, a downside of RZS is that usually an ultracen- their flow behavior, there are different classifications trifuge is used (rather than a benchtop centrifuge) for non-Newtonian inks [488]: dilatant where ηapp and that a DGM is added to slow down the sedimenta- increases with increasing γ̇ [488], pseudoplastic, tion and lead to a more gradual distribution along the where ηapp decreases with increasing γ̇ [488], and cuvette, which has to be removed prior to further pro- thixotropic, where ηapp decreases with time under a cessing. This contrasts to band sedimentation [185] as constant deformation and will start to rebuild once the discussed above. shear force ηapp is removed [488], figure III.3. The ink rheology is strongly influenced by the GRM volume III.2. Inks formulation fraction, ϕ, with their shape and spatial arrangement The production of GRM flakes in dispersion [489], GRM flakes dispersed in a liquid affect its flow following exfoliation and sorting finds a direct use for field resulting in an increase in energy dissipation due applications, such as in polymer composites. In many to fluid-flake and/or flake-flake interactions [489, other cases, the as-prepared dispersion cannot be 490]. These interactions increase with ϕ, restricting used directly. For applications requiring coating and the flakes diffusion into small ‘domains’, formed printing, the dispersion needs a formulation process by the nearest neighbors and η diverges. However, a to obtain an ink. The composition of functional inks theoretical model to predict η of fluids with dispersed is linked with the target deposition/printing process GRMs does not exist yet. Such model is not even [166], see table III.2. There are deposition/coating available for NPs in general, although a significant approaches where the ink formulation is not stringent, research effort has been carried out [491–493]. depending on the nature of the solvent used for We now discuss the main ink requirements and exfoliation and sorting, the deposition can be carried the state of the art of the formulation for the of GRM- out directly with the as-prepared dispersion. This is based inks for the main deposition/printing tech- the case for vacuum filtration and transfer, spray and niques. drop casting, where the most critical parameter is the Coating approaches, such as drop and spin coat- morphology control of the dispersed flakes. For other ing, do not require a structured ink formulation, and, printing processes, such as flexographic, gravure, usually, the as-prepared dispersion can be directly 52 2D Mater. 7 (2020) 022001 C Backes et al Figure III.3. Newtonian and non-Newtonian flow behavior of functional inks. Adapted from [166], © Wiley-VCH Verlag GmbH & Co. KGaA. deposited onto the target substrates. On the contrary, [499, 500] and the inverse of the Ohnesorge number, although the ink for spray coating deposition does not Z (1/NOh) [500]. The Z number is independent on the require a high viscosity, the aerosol formation and the drop velocity (m · s−1) and it is the most used FOM for subsequent drying process need the control of several inkjet printing [501–507]. Different Z values have been processing parameters [494] such as viscosity [495], proposed, ranging from 1 to 91 with different func- flow rate [496], and the distance substrate/spray-noz- tional materials [501–507]. zle [497]. The nature of the solvent, which determines Other parameters for the ink formulation have to its evaporation rate at the substrate surface [496], be considered, such as, e.g. the physical dimensions is also a critical point for a uniform deposition. The of the flakes dispersed in the ink, that can agglomer- solvent vapor pressure is crucial. Solvents having low ate and clogg the print-head. Usually, it is required that vapor pressure, such as NMP (0.5mm Hg @25 °C), one flakes have lateral sizes smaller than ~1/50 of the noz- of the most common for LPE of graphite [164], pro- zle diameter [508] (typically ~100 µm [509]). Other mote long crystal formation time [496]. This allows a parameters to be optimized for the printing/deposi- consistent structural arrangement prior to the trans­ tion process are wetting and adhesion [510] to the sub- ition to the solid phase of the deposited feature [496]. strate, as well as the distance between nozzle and sub- On the contrary, high vapor pressure solvents, such strate (typically 1–3 mm) [511]. as, e.g. isopropyl alcohol (IPA, 33mm Hg @25 °C) and An even more structured inkis needed for flexog- toluene (24 mm Hg @25 °C), promote faster solvent raphy and gravure printing. Although both require a evaporation [1741]. This subsequently reduces the high viscosity ink, there are several differences [166]. ability of the flakes to organize in a ordered manner. In flexography, there is a plate with a raised surface that The consequence will be a decrease of the final device spreads the ink onto the target substrate. Gravure is an performance, such as, e.g. a reduction of the charge intaglio process, where the ink is first transferred onto carriers mobility in FETs [496]. Other sources of insta- an image carrier and then printed on the target sub- bility for the coating are linked with the geometry of strate. As summarised in table III.2 and Ref. [166], gra- the spray coating system, i.e. small nozzle size, or the vure printing is usually faster (~1000 m min−1) than jetting pressure, i.e. low pressures reduce the flow rate, flexo (~500 m min−1), with printed stripes thicker which produce scattered droplets on the substrate and (~1 µm in gravure) than in flexography (<1 µm). eventually a non-uniform coating [496]. A continuous The viscosity is also different: in flexography this is in coating is obtained by high deposition rates [145, 496]. the 50–500 mPa · s range, while in gravure it is higher The uniformity of the coating is also determined by the 100–1000 mPa · s. appropriate substrate/nozzle distance [497]. If this too GRM-based inks for flexography and gravure short, the ink previously deposited onto the substrate were formulated [512, 513]. Ref. [512] reported a flex- can be blown away by the incoming flow. If it is large, ography ink by using CMC as a binder for FLG flakes the solvent can evaporate during the flight time, i.e. in a water/isopropanol solution, obtaining a η ~20 before the droplets reach the substrate. mPa · s. The as-formulated ink was printed onto a flex- The formulation of the ink for an inkjet process ible ITO substrate at 0.4 m s−1, and used as counter requires the control and tuning of various liquid prop- electrode for the realization of a flexible dye-sensitized erties such as ρ, η, and γ [498], which are then sum- solar cell (DSSC) [512]. A FLG/terpineol ink (η ~0.2–3 marized in the following dimensionless figures of Pa · s from the homogeneous dispersion of FLG-ethyl merit (FOM): the Reynolds (NRe) and Weber (NWe) cellulose powder (5–10 wt%) in a mixture of ethanol/ 53 2D Mater. 7 (2020) 022001 C Backes et al 70 WaterEthanol Pluronic108 PVP55000 GumArabic 60 PVP36000 PEO10000 PVP1300000 Sodium CMC 50 Viscosity (mPa s) 40 8 6 4 2 0 20 40 60 80 100 Shear rate (1/s) Figure III.4. Rheology modifiers in 50:50 water/ethanol solution at 1 wt% loading. terpineol (2.5:1 volume%) was reported in Ref. [513]. Figure III.4 shows a selection of polymers at For the formulation of high viscosity GRM-based inks, 1 wt% loading that can be added to water or 50:50 there are several issues to be addressed. First, the sol- water/ethanol solutions to modify the rheology. vents used in LPE (e.g. NMP) are toxic and have very Polymers such as polyvinylpyrrolidone (PVP) can low η (<2 mPa · s). The use of alternatives such as be added, with increased viscosity with increasing volatile alcohols, e.g. ethanol [513] and isopropanol molecular weight (MW). Celluloses which are com- [512], is against the current environmental regula- monly used, provide high viscosity at low loadings, tions, requiring the minimization of volatile organic and impart thixotropic behaviour. E.g. aqueous FLG- compounds [514]. Second, the GRM concentration based inks produced by microfluidization [178] have in these solvents is low (<1 g l−1). Thus, in order to FLG loading  >100 g l−1 and are formulated using 1 produce a functional (e.g. conductive) film, many wt% sodium carboxymethylcellulose (NaCMC) to print passes are required. The increase of GRM fillers provide η up to 1800 mPa · s. could determine a problem with the solvent evapora- NaCMC is not easily dissolved and its preparation tion and in-series printability. In this context, the use can be affected by factors including the NaCMC parti- of high boiling point solvents requires post-processing cle size, rotational speed of the mixer, the rate at which annealing for solvent removal [515], thus posing limi- solutes are introduced, and the solution T [523]. Heat- tations on target substrate. ing the FLG dispersion should be avoided as this may By increasing η, it is possible to obtain a paste, promote flocculation. NaCMC is water-absorbent and needed for screen printing. FLG-based conductive- has high water retention, therefore it clumps. To pre- pastes (i.e. FLG [516], RGO [517–519] and flakes vent this, NaCMC is added until dissolved completely [520, 521]) are emerging as possible alternatives to and stirred to avoid bubbles. Bubbles and clumps need those based on metal NPs. to be removed prior to printing. A paste based on FLG powder dispersed in terpi- The rheological properties of inks were investi- neol, with ethylcellulose as a binder was prepared in gated by monitoring the elastic modulus G′[J m−3  = [516]. The paste contained flakes at a concentration Pa] [524], representing the elastic behavior of the of ~80 g l−1, having a η up to ~10 Pa·s at a shear rate material and a measure of the energy density stored of 10 s−1. The as-deposited paste required a thermal under a shear process [524], and the loss modu- treatment at 300 °C for 30 min to remove ethylcel- lus G″[J m−3  =  Pa] [524], representing the viscous lulose. Another FLG-based paste, having shear thin- behavior and a measure of the energy density lost ning behaviour, was discussed in Ref. [522], which during a shear process due to friction and internal a gelation approach starting from the dispersion motions [524]. Flow curves measured by increas- of expanded graphite in methyl ether dipropyl- ing γ̇ from 1 to 1000 s−1 at a gap of 0.5 mm, because ene glycol in the presence of polyvinyl acetate and this γ̇ range is applied during screen printing. Fig- polyvinylpyrrolidone acting as binders. However, ure III.5(a) plots the steady state viscosity of an ink many parameters still need to be optimized for the produced from microfluidization (100 cycles) con- development of optimal GRM-based inks/pastes for taining 73 wt% flakes as a function of γ̇ . NaCMC screen printing. imparts a drop in viscosity under shearing, from 570 54 2D Mater. 7 (2020) 022001 C Backes et al Figure III.5. (a) Viscosity as a function of shear rate for an ink with 73 wt% flakes, (b) viscosity at 100 s1 for different wt% of FLG flakes. Reproduced from [178]. mPa · s at 100 s−1 to 140 mPa · s at 1000 s−1. This is of surface tension, viscosity, solute concentration thixotropic [525], since the viscosity reduces with γ̇ . and particle size for the ink to be deposited. Viscosity, This behavior is shown by some non-Newtonian volatility and concentration will affect the morphology fluids, such as polymer solutions [526] and biologi- and the thickness of the final film, which can be cal fluids [527]. It is caused by the disentanglement of adjusted by tuning the angular velocity and the time poly­mer coils or increased orientation of polymer coils of processing. The main advantages of spin coating are in the direction of the flow. In contrast, in Newtonian also the low amount of material required (µl) and the liquids, the viscosity does not change with γ̇ [527]. reproducibility of the film features for fixed working Refs. [528, 529] reported that thixotropy in NaCMC parameters. solutions arises from the presence of unsubstituted Ref. [530] produced a FLG dispersion in 1,2-dichlo- (free) OH groups. Thixotropy decreases as the number robenzene (ODCB) due to its compatibility with of OH groups increases [528, 529]. Figure III.5(b) plots P3HT:PCBM by sonication and subsequent cen- the viscosity at hundred s−1 as a function of wt%FLG trifugation to remove unexfoliated graphite. For the flakes (70 microfluidic process cycles). The NaCMC P3HT:PCBM dispersion, 10 g l−1 P3HT and 8 g l−1 polymer (10 g l−1 in water) has a µ ~0.56 Pa · s at hun- PCBM are dissolved in ODCB and mixed with the dred s−1, and drops to 0.43 Pa · s with the addition of 5 FLG-ODCB supernatant. The P3HT:PCBM:FLG is wt% flakes. The loading wt% of flakes affects γ̇ , which then spin-coated on the substrate, which is a stack of increases at 51 wt% and reaches 0.6 Pa · s at 80 wt.%. glass, 100 nm ITO and PEDOT:PSS also deposited by spin coating. The BHJ solar cell is finalised by deposit- III.2.2. Hybrid inks ing a counter electrode of 20 nm Ca and 60 nm Al by III.2.2.1. Graphene/polymer blends evaporation. The critical step of the process is the mix- We describe the formulation of a FLG-polymer- ing of the P3HT:PCBM to the ODCB-FLG superna- fullerene ink that can be processed by spin-coating tant, since the addition of the P3HT:PCBM could cause to fabricate bulk heterojunction (BHJ) solar cells. the destabilisation and flocculation of the ink. To pre- The goal is to improve BHJ solar cells performance in vent this destabilisation, the P3HT:PCBM should be terms of hole mobility, photocurrent efficiency and added gradually and the dispersion immediately stirred photoresponsivity [530]. Spin coating is commonly after mixing. It is also important to not add an excess used to fabricate BHJ solar cells, allowing rapid amount of FLG. In Ref. [530], only 2 wt%FLG is added; deposition of thin films (~hundreds nm thickness) further increases cause aggregation of the FLG sheets, of different materials. In contrast to other printing having a deleterious effect on exciton generation and techniques, it does not have any requirements in term charge separation/transport in the active layer. 55 2D Mater. 7 (2020) 022001 C Backes et al (a) (c) 40 30 20 10 ∆R (%) 0 (b) -10 20 40 80 -20 -30 20 passes 40 passes 80 passes 5 mm -40 -0.4 -0.2 0.0 0.2 0.4 220 MΩ 31 MΩ 3.7 MΩ Strain (%) Figure III.6. Inkjet printing of microfluidized ink after 20 process cycles. (a) Typical jetting behaviour showing good droplet formation and the absence of satellite droplets. (b) Printed serpentine-type resistors after 20, 40 and 80 passes. (c) Resistance change whilst 3-point bending of three devices under tensile and compressive strain. III.2.2.2. Graphene/CNT films to improve the final TCF conductivity. The same pro- Transparent conductive films (TCFs) on flexible cedure is applied for the deposition of the FLG ink on substrates require homogeneity, optical transparency the CNT film. Generally, for a 10  ×  10 cm2 PET film, and stability under mechanical stress. Composites of 40 repetitions of both inks are required to get to tens of graphene and CNTs are promising in this regard. In kΩ □−1 at 80% transparency. this framework the deposition of films by rod coating We find that these coatings can be applied in the [531] fulfils the above-mentioned requirements. The fabrication of CNTs/FLG films as pixel electrodes use of the so-called Meyer bar allows deposition at RT directly on the backplane of an electrophoretic dis- and is fully scalable over large areas. The diameter of play. The CNT/FLG ratio is 40:60% in mass, assuring the wound wire of the bar determines the size of the the uniformity of the film avoiding any aggregation. grooves and controls the thickness of the liquid film, After every 3 coatings the film is washed to remove as well as giving alignment to the material improving surfactants. For a 4  ×  4 inch2 backplane, 150 ml of the the percolation. The key parameters of the ink to water-ethanol solution (50:50 in volume) are used. ensure a homogenous coating are viscosity ~1 mPa · s and a small contact angle ~20–30° [532]. The speed III.2.2.3. Inks with conductive polymers of the deposition can affect the homogeneity creating In order to overcome the obstacles of efficient micro- aggregates. Aggregates can also arise from a fast drying supercapacitors (MSCs) manufacturing, such as of the liquid film. scalability, high cost, and flexibility, new strategies In addition to published work [1742], we found based on processing of FLG nanocomposite inks that a suitable ink can be prepared by bath sonicat- are appealing. Direct printing/transfer techniques, ing (9 h) natural graphite (0.1 g l−1) in aqueous Tri- including spray deposition, inkjet printing, and ton X-100 (0.01 g l−1) followed by centrifugation filtration, offer a promising protocol for production of (10 000 rpm) to remove the unexfoliated flakes. A MSC arrays, which can be processed onto both plastic CNT dispersion was prepared by tip sonication (2 h) and paper substrates at low T and over large areas. As P3-SWNT powder (Carbon Solutions, Inc.) in 1.8 a key component for these techniques, the selected g l−1 aqueous Triton X-100 followed by centrifugation ink should be stable, low cost, and easily printable on (25 000 rpm). These CNTs were purified with nitric appropriate substrates; it should display excellent acid, and left in highly functionalized form contain- electrical properties without the need for aggressive ing 1.0–3.0 atomic% carboxylic acid moieties which post-treatments. High quality FLG obtained from can be derivatized with a variety of functional groups electrochemical exfoliation could meet the above [1743]. TCFs can then be conveniently fabricated on requirements. various flexible substrates such as PET, PEN etc by Two inks were developed based on electrochemi- rod coating. Typically, for a 10  ×  10 cm2 substrate, 80 cally exfoliated FLG (EEG) and conductive polymers µl CNT ink is deposited and spread uniformly on the [533, 534]. In one, conducting polymers, such as substrate using the Meyer bar. The wet film is allowed PEDOT:PSS, are mechanically mixed (bath sonication to dry in air before coating another layer. To remove and stirring) with EEG in isopropanol [533]. The other the surfactant, after 2–3 coatings, the substrate is is a EEG/polyaniline (PANI) composite ink [534], soaked in water-ethanol (50:50 in volume) solution at based on in situ polymerization of aniline in the pres- 60 °C and dried with nitrogen. The washing steps help ence of EEG. First, EEG is prepared and dispersed in 56 2D Mater. 7 (2020) 022001 C Backes et al DMF at the concentration of 2 g l−1, followed by non- persion media, the size of the dispersed materials to covalent functionalization with 1-pyrenesulfonic acid be filtered and the desired approach to carry out film sodium salt to obtain water dispersion of EEG. Then transfer. As the flakes produced by LPE often have lat- the EEG/DMF dispersion is blended with aqueous eral dimensions in 10s-hundreds nm, it is often nec- aniline hydrochloride and gently stirred or sonicated. essary to use filtration membranes with the smallest Afterwards, ammonium persulfate is added, followed possible pore diameters. Mixed cellulose membranes by a vigorous stirring (10 000 rpm, 6 h) to form uni- with 25 nm pore size are compatible with aqueous/sur- form dispersion of EEG/PANI. After washing through factant dispersion media and can be subsequently dis- ultra-centrifugation with water and ethanol for 3 solved using acetone to achieve transfer of the filtered times, the composite is dispersed in ethylene glycol to film. get a stable EEG/PANI ink [534]. FLG films with thickness ~150 nm, transmittance ~50%, roughness ~8 nm and Rs ~ 1–2 kΩ sq−1 were III.3. Printing and deposition of inks prepared as conductive electrodes for biological appli- To develop applications that make use of LPE GRMs, cations [432]. These were also exploited in photonic it is necessary to convert the dispersed material into applications, demonstrating sub-50 fs compressed structures suitable for further studies. In general, flakes pulses from a FLG-mode locked fiber laser [539], may either be deposited on a supporting substrate power scaling lasers [540], and monolithic waveguide (such as glass, metal foil, Si, or plastic film) or act as lasers [541]. To produce these, a 25 mm diameter a filler within some host material (such as polymers, membrane filtration setup was used. 500 µl of disper- metals or other nanomaterials) to produce free- sion with concentration ~0.1 g l−1 was diluted in 1 ml standing composites. To this end, numerous liquid- DI water. The dilution is necessary when the amount phase processing techniques have been developed for of the filtered dispersion is less than 1 ml to ensure uni- the deposition of GRM dispersions in combination form coverage of the membrane. If the dispersion has with other dispersed nanomaterials or polymers. Here, a high surfactant concentration, the presence of foam we outline some approaches used to produce a range on the liquid surface has to be avoided before vacuum of GRM-containing structures and devices such as filtration. FLG is filtered through a 100 nm pore-size vacuum filtration, spray coating, ink-jet and screen nitrocellulose (NC) filter membranes (Millipore). In printing. order to remove the residual surfactant, the film on the membrane can be rinsed by filtration of 20 ml DI III.3.1. Vacuum filtration water. The filtering process requires ~2 h, which will Vacuum filtration is a simple technique whereby a increase with the film thickness prepared. GRM dispersion can be filtered through an underlying These films can be transferred on to glass coverslips porous membrane to obtain nm-to-several micron (25 mm diameter) using the following procedure: 1. thick films [535, 536]. These can be removed from the FLG/NC membrane is cut to the size required and the underlying support to produce a freestanding soaked in deionized water. 2. The FLG/NC is placed film, or transferred to rigid and flexible substrates on the glass substrate with the FLG side face down. 3. and the porous membrane removed e.g. by chemical A sheet of cleanroom paper is pressed by hand on the dissolution [536]. Although this technique may be back of the NC to remove the excess water and trapped unsuitable for the production of GRM-containing air between the film and glass. This is a critical step, as structures for commercial purposes, its low cost and trapped air can leave some FLG non-transferred and, its simplicity makes it ideal for research purposes. therefore, holes or fractures in the final film. The sub- The main advantage is the possibility to wash away strate is then clamped with two bulldog clips between surfactants and solvent residues that usually affect two glass slides (30  ×  30 mm). 4. The sample is left the performance (e.g. conductivity) of the final films. to dry in an oven for ~2 h at 90 °C. 5. The glass slides Films with different thicknesses and composition are removed and the coverslip left to soak in a covered can be produced by tuning the amount of dispersion, beaker of acetone to dissolve the NC (at least 24 h) the pore size of the membrane and the vacuum then soaked in isopropyl alcohol (1 h) and in deion- pressure. GRM dispersions can be filtered on to ized water (1 h). 6. The final FLG/glass sample is then various membranes, selected based on their pore dried in an oven for 2 h at 90 °C. Transfer on flexible size and chemical compatibility. Membrane types substrates can be done with the same procedure. For include cellulose, polycarbonate, anodized alumina, improved transfer it is possible to use a hydraulic press PVDF or PTFE. This allows to produce composites in step 3. such as layered oxides combined with SWNTs for Another method to make large free-standing supercapacitor electrodes [239, 537], and TMDs/ membranes is electrophoretic deposition [542]. A SWNT composites for electrocatalysis [452, 538] can suspension of GO with a concentration of 3 mg ml−1 readily be fabricated. is prepared and used as the electrolyte. Two identical A wide range of filtration membranes are avail- pieces of well-polished Cu sheets (30  ×  50  ×  2 mm) able with different surface chemistries and pore-sizes, are used as anode and cathode. A constant current of the choice of which depends on the nature of the dis- 2 mA is applied between the two electrodes for 2 min. 57 2D Mater. 7 (2020) 022001 C Backes et al The anode coated with GO is then dried under vac- A FLG-NMP ink with a concentration ~1 g l−1 and uum overnight, and then the free-standing GO mem- FLG flakes lateral size up to 600 nm was printed for 50 brane can be peeled off. printing passes with an interdrop distance ~27 µm, inter-layer delay time ~ 120 s, and printer platform T III.3.2. Inkjet printing ~60 °C. Annealing was performed at 80 °C in vacuum The printing of FLG inks is an attractive approach overnight. The strain gauges with a resistance ~100 for developing electronic applications, such as kΩ show had gauge factor(GF ~20 at 25 mm bend- transparent conductive films, printed electrodes and ing radius with a change in resistance ~5% after 1000 sensors. Inkjet printing is the ideal tool to produce bending cycles. FLG-based strain gauges with similar prototype electronic devices at the lab scale. Indeed, properties were printed on planarized PEN using a it enables high resolution (~25 µm) of the pattern FLG-ethanol ink (2 g l−1) to make fully flexible weara- associated with the possibility of easily changing the ble devices. The sensors are aligned with the proximal pattern and the printed material. Several examples of inter-phalangeal joints of the first and second digit of inkjet printing of nanomaterials have been reported the hand, allowing the extension-flexion movements [201, 447, 543, 544]. In order to have a good jettability of these digits to be monitored. 20 printing passes, an ink has to satisfy particular requirements on Z, 22 µm inter drop distance and RT were used with an γ ρ, η and nozzle diameter. If Z is in the 4–14 range, additional annealing at 80 °C in vacuum for 2 h. satellite droplets recombine before reaching the With aqueous-surfactant based inks, one problem substrate, but Ref. [447] demonstrated the possibility is the formation of foam bubbles following agitation of having good printing also outside this range. Most of the print-head. This is observed at surfactant con- lab-scale inkjet printers equipped with disposable centrations  >6 g l−1. The presence of non-conduc- cartridges, (e.g. Fujilm Dimatix 2800, Ceradrop X, tive species (surfactants, polymers) in the resulting LP50) mount cartridges with 50 µm diameter, so printed films will reduce FLG network connectivity the flakes dispersed in the ink have to be at least 1/20 and, thus, conductivity. The ability for microfluidic of the nozzle diameter to avoid printing instability processing to stabilize FLG at low surfactant concen- and nozzle clogging. Dispersions produced by trations (0.5 g l−1) meant foaming does not occur. The LPE followed by centrifugation are particularly suitability of 0.5 g l−1 SDC stabilized FLG inks after 20 suitable for inkjet printing, since they contain flakes process cycles and centrifuged at 10 krpm, 1 h can be with lateral sizes  <2–3 µm [164]. Inks made by tested for inkjet printing (figure III.6(6)). Serpentine- electrochemical exfoliation contain larger flakes even type FLG resistors are printed on to photo quality after centrifugation. These should be filtered before paper (HP Advanced) at room temperature using an printing. interdrop distance of 18 µm, with 20, 40 and 80 print- Inkjet printing can be performed on a variety ing passes (figure III.6(b)). Prints are left to dry at RT of rigid (glass, Si/SiO2) and flexible (PET, PEN, PI) for 60 s between passes. Although minimum optim­ substrates. However, the wettability of ink droplets, ization is required, prints with over 20 layers suffer defined by the contact angle, varies significantly. The bleeding, due to insufficient drying time, and noz- most common procedures for changing the surface zles could become partially clogged. During printing, properties is to reduce the contact angle (improve the head is cleaned by an automated Spit-Purge-Spit wettability) by treating substrates. To increase con- cycle every 180 s. Patterns are printed for 20 passes at tact angles, the substrate can be silanized by spin a time and the nozzles wiped by hand with IPA prior coating hexamethyldisilazane (HMDS) (40 s at to printing more layers. After 20, 40 and 80 passes 1000 rpm, followed by annealing at 80 °C for 2 min) the resistance of the patterns decreases from~220, to [447]. Silanization requires surface hydroxyl groups 31, 3.7 MΩ, respectively, over ~40 mm (track width to bind to. For flexible substrates, a preliminary UV ~0.9 mm) as shown in figure III.6(b). The resistance ozone or oxygen plasma treatment will help to gen- response of these devices is tested with strain (3-point erate surface oxides. This results in more confined bending) and in different gas environments. droplets, less spreading of the ink and higher resolu- Three-point bending is performed with a tion [447]. 300 N/2 kN Deben vertical three point tensile stage. Flexible substrates already optimized for inkjet The samples are bent with an extension rate of can be found on the market i.e. glossy papers, Nov- 0.5 mm min−1 and measured under tension (posi- ele (PET/SiO2/PVA) [545], etc which allow the quick tive strain) and compression (negative strain), to wicking away of solvent, reducing the coffee ring effect a maximum of 5 mm displacement. The electrical thanks to their porosity. resistance of the printed strain gauge varies, depend- Below, we report different examples for the pro- ing on the amount of axial bending strain in the duction of strain sensors on flexible substrates from device. Printed structures at 20 and 40 passes show inkjet-pronted FLG. A flexible integrated plat- very similar resistance versus strain curves (figure form, with four linear strain gauges 3 mm  ×  20 mm III.6(c)) with significant response to  +0.4% strain and a central Wheastone bridge strain gauge of (ΔR ~  +35%) and  −0.4% strain ΔR ~  −25%). 1.5 cm  ×  1.5 cm was printed on Novele substrate. The 80 printed pass sample show significantly lower 58 2D Mater. 7 (2020) 022001 C Backes et al Figure III.7. Schematic of the screen printing process. FLG ink is applied to the screen and the squeegee is lightly passed over the screen to fill the mesh. This is then passed over the screen at a higher pressure, such that the ink is pushed through the mesh onto the substrate. The wet film is then left to dry to produce the final design. Reproduced from [553], with permission from Elsevier. Figure III.8. (a) Demonstration of screen printing, (b) capacitive touchpad design (29 cm  ×  29 cm) printed on paper, (c) the line resolution is 100 µm. Reproduced from [178]. response (ΔR ~  ±10%) in line with the higher sam- of the other. Inkjet printed MoS2/FLG photodetectors ple thickness. These changes are a result of conductiv- (PDs) using LPE inks were demonstrated on PET. The ity variations, due to changes in length, as well as the devices were fabricated using a Dimatix 2800. A 40 µm distance between flakes. The relative change in elec- MoS2 channel contacted by two inkjet printed FLG trical resistance (ΔR/R0) is related to the mechani- electrodes on top. 100 passes of both FLG and MoS2/ cal strain (ε) by the gauge factor (G  =  (ΔR/R0)/ε). NMP based inks were printed at 60 °C using 25 µm These are 93, 85 and 39 for the 20, 40 and 80 print pass interdrop distance and 300 s interdelay time. MoS2 was samples, respectively and thus comparable to other annealed at 90 °C in vacuum for 9 h before printing FLG devices produced by drop-casting or spray coat- FLG. Finally the entire device was annealed overnight ing (G: 75–150 at 0.2% strain) [546]. These gauge fac- at the same conditions. The good performance of the tors are higher than graphite-based ink strain sensors PD showed that the MoS2 properties could be retained (G: 19.3  ±  1.4) [547], and considerably higher than after printing. Similar results were reported in Refs. conventional strain gauges [548] or inkjet-printed [201, 550], confirming that inkjet-printing can become metal strain gauges [549] in which GFactor ~ 2. an efficient method to produce functional devices for a In addition we used inkjet printing to print het- variety of applications including photonics, optoelec- erostructure devices by placing one material on top tronics and energy storage. 59 2D Mater. 7 (2020) 022001 C Backes et al III.3.3. Screen printing pensive, yet highly efficient strain sensors [567, 568]. For flexible electronic devices, e.g. organic E.g., when FLG is added to a lightly cross-linked poly- photovoltaics (OPVs), Rs  <  10 Ω □−1 is required silicone (often referred to as silly putty [1744]), electro- [255], while for printed radio-frequency identification mechanical sensors with GF  >  500 have been realized (RF-ID) antennas one needs a few Ω □−1 [257]. that can measure pulse and blood pressure [567]. To minimize Rs, µm range films are deposited In (opto)electronics, the most widely pursued use using screen printing [551–554]. Screen printing of inks based on FLG and GRM is the fabrication of is a commonly used industrial technique for fast, transparent conductors. The electrical conductiv- inexpensive deposition of films over large areas. It also ity of FLG-based conductive films is typically in the allows patterning to define which areas of the substrate range of  <2–4 S m−1 (corresponding to Rs of  >20–10 receive deposition. The screen is typically a polyester Ω □−1 for a 25 µm film) [166] much lower than the mesh with a cured lacquer stencil on top that provides conductivity of films from metal inks which domi- the desired pattern. Mesh grades vary from ~10 up to nate the field of flexible transparent conductors due 180 threads cm−1. The film or emulsion stencils range to their high conductivities of 107 S m−1 [569]. How- from a minimum of 12 µm to  >300 µm. The ink must ever, since graphite is a much cheaper starting mat­ have high viscosity (>500 mPa s) [555, 556], because erial than noble metals, FLG inks might nonetheless be lower viscosity inks run through the mesh rather than useful for applications that do not require very low Rs dispensing out of it [555]. To achieve this, typical at high transparencies (e.g.  <10 Ω □−1 with transmit- formulations of screen inks contain a conductive filler, tance  >90% for photovoltaics). E.g., electrochromic such as Ag particles,[557] and insulating additives windows, smart windows and electromagnetic shield- [558]at a total concentration higher than 100 g l−1 ings used in every-day life can cope with Rs~ few hun- [558]. Of this, >60 g l−1 consist of the conductive dreds of Ω □−1) [166]. Significant improvements have filler needed to achieve high σ ~ 102 S m−1 [557, 559]. been made with respect to improving both ink form­ A squeegee is used to fill the mesh with ink before a ulations and deposition techniques. E.g. Ref.[178] high pressure is applied to force the ink through the demonstrated the fabrication of blade-coated FLG mesh onto the substrate [558]. Figure III.7 shows a films with Rs 2 Ω □−1 for a 25 µm film (conductiv- schematic of the screen printing process. Ref. [560] ity of 2  ×  104 S m−1) suitable for electrodes in organic provides a good introduction to screen printing. photovoltaics. The highest conductivity (4   ×  104 −1 The printability of inks with ~80 wt%FLG flakes S m ) of an FLG film produced by inkjet printing after 70 microfluidic processing cycles formulated [570] exploited nitrocellulose as stabilizer, which also with 1 wt%NaCMC was tested using a semi-automatic improves the mechanical flexibility. These examples flatbed screen printer (Kippax-2012-BU) and a Nat- demonstrate that FLG-based inks might be suitable graph screen printer (figure III.8(a)), both equipped for low-end applications as transparent conductors if with screens with 120 mesh count per inch. Trials printing and ink formulation can be further improved were made onto (PET) (125 µm thickness, PMX729 and up-scaled. HiFi Industrial Film Ltd) and paper substrates. Since graphene is not a semiconductor, it is not Figure III.8(b) shows a 29 cm  ×  29 cm print on paper suitable as channel material in TFTs. While achiev- with a line resolution ~100 µm (figure III.8(c)). The able mobilities can be as high as 200 cm2 V−1 s−1, the printed pattern (figure III.8(b)) can be used as a capac- on-off ratios are limited to  <5 even in the state-of-the itive touch pad in a sound platform that translates art [571]. In this regard, devices produced from inks touch into audio [561]. of semiconducting LMs are more promising. Indeed, all electrolyte-gated TFTs were reported [572] using III.4. Applications TMD as channel material. In this case, on/off ratios Proof of concept applications have been demonstrated 600 were obtained, but mobilites were low (0.1 cm2 for dispersions and inks produced from a range of V−1 s−1). A potential way forward to further improve LMs in diverse areas from nanocomposites, (opto) the TFT device characteristics was suggested in Ref. electronics and photonics to sensing and energy storage [573]. By optimizing the electrochemical intercalation and conversion. A few examples will be given below. of quartery ammonium molecules, it was possible to For a more detailed and comprehensive description, we produce phase-pure semiconducting flakes with uni- referred to review articles [166, 562–564]. form thickness and relatively large size (0.5–2 µm). In nanocomposite applications, usually no com- This improved the stacking and thus charge transport, plex ink formulations are required and dispersions can resulting in TFTs with on/off ratios and mobilities~ be used without tuning of rheological properties. FLG, 10 cm2 V−1 s−1 outperforming organic electronics. GO and RGO can be used as filler in polymer [564], Semiconducting flakes have also been exploited ceramic or metal composites [564, 565] to improve as active materials in PDs [562]. Ref. [201] uses inkjet mechanical, electrical and thermal properties of the printing to fabricate PDs with interdigitated FLG elec- host material. Aligning the filler can be beneficial to fur- trodes and a MoS2 channel in a lateral heterostructures ther boost the performance [566]. FLG can also be used achieving a tenfold increase compared to the dark as additive in viscoelastic polymers to fabricate inex- conductance. By optimizing the ink formulation and 60 2D Mater. 7 (2020) 022001 C Backes et al printing of water-based dispersions, all inkjet-printed [581], a scalable, high-fidelity manufacturing of FLG PDs were produced on plastic using graphene elec- micro-supercapacitors was demonstrated by a self- trodes and WS2 as semiconductor in a vertical het- aligned printing process utilizing capillary action of erostack with a responsivity ~1 mA W−1 at 514 nm liquid inks in microfluidic channels. An average spe- [550]. NIR light detection was realised in BP integrated cific capacitance ~221 µF cm−2 (at a cyclic voltam- Schottky diodes (responsivity ~of 1.8 mA W−1 at metry scan rate of 100 mV s−1) was achieved. Many 1550 nm) [574] inkjet-printed from an isopropanol/2- other examples exist where GRMs are explored as the butanol mixture. The same device also showed good electrode materials or additives for Li/Na-ion batter- responsivity in the UV/Vis ~164 mA W−1 at 450 nm. ies, scaffold or interfacial layer for Li–metal anodes or Once encapsulated, the BP film showed good long- cathode for Li/Na–O2 batteries [582]. term stability (up to 30 days) also allowing the dem- As illustrated by the examples above, printed onstration that the saturable absorption of BP can be structure from GRM-based inks are explored in many exploited for the generation of ultrafast laser pulses application areas. optimization of dispersions, ink through mode locking [574]. formulations and printing can further boost device Due to their high surface-area to volume ratios and performance. Probably the greatest strength is that high surface sensitivity to the environment, GRMs can similar techniques can be applied to a whole host of be used chemical sensing of various analytes, includ- often cheap and abundant starting materials. Bear- ing gases, ions and biomolecules [575]. Since high- ing in mind that only a few material classes have been performance sensors have been demonstrated from explored to date, while  >1000 LM have been predicted MC or CVD GRMs [575], increasing research activi- (see chapter II.7), it is clear there are many opportuni- ties are now devoted to exploring printed sensors. Ref. ties associated with LPE and further processing of the [576] reported a graphene-based humidity sensor inks. In addition to this intrinsic versatility, a number integrated on a CMOS platform and a gas sensor with of advantages over inks from other nanomaterials high sensitivity (45%), fast response (12 s) and recovery have been identified. E.g., FLG and TMD inks can be (20 s) as well as selectivity to different chemicals based biocompatible [550, 1717]. Furthermore, GRM only on gravure-printed RGO, was shown in Ref. [577]. BP have dangling bonds on the edges, but usually not on was identified as promising material for sensing appli- the basal plane. As illustrated in Ref. [573], this has the cations [190, 578]. However, the degradation under potential to improve charge transport characteristics ambient conditions is a limiting factor [189]. if clean and well-defined interfaces can be created. The The high SSA of GRMs also makes them promis- accessibility of a range of material classes from insula- ing as electrode materials in batteries or supercapaci- tors, conductors and semiconductors in layered struc- tors [563]. Many demonstrations with solution-pro- tures allows one to combine various properties and cessed graphene [563], MoS2 [563] and MXenes were functions in vertical and lateral heterostacks devices. A reported [579]. For such electrodes, advances in 3d few examples have already been demonstrated, such as printing can be exploited. E.g., free-standing anodes all inkjet-printed FLG-BN transistors on textiles [571], were produced from 3d printing a commercial FLG- all-printed electrolyte-gated TFTs with TMDs as chan- based polylactic acid filament and employed in Li-ion nel material, FLGs as electrode and BN as separator batteries and solid-state supercapacitors [580]. In Ref. [572] or vertical PDs [550]. 61 2D Mater. 7 (2020) 022001 C Backes et al IV. Graphene growth on SiC next graphene layer grows, under the buffer layer, the previously grown buffer layer is no longer in interac- IV.1. Sublimation tion with the SiC substrate and turns into a SLG (with a The most commonly used SiC structures for growing linear electronic dispersion), as the newly grown layer graphene are hexagonal 4H-SiC and 6H-SiC polytypes becomes the new buffer layer. Therefore, on the Si-face [583]. Both polytypes are commercially available as a SLG always rests on the buffer layer. SLG on the Si- single crystal wafer of up to 150 mm in diameter. Their face has a defined orientation relative to the SiC crystal   (0 0 0 1) and 0 0 0 1 -faces are polar [583], i.e. they (see figures IV.1(a)  and (b))[585]. are Si- or C-terminated surfaces. The name epitaxial For the 0 0 0 1 -face (C-face) no buffer layer has graphene (or epigraphene—EG) is sometimes used been observed [596]. The graphene layers grow much [584] to reflect the idea that graphene has a quasi- faster than on the Si-face with 5–10 L growing on the epitaxial match with hexagonal SiC lattices, and an C-face compared to one the Si-face in standard Con- exact orientation relation with 4H/6H-SiC(0 0 0 1)- finement Controlled Sublimation [597] (CCS) con- face. The (hetero)epitaxial orientation of EG on the ditions [585]. Contrary to the Si-face where they are Si-face relative to 4H-SiC is shown in figure IV.1. (For a arranged in a graphite stacking (Bernal ABA, or some- review, see [585] and ref. therein). times rhombohedral ABC [598]), the layers on the The decomposition of SiC does not require an C-face are stacked rotationally, alternating azimuthally external source of C to produce graphene. C is pro- rotated layers with 0° and 30° within  ±  few degrees vided by the SiC crystal itself when its surface decom- [585, 599, 600]. The stacking explicitly is not tur- poses at high T [586] (T  >  1000 °C in vacuum [586]). bostratic (random stacking of small grains). Graphite SiC sublimation is intrinsically a wafer scale method (AB) stacking in this case corresponds to stacking faults [584, 587–589]. Since Si has the highest partial vapour that have an occurrence of less than 15%–19% [596]. pressure as shown in figure IV.2, Si leaves the surface, In the Si sublimation process, Si desorbs through resulting in a C rich surface that forms SLG on the SiC steps [601–603] or from defects [604, 605] and terraces substrate. For this reason, the second layer that forms [606] and graphene growth results of a fine kinetic bal- grows under the first one, i.e. the top layer is always the ance of different step evaporation rate, Si sublimation first one to form on the SiC surface. Because SiC is stoi- from and re- absorption onto the surface, and C diffu- chiometric, the amount of graphene formed is related sion [597, 601, 607]. When the SiC surface is in equi- to the amount of Si that leaves the surface by sublima- librium with the Si vapour, the formation of graphene tion, and for the common SiC polytypes used, ~ 3 Si-C is arrested. The key to grow high quality graphene is a bilayers are required to provide enough C atoms to slow growth at high T, where the kinetic energy and the form a SLG. A schematic illustration of the graphene mobility of C and Si atoms are high. It is therefore nec- formation steps is in figure IV.2. essary to slow down the Si escape rate that depends on Graphene growth is driven by the same Si sublima- the partial pressure of Si (PSi) in equilibrium with SiC. tion process on both the Si face and the C face. How- Following figure IV.2, PSi increases by over three ever, the surface reconstruction and growth kinetics orders of magnitude in the range 1200 °C–1600 °C, for the Si terminated and C terminated faces are dif- which is a typical range for graphene growth. ferent, resulting in different graphene orientation Note that the partial pressure of C is less than 10−10 relative to the substrate, different growth rates, mor- atm, at least five orders of magnitude lower than the phologies and electronic properties, as schematically Si vapour pressure (or that of the residual gasses in show in figure IV.1(c). The first C layer grown on the the vacuum chamber), so that the role of gas phase C Si face of SiC polytypes is known as the buffer layer is negligible in the graphitization process. It is reason- [590] (sometimes called the zeroth layer). It has the able to assume that each Si atom that escapes from the graphene atomic structure but it is in interaction with surface leaves a C atom on it to form graphene. Main- the SiC substrate, which is often described as a bonding taining a background gas (either Ar [589] or Si [597] between ~1/3 of the carbon atoms of the buffer layer e.g.) reduces the diffusion out of Si and increases the to the Si dangling bond of the SiC substrate [591, 592]. return probability to the surface, thereby reducing the However, the exact nature of the bonding is not fully Si escape rate and consequently the graphene growth understood as the SiC substrate and buffer layer have (see [597, 589] for details). incommensurate structures [593] and the two lattices In order to better control the growth rate several form a large superlattice unit cell 6  √  3  ×  6  √  3, mak- methods have been employed, either by providing an ing theoretical calculation difficult. The important external source of Si (by adding a flux of silane [609], point here is that the buffer layer has the graphene a Si flux [610] or by sublimating a Si piece nearby), or lattice structure, does not have the electronic disper- by slowing the Si escape by diffusion through a neu- sion of graphene [590]. Instead it is a semiconductor tral gas (Ar atmosphere) [589] or keeping the SiC as with an energy gap  >0.5 eV, as shown by scanning close as possible to equilibrium with its own vapour tunneling spectroscopy (STS) and photoemission by confinement [597]. Alternatively, CVD methods spectroscopy (see section IX.2.4/ARPES for an intro- were employed, where C species are brought externally duction to the technique) [592, 594, 595]. When the [611, 612]. The last three methods are reviewed below. 62 2D Mater. 7 (2020) 022001 C Backes et al Figure IV.1. (a) Orientation of graphane on the Si-face relative to 4H-SiC. (b) LEED diffraction of MLG on 4H-SiC, showing the diffraction spots of SiC, SLG and the buffer layer (6  √  3  ×  6  √  3)R30. (c) Schematics of graphene growth on the top   6H/4H-SiC(0 0 0 1) and bottom 0 0 0 1 surfaces. Adapted from [585], © Springer Nature. IV.1.1. Growth under argon placement in the porous graphite insulation. After The vertical induction heated furnace shown in that, the whole assembly is loaded in the growth cham- figure IV.3 consists of a coil to produce a uniform ber (quartz tube), locked and pumped down. The electro-magnetic field in the semi-close graphite chamber is ready for heating when the pressure reaches platform called crucible on which the T is measured 5  ×  10−7 mbar. with an optical pyrometer, a quartz tube with a To slow the Si sublimation while being close to diameter of 100–200 mm depending on the size of the equilibrium conditions, growth was performed under crucible, porous graphite insulation which is slightly 1 atm of Argon (Ar) [588, 589, 614]. At 2000 °C Si sub- smaller than the quartz tube [607]. The role of the limation in 1 atm of Ar SLG on a large area and thick- radiofrequency (RF) field is to produce Eddy current ness uniformity [614]. Figures IV.4(a) and (b) show in the crucible and that heat up the crucible (figure LEEM images of graphene layers (small domains of IV.3(b)). The long RF coil in the reactor provides a 1–2 µm size and thickness  >5LG) grown in Ar atmos­ uniform electro-magnetic field distribution in order to phere with a better thickness uniformity and quality ensure a uniform T on the substrate. This design allows for graphene grown in Ar [588]. Note that LEEM is treatment of SiC wafers up to 50 mm in diameter. performed instead of conventional SEM, because by Other furnace designs with fully automated T controls counting the number of minima in the LEEM reflec- were also reported [613]. tivity as a function of the electron beam intensity, the As the T distribution in the crucible influences N can be determined at the resolution of the LEEM the thickness uniformity, a symmetric crucible was image. Details about T-time profile can be described as designed for good T uniformity, with optimization setting T ramps (~20 per minute) up to at least 1150 °C by modelling the T distribution (figure IV.3(c)). The and once the desired graphitization T (usually  >1150 SiC substrate position should be in the centre of the °C) is reached keeping this T during the growth time. RF coil. Prior to growth, the substrates are cleaned in Then decreasing it down to RT with a slow ramping organic solvents, then in solutions based on sequen- (70 min−1). tial oxidative desorption and complexing with H2O2– NH4OH–H2O (RCA1) followed by H2O2–NCI–H2O IV.1.2. Near equilibrium confinement controlled (RCA-2) to remove organic contaminations and dip- growth ping in diluted HF. The substrates are then placed in CCS[597] relies on SiC being in near equilibrium the middle of the graphite crucible followed by the conditions with the Si vapour. Graphene growth is 63 2D Mater. 7 (2020) 022001 C Backes et al Figure IV.2. (a) Partial pressure of different vapour spices versus temperature in equilibrium with a crystal on SiC. (b) Schematics of graphene formation by Si sublimation [608] (reprinted with permission from Elsevier). For simplicity, the growth of graphene in (b) refers to the growth on the carbon face and, consequently, there is not buffer layer. Figure IV.3. Sublimation reactor (a) Picture. (b) Sketch of the vertical RF-heated furnace in cross-section [607]. (c) distribution in the crucible. The diameter of the crucible in (c) is 50 mm and the height of the growth cavity (marked by black lines) is 5 mm. Adapted from [607]. controlled by encapsulating the SiC crystals in graphite Details about the full furnace design and operating so as to sequester the evaporated Si (see figure IV.5). conditions can be found in Ref. [597, 615]. The Si A SiC crystal is placed in a closed graphite box (called escape rate that ultimately determines the growth rate, crucible) provided with a very small calibrated hole is defined by the geometry of the hole (diameter 0.5 to of typical diameter 1 mm. As the T of the furnace is 2 mm) [597, 615]. The rate of graphene formation by increased above 1200 °C and Si sublimates from SiC, the CCS can be reduced by a factor  >1000 compared to built-up Si vapour remains confined inside the crucible UHV sublimation, where 1LG grows on the C-face in and escapes slowly through the hole (figure IV.5(b)). ~1 min at T  =  1.200 °C [597]. The graphene formation The SiC chip is therefore in a uniform Si vapour as rates can be reduced by an additional factor of up to 103 defined by the PSi versus T of figure IV.2, ensuring by introducing 1 atm of Ar into the enclosed crucible uniform graphene growth. Because the Si vapour is [597]. The system is compact for fast pumping speed confined, the growth T is substantially increased (by and small thermal inertia, allowing fast heating and ~300 K for CCS [597]) compared to vacuum growth. cooling rate up 150 °C s−1 in the T range from RT to 64 2D Mater. 7 (2020) 022001 C Backes et al Figure IV.4. Graphene layers grown on the Si-face (a) in vacuum, (b) under Ar [588]. Field of view is 20 µm. The lines are identified as BLG by counting the number of minima in the LEEM reflectivity as a function of the electron beam intensity. Adapted from [588, 589]. Figure IV.5. Confinement controlled sublimation. (a) In UHV Si that is sublimated from the SiC surface at high T escapes. (b) In the CCS method, SiC is in quasi-equilibrium with its own vapour (c) picture of the CCS furnace. Adapted from [597]. 2100 °C. The T profile is computer automated [615, and on the background pressure (with or without addi- 616]. The whole process can be performed in less than tional Ar backpressure). Optimum T and growth times two hours, being scalable to full wafer sizes. are specific for each crucible adapted to each type of An alternative method to CCS is to cover the SiC grown graphene. Typically, a 1LG in 20 min at 1520 °C substrate with a piece of graphite, which provides [594] on the Si-face, a graphene buffer layer °C lower some Si vapour confinement. The Si escapes on the than the Si monolayer in the same crucible [594]. On the side providing a non uniform Si vapour gradient that C-face, N is determined by both T (1450 °C–1525 °C) produces narrow tapered stripes of graphene [617]. and time (1 min to several hours for very thick films) To grow EG on hexagonal 4H- or 6H-SiC wafers [597, 615] [606] (see section IV.1.3 and figures IV.5(a)– by CCS [615] the key step after pumping out the cell to (c)). Templated graphene growth of nanostructures below 10−6 mbar, is the Si sublimation at high T (1400 (e.g.SiC sidewalls [618–620]) on non-polar facets (i.e.   °C–1650 °C). The exact T profile depends the geometry other that (0 0 0 1) and 0 0 0 1 ) is finely tuned close to of the graphite crucible and geometry of the hole [615], the buffer layer growth conditions. 65 2D Mater. 7 (2020) 022001 C Backes et al IV.1.3. Carbon-face MLGs have been used as a model system to study The surface state of the substrate prior to growth graphene properties [585]. This stems from a combi- is essential, because rough surfaces have multiple nation of factors: (i) Excellent structural quality (small disordered nano-steps acting as nucleation centres Raman D peak observed, see figure IV.7 and [630]), for graphene growth. The first step is therefore SiC (ii) Flatness of the layers (rms  <  5 pm as measured by surface flattening. For SiC surfaces polished to optical x-ray diffraction [596]) which excludes strain-related quality, the deep polishing scratches can be removed gauge field effects [632] and broadening effects in by SiC etching at high T in a hydrogen environment k-resolved spectroscopy measurements (see figure S1 [621]. Successful flattening is obtained at 1500 °C for in supp of [620]), (iii) Small interaction with the sub- 15 min in a flow rate of 200 sccm in 1 atm of a mixture strate and the environment, particularly for the layers of 5% H2 in Ar [621], which produces a surface with no in the middle of the stack that are quasi-neutral [629, observable scratches in AFM section IX.1.2 and with 633] (charge density n ~ 5  ×  109 cm−2, that is at most 8 a step-terrace structure with atomically flat terraces. meV away from the Dirac point) with high mobilities However, on the 6H(0 0 0 1) face, the average width at RT [628, 633]. (iv) graphene electronic structure for and step height of terraces after treatment do not have all layers [634], due to the rotational stacking, provid- a clear correlation with hydrogen etching, T and time, ing a large signal intensity very much sought in optical but are related to the local miscut angle (the local angle measurements. θ of the surface relative to the on-axis orientation, generally θ  <  0.2° for commercial wafers of on-axis IV.1.4. Silicon-face wafer [621]. Commercial SiC wafers can be purchased MLG on Si-face (see figure IV.4) is the most studied with so-called ‘epiready surfaces’ treated by a chemical as for realization of quantum resistance standards (as mechanical process. These provide excellent starting multiples of h/e2  =  25.812 807 kΩ) at 4.2 K in small surfaces, with rms  <  0.2–0.3 nm, with no need for (<1 Tesla) magnetic fields. With a quantum Hall etching prior to graphene growth. resistance quantization accuracy of 3  ×  10−9 that On the 4H- or 6H-SiC carbon face, continuous rivals the best 2d electron gas standards [612, 635]. As MLG containing 5–10 LG [622] are grown by anneal- growth on the Si-face starts from the SiC steps [601, ing for 15 min at T in the range 1450 °C–1525 °C. 603, 636], BLG (or MLG) tend to form at step edges Repeated cycles produce much thicker films up to 50 when a 1LG fully covers the terrace (see figure IV.4), LG [623, 624]. Shorter times and lower T are required which makes it difficult to produce extended 1LG. to produce 1LG. They appear in patches on the SiC Additionally, the step edges add electronic scattering substrate (see inset of figure IV.7(a)) because of the due to the presence of BLG/MLGH, or to the reduction rapid growth rates on the C-face and the multiple of the carrier concentration on the step [585], which nucleation sites [604, 606, 615, 625].These 1LG have is detrimental to the homogeneity of the quantum excellent electronic properties, despite draping over Hall effect [637]. Better control of growth on the Si- steps and making pleats [604]. face was obtained by providing carefully balanced Sample quality is revealed in AFM (see sec- methane and H2 [638]. Ref. [639] reported MEG tion IX.1.2) and STM images (see section IX.1.4) in draped continuously over the steps by graphitization multiple spectroscopy techniques (see section IX)), of a polymer deposited on the bare SiC. magneto-optical, photoemission, see (section IX), ultrafast optics and electronic transport (see sec- IV.1.5. Step bunching and graphene growth tion IX.3 Raman spectroscopy shows a vanishing D During annealing above 1200 °C, the SiC surface peak attesting the low defect density (see figure IV.7). undergoes microscopic restructuring by forming steps All spectroscopy studies [622, 624, 627–629] confirm and terraces. This process, known as step bunching that the MLG behaves as electronically decoupled, i.e. [1745], is different from atomic surface reconstruction as a stack of graphene layers with a non gapped linear and refers to surface morphology. Step bunching, dispersion down to a few meV from the Dirac point, not i e. the bunching of straight steps on vicinal crystal like graphite and exhibit the quantum Hall effect [604, surfaces, which is governed by energy minimization 625, 630] and high mobility (up to µ  =  39 800 cm2 V−1 on different terraces is a fundamental phenomenon s−1 at charge density n  =  0.19  ×  1012 cm−2 [604]). The in SiC. Graphene formation has been analyzed with se MLGare continuous over the entire SiC, drape over respect to step bunching of SiC, by studying the SiC steps and have rms  <  5 pm as measured by x-ray influence of the crystal structure at the atomic level for diffraction [596]. STM shows extended moiré regions each SiC polytype [640]. As illustrated in figure IV.8(a), [600, 627] providing evidence that adjacently stacked the 4H-SiC polytype has two decomposition energies, graphene layers are rotated creating a superstructure. terraces 4H1 (−2.34 meV) and 4H2 (6.56 meV), On a larger scale, smooth graphene pleats are observed, respectively, and the 6H-SiC (figure IV.8(b)) has three as seen as bright lines in figures IV.6(c) and (e), which distinct terraces, 6H1 (−1.33 meV), 6H2 (6.56 meV) originate from the differential contraction of graphene and 6H3 (2.34 meV), while 3C-SiC (figure IV.8(c)) has and SiC upon cooling. only one kind of terrace, 3C1 (−1.33 meV) [640]. 66 2D Mater. 7 (2020) 022001 C Backes et al Figure IV.6. MLG on C-face topography images. (a) 4  ×  4 nm STM image (reproduced with permission from [626]). (b) 400  ×  400 nm STM image. The blue square represents the size of the STM scan in (a) [626]. (c) 40 µm  ×  40 µm AFM scan. The white lines are graphene pleats. (d) 400 nm long STEM image across areas of different moiré pattern, showing that the top layer is flat (rms  <  50 pm) and continuous (from [627], with permission from AAAS). (e) SEM image of a graphene pleat (from [621], with permission from Elsevier). Figure IV.7. Raman spectroscopy for graphene layers on 4H-SiC C-face (a)SLG, inset AFM image. The black trace is the raw data (SiC  +  Graphene), blue is the bare SiC spectrum; the red trace is the graphene spectrum once the SiC Raman peaks are substracted [631] (b) MLG with the Raman contribution for SiC removed. The inset shows a Lorenzian 2D peak for this 10LG. Figure adapted from [630], © IOP Publishing. The growth process is schematically shown for source of C, will impose the formation of a second 4H-SiC in figure IV.8(d). Since Si and C atoms are layer of graphene (figure IV.8(d), stage 3) since some bonded more weakly in the vicinity of step edges, Si extra C will be released. Therefore, a full coverage of desorbs from these areas faster in comparison with the 4H-SiC substrate surface by just 1LG may be an the terraces [601, 603, 636]. Based on the terrace ener- issue. gies, removing a 4H1 terrace costs less energy and the A similar mechanism of energy minimization is step decomposition velocity is faster (figure IV.8(a)). expected in the 6H-SiC polytype (figure IV.8(b)). As shown in figure IV.8(d), from the edge of the 4H1 As a result, first the step 6H1 will catch step 6H2 terrace on the graphene-free surface, C atoms are and forms two Si-C bilayers. Then step 6H3 will released onto the terrace as Si atoms leave the surface advance and merge with the two-bilayer step. The (stage 1). The C atoms coalesce and nucleate into gra- growth process for 6H-SiC is the same as the 4H-SiC phene islands (stages 1 and 2), which act as a sink for [640]. However, on 3C-SiC all terraces have the subsequently released C atoms. After the 4H1 terrace same decomposition energy [640] (figure IV.8(c)) step catches the 4H2 step, the newly formed two SiC and no energetically driven step bunching should bilayer step provides more C atoms as compared to the be expected. In this polytype, a non-uniformity one-bilayer step and the first graphene layer extends of s­ ublimation may be induced by the presence of along the step edge (stage 2). The large percentage of extended defects such as stacking faults, which are bunched steps with four Si-C bilayers, an increased characteristic of this material [641]. 67 2D Mater. 7 (2020) 022001 C Backes et al Figure IV.8. Stacking sequences and possible terraces on (a) 4H-SiC, (b) 6H-SiC, and (c) 3C-SiC surfaces. Large (blue) and small (red) circles represent Si and C atoms, respectively. Length of arrows indicates different step decomposition velocities. (d) Schematic depiction of the formation process of EG via sublimation of Si from the SiC surface. Adapted from [640], with permission from Elsevier. A capping technique can be employed to local- at the high T of the graphene growth (see figure IV.9(c) ize the step bunching [642]. The cap consists of an bottom). This technique provides large flat terraces amorphous carbon layer evaporated on the bare SiC (figure IV.9(c)-bottom) at a predefined location. The grid evaporated on the bare SiC. Amorphous carbon amorphous C is finally removable by plasma etching is then patterned by plasma etching or lift-off to pro- after annealing to access the graphene layer for further duce typically micron size structures. Upon annealing device patterning [642]. In that case a graphene pro- to produce graphene, the carbon layer pattern pins the tection layer is used (this can be the dielectric of a sub- SiC steps under it so that the SiC steps bunch against sequent top grid). one enclosure wall and align along the amorphous C grid, as shown in figure IV.9(a). Several geometries IV.1.6. Optimization of growth process for scale up have proven effective, and it is reasonable to position Growth conditions can be developed to increase the an edge of the cap perpendicular to the SiC substrate size of the SiC substrate to 20  ×  20 mm2 and ultimately miscut to maximize step bunching against the cap wall. to 100 mm diameter wafers. The furnace is similar The slight SiC surface roughness induced by the mild to that of figure IV.3, but with longer RF-coil with plasma etch used to pattern the carbon layer (see fig- uniform distance between the pipes [643]. The coil ure IV.9(c) top) is smoothed out by the step bunching can be moved up and down to change the T gradient if 68 2D Mater. 7 (2020) 022001 C Backes et al Figure IV.9. SiC Step pinning under an evaporated amorphous carbon grid. (a) Schematics of step pinning (before annealing, upper frame). After annealingat 1350 °C (lower frame), the steps accumulate at one side of the grid in each enclosure providing large terraces at a locations defined by the grid. The arrow point the direction of the flow. (b) AFM image after annealing and (c) topographic profile within the grid (top) before and (bottom) after step bunching annealing, showing that a large terrace has developed. Adapted from [642], with the permission of AIP Publishing. necessary and it is fully automatic. The key to achieve atoms coalesce and nucleate into graphene islands on large area 1LG on SiC is to (a) understand the role of the the step kinks (marked with a circle) and these islands buffer layer [590, 592, 610] and of the step bunching act as a sink for other C atoms. As the step edges are the [607] which takes place during its formation and to main source of C, the growth rate is two times higher monitor the effect of (b) T (in the range 1700 °C–1950 along the step edge compared to that on terraces °C); (c) Ar pressure (in the range 750–950 mbar) (d) and for this reason a graphene layer first covers like a and growth time (from 0 to 15 min). AFM topography ribbon the sidewall of steps (figure IV.10(a)) [602, 618, and phase images show that the formation of the 620, 644].Then growth continues from the edge of a buffer layer (at 1850 °C and Ar pressure 950 mbar) step to the (0 0 0 1) terraces, where the graphene layer can start at any place on the substrate, but preferably interacts with the surface and becomes the buffer on the kink of steps, as shown in figure IV.10(a). The C layer. Studies show that the graphene layer first covers 69 2D Mater. 7 (2020) 022001 C Backes et al Figure IV.10. (a) AFM phase images of initial stage of the buffer layer formation, and (b) LEED pattern of the buffer layer (electron energy 60 eV) grown under Ar (in the reactor of figure IV.3) [643]. Figure IV.11. Size increase of 1LG from 7  ×  7 mm2 to 20  ×  20 mm2. The size of reflectance maps (red scale) are 30 µm  ×  30 µm [643]. the steps with larger terrace [643]. A critical issue is to increase the BLG coverage for both 4H and 6H-SiC fully cover the substrate surface by a buffer layer, before polytypes [643]. lifting it up as a 1LG by growing another buffer under Based on the above results, a growth protocol was it. A good quality buffer layer exhibits a particular defined to optimized SLG coverage. The result, shown pattern in LEED, as illustrated in figure IV.10(b) (see in figure IV.11 indicate the growth of ~99% 1LG also figure IV.10(b), and [584, 592, 611, 645–647]. on 20  ×  20 mm2 SiC substrates and 99.8% 1LG for The pattern for a buffer layer grown in the reactor of 7  ×  7 mm2 substrate (figure IV.11). The SLG cover- figure IV.3 at 1850 °C and 850 mbar is similar over the age is determined by optical reflectance maps. Experi- whole surface. ments showed that the reflectance of graphene on SiC Similarly to the amorphous C corral [642], the normalized to the reflectivity of bare substrate (the buffer layer can stop the step bunching process on contrast) increases linearly ~1.7% per layer for up to 4H and 6H-SiC [648]. This means that the surface 12L, which agrees with theory [649]. The wavelength energy becomes uniform all over the substrate sur- dependence of the contrast in the visible is investigated face after the formation of the buffer layer subse- using the concept of ideal Fermions and compared quently resulting in a uniform and continuous 1LG with existing experimental data for the optical con- coverage, which does not exclude the growth of BLG stants of SLG [649]. at the steps. The T dependence of the buffer layer for- The results for SLG on 4-inch SiC wafer showed mation and SLG coverage is sublinear (that is close that the growth rate on the edge of the wafer is higher to linear, not exponential [643]), which suggests than in the centre, due a faster Si escape at the wafer that the synthesis is surface kinetics limited on both edges [597]. Quasi-equilibrium growth [597] provides SiC polytypes [601, 643]. The results from graphene a way to mitigate these problems. Figure IV.12 shows samples grown Ar indicate that there is an optimal Ar the recipe for growth of SLG on large area substrates, pressure (~850 mbar) yielding under this condition a including on 4 inch SiC wafer. These operation condi- high growth rate. tions assume different T ramps up to 1150 °C and to The time dependence of SLG and BLG growth the growth T, according to figure IV.12. under Ar pressure shows that graphene spreads faster on 4H-SiC substrates than on the 6H polytype from IV.1.7. Growth on SiC off-axis cut surfaces the start of the growth, and that 1LG coverage increases Graphene layers are best grown on 4° and 8° off-axis approximately linearly with time. After the SLG is com- 4H-SiC (0 0 0 1) substrates to have the right terrace plete, increasing annealing time does not significantly width for performing step flow growth [650, 651]. 70 2D Mater. 7 (2020) 022001 C Backes et al Figure IV.12. Temperature-time plot for the growth of SLG on 4 inch substrates under Ar [643]. A general recipe is provided in Ref. [652]. Figure IV.13 3LG, respectively, over the substrate or stacked over shows a side view of the off-axis SiC substrate used for other graphene layers [601]. In fact, like for the (0 0 0 1) graphene growth, where the off-axis surface is tilted at on-axis face, the graphitization starts from the step a specified angle (off-axis angle) from the basal plane edges and propagates gradually to the centre of the ter- surface toward the [1 1 2 0] SiC direction. Controlling race [601–603, 655]. In this case, the C atoms at ter- the tilt angle allows to change step heights, terrace race kinks and step edges have a lower coordination widths and a growth rates. number, thereby leading to an easier breaking of the A direct comparison of graphene formation on bonds at these sites. For this reason, the probability an 8°-off axis (0 0 0 1) 4H-SiC epitaxial layer and on- of SiC decomposition and further surface diffusion axis (0 0 0 1) 6H-SiC [653], performed by uniform of C atoms is increased and graphitization of the SiC graphitization of 1 cm  ×  1 cm SiC substrates at 2000 surface occurs [655]. Regarding the growth kinetics, °C for 30 min (with a base pressure in the chamber anisotropy of the Si desorption rate is always present ~5  ×  10−6 mbar and an argon pressure ~1 atm during and, thus, different SiC steps have different evapora- the growth) shows that the growth of graphene on off- tion rates [601]. In addition, increase in the miscut axis SiC starts from the edge and follows the surface of angle increases the step density, thereby leading to an the large terraces (0.5 to 1 µm width), running parallel increase in graphene thickness at this region [601]. to the original steps of the off-axis wafer. Another interesting finding (see figure IV.14— Off-axis samples present significant step b ­ unching right) is that the presence of the atomic steps leads after high T annealing (rms  =  16 nm), while flat- to the formation of pleats similar to the MEG on the ter surfaces (rms  =  2.4 nm) are obtained on the on- 4H/6H on-axis C- face (~1 to 2 nm high and 10 to axis sample. In comparison with on axis SiC wafers, 20 nm wide) preferentially oriented in the direction off-axis SiC wafers contain a lot of steps due to their perpendicular to the step edges of the SiC terraces (fig- cuting angle. For this reaseon, during annealing they ures IV.14(e) and (f)) [654]. Such a parallel orientation easily bunch. Step bunching is higher in off-axis of pleats is particular to SLG grown on a vicinal SiC ­samples with dense step edge, because etching or ero- surface, while for SLG grown on on-axis 4H-SiC it is sion start from on step edges, which is not the case in characterized by a preferentially hexagonal mesh-like on-axis samples. A more uniform and homogeneous network of pleats interconnected into (often) triangu- graphene coverage was found on on-axis 6H-SiC [653]. lar nodes [585] (see figure IV.14(c)). The 100 to 200 nm wide terraces of the 4H-SiC (0 0 0 1) 8° off-axis samples are covered by FLG spreading like a IV.1.8. Growth on non-polar surfaces SiC carpet for all growth T in the range 1600 °C to 2000 °C The presence of the buffer layer on the Si-face is often (growth under Ar) [654], with an increase in the N as a argued to degrade the electronic properties of the SLG function of T (from 0 to 10 LG) (figure IV.14). above it [656–659]. In quasi -free standing SLG the Similarly to the on –axis surface, the graphitization buffer layer is converted into SLG by intercalation. An rate is much higher for the C-face off axis substrates alternative to intercalation is the direct graphitization while it is almost an order of magnitude smaller for of non-polar SiC surfaces where there is no buffer layer the Si-face on-axis substrate compared to off axis sub- [660, 661], or to master the graphitization on 1LG on strates (figure IV.15). AFM analysis showed the pres- the C-face, where high mobilities are measured [604]. ence of the steps with different heights. An ~0.35 nm According to the crystallography of the hexago- and ~1.1 nm step height can be associated to 1 and nal family of SiC, there are three non-polar planes 71 2D Mater. 7 (2020) 022001 C Backes et al Figure IV.13. (Top panel) Diagram illustrating off-axis surfaces of 4H-SiC and (bottom panel) schematic drawing of the off-axis 4H-SiC surface after graphitization [643]. Figure IV.14. (Left panel) AFM images. Effect of graphitization temperature on the surface morphology of the sample 8°-off axis (0 0 0 1) 4H-SiC: 1600 °C (a), 1700 °C (b) and 2000 °C (c). The phase contrast maps on the same samples are also demonstrated ((d)–(f), respectively). (right panel) STM image of a sample annealed at 2000 °C. Adapted from [654], with permission of The Royal Society of Chemistry. available for graphene formation: the m-plane Synthesis of graphene on on-axis (0 0 1) cubic SiC 1 1 0 0, a-plane 1 1 2 0 and r-plane 1 0 1 2 [661]. Fig- substrates can be performed at T  >1800 °C for 20 min ures IV.16(a) and (b) show the possible polar and in a C rich atmosphere [661]. To provide reliable con- non-polar SiC planes for graphene growth. In addi- trol of graphene formation a low growth rate (~1LG tion, the growth of graphene can be performed using per five minutes) should be maintained, which is real- thermal decomposition of (0 0 1)-oriented cubic ized in a close crucible at a Ar gas pressure of 800 mbar. SiC [661]. However, the commercial production of Optimization of parameters (e.g., lowering T from 3C-SiC substrates is limited, and in contrast to 4H and 2000 °C to 1800 °C) can be effective for obtaining gra- 6H-SiC, large single crystals 3C-SiC are not available. phene layers with desired thicknesses and domain sizes Polycrystalline 3C-SiC are often grown as epilayers on ~100 nm. The optimizacion of the rest of parameters, SiC or on Si-wafers. Figures IV.16(c) and (d) demon- like pressure, flow rate, step bunching, time, etc follow strates the important crystallographic planes in the the same trend that for samples grown on non-polar cubic structure of SiC. faces and that are described in previous paragraphs. 72 2D Mater. 7 (2020) 022001 C Backes et al Figure IV.15. Graphitization thickness versus T at 60 min for 8° off cut Si- and C-face 4H SiC and for Si-face on axis 4H substrate (adapted with permission from [614]). Figure IV.17 shows the LEEM images of graphene has a (6  √  3  ×  6  √  3)R30° periodicity with respect to before and after optimization i.e at T  =  1800 °C in the SiC(0 0 0 1) substrate surface. reactor of figure IV.3. This layer consists of C atoms in a graphene A (0 0 1) cubic SiC substrates with rather flat sur- arrangement [590, 592, 668]. This means that the face and low surface roughness (rms~2 nm) obtained structure is has C hexagons with a C–C distance like in by substrate polishing (rms ~0.6 nm) promotes graphene. The (6  √  3  ×  6  √  3)R30° periodicity with growth of more homogeneous EG with large domain respect to the SiC surface corresponds to a (13  ×  13)   sizes as shown in figure IV.17. Growth on 1 1 0 0 super cell. Thus, the super cell contains 169 graphene 6H-SiC resulted in non uniform coverage exhibiting unit cells. Due to a hybridization of states of the C areas of fragmented graphene and some micrometre atoms of the buffer layer with states from the SiC sub- large areas (figure IV.18—left). Similar results were strate surface, it lacks the typical π-bands of graphene   reported in Ref. [662] for the 1 1 2 0 plane (figure and therefore has no Dirac cone [592]. On the other IV.18—right). hand, the σ-bands of the buffer layer are fully devel- Growth on non-polar SiC surfaces is limited by the oped and indistinguishable from those of graphene vertical growth rate (graphene growth on the polar [592]. These observations are in agreement with theor­ face proceeds laterally) [641]. Such vertical growth etical studies [590, 669]. produces scattered islands of graphene on the sur- The buffer layer (6  √  3) induces electron doping in face with higher density of grain boundaries (see fig- SLG. Typical values for the charge carrier density in SLG ure IV.19). The latter causes a greater amount of Si out- are n ~ 1  ×  1013 cm−2 [589, 671–674]. The buffer layer is diffusion from the substrate, leading to a thicker (up to also partly responsible for the relatively low µ and its T 8LG) subsequent [607] MLG growth on the nonpolar dependence [658]. Furthermore, the reduction in spin faces in comparison to growth on polar faces [607]. transport in SLG is attributed to the buffer layer [656]. As graphene growth on the low-packed non-polar Typical values for µ ~ 1000 cm2 V−1 s−1 at n ~ 1  ×  1013 planes is faster than that on the high densely packed cm−2 and T  =  300 K and µ ~ 2000 cm2 V−1 s−1 at n ~ non-polar planes, it is assumed that the surface density 1  ×  1013 cm−2 and T  =  25 K [584, 589, 658, 675, 676]. is a key factor limiting Si desorption and the graphene The structural similarities between the buffer layer growth rate. and graphene suggest that it should be possible to con- vert the former one into the latter by cutting the bonds IV.1.9. Surface engineering by intercalation between buffer layer and substrate through intercala- Graphene grown on SiC(0 0 0 1), by sublimation tion, as depicted in figure IV.21 schematically for the resides on top of a buffer layer [590, 592], as case of hydrogen. Similar procedures have been stud- schematically depicted in figure IV.20. The buffer layer ied before for SLG on metal surfaces like, e.g. Ni(1 1 1) (denoted also by 6  √  3) is formed at the early stages of [677]. In this case, different metals like Au, Yb, or Cu heating of the samples, i.e. before graphene is formed, were intercalated between the metal substrate and gra- and it has been observed in several experimental phene [677]. Several authors showed that different ele- studies of of SiC(0 0 0 1) surfaces [586, 663–665] ments can be intercalated between the buffer layer and and also in an atomistic simulation of graphene the SiC substrate. E.g., the intercalation of H [645–647, formation on SiC(0 0 0 1) [590, 666]. The buffer layer 658, 678–681], O [682–685], F [686, 687], N [688], 73 2D Mater. 7 (2020) 022001 C Backes et al Figure IV.16. Principal crystallographic planes of (a,b)hexagonal and (c,d) cubic SiC. Figure IV.17. LEEM images of graphene grown on (0 0 1) cubic SiC before (left panel) and after optimization (right panel). In the first case, we have 4–5LG, while optimized growth allows to obtain 1–3 LG with increased domain size up to 10 µm.Field of view is 25 µm. Adapted from [661]. Si [689, 690], Ge [691, 692], Au [693–696], Cu[697], lar processes can be employed to intercalate hydrogen Li [698, 699], Na [700] and subsequent transformation under the buffer layer even if an additional graphene of the buffer layer into graphene have been confirmed. layer already exists on top of it [645, 646, 657, 680, For applications, it is important that the interca- 704–706], although the exact conditions may vary. In lated species does not give rise to electronic states at this case, SLG becomes a quasi-freestanding BLG (see the Fermi level, since these would short-circuit the figure IV.21). graphene layer on top of it. This is the case for hydro- The successful decoupling of the buffer layer and gen which saturates the SiC(0 0 0 1) surface with Si–H conversion into so-called quasi-freestanding SLG ­entities. was confirmed by LEED, ARPES, XPS and LEEM. The doubly filled bonding and empty anti-bond- Figure IV.22(a) shows typical XPS spectra of the C1s ing Si–H states are located below and above the valence core level of various samples [707]. The spectrum of band minimum and valence band maximum, respec- the buffer layer is has 3 components. S1 and S2 are tively, leaving the surface electronically passivated caused by the C atoms of the buffer layer [592]. In [701–703]. A similar situation can be expected for oxy- addition, a component due to C atoms in the SiC gen and fluorine passivation of the SiC(0 0 0 1) surface. substrate is visible. These are also seen in the spec- The first study of the intercalation of hydrogen tra of SLG, BLG and TLG indicating that the buffer under the buffer layer on SiC(0 0 0 1) was reported layer forms the interface. Upon H-intercalation, the in Refs. [645, 646] who annealed SiC(0 0 0 1) sub- buffer layer is converted to SLG which is evident from strates covered by the buffer layer in one atmosphere the disappearance of the components S1 and S2 and of hydrogen at T between 600 °C and 1000 °C. Simi- the formation of a graphene-related component. This 74 2D Mater. 7 (2020) 022001 C Backes et al   Figure IV.18. (left) LEEM image of graphene grown by thermal decomposition at 1900 °C on 1 1 0 0 oriented 6H-SiC substrates,   scale bar 5 micron, adapted from [643]. (right) SEM image of graphene grown on 6H-SiC, scale bar 3 micron 1 1 2 0 . Figure IV.19. Graphene growth mechanism on (a) the polar (Si) face SiC with offcut toward [1 1 2 0] and (b) the nonpolar a/m face with slight offcut along [0 0 0 1]. The growth on the polar face proceeds laterally while the growth on the nonpolar face is limited by the vertical growth rate. The zigzag structure shown on the polar face step edge is a consequence of two possible lateral growth directions. Adapted from [667], copyright American Chemical Society. is also the case for SLG/BLG. The component of the QFBLG on the H-terminated (0 0 0 1) surface of hex- SiC substrate is shifted to lower binding energy due agonal SiC polytypes [645, 646, 658, 704, 707, 708]. to a change of surface band bending. Figure IV.22(a) This is due to the spontaneous polarization of hex- shows examples of the band structure near the agonal SiC [708, 709], which varies with the poly- K-point of the hexagonal Brillouin zone as measured type due to their different ratio of hexagonal to cubic by ARPES [707] for SLG, BLG. stacking sequences [708–710]. For semi-insulating The charge carrier type is changed from electrons 6H- and 4H-SiC(0 0 0 1)-H substrates hole densities of in SLG/BLG on the buffer layer to holes in QFSLG/ 6.2  ×  1012 cm−2 and 8.6  ×  1012 cm−2, respectively, as 75 2D Mater. 7 (2020) 022001 C Backes et al Figure IV.20. Schematic view of graphene on SiC(0 0 0 1). The graphene layer resides on top of the buffer layer. The buffer layer is structurally equivalent to graphene but strongly bound to the SiC substrate. Dangling bonds (db) are present due to the lattice mismatch between graphene and SiC. Drawing not to scale (after [645, 670]). Figure IV.21. Schematic view of the conversion of the buffer layer into quasi-freestanding SLG on H-saturated SiC(0 0 0 1) (also referred to as SiC(0 0 0 1)-H) and of SLG on the buffer layer into quasi-freestanding BLG on SiC(0 0 0 1)-H. After [645, 670]. determined by ARPES in good agreement with theor­ in figure IV.23 [719], initially designed for the hydro- etical predictions [708–710]. In the case of QFSLG on genation of SiC surfaces [703, 719, 720]. The system n-type doped cubic 3C-SiC(1 1 1)-H a slight electron consists of a quartz tube in which the sample can be doping is observed, which agrees with the fact that heated in an atmosphere of purified hydrogen in a this SiC polytype has no spontaneous polarization. In contact-less manner by means of halogen lamps. In this case the excess electrons in QFSLG is due to align- the hydrogenation process, the chamber is first filled ment of the Fermi levels as described by the Schottky with hydrogen and then the sample is raised to the model for metal–semiconductor interfaces [708]. Refs. desired value. Typical process parameters are listed in [658, 704] have provided evidence for the hydrogen table IV.1. at the interface by measuring the Si–H stretch mode Variations of the above procedure have been vibration using FTIR absorption spectroscopy as reported for the synthesis of quasi-freestanding gra- shown in figure IV.22(c). This was later confirmed by phene layers on SiC(0 0 0 1). Ref. [645], reported surface enhanced Raman spectroscopy [711, 712]. preparation T in the range from 600 to 1000 °C at Raman spectroscopy also provided evidence for the atmospheric pressure without mentioning the anneal- conversion of the buffer layer to a QFSLG [658, 705, ing time. Ref. [713] annealed in molecular hydrogen 711, 712]. In the case of QFSLG a narrow 2D line is to convert SLG to QFBLG. They reported a H2 pres­ observed while for QFBLG a broad 2D line indica- sure of 800 mbar, T of 600 to 1200 °C, and process tive for AB stacking is found (see figure IV.22(d)). times between 30 and 120 min. Ref. [721] carried out Electronic transport studies on simple Hall bar struc- the hydrogen annealing directly after the growth of tures and FET have shown that QFSLG on 4H- of graphene in a Hot-Wall CVD reactor. After graphene 6H-SiC(0 0 0 1)-H exhibits a higher charge carrier growth in Ar at 100 mbar at 1650 °C, the sample was mobility (typically ~4000 cm2 V−1 s−1) than SLG on cooled to 1050 °C. At this T, the gas a exchanged to H2 the buffer layer and that it has a considerably reduced T and the pressure increased to 900 mbar. After 30 min, dependence [658, 704, 705, 713–718]. the sample was allowed to cool to 700 °C in H2. Finally Decoupling of the buffer layer from the SiC sub- the system was evacuated and the sample allowed to strate and saturation of the SiC(0 0 0 1) surface with cool to RT. Ref. [715] annealed their buffer layer sam- hydrogen can be carried out in an apparatus depicted ples in H2 at 1013 mbar and temperatures between 76 2D Mater. 7 (2020) 022001 C Backes et al Figure IV.22. (a) C1s core level spectra of the buffer layer, SLG, BLG, =TLG and TLG, on substrated and quasi freestanding (QF). The H-intercalation leads to a disappearance of the buffer layer components (S1, S2) and an increase of the graphene-related component. The signal due to SiC shifts to lower binding energy due to a change of band bending. (b) ARPES intensity maps versus electron energy and momentum of SLG and BLG on the buffer layer as well as QFSLG and QFBLG on H-terminated 6H-SiC(0 0 0 1). (c) Si–H stretch mode of the H-saturated SiC(0 0 0 1) surface below QFSLG and QFBLG. (d) Representative Raman spectra of QDSLG and QFBLG. (a) adapted from [708] and (b) adapted from [707], both © IOP Publishing Ltd; (c) and (d) adapted with permission from [704]. 600 and 1200 °C for 60 min. In Ref. [706], QFSLG was ARPES (see section IX.2.3 for an introduction to the produced by annealing in a furnace at 800 °C and a H2 technique), LEEM and STM (see section IX.1.4 for flow of 0.2 slm for 1 h at an unspecified pressure. The an introduction to the technique) along with FTIR process time was 60 min. The highest mobility was and Raman are ideally suited to characterize the observed for at of 700 °C. In [706] QFBLG by anneal- quasi-freestanding graphene layers on H-terminated ing Hall-bars made from SLG at 1250 °C in H2 at SiC(0 0 0 1). Table IV.2 lists several methods together ~33 mbar was reported. Ref. [680] employed a plasma with the expected observation and relevant references. source as well as an atomic hydrogen source to convert Oxygen intercalation was suggested a means for SLG to QFBLG. They reported that the plasma treat- decoupling the buffer layer from the SiC(0 0 0 1) surface ment induces a disorder. Kinetic Monte Carlo simula- by annealing the buffer layer [682] in one atmosphere tions performed in Ref. [722] have shed light on the of molecular oxygen at 250 °C for 5 s, using a special atomistic mechanism of H intercalation of the buffer oxidation chamber. In the first approach in Ref. [685] layer, in good agreement with experiments. the buffer layer samples was annealed in situ, i.e. directly Surface analytical tools such as LEED, XPS(see sec- in the UHV chamber used for XPS and ARPES analysis, tion IX.2/XPS for an introduction to the technique), at 750 °C in ~10−4 mbar of molecular oxygen. The sec- 77 2D Mater. 7 (2020) 022001 C Backes et al Figure IV.23. Sketch of the apparatus used for intercalating hydrogen between the buffer layer and the SiC(0 0 0 1) substrate [658, 704, 708]. After [719]. Table IV.1. Typical parameters for intercalating hydrogen under the buffer layer [658, 704, 708]. Process Sample temperature in °C Hydrogen pressure in mbar Flow rate in slm Time 6  √  3 to QFSLG 540–560 900–960 0.9–1.0 60–90 SLG to QFBLG 840 900–960 0.9–1.0 60–90 ond procedure in Ref. [685] was carried out ex-situ by Li-compounds. Na partial and inhomogeneous inter- annealing in O2 with best results at 270 °C at 200 mbar. calation occurs already at RT directly after deposition, The Raman spectra, however, demonstrated that both although most of the Na remained on the surface and processes lead to the formation on numerous defects. formed Na droplets. Na intercalation is promoted both In Ref. [683] the oxygen intercalation to transform SLG by electron/photon beam exposure and by moderate into a BLG was carried out by annealing in air at 600 °C annealing at ~100 °C. Annealing at higher T results in for 40 min. A heat up ramp of 50° per minute was used. de-intercalation and Na desorption from the surface They confirmed the decoupling of the buffer layer and [700]. Other elements intercalation would be appeal- the formation of a high quality BLG. ing, such as nitrogen, that is predicted to provide In Ref. [684] buffer layer samples as well as SLG charge neutral QFSLG [726] were annealed up to 500 °C and 650 °C, respectively, in Finally, we review studies on Ge [691, 692] and a water vapour in order to decouple the buffer layer by Au [693–696] intercalation between buffer layer and oxidation of the SiC(0 0 0 1) substrate surface. Raman SiC(0 0 0 1) surface. In both cases, the preparation is spectroscopy indicated numerous defects. However, performed using the following scheme. First, a thin like as for Ref. [683], the QFBLG formed by treating SLG layer (up to 5 monolayers) of the intercalant (Ge or Au) showed a negligible D band. This was also evident from is deposited on the buffer layer using thermal evapora- Hall effect measurements, which indicated µ ~ 790 cm2 tion. Then, the sample is annealed in UHV to induce V−1 s−1 at a rather high hole concentration of 2  ×  1013 intercalation. The properties of the QFSLG formed by cm−2. While oxidation of the SiC interface by oxygen intercalation depend on the amount of material present or water intercalation leads to highly defective QFSLG, at the interface. For Au two phases were observed [693]: this technique is promising for the synthesis of decou- a highly n-type doped one with the Dirac point located pled QFBLG on SiC(0 0 0 1). One may speculate that the ~0.9 eV below the EF for 1 monolayer Au at the interface buffer layer, which is highly corrugated and in which C and a weakly p-type doped phase with the Dirac point atoms are partially sp3 hybridized [723], is more prone at 0.1 eV above EF for one third of a monolayer Au at to be attacked by the oxygen during intercalation. On the interface. Spin-resolved ARPES indicated a Rhasba- the other hand, in the case of SLG, the buffer layer is type spin–orbit coupling in the π-band of QFSLG at protected by the graphene layer on top of it. energies where it interacts with Au bands [696]. In the Intercalation with other elements was also stud- case of Ge [691], an n-type doped QFSLG is obtained ied. Contrary to Si, that intercalates only at T  >  800 when 1 monolayer of Ge resides at the interface while °C through migration at graphene domain boundaries a p-type doped QFSLG is observed when 2 mono­layers and other defects, Li [698, 699] was found to penetrate of Ge are intercalated. The hole and electron concen- into the C layer already at RT through the formation of trations in the QFSLG were ~p   =  4.1  ×  1012 cm−2 and 78 2D Mater. 7 (2020) 022001 C Backes et al Table IV.2. Characterization techniques for QFSLG and QFSLGon H-terminated SiC(0 0 0 1). Method Observation References Process: buffer layer 6  √  3 to QFSLG LEED Strong weakening of the superlattice diffraction spots caused by the buffer layer [645–647] XPS Disappearance of the C1s signals related to the buffer layer and appearance of an [647, 704, 708] asymmetric C1s signal due to QFSLG; Shift of the C1s signal related to the SiC bulk to lower binding energy due to a change of surface band bending (see figure IV.22(a)) XPS Shift of the C1s signal related to the SiC bulk to lower binding energy [647, 704, 708] ARPES Appearance of a graphene π-band at the K-point of the hex. Brillouin zone with the [645–647, 703, 708] Dirac point located above the Fermi level (see figure IV.22(b)) LEEM Transition from a flat LEEM-I(V) curve to a spectrum with one dip [647] STM Topography image without buffer layer moiré structure [668, 707, 715] FTIR Appearance of a sharp Si–H stretch mode signal (see figure IV.22(c)) [658, 704] Raman Disappearance of the broad Raman spectrum of the buffer layer and appearance of [658, 704, 713, 717, 724, 725] sharp G and 2D bands; 2D band symmetric (see figure IV.22(d)); Process: SLG to QFBLG XPS Disappearance of the C1s signals related to the buffer layer and increase of the [645, 680, 704, 708, 713, 721] asymmetric C1s signal due to QFSLG(see figure IV.22(a)) XPS Shift of the C1s signal related to the SiC bulk to lower binding energy (see [645, 680, 704, 708] figure IV.22(a)) ARPES Disappearance of the linear π-band of graphene with the Dirac point below the [645, 704, 708] Fermi level and appearance of the two parabolic π-bands of BLG with the Dirac point located above the Fermi level (see figure IV.22(b)) LEEM Transition from a LEEM-I(V) curve with one dip to a spectrum with two dips [645, 646, 680] STM Topography image without buffer layer Moiré structure [707] FTIR Appearance of a sharp Si–H stretch mode signal (see figure IV.22(c)) [704] Raman Disappearance of the broad Raman spectrum of the buffer layer and change of the [704, 713] symmetric 2D band to an asymmetric 2D band considerably broader than that of QFSLG n  =  4.8  ×  1012 cm−2. This means that the shift of the basal (0 0 0 1) plane (~27°). Upon heating in the 1400 Dirac point with respect to EF is almost symmetric. °C–1600 °C range for graphitization, the SiC steps flow Note that a mixed phase can also be prepared in which to produce the equilibrium facets at that T. E.g., for rib- n- and p-type regions coexist. bons oriented along the graphene armchair direction,    This was employed by Baringhaus et al [692] to   at 1 1 0 7 , 1 1 0 6 , and sym- stable facets are found fabricate pn-, np-, pnp- and npn-junctions which metrically at 1 1 0 7 , 1 1 0 6 [731]. were investigated by 4-point probe transport measure- For growing graphene on natural steps, SiC sur- ments. Evidence for Klein tunnelling of charge carries face treatments are used to organize the step-terrace across such barriers was observed. structure prior to graphene growth. These include [615] annealing the chemically and mechanically IV.1.10. Structured growth processed polished SiC substrate in vacuum, in Ar, in The fact that growth starts at a step edges on the a face-to-face SiC configuration or with a refractory SiC(0 0 0 1) face (see above), can be used to produce capping grid [642]. The key is to obtain a stable step graphene nanostructures for electronics. Masking configuration prior to graphene growth. The T for methods, for selective growth using AlN [727], SiN surface ­structuring are chosen according to the back- [728] and amorphous C masks [642] with tens of nm ground pres­sure to prevent graphene growth, i.e. 1100 precision have been proposed. The template growth °C–1200 °C in vacuum, 1300 °C–1700 °C in Ar or face method on sidewalls of SiC trenches [604, 615, 618, to face. Step bunching by annealing in these controlled 619] is described below. It circumvents the detrimental conditions produces arrays of straight steps and ter- effect of traditional plasma etching [729, 730] and races with uniform width and height (figure IV.24). resulting sidewall nanoribbons show RT ballistic The step structure depends on the polytype [732] the transport properties [620]. local miscut of on-axis SiC wafers (see above), which The idea is to tune the parameters so to stop growth may vary up to  ±0.1° across the whole wafer, and just as graphene covers the step sidewalls. Steps can be other factors such as the heating method. A complete either natural steps obtained by step bunching at the systematic study is still lacking. After treatment, the SiC surface, or the sidewall of trenches of various shape terraces are up to 10–30 µm in width and extended etched in the SiC substrate. The width of the ribbon over several hundreds µm in length (figure IV.24). is then set by the step height and the angle with the Examples of ribbons grown on the sidewall of natu- 79 2D Mater. 7 (2020) 022001 C Backes et al (a) (d) 10 µm 38nm (b) (e) (c) 20nm Figure IV.24. Examples of sidewall ribbons (a)–(d) on natural steps, 20 nm high. (a) Topographic AFM image, (b) lateral force AFM image, revealing graphene (dark) on the sidewall, (c) profile of the steps along the red line in (a) [615]. (d) Example of isolated ribbon more that 80 µm long, 38 nm high (top) lateral force AFM—the ribbon is bright; (bottom) 3d rendition by combining topography and lateral force. (e) Cross-sectional TEM high resolution image of a sidewall trench etched in SiC, recrystallized at a stable facet angle of ~30°, over which MLG is draped. Adapted from Ref. [734], copyright American Chemical Society. ral sidewall steps (obtained by step bunching) and on which the GNRs grow at~1500 °C–1600 °C [618]. The etched trenches are shown in figure IV.24. exact T and time depend on the specifics of system, In a more controllable manner, nano-structured such as the Si escape leak in the furnace CCS method graphene can be grown on various shapes etched in [597, 620]. The sidewall ribbons produced this way are SiC(0 0 0 1) [733]. Trenches can be etched along the ballistic conductors at RT on micrometre distances     4H-SiC 1 1 2 0 , and 4H-SiC 1 1 0 0 directions to pro- [620]. Similar sidewall ballistic ribbons have been pro- duce GNRs along the zigzag or armchair orientation, duced with the same technique, but by a dc current owing to the epitaxial orientation of graphene on SiC annealing of conducting SiC wafers (6H-SiC N-doped) (see figure IV.1(a)). For this, an e-beam or a photore- in an Ar atmosphere of 4  ×  10−5 mbar (sample clamped sist is patterned on SiC(0 0 0 1). Plasma etching is used by graphite contacts) [620]. Well-ordered crystal fac- to etch SiC with the resist as a mask for shallow etching. ets from at~1150 °C, and further annealing to 1300 °C Various recipes have been reported [615, 618–620]E.g., results in growth of extended GNRs [620, 736]. reactive ion etching (RIE) (using SF6 and O2 ratio 20:7) Because the steps in SiC are etched and annealed with etching rates ~0.3 nm s−1, or 43% SF6/23% O2/33% prior to graphene growth, the facets onto which gra- Ar RIE operating at 30 mTorr where the radiofrequency phene grows are smooth and atomically defined. This power was tuned to give a SiC etch rate ~0.8 nm s−1 is demonstrated in cross-sectional transmission elec- [616], allowing fine control of the etch depth. For deeper tron microscopy [734] (figure IV.24(d)), and in ARPES etching, a Ni mask can be used, that is evaporated on the [731] where measurement integrating thousands of resist and lifted up. The Ni mask is removed after RIE parallel GNRs show graphene electronic-bands. etching by an ultrasonic treatment in nitric acid [618]. GRNs of any nominal orientation can in principle The resist pattern is transferred to the SiC as a be produced. By design, tens of thousands of sidewall template for graphene growth, and complex intercon- nano-structures can be produced all at once, follow- nected graphene structures can be designed (see Refs. ing the recipes above [604, 618, 731, 733], in particular [604, 618, 733]), including pillars to produce nano- long straight parallel GNRs [615]. However, sidewalls metric graphene rings [604]. Although trenches of any may in some cases show restructuring after annealing height can be etched, well-formed graphene ribbons [604, 620, 733] with rounding and faceting [597, 615]. are best grown on sidewalls 15–35 nm deep. This is The effects are sensitive to a number of factors such as because shallow trenches (⩽10 nm) tend to get washed the step direction, growth condition (both T, time, and out upon annealing in the CCS furnace and deep also possibly the type of heating), the pinning of steps trenches tend to break into multiple facets revealing (e.g. under amorphous C pads such as in, figure IV.9, or multiple parallel ribbons [615]. by defects), or the polytype. The most prominent fea- In the CCS furnace, the structured SiC chip is then ture of the GNRs, i.e. RT quantized ballistic transport, heated to 1100 °C to allow the vertical etched sidewalls is however a very sturdy result and was observed for to crystallize into the equilibrium facets [735] onto straight as well as curved GNRs [620]. 80 2D Mater. 7 (2020) 022001 C Backes et al IV.1.11. Integration with silicon wafers IV.2. CVD growth on SiC As mainstream electronics and very large scale High quality graphene can be grown by CVD on integration is based on Si technology, schemes to insulating and conductive SiC substrates. In a CVD integrate graphene grown on SiC with Si wafers have process, C atoms are provided externally by C gaseous been devised. Two routes have been investigated. The precursors and deposited on the SiC substrate, which first is based on the standard SOI technology, where is different from Si sublimation. The method enables a thin monocrystalline Si layer ready for CMOS growth directly on the SiC surface on both Si and processing is bonded on top of graphene on SiC. C-faces. A CVD process (see section V) typically involves Figure IV.25 shows the principle of the design. This 3d hydrocarbon precursors: methane, propane or acetylene integration, inspired by the industrial development of delivered to the reactor by a carrier gas (like Ar). It offers 3d integration stacking thin-film electronic devices, the precision of synthesizing a pre-defined number of C realizes the interconnection of the SiC supported layers [611, 714], including a single layer (buffer layer) graphene platform, preserving the integrity of on the Si-face of SiC, and is less sensitive to SiC surface graphene, with a Si wafer, enabling, in principle, the defects than the sublimation method [743]. With CVD full spectrum of CMOS processing. [737]. the nucleation sites for graphene growth are located at The main steps consist in the fabrication of the atomic steps, therefore enabling step-flow epitaxy. ­graphene structures (either from patterned or from A CVD process requires high T. However, the lowest T template growth on SiC sidewalls), evaporation of must be higher that of thermal decomposition of the an alumina film by ALD as a bonding layer between gaseous C precursor (~1000 °C). However, to form the SiC and the oxidized Si wafer, wafer bonding in a graphene of good structural quality requires T ~ 1500 pressure module and finally splitting of the Si wafer by °C to 1800 °C. At that T, SiC substrate decomposition annealing at 400 °C (smartcut) to leave a thin crystal- (i.e. Si sublimation) takes place. Graphene growth by line Si layer on top. The main advantage of the pro- Si sublimation occurs (from ~1300 °C) before the T of cess is that the growth T (1400 °C and above) is not CVD growth is achieved. Therefore, the first C layers will limited by the presence of Si (melting point 1414 °C). be formed by Si evaporation, not by CVD. Also, after the The Si wafer resides on top of SiC and is therefore fully CVD growth is finished, uncontrolled sublimation may accessible. Finally, the stacking of SiC and Si wafers take place, causing additional carbon layers to grow. increases the areal density inspired by the 3d stacked The process presented below [611] is a method layers Very Large Scale Integration Technology. for manufacturing graphene by vapour phase epitaxy The second strategy, so called graphene on-silicon (VPE/CVD), in which SiC substrates may be used (GOS), takes advantage of the heteroepitaxy growth owing to a control of the Si sublimation process. Ref. of 3C-SiC on Si to produce graphane on SiC cov- [638] used a combination of hydrogen and propane to ered Si substrates (3C-SiC is the only SiC polytype balance C supply and etching [638, 740], which pro- known to grow on Si) [738–740] (see figure IV.25). vides excellent samples for quantum Hall effect resist­ The polar Si-terminated 3C-SiC(1 1 1)/Si(1 1 1) sur- ance standards [612]. face grown in UHV shows graphene Bernal stack- Ref. [744] relies on the creation of the flow con- ing with an ­interfacial buffer layer, similarly to the ditions in the reactor that controls the Si sublima- 4H- or 6H-SiC(0 0 0 1) surfaces [738]. Conversely, the tion rate and enables mass transport of hydrocarbon ­C-terminated 3C-SiC(1 1 1)/Si(1 1 0) shows a non-Ber- to the SiC substrate through the argon gas boundary nal stacking, with the absence of an interfacial buffer layer. Laminar gas flow over the SiC surface consists layer, consistent with a C-face termination [738]. of layers moving at different velocities, due to the The quality of these graphene films is mod- shear stress between adjacent gas layers. The Reynolds est as shown by large Raman D peaks. The disorder number (Re) measures the ratio of inertial forces to results from Si diffusion through SiC grain bounda- viscous forces and consequently quantifies the rela- ries (due to a large ~20% lattice mismatch between tive importance of these two forces in a given gas flow Si and 3C-SiC) and the lower T allowed for graphene [745]. Tuning Re (Vd/ν), where V is the gas velocity, d growth (limited by the Si melting point) [741]. ­Process is the characteristic dimension of the reactor and ν is optim­ization is being developed. E.g., growing an epi- the kinematic viscosity, enables the formation of an Ar taxial AlN layer on Si prior to SiC growth reduces Si boundary layer thick enough to confine the Si vapour, ­out-diffusion and helps grow higher quality graphene and slow the Si escape that causes graphene to form. [738]. Catalytic growth of SiC on Si with a NiCu coat- The Ar layer however allows the diffusion of hydrocar- ing [742] allows growth of graphene on predefined bon to the SiC surface and, in consequence, the CVD locations. For electronic application requiring high growth of graphene on the SiC surface. quality graphene, alternative schemes have been When the velocity of Ar atoms above the substrate developed (see figure IV.25(a)). surface is sufficiently high, the flowing gas does not Growing an epitaxial AlN layer on Si prior to SiC inhibit sublimation, because Si escapes in the Ar carry- growth significantly reduces Si out-diffusion and helps ing gas [744]. If the velocity (related to the product of the grow higher quality graphene [738], as well as and pressure in the reactor by the flow through the reactor interface NiCu layer [742]. expressed in litre/min) is reduced below a critical value, 81 2D Mater. 7 (2020) 022001 C Backes et al Figure IV.25. (a) Principle of Si-wafer to graphitized SiC-wafer bonding with the SOI technology, and (b) SEM image of the cross- sectional bonding showing graphene at the interface (after [737]). (c) Growth of graphene on an epilayer of SiC grown on Si(1 0 0), schematics and LEED diffraction, showing rotational order on 3C-SiC(1 0 0) compatible with C-termination, and Bernal stacking on 3C-SiC(1 1 1) compatible with Si-termination. Adapted from Ref. [738], © IOP Publishing Ltd. a so-called ‘stagnant layer’ of Ar is created above the sur- Similarly to the sublimation growth method, the first face, while successive gas layers, starting from the sub- layer to form on the Si-face is the buffer layer. Hydro- strate, move with increasing velocity [745]. The product gen intercalation decouples the CVD grown buffer of flow rate and pressure to obtain this critical condition layer from the SiC substrate forming a QFSLG. The depends on the reactor geometry and is adjusted exper­ use of CVD facilitates the formation of 3LG however, imentally. In the case of the Aixtron -VP508 reactor, an owing to its high structural quality, hydrogen diffusion Ar flow rate ~6 l min−1 and a pressure ~100 mbar were through those layers is practically unattainable. used in order to completely inhibit sublimation. If the QFSLG exhibits much µ than SLG on the Si-face, Ar flow is increased to ~26 l min−1, the stagnant layer and T independent µ. This is desirable for high-speed thickness decreases, enabling again Si sublimation. electronics. The carriers change from electrons for SLG Therefore, by adjusting the Ar flow rate one can regulate on the buffer layer, to holes for QFSLG. H2- intercalated the thickness of the gas layer inhibiting Si escape (the SLG exhibits higher µ than the un-intercalated one. number of atomic gas layers of higher or lower veloc- QFSLG on the Si-face is also largely resistant to ity than a typical velocity to start/stop sublimation), photoresists and solvents applied with standard pro- thereby regulating the sublimation process efficiency. cessing techniques. Hydrogen presence is maintained Graphene was grown by CVD/VPE at 1600 °C on up to 700 °C, high enough to meet the requirements of both Si-face and C-face of nominally on-axis 4H-SiC high-speed electronics and high-T sensing [746, 1746, or 6H-SiC semi-insulating and conductive chemi- 1747]. The intercalation of hydrogen was achieved cally mechanically processed substrates. SiC substrate in situ during the sample cool down, by switching Ar are hydrogen etched prior to growth to remove a few to hydrogen at 1100 °C at ~900 mbar. Cooling in H2 tens of nm upper SiC, damaged in the polishing pro- keeps hydrogen atoms trapped between graphene and cedure. Etching is performed in the same conditions substrate. Prior to unloading the sample, the process as applied for graphene deposition, i.e. T  =  16 000 C, gas is changed back to Ar. P  =  100 mbar. Hydrogen flow is not as critical as etch- X-ray and TEM experiments show an increase in ing time, and was adjusted to 5 min. interlayer spacing in hydrogenated graphene to 3.6 Å– The best graphene was grown at a pressure 3.8 Å, proving that hydrogen goes between the C layer ~30 mbar under anAr laminar flow in hot-wall reac- and the SiC substrate and increases the separation dis- tors, where graphene growth on the 4-inch SiC sub- tance between them, along with decorating Si dangling strate was also developed. Application of CVD to gra- bonds on the Si face [747]. phene manufacturing enables to obtain thick rotated Raman maps are shown in figure IV.26. The I(G)   graphene films on the C-face 0 0 0 1 SiC. It is also pos- and Pos(G) maps demonstrate that the SiC surface sible to obtain a several graphene layers on the Si-face, morphology has terraces and step edges, as expected which is not easy in the case of Si sublimation. from step bunching at high T. The single Lorentzian CVD graphene has been studied for both the Si- fitting and the FWHM (2D)~35 cm−1 indicate that the terminated and C-terminated SiC. However, more areas are SLG or decoupled multiple SLGs. On the steps, attention was directed to SiC(0 0 0 1) since Si-face the I(G) is higher and the 2D peak is blue-shifted and slower growth kinetics allows a better control of the N. broader (FWHM (2D)~62 cm−1), consistent with BLG. 82 2D Mater. 7 (2020) 022001 C Backes et al Figure IV.26. Raman measurements of SLG grown by CVD on SiC (SiC Raman peak subtracted) (a) spectrum, (b) map 20 mm  ×  20 mm (BLG on the step edges is depicted in red), (c) histogram of FWHM(2D) [751]. This process leads to the formation of a step-ter- with a single transecting step exhibit µ~4500–6500 cm2 races structure. The steps remain within the nm range V−1 s−1, and structures with 2–3 transecting steps have and are typically limited to 20 nm in height. The step µ~3500–5000 cm2 V−1 s−1. Improvements in µ are likely edges follow a general parallel alignment. Regardless of to be observed in SLG with large uniform terraces [750]. N, there is an additional layer at the step underneath QFBLG is also p-type doped, typically in the the primary layers. As a consequence, we get BLG at the range between +8 × 1012 cm−2 and +1 × 1013 step edges, [712, 748, 749]. cm−2 for 6H-SiC and between +1.1 × 1013 cm−2 Electrical parameters were measured focusing on and +2.5 × 1013 cm−2 for 4H-SiC. A correlation was large-scale and statistical parameters. The µ and n of observed between µ and n. The highest quality and µ multiple as-grown and hydrogen intercalated samples are associated with the hole concentration approach- were characterized using Hall effect measurements ing a well-defined substrate-dependent values, con- on 10 mm  ×  10 mm in the van der Pauw geometry in sistent with defects augmenting the doping level [714]. ambient conditions. The n and µ were ~ 1.5  ×  1013 The scattering at the step is attributed to a lower n at cm−2 and ~ 1400−1, 800 cm2 V−1 s−1, respectively. After the step (sidewall GNRs were reported to be charge intercalation with hydrogen: n ~ 8  ×  1012 cm−2 and µ neutral [620]), and to disorder near the step edge. ~ 6500 cm2 V−1 s−1 for 10 mm  ×  10 mm MLG [747]. Because step edges are believed to result from SiC As expected, the electrical properties differ on the surface etching in hydrogen prior to graphene growth terraces and if there are step edges crossing the device. the anisotropy calls for a etch shortening. However, The step edges prove more resistive and, therefore, after a milder H2 etching, the transport parameters introduce a significant anisotropy [748]. CVD allows were reported to be worse, in the reverse proportion the fabrication of a large number of devices (here 320 to the steps height. This can be understood by a higher Hall structures). Excellent performances were achieved, step density (for the same substrate miscut angle) with µ  >  7000 cm2 V−1 s−1, measured for devices (sizes interfering with the device size. There is thus a need to of 5  ×  5 µm2 and 10  ×  10 µm2) located on a single ter- optimize the crystal surface with low step height (no race, of widths ranging from 3 to 15 µm. Structures higher than a SiC bi-layer). 83 2D Mater. 7 (2020) 022001 C Backes et al V. CVD, PVD & MBE a single wafer tool, multiple wafers can be loaded (e.g. more than 50 Si wafers) and low cost of ownership. The V.1. Growth on metals second most common growth system, is a single wafer In this section we will discuss the growth of graphene tool [766] which has many advantages as well, flexibil- on metals by : (1) CVD; (2) physical vapour deposition ity in wafer to wafer processing, design of experiments (PVD); (3) diffusion and precipitation; and (4) in the development stage, heating system (lamps ver- molecular beam epitaxy (MBE). In addition, we sus resistive and top versus bottom heating), base pres­ discuss the various growth techniques with respect to sure control and clustering with other processes on a reactor design, precursor delivery, substrates heating multi-chamber platform, rapid heating and cooling options, and growth T. There have been many reports [767, 772, 773], etc. In applications where graphene on growth on many elements and alloys and, while we grain size is not a critical factor, PECVD can also be understand many of the growth mechanism, there is used with the advantage of enabling lower growth T as still a lot more work to be done. Graphene film transfer well as much shorter growth times. This is particularly is discussed in section VI. useful to grow graphene on insulators [774]. The flexibility and pervasive success of CVD has There are a few reports on the growth of graphene made it a process of choice for the growth or deposi- by photothermal CVD [772, 775] where, photochemi- tion of many materials. Growth of graphite films, cal effects might be affecting growth. The principal given its refractory nature, has been a challenge and advantage is fast heating and cooling that could be taken many researchers have resorted to the use of catalysts advantage of in the case of controlled layer growth of [752–755]. These ideas have been expanded to the graphene on substrates with a higher C solubility than growth on metallic substrates by delivering hydrocar- Cu.The rapid cooling can also benefit any effects that bon precursors in a CVD geometry [756, 757, 765]. residual oxidants may have on graphene during cooling. One of the challenges in the growth of graphene on Ni and Co is the precise control of N [758]. SLG is V.1.1. CVD growth on copper difficult to grow Ni [756]. It is also difficult to grown The initial motivation to use Cu as a catalytic surface MLG with a controlled N on Ni and Co, especially to grow graphene was to take advantage of the very when CH4 is delivered at low pressure. However, there low C solubility in Cu up to the melting point of Cu are indications that when the Ni surface is exposed to [776, 777]. This was expected to eliminate the need CH4 at atmospheric pressure and CH4 dissociates into for a diffusion/precipitation mechanism[753, 778]. C on the surface of Ni, a more uniform graphitic film The binary phase diagram for Cu and C is shown in ensues upon cooling [762], growth on Cu led to the figure V.1. The solid solubility of carbon in Cu at 1100 desired result of large SLG. The most striking results °C is ~0.0076 w% at a T slightly above the melting have been achieved on Cu using methane and hydro- point of Cu [777, 779, 780]. Mostly on this basis, it is gen at low pressure, whereby where SLG was grown argued that as CH4 dissociates on the metal surface to on over 95% of the metal surface at ~1000 °C in a few form CHx radicals at high T with the aid of the metal minutes [765]. This process is self-limiting due to the catalyst the resulting carbon atoms arrange in the form very low solubility of C in Cu [332].The transfer onto of graphene on the surface of Cu, thus creating SLG. the substrate of interest is usually achieved by etching This contrasts that on metals with a higher C solubility, the metal subtrates [765] with various chemicals [763, where the C atoms dissolve and diffuse into the metal 764] by electrochemical delamination allowing one with subsequent segregation and precipitation at the to reuse Cu with a significant reduction in cost. Sec- metal surface upon cooling. Growth on Cu is limited tion VI provides a summary of the most commonly largely to one monolayer [765, 781] for a low-pressure used methods. catalytic CVD process. The growth details were studied There are many types of CVD systems. The selec- by using 12C isotopes to try to understand whether tion of the specific system dependens on many fac- graphene was growing directly on the Cu surface or tors. The ultimate objective is to produce a material at precipitating from the bulk of the Cu substrate [781]. a competitive cost that meets the requirements. Here The Raman map of the isotopically labelled graphene is a list of a few CVD approaches that can be consid- transferred on a SiO2 substrate showed that at least ered/have been evaluated for the growth of graphene: macroscopically graphene was growing directly on (1) furnace system (horizontal or vertical) [765]; (2) the surface at high T and not upon cooling as in the single wafer cold wall reactor (resistively heated [766], case of Ni [781]. The growth of single-crystal graphene flash lamps [767] (direct heating, top or bottom, or domains with controlled edges with orientations that through a susceptor); (3) plasma PECVD [768]); (4) range from zig-zag to armchair via a CVD growth– low-pressure CVD (LP-CVD); (5) atmospheric pres­ etching–regrowth process was reported [782]. FLG sure CVD (APCVD) (continuous or single size sub- are also observed on Cu. There are multiple sources strate) [769–771]. In the case of graphene, the first of carbon that lead to the presence of the adlayers. and most common process is a furnace process [765]. First, some of the C dissociated on the surface of Cu The advantages are that the furnace is made of quartz dissolves into Cu and after growth at high T it diffuses which may be compatible with more precursors than to the Cu surface because of the lower solubility as the 84 2D Mater. 7 (2020) 022001 C Backes et al Figure V.1. Cu–C Equilibrium phase diagram at the Cu rich side, L is for Liquid Ref. [790]. Cu cools, and forms FLG under the already grown 1. Cu surface cleaning/preparation prior to graphene [783]. These layers grow under the graphene loading into the growth chamber on Cu as a result of precipitation upon cooling [783]. 2. Surface cleaning and annealing in the growth Even though it was reported that C might diffuse chamber under the growing graphene to form FLG [784], it is 3. Exposure to a hydrocarbon precursor with more difficult to accept this mechanism since it does subsequent adsorption of the same on the Cu not take into account the C dissolved into Cu. If only surface the intentional C from the dissociation of methane 4. Nucleation of graphene by the dissociating CHx were present, then a small percentage of the surface radicals and local supersaturation of C to form area would be covered with adlayers. However, in some graphene nuclei cases many FLG have been observed across the surface 5. Surface diffusion of CHx radicals and of Cu/graphene and this can only be explained by the attachment onto the graphene edges presence of other sources of C, and isothermal growth 6. Dehydrogenation of the CHx radicals attached as a result of supersaturation [785]. It was also reported to the graphene edges and dehydrogenation [791] that the surface of Cu can be covered with a of some of the isolated CHx radicals with high concentration of unwanted C from ambient, subsequent diffusion into the Cu poor cleaning and cleaning using hydrocarbons.This 7. Coalescence of the growing graphene domains can give rise to a high concentration of nuclei as well to form continuous polycrystalline films or as FLG beyond the solubility limit of C in Cu [785]. single crystals. The presence of the uncontrolled and unwanted C is also supported by anecdotal reports indicating Substrate selection, surface preparation, growth that graphene can be grown on metal surfaces in the T and process are critical to achieve films that meet absence of a C precursor. Ref. [786], pointed out the device requirements. Substrate surface quality, surface importance of a high H2/CH4 ratio in the growth of impurities, surface roughness and gas composition SLG. The benefit of a high hydrogen concentration and pressure have large effects on the surface rough- may arise from the details of the reactions at the metal ness, defects (Raman D-band), amount of FLG, and surface, but also from the presence of excess oxygen in electrical properties. Graphene films of varying sizes, the growth chamber. This is highlighted by the fact, on Cu as well as other metal films have been grown that graphene is etched during cooling in the absence in vertical single wafer cold wall reactors, horizontal of H2 or CH4 [787]. The effect of growth parameters furnaces [765], open continuous furnaces as well as such as T, and CH4 flow rate and partial pressure on in plasma enhanced single wafer systems [966]. Each the growth rate, domain size, and surface coverage of one of these approaches requires optimization in graphene led to the definition of a 2-step CVD strategy order to achieve the desired results. Typically, growth for the growth of large single crystal graphene [788]. conditions, such as reactor design, vertical versus High-throughput growth of graphene on Cu foil was horizontal, gas delivery system, precursor selection, achieved at ambient pressure, interesting to reduce heater, low versus high T growth, low pressure versus costs [789]. atmospheric pressure, must be considered and optim­ The graphene growth process on Cu surfaces to ized. The aforementioned factors depend on material first order can be described by the following steps, requirements as well as cost needed by a particular [784, 786, 791–794]: application. 85 2D Mater. 7 (2020) 022001 C Backes et al Intensity [a.u.] 1500 1750 2000 2250 2500 2750 -1 Raman shift [cm ] Figure V.2. (Upper) Schematic diagram of the thermal profile during graphene growth, type of gasses used and steps duration in case of growth SLG at 1000 °C, (Lower Left) Optical image of a graphene domain grown on molten Cu on a Mo substrate. Scale bar is 100 µm. (Lower Right) Raman spectrum of a graphene grown on molten Cu. Typically, graphene is grown on Cu foils [332], Cu T [799, 804] than the Cu melting point, 1084 °C, using thin film on sapphire, and bulk Cu single crystal foils any one of several hydrocarbons (such as methane or [782, 794–796]. The most commonly used Cu foils are ethylene), ethanol hydrogen and argon are also used of a high-purity 99.9999% in a variety of thicknesses [799, 804]. The effect of the H2/CH4 ratio on the shape and sizes. In addition, oxygen [783, 791], carbon and of graphene domains has been widely reported [786, other impurities [785] in or on Cu are have large effects 800, 801]. Low pressure CVD with very low precur- on the growth mechanism. The quality of graphene sor flow rate has been effective in growing large area will depend on the quality of the Cu surface morph­ graphene filmsVarious other methods of heating the ology and crystallographic orientation. The surface of substrate have been use in addition to resistive heating, the metal can be improved by electropolishing [769, such as magnetic inductive heating [805] and a hybrid 770, 786] Many experiments have been performed process combining hot filament and resistive (ther- on Cu annealing [797, 798], heating the substrate to T mal) heating [806], but these techniques have not been above the melting point of Cu,[799, 800], Cu surface widely adopted yet. Growth has also been carried out oxidation, [786, 790, 802], using Cu with controlled in the presence of a plasma in an attempt to reduce the crystal orientation [786] and oxygen free Cu [791]. growth T well below 800 °C. Growth of graphene on bulk single crystals or sin- A typical growth procedure is shown schemati- gle thin films of Cu would be advantageous. It would cally in figure V.2. Prior to growth high purity Cu sub- be ideal to grow high quality graphene directly on a strates (foils) having a thickness ranging from 25 to dielectric surface Ref. [802] diffused C on thin Cu on 35 µm and of a desired surface area are procured. The SiO2 to grow graphene directly on the dielectric sur- are then cleaned in acetic acid for ~10 min and then face. However, at the films grown using this process transferred in de-ionized water for additional 10 min are still not of high enough quality to compete with and dried with N2. After, the Cu foil is introduced in films grown directly on the metal surface. For a more the growth chamber and the chamber is evacuated and extensive discussion on the growth of graphene on back filled with H2 to a desired pressure. CH4 and H2 di­electrics, see section V.3. are usually circulated to purge the lines (50:50 sccm) Growth of graphene on Cu metal has been carried at pressure  <4 mbar prior to substrate heating. The out in horizontal tube furnaces [765] as well as in cold chamber T is then increased to 1000 °C under 20 sccm walled single wafer tools [766, 803] and even at higher of H2 in 70 min (see figure V.2 upper frame). T is then 86 2D Mater. 7 (2020) 022001 C Backes et al Figure V.3. Optical images of (a) continuous and (b) discontinuous graphene films on Cu. kept constant at 1000 °C for an additional 30 min. T because of the large difference in the coefficient of During this last step, the Cu substrate T and micro- thermal expansion between graphene and Cu are also structure are stabilized. Heating of the Cu foil to the visible as narrow lines criss-crossing the surface of growth T can be performed also in Ar and then in H2 the graphene. If Cu of lower quality (purity 99.9% or gas flow at a pressure  <100 mbar. The purpose of 99.999%) is used, e.g. Cu_5N and Cu_3N, the films this step is to promote grain growth of the Cu foil. A are heterogeneous mainly in terms of their thickness. ‘standard’ graphene growth process follows: 5 sccm of In figure V.4 light grey color represents areas with SLG. CH4 or C3H8 are then flowed for several minutes (typi- Atomic resolution images of graphene on Cu as cally ~30 min), after the growth phase, the substrate is measured by STM (see section IX.1) presented in fig- cooled to RT by translating the furnace away from the ure V.5 show the configuration of the carbon atoms substrate. The cooling can be done with the precursor in SLG. The corresponding Fourier transform of flowing, hydrogen flow or under vacuum. The ambi- the STM image (upper right inset) indicates that the ent during cooling is also critical especially in cases maxima correspond to the pattern formed as a result where the base pressure of the chamber is too high. In of the orientation of the SLG in relation to the sub- this case leaking oxygen can etch the graphene at the strate (lattice mismatch). The figure V.5 shows an area highest T. Once T is near RT, the gasses are closed off 3  ×  2.2 nm with a graphene unit cell drawn upper left and the chamber is evacuated and back filled with an inset) exhibiting a lattice constant of 2.44 Å [812]. The inert gas to one atmosphere. Growth of graphene on shape, orientation, edge geometry, and thickness of polycrystalline Cu foils using this process flow usually CVD graphene domains can be controlled by the crys- yields films with a grain size ~10 µm. By applying a tallographic orientations of Cu substrates [1749]. At longer annealing time, higher T, lower pressures, and the inception of the growth phase, during the nuclea- by maintaining optimal hydrocarbon flow or melting tion stage, the Cu roughness plays a key role. The Cu- the Cu, the domain size can reach several millimeters active sites are imperfections on the Cu surface, such as [807]. The lower frames in figure V.2 illustrate that grain boundaries, irregularities or areas of coarseness, domains ~800 µm can obtained systematically opti- or surface impurities. It is at these defects that most of mizing the CVD growth conditions and also show the the graphene nuclei are formed. The higher the den- corresponding Raman spectrum. sity of nucleation sites, the smaller the size of the gra- Optical microscopy with Nomarski contrast, i.e. phene domains. The density of these grain boundaries differential interference contrast (DIC) microscopy, is expected to affect electrical, chemical, and thermal is very useful (see figure V.3) to explore the quality of properties, and it is thus necessary to optimize the the Cu substrate surface as well as the grown graphene. graphene film properties according to the application DIC microscopy can expose cracks and micro-holes requirements [808–811]. in the graphene layer and can provide information on For some applications large (square meters) areas microstructural changes before and after anneal that of graphene may be needed, as it is the case for the func- can be correlated with graphene film quality. tional coating of large pieces of other materials. When Raman spectroscopy has been demonstrated to growing graphene on 500  ×  500 mm2 Cu substrates, be the best characterization technique to confirm the T uniformity across the metal substrate must be the presence of SLG on the Cu substrates [1748]. The controlled since the nucleation and growth character- technique is introduced in section IX.2.1, where repre- istics has a large T component [755]. The control of the sentative spectra of SLG, BLG, etc are also given. Meas- Cu metal foil T is critically important when growing urements performed by SEM plotted in figure V.4 can films at T close to the Cu melting point as shown in identify dark areas interpreted as adlayers of graphene. figure V.6, where part of the Cu foil melted at the edges, In addition, wrinkles formed during cooling from high due to an uneven T distribution across the Cu film. Ref. 87 2D Mater. 7 (2020) 022001 C Backes et al Figure V.4. SEM images of graphene layers on (a) Cu_mono, (b) Cu_6N, (c) Cu_5N and (d) Cu_3N substrates. Scale bars are 1 µm. [794] reported the growth of 5  ×  50 cm2 single-crystal starts by annealing the Cu foil for 5 min at ~935 °C (as graphene on industrial Cu foils. They first transformed measured by a pirometer) and pressure ~7–20 mbar. industrial polycrystalline Cu foils into single-crystal Next, CH4at a flow rate of 15 sccm (CH4/H2 mixture Cu(1 1 1) by thermal annealing using a T-gradient- 4:1) is introduced into the chamber. Under this con- driving technique. Then graphen grains growon the dition a continuous SLG film is grown on Cu in a few Cu(1 1 1) substrate and seamlessly mergeto form a tens of seconds, one order of magnitude faster than in large single-crystal graphene film. thermal CVD using similar growth parameters. The This technique allows the synthesis of large-size SLG coverage is ~94% in films grown in 60 s at 950 °C. single-crystal graphene films, with superior properties The SLG contains very small adlayer flakes, and defects for various high-end applications, especially in elec- (~3  ×  107 cm−2) [775, 813]. The light from the RTP tronics, such as large-scale fabrication of THz devices lamps enhances the catalytic effect of CH4 on the Cu and transparent conductive film to replace ITO. It will surface without generating defects. The grain size is in also enable the growth of various other single-crystal the µm range. Ref. [773] reported a rapid thermal CVD LMson graphene (RT-CVD) suitable for mass-production. In this setup, There are many alternatives for substrate heating, Cu foils are loaded vertically in the reactor, in which ranging from magnetic inductive heating [805] to 24 lamps heat vertical graphite receptors around the hybrid processes, combining hot filament and resis- Cu foil. They obtained a growth rate 7 times higher tive (thermal) heating [806]. Among them, photother- than thermal CVD using hydrogen-free N2 carrier gas mal heating (PT-CVD) provides up to 100 °C min−1 [773]. heating and cooling rates [767, 772] while keeping the It would be desirable to grow graphene at low T for chamber walls cold. The system is based on a halogen ease of process, equipment, contamination, as well as lamp-heated rapid thermal processing (RTP) (for up potential substrate melting issues. Ref. [814] reported to 300 mm wafers), where gas lines (CH4, H2 and Ar) growth of graphene on SiO2/Si, plastic and glass at and their mixing manifolds are incorporated. While close to RT (25–160 °C). However, in most cases the keeping the RTP chamber walls cold. This, together reported reduction in growth Tg/Tmp(C) is very small with the fast growth process, less than 1 min, helps to (⩽0.30) and for most materials the quality of the reduce the contamination from the walls. In this case, grown crystals is higher when the growth T is close the cleaned Cu foil is placed between two silica plates to the melting point. There have been some reports on a photothermally heated Si- or SiC-coated graph- on the growth of graphitic films at low T by CVD ite susceptor. Silica is needed between Cu and Si to [815], plasma assisted [816] and from solid sources prevent reactions between them. The growth process [814, 817] on or under Ni films but not on Cu. Growth 88 2D Mater. 7 (2020) 022001 C Backes et al Figure V.5. Atomic resolution STM of graphene on a Cu mono-crystalline substrate with the Fourier transform (upper right inset). Adapted from [812] with the permission of AIP Publishing. Figure V.6. Higher T near the edges of the substrate causes local Cu melting while the inner part is unaffected. at low T on Cu, Ni or Cu-metal alloys could be of great 781, 818] predominantly by a surface limited process, interest, but the grown graphene could contain many because of the low solubility of C in Cu. The films defects, as in the case of low T crystal growth of many grown using this process are polycrystalline, with a other electronic and optical materials. grain size that depends on the specific growth pro- cess conditions and they can range from less than one V.1.2. CVD growth of large graphene single crystals micron to hundreds of microns. Therefore, in order As in the case of crystalline semiconductor materials, it to increase the grain size, one would have to reduce may be advantageous to grow single crystal graphene the nucleation density on the surface of the metal. films in order to minimize scattering effects by pure Various approaches were developed to achieve this, grain boundaries and grain boundaries that may such as reducing precursor flow by using a Cu pocket react with ambient species such as OH, hydrocarbons, [128, 766, 819, 820]. An example of such pocket is etc. Also, the presence of grain boundaries and their shown in figure V.7(a). tilting has been studied in relation to the mechanical The growth of large graphene single crystals led properties of the synthesized graphene [811]. to the demonstration of very high µ in CVD gra- Graphene has been demonstrated to grow on Cu phene comparable to that of mechanically exfoliated surfaces by a nucleation and growth mechanism [765, flakes [821]. This is an important result since it has 89 2D Mater. 7 (2020) 022001 C Backes et al Figure V.7. (a) Cu ‘pocket’ enclosure. (b) Enclosing flat Cu foil with a quartz cover. (c) growth of graphene on Cu assisted by a continuous oxygen supply. Adapted from [766] (© IOP Publishing Ltd) and [825] (© Springer Nature). led to focused efforts on further understanding of the used a solid source of oxygen, AxOy—e.g. SiO2/Si, to graphene single crystal growth as well as transfer provide a continuous source of oxygen under the Cu [807, 822–824]. substrate while providing the hydrocarbon precur- Further reduction of graphene nucleation den- sor at atmospheric pressure (see figure V.7(c)). This sity was investigated by Ref. [783, 791], who showed process led to a growth rate~60 µm s−1 instead of~0.3 that graphene single crystals on polycrystalline Cu µm s−1 of Ref. [791]; this is a major improvement. of over 1 cm could be grown after treating the surface Another approach to reducing the graphene nuclei of Cu with oxygen. However, this process is time con- density on the Cu surface was performed by treating suming, growth times of over 10 h were used to grow the surface with melamine [827]. This process led to 1 cm single crystals. Oxygen not only passivates the Cu the growth of 1 cm graphene domains with µ reaching surface, leading to reduced nucleation density, it also 25 000 cm2 V−1 s−1 for graphene transferred onto SiO2/ significantly speeds up the growth of graphene crys- Si. Another method was proposed taking advantage of tals by changing the growth regime from diffusion to the residual oxygen in the CVD furnace for increasing edge-attachment-limited. A further explanation on the graphene domain size beyond one millimeter, and the decrease of the nucleation density on the surface simultaneously promote recrystallization of a poly- of Cu was provided by Ref. [785], where they observed crystalline Cu foil in the (1 1 1) structure through a that the surface and the bulk of the Cu foils may be peculiar ‘abnormal grain growth’ mechanism [828]. contaminated with impurities including C which can The process yields aligned high-quality graphene sin- aid in the formation of graphene nuclei. In this case, gle crystals smoothly merging together, with a crystal Cu oxidation prior to graphene growth removes car- alignment maintained over few centimeters (in 1 hr). bons from the Cu surface leading to a lower nuclea- In general, the following critical conditions should tion density. Following the Refs. [791, 826], Ref. [825] be met in order to grow large crystals: (1) suitable 90 2D Mater. 7 (2020) 022001 C Backes et al Figure V.8. Electropolishing set-up. (a) Three-quarter view. (b) Side view, showing the parallel orientation of the copper foil and the counter-electrode. Adapted from [766], © IOP Publishing Ltd. s­ ubstrate, (2) substrate surface cleaning and polishing the evolution of the crystal morphologies grown inside prior to loading in growth chamber, (3) surface pre- Cu enclosures as a function surface orientation, abso- treatment (oxidation, the use of melamine or other lute pressure, hydrogen-to-methane ratio (H2:CH4), surface treatment), (4) enclosing the sample to limit and nucleation density At low pressure and low H2:CH4, precursor flow, and (5) a non-reducing surface treat- circular graphene islands initially form. After exceeding ment at high temperature prior to growth. ~1.0 µm, Mullins-Sekerka instabilities [1750] evolve Substrate preparation is a very important first into dendrites extending hundreds of micrometers in step in order to create a clean and flat surface. Many the 〈1 0 0〉, 〈1 1 1〉, and 〈1 1 0〉 directions on Cu(1 0 0), approaches can be found in literature [583, 819] on Cu(1 1 0), and Cu(1 1 1), respectively, indicating mass mild surface etching using acidic solutions [791, 820, transport limited growth. Increasing H2:CH4 results in 829]. However, the most reproducible surfaces have compact islands that reflect the Cu symmetry. been prepared by electropolishing [766, 807]. The The nucleation density and growth rates are solution used for electropolishing has a volumet- affected by the growth pressure and gas flow rates. ric composition of 45% water, 25% phosphoric acid, While low pressure can help to radically reduce the 25% ethanol and 5% isopropanol. To increase viscos- nucleation density, it can also lead to extremely slow ity, 0.5% (weight) of urea is added. Electropolishing is crystal growth. With a gas flow of~1 sccm CH4, performed by applying a voltage between the sample 900 sccm Ar and 100 sccm H2, the graphene nuclea- (anode) and a Cu plate cathode. A typical electropol- tion density is on the order of several crystals per mm2 ishing procedure is 30 s at 12 V, with a cathode-anode and typical growth rates are ~15 µm min−1, allowing distance of 4 cm. To ensure homogeneous polishing, the synthesis of 1 mm-sized crystals in ~1 h. The use of it is important to keep the electrodes parallel. For this Ni–Cu alloys will enable much higher growth rates due one can employ a Coplin glass staining jar as the elec- to the higher catalytic activity of Ni [831]. Figure V.9 tropolishing vessel (figure V.8(a)), which has grooves illustrates the effect of using a nonreducing atmos­ to keep the foil and the counter electrode in place, phere and sample enclosure on oxidised Cu. ­figure V.8(b). Following the electropolishing, it is When hydrogen annealing is used, the nucleation important to rinse the samples under running DI density is ~10 000 cm−2 (a). Annealing in an Ar atmos­ water to completely remove the viscous electrolyte, phere reduces the nucleation density to ~1000 grains followed by a quick rinse with isopropanol. Finally, the per cm−2 [128, 791]. Substrates annealed in argon and samples are dried with compressed nitrogen. placed inside an enclosure have a nucleation density Owing to the catalytic nature of the Cu surface, ~10 grains per mm−2 or lower, enabling the growth it is important to reduce the carbon precursor flow of large crystals. Figure V.10 shows a schematic dia- impinging on the sample in order to reduce the nuclea- gram of the graphene growth process in a pocket and tion density [755]. One of the most commonly-used the image of a graphene single crystal grown inside approaches is to grow large SLG crystals inside ‘pock- the pocket. The interface between SLG and Cu (1 1 1) ets’ formed from Cu foil, [128, 783, 791, 819, 820]. The oxidises under ambient conditions within a few days, discovery of large graphene crystals inside a Cu pocket making SLG visible and decoupled from Cu. This is led to the investigation of other methods to achieve required for a reliable dry pick-up using h-BN [832]. similar results, for example a Cu tube [807], a sapphire There have been many observations on the shape cap over the Cu surface to minimize evaporation and of graphene domains on different Cu grains having the ‘hydrocarbon’ concentration [830], or placing a varied crystallographic orientation [786, 1751]. The quartz plate on top of a flat Cu foil at a desired distance shape varies depending upon the Cu orientation as from the Cu surface [766]. The process conditions to well as the growth conditions and Cu oxidation state achieve the largest single crystals will have to be optim­ [755, 791]. It has also been reported that growth of ized and perhaps combined with the use of Cu single graphene domains on oriented Cu(1 1 1) foils [794] crystals. Ref. [796] reported a comprehensive study of and Cu(1 1 1) thin films on sapphire [830] under 91 2D Mater. 7 (2020) 022001 C Backes et al Figure V.9. Nucleation density of graphene when using (a) hydrogen annealing, (b) argon annealing (c) argon annealing and sample enclosure. Adapted from [766], © IOP Publishing Ltd. c­ ertain process conditions are aligned. As the domain low pressure conditions (slow growth rates, growth progresses, the domains merge to form single size up to 1 cm) crystals as the Cu surface becomes fully covered by a (2) Growth on Ni–Cu substrates with a localized boundary-free stitching. This is true in the case where precursor delivery (relatively high growth the domains are aligned and the Cu surface is nearly rates) atomically flat [583, 833]. The advantage of this pro- (3) oriented single crystal Cu(1 1 1) foils cess, if large area Cu(1 1 1) films or foils are available, (4) thin Cu(1 1 1) films on sapphire substrates is that the growth time can be significantly reduced in (5) thick predominantly Cu(1 1 1) on sapphire comparison to the single nuclei approach. Efforts have followed by a peeling process so as to been dedicated to the growth large Cu single crystal facilitate sapphire re-use. foils [833] and [831]. More recently, a gas-flow-driven aligned growth of graphene on liquid Cu has been achieved [834]. Another method of preparing large V.1.3. CVD Deterministic seeded growth on Cu area Cu(1 1 1) crystals is to take advantage of the epi- Seeded graphene growth was reported by Ref. [839] taxial nature of Cu on c-plane sapphire and thicken patterned and etched CVD graphene patches were used the thin film using an electrodeposition process, after as nuclei for the subsequent single crystal graphene which Cu(1 1 1) can be peeled off from the sapphire growth. However, because of the polycrystalline nature substrate [835]. Significant activities are ongoing to of the starting graphene film, the resulting crystallites optimize the growth of graphene on controlled Cu were not aligned but each crystal was hexagonal. In surfaces in an effort to grow high quality crystals in a subsequent study [840] pre-annealed foils were production worthy growth systems [836]. patterned with arrays of crosslinked PMMA, then It would be advantageous to be able to grow sin- used as a carbon source for the growth of single- gle crystal graphene from a single nucleus as for bulk crystal arrays. In both cases the array periodicity was single crystals. Wu et al [837] used Ni–Cu alloys to ~20 µm and crystal size of up to ~15 µm. Due to the increase the catalytic activity in conjunction with a much lower nucleation densities required for large- localized precursor delivery technique to grow gra- crystal graphene, carbon-based seeding was found not phene single crystals over 3 cm in diameter. The growth to be robust enough for controlled growth of large- mechanism of graphene on Ni–Cu is more complex periodicity arrays. Metallic nucleation seeds were than on Cu or Ni and further development would chosen instead for large-crystal graphene arrays. Ref. require precise composition identification/selection [841] used Cr as the nucelation sites for localized single in order to ensure single layer growth and full cover- crystal growth of graphene. age by a single graphene grain. Ref. [838] reported the The nucleation seeds can be patterned by perform- synthesis of single-crystal-like SLG on polycrystalline ing optical lithography on electropolished Cu using substrates. The method relies on ‘self-selection’ of the positive photoresist, followed by thermal evapora- fastest-growing domain orientation, which eventually tion of 25 nm Cr and lift-off in acetone. Cr was chosen overwhelms the slower-growing domains and yields a because of its high melting point [1752], ensuring that single-crystal continuous film. thin films are not evaporated during the annealing of In summary, the growth of high quality, large area, the foil and instead form particles that can act as nucle- single crystal graphene can be performed in several ation points on passivated Cu. Spots with a diameter ways on: ~5 µm and a thickness ~25 nm lead to the most repro- ducible results [841]. Smaller seeds or lower film thick- (1) Growth on conditioned Cu substrates ness did not ensure reliable seeding, whereas larger (oxidized or treated with melamine) under spots were sometimes found to initiate the nucleation 92 2D Mater. 7 (2020) 022001 C Backes et al Figure V.10. (a) Typical growth scheme for a low pressure CVD process for SLG on Cu. (b) Optical dark field microscopy image of SLG on Cu foil using the process depicted in (a). Panel (B) is adapted from [832], with permission from AAAS. Figure V.11. Seeded arrays of large-crystal graphene. (a) Array with 200 µm periodicity and 100 µm crystal size. (b) Array with 500 µm periodicity and 350 µm crystal size Adapted from [841]. of several separate crystals. By c­hanging such flow Controlling the graphene edges may be important for ratio, it is possible to control the presence of residual some electronic applications and growth conditions Cr seeds during the growth by changing the CH4:H2 to optimize quality and growth rate(cost) should be flow ratio. After growing graphene using a H2:CH4 investigated. Another important point is the growth of ratio of 20:1, particles of Cr are found at the centre of edge-controlled graphene in the smallest dimensions seeded crystals, whereas when using a H2:CH4 ratio (nm scale) and the treatment of graphene after pat- of 100:1, Cr seeds are removed. Growth was carried terning and etching to set the edge structure. out in a 4-inch Aixtron BM Pro. All gases used in the CVD process had purities of 99.9999%. As shown in V.1.4. Low pressure & low temperature CVD figure V.11(a), arrays with a periodicity of 200 µm It is desirable to grow film at low T not only because show high nucleation control, with less than 10% of of the integration device flows, but also to overcome non-seeded nucleation. At larger periodicities (see fig- problems associated with substrates like Cu. One ure V.11(b)), sporadic nucleation was slightly higher. issue with CVD of graphene on Cu from CH4 is that Well-ordered arrays of crystals with a diameter of sev- the optimum growth conditions are typically at eral hundred microns can be synthesised [841]. T  >  1000 °C, very close to the Cu melting point of We now discuss the growth of single crystal gra- 1084 °C or even above its melting point. Cu has a vapor phene with controlled edge structure. Two types of pressure ranging from 8.1  ×  10−8 atm to 5.7  ×  10−7 graphene crystals have been grown: (1) dendritic, atm for the T range ~ 1000 –1084 °C [843]. During and (2) hexagonal [791]. The shape of the graphene annealing and graphene growth at low background crystals was controlled by the presence (dendritic) or pressures a measurable amount of Cu is transported absence (hexagonal) of oxygen on the surface of the Cu by the flowing gasses away from the Cu surfaces. The for low pressure CVD. Ref. [842] reported the growth presence of Cu on the surface of the quartz tubes in of 100 µm hexagonal graphene crystals on polished Cu the case of a furnace or the metal surfaces in the case foils at one atmosphere, with limited carbon flow, with of cold wall reactor can make the cost of ownership the crystal edges dominated by the zigzag direction. high, since the growth systems has to be cleaned on 93 2D Mater. 7 (2020) 022001 C Backes et al Figure V.12. Raman spectrum at the centre of a SLG domain (black) and averaged across a 20  ×  20 µm region of the film (red). (b) I(D)/I(G)map. (c) FWHM(2D)map. Adapted from [850], with permission from Elsevier. a regular basis to prevent Cu deposits flaking and from 60% to 98% and the grain size and quality of the particle formation. As a result there have been several deposits deteriorated. At 60 °C on glass and PMMA activities to (1) reduce the growth T to address the the Raman signal showed the deposition of highly evaporation and the potential integration of graphene defective graphene layers with nanometric grain size. directly in a device flow and (2) finding alternative Thus, the quality of the films is inferior to that metals or alloys as discussed above to minimize the Cu grown on Cu at higher T using CH4. In addition, the evaporation problem. It should be added that the cost delivery system for the liquids may be more com- and reliability of precursor delivery systems should plicated and more costly, not to mention the toxic- also be considered when designing a CVD. in order ity issues. Ref. [850] reported on the CVD growth of of preference, gases are preferred over liquids and high-quality, large-area predominantly SLG films liquids are preferred over solids. In an effort to reduce with domains o~1 µm2 and FWHM(2D)~35 cm−1 the growth T researchers have used liquid precursors on Cu foils using C2H4at 850 °C. Spectroscopic and such as alcohols [844], or aromatic molecules, such electrical measurements show that these films display as hexane [845, 846], benzene [847] and toluene high uniformity and crystalline quality with reason- [848]. In the case of benzene, high quality mostly able µ~1000 cm2 V−1 s−1 [850]. Figure V.12(a) shows monolayer graphene was synthesized at 300 °C with µ the average Raman spectrum of a ~20  ×  20 µm2 area of 2500 cm2·V−1·s−1 by APCVD [849]. Evaporation of (red) and a discrete spectrum taken at the centre of the catalyst was avoided and oxygen removed from the a SLG domain. I(2D)/I(G) is lower for the averaged deposition chamber to minimize contamination or spectrum as this includes the regions of second- structural disorder of the film. Deposition of graphene ary island growth. Figure V.12(b) plots a I(D)/I(G) at 100 °C was observed by this method but with lower map with brighter areas corresponding to regions quality [1753]. With extended time deposition, BLG of higher defect levels, highlighting the low defect or TLG started to form, indicating that the synthesis is levels throughout the film. The defects that are pre- not self-limiting [817]. At 160 °C continuous graphene sent here are attributed to domain boundaries. Fig- films of reasonable quality (µ~667 cm2·V−1·s−1) were ure V.12(c) plots FWHM(2D) across the same area of deposited on SiO2, however N was barely controlled. the film with brighter regions corresponding to wider At lower T (from 60 to 160 °C) the deposition of FWHM(2D). The majority of the film FWHM(2D) continuous films was not achieved, being the coverage ~35 cm−1, typical of SLG [96, 105]. 94 2D Mater. 7 (2020) 022001 C Backes et al Further characterization was carried out by fabri- solve the equivalent challenging problem of coefficient cating GFETs. Al2O3 pre-patterning of the Cu foil pas- of thermal expansion difference between metal and sivates the areas, giving a patterned graphene film on graphene [859].The growth of graphitic C on Ni, the final substrate. Electrical measurements were per- Co, Pt and other elements that have high C solubility formed with source-drain voltage (VSD) ~20 mV while occurs principally by a diffusion and precipitation the gate voltage was swept to  +60 V and  −60 V. IDS mechanism from the bulk of the metal [753, 754, versus VGS curves in figure V.13 give hole and electron 788]. Depending upon T and microstructure the µ~1100 and 700 cm2 V−1 s−1 at RT. These fall short of C diffusion can be dominated by grain boundaries the best reported values for CVD graphene [851], due or bulk diffusion. Ref. [781] showed using C isotope to the smaller domains (~1 µm) compared to films labeling that while in the case of Cu, graphene growth grown at higher T. is dominated by surface growth, on Ni this takes place Ref. [849] investigated the growth process of gra- by dissolution followed by diffusion and precipitation phene at ~300 °C, using C6H6as precursor at ambient upon cooling to RT. Growth on Ni and Co alloys of Cu pressure. They used a pump purge process multiple on the other hand has given better results, including times to minimize residual oxidizing species in the growth of large area single crystals [1755]. This is due growth chamber and annealed the Cu substrates at to a number of reasons, the principal being that at low 1000 °C under H2 and Ar gasses for 30 min to grow the Ni concentrations in Cu, the catalytic activity of the Cu grains as well as clean the surface through evapora- metal surface is much higher than just on Cu, leading tion of potentially damaged surfaces as well as surface to a higher surface growth rate [837]. While growth adsorbates. This process produced films with 100% on elemental Ni and Co resulted in uncontrollable surface coverage, 97.6% transmittance, and µ~1900 to graphene layers, growth on Cu alloys (Ni_Cu, Co–Cu, 2500 cm2 V−1 s−1. This may become useful for back- etc) may lead to controlled larger single crystals of end-of-line (BEOL) integration of graphene. graphene [837]. The presence of metals with higher Another approach is the use of PECVD. However, catalytic activity could also enable growth at lower T. the typical quality of graphene has not been signifi- cantly better than thermal CVD. PECVD synthesis V.1.6. CVD growth on metal foams on Cu foil was reported via simultaneous removal of The integration of graphene into accessible and native Cu oxide and film growth (<420 °C) [852]. This scalable 3d materials is an issue that is inspiring shows the importance of smoothing Cu and the differ- a growing field of research [860]. Free-standing ences in the deposition on top and bottom sides. The interconnected porous graphene materials, called bottom side near the sample holder suffers a confine- graphene foams (GF), attract interest due to their ment that allows preparing smooth Cu metallic sur- ability to transfer many of the unique properties face and growth of mostly SLG continuous films in one of graphene to a larger scale, with high surface step. However the process is not self-limiting and the area  >850 m2 g−1 high electrical conductivity (>10 thickness was time dependent. Contamination of the S cm−1) and good structural integrity, retaining the process tube with Cu affects the lifetime of furnace asform of the metallic template on which they were well as reproducibility. synthesized [861]. Applications include variable geometry elastic conductors [861], lithium ion V.1.5. CVD graphene growth on Ni batteries [862], EMI shielding [863], supercapacitors Growth of graphitic carbon on Ni was achieved [864, 865] and lithium sulphur batteries [866, 867]. asfar back as the 1960s [746] and to the 1970s [753, Structures such as papers [542] and hydrogels 853]. Using PECVD, vertically oriented graphene [343] can be prepared starting from chemically exfo- nanosheets (VOGNs) on Ni can be prepared­ liated graphene and/or GO [304]. These often suffer [854–856]. Carbon on metals is either soluble or it from poor electrical conductivities (e.g. only 5  ×  10−3 reacts to form a carbide. The majority of efforts on S cm−1) [343] and thermal properties, mainly because growth on metals have focussed on the use of non- of defects and their non-continuous nature. Graphene carbide forming metals such as Cu, Ni, Co, Pt, Ru, Ir, may be grown on templates of virtually any shape by and Au. There have been interesting reports on the CVD and continuous structures can be obtained start- growth of graphene on carbide forming as well [1754]. ing from custom porous templates. Synthesis of gra- In this section we focus on the growth of graphene on phene foams by CVD on metal foam templates was Ni and Co, and similar metals. The solubility of C in reported in Ref. [861]. The authors used commer- Ni and Co is fairly high in comparison to that of C in cially-available Ni foam templates to fabricate tun- Cu [857, 858], and it is much higher than that in Cu able, macroscopic structures consisting of high quality [790] ~1 at% (C in Ni) versus 8–10 ppm (C in Cu). The interconnected graphene sheets. The resulting mat­ higher solubility makes it difficult to grow uniform erials were highly conductive, light and flexible with SLG on the metal surface [756, 761]. In both growth on high surface area, and subsequent infiltration with Ni and Co, one can grow produce patchy SLGs. Even by PDMS led to robust composites that exhibited a resist­ decreasing the Ni or Co thickness and growth T, single ance change when subjected to mechanical deforma- layer control is very challenging. This also does not tion, such as bending or stretching, indicating their 95 2D Mater. 7 (2020) 022001 C Backes et al Figure V.13. (a) Optical micrograph of GNRs on SiO2 used for GFET devices. (b) SEM image of an individual GNR ribbon with second layer islands visible as darker spots. (c) IDS versus VGS curves for GFETs (red) as-fabricated and (black) after vacuum annealing. Adapted from [850], with permission from Elsevier. potential in the field of stretchable electronics. This wrinkles are observed in the foam, arising from the work inspired a lot of new research around graphene mismatch in thermal expansion coefficients between foams, for many different applications, such as lithium Ni and graphene [874].The interior is hollow, indi- storage [347], high sensitivity detection of NH3 (ppm cating that Ni etching was successful. XRD measure- in air at RT and NO2)[868] supercapacitors [869], con- ments on foams etched with HNO3 showed residual ductive scaffolds for the proliferation of neural stem Ni (figure V.14(g)), most likely due to the produced cells [870], superhydrophobic coatings [871] com- by the reaction of Ni with HNO3. These gas bubbles bined with Teflon, or stretchable strain sensors [872] could block the infiltration of etchant to some chan- combined with PDMS. nels of the internal structure. To overcome this, etching The growth process depends on the metal catalyst. was instead performed in FeCl3, which has no gaseous In the case of Cu, the process is self-limited, i.e. growth products on reaction with Ni, only soluble salts. These mostly ceases as soon as the Cu surface is fully covered were [873] washed away in hot HCl, which also served with graphene principally because of the negligible C to complete the etching process. XRD measurements solubility and C diffusivity in Cu [781]. In the case of confirmed that the resulting GF had few traces of Ni Ni, the carbon atoms diffuse into the bulk of the metal when FeCl3 was used at RT (figure V.14(g)), and no due to the high carbon solubility, ~0.9% at 900 °C traces of Ni when it was at 80 °C. However, in this case [781], and graphene is formed both by the isothermal there was some residual Fe (figure IV.14(g)). growth on the surface and carbon precipitation from TEM diffraction measurements confirmed the the bulk upon cooling [115]. The shortcoming of gra- polycrystalline graphite nature although the thickness phene growth on Ni is the poor control on N, which is of the walls was too large to observe the individual lay- strongly influenced by the Ni thickness, T and expo- ers of graphene (figures V.14(e)–(f)). The quality of sure time to the hydrocarbon, and the cooling rate, not the sample was further investigated by Raman spectr­ yielding uniform SLG but rather FLG [791] oscopy (figure V.14(h) The small D peak indicates In Ref. [873] GF were grown on commercially- negligible defects. The 2D is consistent with that of available Ni foam templates by CVD in a hot-wall tube FLG. Thinner GF featuring fewer layers of graphene furnace (diameter 5 cm) using CH4.The templates may also be grown by varying parameters such as the were washed by ultrasonication in dilute hydrochloric amount of CH4, the growth time and/or T and the acid, then de-ionised water, and finally acetone, before cooling rate. However, thinner foams are less robust being placed inside the furnace and heated to 1000 °C and require a PMMA support during etching, so that under a H2 flow of 50 sccm. After annealing under these they do not collapse [861]. conditions for 30 min, the flow of H2 was increased and The structures described above are limited by the CH4 introduced (H2:CH4  =  500:50 sccm). Following a commercially-available templates, which typically 10 min deposition time, the sample was removed from have pore sizes in the range of 200–400 µm. In this case, the furnace and allowed to cool in flowing Ar or N2 for most of the volume is occupied by void space rather at least 2 h. To remove the sacrificial Ni templates, the than graphene, which is a disadvantage in applications samples were immersed overnight either in 17 vol% such as energy storage, where a high volumetric energy nitric acid (HNO3) or in 4.5 vol% iron (III) chloride density is required. Thus, a smaller pore size is often (FeCl3) followed by 10% hydrochloric acid (HCl) desirable. heated to 80 °C. Finally, they were thoroughly rinsed High density scaffolding have been produced, as in deionised (DI) water then allowed to dry completely described in [875] as follows: Ni and Cu metal pow- before further handling. ders with typical particle sizes between 0.5 and 150 µm Figure V.14 shows the characterization of the were purchased from Sigma Aldrich (figure V.15(a)). synthesized GF. From SEM observations, the wall In the case of Cu powder below 1 µm, the powder was thickness is estimated to be 10–20 nm, characteristic additionally mixed with MgCO3 powder to prevent the 96 2D Mater. 7 (2020) 022001 C Backes et al Figure V.14. Characterization of GF from Ni foam precursor. SEM images of (a) Ni foam as received, (b) and (c) GF. (d) Photographs of foam at different stages of the synthesis process. (e) TEM image of GF. (f) Electron diffraction pattern from the foam area in (e). (g) XRD patterns of (i) Ni foam, GF etched with (ii) HNO3, (iii) FeCl3 followed by hot HCl and (iv) hot FeCl3 followed by hot HCl. (h) Raman spectrum of GF, adapted from [873], with permission of Springer. agglomeration of the metal particles and compressed. A similar process was described in Ref. [873], in The powder was placed into a quartz combustion ves- which a combination of Ni and NiO NPs compressed sel inserted into a quartz tube in a horizontal tube fur- into a pellet were used as the template. Despite the pres- nace. An annealing step with a flow of 400 sccm argon ence of NiO, the mechanism for the subsequent growth and 100 sccm hydrogen at high T for 45 min was used of graphene did not change, since during the annealing to connect the metal particles (thus forming the metal process in a high-T (1000 °C) H2 atmosphere NiO was scaffold, see figure V.15(b)) as well as to remove metal reduced to Ni, leading to the same behaviour as for Ni oxides and contamination. The minimum T for pre- templates. After growth, the sacrificial template was paring the scaffold was 600 °C for Ni and 800 °C for removed. A long period (several days) of immersion in Cu. Carbon feedstock was provided by a CH4 flow of the etching solution was required before no more col- 10 sccm under a constant Ar flow of 400 sccm and a our change was observed, indicating that all the Ni had total pressure of 50 mbar and 400 mbar for the case of been removed. the growth on Ni and Cu, respectively. These were the Figure V.17 reports the characterization of the GF minimum pressures yielding closed graphene layers produced in this way. The pellet subjected to high pres­ on the metal templates for all the investigated temper­ sure (~440 kg cm−2) was too compact (figure V.17(a)), atures (figure V.15(c)). while the GF synthesized from pellets subjected to low Figure V.16(a) plots a typical Raman spectrum at pressure (~50 kg cm−2) shows a well-controlled par- 532 nm of GFs grown on Cu at 900 °C and Ni at 600 ticle size distribution (figures V.17(b) and (c)), with °C. This gives Pos(2D) ~2695 cm−1 and 2699 cm−1 for a network of interconnected hollow branches of gra- GFs (Cu) and GFs (Ni), respectively, with FWHM(2D) phene. The pores are in the 1–10 µm range, around 2 ~61 cm−1 and 80 cm−1 for Cu and Ni-grown GFs, orders of magnitude less than the GF grown on com- respectively, indicating FLG. I(D)/I(G) ~ 0.16 and mercially-available Ni foam templates. As estimated ~0.25 for Cu- and Ni-grown GFs, indicating defects. from the SEM images, the thickness of graphene is less For SEM analysis the GFs were freeze-dried at low than 10 nm. Indeed, by TEM, 10–30 layers of graphene pressure in liquid N2 for several hours in order to pre- could be observed (figures V.17(e)–(f)) and the poly- vent the collapse of the foam. A SEM picture of a GF is crystalline graphite nature was confirmed as before by in figures V.16(b) and (c) depicts an optical image of diffraction measurements (figure V.17(g)). To analyze two Ni foams fabricated at different T. the quality of the graphene, Raman spectroscopy was 97 2D Mater. 7 (2020) 022001 C Backes et al Figure V.15. Fabrication principle of graphene foams illustrated by SEM images. (a) Metal particles, in this case Ni, are assembled in a vessel to get (b) a 3d structure. (c) At high T (~600 °C) interconnected metal foams are created; this process is implemented in the CVD annealing step. The scale bars are 20 µm. Adapted from [875], © IOP Publishing Ltd. Figure V.16. (a) Raman spectra of GF grown on Cu (red curve) at 900 °C and Ni (black curve) at 600 °C. (b) SEM micrograph of GF grown at 800 °C after metal etch. (c) Optical image of Ni foams fabricated at 800 °C and 1100 °C using a Ni powder. Scale bar in (b) is 20 µm and in (c) 5 mm. Adapted with permission from [873, 875]. employed at 632.8 nm (figure V.17(d)). The sample Ta, Mo and W) were used to grow graphene by CVD at was first fragmented slightly to ensure that measure- ~1050 °C. However, the Raman spectra showed many ments were taken from the inside of the foam. No sig- defects, as exhibited by the presence of a high intensity nificant D peak was visible in the Raman spectrum, D-band. indicating negligible defects and the 2D peak was con- sistent with FLG. V.1.8. Mechanisms of graphene CVD growth on Cu and Ni V.1.7. CVD growth of graphene on metal alloys To get some insights into the atomic mechanisms Selection of the proper alloy could address some of ruling the growth of on metals by CVD, understanding the shortcomings of Cu. Ref. [876] studied the use of the specific CVD kinetics [881] is crucial for advancing Mo-Ni alloys to grow SLG without the precipitation of graphene processing and achieving a better control diffused C into the bulk, because Mo traps C by forming of graphene thickness and properties. Although the a Mo-carbide. In principle, this approach could be thermodynamics of the synthesis system remains the extended to other metal alloys with the objective of same, based on whether the process is performed at finding a substrate with better CTE matching with atmospheric pressure (AP), low pressure (LP) (0.1  −  1 graphene than Cu, and lower vapor pressure at the Torr) or under ultrahigh vacuum (UHV), the kinetics growth T. Another widely studied alloy is Ni–Cu in are different, leading to a variation in uniformity. order to take advantage of the low solubility of C in Graphene synthesis using a Cu catalyst in APCVD Cu, the higher catalytic activity of Ni and the non- processes at higher CH4concentrations revealed that linear behavior of C solubility on Ni concentration the growth is not self-limiting, which is in contrast in Cu1−xNix [857]. Additionally, some of the metal to previous observations for the LPCVD case [1756]. alloys could be much better catalysts than Cu, enabling Hydrogen acts as an inhibitor for CH4 (and for other growth at much lower T [877, 878]. Refs. [877, 878] widely used precursors like C2H4) on Cu and other grew predominantely SLG by an admixture of Au to metals, supressing deposition onto the surface [882]. polycrystalline Ni. This allows a controlled decrease Besides, hydrogen plays a key role in forming form in graphene nucleation density, at low T ~450 °C. The C—H out of plane defects (as efficient nucleation graphene films exhibit an average I(D)/I(G)~0.24 and points) in CVD graphene on Cu, defining an optimal domain sizes  >220 µm2 were reported via the design (and desired) CH4/H2 ratio [882]. of alloy catalysts. Several theoretical studies predicted that carbon Carbide forming metals were also suggested to to atoms are thermodynamically unfavourable on the Cu grow SLG [879, 880]. Commercially available poly- surface under typical experimental conditions, and the crystalline groups IVB-VIB metals (Ti, Zr, Hf, V, Nb, active species for graphene growth should thus mainly 98 2D Mater. 7 (2020) 022001 C Backes et al Figure V.17. (a) SEM image of GF from pellet of Ni/NiO NPs pressed under ~440 kg cm−2. The resulting pellets are too compact, with only small areas of porosity where NPs resist tight packing. (b) and (c) SEM images of the GF from pellet of Ni/NiO NPs pressed under ~50 kg cm−2. (d) Raman spectrum from HD-GF sample. (e) and (f) TEM images of the HD-GF foam. (g) Electron diffraction pattern from the area in (f). Adapted from [873], with permission of Springer. be CHx instead of atomic carbon [883]. Based on this high T annealing in the presence of CH4 and trans- [882, 883] the nucleation behaviour could be under- formed into vacancy-free SLG. Such atomic insights stood (see figures V.18(A) and (B)), explaining many of the structural evolution at different stages of gra- experimental observations and providing a guide to phene grown on Cu(1 1 1) can help better understand improve graphene simple quality [883]. Other stud- the growth mechanism of CVD graphene, and provide ies were conducted combining ideas from step flow design rules for large scale growth of single crystalline growth augmented by detailed first-principles calcul­ graphene films. ations [792] The kinetics of graphene growth on Cu were also A combination of in situ LT-STM experiments and studied using Raman spectroscopy taking advantage DFT-based calculations (see figure V.18(C)), revealed of the mass difference of 12C and 13C [781, 791] where the growth intermediates of CVD graphene on the time dependence of graphene growth on the sur- Cu(1 1 1) via thermal decomposition of CH4at differ- face of Cu was presented as shown for oxygen rich and ent growth stages, comprising various carbon clusters oxygen free Cu surfaces in figure V.19 (i.e. carbon dimers, carbon rectangles, ‘zigzag’ and On Ni another interesting growth mechanism ‘armchair’-like carbon chains) and defective graphene can take place. This arises from the decomposed on- [884]. Saturation of these carbon clusters on Cu(1 1 1) surface C atoms, catalytically produced from the CVD resulted in the transformation into defective graphene, CH4 [888] (or other precursors like C2H4, among oth- with pseudo-ordered corrugations and vacancies due ers) at high T (see figure V.20(A)). The favourable solu- to lattice mismatching. These defects were healed by bility and diffusivity permits the decomposed C atoms 99 2D Mater. 7 (2020) 022001 C Backes et al Figure V.18. (a) Sketch of graphene formation by both direct chemisorption/deposition on Cu (and precipitation/segregation on Ni) [882] (with permission of the PCCP Owner Societies). (b) Computed optimized structures of representative carbon clusters along the graphene growth-by-aggregation on Cu(1 1 1) and (1 0 0) surfaces [883] (copyright American Chemical Society). (c) Evolution of graphene grown via thermal decomposition of methane on Cu(1 1 1). Top: Schematic illustration of the surface intermediates at different reaction conditions. Bottom: overview of STM images showing (C.a) carbon clusters, (C.b) the boundary between carbon cluster (left) and defective graphene (right), (C.c) defective graphene, and (C.d) vacancy-free graphene, at different reaction stages [884] (copyright American Chemical Society). to diffuse into the Ni bulk during the annealing stage large single-crystal grains on Pt is nondestructive after at high T [889]. Then, the inward carbon atoms pre- electrochemical delamination of the grown graphene cipitate (segregate) on the catalyst during the cooling [108]. The growth of can be designed to control the period due to supersaturation of the diluted carbon graphene edges that range from zigzag to armchair into Ni. This supersaturation-by-cooling leads to the orientations via growth-etching-regrowth by CVD segregation of carbon atoms on the cooled Ni surface, on Pt [782]. Ref. [890] reported the nucleation which will promote the formation of the graphene and growth of graphene islands on Pt(1 1 1) at T ~ domains. It is worth mentioning that this mechanism 1000 K by CVD of ethylene. Using a CVD method has been validated by a complex theoretical computer followed by a repeated etching-regrowth process on modelling protocol accounting for a large variety fac- Pt Ref. [891], single-crystal graphene domains were tors affecting the CVD graphene synthesis (solubility reported up to 13 000 cm2 V−1 s−1 under ambient in Ni, growth time, growth T, cooling rate, Ni thin film conditions. Ref. [892] engineered the grain size of thickness, among others) [886], and will naturally lead SLG grown on Pt with a limitted carbon solubility, to the formation of MLG easier than in other metal by a segregation–adsorption CVD yielding a grain substrates by the nature of the segregation mechanism. size ranging from ~200 nm to ~1 µm, which enabled Another growth mechanism of graphene on Ni(1 1 1) the determination of the scaling laws of thermal was suggested in Ref. [887] at T  <  460 °C. A surface- and electrical conductivities as a function of grain confined nickel-carbide phase coexists with SLG (see size. They concluded that the thermal conductivity figure V.20(B)). This grows by in-plane transforma- of graphene decreases with grain size, while the tion of the carbide along a phase-boundary, unlike electrical conductivity slowly decreases with a other growth processes on other transition metals small grain boundary transport gap ~0.01 eV and and on Ni above 460 °C, where carbon atoms attach to resistivity ~0.3 kΩ µm. CVD-graphene growth on Pt ‘free’ edges of graphene islands [887]. substrates can provide an extra added-value, as it may exhibit catalytic properties towards H2 dissociation V.1.9. Graphene growth on noble metals [893]. Pt is one of most commonly used materials for Other low-interacting substrates have been explored. electrodes due to its thermal stability and chemical Ref. [108] reported the growth of graphene on Pt. This resistance. hand high crystal quality with lowest wrinkle height The thermal expansion coefficient difference of ~0.8 nm and µ  >  7100 cm2 V−1 s−1 under ambient Pt with graphene is smaller than for other transition conditions. The repeatable growth of graphene with metals [894]it is possible to tune the edge termination 100 2D Mater. 7 (2020) 022001 C Backes et al Figure V.19. Effect of O on graphene growth kinetics. SEM images of graphene domains grown on (A) OF-Cu and (B) OF-Cu (O). Isotope-labeled Raman maps of I(2D) on Si for growth at (C) and (D) 1035 °C and (E) and (F) 885 °C. The isotope switching intervals are indicated below each image. (G) Logarithmic plots of graphene domain growth rate dr/dt versus 1/T. The error bars are from calculations of different domains for each case, and the activation energy Ea is extracted from the slope of the linear fit. (J) and (H) Atomic-scale schematics of graphene edge growth on Cu with and without the assistance of O, respectively. (I) DFT calculated energies of different configurations of H attachment, in reference to H in H2. The energy spread of H at the graphene edge is due to the computational uncertainty resulting from the lattice mismatch between graphene and Cu. Adapted from [791] with permission from AAAS. of the crystals via growth–etching–regrowth [108, sisting of heptagon–pentagon pairs of carbon atom 782, 793]. rings, related to small-angle in-plane tilt boundaries. Efforts have been made not only in the growth of Using Ir(1 1 1) as substrate at 750 °C–900 °C, growing millimetre-sized hexagonal single-crystal graphene rotated islands are more faceted than islands aligned films on Pt but also in developping transfer method- with the substrate [895]. Further, the growth velocity ologies. A bubbling method, was used to transfer SLG of rotated islands depends not only on the C adatom grown on Pt to arbitrary substrates, using a nonde- supersaturation but also on the geometry of the island structive process not only to graphene, but also to the edge. Different growth mechanisms were proposed to expensive Pt substrates [108]. There are also studies explain the differences in the graphene edge binding focused on the growth on other noble metals, such as strength to the substrate. Graphene growth by CVD Ir [895], Ru [896] and Rh [896]. Moiré patterns have of ethylene on a Ru(0 0 0 1) surface was also explored. been explored on graphene on Rh(1 1 1) using in situ At low pressures and high T the metal surface facets STM [897]. The synthesis of large-scale uniform gra- into large terraces, leading to ordered graphene layers phene films on high carbon solubility substrates of Rh [900]. Engineering of Ru surfaces with Ar+ sputtering foils at ambient-pressure CVD was reported [898]. to produce subsurface bubbles led to a tailored growth The thickness of graphene can be tuned from multi- of graphene on Ru(0 0 0 1) and to anysitropic oxygen layer to monolayer by increasing the cooling rate in interacalation [896]. the growth process graphene prefers to stack deviat- ing from the Bernal geometry, with the formation V.1.10. CVD growth on Ir (1 1 1) of moiré patterns. Ref. [899] grew SLG on Ir(1 1 1) The substrate Ir(1 1 1)/YSZ/Si(1 1 1) in principle allows by LP CVD. The graphene layer had a very low den- integration with Si technology. However, the growth sity of defects. These are edge dislocation cores con- is ruled by the outermost Ir atoms, and therefore the 101 2D Mater. 7 (2020) 022001 C Backes et al Figure V.20. (A) sketch of graphene CVD growth on a Ni surface [885, 886] through the carbon diffusion and cooling processes. (B) (Top-left) the hexagonal structure of the graphene domain can be identified in the STM image, while the right side shows the quasi-square structure of the carbide Ni2C. The domain boundary between the two phases is in the same atomic plane and without any defect pattern. This indicates that the two phases form a line interface. (Top-right) the carbide and graphene lattices are superimposed in the region indicated by the rectangle in the top-left panel. (Bottom) Top and side views of the optimized atomic structure of the graphene/Ni2C interface [887] (copyright American Chemical Society). process may be generalized to many other transition V.1.11. MBE growth on Pt(1 1 1) metals like Ir [895], Ru [896] or Rh [896]. Several physical evaporation methods employing In the case of Ir and Pt, the interaction with the sublimation from a carbon source have been used, like substrate is rather low, and this presents pros and cons. electron beam evaporator [906, 907], or a resistively- As an advantage graphene can be transferred to other heated piece of graphite [790, 908]. The growth of substrates. However, the main disadvantage is the nanometer-sized SLG on hBNby MBE was reported small domain size, as many rotational domains can be [909]. generated [901, 902]. Work is underway in the search Ref. [910] used a solid carbon source consisting of of optimization methods for enlarging domain sizes, a glassy carbon filament connected by two refractory even to millimeter scales. Ru and Rh are substrates metal bars to a DC current source. Thanks to a 10 times were the graphene-substrate interaction is very strong higher resistivity of glassy carbon versus graphite, it and it is possible to grow large grains on single-crystal is possible to use moderate electrical current ~14–18 substrates, however transfer is difficult [1757, 1758]. A [911]. This makes this method simple and easy to SLG on Ir(1 1 1) thin films on YSZ-buffered install in a UHV system, significantly reducing con- Si(1 1 1) wafers [903] was reported [905]. The sub- tamination. Ref. [910] showed that for MBE growth strate was first cleaned by Ar sputtering (0.8 keV, 5 µA, of graphene, the substrate can be at lower T than in 15 min) and annealing to 900 °C followed by an oxy- other methods [912, 915, 916]. This enables growth on gen treatment (p(O2)  =  5  ×  10−8 to 1  ×  10−7 mbar, T di­electric substrates in an inherent transfer free pro- between 400 °C and 550 °C) and a final flash annealing cess. Also, in situ surface science analysis techniques (to desorb the oxygen) up to 900 °C. (LEED, RHEED, STM, XPS) can be used in a highly SLG was grown using ethylene as a precursor CVD controllable growth environment (growth rate, sub- at T  =  850 °C, starting from very low ethylene par- strate type, substrate temperature, in situ doping). tial pressures (ideally  <1  ×  10−8 mbar for the first The evaporation growth rates are generally very as 20–30 min), then progressively increasing the pressure low as 3  ×  10−4 ML s−1 to avoid a rapid degeneration up to 2  ×  10−7 mbar. This approach prevents the for- of the glassy carbon filament. This requires a sample- mation of multiple nucleation centers, which would to-source distance of tens of millimeters. At this dis- result in the appearance of many rotational domains tance the sample is radiatively heated. (mosaicity). The substrate should first be heated to the An evaporation carbon source can be fabricated setpoint T and only then should ethylene be admitted [913], using a resistively heated piece of glassy carbon into the chamber[905]. Surface processes such as oxi- [914], [915]. dation and overlayer (including graphene) formation Using a 0.3 mm thick glassy carbon plate cut in modify the work function of the metal substrate, the the desired shape (see figure V.21) by laser or water-jet photoelectron yield measured while irradiating the cutting, one gets the carbon source as shown in fig- sample with a pulsed UV source can be used to track ure V.21(b). The glassy carbon filament is mounted the onset of the surface process under study and to on a standard UHV feedthrough, terminated with Ta monitor its evolution until saturation is reached [905]. rods, and fixed with Mo screws. 102 2D Mater. 7 (2020) 022001 C Backes et al Figure V.21. A glassy carbon filament (a) is mounted on an UHV electrical feedthough (b) and resistively heated (c). T is simultaneously measured with an optical pyrometer and a type-C thermocouple located on the back side. The C Source is placed in a deposition UHV sys- of the surface with a surface morphology similar to tem, base pressure 1  ×  10−10 mbar heated using a DC growth from decomposition of hydrocarbons or aro- current. The T calibration of the cell is performed in matic molecules. a separate UHV system with a base pressure 1  ×  10−8 The MBE growth mechanism on Pt(1 1 1) by the mbar. The T in the hottest spot of the filament is meas- evaporation of C atoms can be understood on the basis ured using a dual-color optical pyrometer through an that the evaporated C atoms will not interact with each optical window located in front of the cell. Evapora- other or with gases until they reach the surface. This tion of carbon starts above 2000 °C. Using a DC cur­ guarantees a continuum flux of atomic C on the sur- rent ~14 A (~100 W) the filament is at 2014 °C. A face, heated up to the desired T. The synergistic effect clear correlation of the increase of the partial pressure of surface T with the kinetic energy of the C atoms of atomic mass unit number 12 (a.m.u.) is observed reaching the surface catalyses the initial graphene in a residual gas analyzer (quadrupole) with source nucleation by atomic on-surface aggregation: C atoms temper­ature. The presence of amu 12 can be associated reaching the surface are free to move until finding cor- to the evaporation of monoatomic carbon. The depo- rect (stable) positions on the surface crystal lattice to sition rate is 3  ×  10−4 SLG s−1. bond and the epitaxial graphene growth undergoes by By exposing a clean Pt(1 1 1) substrate to the car- aggregation of on-surface diffusing C atoms or nucle- bon source the formation of carbon structures is ated nanostructures (see figure V.23). observed [910]. First, the glassy carbon filament his T-assisted on-surface diffusion of C nucleation cent- held at T  >  2000 °C for degassing for 15 min. The ers, able to collide with others, favours graphene forma- pressure is kept  <1  ×  10−8 mbar. Before growth, the tion, which will epitaxially grow in their most stable gra- Pt single crystal is cleaned by cycles of Ar+ sputter- phene domains. The CI-NEB approach [916, 917] yields ing and annealing at 850 °C. The cycles are repeated a minimum diffusion barrier of a C atom on Pt(1 1 1) until a sharp LEED pattern is observed, where only Pt of 0.45 eV, and a C  +  C  →  C2 on-surface aggrega- peaks appears and/or STM images showing clean and tion reaction barrier  <0.8 eV (with a net enthalpy gain flat extended regions. Then, one places the Pt (1 1 1) ~3 eV), to be compared with barriers of 0.6 and 1.1 eV, substrate at ~650 °C in front of the carbon source at respectively, reported for the Cu(1 1 1) surface [918], 2000 °C with a distance between them of 20 mm. After or the 0.58 eV for the atomic O diffusion barrier on 30 min 0.5 ML SLG is attained. During the process the Pt(1 1 1) [919]. These energy barrier values are easy to pressure is ~7  ×  10−9 mbar. The procedure works overcome in seconds at the typical surface temperatures equally well if the sample is annealed after growth. in the MBE process on Pt(1 1 1). In addition, the gra- The graphene formation is demonstrated by in situ phene epitaxial growth-by-aggregation will be favoured LEED and STM. The LEED diagram in figure V.22(a) as the nanopatches increase their size, which will be able shows the spots corresponding to the Pt(1 1 1) plus a to stabilize the incorporation of new diffusing C atoms surrounding ring characteristic of graphene. Brighter and larger nucleated C nanostructures. spots along the ring are indicative of the preferential orientations of the carbon periodicity with respect V.1.12. MBE growth on Au to the Pt crystal orientation. Figure V.22(b) shows The use of Au as a substrate is difficult, since the T an STM image measured on the graphene areas. The needed to obtain graphene is too close to Au melting bigger periodicity (bright bumps) of the image cor- point [906, 907]. A solid-carbon source can be used to responds to one of the possible moirés. The smaller decrease the T. periodicity (diagonal lines) is the atomic corrugation The Au (1 1 1) substrate is kept at 550 °C, with a of the graphene. Graphene covers extended regions well degassed carbon filament. This step can last for 103 2D Mater. 7 (2020) 022001 C Backes et al Figure V.22. (a) LEED pattern on SLG/Pt(1 1 1). (b) STM image showing a moiré and the atomic corrugation of graphene. Adapted from [910], with permission from Elsevier. days because the freshly new filaments are porous con- reachable by SuMBE, does not lead to immediate C60 taining many impurities. The degas pressure needs to cage rupture. be  <5  ×  10−9 mbar when the glassy carbon filament is The ultimate evidence of the presence of defected at ~1600 °C (12 A, 83 W).If this step is not performed nanometric graphene islands comes both STM and correctly, the samples get covered by an amorphous Raman spectroscopy [923]. This process is expected to layer and the substrate has to be cleaned again. be self-limiting and to stop completely as soon as the In case of graphene growth by supersonic molecu- Cu surface is entirely covered by a monolayer of carbon lar beam epitaxy (SuMBE) from C60 [920, 921], the atoms [923]. two major issues to take into account are the substrate Ref. [923] reported a coating of graphene-like type and kinetic energy (KE). The main steps are the material at 645 °C using H2 as carrier gas. These following: nanoislands also contain pentagons, which come from the original buckyball structures. Thus, at this stage • Surface preparation. Numerous (up to 40) Ar this is just a proof of principle approach. The technique ion sputtering (0.5 keV) and annealing cycles of might be applied to a wide range of substrates, such as the Cu substrate (T  >  700 °C for optimal LEED semiconductors and insulators, avoiding transfer. diffraction pattern) to avoid the presence of unwanted contaminants, such as oxygen, sulphur V.1.13. UHV-thermal decomposition on Cu foil, or remaining carbons, during growth. Cu(1 1 1), Pt(1 1 1) and Ir(1 1 1) • Carrier gas choice. He or H2The use of noble gases This method consists in the growth of graphene in a prevents strong interactions between carrier gas UHV environment, using surface decomposition of and the substrate and does not affect C60 internal C60 [924] as a carbon source. The same method applies dynamics, which is frozen unlike ordinary heating to Pt(1 1 1) and Ir (1 1 1), Cu(1 1 1)single crystals [901, that increases the molecular vibrations. 912, 925] and Cu foil [926]. • Aerodynamical acceleration of the highly The combination of a UHV environment together diluted (less than 1% of the mixture) C60 beam with the use of C60 molecules provides some advan- by isentropic expansion of the flux out of the tages with respect to conventional CVD [927, 928]. On injection cell into vacuum through a nozzle. one hand, this procedure is self-limited, which assures By changing the carrier gas and the seeding the growth of SLG. On the other hand, the controlled parameters (source T, gas inlet pressure), fullerene and clean environment (UHV chamber with a base KE can be tuned, being inversely proportional pressure ~10−10 mbar) results in samples almost free to the carrier gas mass. KE ~ 10–15 eV using He of impurities with a lower growth T than CVD. carrier gas up to 30–40 eV using H2. Ref. [926] used commercial C60 (Sigma, 98% • Collimation of the expanded flux and impact of the purity). To control the evaporation rate, a molecular organic molecules on the Cu reconstructed surface, evaporator is needed, that can be either commercial keeping the substrate at RT. or homemade, as in this case. Figure V.24 shows a • Thermal activated growth of graphene islands by homemade one with a Ta crucible on which a type- increasing T to 645 °C. K thermocouple is spot-welded. The molecules are placed inside the crucible. Acurrent is passed through Refs. [922, 923] showed that substrate T must be the Cu rods and the molecules are evaporated at 450 raised to 645 °C to synthesize graphene islands as C60 °C–500 °C. Molecules are outgassed in UHV condi- high-energy deposition on Cu, even at the highest KE tions prior to growth. To this aim, C60 is heated 10 °C 104 2D Mater. 7 (2020) 022001 C Backes et al Figure V.23. Sketch of the coexisting processes leading to the growth of graphene on a Pt(1 1 1) surface by MBE. Figure V.24. Homemade C60 evaporator. above the growth T, until the initial pressure is The simplest and more efficient way of improving the restored (~10−10 mbar) after elimination of impu- thermal contact is the use of thicker Cu substrates. rities (water, CO, CO2). Usually the crucible is spot In situ surface techniques such as LEED, AES or welded to two pieces of steel that are clamped to Cu STM can provide valuable information with no need rods. It is important to have a good electrical contact to expose the sample to the air, thus avoiding surface between Ta crucible and rods, otherwise the whole contamination. Figure V.26 is a LEED pattern of a evaporator will be heated and the result will be a rise Cu foil after cleaning, where multiple spots, corre­ in the base pressure of the chamber and, therefore, a sponding to different grain orientations, are observed. dirty evaporation. Once the substrate is clean and the molecules Cu, used as a substrate, is cleaned with several sput- purified, the growth consists in exposing the sub- tering and annealing cycles. A standard prep­aration strate to the molecules at the chosen growth T. Typi- consists in five cycles of 10 min Ar sputtering and cal parameters are: evaporation T ~ 450 °C, substrate 10 min annealing by electron bombardment at 800 size ~15  ×  20 mm2, 10 cm evaporator/sample distance °C. For the first cycle the sample current is 10 µA. For and growth time ~90 min. Coverage can be tuned by the last cycle, the sample current is reduced to 5 µA to changing the exposure time. An important step is to prevent large surface roughness. With this sample cur­ have initially the substrate at the growth T before the rent value a surface roughness of 0.4 nm is obtained. C60 molecules start to sublimate, avoiding any C60 T  >  800 °C may promote the diffusion of impurities arriving to the sample surface when this is still cold. from the bulk to the surface as well as an excessive Cu Different growth trials have been made by first depos- sublimation, leading to higher roughness. iting the molecules on the cold surface and making a Figure V.25(a) shows a Cu foil mounted on a typi- post-annealing, but this procedure shows less satis- cal sample holder by using two lateral Ta wires to sup- factory results. Once the evaporation is finished, it is port it. Other fixing procedures, like the use of glue or important to keep the sample hot until the evaporator carbon tape, are not possible in UHV. The way the sam- gets cold. ple is mounted is important because the sample might This procedure provides polycrystalline SLG on bend during annealing if the thermal contact with the Cu. The next step involves sample characterization. sample holder is not homogenous (figure V.25(b)). Figures V.27(a) and (b) exhibit LEED patterns taken at 105 2D Mater. 7 (2020) 022001 C Backes et al Figure V.25. (a) Mounting of a Cu foil in a Ta sample holder. (b) Bending of the 25 µm Cu foil produced by Cu thermal expansion. Figure V.26. LEED pattern at 107 eV of a Cu foil after cleaning and prior to the C60 evaporation. of 50 and 56 eV with a typical ring of a polycrystalline of the typical moiré patterns that depend on the sub- graphene surface. AFM images (figures V.27(c) and strate can be seen. These appear due to the mismatch (d)) show very large and flat Cu terraces (hundreds of the graphene and metal network, meaning that only of nanometers) free of contaminants where graphene SLG has grown. This leads to a clean graphene not only wrinkles cross all over the area, assuring a complete because it takes place in a UHV environment but also coverage. Figure V.27(e) shows a Raman spectrum at because the precursor contains only C atoms. 532 nm, after background subtraction, taken in the The same method works for other metals like Pt or area delimited in the optical image that appears in the Ir. Table V.1 shows the differences between the growths inset. The spectrum exhibits a symmetric and nar- on the various substrates. row 2D peak located at ~2703 cm−1 with FWHM(2D) There is still a lack of a precise atomic mech­ ~28 cm−1, compatible with SLG [105], with a signifi- anism rationalizing the thermally-assisted on-surface cant D peak that could be due to grain boundaries and decomposition of fullerenes on metals, whose founda- defects. tions still remain unclear. The process needs the syn- As the solubility of carbon in Cu decreases with ergistic effect of surface T and availability of enhanced decreasing T with this method the solubility is reduced reactivity sites on the metal surface: such as atomic four times with respect to CVD [790]. For this rea- metal vacancies, kink/add surface metal atoms or son, with this process mainly SLG is growth, without metal surface steps. The combination of these factors contrib­ution of BLG. favours the anchoring of the C60 molecules to the sur- The growth of graphene on single crystals by ther- face strongly enough to disrupt their stable geometri- mal decomposition of C60 does not differ much from cal arrangement (see figure V.29). A strong interaction the polycrystalline ones. All surfaces must be clean, of C60 with the sites with enhanced-reactivity could several sputtering and annealing cycles are mandatory, lead to an efficient weakening of the C–C bonds within otherwise defective samples are obtained. Figure V.28 the cage-like configuration contacting with the sur- shows three STM images of graphene grown by on face, in particular, of the C–C bridges participating in Cu(1 1 1) and Pt(1 1 1), Ir(1 1 1) for comparison. Some pentagon-rings of the C60 molecules, which store less 106 2D Mater. 7 (2020) 022001 C Backes et al Figure V.27. LEED patterns with a beam energy of (a) 50 eV and (b) 56 eV, where the ring of polycrystalline graphene is observed. (c) and (d) AFM topography images where Cu flat terraces are observed. Wrinkles crossing the terraces (pointed by black arrows). (e) Raman spectrum at 532 nm. The inset shows an optical image with Cu grains visible. Figure V.28. STM images of SLG grown from C60 on: (a) Pt(1 1 1), (6  ×  6)nm, 0.11 V, 2.3 nA. (b) Ir(1 1 1), (15  ×  15)nm, 1.2 V, 2.6 pA and (c) Cu(1 1 1), (27  ×  27 nm), −0.4 V, 2.7 pA. In every image different moiré superstructures are shown. energy than those connecting hexagon-rings. The pro- fuse on the metal surface by effect of the T.The epitax- cess will start with the molecular cage rupture toward ial graphene growth will undergo by aggregation of the the on-surface disaggregation of the molecules into C mentioned on-surface diffusing C atoms or C-based atoms or other flatter C-based structures, free to dif- nanostructures (see figure V.29), which will lead to the 107 2D Mater. 7 (2020) 022001 C Backes et al formation of the most stable graphene domains on the Table V.1. T and typical evaporation times needed to grow SLG from thermal decomposition of C60 for different substrates (at a metal substrates. sample-evaporator distance ~10 cm). V.2. Graphene growth on semiconductors T sample during Time to Working conditions in CMOS fabs are extremely Substrate evaporation (°C) have SLG restrictive in particular regarding metal contamination Cu foil 800 90′ [1718] Transfer technologies are being improved to Cu(1 1 1) 900 30′ provide a very clean environmentDirect growth on Pt(1 1 1) 850 30′ semiconductors would be the ideal alternative. Ir(1 1 1) 900 30′ V.2.1. Graphene growth on Ge It would be desirable to grow graphene on high quality the large area graphene film to the desired substrate single crystals. Graphene on SiC is of high quality and for applications that normally induces contamination, may satisfy some applications. However, these are wrinkles or even breakage of graphene. In this limited to the use of SiC as a substrate. Other than on scenario, the game changing breakthrough would SiC, most of the high-quality isolated graphene has be the development of processes to deposit directly been grown on metal surfaces because of their high high quality graphene films and single crystals on catalytic activity and limited C solubility. While as a arbitrary substrates avoiding the transference step single crystal is not widely used in the semiconductor and, if possible, at lower T than catalytic approaches. industry, it is used as an alloy with Si (SiGe) in the The first route involves the direct CVD synthesis on sources and drain of advanced transistors and it could insulators or dielectrics at T  >  1000 °C with examples also be deposited selectively onto Si. With respect to of synthesis surpassing 1500 °C [943]. Along the its ability to be a good graphene substrate, Ge like high T, atmospheric pressure and CH4 as precursor Cu has an extremely low C solubility and it does not diluted typically in Ar and/or H2 are the most common form a carbide, [930] thus potentially making it a parameters. Synthesis of SLG and FLG films was good candidate for graphene growth. Ref. [931] took on reported on Si3N4 [938, 939], sapphire (Al2O3) advantage of this and found CVD growth conditions [940–943] and silicon oxides in different structures for SLG. Additional work was performed on the [944–946] with final Rs between 5 kΩ [944] and 800 growth of graphene using other techniques such as Ω [945]. FLG grains were also deposited on h-BN APCVD, [932] and MBE, [933] on Ge(1 1 0), Ge(1 0 0) flakes at 1000 °C in atmospheric pressure CVD [947] and Ge(1 1 0)/Si(1 1 0) and Ge(1 0 0)/Si(1 0 0). and on MgO nanocrystal powder by LP CVD at lower Although Ge is more ‘compatible’ in Si device flows, T between 325 and 875 °C [948]. Other substrates there are still challenges: transfer and crystallinity. include SiC [611] and SrTiO3 [949] with µ ~1800 These could be overcome by developing growth and 1050 cm2·V−1·s−1. Even though, the typical selective techniques on isolated Ge/Si post transistor grain size is smaller (between 0.5 and 10 µm) and fabrication. Ref. [934, 935] reported the CVD growth the amount of defects higher in comparison with the of graphene on monocrystalline Ge(1 0 0) on Si(1 0 0), graphene synthesized with the catalytic approach, the the preferred orientation of Si for CMOS technology. final properties of the deposited graphene resemble The growth took place between 900 and 930 °C with transferred material grown on catalytic metals [945]. a ramp-up rate of 1–20 °C min−1 and CH4 flow in the (0.4–5 sccm) range. The optimal roughness of V.3.2. SiO2 the substrate was achieved when the step growth was SiO2 whether in bulk form as transparent quartz or preceded by annealing at 750 °C–800 °C in pure H2 to fused silica or on thin films grown on Si wafers (SiO2/ reduce native oxides in situ [936, 937]. Ref. [934, 935] Si), are the most frequently used insulators to directly also grew SLG on Ge(1 0 0)/Si(1 0 0) under laminar growth graphene. In the case of the transparent conditions with a Aixtron VP508 horizontal CVD substrates, the typical features of these films include hot wall reactor. The SLG growth was perfromed at transparency from 80% to 95%, nearly maintaining 890 °C–910 °C with a ramp-up rate ~1 –10 °C min−1 the transmittance of the substrate itself in some cases, and a pressure ~850–990 mbar. CH4 at flow rate ~0.1– and tunable Rs depending on process parameters 0.5 sccm was used in an Ar flow ~3 l min−1. The step [950–954] with values ranging from 1 kΩ sq−1 [954] growth was preceded by substrate annealing at 800 to few kΩ sq−1 [952]. Those transparent materials have °C–850 °C in H2. demonstrated potential for a broad range of daily life applications as heating elements in smart windows, V.3. Graphene growth on insulators high performance defoggers with fast response (one V.3.1. High temperature CVD growth minute) [951], anti-icing glasses or as mechanical The production of large area polycrystalline graphene strain or rupture sensors (demonstrated GF ~600 and graphene single crystals has been demonstrated via [955]), to name a few. The main limitation to transfer chemical vapor deposition (CVD) on metal catalyst. this knowledge to the industrial environment is the Currently, the major drawback is the transference of high growth T. 108 2D Mater. 7 (2020) 022001 C Backes et al Figure V.29. Pictorial sketch corresponding to the thermal PVD decomposition of C60 on the mentioned Cu foils, and Cu, Pt and Ir single-crystal surfaces, as well as the coexisting subsequent processes leading to the growth of graphene. instrumentation OPTIC FIBER GAS DELIVERY ECR MAGNET GAS DELIVERY MW PLASMA SYSTEM SYSTEM FLOW POWER SOURCE CONTROLLER DISCHARGE H2 OPTIC FIBER MW CHAMBER (2.45GHz) C2H2 H2 e- H e- ECR ECR DISCHARGE - MAGNET C2HY H+ H + MAGNET REACTION C2H2 PLASMA H+ CHAMBER e- CHAMBER e- e- e- Ion + Ion + BIAS + + e- C 2H 2 CH+ - - BIAS REACTION CHAMBER (electrode) H C 2H2 H2 C2HY THERMAL `PUMPING SUBSTRATE SETUP SYSTEM PUMPING CONTROLLERS: SYSTEM BIAS, THERMAL Figure V.30. Schematic illustration (left) and picture (right) of a r-(ECR-CVD) Plasma system. This consists of a gas delivery system with flow controllers, an ASTEX AX 2000 microwave power source with optic fiber coupling the power to the discharge plasma chamber with an ECR magnet surrounding it. The activated gases are transported by forced convection, with two electrodes attracting electrons and ions. The substrate is independently heated. A two stage pumping system generates the vacuum. Adapted from [816], © IOP Publishing Ltd. Given that Si based electronic technology is the than catalytic CVD. In order to manage the stability established standard, it would be also extremely inter- issue, PECVD has been considered as an alternative esting to grow graphene on SiO2/Si wafers so as to inte- approach to direct deposit graphene at lower T [962]. grate graphene in Si-based devices. Many approaches have been developed in this area, but the temperature V.3.3. PE CVD on SiOx and Si/SiO2 required, typically over 1100 °C [944, 945, 951, 954, PECVD enables catalyst free direct growth by the 956–958], poses some problems derived from the poor activation of molecules in gas phase and reduces the thermal stability of SiO2. Among them, desorption of high T in conventional CVD or in attempts to use SiOx from the exposed surface [956] and diffusion of pyrolysis [938, 939, 941, 942, 944–948]. The most carbon or silicon species from the interface) [951] are important drawback is the resulting small grain size. well known phenomena that play an undesirable role PECVD has been reported on non-metallic sub- during synthesis, affecting the structure and quality of strates, resulting in films with small grain sizes -from the deposited graphene. These effects are more pro- 2 to 30 nm-, and amorphous carbon [816, 960, 961]. nounced in thin oxide layers (less than 90 nm) [951] Ref. [774] reported an enlargement of grain size by a that, in turn, are an important issue in CMOS tech- ‘two-step’ growth strategy in remote electron cyclo- nologies [959]. Confinement of the substrate was sug- tron resonance plasma assisted chemical vapor deposi- gested to eliminate desorption [956]. Even so, the final tion, r-(ECR-CVD), figure V.30. The entire protocol is µ ranged from 410 to 760 cm2·V−1·s−1much lower shown in figure V.31. 109 2D Mater. 7 (2020) 022001 C Backes et al Figure V.31. ‘Two step’ synthesis protocol schematics. The various parameters (time (t), T and partial V.3.4. PECVD on quartz substrates pressures of C2H2 and H2, PC2H2 and PH2) were tuned Following the above protocol graphene was deposited ‘in situ’ in the nucleation (t1, T1, P1C2H2/P1H2) of on quartz as depicted in figure V.32. AFM images of high quality graphitic seeds and growth from the seeds lateral force contrast are shown in figures V.32(a)– edges (t2, T2, P2C2H2/P2H2) stages. A post-growth (c). Raman analysis indicates the formation of annealing is applied in UHV, improving the final prop- predominantly highly defective SLG at 650 °C, erties of the film. The detailed procedure is usually as figure V.32(d). follows: Figure V.33 presents an AFM characterization of continuous film, confirmed by four point probe meas- • Evacuation down to 10−5 mbar, introduction urements, with a RS ~ 3.4 kΩ·sq−1. The coalescence of molecular H2 at a P1H2  =  10−2 mbar (typical flow graphene grains follows two different trends (figures of H2  =  55 sccm) and increase T up to T1  =  500 V.33(a) and (b)). Hexagonal graphene domains of °C–700 °C in 40–60 min (t0, figure V.31) for similar orientation often generate a smooth lateral stabilization. Subsequently, annealing the substrate merging, presenting boundaries without linear defects for 5 min. In some cases is interesting to bias the (figure V.33(b), black arrows) [611, 931, 962–965]. On substrate with a positive potential (0 V–50 V) if the the other hand, rough lateral merging also takes place auto polarization due to the plasma is not positive. at many points, where linear defects occur ­(figure • Step 1 (NUCLEATION): Introduction of V.33(d), pink line). On these defects, carbon species C2H2, at a P1C2H2  =  10−4 mbar, typical accumulate as in vertical graphene deposition [966] flow  =  0.25:0.20 sccm and turn on the plasma that indicates that the process is not completely self- (100 W) during 5 min (t1, figure V.31). limiting. From UV–vis-NIR spectrophotometry • Step 2 (GROWTH): Modify T2 in a range of 30 transmittance (figure V.33(c)), it is also clear that there °C–100 °C or P2C2H2/P2H2 in the 0.1–0.9 · 10−4 is an influence of this accumulation (values ~95%). mbar range (between 10%–20%), maintaining it After growth, the sample can be annealed in UHV for few hours (t2, figure V.31), until the substrate to improve the final properties. Ref. [774] increased T is completely covered. As a general rule, P2C2H2 is stepwise up to 600 °C. The height and density of the modified, due to its minimal relevance in the final linear defects is lower after annealing. RS decreased pressure. from ~3.4 kΩ·sq−1 to ~900 Ω·sq−1, with transmit- • Turn the plasma and bias off and cut the flow of tance  >  92%. The protocol is scalable it is likely accel- precursor (P2C2H2), cooling down in hydrogen erated by changing plasma power and pressure. atmosphere (P2H2) during 40 min (t3, figure V.31) The r-(ECR-CVD) mode activated by microwave preventing oxidation. Post growth annealing in (MW) source provides higher efficiency in the disso- UHV at 650 °C, to desorb some chemical species ciation of gases than other plasma and it is especially attached to grain boundaries and to improve the efficient in the dissociation of H2. It is one order of coalescence among the grains. magnitude more efficient than other DC (continu- 110 2D Mater. 7 (2020) 022001 C Backes et al a) b) c) 100nm 2D D d) e) C1s f) G D G 2D Intensity (a.u.) Intensity (a.u.) 1000 2000 3000 290 288 286 284 282 280 278 1000 2000 3000 Raman shifts (cm-1) Binding Energy (eV) Raman shifts (cm-1) Figure V.32. Graphene on quartz at H2/C2H2  =  55/0.25:0.20 (sccm), PT  =  5.4  ×  10−2 mbar, P  =  100 W for 5 min nucleation time, 9 h growth. AFM lateral force image of the sample grown at (a) T  =  650 °C and, (b) high magnification lateral force image of a single flake from sample and below the corresponding atomic lattice periodicity of the flake along with the Fast-Fourier transform in the inset of the periodicity image in (a)–(c) AFM lateral force image of the sample grown at T  =  700°. (d) and (f) Individual Raman spectra acquired at grain positions, for samples in (a) and (c), respectively. (e) High resolution XPS C1s core level spectrum (black dots) of the sample grown at 650 °C. The C-sp2 graphitic component (blue, 284.6 eV), C–H (green, 285.6 eV) likely from H-terminated edges, and C–O (cyan, 286.5 eV) species. The component at (magenta, 283 eV) can be ascribed to remainders of the growth process (e.g. C2H4Si at 282.5 eV) or to the interaction with substrate (e.g. SiC C(1s) is 282.5–283.5 eV). Adapted from [774], © IOP Publishing Ltd. ous polarization) and RF (radiofrequency) plasma phitization, the key is to favor the reaction at the gra- sources. Along with the activated species and neu- phene edge instead of reaction with substrate by care- tral radicals, ions and electrons coexist, being mostly fully playing with the local saturation of carbon. In the highly energetic ions a potential cause of etching figure V.32, a deviation in T from the established recipe of films. From a geometric point of view, a minimum affects the nucleation density, growth rate, microstruc- distance between the plasma zone and the substrate ture and N. Figure V.32(a) shows single layer grains (20 cm) should be respected (‘remote plasma activa- at 650 °C. At 700 °C, figure V.32(c) more flakes of few tion’). Moreover, the inhomogeneous plasma activa- layers are predominant. Side chemical and topological tion (plasma ball) results in different deposition rates effects take place, as covalent bonding with substrate on different sites. For this reason, to make de process or etching, hindering the desired van der Waals inter- reproducible the position of the sample in the holder action. Formation of carbide phases or volatilization is critical and must be respected. Apart from geomet- of oxide species (CO, H2O, OH, SiO...) can be pro- ric considerations, one parameter that can minimize duced. There are interdependence between pressure the presence of remaining ions reaching the sample is and T, as the unbalanced pressure has similar effects, as the positive bias of the substrate [967–969]. The acti- observed in figure V.34 (variation of 20% of gas flow). vation of the substrate is also minimized preventing If hydrocarbon flow is below the critical value at a given high nucleation density. The last fundamental plasma T, there is no deposition in the nucleation step. If it is parameter is the power; the higher the power, the over, the deposited nuclei suffer from amorphization. grater is the gas activation. However, as this protocol In the second step, a similar effect occurs with (T2) or is a competitive process between etching and growth, partial pressures relation, P2(C2H2)/(P2H2), resulting assessing the role of power is not straightforward. The in secondary nucleation or etching of the deposited increase of power itself does not reflect a proportional material. In figures V.34(a) and (c) a deposition of sub- increase of all species. monolayer graphene can be seen in topography and A T increase keeping the other parameters fixed, friction images. In figures V.34(b)–(d) more few layer results in an increase of the reaction rate but at the flakes are deposited increasing pressure. expense of an enhanced nucleation density due to The precursors and atomic hydrogen, also play the present dangling bonds in freshly activated oxide a critical role. An appropriate H/CxHy ratio in the surfaces. As a minimum T  >  400 °C is needed for gra- atmosphere is necessary to grow crystalline material 111 2D Mater. 7 (2020) 022001 C Backes et al a) b) 0nm 7nm 0nm 5nm 6 100 c) 5 4 d) Z[nm] 95 3 2 Transmittance % 90 1 0 85 0 100 200 300 400 500 600 X[nm] 80 1.5 75 Quartz Quartz+Graphene Z[nm] 1 70 Graphene Graphene after anneal. 0.5 65 0 60 0 20 40 60 80 100 500 1000 1500 2000 X[nm] Wavelenght (nm) Figure V.33. Characterization of continuous film T  =  650 °C, H2/C2H2  =  55/0.25:0.20 (sccm), PT  =  5.4  ×  10−2 mbar: P  =  100 W. (a) and (b) AFM topographic images of graphene deposition on quartz for 5 min nucleation time and 10 h growth. (c) Transmittance spectra before (95%) and after (92%) post-growth annealing. (d) Height profiles taken along the corresponding lines in (b) (film thickness, cyan). [970]. There is a competition between growth from highly favored over the nucleation on substrate. Even CxHy radicals and etching of amorphous deposits by so, some limitations continue as the process is slow atomic H [971, 972]. The etching rate depends on T and not completely self-limiting as can be seen in and N. ­figure V.34(d) where several layers conforming a mul- Ref. [774] found that C2H2 is a suitable precursor tilayer deposit (rose petals shape) is apparent. The ori- for low T, and no relevant differences between the films gin of this multilayer deposits is the slight reduction of deposited with CH4 and C2H2 diluted in hydrogen, the quartz surface. once the appropriate H/CxHy relation and T) settings The reduction process of the quartz surface has are selected. For C2H2 the activation is much higher been computationally simulated by modelling the due to combination of plasma and thermal activation quartz [1 1   −  2 0] facet (x-cut), assessing the activation (that does not happen with CH4) [973]. This enhances energies for the most probable reaction paths within the reaction rate because the plasma environment is the CI-NEB approach [916, 917] (see figure V.35(a)). full of activated carbon dimmers, postulated as the Protonation of surface oxygen with atomic hydrogen main precursor, or at least the main intermediate from plasma and subsequent water vapour release are product, in the case of the synthesis of graphene, due to highly probable in this surface at Si synthesis T[774]. their higher mobility [974, 975]. The passivation of the silicon enriched surface with Ref. [774] used fused silica at high T up to 700 °C. hydrogen prior growth is also favoured because the Si However, some kind of modification was detected in detachment is more energy demanding. Simultane- the surface and changed to quartz. ously to the different steps along the reduction of the Dangling bonds in freshly activated surfaces quartz surface (figure V.35(a)), plasma products, act- (oxides or insulators) promote strong chemical ing as reaction precursors toward the formation of interaction between the surface and gases. To over- graphene, start to adsorb on the surface, which will come this limitation the two step process was devel- lead to their decomposition and dehydrogenation, oped Ref. [774] where the edge growth from seeds is favouring the nucleation at specific enhanced-reac- 112 2D Mater. 7 (2020) 022001 C Backes et al Figure V.34. Graphene on quartz T  =  650 °C, PT  =  5.4  ×  10−2 mbar: P  =  100 W. t1  =  5 min. t2  =  690 min. Topographic (top row) and corresponding lateral force (bottom row) AFM images of samples grown at increasing pressure from (a) and (c) H2/C2H2  =  55 0.25:0.20 (sccm) to (b) and (d) H2/C2H2  =  55/0.30:0.25 (sccm). The increase in the number of FLG with increasing pressure is seen. Figure V.35. (a) Side view of the different DFT-computed steps along the reduction process of the [1 1   −  2 0] quartz surface (adapted from [774], © IOP Publishing Ltd): starting from the clean hydroxylated surface [976] across subsequent protonation barrierless steps and detachment of H2O molecules (see inset), which lead to slight surface reconstructions, and H-passivation of the surface. (b) sketch corresponding to the coexisting processes leading to the growth of graphene on a quartz single-crystal surface from the different plasma products by PECVD: adsorption, decomposition and dehydrogenation of the reaction precursors, toward graphene nucleation and formation. 113 2D Mater. 7 (2020) 022001 C Backes et al a) C2H2 flow (0.9 sccm-0.8 sccm) 1.4 1.2 height [nm] 1 0.8 0.6 0.4 0.2 2.58 nm 0 0 50 100 distance [nm] 1 0.8 height [nm] 0.6 0.4 0.2 0 0 50 100 distance [nm] b) C2H2 flow (0.8 sccm-0.7 sccm) 1 0.8 height [nm] 0.6 0.4 0.2 0 0.00 nm 0 50 100 distance [nm] 1 0.8 height [nm] 0.6 0.4 0.2 0 0 50 100 distance [nm] Figure V.36. AFM topographic images of graphene nuclei and corresponding profiles at 550 °C, PT  =  5.4  ×  10−2 mbar, P  =  200 W, H2  =  50 sccm, t1  =  5 min, t2  =  150 min. (a) C2H2 flow: 0.9 sccm in the nucleation and 0.8 sccm in the growth step. (b) C2H2 flow: 0.8 sccm in the nucleation and 0.7 sccm in the growth step. Adapted from [981], with permission of The Royal Society of Chemistry. tivity sites, depending on the surface reduction stage Figure V.36 is one example of graphene nuclei (see figure V.35(b)). Graphene nanopatches will grow obtained at 550 °C following the two step protocol from the incorporation of dehydrogenated carbon using two C2H2 flows. The increase of C2H2 flow in atoms, as well as from nucleation centers catalysed by the nucleation step from 0.8 to 0.9 promotes a moder- the surface. The coexistence of these processes with the ately increase of the nucleation density for a particular plasma atomic hydrogen maintains the H-passivation nucleation time (~30 nuclei µm−2). Also, the increase and de-passivation (by surface-assisted H2 formation) of C2H2 flow from 0.7 to 0.8 sccm increases the grain equilibrium in the surface promoting the plasma pre- size during a given growth time. The height profiles cursors into graphene. in figures V.36(a) and (b) evidenced that, despite the different graphene nuclei sizes, for both experimental V.3.5. PECVD on Si/SiO2 conditions, the thickness of the nuclei remained below PECVDis as an alternative approach to directly 1 nm, resembling the features of SLG on quartz [774]. deposit graphene at low T. Several studies were During the synthesis, the SiOx surface is not affected in reported on the grown of graphene by PECVD over terms of surface roughness. Si/SiO2 with thick (250–300 nm) and medium oxide After nucleation, the enlargement of the nuclei layers [977, 978], resulting in BLG or FLG [979, 980]. in figure V.37(a) follows to complete the deposition Ref. [981] addressed the direct growth of graphene on of a continuous film. Figure V.37(a) shows the sub- Siwafers with ultrathin native oxide by r-(ECR-CVD) monolayer coverage of the substrate. Longer t2 (seven using the two-step protocol discussed throughout this hours) enables the creation of graphene domains up section to separately control nucleation and growth to 300 nm in lateral size. Flake thickness, as shown in stages [977, 774], enabling tuning of grain size and ­figure V.37(b), corresponds typically to SLG. The coa- thickness. lescence of the domains occurred without vertical 114 2D Mater. 7 (2020) 022001 C Backes et al a) b) 1.2 1 0.8 Z[nm] 0.6 0.4 0.2 0 200nm 0 200 400 X[nm] 550ºC submonolayer c) d) 550ºC monolayer 550ºC monolayer anneal. D G Intensity (a. u.) 2D D´ 1200 1500 1800 2100 2400 2700 3000 Raman Shift (cm -1) Figure V.37. (a) AFM topographic image of graphene nuclei at 550 °C, H2/2H2  =  50/0.9:0.8 (sccm), PT  =  5.4  ×  10−2 mbar: P  =  200 W, t1  =  5 min t2  =  7 h. Vertical scale: 0–5 nm (b) Corresponding profile from a graphene flake in (a). (c) Continuous film, t2  =  10 h and same conditions than (a). Vertical scale: 0–5 nm. (d) Corresponding Raman spectrum of sample (a) (blue) and (c) before (black) and after annealing (red). growth at the grain boundaries as seen in figure V.37(c) in vacuum at 10−6 mbar, lower than reported for low T but the surface of the film is not completely flat. Inho- procedures [977]. After annealing, I(D)/I(G)decreases mogeneous stitching between grains is apparent, from ~1.1 to 0.9. This could be related to H desorp- along with lack of continuity in some others (darker tion from the film inducing sp2 hybridization in the contrast). Figure V.37(d) presents the corre­sponding carbon atoms [774]. The decrease of Rsafter annealing Raman spectra before and after annealing. The Raman could be also related to the same reason: H desorption spectrum from submonolayer coverage shows the from grains and boundaries and conversion of carbon characteristic graphene peaks. I(2D)/I(G)~1 and 2 and hybridization [774, 982, 983]. FWHM(2D) 38 cm−1. The high I(D)/I(G)~ 1 is due to the high density of grain boundaries and defects. Along V.3.6. PECVD of vertical structures with grain boundaries, another source of defects could PECVD can be used to synthesize graphene on arise from the hydrogen enriched atmosphere during insulators. In Ref. [854–856] a cold wall chamber was deposition, which functionalizes the graphene grains pumped by a turbo pump and an RF plasma power and boundaries, inducing some re-hybridization in of up to 1000 W is applied on top of the chamber, the corresponding carbon atoms [982]. The Raman through a matching network. H2 and CH4 are used spectrum in black from the continuous film in fig- as precursors (figure V.38(a)). Different growth ure V.37(c) shows similar results. The continuity of conditions were investigated and the highest film the film in figure V.37(c) was also confirmed by four uniformity was achieved by using a 1000 W RF plasma, point probe measurements (Rs~15 kΩ·sq−1) This a T  =  750 °C and a growth time of 20 s. Figure V.38(b) high value can be related to these aspects mentioned: shows an AFM image of the grown film directly on incomplete stitching in some grain boundaries due to Si/SiO2, highlighting a film roughness of 0.624 nm. randomly oriented grains and functionalization [972]. The film thickness was 3.33 nm (figure V.38(c)), Thisdecreased to 8 kΩ·sq−1 after annealing at 650 °C indicating MLG. 115 2D Mater. 7 (2020) 022001 C Backes et al Figure V.38. (a) Schematic PECVD system used to grow graphene on Si/SiO2. Figure reproduced from [856], with permission from Elsevier. (b) AFM image of the grown graphene film (c) Height profile. V.3.7. CVD graphene by sacrificial metal layers on Si/ the chamber for an hour. While pumping, 5 sccm H2 SiO2 is admitted to remove moist and air remains from By replacing Cu foil with a thin sacrificial metallic film the chamber. Thereafter, the chamber is heated up to one may arrive at an alternative route for graphene 700 °C in H2 (5 sccm, 0.5 mBar, 20 °C min−1). At 700 °C deposition on dielectrics. The proposed approach is the CVD chamber is vacuumed and CH4:H2 (1:1) gas based on the fact that in the CVD process, graphene mix injected (8 mBar, static atmosphere). T is risen (10 grows on the Cu-substrate interface provided that °C min−1) up to 950 °C and kept for 5 min and then the carbon precursors have access to it [802, 984]. Thus, chamber is cooled down to 700 °C in one hour. At 700 by removing graphene from the Cu surface by RIE and °C the CH4–H2 gas mix is replaced with H2 (10 mBar) then removing the Cu catalyst by wet etching, one can and the chamber is cooled down to RT overnight (see obtain graphene on a silica substrate [802, 984]. also [984, 987]). Ref. [984] used up to 300 nm thick thermally evap- Surface melting is a key process that governs the orated sacrificial layers of 99.999% pure Cu on silica deposition of graphene [984]. Although the melting or Si/ SiO2. Cu evaporation does not require UHV, T of bulk Cu is 1084 °C, the melting of low dimen- although the higher vacuum reduces the oxidation of sional systems takes place in a much lower range [988]. the deposited Cu film. Ref. [984] indicated that a vac- Therefore, although the hydrocarbon catalysis starts uum level ~10−5 Bar is sufficient. Since the surface of ~800 °C–900 °C [989, 990], in Ref. [984], CH4 was the deposited Cu layer rapidly oxidizes in an ambient injected into the chamber at 700 °C. At this T, cata- atmosphere [985], it is advisable to start the CVD pro- lytic decomposition of CH4involves mobile atoms of cess soon after evaporation to minimize the amount of molten Cu. Since the carbon solubility in Cu is very low Cu oxide. Otherwise, Cu oxide can be removed by treat- (<10 ppm at 950 °C) [790], the migration of the car- ment of the deposited layer in acetic acid [985, 986]. bon atoms bound to Cu atoms is mainly restricted to Ref. [984] placed the substrate into the CVD chamber the surface [802, 881, 882]. When a few hundred nm immediately after Cu deposition and pumped down thick Cu film melts, melting of each Cu grain results 116 2D Mater. 7 (2020) 022001 C Backes et al in well-defined grain boundaries [802]. Highly mobile substrates (SiO2/Si, glass and PMMA). Then the stack Cu atoms carry carbon atoms along the molten grain heated at low T. Ref. [814] suggested that carbon dif- boundaries allowing them to penetrate into the Cu fuses through the Ni film boundaries to the interface. film and arrive at the substrate surface [802, 984]. This At 160 °C continuous graphene films of reasonable effect eventually results in graphene synthesis on sub- quality (~667 cm2·V−1·s−1) was deposited on SiO2, strate-Cu interface. After reaching the maximum T ~ however N was barely controlled. At lower T (from 950 °C the reactor cools down very slowly, i.e. the car- 60 °C to 160 °C) the deposition of continuous films is bon precursor (CH4–H2 mixture) is available for gra- avoided being the coverage from 60% to 98% and the phene synthesis at the surface of the molten Cu grains grain size and quality of the deposits deteriorate. At 60 until the Cu solidifies at 700 °C. °C on glass and PMMA the Raman signal shows the The Cu film thickness is an important param­ deposition of highly deteriorate graphene layers with eter [984]. When the thickness of an original Cu film nanometric grain size. is  <100 nm the molten Cu film recedes forming mainly Another route is the use of alloys as catalysts, an round droplets of submicron radii covered by graphene, approach well-known in heterogeneous catalysis to with no graphene between those Cu droplets. When the tune the reactivity and selectivity of precursor mol- thickness is ~300 nm the Cu film remains solid after the ecules. Low T (~450 °C) SLG (74%) was demonstrated process, resulting in a grainy but continuous graphene via the catalysts alloying (sputter deposited polycrystal- film to the Cu-substrate interface [984]. line Ni films covered with evaporated Au. The initial Au The topmost layer of graphene can be removed top-layer dissolves into the Ni upon annealing) [878]. from Cu by oxygen plasma (20 sccm/100 W/1 min), The nucleation density is reduced and the graphene while the remaining Cu protects the graphene layer at domains exhibit lateral dimensions  >15 µm. The main the Cu-substrate interface. The remaining Cu is etched limitation is the lack of control in the deposition of by ferric chloride solution. If the topmost graphene is monolayer films without multilayer areas [877]. not removed, it will usually stay on the substrate sur- Pd–Coas a catalyst was introduced to realize low T face as small, arbitrary ribbons and fragments. graphene growth on glass substrate by using remote Raman characterization in figures V.39(c) and (d) RF PECVD [877]. FLG films were formed with high shows a small D-peak and the absence of amorphous transmittance ~88.8% and Rs~66 kΩ·sq−1 at T as low carbon [991]. The SEM image in figure V.39(b) indicates as 400 °C, in contrast with Ni catalyst in which no gra- that the material is grainy and consist of µm and sub µm phitic layer was formed. High decomposition rate of flakes. Ref. [802] reported a rather poor µ ~300 cm2 V−1 hydrocarbon gases and formation of nanosize aggre- s−1. However, the linear and nonlinear optical proper- gates enhances the carbon incorporation into PdCo ties of directly deposited graphene are somewhat com- alloys and as a consequence, a graphene layer is formed parable to graphene grown on Cu foil [992]. at low T. The graphene may be further improved by One major part of the process is that of the sub- optimizing Co composition and CVD conditions. strate surface. Similarly to controlling the water droplet behaviour by modifying the surface wettability [993], V.3.8. Conversion of amorphous carbon and other surface modifications can be used to control molten carbon sources on sacrificial layers Cu [987]. In figures V.40(a)–(c) one can observe a grat- For this purpose, one can employ liquid and solid ing structure that has been under the Cu film during carbon precursors for graphene synthesis [814, 847, CVD. The continuous Cu film has first molten and 994–997]. Regarding the substrate, due to higher receded in between grating lines. Under those Cu lines, carbon solubility and lower reaction T, Ni can be graphene lines are located. When the width of the grat- considered a more aggressive catalyst in comparison ing lines is sub-µm, the lines are no more discrete but to Cu. Its use may be advantageous when solid carbon sometimes merge together [987]. precursors are chosen. Ni-catalysis is activated at a few Deposition on thin Cu film does not offer an ideal hundred degrees lower T than Cu [182, 802, 814, 984, route for graphene synthesis. An important drawback 994–997] and may open pathway for direct deposition is the very high, 900 –1000 °C deposition T. This makes on bendable substrates [881, 882, 995, 998]. The graphene deposition impossible on some substrates drawback of Ni is the high C solubility (~0.6 w% Ni because (i) substrate melting (e.g. polymer substrates), at 1326 °C [999] that makes difficult to control the (ii) recrystallization of a thin film (e.g. amorphous tita- deposition process. nium dioxide) or (iii) thin film-substrate thermal expan- Typically, the amount of carbon precursors deter- sion mismatch, which may cause thin film mat­erial to mines number of carbon atoms available for the catal- crack out from the substrate surface. Moreover, high syn- ysis, while the temporal evolution of the process T thesis T may start a chemical reaction between the metal governs N [1000]. The cooling rate is important since and the substrate (e.g. Si may react with Cu or Ni). it defines either the dissolved C will stay inside Ni or Another approach demonstrated that even at near be squeezed on a Ni surface forming SLG or FLG films RT (60 –160 °C) graphene can be deposited from [1000], while the amount C dissolved in the Ni is deter- graphite powder by using Ni as catalyst [814]. A thick mined by catalyst amount. Ref. [1001] minimized the Ni polycrystalline film was firstly deposited over the thickness of Ni to decrease the amount of dissolved 117 2D Mater. 7 (2020) 022001 C Backes et al Figure V.39. SEM images of sample with (a) and without (b) and (c) Cu catalyst remains after CVD. Raman spectra with (c) and without (d) Cu catalyst. (e) High resolution SEM image from border area where Cu remains are removed. Adapted from [984], with permission from Elsevier. Figure V.40. (a)–(c) SEM images of sample before CVD, after CVD and after Cu removal in FeCl3. (d) and (e) Raman mapping reveals graphene on the grating. (f) Raman spectrum of the grating area is comparable to that grown outside. Adapted from [987], with permission from Elsevier. carbon because precise T control of the bulky hot wall CVD chamber varied from 600 to 900 °C and resulted CVD of was hardly possible. Ref. [1001] used 10 nm pyrolyzed photoresist film (PPF) with and without Ni thick Ni film on a silica substrate as a catalyst and the [1001]. Since during the CVD process some part of the carbon precursor was nLOF photoresist (AZ nLOF resist layer is expected to evaporate, the Ni film was 2070, negative tone resist diluted with AZ EBR 70 deposited under the nLOF layer. Thus the Ni layer did thinner). The pyrolysis T of the photoresist film in the not prevent the evaporation from the resist layer. 118 2D Mater. 7 (2020) 022001 C Backes et al Figure V.41. (a) optical microscope and (b) SEM image of PPF with Ni particles fabricated at 800 °C. SEM reveals non-uniform size and shape of Ni particles (bright particles). Raman spectra shows the evolution of the carbon film as a function of temperature (c) without and (d) with Ni in the process. Adapted from [1001]. Table V.2. Methods, substrates, precursors and optimal growth conditions. Method Substrate Precursor Temperature Pressure Lateral grain size CVD • Cu Foil CH4, C3H8 1000 °C  < 0.5 mbar–1 bar 10 µm • Partially oxidized T  <  1084 °C Few mm-cm Cu foil enclosed CVD Cu foil C2H4 850 °C 2 mbar 1 µm CVD Ir(1 1 1)/YSZ/Si(1 1 1) C2H4 850 °C 1 · 10 —2 · 10 −8 −7 mbar Up to a few mm CVD Ge(1 1 0) on Si(1 1 0) CH4 900 °C–930 °C 850–990 mbar <1 µm CVD Ni Foam CH4 1000 °C 50 mbar Depends on metal foam structure CVD Cu Foam CH4 1000 °C 400 mbar Depends on metal foam structure CVD Cu film on Silica wafer CH4 950 °C 8 mbar <1 µm PT-CVD Cu CH4 950 °C 7–20 mbar 1 µm MBE Au (1 1 1) Glassy carbon filament 550 °C UHV Tens of nm PVD Cu (1 1 1), Pt(1 1 1), C60 800 °C–900 °C UHV Tens of nm Rh(1 1 1) single crystal or polycrystalline PECVD Glass or quartz C2H2 600 °C–720 °C 0.1–0.9 · 10−4 mbar 500 nm Similarly to Cu, Ni will also melt and recede dur- rier enabled the growth of SLG at 600 °C, with domain ing CVD. Since the thickness of the original Ni film size  >50 µm and quality that equals the CVD grown was only 10 nm the particle size is in sub-µm range. films at similar T. Figures V.41(a) and (b) show the appearance of the The direct formation of graphene on various Ni particles on PPF pyrolyzed in 800 °C. In Ref. dielectric surfaces via a single-step rapid thermal pro- [1001] the T dependence to the PPF graphitization cessing (RTP) of substrates coated with amorphous was observed by Raman spectroscopy. Most signifi- carbon and nickel (Ni) thin films has been reported cant changes in the structure of the PPF with Ni took [1002]. Here, high-quality graphene was obtained uni- place For T  =  800 °C or above (see figures V.41(c) formly on the whole surface of wafers with a controlled and (d)) [1001]. A 2D peak appeared at 2705 cm−1. N. SLGshoewd a low Rs and a high optical transmit- The D and G peaks became narrower, indicating tance in the visible. higher degree of crystallization in to comparison to PPF pyrolyzed in 700 °C and lower T or without Ni V.4. Brief comparison of all methods catalyst. There is a wealth of methods to grow graphene on a One approach to control the growth process and variety of substrates. Some are more suitable to grow improve the quality of graphene grown by catalytic on catalytic substrates while others are better for transformation of solid carbon sources was the intro- insulators. Most of them exploit a variety of precursors. duction of a carbon diffusion barrier (Al2O3) between Table V.2 summarizes methods, substrates, precursors Ni catalyst and carbon layer [817]. At thin Al2O3 bar- and optimal growth conditions. 119 2D Mater. 7 (2020) 022001 C Backes et al VI. Graphene transfer, placement and metal etching solution such as ammonium persul- decoupling from substrate fate (APS), ~2.0 g in 150 ml of DI H2O, to etch Cu as shown in figure VI.1 and enable transfer of graphene VI.1. Wet transfer onto a dielectric substrate. After Cu is dissolved, the Many advances have been reported on the transfer remaining PMMA/graphene stack is lifted with the of nanolayered materials such as graphene since the desired substrate (PET, SiO2/Si, or other) substrate and isolation of graphene on silicon dioxide and the growth transferred to a container with DI H2O for washing. of graphene on metal substrates [759, 765, 1003, The procedure is repeated a few times to ensure that 1004]. In 2007, Gölzhäuser et al were first to disclose a all of the etching solution is removed. After washing, method to transfer nanolayers. Patents for this method the PMMA/graphene stack is removed from the water were issued in 2013 and 2017 (Gölzhäuser; Armin, bath and left to dry. There are many implementations Nottbohm; Christoph, and Beyer; Andre, US Patent of this process and thus a lot of room for improvement No. 8,377,243 B2 (19 February 2013), A. Gölzhäuser, as the processes are transferred to a pilot line or pro- C.T. Nottbohm, and A. Beyer, European Patent No. duction environments. The PMMA/graphene/tar- EP2144711B1 (31 May 2017)); this process is a wet get substrate is then first transferred in a beaker with transfer procedure very similar to the one published in acetone for PMMA removal for 1 h, second to a beaker the papers above. The nanolayer can be a single layer with isopropyl alcohol for 5 min and then dried with (SLG), bilayer (BLG) or multilayer (MLG) graphene nitrogen, leaving the GRM on the target substrate. but this aspect is not substantial for the purpose of the description of the transfer process. GRMs are covered VI.1.1.2. Au coated mica substrate with a polymeric layer that acts as surface protection as The transfer of a CNM grown on a Au coated mica well as a carrier after the metallic substrate is removed substrate follows a slightly different path and can be by a chemical etchant. The polymeric layer, PMMA, applied also to the transfer to many target substrate with a thickness of several hundred nm, is used to [75, 76, 78, 1006]. The procedure is as follows: coat the surface of GRM. One of the advantages of PMMA (950 K, AR-P 671.04, Allresist) is spincoated this process is that the polymeric layer conforms with at 4000 rpm for 30 s, followed by a soft bake at 90 °C the GRM surface without macroscopic damage. The for 5 min. Optionally, a double layer of PMMA can be resulting polymer/GRM/substrate structure is then applied: (i) a first layer (50 K, AR-P 631.09, Allresist) is dipped into a liquid etchant to dissolve the substrate or, spun to a thickness of 130 nm at 2000 rpm for 30 s and alternatively, detached from the substrate. The latter baked on a hotplate at 90 °C for 5 min; (ii) a second can be achieved by electrochemical methods. After layer of PMMA (950 K, AR-P 671.04, Allresist) has the substrate is removed/etched away or ‘detached’ an a higher molecular weight to provide mechanical optically visible, several hundred nm thick polymer, stability. The latter is spun to a nominal thickness of with the GRM is left floating on the etchant surface. 310 nm at 4000 rpm for 30 s and also baked at 90 °C This ‘hybrid layer’ is then transferred onto a second for 5 min. The advantage of the additional PMMA desired substrate, typically a dielectric surface. Finally, layer with ‘50 K’ is that it can be removed more cleanly the thick polymeric layer is dissolved or removed and than PMMA with a higher molecular weight in direct the GRM remains on the target substrate. contact with the GRM. After the formation of the PMMA transfer medium, the edges of the sample are VI.1.1. Etching of a metallic substrate cut in order to ease the release of the PMMA/GRM VI.1.1.1. Cu foil substrate from the mica substrate. Then, the sample floats on The wet transfer of CVD graphene on metals like Cu the liquid level of I2/KI etching bath (I2:KI:H2O with and Ni or its alloys by etching the metal is now broadly ratio of 1 g:2 g:10 ml) for ~10 min. Au is slightly etched used [759, 764, 1003, 1004]. from its lateral interface The separation of polymer/ Once the graphene is grown on a large Cu metal GRM/Au from mica is attained by dipping into water. substrate (foil) by CVD, the desired sample size is cut To remove the Au layer, PMMA/GRM/Au floats on and transferred onto a 125 µm-thick PET foil, which the liquid level of I2/KI etching bath for ~20 min. acts as a mechanical support. Then a A4-950K poly- After wards the Au layer is completely removed, the methyl methacrylate (PMMA) resist is spin coated on polymer/GRM is transferred to a fresh water bath for the GRM/Cu/PET stack at 4000 rpm for 40 s to achieve rinsing. The hybrid transfer layer can be taken out of the desired thickness, up to ~400 nm, after which the solution and placed onto a target substrate following PET support is removed. Since graphene grows on both the same protocols used to rinse it as detailed in the sides of the Cu foil, that on the uncoated by PMMA previous paragraph. has to be removed to prevent it from being combined If the new host substrate is not solid and the with graphene under the PMMA. This graphene can GRM is to be free-standing on a grid-like substrate, be removed from the Cu surface using a reactive ion removal of PMMA can be done by critical point drying etch (RIE) process, typically using O2plasma at 20 [78, 1009] (CPD-) that avoiding the damaging effects W for 20 s and at 200 mTorr. The PMMA/graphene/ of surface tension. The mounted sample is immersed Cu stack is then immersed/placed on the surface of a into acetone. After 60 min, the chamber is cooled with 120 2D Mater. 7 (2020) 022001 C Backes et al Figure VI.1. Transfer of SLG from a Cu substrate to a dielectric surface. The top-right and bottom-left insets are the optical micrographs of SLG transferred on SiO2/Si with ‘bad’ and ‘good’ transfer, respectively. The bottom-right is a photograph of a 4.5  ×  4.5 cm2 SLG on quartz substrate. Adapted from [1005], copyright American Chemical Society. liquid CO2 and then liquid CO2 is introduced into the Ref. [1009] systematically studied the effect of avoiding turbulence. Liquid CO2 substitutes acetone annealing of transferred graphene by PMMA by FTIR according to a predefined time (15 or 20 min). Finally and XPS in oxidative environments (CO2, O2 and the dryer heats and pressurizes CO2 to its critical point, NO2) and found it more effective than in reducing and gas CO2 bleeds off to leave nanolayer dry. A scheme ambients, such as forming gas or vacuum [1010]. More of the procedure is shown in figure VI.2. specifically, they found that annealing in an atmos­ A HIM [1007] image of a CNM, placed onto a Cu phere of CO2 at 500 °C in a few minutes is very effective grid with hexagonal pores of 40 µm, after removing in removing PMMA (mainly carbon) without damage PMMA without CPD, is show in figure VI.2. Note the to the graphene. Atomic level cleaning of PMMA was appearance of wrinkles. The effect of critical point dry- also performed using radiolized water at high T. The ing can be seen in figure VI.3(a) that shows a flat CNM cleaning process was studied by [1011]. The radiolized on a grid with 40 µm openings. Figure VI.4(b) shows a water was obtained via interaction of the electron very large free-standing CNM transferred onto a hex- beam in the TEM with water at high T. In this process agonal grid with openings of ~0.5 mm [1007]. In both of radiolysis, as the electron beam interacts with water images, no folds and wrinkles are visible. and the graphene surface forms atomic hydrogen and An additional option of removing the PMMA is hydroxyl radicals at high T. This process was reported the inverted floating method (IFM) [1008]. Therein, at the submicron scale and will have to be demon- only the PMMA covered side of the sample is in con- strated at much higher scales before considering it as tact with acetone. Large graphene membranes were a viable practical process, although it is contributing to reported, covering circular holes in SiO2 substrates. the basic understanding of graphene surface cleaning. This approach is most effective in circular support structures and it requires full coverage of the substrate VI.1.2. Electrochemical delamination openings. In contrast to IFM, CPD avoids surface ten- The use of PMMA as a carrier layer followed by sion related forces on the membrane. Therefore, if chemical etching of the metal substrate is an effective available, CPD seems to be a better choice for remov- way of graphene transfer. This process, however, can ing the PMMA layer. be time consuming and the whole metal substrate is 121 2D Mater. 7 (2020) 022001 C Backes et al Figure VI.2. Top: Schematic of GRM transfer using polymeric media. Bottom, left: Optical micrograph of CNM transferred onto SiO2/Si showing interference contrast; bottom right: HIM of CNM transferred onto a metal grid without critical point drying. Ruptures and faults, as well as large intact areas of the freestanding nanolayer are visible. Adapted from [75, 1009]. etched away. The latter makes the process costly and the weakly bound graphene from the metal substrate, environmentally unfriendly. It would be desirable to Cu, Pt, etc [779, 1013, 1015]. K2S2O8also etches Cu use a process whereby the graphene film is removed and thus is not totally non-destructive. However, at from the substrate without significant metal loss low concentrations one might be able to use it and still and without damaging the overlying graphene layer. reuse the Cu substrate. Electrochemical delamination was introduced to Using a custom-made mechanism (figure VI.4) accomplish this task. Ref. [779] reported this processes (utility model P.411053) the Cu/Graphene/PMMA for graphene transfer from Cu substrates and in Ref. stack is placed in KCl, 1 mol dm−3, at a rate of 1 mm [108] for the delamination of graphene from Pt. This s−1 or an aqueous solution of K2S2O8 (0.05 mM) to technique is nondestructive not only to graphene perform the delamination. The graphene/Cu cath- but also for the metal and preserves the quality of ode is kept as the negatively polarized (from 4 to 10 V) the substrate so that it can be reused [108, 779, 780, and the anode can be made of a number of materials 896, 1012]. After the growth process, the Cu foil with including glassy carbon. Hydrogen bubbles appear graphene on top is spin-coated with a thin layer of then at the graphene/Cu interface due to the reduc- PMMA (4% in anisole, layer thickness ~350 nm) and tion of water molecules and allow graphene to gently cut into suitable pieces. PMMA also serves as a support detach. The current depends on the Cu foil size and can preventing the graphene film from collapsing during be adjusted to control the delamination rate to ~1 mm Cu removal. A cleaner support for graphene transfer s−1. was suggested:rosin [1012]. As shown in figure VI.4, In order to obtain the best results, the samples are this technique takes advantage of the electrochemical gradually immersed in the electrolyte solution at an process of metals in various solutions of e.g. (1) KOH angle ~45° as in figure VI.5. At this angle the graphene/ [1013],(2) NaOH [1014], (3) K2S2O8 etc to dissociate PMMA stack slowly moves towards (floats) the top of 122 2D Mater. 7 (2020) 022001 C Backes et al Figure VI.3. HIM image of large freestanding CNMs transferred onto hexagonal TEM grids. (left) The size of each hexagon is 50 µm and the grid is uniformly covered by the CNM.Scale bar is 50 µm. Right) The size of the hexagon is 500 µm and it is completely covered by CNM, no wrinkles or folds seen. Scale bar is 200 µm. Adapted from [1007]. the solution and the Cu foil moves toward the bottom (see section IX.2/ARPES) to determine the chemical of the glass container. After the delamination process, composition of graphene surface before and after heat the detached Graphene/PMMA stack is rinsed with treatment. A large O1s [812] peak on the XPS graphene DI water, transferred onto the substrate, heated to 130 spectrum before the heating is apparent in figure VI.7. °C, and treated with acetone to remove PMMA. Next, An enhancement of the intensity of the characteristic graphene samples are dried for 1 h at high T (~350 °C) C1s XPS peak located at 288.9 eV, which represents the in high vacuum [851, 1010] or high pressure [1016] C–C=O and O=C–O groups [812], is a signature of furnace to be sure that the organic residues and impu- an incomplete PMMA removal. After heat treatment rities are removed, although other methods claim no in 350 °C in high-vacuum, the C1s and O1s XPS peaks polymer residue [851, 1017]. After the delamination in figure VI.7 are the same as those corresponding to process, the detached Graphene/PMMA is rinsed with graphene samples that had never had been into contact DI (deionized) water [1011], transferred onto the with PMMA, confirming that the removal of the poly- substrate, heated to 130 °C, and treated with acetone mers is complete. to remove PMMA. Additionally, UV-irradiation was An additional method for the confirmation of the suggested to facilitate PMMA rinsing [1018]. When PMMA residue removal is scanning near-field optical substrates other than Cu are used to grown graphene, microscopy (s-SNOM) [1023, 1024] (see section IX.1/ such as Ir, an additional step before the main electro- AFM for an introduction to the technique). Topog- chemical delamination process is needed. It was shown raphy and near-field images recorded on graphene that an effective transfer could be achieved in SLG/Ir samples on SiO2/Si before and after heat treatment are samples after treatment with a 0.1 M tetraoctylammo- shown in figure VI.8. After heating the graphene sur- nium-bromide TOABr/acetonitrile solution prior to face looks much cleaner, the wrinkles smoothen and the electrochemical delamination [1019, 1020]. Fig- the particles derived probably from PMMA are fewer. ure VI.5 presents the scheme of the graphene transfer As the probing depth of s-SNOM is larger compared to process. Figure VI.6, shows the iteration of the transfer the graphene thickness, the recorded optical contrasts process to place MLG on PMMA [1021]. Ref. [1015] show not only the local defects in the graphene layer found that the delamination of graphene from Pt and but also differences in the substrate composition and Cu electrodes (substrates) increases with Na(OH) in the distribution of the PMMA contamination concentration, making this attractive as practical pro- A spectroscopic method based on imaging an area cess to remove graphene from metal surfaces. More at different wavelengths was used to test whether the work is needed to bring this process from laboratory contamination particles randomly spread across the to ‘production’. graphene samples are indeed PMMA molecules. As the One of the most difficult challenges are residues dielectric function depends on the wavelength [1025], left by the PMMA used as the carrier µ. Ref. [1022] the scattering coefficients (amplitude and phase) of studied the surface of CVD graphene before and after the investigated materials varies with the tuning of the transfer and compared it to HOPG. They found that laser wavelength [1026]. Imaging at 1150 cm−1 illumi- there are significant organic residues that limit µ [851]. nation frequency, on resonance with the SiO2 vibration It is difficult to completely remove all impurities using (asymmetric stretching vibration of the oxygen atoms only the organic solvent and the PMMA residue from bridging the Si-O tetrahedral groups) the optical con- the graphene surface. High-T treatment in high-vac- trast between graphene/SiO2 and PMMA particles uum is additionally needed for a more effective and is the largest. The near-field IR images recorded on a cleaner transfer. To examine the effect of T during this PMMA particle in figure VI.9 show a stronger phase at treatment, x-ray photoelectron spectroscopy was used 1740 cm−1 due to the carbonyl band absorption at this 123 2D Mater. 7 (2020) 022001 C Backes et al Figure VI.4. Electrochemical exfoliation of graphene from Cu. (a) Schematic diagram of electrochemical cell used for exfoliation, (b)–(d) optical images showing the ‘whole film’ peeling of PMMA-covered graphene form Cu. (e) Graphene film wet transferred onto a 4 inch wafer. (f) Time needed for complete film exfoliation of a 5 cm  ×  5 cm foil as a function of voltage and electrolyte concentration (at 5 V, red line). Adapted from [779], copyright American Chemical Society. wavelength [1027], while the optical phase on the gra- Another way to ensure a clean graphene surface phene and substrate is larger at 1150 cm−1, matching is to coat it with a metal or an oxide to eliminate the the SiO2 phonon absorption [1028]. For both wave- adhesion of any organic compounds. Ref. [1017] used lengths, the optical amplitude (scattering efficiency) Au as interface layer between graphene and PMMA is stronger on the inorganic materials compared to and demonstrated a decrease in contact resistance PMMA, due to the lower reflectivity coefficient on the of a factor of two. Ref. [1030] used Ti as an interface polymer [1029]. These measurement together with material to clean the graphene surface after transfer. Ti the XPS strongly suggest that the particle residue is reacts with the residues on the graphene upon deposi- PMMA. tion. The ‘reacted’ Ti is then removed using a dilute Ref. [1012] studied the use of rosin (C19H29COOH) HF solution. This also led to a decrease in contact as a support layer. The advantage is a high solubility in resistance. organic solvents, weak interaction with graphene, yet strong enough to enable support and transfer. These VI.1.2.1. Chemical delamination potentially make it a more suitable support in compar- There are two main techniques that have been ison to PMMA, as for the XPS data comparison of the evaluated to delaminate graphene from metals transferred graphene using the two different mat­erials chemically: (1) Tetraalkylammonium (TOA)- [1012]. Therefore, through a combination of sup- assisted electrochemical delamination and (2) water port material selection and surface treatment process/ delamination. The first is is a modified electrochemical cleaning one can in principle create a graphene surface delamination (‘bubbling’) method which enables that is clean enough for many applications. The clean- the exfoliation of SLG and FLG or h-BN from liness needed will depend on the application. substrates which are more strongly interacting than 124 2D Mater. 7 (2020) 022001 C Backes et al Figure VI.5. High-speed electrochemical delamination graphene transfer. Adapted from [1021], with the permission of AIP Publishing. Figure VI.6. (a) Images of stacks of 2, 4, 9, 12, 24, 37, 50 LGs on PMMA. (b) Optical transmittance of MLG stack. Cu [1031]. The TOA pretreatment was introduced in other reference electrode, in which case one needs to Ref. [1019] to enable the subsequent delamination of correct the bias between WE and RE to account for the Graphene on Ir. Tetraalkylammonium compounds potential offset of the RE). The Counter Electrode (CE) are weaken the interactions of sp2 materials with the is a Pt wire. The WE and the RE should be placed as underlying metal [1019, 1020, 1032], thus facilitating close as possible facing each other (avoiding contact); the subsequent detachment from the substrate. the CE should be facing the WE (with the RE between TOA+ intercalation is a 3-electrode electrochemical the two) and should be immersed in such a way to procedure which requires the use of a potentiostate to expose a large area. The operation of the vessel follows ensure a fine control on the bias between working and several steps. Fill the vessel with 10–20 ml of TOABr/ reference electrode. In the following, reference will be acetonitrile solution (0.1 M concentration, i.e. 5.46 g of made to graphene grown on Ir, but the same protocol TOABr every 100 ml of solution). Subsequently, degas- applies to many other LMs, including h-BN/Rh and sing the solution with Ar or N2 for 15 min is required. MoSe2. After, leave the Ar/N2 capillary tube above the vessel, in The experimental set up of the electrochemical cell order to maintain an Ar or N2 atmosphere and prevent can be seen in figure VI.10. The cell is open to the air air from dissolving into the solution. Then, apply  −1.9 or Ar atmosphere. The Working Electrode (WE) is the V between WE and CE for 10 min to promote TOA+ graphene/Ir sample. It is fixed to one corner by an alli- to the graphene/metal interface. A partial discharge gator clip. The reference electrode (RE) is an Ag wire of the sample for 20 s at  −0.2 V follows to obtain Ir/ (alternatively one can use a standard Ag/AgCl or any TOA/graphene stack. This step is needed to prevent the 125 2D Mater. 7 (2020) 022001 C Backes et al Figure VI.7. High resolution XPS C1s (a) and O1s (b) spectrum of transferred graphene samples before heating (red line), after heating (blue line) and graphene samples without contact with PMMA (black line). Figure VI.8. AFM and near-field IR images recorded from samples before (a) and after (b) heat treatment. Scan size for each measurement was 10  ×  10 µm2. negatively charged graphene from reacting with the electrolysis process. This process enables the reuse of PMMA and facilitate subsequent PMMA removal. the growth substrate without significant metal loss. After TOA-pretreatment, the samples must be rinsed in acetonitrile, dried in N2 atmosphere, and VI.1.3. Electrochemical oxidation by cyclic spin-coated (or drop-coated) with PMMA. The sam- voltammetry ples can be then exfoliated according to the electro- Electrochemical oxidation can be described in terms chemical delamination explained above [1021]. of the formation of a metal-oxide layer at the interface Water intercalation has also been found to be an between graphene and metal. A rational optimization of effective method of decreasing the bonding between the parameters of this process follows the methodology graphene and underlying metal substrate [1033]. In described below. The voltammetric profile of Pt in a this process, water is intercalated between metal sub- non-adsorbing electrolytes is known [1034], but the strate, e.g. Pt, and graphene. The intercalated graphene conclusions can be extended to other metals. It was film/metal substrates is then placed on a hydrophilic reported [1035] that the processes that occur at the SAM surface, dry bonded and transferred through an metal surface can be described by the reaction 126 2D Mater. 7 (2020) 022001 C Backes et al Figure VI.9. Sequential imaging of a PMMA particle at 1130 cm−1 and 1740 cm−1. Scan size: 200  ×  200 nm. Figure VI.10. 3-electrode setup used in TOA intercalation and voltammetry experiments. Adapted from [1020], with permission from AAAS. H2 O + (∗) ↔ OHads (∗) + H+ + 1e− [1037]. There is a N2 gas inlet and outlet to assure an inert environment during the electrochemical oxida- where * stands for a free Pt adsorption site. In a first tion. A Pt wire (99.99% purity) serves CEand all the step, the water molecules interact with the graphene/ potentials are quoted with respect to the Ag/AgCl (RE. Pt electrode, intercalating through graphene, then The Ag electrode is specific for the cell (Ag/AgCl elec- dissociate in the Pt surface forming Pt2(OHads). A trode driref-25H, WPI). The WE is the Pt(1 1 1) sample charge ~101.50 µC cm−2 is transferred, corresponding covered with a SLG. The WE area exposed to the solu- to the formation of Pt2(OHads) with a coverage close tion is 0.33 cm2 in all experiments. The electrochemi- to 1 monolayer. Even though SLG presents a barrier to cal cell is set into a Faraday cage in order to reduce the direct access to the metal surface, the water molecules signal-noise ratio. The electrolyte employed is 0.1 M are able to diffuse onto the metallic substrate via HClO4 and the water is purified with a M ­ illipore Milli-Q surface defects [1036]. The intercalated water between purification system. A scan rate of 0.05 V · s−1 is selected SLG and metal surface is responsible for the oxidation to control the oxidation reaction. The anodic potential of the outermost Pt atoms when an appropriate is varied to obtain the optimal conditions to decouple potential is applied. graphene from Pt avoiding side reactions, such as the Voltammetric measurements were performed evolution of CO2 and O2, and defects in the graphene under N2 in a three-electrode set up (figure VI.11) network. The sample is placed on the electrochemical 127 2D Mater. 7 (2020) 022001 C Backes et al Figure VI.11. Electrochemical cell used for the electrochemical oxidation of Pt in the Graphene /Pt (1 1 1) stack. RE, is the Ag/AgCl reference electrode, CE the Pt counter electrode and WE the working electrode (i.e. stack graphene/ Pt(1 1 1). The cell is purged with N2 and installed inside a Faraday cage. cell for 1.5 h soaked into the previously deoxygenated semi-dry transfer. This is the process of transferring acid solution so water molecules are able to diffuse graphene without complete immersion of substrate onto the metallic substrate via surface defects [1036]. and graphene in an aqueous environment [115]. In Figure VI.12 shows four cyclic voltammograms with particular this refers to the use of melt techniques different experimental conditions used to decouple whereby a liquid interfacial layer is solidified by an area ~0.33 cm2 of graphene from Pt(1 1 1).The first various means [1759]. Both methods are large-area- protocol (A, figure VI.12(a)) is an anodic potential scan compatible [115], and are capable of producing from 0.36 to 1 V followed by a cathodic potential scan transferred layers that are optically transparent between 1 V to 0.7 V. The second B, figure VI.12(b)) is and electrically conductive whilst retaining areal similar, but with a higher applied potential (1.05 V). In uniformity with only a few percent variation. UVA and the third cycle (C, figure VI.12(c)) the potential is even HPL are stable under bending stress and fatigue with higher (1.45 V). In the last one (D, figure VI.12(d)) two a resistance towards micro-crack formation compared cycles are involved: a complete oxidation-reduction in to ITO and FTO platforms [1760]. the range 0.45–1.1 V and a subsequent oxidation up Figure VI.13(a) outlines the HPL process [1040]. to 1.1 V, followed by reversing the potential finishing Thermally activated ethylene vinyl acetate (EVA) at 0.7 V. The applied potential in A is not high enough treated PETsubstrate were used, however other sup- to decouple graphene. D is the most aggressive and the ports such as thermal release tape (TRT), with dif- sample is decoupled but also damaged. B and C accom- ferent adhesive strengths ranging from 2.5 to 7.0 plishthe best decoupling, meaning that every method (N/20 mm), can be also used. First, the as-synthesized in between those ranges will be appropriate to decouple graphene-on-Cu stack is sandwiched between the graphene from Pt. polymer and a 125 µm-thick PET substrate, the acting Table VI.1 summarizes the potentials and scan rates as mechanical support. The sandwich structure then for the shown cyclic voltammograms. It also includes traverses a dual roller laminator, which is, contingent the parameters to oxidize Pt and the charge involved on the laminate used, either cold [1761] or heated to in the oxidation process, as well as the charge to get 1 ~100 °C [1040]. Following lamination, the backside monolayer of Pt2OH. The presence of SLG delays the laminate, i.e. the side in contact with the heated stage oxidation process compared with the clean Pt(1 1 1) during CVD, is mechanically detached. This exposed sample, requiring higher overpotentials to attain a graphene-coated-Cu face can then be ashed to remove comparable oxidation charge. The charge obtained by residual organics along with the defective graphene integrating the anodic peaks for the other treatments is layer. The Cu catalyst is then etched, in this case, using lower in comparison with the target. The process starts (NH4)2S2O8 in de-ionised (DI) water (1 M) for 12 h. when the oxide coverage is much less than one mono­ The transferred films are subsequently rinsed in DI layer (~0.25 of monolayer). Table VI.1 also includes water and dried in high-purity N2. Up to 0.1 m2 areas the parameters to decouple graphene from Ir(1 1 1). were transferred using this method, though larger areas and subsequent integration into roll-to-roll process VI.2. Semi-dry transfer: hot press lamination and lines are easily accommodated. EVA is also suitable for UV assisted transfer flexible substrates, which cannot withstand solvents UV adhesive (UVA) and hot-press lamination such as acetone and/or isopropyl alcohol [1762]. Fol- (HPL) [1040], which can be broadly classified as lowing transfer, the resulting polymer/graphene stack 128 2D Mater. 7 (2020) 022001 C Backes et al Figure VI.12. Cyclic voltammetries for graphene on Pt(1 1 1) in 0.1M aqueous HClO4 solution. Four different voltammetric treatments have been applied (A, B, C, D), being the scan rate 0.05 V · s−1 in all the cases. The right panels plot the potential programs applied to the SLG/Pt electrodes, producing the corresponding cyclic voltammograms of the left panels. Table VI.1. Parameters for potential-controlled electrochemical oxidation of Pt(1 1 1) and Ir(1 1 1) with SLG on top. Ei stands for the initial potential, Esup for the upper limit of the potential, Ef for the final potential, v for the scan rate and Q for the charge. Ei (V) Esup (V) Ef (V) v (V · s−1) Q (µC · cm2) 1 ML Pt2OH 120 SLG/Pt(1 1 1) A 0.36 1 0.7 0.05 10 B 0.36 1.05 0.7 0.05 32 C 0.36 1.45 0.7 0.05 106 D 0.46 1.1 0.7 (cycle and a half) 0.05 24 Pt(1 1 1) B 0.36 1.05 0.6 0.05 101.5 Gr/Ir(1 1 1) 0.2 1.0 0.9 0.05 140.8 remains attached to the target substrate. Repeated following a preliminary casting at 500 rpm for 10–20 s. lamination, at RT or elevated T, can be undertaken to This coated substrate is then placed in contact with further remove absorbates, bubbles, voids and non- the as-grown graphene-on-catalyst. This sandwich is conformalities in the transfer [1719, 1720]. In the case subsequently compressed at 0.2 MPa using a cold-roll of TRT, the support-coated substrate is finally baked at laminator, to avoid air pocket formation at the inter- 120 °C to remove the TRT. face. The UVA is subsequently cured by exposing the Figure VI.14(b) outlines the UVA transfer [1040]. backside of the PET to a UV optical source (365 nm, The UV adhesive is first coated onto a polymer sub- 20–25 mW m−2) for 10–15 min. The Cu foil is etched strate by spin coating at 5000–8000 rpm for 30–60 s, in aqueous ammonia persulfate for 12 h, rinsed in DI 129 2D Mater. 7 (2020) 022001 C Backes et al Figure VI.13. Graphene transfer by (a) HPL and (b) UVA, and (b) corresponding optical images and (c) and (d) 550 nm optical transmission maps. Adapted from [1038–1040]. water and dried in high-purity N2. All at RT, making the attributed to Cu etchant exposure [1763]. Such peaks approach applicable to a wide range of polymer sub- were absent for HPL transfers [1764]. S is a strong strates. Figure VI.14(c) shows typical optical micro- dopant of graphitic carbons [1765]. graphs of transferred samples that exhibit long-lasting The 550 nmtransmittance of UVA transfers tends adhesion between substrate and graphene[1040]. to increase with time (~80%) [1040]. No observable change was reported for HPL transfers, likely due to VI.2.1. Benchmarking and ageing effects the absence of doping. The metallic component of may Ref. [1040] reported that the optical transmittance be central to the doping temporal stability [1039]. (%T) for HPL and UVA transfers was ~10% and 12% HPL transfers were reported with a time invariant lower than the uncoated substrates. This was assigned Rs (5.0  →  5.2 kΩ sq−1), whereas Rs of the UVA trans- [1038–1040] to the transferred graphene and the fers increased from an initial 2.2 kΩ sq−1 to 3.5 kΩ augmented optical properties of the interfacial binding sq−1, after 200 h, suggesting the need for an hermetic layer. The spatially averaged 550 nm transmittance of capping layer [1766]. Such deleterious increase in Rs in HPL and UVA is ~58.6%  ±  3.6% and 76.5%  ±  3.8%, due to time and ambient unstable Sdoping associated respectively [1040]. Both techniques result in with the necessary Cu-etching. Rs doped graphene equivalent areal uniformity with  <4.0% variation. remains approximately one order of magnitude larger The increase in absorption is likely due to folding and than achieved by doping strategies reported elsewhere wrinkling of graphene during transfer. Ref. [1040] [1051, 1041, 1042]. This may be, in part, associated reported a rms of HPL ~26% higher than UVA. The with the macroscale measurements undertaken, and agglomerates, 5.5  ±  5.6 µm (UVA) and 7.9  ±  3.9 µm the necessarily higher areal density of defects within (HPL), are most likely adhesive residues. the measured channels. Using PMMA transferred graphene on quartz, To attain a robust mechanical interface, graphene with Cr/Au van der Pauw structures on the order requires proximal contact to the substrate [1767]. In of tens of micron in channel size, graphene showed both transfer approaches, the as-grown graphene-on- Rs~5.47  ±  1.20 kΩ sq−1. Rs for HPL transfers was catalyst achieves contact the EVA melt and to low-vis- ~9.9  ±  3.8 kΩ sq−1, whereas UVA showed lower val- cosity UVA prior to curing. Graphene transferred in this ues ~3.5  ±  2.3 kΩ sq−1 [1040]. Elemental analysis way showed lower surface energy and higher adhesion revealed significant (S) and (O) peaks in UVA transfer, (UVA: 4.40  ±  1.09 N m−1, HPL: 0.60  ±  0.26 N m−1) 130 2D Mater. 7 (2020) 022001 C Backes et al Figure VI.14. (a) Change in normalised resistance (ΔR/R0) with bend radius. (b) 〈ΔR/R0〉 for all bending diameters for UVA, HPL, PMMA and ITO. (c) peel strengths for transfer methods. Inset: Typical AFM height maps of fatigued transfers. Adapted from [1038–1040]. compared to PMMA transfers (PMMA: 0.44  ±  0.06 N UVA and HPL increased from ~46 kΩ to 74 kΩ, with m−1) [1040]. less than  ±0.5 kΩ difference between bent and relaxed To assess the bending fatigue performance, trans- states [1040], highlighting the robustness in maintain- fers were assessed during 104 bend cycles with the ing electrical continuity even after many thousands differential resistance extracted after each (figure cycles. The is was~60 times higher than ITO/PET. VI.14(b)) [1040]. ITO on PET showed a 95-fold The differential resistance as a function of angle increase in differential resistance, up to ~190 kΩ. After and bend radius was also investigated (figure VI.14(b)) such testing, the fatigued ITO showed a significant [1040]. All transfers demonstrated a low normalized difference in resistance between its bent and relaxed resistance even at bend angles ~100° (1.05, PMMA; state. The differential resistance of graphene, in both 0.94, UVA; and 1.04, HPL), almost two orders of mag- 131 2D Mater. 7 (2020) 022001 C Backes et al Figure VI.15. (A) Graphene crystal onCu grown as described in section VI.1.2. after exposure to ambient conditions for several days. (B) Schematic of dry transfer. (C) h-BN/Graphene/h-BN fabricated using contamination free dry transfer. Panel (B) is adapted from [832], with permission from AAAS. Panel (C) is adapted from [823], copyright American Chemical Society. nitude less than optoelectronically comparable ITO and baked for 10 min at 110 °C. Then, h-BN is exfoli- on PET (79.8) [1768]. The mechanical robustness ated on the PVA and large, thin and defect free flakes of transferred graphene is, regardless of the trans- are selected and cut out in a 5 mm  ×  5 mm square. fer method, much improved over ITO. UVA trans- The polymer stack can be lifted off the scotch tape fer showed consistently lower resistance for all bend using tweezers and placed on a 3 mm thick cush- angles, though HPL graphene repeatedly showed the ion of PDMS, resting on a glass slide. As depicted in lowest for bend angles  >40°. These bend angle experi- ­figure VI.15(b), the h-BN is aligned with a graphene ments suggest that UVA and HPL are somewhat more flake brought into contact and heated to 125 °C. Dur- robust than PMMA in high-bend-angle applications, ing the cool down (at ~75 °C), the polymer and Cu foil such as e-paper and wearable sensors. For bend radii are separated again, resulting in an h-BN flake covered Rb  =  1  −  5 mm Ref. [1039] reported that ITO nor- with graphene on the polymer stamp. Next, an arbi- malised resistance increased ~20/90 at Rb  =  5/1 mm. trary substrate, e.g. an h-BN flake on Si/SiO2, is aligned Conversely, the change in the normalised resistance of with the h-BN/graphene stack, brought into contact at all graphene transfers, including PMMA, were much 125 °C and peeled off the PDMS. Subsequently, stack smaller (PMMA: 0.5  −  0.9, UVA: 0.5  −  0.8, and HPL: of SiO2/h-BN/graphene/h-BN/PMMA/PVA is placed 0.2  −  1.2) and showed no measurable dependence in hot water, acetone and isopropanol, in order to dis- on Rb for the bend radii considered. UVA and HPL solve the polymers. Figure VI.15(c) depicts such an approaches seem to offer a viable solution to the lim- h-BN/graphene/h-BN sandwich on a SiO2 substrate. ited flexibility, with maintained conductivity, of trans- parent electrodes used in flexible electronics. VI.4. Graphene/PMMA sandwich structures T-peel tests were also conducted to explore the level Removing PMMA after transfer is not allways necessary of adhesion between graphene and flexible substrate as there are potential benefits of building graphene/ [1769]. A low peel strength represents weak adhesion PMMA heterostructures [1040]. PMMA is a colorless between laminate and substrate, while a high peel polymer, commonly used as a glass replacement strength suggests that graphene is strongly adhering. (acryl glass). Very thin, transparent PMMA could be As depicted in figure VI.14(c), UVA-transfers showed beneficial. The dielectric constant of PMMA is 2.6 the highest peel strength (4.39  ±  1.09 N m−1), fol- and comparable to fused silica (~3.5) [895], making lowed by HPL (0.60  ±  0.26 N m−1) and PMMA PMMA a decent insulating layer between multiple (0.44  ±  0.06 N m−1). graphene layers. PMMA has been used to prevent the contact of graphene with water during transfer, VI.3. Dry transfer using h-BN therefore enabling the use of water-sensitives substrate A contamination- free transfer process utilizes and in applications to fabricate flexible, air-stable, low- exfoliated hBN flakes to pick up graphene from Cu voltage GFETs [1043]. A polymeric bilayer consisting [823, 832]. Figure VI.16(a) is an optical microscopy of PBU (bottom) and PMMA (top) can be used as image of a graphene crystal on Cu, exposed to a supporting layer for graphene transfer without ambient conditions for a few days, resulting in the affecting its promising intrinsic characteristics [1770]. oxidation of the interface between Cu and graphene. The Dirac voltage and field-effect µ of GFETs with The oxidation of the interface does not only make the the polymeric layers kept the intrinsic properties graphene optically visible, but also decouples it from of graphene unaltered in air for more than 130 the substrate. days, whereas GFETs without polymers underwent For the transfer process, a polymer stack consist- significant and rapid degradation. High resolution ing of PDMS, PVA and PMMA is prepared. First, a positive tone electron beam resists, ARP-661.08 and glass slide is covered with scotch tape and drop coated ARP-672.08 (by Allresist(tm)), have been used to build with a 13% aqueous solution of PVA and baked at 95 PMMA/Graphene heterostructures [1044, 1045]. °C for 5 min. Subsequently, the slide is spin coated ARP-661.08 has molecular mass ~600 000 g mol−1 and with PMMA (4%, 50 k, in ethyl lactate) at 1000 rpm requires chlorobenzene as a solvent or a thinner, while 132 2D Mater. 7 (2020) 022001 C Backes et al Figure VI.16. Free standing 500 nm thick and 10 mm in diameter graphene/PMMA structure. The film is transparent and thus convenient for optical experiments. ARP-672.08 has molecular mass ~950 000 g mol−1 and temper­ature of PMMA from soft polymer to acryl requires anisole as a thinner. Both thinners consists glass is ~100 °C [904]. Similar recipes can be used of a benzene like structure and are small size weight to transfer the PMMA/Graphene hybrid or as a free ~110 g mol−1). The higher molecular mass of ARP- standing. Elastic and thin graphene/PMMA films can 672.08 can increase the resist film durability, which can be stacked [1044]. Figure VI.16 shows a free stand- be beneficial if the resist film needs to be transferred or ing 500 nm graphene/PMMA sandwich placed on an made free standing. When the PMMA/Graphene film Al aluminum plate with 10 mm diameter hole. The is deposited on a substrate, long PMMA molecules sample is prepared by using ARP-672.08 resist (4.5% can be reduced in size by electron beam lithography diluted with anisole). Similar 1 µm thick free standing [904, 1046]. This may open up an interesting route for films were fabricated using ARP-661.08 (4.5% diluted graphene patterning [1046]. with chlorobenzene) [1045]. PMMA layers with thickness ~600–800 nm have Multi-layered structures, where graphene is sand- been used in heterostructures [1044]. After spin coat- wiched between two PMMA layers, can be prepared ing, Cu/graphene/PMMA was baked at 60 °C for and used as EMI shielding layer by absorbing  >50% of 10 min to evaporate most solvents. The transition electromagnetic waves at 30 GHz [1044]. 133 2D Mater. 7 (2020) 022001 C Backes et al VII. Growth and transfer of other layered industry. Borazine is a volatile liquid with a relatively materials high vapor pressure (280 mbar at 25 °C), and is air and moisture sensitive [1050]. Therefore, its use requires a VII.1. Hexagonal BN specific set-up to have an accurate control of the gas BN, has been extensively studied as a dielectric material flow. To avoid gas flash in the CVD chamber and have for LM-based devices It has a large bandgap (>6 eV), good control the flow rate, borazine is maintained at atomic flatness, chemical inertness and small lattice 0 °C with a chiller to control the vapor pressure and a mismatch (1.8% larger) with respect to graphene system of mass flow controllers is used to transport the [1047]. H-BN is a natural dielectric to GRMs as SiO2 gas to the growth chamber. A cold trap was installed and Si3 N4 are to Si-based devices. The quality and to prevent the damage of the scroll pump with structural characteristics of the hBN film can impact potentially non-decomposed borazine. The process the performances of GRM-based devices. h-BN bulk follows the typical cycle of heating, pre-annealing single crystals, are very small, millimiters at most, and (30 min under H2 at 1000 °C) of the metallic substrate, do not satisfy manufacturing requirements. In order usually polycrystalline Cu or Ni, growth at 1000 °C, a to meet future, GRM based device manufacturing regulated pressure of 0.3 mbar, under 50 sccm of H2, requirements, it is necessary to develop processes with Pborazine  =  10−2 mbar, for a time between 10 min capable of providing large area single crystal or and 2 h, and slow cooling under H2 flow. polycrystalline films that can meet both device and The role of the underlying metal, Ni and Cu, high volume production needs. The growth process microstructure on the BN film morphology was stud- selection depends on the materials requirements. ied. ­Figure VII.2(a) shows an optical image of a BN Up to now most devices have used multilayer h-BN film grown on Cu and transferred onto a 90 nm SiO2/ with only a few using monolayer films. It is likely Si. Yellow and pink areas correspond to folded edges that many applications will require multilayer films of BN. The thickness of the film determined from the rather than 1L-h-BN, and some applications can use AFM profile is ~70 nm, with a 20 nm roughness ­(figure use polycrystalline films. The exact requirements will VII.2(b)). This behaviour is confirmed by TEM obser- guide the selection of the film growth/deposition vation of the sample (figure VII.2(c)). The electron process. diffraction pattern, figure VII.2(d), shows several six- Hexagonal hBN has been grown on metals, fold symmetry patterns and diffuse rings, indicating di­electrics and graphitic surfaces by various techniques that the film is crystalline and turbostratic. This result and under different conditions. The basic growth is in agreement literature in the same range of T and techniques used so far are the following: (1) growth borazine partial pressure [1051, 1052], for which a on dielectrics and graphitic surfaces by CVD and PA- non self-limited growth, with a turbostratic stacking MBE; (2) growth on metals by a diffusion and precipi- is expected. Due to its non-flatness and turbostratic tation as well a surface limited processes; (3) Growth structure, this hBN useful as a graphene substrate, as by ion beam assisted deposition and by physical vapor well as being good candidate to be used as dielectric deposition. Each of these have advantages and disad- compatible with GRMs as capping material in a device. vantages. When considering the use of any material Higher crystallinity is achieved if h- BN is grown for devices it would be desirable to deposit or grow on Ni [1053–1056] as compared to the growth on Cu. it at low T on a desired surface as done for mat­erials Few investigations were performed to elucidate the integrated in Si-based devices. In the case of GRMs it influence of the underlying Ni grain orientation on the has been difficult so far to integrate/grow/deposit the BN growth kinetics and coverage but there is hardly m directly on the desired surface in a fully integrated any record in terms of crystallinity, morphology and flow. This issue has necessitated the development of thickness of the BN film [1057, 1058]. In order to fully specific film transfer processes which will be discussed characterize the influence of the grain orientations a the end of this section. Therefore, the selection of a of the substrate on the BN film structure, combined growth process will depend on whether the materials SEM and TEM imaging and electron diffraction of requirements can be met to satisfy the specifications of both BN film and underlying Ni substrate are used. the individual application and on whether the material ­Figure VII.3(a) is a SEM image of the as-grown BN will be transferred or grown/deposited on the device on its substrate showing an area at the intersection of being fabricated. three differently oriented Ni grains. Their respective crystalline orientation is determined by Electron Back VII.1.1. Growth of h-BN on metal surfaces ­Scattering Diffraction (EBSD) in figure VII.3(b). Two h-BN was grown on Cu foils by LPCVD in a horizontal different surface aspects of the film can be identified reactor using ammonia borane as BN source and H2 as and are dependent on the Ni orientation. In the blue carrier gas [1049]. The growth of h-BN on metals can grain with a (1 1 1) orientation, the film surface struc- be carried out in a hot-wall low-pressure ~10−3 mbar ture is striated and nearly smooth, while for other ori- reactor, vertical or horizontal, as for figure VII.1. This entations (0 0 1), red grain, the surface appears to be process can be redesigned to a vertical furnace batch very rough. Direct transfer on the TEM grid enabled type reactor commonly used by the semiconductor TEM observations of the same area. The TEM image 134 2D Mater. 7 (2020) 022001 C Backes et al Figure VII.1. Schematic diagram of vertical CVD growth chamber for hBN growth on metal. Figure VII.2. (a) Optical image of BN film grown on Cu and transfer red onto 90 nm SiO2/Si, scale bar is 50 µm. (b) AFM profile showing the 90 nm thickness and the high roughness. (c) TEM image with scale bar 4 µm and (d) electron diffraction pattern showing the rough/turbostratic nature of the ML-h BN film grown on Cu with scale bar 5 nm. Figure VII.3. (a) SEM image and (b) corresponding EBSD mapping of BN on Ni. (b) TEM image of the BN film transferred polymer-free on holey amorphous carbon TEM grid. HR TEM images and corresponding electron diffractions of (c) and (d) a flat BN area and (e) and (f) a rough BN area with the schematic of the growth process: Van der Merwe for Ni(1 1 1) and Volmer Weber for Ni(1 1 0) and Ni(0 0 1). 135 2D Mater. 7 (2020) 022001 C Backes et al of figure VII.3(c) reveals the effect of Ni grain bounda- and is surrounded by ‘wire’ regions [1062, 1067] (figure ries on the BN film and confirms the presence of dif- VII.4(c)).The high T CVD growth of h-BN nanomesh ferent BN morphologies depending on the underlying on Rh(1 1 1) single crystal or Rh(1 1 1) single crystalline Ni grain orientation: flat and rough [1059]. As on Ni films on YSZ-buffered Si(1 1 1) wafers [1063] proceeds (1 1 1) hBN displays an almost perfect lattice match, as follows. The substrate is first cleaned in UHV by Ar+ the BN growth is highly oriented resulting in the for- sputtering (1.0 keV, 3.5 µA cm−2 sputter current) for mation of a flat h-BN as shown in figure VII.3(c). By ~30 min, and annealed to 700 °C. The substrate is kept recording electron diffraction patterns in different at 700 °C for 5 min, and oxygen (P(O2)  =  5–8  ×  10−7 areas as shown in figure VII.3(d), the single crystalline mbar) is dosed for 10 min. This is followed by a flash- character of the h-BN film over the whole Ni (1 1 1) annealing treatment up to 900 °C for 5 min. The sub- grain is assessed. For (0 0 1) Ni grain orientations, the strate is then cooled to RT. The whole preparation cycle h-BN growth domains nucleate at different places on (sputtering  +  annealing  +  O2 dosing  +  flash-anneal- the Ni surface and grow independently from each other. ing) must be repeated twice to ensure high quality of This process is responsible for the island morphology 1L-h-BN on Rh(1 1 1). At the third treatment cycle, no characterized by thickness variations from an island to O2 dosing is needed. The substrate is directly flashed another, discontinuity between the islands (darker and to 900 °C for 5 min after annealing. Then the sub- brighter contrast on TEM image figure VII.3(e)) and strate T is stabilized at 800 °C, and borazine is dosed at multiple orientation domains (several rotated electron P  =  5  ×  10−4 mbar for 5 min. Afterwards the sample is diffraction patterns, figure VII.3(f)). cooled to RT with an initial cooling rate of 10 °C min−1. This detailed analysis presented above on the The YSZ/Si(1 1 1) substrates may be coated with a large growth of h-BN on polycrystalline Cu and Ni provides number of unary [903] and binary transition metals, insights and guidelines for further optimization of the which allows the tailoring of the growth products. The h-BN conditions leading to planar and single crystal intercalation of other metals, such as like Au, was stud- layers needed for many applications. ied [1069]. On the binary transition metal PtRh it was The growth of h-BN at atmospheric pressure CVD found that the nanomesh unit cell may be tuned from (APCVD) may reduce the production costs of h-BN. 3.2 to 2.7 nm when increasing the substrate lattice con- Ref. [1062] synthesized h-BN by APCVD on poly- stant from Rh to PtRh [1070]. crystalline Ni. Depending on growth conditions, the thickness of the obtained h-BN film is between ~5 VII.1.2. Growth of h-BN on graphitic and dielectric and 50 nm with grain size limited by the size of the Ni surfaces single crystal grains. The controlled growth of h-BN It would be advantageous to to grow h-BN on dielectrics on polycrystalline Pt foils by APCVD with ammonia or graphitic surfaces in the absence of a metal catalyst borane as the precursor was also reported [1061]. 1L, to satisfy device integration requirements for many 2L and FL-h-BN domains and large-area films were applications. Direct growth of h-BN layers on graphite selectively obtained on Pt by changing the concentra- substrates was achieved using high T plasma-assisted tion of ammonia borane. 1L and 2L h-BN obtained are MBE [1071]. The h-BN grown on HOPG is hexagonal uniform with high quality and smooth surfaces. with no chemical intermixing with the graphitic carbon In principle the quality of h-BN can be significantly layers. This processes seems to provide high qualtiy improved by growing on single crystal metals. h-BN commensurated growth of h-BN on graphite. on Rh(1 1 1) (h-BN nanomesh) was first reported There are reports on the direct growth of [1062]as well as on Rh(1 1 1)/YSZ-buffered/Si(1 1 1) Graphene/h-BN heterostrutures. Ref. [1072, 1073] [1063]. This substrate formation strategy may also be took advantage of the solubility of C in Ni to grow a considered as a model system for growth of 1L- h-BN graphene/h-BN heterostructure by a combination of on transition metals (catalyst) surfaces at T  <  1000 K. C precipitation and MBE of h-BN on Ni/MgO(1 1 1). BN nanomesh has been extensively studied, starting The growth process was performed first by saturat- from pure surface science [1062] to applications as ing Ni(1 1 1)/Mg(1 1 1) with C, then growing h-BN on electrodes in electrochemistry [1064, 1065] and exfo- the exposed surface at high T (>730 C). The process liation [1066]. It was used as a template for molecular can yield FLG/FLG-hBN heterostructures. The great- assemblies in vacuum [1067], and is stable in air and in est difficulty is the control of thickness of both gra- liquids [1065, 1067]. It was furthermore used to study phene and h-BN at the 1L level. While h-BN may be the effects of hydrogen [1064, 1068] and metal interca- desired as a FL film, in most applications the graphene lation [1069]. layer number would have to be precisely controlled The h-BN nanomesh is a corrugated honeycomb (N  =  1  ±  0, N  =  2  ±  0, etc). superstructure and is the product of a high T CVD Ref. [1074] grew wafer-scale and high-quality growth using (HBNH)3 as a precursor. The highly regu- ML-h-BN on sapphire by LPCVD. Spectroscopic and larly corrugated structure with a lateral periodicity of microscopic measureements revealed that the h-BN 3.2 nm is determined by the mismatch between hBN film had a wrinkle-free surface and AA′ stacking, with and Rh. The unit cell consists of 13  ×  13 BN units on a single orientation and a well-oriented superstructure 12  ×  12 Rh atoms. It has one ‘pore’ with 2 nm diameter, with respect to sapphire. The electron transport in a 136 2D Mater. 7 (2020) 022001 C Backes et al Figure VII.4. (a) 800 °C hot 4 inch wafer in the growth chamber. (b) Low energy electron diffraction (LEED) pattern of h-BN/ Rh(1 1 1) at 60 eV kinetic energy. (c) STM image of the same surface with Ut  =  −1.2 V and It  =  0.5 nA. Figure VII.5. IBAD set up. GFET on h-BN/SiO2 was improved compared to SiO2. VII.1.3. Growth of h-BN by ion beam assisted deposi- Ref. [1076] claims that these h-BN films can be trans- tion on metals and dielectrics ferred to any substrate making LMhs and LM-based An alternative method of hBN growth is the physical electronic devices possible. vapor deposition (PVD) using a solid boron and h-BNfilms were also successfully grown on (0 0 nitrogen gas. This is not self-limiting, therefore 0 1) sapphire by metalorganic vapor phase epitaxy the h-BN thickness can be controlled through the (MOVPE) [1075]. These were grown using triethylbo- deposition time. This technique allows the deposition ron, B(C2H5)3, and ammonia (NH3) at various V/III of thick layers of amorphous BN or cubic BN with ratios, ranging from 210 to 2100. The films grown at different orientations and textures on almost any type V/III ratios above 1280 showed two XRDpeaks, one of substrate, both metallic or insulator. from the (0 0 0 2) plane and the other from the (0 0 0 Among the possible PVD techniques, ion beam 4) plane. These films also exhibited a Raman hBN peak assisted deposition (IBAD) is specially versatile at 1366 cm−1. In contrast, the films grown using V/III because it permits independent control of B and N below 640 revealed an XRD pattern consistent with a fluxes, and fine tuning of the ion energy. The technique turbostratic structure. relies on electron beam evaporation of boron in vac- 137 2D Mater. 7 (2020) 022001 C Backes et al 4 2 Photoemission B (1s) N (1s) hν= 620eV 3 Intensity (x10 cps) 5 2 1 1 194 192 190 188 402 400 398 396 394 Binding energy (eV) Figure VII.6. XPS from a stoichiometric h-BN produced by IBAD. uum and the simultaneous bombardment with N ions, thickness of the film, its homogeneity, the interaction [1076–1080]. Figure VII.5, depicts a typical exper­ with the substrate, and roughness and surface termi- imental set up. BN is obtained in a vacuum by reaction nation of the host substrate. These parameters deter- between the evaporated B and the N+ 2 ions. By using mine best suited transfer for a particular growth on an this technique, one can control the quality, texture specific substrate as it rules the basic properties of the and thickness of the h-BN layer by proper selection of grown film, (mechanical, defect density…) and inter- growth parameters such as B/N impinging ratio, evap- action of film with growth surface and temporary sup- oration rate, deposition time, ion energy and substrate port. T. The composition of BN is adjusted by controlling the incoming B and N fluxes. The B/N ratio can then be VII.1.5. Wet transfer measured by EDX, XPS, as shown in figure VII.6, and Wet transfer includes chemical etching, and XANES, electrochemical delamination, aided or not by a Typical growth rates range from 0.1 to 10 nm chemical intercalation min−1. Since the growth is not self limited in this pro- cess the thickness is controlled by the various deposi- VII.1.6. Electrochemical tion variables including deposition time. The quality Using the bubbling method, Ref. [1061] achieved of BN is determined principally by the substrate T and nondestructive transfer of h-BN from Pt to arbitrary the ion energy, which controls the orientation of the substrates and the repeated use of Pt for h-BN growth, basal planes, the defect density and surface rough- which not only reduces environmental pollution but ness. A review of these parameters with regards to the also decreases the production cost of h-BN. formation of high quality h-BN: orientation, crystal- A PMMA layer is first spin coated on BN/Cu. For lite size and density of defects is given in Ref. [1081]. electrochemical delamination process, an aqueous The deposited materials have been thoroughly char- solution of KCl (1M) is employed as electrolyte, with acterized using spectroscopic analytical tools [1079, a glassy Pt anode and the PMMA/BN/Cu cathode is 1082–1085]. One of the most important properties is negatively polarized (5 V). The formation of H2 bub- the extreme flatness, with a surface roughness ~0.1 nm bles at the BN/Cu interface leads to the separation of for any film thickness. PMMA/BN and Cu substrate. The film is then rinsed with DI water, deposited on the target substrate and, at VII.1.4. Transfer of h-BN 20 °C/10 min, heated to 120 °C to remove excess water The goal of the transfer process is to get GRMs on and to stretch it. After one night, the PMMA film is suitable host substrates. removed with acetone at RT. An example of BN trans- Transfer processes are based on the use of a tempo- fer from Cu onto SiO2/Si and TEM grid is shown in rary support, most of the time a polymer, as described figures VII.7(a) and (b). Sizes 1  ×  1 cm2 are achieved. in section VI for graphene. Several methods have been developed: (1) wet transfer, involving the use of a liq- VII.1.7. Chemical intercalation uid aqueous solution, and (2) dry transfer, relying on The previously described electrochemical delamination solid adhesive layers. The second approach ensures the does not work on Ni due to the stronger BN/Ni absence of water between layers, which can be detri- interaction compared to BN/Cu TOA-assisted chemical mental to device properties. Table VII.1 scrutinizes delamination, described in section VI for graphene, was advantages and drawbacks or difficulties inherent to tested and new strategies are required in order to remove both kinds of transfer processes. the ions residues. In this particular case, chemical etching Several parameters can impact the feasibility and (see section VI.1) is used to remove the Ni substrate, quality of transfer, including the roughness and the without any polymer [763]. The Ni substrate is etched 138 2D Mater. 7 (2020) 022001 Table VII.1. Comparison of the various procedures developed to date for transferring CVD grown BN. Special Re-use of the Method Metals Potential contamination equipment substrate Time Key parameters Other WET Chemical etching Cu [1049, 1087], Ni [1054, Etchant, polymer, metal resi- No Several Etchant nature and 139 1058, 1062] dues, interfacial water hours concentration Electrochemical de- Cu [1088], Pt [1061, 1189], Electrolyte, polymer, metal 5 V DC Yes (smoothed) A few Voltage, electrolyte na- Possible damage due to gas bub- lamination (ECD) [108, 1090] Fe [1091] residues, interfacial water generator minutes ture and concentration bles produced at the interface TOA treatment  +  ECD Rh [1066], Ni ions, electrolyte, polymer, 3 electrodes Yes ~20 min Voltage, electrolyte na- metal residues, interfacial water set-up ture and concentration Lift-off transfer Cu [1092] NaOH residues, polymer, metal Yes 2–3 h at Interface oxidation residues, interfacial water 60 °C DRY PMMA/PVA [832] Potentially any (depends on PMMA Heating plate Yes (if no BN ~2 h PMMA adhesion Semi-dry transfer: water at the the interaction BN—substrate) remaining) very end Scotch tape Adhesive residues A few Scotch tape choice Difficult release/deposition minutes C Backes et al 2D Mater. 7 (2020) 022001 C Backes et al Figure VII.7. Examples of large areas BN film transferred on (a) 90 nm SiO2/Si, (b, c) TEM grid, fromCu and Ni growth substrates. Figure VII.8. (a) Two zone furnace for TAC of Mo to MoS2. (b) Photographs showing different thicknesses MoS2 films on fused quartz (top) and SiO2 (bottom) substrates. Adapted from [1093], with permission from Elsevier. with a commercial solution (TFB Nickel Etchant, VII.2. Layered semiconductors Transene) for one night. The BN film is rinsed with VII.2.1. Thermal assisted conversion DI water, deposited on the target substrate and slowly While CVD provides a feasible route towards heated to 120 °C to remove excess water. This polymer- deposition of GRMs, other approaches offer simpler, free technique does not create any contamination scalable methods to grow thin films, at the expense of andpreserves the film integrity. An example of BN crystallinity. Thermally Assisted Conversion (TAC) transfer from Ni substrate onto SiO2/Si and TEM grid [1093] of transition metals is one such technique. is show in figures VII.7(c) and (d), respectively. A size Figure VII.8 a shows a typical TAC set up. Thin films ~1  ×  1 cm2 is achieved, and a full TEM grid coverage. of the appropriate transition metal are deposited via sputter coating or electron beam evaporation onto VII.1.8. Dry transfer appropriate substrates (often SiO2 or sapphire) and Dry transfer uses PVA/PMMA or scotch tape as the then exposed to chalcogen vapour at elevated T (400 adhesive material. Most of the time using an adhesive –1000 °C). Controlling the thickness of the initial strong enough to pick up the film is necessary but not metal layer allows some control of the thickness sufficient, as the redeposition on the targeted substrate of the final TMD film, but achieving consistent 1L may be hard to achieve. If the adhesive material is remains a problem. Additionally, TAC films are transparent, it can be done under a microscope, which polycrystalline in nature [1093, 1094]. Thinner films permits to target precisely the area to pick up and preferably align parallel to the substrates while thicker where to deposit it. films form vertically standing grains, with crossover Ref. [1086] proposed a method for wafer scale dry point roughly at 10 nm [1095]. These features limit transfer of thick h-BN films and other LMs and LMHs. the practical usage of TAC in certain applications The is based on controlled crack propagation between such as electronic devices, as the abundance of grain thick layer and saffire growth substrate. For that pur- boundaries compromises the electrical properties. pose a Ni film was deposited onto the h-BN film. An However, they are suited to other areas, such as gas adhesive material is then placed on top of the Ni layer sensing and electrochemical applications [1093, 1094, and the Ni and h-BN film are detached from the saffire 1096–1099]. wafer. Then another Ni layer is grown on fresh h-BN A variety of TMD films via TAC have been grown surface after the removal of the saffire wafer. Another and employed across a wide range of applications adhesive tape is then placed onto the new Ni film and [1093, 1094, 1096–1099]. Thin films of Mo or W of repeated splitting of the Ni-h-BN-Ni proceeds up to varying thickness (typically 0.5–20 nm) were depos- reaching a single layer h-BN. ited on to SiO2 and heated to 750 °C in a quartz tube 140 2D Mater. 7 (2020) 022001 C Backes et al Figure VII.9. (a) and (b) Raman spectra and (c) PL of TAC MoS2 grown from Mo films of differing thickness. (d) Mo 3d peak for MoS2 grown from 1 nm Mo. (e) W 4f peak for WS2 grown from a 20 nm W film. Adapted from [1093], with permission from Elsevier. (f) HRTEM image of WS2 polycrystalline film. Adapted from [1100], copyright American Chemical Society. furnace under Ar flow (150 sccm), see figure VII.8(b), features consistent with a M:S stoichiometric ration of and annealed under these conditions for 30 min. Fol- 1:1.9, indicating signifying an almost ideal MoS2 film. lowing this, S powder was heated to 113 °C in a sepa- A small amount (<5%) of surface oxide was also pre- rate heating zone upstream of the samples to generate sent in the form of MoO2 and MoO3. The W 4f core S vapour. Sulphurization was carried out for 20 min, level is typical of WS2 with minimal oxides or unre- after which the supply of S vapour was shut off and the acted S on the surface. samples were annealed for a further 20 min. They were One strength of TAC production of TMD films then cooled to RT by switching off the heating over is the ability to pre-pattern the starting metal prior several hours under Ar flow. to conversion. This eliminates subsequent process- Raman spectra and PL (see section IX.2 for an ing steps which would be necessary if one were to introduction to these techniques) of MoS2 films grown use lithography to pattern a pre-existing TMD film. via TAC from different initial thicknesses (0.5–20 nm) MoS2 devices were fabricated by sputter-coating of of Mo are shown in figure VII.9(a)–(c). Of note is the Mo through a shadow mask, conversion to MoS2 decreasing separation between the characteristic A1 and then contacting with a second metal deposi- and E′ peaks and increasing PL intensity with decreas- tion with a second shadow mask defining the con- ing film thickness. This might suggest that, although tacts [1093, 1096]. A photograph of the resultant these samples are polycrystalline few layer films, thin- device is shown in figure VII.10(a). This device was ner samples exhibit significantly more monolayer employed as a gas sensor for NH3 (figure VII.10(b)), character [1093]. This is of particular interest if one whereby adsorption of NH3 onto the MoS2 film sur- wishes to tune the characteristics of such TAC films to face causes a change in resistance, figure VII.10(c). suit a particular application. Figures VII.9(d) and (e) This was used to generate a calibration curve, dem- show the Mo 3d and W 4f spectral regions for MoS2 onstrating the sensor response of the device in and WS2 films. The Mo 3d core level exhibits spectral figure VII.10(d) [1096]. 141 2D Mater. 7 (2020) 022001 C Backes et al Figure VII.10. (a) MoS2 device with Ti/Au contacts. (b) NH3 gas sensing device. (c) Sensor response plot showing percentile resistance change versus time for the MoS2 film with a bias voltage of 0.5 V. NH3 exposures range from 300 ppb to 2 ppm. (d) Resistance change (black solid circles) and signal-to-noise ratios (gray open boxes) as a function of NH3 concentration. Inset: full calibration curve ranging from 300 ppb to 200 ppm. Adapted from [1096], © Wiley-VCH Verlag GmbH & Co. KGaA. Heterojunction devices with large (>tens mm) low thermal conductivity [1112], phase-change mem- active layers were fabricated from TAC films for a range ories [1122], and high-gain photodetectors [1123]. of applications [1099–1103]. The grown layers by physical vapour transport are β-In2Se3. The presence of β-In2Se3 is confirmed using VII.2.2. Physical vapour transport a combination of Raman spectroscopy XRD, elemen- InSe is attracting increasing interest for its desirable tal analysis and PL. In general, Raman spectroscopy electronic and optical properties [1104]. More can help identify the crystalline phase of In2Se3 due to generally, compounds with different stoichiometry, the presence of distinct vibrational modes for the dif- InxSey, and their different polytype phases, e.g. α, ferent phases [1120]. β-In2Se3 flakes grown by physical β, γ, have physical properties relevant for several vapor transport are optically active with a RT PL emis- applications in electronics, thermoelectrics and sion that is blue-shifted relative to that of exfoliated optoelectronics [1105–1108]. They have band gaps layers of α-In2Se3 and γ-InSe [1117, 1124]. Also, the in the NIR to visible range, 1.25–2 eV, high µ at RT grown layers have distinct Raman lines [1117, 1120, (>0.1 m2 V−1 s−1) and an interesting ‘Mexican hat’ 1121]. More importantly, the RT PL peak of β-In2Se3 valence band energy dispersion [1109–1113], sensitive is strongly sensitive to the thickness due to quantum to the layer thickness [1110, 1111, 1113] and/or an confinement of carriers by the boundary of the layers externally applied electric field [1109]. and light electron mass [1117]. Several methods have been employed for the syn- The synthesis of the In2Se3 layers by physical vapor thesis of In-Se compounds with different stoichiomet- transport involves three separate steps ric ratios, such as InSe, In2Se3, and In4Se3, and their For the growth of β-In2Se3 layers by PVT, ground polytype phases [1114]. The synthesis of 1L was dem- powder of Bridgman-grown [1125] γ-polytype InSe onstrated for the α-[1115, 1116] and β-phase [1117, was used as source material, see figure VII.12(a). Bulk 1118] of In2Se3, and for γ-InSe [1119, 1120]. Bridgman-grown γ-InSe crystals were ground into Figure VII.11 shows the crystal structures of α- and fine-grained powder using a pestle and a porcelain β-In2Se3 and γ-InSe. Electronic, thermal and opti- mortar. Then the powder was loaded into a quartz cal properties of these flakes are still largely unknown beaker and mixed with ethanol (96% purity) in a 1:4 [1118, 1121], although crucial to the implementation weight ratio of InSe:ethanol, which acted as a dispers- in several technologies, including molecular sensing ing liquid environment. The obtained suspension was [1106], high-performance thermoelectric devices with sonicated for 18 h at RT (f   =  21.2 kHz and P  =  100 142 2D Mater. 7 (2020) 022001 C Backes et al Figure VII.11. Side view of the crystal lattice and unit cell of γ-InSe (left), α-In2Se3 (middle) and β-In2Se3 (right). Blue and red spheres correspond to Se- and In-atoms, respectively. The primitive unit cell of γ-InSe contains three layers, each consisting of four closely-packed, covalently bonded, atomic sheets in the sequence Se–In–In–Se. For both α- and β-In2Se3 phases, the unit cell contains three layers, each consisting of five closely-packed, covalently bonded, atomic sheets in the sequence Se–In–Se–In–Se. In α-In2Se3 the outer Se-atoms in each layer are aligned, whereas in β-In2Se3 they are located into the interstitial sites of the Se-atoms in the neighboring layers, thus leading to a smaller volume in β-In2Se3. (a) (b) furnace Ar pump γ-InSe fine-grained powder (c) 600 (d) α ( o C) 500 400 Temperature 300 200 Si/SiO2 γ -InSe o α = 3 C/min 100 substrate powder α 0 60 120 180 240 300 360 420 480 540 Time (mins) Figure VII.12. (a) Optical image of fine-grained powder of γ-InSe. (b) Tube furnace system used for the growth of In2Se3. (c) The source material and the substrate in the quartz tube. (d) Program profile of the furnace heater. dB). The resulting solution was dried in a porce- The β-In2Se3 layers were grown on different sub- lain beaker at RT. The dried powder was rinsed with strates such as Si/SiO2, mica and HOPG [1117]. Prior DI water five times and finally dried at 50 °C for 24 h. to growth, the Si/SiO2 substrate (7 mm wide and 3 cm Then the powder was ground again using a pestle and long) was cleaned in hot acetone (T  =  60 °C) and a porcelain mortar. Finally, the InSe powder was sieved ultrasonicated in methanol and isopropanol (10 min (200  ×  200 µm2). in each solvent) and then cleaned by oxygen plasma at 143 2D Mater. 7 (2020) 022001 C Backes et al (a) (b) (c) 100 µm 5 µm 100 µm 20 µm Figure VII.13. Optical micrographs of β-In2Se3 layers grown on (a) SiO2/Si, (b) mica, and (c) graphite (T  =  600 °C, t  =  9 h and Ar flow rate ~150 sccm). Different colors correspond to different thicknesses of the layers. This ranges from hundreds of nanometers down to 2.8 nm. Adapted from [1117], © IOP Publishing Ltd. (a) (b) (c) 10 µm 10 µm 10 µm Figure VII.14. Optical micrographs of β-In2Se3 rods (a) grown on mica at T  =  620 °C for t  =  9 h under an Ar flow rate of 150 sccm, (b) grown on mica at T  =  570 °C for t  =  6 h under an Ar flow rate of 150 sccm, and (c) grown on Si/SiO2 at T  =  600 °C for t  =  6 h under an Ar flow rate of 170 sccm. T  =  100 °C for t  =  10 min to remove ambient adsorb- slightly facetted films with lateral size between 1 and ates from its surface. The cleaned substrate was imme- 15 µm (figure VII.13(a)), while highly facetted hex- diately loaded into the tube furnace for the growth. agonal films with lateral size ~100 µm grew on mica, Mica and graphite were cleaved prior to growth and see figure VII.13(b). On graphite, films with arbitrary the fresh surface was used as substrate. shape grew (figure VII.13(c)) with preferential growth A tube furnace system was used to grow the at step edge bunches on the substrate. The thickness, In2Se3 layers (figure VII.12(b)). The system com- t, was measured by AFM. It ranges from hundreds of prised a Carbolite furnace, a 1 m long quartz tube, a nanometers down to 2.8 nm on Si/SiO2. rotary pump and an Ar flow controller (Hastings For the same growth conditions (T  =  600 °C, mass flow controller). γ-InSe powder (~300 mg) was t  =  9 h and Ar flow  =  150 sccm) β-In2Se3 layers were loaded into a ceramic boat and placed at the center grown on mica with layer thickness between 80 and of the tube furnace. The substrate was placed in the 5 nm and on graphite from hundreds nanometer down downstream side 6–10 cm away from the source mat­ to 10 nm, (see figure VII.13(c)). Figures VII.14(a)–(c) erial, see figure VII.12(c). The pressure was lowered show some examples of growth under different con- ~ 6  ×  10−3 mbar using the rotary pump. Then an Ar ditions. ‘Hay’-like growth was obtained on mica at gas was intoduced at a rate of 150 sccm for 2 h provid- T  =  620 °C for t  =  9 h for an Ar flow of 150 sccm, see ing P  =  1.6 mbar. Thereafter, the system was heated figure VII.14(a). Nano rods were grown on mica and from T  =  22 to 600 °C at a rate of 3 °C per minute Si/SiO2, respectively, at 570 °C or under 170 sccm Ar and kept at 600 °C for 6 h or longer (up to 9 h), see fig- flow rates (figures VII.14(b) and (c)). The growth of ure VII.12(d). The vaporized In- and Se-atoms were different polytype phases of In-Se is highly depend- carried by the Ar gas and deposited on the substrate. ent on the growth T [1114]. This results in a significant The system was then allowed to cool naturally to RT. advantage of this method since, due to the T gradient Typically, the source material can be reused for a sec- within the tube furnace, it is possible to grow different ond cycle of growth. phases and stoichiometries, including γ-InSe, during a Figures VII.13(a)–(c) show the as-grown β-In2Se3 single run [1120]. layers on Si/SiO2, mica and graphite, respectively. The The electronic and vibrational properties of the as- growth on Si/SiO2 substrates produces near-circular, grown films of different t were compared with those 144 2D Mater. 7 (2020) 022001 C Backes et al (a) β - In2Se3 (b) A1 α -In 2 Se 3 Normalized PL Intensity T = 300K γ -In Se β -In 2 Se 3 Normalized Raman Int. t = 2.8 nm 10 nm T = 300K 22 nm 37 nm 1.2 1.4 1.6 1.8 70 nm Energy (eV) 100 150 200 1.7 (c) β -In2Se3 α - In Se Norm. PL Int. β -In 2 Se 3 Normalized Raman Int. 2 3 6 nm 1.6 50 nm 11 nm E ( eV ) 18 nm 1.5 1.2 1.5 1.8 Energy (eV) 27 nm 1.4 α -In 2 Se 3 48 nm 1.3 γ -InSe 61 nm 100 150 200 0 20 40 60 80 Raman Shift (cm ) -1 t (nm) Figure VII.15. (a) Normalized Raman spectra of as-grown β-In2Se3 layers (top) and exfoliated flakes of α-In2Se3 (bottom) at T  =  300 K (P  =  0.1 mW and λ  =  633 nm). (b) Normalized µPL spectra of bulk γ-InSe, β-In2Se3 and α-In2Se3 at T  =  300 K (λ  =  633 nm and P  =  0.1 mW). (c) Measured and calculated dependence of the peak energy, E, of the PL emission on layer thickness t of as-grown β-In2Se3 (black dots) and exfoliated Bridgman-grown α-In2Se3 (blue dots) and γ-InSe (magenta dots) flakes. Inset: Normalized PL spectra of β-In2Se3 with t ~ 50 nm and 6 nm at T  =  300 K. Adapted from [1117], © IOP Publishing Ltd. of thin α-In2Se3 and γ-InSe layers exfoliated with tial of infinite height, i.e. E = E3d + π 2 2 /2t 2 µc , adhesive tape from Bridgman-grown ingots [1125, where E3d is the band gap for the bulk and µc is the 1126]. The α- and β-phases of In2Se3 are ­characterized electron–hole reduced mass for motion along the by distinct Raman modes [1117, 1120, 1121]. For the c-axis. For β-In2Se3, the best fit to the measured E ver- as-grown β-In2Se3 films, the Raman peaks are cen- sus t, gives µc  =  0.04 me and E3d = 1.428 eV, where tered at ~110, 175, and 205 cm−1 at T  =  300 K (­figure me is the electron mass in vacuum. These values differ VII.15(a)), corresponding to the intralayer vibra- from α-In2Se3, i.e. µc  =  0.08me and E3d = 1.390 eV, tional A1-modes (110 and 205 cm−1) and the Eg-mode and γ-InSe, i.e. µc  =  0.05me and E3d = 1.250 eV. (175 cm−1) of bulk β-In2Se3 [1118, 1127]. For exfoli- The comparison of the PL peak energies for the ated flakes of α-In2Se3 the Raman modes are narrower β-In2Se3 layers with those for α-In2Se3, and γ-InSe and centered at ~104, 181, and 200 cm−1 at T  =  300 K. [1116, 1119, 1124] indicates distinct spectral ranges for β-In2Se3 flakes exhibit RT PL. RT normalized PL spec- these In-Se compounds and a wider spectral tunabil- tra of as-grown β-In2Se3 with t ~ 50 nm are compared ity for β-In2Se3 and γ-InSe, among the largest within with bulk α-In2Se3 and γ-InSe in figure VII.15(b). the wide family of LMs. The larger quantum shift and The PL emission for the β-In2Se3 films is peaked at lighter electron–hole reduced mass for the β-phase higher energy compared to α-In2Se3 and γ-InSe, and compared to the α-phase are assigned to the closer it blue-shifts with decreasing layer thickness (inset of spacing of the layers in the β-phase (Δc/c ~  −1.5%) ­figure VII.15(c)). (figure VII.11), leading to a stronger inter-layer cou- Figure VII.15(c) shows the dependence of the PL pling. On the other hand, the larger value of E3d in peak energy, E, at RT on t, as obtained from several PL β-In2Se3 compared to α-In2Se3 cannot be accounted and AFM studies of as-grown β-In2Se3 layers (black for by a simple argument and may involve a redistribu- dots) and α-In2Se3 and γ-InSe flakes [1124] exfoliated tion of electronic charge between the Se- and In-atoms from Bridgman-grown crystals [1117, 1125] (blue in neighboring vdW layers [1114]. This may be also and magenta dots). The PL peak energy undergoes a affected by the presence of crystal defects, such as In- blue-shift to higher photon energies with decreasing and Se-vacancies (VIn and VSe), which tend to form in layer thickness consistent with quantum confinement In2Se3 due to the misvalency between the III- and VI- of photo-excited carriers along the c-axis [1117]. This atoms [1114]. Vacancy ordering and bond relaxation energy shift can be modelled by a quantum well poten- due to vacancies are expected to modify atomic orbit- 145 2D Mater. 7 (2020) 022001 C Backes et al Figure VII.16. (a) Schematic set-up for CVD of MoS2. (b) mechanism by which MoS2 is deposited on the target substrate. Adapted from [1098] CC BY 4.0. als and increase the band gap of bulk layers [1128], and 20 min before annealing at 750 °C under just Ar flow may play a more important role in LMs [1129]. (150 sccm) for a further 20 min. Imaging of the resultant deposit shows a film VII.2.3. CVD growth consisting of single crystal domains joined together As is the case with graphene, CVD of LMs is a to form a continuous film (figure VII.17(a)). Raman synthesis route which offers a compromise between spectroscopy shows the A1 and E′ peaks characteristic film quality and large area coverage and uses techniques of MoS2 [1098] (Figure VII.17(b)), while the spectral compatible with CMOS processing. CVD of LMs contributions to the Mo 3d region of the XPS spectrum other than graphene typically requires two precursor are consistent with that of MoS2 with a small amount materials which adds to the complexity of depositing of oxide present [1098] (Figure VII.17(c)). The A1 to uniform, 1L films. Precursors can be sourced as solids E′ peak separation ~18 cm−1 and the strong PL (Figure which sublime under heating with these vapours then VII.17(d)) are evidence for 1L-MoS2 [1098]. reacting to complete the CVD process, or as gaseous. CVD of MoS2 on c-plane sapphire using solid inor- Gaseous precursors allow much greater control of ganic precursors [1130–1132] has enabled an efficient parameters, such as flow rate, but are often toxic or route toward producing high quality 1L-MoS2 films highly flammable, such as H2S and H2Se [1008]. with lateral dimensions (>cm2) beyond that of the mechanically exfoliated flakes (less than 100 µm) with VII.2.4. Solid precursors electronic properties matching those of mechanically Figure VII.16 describes the microreactor set-up for exfoliated samples [1133]. CVD of MoS2 in which flakes of MoO3 and S powder However, solid precursors such as S and MoO3 are used as solid precursors to supply Mo and S, exhibit negligible vapor pressure at RT. This leads to respectively. LPE MoO3 flakes in IPA were drop-cast the requirement of positioning the precursors inside on a SiO2 chip placed face-down on top. This assembly the heated zone of the reaction chamber to reach T usu- was heated to 750 °C in a quartz tube furnace under ally above 200 °C for S and 500 °C for MoO3 with fixed Ar flow (150 sccm) with S powder heated to 120 °C in distances from the position of the substrates as shown a separate heating zone upstream of the microreactor. in figure VII.18 with unseeded or seeded substrates. The samples were exposed to the resulting S vapour for Since the available composition of the precursors in 146 2D Mater. 7 (2020) 022001 C Backes et al Figure VII.17. (a) SEM image showing continuous MoS2 film. Scale bar is 2 µm. (b) Average Raman spectrum measured across a 20  ×  20 µm area of the film. (c) Mo 3d core level showing contributions consistent with MoS2. (d) PL measured across a 1L region (green) and 2L island (pink). Adapted from [1098]. b c furnace 600 - 700 °C a Ar d S MoO3 c-sapphire Figure VII.18. (a) CVD setup for deposition of 1L MoS2 using solid precursors, MoO3 and S. (b) MoS2 deposited on c-plane sapphire. (c) 1L-MoS2 covering c-plane sapphire without complete coverage, scale bar is 20 µm. (d) 1L MoS2 with complete coverage, bar is 10 µm. Adapted from [1130], copyright American Chemical Society. the atmosphere of the reaction chamber depends on VII.2.5. Gas precursors the evaporation from the time-dependent masses and In order to achieve a better control over the gas phase surface areas of the solid precursors, it is challenging to composition of the precursors in the chamber, gas consistently control the flow of the precursor vapours sources with high RT vapor pressures (>0.1 mbar) reaching the substrates over the period that affects the can be used, which allow the precursors to be stored uniformity and stoichiometry of the resulting film. in a T-and pressure-controlled bubbler, outside the There is also a practical issue of over-consumption heated zoned of the reaction chamber where the of the precursor materials that may lead to unwanted vapour pressure and mass flow rate can be controlled deposition of residual by-products on the substrates independently, see figures VII.19 and VII.20. and walls of the reaction chamber which also hinders A large range of gas precursors containing Mo the growth uniformity and technological feasibility. and S were developed for semiconductor and cata- 147 2D Mater. 7 (2020) 022001 C Backes et al 6 5 vapor pressure (mbar) 4 3 2 1 0 0 10 20 30 40 50 60 temperature (°C) Figure VII.19. Equilibrium vapor pressure of Mo(CO)6 as a function of T based on the measured thermodynamic constants in Ref. [1134]. lyst industries. They include Mo(CO)6 [1135], MoCl5 ability. Ref. [1140] developed a method to grow a wafer [1136], (NH4)2MoS4 [1137] for molybdenum and H2S scale polycrystalline 1L-MoS2 films on SiO2/Si with [1136], diethyl sulphide [1138] for S. Often these tech- Mo(CO)6 and (C2H5)2S (Diethyl sulfide; DES) with niques using gas precursors are named Metal organic relatively low growth T (~550 °C), albeit with a rather CVD, MOCVD, even if the precursors cannot be con- long growth time to achieve a continuous film (~26 h) sidered proper metal organics. in low vacuum conditions (7.5 Torr). Ref. [1141] pro- Significant progress has been made using MOCVD duced average-thickness controlled MoS2 1L-to-ML in the growth of large-area, reproducible, high-qual- on cm scale SiO2/Si and sapphire after a thermody- ity TMDs. However, we still have to overcome the namic modelling of vapor phase equilibrium reac- challenge of single crystal continuous film growth tion of the precursors Mo(CO)6 and H2S. The growth of MoS2 by MOCVD in a reasonable growth time pressure (~1 atm to 1 Torr) and T (400 –850 °C) were (~1 h). Basic nucleation and growth processes will considered as a main factor in determining the range have to be developed and understood in more detail of growth parameters for the synthesis of high quality before this problem can be solved. More work needs MoS2 films of a good stoichiometry without any car- to be done on understanding the basic growth mech­ bon residues originating from the carbon-containing anisms which may include in situ growth observations Mo precursor. to determine the kinetics and rate limiting steps for Mo(CO)6 and H2S were used in Ref. [1142] as systematic MOCVD control of film morphology and model precursors. The choice was based on the fact micro-nanostructures. Ref. [1139] reported the diffu- that both are readily available, well-established high sion-controlled growth of WSe2 on sapphire, where a purity (>99.9%) MOCVD sources with know ther- two-step process was used to nucleate WSe2 followed mochemical properties [1141, 1143–1147]. by growth from the seeds using process conditions to In designing how the sources are delivered, a minimize additional nuclei while growing single lay- solid precursor such as Mo(CO)6 is to be loaded into ers. This is an example of many growth studies that an inert, stainless-steel bubbler immersed inside a T need to be performed in order to achieve large area controlled bath. Ideally, the evaporation of Mo(CO)6 single-crystal TMDs. While this study has shed some is considered to be fast enough to always completely light on WSe2 growth process, single 1L crystals across saturate the bubbler with equilibrium vapor pressure the whole wafer have not been achieved yet. Another (figure VII.19) which would determine the concen- growth process that could be used is atomic layer tration of the precursor exiting the bubbler with the epitaxy, however at this time there does seem to be a carrier gas. However, the common problem with solid significant effort in this direction, perhaps because a precursors is that the actual vapor transported out of proper substrate has not been identified yet. the bubbler quickly decreases over multiple growth So far, the growth of 1L-MoS2 by MOCVD was runs because of increased dead volume of the precur- demonstrated by a few groups with a varying degree of sor and decrease in the available precursor surface area coverage, crystallinity, thickness uniformity, and scal- for an efficient evaporation (at least ~1 µl (min · g)−1). 148 2D Mater. 7 (2020) 022001 C Backes et al furnace 750 - 1000 °C Mass. spec Ar/H2 H2S Ar Mechanical Bubbler pump Mo(CO)6 -25 to 30°C sapphire substrates Vapor Pressure exhaust 0.1 mbar to 1 mbar Figure VII.20. MOCVD setup, with a reverse bubbler configuration for Mo(CO)6. In order to tackle this issue, a reverse bubbler configu- cylinder and near the CVD furnace are installed. They ration (see figure VII.20) and packing of precursor are connected to a sound alarm to alert the appropriate with an inert silica beads ~2 mm are used to increase emergency services immediately. In addition, experi- effective dwell time of the carrier gas and the precursor ments thorough He leak test of the assembled CVD surface area [1148] by an order of magnitude. These system need to the conducted before performing the approaches have been previously employed for solid experiments involving H2S to prevent anyone in the metalorganic precursors to ensure efficient evapora- work area from exposure. tion [1148]. The evaporation rate of the Mo(CO)6 pre- The CVD system used, as illustrated in cursor should now be fast (at least ~1 µl (min · g)−1) ­figure VII.20, consists of a horizontal hot wall filament enough to assume an instantaneous phase equilibrium three-zone, quartz tube system (Carbolite HZS 12/- between gas and solid phases of Mo(CO)6 within the /600) commonly used for laboratory scale CVD syn- bubbler while the Mo(CO)6 precursor vapour is car- thesis as one of the most economical ways to produce ried into the reaction zone of the system by an inert cm scale thin films. The quartz tube is coupled to KF- gas (specifically Ar). In this case, the partial pressure standard (DIN 28403, ISO 2861) high-vacuum com- of Mo(CO)6 exiting the bubbler can be estimated by patible flanges which can be evacuated by a mechani- the vapour pressure of Mo(CO)6 at the T of the bub- cal pump to ~10−3 mbar. All the gas lines are made of bler (Tb). Tb is preferably kept below the RT to prevent stainless connections that prevent any outgassing and unwanted condensation of the precursor material extraneous chemical reactions. The size of the sample upstream, before reaching the heated zone. Typically, a is limited by the diameter of the tube and length of the bubbler equipped with a chiller can reach a minimum heating zone. bubbler T~–30 °C. MOCVD MoS2 was deposited onto single crys- The actual amount of Mo(CO)6 introduced to tal c-plane sapphire substrates pre-annealed in air at the reaction chamber can be monitored by mass 1000 °C for 1 h to yield atomically flat terraces [1130]. spectr­oscopy of the atmosphere inside the chamber. This substrate has yielded much better growth morph­ Although calibration of the absolute concentration ology than other amorphous substrates, such as SiO2/ of the precursor is somewhat challenging, the relative Si. The presence of an atomically flat surface is likely concentration can be obtained for every run of the to be one of the prerequisites layer-by-layer epitaxy of experiment serving as an indicator for the amount of LMs [1130, 1149, 1150]. Typically, c-plane sapphire the precursor remaining in the bubbler. At the outlet substrates are cut into square pieces of 1 by 1 cm by first of the bubbler, in order to abruptly turn on and off scratching them with a diamond tipped pen and snap- the flow of Mo(CO)6, a solenoid valve was installed, ping them into separate pieces by hand. Before the pre- which can be automatically controlled, so that the pre- annealing procedure at 1000 °C in air, they are sub- cursor mass flow from the beginning to the end of the sequently cleaned in the IPA, acetone, and DI bathes MOCVD growth cycle is timed precisely. inside a bath sonicator for at least 10 min. As for the S source, mass flow control is more Ref. [1141] computed the thermodynamic ‘map’ straight-forward as H2S is in gaseous state at ambient of all possible phases resulting from the vapor reac- conditions and high purity (99.999%) gas can be pro- tions for different compositions of Mo(CO)6:H2S:H2, vided readily from commercially available compressed T, and pressures. They found a set of growth param­ gas cylinders. However, extreme caution should be eters with a high growth T ~850 °C, presence of exercised since the H2S gas is a highly flammable and H2(99%), and high growth pressure ~1 atm leading toxic gas with an acceptable exposure limit of 20 ppm to a good morphology with a large 1L domain size  >1 (set by the Agency for Toxic Substances and Disease µm and stoichiometry predicted from the equilibrium Registry, USA). In this regard, H2S gas level detectors thermodynamic modelling of the vapor phase. How- both in the ventilated gas cabinet storing the H2S gas ever, vapour phase reactions may not translate very 149 2D Mater. 7 (2020) 022001 C Backes et al Tgrowth = 750 ˚C 850 ˚C 900˚C 950˚C 1000˚C Figure VII.21. Dependent growth morphology illustrated by AFM height images of as-grown MoS2 on sapphire samples produced at different using Mo(CO)6 and H2S as model precursors. Scale bar, 250 nm. Adapted from [1142], copyright American Chemical Society. Figure VII.22. Nucleation density versus 1/T for the samples in figure VII.11. Adapted from [1142], copyright American Chemical Society. well to what takes place at the substrate surface level, as saturated vapour of Mo(CO)6 at the bubbler T~30 the non-equilibrium dynamics, mass-transport, pres- °C was introduced into the CVD system (a quartz ence of boundary layers due to the fluid dynamic con- tube three-zone hot-wall furnace) through Ar car- dition, and surface reactivity also play an important rier gas. c-plane (1 0 0 0) monocrystalline sapphire role in the final surface mediated growth mechanism substrates were placed in the center of the furnace [171, 765, 1151]. by loading them onto a larger sapphire plate (which E.g. excessive vapour phase reactions must be serves as a boat) and were heated to the growth T avoided, as they may lead to a significant formation of (750 °C–1000 °C) and annealed for 15 min in Ar MoS2 particles rather than substrate assisted epitaxy of (75 sccm), followed by added flow of the S precursor, atomically thin layers [1152–1155]. In order to enhance H2S (1.5–20 sccm), before introducing Mo(CO)6 to the heterogeneous, surface reaction over the vapour initiate the MoS2 growth in the S rich conditions that phase ones, generally speaking, higher T (>800 °C) prevent the creation of vacancies due to S removal and pressure (>850 Torr) are preferred [1151]. within the as-grown MoS2. At the end of the growth H2 during MOCVD is used as a reducing agent (typical growth time ~30 min), the Mo(CO)6 flow to prevent the oxidation of MoS2 and deposition of was shut-off by a solenoid valve and the sample was carbon based species. However, H2 is detrimental to cooled naturally to RT with Ar and H2S gases still the grain size, as the nucleation concentration could flowing through the quartz tube. All the stages of increase by an order of magnitude increasing H2 the MOCVD growth were performed under atmos- [1140]. Performing the growth at high T, low vacuum pheric pressure. environment (~800 °C and 10 mbar) in the presence After each growth run, the inner wall of the quartz of H2 would lead to Si contamination of substrate hin- tube was rinsed with DI water and IPA and blow-dried dering the growth. with N2. Afterwards, it was annealed at 1100 °C in Ar In turn, the growth conditions based on Ref. (100 sccm) for 2 h and in air for 2 h order to eliminate [1141] were used for initial growth studies. The all the residual precursor and by-product materials. 150 2D Mater. 7 (2020) 022001 C Backes et al Figure VII.23. (a) Optical absorption and (b) Raman spectra at 532nm from 1L-MoS2 grown at 950° C. Adapted from [1142], copyright American Chemical Society. In order to optimize the growth morphology Figures VII.25(a) and (b) showoptical microscopy toward large lateral size (>1 µm) and crystalline images of MoS2 films grown on sapphire, for two differ- 1L-MoS2, the growth parameter space was explored. ent Mo(CO)6:H2S ratios, 4:1 and 1:1 respectively. The The T dependent growth of MoS2 on sapphire in fig- core-level XPS spectra of Mo 3d, S 2s, and S 2p demon- ure VII.21 shows the AFM height images of for the strate that the low Mo(CO)6:H2S condition provides conditions that yield low overall coverage without a good stoichiometric film of MoS2 with only small merging of individual nuclei. The MoS2 nuclei form (<1%) amount Mo in different chemical state (Mo5+) randomly over the surface. than MoS2 (Mo4+), while the high Mo(CO)6:H2S The nucleation concentration exponentially condition leads to undesired deposition of thick Mo decreases with increasing temperature in an Arrhenius particles likely oxidized in 5+  state, with a poor Mo:S fashion (figure VII.22). This is consistent with a sur- stoichiometry consistent with the thermochemistry of face-mediated process, where the apparent activation the Mo, O, S, H system [1147]. energy is linked with various activated surface reac- tions such as adsorption, attachment, and diffusion VII.2.6. Reduction of the nucleation density using [793, 1156]. alkali metal halides Raman and absorption spectroscopy (figure So far MOCVD has produced 1L-MoS2 with average VII.23) demonstrate the high optical quality of the as- domain size  <1 µm[1157, 1158]. This presents grown 1L-MoS2, comparable to exfoliated and CVD a challenge towards production of single-crystal deposited MoS2 [1130]. E.g. the A, B, C, and D exci- 1L-MoS2 over the length scale of the substrates. In tonic peaks are visible [1130, 1142], a clear optical sig- order to improve the domain size, one could employ nature of MoS2. Moreover, the wavenumber difference alkali metal halide salts such as NaCl, KI, and KBr. between the two prominent Raman peaks, relating to The use of thes salts was documented in MOCVD E′ and A1′ modes is ~19.7 cm−1, close to the standard experiments to improve domain sizes, involving value of 1L- MoS2. Nevertheless, the grain sizes even precursor combinations of Mo(CO)6 and (C2H5)2S for high growth T (1000 °C) are less than hundreds of [1140], MoCl5, and (CH3)2S, [1159] and conventional nm for these growth conditions. This makes further CVD [1160] on SiO2/Si. A metal halide, such as characterization of individual flakes difficult and the NaCl or KI powders (~0.01–1 mg), was placed in the polycrystallinity of the film may lead to a significant upstream area of the furnace either in a boat or plate degradation and non-uniformity in electronic quality of alumina [1142]. Mo(CO)6 was introduced at ~600 for wafer-scale applications. °C for 30 min before introducing H2S at higher T The effect of partial and overall pressures of the (750 °C–1050 °C). A nucleation suppression effect gas species, Mo(CO)6:H2S:H2 on grain size was inves- is observed [1140], where 1L-flakes  >  10 µm can be tigated. Using XPS, it was found that for low growth achieved. Analysis by optical absorption, Raman, PL pressure (10 mbar or below), the MoS2 film becomes spectroscopy and XPS in figure VII.26 indicate that this contaminated with Si (figure VII.24) likely originat- MoS2 exhibits a direct-band gap without significant ing from the quartz tube, with the reducing environ­ defects, strain, impurities and doping comparable ment originating from H2 in the carrier gas and from to those of 1L-MoS2 grown by conventional CVD. the decomposition of H2S, not ideal for a clean, growth Furthermore, the crystal facets are highly aligned in of 1L-MoS2. In addition, larger amounts of Mo(CO)6 one direction, demonstrating a successful lattice- compared to H2 tend to yield bulk particles of non-sul- orientation matched epitaxial deposition of MoS2 phurized, Mo oxides as probed by XPS (figure VII.25). on single-crystal sapphire. T dependent field-effect 151 2D Mater. 7 (2020) 022001 C Backes et al Pgrowth = 1 atm Pgrowth = 8.5 mbar normalized intensity O KLL Mo 3p C 1s C KLL Mo 3s S 2p Si 2s Si 2p 2s Al 2p Al 1200 1000 800 600 400 200 binding energy (eV) Figure VII.24. XPS survey spectra of MoS2/sapphire samples from the two growth runs with different chamber pressures (1 atm and 8.5 mbar). XPS measurements were performed with PHI Versaprobe II equipped with x-ray source, Al Kα; hv  =  1486.6 eV. a c Mo4+ 3d5/2 e 600 S 2p3/2 500 1200 Intensity (counts/s) Intensity (counts/s) 400 Mo4+ 3d3/2 800 300 S 2p1/2 200 400 S 2s 100 Mo5+ 3d3/2Mo 3d5/2 5+ 0 0 242 240 238 236 234 232 230 228 169 168 167 166 165 164 163 162 Binding Energy (eV) Binding Energy (eV) b d f 150 600 Mo5+ 3d5/2 S 2p3/2 500 Intensity (counts/s) 100 Intensity (counts/s) 400 Mo5+ 3d3/2 S 2p1/2 Mo4+ 3d3/2 300 Mo4+ 3d5/2 50 200 100 S 2s 0 0 244 242 240 238 236 234 232 230 228 169 168 167 166 165 164 Binding Energy (eV) Binding Energy (eV) Figure VII.25. Influence of Mo(CO)6:H2S flow rate ratio on MoS2 film composition. (a) and (b) Optical images of two different Mo(CO)6 Ar carrier and H2S mass flow rates of (a) 15 (Mo(CO)6 flow rate of 0.0053 sccm) and 15 sccm, and (b) 5 (Mo(CO)6 flow rate of 0.0018 sccm) and 20 sccm with growth T~ 950 °C, bubbler T~ 30 °C, H2 ~10 sccm, growth pressure ~1 atm, and growth time ~30 min. Scale bar, 25 µm. (c) and (d) Mo 3d and S 2s core-level XPS spectra for the corresponding growth conditions of (a) and (b). (f) S 2p core-level XPS spectra for the corresponding growth conditions of (a) and (b). XPS measurements were performed with PHI Versaprobe II equipped with a Al Kα x-ray source; hv  =  1486.6 eV. transistor characterization using polymer electrolyte the MOCVD precursor. Refs. [1161, 1162] suggested gating demonstrated a promising n-type µ~100 cm2 that the presence of alkali metals leads to decrease in V−1 s−1, similar to 1L-MoS2 deposited by conventional the melting point of oxide-based Mo precursors and CVD [1149]. even the melting point of a substrate [1163], increasing It is currently not clear exactly how the domain the reactivity. This in turn could lead to the increased enlargement or the suppression of nucleation by the diffusion and reduction in the nucleation density. Na metal halides occurs. The desiccating effect of NaCl oxides or compounds of sodium and molybdenum salt as suggested by Ref. [1140] is not likely to be the oxides (NaxMoyOz) [1160, 1162, 1164] may also play dominant factor since even a very small amount of salt a role, as they were found at the interface between (<10 µg) could be used, where it would decompose as-grown 1L-MoS2 and substrate [1164]. It was also before the growth reaction. Ref. [1159] suggested that found that after growth, where the salt was initially salt helps to reduce the carbon contaminant, which present, the halogen atoms evaporated away from the might hinder crystallization, from organic parts of substrate surface while the alkali metals remained 152 2D Mater. 7 (2020) 022001 C Backes et al a c 0.030 C 425.4 nm 0.025 D 301 nm Absorbance 0.020 0.015 B 0.010 607 nm A 656 nm 0.005 0.000 300 400 500 600 700 wavelength (nm) as-grown MoS2/sapphire b d 666 nm A1g transferred on SiO2/Si 20 cm-1 E’2g sapphire intensity Counts 25 nm sapphire 360 370 380 390 400 410 420 600 650 700 -1 raman shift (cm ) wavelength (nm) Figure VII.26. (a) Optical microscopy of 1L-MoS2 with NaCl placed upstream. Scale bar: 25 µm. (b) Raman spectroscopy on a single crystalline 1L-MoS2 area (c) optical absorbance. (d) PL spectra from single crystalline 1L-MoS2 flake as-grown on sapphire and transferred onto SiO2/Si. Adapted from [1142], copyright American Chemical Society. [1142]. Na2MoO4 alone (without any halogens) is and dimethyl disulphide [1172], Mo(NMe2)4 and effective precursor for growth of large domain sizes 1,2-ethanedithiol [1173], on various insulating [1165]. Large area MoS2 growth of large domain sizes substrates such as SiO2/Si, sapphire, and mica. was acieved on Na doped glass substrates [1164, 1167]. However, often the end products are misaligned Na especially impacts the growth of large crystalline nanocrystalline MoS2, with varying degree of thickness domains at a wide range of growth T (720–1050 C). likely resulting from the low growth T ~300 – 600 The exact role of the alkali metals remains to be inves- °C). Although it is more challenging to prepare and tigated, as there are numerous competing atomistic control the stoichiometry of the product, MOCVD of processes, such as surface diffusion, nucleation site MoS2 using single-source precursors, tetrakis(diethyl- activation, attachment kinetics that control nucleation dithiocarbamato)molybdenum(IV) [1174] and and growth, which all can influence the nucleation Mo(S-t-Bu)4 [1175] were also investigated. density and domain size. This be extended to perform MOCVD of verti- 7.2.8. Brief comparison methods for the growth 2D cal and horizontal heterostructures of TMDs, such semiconductors as MoWS2 and MoWSe2, using currently available In the table VII.2, we summarize the methods for metalorganic sources, such as W(CO)6, and dimethyl the growth of LM semiconductors presented in this selenide [1168]. In addition, doping of TMDs may also chapter. be achieved by introducing dopant precursors, as done for MOCVD of other materials systems [1169, 1170]. VII.3. Layered materials heterostructures Low toxicity, low-cost precursors should also be con- Typically, heterostructures of semiconductor sidered, in order to achieve a safer and more accessible compounds have been fabricated using isostructural production of TMDs. substrates with nearly lattice matched substrates. One of the advantages of LMs is that the lattice matching VII.2.7. ALD requirements needed for cubic materials systems is Atomic layer deposition (ALD) techniques were relaxed [1149]. Lateral heterostructures are also a presented as a modified CVD to produce thin layers focus of research [1176]. Growth of lateral and vertical of MoS2 using the precursor pairs—MoCl5 and heterostructure could enable a number of devices H2S [1171], Mo(CO)6 and H2S [1157], Mo(CO)6, [1177]. Many groups have fabricated and studied 153 2D Mater. 7 (2020) 022001 C Backes et al Table VII.2. Basic conditions for growth of TMDs. Method Material Substrate Precursors Temperature Pressure Reference CVD MoS2 SiO2 MoO3 and S 750 °C Atmospheric [1008] MOCVD MoS2 Sapphire Mo(CO)6 and H2S 600 °C–700 °C Atmospheric [1142] TAC MoS2, WS2 Quartz, SiO2 Mo, W, S 750 °C Atmospheric [1093] PVT β-In2Se3 SiO2, mica, graphite γ-InSe 600 °C 1.6 mbar [1117] aligned and unaligned LMHs with the objective of VII.3.1. Graphene/h-BN stacks making new devices and scale transistors beyond There are several types of LMHs that can be envisioned the CMOS limit [1178–1183]. Many of these LMHs for future devices: (1) vertical, (2) lateral, (3) twisted were fabricated using exfoliation and transfer rather (controlled and uncontrolled). Graphene and h-BN than direct growth. It would be desirable to use CVD have been grown by a number of catalytic CVD to fabricate them. In the case of h-BN and graphene, [1203, 1204] and plasma enhanced techniques [963]. because of their refractory nature, a catalyst is used Catalyst-free graphene growth yields poor crystal to grow them. HBN or graphene would be ideal shapes and crystallinity, with low rates [947, 1204, substrates for other LMs. Thanks to its atomically 1205]. Ref. [836] used a catalyst-free CVD approach flat surface, low interlayer electronic coupling and to synthetize crystalline graphene on exfoliated h-BN, almost perfect reticular matching, h-BN is an ideal with noticeable growth rates on single-crystal on-axis substrate for graphene [1184]. Many of the researched SiC substrates (either Si- or C-face). SiC was selected LMH devices also include TMD semiconductors because of its high thermal stability [1206], in view [112, 1185]. However, these materials have challenges of the high growth T, typically  >1200 °C, needed for of their own as discussed in section VII.2 In order to a non-catalytic growth approach. The SiC(0 0 0 1) create LMHs containing semiconducting TMDsor samples were diced from commercially available BLG, metals (TMD and graphene), and dielectrics wafers and then cleaned via wet chemistry, prior to (h-BN), a significant amount of effort is needed either mechanical exfoliation of h-BN. They were immersed via repeated transfer of the individual LMs or direct in an ultrasonic bath of acetone and then IPA for 3 min growth [1177]. each followed by a DI water. They were then cleaned LMHs of cubic systems were grown by MBE and in HF (48% solution) diluted in DI water (1:10 HF: CVD, MBE being the work horse for bulk III-V het- DI-H2O) for 1 min to remove the native oxide and erostructures today. Both CVD and MBE were used thoroughly rinsed in DI water. The samples were then for the growth of TMDs [1186–1189] while CVD [765, subjected to an oxygen plasma (O2 flow, 80 W, 5 min) 1038] has advantages for graphene [765, 1038] and to completely remove the organic or carbonaceous h-BN [836, 1054, 1055, 1087, 1190, 1191]. CVD may contaminants. The SiC surfaces, which had polishing be the most suitable technique for the scalable synthe- scratches in the as purchased form, were hydrogen sis of highly-crystalline LMHs even though the selec- etched in the same vertical cold-wall reactor (Aixtron, tion of the final deposition equipment and preferred HT-BM) for 4 min in an atmosphere of high purity precursor types have not been finalized yet. Such CVD (6N) H2 and Ar at 1230 °C and a total pressure of approach is not trivial as weak interlayer interactions 450 mbar using 500 sccm flow rate for each gas. AFM favor 3d island growth [1192]. The formation of multi- was used to measure the SiC surface roughness and layer islands is typically avoided by using high chalco- only substrates with atomically flat surfaces were used genide to metal ratios adopting short reaction times, for growth. PDMS stamps were used for mechanical which leads, however, to isolated crystals [1192–1196]. exfoliation of h-BN on the hydrogen etched SiC. To this Growth LMHs, vertical or lateral, is challenging. In this end, two PDMS stamps with dimension of 2  ×  2 cm2 section we will discuss a few materials options[1192]. were adopted and the mechanical exfoliation process There are several reports targeting the Controlled was repeated more than 40 times. The clean SiC lateral and vertical TMDs heterostructure growth substrate was then pressed on the PDMS stamp for 20s. [1197–1199]. Controlled grown of lateral heterostruc- This process yielded h-BN flakes with lateral size up to tures of TMDs would enable quantum engineered 150 µm. The PDMS residue on top of the h-BN/SiC devices, through band alignments, enabling new gen- was removed by using first acetone for 3 min and then eration of electronic devices [1177, 1200]. 1L-TMDs IPA while gently shaking the beakers. Ultra-sonication can be grown sequentially to construct in-plane LMHs, was not adopted, since it causes detachment of the otherwise not achievable through other processes such mechanically exfoliated h-BN flakes from the SiC as transfer of grown or exfoliated films. Ref. [1200] substrate. Finally, the samples were treated with a 25 demonstrated a direct synthesis of lateral LMHs of W oxygen plasma for 30 min. This last step was found MoS2  −  WS2 and MoSe2  −  WSe2 by using ambient- to be crucial for obtaining single crystal graphene on pressure CVD with aromatic molecules as seeding pro- h-BN as it allows for removal of adventitious carbon moters. Vertical LMHs, while simpler, will require sig- contaminations [836]. The h-BN/SiC substrate nificant effort is reducing 3d growth [1201, 1202]. was then placed in the Aixtron HT-BM system and 154 2D Mater. 7 (2020) 022001 C Backes et al Figure VII.27. Growth chamber of the Aixtron HT-BM vertical cold-wall reactor used for growth of graphene on h-BN. positioned in the central recess of the heater (see 1211–1213] TMDs using a variety of precursors using figure VII.27) for subsequent graphene growth. similar CVD systems, although this is far from the The h-BN films on SiC were then annealed in an optimal or desired approach. Direct growth of WS2 8 mbar H2 pressure, with hydrogen flow of 1000 sccm on other LMs resulted in isolated grains at the micron at 1150 °C for 10 min. [836]. Single crystal graphene scale [1194] or uncontrollable growth of FL-WS2 on was then grown under the following conditions: graphene [1195] and isolated grains on h-BN [1196]. 150 sccm H2, 1000 sccm Ar and 5 sccm of methane What is critical in the initial crystal growth process (CH4) for 30 min. The H2:CH4 ratio was found to be development, in addition to identification of the a critical parameter for the growth of high quality materials and the equipment and precursor selection graphene. H2:CH4 ratios of 1:1 yielded polycrystal- and design, is the understanding of nucleation and line graphene with rounded edges, and a ratio of 1:30 growth of controlled TMD layers. Some inroads in the yielded hexagonally shaped single-crystal graphene basic understanding for both MBE as well as CVD were domains. The optimum growth T was found to be reported in Refs. [1201, 1202, 1214]. 1150 °C. Lower growth rates were achieved at 1000 A CVD approach to obtain continuous 1L-WS2 °C [836], but little to no growth was observed below on h-BN and graphene was reported [1209] following this T. Above 1150 °C, growth was also achieved, similar growth procedure used for MoS2 [1189, 1215]. but clustering of amorphous carbon was also often Ref. [1216] used a similar method to grow MoS2/h-BN observed [836]. Growth was performed at different LMHs, as well as growing h-BN by CVD as shown in pressures ranging from 7 to 150 mbar. This yielded figure VII.28. smaller (<0.5 µm) and bigger (>2.0 µm) domains Sulfurization of WO3 powder (Sigma Aldrich, for low and high pressure conditions respectively. 99.995%), was carried out within a horizontal hot- Carbonaceous clusters were occasionally observed at wall furnace. This is is basically a two-zone system the higher pressures (150 mbar). For this reason, the with WO3 within a high T zone, and Sulphur (Sigma growth pressure should be limited to 25 mbar to pre- Aldrich, 99.99%) in a quartz crucible in a cooler zone, vent the formation of such clusters. Lower hydrogen of the furnace as shown in figure VII.29. The system flows (i.e. 100 sccm) and longer growth times were is connected to a scroll pump, which ensures a base also be used to obtain continuous films. With hydro- vacuum ~5  ×  10−2 mbar. The h-BN and graphene gen flows  >150 sccm only partial growth of hexago- substrates are prepared through MC on either quartz nal single-crystals could be achieved. At the end of the or SiCwith variable (up to 1 cm2) surface area. Quartz growth process at high T, the system was cooled (300 substrates are cleaned with a piranha solution (3:1 °C min−1) in the same atmosphere of Ar and H2 used H2SO4:H2O2) for 15 min and acetone and 2-propanol during growth. The sample was unloaded from the semiconductor grade for 3 min each before flake exfo- chamber for T  <  120 °C. liation. SiC substrates are also threated with 5% HF in water for 1 min to remove the SiOx on the surface. After VII.3.2. TMD/h-BN and TMD/grapheneLMHs the exfoliation of h-BN, the substrates (either quartz There are many reports on growth of sulfides or SiC) are treated with an oxygen plasma (5 min, 80 [1196, 1197, 1207–1210] and selenide [1139, 1176, Watt) to remove possible scotch tape residue. 155 2D Mater. 7 (2020) 022001 C Backes et al Figure VII.28. Schematic of two fabrication processes of MoS2/h-BN LMHs. The first consists in transferring the CVD grown film from a Cu foil to the center of a Si/SiO2 chip, leaving four corners of the substrate to be bare SiO2. This substrate is then loaded face up into the CVD system to grow MoS2 on it. The MoS2 domains will cover both h-BN and SiO2 areas. The second approach is through a two step PMMA assisted process. The CVD grown h-BN film is first transferred onto the center of a SiO2/Si chip, followed by stacking another layer of CVD grown MoS2 on the whole substrate surface, obtaining transferred MoS2 domains on h-BN in the central area. Two of the transferred bundaries of h-BN on the SiO2 surface in the first fabrication method are labeled A and B. Adapted from [1216], copyright American Chemical Society. WS2 can also be grown on polycrystalline or single to control evaporation. It is important that the growth crystal CVD graphene [766, 841] transferred on quartz reaction occurs under S rich conditions. To ensure this, or SiC cleaned as described above. For WS2 growth on two aspects are crucial: graphene on SiC, fresh as-grown samples are used. Both are cleaned for 3 min in acetone and for 3 min 1) S and WO3 powders have to be in a ratio in 2-propanol. Crucibles are cleaned with 2-propanol between 50:1 and 100:1. If not, the reaction wet tissue and subsequently annealed in the furnace typically stops at the intermediate step of WO2 used for the growth at 1100 °C for 60 min in 200 sccm cluster formation. Optimum weights ~100 mg Ar before each growth. This ‘empty run’ ensures hav- of S and 1 mg of WO3 were reported. ing both a reactor and clean crucibles. After annealing 2) S has to be delivered when WO3 has reached a usually the crucibles are electrostatically charged. This stable optimal T. To achieve this, this process can alter the weighing of the reactants. The electric can be followed. Once crucibles and target field interacts with the WO3 powder, which sticks to sample are in place within the reactor as shown the crucible and spreads around in a not reproducible in figure VII.29(b), the growth can start. The way. Hence, the crucible needs to be grounded before ramp-up phase has two steps: (i) ramping up weighing. to 350 °C at 10 °C min−1 with no gas flowing There are many important details related to oxide to favor sample outgassing; (ii) ramping up powder preparation, loading and placement in the to 900 °C (still at 10 °C min−1) while the Ar horizontal furnace. The substrate and WO3 are placed flow is set at 500 sccm (tube pressure equals in the same flat 1-inch wide crucible, next to each 4 mbar) in order to keep the S powder solid. other, but well separated to avoid accidental deposi- The purity of the Ar gas is 99.9999%. Once the tion of the powder on the substrate. The WO3 powder hot zone reaches the working T~ 900 °C, the has to be spread uniformly on the quartz crucible after Ar flow is reduced to 8 sccm (i.e. tube pressure weighing in order to maximize the reaction surface. ~0.5 mbar) to ensure S delivery. The heating The distance, typically ~1 cm, between substrate and belt wrapped around the S zone has to be WO3 powder is also important to control the sur- turned on in order to reach 200 °C when the face coverage. The S powder is placed outside the hot hot zone reaches the working T. This ensures S zone of the oven in the Ar upstream part, ~20 cm away feeding for ~30 min. The total process is ~1 h from the furnace (see figure VII.29). A heating belt is long to anneal the surface of the sample after wrapped around the tube region where the S is placed the growth reaction. The sample is then cooled 156 2D Mater. 7 (2020) 022001 C Backes et al Figure VII.29. (a) Scheme of the furnace used to grow continuous films of WS2 on graphene and h-BN; (b) picture of the furnace. Adapted from [1217], with permission of The Royal Society of Chemistry. Figure VII.30. Different method used to create TMD LMHs, (a) Co- conversion of both metal films together to form TMDs. (b) Sequential conversion of metal films. (c) LMH formation through stacking pre-converted films. to RT by shutting the power to the furnace 1218]. Lateral LMHs between different TMDs show and, once 400 °C are reached, the furnace lid is atomically sharp interfaces with an abrupt transition opened to speed up cooling. between constituent materials, without voids and other defects. Using direct growth instead of material transfer has the benefit of a residue-free LMHs and VII.3.3. TMD/TMD LMHs scalability to large areas. CVD and MOCVD can also be used for growing Lateral and also vertical 1L-LMHs both based on different types of TMD LMHs [1176, 1198, 1199, the same metal such as MoS2 and MoSe2 [1218], WS2 157 2D Mater. 7 (2020) 022001 C Backes et al Figure VII.31. (a) Legend for Raman maps. (b) and (c) Raman map of MoS2/WS2 A1g peak position for sequentially and co- 1 converted films. (c) and (d) Raman map of MoS2/WS2 E2g peak position for sequentially and co-converted films. (e) and (f) Raman spectra of MoS2/WS2 for sequentially and co-converted films respectively. and WSe2 [1218], as well as different metals, such as lowed by a second metal deposition and TAC. The WSe2 and MoSe2 [1176], MoS2 and WS2 [1198], MoSe2 third is the large scale mechanical transfer of one TAC and WSe2 [1176], and WSe2 and MoS2 [1199] have film on top of another (figure VII.30(c) reported using CVD based on solid sources. Control Raman spectroscopy was employed to analyse over gas flow during the CVD process can also allow the interface between the two TAC films in a LMH. the growth of superlattices with well-controlled spatial It was used to analyse the effect of co-converting the modulation [1220, 1221]. two metal films versus sequentially converting them. Some of these materials can be used as lateral p-n This experiment was performed with MoS2 and WS2 junctions without employing in situ doping, as e.g. as they both can be sulfurised in the same T range, 800 in the case of MoS2 and WSe2, where the boundary °C. The co-sulphurized sample had both metals sput- between the two materials serves as a p-n junction tered and converted to the equivalent TMD together, [1222] and can be used as a LED. Theory also predicted while the sequential sample had the Mo deposited novel behaviours, such as efficient charge separation and converted, then the W deposited and converted. [1222] and spin/valley filtering [1223]. Shadow masks were used to give the cross structure While these methods yielded perfectly stacked indicated in figure VII.31(a). Both samples were then materials, they are limited to very small lateral dimen- Raman mapped, with a spectrum taken every 300 nm sions and a few selected material sets. TAC methods over a 10 µm  ×  10 µm area. All spectra are normalised are an alternative for large scale production of TMD/ to the SiO2 peak ~520 cm−1. This data was then used TMD heterostacks. TAC opens various avenues for the in figures VII.31(d) and (e), which show the frequency formations of large scale LMHs as in figure VII.30. Fig- position map of the MoS2 E2g 1 peak. For the MoS2 area ure VII.30(a)) shows the subsequent deposition of two −1 the peak is at ~383 cm (blue) in both maps. The WS2 different metals by PVD followed by TAC. This works and SiO2 areas, with no peak in this frequency region, for sulfides or selenides of metal in which the conver- show only random noise. In the overlap region, for the sion takes place at a similar T. The second example in sequentially sulphurized sample figure VII.31(d), the figure VII.30(b) is the growth of a TMD by TAC fol- maximum intensity is very close to the same frequency 158 2D Mater. 7 (2020) 022001 C Backes et al as the MoS2 area. However, in the co-sulphurized sam- peak of MoS2 and WS2 will both shift towards an inter- ple figure VII.31(e), there is a noticeable shift to a lower mediate value, as observed. For the sequential conver- wavenumber as shown by the green colour. This is an sion there is a small shift of both peaks towards each indication of increased alloying at the interface in the other by ~1 cm−1. For the co-conversion the peaks co-sulphurized sample. A similar effect was seen in Ref. each shift by ~4 cm−1 towards each other, indicating [1224] where they saw increasing shift of the MoS E2g 1 more alloying in the co-conversion case. to lower wavenumber with increasing alloying with W. Even though large scale LMH formation by co- To confirm this, the A1g peaks of MoS2 and WS2 conversion was successful [1224], the electrical per- were analysed, position maps are shown in fig- formance was poor probably due to the roughness of ures VII.31(b) and (c). These peaks are both in the same the films and some level of alloying during the TAC frequency region ~400–430 cm−1 so to distinguish process. Co-sulfurisation (figure VII.31(a)) was used between changing peak intensities and peaks shifting to create catalyst for hydrogen evolution, [1225] while frequency, these peaks were fitted. What is expected in the subsequent sulphurization of Mo and W (figure the case of alloying between the layers is that the A1g (b)) was employed to create PDs [1226]. 159 2D Mater. 7 (2020) 022001 C Backes et al VIII. Functionalization of GRMs of graphene using graphite as starting material has gained significant interest since it allows (i) the A number of strategies have been developed for the generation of exfoliated graphite sheets in large functionalization of GRMs. On one hand, this is driven quantities compared to other exfoliation techniques by the curiosity of chemists to master controlled surface [375, 377, 1230], (ii) access to solution-processable modification. On the other hand, functionalization is graphene [373, 374], (iii) access to modified physical of great practical importance, as it has the potential to properties compared to pristine graphite [1231, 1232], tune the properties according to the needs of specific and (iv) the investigation of the intrinsic chemical and applications. In this section, an overiew is given in physical properties of graphane [160, 1230]. Graphite particular focusing on the practical requirements for can be activated by reductive charging using an alkai successful modification of GRMs. A broader overview, metal, which subsequently leads to a Coulomb driven, is also given in Refs. [160, 165, 1227–1229]. solvent-based exfoliation [160, 373, 375], as sketched Typically, functionalization is classified as nonco- in figure VIII.1. The intermediately generated material valent or covalent as outlined in Ref. [160] and brifly is negatively charged and commonly referred to summarized below. Noncovalent involves the adsorp- graphenide [1233]. It can be reoxidized by benzonitrile tion of other species without covalent bond formation. or other chemical agents [1234], yielding SLG It is therefore often considered as ‘non-destructive’, and FLG [375], or trapped by a variety of different since the lattice is not altered, or referred to as phys- electrophiles, yielding covalently functionalized isorption. However, physisoprtion is not a precise term graphene with a variety of functional entities. By this for noncovalent functionalization. This is because route, hydrogenated [369, 1235, 1236], alkylated [368, not only van der Waals interactions can occur. Often, 1237–1239] and arylated graphene [370, 1240, 1241] more specific, stronger noncovalent interactions are are accessible in a one pot synthesis. exploited, e.g. dipole and ionic interactions [1771], The entire functionalization sequence has to be or π–π stacking, which is specific to graphene with a carried out without oxygen and moisture under inert conjugated sp2 carbon network. The term covalent gas atmosphere in a glove box, otherwise a substantial modification is used when covalent bonds are formed amount of side products is formed [375, 1247, 1242, [1329]. Often, this is referred to as chemisorption. For 1243]. Reductive graphene functionalization repre- graphene this includes various cases: from new C–C sents a versatile protocol which can be applied to gra- bonds on the basal plane leading to a rehybridization phene in dispersion as well as to SLG on substrates. The from sp2 to sp3 [1772, 1773], to modifications of defects latter approach can be used to elucidate topological (e.g. edges [1773–1776]) or the production and deri- and mechanistic details of the reaction [1237, 1244]. vatization of GO [156, 160, 1799]. A variety of experimental conditions are reported In all cases, the properties of the parent material in the literature [160, 1777–1779]. From the starting are modified, but in a different way and to a different graphite sources, to the composition of the GICs, the extent. E.g., the grafting of other species on the sur- employed solvents and/or the used equivalent number face can be used to form functional architectures or to of the organic addends, to name a few. We will over- adjust the compatibility with other materials in com- view different aspects of fundamental importance posites. Functionalized GRMs typically also display a for the successful covalent functionalization of gra- better dispersibility (higher concentration), especially phene, starting from reductively activated bulk graph- in solvents that can otherwise not be used. In terms ite: (a) Type of starting graphite, (b) solvents used for of electronic properties, noncovalent functionaliza- the exfoliation, and (c) type of GIC and electrophilic tion can be considered as a ‘milder approach’: while trapping reagent [371, 1238]. Covalently functional- it can be used to controllably dope graphene, it does ized graphene, without initial reductive activation of not harm the conjugated basal plane network and thus graphite, can be accessed by a variety of other synthesis does not alter the intrinsic electronic properties dra- approaches [1245–1247]. matically. In turn, covalent functionalization of GRMs essentially creates new materials with distinct proper- VIII.1.1.1. Type of starting graphite ties. E.g., covalent functionalization of graphene is a In the context of the reductive graphite exfoliation/ tool to open a band gap, but at the cost of reducing µ functionalization protocol, almost any type of [1780, 1781]. graphite, can be used. A variety of graphite starting materials with different physical and morphological VIII.1. Covalent functionalization of graphene properties were screened [1248] and based on that VIII.1.1. Reductive bulk functionalization their covalent functionalization studies focused on The covalent functionalization of carbon allotropes, 3 types of graphite [1238]: (a) Natural flake graphite especially graphene, is a major research topic in (Kropfmühl AG, Passau) with a particle size of 18 µm the growing field of functional nanomaterials. The and a low intrinsic density of defects as expressed by development of novel architectures built-up from I(D)/I(G) ~ 0.2, with a regular stacking order, (b) graphene requires the availability of a cheap source with PEX 10 (Future Carbon AG)—an expended powder low polydispersity. Chemical bulk functionalization graphite with grain size of ~3–5 µm and I(D)/I(G) ~ 160 2D Mater. 7 (2020) 022001 C Backes et al Figure VIII.1. Reductive exfoliation/functionalization of graphite. 0.3 and, (c) SGN18 (Future Carbon AG)—a spherical, The type of graphite activation used prior to the defect-rich graphite with I(D)/I(G) ~ 0.4, with a grain addition of the trapping electrophile has a significant size ~20 µm. influence on the outcome of the functionalization In general, the chemical reactivity of graphite with sequence and the obtained graphene architectures respect to a reductive covalent functionalization sce- [1235]. The hydrogenation of graphite the GIC (d) nario strongly depends on the nature and in particular yields polyhydrogenated graphene with the highest the morphology of the starting material. It is key that hydrogen content when alcohols are used as trapping the starting material is dry and free of oxygen. There- reagents [1235], whereas for the Birch-type reductive fore, a thermal annealing step (300 °C, vacuum, 4–5 d), activation (b) the highest hydrogen content is obtained is used to remove any oxygen inclusion. For materials for water [369]. This can be explained by the nature of with intrinsically large graphene basal planes (e.g. nat- the solvent. In the latter case, ammonia is used at  −75 ural flake graphites) a reductive charging/exfoliation °C and interacts with the trapping water (formation is very difficult. The intrinsic flake size can be reduced of ammonium hydroxide in an equilibrium reaction). by grinding with NaCl, in particular, the graphite is This underlines that the applied reductive activation ground with 5 times the amount of NaCl in a mortar conditions of graphite have a fundamental influence for 20 min. After the elution of NaCl with distilled on the outcome of the reaction. water and drying in vacuum, smaller and easier dis- persible flakes can be obtained. VIII.1.1.2. Solvent preparation For the reductive charging/activation of graphite, 4 Water- and oxygen-free solvents are mandatory for approaches can be pursued: (a) Electron transfer rea- the successful exfoliation of the reductively activated gents [1239], (b) Birch-type based reduction [1249– graphite starting material and in order to suppress side 1252] in liquid ammonia with alkali metals [374], (c) reactions during the functionalization sequence [375, reduction with sodium/potassium alloy [1253], and 1242, 1243]. The solvent has to be stable against the (d) the generation of alkali metal/GICs [373, 1254]. In intermediately generated negatively charged graphene the latter two cases, the balanced combination of inter- sheets (graphenides). The most suitable solvents with calation-driven layer expansion in combination with respect to these prerequisites are tetrahydrofuran a Coulomb-repulsion-driven interaction facilitates (THF) [1242], 1,2-dimethoxymethane (DME) exfoliation and the wet chemical functionalization of [1255] and N,N′-dimethylacetamide (DMAc) [1256]. the electronically activated graphene sheets. The solid- Although widely used for the dispersion of negatively state-based graphite intercalation process (d) allows for charged CNTs and graphene, DMSO is thought to a fine tuning of the different alkali metal to carbon stoi- partially quench the respective negatively charged chiometries from KC4 to KC24 [370]. K is used to reduce carbon allotrope, yielding methylated side products graphite and to generate the respective GICs. The metal [1242, 1256]. is directly mixed with graphite in an inert gas atmos­ Regardless of solvent used, the most crucial step phere (glove box  <0.1 ppm O2, <0.1 ppm H2O) at 150 for the wet-chemical reductive functionalization of °C overnight. By changing the alkali metal amount, dif- graphite is the initial solvent purification. Stabilizers, ferent intercalation compound stages can be obtained water and/or oxygen traces can influence the reaction e.g. KC4, KC8, KC16, and KC24. The reductive activation outcome and may lead to undesired and uncontrolla- of flakes, deposited on substrates, or CVD SLG can be ble side reactions [1242]. Ref. [1257] established a mul- carried out by applying of a droplet of liquid Na/K alloy tistep solvent purification protocol using molecular (1:3) in 1,2-dimethoxymethane (DME). The deep-blue sieve. Ethers (THF, DME) must be stored under light Na/K alloy solution can be prepared inside a glovebox exclusion in order to suppress any auto oxidation pro- and stirred for 3 d in an Ar atmosphere. After the gra- cesses. The water content of the solvent can be deter- phene activationfollowed by the addition of the trap- mined by a Karl-Fischer titration. A sufficiently low ping electrophile, the remaining reagents can be rinsed water content value, THF:  <  2 ppm, DME:  <  0.5 ppm, off by dry and degassed DME. allows to proceed with the next solvent treatment step: 161 2D Mater. 7 (2020) 022001 C Backes et al Figure VIII.2. (a) Molecular structure of DBT. (b) STM image of DBT assembly on graphite, showing also the unit cell; the possible molecular packing is also schematized. The position of BF4-counterions is indicative. Lattice parameters: a  =  3.9  ±  0.1 nm: b  =  1.0  ±  0.1 nm: a  =  89°  ±  2°: A  =  3.9  ±  0.1 nm2. (c) and (d) STM height (c) and current (d) images of DBT self-assembled on SLG. The quality of the image is not as good as on graphite, as a result of the few Å roughness of the underlying SiOx, causing a blurring effect. (e) Schematic representation of the electrochemical setup used for grafting. WE  =  graphene; CE  =  Pt; RE  =   Ag/AgCl (3 M KCl). Adapted from [1263], copyright American Chemical Society. the removal of oxygen traces by a freeze-pump-thaw functionalization step, mixed functionalized graphene degassing. For this purpose, the solvent is transferred derivatives are accessible [1244]. However, there is into a Schlenk flask and frozen by the aid of liquid a huge dispersion of the experimental conditions in nitrogen. The flask is connected to a vacuum line in literature [368, 1239, 1258, 1777, 1778]. In this sense, order to remove the gas phase. Afterwards, the solvent a complete study surveying all the reaction parameters is allowed to warm to RT (gas bubble evolution from and determining the most effective synthetic routes the solution). Subsequently, the solvent is frozen again is highly desired. Even more important, a deeper and the gas phase is removed in vacuo (~10−2 mbar). understanding of the correlation between bulk This procedure is iteratively repeated several times approaches, aiming at functionalized SLG and direct (4–6 cycles) until no gas evolution is detected during functionalization is of utmost interest [368, 370, 371, the thawing process. In order to remove any residual 1227, 1238, 1253]. Ref. [371] compared 3 reductive water traces, a distillation over Na/K alloy is carried functionalization pathways (phenyldiazonium- and out. Here, a Na/K alloy (1:3) is freshly prepared and bis-(phenyl)-iodonium salts as well as aryliodide) added dropwisely to the solvent. A water separator and the reductive alkylation route using hexyl- and connected to the distillation apparatus and filled with dodecyliodide. The use of iodide derivatives with dried molecular sieves allows additional drying of the spherical graphite as starting materials is the most solvent. Preferably, the solvent is stored in an Arfilled efficient to achieve functionalized materials. A single glove box in order to prevent any further contamina- electron transfer (SET) process from the charged tion. The purity of the solvent can be checked upon graphenides towards the electrophilic trapping reagent addition of GIC. Here, the bronze color of the GIC can be expected as the primary reaction step. This must be preserved. leads to the generation of radicals in the proximity of the graphene sheet, with subsequent covalent bond VIII.1.1.3. Type of GIC and electrophilic trapping formation [1260, 1261]. In this line, the diazonium/ reagent iodonium coupling reactions are more expedite than Several electrophiles have been utilized in solid support the alkylation, because the generated phenyl radicals and in homogeneous dispersion functionalization of are significantly less stable than the alkyl ones due graphenides, like the diazo coupling [1253], iodonium to the much higher bond strength of aromatic C–H coupling [1240, 1258], alkylation [368, 1237–1239], bonds [1262]. This higher stability, as well as improved arylation [370, 1241], hydrogenation [369, 1235, solubility, of the resulting functionalized graphene in 1236], halogenation [1259], and silylation [1255]. In DME, plays the major role during the functionalization addition, by an iterative repetition of the charging/ process, favoring the reaction of hexyliodide with 162 2D Mater. 7 (2020) 022001 C Backes et al new graphenide activated surfaces, leading to achieved after intercalation/activation of graphite via higher functionalization and explaining the narrow GICs, thermal or photo-induced cycloadditions of SLG Raman I(D)/I(G) distributions [371]. In contrast, with nitrene and carbene intermediates are employed faster reactions (i.e. diazonium) lead to broader [1263]. However their long (~3 h) reaction time and/ distributions, probably related with a clustering of the or low yield (<10% surface functionalization) hinders functionalities [371]. The open question regarding a facile and useful covalent functionalization. the role of remaining negative charges after the initial Functionalization can be enhanced using elec- electrophile addition step was also addressed [1242]. trochemistry [1263]. The electrochemical potential According to Ref. [371] the role of the remaining applied can shift the EF of graphene, graphite, glassy negative charges can be neglected. This is in contrast carbon or any conductive substrate, increasing their to unfunctionalized graphenides. The exposure of reactivity as compared to a direct attack of the covalent these air-sensitive intermediates towards air, leads to a sp2 bonds with aggressive chemicals. Ref. [1263] dem- covalent framework modification with an attachment onstrated that electrochemistry can also be used to of OH- and H- groups [375, 1243]. functionalize SLG with organic molecules, and that the Refs. [368–374] demonstrated that GICs are suit- functionalization can be controlled on the nm scale. able for the covalent functionalization of graphene. In In contrast to earlier work on electrochemical func- typical covalent functionalization sequences, the nega- tionalization of graphite/graphene using diazonium tively charged graphene layers first act as reductants for salts [1264, 1265], a two-steps process was used that electrophiles, subsequently attacked by the intermedi- allows the independent control over the adsorption of ately generated organic radicals or H-atoms, yielding the molecules on graphene as ruled by supramolecu- the covalently modified graphene architectures. This lar interactions, and their successive covalent grafting wet-chemical functionalization concept is facilitated (figure VIII.2). The molecule used is 4-docosyloxy- by the fact that, due to Coulomb repulsion, the nega- benzenediazonium tetrafluoroborate (DBT), an aryl tively charged graphenide layers within the solid GICs diazonium salt comprising a long (C12) aliphatic can be dispersed in suitable organic solvents [373]. One chain and a diazonium grafting unit. The alkoxy chains fundamental question is whether all negative charges promote the physisorption of DBT forming ordered of the graphenide intermediates can be controlled or patterns. Meanwhile, the highly reactive diazonium completely removed in such redox reactions [376]. salt head group allows the covalent attachment of DBT Only complete oxidation is expected to avoid reactions onto graphene, disrupting its sp2 lattice, modifying its with moisture and oxygen during workup, leading to optical and electronic properties [1263]. side products with undesired and additional oxygen- In Ref. [1263], DBT was physisorbed from solu- and hydrogen functionalities. More importantly, the tion onto SLG, allowing the molecules to self-assemble controlled removal of all negative charges from the on the surface into ordered monolayers. The maximal solvent-exfoliated graphenide intermediates with a amount of molecules deposited on the SLG surface suitable oxidation reagent allows for the bulk produc- depends on the packing density of the DBT monolayer, tion of defect-free graphene. Ref. [375] reported that and could be as low as few ng cm−2. After drying in air, a the treatment of K intercalated graphite with benzoni- polydimethylsiloxane (PDMS) circular mask was fixed trile (PhCN), leads to a quantitative discharging of the on the top of SLG along with an electrochemical setup individual graphenide sheets upon the formation of composed of a three-electrode cell: a Pt wire as CE, Ag/ the colored radical anion PhCN•- (figure II.18), which AgCl as RE, and the target substrate used as WE. The can be monitored by the accompanying exhaustive substrate was high-quality SLG obtained by CVD on and Coulomb force-driven migration of the interlayer Si, or on polymeric substrates. The PDMS mask con- potassium counterions (K+) from GICs into the sur- fined the electrolytic solution laterally giving a fixed rounding benzonitrile phase. The suppression of any reaction area ~0.1 cm2 on the CVD graphene, making reactions of dispersed graphenides with moisture and it possible to perform electrochemical treatments with air takes place when no treatment with benzonitrile is very small amounts of solution (10 µl). An acidic solu- provided, followed by generation of SLG. This could be tion (0.1 M H2SO4) was then deposited inside the mask confirmed by Raman spectroscopy and AFM of exfo- as the electrolyte, and DBT molecules were not soluble liated material on Si/SiO2, treated with benzonitrile, in this acidic aqueous solution. The grafting process leading to SLG flakes over a lateral grain size of 10–18 was done by ramping the voltage from  +0.2 to  −0.7 µm. This represents a rather mild and inexpensive V at 100 mV s−1. After the electrochemical reaction, method for the wet-chemical graphene production functionalized SLG was sequentially washed by DI on a small scale. This exfoliation approach was also water and dried under a gentle flow of nitrogen gas. extended to water as solvent [377]. The versatility of this approach was demonstrated by using it on different carbon-based materials, graph- VIII.1.2. Electrochemical functionalization ite, glassy carbon and SLG on SiO2, PET, or quartz The high chemical stability of the SLG basal plane [1263] utilizing a homemade electrochemical setup requires highly reactive conditions for its covalent (figure VIII.2(d)). Besides the easy packing of ali- modification. While reductive functionalization can be phatic chains, the two-step approach could be used in 163 2D Mater. 7 (2020) 022001 C Backes et al Figure VIII.3. Preparation of a vertical casting system for the evaluation of a graphene dispersion. principle also with more complex patterns, e.g. based reactions [375]. In both cases, it is important to work on arrays formed by adsorption of alternating com- under dry and inert conditions to achieve maximum plementary building blocks, or nanoporous organic control over the grafted functional groups. Both frameworks, paving the way to a versatile, ordered and functionalization strategies can in principle be applied simple route to functionalize such a technologically to a range of graphite/graphene precursors, even important, yet poorly reactive material. The maxi- though electrochemical functionalization is often used mal amount of molecules deposited on the graphene for CVD-grown graphene or graphene grown from surface depends on the packing density of the DBT SiC [1778], while reductive bulk functionalization is monolayer and can be tuned using molecules of dif- mostly employed for graphite powders [1238]. ferent packing density [193]. Combining the DBT A challenge is the control over the degree of func- surface density (as measured by STM) and the DBT tionalization and the homogeneity of surface modifi- molecular weight of 401 g/mol (with no diazonium cation. In the case of reductive bulk functionalization group), a DBT coating ~35 ng cm−2 could be estimated via GICs and graphenide dispersions, the functional- for a uniform, perfect coverage, although the defects ized flakes are extremely polydisperse, in par­ticular observed in the grafted layer will significantly modify when using natural graphite as starting material this theoretical value [193]. [1238]. Electrochemical functionalization should in Other examples of electrochemical functionali- principle offer a better control since the charge den- zation have been reported. E.g., Ref. [1266] showed sity can be controlled via the applied potential. How- that amidation can be achieved directly after electro- ever, as discussed in Ref. [1263], in practice, sp3 defects chemical exfoliation in nitrogen containing electro- and functional groups are spatially poorly controlled lytes such as (NH4)2SO4, and NH4NO3. RGO was also resulting in a highly disordered morphology. This functionalized electrochemically with catechol deriva- has been overcome by combining electrochemical tives [1267]. Electrodeposition of Au NPs selectively functionalization with self-assembly of the addends at edges of CVD graphene was demonstrated [1268]. [1263], a strategy that has not yet been successfully In Ref. [1269], it is highlighted that arylmethyl groups shown in reductive bulk functionalization. A down- can be reversibly grafted electrochemically onto gra- side of electrochemical functionalization is that the phene with controlled functionalization by adjusting parameter space that can be explored is even wider the electrochemical conditions, rather than the con- than in reductive bulk functionalization, including centration of the reagent α-naphthylacetic acid. parameters such as applied potential in combination with the electrolyte. Due to the required presence of VIII.1.3. Reductive bulk versus electrochemical func- the electrolytes, it is probably even more challenging tionalization to suppress unwanted side-reactions than in reductive Reductive bulk and electrochemical functionalization bulk functionalization. of graphene both rely on the principle that the In summary, both of the discussed covalent reduc- reactivity of graphite can be enhanced though tive functionalization routes have seen remarkable activation introducing negative charges on the progress. Currently, reductive bulk functionalization lattice which then readily react with electrophiles through chemically charged species is more widely [160]. Negatively charged graphene/graphite is used and the more versatile strategy enabling grafting highly reactive, so that it is difficult to suppress side- of a range of functional groups via electrophilic addi- 164 2D Mater. 7 (2020) 022001 C Backes et al tion reactions [160]. Compared to this, electrochemi- Hagihara coupling reaction [1281] of 1,3,5-triethy- cal functionalization is still in its infancy, but definitely nylbenzene and halogenated aromatic monomers, holds great promise due to the better control of the by using RGO-Br as structural-directing template. charge density on the graphite/graphene. 1,3,5-triethynylbenzene was selected as key monomer, mixed with another monomer of aryl halides (such as VIII.2. Functionalization of GO 2,5-dibromothiophene, 2,5-dibromo- 1,3-thiazole, VIII.2.1. Covalent derivatization and 2,6-dibromopyridine) or halogenated BODIPY GO is rich in polar oxygen-containing functional (2,6-diiodo-1,3,5,7-tetramethyl-8-phenyl-4,4-dif- groups that make it highly soluble in a wide range of luoroboradiazaindacene) in the RGO-Br dispersion. solvents [216]. The different functional groups (i.e. The above mixture polymerized at 80 °C for 72 h cat- hydroxyl, epoxy, carboxylic, lactones, phenols, etc) can alyzed by tetrakis-(triphenylphosphine) palladium be exploited to prepare chemically functionalized GO ([Pd(PPh3)4]), copper iodide (CuI), and triethylamin for a wide range of applications [160, 1270]. Due to the (Et3N), forming a series of CMPs on graphene-based presence of the different oxygenated functional groups template (RGO@CMPs). The as-prepared insolu- and their high chemical reactivity, simultaneous ble precipitated polymer networks were filtered and reactions of these functions may occur, leading to washed several times with chloroform, water, and ace- uncontrolled GO derivatives. Most of the reactions tone to remove any unreacted monomers or catalyst reported in literature concern functionalization of residues. Further purification of the polymer networks carboxylic functions via amidation, using a variety was carried out by Soxhlet extraction with acetone for of amine derivatives [1226, 1271]. However, other 48 h. The powder was dried in vacuum (0.1 mbar) for functional groups can participate. E.g., epoxy 24 h at 60 °C to obtain the final product. ring opening can take place in the conditions of This method not only offers a new way to polym- amidation, as epoxides are highly reactive towards erize on RGO surfaces through the Sonogashira-Hagi- nucleophiles like amines [1272–1278] so that precise hara coupling reaction, but also paves the way for ver- control is lacking. To clarify the chemical reactivity of satile reactions on the RGO surface in organic solvent. carboxylic acids and epoxides, Ref. [1279] elucidated GRMs are ideal for energy applications because the structure of GO in the presence of amines and the of their large accessible specific surface areas and high reactivity of the functional groups towards amines. electrical conductivity [563].The SLG theoretical spe- Magic angle spinning solid state NMR allowed to cific surface area is ~2600 m2 g−1 [1282]. However, a prove the absence of reactive COOH groups. As a common problem is the restacking of SLG sheets, i.e. consequence, the amidation reaction of carboxylic the aggregation during wet-processing, resulting in a groups with amine derivatives did occur to a negligible decrease of accessible specific surface area [1283]. E.g. extent, while the main reaction consists in epoxide the reduction of unmodified GO typically leads to ring opening. This is also the case in the presence the irreversible aggregation of RGO sheets [1284]. In of coupling reagents necessary to form the amide order to overcome this issue, the bridging of GO lay- bonds. Therefore, the reaction between GO and ers into macroscopic porous structures is a key step to amine functions involves mostly epoxy groups, and modify SLG-based multifunctional materials, which not carboxylic acids, present in a small amount. This could preserve many of the unique properties of the situation cannot be generalized to all types of GO. Ref. individual SLGs [1285]. In particular, the high specific [1279] underlined the importance to well characterize surface area and porous structures can provide space the starting material and use the appropriate controls for hosting electrolyte ions, therefore increasing the (e.g. the conditions for amidation versus the epoxide performance in supercapacitors [1286]. In this con- ring opening reaction) to prove the reactivity of the text, graphene-based networks through covalent link- different oxygenated functions on the surface of GO. A age between individual SLGs is one among the greatest better understanding of the reactivity of GO is essential challenges, with the aim to achieve a high control over for controlled derivatization. the structures with a nanoscale precision and conse- Although polymerization on the GO surface in quently tuning of the material’s physical properties polar solvents, e.g. water, has been widely studied, it [1287–1289]. remains a great challenge to achieve polymerization A method for the facile preparation of graphene- reactions on RGO in organic solvents due to the lim- based covalent networks (G3DCNs) with adjustable ited functional groups for dispersing RGO in organic interlayer distance upon covalent functionalization solvents. RGO was produced from the reduction of of GO has been reported [1290]. This relies on the GO by hydrazine hydrate [1280]. We found that RGO condensation and ring-opening reactions of the car- can then be functionalized using p -bromobenzene bonyl and epoxy groups on GO with benzidine (BZ) at diazonium salt under aqueous conditions to ­produce different T. In Ref. [1290] the controlled polymeriza- p -bromobenzene functionalized RGO (RGO-Br). tion of BZ with GO, to obtain the G3DCNs, was per- This was dispersible in DMF for further reaction. formed under catalyst- and template-free conditions. Afterwards, conjugated microporous polymers By varying the reaction temperature, BZ monomers (CMPs) were prepared through the Sonogashira- or polymerized BZ units can bridge GO sheets to form 165 2D Mater. 7 (2020) 022001 C Backes et al covalent networks with tunable interlayer spacing. lytic activity of horseradish peroxidase (HRP). GO was The reduced form of G3DCNs (RG3DCNs) was used functionalized with two reducing substrates of HRP: to develop high-performance supercapacitors, tak- coumarin (7-hydroxy azido coumarin) and catechol ing advantage of the high specific surface areas (280 (3,4-dihydroxy benzoic acid) derivatives. The kinetic m2 g−1) combined with N doping obtained through of the biodegradation process was enhanced in com- chemical functionalization. A three-electrode configu- parison to unmodified GO. ration gave~460 F g−1 at a current density ~0.5 A g−1, while a two-electrode configuration resulted in VIII.2.2. Noncovalent derivatization of GO for ther- ~156 F g−1 at a current density ~1 A g−1 in combina- moplastics tion with a stability over 5000 cycles. This combination When GRMs are incorporated into a polymer matrix, of physicochemical stability and high supercapacitor one of the most important problems is the lack of performance demonstrates that integration of GO affinity between them, due to differences in polarity into covalent tuneable structures is a promising route [1294]. Just a few polymers, such as polyaniline and to obtain multifunctional graphene based hybrids. Kevlar, contain aromatic rings in their structure Most methods towards porous carbons mainly [1783, 1784]. π–π interactions between these chains focused on a hard template and activation approaches and SLG/GO can take place [1295], however most of [1782], which suffer from poor rational design of the the polymers do not have these moieties that facilitate micropores and heteroatom components. In general, interaction with SLG/GO. the preparation of layered porous carbon remains a Re-agglomeration or bad interaction between challenge. In Ref. [1291, 1292] the as-prepared porous them and polymer chains leads to creation of defects polymers (RGO@CMPs) were directly pyrolyzed at X and discontinuity regions, and loss of mechanical per- °C (X  =  700, 800, 900 and 1000) for 2 h under inert formance, such as a decrease of the elastic modulus at atmosphere, resulting in heteroatom-doped (B, N or break [1695]. This affects the properties of the SLG/ S) porous carbon materials. This method not only GO and the composite does not show the desired per- provides a new way for preparation of porous carbons formance, i.e. electrical conductivity, thermal manage- without using any inorganic templates, but also offers ment, gas barrier properties, flame retardance or other a rational approach to control the heteroatoms by due to the creation of these agglomerates. choosing different heteroatom-contained monomers. To solve these problems, functionalization is Chemical functionalization of GO is a key step to needed to change the surface chemistry and improve prepare covalent conjugates that can be exploited not the compatibility with the polymeric matrix. Another only in for energy applications or in the biomedical effect of functionalization is the reduction of agglom- field, but also in the studies related to the impact of eration during the preparation of composites, due to such materials on health (i.e. biodistribution and bio- the electrostatic repulsion that can be produced [1296, degradation). In the context of pharmacokinetic stud- 1297]. Non-covalent functionalization can enhance ies, GO needs to be labelled with molecules that can dispersability, compatibility and binding capacity be traced once the conjugates are injected into a body with the matrices, with less impact on the structural [289]. For this purpose, radiolabeling is extremely properties of the graphitic sheets than covalent func- powerful, as the material can be followed using dif- tionalization, as this creates sp3 carbon centers [160]. ferent imaging techniques and many radionuclei can Non-covalent functionalization helps in networking be exploited. GO has been modified with the chelat- or connecting the molecules without forming cova- ing agent DOTA (1,4,7,10-tetraazacyclododecane- lent bonds. This process requires physical adsorption 1,4,7,10-tetraacetic acid) for complexation of the radi- of suitable molecules by forming van der Waals bonds onuclide [1 1 1]In to assess the in vivo biodistribution between functional groups and SLG, such as π–π inter- in mice [1293]. GO was initially functionalized with actions, electrostatic attraction, adsorption of sur- triethylene glycol diamine to open the epoxy rings and factants and polymer wrapping [1298]. introduce amino functions on its surface. Then, the When organic molecules or hydrophobic character amino groups were derivatized with an isothiocyanate polymers are the functionalities, van der Waal forces DOTA reagent. Structural characterization of both GO are created between them and SLG/GO [282, 1785]. and GO-DOTA revealed that the thickness of GO was When the functional groups are molecules with an increased from 1L to FL after functionalization. GO- extended π-system, the predominant interactions are DOTA was subsequently radiolabelled with [1 1 1]In π-π based. Besides, hydrogen bonds and ionic inter- and injected into mice to follow its organ distribution, actions can be involved due to the presence of oxygen accumulation and elimination. An appropriate func- groups, especially in the case of GO, and in minimum tionalization of GO can also be useful to modulate its percentage in the case of RGO or SLG [282]. biodegradability. Covalent functionalization was con- For the design of the non-covalent functionali- ceived by designing surface-functionalized GO with zation strategies, an important decision regards the the capacity to be degraded more effectively compared polymer matrix and the processing method. In the to unmodified GO [463] The surface of GO was tai- case of thermosets, most of the processes consist on lored with different ligands able to enhance the cata- liquid phase polymerization [1786]. For thermoplastic 166 2D Mater. 7 (2020) 022001 C Backes et al matrixes, most of the polymers are processed by melt of silane molecules and the hydroxyl groups of the mixing, and just in a few cases in situ polymerization graphene surface. It is also important to select a gra- can be an alternative [1786]. The polarity of the matrix, phene material with lower number of hydroxyl moie- from very nonpolar such as the polyolephines, to polar ties. Several silanes can be used, e.g. 3-(aminopropyl) such as polycarbonate needs to be taken into account. triethoxysilane, 3-glycidoxypropyltrimethoxy silane, Besides, the groups on the surface of GO or RGO are tris[3-(trimethoxysilyl)propyl]isocyanurate. Covalent of crucial importance for the functionalization strat- silanization is a common strategy for the functionali- egy. In this regards, XPS is a useful tool to understand zation of GRMs for their use in polymers [1308]. For the distribution and binding of the moieties and func- this, the GRMs need a high number of hydroxyl groups. tional groups over the GRM surface. This functionalization usually is done at RT, or under Many different organic molecules were used in reflux conditions in organic solvents such as toluene, literature [194, 218, 282, 1284, 1299–1302]. The selec- ethanol [1309]. Silanization of OH groups can also be tion is based on availability on the market, or ease to accelerated using microwaves: coupling microwaves synthetize them and scale up. For the preparation of (energy source) with graphite (support) is responsible thermoplastics and thermoset composites, even at lab- for a high T gradient leading to increased reaction rates scale, at least hundred g are needed. For scale-up at lab as compared to conventional procedures. or pilot level, few kgs are needed. A typical amphiphilic The most commonly used strategy for noncova- non-ionic surfactant, commercially available as Triton lent functionalization exploits π-π interactions. This X-100 (polyoxyethylene octyl phenyl ether, POPE) can has been discussed in several reviews and books [1327, be used to increase the compatibility of GRMs and the 1310]. Some organic molecules can be used for non- polymer matrix [1294]. covalent functionalization to improve the dispersion Aqueous surfactant solutions of RGO above the in polymers: i.e. melamine [1311], chitosan [1312], critical micelle concentration (CMC) are enough to glycidyl 2-methylphenyl ether, anisole, 4-ethyny- ensure physical adsorption and non-covalent adsorp- lanisole among others [282]. These were dissolved in tion. Aqueous dispersions of functionalized mat­erials organic solvents of slightly acidic solutions in the pres- remain stable during long (more than one month) ence of GRMs and agitated under bath sonication for periods, while pristine graphene suspensions may sed- 15–30 min, then filtered and dried [282]. iment in less than one day [1303], due to low viscous To establish the success of the functionalization and agglomeration problems. For laterally large (>20 in the dispersion of GRMs in thermoset composites, µm) RGO, sonication is needed, in bath or tip, or even a fast method is vertical casting (figure VIII.3). This a combination of both processes. For RGO of lateral employs two polished surfaces (e.g. glass) treated with dimension  >50 mm, two cycles of 10 min of tip and release agent or covered with peel ply. Few silicone 10 min in bath is our recommended process to achieve double tapes of 1 mm are glued at the same distance dispersion without significant decrease in lateral size. (e.g. 2 cm) to the polished surface. Another silicone After filtration and drying of the non-covalently func- tape is glued perpendicular to the other tapes to the tionalized RGO, the GRM can be dispersed in the poly- bottom surfaces. This system is used as the mould, and meric matrix. the epoxy-graphene dispersions are introduced. Polyvinylpyrrolidone, SDS, SBDS, CTAB and In the case of thermoplastics, a polymer adapted other surfactants can be added to water suspensions microtome which allows to obtain flat and smooth of GRMs to stabilize them thanks to the electrostatic slices or ultrathin slices of the composite is needed repulsions created [1294, 1304]. In most cases, inter- [1311] and cryo-microtomography, working at action occurs in the absence of ultrasound, which is temper­atures between   −15 to  −185 °C, to avoid dis- required to increase the accessible surface area and turbing the positions or relocating NPs along the sam- it is especially helpful when lateral size is medium or ple [1313, 1316]. The analysis of the graphene com- large (>20 µm). The same strategy has been used to posites’ mechanical and electrical properties is key to improve the dispersion of RGO in plasticized poly- understand if the functionalization works properly, to mers and in epoxy resins before curing [1297, 1305]. improve the dispersion and interface. The results of the dispersion have a strong influence on By optical microscopy, even at low resolution, it is the final properties of the composites. Epoxy compos- possible to characterize the effectiveness of function- ites prepared with this functionalized graphene mat­ alization, i.e. the homogeneity of the dispersion of GO erials have shown better interaction to the matrix in in an epoxy composite, and the rate of decrease of large SEM micrographs and improved mechanical behavior size aggregates. at very low loadings [1294, 1300, 1306, 1307]. Silanes can also be used for non-covalent func- VIII.3. Noncovalent functionalization of graphene tionalization of GO. From our experience, mild con- VIII.3.1. Functionalization in dispersion ditions [i.e. ethanol dispersions of GO and selected Graphene can be exfoliated in water with the help of silane are mixed under bath sonication, and filtered amphiphilic surfactants [182, 184]. Most published in air, or oven dried  <80 °C] are favoured to avoid works use ‘conventional’ aliphatic surfactants (i.e. covalent reactions with GO between the alkoxy groups soaps) [182, 184]. However, stable dispersions can 167 2D Mater. 7 (2020) 022001 C Backes et al also be obtained using small polyaromatic dyes as (0.8%). YW of LPE in docosane is ~1.35%, while doc- surfactants [194, 214, 215]. Thanks to their aromatic osanol and docosan-1-amine have lower performance core, these molecules can adsorb strongly on the (1% and 1.1%, respectively). The thermodynamic graphene surface, forming also ordered layers [193], analysis in Ref. [235] suggests that aliphatic chains noncovalently functionalizing SLG/FLG. They have functionalized with carboxylic acid groups promote also a strong and unique absorption (molar absorption the stabilization of exfoliated SLG and FLG in NMP coefficient ε  >  15 000 l g−1 m−1) and emission due to synergistic interactions between DSAs, surface, spectrum in the visible range, which allows to monitor and polar solvent. their interaction with graphene in solution and in solid Ref. [231] also reported that the performance of [216, 217]. Many of these molecules are also low cost linear alkanes exposing a carboxylic acid head group dyes, already widely used in large-scale compounding as stabilizers directly depends on the length of the of polymers, e.g. as industrial additives and colorants linear alkane chain. 5 linear modules were explored: [218]. These molecules can exfoliate not only graphene hexanoic acid (C6CA), lauric acid (C12CA), stearic but also a wide range of LMs, such as BN, WS2, MoS2, acid (C18CA), lignoceric acid (C24CA) and melissic selenides and tellurides with concentrations up to acid (C30CA), whose different adsorption energies 0.54 g l−1 [219]. on graphene and tendency to form tightly packed self- Several pyrene derivatives with a varying num- assembled monolayers on such a surface affect their ber of polar functionalities were used as exfoliation performances as stabilizers. The analysis of carboxylic agents [194]. A significant part of the solubilized acid assisted LPE revealed that the concentration of material was composed of SLG (up to 22%), with graphene dispersions prepared in NMP, ODCB and most of the remaining material composed of FLG TCB increases linearly with the length of the aliphatic (50% to 60%  <7L). The total concentration depends chain. The dependence of YW with the length of the ali- on the polar functionalization present on the pyrene phatic chain was interpreted by means of a thermody- core. Molecular dynamics calculations revealed that namic model of molecular self-assembly on graphene a critical factor is a one molecule thick solvent layer [231]. This showed that the shorter the aliphatic chain, present between dye and SLG [194]. The amphiphilic the larger is the (rotational and translational) entropic molecule changes its orientation when approaching cost of forming a 2d layered structure. These results the surface to slide into this layer, and the asymmet- suggest that a model based on molecular mechanics ric shape of the dyes facilitates this step. In these gra- for the energetics and a statistical mechanic treatment phene organic hybrid systems, colloidal stabilization of entropy, could be used to predict the efficiency of is achieved through electrostatic repulsion between supramolecular building blocks as stabilizers and charges introduced by the surfactant, and can be over- guide the chemical design of next generation stabiliz- come by changing pH or adding salts [194]. ers. Nevertheless, the role of kinetics cannot be fully ruled out. VIII.3.2. Exfoliation in organic solvents Ref. [234] reported that alkoxy-substituted pho- The use of small organic molecules acting as stabilizing tochromic molecules can act as photo-addressable agents is expected to promote LPE of graphite when stabilizers to enhance YW in an upscalable molecule- molecules have a stronger affinity to graphite/ assisted LPE-based method. The large conformational graphene than the solvent/graphene interactions. A change associated with the trans–cis photochemical good starting point in terms of molecular design relies isomerization of alkyl-substituted azobenzenes can on the use of alkanes, which exhibit a high affinity (~2 be used to improve YW. The simultaneous use of UV kcal mol−1 per each methylene unit) for the basal plane light, promoting the trans-to-cis isomerization, as well of graphite/graphene [1315]. Ref. [230] reported that as thermal annealing at 40 °C and mechanical forces arachidic acid (C19CA) can promote the exfoliation generated by sonication, both favoring cis-to-trans of graphite in NMP. The addition of C19CA does isomerization of 4-(decyloxy)azobenzenes, promotes not affect the quality and structure of the resulting the exfoliation of graphite in liquid media. The most graphene, as compared to NMP alone, highlighting effective exfoliation is obtained with azobenzene mol- the non-invasive nature of the process, but alkyl chain ecules irradiated with UV light in NMP at 40 °C, with based stabilizers led to an increase of SLG (25%) and a concentration of exfoliated graphene ~110 mg ml−1. FLG (70% for 2–6 layers) [230]. This corresponds to an ~80% increase in YW when The use of α-functionalized alkanes as stabilizers compared with pure NMP (63 mg ml−1). By deposit- during LPE allows one to increase YW in liquid media. ing the hybrid film onto Au pre-patterned SiO2 sub- To understand the role of the functional group in α- strates, light- responsive thin hybrid films, formed substituted alkanes, Ref. [235] used the C21H43 alkyl in a one-step co-deposition process, can be realized, chain as a scaffold and decorated it with simple, yet whose conductivity can be reversibly modulated by the chemically distinctive groups: methyl, alcohol, amine, trans-to–cis photoisomerization of the azobenzenes. and carboxylic acid. The most effective exfoliation was By combining this approach with cost-effective tech- obtained with docosanoic acid, with YW ~ 1.6%, with niques, such as ink-jet printing, more complex respon- a ~100% increase when compared to control samples sive device designs and architectures may be realized. 168 2D Mater. 7 (2020) 022001 C Backes et al VIII.3.3. Exfoliation with aromatic dyes in chloroform and THF, and further processing in polymers Graphene can be exfoliated in water, using surfactants, or in high-boiling solvents such as DMF or NMP. For an effective, technologically competitive application of GRMs as additives in (nano)composites for electronics or structural applications it may be preferable to solubilize graphene in low-boiling volatile solvents, like CHCl3 or THF. However, the use of amphiphilic surfactants is not suitable to solubilize graphene in organic solvents whose polarity is low [193]. Ref. [193] used perylene diimide (PDI) molecules soluble in organic solvents as ‘apolar surfactant’ for exfoliation and successive processing in polymeric commercial films of poly(vinyl chloride) (PVC), to make them conductive. The PDI has an extended poly­ aromatic core (which can interact via π–π stacking with the GRM) and flexible side groups, with low, but tunable polarity that make these molecules soluble in a range of organic solvents (figure VIII.4). Transparent sheets of PVC (thickness 182  ±  4 µm) were dipped into chloroform dispersions of FLG and PDI at RT, a procedure already used for rubber [568]. They were then dried in air, to let the chloroform evap- orate. The swelling process was completed in  <3 min, and gave an increase of ~80% in volume. No further change of the swollen area/size was observed when the polymer films were left in the dispersions for  >3 min. Figure VIII.4. (a) General formula of the PDI used for After the swelling treatment, the PVC samples showed exfoliation. (b) STM image of the typical packing of PDI a dark color due to the presence of the FLG as well as a adsorbed on graphite. (c) Average distance of the side groups bright fluorescence due to the PDI. of the PDI from SLG, from molecular modeling. In the inset, a typical snapshot of the simulation. Adapted from [193], SEM images showed a dense coating of rectangular © 2016 WILEY-VCH Verlag GmbH & Co. KGaA. and polygonal shapes covering the surface. The sheets were not just deposited on the surface, but embedded into it, in some case reaching some microns (<10 µm). be a reliable, stable way of halogenation, allowing SLG Control experiments by swelling in PDI only gave sur- doping without losing transparency [1316]. face covered by PDI crystals, and no improvement of In Ref. [1316] Rs ~ 30 Ω sq−1 using up to 5LG was electrical conductivity. Typically, when deposited by achieved, while keeping a high transparency  ⩾85%. conventional solution processing, GRM flakes tend Depending on whether the doping agent is an electron to lay flat on a substrate [272, 277]. Due to the swell- acceptor or donor, p- or n-type doping can be attained. ing, the flakes penetrate into the polymer in an iso- To achieve a high doping level in the range 1013 cm−2, tropic way, in some cases protruding out of the surface thionyl chloride (SOCl2) is recommended. SOCl2 [193]. An improvement of material hydrophobicity by induces p-type doping due to the higher electronega- ~40% was observed in all cases, with measurements of tivity of Cl compared to C. water contact angle increasing from 72° (blank PVC) CVD-grown graphene is first transferred on glass. to  >100°. Rs decreased significantly, going from com- Thermal release tape was used for the transfer process pletely insulating to ~105 Ω sq−1, a value useful, e.g. for [1316]. Different approaches for doping with thionyl applications in antistatic coatings [1787]. chloride have been tested: last layer doping, where the last or top layer of 1 to 5LG was doped after all layers VIII.3.4. Functionalization on substrate were transferred in a staggered fashion, or interlayer VIII.3.4.1. Doping of CVD graphene via noncovalent doping, where the top layer after each transfer of SLG/ interaction with thionylchloride FLG is doped creating a doped surface in between the Halogenation, i.e. doping of graphene with halogens, is layers. The doping is achieved as follows [1316]: SOCl2 a process in which halogens, such as fluorine, chlorine, treatments are performed in a dry chamber by plac- bromine and iodine, are either used to replace one ing graphene/glass and 1 ml of liquid SOCl2 (avoiding of the carbon atoms or are attached to the graphene direct contact) at 105 °C for 60 min. Doping of multi- surface [1316–1320]. As for all doping or chemical layer samples was performed by repeating the SOCl2 modification, processes stability is of importance. treatment after transferring and stacking each SLG/ Here the focus is on chlorination, as this has proven to FLG. 169 2D Mater. 7 (2020) 022001 C Backes et al Figure VIII.5. (a) Comparison of Rs of chlorinated and non-chlorinated CVD graphene as function ofN. (b) Transmittance for 1 to 5L G. Adapted from [1316]. Figure VIII.6. (a) 514.5nm Raman spectra of pristine, hydrogenated and chlorinated graphene after hydrogenation. (b) 514.5nm Raman spectra of chlorinated, Cl-SLG, and hydrogenated and chlorinated graphene (Cl-(H-SLG)) directly and after doping and after three months in ambient conditions. Rs is then measured in a van der Pauw configura- This shows that SOCl2 molecules selectively chlorinate tion (see section IX.3). Ref. [1316] suggested to use a defective sites in graphene. However, defects are always large sample area ~4  ×  4 mm2 to obtain a representa- counter-productive for achieving low Rs and high µ. tive Rs. Figure VIII.5(a) plots Rs as function of N for Thus, a compromise between doping and Rs has to be interlayer doped and undoped samples. Rs  =  1/neµ, made. It is therefore more favorable to abstain from the where n is the doping level, e the elementary charge introduction of defects by mild H plasma in order to and µ the mobility. The mobility is affected by defects, achieve lower Rs, instead of inducing structural defects therefore µ ~ 1/nD, where nD is the defect concentra- to which Cl will attach and increase the doping level. tion. It is thus important to know doping and defect The most critical point is the doping stability with concentration. SOCl2 allows a nucleophilic substitu- time. Figure VIII.6 shows the Raman spectra of pris- tion by Cl atoms, which selectively occurs on defective tine, chlorinated and defective graphene (induced by sites, without affecting the existing sp2 carbon bonds, H plasma) measured directly after chlorination and therefore, without introducing any new defects [1316]. after 3 months. Pristine graphene shows a Pos(G) In order to verify this doping mechanism a controlled and FWHM(G) ~1583 cm−1 and 14 cm−1, while amount of defects was introduced in graphene by H FWHM(2D) ~31 cm−1. I(D)/I(G)  <  0.1. Therefore, plasma [1316]. the doping level is estimated ~100 meV. Hydrogen- The quality of graphene after doping can be ated graphene on the other hand shows D, D′ and assessed by Raman spectroscopy. It is important to fol- D  +  D′ peaks as result of defect introduction via H low the correct procedure to extract the correct dop- plasma treatment with I(D)/I(G) ~ 1.65. The doping ing level and defect density [1321, 1322]. Graphene is estimated ~107 meV. This gives nD  =  4 * 1011 cm−2. with controlled amount of defects induced by mild H This confirms the presence of defects introduced by plasma showed a higher doping compared to a pristine, the plasma treatment with no change in doping. After reference sample doped with SOCl2 under the same chlorination of the plasma treated sample the doping conditions, but without being subjected to H plasma. is ~380 meV. This is higher compared to that achieved 170 2D Mater. 7 (2020) 022001 C Backes et al Figure VIII.7. (a) Raman spectra for high and low packing density perylene1 SAMs on as-transferred and pre-annealed SLG. Inset: Structure of perylene1. Water contact angles on (b) bare SLG, (c) low and (d) high packing density of perylene1 on SLG. Adapted from [1323] with permission of The Royal Society of Chemistry. by doping pristine SLG, suggesting that SOCl2 mole- for doped pristine graphene and ~380 meV for doped cules selectively chlorinate defective sites. The Raman hydrogenated graphene. After three months these are spectrum of chlorinated graphene after plasma treat- ~250 meV and ~350 meV. This shows that doping is ment shows a reduction I(D). This can be explained stable over time. by the doping dependence of the D peak [1321]. This UV–vis spectroscopy can be used to assess the gives nD ~3 * 1011 cm−2 consistent with that of the optical properties of the chlorinated samples. Fig- hydrogenated sample before chlorination. Therefore, ure VIII.5(b) reports the transmittance of 1 to 5LG although the chlorination process is able to increase interlayer doped graphene using thionyl chloride. The doping, the defect density remains unchanged, in transmittance is still ~85% even for the 5LG interlayer agreement with the chemistry of the SOCl2 doping doped sample, making this attractive for applications process. where low Rs and high transparency are needed. Considering that after chlorination, one chlorine Solution based processes for halogenation of gra- atom is attached to each defect and taking a charge phene were also proposed (see e.g. [1318, 1319]), Ref. transfer of 0.57 electrons [1317], i.e. each Cl takes 0.57 [1320] used CVD grown FLG on Ni and transferred it electrons from graphene to induce p-doping, the dop- on a glass substrate to induce doping by bromine vapor ing can be estimated from the defect concentration. As at RT in inert atmosphere: a glovebox with  <0.1 ppm a result, the estimated defect density evaluated in hydro- of O2 and H2O and nitrogen filled glovebag with genated graphene corresponds to a doping ~65 meV. 2000 ppm of O2 and 2500 ppm of H2O. Ref. [1320] This is the doping level introduced via additional defects reported Rs ~ 180 Ω sq−1 with a reduction in transmit- by plasma etching. The doping of a chlorinated pris- tance ~2%–3% with respect to pristine SLG. A drop of tine graphene sample was ~304 meV. This is the doping Rs from 1548 Ω sq−1 to 602 Ω sq−1 after exposure to Br introduced directly via chlorination. Adding these two for 60 min was reported [1320]. contributions to the doping coming from only chlorina- tion, the final doping level is ~370 meV, in agreement VIII.3.4.2. Functionalization with polycyclic aromatic with the ~380 meV of the hydrogenated and chlorinated hydrocarbons sample estimated from the Raman spectra measured Non-covalent functionalization of CVD graphene directly after plasma treatment and doping with SOCl2. transferred to SiO2 was achieved via drop-casting of To check stability, figure VIII.6(b) compares the an aqueous solution of perylene bisimide 1 [1323]. spectra of chlorinated (Cl-SLG) and chlorinated The packing density of the perylene self-assembled hydrogenated graphene (Cl-(H-SLG)) recorded after monolayer (SAM) was related to the cleanliness three months. The initial doping levels are ~304 meV of the SLG surface. Vacuum annealing prior to 171 2D Mater. 7 (2020) 022001 C Backes et al Figure VIII.8. STM images of (a) as-transferred (top) and subsequently annealed (bottom) CVD-SLG on SiO2 and (b) after wet- chemical deposition of perylene1 on the same annealed substrate. (c) line profiles from (b) and (d) zoomed in area with molecular structure overlay. Adapted from [1323] with permission of The Royal Society of Chemistry. functionalization removes polymer and leads to together with the perylene cores perpendicular to the a much higher denser perylene packing density SLG surface, figure VIII.8. This is at odds with the com- than on untreated SLG [1323]. This is illustrated mon belief [1324–1326] that organic molecules with when the Raman spectrum and water contact angle large conjugates systems will adsorb via π–π interac- measurements are considered (figure VIII.7). The tions between the conjugates core and SLG/FLG. This increased relative intensity of the perylene Raman suggests that the method of deposition has a large peaks (pale shading) when compared to the SLG peaks bearing on the adsorption mechanism, as the liquid (dark shading) for the pre-annealed SLG versus the deposition process discussed here results in signifi- as-transferred one is indicative of a greater quantity of cantly more neighboring molecule-molecule interac- perylene on the surface. Likewise, high density packed tions [1323] than another techniques, such as thermal perylene on graphene has the smallest water contact evaporation. angle (58 °) compared to the as-transferred graphene This non-covalent functionalization was taken a sample (89 °) and low density perlylene sample (67 °). step further by chemically modifying the functional Again, this increased hydrophilicity indicates a greater end groups, as well as showing that perylene func- number of molecules on the SLG surface for the pre- tionalization can successfully be carried out prior annealed film. to polymer transfer to arbitrary substrates, creating STM can be used to estimate the SLG cleanliness a functional layer transfer (FLaT) [1327]. The lat- before and after annealing, figure VIII.8. An image of ter finding is important, as it allows perylene SAMs the resultant high packing density perylene SAM is in of very high packing density to assemble on pris- figure VIII.8(b). Analysis of the periodicity of the line tine SLG without having to anneal the as-transferred profiles indicates that the perylene molecules pack SLG to remove any polymer residues as depicted in 172 2D Mater. 7 (2020) 022001 C Backes et al Figure VIII.9. (a) Schematic FLaT, in which perylene fucntionalised graphene is transferred b-d) XPS spectra of perylene on CVD-SLG for each derivitization step. (b) As-transferred perylene1 on SLG. Inset: Partial structure of perylene1 with corresponding nitrogen contributions. (c) Perylene1 on SLG after reaction with HATU and EDA. The two new contributions are from HATU residue. Inset: Chemical structure of HATU. (d) After annealing of derivatized sample at 220 °C. The HATU contributions reduce, but the amine signal increases. Inset: Partial structure of perylene1 after reaction, depicting corresponding nitrogen contributions. Adapted from [1327] with permission of The Royal Society of Chemistry. Figure VIII.10. Noncovalent chemical functionalization of SLG. (a) Schematic of the fabrication of NH2-CNM/SLG heterostructures via mechanical stacking of the individual sheets. (b) Microscope image of the fabricated FET with Au contacts in the Hall bar geometry. (c) Mapping of the Raman 2D peak on the active device area. (d) RT homogeneous electrical field effect at different side contacts of the FET as a function of VG. (e) µ of heterostructure and reference devices. (f) Quantum oscillations in a magnetic field at low T confirming high electronic quality of SLG in the heterostructures. Adapted from [1328] © Wiley-VCH Verlag GmbH & Co. KGaA. 173 2D Mater. 7 (2020) 022001 C Backes et al figure VIII.9(a). Reacting ethylene-diamine (EDA) of the quantum Hall effect at low T, figure VIII.10(f), with the carboxylic acid groups of the perylene1 demonstrate the high electronic quality in the SAM using O-(7-azabenzotriazole-1-yl)-N,N,N,N′- engineered NH2-CNM/graphene heterostructures tetramethyluronium hexafluorophosphate (HATU) [107, 1328]. was carried out to investigate the possibility of con- The NH2-CNM/graphene FETs are promising for verting the carboxylic acid groups to amines [1327]. biosensors. As amino groups can further be function- This would allow the use of perylene bisimide 1 as alized with respective receptors, one could use NH2- anchor in SLG-based biosensing. CNM/SLG FETs as analyte-specific electrochemical Figures VIII.9(b)–(d) show that after reacting per- sensors. GFET biosensors have a very high sensitiv- ylene molecules with HATU and EDA new contrib­ ity up to the fM level [285]. However, this is difficult utions can be seen in the N1s core level XPS spectra. to achieve in combination with high selectivity and The relative intensity of the peak component arising specificity of the biomolecular binding events. While due to the presence of amine groups increases (pink), any covalent SLG functionalization may affect the along with other contributions likely due to HATU electronic quality, a functionalization via physisorp- residues (grey). Annealing the derivitized sample at tion typically involves larger distances leading to the 220 °C was found to reduce the relative intensity of reduced sensitivity. Woszczyna et al [1329] used 1 nm these peaks, indicating that the any unreacted phy- dielectric and specifically functionalized CNMs to sisorbed surface contaminants are removed [1327]. overcome these problems. It is also promising to use These results show that it is possible to use perylene them in graphene-based electronics as a complemen- functionalized graphene as the basis for further chemi- tary dielectric, similar to hBN, for engineering top- cal modification, towards biosensing applications. gate electrodes, for field-effect tunneling transistors or The non-covalent functionalization, can then be, in flexible electronic applications. if needed, transformed also in covalent functionaliza- tion, triggering the covalent grafting of molecules self- VIII.4. Defect functionalization of GRM assembled on graphene using, e.g., an electrochemical VIII.4.1. Edge modification of GRM in dispersion pulse (see section VIII.1 Electrochemical Functionali- Edge selective functionalization would be of zation) [1263]. particular interest to obtain engineered samples able to further react or assemble in a controlled way. Edge VIII.3.4.3. Noncovalent functionalization with CNMs functionalization is in principle driven by the reactivity Noncovalent functionalization can be used to fabricate of the carbon atoms localized at the peripheral edges, heterostructures of graphene and CNMs [1328]. which differs from the inertness of the C atoms on Figure VIII.10(a) shows a route to noncovalent the basal plane due to the stability of the extended chemical functionalization via assembly of all- delocalized π-system. In fact, edges may be regarded carbon vertical heterostructures consisting of amino- as defects in the graphene structure, which can be terminated CNMs (NH2-CNM) on SLG produced exploited for its functionalization without severely by CVD on Cu foils [1328]. The chemically active disrupting the π-conjugated network [1329, 1330]. amino groups of NH2-CNMs are located in these Edge functionalization may also be carried out on non heterostructures in close vicinity to the SLG plane, as graphene LMs in general [1246, 1331]. they are separated from it only by the ~1 nm dielectric The diazonium chemistry of carbon nanostruc- CNM [81]. As shown by complementary spectroscopy, tures is one of the most convenient methods to cova- microscopy and electric transport measurements, lently functionalize the edges of graphitic materials the pristine SLG properties in these heterostructures [1332–1335]. This reaction proceeds through a free- remain unaffected [1328], which opens avenues radical mechanism as a result of transfer of a delo- for implementations in graphene-based electronic calized electron from a graphitic substrate to the aryl devices. In figure VIII.10(b), an optical microscope diazonium cation to form an aryl radical after the image of an NH2-CNM/graphene FET in the Hall release of nitrogen [1246] (figure VIII.11). bar geometry is presented. The structural quality of The standard protocol [1330], which generates the the device characterized by Raman mapping (figure reactive diazonium species in situ from a stable anilinic VIII.10(c)) shows homogeneity. Figure VIII.10(d) compound in the presence of an alkyl nitrite (radical presents RT electric-field effect measurements as a initiator), was followed by performing the reaction in function of back-gate voltage, VBG, at 4 side contacts, DMF with different equivalents of the aniline deriva- figure VIII.10(b), P1–P4. The electrical characteristics tive per carbon atom. In Ref. [1337], graphite flakes are homogeneous on the scale ~3500 µm2. µ is nearly with lateral size ~10–50 µm and thickness ~10 nm, a factor of two higher (~3000 cm2 V−1 s−1) in the were first dispersed in DMF by bath sonication. Then, heterostructure devices in comparison to the starting 4-aminophenol was added and the mixture sonicated. SLG (~1500 cm2 V−1 s−1), figure VIII.10. This may After that, isopentyl nitrite (6 equiv. per carbon atom) results from a reduction of charge impurities at the was slowly added to the dispersion and the reaction SLG/SiO2 and SLG/ambient interfaces. Shubnikov–de mixture was kept at 80 °C for two different reaction Haas oscillations accompanied with resistivity plateaus times, 24 and 48 h. After cooling to RT, the reaction was 174 2D Mater. 7 (2020) 022001 C Backes et al Figure VIII.11. Free-radical mechanism for edge-selective functionalization using diazonium species. Adapted from [1336], copyright American Chemical Society. Figure VIII.12. Edge-selective functionalization with 4-aminophenol. quenched by pouring the mixture into distilled water tives. This opens new possibilities to modulate physical and filtration of the GRM dispersion through a PTFE and electronic properties for diverse applications, such membrane. The filtered cake was redispersed in DMF as catalysis and opto-electronics [1335, 1339]. Simi- by ultrasonication and filtered through a PTFE mem- lar functionalization can be used to obtain building brane. This sequence was repeated twice with DMF, blocks for the design and manufacturing of functional distilled water, methanol and diethyl ether, before materials. obtaining the product after drying (figure VIII.12) [1337]. VIII.4.2. Functionalization of SLG on substrate after Functionalization of the flakes was confirmed by introduction of defects a dispersibility test in an organic solvent, where the Ref. [1340] functionalized graphene on SiC with functional groups present good solubility. Pristine and organic molecules forming covalent bonds. This is functionalized flakes were dispersed in IPA at a con- alternative to wet-chemistry protocols that reported centration of 0.1 mg ml−1 and sonicated for 30 min. covalent modification methodologies incompatible After 2 days, chemically modified flakes with 2 and 4 with some functional groups due to harsh conditions equiv. of 4-aminophenol during 24 and 48 h, respec- (such as high thermodynamic force, generally exceeding tively, precipitated from the dispersion. The function- 20 kcal mol−1, or long reaction times, even more than alized sample prepared with 4 equiv. of the anilinic 12 h) used e.g. diazonium salt, click chemistry or surface compound using a reaction time of 24 h displayed bet- reactions [1227, 1232, 1244, 1253, 1341–1344]. ter dispersibility and was the only sample that did not Ref. [1340] reported the controlled a­ tom-molecule precipitate after a few days in IPA, which suggests the substitution in ultra high vacuum (UHV, P  <  2  × presence of a higher concentration of hydroxyl groups. 10−10 mbar) by using p-aminophenol molecules, The chemical composition of pristine and chemi- consisting of an aromatic ring with an amino and a cally modified flakes was investigated by XPS. Upon hydroxyl group, linking the amino group to the gra- functionalization, both the C–C sp2 content and the phene network and leaving free the hydroxyl group. π-π* shake-up band decrease due to the disruption This method was also successfully carried out on gra- of the delocalized π conjugation in the gra- phene grown on Ir. phitic structure, while the content of C–OH/C–N The substrate was degassed by UHV annealing at and  >C=O/C  =  N increased. The O/C ratio increases 200 –400 °C (details depending on the type of sam- from 1.7% to 8.2%, while the N concentration is ~3%– ple) in order to remove physisorbed contaminations 5%, ascribed to adsorbed molecules or residual sol- [1340]. This thermal treatment was performed by elec- vents [1332, 1338]. tron irradiation in a heating stage placed at the manip- The degree and selectivity of edge functionaliza- ulator, while the annealing T was monitored with an tion is influenced by N, edge states, degree of exposure IR pyrometer (emissivity ~0.9). After degassing, the of the interior basal planes and defects [1330, 1332]. cleanness of the sample was confirmed by STM [1340]. In Ref. [1337], edge selective functionalization was The next step consists in creating single atom maximized for 4-aminophenol and these conditions vacancies (SAV) by bombarding SLG with low energy can be extended to para-substituted aniline deriva- ions (Ar+). For this procedure, the sample needs to 175 2D Mater. 7 (2020) 022001 C Backes et al Figure VIII.13. STM images of the p-aminophenol molecules covalently anchored to the monovacancies created by the soft irradiation of the surface with argon ions in both types of samples: SLG (upper panel) and QFSLG (lower panel). The graphene on SiC was grown as for Ref. [611]. Note that in the images of the SLG sample one can see the (6  √  3  ×  6  √  3)R30° reconstruction from the SiC buffer layer, whereas in the QFSLG sample only the graphene network is seen since intercalated hydrogen has decoupled the buffer layer. The molecules appear as protuberances inserted into the graphene network. Partially adapted from [1340]. be placed close to the ion gun (10 cm maximum) at a bonding in the latter case. The thermal stability of the 0°–15° incidence. The acceleration of the elec- bond was tested up to T  =  250 °C, figure VIII.14(b)). trons inside the Specs IQE 11/35 ion gun should be Even after annealing at this T, the XPS N 1s does not 100–140 eV, 10 mA, having a controlled Ar pressure shift indicating the absence of breaking up of the A-AP ~1  ×  10−7 mbar. The most important parameter is molecule and that the covalent linked N remains in the the exposure time, that varies between 30 s−2 min for graphene network. producing the desired density of mono-vacancies Functionalization reaction mechanisms have [1345–1348]. been computed when the p-AP approaches the SAV After irradiation, the sample is subjected to a thermal either by the –OH or –NH2 terminating groups flash to remove the physisorbed Ar. This flash should not [1340] within the CI-NEB approach [916, 917] to get be longer than 5 min, reaching different T for each type a converged minimum energy path. For approach of sample: 550 °C for SLG and 250 °C for QFSLG. It is by the –NH2 group, the molecular incorporation important to check the mono-vacancies by STM as soon undergoes by a double (–NH2) dehydrogenation (see as possible. Defects are very reactive and they could get ­figure VIII.15). This process is efficiently catalysed passivated with air contamination [1349, 1350]. by the SAV on the surface yielding energy barriers After modifying the graphene network, the evap- ~ 1.3 and 0.7 eV for the first and the subsequent second oration of organic molecules (P-aminophenol) fol- dehydrogenation, respectively, and a total enthalpy lowed. They need to be purified by turbo pumping for gain ~ 3 eV. The net enthalpy in the double (–NH2) at least 6 h to remove impurities. Following this strat- dehydrogenation (~1.5 eV) is significantly higher than egy, Ref. [1340] tested by STM that the organic mole- that corre­sponding to a single dehydrogenation of the cules filled most of the SAV defects, as for ­figure VIII.13. OH group, pointing to a favoured incorporation of N atoms in graphene. VIII.4.3. QFSLG In order to check the covalent nature of the bonding VIII.5. Decoration with nanoparticles of the organic molecules to the graphene lattice several The large surface area, electrical conductivity and experiments were conducted. Figure VIII.14(a) shows mechanical strength make SLG a promising substrate the XPS N 1s peaks corresponding to a p-aminophenol to be coupled with NPsfor synergistic benefits dosing of a sample with and without previous SAV in applications such as catalysis, energy storage, defects. This points out the covalent nature of the plasmonics and optoelectronics or solid state hydrogen 176 2D Mater. 7 (2020) 022001 C Backes et al Figure VIII.14. (a) N 1s core level spectra after dosing p-AP on a SLG surface without and with atomic vacancies. The N 1s binding energy corresponds to that of a C atom substitutional in the graphene-network. (b) thermal estability of the N 1s peak at RT and after annealing at ~250 °C. storage [1351]. Depending on the Fermi level/work strength of the nanocarbons. Fulleride, nanotubide function of SLG and metal NPs, interfacial charge and graphenide can be generated following reduction transfer can occur from SLG to NPs or vice versa. using alkali metal/liquid NH3 [1233, 1353, 1354] Hence, decorating with NPs is an efficient method for alkali metal naphthalide/THF [1355, 1356], or graphene doping with electrons or holes. To enhance electrochemically in non-aqueous electrolytes [1357]. interfacial charge transfer, a strong interaction All reductive charging techniques insert electrons between graphene and NPs is required [1351]. SLG can into the carbon π* orbitals, shifting EF, resulting in be decorated with NPs either by growing NPs directly increased reactivity. Subsequent chemical reaction on its surface, by mixing it with NPs, or by depositing may take place via redox or electrophilic addition NPs produced in the gas phase in UHV conditions. reactions that are believed to involve SET (i.e. radical based reactions) [1261, 1358–1362]. Due to the high VIII.5.1. Graphene-metal NPs reactivity of reduced nanocarbons towards oxygen and Graphene and other surfaces can be decorated with moisture in the atmosphere, these materials should be NPs via electroless plating [1352]. CNTs/SLG act as handled using inert gas filled glove boxes or Schlenk reducing agents towards metal salts. The Fermi level techniques [1363]. of nanocarbons allows them to spontaneously reduce Ref. [376] used graphenides to reduce a series of Au3+ and Pt2+ salt solutions to form their respective Cu, Zn and Mg salts/complexes, to generate metal NPs M0(metal) NPs on the nanocarbon surface [1352]. on the SLG surface. KC8 and KC24 potassium-graphite To extend this to other metal NPs, reductive charging intercalation compounds (K-GICs) were produced protocols can be used to increase the reducing via the vapor transport [1363] from natural graphite, 177 2D Mater. 7 (2020) 022001 C Backes et al Figure VIII.15. (a) Schematic representation of the approaching of a generic NH2-terminated organic molecule towards a graphene SAV (with a magnetic moment of 1 µB). (b) For the case of p-AP molecule, most representative geometries for initial state, transition state and final state along the minimum energy path (MEP) for: (top) the first single dehydrogenation of the molecular –NH2 group, and (bottom) the second single dehydrogenation of the remaining molecular –NH group and the subsequent incorporation of the molecule via the N atom into the graphene lattice via the graphene SAV. giving the corresponding characteristic bronze stage 1 amine/imine groups of PANi and assembled into EEG- KC8 and steel blue stage 2 KC24 compounds. For each PANi via electrostatic interaction and hydrogen bond- experiment, anhydrous NMP was added to 10 mg ing. Protonic acid doping of PANi with HCl generated K-GIC in a Young’s tap Schlenk tube. The sample EEG-NPs with sandwich-like nanostructures. was removed from the glove box, mildly sonicated for Plasmonic coupling of AuNP creates intense hot 30 min and returned to the glove box for subsequent spots with large electromagnetic field-enhancements reactions. within the cavity formed by the two metallic surfaces MX2 (M  =  Mn, Zn and Cu) salts and Cu mesi- [1721]. However, this localized field is extremely sensi- tylene (CuMes) were dissolved in NMP to yield tive to morphological fluctuations and subtle changes 0.1 M stock solutions. Aliquots were added to the in the dielectric properties of the cavity contents. SLG KCx dispersion to give the desired stoichiometry of was used as an ultrathin spacer between AuNPs and a metal to K. Typically, reactions are performed using Au substrate to create plasmonic field enhancements the exact stoichiometry as the number of charges [1721]. By gating, the possibility to produce plasmon that are available for the reduction, e.g. for M2+ salts, tuning was demonstrated. M2+:K  =  0.5:1, M+ salts, M+:K  =  1:1, etc. It should NPs can also be grown on SLG by gas phase syn- be noted that the reduction potential of the metal thesis [1366]. This approach was used to grow clusters salt/complex is shifted dependent on its concen- (hence the name Ion Cluster Source or ICS [1367]). tration as defined by the Nernst equation [1364], The ability to control the mean NP size (from 1 nm EMn+ /M /M = EM 0 n+ /M + ((RT/nF) ln [M n+ ]), where to few tens nm) with a quite narrow size distribution 0 E is the standard reduction potential, and n is the while keeping a high chemical composition has made number of electrons involved. The reaction was left to ICS a versatile tool for the fabrication of NPs [1366]. stir for 72 h before being removed from the glove box, The replacement of the single magnetron of the ICS by washed and filtered under vacuum with 100 ml each of multiple magnetrons (Multiple Ion Cluster Source or NMP, water, chloroform and ethanol, and left to dry in MICS), offers the advantage of the synthesis of com- air [376]. plex highly crystalline NPs with core–shell and core– Another approach for NP decoration is anchor- shell–shell structures [1368–1370], hence providing ing them to a polymer matrix surrounding SLG/FLG multiple functionalities to the decorated material [1365]. EEG was prepared in Ref. [1365] with a sodium [1371, 1372]. methanesulfonate aqueous solution as electrolyte. It One of the important issues in this fabrication was then dispersed in DMF and mixed with polyani- method is the vacuum quality in terms of base pres­ line (PANi, emeraldine base). The strong π–π interac- sure and purity of the injected gases. A good base pres­ tion between PANi and EEG facilitated anchoring of sure is needed (<5 · 10−9 mbar–5 · 10−7 Pa) in order to PANi, achieving controlled surface functionalization. avoid contaminations. Ref. [1373] showed that small Subsequently, various colloidal NPs (Si, Fe3O4 and Pt) amounts of oxygen could modify the atomic struc- were added into the above PANi-functionalized EEG ture of Au clusters. One can assume that other con- dispersion. During this process, NPs were bound to the taminants, such as C, N, He, etc could also affect the 178 2D Mater. 7 (2020) 022001 C Backes et al Figure VIII.16. Representative Cs-corrected STEM-HAADF images of (a) Ih, (c) Dh and (e) fcc NPs and (b–f) their corresponding size distributions for Ih, Dh and fcc, for Au NPs grown by MICS. The FFT and the model of each structure are shown as insets. The mean diameters are also given. Figure VIII.17. AFM of Au NPs (a) on SiOx and (b) on SLG after a deposition time of 4 s and 3 s, respectively, to illustrate the importance of the sticking coefficient. growth and properties of NPs. Therefore, it is crucial ally, the purity of Au was ~99.99% and a systematic to obtain the best base pressure in the vacuum vessel. pre-sputtering of the targets was performed for 4 min For the fabrication of the NPs in figure VIII.16, a base to remove possible contaminants adsorbed on the tar- pressure below 5 · 10−10 in the MICS set-up was meas- get surfaces. The samples were handled following basic ured [1368–1370]. To ensure a minimum contamina- precautions like using plastic tools, to avoid metallic tion, all gas pipes used to inject Ar and He are made of contamination, [1374] flushing the fresh Si substrates stainless steel and vacuum sealed. The purity of Ar and with nitrogen, to remove dust, and depositing  <4% of He was ~99.5% and  >99.5% respectively. Addition- a monolayer of NPs to avoid NPs interactions [1375]. 179 2D Mater. 7 (2020) 022001 C Backes et al Figure VIII.18. (a) Increment of carriers versus NPs density using UHV and wet chemistry Au salts methods, (b) surface potential versus density of Au, NPs using the UHV method or the wet chemistry Au salts method. After [1378]. The structure and properties of NPs grown by gas smaller contact area between the UHV generated NPs phase synthesis is dependent on the parameters used in campared to those synthesized by wet chemistry, since the cluster sources. Ref. [1376] reported that the power the cleaner UHV synthesis does not result in debris any applied to the magnetron and the positioning of the mag- debris in the NPs neibourhood. netron inside the cluster source that defines the residence time of the NPs in the cluster source (and hence their VIII.5.2. GRM as support to grow Mg NPs size) can be used to tune the proportion of icosahedral For the synthesis MgH2NP a support is needed (Ih), decahedral (Dh) and face-centered cubic (fcc) iso- to better control the NP dimensions. GRMs are a mer structures. Therefore, a detailed description of the possible alternative to other substrates such as porous parameters is mandatory for any attempt to reproduce carbon, CNTs or other carbon nanostructures [280, the growth of NPs by using the gas phase synthesis route. 1379–1386]. Ref. [1387] used graphite flakes as The next step is the deposition of the NPs on SLG. support and anhydrous tetrahydrofuran (>99%) and a series of samples with different coverages are needed n-dibutyl-Mg. High purity hydrogen gas (99.999%) to evaluate the evolution of the SLG properties with and graphite flakes with 12, 6 and 1.6 nm thicknesses increasing amounts of NPs. When using gas aggregation and tens of microns in lateral size were used. Material sources, the NP density is controlled by the deposition processing was performed in a glove box filled with time. Once the number of NPs per time unit is known, high purity argon (O2 and H2O  <  1 ppm) to avoid one can calculate the time needed for a given coverage. oxygen contamination of the reactants preventing the However, if the calibrations are made on a different sub- formation of the Mg hydride. strate, the different sticking coefficient will modify this MgH2 was synthesized by decomposition of value. The sticking coefficient of metallic NPs on SLG is 3 n-dibutyl-Mg leading to the direct formation of MgH2 times SiOx as depicted in figure VIII.17). and under an inert atmosphere [1252] at 110 °C. Charge transfer effects were explored between the A rapid formation of MgH2 is expected for higher T AuNPs and graphene and compared with the doping following the reaction [1388]: achieved using other methods [1377], figure VIII.18. (C4 H9 )2 Mg → 2C4 H8 + MgH2 . (VIII.1) Depending on the functionalization method, differ- ent results are observed. When the decoration is driven In a typical synthesis protocol, n-dibutyl-Mg was through direct growth of NPs by chemical methods mixed to THF or C6H12 in different proportions. (red line, HAuCl4 [1377]) an increase of charge carrier The mixture was placed in a stainless steel reactor concentration and a decrease in the surface potential and porous carbon, MWNTs or FLGs added. THF is as a function of the coverage is observed. This trend is slightly polar (0.207), while C6H12 non-polar (0.0006). not observed when the decoration is carried out with For this reason, different solubility in relation to the AuNPs deposited in UHV conditions (blue line). The solvent used are expected. In addition, MgH2 is highly effect of the latter method, where the AuNPS arrive reactive toward organic molecules [1389]. Then the already formed, is much subtler. This is the result of a process of thermal degradation of n-dibutyl-Mg rea- 180 2D Mater. 7 (2020) 022001 C Backes et al Figure VIII.19. TEM images showing wrinkles and crumpling. Figure VIII.20. Schematic of basal-plane functionalization of MoS2. After intercalation with n-butyllithium, the negatively charged MoS2 is dispersed in water by mild bath-type sonication leading to an efficient exfoliation into individual sheets. The charges on the MoS2 are quenched by the addition of 4-methoxyphenyldiazonium tetrafluoroborate obtaining the functionalized product. Adapted from [379], copyright American Chemical Society. gent is expected to be susceptible to the synthetic sol- The XPS analysis showed Mg in a completely oxi- vent utilized. Hence the use of different solvents in dized form, because it difficult to introduce the sample principle should lead to different MgH2 structures and in the XPS instrument avoiding exposure to atmos­ hydrogen performances [1390]. phere. The oxygen is ~34%. However, if oxygen is con- Hydrogen was introduced in the reactor at a pres­ sidered as exogenous due to atmosphere exposure, the sure of 10 or 20 bar and the solution heated at 170 °C quantification can be limited to C and Mg core lines. and stirred at 50 rpm overnight [1391]. After 12 h, a The Mg concentration corresponds to ~27%. black/gray precipitate was obtained from the decom- Lower reactant concentrations led to better results position of n-dibutyl-Mg. The suspension was then in terms of NP size. MgH2 NPs with size  >5 nm were transferred in appropriate vessels under inert atmos­ obtained after optimization. Nowadays the use of phere and the main part of the solvent was removed hydrogen for fuel cells is limited by the difficulties of by using a centrifuge at 5000 rpm (~2000 g). To further synthesizing efficient storage materials. Mg offers a dry the material and remove the remaining organic compromise between good hydrogen storage (7.6 residuals, the material was placed in a dynamic vac- wt%) and desorption T if compared to other mat­erials. uum ~10−2 mbar for ~5 h. The exsiccated material was In addition, Mg is an interesting material because it is placed in sealed vials, to avoid Mg oxidation, for the low cost and non-toxic. For Mg/MgH2 the enthalpy of characterization [1387, 1392]. reaction is ~75 kJ mol−1 H2, corre­sponding to a H des- Stabilization of MgH2NPs depends on the orption T ~350 °C [1398, 1399]. However the thermo- interaction between NPs and the support surface dynamic properties of this material can be improved [1393–1395]. Defects and functional groups play an lowering the Mg-H bond formation enthalpy. For important role and a description of synthesis recipes practical application in automotive, the metal-hydro- should always be accompanied by a detailed descrip- gen formation enthalpy should be ~30 kJ mol−1 H2 tion of the material. The TEM analysis showed crum- [1400, 1401]. It has been demonstrated both theor­ pled surfaces (figure VIII.19). Curvature leads to etically and experimentally that the reduction of the higher reactivity of FLG towards chemical elements Mg particles affects the bond energy with hydrogen such as hydrogen or Mg [1394, 1396, 1397]. Wrinkles [280, 1402], although it is not easy to confine the Mg may then play an active role in limiting the formation particle size to the nanometer scale. The possibility of MgH2NPs in the solvent and enhancing the decora- to synthesize Mg NPs on the nanometer size is then tion of the graphene surface. of great interest for both automotive and stationary 181 2D Mater. 7 (2020) 022001 C Backes et al applications. Materials on these size exhibit hydrogen Washing with ~100 ml isopropanol to remove desorption T~ 160, 140 °C [1387]. organic side-products, such as the correspondent biphenyl, and washing with distilled water yielded the VIII.6. Functionalization of other LMs functionalized MoS2 product after drying under vac- VIII.6.1. Covalent functionalization of MoS2 uum (~ 10−2 mbar) at RT. The formation of a covalent VIII.6.1.1. Reductive covalent functionalization C-S bond was evidenced by the appearance of a C-S Functionalization of MoS2 can be achieved by stretching vibration in the IR spectra [379, 380], a new ligand conjugation of thiols at S vacancy sites either component in the S XPS core level spectra at higher introduced by ion irradiation [1403] or naturally binding energy [379, 380] and solid state NMR [380]. occurring after chemical exfoliation [1404, 1405]. Such By XPS [379, 380] and thermogravmetric analysis cou- methods are interesting but limited in their utility. pled to mass spectrometry [379], it was possible to esti- Ref. [380] reported the first step toward a general mate the degree of functionalization. Typically, 10–20 route to functionalize TMDs by grafting functional atom% S bears a functionality [379, 380]. The degree groups to the exposed S atoms, with a sequence, of functionalization was shown to depend on the ini- based on intercalation, chemical-exfoliation and tial intercalation conditions: when a milder intercala- subsequent quenching of the negative charges on the tion with a low n-BuLi to MoS2 stoichiometry ratio is MoS2 by organic halides or other strong electrophiles used (three-fold molar excess of MoS2), the degree of [379, 380]. In this functionalization sequence functionalization is ~ 10 atom% [379], whereas it can intercalated, chemically exfoliated MoS2 is reacted reach up to ~ 20 atom% when n-BuLi is used in excess with electrophiles such as halides and diazonium (three-fold molar excess of n-BuLi) [379, 380]. salts (figure VIII.20). The use of chemically exfoliated Here we provide a protocol for the fabrication of MoS2 is beneficial due to the good exfoliation in MoS2-templated conjugated microporous polymers water making LM both sides accessible. Chemically- (M-CMPs) by growing nitrogen-rich CMPs shells exfoliated MoS2 can be obtained by reacting it with n- on both sides of 4-iodophenyl-functionalized MoS2 butyllithium (n-BuLi). This leads to the formation of a templates [1408]. The synthesis strategy for M-CMPs Li intercalation compound associated with a widening is illustrated in figure VIII.21. First, chemically-exfo- of the interlayer distance between the individual MoS2 liated MoS2 (CE-MoS2) was achieved by reacting bulk layers and a charge transfer from n-BuLi to the MoS2. MoS2 with n-butyllithium (n-BuLi) [379, 380, 1408]. As a result of the charges residing on the MoS2 flakes, Then, it was functionalized with 4-iodophenyl diazo- a destablilization of the semiconducting 2H-polytype nium salt under aqueous conditions [379, 380, 1408]. is observed due to an effective change of electron The obtained 4-iodophenyl-functionalized MoS2 density in the d orbitals of the transition metal [1406]. (MoS2-I) can be well-dispersed in various organic Chemical exfoliation after Li intercalation thus usually solvents, such as toluene and DMF. Next, the aryla- results in the formation of the 1T-polytype of MoS2 cetylene building block 1,3,5-triethynylbenzene mixed [1406]. It depends on the intercalation conditions to with an aryl di- or trihalide was reacted with MoS2-I which extent this transformation occurs [379, 1407]. in anhydrous DMF in the presence of Pd(PPh3)4, CuI In contrast to negatively charged graphenides, the and Et3N under inert atmosphere. This Sonogashira- resultant material is reasonably stable under ambient Hagihara cross-coupling reaction was carried out at conditions and can be dispersed in water [1406]. Due 100 °C for 3 days under vigorous stirring and yielded to this, the preparation of solvents used for the initial an insoluble, crude product that was collected by fil- intercalation process is easier compared to graphene. tration and purified by Soxhlet extraction with THF Solvents only have to be dried and not necessarily freed for 2 days. Sandwich-like M-CMPs with high specific from oxygen. For this purpose, it is recommended surface areas and hierarchically porous structure were to distil and reflux the solvents (e.g. n-hexane and obtained after vacuum drying. As-prepared porous cyclohexane) two times over a Na wire under Ar atmos­ polymer-MoS2 sandwiches can be converted into the phere, prior to the use of the intercalation reagent. corresponding hierarchically porous MoS2/nitrogen- Since MoS2 has long been considered to be unreac- doped porous carbon hybrids by direct pyrolysis of tive and almost inert even after intercalation [159], Ref. the M-CMPs flakes. The hybrids are characterized [379] used very aggressive electrophiles: diazonium by high specific surface areas and aspect ratios and salts to quench the negative charges on the MoS2 and showed a promising oxygen reduction reaction (ORR) functionalize the material. After the final sonication and supercapacitor performance, due to the maxi- step of the purified intercalate 4-methoxyphenyldiazo- mized interfacial interaction between nitrogen-doped nium tetrafluoroborate was dissolved in distilled water porous carbon and MoS2 layers. and added to the dilute yellowish/brownish MoS2 dis- A pathway towards covalent functionalization of persion dropwisely under exposure to light. After addi- TMDs was established through the reaction between tion of only a few drops of reagent, a black precipitate liquid-exfoliated MoS2 and various metal acetate salts formed. The reaction mixture was stirred overnight [1409]. This involves functionalization of the semi- and filtered. Material which passed through the mem- conducting 2H-MoS2 polytype. Generating it requires brane in the first step was collected and filtered again. treatment of the 2H material with n-butyllithium, 182 2D Mater. 7 (2020) 022001 C Backes et al Figure VIII.21. Idealized formula scheme depicting the chemical exfoliation of bulk MoS2 and subsequent functionalization with 4-iodophenyl substituents under formation of MoS2-I, as well as the preparation of MoS2-templated conjugated microporous polymers (M-CMPs) and the corresponding MoS2/nitrogen-doped porous carbon (M-CMPs-T) hybrids. (i) monomers: 1,3,5-triethynylbenzene and 2,5-dibromopyridine, 2,5-dibromopyrazine, or 2,4,6-trichloro-1,3,5-triazine, argon, Pd(PPh3)4, CuI, Et3N, DMF, 100 °C, 3 d; (ii) argon, heating rate: 10 °C min−1, pyrolysis temperature: 700, 800, or 900 °C, 2 h. necessitating safety steps when handling this harsh IPA/H2O solvent. From our experience, a stable disper- substance. Direct functionalization of 2H-TMDs sion can be obtained by sonicating this precipitated removes the requirement to use this treatment. Acetate MoS2 a second time. However, in such cases, the degree functionalization of LPE MoS2 was shown to stabilize of functionalization was again lower than reported. the exfoliated flakes in IPA and acetone, rather than The number of functionalization procedures avail- having to use toxic solvents such as NMP and CHP able is limited. These mainly use exfoliated metal- [1409]. Dispersions of exfoliated 2H-MoS2 in IPA were lic 1T-MoS2 and in some cases ligand conjugation to mixed with a given metal acetate dissolved in IPA. This semiconducting 2H-MoS2. All these works [1229, mixture was then sonicated for 30 min and the func- 1404, 1788] do not focus and discuss on the location tionalized 2H-MoS2 was separated and collected by and how the functionalization takes place. To get this centrifugation and washing [1409]. information a different approach has been devel- XPS analysis was performed on the thin films of oped for functionalizing exfoliated 2H-MoS2 with the material formed via vacuum filtration [1409]. The Mo organic electron donor of 1,2-dithiolane-based pyrene 3d and S 2p spectral regions shown in figure VIII.22 derivative [1412]. These works demonstrated the pos- demonstrate functionalization of the flakes after treat- sibility of obtaining MoS2-pyrene nanohybrids, show- ment with Cu(OAc)2. The Mo 3d core level displays a new ing a preferential edge functionalization [1412]. This doublet associated with MoS2, shifted to higher binding primarily occurs via sulphur addition along partially energies than the unfunctionalized 2H-MoS2 polytype, saturated zig-zag MoS2-S edges, preserving intact the due to functionalization. The S2p core level exhibits a basal structure of functionalized MoS2. A natural con- new spectral feature for the same reasons. Re-dispersing sequence of this finding is that chemical functionali- dried functionalized 2H-MoS2 flakes in NMP removed zation of exfoliated MoS2 should be similar to that of the functional groups [1409]. Figures VIII.22(c) and bulk material, where edge functionalization is likely to (f) show that the material returns to its original state be predominant [1412, 1732]. Thus, this offers the pos- ­without any permanent changes in the spectral features, sibility of direct transfer of functionalization strategies indicating no permanent structural or chemical modifi- from bulk to exfoliated MoS2, e.g. transition metal cation during functionalization with M(OAc)2. doping of edge sites to improve catalytic behavior. The most critical step is to provide sufficient The interfacing of TMDs with photoactive species energy when the liquid exfoliated MoS2 in IPA is mixed via robust covalent bonding needs to be exploited via with the acetate salts. This was achieved by additional the formation of hybrid materials. This is particularly sonication using a tapered microtip with high local true in the context of their ability to function as donor– energy input [243]. When using a larger diameter tip acceptor systems upon photoillumination. This goal has (required to produce larger quantities), the degree of been achieved via the realization of novel donor–accep- functionalization (i.e. the number of S atoms on the tor hybrids based on the covalent functionalization of surface decorated by the functional group) drops MoS2 and WS2 with 1,2-dithiolane-modified carbon to  <20%. Furthermore, depending on the purity of nanodots (CNDs) [1413]. The excitation of CND-MoS2 starting MoS2 and water content in the IPA, the initial and CND-WS2 at 370 nm showed ultrafast energy trans- MoS2 dispersion is sometimes not colloidally stable fer from CNDs to both MoS2 and WS2, while at 425 nm, and the MoS2 precipitates from the dispersion while charge transfer in CND-MoS2 but not in CND-WS2 was leaving a deep blue colored supernatant. This is cur­ detected [1413]. These excited state electron-transfer rently not understood, but likely related to MoOx processes offer good opportunities for donor–acceptor- impurities in the powder that form complexes with the type hybrids for energy harvesting applications. 183 2D Mater. 7 (2020) 022001 C Backes et al a) b) c) Mo 3d Mo 3d Mo 3d Cu Ni MoS2 Zn 33% Intensity (a.u.) Intensity (a.u.) Intensity (a.u.) MoS2 MoS2 20% 19% func. + func. + MoO3 MoO3 MoO2 20% 12% S 2s S 2s func. + 5% S 2s 33 % 48 % MoO3 func S 2s func S 2s MoO2 56% MoO2 6% 28 % 17 % 1% 1% 245 240 235 230 225 240 235 230 225 220 240 235 230 225 Binding Energy (eV) Binding Energy (eV) Binding Energy (eV) d) func: 50% S 2p e) func: 30% S 2p f) func. 12% S 2p ∆ 6.5 eV ∆ 6.7 eV ∆ 7.9 eV Intensity (a.u.) Intensity (a.u.) Intensity (a.u.) Cu@ MoS2 Zn@ MoS2 MoS2 MoS2 MoS2 Ni@ MoS2 175 170 165 160 175 170 165 160 155 175 170 165 160 155 Binding Energy (eV) Binding Energy (eV) Binding Energy (eV) Figure VIII.22. Fitted XPS Mo 3d and S 2p core level peaks for (a, d) 2H-MoS2 (a), (b, e) 2H-MoS2-Cu(OAc)2, and (c,f) defunctionalized 2H-MoS2-Cu(OAc)2. Adapted from [1409], © Wiley-VCH Verlag GmbH & Co. KGaA. Figure VIII.23. (a) Topography of MoS2 triangles after 27 cycles of ALD after seeding with a perylene bisimide derivative. The flakes are higher than the surrounding substrate. (b) Line profile along the line marked in (a). The step height is 1.9 nm. (c) Schematic representation of the structure of (a) as indicated by the scan. Adapted from [1414] with permission of The Royal Society of Chemistry. VIII.6.2. Noncovalent functionalization of MoS2 successful for N  >  2, was unsuccessful on 1L. This was Perylene functionalization of CVD-grown MoS2 and attributed to the electronic structure of 1L-TMDs, WS2 was carried out in a similar manner as for SLG and their cleanliness [1414]. After functionalization [1323, 1327]. MoS2 and WS2 grown on SiO2 were with perylene1 the ALD growth of Al2O3 takes place coated with aqueous perylene solutions creating uniformly across the sample. Thus the non-covalent a non-covalent functionalization. It was reported functionalization provides a robust mechanism to that the resulting perylene layer was suited to seed seed thin films of dielectric materials on 1L-TMDs, ALD growth of Al2O3 on 1L TMD flakes (figure a requirement for many device applications and for VIII.23) [1414]. ALD directly on TMD flakes, while further functionalization. 184 2D Mater. 7 (2020) 022001 C Backes et al IX. Characterization methods infinite sum of light beams coming from the multiple reflections in the central SiO2 layer (see figure IX.2). IX.1. Microscopies Anytime that a light beam reaches an interface the IX.1.1. Optical microscopy Fresnel equations are applied, considering both the IX.1.1.1. Quantitative analysis of the optical real and the imaginary part of the refractive index ­microscopy images and accounting for the phase shift between the dif- Many optical characterization methods rely on the ferent light beams. E.g. the phase shift between the quantitative analysis of optical microscopy images reflected beams at the air/SiO2 and those transmitted [1, 1375, 1417–1419, 1421]. One can extract the through the air/SiO2 (going across the SiO2 then get- contrast difference between GRM and substrate by ting reflected at the SiO2/Si interface then going across analysing the red, green and blue channels of the the SiO2 and finally being transmitted at the SiO2/air digital images. For each camera channel the reflected interface) is 2Φ2 (see the sketch in figure IX.2) with intensity at the substrate and GRM can be measured Φ2  =  2πn2d2 · cos(θ2)/λ. All this being considered, and the contrast difference can be directly calculated. the total amplitude of the reflected light by the SiO2/Si Figure IX.1(a(l)) shows an example of a collection substrate (r) is: of epi-illumination (light reflected by the surface of the  ∞     sample and then detected) mode optical microscopy −2iΦ2 −2iΦ2 m r = r02 + t02 e r23 t20 1 + r23 r20 e images of MoS2 flakes with different N (previously m=1 determined by AFM) where the thickness dependent  (IX.2) colour is evident. The contrast difference requires an where rij and tij are the amplitude of the beam initial calibration, using an alternative thickness deter- reflected and transmitted at the interface between mination method, and it might slightly change from the media i and j . These coefficients rij and tij can be one experimental setup to another one as it relies on obtained directly from the Fresnel law. Considering the spectral response of the employed digital cam- that rij  =  −rji and tijtji  −  rijrji  =  1 and summing the era, the microscope lamp emission spectrum, and the geometrical series numerical aperture of the microscope objective. r02 + r23 e−2iΦ2 r= (IX.3) . 1 + r02 r23 e−2iΦ2 IX.1.1.2. Multispectral imaging Multi-spectral imaging characterization of GRMs Then the intensity of the light reflected by the SiO2/Si was introduced by Refs. [1418, 1419, 1420] to identify is Isubstrate  =  |r|2. N for graphene flakes on SiO2/Si substrates and it has    r02 + r23 e−2iΦ2 2 been rapidly adapted to characterize and identify other (IX.4)  Isubstrate =  . 1 + r02 r23 e−2iΦ2  GRMs [1416, 1421–1425]. The method involves the acquisition of optical images using narrow bandpass Under normal incidence assumption this expression filters to select the illumination wavelength. The is simplified because rij  =  (ñi  −  ñj)/(ñi  +  ñj) and optical contrast, CT, is typically extracted from these Φ2  =  2πn2d2/λ. When a GRM is placed on the SiO2/ images in order to obtain a quantity that does not Si substrate, the intensity of the light reflected by the depend on the illumination intensity. CT is typically stack can be calculated in a similar way as Isubstrate: defined as:   r ei(Φ1 +Φ2 ) + r e−i(Φ1 −Φ2 ) + r e−i(Φ1 +Φ2 ) + r r r ei(Φ1 −Φ2 ) 2   01 12 23 01 12 23  IGRM =  i(Φ +Φ ) . IGRM − Isubstrate e 1 2 + r01 r12 e −i(Φ 1 −Φ 2 ) + r01 r23 e −i(Φ 1 +Φ 2 ) + r12 r23 e i(Φ 1 −Φ 2 )  C= (IX.1)  IGRM + Isubstrate (IX.5) where IGRM and Isubstrate are the reflected intensities Although the previous approach is general and from GRM and substrate. A quantitative analysis can it can be applied to stacks with an arbitrary number be done on the basis of a Fresnel law based model, of media, it might become cumbersome to calculate where the system is modelled as a stack of different the intensity of the light reflected by stacks with more optical media under monochromatic illumination in than 3 media. The transfer matrix formalism is the a normal incidence configuration [1418, 1419, 1426]. most appropriate approach to analyse the light propa- Hereafter the subscripts 0, 1, 2 and 3 will be used to gation in systems formed by a stack of many different refer to air, GRM, SiO2 and Si. First we focus on the media. From the previous expressions one can deduce case where only air, SiO2 and Si are considered. The Si that if one determines the GRM thickness (e.g. with layer is modelled as a semi-infinite slab whose optical AFM) it is possible to use the optical contrast meas- properties are determined by its complex refractive ured at different illumination wavelengths to deter- index ñ3(λ) that strongly depends on the wavelength mine the GRM refractive ñ  =  n  −  iκ [1419, 1421, (λ) [1431, 1432, 1426]. The SiO2 layer, with a thickness 1424, 1428–1430]. On the other hand, if the GRM d2, is modelled with its refractive index n2(λ) that also refractive index is known one could determine its depends on the wavelength [1418, 1419, 1426]. thickness directly from the measurement of the opti- The total amplitude of the light beam reflected cal contrast acquired at different wavelengths with no by the SiO2/Si substrate (r) can be obtained from the need of AFM (figure IX.3). 185 2D Mater. 7 (2020) 022001 C Backes et al Figure IX.1. (a)–(l) Color optical images in epi-illumination mode of MoS2 flakes with N  =  1–12 a 300 nm SiO2/+  Si. The scale bar is 5 µm. (m) Thickness dependent contrast difference extracted from the red, green and blue channels. Figure reproduced from [1416], © Tsinghua University Press and Springer. IX.1.1.3. Hyper-spectral imaging wavelength (figure IX.4(b)). Spectral information In this technique the illumination wavelength is of a certain sample region can be directly obtained selected with a tuneable light source, e.g. based on a by plotting all the elements of the matrix along the halogen lamp and a monochromator, instead of using wavelength dimension λ for fixed X and Y coordinates narrow band pass filters [1431]. Figure IX.4 shows a (figure IX.4(b)). Figure IX.4(c) shows a sketch of two schematic setup for transmission mode hyperspectral spectra extracted from two regions in the sample: imaging. Epi-illumination hyperspectral imaging substrate (red) and GRM (blue). The GRM transmit- (or reflection mode hyperspectral imaging) can be tance (Tr) can be obtained by dividing both spectra: also done with minimum modifications of the setup Tr  =  IGRM/Isubs (inset in figure IX.4(c)), where IGRM is [1432]. the intensity acquired on the GRM and Isubs that of Hyperspectral imaging is carried out by sweep- the substrate. The absorbance A can be obtained from ing the excitation wavelength in steps and acquiring a as: A  =  −log10 (Tr). microscope image of the sample for each wavelength. Figure IX.8 shows an example of characterization The acquired images are then arranged forming a of GRM by means of hyperspectral imaging. The pan- 3d matrix, being the first two matrix indexes the X els (a) to (f) show transmission mode optical images and Y spatial coordinates and the third index (λ) the of a MoS2 flake for different wavelengths. From the 186 2D Mater. 7 (2020) 022001 C Backes et al Figure IX.2. Sketch of the optical beam path transmitted and reflected at the different interfaces in a multilayer structure air/SiO2/Si to understand the role of the different optical paths on the optical contrast. Adapted with permission from [1427]. Figure IX.3. (a) Optical images of a TaSe2 flakes with regions with different thicknesses, acquired at different wavelengths. (b) Optical contrast as a function of thickness for different wavelengths. The experimental data (circles) can be reproduced by the Fresnel law based model by using the refractive index for bulk TaSe2 (solid lines). Reproduced from [1416], © Tsinghua University Press and Springer. collection of hundreds of images from 400 to 1000 nm Kramers-Kronig constrained analysis of the spectra one can extract the absorbance, figure IX.5(g). of GRMs on transparent substrates [1198]. From the resulting dielectric function, the refractive index can IX.1.1.4. Micro reflectance/transmittance spectroscopy be calculated [1198]. Figure IX.8 shows an example In this method one illuminates the sample with white of determination of the dielectric function from the light and the reflected/transmitted light is analysed reflectance spectra of 1-4L TMDs. From the resulting with a spectrometer (see figure IX.6) [1198, 1310, 1428, dielectric function, the absorption spectra refractive 1429, 1433, 1434]. The acquired reflection/transmission index can be calculated [1198]. spectra can be analysed with the Fresnel law to extract Another feature of micro-reflectance/transmit- information on the GRM optical properties. In tance measurements is their speed, each spectrum can figure IX.7 the refractive index of 1L-MoS2 is determined be acquired in ~1 s, allowing one to characterize large by fitting the measured optical contrast on different scale samples (e.g. films grown by CVD, cm2 area) at SiO2/Si substrates to the Fresnel law based model, using hundreds of different locations to get an insight about the refractive index of 1L-MoS2 as a fitting parameter uniformity [1429]. Figure IX.9 shows a differential (panel a) [1428]. The resulting values of the real and reflectance spectrum acquired at one position on a imaginary part of the refractive index are displayed in 1L-MoS2 grown on sapphire. It also plots the histo- figure IX.7(b) and compared with bulk MoS2. grams of the energy and FWHM of the A and B exci- One can also obtain the dielectric functions tons measured at 550 different locations to character- from the measured reflectance spectra by means of a ize the sample homogeneity. 187 2D Mater. 7 (2020) 022001 C Backes et al is a powerful tool to perform fast measurements in one spot, due to its high spectral resolution and low measurement time. The short measurement time allows one to perform point measurements at hundreds of different locations[1429]. Multi-spectral and hyperspectral imaging are very interesting to study large areas as in one shot ~10 000 µm2 can be measured at once, allowing for the acquisition of spatially resolved maps with diffraction limited spatial resolution [1435]. IX.1.2. AFM IX.1.2.1. LPE GRMs AFM provides fast (<10 min/measurement) and reliable characterization of the lateral size and thickness distribution of flakes deposited on different substrates, allowing one to investigate large areas (hundreds of µm2) and to collect data from hundreds of flakes (figure IX.11(c)). The size distribution can be roughly modelled using a log-normal distribution, as observed for a range of GRMs exfoliated in the liquid including graphene [178], hBN [1437], MoS2 [185], WS2 [186], GaS [188] and black phosphorus [190]. More extensive statistical studies, performed on GO flakes using AFM, SEM and fluorescent microscopy, demonstrated that the fragmentation process can be described by a series of rupture-like breakup events, yielding two different populations of flakes featuring different size distributions [136]. When working with LPE samples for AFM analy- sis, it is critical to avoid re-aggregation during solvent evaporation. For this, drop-casting the dispersion on pre-heated wafers (10 µl per 0.5  ×  0.5 cm2 wafer) is often used [245]. The solvent evaporates and bub- bles are formed, resulting in more uniform deposition compared to drop casting at lower T. The wafer should be heated to ~50 °C–70 °C above the boiling point of the solvent. In the case of surfactant-based dispersions, it is recommended to dilute the sample with water (rather than surfactant) prior to deposition and wash the wafer thoroughly with water and isopropanol (~5 ml Figure IX.4. (a) Schematic setup for transmission mode each) to remove residual surfactant. Residual sur- hyperspectral imaging. (b) Data acquisition process: optical factant can make thickness measurements very tedi- images are acquired while the wavelength is varied. The resulting dataset is arranged in a 3d matrix. (c) Spectral ous, especially for very small flakes that are more dif- information extracted from the dataset. The inset shows a ficult to distinguish from the surfactant. In this case, transmittance spectrum obtained by normalizing the GRM phase images can provide a guide as they usually give spectrum to that of the substrate. Adapted from [1431], © IOP Publishing Ltd. a good contrast between different materials. If prob- lems with residual surfactant persist, the substrates can be soaked in water overnight without significant loss IX.1.1.5. Comparison between multispectral, hyper- of flakes. Deposition from high boiling point solvents, spectral and micro-transmittance such as NMP, may be challenging, and re-aggregation, Figure IX.10 compares different optical microscopy residual solvent, or polymerized NMP covering flakes characterization methods, studying the same are observed. In order to overcome this problem, it is system: multi-spectral, hyper-spectral and micro advisable to transfer the material exfoliated in NMP to transmittance of 1-3L-MoS2 [1421, 1431, 1434]. IPA by centrifugation prior to AFM [190]. Table IX.1 compares the area characterized, the spectral Si substrates with a 200–300 nm SiO2 enable flakes resolution, and the time to carry out one measurement. to be seen with an optical microscope/optical [1419]. This shows that micro transmittance/reflectance This is a useful guide to identify regions of interest for 188 2D Mater. 7 (2020) 022001 C Backes et al Figure IX.5. (a)–(f) Transmission mode images of a MoS2 flake acquired at different wavelengths, selected with a tuneable monochromatic light source. (g) Absorbance versus wavelength, extracted from the sequence of MoS2 images on a flake with regions of different thicknesses (N  =  1–6), for different wavelengths. Adapted from [1431], © IOP Publishing Ltd. Figure IX.6. (a) Setup to characterize GRMs with micro-reflectance. (b) CVD grown MoS2 studied by micro-reflectance. Two spectra are acquired on (A) MoS2 and on (B) the SiO2/Si substrate and the optical contrast is extracted from them. Adapted from [1428], CC BY 4.0. 189 2D Mater. 7 (2020) 022001 C Backes et al Figure IX.7. (a) Optical contrast for 1L-MoS2 on SiO2/Si with different SiO2 thicknesses. The datapoints are fitted to a Fresnel law using the refractive index of 1L-MoS2 as fitting parameter. (b) refractive index of 1L-MoS2, compared with the bulk value. Adapted from [1428]. Figure IX.8. (a)–(d) Reflectance spectra for 1L-MoSe2, WSe2, MoS2, WS2. (e)–(h) Real part and (i)–(l) imaginary part of the dielectric function. Adapted from [1198], © Springer Nature. 190 2D Mater. 7 (2020) 022001 C Backes et al Figure IX.9. (a) Differential reflectance of 1L-MoS2 grown on Sa. (b) and (c) Histograms of A and B exciton energy for 550 flakes. (d) and (e) Histograms of FWHM of A and B excitons. Adapted from [1429], CC BY 4.0. imaging. Measured, apparent AFM heights of LPE be conducted in order to make sure that the thin- GRMs are overestimated due to the presence of resid- nest objects measured are really SLG [176, 185]. The ual solvent. In general, accurate height measurements apparent SLG height can then be used to convert the on inhomogeneous surfaces (such as flakes deposited apparent measured AFM thickness to N. For statisti- on a substrate) are challenging due to contributions cal analysis, it is recommended to record the height of from capillary forces and adhesion, which depend at least 100–150 flakes. If the thickness varies across on the material and measurement parameters [1438, the flake, the mean value should be taken. From such 1439]. In addition, crosstalk between electrostatic and statistical analysis, population histograms can be con- topographic signals [1214, 1215] can occur, leading to structed. These are typically log-normal [185, 186, a bias of the height.To overcome these problems and 188, 190, 1437] (also in the case of flake length). If this obtain N from apparent height values, a step height is not the case, the counting and/or imaging may be analysis procedure was developed in Refs. [176, 185, biased. e.g., if reaggregated rather than individually 186, 190]. Incompletely exfoliated flakes showing deposited flakes are included in the counting, this will clear terraces are first examined and the height of a lead to a deviation from the log-normal shape at the various steps are recorded. These terrace step heights thicker end of the thickness distribution histogram. will always be a multiple of the apparent 1L thickness. On the contrary, if surfactant and solvent residues are The apparent measured SLG height can be 1–2 nm, included in the counting, a deviation is obtained on much greater than the theoretical thickness [176, the smaller end of the length distribution histograms. 185, 188, 190, 240, 1407]. Raman and PL analysis can From these histograms and the statistical analysis, the 191 2D Mater. 7 (2020) 022001 C Backes et al Figure IX.10. Comparison between different methods to measure the optical properties of 1L, 2L,3L-MoS2. In multispectral imaging narrow bandwidth filters (10 nm FWHM) are used to select the wavelength (the wavelength resolution is limited by the number of filters). In hyperspectral imaging, the illumination is provided through a white light source connected to a monochromator (2–3 nm resolution). In micro-transmittance, white light is collected through an optical fiber (acting as a confocal pinhole) and sent to a CCD spectrometer (~1 nm resolution). Adapted from [1436]. Table IX.1. Comparison between different optical microscopy IX.1.2.2. Grown and transferred GRMs methods. When referring to grown materials measured heights Spectral Measure- can be affected by artefacts, i.e. crosstalk between Method Area resolution ment time electrostatic and topographic signals [1443, 1444], or can be overestimated due to the presence of Multi-spectral ~10 000 µm2 ~20–100 nm 10–30 min adsorbates between substrate and 1L-GRMs [1445]. Hyper-spectral ~10 000 µm2 ~1–5 nm 30–60 min By combining two of the most employed measuring Micro trans- ~1–4 µm2 ~1 nm ~0.1–5 s modes (contact and dynamic), these can be mittance minimized and more accurate height values obtained. Recording secondary signal channels such as phase (in arithmetic number mean for length and thickness is dynamic mode) or lateral force (in contact mode) is obtained. useful in revealing information such as different local Since AFM can be used to measure both flake mechanical and/or electrical properties [1446] or thickness and lateral dimensions, for each flake of a different polymorphic structures [1447] sometimes given thickness, the volume can be estimated as thick- difficult to detect, or even hidden, in topographic ness (N)  ×  length(L)  ×  width(w). This allows for the images [1443–1445, 1447]. Dynamic and contact calculation of the volume-fraction-weighted mean modes measurements can be performed at RT,   layer number, NVf = N2 LW/ NLW, where purging the chamber with N2 to decrease the relative the summations are over all flakes. This is an alterna- humidity and minimize capillary forces between tip tive measure of thickness which reflects the fact that and sample surface. Details about measuring modes mass the tends to be concentrated in thicker flakes in AFM can be found in Refs. [1448, 1449]. (the difference between arithmetic and volume frac- Figure IX.12 shows representative images of gra- tion weighted mean is akin to the difference between phene layers grown on different substrates acquired number-average-molecular-weight and weight-aver- in dynamic mode. In this mode, which employs the age-molecular-weight in polymer physics [1440]). oscillation amplitude as feedback parameter [1448], Arithmetic and volume fraction weighted mean values the durability of the AFM probes (i.e. tip sharpness) is are typically related linearly, therefore both are an ade- usually larger than in contact mode, since the tip gen- quate measure of thickness [236]. tly [1450] taps the surface during the scanning [1446, Conductive SPM, like Kelvin probe microscopy 1451]. Probes with nominal force constant values ~3 (KPM), beside imaging capabilities, can be used to nN nm−1 and resonance frequencies ~75 kHz are used. investigate charge transfer in exfoliated GRMs [277], The most characteristic features are the pleats, com- while conductive AFM, providing localized current monly referred to as wrinkles, figure IX.12. These cor- injections, can be used to locally modify the GRM respond to pleated/folded SLG due to released stress chemistry, e.g. directly drawing conducive paths caused by the differences in the thermal expansion in GO films on SiO2 [1441].This can also be used to coefficients between SLG and the substrate, during probe other electronic properties, such as piezoelectric cooling after growth [1452, 1453], or to local accumu- effects [1442]. lation of carbon material at defects between ­coalescing 192 2D Mater. 7 (2020) 022001 C Backes et al Figure IX.11. (a) SEM micrograph of bulk BN flakes. (b) Solutions of exfoliated BN in isopropanol. (c) AFM image of BN flakes on SiO2. (d) Individual BN flake. (e) Histogram distribution of flake sizes. From [1437], with permission of The Royal Society of Chemistry. Figure IX.12. AFM topographic images measured in dynamic mode of SLG grown on (a) Cu foil by PDV, (b) SiC(0 0 0 1 by CVD, (c) quartz by r-(ECR-CVD) [774] z-scale: 0–12 nm. grains of different orientation [630]. The maximum in figure IX.13(a) illustrate this case, where circular scanned areas are limited by the morphology of the brighter patches corresponding to BLG areas and are substrate. For Cu foils this is usually  <5  ×  5 µm2, due distinguished in phase imaging due to their differ- to the high overall roughness that can exceed several ent dissipative properties arising from mechanical hundreds of nm for those areas. For graphene on SiC, and electrostatic difference as a function of N [1454]. the scanned area is only limited by the piezo range Damaged areas in transferred layers (figure IX.13(b)) thanks to the flatness of the SiC wafer [627, 630] (for or covered areas in sub-SLG growth (figure IX.13(c)) chemical mechanical polished on-axis wafers). can be identified in phase images. However, phase In dynamic mode, phase images can be acquired images can also be misleading since the contrast is simultaneously with topography. Phase contrast very sensitive to the measuring parameters (oscillation is generally ascribed to differences in local energy amplitude and control setpoint [1455, 1456]). This dissipation and can reveal local mechanical and/ is shown in figure IX.14, where phase contrast can be or electrical properties [1446, 1451]. As shown in turned on and off at will, and even reversed, by chang- ­figure IX.13, this signal can be very useful to identify ing the measured amplitude. There is no fixed recipe areas of different thickness. Top and bottom images that ensures a good phase contrast, and tuning of the 193 2D Mater. 7 (2020) 022001 C Backes et al Figure IX.13. Simultaneous topographic (top row) and phase (bottom row) AFM images in dynamic mode of (a) and (b) CVD- graphene transferred to SiO2. (c) Flakes grown by r-(ECR-CVD) on quartz at submonolayer coverage. Figure IX.14. Simultaneous topographic (top row) and phase (bottom row) AFM images measured in dynamic mode of graphene flakes grown by r-(ECR-CVD) on quartz at submonolayer coverage. (a) Phase contrast between graphene and substrate disappears in the center of the image after changing the amplitude setpoint. (b) Phase contrast between graphene and substrate is reversed at the bottom of the image by increasing the amplitude setpoint. 194 2D Mater. 7 (2020) 022001 C Backes et al Figure IX.15. Topographic images of graphene flakes on quartz measured in (a) dynamic and (b) contact mode. Bottom: height profiles from the corresponding lines marked in the above images. The thickness measured on the same sample is different, depending on the mode used. measuring parameters is necessary. Small amplitudes since they usually present a low (<0.1) friction coef- and amplitude set-points  >  50% of the cantilever free ficient [1459], which yields images like figure IX.16. amplitude usually enhance the phase contrast [1456]. The SLG areas are distinguished in lateral force The main disadvantage of using dynamic mode stems images, facilitating flakes location, coverage quantifi- from the possible crosstalk between tip and sample cation and size analysis. With lateral force images it is surface. This is particularly critical in heterogeneous also possible to visualize the internal structure of SLG surfaces, with areas presenting different mechanical, grains, masked in topographic images due to surface electrostatic and/or hydrophobic/hydrophilic nature, roughness. In figure IX.16, lateral force images reveal and can lead to wrong N determination [1443, 1457]. the multilayer structure of some flakes. This is illustrated in figure IX.15 graphene on quartz To better visualize these internal features in gra- at submonolayer coverage. The thickness of the flakes phene flakes, soft tips (0.05 nN nm−1) and forces measured in dynamic mode is ~3 nm, almost twice between 5 and 10 nN are recommended to avoid wear that in contact mode. This is due to the influence of [1455]. Lattice-resolved lateral force images can also uncompensated electrostatic forces, particularly reveal the lattice periodicity, figure IX.17. To acquire affecting samples grown on insulating substrates. Ref. these images, small scans are performed (5–20 nm [1443] suggested to use KPM to avoid this problem lateral sizes), at frequencies ~2–3 Hz, with low feed- [1452]. However, when dealing with insulating sub- back parameters. Applied forces are ~5–15 nN. These strates, like in figure IX.15, it is not so straightforward, images do not correspond to true atomic resolution and the alternative is switching to contact mode (figure like routinely measured in STM, but lattice orientation IX.15(b)). In contact mode, the lifetime of the probes is and average lattice parameters can be determined. usually shortened compared to dynamic mode [1448] The AFM tip can also be used to displace the sam- (tip radius increases after successive scans and resolu- ples. Ref. [774] swept the tip to determine N in fully tion worsens), but height artefacts as those shown in covered samples. This should be used as last resort, figure IX.15 are seldom encountered. Commercial after morphological characterization, since the tip can probes of low force constant (0.05–1 nN nm−1) are be irreversibly damaged due to the high (>100 nN) used to minimize damage to the sample. Similar to forces needed to remove the material. This damage can the dynamic mode, secondary signals can be recorded be visualized by recording force-distance curves prior when measuring in contact [1458]. Lateral force imag- and after sweeping. Figure IX.18(a) compares these two ing can reveal differences in friction between tip and curves, and shows that adhesion is one order of mag- surface [1458]. This is particularly interesting GRMs nitude larger after the experiment. This is mainly due 195 2D Mater. 7 (2020) 022001 C Backes et al Figure IX.16. Simultaneous (a) topographic and (b) lateral force images of graphene flakes on quartz measured in contact mode. Dotted circles mark flakes location, hardly distinguishable in topography, but resolved in lateral force images. Figure IX.17. Simultaneous (a) topographic and (b) lateral force images of graphene flakes on quartz in contact mode. The multilayer structure of some flakes can be distinguished in lateral force images. (c) High resolution lateral force image. Right: high resolution image acquired at the marked area. Inset: fast Fourier transform (FFT) of the high-resolution image to better observe the lattice periodicity. to the increase in tip radius, which translates in a wors- (256/128) and increasing gradually the applied force. ening of the resolution [1460, 1461]. Taking this into By controlling the topography and lateral force signals, account, sweeping experiments should performed as one can detect the onset of the removal of the material. follows. 1) an area is scanned either in dynamic or con- Once the critical applied force is reached, the scanning tact mode to check the morphology prior to sweeping. 2) is repeated during several images to ensure the com- contact mode is selected, and the same area (2–5 µm plete displacement of the material. 3) To visualize the lateral size) is repeatedly scanned while increasing the swept area it is recommended to switch to dynamic scan rate (5–10 Hz), decreasing the number of points mode, to avoid spreading the accumulated material at 196 2D Mater. 7 (2020) 022001 C Backes et al Figure IX.18. (a) Force versus distance curves acquired before (black) and after (blue) sweeping. Only the retract branch is plotted to better visualize the increase in the adhesion force. (b) Topographic image showing the rectangular area with SLG removed. From line profiles taken across the edge, the thickness can be obtained as in as pointed out scanning the inset in (b). the edges of the scanned area. To acquire final image IX.19(a) bottom), a fingerprint of the presence of BLG is better if the scan is performed at 90°, since debris [1462, 1464]. The dark contrast corresponds to SLG are accumulated at the lateral edges (fast scan direc- and the brighter contrast areas, at the step bunches, tion). Figure IX.18 (b) summarizes the results of this indicate BLG. Small islands (300–500 nm in diameter) procedure. decorating the terraces are assigned to BLG from their Depending on applied force, the material swept by contrast. This contrast observed in surface potential the tip can be either the contaminant layer adsorbed as a function of N is ascribed to different work func- on SLG exposed to air, if forces are restricted to few nN, tion values for different N, due to different substrate or the complete layer for forces up to hundreds nN. induced doping and different energy dispersions of KPM can give information on thickness, layer- SLG and BLG [749, 1465, 1466]. dependent distribution of charges, electrical potential When measuring in ambient conditions, aging of and work function [1462]. Measurements are per- the sample can affect the contrast in surface poten- formed in dynamic mode, using the amplitude as feed- tial. This is shown in figure IX.19(b), where the same back channel for topographic determination, and the sample is measured after growth and several days later. retrace mode for surface potential acquisition, at a lift In addition to the presence of small clusters (2–5 nm distance between 15–20 nm. NT-KP tips from Next- high) in the topographic image, due to adsorbates, the Tip S.L. provide very good resolution for KPM [1463]. corresponding surface potential image shows negligi- Surface potential maps can be used to discriminate ble contrast between terraces and steps bunches. This between N, figure IX.19, for graphene grown on SiC. is due to the passivation layer that screens any electro- While the topography image (figure IX.19(a) top) is static difference between SLG and BLG areas. Anneal- dominated by SiC terraces, a contrast is observed at ing in controlled atmosphere above 150 °C is enough the step bunches in surface potential images (figure to recover the original KPM signal. 197 2D Mater. 7 (2020) 022001 C Backes et al Figure IX.19. Simultaneous topographic (top) and surface potential (bottom) images acquired on graphene on SiC. (a) Measurements performed after sample growth. (b) After several days, with the sample kept in ambient conditions. KPM was also employed to investigate the changes eral curves are acquired on different locations, and in work function upon controlled doping [1377]. By measurements are repeated altering the sample order, definition, the measured surface potential (VSP) is to ensure reproducibility of measured differences in related to the work function (ΦS) [1467] as: surface potential. Similar curves are measured in the reference sample after each round. Figure IX.20(b) eVSP = ΦT − ΦS (IX.6) compares the results of gradual doping with Au NPs, where ΦT is the work function of the tip, and e is the reflected in a gradual variation of the surface potential electron charge. Therefore, changes VSP potential measured by KPM [1377]. for differently doped samples reflect changes in ΦS. The critical point in these studies is ensuring that the IX.1.3. Transmission electron microscopy tip conditions (particularly, ΦT) do not change. To TEM techniques can be used to investigate individual monitor this, a reference sample of known ΦS (i.e. flakes from solutions, measuring their lateral HOPG, Au) is measured between samples. size and thickness, the characterization of their To preserve tip conditions, these comparative stud- crystal structure and chemical composition. Cross ies can be done using Kelvin probe force spectroscopy section TEM can also be used to give information (KPFS) instead imaging acquisition [1468, 1469]. about N and their interaction with the substrate KPFS consists of applying a varying DC bias, to the [688] (see e.g. figure IV.25(e). Since TEM provides tip, located above a single spatial location, while moni- a higher resolution compared with SEM and higher toring the dynamic response of the cantilever using throughput compared with AFM, low resolution heterodyne detection [1469]. The force-voltage or TEM is frequently used to analyse the shape and lateral frequency-voltage (depending on the detection mode size of LPE GRMs statistically [131, 174, 176, 177, used [1457]) curves have parabolic voltage depend­ 184–186, 188, 190, 239–242, 244, 1470]. To perform ence, and the position of the maximum of the fitted these measurements the dispersion is drop-cast parabola yields VSP. Two representative curves are onto a TEM grid. Here, it is important to use dilute shown in figure IX.20(a), for SLG doped with Au NPs, dispersions (i.e. optically transparent if the material for different NP concentration [1377]. The difference absorbs in the visible) to avoid reaggregation. For in the maximum position of the curves corresponds statistical length analysis, the longest dimension is to a difference in the EF[1377]. For comparison, sev- measured and denoted as length and the direction 198 2D Mater. 7 (2020) 022001 C Backes et al Figure IX.20. (a) Frequency versus voltage curves acquired on SLG on SiO2 doped with HAuCl4 for two concentrations [1377]. VSP corresponds to the maximum of the parabolas, marked with dashed vertical lines. (b) The variation of the measured VSP reflects the tuning of EF upon doping with Au NPs nanoparticles. (Reproduced from [1377], with permission from Elsevier.) Figure IX.21. TEM analysis of graphene flakes exfoliated in NMP and different Alkanes. (a) Histogram of N. (b) HREM of folded edge. (c) Flake size distribution; data fitted with a log normal function. (d) TEM micrograph of graphene flake. From [235], with permission of The Royal Society of Chemistry. 199 2D Mater. 7 (2020) 022001 C Backes et al perpendicular is denoted as width [131, 174, 176, 177, Today’s high-end TEMs are equipped with a corrector 184–186, 188, 190, 239–242, 244, 1470]. Depending for the spherical aberration (Cs), and usually operate on the expected flake size, continuous film grids can at acceleration voltages between 300 kV and 80kV. be beneficial to avoid small flakes falling through the The resolution is ~0.8 Å at 300 kV and ~1.9 Å at 80 kV, holes. When drop-casting, the best results are obtained which allows to resolve single atomic columns of many when the grid is placed on a filter paper to wick away bulk crystalline samples. For SLG, the pristine lattice access solvent [131, 174, 176, 177, 184–186, 188, 190, withstands 80 kV for a sufficiently high electron dose 239–242, 244, 1470]. During image acquisition, it for HRTEM imaging with good signal-to-noise ratio is important to adjust the field of view according to (SNR), however edges and pores will be more quickly flake size. Without size selection, samples can be very altered and much lower voltages are required [1478]. polydisperse with lateral sizes ranging from 20 nm-a At voltages below 80 kV, the resolving power of Cs- few µm. This is extremely challenging and requires corrected microscopes is however not sufficient as it is recording higher magnification images as to not bias limited by the chromatic aberration Cc of the objective the statistics towards larger, more easily discernible lens. In recent years this limitation has been tackled flakes in wide-view images. Similar to the AFM by two approaches: either the effect of chromatic statistics, the log-normal shape of the histogram can aberration is reduced due to a reduced energy width be used as a guide whether counting/imaging are of the primary electron beam by a cold field emission biased [186, 188, 190]. In such cases, a comparison of gun [1479] or by a monochromator (e.g. [1480, the statistically determined mean length from AFM 1481]), or, most efficiently, by a chromatic aberration and TEM [186, 188, 190] suggests that ~150 counts are corrector [1482]. Ref. [1482] reported a microscope, sufficient to obtain robust mean values. the so-called Sub-Angstroem Low-Voltage Electron In addition, thickness determination can be microscope (SALVE), equipped with a Cc/Cs corrector achieved from high-resolution electron microscopy optimized for voltages between 80 and 20 kV. Even (HREM) images. This is done by counting the number at 20 kV, this microscope can resolve individual C of (0 0 0 2) lattice fringes (see figure IX.21(b)) exposed atoms with a resolution  <  0.14 nm [1482] over a wide in folded flakes [235] or by analyzing the electron dif- field of view of 4096  ×  4096 pixel. With the SALVE fraction pattern intensities from individual crystal microscope, sub-Å resolution resolution is achieved [235], providing statistical qualification of the exfo- for voltages between 40 kV (0.92 Å) and 80 kV (0.76 Å), liation process [235]. Additional functional properties and the achievable contrast for the same electron dose can be engineered by decoration with either organic is increased [1473, 1484]. moieties or NPs. In this case, STEM imaging and ele- Example low-voltage HRTEM images of GRMs are mental mapping can provide nanoscale characteriza- shown in figure IX.22. The diffractograms of SLG and tion of the decoration and of the interface between MoS2 (30 kV images) feature reflections up to the third NPs and GRMs [235, 288]. Electron beam damage can and fourth order. A good agreement between image be limited by reducing the accelerating voltage, mak- calculation and experiments can be seen (see linescans ing it possible to use TEM for the characterization of A6/B6). blends between GRMs and organic complexes [288]. The tunable low acceleration voltage allows for its optimal choice for each sample. The chosen one is IX.1.3.1. Low-energy HRTEM a compromise between the resolving power (which For structural characterization on the atomic level, is higher at higher voltages) and knock-on damage. aberration-corrected high-resolution transmission In addition, especially non-conducting material, suf- electron microscopy (HRTEM) is the method of fer from ionization effects, which usually increase at choice. However, for very thin samples such 1L-GRMs, lower voltages [1473, 1484]. Sophisticated sample the interactions of energetic electrons with the prep­aration methods are then needed to reduce these material during imaging may result in permanent effects. The following techniques provide good results: changes in the structure, also referred to as radiation the production of clean surfaces [1485], sandwiching damage. This can be both an obstacle for imaging, as the radiation-sensitive material between two SLGs well as a useful tool for manipulating matter at the [1486], and isotope substitution [1487]. To summa- atomic scales [1471–1474]. The best investigated rize, the choice of the acceleration voltage depends damage mechanism is the so-called knock-on damage on the trade-off between resolution, knock-on dam- [1475], where the impact of fast electrons gradually age and ionization. Appropriate sample preparation removes atoms from the sample. From kinematic is crucial for reaching the voltage-dependent resolu- considerations [1476], it follows that these alterations tion limits. The Cc/Cs-corrected technology for TEM can be prevented if a certain threshold energy is is regarded as a major improvement of the current undercut [1477]. Depending on the GRM, acceleration technology which can be judged from the single-shot voltages of 80 kV and less are required [1478]. At such images of SLG in figure IX.23, at 80 and 30 kV from the voltages, in conventional TEMs without aberration new (Cc/Cs) technology compared with the old (Cs) correction, atomic resolution is not achievable as technology (all single-shot images). Reliable informa- the resolution is worse than 0.35 nm already at 80 kV. tion from HRTEM images about pore sizes distribu- 200 2D Mater. 7 (2020) 022001 C Backes et al Figure IX.22. Experimental and calculated Cc/Cs-corrected HRTEM images of (A) SLG [1482] and (B) MoS2 at 30 kV. (A1) and (B1) show the Fourier transforms of (A2) and (B2). The outmost reflections demonstrate a resolution ~1 Å. (A3) and (B3) show atomic resolution with magnified areas. The SNR can be improved by averaging the experimental images (A4) and (B4). The results are in good agreement with simulated HRTEM images (A5) and (B5) [1483]. The line scans (A6) and (B6) show the intensity profiles along the marked lines in the experimental and the simulated images. Adapted from [1482], copyright by the American Physical Society. tions in GRMs, or edge structures can now be obtained structural and electronic studies at the atomic level. at lower voltages with higher resolution and higher When the graphene samples are measured in ambient contrast [1482]. In addition, due to the wide field of conditions or after ex situ manipulations, AFM view of 4096  ×  4096 pixels, this microscope is also is a preferred tool as force maps are more directly very useful for dynamic studies and manipulations at interpreted in terms of topographic features of the the atomic scale. sample. STM images, in contrast, are proportional to the tunnel current which is sensitive to the electronic IX.1.4. STM structure of the tip and surface [1488, 1489]. When STM is a powerful tool for graphene characterization. samples are characterized in ambient conditions, It is generally used inside UHV chambers for in situ data acquisition can be quite difficult as atmospheric 201 2D Mater. 7 (2020) 022001 C Backes et al Figure IX.23. SLG HRTEM images obtained at 30 kV (left) and 80 kV (middle) in the Cc/Cs-corrected SALVE microscope and at 80 kV (right) in the Cs-corrected TITAN-microscope. The inserts show the corresponding Fourier transform patterns. The (1  1 2 0), (0  2 2 0) and (1 3 2 0) reflections are encircled in red, blue, green, and purple, respectively. Adapted from [1482], copyright by the American Physical Society. contaminants may intrude between tip and surface by STM images. The most important 1d defective as they diffuse, thus provoking current instabilities structures are wrinkles and graphene edges between and hindering high-quality data acquisition. After different (rotational) domains. Figure IX.24 shows a graphene exposure to ambient conditions, annealing representative STM image of a graphene on Pt(1 1 1) in vacuum for a short period of time at ~ 250 °C before sample together with a representative LEED pattern of measurement is recommended in order to remove the same system [1035]. In the constant-current topo- possible adventitious adsorbates. With respect to the graphic image, a graphene pleat and some graphene acquisition mode, topography images at constant nanobubbles can be seen. Profiles along this pleat show current are preferred as they are typically more an apparent height ~ 2–2.5 nm. The origin of these stable and avoid tip crashes during measurements. structures is the difference between thermal expansion Nevertheless, the use of constant-height or current- coefficients of graphene and Pt, which results in com- error images can also be useful in some cases for pressive strain during cooling. Graphene nanoscale obtaining atomically-resolved images of local features dome-like structures, also known as nanobubbles, can [1490, 1491]. also be resolved with STM images. In figure IX.24 (a) The atomic lattice of graphene can routinely be one is encircled in yellow and in figure IX.24(b) one resolved with STM. A hexagonal lattice with a perio- is shown with atomic resolution. Structures submitted dicity of 2.44 Å is normally observed for graphene to compressive strain, like graphene pleats and nano- and graphite samples. When imaging graphene, the bubbles, have been proposed to induce high pseudo- smaller periodicity (1.42 Å) of the honeycomb lat- magnetic fields, which lead to Landau quantization of tice can sometimes also be resolved. Low voltages and the electrons when measured with STM at cryogenic high currents (short tip-sample distances) are rec- T [632]. ommended for obtaining atomic resolution images, In figure IX.25 different point defects of a SLG/ with typical values ranging between 50–200 mV and Pt(1 1 1) system are depicted. Figure IX.25(a) shows 1–4 nA, respectively. Graphene superlattices of larger a single atom defect which can be explained in terms scale (1–2 nm) can also be resolved with STM images of an in-lattice heteroatom inclusion,–most likely a N STM is the ideal tool for local measurements needing impurity coming from the residual gas in the cham- ultrahigh spatial resolution. STM has demonstrated its ber [1492] Figure IX.25(b) displays a multiatomic capabilities for resolving low-dimensional structural vacancy which induced a reconstruction of a region ~ defects with unmatched atomic resolution. There are 1–2 nm. Although a dome-like local configuration can several defective structures that have been described be expected one must take with caution the fact that 202 2D Mater. 7 (2020) 022001 C Backes et al Figure IX.24. (a) STM image of a 200  ×  200 nm2 area of graphene grown on Pt(1 1 1), I  =  0.15 nA, V  =  750 mV. A SLG covers the whole surface. In the image a wrinkle/pleat and several nanobubbles, one of them is encircled in yelow, can be seen. The inset shows the characteristic LEED pattern of SLG on Pt(1 1 1). (b) High resolution STM image of a graphene nanobubble (20  ×  20 nm2, I  =  0.6 nA, V  =  310 mV) and few point-like atomic defects with atomic resolution. Figure a is adapted from [1035], with permission from Elsevier. Figure IX.25. 5  ×  5 nm2 images showing graphene point defects on Pt(1 1 1). (a) 10 mV, 3.9 nA. (b) 10 mV, 2 nA. (c) 10 mV, 3.9 nA. this region appears brighter, thus higher, in the STM respect to the metal substrate. Figure IX.27 presents images as electronic effects are normally occurring the STM images of four moiré superstructures on in the defective regions of graphene. Figure IX.25(c) SLG/Pt(1 1 1) that can be described with these struc- shows a strain-induced lattice dislocation involving at tural parameters (see blue hexagons overlaid onto it). least two unit cells of graphene. In every STM image, two periodicities can be seen: a short one, corresponding to atomic resolution, and a IX.1.4.1. Graphene on metals long one, corresponding to the moiré superstructure. The formation of graphene has been mostly studied Depending on the reactivity of the metal, the by STM on hexagonal metal surfaces, like the (0 0 0 1) interaction between graphene and metal substrate for the HCP crystallographic structures, Ru, Co, Re, can range from van der Waals physisorption to strong and the (1 1 1) for the FCC metals, Ni, Cu, Rh, Pd, Ir, Pt bonded chemisorption. The interaction between sub- and Au, but it has also been observed to grow on non- strate and graphene overlayer drives the measured hexagonal crystallographic surfaces such as Pt(1 0 0) STM corrugation and the mean distance from the sub- [1031]. strate [901]. Typical corrugations of moiré supperlat- Moiré superstructures [902] arise from the elec- tices range between 0.2–1.2 Å for graphene grown on tronic interference between two periodicities with a metal surfaces, and typical corrugations for the atomic difference in lattice parameter and/or angle. In the lattice range between 0.1–0.3 Å [901]. present case moirés are originated by the interference Figure IX.28 presents a series of defective struc- between graphene lattice and metal surface beneath it. tures appearing on the SLG/Pt(1 1 1). This figure shows Figure IX.26 shows a ball and stick model of a moiré how graphene adjusts itself to the lowest energy con- pattern for describing the different distances and formation when grown on surfaces, thus deforming its angles occurring in moiré superstructures. With these atomic structure to fit to the substrate and other gra- parameters, models minimizing lattice mismatch can phene grains in the vicinities. Figure IX.28(a) shows be used for describing the structure of graphene with two moiré domains lying in the same Pt(1 1 1) terrace, 203 2D Mater. 7 (2020) 022001 C Backes et al Figure IX.26. Diagram of the model represented for a moiré superstructure. Pt atoms are represented by blue spheres, whereas   the hexagonal lattice of graphene is represented by black spheres. The angle between the black dotted line and the Pt 1 1 0 surface direction (x axis) represents the crystallographic angle, Φ  =  25.1° for this case. The orange spheres are C atoms with the lowest mismatch for a given Φ, which define the moiré unit cell indicated by the black rhombus. The angle between the orange dashed line   and the Pt 1 1 0 direction is the moiré apparent angle (Ω). The white arrow in the inset represents the mismatch. Adapted from [902], copyright American Chemical Society. connected by an amorphous grain boundary. The reveals electronic highly localized states in one of the strain accumulated between the two domains induces graphene sublattices, and thus the brighter aspect in the appearance of an edge. Figure IX.28(b) shows that the STM images of the contact region. Theory predicts, graphene can, in some cases, run across an atomic step and STM images confirm, that this state is mainly con- without forming a grain boundary in a ‘carpet-like’ fined on the first C atomic lines of the edge [1590]. fashion (covering unperturbed the higher and lower part of the step while running through it). In this case IX.1.4.2. Silicon carbide the only effect is an out-of-plane bending of SLG. In Graphene grown on SiC(0 0 0 1) shows very interesting figure IX.28(c) a pseudo-periodic graphene grain electronic properties at the nanoscale. When inspected boundary is presented. The structure involves several, with STM, it can appear transparent, depending on unresolved, vacancies and lattice deformations. Inter- the bias voltage. Used: The honeycomb graphene estingly, in cases like this, the grain boundary can be structure is ‘visible’ only within a small range of pseudocrystalline (contrary to figure IX.28(a)) and scanning conditions which typically involve voltages adopt the periodicity of one of the two moirés that bias near EF (see figure IX.30) [659]. SLG/SiC(0 0 0 1) reaches the interface. present quantum interferences between electrons of Figure IX.29 presents a different 1d structure different sublattices, which result in a  √3  ×  √3R30° occurring on graphene on metals: the interface lattice appearance of graphene lattice near defective between graphene and a metal step (in particular a structures. All these characteristics makes G/ Pt(1 1 1) surface). In this case a crystalline structure SiC(0 0 0 1) an ideal system to test the fundamental emerges permitting to perform a combined exper­ electronic properties of this pure sp2 compound with imental-theoretical approach to study the contact STM, although typically low T STM is needed to region formed along the interface at the atomic scale. decrease electronic broadening effects. The, so called, Figure IX.29(a) shows a high-resolution STM image C-terminated SiC face or SiC(0 0 0  1) develops MLG. of atomically resolved border-like edge of (√7  ×  √7) Its layers are weakly bound and stacked in different R19° graphene with a Pt(1 1 1) step. Combining STM orientations forming graphene-graphene moirés with images with DFT simulations, the atomic and elec- different orientations and periodicities [600, 659]. tronic structure of Pt-graphene edges can be fully Graphene grown on the Si terminated surface characterized (see figure IX.29(b)). This approach SiC(0 0 0 1) can show different surface reconstruc- 204 2D Mater. 7 (2020) 022001 C Backes et al Figure IX.27. High resolution atomically resolved STM images of periodically modulated graphene structures. The red line indicates the graphene orientation with respect to the black line, which indicates the Pt [1 1 0] surface direction. The blue hexagon denotes the resulting moiré structure. All images are 5  ×  5 nm2. V ~  −250, +250 mV; I ~ 1, 3 nA. Adapted from [902], copyright American Chemical Society. Figure IX.28. Examples of defective structures of graphene. (a) edge between two rotational domains (moirés) 10  ×  10 nm2 1.9 nA, 10 mV. (b) Graphene running through a Pt step in a carpetlike fashion 7.5  ×  7.5 nm2, 8 nA, 2 mV. (c) Pseudo periodic edge, the grain boundary follows the periodicity of the moiré appearing in the lower part of the image. 18.4  ×  9 nm2, 2 nA, 10 mV. tions (see section IV). There are three main surface graphene [1493]. Third, BLG and, more generally, terminations that can be distinguished by STM. First, MLG. When SiC samples are annealed at very high T, (6  √  3  ×  6  √  3)R30°/SiC(0 0 0 1) or buffer layer ­(figure Si depletion can be high and promote MLG growth. IX.31(a)) is a carbon rich termination which is usually STM measurements of MLG (figure IX.31(c)) have described as sp2 carbon lattice covalently bound to the lower corrugation and, in some cases, the underlying uppermost SiC slab. Second, SLG, figure IX.31(b), is superperiodicity disappears and a graphene–graphene characterized by a (6  √  3  ×  6  √  3)R30° superperi- moiré superlattice dominates. Similar MLG are also odicity overlaid with an atomic scale graphene lattice. found in the SiC(0 0 0   −  1) surface [1494]. SLG can When measuring SLG at high voltage bias one can nor- be decoupled from the SiC substrate by H intercala- mally measure the atomic features of the SiC beneath tion. Figure IX.31(d) shows a STM image of hydro- 205 2D Mater. 7 (2020) 022001 C Backes et al Figure IX.29. Graphene-metal interface. (a) High resolution, atomically resolved, STM images of SLG/Pt(1 1 1)-Pt(1 1 1) interface. These crystalline edges are energetically favoured and have the same orientation as the graphene moiré superstructures. 12.6  ×  6.8 nm2, V  =  40.2 mV, I  =  5.2 nA. The inset shows the chiral vectors of this particular edge. (b) Ball and stick model of the DFT relaxed structure (right-hand part of the image) compared with the STM image of the SLG-Pt(1 1 1) edge boundary. The atomic positions of the calculations reproduce with great accuracy the protrusions in the STM images. Adapted from [1490], copyright American Chemical Society. Figure IX.30. STM images of the same region showing the bias dependence of SLG/SiC(0 0 0 1). When scanned at bias near EF (100 mV) the main feature is the graphene lattice, while scanned at higher bias (400 mV) the subsurface buffer layer structure is revealed. 7  ×  7 nm2. gen intercalated graphene. In this case, the underlying and measurements of its thermal conductivity in the (6  √  3  ×  6  √  3)R30° superperiodicity is not appear- range of a few thousands W/mK [1499]. Due to the ing in the STM images and only the graphene lattice strong covalent sp2 bonding, heat transport in carbon can be resolved. The low surface corrugation measured materials is usually dominated by phonons [1498]. with STM on the hydrogen intercalated samples might However, lattice defects, impurities, interfaces as well be related to the high µ reported for these samples. as the presence of sp3 bonds can limit heat conduction The (0 0 0   −1) face shows ordered graphene over [1498]. Several techniques have been proposed for mm scales, easily observed by STM on doped SiC. This characterizing the GRM thermal properties [1500]. allows atomic moiré and spectroscopic studies at low T However, accurate and reliable measurements remain and high magnetic field [595, 600, 627, 630]. challenging. Electrical methods and the Raman optothermal measurements were used to determine IX.1.5. Scanning thermal microscopy the thermal conductivity of both suspended and Heat transfer properties of graphene are a topic of supported SLG [1498, 1501, 1502]. The main high interest [1495–1498] based on calculations limitations are the rather complex setup and limited 206 2D Mater. 7 (2020) 022001 C Backes et al Figure IX.31. Atomically resolved STM images of the SiC surface with different terminations. (a) 6  √  3  ×  6  √  3)R30°1000 mV, 200 pA. (b) SLG  −100 mV, 200 pA. (c) MLG, −100 mV, 100 pA. (d) Hydrogen intercalated graphene. V  =  243 mv; I  =  1.33 nA. Images (a)–(c) are 10  ×  10 nm2. d is 6  ×  6 nm2. spatial resolution. Optical techniques are limited scanning on less conductive areas, since less heat is to a few hundreds nm by the diffraction limit of the carried away by the material under investigation. The incident light, while electrical measurements provide change in the sensor resistance is monitored by meas- a result representing the average of the conductivity on uring the bridge voltage, which is sent to the control- the whole specimen. ler to display the thermal maps, subsequently con- Scanning thermal microscopy (SThM) [1503, verted into T by using the Wheatstone bridge formula 1504] is an interesting technique to measure heat [1789], and the temperature coefficient of the probe transfer in GRMs, requiring limited sample prep­ ~8.92 × 10−4 K−1 . [1790] aration and with lateral resolution of some tens nm RGO flakes were studied by SThM [334]. These were or less. Although a quantitative determination of the measured both ‘as-received’ (RGO) or after a annealing thermal conductivity of GRMs by SThM [334, 1504] at 1700 °C in vacuum for 1 h (referred to as RGO_1700) is very challenging, it was demonstrated to be possi- to gain insight into the evolution of the heat dissipa- ble in at least some specific cases and with particular tion properties, due to the simultaneous reduction of probes, e.g. by electrically heating a suspended SLG disorder and defectiveness upon annealing [334]. SEM and by using the so called null point technique, adopt- and AFM analyses did not evidence significant changes ing probes integrated with a thermocouple junction in the morphology upon annealing. XPS showed a [1505]. strong reduction of oxygen, by the elimination of car- SThM can be performed by using resistive probes, boxylic and carbonyl groups. A considerable reduction where a Pd film acts both as heater and T sensor (the of defects was observed by Raman spectroscopy which so-called Pd probes) [1504]. A 40 nm [1506] Pd film, also indicated an increase of the graphitic stacking order deposited near the apex of the Si3N4 tip, acts as the ele- [334]. This result was further confirmed by x-ray diffrac- ment of a Wheatstone bridge. The resistor is heated by tion experiments which revealed an increased ordering Joule effect and, when the tip is in contact with more both along and perpendicular to the planes. Figure IX.32 thermally conducting regions of the sample, its T (and plots the SThM maps for (a) RGO and (b) RGO_1700 therefore resistance) decreases, while it increases when on Si/SiO2. Topography artefacts are o ­ bservable owing 207 2D Mater. 7 (2020) 022001 C Backes et al Figure IX.32. SThM maps of (a) RGO and (b) RGO_1700 on Si/SiO2 and (c) RGO and (d) RGO_1700 on PET. Adapted from [334], with permission from Elsevier. Figure IX.33. (a) Summary of T difference between Si/SiO2 and RGO (red) or RGO_1700 (blue). (b) The same as in (a) but for PET substrate. Adapted from [334], with permission from Elsevier. to the dependence of the tip-sample contact on steep by comparing (a) and (b), Ref. [334] concluded that edges, wrinkles and bulges on the surface [334, 1507]. the annealed flake features a higher thermal conduc- The tip heater T in panel (a) shows that on the flat areas tivity than the as-received ones, due to the reduced of the sample (see e.g. the masked region), T is basi- disorder and defectiveness upon annealing. Panel (c) cally the same as when the tip is scanning the substrate and (d) show an analogous result, but with the flakes (Tsub − TRGO = ∆T = (8 ± 14) × 10−3 K ). This supported by the less (one order of magnitude) ther- does not mean that the two materials have the same mally conducting PET. In this case, the heater T on thermal conductivity, but only that their apparent the RGO flake is now lower than that of the substrate thermal conductance is the same. On the other hand, (∆T = (103 ± 18) × 10−3 K ) and also in this case ∆T in panel (b) it can seen that for the annealed flakes T is increases when looking at the RGO_1700 sample, where now lower when the tip is on the flake than when it is it goes to ∆T = (143 ± 13) × 10−3 K . The results on on the substrate, ∆T = (42 ± 15) × 10−3 K . Thus, PET confirm that RGO_1700 flakes are more thermally 208 2D Mater. 7 (2020) 022001 C Backes et al Figure IX.34. SThM maps of (a) 1LG, (b) 2LG, and (c) 4LG, on Si/SiO2. (d) Sensor T difference as a function of N. The dashed line is a guide for the eye. conducting and also show that the use of a less conduct- ference between substrate and sample was found to be ing substrate enhances the sensitivity of the measure- ∆T = (71 ± 42) × 10−3 K [1790]. Figure IX.34(b) ment. This is due to the decrease of spreading resistance reports a similar image for 2LG. Apart from topologi- [1508] of the substrate occuring when enlarging the cal features giving rise to high-T spots, the flat areas of heat flow area, being larger when the thermal conduc- the surface feature on average a lower T than the sub- tivity of the substrate is smaller [334]. Thus, the T differ- strate, with a T difference (averaged on the red masked ence substrate-flake is also expected to be higher. area) increased to ∆T = (100 ± 48) × 10−3 K , sug- Figure IX.33 reports a summary of the SThM gesting that 2LG has a higher thermal conductance results obtained in terms of T differences between than 1LG. Figure IX.34(c) shows the 4LG case, in substrate and flakes for several samples on substrates. which both impurities and grain boundaries are vis- Although these are small and the error bars rather ible due to the topography-related T readings. The rep- large, the result is reproducible on both substrates, resentative T on the sample flat areas is much lower showing that the annealing process improves the than on the substrate. The T difference between sub- thermal conduction of flakes. No particular depend­ strate and 4LG is now higher than for the two previous ence on flake thickness of the flakes is observed, likely samples. ∆T = (245 ± 46) × 10−3 K , indicating that due to the fact that the in-plane component of the the thermal conductance of 4LG is further increased thermal conductivity is at least two orders of mag- with respect to 1LG and 2LG. Figure IX.34(d) summa- nitude higher than the out-of-plane one, and a con- rizes the results obtained by performing several SThM siderable amount of heat could be spread along the measurements as a function of N. Even though the planes. error bars for 1 and 2LG overlap, the thermal conduct- SThM was also applied on CVD 4LG, BLG, SLG ance increases with N. supported by SiO2/Si (figure IX.34). Figure IX.34(a) A similar trend was reported for mechanically shows a thermal map os SLG on SiO2/Si. The high-T exfoliated flakes using either the resistive Si [1509] or spots are due to impurities and the corresponding T Pd probes [1506]. The topography of CVD graphene, value is due to a topography artefact [1507]. When with grain boundaries and crystallites, is considerably averaging over the flat areas (with no topography arte- different from mechanically exfoliated flakes, that are facts) the measured tip T is lower than on the substrate. much flatter and more regular. This affects also the By averaging the red masked area on one 1LG and on heat dissipation properties and the thermal response a similar area on the substrate (not shown), the T dif- of the SThM probe. 209 2D Mater. 7 (2020) 022001 C Backes et al IX.2. Spectroscopies IX.2.1.1. Graphene IX.2.1. Raman A typical Raman spectrum of defect-free SLG consists Raman spectroscopy is an integral part of graphene of two main peaks, the G peak located at ~1580 cm−1 research. It may be used to determine the number and and the 2D peak at ~2700 cm−1 [96], figure IX.35. In orientation of layers, the quality and types of edge, defective graphene, additional peaks such as the D, D′ and the effects of perturbations, such as electric and and their combination D  +  D′ appear [96, 1511, 1524, magnetic fields, strain, doping, disorder an functional 1722], figure IX.35 bottom spectrum. 2D and 2D′ are groups [96]. second order peaks involving two phonons and are This, in turn, provides insight into all sp2-bonded thus always present even in the absence of defects. The carbon allotropes, because graphene is their funda- fitting parameters, such as peak position Pos, FWHM, mental building block. Ref. [96] reviewed the state of height (I) and area (A) are important to interpret the art, future directions and open questions in Raman Raman data. E.g.: I(G) for G peak height, A(G) for G spectroscopy of graphene. It described the key physi- peak area, Pos(G) for G peak position and FWHM(G) cal processes, such as the various types of resonances at for G peak full-width at half-maximum. play, and the role of quantum interference. It outlined The lineshape of the 2D peak can be used to dis- the basic concepts and notations, and a terminology tinguish between different N. For SLG it can be fitted able to describe any result in literature for GRMs. Here with a single Lorentzian, while in BLG it splits into we briefly describe some of the key concepts, referring four components as result of the evolution of the the reader to Ref. [96] for a more complete overview. band structure [105]. For N  >  10 the 2D peak shape An ideal characterization tool should be fast and is similar to that of graphite. For graphene on the car- non-destructive, offer high resolution, give struc- bon (0 0 0   −  1)face of SiC, each layer has the electronic tural and electronic information, and be applica- structure of a single layer, due to the rotational stack- ble at both laboratory and mass-production scales. ing, and the Raman 2D peak is a single Lorentzian in Raman spectroscopy fulfils all these requirements. most cases [1517], see figure IV.7(b). The C and LBMs The Raman spectrum of graphite was first recorded can be observed for GRMs. Their positions change 50 years ago [1510] and, by the time the Raman spec- with N. They can also be used to calculate coupling trum of graphene was first measured [105], Raman constants related to in-plane and out-of-plane Young’s spectroscopy had become one of the most popular moduli [1514]. Figure IX.36(b)) shows the low fre- techniques for the characterization of disordered quency Raman spectra of different graphene layers. and amorphous carbons, fullerenes, nanotubes, dia- While Pos(G) is nearly constant, Pos(C) changes and monds, carbon chains and polyconjugated molecules can be used to determine N. [1511]. Raman techniques are particularly useful for Raman spectroscopy can also be used to deter- graphene, because the absence of a bandgap makes all mine the doping level [1321, 1518, 1519]. There are wavelengths of incident radiation resonant, thus the two major effects: (i) a change of the equilibrium lat- Raman spectrum contains information about both tice parameter with a consequent stiffening/softening atomic structure and electronic properties. Reso- of the phonons, and (ii) the onset of effects beyond nance could also be reached by ultraviolet excitation the adiabatic Born–Oppenheimer approximation [1722] either with the M-point Van Hove singularity that modify the phonon dispersion close to the Kohn or in the case of bandgap opening, such as in fluori- anomalies (KA) [1518, 1520]. nated graphene. In doped samples Pos(G) increases, while Raman spectra are typically measured in the high FWHM(G) decreases for both electron or hole dop- frequency range  >100 cm−1 because of the cut-off ing [1518, 1521–1523]. The increase in Pos(G) is due limit of Raman filters (edge or notch filters). In the low to the non-adiabatic KA removal, while the decrease frequency (<100 cm−1) range interlayer interaction in FWHM(G) is due to Pauli blocking of the pho- can be estimated by the shear (C) and layer breathing non decay channel into electron hole pairs due to the modes (LBMs), which can also be used to determine N increase in EF [1522, 1518]. The 2D peak instead shifts [1514, 1515]. Traditionally, a triple stage spectro­meter up for hole doping and down for electron doping, due is used to measure low frequency Raman spectra, to the change of the equilibrium lattice parameter. but its low throughput and complexity for operation Therefore, Pos(2D) can be used to determine the type restrictsed its utilization. A new method was devel- of doping. Doping also has an effect on I(2D)/I(G) oped to get low frequency spectra easily, using volume and A(2D)/A(G). Both are maximum for zero dop- Bragg grating filters to reject a strong laser line down ing and decrease for increasing doping [1522, 1518]. to 5 cm−1 [1516, 1514]. When measuring GRMs, the Figure IX.37 summarizes the doping dependence of damage threshold for the laser power should be con- all the fitting parameters. Using Pos(2D) one can dis- sidered. Typically, the power should be kept  <1 mW tinguish between electron and hole doping, and then for SLG, and few hundreds µW for TMDs, depend- use all the remaining fitting parameters and Pos(2D) ing on the material. Some materials should be isolated to estimate the amount of doping. The charge carrier from air with a protection layer or vacuum to avoid concentration is related to the Fermi level by [1518]: photo induced oxidation [189]. EF(n)  =  ħ|vF|  √  (π n), where ħ  =  h/2π is the reduced 210 2D Mater. 7 (2020) 022001 C Backes et al Figure IX.35. Typical Raman spectra of defect free (top) and defective (bottom) graphene. Figure taken from [96], with permission of Springer. Planck constant, vF  =  1.1 * 106 m s−1 is the SLG Fermi point-like defects, separated from each other by LD, velocity equation (IX.7) can be restated in terms of defect Ref. [1524] introduced a three-stage classification density, nD, given by nD (cm−2 ) = 1014 /πL2D, as of disorder, leading from graphite to amorphous car- [1322]: bons, that allows to simply assess all the Raman spectra   I(D) of carbons: nD cm−2 = 7.3 × 109 EL4 (eV) (IX.8) . I(G) • Stage 1: graphene to nanocrystalline graphene. Equations (IX.7) and (IX.8) are limited to Raman- • Stage 2: nanocrystalline graphene to low-sp3 active defects. Perfect zig-zag edges [1526], charged amorphous carbon. impurities [1518, 1527], intercalants [1528], uniaxial • Stage 3: low-sp3 amorphous carbon to high-sp3 and biaxial strain [1275, 1529–1534] do not generate amorphous carbon. a D peak. Equations (IX.7) and (IX.8) are derived assuming Here we focus on stage 1, the most relevant when negligible EF shift. However, most samples in literature considering the vast majority of publications dealing show doping levels ~200–500 meV. I(D) depends on with graphene production, processing and applica- the doping level [1321]: I(D) decreases for increasing tions. As shown in figure IX.38, in stage 1 the Raman doping. This needs to be taken into account when esti- spectrum evolves as follows: (a) the D peak appears, mating the defect density. Equations (IX.7) and (IX.8) I(D)/I(G) increases; (b) the D’ peak appears; (c) all can be modified for samples with n ­ on-negligible dop- peaks broaden; (d) the D  +  D′ peak appears; (e) at the ing [1321]: end of stage 1, the G and D′ peaks are so wide that it ï ò   (1.2 ± 0.3) × 103 I(D) −1 −(0.54±0.04) is sometimes more convenient to consider them as a L2D nm2 = {EF [eV]} EL4 [eV] I(G) single, up-shifted, wide G band at ~1600 cm−1 [1524]. (IX.9) For a sample with rare defects I(D) is proportional   I(D) (0.54±0.04) to the total number of point defects probed by the laser nD cm−2 = (2.7 ± 0.8) × 1010 EL4 [eV] {EF [eV]} . I(G) spot, giving rise to I(D)/I(G) ~ 1/L2D, where LD is the (IX.10) average inter-defect distance. Taking into account the Equations (IX.9) and (IX.10) are valid for samples excitation energy dependence of the peak areas and with a defect concentration corresponding to stage 1, intensities the inter-defect distance can be expressed and for EF  <  EL/2, which is by far the most relevant for by [1322, 1524]: graphene production and applications. ï ò In order to extract the defect density correctly the   4.3 × 103 I(D) −1 L2D nm2 = 4 (IX.7) . flowchart presented in figure IX.39 should be followed. EL (eV) I(G) After recording a Raman spectrum, one first has EL is the laser excitation energy in eV. This relation to check whether a D peak is present. In case no D is valid for LD  ⩾  10 nm (stage 1). By considering peak is present, the standard procedure of estimating 211 2D Mater. 7 (2020) 022001 C Backes et al Figure IX.36. (a) Raman spectrum of 2D peak as function of N [78]. (b) and G peaks as function of N. (c) Pos(C) and Pos(G) as a function of 1/N [1514]. doping from the Raman fitting parameters, such as decrease of I(D) with doping is seen. The red spectrum Pos(G), FWHM(G), Pos(2D), I(2D)/I(G) and for the undoped case gives I(D)/I(G)  =  2.8. Using A(2D)/A(G), can be used. The presence of a D peak, equations (IX.7) and (IX.8) gives nD ~ 3.1  ×  1011 leads to the next step, where one needs to check if cm−2 LD  =  10 nm. The black line gives I(D)/I(G) ~ the sample is doped or not, e.g. by looking at the rel- 1.24. Taking the doping level ~500 meV into account evant Raman peak fitting parameters. If the sample nD  =  3.4  ×  1011 cm−1 and LD  =  10 nm, are obtained has doping  ⩽100 meV, equation (IX.8), can be used using equations (IX.9) and (IX.10). to estimate the defect density. If however the sample Strain can be present in a material compressed or has non-negligible dopingequation (IX.10) has to be stretched out of equilibrium. Strain can be either com- used. E.g. Figure IX.40 shows spectra of a defective SLG pressive or tensile. While compressive strain causes at low, EF  =  100 meV, and high, EF ~ 500 meV, dop- an upshift or stiffening of the Raman peaks, tensile ing. By analysing the spectra of doped and defected strain causes a downshift or softening. Strain can be SLG (black line in figure IX.40) Pos(G)  =  1599 cm−1, either isotropic, i.e. biaxial strain, or along a certain FWHM(G)  =  12 cm−1 and I(2D)/I(G)  =  0.94 are axis, i.e. uniaxial. Biaxial strain does not cause any obtained. This allows us to estimate ~500 meV for change in peak shape, while uniaxial strain causes the doping level of the sample using the graphs splitting of the G peak into two components G+ and from Ref. [1518]. The 2D peak is upshifted with G− for ε  ⩾  0.5% [1529], figure IX.41. The G peak is Pos(2D)  =  2651 cm−1 suggesting hole doping. A due to a doubly degenerate E2g mode [105]. By apply- 212 2D Mater. 7 (2020) 022001 C Backes et al Figure IX.37. Doping dependence of Raman fitting parameters (a) Pos(G), (b) FWHM(G), (c) Pos(2D), (d) I(2D)/I(G) versus electron concentration (×1013 cm−2) and (e) [A(2D)/A(G)]0.5 as a function of EF. Adapted from Ref. [1518], with permission of Springer and from Ref. [1519], copyright by the American Physical Society. ing uniaxial strain this splits into two components, one IX.2.1.2. HBN along the strain direction and one perpendicu- Two Raman peaks in hBN are observed with the lar to it. While Pos(2D) changes as δPos(2D)/δε excitation of visible laser light, the C peak near 50 cm−1 ~  −64 cm−1/%, the G peak changes as δPos(G+)/δε and a high frequency E2g mode near 1366 cm−1 [1537]. ~  −10.8 cm−1/% and δPos(G−)/δε ~  −31.7 cm−1/% The C peak strongly depends on N so that it can be used [1529] in uniaxial, tensile strain experiments. Ref. to estimate N [1048, 1538, 96]. The high frequency E2g [1529] reported δPos(2D)/δε ~  −144 cm−1/% and Raman peak weakly depends on N. The peak positions δPos(G)/δε ~  −58 cm−1/% for biaxial strain. To esti- for the mono layer and bulk are ~1368 cm−1 and mate the strain, the difference between fitted peak ~1366 cm−1 [1048, 1539–1541] positions and unstrained reference peak positions needs to be derived. Then, by knowing how much the IX.2.1.3. Transition metal dichalcogenides peaks shift with strain, i.e. δPos(2D)/δε ~  −64 cm−1/% Many TMDs are composed of three atomic layers in case of the 2D peak for uniaxial strain, one can cal- X–M–X, represented by MX2, as well as ordering, culate the strain. To distinguish and decouple the com- 2H (Hexagonal symmetry), 3R (rhombohedral bined effects of doping and strain a thorough analysis symmetry) and 1T (tetragonal symmetry).The most of all the fitting parameters is necessary [1527]. common polytype is 2H, e.g., MoS2, MoSe2, WS2, 213 2D Mater. 7 (2020) 022001 C Backes et al Figure IX.38. Representative Raman spectra of ion bombarded SLG measured at EL  =  2.41 eV [1322]. Figure IX.39. Flowchart to estimate defect density in doped and undoped samples. and WSe2. For 2H-MX2, two prominent Raman MoS2 broaden the E′ and A1′ peaks and E′ becomes peaks appear in the high frequency range, A1′ and E′ asymmetric towards lower frequencies [1545]. A modes. A1′ is an out-of-plane vibrational mode and peak ~ 200 cm−1 appears due to the defect induced E′ is an in-plane one [1543, 1544]. The peak positions longitudinal acoustic (LA) phonon at the M point depend on N so that the difference of their positions [1545]. As uniaxial strain increases on 1L-MoS2, may be used to estimate N. However, one needs to the A1′ peak shows no measurable shift while the be careful since the dielectric environment can also E′ peak splits into two, E′+ and E′− [1546]. The E′− change this difference, for a fixed N. Typically, A1′ peak shifts ~ 4.5 cm−1/% strain and E′+ ~1.0 cm−1/% decreases with decreasing N, and E′ has the opposite strain. A1′ is not sensitive to strain, while it is sensitive trend. In the low frequency range, C and LBMs are to doping. By increasing doping of 1L-MoS2, A′ observed and can be used to estimate the coupling redshifts and broadens due to electron–phonon strength between layers and N [96]. Defects in interactions [1542]. 214 2D Mater. 7 (2020) 022001 C Backes et al Figure IX.40. Raman spectra of a defective graphene sample (a) EF  ⩽  100 meV and (b) EF ~ 500 meV. Excitation wavelength: 633 nm. Adapted from [1321], copyright American Chemical Society. Figure IX.41. Change in Pos(G) and Pos(2D) with increasing tensile, uniaxial strain. Adapted from [1529], copyright American Chemical Society. 215 2D Mater. 7 (2020) 022001 C Backes et al IX.2.2. Absorbance and fluorescence extinction coefficient at these spectral positions can be IX.2.2.1. Transition metal dichalcogenides used as a robust measure for Co over a broad range of UV/Vis transmission spectroscopy can provide sizes. In addition, extinction spectra do not only pro- information GRM flakes, as inorganic LMs have vide insight in lateral size and dispersed concentration, well documented excitonic transitions [1547] but also in thickness. N can be quanti­tatively related to characteristic for the material and even polytype or the peak position/energy of the A-exciton (obtained phase [378, 379, 1407]. Information on concentration from an analysis of the second derivative), as plotted in can be extracted [185, 186, 236, 1407, 1548]. However, figure IX.42(H). for this to be performed reliably, one needs to be aware PL is also a very powerful tool to study GRMs, as it that when transmittance mode is used, extinction provides information on the band gap and its direct/ and not absorbance spectra are obtained. Extinction indirect nature [1550, 1551]. In PL measurements, is related to transmittance, Tr, via Tr  =  10−Ext, where GRMs are excited with a laser with higher energy than Ext  =  εcCol, with εc the extinction coefficient, Co the the bandgap to create electron–hole pairs. The emis- flakes concentration and l as the path length. This sion yield upon radiative recombination is studied means that the quantity that is measured contains with a spectrometer. The excitation laser is notch fil- information related to both absorbance and scattering tered at the entry of the spectrometer to remove the of light [185, 188, 190, 236, 240, 1549]. Although Rayleigh contribution, which is much larger than the scattering component of the extinction is size the PL yield and might saturate the detector. Fig- dependent [185, 188, 190, 236, 240, 1549], it is also ure IX.43(a) compares PL spectra in MoS2 flakes with the case that the absorbance coefficient is dependent different N. The peaks correspond to the recombina- on size, due to edge effects [185, 186, 188, 190]. Hence, tion of excitons. A and B are related to direct bandgap the overall extinction coefficient is size-dependent transitions at the K point of the Brillouin zone and I and it is necessary to determine it gravimetrically for corresponds to an indirect bandgap transition along dispersions with a range of lateral sizes and thicknesses ΓK, that becomes favourable for MLs. Therefore, PL to achieve an accurate Co determination. allows one to observe quantum confinement in MoS2, The change in extinction and absorbance spectra resulting in a N dependent bandgap. PL also makes evi- with lateral size can be used to establish metrics for dent the direct-to-indirect transition that occurs due lateral size once the sizes are calibrated by statistical to the interlayer interaction in MoS2. While 1L-MoS2 microscopy. Such relationships were established for has direct bandgap with a bright PL emission, ML MoS2 [185], WS2 [186], GaS [188] and black phos- flakes are indirect bandgap with a much lower PL yield. phors [190]. Changes in LM band structure as a result of IX.2.2.2. HBN confinement are reflected in changes in peak Spectroscopy can be used to investigate and to position of excitonic transitions. This means that benchmark the optical and structural properties of information on N is encoded in optical extinction both bulk and exfoliated hBN. The hyperspectral spectra [185, 186, 188]. capabilities of cathodoluminescence at 10 K are An example of the spectral changes with size and particularly suited [1552, 1553], figure IX.44. This thickness for LPE MoS2 and WS2 is in figure IX.42 figure shows a typical cathodoluminescence spectrum [245]. Figures IX.42(A) and (C) plot optical extinction at 10 K on a sample prepared from a reference single spectra of MoS2 (A) and WS2 (C) with different mean crystal [1554] together with monochromatic images flake sizes and thicknesses. Due to the power law non- recorded at energies related to the different emission resonant scattering background (>680 nm for MoS2, bands. It displays emission lines in a broad energy >650 nm for WS2), peak positions are best found from range from almost 6 to 3 eV. the second derivative. Second derivative spectra of Luminescence properties of hBN are governed by a subset of the samples in the region of the A-exciton strong excitonic effects resulting in recombinations at are shown in figures IX.42(B) and (D). Peak intensity energies below the bandgap [1791]. It is according to ratios can be used to express the changes associated ab initio calculations [1791], hBN is a large band gap with edge effects. Figures IX.42(E) and (F) plot peak material with an indirect band gap around 6 eV and intensity ratios as function of mean flake length for a direct band gap ~6.5 eV [1791]. Excitonic recombi- MoS2 and WS2, showing that both materials behave nations are near band edge (NBE) in the deepest UV similarly. This means that the flake size for both mat­ region in figure IX.44. Among them, a series of peaks erials can be quantitatively linked to the length via at the shortest wavelength (210–215 nm) is referred identical equations [245]. Due to the changes in spec- to as the S series [1048, 1555]. The excitonic nature of tral shape, extinction coefficients are also dependent on these recombinations is supported by theoretical stud- size. This is more or less severe depending on spectral ies [1556, 1557], but their fine structure splitting and position. E.g. as plotted in figure IX.42(G), the extinc- their correlation with absorption data are still debated tion coefficient of the A-exciton for both mat­erials is [1555, 1558–1561]. strongly length dependent. However, this is not the Other features can be observed in the 210– 250 nm case at 345 nm for MoS2 and 235 nm for WS2 so that the region, and tentatively assigned to NBE [1048]. 216 2D Mater. 7 (2020) 022001 C Backes et al A B 100 Min Extinction (L g-1 cm-1) SIZE 80 d2Ext/dE2 60 A 40 MaxHE 20 MoS2 MoS2 0 200 400 600 800 1.75 1.80 1.85 1.90 1.95 Wavelength (nm) Energy (eV) E F 2 MoS2 MoS2 Ext. Ratio MaxHE/Min Ext. Ratio A/Min 1.8 1 WS2 WS2 1.6 1.4 0.3 1.2 0.1 1 100 1000 100 1000 (nm) (nm) C D Min Extinction (L g-1 cm-1) SIZE 40 d2Ext/dE2 A 20 MaxHE WS2 WS2 0 200 400 600 800 1.9 2.0 2.1 Wavelength (nm) Energy (eV) G H 80 A-exciton Energy (eV) 345 nm 2.0 ε (L g-1 cm-1) 60 235 nm 40 1.9 20 A-exc. A-exc. 0 1.8 100 1 10 (nm) Figure IX.42. (A) and (C) Optical extinction spectra of LCC separated MoS2 (A) and WS2 (C). Peaks relevant for the analysis are indicated. (B) and (D) Second derivatives of the A-exciton versus energy for MoS2 (B) and WS2 (D) after smoothing the second derivative with Adjacent Averaging. The solid lines are fits to the second derivative of a Lorentzian to assess peak positions/energies. (E) and (F) peak intensity ratios as a function of 〈L〉. Data for MoS2 and WS2 fall on the same curve. Hence, the same equation can be used to quantify the flakes' length. (E) peak intensity ratio at the A-exciton/local minimum. (F) peak intensity ratio at the high energy maximum/local minimum. (G) Extinction coefficient at different spectral positions as function of flakes' length. At some spectra positions (such as the A-exciton), extinction coefficients are size dependent, while at others (345 nm for MoS2 and 235 nm for WS2) this is not the case. (H) A-exciton peak energies (from second derivatives) as function of 〈N〉. Adapted with permission from [245]. At larger wavelengths than the S series, the most docu- ure IX.44, attributed to donor-acceptor pairs (DAP) mented luminescence feature is the so-called D series recombination processes [1564]. (220–227 nm) (see figure IX.44) related to the presence The series of CL monochromatic images recorded of extended structural defects. These emissions arise from the same zone, figure IX.44, illustrates the dis- from recombinations of excitons trapped at structural tinct origins of the S and D features. The image, related defects [1048], such as stacking defects, dislocations, to the intrinsic emissions (S series) and recorded by or grain boundaries [1553, 1562, 1563]. Finally, a weak collecting photons at 215  ±  3 nm, is almost homoge- and broader emission is often observed at 233 nm, fig- neous, except along parallel black lines, due to thick- 217 2D Mater. 7 (2020) 022001 C Backes et al Figure IX.43. (a) PL spectra on MoS2 flakes with N  =  1–6. The spectra show peaks corresponding to excitonic features, labelled A, B and I. (b) N dependence of the lowest energy PL peak peak. Adapted from [1550], copyright American Physical Society. Figure IX.44. Upper: typical cathodoluminescence spectrum of hBN in the 200–1000 nm range acquired at 10 K on a sample prepared from a reference single crystal [1554].Lower: SEM image (black) with the corresponding monochromatic CL images recorded in the same area at  ±3 nm around 215 nm(blue), 227 nm(green) and 303 nm(orange). Scale bar: 1 µm. Reproduced from [1048], © IOP Publishing Ltd. 12 8 4 0 Figure IX.45. Left: SEM image of a hBN crystal flake. Right: Catholuminescence mapping of the D/S ratio of the integrated intensity of the D band relatively to that of the S band. Adapted from [1048], © IOP Publishing Ltd. 218 2D Mater. 7 (2020) 022001 C Backes et al Wavelength (nm) Wavelength (nm) 225 220 215 210 225 220 215 210 bulk crystal CL intensity (a.u.) 90L CL corrected intensity (a.u.) 55L 25L 20L 5,4 5,5 5,6 5,7 5,8 5,9 6,0 8L Photon energy (ev) 6L 5.4 5.5 5.6 5.7 5.8 5.9 6.0 Photon energy (eV) Figure IX.46. Left: series of CL spectra recorded at 10 K on MC flakes with different thicknesses from bulk to 6L and transferred on to Si/SiO2. N is determined from both optical and AFM measurements. Right: CL spectra on 6L exfoliated flakes and from a grown single crystal [1554] (bottom spectrum). Adapted from [1555] with permission of The Royal Society of Chemistry. ness variations along the sample (see the SEM image of or hBN phonos [1566]. In figure IX.44, the CL image the same area). The CL image taken from the D series recorded from the ZPL emission at 302.8  ±  3 nm (227  ±  3 nm) displays several continuous lines cross- reveals a spatial distribution which is typical of chemi- ing the sample in random directions at places where cal impurities. The luminescence from the deep structural defects are present. It was proposed to use defects is distributed as bright spots, which confirms the ratio of the integrated intensity of the D band rela- their distinct origin from D and S series. The observed tive to that of the S band, as an indicator of structural luminescence spots unveil a discrete and random dis- defects [1563]. Typically in defect-free areas, D/S can tribution of impurity centers throughout the sample. be as low as 0.3, while it can increase up to two order Similar spots were reported and identified as single scales of magnitude at the location of structural photon emitters, indicating that they correspond to defects. Therefore, as illustrated in figure IX.45, hyper- isolated atomic centers randomly diluted in the crystal spectral mapping of D/S is a valuable tool to get, in a lattice [1567]. Sharp emissions from deep defects have non destructive way, the spatial distribution of struc- been identified as single emitters in the visible and the tural defects in hBN and assess its crystalline quality NIR after introducing defects in hBN [1568]. [1048]. The dependence on N of NBE emission has been At larger wavelengths (>250 nm), the lumines- investigated on MC flakes [1555]. The samples were cence spectrum of figure IX.44 reveals deep defect free of defects with no detectable D band. Typical emissions. A luminescence signal with a maximum at thickness evolution of the CL spectra is shown in fig- 302.8 nm is regularly detected in various hBN sources ure IX.46. The S series, characteristic of bulk hBN, [1048]. This sharp: its linewidth  <0.075 nm, the spec- progressively vanishes in the thinnest layers in favor of tral resolution of the experiment. It has been identi- a single peak ~ 5.909 eV in 6L samples. Such an effect fied as the zero-phonon emission line (ZPL), labelled was observed in different kinds of samples, includ- here α, of impurities introduced in the lattice during ing reference single crystals and commercial sources, synthesis, probably C (or possibly O) as revealed by a which points to an intrinsic behaviour and can be con- detailed SIMS analysis [1554]. The β and γ peaks are sidered as a signature of low dimensionality effects on assigned to phonon replicas, with an energy ~195 the intrinsic luminescence of hBN. With the support meV, involving a local vibrational mode (LVM) [1565] of combined ab initio and tight binding calculations, 219 2D Mater. 7 (2020) 022001 C Backes et al Figure IX.47. XPS spectra of three samples: (top) Pt(1 1 1), (middle) SLG/Pt(1 1 1), (bottom) contaminated SLG/Pt(1 1 1). Grey (blue) area represents the zone where C1s (O1s) related peaks appear. Figure IX.48. C1s XPS spectra of (a) SLG/ Pt(1 1 1) (b) SLG/Cu(1 1 1) and (c) SLG/Cu polycrystalline. the remaining line observed in the thinnest layers the effective binding energy of the corresponding atom. was assigned to direct bright excitons, as theoretically The energy and intensity of the emitted photoelectrons expected for 1L [1556]. can be analysed to identify the elemental and chemical composition in the parts per thousand [1569, 1570]. IX.2.3. XPS Photoelectrons originate from depths below 10 nm, XPS measures the kinetic energy of photoelectrons therefore the information obtained is surface- emitted from a sample irradiated with x-rays, which sensitive [1569]. The XPS equipment has to detect transfer their energy to a core-level electron. This electrons, thus the sample has to be introduced in a electron is emitted out of the sample surface from its UHV environment. The area illuminated is typically initial state with a kinetic energy dependent on the ~0.5–1 mm when a x-ray monochromator is used incident x-ray photon energy and the binding energy [1569], meaning that the information corresponds of the atomic orbital from which it is originated, to an average on the irradiated area. The best energy Ekinetic  =  Ephoton  −  E_binding-efective_wf, where Ekinetic is resolution that one may have is ~ 0.4 eV in standard the Kinetic Energy of the emitted electron, Ephoton is the machines. However, this may increase to ~0.2 eV in energy of the incident photon and E_binding-efective_wf is synchrotron radiation XPS facilities. 220 2D Mater. 7 (2020) 022001 C Backes et al Figure IX.49. (a) C1s XPS spectrum of SLG/ Pt(1 1 1) after exposure to air and water. New components related with epoxy and hydroxyl groups appear as a tail of the main C-sp2 one. (b) Schematic representation of N and C configuration of heteroatoms on a graphene flake related with the XPS peaks assignments. C1 (light blue): sp2 C atoms; C2 (purple): C–N pyridine ring; N1 (dark blue): N atoms in the pyridine ring interacting with the metal substrate (Cu in this example); N2 (orange): substitutional N atoms in the graphene network. Adapted from [1577] with permission of the PCCP Owner Societies. When analysing graphene based samples there are The chemical state of the C atoms of graphene can three main figures that can be extracted from the XPS be also derived from XPS. The characteristic binding spectra: presence of contaminants (or heteroatoms), energy of C 1s when C is in a sp2 configuration (either coverage, and chemical state of the C atoms. graphene or HOPG) is ~284.5 eV [1571] (the precise The presence of contaminants (heteroatoms react- value depends on the analyser) [1572]. When SLG is ing with graphene or other elements) can be derived grown on substrates, such as metals or semiconduc- from a survey spectrum of the sample, i.e. measuring tors, the C 1s binding energy might slightly change in a wide range all the relevant peaks. This usually cov- due to the interaction between graphene and substrate ers from the EFto ~50 eV less than the x-ray photon [1569]. C-sp2 is ubiquitous as many contaminants energy, and shows core-level peaks from all elements include carbonaceous species that are difficult to dis- on the sample surface. The amount of impurities can cern from graphene. be determined by integrating the area of the different Bonding of C with other atoms leads to different levels and comparing the ratio of the different peaks. binding energies, therefore a precise determination The analytic expressions for deriving the coverage of of the C1s binding energy may give the type of bond- impurities or graphene patches with respect to a sub- ing. This is known as core level shift, and most of the strate can be found in Refs. [1569, 1570]. values are tabulated [1570]. C-sp3 appears ~285.0 eV The upper spectrum in figure IX.47 was recorded [1573–1576]. To derive the different components of an from a clean Pt(1 1 1) surface. Only Pt peaks are vis- XPS peak, a curve fitting is necessary. ible. The middle one belongs to the same sample Figure IX.48 shows three C1S XPS spectra of SLG after SLG growth. In addition to the Pt peaks, a new grown on different substrates by thermal decomposi- peak related to carbon (C1s) appears. If the sample is tion of fullerenes in a UHV environment on Pt(1 1 1) exposed to air, it can get contaminated, mainly with O [1035] Cu(1 1 1) and polycrystalline Cu. The peaks that binds to SLG as epoxy or hydroxyl groups [935, are fitted with two components. The purple and nar- 1035, 806]. The lower XPS survey spectrum shows row one is related with C-sp2. This component is fit- in addition to the Pt and C peaks the oxygen peak ted using a Doniach-Sunjic curve, which is the math- (O1s). This technique is surface-sensitive. When- ematical function that better fits an XPS core level ever the sample gets something on top, i.e. graphene, peak [1569]. In this curve, the asymmetry parameter O, etc, the intensity of its main peaks decreases. In is 0.068 eV [1569]. The Lorentzian FWHM is ~0.13 eV. figure IX.47 the Pt4f peaks intensity decrease from The binding energies for the C-sp2 are 284.1, 284.6, the upper spectrum to the lower one, as SLG is grown, 284.3 eV for SLG on Pt(1 1 1), Cu(1 1 1) and polycrys- and then exposed to air. talline Cu respectively. Another small component is 221 2D Mater. 7 (2020) 022001 C Backes et al Table IX.2. Binding energies of graphene C atoms in different chemical states. C1, C2, N1 and N2 are related with the scheme of figure IX.49(b) and represent C-sp2 (C1), C-N pyridine (C2), N atoms in the pyridine ring interacting with the metal (N1) and substitutional N atoms in the graphene network (N2). In addition, binding energies of C-metal, sp3 configuration, radical-N, epoxy and hydroxyl groups are also given. The values may slightly change depending on substrate. O–C= C1 C2 N1 N2 R–N–C C-metal sp3 C–OH C–O C=O C/C–C=O Binding energy (eV) 284.4 285.9 398.0 400.6 399.5 283.2 284.7 285.4 286.7 288.5 290.5 Ref [1577] [1577] [1577] [1577] [1340] [1577] [1035] [926] [926] [926] [926] needed to complete the fitting, with a binding energy IX.2.3.1. Graphene grown on SiC that depends on the substrate. This can be assigned to XPS characterization of C-rich SiC(0 0 0 1) surfaces C-sp3 [1573–1575].This component appears in defec- can be used to extract the degree of graphitization. tive SLG due to different domains and grain bounda- SiC surfaces prepared in UHV conditions present ries [902]. In figure IX.48 the fitted values are: 284.7, a series of surface reconstructions depending on the 285.4, 284.8 eV for SLG on Pt(1 1 1), Cu(1 1 1) and C/Si stoichiometry [1579]. For T  >  1350 K, Si depletion polycrystalline Cu. The width of these peaks (0.13 eV) from the surface eventually leads a reconstruction may increase when the graphene surface is damaged, presenting a (6  √  3  ×  6  √  3)R30° LEED pattern [592, contaminated or oxidized. 645, 659, 1580, 1581]. The atomic structure is normally Figure IX.49(a) shows an example C1s peak of SLG described as a graphene-like honeycomb carbon layer exposed to air and water. This is a broad with many highly buckled and partially covalently bonded to the components. The purple one at 284.1 corresponds to uppermost Si layer [1582] (see section IV). C-sp2 and the yellow one at 284.7 to C-sp3 (as in the This covalent interaction alters the electronic prop- clean sample in figure IX.48(a)). Three new comp­ erties of the graphene-like layer and it becomes a semi- onents related to surface contaminants are also pre- conductor [594]. XPS is presented in figure IX.50. The sent: the light blue at 285.1 assigned to C–OH, the red lower spectrum is an overview on a (6  √  3  ×  6  √  3) one at 286.5 assigned to C–O and the dark blue at 288.5 R30° surface. The C1s peak (284.8 eV) has higher inten- assigned to C=O [806, 1573]. They can be fitted using sity than the two Si-related peaks: Si2p (101.4 eV) and Gaussian-Lorentzians. These epoxy groups (CO bond- Si2s (152.0 eV) [926, 1570]. The C1s/Si2p ratio is ~1.15 ing or oxidized species) and hydroxyl groups appear as for 1486.6 eV, indicating that the surface is C-rich [659]. a tail of the main peak at higher binding energies and A SLG/SiC(0 0 0 1) spectrum is presented in they are a fingerprint of the ‘cleanliness’ of a graphene ­figure IX.50 for comparison. The upper black spectrum sample. A soft annealing in vacuum will remove these corresponds to a synchrotron radiation-based overview type of contaminants. Table IX.2 shows typical bind- of a sample annealed at 1400 K recorded with a photon ing energies of some of the most important oxygen- energy of 600 eV [1583] The C1s peak dominates, with ated species (epoxy and hydroxyl) bonded to C The C1s/Si2p ~ 2.53, which indicates a higher C concentra- tion on the surface than in the (6  √  3  ×  6  √  3)R30. binding energy may slightly change (within  ±0.5 eV) The good resolution of the XPS in figure IX.51 allowed depending on system or substrate used. to decompose the peaks into their curve-components, It is important not to mistake the contamination which results from a convolution of a purely Lorentzian of graphene with GO. RGO can be derived from GO by one (FWHM, 0.12–0.2 eV) with a Gaussian distribution reducing it with chemical reducing agents. This pro- (FWHM, 0.4–0.7 eV) [1584]. cess can be followed by XPS by evaluating the ratio of C chemistry is very rich and the interpretation of the component ~286.5 eV (mainly C–O bonds) with the C1s spectra is complicated. One could decompose the C-sp2 [1578]. the C1s peak of SLG/SiC(0 0 0 1) in five components. Heteroatoms. In some cases one may be inter- That at 284.83 eV can be assigned to C-sp2, and is the ested in the binding energy not of the C atoms, but biggest contribution to the C1s core level peak, with of other elements covalently linked to graphene. The 58.2% of the total area. This is used for calibration of most important case corresponds to N heteroatoms. the the Lorentzian and Gaussian widths. The comp­ Its binding energy strongly depends on the adsorption onents at lower binding energies (283.98 eV) can be configuration inside the graphene network. A scheme attributed to carbides [1585], due to the C in the SiC C and N atoms configuration in relation with the XPS bulk [588, 589]. There are three other components components is shown in figure IX.49(b). Four possible at higher binding energies, which can in principle be configurations are shown: C-sp2 (C1, light blue circle), assigned to different configurations of C in the rear- C in a pyridine ring (C2, purple circle), N in the pyri- rangement that takes place on surface reconstruc- dine ring interacting with the metal substrate (Cu for tion [589]. These can be assigned to surface-related this case) (N1, dark blue circle) and a substitutional N peaks: C-Si 1 at 285.26 eV (orange component in fig- in the graphene network (N2, orange circle). The XPS ure IX.51(a)), C–Si 2 at 285.73 eV (pink component binding energies in these particular configurations are in figure IX.51(a)) and S. C-Si 3 at 286.4 eV (bordeaux in table IX.2. component in figure IX.51(b)). 222 2D Mater. 7 (2020) 022001 C Backes et al Figure IX.50. Lower spectrum: (6  √  3  ×  6  √  3)R30° surface using an Al Kα anode (1486.6 eV) as x-ray source. Upper spectrum: synchrotron radiation XPS overview of SLG/SiC(0 0 0 1) annealed at 1400 K for 600 eV photons. Figure IX.51. (a) XPS Spectrum of Si2p peak of SLG/SiC annealed at 1400 K recorded with a photon energy of 150 eV. (b) XPS spectrum of C1s peak of SLG/SiC annealed at 1400 K recorded at 400 eV. The Si2p curve can be decomposed in five Si dou- of the electronic wavefunctions in the (x,y ) plane, blets. Based on H adsorption it is possible to distin- the band structure of GRMs can be fully determined guish surface-related from bulk-related components. through angle-resolved photoemission spectroscopy Three components present intensity decrease upon H (ARPES) [1586–1588]. This technique takes advantage exposition. The intensity of the components at 100.4, of the photoelectric effect [1589], hence UV or soft 101.45 and 101.85 eV decrease ~15% after H exposure, x-ray radiation is used to extract electrons from the while the other two at 100.9 and 102.4 eV present a investigated sample. small (3%–4%) increase [1585]. This suggests that the The photons can be generated by laboratory components which decrease and those that increase sources, such as He or Hg lamps, or by synchrotron are related to surface and bulk Si atoms respectively. radiation. During photoemission, the crystal momen- tum of the electron inside the material is conserved in IX.2.4. ARPES the plane parallel to the samples’s surface. As a conse- ARPES is a technique that extracts the information quence, one can relate the electron’s crystal momen- retained by the electrons, emitted from a solid surface tum with the photoemission angle and determine upon excitation with a photon. Due to the confinement the band structure. A typical experimental set-up for 223 2D Mater. 7 (2020) 022001 C Backes et al Figure IX.52. ARPES set up scheme. Electromagnetic radiation provides enough energy for electrons to the sample to leave and travel towards the analyzer. ARPES is sketched in figure IX.52. The radiation is takes into account the angular dependence of the shone over the sample so that the photon beam and photoemitted electrons. the analyzer lay in the same plane. The photoemitted When the electrons leave the material, they end electrons are collected, within a certain angular accept- in a free particle state. Considering the scheme in ance, through a series of focusing electrostatic lenses ­figure IX.53 and assuming the conservation of the into a hemispherical analyzer. Here, only the electrons electron momentum parallel to the surface, the follow- with a certain kinetic energy E0 travel in the middle of ing expression holds [1586]: the hemispherical capacitor. The width of the energy … window about E0 is called pass energy, and it is a rel- 2mEkin (IX.12) k|| int = k||ext = sin(θ) evant parameter in determining the energy resolution 2 of the instrument. The other important parameter is where θ is the emission angle and Ekin is the kinetic the radius of the hemispherical capacitor. Finally, the energy. When considering the perpendicular electrons impinge onto the detector, which in most component, the description is not this simple, since cases is a fluorescent screen. A charge-coupled device it is not conserved when crossing the crystal surface (CCD) camera records the light intensity emitted by [1586]. An expression to determine it is [1586]: the detector. … » In the UV, soft x-ray regime, the momentum car- 2m (IX.13) k⊥ int = Ekin cos2 (θ) + V0 ried by the photon pph = hυ/c is negligible com- 2 pared to the crystal momentum of the electron where V0 is the inner potential, defined as the energy k  pph, and this ensures the identity between the difference between the vacuum level and the top of the final and initial electron momentuma, i.e. kf = ki . valence band. Hence, the von Laue equation kf = ki + G is fulfilled, The resolution on the measurement of k|| is: where G is a base vector of the crystal’s reciprocal  Å ã Å ã2 space. Experiments are usually performed using pho- ∆E 2 ∆θ (IX.14) ∆k || = k|| + ton energy  <100 eV corresponding to a momentum 2Ekin tan(θ) p −1  ∼ 0.05 Å . In the so-called sudden approx­imation, where ∆θ is the angular acceptance of the analyzer. the electron is extracted instantaneously  −from the If the electron signal is spin-filtered, e.g. by letting material with no time to interact with the remaining it passing through a Mott detector, it is possible to N  −  1 electrons [1586]. This allows one to consider reveal the contribution of a single spin component the obtained dispersion as the non-interacting state of projected onto the detector’s plane [1590] on a par- the system, or the one particle dispersion. As a conse- ticular band. This technique is called spin-resolved quence, for GRMs, the emitted photocurrent takes the ARPES (SARPES) and it is fundamental for determin- form [1792]: ing the spin-polarization of electronic bands.  2 A recent development in ultra-fast lasers gave the I (k, E, hυ) ∝ Mi,f  A (k, E) f (E, T) + B (IX.11)   opportunity to realize facilities for pump-probe exper- where Mi,f  is the matrix element of the transition iments that can access the time scales of the quasipar- from the initial (i) to the final (f ) state, A(k, E) is ticle dynamics in solid-state materials [1794]. Time- the spectral function, f (E, T) is the Fermi–Dirac resolved ARPES (Tr-ARPES) is being used more and distribution and B is a background term coming from more to investigate the carrier dynamics in GRMs as inelastically scattered electrons. The spectral function well as charge transfer in LMhs. 224 2D Mater. 7 (2020) 022001 C Backes et al Figure IX.53. Momentum component conservation at the crystal vacuum interface. Only k|| is conserved. IX.2.4.1. ARPES on graphene and vertical LMHs A large amount of photoemission measurements have been carried out on graphene since the early 1970s [586, 663, 664], and ARPES measurements have since mapped the dispersion of the π-bands over the entire BZ. The energy loss due to quasiparticle excitation, the electron phonon interaction in π and σ bands [671, 1591–1593] were investigated. Graphene on metals [1594], exfoliated, suspended and GNRs [592, 618, 620] have been measured. Perhaps the most inves- tigated graphene with ARPES is that grown on SiC, because it has cm2 area coverage, and because of its flatness (figure VII.10) [592, 671, 673, 679, 1591]. The electronic properties of buffer, 1LG, 2LG, 3LG have been characterized by means of ARPES [598]. Quasi-free standing graphene on SiC has been investi- gated [645, 647, 678]. This method can be applied also Figure IX.54. Micrograph of a single WS2 crystal. Areas with different contrast are labeled MLG, BLG and WS2, to other LMs, such as TMDs [1595], and their combi- respectively. Red arrows indicate the crystallographic nation in vertical LMHs [1217, 1596]. The possibility orientation of WS2 crystal. Adapted from [1217] with of studying the electronic properties of such a struc- permission of The Royal Society of Chemistry. ture allows for understanding how the layers interact with each other. highlighted by orange dots, corresponding to DFT The band structure of WS2/SLG can be retrieved by bands on the graphene single cell. Calculated gra- means of µ-ARPES on a single WS2 crystal. µ-ARPES phene σ-bands are not superimposed, since they are (or micro-spot ARPES) is a technique capable of meas- not detectable due to their low intensity. The bands uring the electronic properties of a material with a spa- visible in Γ belong to WS2, as also indicated by DFT. At tial resolution ~1 µm. The price for the higher spatial the points where the bands of graphene and WS2 cross resolution is paid in terms of energy resolution, which is (indicated by green arrows in the panel), no apparent slightly worse. Figure IX.54(a) show a LEEM image of a splitting or gap is observed. In order to confirm this, single triangular WS2 crystal on graphene on SiC [1217]. DFT calculations were carried out, the result of which Overlapping ARPES measurements with DFT is summarized in figure IX.55(b), where the bands minimizes the error in interpreting the measured ‘unfolded’ onto graphene’s BZ are displayed for bet- signal. Figure IX.55(a) reports the graphene π- and ter readability. π*-bands measured via µ-ARPES on the crystal When crystals with different lattice parameter shown in figure IX.54. The bands are well visible and are superimposed, they can give rise to a superlattice. 225 2D Mater. 7 (2020) 022001 C Backes et al Figure IX.55. Band structure of WS2/MLG. (a) µARPES measured on a single WS2 triangle with photons~70 eV. (b) DFT band structure evaluated on the WS2/graphene supercell depicted in panel (c) and unfolded into graphene’s BZ. (c) Coincidence supercell (7  ×  7) over (9  ×  9) of WS2/graphene. (d) Experimental ARPES CESs recorded with p-polarized photons ~ 27.5 eV (e) Experimental ARPES band structure of WS2/EG measured with He I light along the path indicated in the inset by the red line. DFT bands including spin–orbit effects are overlapped to the raw data. On the right: zoom-in of the region around K WS2 (green-dashed line in panel (e)). Both DFT and experimental bands fit are superimposed to the raw data. The red-dashed line is on the graphene’s π-bands. Adapted from [1217] with permission of The Royal Society of Chemistry. When multiples of the different unit cells match in values of the holes. The result along the ΓKM direc- size, they form a supercell which behaves as a new unit tion is displayed on the right side of figure IX.55(e), cell of a hybrid crystal. This new periodicity alters the representing the zoom-in of the region framed with a overall band structure, opening minigaps in high sym- green-dashed line in the panel. This gives mh1 ~ 0.39me metry points. The smallest coincidence lattice for the for the low energy band and mh2 ~ 0.53me for the high WS2/graphene system was found to be the (7  ×  7) WS2 energy band, confirming the asymmetry reported in unit cells (u.c.) on (9  ×  9) of graphene, figure IX.55(c). Ref. [1597]. The spin–orbit splitting of the WS2 bands DFT calculations were carried out on this supercell and in K was retrieved from integrated energy distribu- as a result, no mini-gap opening was predicted, unlike tion curves to be 462  ±  5 meV. This is ~10% larger what observed for MoS2 on graphene [1596]. To better than what was measured for 1L-WS2 on Au(1 1 1) and visualize the relation between the graphene π-bands Ag(1 1 1) [1598, 1599] and ~7% larger than the highest and the WS2 bands, constant energy surfaces (CESs) value reported so far [1599]. This is comparable with are shown in figure IX.55(d), extracted from the Fermi measurements on bulk WS2 [1594]. surface (FS) at binding energies indicated in the fig- ure. The CESs in this case are small volumes in k-space IX.3. Electrical characterization integrated over #250 meV, corresponding to the reso- IX.3.1. Four probe configuration lution of the instrument. The data were acquired with The basic electronic properties of a homogeneous photon energy 27.5 eV in order to maximize the inten- metallic diffusive material are captured in the primary sity of the WS2 bands with respect to graphene. parameters: resistivity ρ (Ω · cm), n (cm−3) and µ (cm2 ARPES measurements with He I radiation of V−1 s−1). For a LM, the resistivity is replaced by RS (Ω), 21.2 eV are presented in figure IX.55(e), together with i.e the resistance that would have a (homogeneous) the DFT-calculated bands including spin–orbit cou- sheet of material with equal length and width. For a pling. The image was obtained by scanning the BZ sheet of thickness t, RS  =  ρ/t. Similarly the surface along the red line traced within the green hexagon in charge density ns (cm−2) is ns  =  n t where σ is the the inset. In this image the high symmetry points are conductivity. σ/(ne)  =  1/(nseRS). It is usual to define for the WS2 BZ, whereas for panels (a) and (b) they µ  =  σ/(nse)  =  1/(nseRS). It is common in the GRM refer to graphene’s BZ. Single spectra were measured literature for RS to be written as ρxx. perpendicular to the red line. The experimental data Resistances are measured in a four-probe con- are fitted in the proximity (±~0.1 Å−1) of K with a figuration, where a current I is applied on 2 probes parabolic function in order to extract the effective mass (source and drain) and the voltage drop V measured 226 2D Mater. 7 (2020) 022001 C Backes et al Figure IX.56. (a) 4-probe configuration. (b) Hall bar. (c) Van der Pauw. (d) TLM configuration; (e) FET, showing a SLG with a top gate deposited on a dielectric, and source and drain contacts. (f) f coefficient of the Van der Pauw method. on two others following the schemes in figure IX.56. (RAB,CD + RBC,DA ) / (RAB,CD + RBC,DA ) This removes the contribution from contacts, i.e. = f arccos h [exp ((Ln2/f ) /2)] . that of the leads (wires), to the measurement appara-  (IX.15) tus, so that the measured resistance V/I is intrinsic to f is plotted in figure IX.56(f). The Van der Pauw the sample. In a linear configuration (figure IX.56(a) configuration allows measuring a sheet of arbitrary and Hall bar figure IX.56(b), RS  =  (VCD/IAB)·(W/L), shape, like a cross e.g. where W and L are the width of the sheet and dis- Standard techniques [1601] are used to measure tance between the voltage probes C and D, respec- the SLG resistance.This is usually in the range a few tively. In a Van der Paw configuration [1600] (figure tens Ω to a few 100 kΩ. In the low current range, low- IX.56(c)), contacts are attached to the edges of a sheet frequency (typically a few Hz to a few kHz) AC meas- and RS  =  π(RAB,CD  +  RBC,DA)/[2Ln(2)]. f (RAB,CD/RB urements with a lock-in amplifier, either in current or C,DA), where RAB,CD is the resistance measured inject- voltage source mode, provide the best SNR. I-V char- ing the current between A and B and the voltage probes acteristics are measured in DC mode, or by recording are C and D. RBC,DA is measured by permutation. The the current change (Iac) induced by modulating the f is a function of RAB,CD/RBC,DA only, and satisfies the voltage around a DC bias (V  =  Vdc  +  Vac). In all cases, ­condition [1600]: the current must be kept low enough not to heat the 227 2D Mater. 7 (2020) 022001 C Backes et al Figure IX.57. σ versus gate voltage at 20 K in UHV, for SLG and three different K doping concentrations. Lines are fits to equation (IX.16) and give µ, and the crossing of the lines defines the points of the residual conductivity and the gate voltage at minimum conductivity (σres, Vg,min) for each data set [1599]. sample, or to induce changes. At cryogenic T, typical as shown in figure IX.57. Here nimp is the charge currents range between nA to hundreds µA. High cur­ impurity concentration, K/nimp has the unit of rents (~mA per µm width) can be used to anneal SLG mobility and σres reflects the fact that SLG conducts to clean up impurities [1602, 1603] or to produce nar- even at zero carrier density [1610]. row constrictions tens of nm wide in suspended SLG In the previous examples ns is modulated with an [1604]. electrostatic gate. In principle this is a parallel plate Ultra-high frequency measurements (>100 GHz capacitor, with one plate being the gate and the other [1605–1607]) or high precision resistance measure- SLG. Either SLG is deposited on the dielectric or vice- ments for quantum Hall effect [612, 635] require spe- versa. In the former case, the gate below SLG (bottom cialized equipment. gate) often consists of conducting Si (the gate) covered The charge density is determined either from with a dielectric (SiO2) onto which SLG is deposited. the Hall effect, or from electrostatic field effect. The All devices fabricated on the substrate are subjected Hall voltage, VH, is measured perpendicular to both to the same field when a voltage VG is applied to the current and magnetic field. Typically Hall bars such Si gate. Top gates are realized by depositing a dielectric as in figure IX.56(b) are used. With the magnetic (Al2O3, HfO, SiN, h-BN, etc) on SLG, e.g. by ALD or field B perpendicular to SLG and current from A to thermal evaporation. Sputtering is not recommended B, VH  =  VCE  =  VDF. For SLG, nS  =  B/(eρxy), where since it damages SLG [1611]. This is followed by coat- ρxy  =  (VH/I). In the van de Pauw geometry, cur­ ing with a metal (Au, Al). Top gates are usually pat- rent leads and voltage probed are placed in cross terned and can be addressed individually to switch ­configuration, such that ρxy  =  VAC,BD  =  VBD,AC fol- one device at a time. The gate can also be brought close lowing figure IX.56(c). to SLG from the side in the same plane as SLG (side- In the simplest approximation, µ is deduced gate). If the gate itself is also made of SLG, this allows from resistance and charge density measurements: patterning the SLG device and the gate all at once µ  =  ρxy/(Rsq B). In the simplest picture of short [730, 1612]. range scatters, the Boltzmann conductivity σ is inde- In the top/bottom gate configuration, the simplest pendent of energy, i.e. independent of n [1608]. In assumption gives a charge variation Δns induced on this picture, the usual definition σ  =  nseµ, would SLG as: give µ with unphysically large values at charge neu- ∆ns = (ε0 ε/td ) Vg trality. In reality, σ varies with ns, being minimum (IX.17) at and around charge neutrality. This calls for other where ε and td are the dielectric constant and thickness scattering mechanisms, like charge impurity scatter- of the dielectric. In most FET configurations the SLG ing or resonant scattering [1608]. A better descrip- channel is provided with only 2-contacts (plus the gate, tion of the relation between σ and n was given in see figure IX.56(e)), therefore the contact resistance Refs. [1608, 1599]: is included in the measurements, which results in   underestimating µ. More precise methods have been (IX.16) σ = σres + Ke ns /nimp  developed [1613, 1614]. For sufficiently large bias 228 2D Mater. 7 (2020) 022001 C Backes et al Figure IX.58. Shubnikov–de Haas oscillations of the magnetoresistance in (a) and (b) SLG on SiO2 [1617] and (c), (d) MLG on 4H-SiC(0 0 0 −1) (C-face)[1620] for different T from 4 to 70 K. (b) The minima in ρxx are plotted as a function of 1/B (so called fan plots) to determine the charge density at a particular gate voltage. (d) Same analysis for the data in (c), but plotting the maxima in ρxx. voltage, the n may vary along the SLG channel, an even Standard electrical measurements require con- change sign, due to the SLG ambipolar nature, leading tacts on SLG. Besides the fundamental interest of how to a complex interplay between VSDand VG (top, to inject from a 3d to a 2d material, from planar (2d) bottom or both) [1615]. Equation (IX.12) does not or edge (1d) contacts, while matching energy and take into account the variation of chemical potential momentum (at charge neutrality the SLG Fermi sur- with n, which introduces an extra capacitance term face is reduced to points in k-space), it is technically e2D(EF) [1616], the so-called quantum capacitance difficult to realize contacts to SLG of low (in the 10 s per unit area, with D(EF) the density of states, which Ω) resistance. Because there are no dangling bonds cannot be neglected close to the Dirac point [1616]. to adhere to, metals like Au tend to bead up on SLG ns can also be determined from magnetoresistance [1623], which requires an adhesion layer, like Pd, Ti or measurements in high (generally several Tesla) magnetic Cr. Low resistance contacts (50–100 Ω µm) are a com- fields, and at low (cryogenic) T by analyzing the periodic bination of mechanical adhesion and charge transfer oscillations of the magnetoresistance (Shubnikov–de to SLG, due to a large metal-SLG work function differ- Haas oscillations) as a function of 1/B [107, 622, 1617]. ence (to increase the SLG conductivity under the con- Examples are given in Figure IX.58 for SLG on SiO2 and tact by heavily doping it) [1624–1626]. Metal induced MLG on 4H-SiC(0 0 0   −  1). The 1/B oscillation period doping can create p-n junctions from the SLG under is BF  =  (h/4πe)kF2 with kF2  =  nsπ. With increasing field, the contact to the channel, which have to be taken into the oscillations develop into the quantum Hall effect, account in the resistance analysis [1625, 1626]. and the quantification of Hall resistivity also give access Contacts are usually defined by lift-off after devel- to n [107, 1617]. These measurement techniques can be opment of the lithographic resist. Resists commonly useful in case of mixed (electron and hole) doping in the used are positive e-beam resist (e.g. 200 nm PMMA vicinity of the Dirac point [1618, 1619]. spun onto 200 nm MMA), or photoresist (e.g. S1813 In semiconductors a current saturation regime ~ 1.8 µm thick). Good contacts are obtained with is reached by applying a sufficiently large VSD, and µ 5–10 nm Cr, Pd, Ni or Ti (or combination of these met- can be deduced from the current saturation value als, Ti/Pd 2 nm/10 nm) evaporated on SLG, followed [1621]. However, this regime is difficult to reach in by 30–50 nm Au, by e-beam or thermal evaporation. SLG because of the requirement of extremely high VSD, The contact resistance, Rc, can be measured by the which will cause inhomogeneity in n along the channel transfer length method (TLM), see figure IX.56(d). [1615, 1622]. A SLG strip is provided with a series of contacts. The 229 2D Mater. 7 (2020) 022001 C Backes et al 2-point resistance RTLM  =  Rc(W)  +  2Rs L/W is meas- spectroscopy (THz-TDS)[1635–1639, 1641–1643]. ured and plotted as a function of distance L between This method has been thoroughly benchmarked the contacts, with W the channel width. The advantage against van der Pauw electrical measurements with is that this apporach fast and involves only resistance fixed electrodes [1638, 1643] as well as scanning micro measurements. However this assumes that contacts four-point probes [1636, 1637, 1795]. the same, and Rs constant along the graphene channel THz-TDS is based on the fact that the absorption (i.e, homogeneous scattering, homogeneous gate or in the THz range is highly dependent on the complex- bias induced charge density). This may be question- valued and frequency-dependent electrical conduc- able very close to the contacts, due to metal-induced tivity [1644]. This is because the absorption of light doping. Alternatively, Rc can be derived from Kelvin in the THz frequency range is dominated by low- probe microscopy, by measuring locally with an AFM energy intra-band transitions, as opposed to the inter- tip the voltage drop from contacts to channel [1627]. band transitions that give rise to graphene’s constant The lowest intrinsic Rc measured are in the range absorption of light in the visible spectrum [1645]. ρc  =  Rc · W  =  50–100 Ω µm [1607]. Ref. [1628] devel- The electrical field attenuation caused by the oped edge contacts, where lateral contacts inject cur­ charge carriers in the thin film can be established by rent through the SLG edges. RC are of the same order of dividing the electrical field, Ẽgraphene , through the sam- magnitude [1628]. ple area, with a reference, Ẽref , from an area without the Deposition of gate dielectrics is a challenge for the conducting film, which is then related to the conduc- same reason that metals do not adhere on SLG. ALD or tivity through [1641]: evaporation of high K dielectrics (Al2O3, HfO2, SiO2) [1629] or Si3N4 [1630] have best RF performances (in Ẽgraphene (ω) 1 + nsub (IX.18) = terms of cut-off frequency) [585], [1631], and hBN Ẽref (ω) 1 + nsub + Z0 σ̃ (ω) encapsulation provides protection [1184, 1632]. SLG was the first measured LM that was exposed, where nsub is the refractive index of the substrate, Z0 is i.e. not buried at the interface between two semi- the free-space impedance, and σ̃(ω) is the frequency- conductors. While this presents great opportunities dependent (complex-valued) Drude conductivity, (spectroscopy in particular), exposure to the ambi- σdc (IX.19) σ̃(ω) = . ent degrades the SLG pristine properties, by doping 1 − iωτ and scattering by impurities. The transport properties The THz radiation is focused into a 300 µm spot, drift with a time constant ~1 h or so, and are recovered limited by the long wavelength in the THz frequency after annealing (e.g. ~1 h at 420 K in He) [1633]. SLG range, which enables the average sheet conductivity protection is therefore required (top layer, or measure- (or resistivity) of a graphene film to be determined. ment in neutral or vacuum environment), unless the In contrast, electrical measurements probe the exposure provides useful counter doping (like for the device conductance (or resistance), which can differ measurement of the quantum Hall effect [625]). significantly from the average conductivity for the SLG is a flexible membrane, therefore the rough- same device area, in case of spatially non-uniform ness of the substrate matters. Sharp ripples or bubbles conductivity [1637, 1642]. introduce field gauge effects [450, 632], equivalent Figure IX.59(a) is a conceptual schematic of a THz to high magnetic field of up to several hundred Tesla absorption setup, with a femtosecond laser pulse hit- [632], with consequences on transport [450]. ting a photoconductive antenna (PCA). The result- It is relevant for graphene nano/microstructures of ing THz pulse is focused by a lens, and transmitted size L  ×  W to keep in mind that the energy scale is set through a thin conductive film. Finally, after refocus- » by E = kvF = vF nΠ mΠ L + W , so that the energy sep- ing by another lens, the attenuated THz pulse is picked aration between modes in charge neutral SLG is ~20 K up by a PCA operating as a detector. Figure IX.59(b) for L  =  1 µm. Therefore, submicrometer structures of is a photo of a Si wafer covered with graphene, which SLG at low T are effectively quantum dots [620]. is raster scanned across the focal plane of the beam 1d ballistic transport was measured in GRNs line, while acquiring THz-spectra at a rate ~1 Hz. In at RT [620]. In this case the nature of the contacts figure IX.59(c) the time-resolved electrical field is (invasive or not) determines the voltage. For invasive shown. The initial peak is followed by smaller echo contacts the electron flow is interrupted by the volt­ peaks, spaced by the round trip time of flight for age probe which then acts like a current injector, and internal reflections inside the substrate. Using equa- a resistance close to the quantum of conductance tion (IX.18) the real (green triangles) and imaginary Rq  =  (e2/h)−1  =  25.812 kΩ is measured independent (green circles) parts of the conductivity can be plot- of distance between voltage probes, even in a 4-probe ted against frequency (figure IX.59(d)), and fitting the configuration [620, 1634]. Drude conductivity (equation (IX.19)) yields the low frequency (DC) limit of the conductivity, σDC , and the IX.3.2. THz-TDS mapping of electrical properties momentum relaxation time (elastic scattering time), Rs, µ and n can also be measured quickly, accurately τ [1641]. Under the assumption of diffusive trans- and non-destructively using terahertz time-domain port, the semiclassical Boltzmann transport equation, 230 2D Mater. 7 (2020) 022001 C Backes et al Figure IX.59. (a) Schematic of THz-TDS setup. (b) Photo of equivalent THz beam line, with a graphene-coated Si wafer held by a motorized raster scanner. (c) THz direct pulse and weaker echoes from multiple internal reflections. The black curve is from an uncoated reference area, while the red curve is attenuated by the conducting sample. (d) Conductivity spectrum of graphene grown on single crystal, which closely follows the Drude model (equation (IX.18)). The fitting parameters (σDC , τ ) can be used to calculate n and µ. For large-area conductivity mapping, commercial spectrometers that cover up to 2–3 THz are used, while a more sophisticated air-THz setup allows spectra up to 15–20 THz to be recorded [1637]. Figure IX.60. Examples of THz-TDS electrical characterization of graphene. Conductivity of graphene coated Si wafer mapped before (a) and after (c) fabrication of 49 electrical devices to test how individual process steps affect the conductivity [1641]. (b) wafer after laser-etching and physical mask deposition of metal contacts [ 1796]. Histogram (inset) of the sheet conductivity across the wafer before and after fabrication. (d) and (e) µ and n for a Si wafer determined by equations (IX.20), with histograms below showing scale and distribution of n and µ [1638]. (f) Investigation of how CVD growth T affect large-scale uniformity of sheet conductivity of graphene transferred to 4″ Si wafers [1641]. 231 2D Mater. 7 (2020) 022001 C Backes et al Figure IX.61. (a) Terahertz resistivity maps (0.9–1.0 THz) of two A4 sized sheets of doped graphene on PET with (a) high and (c) lower uniformity (color scale is from 200 Ω to 500 Ω. The histograms corresponding to the white dashed rectangles are shown in (b), where sheet 1 shows a very narrow peak, while sheet 2 exhibits a bimodal distribution. (d) THz conductivity map of a 10 × 10 mm2 device, by THz and van der Pauw measurements. After Drude-fitting, the DC Rs is ~267 ± 12Ω for the shown device, as for the histogram. The measurements for 7 such devices agree well, except device 4, which had a visible scratch. With permission from [1643], The Optical Society. σ = (2e2 /h)kF vF τ in combination with the funda- where e is the electron charge, vF is the Fermi velocity, mental relation σ = neµ yields expressions for n and and  is the reduced Planck constant. Therefore, µ that can be calculated from the fitting parameters the average µ and n can be spatially mapped with a (σDC , τ ): resolution approximately equal to the THz spotsize dTHz = 300 µm , by recording a full spectrum in every e3 vF2 τ 2 π2 σDC 2 pixel, and computing equation (IX.20) with the Drude (IX.20) µ= , n = , fitting parameters. π2 σDC e4 vF2 τ 2 232 2D Mater. 7 (2020) 022001 C Backes et al Figure IX.62. Comparison of (a) transmission and (b) reflection mode measurements on the same four samples, with comparable spatial conductivity pattern and (c) overlapping statistical distributions of measured conductivity (d) Direct comparison of van der Pauw four terminal measurements with reflection (diamonds) and transmission (crosses) measurements, showing good agreement except sample 1, which was less homogeneous compared to samples 2–4 [1642]. Figures IX.60(a)–(c) shows the conductivity map and (c), typically within 5%–10% of the DC Rs, as can of the same CVD grown graphene film before and after also be seen by noting the small drop in σ(ω) from 0 to definition of 49 test devices, verifying that the laser 1.0 THz in figure IX.59(d). ablation/shadow mask device fabrication method The histogram in figure IX.61(b) shows a very nar- has little impact on the device characteristics, due to row distribution for sheet 1, and a broader, bimodal the complete absence of solvents and resists [1640, peak for sheet 2, consistent with the difference in the 1642]. Figures IX.60(d) and (e) show the µ and n visual appearance of the THz resistivity maps (a and c). maps obtained by the methodology described above, The histogram of DC Rs after fitting the Drude equa- with the histograms below showing both the scale and tion, equation (IX.19), to every pixel, shows a nar- distributions of values within a rectangular zone indi- row distribution across the ~750 cm2 graphene area cated with dark dashed lines. Figure IX.60(f) shows (dashed rectangle). Figure IX.61(d) is a THz map of how the method can be used to map the large-scale a cm-sized device with the corresponding histogram changes in uniformity of electrical key parameters of DC Rs. Rs by van der Pauw and THz mapping show with growth T in a CVD process, indicating that higher excellent agreement, however, with larger statistical T tends to give more uniform growth and larger usable spread for the THz because of the avoidance of physi- areas of graphene [1641]. cal contact, and the greater number of measurement The THz mapping technique can be used for points. One of the devices (device 4) had a scratch, graphene on flexible substrates [1641, 1643]. Gra- which typically offsets electrical measurements in phene grown on Cu, transferred to PET and doped unpredictable ways, while having little impact on the by AuCl3 showed very high uniformity across 15 inch THz-measurements [1641]. (25  ×  30 cm2) PET sheets. The THz-resistivity in the THz-TDS electrical characterization can be done frequency range (0.9–1.0 THz) is in figures IX.61(a) in transmission mode, the most used in literature 233 2D Mater. 7 (2020) 022001 C Backes et al Figure IX.63. (a) Schematic of suspended SLG over hole for AFM nanoindentation. (b) SEM images of suspended SLG over holes. The border of the SLG-covered area is indicated by a dashed line. Wrinkles often present in the transferred SLG can be seen. (c) Force-displacement curve of SLG in AFM nanoindentation. The red line is a fit to equation (IX.22). Inset:AFM topography images of suspended SLG before and after fracture. Scale bars in (b) 2 µm and (c) 1 µm Adapted from [1653], with permission from AAAS. [1641], or in reflection mode [1798]. Comparison IX.4. Mechanical characterization of THz conductivity maps recorded in transmis- GRMs compared to existing materials, combine high sion (see figure IX.62(a)) and with a commercial stiffness, strength, strain-to-failure and high flexibility reflection mode THz-system (DasNano, see fig- [1646]. SLG compared to existing membranes has ure IX.62(b)) showed negligible difference [1642], high ductility, as can be stretched up to 30%, however. and good agreement with electrical measurements, Beyond that point it fractures in a brittle manner for the same samples. Despite the large CVD graphene like glass [1647]. However, up to date, experimental samples not being encapsulated, the measurements evidence under uniaxial tension that confirms agreed well within the statistical errors, as clear from the above is missing, as these properties have been the ­conductivity histograms in figure IX.62(c) and predicted via theoretical or modelling [1647, 1648]. the direct comparison between electrical four point The main reason for the lack of experimental data is probe sheet conductance with THz sheet conductivity associated with the sample quality (and size) and the measurements for both transmission and reflection damage inflicted by handling and subsequent transfer mode systems in figure IX.62(d). onto the loading devices. E.g. MC of graphite produces For transmission mode THz-TDS it is important flakes of at best 100  ×  100 µm2 depending on substrate that the substrate has high resistivity, to reduce THz interaction [1007]. Suspended SLG samples over absorption. The highly doped Si commonly used as a relatively large areas (~ 1 mm2) as required for classical backgate, can be replaced by a custom-made substrate tensile experiments are difficult to produce [1007]. Wet with a THz-transparent gate [1639] to enable THz- methods for transferring CVD graphene from a metal TDS and electrostatic gating on the same samples. The substrate to arbitrary target substrates lead to samples advantage of reflection mode is that there is no need with defects such as wrinkles, folds and tears, as well for using high resistive substrates, while transmission as residues, voids and polymer contaminants that can mode can measure buried conductive layers at any induce uncertainties in the measurements. Normally, depth. optimization work has to be done in order to minimize THz-TDS is a fast, clean, and accurate way of meas- the influence of defects and select the gauge area of the uring the transport characteristics of graphene on membrane to be tested. Due to limitations posed by the many substrates. Apart from Si and PET substrates, the accuracy in force sensing in the nN range and sample method was used for graphene on SiC [1643, 1797] and gripping techniques, there is a lack of specialized graphene on Ge (on Si). As clear from figure IX.59(b) instrumentation for in plane tensile testing of GRMs. there is no limit for the size of samples, as a THz-TDS This has triggered intense research for applying either measurement head could easily be integrated in a roll- AFM based, or thin film metrology techniques [1646, to-roll production line, giving instant feedback on the 1649]. A comparison of the measured mechanical quality of graphene. The technique can also measure properties with bulk materials is still an open question. buried and encapsulated conducting layers, which Measuring techniques can be categorized as direct provides essential information on the impact of inter- and indirect. In direct methods the force or strain is mediate process steps on the electrical characteristics applied directly to the GRM membrane. In indirect before device fabrication. methods, Raman spectroscopy can be used for assess- 234 2D Mater. 7 (2020) 022001 C Backes et al ing the mechanical response of GRMs supported or Table IX.3. Indicative values of Young’s modulus and breaking strength for 1L-GRMs measured by AFM nanoidentation. embedded into polymers, where the strain is applied by shear forces from the deformed substrate. Other E2D max σ2D Strain-at- methodologies for applying strain are budge test and 1L-GRM (N m−1) (N m−1) break (%) electrostatic actuation. Graphene (MC) [1652] 340 (50) 42 (4) 25 Graphene (CVD) 328 (15) 39.5 — IX.4.1. AFM nanoindentation MoS2 (MC) 130 16.5 — The elastic properties of suspended SLG can be MoS2 (CVD) [1657] 123 — — measured by nanoindentation using an AFM. The WS2 (CVD) [1657] 137 — — samples are made either by MC of graphite or by BN [1087] 220–510 8.8 — transferring CVD grown SLG onto pre-patterned substrates with holes or trenches. Usually an AFM tip applies a point load at the center of GRM sheets constant [1649, 1654]. The microindentation experi- suspended over a hole on a pre-patterned substrate. ment is in essence biaxial. The conversion of biaxial Because of the high strength of SLG, higher than steel deflection data to uniaxial stress–strain curves by and Kevlar [1650], cantilevers with diamond tips assuming zero bending stiffness is problematic, and are preferable for these experiments. For a doubly not fully representative of true uniaxial tensile loading clamped SLG ribbon, a wedge indentation tip was used [1652]. to introduce a uniform tensile strain [1651]. During AFM indentation experiments were conducted on nanoindentation the force, F, versus deformation, δ, graphene on SiC [1014] and CVD grown SLG of differ- of the suspended layer (at the point where the load ent grain size [1653]. Ref. [1653] showed that the elas- is applied, figure IX.63(a) and (b)) is continuously tic stiffness of CVD-SLG is similar to that of MC SLG, recorded up to fracture. By employing third order provided that the sample is free of pleats and ripples. elasticity theory and a number of assumptions based The mechanical properties of 1L-MoS2 [1655, on classical membrane mechanics, the conversion of 1656] and h-BN [1087] were measured by AFM nanoindentation force-deflection measurements into nanoindentation. Similar modelling approaches were an axial stress–strain curve for a free standing SLG pursued for the interpretation of the force-displace- membrane was performed (figure IX.63(c)) [1652]. ment traces. In table IX.3 the elastic properties of By this procedure, the 2d Young’s modulus, E2D, various GRMs measured by AFM nanoindentation are max summarized. breaking stress, σ2d , and strain at break, εmax were estimated [1652]. In order to obtain the corresponding bulk parameters, these quantities were divided by the IX.4.2. Bulge testing interlayer spacing in graphite h  =  0.335 nm (assumed The elastic properties of GRMs can be assessed as SLG thickness) yielding values of Young’s modulus by bulge [1658] or pressurized blister test [1659], ~1 TPa and of tensile strength of ~140 GPa [1652]. commonly applied for thin film testing. The advantage To convert the experimental nanoindentation of the bulge test is the ability to measure not only E but data to uniaxial stress–strain curves, GRMs are nor- also ν, whereas, in the case of AFM nanoidentation, ν mally modelled as a membrane (thin plate) of negli- should be known in advance (see equation (IX.22)) gible bending stiffness and nonlinear elastic properties [1659]. Another advantage is that the stress [1652]. For doubly clamped beam shaped membranes concentration caused by the AFM tip is avoided by (GNRs) [1649, 1651] the force-displacement behav- the application of a uniform pressure. In Ref. [1660] iour can be expressed as [1649]: suspended MC SLG over microcavities (diameter ~ 5 ï  t 3 ò Å ã µm) etched in a SiO2 substrate (figure IX.64(a)). The 12.32 8w δ 3 F = 30.78w E2D + σ0 δ + tE2D . SLG membrane remains flat due to the balance of l l 3 l  internal (atmospheric) and external pressures (figure (IX.21) For circular shaped membranes the corresponding IX.64(b)). It is adhered to the substrate by van der relation is as follows [1652, 1653]: Waals forces and can confine gas molecules (e.g. N2). ï ò Å ã3 The SLG membrane inflates forming a blister, when a 4πa  t 3 δ pressure difference across its surface is applied (figure F= E 2D + πσ 0 δ + tE2D q . 3 (1 − ν 2 ) α α IX.64(c)). AFM was used to define the SLG membrane  (IX.22) shape and measure its maximum deflection δ at the In equations (IX.21) and (IX.22), σ0 is the pre- center of the blister and its radius α (figure IX.64(d)). tension in the membrane due to the preparation pro- The deflection can be correlated with the pressure cedure, t is the thickness and ν is the Poisson’s ratio difference, ∆p, across the SLG membrane (figure (for SLG this is taken as 0.165, the Poisson’s ratio for IX.64(e)) using Hencky’s solution [1661, 1662], graphite in the basal plane [1652, 1654]). In equa- which gives the deformation of the membrane for tion (IX.21), w , l are the width and length of the GNR, the geometrically nonlinear response of a clamped while in equation (IX.22) a is the hole radius and circular elastic membrane subjected to a pressure 3 q = a/(1.05 − 0.15ν − 0.16ν 2 ) is a dimensionless difference, ∆p, provided that the bending stiffness 235 2D Mater. 7 (2020) 022001 C Backes et al Figure IX.64. (a) Optical images showing flakes with regions of 2–5 L (top) and 1 and 3 L (bottom). Schematic illustration of a SLG-sealed microcavity (b) at rest where Δp   =  0, the pressure inside the microcavity is equal to the external pressure p ext, and (c) at higher Δp , (d) an AFM image showing the deformed shape of a SLG membrane with Δp   =  1.25 MPa and (e) deflection versus position for various levels Δp (cyan). The dashed black line is the shape obtained from Hencky’s solution for Δp   =  0.41 MPa. The deflection is measured by an AFM along a line that passes through the centre of the membrane. Adapted from [1660], © Springer Nature. Figure IX.65. (a) K(y )δ3/α4 versus Δp for SLG membranes before delamination (black symbols) and after delamination (magenta), (b) nEt versus number of layers n. Closed shapes are for the fitted lines; open shapes are for nEt, with Et  =  E2D  =  347 N m−1. of the membrane is negligible. Therefore, ∆p can be band gap engineering [1663]. The strong adhe- written as [1660]: sion, comparable to solid–liquid adhesion [1660], between 1L-GRMs and an atomically flat substrate ∆p = K (ν) Eδ 3 /α4 (IX.23) (e.g. SLG/SLG, SLG/hBN, MoS2/MoS2) results in where K (ν) is a constant that depends on ν. atomically clean interfaces since any contaminants By plotting K (ν) δ 3 /α4 versus ∆p, E can be (water and hydrocarbons) are repelled by the van der estimated as the slope of the linear dependence. In Waals forces between the layers forming microbub- ­figure IX.65(a) the experimental data for K(ν) δ3/α4 bles [1664]. These bubbles have constant volume, and versus Δp are shown for SLG. A linear fit to equa- their shape is determined by the competition between tion (IX.23) gives E2D  =  347 N m−1, in good agree- van der Waals adhesion of the crystal to the substrate ment with the AFM nanoindentatation measurements and the elastic energy needed to deform it [1665]. In [1008, 1653]. Ref. [1660] measured E of up to 5LG. The ­figure IX.66(a), typical force-displacement curves results are summarized in figure IX.65(b) and show (solid lines) recorded from bubbles of different sizes that E2D is quantized to the SLG E2D (347 N m−1). These are presented. The symbols correspond to numerical findings are based on a continuum model in which SLG fits by taking into account the changes in area of con- is considered as a membrane with negligible bending tact and the increase in bubble’s radius [1665]. The stiffness and the crystal is of high structural quality i.e. data in figure IX.66(b) correspond to SLG bubbles free of ripples, folds and defects. for which the fitting curve to the experimental data is Bulge testing has also been used for 1L and ML- drawn by assuming a SLG axial stiffness ~420  ±  20 N MoS2 as a way to apply biaxial strains up to 7% for m−1 [1665]. Similarly, the stiffness of 1L-MoS2 was 236 2D Mater. 7 (2020) 022001 C Backes et al Figure IX.66. (a) Experimental force-deflection curves (solid lines) and numerical fits (symbols) for two SLG and 1L-MoS2 bubbles of different sizes. (b) A force-deflection curve (solid curve) and corresponding numerical fits for three values of Y  =  E2D. Table IX.4. Fracture toughness of 1LG, 2LG, MLG ML-hBN. under uniaxial tension on MC BLG and measured its Fracture toughness Mechanical fracture toughness. In table IX.4 the measured fracture √ toughness for various SLG and MLG and ML-hBN are Sample MPa m testing/device summarized. CVD SLG [1667] 4.00 Tensile/MEMS GRMs are modelled as membrane-like materials device (SEM) i.e. with approximately zero bending stiffness [1672]. BLG [1671] 21.25 Tensile/MEMS Therefore, the effect of bending modulus is usually device (SEM) ignored [1652, 1653]. Refs. [1649, 1655] explored the MLG [1428] 12.00 Tensile/MEMS effect of bending modulus as being proportional to the device (TEM) third power of the crystal thickness (see first term in ML-hBN [1428] 5.50 Tensile/MEMS equation (IX.21)) in AFM nanoindentation of MoS2 device (TEM) crystals with various thicknesses. The force-displace- ment curves for the FL crystals were strongly nonlin- ear, while for number of layers N  >  10 were linear. This estimated as 210  ±  20 N m−1. Both values are some- is important when considering membrane-to-plate what higher than those given in table IX.3. like transition for N  >  10. The determination of bend- IX.4.3. Fracture toughness and bending stiffness ing stiffness of either 1L or FL-GRMs has been mainly Large-scale applications of GRMs, beyond microscale, based on various types of theoretical calcul­ ation require the knowledge of fracture toughness, rather [1673]. Ab initio or first-principles quantum mechani- than the intrinsic strength, which corresponds to cal calculations and empirical potential calcul­ations a uniform breaking of atomic bonds in a perfect were used to determine the bending modulus of GRM crystal. Fracture toughness is a property that SLG [1646, 1673], with higher values (mean value ~ describes the ability of a material containing a crack 1.51 eV) in comparison with those derived by empiri- or a defect to resist fracture [1666]. The absence of cal potentials (mean value ~ 1.12 eV). These models reliable tensile testing devices with force sensing at treat SLG as an isotropic body. To introduce anisotropy the pN or nN range and the cumbersome fabrication/ and to incorporate bending effects, a finite elasticity transferring procedure has hindered progress. Thus, model was proposed [1674]. This provides the means the direct determination of fracture toughness to introduce (and handle) additional degrees of free- has been challenging for GRMs. Fracture data for dom through dependence on the shift vector which CVD SLG were measured using in situ MEMS- separates the two Bravais sublattices of SLG [1674]. based uniaxial tension devices in a Scanning [1667], Continuum mechanics was applied to describe the Transmission [1428] electron microscope and AFM SLG flexural deformation, leading to a nonlinear von- nanoindentation [1668, 1669]. Ref. [1667] conducted Karman plate theory with two elastic bending moduli in situ tensile testing of centre-cracked CVD SLG in [1646, 1674, 1675]. However, from the experimental a SEM and obtained fracture toughness based on the point of view, Ref. [1676] criticized the use of the con- classical Griffith’s fracture theory [1670]. Ref. [1667] tinuum mechanics shell model. By exploring the nm conducted in situ fracture toughness testing in a scale rippling of SLG, Ref. [1676] claimed that phe- HRTEM on MLG having V/U-shaped single-edge nomenology fails to predict correctly E2D, as well as the notches. For comparative purposes fracture toughness SLG thickness of 0.335 nm, since classical continuum testing on ML-h BN samples was also conducted. Ref. plate theory assumes that bending always induces in- [1671] performed fracture mechanics experiments plane stretching. 237 2D Mater. 7 (2020) 022001 C Backes et al Figure IX.67. Bending stiffness of SLG by (a) applying controlled forces using an IR laser and (b) tracking the motion of a rotated device under thermal fluctuations. (c) Stacked histogram of bending stiffness measured by the methods in (a) and (b). The red arrow points to the bending stiffness ~1.2 eV calculated theoretically. Scale bars are 10 mm. Adapted from [1680], © Springer Nature. A few attempts have been made to measure exper­ lever and the bending rigidity was measured (figure imentally the bending stiffness of FLG and MoS2 and IX.67(a)). Using the thermal fluctuations of the SLG confirmed the dependence on the third power of cantilevers and based on the equipartition theorem thickness in accordance with the membrane theory of classical statistical mechanics, the cantilever spring of shells with pure bending [1673]. Ref. [1677] per- constant was also measured independently (figure formed AFM nanoindentation in FLG membranes IX.66(b)). As shown in figure IX.67(c) both methods onto a circular hole, while Ref. [1678], fabricated and yield a high bending stiffness for SLG, ~103 to 104 eV. measured the bending stiffness of convex-buckled sus- These values are orders of magnitude higher than pended BLG ribbons. By exploiting the abrupt switch- 1.2 eV, predicted from simulations [1673] and meas- ing from convex to concave geometry of the ribbons, urements of the phonon modes in graphite [1681]. under electrostatic actuation, they measured a bend- This high value is attributed to both thermal and static ing rigidity ~35.5 eV. Ref. [1679] proposed a different rippling stiffening ultrathin crystalline membranes in experimental approach to measure the bending stiff- a manner similar to how a flat sheet of paper becomes ness of 1L-GRMs. The method is based on atomic more rigid when crumpled [1680]. mapping of the crystal lattice within a fold by using HRTEM for SLG and scanning TEM imaging for MoS2 IX.4.4. Raman spectroscopy and WSe2. They confirmed the applicability of the lin- Information on GRM mechanical properties may ear elastic shell model and derived the bending rigidity. be obtained by Raman spectroscopy combined with They found that bending rigidity for TMDs is in the various types of external loadings (uniaxial, biaxial, range ~10–16 eV, five to six times that of SLG, in agree- hydrostatic, etc) [1529, 1682–1685]. Due to the ment with theory [1679]. anharmonicity of the interatomic potentials phonon Ref. [1680] measured the bending modulus of frequencies soften upon tension and harden under CVD SLG. They produced cantilever structures compression [1529]. As a result, Raman peak positions (length, 10–100 µm and width, 10 µm) and measured red (blue) shifted under tensile (compressive) loading. their spring constants by assuming them to be related The magnitude of shift depends on the interatomic to the bending rigidity. By using the photon pressure force constants and the mode eigenvectors relative to from an infrared laser, the spring constant of the canti­ the directions of the strain axis [1529]. However, in 238 2D Mater. 7 (2020) 022001 C Backes et al Figure IX.68. Pos(2D) as a function of uniaxial compressive strain for SLG flakes of various aspect ratios. Adapted from [1703] © Springer Nature. GRM many other effects can shift peaks, e.g. doping of the strain variation of the 2D peak depends on the [1518], hence caution needs to be taken when using shifting of the Dirac cones and the anisotropic phonon Raman spectroscopy to derive these information. softening which, in turn, cause alteration in the scat- Several groups investigated the effect of uniaxial terings paths participating in the double resonance strain in tension or compression by bending flexible mechanism for the 2D band [1534, 1696–1697]. substrates (e.g. PDMS) or plastic beams (e.g. PMMA) Under compression, by monitoring the Pos(2D) on which GRMs are deposited without slippage [1529, shift of rectangular SLG embedded into polymer flakes 1536, 1546, 1686–1693, 1723]. The samples can be sup- of various sizes the critical strain-to-failure was deter- ported or embedded within a thin polymer film. In mined [1692]. These were found to be independent of these experiments the application of uniaxial stress/ flake size at a mean value ~  −0.60% corre­sponding strain leads to the development of interfacial shear to a yield stress up to ~  −6 GPa [1691]. CVD SLG between the GRM membrane and the underlying or can be considerably wrinkled [1698]. Upon uniaxial surrounding polymer, transmitted into a normal stress deformation the Raman response for the 2D peak [1694]. Care should be taken for the examined area of was  <25% of that of flat exfoliated flakes [1698]. In all the flake to be free of residual stresses [1694]. Pos(G) cases in which SLG is either supported on or embed- is an indicator that the applied strain is transferred to ded in a polymer matrix, parameters such as interfacial the interatomic bonds, provided one can ensure that shear strength, onset of interfacial sliding and critical other effects, also leading to peak shifts, are not pre- compressive strain-to-failure may be extracted via sent. Under uniaxial strain the doubly degenerate E2g in situ Raman spectroscopy under tensile or compres- optical mode splits in two components, one polar- sive loading [1699, 1700]. ized along the strain and the other perpendicular to it. Raman spectroscopy was also used to investigate This leads to the splitting of the G peak into two bands, SLG bubbles as a function of strain [1531, 1658, 1701]. which are usually called G+ and G− [1529]. Both peaks These are formed by the pressure difference on both red shift with increasing tensile strain, and their split- sides of SLG during the deposition of flakes over aper- ting increases, in agreement with first-principles tures of various size and shape. SLG membranes can sup- calculations. Their relative intensities are found to port pressures up to 14 bar, corresponding to reversible depend on light polarization, which provides a useful strains up to 2% [1701]. The advantage of intentionally tool to probe the graphene crystallographic orienta- pressurized membranes is to avoid the influence of the tion with respect to the strain. A detailed explanation substrate, excluding complications from SLG-substrate of these effects is also given in section IX.2/Raman and interactions. Biaxial strain enables the determination figure IX.41. of the Grüneisen parameters. Ref. [1702] determined In tension, the shift of the main Raman modes is the SLG E2D by comparing the strain induced on pres- linear up to 1.5% [1695]. The 2D peak position down- surized balloons and the response of the G peak, with shifts with uniaxial strain at a much higher rate of numerical simulations. The estimated E2D as 2.4 and 2.0 ~  −  60 cm−1/% [1529, 1688]. An analogous but not TPa, is much larger than previous values [1652, 1660]. as pronounced splitting to two distinct components This discrepancy was attributed to the small attainable 2D− and 2D+ was observed [1534, 1536] depending on strain (~0.2%), since E may depend on the strain range laser wavelength. The strain-induced splitting and the being larger in small strain ranges. red shift depend on the strain axis direction relative to In Ref. [1689] an effective bending stiffness of SLG the crystallographic axis. The underlying mechanism embedded into a polymer cantilever was estimated. 239 2D Mater. 7 (2020) 022001 C Backes et al Figure IX.69. (a) Schematic of blister geometry. The optical path of the Raman signal and excitation is also depicted. SLG suspended over the hole (Inset), (b) enhancement factor for G and 2D. (c) I(G) and I(2D) versus distance from the centre of the hole with SLG at rest (0 kPa) and at a higher pressure difference. (d) Third power of SLG blister at its center versus pressure difference. From the slope of the linear fit, E is determined. Flakes of various aspect ratios (length to width) were for a SLG-air-Si stack (figure IX.69(a)) that takes embedded on the surface of a PMMA bar covered by into account the variation of I(2D) and I(G) (figure a 200 nm PMMA layer. SLG was modelled as a thin IX.69(b)). Both oscillate between a minimum and plate and subjected to uniaxial compression [1689]. a maximum which depend on the height of the blis- SLG was buckled at a critical strain determined by ter at various pressure differences (figure IX.69(c)) the response of the 2D peak (figure IX.68). Pos(2D) [1704]. The enhancement factor of I(D) and I(G) was relaxed after an abrupt uptake, while the strain at the correlated by a multireflection model [1705] with htot, onset of the Pos(2D) relaxation defines the critical the total distance between SLG and the underlying Si strain for each flake, see figure IX.68. For flakes with (figure IX.69(a) and (c)), and was used for the metrol- lengths smaller than the critical length (twice the dis- ogy of the blister height. Following Hencky’s model tance it needs for strain to increase to the externally (equation (IX.23)), the third power of the height at applied value) the externally applied strain is not fully the centre point of the blister is linearly dependent on transmitted to the flake [1687, 1694, 1698]. For flake the pressure difference across SLG (figure IX.69(d)) lengths higher than the critical length, the critical resulting in a slope from which the suspended SLG E2D strain for buckling is ~  −  0.6% [1692]. By consider- was estimated ~352 Nm−1, in agreement with values ing the critical buckling strain for an embedded flake derived from direct methods (see table IX.3). in the classical Euler regime [1672, 1689], the effective Instead of using the pressure difference, the bulg- bending rigidity of SLG fully embedded in a polymer ing test can be also implemented with electrostatic matrix was estimated ~70 MeV. This means that SLG actuation of SLG suspended onto holes with large encapsulation into a matrix affects the resistance to diameters (7.5–30 µm) over a gating chip forming a bending and paves the way for the development of parallel plate capacitor with an inter-plate distance, novel composite materials with higher values of com- d, (figure IX.70(a)) [1706]. By applying a gate voltage, pression strength compared to conventional ones. This SLG deflects subjecting to a pressure which defines the value, however, is six orders of magnitude higher than radial in plane stress and corresponds to a radial strain. in air and one order of magnitude higher than CVD Therefore E2D can be estimated (figure IX.70(d)). SLG in water [1680]. This could imply that encapsula- The radial strain depends on the height of the mem- tion into a matrix or in water, affects the resistance of brane dome at the centre deflection point, measured SLG to bending. by employing interferometric profilometry (figure The bulging test combined with Raman spectr­ IX.70(b). A number of different devices were tested oscopy was used [1704] on a multireflection model and the estimated E are presented in the histogram of 240 2D Mater. 7 (2020) 022001 C Backes et al the Amonton’s law. This behaviour was attributed the lower bending rigidity of 3LG, enabling higher surface conformation and stronger adhesion. Several diamond anvil cell experiments were con- ducted on MC 1L and FLG on Si/SiO2 or Cu, employing different pressure transmitting media [1529, 1710]. The choice of transmitting medium, e.g. polar (4:1 metha- nol-ethanol) or non-polar (e.g Fluorinert), is impor- tant due to pressure mediated doping that occurs from the substrate. Assuming that SLG follows the pressure- induced substrate contraction, its compression is deter- mined by the bulk modulus of the substrate (140 GPa for Cu [1710]). In Ref. [1710] suggested that the expected value of the pressure dependent wavenumber shift is ~15 cm−1/GPa (for SiO2/Si, which is more compress- ible, this is ~21.4 cm−1/GPa). The significant deviation was ascribed to the non-ideal adherence of SLG on Cu. From the applications point of view, it is impor- tant to understand how atomically thin membranes respond to mechanical deformations at the nanoscale. Sample preparation and handling hinders systematic studies, since a defect-free gauge area of the sample is difficult to be placed and gripped onto the loading device. The development of specific instruments with low force sensing and capable of applying in-plane stress in GRM membranes is necessary to capture their full response under axial deformation up to failure and to measure directly the elastic (e.g. stiffness) and inelastic properties (e.g. fracture strength and strain- to-failure). Fundamental questions have been raised on the effects of defects, out-of-plane deformations and wrinkling in either elastic or inelastic properties of GRMs [1669, 1711].Therefore, further research is needed in order for novel applications to take full advantage of these superior properties. Conclusions We presented the main techniques for production, Figure IX.70. (a) Device schematic. Inset: SEM image of transfer and functionalization of graphene, related a representative free-standing SLG membrane (scale bar, 8 µm). (b) Cross-sections of SLG membrane at various layered materials, and heterostructures. We also applied voltages. Height data obtained from interferometric summarized the key characterization techniques. We profilometry corresponding to these cross-sections are in the provided key practical details and procedures, to help inset. Also shown is a 3d view of the data at Vg  =  400 V. (c) the reader reproduce the results. We hope this work AFM measurements of SLG membrane with nm-scale static wrinkles (left, scale bar, 100 nm). A cross section of the AFM will stimulate both fundamental and practical interest data is shown in the bottom panel. Wrinkling is also evident in GRMs, and facilitate their use in a variety of novel on the high-angle tilted SEM image (right, scale bar, 1 µm). applications. (d) Histogram for all measured CVD SLG devices. Acknowledgment figure IX.70(d) and are significantly lower than those from other techniques (table IX.3). We acknowledge funding from the European Ref. [1707] employed Raman spectroscopy using Commission Graphene Flagship Core1 (grant the aforementioned SLG-sealed micro-chambers agreement 696656) and Core2 (grant agreement under variable external pressure to determine the pres­ 785219). We thank Dr Elena López-Elvira for editing this sure induced sliding friction between SiO2 and 1L-, 2L- manuscript. and 3LG. Tensile radial strain ~ 0.6% and compressive tangential strain ~  −0.3% were achieved. A violation ORCID iDs of the Amonton’s law [1708, 1709] for 1L and 2LG was found. In 3LG, however, the sliding friction is directly Mar Garcia-Hernandez https://orcid.org/0000- proportional to the applied load in accordance with 0002- 5987-0647 241 2D Mater. 7 (2020) 022001 C Backes et al References [25] Schwab M G et al 2015 Bottom-up synthesis of necklace-like graphene nanoribbons Chemistry 10 2134–8 [1] Chen L, Hernandez Y, Feng X and Müllen K 2012 From [26] Yang X, Dou X, Rouhanipour A, Zhi L, Räder H J and Müllen K nanographene and graphene nanoribbons to graphene sheets: 2008 Two-dimensional graphene nanoribbons J. Am. Chem. chemical synthesis Angew. Chem, Int. Ed. Engl. 51 7640–54 Soc. 130 4216–7 [2] Bai J and Huang Y 2010 Fabrication and electrical properties of [27] Schwab M G, Narita A, Hernandez Y, Balandina T, Mali K S, graphene nanoribbons Mater. Sci. Eng. R 70 341–53 De Feyter S, Feng X and Müllen K 2012 Structurally defined graphene nanoribbons with high lateral extension J. Am. [3] Ma L, Wang J and Ding F 2013 Recent progress and challenges Chem. Soc. 134 18169–72 in graphene nanoribbon synthesis ChemPhysChem 14 47–54 [28] Watson M D, Fechtenkötter A and Müllen K 2001 Big is [4] Terrones M et al 2010 Graphene and graphite nanoribbons: beautiful-’aromaticity’ revisited from the viewpoint of morphology, properties, synthesis, defects and applications macromolecular and supramolecular benzene chemistry Nano Today 5 351–72 Chem. Rev. 101 1267–300 [5] Barone V, Hod O and Scuseria G E 2006 Electronic structure [29] Wu J, Pisula W and Müllen K 2007 Graphenes as potential and stability of semiconducting graphene nanoribbons Nano material for electronics Chem. Rev. 107 718–47 Lett. 6 2748–54 [30] Stille J K 1981 Step-growth polymerization J. Chem. Educ. [6] Wakabayashi K, Sasaki K-I, Nakanishi T and Enoki T 2010 58 862 Electronic states of graphene nanoribbons and analytical [31] Zhang W et al 2017 Monitoring the on-surface synthesis of solutions Sci. Technol. Adv. Mater. 11 054504 graphene nanoribbons by mass spectrometry Anal. Chem. [7] Nakada K, Fujita M, Dresselhaus G and Dresselhaus M S 1996 89 7485–92 Edge state in graphene ribbons: nanometer size effect and edge [32] Narita A et al 2013 Synthesis of structurally well-defined and shape dependence Phys. Rev. B 54 17954–61 liquid-phase-processable graphene nanoribbons Nat. Chem. [8] Obradovic B, Kotlyar R, Heinz F, Matagne P, Rakshit T, 6 126–32 Giles M D, Stettler M A and Nikonov D E 2006 Analysis of [33] Carothers W H 1936 Polymers and polyfunctionality Trans. graphene nanoribbons as a channel material for field-effect Faraday Soc. 32 39 transistors Appl. Phys. Lett. 88 142102 [34] Odian G 2004 Principles of Polymerization (New York: Wiley) [9] Wang J, Zhao R, Yang M, Liu Z and Liu Z 2013 Inverse [35] Lee J, Cho H-J, Jung B-J, Cho N S and Shim H-K 2004 relationship between carrier mobility and bandgap in Stabilized blue luminescent polyfluorenes: introducing graphene J. Chem. Phys. 138 084701 polyhedral oligomeric silsesquioxane Macromolecules [10] Son Y-W, Cohen M L and Louie S G 2006 Energy gaps in 37 8523–9 graphene nanoribbons Phys. Rev. Lett. 97 216803 [36] Setayesh S, Grimsdale A C, Weil T, Enkelmann V, Müllen K, [11] Yang L, Park C-H, Son Y-W, Cohen M L and Louie S G Meghdadi F, List E J W and Leising G 2001 Polyfluorenes with 2007 Quasiparticle energies and band gaps in graphene polyphenylene dendron side chains: toward non-aggregating, nanoribbons Phys. Rev. Lett. 99 186801 light-emitting polymers J. Am. Chem. Soc. 123 946–53 [12] Fujita M, Wakabayashi K, Nakada K and Kusakabe K 1996 [37] Talirz L, Ruffieux P and Fasel R 2016 On-surface synthesis Peculiar localized state at zigzag graphite edge J. Phys. Soc. of atomically precise graphene nanoribbons Adv. Mater. Japan 65 1920–3 28 6222–31 [13] Wang X and Dai H 2010 Etching and narrowing of graphene [38] Grill L, Rieder K H, Moresco F, Rapenne G, Stojkovic S, from the edges Nat. Chem. 2 661 Bouju X and Joachim C 2007 Rolling a single molecular wheel [14] Bai J, Duan X and Huang Y 2009 Rational fabrication of at the atomic scale Nat. Nanotechnol. 2 95–8 graphene nanoribbons using a nanowire etch mask Nano Lett. [39] Huang H, Wei D, Sun J, Wong S L, Feng Y P, Neto A H C and 9 2083–7 Wee A T S 2012 Spatially resolved electronic structures of [15] Han M Y, Özyilmaz B, Zhang Y and Kim P 2007 Energy band- atomically precise armchair graphene nanoribbons Sci. Rep. gap engineering of graphene nanoribbons Phys. Rev. Lett. 2 983 98 206805 [40] Fairbrother A, Sanchez-Valencia J-R, Lauber B, Shorubalko I, [16] Palma C-A and Samorì P 2011 Blueprinting macromolecular Ruffieux P, Hintermann T and Fasel R 2017 High vacuum electronics Nat. Chem. 3 431 synthesis and ambient stability of bottom-up graphene [17] Kosynkin D V, Higginbotham A L, Sinitskii A, Lomeda J R, nanoribbons Nanoscale 9 2785–92 Dimiev A, Price B K and Tour J M 2009 Longitudinal [41] Chen Z et al 2016 Synthesis of graphene nanoribbons by unzipping of carbon nanotubes to form graphene ambient-pressure chemical vapor deposition and device nanoribbons Nature 458 872 integration J. Am. Chem. Soc. 138 15488–96 [18] Jiao L, Zhang L, Wang X, Diankov G and Dai H 2009 Narrow [42] Chen Z et al 2017 Chemical vapor deposition synthesis graphene nanoribbons from carbon nanotubes Nature and terahertz photoconductivity of low-band-gap N  =  9 458 877 armchair graphene nanoribbons J. Am. Chem. Soc. [19] Jiao L, Wang X, Diankov G, Wang H and Dai H 2010 Facile 139 3635–8 synthesis of high-quality graphene nanoribbons Nat. [43] Greenwood A E A N 1997 Chemistry of elements ISBN: Nanotechnol. 5 321 9780750633659 https://www.elsevier.com/books/chemistry- [20] Elías A L, Botello-Méndez A R, Meneses-Rodríguez D, Jehová of-the-elements/greenwood/978-0-7506-3365-9 González V, Ramírez-González D, Ci L, Muñoz-Sandoval E, [44] Kimouche A, Ervasti M M, Drost R, Halonen S, Harju A, Ajayan P M, Terrones H and Terrones M 2010 Longitudinal Joensuu P M, Sainio J and Liljeroth P 2015 Ultra-narrow cutting of pure and doped carbon nanotubes to form graphitic metallic armchair graphene nanoribbons Nat. Commun. nanoribbons using metal clusters as nanoscalpels Nano Lett. 6 10177 10 366–72 [45] Chen Z et al 2017 Lateral fusion of chemical vapor deposited [21] Cai J et al 2010 Atomically precise bottom-up fabrication of N  =  5 armchair graphene nanoribbons J. Am. Chem. Soc. graphene nanoribbons Nature 466 470–3 139 9483–6 [22] Narita A, Wang X-Y, Feng X and Müllen K 2015 New [46] Sakaguchi H, Kawagoe Y, Hirano Y, Iruka T, Yano M and advances in nanographene chemistry Chem. Soc. Rev. Nakae T 2014 Width-controlled sub-nanometer graphene 44 6616–43 nanoribbon films synthesized by radical-polymerized [23] Wang X-Y, Narita A and Müllen K 2017 Precision synthesis chemical vapor deposition Adv. Mater. 26 4134–8 versus bulk-scale fabrication of graphenes Nat. Rev. Chem. [47] Zhang H et al 2015 On-surface synthesis of rylene-type 2 0100 graphene nanoribbons J. Am. Chem. Soc. 137 4022–5 [24] Dössel L, Gherghel L, Feng X and Müllen K 2011 Graphene [48] Talirz L et al 2017 On-surface synthesis and characterization nanoribbons by chemists: nanometer-sized, soluble, and of 9-atom wide armchair graphene nanoribbons ACS Nano defect-free Angew. Chem, Int. Ed. Engl. 50 2540–3 11 1380–8 242 2D Mater. 7 (2020) 022001 C Backes et al [49] Chen Y-C, de Oteyza D G, Pedramrazi Z, Chen C, Fischer F R [71] Llinas J P et al 2017 Short-channel field-effect transistors and Crommie M F 2013 Tuning the band gap of graphene with 9-atom and 13-atom wide graphene nanoribbons Nat. nanoribbons synthesized from molecular precursors ACS Commun. 8 633 Nano 7 6123–8 [72] Bennett P B, Pedramrazi Z, Madani A, Chen Y-C, de [50] Nguyen G D et al 2016 Bottom-up synthesis of N  =  13 sulfur- Oteyza D G, Chen C, Fischer F R, Crommie M F and Bokor J doped graphene nanoribbons J. Phys. Chem. C 120 2684–7 2013 Bottom-up graphene nanoribbon field-effect transistors [51] Kawai S, Saito S, Osumi S, Yamaguchi S, Foster A S, Spijker P Appl. Phys. Lett. 103 253114 and Meyer E 2015 Atomically controlled substitutional boron- [73] Angelova P et al 2013 A universal scheme to convert doping of graphene nanoribbons Nat. Commun. 6 8098 aromatic molecular monolayers into functional carbon [52] Cloke R R, Marangoni T, Nguyen G D, Joshi T, Rizzo D J, nanomembranes ACS Nano 7 6489–97 Bronner C, Cao T, Louie S G, Crommie M F and Fischer F R [74] Matei D G et al 2013 Functional single-layer graphene sheets 2015 Site-specific substitutional boron doping of from aromatic monolayers Adv. Mater. 25 4146–51 semiconducting armchair graphene nanoribbons J. Am. Chem. [75] Nottbohm C T, Turchanin A, Beyer A and Gölzhäuser A 2009 Soc. 137 8872–5 Direct e-beam writing of 1 nm thin carbon nanoribbons J. Vac. [53] Zhang Y, Zhang Y, Li G, Lu J, Lin X, Du S, Berger R, Feng X, Sci. Technol. B 27 3059 Müllen K and Gao H-J 2014 Direct visualization of atomically [76] Nottbohm C T, Turchanin A, Beyer A, Stosch R and precise nitrogen-doped graphene nanoribbons Appl. Phys. Gölzhäuser A 2011 Mechanically stacked 1 nm-thick carbon Lett. 105 023101 nanosheets: ultrathin layered materials with tunable optical, [54] Bronner C, Stremlau S, Gille M, Brauße F, Haase A, Hecht S chemical, and electrical properties Small 7 874–83 and Tegeder P 2013 Aligning the band gap of graphene [77] Rhinow D, Weber N-E and Turchanin A 2012 Atmospheric nanoribbons by monomer doping Angew. Chem, Int. Ed. Engl. pressure, temperature-induced conversion of organic 52 4422–5 monolayers into nanocrystalline graphene J. Phys. Chem. C [55] Cai J et al 2014 Graphene nanoribbon heterojunctions Nat. 116 12295–303 Nanotechnol. 9 896–900 [78] Turchanin A, Beyer A, Nottbohm C T, Zhang X, Stosch R, [56] Marangoni T, Haberer D, Rizzo D J, Cloke R R and Fischer F R Sologubenko A, Mayer J, Hinze P, Weimann T and 2016 Heterostructures through divergent edge reconstruction Gölzhäuser A 2009 One nanometer thin carbon nanosheets in nitrogen-doped segmented graphene nanoribbons with tunable conductivity and stiffness Adv. Mater. 21 1233–7 Chemistry A 22 13037–40 [79] Turchanin A, Weber D, Büenfeld M, Kisielowski C, Fistul M V, [57] Ruffieux P et al 2016 On-surface synthesis of graphene Efetov K B, Weimann T, Stosch R, Mayer J and Gölzhäuser A nanoribbons with zigzag edge topology Nature 531 489–92 2011 Conversion of self-assembled monolayers into [58] de Oteyza D G et al 2016 Substrate-independent growth of nanocrystalline graphene: structure and electric transport ACS atomically precise chiral graphene nanoribbons ACS Nano Nano 5 3896–904 10 9000–8 [80] Weber N-E, Wundrack S, Stosch R and Turchanin A 2016 [59] Bronner C et al 2018 Hierarchical on-surface synthesis of Direct growth of patterned graphene Small 12 1440–5 graphene nanoribbon heterojunctions ACS Nano 12 2193–200 [81] Turchanin A and Gölzhäuser A 2016 Carbon nanomembranes [60] Chen Y-C, Cao T, Chen C, Pedramrazi Z, Haberer D, de Adv. Mater. 28 6075–103 Oteyza D G, Fischer F R, Louie S G and Crommie M F 2015 [82] Turchanin A, Käfer D, El-Desawy M, Wöll C, Witte G and Molecular bandgap engineering of bottom-up synthesized Gölzhäuser A 2009 Molecular mechanisms of electron- graphene nanoribbon heterojunctions Nat. Nanotechnol. induced cross-linking in aromatic SAMs Langmuir 25 7342–52 10 156–60 [83] Turchanin A, Schnietz M, El-Desawy M, Solak H H, [61] Jacobse P H, Kimouche A, Gebraad T, Ervasti M M, David C and Gölzhäuser A 2007 Fabrication of molecular Thijssen J M, Liljeroth P and Swart I 2017 Electronic nanotemplates in self-assembled monolayers by extreme- components embedded in a single graphene nanoribbon Nat. ultraviolet-induced chemical lithography Small 3 2114–9 Commun. 8 119 [84] Kankate L, Turchanin A and Gölzhäuser A 2009 On the release [62] Nguyen G D et al 2017 Atomically precise graphene of hydrogen from the S–H groups in the formation of self- nanoribbon heterojunctions from a single molecular assembled monolayers of thiols Langmuir 25 10435–8 precursor Nat. Nanotechnol. 12 1077–82 [85] Matei D G, Muzik H, Gölzhäuser A and Turchanin A 2012 [63] Gröning O et al 2018 Engineering of robust topological Structural investigation of 1,1′-biphenyl-4-thiol self- quantum phases in graphene nanoribbons Nature 560 209– assembled monolayers on Au(1 1 1) by scanning tunneling 13 microscopy and low-energy electron diffraction Langmuir [64] Rizzo D J, Veber G, Cao T, Bronner C, Chen T, Zhao F, 28 13905–11 Rodriguez H, Louie S G, Crommie M F and Fischer F R 2018 [86] Love J C, Estroff L A, Kriebel J K, Nuzzo R G and Topological band engineering of graphene nanoribbons Whitesides G M 2005 Self-assembled monolayers of Nature 560 204–8 thiolates on metals as a form of nanotechnology Chem. Rev. [65] Vandescuren M, Hermet P, Meunier V, Henrard L and 105 1103–70 Lambin P 2008 Theoretical study of the vibrational edge [87] Geyer W, Stadler V, Eck W, Zharnikov M, Gölzhäuser A modes in graphene nanoribbons Phys. Rev. B 78 195401 and Grunze M 1999 Electron-induced crosslinking of [66] Gross L, Mohn F, Moll N, Liljeroth P and Meyer G 2009 The aromatic self-assembled monolayers: negative resists for chemical structure of a molecule resolved by atomic force nanolithography Appl. Phys. Lett. 75 2401–3 microscopy Science 325 1110–4 [88] Turchanin A and Gölzhäuser A 2012 Carbon nanomembranes [67] Giessibl F J 2000 Atomic resolution on Si(1 1 1)-(7  ×  7) by from self-assembled monolayers: functional surfaces without noncontact atomic force microscopy with a force sensor based bulk Prog. Surf. Sci. 87 108–62 on a quartz tuning fork Appl. Phys. Lett. 76 1470–2 [89] Chesneau F, Hamoudi H, Schüpbach B R, Terfort A and [68] Bartels L, Meyer G, Rieder K H, Velic D, Knoesel E, Hotzel A, Zharnikov M 2011 Modification of self-assembled monolayers Wolf M and Ertl G 1998 Dynamics of electron-induced of perfluoroterphenyl-substituted alkanethiols by low-energy manipulation of individual CO molecules on Cu(1 1 1) Phys. electrons J. Phys. Chem. C 115 4773–82 Rev. Lett. 80 2004–7 [90] Amiaud L, Houplin J, Bourdier M, Humblot V, Azria R, [69] Denk R et al 2014 Exciton-dominated optical response of Pradier C M and Lafosse A 2014 Low-energy electron induced ultra-narrow graphene nanoribbons Nat. Commun. 5 4253 resonant loss of aromaticity: consequences on cross-linking in [70] Li G, Yoon K-Y, Zhong X, Zhu X and Dong G 2016 Efficient terphenylthiol SAMs Phys. Chem. Chem. Phys. 16 1050–9 bottom-up preparation of graphene nanoribbons by mild [91] Mrugalla A and Schnack J 2014 Classical molecular dynamics Suzuki-Miyaura polymerization of simple triaryl monomers investigations of biphenyl-based carbon nanomembranes Chemistry A 22 9116–20 Beilstein J. Nanotechnol. 5 865–71 243 2D Mater. 7 (2020) 022001 C Backes et al [92] Zharnikov M, Geyer W, Gölzhäuser A, Frey S and Grunze M [114] Zheng Z, Nottbohm C T, Turchanin A, Muzik H, Beyer A, 1999 Modification of alkanethiolate monolayers on Au- Heilemann M, Sauer M and Gölzhäuser A 2010 Janus substrate by low energy electron irradiation: alkyl chains and nanomembranes: a generic platform for chemistry in two the S/Au interface Phys. Chem. Chem. Phys. 1 3163–71 dimensions Angew. Chem, Int. Ed. Engl. 49 8493–7 [93] Zharnikov M and Grunze M 2002 Modification of thiol- [115] Bonaccorso F, Lombardo A, Hasan T, Sun Z, Colombo L and derived self-assembling monolayers by electron and x-ray Ferrari A C 2012 Production and processing of graphene and irradiation: scientific and lithographic aspects J. Vac. Sci. 2d crystals Mater. Today 15 564–89 Technol. B 20 1793 [116] Cano-Márquez A G, Rodríguez-Macías F J, Campos- [94] Zhang X, Vieker H, Beyer A and Gölzhäuser A 2014 Delgado J, Espinosa-González C G, Tristán-López F, Fabrication of carbon nanomembranes by helium ion beam Ramírez-González D, Cullen D A, Smith D J, Terrones M lithography Beilstein J. Nanotechnol. 5 188–94 and Vega-Cantú Y I 2009 Ex-MWNTs: graphene sheets and [95] Beyer A, Turchanin A, Nottbohm C T, Mellech N, Schnietz M ribbons produced by lithium intercalation and exfoliation of and Gölzhäuser A 2010 Fabrication of metal patterns on carbon nanotubes Nano Lett. 9 1527–33 freestanding graphenoid nanomembranes J. Vac. Sci. Technol. [117] Varela-Rizo H, Rodriguez-Pastor I, Merino C, Terrones M B 28 C6D5–10 and Martin-Gullon I 2011 Graphene oxide nanoplatelets of [96] Ferrari A C and Basko D M 2013 Raman spectroscopy as a different crystallinity synthesized from helical-ribbon carbon versatile tool for studying the properties of graphene Nat. nanofibers and multiwall carbon nanotubes J. Mater. Res. Nanotechnol. 8 235–46 26 2632–41 [97] Kim K, Lee Z, Regan W, Kisielowski C, Crommie M F and [118] Yeon Y, Lee M Y, Kim S Y, Lee J, Kim B, Park B and In I 2015 Zettl A 2011 Grain boundary mapping in polycrystalline Production of quasi-2D graphene nanosheets through graphene ACS Nano 5 2142–6 the solvent exfoliation of pitch-based carbon fiber [98] Warner J H, Rümmeli M H, Gemming T, Büchner B and Nanotechnology 26 375602 Briggs G A D 2009 Direct imaging of rotational stacking faults [119] Lee M, Lee J, Park S Y, Min B, Kim B and In I 2015 Production in few layer graphene Nano Lett. 9 102–6 of graphene oxide from pitch-based carbon fiber Sci. Rep. [99] Kotakoski J, Krasheninnikov A V, Kaiser U and Meyer J C 5 11707 2011 From point defects in graphene to two-dimensional [120] Botas C, Álvarez P, Blanco C, Santamaría R, Granda M, Ares P, amorphous carbon Phys. Rev. Lett. 106 105505 Rodríguez-Reinoso F and Menéndez R 2012 The effect of the [100] Jung I, Dikin D A, Piner R D and Ruoff R S 2008 Tunable parent graphite on the structure of graphene oxide Carbon electrical conductivity of individual graphene oxide sheets 50 275–82 reduced at ‘low’ temperatures Nano Lett. 8 4283–7 [121] Bernal J D 1924 The structure of graphite Proc. R. Soc. A [101] Eda G, Mattevi C, Yamaguchi H, Kim H and Chhowalla M 106 749–73 2009 Insulator to semimetal transition in graphene oxide J. [122] Lipson H and Stokes A R 1942 The structure of graphite Proc. Phys. Chem. C 113 15768–71 R. Soc. A 181 101–5 [102] Loh K P, Bao Q, Eda G and Chhowalla M 2010 Graphene [123] Franklin R E 1951 Crystallite growth in graphitizing and non- oxide as a chemically tunable platform for optical applications graphitizing carbons Proc. R. Soc. A 209 196–218 Nat. Chem. 2 1015–24 [124] Marsh J A G H 1979 14th Biennial Conf. on Carbon ed ACS [103] Parvez K, Yang S, Hernandez Y, Winter A, Turchanin A, (Penn State University) p 117 www.tib.eu/en/search/id/ Feng X and Müllen K 2012 Nitrogen-doped graphene and its TIBKAT%3A041557468/Contributions-for-the-14th- iron-based composite as efficient electrocatalysts for oxygen Biennial-Conference/ reduction reaction ACS Nano 6 9541–50 [125] Meyer M I R A 1999 Stress Graphitization in Chemistry and [104] Wu Z-S, Winter A, Chen L, Sun Y, Turchanin A, Feng X and Physics of Carbon (New York: Dekker) Müllen K 2012 Three-dimensional nitrogen and boron [126] Loh G C and Baillargeat D 2013 Graphitization of amorphous Co-doped graphene for high-performance all-solid-state carbon and its transformation pathways J. Appl. Phys. supercapacitors Adv. Mater. 24 5130–5 114 033534 [105] Ferrari A C et al 2006 Raman spectrum of graphene and [127] Rodriguez-Reinoso F 1989 Comite internacional para graphene layers Phys. Rev. Lett. 97 187401 la caracterizacion y la terminologia del carbon Carbon [106] Meyer J C, Geim A K, Katsnelson M I, Novoselov K S, 27 305–12 Obergfell D, Roth S, Girit C and Zettl A 2007 On the [128] Li X, Magnuson C W, Venugopal A, Tromp R M, Hannon J B, roughness of single- and bi-layer graphene membranes Solid Vogel E M, Colombo L and Ruoff R S 2011 Large-area State Commun. 143 101–9 graphene single crystals grown by low-pressure chemical [107] Novoselov K S, Geim A K, Morozov S V, Jiang D, vapor deposition of methane on copper J. Am. Chem. Soc. Katsnelson M I, Grigorieva I V, Dubonos S V and Firsov A A 133 2816–9 2005 Two-dimensional gas of massless Dirac fermions in [129] Wassei J K, Mecklenburg M, Torres J A, Fowler J D, Regan B C, graphene Nature 438 197–200 Kaner R B and Weiller B H 2012 Graphene: chemical vapor [108] Gao L et al 2012 Repeated growth and bubbling transfer of deposition of graphene on copper from methane, ethane and graphene with millimetre-size single-crystal grains using propane: evidence for bilayer selectivity (Small 9/2012) Small platinum Nat. Commun. 3 699 8 1289 [109] Wang D-Y et al 2013 Clean-lifting transfer of large-area [130] Novoselov K S, Jiang D, Schedin F, Booth T J, Khotkevich V V, residual-free graphene films Adv. Mater. 25 4521–6 Morozov S V and Geim A K 2005 Two-dimensional atomic [110] Lercel M J, Redinbo G F, Pardo F D, Rooks M, Tiberio R C, crystals Proc. Natl Acad. Sci. 102 10451–3 Simpson P, Craighead H G, Sheen C W, Parikh A N and [131] Hernandez Y V et al 2008 High-yield production of graphene Allara D L 1994 Electron beam lithography with monolayers by liquid-phase exfoliation of graphite Nat. Nanotechnol. of alkylthiols and alkylsiloxanes J. Vac. Sci. Technol. B 3 563–8 12 3663–7 [132] Botas C, Pérez-Mas A M, Álvarez P, Santamaría R, Granda M, [111] Ferrari A C et al 2015 Science and technology roadmap for Blanco C and Menéndez R 2013 Optimization of the size and graphene, related two-dimensional crystals, and hybrid yield of graphene oxide sheets in the exfoliation step Carbon systems Nanoscale 7 4598–810 63 576–8 [112] Geim A K and Grigorieva I V 2013 Van der Waals [133] Cai M, Thorpe D, Adamson D H and Schniepp H C 2012 heterostructures Nature 499 419–25 Methods of graphite exfoliation J. Mater. Chem. 22 24992 [113] Zheng Z, Zhang X, Neumann C, Emmrich D, Winter A, [134] Bourlinos A B, Georgakilas V, Zboril R, Steriotis T A and Vieker H, Liu W, Lensen M, Gölzhäuser A and Turchanin A Stubos A K 2009 Liquid-phase exfoliation of graphite towards 2015 Hybrid van der Waals heterostructures of zero- solubilized graphenes Small 5 1841–5 dimensional and two-dimensional materials Nanoscale [135] González Z, Botas C, Blanco C, Santamaría R, Granda M, 7 13393–7 Álvarez P and Menéndez R 2013 Graphite oxide-based 244 2D Mater. 7 (2020) 022001 C Backes et al graphene materials as positive electrodes in vanadium redox [157] Moon I K, Lee J, Ruoff R S and Lee H 2010 Reduced graphene flow batteries J. Power Sources 241 349–54 oxide by chemical graphitization Nat. Commun. 1 1–6 [136] Liscio A, Kouroupis-Agalou K, Betriu X D, Kovtun A, [158] Park S and Ruoff R S 2009 Chemical methods for the Treossi E, Pugno N M, De Luca G, Giorgini L and Palermo V production of graphenes Nat. Nanotechnol. 4 217–24 2017 Evolution of the size and shape of 2D nanosheets during [159] Benavente E, Santa Ana M A, Mendizábal F and González G ultrasonic fragmentation 2D Mater. 4 025017 2002 Intercalation chemistry of molybdenum disulfide [137] Xia Z Y, Pezzini S, Treossi E, Giambastiani G, Corticelli F, Coord. Chem. Rev. 224 87–109 Morandi V, Zanelli A, Bellani V and Palermo V 2013 The [160] Eigler S and Hirsch A 2014 Chemistry with graphene and exfoliation of graphene in liquids by electrochemical, graphene oxide—challenges for synthetic chemists Angew. chemical, and sonication-assisted techniques: a nanoscale Chem, Int. Ed. Engl. 53 7720–38 study Adv. Funct. Mater. 23 4684–93 [161] Grayfer E D, Kozlova M N and Fedorov V E 2017 Colloidal [138] Ghosh P K B A S 1997 Elements of Prospecting for Non-Fuel 2D nanosheets of MoS2 and other transition metal Mineral Deposits (Allied Publisher Ltd) dichalcogenides through liquid-phase exfoliation Adv. Colloid [139] Peng W, Li H, Hu Y, Liu Y and Song S 2016 Characterisation of Interface Sci. 245 40–61 reduced graphene oxides prepared from natural flaky, lump [162] Nicolosi V, Chhowalla M, Kanatzidis M G, Strano M S and and amorphous graphites Mater. Res. Bull. 78 119–27 Coleman J N 2013 Liquid exfoliation of layered materials [140] Peng W, Li H, Hu Y, Liu Y and Song S 2016 Does silicate Science 340 1420 mineral impurities in natural graphite affect the [163] Yi M and Shen Z 2015 A review on mechanical exfoliation characteristics of synthesized graphene? Mater. Res. Bull. for the scalable production of graphene J. Mater. Chem. C 74 333–9 3 11700–15 [141] Colonna S, Monticelli O, Gomez J, Novara C, Saracco G and [164] Tao H, Zhang Y, Gao Y, Sun Z, Yan C and Texter J 2017 Scalable Fina A 2016 Effect of morphology and defectiveness of graphene- exfoliation and dispersion of two-dimensional materials—an related materials on the electrical and thermal conductivity of update Phys. Chem. Chem. Phys. 19 921–60 their polymer nanocomposites Polymer 102 292–300 [165] Ciesielski A and Samorì P 2016 Supramolecular approaches [142] Colonna S, Monticelli O, Gomez J, Saracco G and Fina A to graphene: from self-assembly to molecule-assisted liquid- 2017 Morphology and properties evolution upon ring- phase exfoliation Adv. Mater. 28 6030–51 opening polymerization during extrusion of cyclic butylene [166] Bonaccorso F, Bartolotta A, Coleman J N and Backes C 2016 terephthalate and graphene-related-materials into thermally 2D-crystal-based functional inks Adv. Mater. 28 6136–66 conductive nanocomposites Eur. Polym. J. 89 57–66 [167] Gutiérrez M and Henglein A 1989 Preparation of colloidal [143] Galindo B, Alcolea S G, Gómez J, Navas A, Murguialday A O, semiconductor solutions of MoS2 and WSe2 via sonication Fernandez M P and Puelles R C 2014 Effect of the number Ultrasonics 27 259–61 of layers of graphene on the electrical properties of TPU [168] Kang J, Sangwan V K, Wood J D and Hersam M C 2017 polymers IOP Conf. Ser.: Mater. Sci. Eng. 64 012008 Solution-based processing of monodisperse two-dimensional [144] Gomez J 2017 Preparation and processing of large lateral size nanomaterials Acc. Chem. Res. 50 943–51 graphene materialsin epoxy matrix composites Graphene [169] Abellán G, Martí-Gastaldo C, Ribera A and Coronado E Week (Athens, Greece) 2015 Hybrid materials based on magnetic layered double [145] Casaluci S, Gemmi M, Pellegrini V, Di Carlo A and hydroxides: a molecular perspective Acc. Chem. Res. Bonaccorso F 2016 Graphene-based large area dye-sensitized 48 1601–11 solar cell modules Nanoscale 8 5368–78 [170] Naguib M, Mochalin V N, Barsoum M W and Gogotsi Y [146] Villaro E 2013 Preparación y caracterización de 2014 25th anniversary article: MXenes: a new family of two- nanocompuestos de grafeno en una suspensión de partículas dimensional materials Adv. Mater. 26 992–1005 de PVC (plastisol) MsD Thesis Facultad de Ciencias UNED [171] Kim H, Saiz E, Chhowalla M and Mattevi C 2013 Modeling of [147] Sun L and Fugetsu B 2013 Mass production of graphene oxide the self-limited growth in catalytic chemical vapor deposition from expanded graphite Mater. Lett. 109 207–10 of graphene New J. Phys. 15 053012 [148] Sierra U, Álvarez P, Blanco C, Granda M, Santamaría R and [172] Ciesielski A and Samori P 2014 Graphene via sonication Menéndez R 2015 New alternatives to graphite for producing assisted liquid-phase exfoliation Chem. Soc. Rev. 43 381–98 graphene materials Carbon 93 812–8 [173] Parvez K, Yang S, Feng X and Müllen K 2015 Exfoliation of [149] Sierra U, Álvarez P, Blanco C, Granda M, Santamaría R and graphene via wet chemical routes Synth. Met. 210 123–32 Menéndez R 2016 Cokes of different origin as precursors of [174] Coleman J N et al 2011 Two-dimensional nanosheets graphene oxide Fuel 166 400–3 produced by liquid exfoliation of layered materials Science [150] Alvarez P, Díez N, Santamaría R, Blanco C, Menéndez R and 331 568–71 Granda M 2012 Novel coal-based precursors for cokes with [175] Zhong Y L, Tian Z, Simon G P and Li D 2015 Scalable highly oriented microstructures Fuel 95 400–6 production of graphene via wet chemistry: progress and [151] Oberlin A 1984 Carbonization and graphitization Carbon challenges Mater. Today 18 73–8 22 521–41 [176] Paton K R 2014 Scalable production of large quantities of [152] Fitzer E, Kochling K H, Boehm H P and Marsh H 1995 defect-free few-layer graphene by shear exfoliation in liquids Recommended terminology for the description of carbon as Nat. Mater. 13 624–30 a solid (IUPAC Recommendations 1995) Pure Appl. Chem. [177] Varrla E, Paton K R, Backes C, Harvey A, Smith R J, 67 473–506 McCauley J and Coleman J N 2014 Turbulence-assisted shear [153] Del Rio-Castillo A E, Merino C, Díez-Barra E and Vázquez E exfoliation of graphene using household detergent and a 2014 Selective suspension of single layer graphene kitchen blender Nanoscale 6 11810–9 mechanochemically exfoliated from carbon nanofibres Nano [178] Karagiannidis P G et al 2017 Microfluidization of graphite Res. 7 963–72 and formulation of graphene-based conductive inks ACS [154] Lobato B, Merino C, Barranco V and Centeno T A 2016 Large- Nano 11 2742–55 scale conversion of helical-ribbon carbon nanofibers to a [179] León V, Quintana M, Herrero M A, Fierro J L G, Hoz A D L, variety of graphene-related materials RSC Adv. 6 57514–20 Prato M and Vázquez E 2011 Few-layer graphenes from [155] Vera-Agullo J, Varela-Rizo H, Conesa J A, Almansa C, ball-milling of graphite with melamine Chem. Commun. Merino C and Martin-Gullon I 2007 Evidence for growth 47 10936 mechanism and helix-spiral cone structure of stacked-cup [180] Qian W, Hao R, Hou Y, Tian Y, Shen C, Gao H and Liang X carbon nanofibers Carbon 45 2751–8 2009 Solvothermal-assisted exfoliation process to produce [156] Hummers W S and Offeman R E 1958 Preparation of graphene with high yield and high quality Nano Res. graphitic oxide J. Am. Chem. Soc. 80 1339 2 706–12 245 2D Mater. 7 (2020) 022001 C Backes et al [181] Tang Z, Zhuang J and Wang X 2010 Exfoliation of graphene dispersions of single-walled carbon nanotubes J. Phys. Chem. from graphite and their self-assembly at the oil–water B 107 8799–804 interface Langmuir 26 9045–9 [203] Yau H C, Bayazit M K, Steinke J H G and Shaffer M S P 2015 [182] Guardia L, Fernandez-Merino M J, Paredes J I, Solis- Sonochemical degradation of N-methylpyrrolidone and its Fernandez P, Villar-Rodil S, Martinez-Alonso A and influence on single walled carbon nanotube dispersion Chem. Tascon J M D 2011 High-throughput production of pristine Commun. 51 16621–4 graphene in an aqueous dispersion assisted by non-ionic [204] Ciesielski A et al 2016 Modifying the size of ultrasound- surfactants Carbon 49 1653–62 induced liquid-phase exfoliated graphene: from nanosheets [183] Green A A and Hersam M C 2009 Solution phase production to nanodots ACS Nano 10 10768–77 of graphene with controlled thickness via density [205] Solomon H M, Burgess B A, Kennedy G L and Staples R E differentiation Nano Lett. 9 4031–6 1995 1-methyl-2-pyrrolidone (nmp): reproductive and [184] Lotya M et al 2009 Liquid phase production of graphene by developmental toxicity study by inhalation in the rat Drug exfoliation of graphite in surfactant/water solutions J. Am. Chem. Toxicol. 18 271–93 Chem. Soc. 131 3611–20 [206] Capasso A, Del Rio Castillo A E, Sun H, Ansaldo A, [185] Backes C et al 2014 Edge and confinement effects allow in Pellegrini V and Bonaccorso F 2015 Ink-jet printing of situ measurement of size and thickness of liquid-exfoliated graphene for flexible electronics: an environmentally-friendly nanosheets Nat. Commun. 5 4576 approach Solid State Commun. 224 53–63 [186] Backes C et al 2016 Production of highly monolayer enriched [207] Zhou K-G, Mao N-N, Wang H-X, Peng Y and Zhang H-L 2011 dispersions of liquid-exfoliated nanosheets by liquid cascade A mixed-solvent strategy for efficient exfoliation of inorganic centrifugation ACS Nano 10 1589–601 graphene analogues Angew. Chem, Int. Ed. Engl. 50 10839–42 [187] Lee K P, Chromey N C, Culik R, Barnes J R and Schneider P W [208] May P, Khan U, Hughes J M and Coleman J N 2012 Role of 1987 Toxicity of N-methyl-2-pyrrolidone (NMP): solubility parameters in understanding the steric stabilization teratogenic, subchronic, and two-year inhalation studies of exfoliated two-dimensional nanosheets by adsorbed Fundam. Appl. Toxicol. 9 222–35 polymers J. Phys. Chem. C 116 11393–400 [188] Harvey A et al 2015 Preparation of gallium sulfide nanosheets [209] Liang Y T and Hersam M C 2010 Highly concentrated by liquid exfoliation and their application as hydrogen graphene solutions via polymer enhanced solvent exfoliation evolution catalysts Chem. Mater. 27 3483–93 and iterative solvent exchange J. Am. Chem. Soc. 132 17661–3 [189] Favron A, Gaufres E, Fossard F, Phaneuf-Lheureux A-L, [210] Bourlinos A B, Georgakilas V, Zboril R, Steriotis T A, Tang N Y W, Levesque P L, Loiseau A, Leonelli R, Francoeur S Stubos A K and Trapalis C 2009 Aqueous-phase exfoliation and Martel R 2015 Photooxidation and quantum of graphite in the presence of polyvinylpyrrolidone for the confinement effects in exfoliated black phosphorus Nat. production of water-soluble graphenes Solid State Commun. Mater. 14 826–32 149 2172–6 [190] Hanlon D C et al 2015 Liquid exfoliation of solvent-stabilised [211] Liu W S, Zhou R, Zhou D, Ding G G, Soah J M, Yue C Y and few-layer black phosphorus for applications beyond Lu X H 2015 Lignin-assisted direct exfoliation of graphite to electronics Nat. Commun. 6 8563 graphene in aqueous media and its application in polymer [191] Guo Y, Zhou S, Bai Y and Zhao J 2017 Defects and oxidation composites Carbon 83 188–97 of group-III monochalcogenide monolayers J. Chem. Phys. [212] Ahadian S et al 2015 Facile and green production of aqueous 147 104709 graphene dispersions for biomedical applications Nanoscale [192] Wang S, Yi M and Shen Z 2016 The effect of surfactants and 7 6436–43 their concentration on the liquid exfoliation of graphene RSC [213] Irin F, Hansen M J, Bari R, Parviz D, Metzler S D, Adv. 6 56705–10 Bhattacharia S K and Green M J 2015 Adsorption and removal [193] Liscio A et al 2017 Exfoliation of few-layer graphene in of graphene dispersants J. Colloid Interface Sci. 446 282–9 volatile solvents using aromatic perylene diimide derivatives [214] Schlierf A, Samorì P and Palermo V 2014 Graphene–organic as surfactants ChemPlusChem 82 358–67 composites for electronics: optical and electronic interactions [194] Schlierf A et al 2013 Nanoscale insight into the exfoliation in vacuum, liquids and thin solid films J. Mater. Chem. C mechanism of graphene with organic dyes: effect of charge, 2 3129 dipole and molecular structure Nanoscale 5 4205–16 [215] Englert J M, Röhrl J, Schmidt C D, Graupner R, [195] Svedberg T, Pederson K O and Bauer J H 1940 The Hundhausen M, Hauke F and Hirsch A 2009 Soluble Ultracentrifuge (London: Oxford University Press) graphene: generation of aqueous graphene solutions aided [196] Hansen C 1967 The Three Dimensional Solubility Parameter by a perylenebisimide-based bolaamphiphile Adv. Mater. and Solvent Diffusion Coefficient and Their Importance in 21 4265–9 Surface Coating Formulation (Copenhagen: Danish Technical [216] Melucci M et al 2012 Graphene-organic hybrids as Press) processable, tunable platforms for pH-dependent [197] Hughes J M, Aherne D and Coleman J N 2013 Generalizing photoemission, obtained by a new modular approach J. solubility parameter theory to apply to one- and two- Mater. Chem. 22 18237–43 dimensional solutes and to incorporate dipolar interactions J. [217] Melucci M, Treossi E, Ortolani L, Giambastiani G, Morandi V, Appl. Polym. Sci. 127 4483–91 Klar P, Casiraghi C, Samorì P and Palermo V 2010 Facile [198] Shen J et al 2015 Liquid phase exfoliation of two-dimensional covalent functionalization of graphene oxide using materials by directly probing and matching surface tension microwaves: bottom-up development of functional graphitic components Nano Lett. 15 5449–54 materials J. Mater. Chem. 20 9052 [199] Kim J et al 2015 Direct exfoliation and dispersion of two- [218] Schlierf A, Cha K, Schwab M G, Samori P and Palermo V 2014 dimensional materials in pure water via temperature control Exfoliation of graphene with an industrial dye: teaching an Nat. Commun. 6 8294 old dog new tricks 2D Mater. 1 035006 [200] Zhi C Y, Bando Y, Tang C C, Kuwahara H and Golberg D [219] Yang H, Withers F, Gebremedhn E, Lewis E, Britnell L, 2009 Large-scale fabrication of boron nitride nanosheets Felten A, Palermo V, Haigh S, Beljonne D and Casiraghi C and their utilization in polymeric composites with 2014 Dielectric nanosheets made by liquid-phase exfoliation improved thermal and mechanical properties Adv. Mater. in water and their use in graphene-based electronics 2D 21 2889 Mater. 1 011012 [201] Finn D J, Lotya M, Cunningham G, Smith R J, McCloskey D, [220] Bianco A 2013 Graphene: safe or toxic? the two faces of the Donegan J F and Coleman J N 2014 Inkjet deposition of medal Angew. Chem, Int. Ed. Eng. 52 4986–97 liquid-exfoliated graphene and MoS2 nanosheets for printed [221] Feng X L, Marcon V, Pisula W, Hansen M R, Kirkpatrick J, device applications J. Mater. Chem. C 2 925–32 Grozema F, Andrienko D, Kremer K and Müllen K 2009 [202] Niyogi S, Hamon M A, Perea D E, Kang C B, Zhao B, Pal S K, Towards high charge-carrier mobilities by rational design of Wyant A E, Itkis M E and Haddon R C 2003 Ultrasonic the shape and periphery of discotics Nat. Mater. 8 421–6 246 2D Mater. 7 (2020) 022001 C Backes et al [222] Li C, Liu M Y, Pschirer N G, Baumgarten M and Müllen K [244] O’Neill A, Khan U and Coleman J N 2012 Preparation of high 2010 Polyphenylene-based materials for organic concentration dispersions of exfoliated MoS2 with increased photovoltaics Chem. Rev. 110 6817–55 flake size Chem. Mater. 24 2414–21 [223] Fujigaya T and Nakashima N 2008 Methodology for [245] Backes C, Hanlon D, Szydlowska B M, Harvey A, Smith R J, homogeneous dispersion of single-walled carbon nanotubes Higgins T M and Coleman J N 2016 Preparation of liquid- by physical modification Polym. J. 40 577–89 exfoliated transition metal dichalcogenide nanosheets with [224] Dong X, Shi Y, Zhao Y, Chen D, Ye J, Yao Y, Gao F, Ni Z, Yu T controlled size and thickness: a state of the art protocol J. Vis. and Shen Z 2009 Symmetry breaking of graphene monolayers Exp. 118 e54806 (https://doi.org/10.3791/54806) by molecular decoration Phys. Rev. Lett. 102 135501 [246] Ueberricke L, Coleman J N and Backes C 2017 Robustness of [225] Jang J-H, Rangappa D, Kwon Y-U and Honma I 2010 Direct size selection and spectroscopic size, thickness and monolayer preparation of 1-PSA modified graphene nanosheets by metrics of liquid-exfoliated WS2 Phys. Status Solidi b supercritical fluidic exfoliation and its electrochemical 254 1700443 properties J. Mater. Chem. 21 3462–6 [247] Panagiotou T, Mesite S, Bernard J, Chomistek K and [226] Parviz D, Das S, Ahmed H S T, Irin F, Bhattacharia S and Fisher R 2008 Production of polymer nanosuspensions Green M J 2012 Dispersions of non-covalently functionalized using microfluidizerTM processor based processes NSTI graphene with minimal stabilizer ACS Nano 6 8857–67 Nanotechnol. 1 688–91 (www.nsti.org/) [227] Zhang L, Zhang Z J, He C Z, Dai L M, Liu J and Wang L X 2014 [248] Goldberg S 2008 Mechanical/physical methods of cell Rationally designed surfactants for few-layered graphene disruption and tissue homogenization 2D PAGE: Sample exfoliation: ionic groups attached to electron-deficient pi- Preparation and Fractionation pp 3–22 conjugated unit through alkyl spacers ACS Nano 8 6663–70 [249] M.I. Corporation (www.microfluidicscorp.com) [228] Sampath S, Basuray A N, Hartlieb K J, Aytun T, Stupp S I and [250] Lajunen T, Hisazumi K, Kanazawa T, Okada H, Seta Y, Stoddart J F 2013 Direct exfoliation of graphite to graphene Yliperttula M, Urtti A and Takashima Y 2014 Topical drug in aqueous media with diazaperopyrenium dications Adv. delivery to retinal pigment epithelium with microfluidizer Mater. 25 2740–5 produced small liposomes Eur. J. Pharm. Sci. 62 23–32 [229] Chua L L, Zaumseil J, Chang J F, Ou E C W, Ho P K H, [251] Tang S Y, Shridharan P and Sivakumar M 2013 Impact Sirringhaus H and Friend R H 2005 General observation of process parameters in the generation of novel aspirin of n-type field-effect behaviour in organic semiconductors nanoemulsions—comparative studies between ultrasound Nature 434 194–9 cavitation and microfluidizer Ultrason. Sonochem. 20 485–97 [230] Ciesielski A et al 2014 Liquid-phase exfoliation of graphene [252] Jafari S M, He Y and Bhandari B 2007 Production of sub- using intercalating compounds: a supramolecular approach micron emulsions by ultrasound and microfluidization Angew. Chem, Int. Ed. Eng. 53 10355–61 techniques J. Food Eng. 82 478–88 [231] Haar S et al 2015 A supramolecular strategy to leverage the [253] Panagiotou T, Bernard J M and Mesite S V 2008 liquid-phase exfoliation of graphene in the presence of Deagglomeration and dispersion of carbon nanotubes using surfactants: unraveling the role of the length of fatty acids microfluidizer high shear fluid processors Conf. and Expo Small 11 1691–702 Proc. Nano Science and Technology Institute pp 1–5 [232] Haar S, El Gemayel M, Shin Y Y, Melinte G, Squillaci M A, [254] dos Reis Benatto G A, Roth B, Madsen M V, Hösel M, Ersen O, Casiraghi C, Ciesielski A and Samori P 2015 Søndergaard R R, Jørgensen M and Krebs F C 2014 Carbon: Enhancing the liquid-phase exfoliation of graphene in the ultimate electrode choice for widely distributed polymer organic solvents upon addition of n-octylbenzene Sci. Rep. solar cells Adv. Energy Mater. 4 1400732 5 16684 [255] Lucera L, Kubis P, Fecher F W, Bronnbauer C, Turbiez M, [233] Conti S et al 2016 Perchlorination of coronene enhances its Forberich K, Ameri T, Egelhaaf H-J and Brabec C J 2015 propensity for self-assembly on graphene ChemPhysChem Guidelines for closing the efficiency gap between hero 17 352–7 solar cells and roll-to-roll printed modules Energy Technol. [234] Döbbelin M et al 2016 Light-enhanced liquid-phase 3 373–84 exfoliation and current photoswitching in graphene- [256] Nisato G, Lupo D and Ganz S 2016 Organic and Printed azobenzene composites Nat. Commun. 7 11090 Electronics: Fundamentals and Applications (Pan Stanford) [235] Haar S et al 2016 Liquid-phase exfoliation of graphite into [257] Huang X, Leng T, Zhang X, Chen J C, Chang K H, Geim A K, single- and few-layer graphene with alpha-functionalized Novoselov K S and Hu Z 2015 Binder-free highly conductive alkanes J. Phys. Chem. Lett. 7 2714–21 graphene laminate for low cost printed radio frequency [236] Backes C, Paton K R, Hanlon D, Yuan S, Katsnelson M I, applications Appl. Phys. Lett. 106 203105 Houston J, Smith R J, McCloskey D, Donegan J F and [258] Welham N J, Berbenni V and Chapman P G 2003 Effect Coleman J N 2016 Spectroscopic metrics allow in situ of extended ball milling on graphite J. Alloys Compd. measurement of mean size and thickness of liquid-exfoliated 349 255–63 few-layer graphene nanosheets Nanoscale 8 4311–23 [259] Servant A, Leon V, Jasim D, Methven L, Limousin P, [237] Duncan B and Perry C 1977 Uncancelled hypnotic Fernandez-Pacheco E V, Prato M and Kostarelos K 2014 suggestions: initial studies Am. J. Clin. Hypn. 19 166–76 Graphene-based electroresponsive scaffolds as polymeric [238] Varrla E, Backes C, Paton K R, Harvey A, Gholamvand Z, implants for on-demand drug delivery Adv. Healthcare Mater. McCauley J and Coleman J N 2015 Large-scale production of 3 1334–43 size-controlled MoS2 nanosheets by shear exfoliation Chem. [260] Gossner C M-E, Schlundt J, Ben Embarek P, Hird S, Lo-Fo- Mater. 27 1129–39 Wong D, Beltran J J O, Teoh K N and Tritscher A 2009 The [239] Hanlon D et al 2014 Production of molybdenum trioxide melamine incident: implications for international food and nanosheets by liquid exfoliation and their application in high- feed safety Environ. Health Perspect. 117 1803–8 performance supercapacitors Chem. Mater. 26 1751–63 [261] Rodríguez A M, Muñoz-García A B, Crescenzi O, Vázquez E [240] Harvey A et al 2016 Production of Ni(OH)2 nanosheets and Pavone M 2016 Stability of melamine-exfoliated by liquid phase exfoliation: from optical properties to graphene in aqueous media: quantum-mechanical insights at electrochemical applications J. Mater. Chem. C 4 11046–59 the nanoscale Phys. Chem. Chem. Phys. 18 22203–9 [241] Khan U, O’Neill A, Lotya M, De S and Coleman J N 2010 [262] Mottier A et al 2016 Surface area of carbon nanoparticles: a High-concentration solvent exfoliation of graphene Small dose metric for a more realistic ecotoxicological assessment 6 864–71 Nano Lett. 16 3514–8 [242] Lotya M, King P J, Khan U, De S and Coleman J N 2010 High- [263] Bramini M, Sacchetti S, Armirotti A, Rocchi A, Vázquez E, concentration, surfactant-stabilized graphene dispersions León Castellanos V, Bandiera T, Cesca F and Benfenati F 2016 ACS Nano 4 3155–62 Graphene oxide nanosheets disrupt lipid composition, Ca2+ [243] Ashokkumar M 2016 Handbook of Ultrasonics and homeostasis, and synaptic transmission in primary cortical Sonochemistry (Singapore: Springer) neurons ACS Nano 10 7154–71 247 2D Mater. 7 (2020) 022001 C Backes et al [264] Rauti R et al 2016 Graphene oxide nanosheets reshape Palermo V and Bianco A 2015 Dispersibility-dependent synaptic function in cultured brain networks ACS Nano biodegradation of graphene oxide by myeloperoxidase Small 10 4459–71 11 3985–94 [265] Frišcic T 2010 New opportunities for materials synthesis [288] Russier J, Treossi E, Scarsi A, Perrozzi F, Dumortier H, using mechanochemistry J. Mater. Chem. 20 7599 Ottaviano L, Meneghetti M, Palermo V and Bianco A 2013 [266] Abdelkader A M and Kinloch I A 2016 Mechanochemical Evidencing the mask effect of graphene oxide: a comparative exfoliation of 2D crystals in deep eutectic solvents ACS study on primary human and murine phagocytic cells Sustain. Chem. Eng. 4 4465–72 Nanoscale 5 11234–47 [267] Abbott A P, Boothby D, Capper G, Davies D L and [289] An S S, Wu S-Y and Hulme J 2015 Current applications of Rasheed R K 2004 Deep eutectic solvents formed between graphene oxide in nanomedicine Int. J. Nanomed. 10 9–24 choline chloride and carboxylic acids: versatile alternatives to [290] Eftekhari A, Shulga Y M, Baskakov S A and Gutsev G L 2018 ionic liquids J. Am. Chem. Soc. 126 9142–7 Graphene oxide membranes for electrochemical energy [268] Zhang Q, De Oliveira Vigier K, Royer S and Jérôme F 2016 storage and conversion Int. J. Hydrog. Energy 43 2307–26 Deep eutectic solvents: syntheses, properties and applications [291] Li F, Jiang X, Zhao J and Zhang S 2015 Graphene oxide: a Chem. Soc. Rev. 41 7108 promising nanomaterial for energy and environmental [269] Shahbaz K, Mjalli F S, Vakili-Nezhaad G, Alnashef I M, applications Nano Energy 16 488–515 Asadov A and Farid M M 2016 Thermogravimetric [292] Wei Y, Zhang Y, Gao X, Ma Z, Wang X and Gao C 2018 measurement of deep eutectic solvents vapor pressure J. Mol. Multilayered graphene oxide membranes for water treatment: Liq. 222 61–6 a review Carbon 139 964–81 [270] Ruß C and König B 2012 Low melting mixtures in organic [293] Joshi R K, Alwarappan S, Yoshimura M, Sahajwalla V and synthesis – an alternative to ionic liquids? Green Chem. Nishina Y 2015 Graphene oxide: the new membrane material 14 2969 Appl. Mater. Today 1 1–12 [271] 2010 Contents Chem. Soc. Rev. 39 3339 [294] Shao J-J, Lv W and Yang Q-H 2014 Self-assembly of graphene [272] Treossi E, Melucci M, Liscio A, Gazzano M, Samorì P and oxide at interfaces Adv. Mater. 26 5586–612 Palermo V 2009 High-contrast visualization of graphene [295] Padmajan Sasikala S, Lim J, Kim I H, Jung H J, Yun T, Han T H oxide on dye-sensitized glass, quartz, and silicon by and Kim S O 2018 Graphene oxide liquid crystals: a frontier fluorescence quenching J. Am. Chem. Soc. 131 15576–7 2D soft material for graphene-based functional materials [273] Boehm H P, Clauss A, Fischer G O and Hofmann U 1962 Chem. Soc. Rev. 47 6013–45 Dünnste kohlenstoff-folien Z. Nat.forsch. 17B 150–3 [296] Brodie B C 1859 On the atomic weight of graphite Phil. Trans. [274] Boehm H P, Clauss A, Fischer G and Hofmann U 1962 Surface R. Soc. 149 249 properties of extremely thin graphite lamellae Proc. of the [297] Staudenmaier L 2016 Verfahren zur darstellung der Fifth Conf. on Carbon (Elsevier) pp 73–80 graphitsaure Ber. Deut. Chem. Ges. 31 1481 [275] Boehm H-P and Stumpp E 2007 Citation errors concerning [298] Marcano D C, Kosynkin D V, Berlin J M, Sinitskii A, Sun Z Z, the first report on exfoliated graphite Carbon 45 1381–3 Slesarev A, Alemany L B, Lu W and Tour J M 2010 Improved [276] Dreyer D R, Park S, Bielawski C W and Ruoff R S 2010 The synthesis of graphene oxide ACS Nano 4 4806–14 chemistry of graphene oxide Chem. Soc. Rev. 39 228–40 [299] Kim S, Zhou S, Hu Y, Acik M, Chabal Y J, Berger C, de [277] Liscio A, Veronese G P, Treossi E, Suriano F, Rossella F, Heer W, Bongiorno A and Riedo E 2012 Room-temperature Bellani V, Rizzoli R, Samorì P and Palermo V 2011 Charge metastability of multilayer graphene oxide films Nat. Mater. transport in graphene–polythiophene blends as studied by 11 544–9 kelvin probe force microscopy and transistor characterization [300] Wei Z et al 2010 Nanoscale tunable reduction of graphene J. Mater. Chem. 21 2924 oxide for graphene electronics Science 328 1373–6 [278] Chen D, Feng H and Li J 2012 Graphene oxide: preparation, [301] Wu X, Sprinkle M, Li X, Ming F, Berger C and de Heer W A functionalization, and electrochemical applications Chem. 2008 Epitaxial-graphene/graphene-oxide junction: an Rev. 112 6027–53 essential step towards epitaxial graphene electronics Phys. [279] Wang Y, Li Z, Wang J, Li J and Lin Y 2001 Graphene and Rev. Lett. 101 026801 graphene oxide: biofunctionalization and applications in [302] Sanchez V C, Jachak A, Hurt R H and Kane A B biotechnology Trends Biotechnol. 29 205–12 2012 Biological interactions of graphene-family [280] Zlotea C, Oumellal Y, Hwang S-J, Ghimbeu C M, de Jongh P E nanomaterials: an interdisciplinary review Chem. Res. and Latroche M 2015 Ultrasmall MgH2 nanoparticles Toxicol. 25 15–34 embedded in an ordered microporous carbon exhibiting [303] De Marzi L, Ottaviano L, Perrozzi F, Nardone M, Santucci S, rapid hydrogen sorption kinetics J. Phys. Chem. C 119 18091– De Lapuente J, Borras M, Treossi E, Palermo V and Poma A 8 2014 Flake size-dependent cyto and genotoxic evaluation of [281] Chung C, Kim Y-K, Shin D, Ryoo S-R, Hong B H and Min D- graphene oxide on in vitro A549, Caco2 and vero cell lines H 2013 Biomedical applications of graphene and graphene J. Biol. Regul. Homeost. Agents 28 281–9 oxide Acc. Chem. Res. 46 2211–24 [304] Ardini M et al 2016 Supramolecular self-assembly of [282] Georgakilas V, Tiwari J N, Kemp K C, Perman J A, graphene oxide and metal nanoparticles into stacked Bourlinos A B, Kim K S and Zboril R 2016 Noncovalent multilayers by means of a multitasking protein ring Nanoscale functionalization of graphene and graphene oxide for energy 8 6739–53 materials, biosensing, catalytic, and biomedical applications [305] Panzavolta S et al 2014 Structural reinforcement and failure Chem. Rev. 116 5464–519 analysis in composite nanofibers of graphene oxide and [283] Nanda S S, Papaefthymiou G C and Yi D K 2015 gelatin Carbon 78 566–77 Functionalization of graphene oxide and its biomedical [306] Pei S, Wei Q, Huang K, Cheng H-M and Ren W 2018 Green applications Crit. Rev. Solid State Mater. Sci. 40 291–315 synthesis of graphene oxide by seconds timescale water [284] Lee J, Kim J, Kim S and Min D-H 2016 Biosensors based on electrolytic oxidation Nat. Commun. 9 145 graphene oxide and its biomedical application Adv. Drug [307] Morimoto N, Kubo T and Nishina Y 2016 Tailoring the Delivery Rev. 105 275–87 oxygen content of graphite and reduced graphene oxide for [285] Liu Y, Dong X and Chen P 2012 Biological and chemical specific applications Sci. Rep. 6 21715 sensors based on graphene materials Chem. Soc. Rev. [308] Tang L-C, Wan Y-J, Yan D, Pei Y-B, Zhao L, Li Y-B, Wu L-B, 41 2283–307 Jiang J-X and Lai G-Q 2013 The effect of graphene dispersion [286] Kucki M et al 2016 Interaction of graphene-related materials on the mechanical properties of graphene/epoxy composites with human intestinal cells: an in vitro approach Nanoscale Carbon 60 16–27 8 8749–60 [309] Wu G and Gao W 2015 GO/rGO as Advanced Materials for [287] Kurapati R, Russier J, Squillaci M A, Treossi E, Ménard- Energy Storage and Conversion, Graphene Oxide (Berlin: Moyon C, Del Rio-Castillo A E, Vazquez E, Samorì P, Springer) pp 97–127 248 2D Mater. 7 (2020) 022001 C Backes et al [310] Chee S Y, Poh H L, Chua C K, Šaněk F, Sofer Z and Pumera M [332] Zhou Y, Bao Q, Tang L A L, Zhong Y and Loh K P 2009 2012 Influence of parent graphite particle size on the Hydrothermal dehydration for the ‘green’ reduction of electrochemistry of thermally reduced graphene oxide Phys. exfoliated graphene oxide to graphene and demonstration of Chem. Chem. Phys. 14 12794 tunable optical limiting properties Chem. Mater. 21 2950–6 [311] Gómez J, Villaro E, Navas A and Recio I 2017 Testing the [333] Bagri A, Mattevi C, Acik M, Chabal Y J, Chhowalla M and influence of the temperature, RH and filler type and content Shenoy V B 2010 Structural evolution during the reduction of on the universal power law for new reduced graphene oxide chemically derived graphene oxide Nat. Chem. 2 581–7 TPU composites Mater. Res. Express 4 105020 [334] Tortello M et al 2016 Effect of thermal annealing on the [312] Singh G, Divakar Botcha V, Sutar D S, Talwar S S, Srinivasa R S heat transfer properties of reduced graphite oxide flakes: a and Major S S 2015 Graphite mediated reduction of graphene nanoscale characterization via scanning thermal microscopy oxide monolayer sheets Carbon 95 843–51 Carbon 109 390–401 [313] Zhang H, Liu Y, Huo S, Briscoe J, Tu W, Picot O T, Rezai A, [335] Cheng M, Yang R, Zhang L, Shi Z, Yang W, Wang D, Xie G, Bilotti E and Peijs T 2017 Filtration effects of graphene Shi D and Zhang G 2012 Restoration of graphene from nanoplatelets in resin infusion processes: problems and graphene oxide by defect repair Carbon 50 2581–7 possible solutions Compos. Sci. Technol. 139 138–45 [336] López V, Sundaram R S, Gómez-Navarro C, Olea D, [314] Ambrosetti G, Grimaldi C, Balberg I, Maeder T, Danani A and Burghard M, Gómez-Herrero J, Zamora F and Kern K 2009 Ryser P 2010 Solution of the tunneling-percolation problem Graphene monolayers: chemical vapor deposition repair in the nanocomposite regime Phys. Rev. B 81 155434 of graphene oxide: a route to highly-conductive graphene [315] Aleksandrzak M, Adamski P, Kukułka W, Zielinska B monolayers (Adv. Mater. 46/2009) Adv. Mater. 21 4683–6 and Mijowska E 2015 Effect of graphene thickness on [337] Acik M, Lee G, Mattevi C, Pirkle A, Wallace R M, photocatalytic activity of TiO2-graphene nanocomposites Chhowalla M, Cho K and Chabal Y 2011 The role of oxygen Appl. Surf. Sci. 331 193–9 during thermal reduction of graphene oxide studied [316] Su C and Loh K P 2013 Carbocatalysts: graphene oxide and its by infrared absorption spectroscopy J. Phys. Chem. C derivatives Acc. Chem. Res. 46 2275–85 115 19761–81 [317] Casco M E et al 2014 CO2 adsorption on crystalline graphitic [338] Su C-Y, Xu Y, Zhang W, Zhao J, Liu A, Tang X, Tsai C-H, nanostructures J. CO2 Util. 5 60–5 Huang Y and Li L-J 2010 Highly efficient restoration of [318] Huang X, Qi X, Boey F and Zhang H 2012 Graphene-based graphitic structure in graphene oxide using alcohol vapors composites Chem. Soc. Rev. 41 666–86 ACS Nano 4 5285–92 [319] Jeon I-Y et al 2013 Facile, scalable synthesis of edge- [339] Wang Z-L, Xu D, Huang Y, Wu Z, Wang L-M and Zhang X-B halogenated graphene nanoplatelets as efficient metal-free 2012 Facile, mild and fast thermal-decomposition reduction eletrocatalysts for oxygen reduction reaction Sci. Rep. of graphene oxide in air and its application in high- 3 1810 performance lithium batteries Chem. Commun. 48 976–8 [320] Jabari Seresht R, Jahanshahi M, Rashidi A and Ghoreyshi A A [340] Chabot V, Higgins D, Yu A, Xiao X, Chen Z and Zhang J 2014 2013 Synthesize and characterization of graphene nanosheets A review of graphene and graphene oxide sponge: material with high surface area and nano-porous structure Appl. Surf. synthesis and applications to energy and the environment Sci. 276 672–81 Energy Environ. Sci. 7 1564 [321] Blanco-Villalba I R J F, Gómez J and Hübner C 2017 [341] Wang M, Duan X, Xu Y and Duan X 2016 Functional three- Improving the processability of graphene nanoplatelets in dimensional graphene/polymer composites ACS Nano polyamide 6 during melt compounding extrusion PPS2017 10 7231–47 (Dresden, Germany) [342] Sun H, Xu Z and Gao C 2013 Multifunctional, ultra-flyweight, [322] Park S, An J, Potts J R, Velamakanni A, Murali S and Ruoff R S synergistically assembled carbon aerogels Adv. Mater. 2011 Hydrazine-reduction of graphite- and graphene oxide 25 2554–60 Carbon 49 3019–23 [343] Xu Y, Sheng K, Li C and Shi G 2010 Self-assembled graphene [323] Bo Z, Shuai X, Mao S, Yang H, Qian J, Chen J, Yan J and Cen K hydrogel via a one-step hydrothermal process ACS Nano 2014 Green preparation of reduced graphene oxide for 4 4324–30 sensing and energy storage applications Sci. Rep. 4 4684 [344] Zhao J, Ren W and Cheng H-M 2012 Graphene sponge for [324] Dreyer D R, Murali S, Zhu Y, Ruoff R S and Bielawski C W efficient and repeatable adsorption and desorption of water 2011 Reduction of graphite oxide using alcohols J. Mater. contaminations J. Mater. Chem. 22 20197 Chem. 21 3443–7 [345] Cong H-P, Ren X-C, Wang P and Yu S-H 2012 Macroscopic [325] Agharkar M, Kochrekar S, Hidouri S and Azeez M A 2014 multifunctional graphene-based hydrogels and aerogels Trends in green reduction of graphene oxides, issues and by a metal ion induced self-assembly process ACS Nano challenges: a review Mater. Res. Bull. 59 323–8 6 2693–703 [326] Sarkar S and Basak D 2013 The reduction of graphene oxide [346] Qiu L, Liu J Z, Chang S L Y, Wu Y and Li D 2012 Biomimetic by zinc powder to produce a zinc oxide-reduced graphene superelastic graphene-based cellular monoliths Nat. oxide hybrid and its superior photocatalytic activity Chem. Commun. 3 1241 Phys. Lett. 561–2 125–30 [347] Xia Z, Wei D, Anitowska E, Bellani V, Ortolani L, Morandi V, [327] Chua C K and Pumera M 2014 Chemical reduction of Gazzano M, Zanelli A, Borini S and Palermo V 2015 graphene oxide: a synthetic chemistry viewpoint Chem. Soc. Electrochemically exfoliated graphene oxide/iron oxide Rev. 43 291–312 composite foams for lithium storage, produced by [328] Pei S, Zhao J, Du J, Ren W and Cheng H-M 2010 Direct simultaneous graphene reduction and Fe(OH)3 condensation reduction of graphene oxide films into highly conductive Carbon 84 254–62 and flexible graphene films by hydrohalic acids Carbon [348] Xia Z Y et al 2014 Synergic exfoliation of graphene with 48 4466–74 organic molecules and inorganic ions for the electrochemical [329] Zhao J, Pei S, Ren W, Gao L and Cheng H-M 2010 Efficient production of flexible electrodes ChemPlusChem 79 439–46 preparation of large-area graphene oxide sheets for [349] Vickery J L, Patil A J and Mann S 2009 Fabrication of transparent conductive films ACS Nano 4 5245–52 graphene-polymer nanocomposites with higher-order three- [330] Storm M M, Johnsen R E and Norby P 2016 In situ x-ray dimensional architectures Adv. Mater. 21 2180–4 powder diffraction studies of the synthesis of graphene oxide [350] Nardecchia S, Carriazo D, Ferrer M L, Gutiérrez M C and del and formation of reduced graphene oxide J. Solid State Chem. Monte F 2013 Three dimensional macroporous architectures 240 49–54 and aerogels built of carbon nanotubes and/or graphene: [331] Hu K, Xie X, Szkopek T and Cerruti M 2016 Understanding synthesis and applications Chem. Soc. Rev. 42 794–830 hydrothermally reduced graphene oxide hydrogels: from [351] Mao J, Iocozzia J, Huang J, Meng K, Lai Y and Lin Z 2018 reaction products to hydrogel properties Chem. Mater. Graphene aerogels for efficient energy storage and conversion 28 1756–68 Energy Environ. Sci. 11 772–99 249 2D Mater. 7 (2020) 022001 C Backes et al [352] Botas C, Carriazo D, Singh G and Rojo T 2015 Sn–and fully exfoliated individual graphene flakes in low boiling SnO2–graphene flexible foams suitable as binder-free anodes point solvents Soft Matter 8 7882–7 for lithium ion batteries J. Mater. Chem. A 3 13402–10 [373] Pénicaud A and Drummond C 2012 Deconstructing graphite: [353] Quesnel E et al 2015 Graphene-based technologies for energy graphenide solutions Acc. Chem. Res. 46 129–37 applications, challenges and perspectives 2D Mater. 2 030204 [374] Kelly K F and Billups W E 2013 Synthesis of soluble graphite [354] Gutiérrez M C, Ferrer M L and del Monte F 2008 Ice- and graphene Acc. Chem. Res. 46 4–13 templated materials: sophisticated structures exhibiting [375] Vecera P, Holzwarth J, Edelthalhammer K F, Mundloch U, enhanced functionalities obtained after unidirectional Peterlik H, Hauke F and Hirsch A 2016 Solvent-driven freezing and ice-segregation-induced self-assembly Chem. electron trapping and mass transport in reduced graphites to Mater. 20 634–8 access perfect graphene Nat. Commun. 7 12411 [355] Botas C, Carriazo D, Zhang W, Rojo T and Singh G 2016 [376] Hodge S A, Tay H H, Anthony D B, Menzel R, Buckley D J, Silicon-reduced graphene oxide self-standing composites Cullen P L, Skipper N T, Howard C A and Shaffer M S P 2014 suitable as binder-free anodes for lithium-ion batteries ACS Probing the charging mechanisms of carbon nanomaterial Appl. Mater. Interfaces 8 28800–8 polyelectrolytes Faraday Discuss. 172 311–25 [356] Lee J K, Smith K B, Hayner C M and Kung H H 2010 Silicon [377] Bepete G, Anglaret E, Ortolani L, Morandi V, Huang K, nanoparticles-graphene paper composites for Li ion battery Pénicaud A and Drummond C 2016 Surfactant-free single- anodes Chem. Commun. 46 2025 layer graphene in water Nat. Chem. 9 347 [357] Zhou X, Yin Y-X, Wan L-J and Guo Y-G 2012 Facile synthesis [378] Eda G, Yamaguchi H, Voiry D, Fujita T, Chen M and of silicon nanoparticles inserted into graphene sheets as Chhowalla M 2011 Photoluminescence from chemically improved anode materials for lithium-ion batteries Chem. exfoliated MoS2 Nano Lett. 11 5111–6 Commun. 48 2198 [379] Knirsch K C et al 2015 Basal-plane functionalization of [358] Botas C, Álvarez P, Blanco C, Santamaría R, Granda M, chemically exfoliated molybdenum disulfide by diazonium Gutiérrez M D, Rodríguez-Reinoso F and Menéndez R 2013 salts ACS Nano 9 6018–30 Critical temperatures in the synthesis of graphene-like [380] Voiry D, Goswami A, Kappera R, SilvaCecilia de materials by thermal exfoliation–reduction of graphite oxide Carvalho Castro E, Kaplan D, Fujita T, Chen M, Asefa T Carbon 52 476–85 andChhowalla M 2015 Covalent functionalization of [359] Li G, Sun J, Hou W, Jiang S, Huang Y and Geng J 2016 Three- monolayered transition metal dichalcogenides by phase dimensional porous carbon composites containing high engineering Nat. Chem. 7 45–9 sulfur nanoparticle content for high-performance lithium– [381] Pagona G, Bittencourt C, Arenal R and Tagmatarchis N 2015 sulfur batteries Nat. Commun. 7 10601 Exfoliated semiconducting pure 2H-MoS2 and 2H-WS2 [360] Ajuria J, Arnaiz M, Botas C, Carriazo D, Mysyk R, Rojo T, assisted by chlorosulfonic acid Chem. Commun. 51 12950–3 Talyzin A V and Goikolea E 2017 Graphene-based lithium ion [382] Yang S, Lohe M R, Müllen K and Feng X 2016 New-generation capacitor with high gravimetric energy and power densities J. graphene from electrochemical approaches: production and Power Sources 363 422–7 applications Adv. Mater. 28 6213–21 [361] Nair R R, Wu H A, Jayaram P N, Grigorieva I V and Geim A K [383] Maccaferri G, Zanardi C, Xia Z, Kovtun A, Liscio A, Terzi F, 2012 Unimpeded permeation of water through helium-leak- Palermo V and Seeber R 2017 Systematic study of the tight graphene-based membranes Science 335 442–4 correlation between surface chemistry, conductivity and [362] Boretti A, Al-Zubaidy S, Vaclavikova M, Al-Abri M, electrocatalytic properties of graphene oxide nanosheets Castelletto S and Mikhalovsky S 2018 Outlook for graphene- Carbon 120 165–75 based desalination membranes npj Clean Water 1 5 [384] Su C-Y, Lu A-Y, Xu Y, Chen F-R, Khlobystov A N and [363] Liu G, Jin W and Xu N 2015 Graphene-based membranes Li L-J 2011 High-quality thin graphene films from fast Chem. Soc. Rev. 44 5016–30 electrochemical exfoliation ACS Nano 5 2332–9 [364] Sun P, Wang K and Zhu H 2016 Recent developments in [385] Parvez K, Li R, Puniredd S R, Hernandez Y, Hinkel F, Wang S, graphene-based membranes: structure, mass-transport Feng X and Müllen K 2013 Electrochemically exfoliated mechanism and potential applications Adv. Mater. 28 2287– graphene as solution-processable, highly conductive 310 electrodes for organic electronics ACS Nano 7 3598–606 [365] An D, Yang L, Wang T-J and Liu B 2016 Separation [386] Rao K S, Sentilnathan J, Cho H-W, Wu J-J and Yoshimura M performance of graphene oxide membrane in aqueous 2014 Soft processing of graphene nanosheets by glycine- solution Ind. Eng. Chem. Res. 55 4803–10 bisulfate ionic-complex-assisted electrochemical exfoliation [366] Morelos-Gomez A et al 2017 Effective NaCl and dye rejection of graphite for reduction catalysis Adv. Funct. Mater. of hybrid graphene oxide/graphene layered membranes Nat. 25 298–305 Nanotechnol. 12 1083 [387] Parvez K, Wu Z-S, Li R, Liu X, Graf R, Feng X and Müllen K [367] Kim H W et al 2013 Selective gas transport through few- 2014 Exfoliation of graphite into graphene in aqueous layered graphene and graphene oxide membranes Science solutions of inorganic salts J. Am. Chem. Soc. 136 6083–91 342 91–5 [388] Yang S, Brüller S, Wu Z-S, Liu Z, Parvez K, Dong R, [368] Englert J M, Knirsch K C, Dotzer C, Butz B, Hauke F, Richard F, Samorì P, Feng X and Müllen K 2015 Organic Spiecker E and Hirsch A 2012 Functionalization of graphene radical-assisted electrochemical exfoliation for the scalable by electrophilic alkylation of reduced graphite Chem. production of high-quality graphene J. Am. Chem. Soc. Commun. 48 5025 137 13927–32 [369] Schäfer R A, Englert J M, Wehrfritz P, Bauer W, Hauke F, [389] Wang J, Manga K K, Bao Q and Loh K P 2011 High- Seyller T and Hirsch A 2013 On the way to graphane— yield synthesis of few-layer graphene flakes through pronounced fluorescence of polyhydrogenated graphene electrochemical expansion of graphite in propylene carbonate Angew. Chem, Int. Ed. Engl. 52 754–7 electrolyte J. Am. Chem. Soc. 133 8888–91 [370] Vecera P, Edelthalhammer K, Hauke F and Hirsch A 2014 [390] Cabana J, Monconduit L, Larcher D and Palacín M R 2010 Reductive arylation of graphene: Insights into a reversible Beyond intercalation-based li-ion batteries: the state of the carbon allotrope functionalization reaction Phys. Status Solidi art and challenges of electrode materials reacting through b 251 2536–40 conversion reactions Adv. Mater. 22 E170–92 [371] Abellán G, Schirowski M, Edelthalhammer K, Fickert M, [391] Ohzuku T 1993 Formation of lithium-graphite intercalation Hauke F and Hirsch A 2017 Unifying aspects of the reductive compounds in nonaqueous electrolytes and their application covalnet graphene functionalization J. Am. Chem. Soc. as a negative electrode for a lithium ion (shuttlecock) cell J. 139 5175  −  82 Electrochem. Soc. 140 2490 [372] Catheline A, Ortolani L, Morandi V, Melle-Franco M, [392] Tarascon J M and Armand M 2001 Issues and challenges Drummond C, Zakri C and Penicaud A 2012 Solutions of facing rechargeable lithium batteries Nature 414 359–67 250 2D Mater. 7 (2020) 022001 C Backes et al [393] Weller T E, Ellerby M, Saxena S S, Smith R P and Skipper N T [416] Zhou X-F, Dong X, Oganov A R, Zhu Q, Tian Y and Wang H-T 2005 Superconductivity in the intercalated graphite 2014 Semimetallic two-dimensional boron allotrope with compounds C6Yb and C6Ca Nat. Phys. 1 39–41 massless dirac fermions Phys. Rev. Lett. 112 085502 [394] Zhong Y L and Swager T M 2012 Enhanced electrochemical [417] Revard B C, Tipton W W, Yesypenko A and Hennig R G 2016 expansion of graphite for in situ electrochemical Grand-canonical evolutionary algorithm for the prediction functionalization J. Am. Chem. Soc. 134 17896–9 of two-dimensional materials Phys. Rev. B 93 054117 [395] Abdelkader A M, Kinloch I A and Dryfe R A W 2014 [418] Luo W, Ma Y, Gong X and Xiang H 2014 Prediction of silicon- Continuous electrochemical exfoliation of micrometer-sized based layered structures for optoelectronic applications J. Am. graphene using synergistic ion intercalations and organic Chem. Soc. 136 15992–7 solvents ACS Appl. Mater. Interfaces 6 1632–9 [419] Luo X, Yang J, Liu H, Wu X, Wang Y, Ma Y, Wei S-H, Gong X [396] Yu P 1999 Determination of the lithium ion diffusion and Xiang H 2011 Predicting two-dimensional boron–carbon coefficient in graphit J. Electrochem. Soc. 146 8 compounds by the global optimization method J. Am. Chem. [397] Abdelkader A M, Cooper A J, Dryfe R A W and Kinloch I A Soc. 133 16285–90 2015 How to get between the sheets: a review of recent works [420] Yang J-H, Zhang Y, Yin W-J, Gong X G, Yakobson B I and on the electrochemical exfoliation of graphene materials from Wei S-H 2016 Two-dimensional SiS layers with promising bulk graphite Nanoscale 7 6944–56 electronic and optoelectronic properties: theoretical [398] Abdelkader A M, Patten H V, Li Z, Chen Y and Kinloch I A prediction Nano Lett. 16 1110–7 2015 Electrochemical exfoliation of graphite in quaternary [421] Wu X, Dai J, Zhao Y, Zhuo Z, Yang J and Zeng X C 2012 Two- ammonium-based deep eutectic solvents: a route for the mass dimensional boron monolayer sheets ACS Nano 6 7443–53 production of graphane Nanoscale 7 11386–92 [422] Zhang Z, Mannix A J, Hu Z, Kiraly B, Guisinger N P, [399] Kamali A R, Feighan J and Fray D J 2016 Towards large scale Hersam M C and Yakobson B I 2016 Substrate-induced preparation of graphene in molten salts and its use in the nanoscale undulations of borophene on silver Nano Lett. fabrication of highly toughened alumina ceramics Faraday 16 6622–7 Discuss. 190 451–70 [423] Zhang Z, Penev E S and Yakobson B I 2016 Two-dimensional [400] Kamali A R and Fray D J 2015 Large-scale preparation of materials: polyphony in B flat Nat. Chem. 8 525–7 graphene by high temperature insertion of hydrogen into [424] Hassoun J et al 2014 An advanced lithium-ion battery based graphite Nanoscale 7 11310–20 on a graphene anode and a lithium iron phosphate cathode [401] Kim H-K, Kamali A R, Roh K C, Kim K-B and Fray D J 2016 Nano Lett. 14 4901–6 Dual coexisting interconnected graphene nanostructures [425] Kang J, Wells S A, Wood J D, Lee J-H, Liu X, Ryder C R, Zhu J, for high performance supercapacitor applications Energy Guest J R, Husko C A and Hersam M C 2016 Stable aqueous Environ. Sci. 9 2249–56 dispersions of optically and electronically active phosphorene [402] Kamali A R 2016 Eco-friendly production of high quality Proc. Natl Acad. Sci. 113 11688–93 low cost graphene and its application in lithium ion batteries [426] Bunch J S, Yaish Y, Brink M, Bolotin K and McEuen P L 2005 Green Chem. 18 1952–64 Coulomb oscillations and hall effect in quasi-2D graphite [403] Zhuang H L and Hennig R G 2014 Computational discovery, quantum dots Nano Lett. 5 287–90 characterization, and design of single-layer materials J. Miner. [427] Hasan T, Torrisi F, Sun Z, Popa D, Nicolosi V, Privitera G, Met. Mater. Soc. 66 366–74 Bonaccorso F and Ferrari A C 2010 Solution-phase [404] Gould T, Lebègue S, Björkman T and Dobson J F 2016 2D exfoliation of graphite for ultrafast photonics Phys. Status Structures Beyond Graphene, Semiconductors and Semimetals Solidi b 247 2953–7 (New York: Elsevier) pp 1–33 [428] Liu L, Shen Z, Yi M, Zhang X and Ma S 2014 A green, rapid [405] Lebègue S, Björkman T, Klintenberg M, Nieminen R M and and size-controlled production of high-quality graphene Eriksson O 2013 Two-dimensional materials from data sheets by hydrodynamic forces RSC Adv. 4 36464 filtering and ab initio calculations Phys. Rev. X 3 031002 [429] Pattammattel A and Kumar C V 2015 Kitchen chemistry [406] Ashton M, Paul J, Sinnott S B and Hennig R G 2017 Topology- 101: multigram production of high quality biographene in a scaling identification of layered solids and stable exfoliated 2D blender with edible proteins Adv. Funct. Mater. 25 7088–98 materials Phys. Rev. Lett. 118 106101 [430] Yi M and Shen Z 2014 Kitchen blender for producing high- [407] Mounet N et al 2018 Two-dimensional materials from high- quality few-layer graphene Carbon 78 622–6 throughput computational exfoliation of experimentally [431] Damm C, Nacken T J and Peukert W 2015 Quantitative known compounds Nat. Nanotechnol. 13 246–52 evaluation of delamination of graphite by wet media milling [408] Zhuang H L, Kent P R C and Hennig R G 2016 Strong Carbon 81 284–94 anisotropy and magnetostriction in the two-dimensional [432] Fabbro A et al 2016 Graphene-based interfaces do not alter Stoner ferromagnet Fe3GeTe2 Phys. Rev. B 93 054429 target nerve cells ACS Nano 10 615–23 [409] Zhang S, Yan Z, Li Y, Chen Z and Zeng H 2015 Atomically thin [433] Ibrahem M A, Lan T-W, Huang J K, Chen Y-Y, Wei K-H, Li L-J arsenene and antimonene: semimetal-semiconductor and and Chu C W 2013 High quantity and quality few-layers indirect-direct band-gap transitions Angew. Chem, Int. Ed. transition metal disulfide nanosheets from wet-milling Engl. 127 3155–8 exfoliation RSC Adv. 3 13193 [410] Schusteritsch G, Uhrin M and Pickard C J 2016 Single-layered [434] Knieke C, Berger A, Voigt M, Taylor R N K, Röhrl J and hittorf’s phosphorus: a wide-bandgap high mobility 2D Peukert W 2010 Scalable production of graphene sheets by material Nano Lett. 16 2975–80 mechanical delamination Carbon 48 3196–204 [411] Cahangirov S, Topsakal M, Aktürk E, Şahin H and Ciraci S [435] Yao Y, Lin Z, Li Z, Song X, Moon K-S and Wong C-P 2012 2009 Two- and one-dimensional honeycomb structures of Large-scale production of two-dimensional nanosheets J. silicon and germanium Phys. Rev. Lett. 102 236804 Mater. Chem. 22 13494 [412] Rasmussen F A and Thygesen K S 2015 Computational 2D [436] Zhao W, Fang M, Wu F, Wu H, Wang L and Chen G 2010 materials database: electronic structure of transition-metal Preparation of graphene by exfoliation of graphite using wet dichalcogenides and oxides J. Phys. Chem. C 119 13169–83 ball milling J. Mater. Chem. 20 5817 [413] Ataca C, Şahin H and Ciraci S 2012 Stable, single-layer MX2 [437] Shen Z, Li J, Yi M, Zhang X and Ma S 2011 Preparation of transition-metal oxides and dichalcogenides in a honeycomb- graphene by jet cavitation Nanotechnology 22 365306 like structure J. Phys. Chem. C 116 8983–99 [438] Yi M, Li J, Shen Z, Zhang X and Ma S 2011 Morphology and [414] Miró P, Audiffred M and Heine T 2014 An atlas of two- structure of mono- and few-layer graphene produced by jet dimensional materials Chem. Soc. Rev. 43 6537–54 cavitation Appl. Phys. Lett. 99 123112 [415] Bahmann S and Kortus J 2013 EVO—evolutionary algorithm [439] Yi M, Shen Z, Zhang W, Zhu J, Liu L, Liang S, Zhang X and for crystal structure prediction Comput. Phys. Commun. Ma S 2013 Hydrodynamics-assisted scalable production of 184 1618–25 boron nitride nanosheets and their application in improving 251 2D Mater. 7 (2020) 022001 C Backes et al oxygen-atom erosion resistance of polymeric composites [463] Coombs D H and Watts N R M 1985 Generating sucrose Nanoscale 5 10660 gradients in three minutes by tilted tube rotation Anal. [440] Del Rio Castillo A E et al 2018 High-yield production of 2D Biochem. 148 254–9 crystals by wet-jet milling Mater. Horiz. 5 890–904 [464] Brakke M K and Daly J M 1965 Density-gradient [441] Bonaccorso F, Zerbetto M, Ferrari A C and Amendola V 2013 centrifugation: non-ideal sedimentation and the interaction Sorting nanoparticles by centrifugal fields in clean media J. of major and minor components Science 148 387–9 Phys. Chem. C 117 13217–29 [465] Cohen R and Claverie J-M 1975 Sedimentation of [442] O’Connell M J et al 2002 Band gap fluorescence from generalized systems of interacting particles. II. Active enzyme individual single-walled carbon nanotubes Science 297 593–6 centrifugation? theory and extensions of its validity range [443] Hasan T, Tan P H, Bonaccorso F, Rozhin A G, Scardaci V, Biopolymers 14 1701–16 Milne W I and Ferrari A C 2008 Polymer-assisted isolation of [466] Smale G and Sasse J 1992 RNA isolation from cartilage using single wall carbon nanotubes in organic solvents for optical- density gradient centrifugation in cesium trifluoroacetate: an quality nanotube-polymer composites J. Phys. Chem. C RNA preparation technique effective in the presence of high 112 20227–32 proteoglycan content Anal. Biochem. 203 352–6 [444] Maragò O M et al 2008 Optical trapping of carbon nanotubes [467] Arnold M S, Green A A, Hulvat J F, Stupp S I and Hersam M C Physica E 40 2347–51 2006 Sorting carbon nanotubes by electronic structure using [445] Tan P H, Hasan T, Bonaccorso F, Scardaci V, Rozhin A G, density differentiation Nat. Nanotechnol. 1 60–5 Milne W I and Ferrari A C 2008 Optical properties of [468] Arnold M S, Stupp S I and Hersam M C 2005 Enrichment nanotube bundles by photoluminescence excitation and of single-walled carbon nanotubes by diameter in density absorption spectroscopy Physica E 40 2352–9 gradients Nano Lett. 5 713–8 [446] Maragó O M et al 2010 Brownian motion of graphene ACS [469] Bonaccorso F, Hasan T, Tan P H, Sciascia C, Privitera G, Nano 4 7515–23 Di Marco G, Gucciardi P G and Ferrari A C 2010 Density [447] Torrisi F et al 2012 Inkjet-printed graphene electronics ACS gradient ultracentrifugation of nanotubes: interplay of Nano 6 2992–3006 bundling and surfactants encapsulation J. Phys. Chem. C [448] Capasso A, Matteocci F, Najafi L, Prato M, Buha J, Cinà L, 114 17267–85 Pellegrini V, Carlo A D and Bonaccorso F 2016 Few-layer [470] Vieira O V, Laranjinha J A N, Madeira V M C and Almeida L M MoS2 flakes as active buffer layer for stable perovskite solar 2014 Rapid isolation of low density lipoproteins in a cells Adv. Energy Mater. 6 1600920 concentrated fraction free from water-soluble plasma [449] Li D, Del Rio Castillo A E, Jussila H, Ye G, Ren Z, Bai J, Chen X, antioxidants J. Lipid Res. 37 2715–21 Lipsanen H, Sun Z and Bonaccorso F 2016 Black phosphorus [471] Li S, Zhu F, Meng F, Li H, Wang L, Zhao J, Yue Q, Liu J and polycarbonate polymer composite for pulsed fibre lasers Appl. Jia J 2013 Separation of graphene oxide by density gradient Mater. Today 4 17–23 centrifugation and study on their morphology-dependent [450] Morozov S V, Novoselov K S, Katsnelson M I, Schedin F, electrochemical properties J. Electroanal. Chem. 703 135–45 Ponomarenko L A, Jiang D and Geim A K 2006 Strong [472] Kang J, Seo J-W T, Alducin D, Ponce A, Yacaman M J and suppression of weak localization in graphene Phys. Rev. Lett. Hersam M C 2014 Thickness sorting of two-dimensional 97 016801 transition metal dichalcogenides via copolymer-assisted [451] Khattab I S, Bandarkar F, Fakhree M A A and Jouyban A 2012 density gradient ultracentrifugation Nat. Commun. 5 5478 Density, viscosity, and surface tension of water  +  ethanol [473] Zhu J, Kang J, Kang J, Jariwala D, Wood J D, Seo J-W T, mixtures from 293 to 323 K Korean J. Chem. Eng. 29 812–7 Chen K-S, Marks T J and Hersam M C 2015 Solution- [452] Gholamvand Z, McAteer D, Harvey A, Backes C and processed dielectrics based on thickness-sorted two- Coleman J N 2016 Electrochemical applications of two- dimensional hexagonal boron nitride nanosheets Nano Lett. dimensional nanosheets: the effect of nanosheet length and 15 7029–36 thickness Chem. Mater. 28 2641–51 [474] Kang J, Sangwan V K, Wood J D, Liu X, Balla I, Lam D and [453] Vega-Mayoral V, Backes C, Hanlon D, Khan U, Hersam M C 2016 Layer-by-layer sorting of rhenium Gholamvand Z, O’Brien M, Duesberg G S, Gadermaier C and disulfide via high-density isopycnic density gradient Coleman J N 2016 Photoluminescence from liquid-exfoliated ultracentrifugation Nano Lett. 16 7216–23 WS2 monomers in poly(vinyl alcohol) polymer composites [475] Brakke M K 1953 Zonal separations by density-gradient Adv. Funct. Mater. 26 1028–39 centrifugation Arch. Biochem. Biophys. 45 275–90 [454] Sun H et al 2017 How much does size really matter? Exploring [476] Sun X, Luo D, Liu J and Evans D G 2010 Monodisperse the limits of graphene as Li ion battery anode material Solid chemically modified graphene obtained by density gradient State Commun. 251 88–93 ultracentrifugal rate separation ACS Nano 4 3381–9 [455] Woomer A H, Farnsworth T W, Hu J, Wells R A, Donley C L [477] Huebler A, Hahn U, Beier W, Lasch N and Fischer T 2002 High and Warren S C 2015 Phosphorene: synthesis, scale-up, and volume printing technologies for the production of polymer quantitative optical spectroscopy ACS Nano 9 8869–84 electronic structures 2nd Int. IEEE Conf. on Polymers and [456] Majekodunmi S O 2015 A review on centrifugation in the Adhesives in Microelectronics and Photonics. POLYTRONIC pharmaceutical industry Am. J. Biomed. Eng. 5 67–78 2002. Conf. Proc. (Cat. No.02EX599) (IEEE) [457] Williams J W, Van Holde K E, Baldwin R L and Fujita H [478] Kawahara T 2003 J. Sol-Gel Sci. Technol. 27 301–7 1958 The theory of sedimentation analysis Chem. Rev. [479] Bergsmann F K M, Bauer G, Domnick R and Walter H 2003 58 715–44 The 46th Annual Technical Conf.—Society of the Vacuum [458] Bildirici L and Rickwood D 2000 Fractionation of Coaters (San Francisco, USA, 4–8 May 2003) p 566 differentiating cells using density perturbation J. Immunol. [480] Maaninen A, Tuomikoski M, Kivimäki L, Kololuoma T, Methods 240 93–9 Välimäki M, Leinonen M and Känsäkoski M 2005 Roll to roll [459] Patel D and Rickwood D 1995 Optimization of conditions for fabrication technologies for optoelectronic and electronic specific binding of antibody-coated beads to cells J. Immunol. devices and sensors Proc. SPIE Methods 184 71–80 [481] Sirringhaus H 2006 Device physics and reliability of solution [460] Cacace M G, Landau E M and Ramsden J J 1997 The processed organic TFTs Flexidis Training Workshop (Stuttgart, Hofmeister series: salt and solvent effects on interfacial Germany, 2006) phenomena Q. Rev. Biophys. 30 241–77 [482] Schrödner M, Schultheis K, Stohn R, Schache H, [461] Collins K D and Washabaugh M W 1985 The Hofmeister Blankenburg L and Roth H 2006 Technologies for Polymer effect and the behaviour of water at interfaces Q. Rev. Biophys. Electronics (Frankfurt, Germany, 2006) 18 323 [483] Blayo A and Pineaux B 2005 Printing processes and their [462] Kaminsky M 1957 Ion-solvent interaction and the viscosity of potential for RFID printing Proc. of the 2005 Joint Conf. on strong-electrolyte solutions Discuss. Faraday Soc. 24 171 Smart Objects and Ambient Intelligence Innovative Context- 252 2D Mater. 7 (2020) 022001 C Backes et al Aware Services: Usages and Technologies—sOc-EUSAI ‘05 [509] Perelaer J, de Gans B J and Schubert U S 2006 Ink-jet printing (ACM Press) and microwave sintering of conductive silver tracks Adv. [484] Jabbour G E, Radspinner R and Peyghambarian N 2001 Mater. 18 2101–4 Screen printing for the fabrication of organic light-emitting [510] de Gennes P G 1985 Wetting: statics and dynamics Rev. Mod. devices IEEE J. Sel. Top. Quantum Electron. 7 769–73 Phys. 57 827–63 [485] Park J and Moon J 2006 Control of colloidal particle deposit [511] Saunders R E and Derby B 2014 Inkjet printing biomaterials patterns within picoliter droplets ejected by ink-jet printing for tissue engineering: bioprinting Int. Mater. Rev. 59 430–48 Langmuir 22 3506–13 [512] Baker J, Deganello D, Gethin D T and Watson T M 2014 [486] Pekarovicova A and Husovska V 2016 Printing Ink Flexographic printing of graphene nanoplatelet ink to Formulations, Printing on Polymers (New York: Elsevier) pp replace platinum as counter electrode catalyst in flexible dye 41–55 sensitised solar cell Mater. Res. Innov. 18 86–90 [487] The colloidal domain: where physics, chemistry, biology, and [513] Secor E B, Lim S, Zhang H, Frisbie C D, Francis L F and technology meet Choice Rev. 37 0334 Hersam M C 2014 Gravure printing of graphene for large- [488] Viswanath D S, Ghosh T K, Prasad D H L, Dutt N V K and area flexible electronics Adv. Mater. 26 4533–8 Rani K Y Correlations And Estimation of Pure Liquid Viscosity [514] Yamazaki K 1993 Report on 1993 TAPPI coating conference (Viscosity of Liquids) (Berlin: Springer) pp 135–405 Japan Tappi J. 47 1213–22 [489] Willenbacher N and Georgieva K 2013 Rheology of Disperse [515] Kim Y-J, Yoo H, Lee C-H, Park J B, Baek H, Kim M and Systems, Product Design and Engineering (New York: Wiley) Yi G-C 2012 Position- and morphology-controlled ZnO pp 7–49 nanostructures grown on graphene layers Adv. Mater. [490] Hiemenz P C and Rajagopalan R 1997 Principles of Colloid 24 5565–9 and Surface Chemistry (Boca Raton, FL: CRC Press) [516] Hyun W J, Secor E B, Rojas G A, Hersam M C, Francis L F [491] Garg J et al 2008 Enhanced thermal conductivity and viscosity and Frisbie C D 2015 All-printed, foldable organic thin-film of copper nanoparticles in ethylene glycol nanofluid J. Appl. transistors on glassine paper Adv. Mater. 27 7058–64 Phys. 103 074301 [517] Liu B and Wang S 2012 Preparation of patterned biological [492] Mahbubul I M, Saidur R and Amalina M A 2012 Latest film for nanolithography J. Biosens. Bioelectron. 01 developments on the viscosity of nanofluids Int. J. Heat Mass [518] Qian M, Feng T, Ding H, Lin L, Li H, Chen Y and Sun Z 2009 Transfer 55 874–85 Electron field emission from screen-printed graphene films [493] Mishra P C, Mukherjee S, Nayak S K and Panda A 2014 A brief Nanotechnology 20 425702 review on viscosity of nanofluids Int. Nano Lett. 4 109–20 [519] Zhang D W, Li X D, Li H B, Chen S, Sun Z, Yin X J and [494] Bose S, Keller S S, Alstrøm T S, Boisen A and Almdal K 2013 Huang S M 2011 Graphene-based counter electrode for dye- Process optimization of ultrasonic spray coating of polymer sensitized solar cells Carbon 49 5382–8 films Langmuir 29 6911–9 [520] Wróblewski G and Janczak D 2012 Screen printed, [495] Xing L-L, Glass J E and Fernando R H 1997 Spray application transparent, and flexible electrodes based on graphene of waterborne coatings ACS Symp. Series (American nanoplatelet pastes Proc. SPIE Chemical Society) pp 265–95 [521] Xu Y, Schwab M G, Strudwick A J, Hennig I, Feng X, Wu Z and [496] Owen J W, Azarova N A, Loth M A, Paradinas M, Coll M, Müllen K 2013 Screen-printable thin film supercapacitor Ocal C, Anthony J E and Jurchescu O D 2011 Effect of device utilizing graphene/polyaniline inks Adv. Energy Mater. processing parameters on performance of spray-deposited 3 1035–40 organic thin-film transistors J. Nanotechnol. 2011 1–6 [522] Arapov K, Rubingh E, Abbel R, Laven J, de With G and [497] Perfetti G, Alphazan T, van Hee P, Wildeboer W J and Friedrich H 2015 Conductive screen printing inks by gelation Meesters G M H 2011 Relation between surface roughness of graphene dispersions Adv. Funct. Mater. 26 586–93 of free films and process parameters in spray coating Eur. J. [523] Whistler R and BeMiller J 1993 Industrial Gums : Pharm. Sci. 42 262–72 Polysaccharides and Their Derivatives (New York: Academic) [498] Singh M, Haverinen H M, Dhagat P and Jabbour G E 2010 pp 558–9 Inkjet printing-process and its applications Adv. Mater. [524] Mezger T G 2006 The Rheology Handbook: For Users of 22 673–85 Rotational and Oscillatory Rheometers (Vincentz Network) [499] Bergeron V, Bonn D, Martin J Y and Vovelle L 2000 Nature [525] Benchabane A and Bekkour K 2008 Rheological properties of 405 772–5 carboxymethyl cellulose (CMC) solutions Colloid Polym. Sci. [500] Squire L C 1967 An Introduction to Fluid Dynamics ed 286 1173–80 G K Batchelor (Cambridge: Cambridge University Press) [526] Nijenhuis K, McKinley G, Spiegelberg S, Barnes H, Aksel N, 616p (Figures. 15s J. R. Aeronaut. Soc. 71 801–2) Heymann L and Odell J 2007 Non-Newtonian Flows, Springer [501] de Gans B-J, Kazancioglu E, Meyer W and Schubert U S Handbook of Experimental Fluid Mechanics (Berlin: Springer) 2004 Ink-jet printing polymers and polymer libraries using pp 619–743 micropipettes Macromol. Rapid Commun. 25 292–6 [527] Irgens F 2014 Rheology and Non-Newtonian Fluids (Berlin: [502] Dong H, Carr W W and Morris J F 2006 An experimental Springer) study of drop-on-demand drop formation Phys. Fluids [528] deButts E H, Hudy J A and Elliott J H 1957 Rheology of 18 072102 sodium carboxymethylcellulose solutions Ind. Eng. Chem. [503] Fromm J E 1984 Numerical calculation of the fluid dynamics 49 94–8 of drop-on-demand jets IBM J. Res. Dev. 28 322–33 [529] Elliot J H and Ganz A J 1974 Some rheological properties of [504] Hsiao W-K, Hoath S D, Martin G D and Hutchings I M 2009 sodium carboxymethylcellulose solutions and gels Rheol. Acta Ink jet printing for direct mask deposition in printed circuit 13 670–4 board fabrication J. Imaging Sci. Technol. 53 050304 [530] Robaeys P et al 2014 Enhanced performance of [505] Jang D, Kim D and Moon J 2009 Influence of fluid physical polymer:fullerene bulk heterojunction solar cells upon properties on ink-jet printability Langmuir 25 2629–35 graphene addition Appl. Phys. Lett. 105 083306 [506] Jung S and Hutchings I M 2012 The impact and spreading of a [531] Hu L, Kim H S, Lee J-Y, Peumans P and Cui Y 2010 Scalable small liquid drop on a non-porous substrate over an extended coating and properties of transparent, flexible, silver time scale Soft Matter 8 2686 nanowire electrodes ACS Nano 4 2955–63 [507] Shin P, Sung J and Lee M H 2011 Control of droplet [532] Dan B, Irvin G C and Pasquali M 2009 Continuous and formation for low viscosity fluid by double waveforms scalable fabrication of transparent conducting carbon applied to a piezoelectric inkjet nozzle Microelectron. nanotube films ACS Nano 3 835–43 Reliab. 51 797–804 [533] Liu Z, Parvez K, Li R, Dong R, Feng X and Müllen K [508] van Osch T H J, Perelaer J, de Laat A W M and Schubert U S 2014 Transparent conductive electrodes from graphene/ 2008 Inkjet printing of narrow conductive tracks on PEDOT:PSS hybrid inks for ultrathin organic photodetectors untreated polymeric substrates Adv. Mater. 20 343–5 Adv. Mater. 27 669–75 253 2D Mater. 7 (2020) 022001 C Backes et al [534] Liu Z, Wu Z-S, Yang S, Dong R, Feng X and Müllen K 2016 [556] Tobjörk D and Österbacka R 2011 Paper electronics Adv. Ultraflexible in-plane micro-supercapacitors by direct Mater. 23 1935–61 printing of solution-processable electrochemically exfoliated [557] Merilampi S, Laine-Ma T and Ruuskanen P 2009 The graphene Adv. Mater. 28 2217–22 characterization of electrically conductive silver ink patterns [535] Zaugg C A et al 2013 Ultrafast and widely tuneable vertical- on flexible substrates Microelectron. Reliabil. 49 782–90 external-cavity surface-emitting laser, mode-locked by a [558] Birkenshaw J W Printing processes The Printing Ink Manual graphene-integrated distributed Bragg reflector Opt. Express (Netherlands: Springer) pp 14–85 21 31548–59 [559] Hyun W J, Lim S, Ahn B Y, Lewis J A, Frisbie C D and [536] Bianchi V et al 2017 Terahertz saturable absorbers from liquid Francis L F 2015 Screen printing of highly loaded silver inks phase exfoliation of graphite Nat. Commun. 8 15763 on plastic substrates using silicon stencils ACS Appl. Mater. [537] Higgins T M et al 2014 Effect of percolation on the Interfaces 7 12619–24 capacitance of supercapacitor electrodes prepared from [560] Abbott S, Church T, Parker D and Harris A 2008 How to be a composites of manganese dioxide nanoplatelets and carbon Great Screen Printer ed S Abbott (MacDermid Autotype Ltd) nanotubes ACS Nano 8 9567–79 [561] Novalia (www.audioposter.com/) [538] McAteer D, Gholamvand Z, McEvoy N, Harvey A, O’Malley E, [562] Hu G, Kang J, Ng L W T, Zhu X, Howe R C T, Jones C G, Duesberg G S and Coleman J N 2016 Thickness dependence Hersam M C and Hasan T 2018 Functional inks and printing and percolation scaling of hydrogen production rate in MoS2 of two-dimensional materials Chem. Soc. Rev. 47 3265–300 nanosheet and nanosheet-carbon nanotube composite [563] Bonaccorso F, Colombo L, Yu G H, Stoller M, Tozzini V, catalytic electrodes ACS Nano 10 672–83 Ferrari A C, Ruoff R S and Pellegrini V 2015 Graphene, related [539] Purdie D, Popa D, Wittwer V J, Jiang Z, Torrisi F and two-dimensional crystals, and hybrid systems for energy Ferrari A C 2014 Sub-50 fs compressed pulses from a conversion and storage Science 347 1246501 graphene-mode locked fiber laser CLEO: 2014 OSA [564] Mohan V B, Lau K-T, Hui D and Bhattacharyya D 2018 [540] Jiang Z, Bonacchini G E, Popa D, Torrisi F, Ott A K, Graphene-based materials and their composites: a review on Wittwer V J, Purdie D and Ferrari A C 2014 Graphene production, applications and product limitations Composites saturable absorber power scaling laser CLEO: 2014 OSA B 142 200–20 [541] Yingying R, Brown G, Mary R, Demetriou G, Popa D, Torrisi F, [565] Miranzo P, Belmonte M and Osendi M I 2017 From bulk Ferrari A C, Feng C and Kar A K 2015 7.8 GHz graphene- to cellular structures: a review on ceramic/graphene filler based 2 µm monolithic waveguide laser IEEE J. Sel. Top. composites J. Eur. Ceram. Soc. 37 3649–72 Quantum Electron. 21 395–400 [566] Wang F, Wang H and Mao J 2019 Aligned-graphene [542] Dikin D A, Stankovich S, Zimney E J, Piner R D, composites: a review J. Mater. Sci. 54 36–61 Dommett G H B, Evmenenko G, Nguyen S T and Ruoff R S [567] Boland C S et al 2016 Sensitive electromechanical sensors 2007 Preparation and characterization of graphene oxide using viscoelastic graphene-polymer nanocomposites Science paper Nature 448 457–60 354 1257–60 [543] Kim B, Jang S, Geier M L, Prabhumirashi P L, Hersam M C [568] Boland C S et al 2014 Sensitive, high-strain, high-rate bodily and Dodabalapur A 2014 High-Speed, inkjet-printed motion sensors based on graphene-rubber composites ACS carbon nanotube/zinc tin oxide hybrid complementary ring Nano 8 8819–30 oscillators Nano Lett. 14 3683–7 [569] Layani M, Kamyshny A and Magdassi S 2014 [544] Secor E B, Prabhumirashi P L, Puntambekar K, Geier M L Transparent conductors composed of nanomaterials and Hersam M C 2013 Inkjet printing of high conductivity, Nanoscale 6 5581–91 flexible graphene patterns J. Phys. Chem. Lett. 4 1347–51 [570] Secor E B, Gao T Z, Islam A E, Rao R, Wallace S G, Zhu J, [545] Novacentrix (www.novacentrix.com) Putz K W, Maruyama B and Hersam M C 2017 Enhanced [546] Rinaldi A, Proietti A, Tamburrano A, De Bellis G, Mulattieri M conductivity, adhesion, and environmental stability of and Sarto M S 2014 Multilayer Graphene-based films for printed graphene inks with nitrocellulose Chem. Mater. strain sensing 14th IEEE Int. Conf. on Nanotechnology (IEEE) 29 2332–40 [547] Bessonov A, Kirikova M, Haque S, Gartseev I and Bailey M J A [571] Carey T, Cacovich S, Divitini G, Ren J, Mansouri A, 2014 Highly reproducible printable graphite strain gauges for Kim J M, Wang C, Ducati C, Sordan R and Torrisi F 2017 flexible devices Sensors Actuators A 206 75–80 Fully inkjet-printed two-dimensional material field-effect [548] Bao M-H 2000 Capacitive pressure transducers and heterojunctions for wearable and textile electronics Nat. accelerometers Micro Mechanical Transducers—Pressure Commun. 8 1202 Sensors, Accelerometers and Gyroscopes (Amsterdam: Elsevier) [572] Kelly A G et al 2017 All-printed thin-film transistors from pp 319–52 networks of liquid-exfoliated nanosheets Science 356 69–73 [549] Zlebic C, Zivanov L, Menicanin A, Blaz N and [573] Lin Z et al 2018 Solution-processable 2D semiconductors for Damnjanovic M 2016 Inkjet printed resistive strain gages on high-performance large-area electronics Nature 562 254–8 flexible substrates Facta Universitatis—Series: Electronics and [574] Hu G et al 2017 Black phosphorus ink formulation for inkjet Energetics vol 29 pp 89–100 printing of optoelectronics and photonics Nat. Commun. [550] McManus D et al 2017 Water-based and biocompatible 8 278 2D crystal inks for all-inkjet-printed heterostructures Nat. [575] Anichini C, Czepa W, Pakulski D, Aliprandi A, Ciesielski A and Nanotechnol. 12 343 Samorì P 2018 Chemical sensing with 2D materials Chem. [551] Sommer-Larsen P, Jørgensen M, Søndergaard R R, Hösel M Soc. Rev. 47 4860–908 and Krebs F C 2013 It is all in the pattern-high-efficiency [576] Santra S, Hu G, Howe R C T, De Luca A, Ali S Z, Udrea F, power extraction from polymer solar cells through high- Gardner J W, Ray S K, Guha P K and Hasan T 2015 CMOS voltage serial connection Energy Technol. 1 15–9 integration of inkjet-printed graphene for humidity sensing [552] Krebs F C, Espinosa N, Hösel M, Søndergaard R R and Sci. Rep. 5 17374 Jørgensen M 2013 25th anniversary article: rise to power— [577] Huang L, Wang Z, Zhang J, Pu J, Lin Y, Xu S, Shen L, Chen Q OPV-based solar parks Adv. Mater. 26 29–39 and Shi W 2014 Fully printed, rapid-response sensors based [553] Lee D H, Choi J S, Chae H, Chung C H and Cho S M 2009 on chemically modified graphene for detecting NO2 at room Screen-printed white OLED based on polystyrene as a host temperature ACS Appl. Mater. Interfaces 6 7426–33 polymer Curr. Appl. Phys. 9 161–4 [578] Cho S-Y, Lee Y, Koh H-J, Jung H, Kim J-S, Yoo H-W, Kim J and [554] Jung S, Hoath S D, Martin G D and Hutchings I M 2015 Inkjet Jung H-T 2016 Superior chemical sensing performance of Printing Process for Large Area Electronics, Large Area and black phosphorus: comparison with MoS2 and graphene Adv. Flexible Electronics (New York: Wiley) pp 315–44 Mater. 28 7020–8 [555] Khan S, Lorenzelli L and Dahiya R S 2015 Technologies for [579] Sun S, Liao C, Hafez A M, Zhu H and Wu S 2018 Two- printing sensors and electronics over large flexible substrates: dimensional MXenes for energy storage Chem. Eng. J. a review IEEE Sens. J. 15 3164–85 338 27–45 254 2D Mater. 7 (2020) 022001 C Backes et al [580] Foster C W, Down M P, Zhang Y, Ji X, Rowley-Neale S J, [601] Hupalo M, Conrad E H and Tringides M C 2009 Growth Smith G C, Kelly P J and Banks C E 2017 3D printed graphene mechanism for epitaxial graphene on vicinal 6H–SiC (0 0 0 1) based energy storage devices Sci. Rep. 7 42233 surfaces: a scanning tunneling microscopy study Phys. Rev. B [581] Hyun W J, Secor E B, Kim C-H, Hersam M C, Francis L F 80 041401 and Frisbie C D 2017 Scalable, self-aligned printing of [602] Borovikov V and Zangwill A 2009 Step-edge instability during flexible graphene micro-supercapacitors Adv. Energy Mater. epitaxial growth of graphene from SiC (0 0 0 1) Phys. Rev. B 7 1700285 80 121406 [582] Shi L and Zhao T 2017 Recent advances in inorganic 2D [603] Ohta T, Bartelt N C, Nie S, Thurmer K and Kellogg G L 2010 materials and their applications in lithium and sodium Role of carbon surface diffusion on the growth of epitaxial batteries J. Mater. Chem. C 5 3735–58 graphene on SiC Phys. Rev. B 81 121411 [583] Nguyen V L et al 2015 Seamless stitching of graphene domains [604] Hu Y, Ruan M, Guo Z L, Dong R, Palmer J, Hankinson J, on polished copper (1 1 1) foil Adv. Mater. 27 1376–82 Berger C and de Heer W A 2012 Structured epitaxial [584] Berger C et al 2004 Ultrathin epitaxial graphite: 2D graphene: growth and properties J. Phys. D: Appl. Phys. electron gas properties and a route toward graphene-based 45 154010 nanoelectronics J. Phys. Chem. B 108 19912–6 [605] Hite J K, Twigg M E, Tedesco J L, Friedman A L, Myers- [585] Berger C, Conrad E H and de Heer W A 2018 Epigraphene in Ward R L, Eddy C R and Gaskill D K 2011 Epitaxial graphene Physics of Solid Surfaces (Landolt-Borstein Encyclopedia) ed nucleation on C-face silicon carbide Nano Lett. 11 1190–4 P Chiaradia and G Chiarotti pp 727–807 [606] Zhang R, Dong Y L, Kong W J, Han W P, Tan P H, Liao Z M, [586] Van Bommel A J, Crobeen J E and Van Tooren A 1975 LEED Wu X S and Yu D P 2012 Growth of large domain epitaxial and Auger electron observations of the SiC(0 0 0 1) surface graphene on the C-face of SiC J. Appl. Phys. 112 104307 Surf. Sci. 48 463–72 [607] Yazdi G R, Iakimov T and Yakimova R 2016 Epitaxial [587] de Heer W A, Berger C and First P N 2006 Patterned thin films graphene on SiC: a review of growth and characterization graphite devices and methods for making the same US Patent Crystals 6 53 7015142 (provisional Application No. 60/477997 filed June 12 [608] Råback P, Yakimova R, Syväjärvi M, Nieminen R and 2003, Issued March 21 2006) Janzén E 1999 A practical model for estimating the growth [588] Virojanadara C, Syvajarvi M, Yakimova R, Johansson L I, rate in sublimation growth of SiC Mater. Sci. Eng. B Zakharov A A and Balasubramanian T 2008 Homogeneous 61–62 89–92 large-area graphene layer growth on 6H–SiC (0 0 0 1) Phys. [609] Tromp R M and Hannon J B 2009 Thermodynamics and Rev. B 78 245403 kinetics of graphene growth on SiC(0 0 0 1) Phys. Rev. Lett. [589] Emtsev K V et al 2009 Towards wafer-size graphene layers by 102 106104 atmospheric pressure graphitization of silicon carbide Nat. [610] Riedl C, Starke U, Bernhardt J, Franke M and Heinz K 2007 Mater. 8 203–7 Structural properties of the graphene-SiC(0 0 0 1) interface [590] Varchon F et al 2007 Electronic structure of epitaxial as a key for the preparation of homogeneous large-terrace graphene layers on SiC: effect of the substrate Phys. Rev. Lett. graphene surfaces Phys. Rev. B 76 245406 99 126805 [611] Strupinski W et al 2011 Graphene epitaxy by chemical vapor [591] Emery J D, Detlefs B, Karmel H J, Nyakiti L O, Gaskill D K, deposition on SiC Nano Lett. 11 1786–91 Hersam M C, Zegenhagen J and Bedzyk M J 2013 Chemically [612] Lafont F et al 2015 Quantum Hall resistance standards from resolved interface structure of epitaxial graphene on graphene grown by chemical vapour deposition on silicon SiC(0 0 0 1) Phys. Rev. Lett. 111 215501 carbide Nat. Commun. 6 6806 [592] Emtsev K V, Speck F, Seyller T, Ley L and Riley J D 2008 [613] Ostler M, Speck F, Gick M and Seyller T 2010 Automated Interaction, growth, and ordering of epitaxial graphene preparation of high-quality epitaxial graphene on 6H– on SiC{0 0 0 1} surfaces: A comparative photoelectron SiC(0 0 0 1) Phys. Status Solidi b 247 2924–6 spectroscopy study Phys. Rev. B 77 155303 [614] Yakimova R, Virojanadara C, Gogova D, Syvajarvi M, Siche D, [593] Conrad M et al 2016 Wide band gap semiconductor from Larsson K and Johansson L I 2010 Analysis of the formation a hidden 2D incommensurate graphene phase Nano Lett. conditions for large area epitaxial graphene on SiC substrates 17 341–7 Silicon Carbide Relat. Mater. 645–8 565–8 [594] Nevius M S, Conrad M, Wang F, Celis A, Nair M N, Taleb- [615] Berger C et al Epitaxial graphene on SiC: 2D sheets, Ibrahimi A, Tejeda A and Conrad E H 2015 Semiconducting selective growth and nanoribbons Growing Graphene on graphene from highly ordered substrate interactions Phys. Semiconductors ed C C N Motta and F Iacopi (PanStanford) Rev. Lett. 115 136802 (arXiv:1611.089372016) [595] Rutter G M, Guisinger N P, Crain J N, Jarvis E A A, Stiles M D, [616] Palmer J 2014 Pre-growth structures for nanoelectronics Li T, First P N and Stroscio J A 2007 Imaging the interface of of EG on SiC School of Physics—Georgia Institute of epitaxial graphene with silicon carbide via scanning tunneling Technology microscopy Phys. Rev. B 76 235416 [617] Camara N, Huntzinger J R, Rius G, Tiberj A, Mestres N, Perez- [596] Hass J, de Heer W A and Conrad E H 2008 The growth Murano F, Godignon P and Camassel J 2009 Anisotropic and morphology of epitaxial multilayer graphene J. Phys.: growth of long isolated graphene ribbons on the C face of Condens. Matter 20 323202 graphite-capped 6H–SiC Phys. Rev. B 80 125410 [597] de Heer W A, Berger C, Ruan M, Sprinkle M, Li X, Hu Y, [618] Sprinkle M, Ruan M, Hu Y, Hankinson J, Rubio-Roy M, Zhang B, Hankinson J and Conrad E H 2011 Large area and Zhang B, Wu X, Berger C and de Heer W A 2010 Scalable structured epitaxial graphene produced by confinement templated growth of graphene nanoribbons on SiC Nat. controlled sublimation of silicon carbide Proc. Natl Acad. Sci. Nanotechnol. 5 727–31 108 16900–5 [619] Ruan M 2012 Structured epitaxial graphene for electronics, [598] Coletti C et al 2013 Revealing the electronic band structure of PhD School of Physics, Georgia Institute of Technology, trilayer graphene on SiC: an angle-resolved photoemission Atlanta study Phys. Rev. B 88 155439 [620] Baringhaus J et al 2014 Exceptional ballistic transport in [599] Hass J, Varchon F, Millan-Otoya J E, Sprinkle M, Sharma N, epitaxial graphene nanoribbons Nature 506 349–54 De Heer W A, Berger C, First P N, Magaud L and Conrad E H [621] Cambaz Z G, Yushin G, Osswald S, Mochalin V and Goyotsi Y 2008 Why multilayer graphene on 4H-SiC(0 0 0(1)over- 2008 Noncatalytic synthesis of carbon nanotubes, graphene bar) behaves like a single sheet of graphene Phys. Rev. Lett. and graphite on SiC Carbon 46 841–9 100 125504 [622] Berger C et al 2006 Electronic confinement and coherence in [600] Miller D L, Kubista K D, Rutter G M, Ruan M, de Heer W A, patterned epitaxial graphene Science 312 1191–6 First P N and Stroscio J A 2010 Structural analysis of [623] Maysonnave J et al 2014 Terahertz generation by dynamical multilayer graphene via atomic moireacute interferometry photon drag effect in graphene excited by femtosecond Phys. Rev. B 81 223108 optical pulses Nano Lett. 14 5797–802 255 2D Mater. 7 (2020) 022001 C Backes et al [624] Sadowski M L, Martinez G, Potemski M, Berger C and de [648] Kruskopf M et al 2012 Epitaxial graphene on SiC: Heer W A 2007 Magnetospectroscopy of epitaxial few-layer modification of structural and electron transport properties graphene Solid State Commun. 143 123–5 by substrate pretreatment J. Phys.: Condens. Matter 27 185303 [625] Wu X S, Hu Y K, Ruan M, Madiomanana N K, Hankinson J, [649] Ivanov I G, Hassan J U, Iakimov T, Zakharov A A, Yakimova R Sprinkle M, Berger C and de Heer W A 2009 Half integer and Janzen E 2014 Layer-number determination in graphene quantum Hall effect in high mobility single layer epitaxial on SiC by reflectance mapping Carbon 77 492–500 graphene Appl. Phys. Lett. 95 223108 [650] Camara N, Jouault B, Caboni A, Jabakhanji B, Desrat W, [626] Hass J 2008 Structural characterization of epitaxial graphene Pausas E, Consejo C, Mestres N, Godignon P and Camassel J on silicon carbide PhD School of Physics, Georgia Institute of 2010 Growth of monolayer graphene on 8 degrees off-axis Technology 4H-SiC (0 0 0   −  1) substrates with application to quantum [627] Miller D L, Kubista K D, Rutter G M, Ruan M, de Heer W A, transport devices Appl. Phys. Lett. 97 093107 First P N and Stroscio J A 2009 Observing the quantization of [651] Eddy C R and Gaskill D K 2009 Silicon carbide as a platform zero mass carriers in graphene Science 324 924–7 for power electronics Science 324 1398–400 [628] Orlita M, Faugeras C, Grill R, Wysmolek A, Strupinski W, [652] Giannazzo F, Deretzis I, Nicotra G, Fisichella G, Ramasse Q M, Berger C, de Heer W A, Martinez G and Potemski M 2011 Spinella C, Roccaforte F and La Magna A 2014 High Carrier scattering from dynamical magnetoconductivity in resolution study of structural and electronic properties of quasineutral epitaxial graphene Phys. Rev. Lett. 107 216603 epitaxial graphene grown on off-axis 4H-SiC (0 0 0 1) J. Cryst. [629] Sprinkle M et al 2009 First direct observation of a nearly ideal Growth 393 150–5 graphene band structure Phys. Rev. Lett. 103 226803 [653] Sonde S, Giannazzo F, Huntzinger J R, Tiberj A, Syvajarvi M, [630] de Heer W A et al 2010 Epitaxial graphene electronic structure Yakimova R, Raineri V and Camassel J 2010 Uniformity of and transport J. Phys. D: Appl. Phys. 43 374007 epitaxial graphene on on-axis and off-axis SiC probed by [631] Kunc J, Hu Y, Palmer J, Berger C and de Heer W A 2013 raman spectroscopy and nanoscale current mapping Silicon A method to extract pure Raman spectrum of epitaxial Carbide Relat. Mater. 645–8 607–10 graphene on SiC Appl. Phys. Lett. 103 201911 [654] Vecchio C, Sonde S, Bongiorno C, Rambach M, Yakimova R, [632] Levy N, Burke S A, Meaker K L, Panlasigui M, Zettl A, Raineri V and Giannazzo F 2011 Nanoscale structural Guinea F, Neto A H C and Crommie M F 2010 Strain-induced characterization of epitaxial graphene grown on off-axis pseudo-magnetic fields greater than 300 tesla in graphene 4H-SiC (0 0 0 1) Nanoscale Res. Lett. 6 269 nanobubbles Science 329 544–7 [655] Penuelas J, Ouerghi A, Lucot D, David C, Gierak J, Estrade- [633] Orlita M et al 2008 Approaching the dirac point in high- Szwarckopf H and Andreazza-Vignolle C 2009 Surface mobility multilayer epitaxial graphene Phys. Rev. Lett. morphology and characterization of thin graphene films on 101 267601 SiC vicinal substrate Phys. Rev. B 79 033408 [634] Sadowski M L, Martinez G, Potemski M, Berger C and de [656] Maassen T, van den Berg J J, Huisman E H, Dijkstra H, Heer W A 2006 Landau level spectroscopy of ultrathin Fromm F, Seyller T and van Wees B J 2013 Localized states graphite layers Phys. Rev. Lett. 97 266405 influence spin transport in epitaxial graphene Phys. Rev. Lett. [635] Tzalenchuk A, Lara-Avila S, Kalaboukhov A, Paolillo S, 110 067209 Syvajarvi M, Yakimova R, Kazakova O, Janssen T J B M, [657] Shtepliuk I, Khranovskyy V and Yakimova R 2016 Combining Fal’ko V and Kubatkin S 2010 Towards a quantum resistance graphene with silicon carbide: synthesis and properties—a standard based on epitaxial graphene Nat. Nanotechnol. review Semicond. Sci. Technol. 31 113004 5 186–9 [658] Speck F, Jobst J, Fromm F, Ostler M, Waldmann D, [636] Borovikov V and Zangwill A 2009 Step bunching of vicinal Hundhausen M, Weber H B and Seyller T 2011 The quasi- 6H–SiC{0 0 0 1} surfaces Phys. Rev. B 79 121406 free-standing nature of graphene on H-saturated SiC(0 0 0 1) [637] Chua C et al 2014 Quantum Hall effect and quantum point Appl. Phys. Lett. 99 122106 contact in bilayer-patched epitaxial graphene Nano Lett. [659] Riedl C, Coletti C and Starke U 2010 Structural and electronic 14 3369–73 properties of epitaxial graphene on SiC(0 0 0 1): a review [638] Michon A, Vezian S, Roudon E, Lefebvre D, Zielinski M, of growth, characterization, transfer doping and hydrogen Chassagne T and Portail M 2013 Effects of pressure, intercalation J. Phys. D: Appl. Phys. 43 374009 temperature, and hydrogen during graphene growth on [660] Ostler M, Deretzis I, Mammadov S, Giannazzo F, Nicotra G, SiC(0 0 0 1) using propane-hydrogen chemical vapor Spinella C, Seyller T and La Magna A 2013 Direct growth deposition J. Appl. Phys. 113 203501 of quasi-free-standing epitaxial graphene on nonpolar SiC [639] Mattias K et al 2016 Comeback of epitaxial graphene for surfaces Phys. Rev. B 88 085408 electronics: large-area growth of bilayer-free graphene on SiC [661] Hens P, Zakharov A A, Iakimov T, Syvajarvi M and 2D Mater. 3 041002 Yakimova R 2014 Large area buffer-free graphene on non- [640] Yazdi G R, Vasiliauskas R, Iakimov T, Zakharov A, Syvajarvi M polar (0 0 1) cubic silicon carbide Carbon 80 823–9 and Yakimova R 2013 Growth of large area monolayer [662] Jabakhanji B, Camara N, Caboni A, Consejo C, Jouault B, graphene on 3C–SiC and a comparison with other SiC Godignon P and Camassel J 2012 Almost free standing polytypes Carbon 57 477–84 graphene on SiC(0 0 0   −  1) and SiC(1 1   −  2 0) Heterosic [641] Yakimova R and Yazdi G R 2015 private commnication Wasmpe 711 235–41 [642] Palmer J, Kunc J, Hu Y K, Hankinson J, Guo Z L, Berger C and [663] Forbeaux I, Themlin J M and Debever J M 1998 De Heer W A 2014 Controlled epitaxial graphene growth Heteroepitaxial graphite on 6H–SiC(0 0 0 1): interface within removable amorphous carbon corrals Appl. Phys. Lett. formation through conduction-band electronic structure 105 023106 Phys. Rev. B 58 16396–406 [643] Yakimova R 2017 unpublished [664] Forbeaux I, Themlin J M, Charrier A, Thibaudau F and [644] Norimatsu W and Kusunoki M 2014 Growth of graphene Debever J M 2001 Solid-state graphitization mechanisms of from SiC{0 0 0 1} surfaces and its mechanisms Semicond. Sci. silicon carbide 6H–SiC polar faces Appl. Surf. Sci. 162– Technol. 29 064009 3 406–12 [645] Riedl C, Coletti C, Iwasaki T, Zakharov A A and Starke U 2009 [665] Starke U 2003 Atomic Structure of SiC Surfaces (Recent Major Quasi-free-standing epitaxial graphene on SiC obtained by Advances in Silicon Carbide) ed W J Choyke et al (Berlin: hydrogen intercalation Phys. Rev. Lett. 103 246804 Springer) [646] Riedl C, Coletti C, Iwasaki T and Starke U 2010 Hydrogen [666] Deretzis I and La Magna A 2016 Simulating structural intercalation below epitaxial graphene on SiC(0 0 0 1) Mater. transitions with kinetic Monte Carlo: the case of epitaxial Sci. Forum 645–8 623–8 graphene on SiC Phys. Rev. E 93 033304 [647] Coletti C, Emtsev K V, Zakharov A A, Ouisse T, Chaussende D [667] Daas B K, Omar S U, Shetu S, Daniels K M, Ma S, and Starke U 2011 Large area quasi-free standing monolayer Sudarshan T S and Chandrashekhar M V S 2012 Comparison graphene on 3C–SiC(1 1 1) Appl. Phys. Lett. 99 081904 of epitaxial graphene growth on polar and nonpolar 6H–SiC 256 2D Mater. 7 (2020) 022001 C Backes et al faces: on the growth of multilayer films Cryst. Growth Des. on SiC (0 0 0 1) by fluorine intercalation from a molecular 12 3379–87 source ACS Nano 5 7662–8 [668] Goler S et al 2013 Revealing the atomic structure of the buffer [688] Masuda Y, Norimatsu W and Kusunoki M 2015 Formation of layer between SiC(0 0 0 1) and epitaxial graphene Carbon a nitride interface in epitaxial graphene on SiC (0 0 0 1) Phys. 51 249–54 Rev. B 91 104508 [669] Mattausch A and Pankratov O 2007 Ab initio study of [689] Xia C, Watcharinyanon S, Zakharov A A, Yakimova R, graphene on SiC Phys. Rev. Lett. 99 076802 Hultman L, Johansson L I and Virojanadara C 2012 Si [670] Seyller T 2012 Epitaxial graphene on SiC(0 0 0 1) Graphene intercalation/deintercalation of graphene on 6H–SiC(0 0 0 1) Nanoelectronics: Metrology, Synthesis, Properties and Phys. Rev. B 85 045418 Applications ed H Raza (Berlin: Springer) pp 135–59 [690] Wang F et al 2012 Silicon intercalation into the graphene-SiC [671] Bostwick A, Ohta T, Seyller T, Horn K and Rotenberg E 2007 interface Phys. Rev. B 85 165449 Quasiparticle dynamics in graphene Nat. Phys. 3 36–40 [691] Emtsev K V, Zakharov A A, Coletti C, Forti S and Starke U [672] Rollings E, Gweon G H, Zhou S Y, Mun B S, McChesney J L, 2011 Ambipolar doping in quasifree epitaxial graphene Hussain B S, Fedorov A, First P N, de Heer W A and Lanzara A on SiC(0 0 0 1) controlled by Ge intercalation Phys. Rev. B 2006 Synthesis and characterization of atomically thin 84 125423 graphite films on a silicon carbide substrate J. Phys. Chem. [692] Baringhaus J, Stöhr A, Forti S, Starke U and Tegenkamp C Solids 67 2172–7 2015 Ballistic bipolar junctions in chemically gated graphene [673] Ohta T, Bostwick A, Seyller T, Horn K and Rotenberg E 2006 ribbons Sci. Rep. 5 9955 Controlling the electronic structure of bilayer graphene [693] Gierz I et al 2010 Electronic decoupling of an epitaxial Science 313 951–54 graphene monolayer by gold intercalation Phys. Rev. B [674] Ohta T, Bostwick A, McChesney J L, Seyller T, Horn K and 81 235408 Rotenberg E 2007 Interlayer interaction and electronic [694] Premlal B, Cranney M, Vonau F, Aubel D, Casterman D, De screening in multilayer graphene investigated with angle- Souza M M and Simon L 2009 Surface intercalation of gold resolved photoemission spectroscopy Phys. Rev. Lett. underneath a graphene monolayer on SiC(0 0 0 1) studied by 98 206802 scanning tunneling microscopy and spectroscopy Appl. Phys. [675] Jobst J, Waldmann D, Speck F, Hirner R, Maude D K, Lett. 94 263115 Seyller T and Weber H B 2010 Quantum oscillations and [695] Pillai P B, Desouza M, Narula R, Reich S, Wong L Y, Batten T quantum Hall effect in epitaxial graphene Phys. Rev. B and Pokorny J 2015 Decoupling of epitaxial graphene via gold 81 195434 intercalation probed by dispersive Raman spectroscopy J. [676] Jobst J, Waldmann D, Speck F, Hirner R, Maude D K, Seyller T Appl. Phys. 117 183103 and Weber H B 2011 Transport properties of high-quality [696] Marchenko D, Varykhalov A, Sánchez-Barriga J, Seyller T and epitaxial graphene on 6H–SiC(0 0 0 1) Solid State Commun. Rader O 2016 Rashba splitting of 100 meV in Au-intercalated 151 1061–4 graphene on SiC Appl. Phys. Lett. 108 172405 [677] Farı́as D, Rieder K H, Shikin A M, Adamchuk V K, Tanaka T [697] Yagyu K, Tajiri T, Kohno A, Takahashi K, Tochihara H, and Oshima C 2000 Modification of the surface phonon Tomokage H and Suzuki T 2014 Fabrication of a single layer dispersion of a graphite monolayer adsorbed on Ni(1 1 1) graphene by copper intercalation on a SiC(0 0 0 1) surface caused by intercalation of Yb, Cu and Ag Surf. Sci. 454–6 437– Appl. Phys. Lett. 104 053115 41 [698] Virojanadara C, Watcharinyanon S, Zakharov A A and [678] Forti S, Emtsev K V, Coletti C, Zakharov A A, Riedl C and Johansson L I 2010 Epitaxial graphene on 6H–SiC and Li Starke U 2011 Large-area homogeneous quasifree standing intercalation Phys. Rev. B 82 205402 epitaxial graphene on SiC(0 0 0 1): electronic and structural [699] Virojanadara C, Watcharinyanon S, Zakharov A A, characterization Phys. Rev. B 84 125449 Yakimova R and Johansson L I 2012 Studies of Li intercalation [679] Bostwick A, Speck F, Seyller T, Horn K, Polini M, Asgari R, into epitaxial graphene on SiC(0 0 0 1) Mater. Sci. Forum MacDonald A H and Rotenberg E 2010 Observation of 717–20 653–6 plasmarons in quasi-freestanding doped graphene Science [700] Xia C, Watcharinyanon S, Zakharov A A, Johansson L I, 328 999–1002 Yakimova R and Virojanadara C 2013 Detailed studies of Na [680] Virojanadara C, Zakharov A A, Yakimova R and Johansson L I intercalation on furnace-grown graphene on 6H–SiC(0 0 0 1) 2010 Buffer layer free large area bi-layer graphene on Surf. Sci. 613 88–94 SiC(0 0 0 1) Surf. Sci. 604 L4–7 [701] Seyller T 2005 Hydrogen-saturated SiC-surfaces: model [681] Watcharinyanon S, Virojanadara C, Osiecki J R, Zakharov A A, systems for studies of passivation, reconstruction, and Yakimova R, Uhrberg R I G and Johansson L I 2011 Hydrogen interface formation 5th European Conf. on Silicon Carbide and intercalation of graphene grown on 6H–SiC(0 0 0 1) Surf. Sci. Related Materials, ECRSCRM2004 vol 483–5 pp 535–40 605 1662–8 [702] Seyller T et al 2006 Structural and electronic properties of [682] Oida S, McFeely F R, Hannon J B, Tromp R M, Copel M, graphite layers grown on SiC(0 0 0 1) Surf. Sci. 600 3906–11 Chen Z, Sun Y, Farmer D B and Yurkas J 2010 Decoupling [703] Sieber N, Mantel B F, Seyller T, Ristein J, Ley L, Heller T, graphene from SiC(0 0 0 1) via oxidation Phys. Rev. B Batchelor D R and Schmeisser D 2001 Electronic and 82 041411 chemical passivation of hexagonal 6H–SiC surfaces by [683] Oliveira M H Jr, Schumann T, Fromm F, Koch R, Ostler M, hydrogen termination Appl. Phys. Lett. 78 1216–8 Ramsteiner M, Seyller T, Lopes J M J and Riechert H 2013 [704] Speck F, Ostler M, Röhrl J, Jobst J, Waldmann D, Formation of high-quality quasi-free-standing bilayer Hundhausen M, Ley L, Weber H B and Seyller T 2010 Quasi- graphene on SiC(0 0 0 1) by oxygen intercalation upon freestanding graphene on SiC(0 0 0 1) Mater. Sci. Forum annealing in air Carbon 52 83–9 645–8 629–32 [684] Ostler M, Fromm F, Koch R J, Wehrfritz P, Speck F, Vita H, [705] Daniels K M, Jadidi M M, Sushkov A B, Boyd A K, Nath A, Böttcher S, Horn K and Seyller T 2014 Buffer layer free Drew H D, Murphy T E, Myers-Ward R L and Gaskill D K graphene on SiC(0 0 0 1) via interface oxidation in water 2016 Narrow terahertz plasmon resonance of quasi- vapor Carbon 70 258–65 freestanding bilayer epitaxial graphene 2016 74th Annual [685] Ostler M, Koch R J, Speck F, Fromm F, Vita H, Hundhausen M, Device Research Conf. pp 1–2 Horn K and Seyller T 2012 Decoupling the graphene buffer [706] Tanabe S, Sekine Y, Kageshima H and Hibino H 2012 layer from SiC(0 0 0 1) via interface oxidation Mater. Sci. Electrical characterization of bilayer graphene formed by Forum 717–20 649–52 hydrogen intercalation of monolayer graphene on SiC(0 0 0 1) [686] Walter A L et al 2011 Highly p-doped epitaxial graphene Japan. J. Appl. Phys. 51 02BN02 obtained by fluorine intercalation Appl. Phys. Lett. 98 184102 [707] Mammadov S, Ristein J, Krone J, Raidel C, Wanke M, [687] Wong S L, Huang H, Wang Y, Cao L, Qi D, Santoso I, Chen W Wiesmann V, Speck F and Seyller T 2017 Work function of and Wee A T S 2011 Quasi-free-standing epitaxial graphene graphene multilayers on SiC(0 0 0 1) 2D Mater. 4 015043 257 2D Mater. 7 (2020) 022001 C Backes et al [708] Mammadov S, Ristein J, Koch R J, Ostler M, Raidel C, [728] Puybaret R, Hankinson J, Palmer J, Bouvier C, Ougazzaden A, Wanke M, Vasiliauskas R, Yakimova R and Seyller T 2014 Voss P L, Berger C and de Heer W A 2015 Scalable control Polarization doping of graphene on silicon carbide 2D Mater. of graphene growth on 4H–SiC C-face using decomposing 1 035003 silicon nitride masks J. Phys. D: Appl. Phys. 48 152001 [709] Ristein J, Mammadov S and Seyller T 2012 Origin of doping [729] Han M Y, Brant J C and Kim P 2010 Electron transport in in quasi-free-standing graphene on silicon carbide Phys. Rev. disordered graphene nanoribbons Phys. Rev. Lett. 104 056801 Lett. 108 246104 [730] Stampfer C et al 2011 Transport in graphene nanostructures [710] Sławińska J, Aramberri H, Muñoz M C and Cerdá J I 2015 Frontiers Phys. 6 271–93 Ab initio study of the relationship between spontaneous [731] Hicks J et al 2013 A wide-bandgap metal-semiconductor- polarization and p-type doping in quasi-freestanding metal nanostructure made entirely from graphene Nat. Phys. graphene on H-passivated SiC surfaces Carbon 93 88–104 9 49–54 [711] Ferralis N and Carraro C 2014 Evolution of interfacial [732] Yakimova R, Iakimov T, Yazdi G R, Bouhafs C, Eriksson J, intercalation chemistry on epitaxial graphene/SiC by surface Zakharov A, Boosalis A, Schubert M and Darakchieva V enhanced Raman spectroscopy Appl. Surf. Sci. 320 441–7 2014 Morphological and electronic properties of epitaxial [712] Melios C, Spencer S, Shard A, Strupiński W, Silva S R P and graphene on SiC Physica B 439 54–9 Kazakova O 2016 Surface and interface structure of quasi-free [733] Ruan M, Hu Y, Guo Z, Dong R, Palmer J, Hankinson J, standing graphene on SiC 2D Mater. 3 025023 Berger C and de Heer W A 2012 Epitaxial graphene on silicon [713] Robinson J A, Hollander M, LaBella M, Trumbull K A, carbide: introduction to structured graphene MRS Bull. Cavalero R and Snyder D W 2011 Epitaxial graphene 37 1138–47 transistors: enhancing performance via hydrogen [734] Palacio I et al 2014 Atomic structure of epitaxial graphene intercalation Nano Lett. 11 3875–80 sidewall nanoribbons: flat graphene, miniribbons, and the [714] Ciuk T and Strupinski W 2015 Statistics of epitaxial graphene confinement gap Nano Lett. 15 182–9 for Hall effect sensors Carbon 93 1042–9 [735] Norimatsu W and Kusunoki M 2010 Formation process of [715] Murata Y, Mashoff T, Takamura M, Tanabe S, Hibino H, graphene on SiC (0 0 0 1) Physica E 42 691–4 Beltram F and Heun S 2014 Correlation between morphology [736] Baringhaus J, Aprojanz J, Wiegand J, Laube D, Halbauer M, and transport properties of quasi-free-standing monolayer Hubner J, Oestreich M and Tegenkamp C 2015 Growth and graphene Appl. Phys. Lett. 105 221604 characterization of sidewall graphene nanoribbons Appl. [716] Pallecchi E, Lafont F, Cavaliere V, Schopfer F, Mailly D, Phys. Lett. 106 043109 Poirier W and Ouerghi A 2014 High electron mobility in [737] Dong R, Guo Z L, Palmer J, Hu Y K, Ruan M, Hankinson J, epitaxial graphene on 4H–SiC(0 0 0 1) via post-growth Kunc J, Bhattacharya S K, Berger C and de Heer W A 2014 annealing under hydrogen Sci. Rep. 4 4558 Wafer bonding solution to epitaxial graphene-silicon [717] Tanabe S, Takamura M, Harada Y, Kageshima H and integration J. Phys. D: Appl. Phys. 47 904001 Hibino H 2012 Quantum Hall effect and carrier scattering [738] Suemitsu M, Jiao S, Fukidome H, Tateno Y, Makabe I and in quasi-free-standing monolayer graphene Appl. Phys. Nakabayashi T 2014 Epitaxial graphene formation on Express 5 125101 3C–SiC/Si thin films J. Phys. D: Appl. Phys. 47 094016 [718] Tanabe S, Takamura M, Harada Y, Kageshima H and [739] Aristov V Y et al 2010 Graphene synthesis on cubic SiC/Si Hibino H 2014 Effects of hydrogen intercalation on transport wafers. Perspectives for mass production of graphene-based properties of quasi-free-standing monolayer graphene Japan. electronic devices Nano Lett. 10 992–5 J. Appl. Phys. 53 115101 [740] Ouerghi A et al 2011 Sharp interface in epitaxial graphene [719] Sieber N 2002 Wasserstoff- und Sauerstofftabilisierte layers on 3C–SiC(1 0 0)/Si(1 0 0) wafers Phys. Rev. B 83 205429 6H–SiC{0 0 0 1}-Oberflächen—Eine Studie Chemischer, [741] Zarotti F, Gupta B, Iacopi F, Sgarlata A, Tomellini M and Struktureller und Elektronischer Eigenschaften (Friedrich- Motta N 2016 Time evolution of graphene growth on SiC as a Alexander-Universität Erlangen-Nürnberg) function of annealing temperature Carbon 98 307–12 [720] Sieber N, Seyller T, Ley L, James D, Riley J D, Leckey R C G and [742] Iacopi F, Mishra N, Cunning B V, Goding D, Dimitrijev S, Polcik M 2003 Synchrotron x-ray photoelectron spectroscopy Brock R, Dauskardt R H, Wood B and Boeckl J 2015 A study of hydrogen-terminated 6H–SiC{0 0 0 1} surfaces Phys. catalytic alloy approach for graphene on epitaxial SiC on Rev. B 67 205304 silicon wafers J. Mater. Res. 30 609–16 [721] Emery J D, Wheeler V H, Johns J E, McBriarty M E, Detlefs B, [743] Strupinski W, Grodecki K, Caban P, Ciepielewski P, Jozwik- Hersam M C, Gaskill D K and Bedzyk M J 2014 Structural Biala I and Baranowski J M 2015 Formation mechanism of consequences of hydrogen intercalation of epitaxial graphene graphene buffer layer on SiC(0 0 0 1) Carbon 81 63–72 on SiC(0 0 0 1) Appl. Phys. Lett. 105 161602 [744] Strupinski W 2010 Method of graphene manufacturing PAT- [722] Deretzis I and Magna A L 2013 Process simulation of 213291 US 9,067,796 B2; JP 5662249; ZL 201180027996.X; hydrogen intercalation in epitaxial graphene on SiC(0 0 0 1) Korea 10-1465452; EP 11168749.7 Phys. Status Solidi b 250 1478–82 [745] Weber M E 1974 Momentum, Heat and Mass Transfer 2nd edn [723] Deretzis I and La Magna A 2011 Role of covalent and metallic ed C O Bennett and J E Myers (New York: McGraw-Hill) p 810 intercalation on the electronic properties of epitaxial Weber M E 1974 AIChE J. 20 1037 graphene on SiC(0 0 0 1) Phys. Rev. B 84 235426 [746] Ciuk T, Petruk O, Kowalik A, Jozwik I, Rychter A, Szmidt J and [724] Fromm F, Oliveira J M H, Molina-Sánchez A, Hundhausen M, Strupinski W 2016 Low-noise epitaxial graphene on SiC Hall Lopes J M J, Riechert H, Wirtz L and Seyller T 2013 effect element for commercial applications Appl. Phys. Lett. Contribution of the buffer layer to the Raman spectrum of 108 223504 epitaxial graphene on SiC(0 0 0 1) New J. Phys. 15 043031 [747] Tokarczyk M, Kowalski G, Możdżonek M, Borysiuk J, [725] Giannazzo F, Hertel S, Albert A, La Magna A, Stępniewski R, Strupiński W, Ciepielewski P and Roccaforte F, Krieger M and Weber H B 2014 Electrical Baranowski J M 2013 Structural investigations of nanocharacterization of epitaxial graphene/silicon carbide hydrogenated epitaxial graphene grown on 4H–SiC (0 0 0 1) Schottky contacts Silicon Carbide and Related Materials 2013, Appl. Phys. Lett. 103 241915 Parts 1 and 2 ed H Okumura et al (Stafa-Zurich: Trans Tech [748] Ciuk T, Cakmakyapan S, Ozbay E, Caban P, Grodecki K, Publications Ltd) pp 1142–5 Krajewska A, Pasternak I, Szmidt J and Strupinski W 2014 [726] Caffrey N M, Armiento R, Yakimova R and Abrikosov I A 2015 Step-edge-induced resistance anisotropy in quasi-free- Charge neutrality in epitaxial graphene on 6H–SiC(0 0 0 1) standing bilayer chemical vapor deposition graphene on SiC via nitrogen intercalation Phys. Rev. B 92 081490 J. Appl. Phys. 116 123708 [727] Rubio-Roy M, Zaman F, Hu Y K, Berger C, Moseley M W, [749] Melios C, Panchal V, Giusca C E, Strupiński W, Silva S R P Meindl J D and de Heer W A 2010 Structured epitaxial and Kazakova O 2015 Carrier type inversion in quasi-free graphene growth on SiC by selective graphitization using a standing graphene: studies of local electronic and structural patterned AlN cap Appl. Phys. Lett. 96 082112 properties Sci. Rep. 5 10505 258 2D Mater. 7 (2020) 022001 C Backes et al [750] Habibpour O, He Z S, Strupinski W, Rorsman N and Zirath H thermal chemical vapor deposition growth of graphene 2017 Wafer scale millimeter-wave integrated circuits based on Carbon 50 668–73 epitaxial graphene in high data rate communication Sci. Rep. [773] Ryu J et al 2014 Fast synthesis of high-performance graphene 7 41828 films by hydrogen-free rapid thermal chemical vapor [751] Strupinski W 2017 unpublished deposition ACS Nano 8 950–6 [752] Karu A E and Beer M 1966 Pyrolytic formation of highly [774] Muñoz R, Munuera C, Martínez J I, Azpeitia J, Gómez- crystalline graphite films J. Appl. Phys. 37 2179–81 Aleixandre C and García-Hernández M 2016 Low [753] Shelton J C, Patil H R and Blakely J M 1974 Equilibrium temperature metal free growth of graphene on insulating segretation of carbon to a nickel(1 1 1) surface: a surface phase substrates by plasma assisted chemical vapor deposition 2D transition Surf. Sci. 43 493–520 Mater. 4 015009 [754] Eizenberg M and Blakely J M 1979 Carbon interaction with [775] Riikonen J, Kim W, Li C, Svensk O, Arpiainen S, Kainlauri M nickel surfaces: monolayer formation and structural stability and Lipsanen H 2013 Photo-thermal chemical vapor J. Chem. Phys. 71 3467–77 deposition of graphene on copper Carbon 62 43–50 [755] Li X, Magnuson C W, Venugopal A, An J, Suk J W, Han B, [776] Mathieu G, Guiot S and Cabane J 1973 Solubility of carbon in Borysiak M, Cai W, Velamakanni A and Zhu Y 2010 Graphene silver, copper and gold Scr. Metall. 7 421–5 films with large domain size by a two-step chemical vapor [777] McLellan R B 1969 Solubility of carbon in solid gold copper deposition process Nano Lett. 10 4328–34 and silver Scr. Metall. 3 389–91 [756] Reina A, Jia X, Ho J, Nezich D, Son H, Bulovic V, [778] Eizenberg M and Blakely J M 1979 Carbon monolayer phase Dresselhaus M S and Kong J 2009 Large area, few-layer condensation on Ni(1 1 1) Surf. Sci. 82 228–36 graphene films on arbitrary substrates by chemical vapor [779] Wang Y, Zheng Y, Xu X F, Dubuisson E, Bao Q L, Lu J and deposition Nano Lett. 9 30–5 Loh K P 2011 Electrochemical delamination of CVD-grown [757] Chae S J et al 2009 Synthesis of large-area graphene layers on graphene film: toward the recyclable use of copper catalyst poly-nickel substrate by chemical vapor deposition: wrinkle ACS Nano 5 9927–33 formation Adv. Mater. 21 2328–33 [780] Ciuk T, Pasternak I, Krajewska A, Sobieski J, Caban P, [758] Cabrero-Vilatela A, Weatherup R S, Braeuninger-Weimer P, Szmidt J and Strupinski W 2013 Properties of chemical Caneva S and Hofmann S 2016 Towards a general growth vapor deposition graphene transferred by high-speed model for graphene CVD on transition metal catalysts electrochemical delamination J. Phys. Chem. C 117 20833–7 Nanoscale 8 2149–58 [781] Li X, Cai W, Colombo L and Ruoff R S 2009 Evolution of [759] Kim K S, Zhao Y, Jang H, Lee S Y, Kim J M, Kim K S, Ahn J H, graphene growth on Ni and Cu by carbon isotope labeling Kim P, Choi J Y and Hong B H 2009 Large-scale pattern Nano Lett. 9 4268–72 growth of graphene films for stretchable transparent [782] Ma T, Ren W, Zhang X, Liu Z, Gao Y, Yin L C, Ma X L, Ding F electrodes Nature 457 706–10 and Cheng H M 2013 Edge-controlled growth and kinetics [760] De Arco L G, Zhang Y, Schlenker C W, Ryu K, Thompson M E of single-crystal graphene domains by chemical vapor and Zhou C W 2010 Continuous, highly flexible, and deposition Proc. Natl Acad. Sci. USA 110 20386–91 transparent graphene films by chemical vapor deposition for [783] Hao Y F et al 2016 Oxygen-activated growth and bandgap organic photovoltaics ACS Nano 4 2865–73 tunability of large single-crystal bilayer graphene Nat. [761] Ramón M E et al 2011 CMOS-compatible synthesis of large- Nanotechnol. 11 426–31 area, high-mobility graphene by chemical vapor deposition of [784] Nie S, Wu W, Xing S R, Yu Q K, Bao J M, Pei S S and acetylene on cobalt thin films ACS Nano 5 7198–204 McCarty K F 2012 Growth from below: bilayer graphene on [762] Gong Y P et al 2012 Layer-controlled and wafer-scale copper by chemical vapor deposition New J. Phys. 14 093028 synthesis of uniform and high-quality graphene films on a [785] Braeuninger-Weimer P, Brennan B, Pollard A J and polycrystalline nickel catalyst Adv. Funct. Mater. 22 3153–9 Hofmann S 2016 Understanding and controlling Cu- [763] Lin W-H, Chen T-H, Chang J-K, Taur J-I, Lo Y-Y, Lee W-L, catalyzed graphene nucleation: the role of impurities, Chang C-S, Su W-B and Wu C-I 2014 A direct and polymer- roughness, and oxygen scavenging Chem. Mater. 28 8905–15 free method for transferring graphene grown by chemical [786] Vlassiouk I et al 2013 Graphene nucleation density on copper: vapor deposition to any substrate ACS Nano 8 1784–91 fundamental role of background pressure J. Phys. Chem. C [764] Liang X L et al 2011 Toward clean and crackless transfer of 117 18919–26 graphene ACS Nano 5 9144–53 [787] Li X 2009 personal communication Austin, TX [765] Li X et al 2009 Large-area synthesis of high-quality and [788] Li X et al 2010 Graphene films with large domain size by uniform graphene films on copper foils Science 324 1312–4 a two-step chemical vapor deposition process Nano Lett. [766] Miseikis V et al 2015 Rapid CVD growth of millimetre-sized 10 4328–34 single crystal graphene using a cold-wall reactor 2D Mater. [789] Gao L et al 2010 Efficient growth of high-quality graphene 2 014006 films on Cu foils by ambient pressure chemical vapor [767] Huang L, Chang Q H, Guo G L, Liu Y, Xie Y Q, Wang T, Ling B deposition Appl. Phys. Lett. 97 183109 and Yang H F 2012 Synthesis of high-quality graphene films [790] López G A and Mittemeijer E J 2004 The solubility of C in on nickel foils by rapid thermal chemical vapor deposition solid Cu Scr. Mater. 51 1–5 Carbon 50 551–6 [791] Hao Y et al 2013 The role of surface oxygen in the growth of [768] Kim Y, Song W, Lee S Y, Jeon C, Jung W, Kim M and Park C Y large single-crystal graphene on copper Science 342 720–3 2011 Low-temperature synthesis of graphene on nickel foil [792] Artyukhov V I, Liu Y and Yakobson B I 2012 Equilibrium at by microwave plasma chemical vapor deposition Appl. Phys. the edge and atomistic mechanisms of graphene growth Proc. Lett. 98 263106 Natl Acad. Sci. USA 109 15136–40 [769] Vlassiouk I, Fulvio P, Meyer H, Lavrik N, Dai S, Datskos P and [793] Kim H, Mattevi C, Calvo M R, Oberg J C, Artiglia L, Agnoli S, Smirnov S 2013 Large scale atmospheric pressure chemical Hirjibehedin C F, Chhowalla M and Saiz E 2012 Activation vapor deposition of graphene Carbon 54 58–67 energy paths for graphene nucleation and growth on Cu ACS [770] Luo Z T, Lu Y, Singer D W, Berck M E, Somers L A, Nano 6 3614–23 Goldsmith B R and Johnson A T C 2011 Effect of substrate [794] Xu X et al 2017 Ultrafast epitaxial growth of metre-sized roughness and feedstock concentration on growth of single-crystal graphene on industrial Cu foil Sci. Bull. wafer-scale graphene at atmospheric pressure Chem. Mater. 62 1074–80 23 1441–7 [795] Nie S, Wofford J M, Bartelt N C, Dubon O D and McCarty K F [771] Liu L et al 2012 A systematic study of atmospheric pressure 2011 Origin of the mosaicity in graphene grown on Cu(1 1 1) chemical vapor deposition growth of large-area monolayer Phys. Rev. B 84 155425 graphene J. Mater. Chem. 22 1498–503 [796] Jacobberger R M and Arnold M S 2013 Graphene growth [772] Tan Y Y, Jayawardena K, Adikaari A, Tan L W, Anguita J V, dynamics on epitaxial copper thin films Chem. Mater. Henley S J, Stolojan V, Carey J D and Silva S R P 2012 Photo- 25 871–7 259 2D Mater. 7 (2020) 022001 C Backes et al [797] Yan Z, Lin J, Peng Z W, Sun Z Z, Zhu Y, Li L, Xiang C S, carbon diffusion barriers for uniform, high-quality graphene Samuel E L, Kittrell C and Tour J M 2012 Toward the synthesis growth from solid sources Nano Lett. 13 4624–31 of wafer-scale single-crystal graphene on copper foils ACS [818] Li X et al 2010 Graphene films with large domain size by Nano 6 9110–7 a two-step chemical vapor deposition process Nano Lett. [798] Wang H, Wang G Z, Bao P F, Yang S L, Zhu W, Xie X and 10 4328  −  34 Zhang W J 2012 Controllable synthesis of submillimeter [819] Gan L and Luo Z 2013 Turning off hydrogen to realize seeded single-crystal monolayer graphene domains on copper foils growth of subcentimeter single-crystal graphene grains on by suppressing nucleation J. Am. Chem. Soc. 134 3627–30 copper ACS Nano 7 9480–8 [799] Wu Y A, Fan Y, Speller S, Creeth G L, Sadowski J T, He K, [820] Fang W, Hsu A L, Song Y, Birdwell A G, Amani M, Dubey M, Robertson A W, Allen C S and Warner J H 2012 Large single Dresselhaus M S, Palacios T and Kong J 2014 Asymmetric crystals of graphene on melted copper using chemical vapor growth of bilayer graphene on copper enclosures using low- deposition ACS Nano 6 5010–7 pressure chemical vapor deposition ACS Nano 8 6491–9 [800] Wu B, Geng D C, Xu Z P, Guo Y L, Huang L P, Xue Y Z, [821] Petrone N, Dean C R, Meric I, van der Zande A M, Huang P Y, Chen J Y, Yu G and Liu Y Q 2013 Self-organized graphene Wang L, Muller D, Shepard K L and Hone J 2012 Chemical crystal patterns Npg Asia Mater. 5 e36 vapor deposition-derived graphene with electrical [801] Robinson Z R, Ong E W, Mowll T R, Tyagi P, Gaskill D K, performance of exfoliated graphene Nano Lett. 12 2751–6 Geisler H and Ventrice C A 2013 Influence of chemisorbed [822] Banszerus L, Janssen H, Otto M, Epping A, Taniguchi T, oxygen on the growth of graphene on Cu(1 0 0) by chemical Watanabe K, Beschoten B, Neumaier D and Stampfer C 2017 vapor deposition J. Phys. Chem. C 117 23919–27 Identifying suitable substrates for high-quality graphene- [802] Su C-Y et al 2011 Direct formation of wafer scale graphene based heterostructures 2D Mater. 4 025030 thin layers on insulating substrates by chemical vapor [823] Banszerus L, Schmitz M, Engels S, Goldsche M, Watanabe K, deposition Nano Lett. 11 3612–6 Taniguchi T, Beschoten B and Stampfer C 2016 Ballistic [803] Bointon T H, Barnes M D, Russo S and Craciun M F 2015 transport exceeding 28 µm in CVD grown graphene Nano High quality monolayer graphene synthesized by resistive Lett. 16 1387–91 heating cold wall chemical vapor deposition Adv. Mater. [824] Schmitz M, Engels S, Banszerus L, Watanabe K, Taniguchi T, 27 4200–6 Stampfer C and Beschoten B 2017 High mobility dry- [804] Geng D, Wu B, Guo Y, Huang L, Xue Y, Chen J, Yu G, Jiang L, transferred CVD bilayer graphene Appl. Phys. Lett. Hu W and Liu Y 2012 Uniform hexagonal graphene flakes 110 263110 and films grown on liquid copper surface Proc. Natl Acad. Sci. [825] Xu X et al 2016 Ultrafast growth of single-crystal graphene 109 7992–6 assisted by a continuous oxygen supply Nat. Nanotechnol. [805] Piner R et al 2013 Graphene synthesis via magnetic inductive 11 930–5 heating of copper substrates ACS Nano 7 7495–9 [826] Magnuson C W, Kong X H, Ji H X, Tan C, Li H F, Piner R, [806] Hawaldar R, Merino P, Correia M R, Bdikin I, Gracio J, Ventrice C A and Ruoff R S 2014 Copper oxide as a ‘self- Mendez J, Martin-Gago J A and Singh M K 2012 Large-area cleaning’ substrate for graphene growth J. Mater. Res. high-throughput synthesis of monolayer graphene sheet by 29 403–9 hot filament thermal chemical vapor deposition Sci. Rep. [827] Lin L, Li J Y, Ren H Y, Koh A L, Kang N, Peng H L, Xu H Q and 2 682 Liu Z F 2016 Surface engineering of copper foils for growing [807] Chen S, Ji H, Chou H, Li Q, Li H, Suk J W, Piner R, centimeter-sized single-crystalline graphene ACS Nano Liao L, Cai W and Ruoff R S 2013 Millimeter-size single- 10 2922–9 crystal graphene by suppressing evaporative loss of Cu [828] Reckinger N, Tang X, Joucken F, Lajaunie L, Arenal R, during low pressure chemical vapor deposition Adv. Mater. Dubois E, Hackens B, Henrard L and Colomer J-F 2016 25 2062–5 Oxidation-assisted graphene heteroepitaxy on copper foil [808] Cummings A W, Duong D L, Nguyen V L, Tuan D V, Nanoscale 8 18751–9 Kotakoski J, Vargas J E B, Lee Y H and Roche S 2014 Charge [829] Kim S M, Hsu A, Lee Y-H, Dresselhaus M, Palacios T, Kim K K transport in polycrystalline graphene: challenges and and Kong J 2013 The effect of copper pre-cleaning on opportunities Adv. Mater. 26 5079–94 graphene synthesis Nanotechnology 24 365602 [809] Isacsson A, Cummings A W, Colombo L, Colombo L, [830] Kim Y, Moyen E, Yi H M, Avila J, Chen C Y, Asensio M C, Kinaret J M and Roche S 2017 Scaling properties of Lee Y H and Pribat D 2018 Synthesis of high quality graphene polycrystalline graphene: a review 2D Mater. 4 012002 on capped (1 1 1) Cu thin films obtained by high temperature [810] Yazyev O V and Louie S G 2010 Electronic transport in secondary grain growth on c-plane sapphire substrates 2D polycrystalline graphene Nat. Mater. 9 806–9 Mater. 5 035008 [811] Grantab R, Shenoy V B and Ruoff R S 2010 Anomalous [831] Jin S et al 2018 Colossal grain growth yields single-crystal strength characteristics of tilt grain boundaries in graphene metal foils by contact-free annealing Science 362 1021–5 Science 330 946–8 [832] Banszerus L, Schmitz M, Engels S, Dauber J, Oellers M, [812] Wlasny I, Dabrowski P, Rogala M, Kowalczyk P J, Pasternak I, Haupt F, Watanabe K, Taniguchi T, Beschoten B and Strupinski W, Baranowski J M and Klusek Z 2013 Publisher’s Stampfer C 2015 Ultrahigh-mobility graphene devices from note: ‘role of graphene defects in corrosion of graphene- chemical vapor deposition on reusable copper Sci. Adv. coated Cu(1 1 1) surface’ [Appl. Phys. Lett. 102 111601] Appl. 1 e1500222 Phys. Lett. 102 149901 [833] Brown L et al 2014 Polycrystalline graphene with single [813] Kim W, Li C, Chekurov N, Arpiainen S, Akinwande D, crystalline electronic structure Nano Lett. 14 5706–11 Lipsanen H and Riikonen J 2015 All-graphene three- [834] Xue X, Xu Q, Wang H, Liu S, Jiang Q, Yu Z, Zhou X, Ma T, terminal-junction field-effect devices as rectifiers and Wang L and Yu G 2019 Gas-flow-driven aligned growth of inverters ACS Nano 9 5666–74 graphene on liquid copper Chem. Mater. 31 1231  −  6 [814] Kwak J et al 2012 Near room-temperature synthesis of [835] Yu H K, Balasubramanian K, Kim K, Lee J L, Maiti M, transfer-free graphene films Nat. Commun. 3 645 Ropers C, Krieg J, Kern K and Wodtke A M 2014 Chemical [815] Addou R, Dahal A, Sutter P and Batzill M 2012 Monolayer vapor deposition of graphene on a ‘peeled-off’ epitaxial graphene growth on Ni(1 1 1) by low temperature chemical Cu(1 1 1) foil: a simple approach to improved properties ACS vapor deposition Appl. Phys. Lett. 100 021601 Nano 8 8636–43 [816] Muñoz R and Gómez-Aleixandre C 2013 Fast and non- [836] Mishra N, Miseikis V, Convertino D, Gemmi M, Piazza V and catalytic growth of transparent and conductive graphene-like Coletti C 2016 Rapid and catalyst-free van der Waals epitaxy carbon films on glass at low temperature J. Phys. D: Appl. Phys. of graphene on hexagonal boron nitride Carbon 96 497–502 47 045305 [837] Wu T et al 2016 Fast growth of inch-sized single-crystalline [817] Weatherup R S, Baehtz C, Dlubak B, Bayer B C, Kidambi P R, graphene from a controlled single nucleus on Cu–Ni alloys Blume R, Schloegl R and Hofmann S 2013 Introducing Nat. Mater. 15 43–7 260 2D Mater. 7 (2020) 022001 C Backes et al [838] Vlassiouk I V et al 2018 Evolutionary selection growth of [860] Zeng M, Wang W-L and Bai X-D 2013 Preparing three- two-dimensional materials on polycrystalline substrates Nat. dimensional graphene architectures: Review of recent Mater. 17 318–22 developments Chin. Phys. B 22 098105 [839] Yu Q et al 2011 Control and characterization of individual [861] Chen Z, Ren W, Gao L, Liu B, Pei S and Cheng H-M 2011 grains and grain boundaries in graphene grown by chemical Three-dimensional flexible and conductive interconnected vapour deposition Nat. Mater. 10 443–9 graphene networks grown by chemical vapour deposition [840] Wu W, Jauregui L A, Su Z, Liu Z, Bao J, Chen Y P and Yu Q Nat. Mater. 10 424–8 2011 Graphene: growth of single crystal graphene arrays by [862] Li N, Chen Z, Ren W, Li F and Cheng H M 2012 Flexible locally controlling nucleation on polycrystalline Cu using graphene-based lithium ion batteries with ultrafast charge chemical vapor deposition (Adv. Mater. 42/2011) Adv. Mater. and discharge rates Proc. Natl Acad. Sci. 109 17360–5 23 4897–7 [863] Chen Z, Xu C, Ma C, Ren W and Cheng H-M 2013 [841] Miseikis V, Bianco F, David J, Gemmi M, Pellegrini V, Lightweight and flexible graphene foam composites for high- Romagnoli M and Coletti C 2017 Deterministic patterned performance electromagnetic interference shielding Adv. growth of high-mobility large-crystal graphene: a path Mater. 25 1296–300 towards wafer scale integration 2D Mater. 4 021004 [864] Cao X, Shi Y, Shi W, Lu G, Huang X, Yan Q, Zhang Q and [842] Luo Z, Kim S, Kawamoto N, Rappe A M and Johnson A T C Zhang H 2011 Preparation of novel 3D graphene networks 2011 Growth mechanism of hexagonal-shape graphene flakes for supercapacitor applications Small 7 3163–8 with zigzag edges ACS Nano 5 9154–60 [865] Zhang W et al 2017 Nitrogen-superdoped 3D graphene [843] Alcock C B, Itkin V P and Horrigan M K 1984 Vapour pressure networks for high-performance supercapacitors Adv. Mater. equations for the metallic elements: 298–2500 K Can. Metall. 29 1701677 Q. 23 309–13 [866] Hu G, Xu C, Sun Z, Wang S, Cheng H-M, Li F and Ren W 2015 [844] Guermoune A, Chari T, Popescu F, Sabri S S, Guillemette J, 3D graphene-foam-reduced-graphene-oxide hybrid nested Skulason H S, Szkopek T and Siaj M 2011 Chemical vapor hierarchical networks for high-performance Li-S batteries deposition synthesis of graphene on copper with methanol, Adv. Mater. 28 1603–9 ethanol, and propanol precursors Carbon 49 4204–10 [867] Zhu L, Peng H-J, Liang J, Huang J-Q, Chen C-M, Guo X, [845] Srivastava A, Galande C, Ci L, Song L, Rai C, Jariwala D, Zhu W, Li P and Zhang Q 2015 Interconnected carbon Kelly K F and Ajayan P M 2010 Novel liquid precursor-based nanotube/graphene nanosphere scaffolds as free-standing facile synthesis of large-area continuous, single, and few-layer paper electrode for high-rate and ultra-stable lithium–sulfur graphene films Chem. Mater. 22 3457–61 batteries Nano Energy 11 746–55 [846] Yao Y G, Li Z, Lin Z Y, Moon K S, Agar J and Wong C P 2011 [868] Yavari F, Chen Z, Thomas A V, Ren W, Cheng H-M and Controlled growth of multilayer, few-layer, and single-layer Koratkar N 2011 High sensitivity gas detection using a graphene on metal substrates J. Phys. Chem. C 115 5232–8 macroscopic three-dimensional graphene foam network Sci. [847] Li Z, Wu P, Wang C, Fan X, Zhang W, Zhai X, Zeng C, Li Z, Rep. 1 166 Yang J and Hou J 2011 Low-temperature growth of graphene [869] Chen W, Fan Z, Zeng G and Lai Z 2013 Layer-dependent by chemical vapor deposition using solid and liquid carbon supercapacitance of graphene films grown by chemical vapor sources ACS Nano 5 3385–90 deposition on nickel foam J. Power Sources 225 251–6 [848] Zhang B, Lee W H, Piner R, Kholmanov I, Wu Y P, Li H F, [870] Li N, Zhang Q, Gao S, Song Q, Huang R, Wang L, Liu L, Dai J, Ji H X and Ruoff R S 2012 Low-temperature chemical Tang M and Cheng G 2013 Three-dimensional graphene vapor deposition growth of graphene from toluene on foam as a biocompatible and conductive scaffold for neural electropolished copper foils ACS Nano 6 2471–6 stem cells Sci. Rep. 3 1604 [849] Jang J, Son M, Chung S, Kim K, Cho C, Lee B H and Ham M-H [871] Singh E, Chen Z, Houshmand F, Ren W, Peles Y, Cheng H-M 2015 Low-temperature-grown continuous graphene films and Koratkar N 2012 Superhydrophobic graphene foams from benzene by chemical vapor deposition at ambient Small 9 75–80 pressure Sci. Rep. 5 12402 [872] Jeong Y R, Park H, Jin S W, Hong S Y, Lee S-S and Ha J S [850] Wirtz C, Lee K, Hallam T and Duesberg G S 2014 Growth 2015 Highly stretchable and sensitive strain sensors using optimisation of high quality graphene from ethene at low fragmentized graphene foam Adv. Funct. Mater. 25 4228–36 temperatures Chem. Phys. Lett. 595 192–6 [873] Christian L O M, Liscio F, Rizzoli R, Palermo V and [851] Chan J, Venugopal A, Pirkle A, McDonnell S, Hinojos D, Morandi V 2017 Synthesis of high-density graphene foams Magnuson C W, Ruoff R S, Colombo L, Wallace R M and using nanoparticle templates GraphITA ed V Morandi and Vogel E M 2012 Reducing extrinsic performance-limiting L Ottaviano (Bologna: Springer) pp 185–96 factors in graphene grown by chemical vapor deposition ACS [874] Wang L, Li X, Guo T, Yan X and Tay B K 2014 Three- Nano 6 3224–9 dimensional Ni(OH)2 nanoflakes/graphene/nickel foam [852] Boyd D A et al 2015 Single-step deposition of high-mobility electrode with high rate capability for supercapacitor graphene at reduced temperatures Nat. Commun. 6 6620 applications Int. J. Hydrogen Energy 39 7876–84 [853] Tontegode A Y 1991 Carbon on transition metal surfaces Prog. [875] Drieschner S, Weber M, Wohlketzetter J, Vieten J, Surf. Sci. 38 201–429 Makrygiannis E, Blaschke B M, Morandi V, Colombo L, [854] Cai M, Outlaw R A, Quinlan R A, Premathilake D, Butler S M Bonaccorso F and Garrido J A 2016 High surface area and Miller J R 2014 Fast response, vertically oriented graphene foams by chemical vapor deposition 2D Mater. graphene nanosheet electric double layer capacitors 3 045013 synthesized from C2H2 ACS Nano 8 5873–82 [876] Dai B, Fu L, Zou Z, Wang M, Xu H, Wang S and Liu Z 2011 [855] Miller J R, Outlaw R A and Holloway B C 2010 Graphene Rational design of a binary metal alloy for chemical vapour double-layer capacitor with ac line-filtering performance deposition growth of uniform single-layer graphene Nat. Science 329 1637–9 Commun. 2 522 [856] Miller J R, Outlaw R A and Holloway B C 2011 Graphene [877] Kim E S, Shin H-J, Yoon S-M, Han G H, Chae S J, Bae J J, electric double layer capacitor with ultra-high-power Gunes F, Choi J-Y and Lee Y H 2011 Low-temperature performance Electrochim. Acta 56 10443–9 graphene growth using epochal catalyst of PdCo alloy Appl. [857] Nicholson M E 1962 Solubility of carbon in nickel-copper Phys. Lett. 99 223102 alloys at 1000 degrees C Trans. Metall. Soc. Aime 224 533 [878] Weatherup R S, Bayer B C, Blume R, Ducati C, Baehtz C, [858] Hasebe M, Ohtani H and Nishizawa T 1985 Effect of magnetic Schlögl R and Hofmann S 2011 In situ characterization of transition on solubility of carbon in bcc Fe and fee Co–Ni alloy catalysts for low-temperature graphene growth Nano alloys Metall. Trans. A 16 913–21 Lett. 11 4154–60 [859] Amato G, Beccaria F and Celegato F 2017 Growth of strained, [879] Zou Z Y, Fu L, Song X J, Zhang Y F and Liu Z F 2014 Carbide- but stable, graphene on Co Thin Solid Films 638 324–31 forming groups IVB-VIB metals: a new territory in the 261 2D Mater. 7 (2020) 022001 C Backes et al periodic table for CVD Growth of graphene Nano Lett. [901] Martínez J I, Merino P, Pinardi A L, Gonzalo O-I, López M F, 14 3832–9 Méndez J and Martín-Gago J A 2016 Role of the pinning [880] Zou Z Y, Song X J, Chen K, Ji Q Q, Zhang Y F and Liu Z F 2015 points in epitaxial graphene moiré superstructures on the Uniform single-layer graphene growth on recyclable tungsten Pt(1 1 1) surface Sci. Rep. 6 20354 foils Nano Res. 8 592–9 [902] Merino P, S?vec M, Pinardi A L, Otero G and Martín-Gago J A [881] Bhaviripudi S, Jia X, Dresselhaus M S and Kong J 2010 Role 2011 Strain-driven moiré superstructures of epitaxial of kinetic factors in chemical vapor deposition synthesis of graphene on transition metal surfaces ACS Nano 5 5627–34 uniform large area graphene using copper catalyst Nano Lett. [903] Gsell S, Fischer M, Schreck M and Stritzker B 2009 Epitaxial 10 4128–33 films of metals from the platinum group (Ir, Rh, Pt and Ru) [882] Losurdo M, Giangregorio M M, Capezzuto P and Bruno G on YSZ-buffered Si(1 1 1) J. Cryst. Growth 311 3731–6 2011 Graphene CVD growth on copper and nickel: role of [904] Zeller P, Dänhardt S, Gsell S, Schreck M and Wintterlin J 2012 hydrogen in kinetics and structure Phys. Chem. Chem. Phys. Scalable synthesis of graphene on single crystal Ir(1 1 1) films 13 20836 Surf. Sci. 606 1475–80 [883] Zhang W, Wu P, Li Z and Yang J 2011 First-principles [905] Hemmi A, Cun H, Roth S, Osterwalder J and Greber T 2014 thermodynamics of graphene growth on Cu surfaces J. Phys. Low cost photoelectron yield setup for surface process Chem. C 115 17782–7 monitoring J. Vac. Sci. Technol. A 32 023202 [884] Niu T, Zhou M, Zhang J, Feng Y and Chen W 2013 Growth [906] Marti´nez-Galera A J, Brihuega I N and Go´mez- intermediates for CVD graphene on Cu(1 1 1): carbon clusters Rodri´guez J M 2011 Ethylene irradiation: a new route to and defective graphene J. Am. Chem. Soc. 135 8409–14 grow graphene on low reactivity metals Nano Lett. 11 3576–80 [885] Fogarassy Z, Rümmeli M H, Gorantla S, Bachmatiuk A, [907] Nie S, Bartelt N C, Wofford J M, Dubon O D, McCarty K F Dobrik G, Kamarás K, Biró L P, Havancsák K and Lábár J L and Thürmer K 2012 Scanning tunneling microscopy study 2014 Dominantly epitaxial growth of graphene on Ni (1 1 1) of graphene on Au(1 1 1): Growth mechanisms and substrate substrate Appl. Surf. Sci. 314 490–9 interactions Phys. Rev. B 85 205406 [886] Al-Shurman H N K M 2014 Chemical vapor deposition [908] Lippert G, Dabrowski J, Lemme M, Marcus C, Seifarth O and (CVD) growth of graphene films Proc. of the 2014 COMSOL Lupina G 2011 Direct graphene growth on insulator Phys. Conf. in Boston (Elsevier) pp 27–49 Status Solidi b 248 2619–22 [887] Lahiri J, Miller T, Adamska L, Oleynik I I and Batzill M 2011 [909] Garcia J M et al 2012 Graphene growth on h-BN by molecular Graphene growth on Ni(1 1 1) by transformation of a surface beam epitaxy Solid State Commun. 152 975–8 carbide Nano Lett. 11 518–22 [910] Hernández-Rodríguez I, García J M, Martín-Gago J A, de [888] Reina A and Kong J 2012 Graphene growth by CVD methods Andrés P L and Méndez J 2015 Graphene growth on Pt(1 1 1) Graphene Nanoelectronics (Berlin: Springer) pp 167–203 and Au(1 1 1) using a MBE carbon solid-source Diamond [889] Baraton L, He Z B, Lee C S, Cojocaru C S, Châtelet M, Relat. Mater. 57 58–62 Maurice J L, Lee Y H and Pribat D 2011 On the mechanisms of [911] Martinez L N P J M G 2008 Devices with graphene layers precipitation of graphene on nickel thin films Europhys. Lett. [912] Otero G et al 2010 Ordered vacancy network induced by 96 46003 the growth of epitaxial graphene on Pt(1 1 1) Phys. Rev. Lett. [890] Cushing G W, Johánek V, Navin J K and Harrison I 2015 105 216102 Graphene growth on Pt(1 1 1) by ethylene chemical vapor [913] Pfeiffer L N 2008 Deposition of carbon-containing layers deposition at surface temperatures near 1000 K J. Phys. Chem. using vitreous carbon source C 119 4759–68 [914] Harris P J F 1997 Structure of non-graphitising carbons Int. [891] Ma T, Ren W, Liu Z, Huang L, Ma L-P, Ma X, Zhang Z, Peng L- Mater. Rev. 42 206–18 M and Cheng H-M 2014 Repeated growth–etching–regrowth [915] Garcia P W J M 2012 System and methods using a glassy for large-area defect-free single-crystal graphene by chemical carbon heater vapor deposition ACS Nano 8 12806–13 [916] Giannozzi P et al 2009 Quantum espresso: a modular and [892] Ma T et al 2017 Tailoring the thermal and electrical transport open-source software project for quantum simulations of properties of graphene films by grain size engineering Nat. materials J. Phys.: Condens. Matter 21 395502 Commun. 8 14486 [917] Henkelman G and Jónsson H 2000 Improved tangent [893] Vlassiouk I, Regmi M, Fulvio P, Dai S, Datskos P, Eres G estimate in the nudged elastic band method for finding and Smirnov S 2011 Role of hydrogen in chemical vapor minimum energy paths and saddle points J. Chem. Phys. deposition growth of large single-crystal graphene ACS Nano 113 9978–85 5 6069–76 [918] Wu Z W P, Li Z, Yang J and Hou J G 2010 Communication: [894] Choi J-K et al 2014 Growth of wrinkle-free graphene on coalescence of carbon atoms on Cu (1 1 1) surface: emergence texture-controlled platinum films and thermal-assisted of a stable bridging-metal structure motif J. Chem. Phys. transfer of large-scale patterned graphene ACS Nano 9 679–86 133 071101 [895] Rogge P C, Nie S, McCarty K F, Bartelt N C and Dubon O D [919] Bogicevic A, Strömquist J and Lundqvist B I 1998 First- 2014 Orientation-Dependent Growth Mechanisms of principles diffusion-barrier calculation for atomic oxygen on Graphene Islands on Ir(1 1 1) Nano Lett. 15 170–5 Pt(1 1 1) Phys. Rev. B 57 R4289–92 [896] Jin L, Fu Q, Zhang H, Mu R, Zhang Y, Tan D and Bao X [920] Taioli S, Garberoglio G, Simonucci S, Beccara S A, Aversa L, 2012 Tailoring the growth of graphene on Ru(0 0 01 ) Nardi M, Verucchi R, Iannotta S, Dapor M and Alfè D 2013 via engineering of the substrate surface J. Phys. Chem. C Non-adiabatic ab initio molecular dynamics of supersonic 116 2988–93 beam epitaxy of silicon carbide at room temperature J. Chem. [897] Dong G and Frenken J W M 2013 Kinetics of graphene Phys. 138 044701 formation on Rh(1 1 1) investigated by in situ scanning [921] Verucchi R, Aversa L, Nardi M V, Taioli S, Beccara S A, Alfè D, tunneling microscopy ACS Nano 7 7028–33 Nasi L, Rossi F, Salviati G and Iannotta S 2012 Epitaxy [898] Liu M, Zhang Y, Chen Y, Gao Y, Gao T, Ma D, Ji Q, Zhang Y, of nanocrystalline silicon carbide on Si(1 1 1) at room Li C and Liu Z 2012 Thinning segregated graphene layers on temperature J. Am. Chem. Soc. 134 17400–3 high carbon solubility substrates of rhodium foils by tuning [922] Taioli S 2014 Computational study of graphene growth the quenching process ACS Nano 6 10581–9 on copper by first-principles and kinetic Monte Carlo [899] Coraux J, N’Diaye A T, Busse C and Michely T 2008 Structural calculations J. Mol. Modeling 20 2260 coherency of graphene on Ir(1 1 1) Nano Lett. 8 565–70 [923] Tatti R, Aversa L, Verucchi R, Cavaliere E, Garberoglio G, [900] Günther S, Dänhardt S, Ehrensperger M, Zeller P, Schmitt S Pugno N M, Speranza G and Taioli S 2016 Synthesis of single and Wintterlin J 2012 High-temperature scanning tunneling layer graphene on Cu(1 1 1) by C60 supersonic molecular microscopy study of the ordering transition of an amorphous beam epitaxy RSC Adv. 6 37982–93 carbon layer into graphene on Ruthenium(0 0 0 1) ACS Nano [924] Kroto H W, Heath J R, O’Brien S C, Curl R F and Smalley R E 7 154–64 1985 C60: buckminsterfullerene Nature 318 162–3 262 2D Mater. 7 (2020) 022001 C Backes et al [925] Pinardi A L et al 2014 Vacancy formation on C60/Pt [950] Sun J, Chen Y, Priydarshi M K, Gao T, Song X, Zhang Y (1 1 1): unraveling the complex atomistic mechanism and Liu Z 2016 Graphene glass from direct CVD routes: Nanotechnology 25 385602 production and applications Adv. Mater. 28 10333–9 [926] Azpeitia J et al 2017 High-quality PVD graphene growth by [951] Chen Z et al 2016 Fast and uniform growth of graphene fullerene decomposition on Cu foils Carbon 119 535–43 glass using confined-flow chemical vapor deposition and its [927] Muñoz R and Gómez-Aleixandre C 2013 Review of CVD unique applications Nano Res. 9 3048–55 synthesis of graphene Chem. Vap. Depos. 19 297–322 [952] Sun J et al 2015 Direct chemical vapor deposition-derived [928] Yan Z, Peng Z and Tour J M 2014 Chemical vapor deposition graphene glasses targeting wide ranged applications Nano of graphene single crystals Acc. Chem. Res. 47 1327–37 Lett. 15 5846–54 [929] Castro Neto A H, Guinea F, Peres N M R, Novoselov K S and [953] Liu Q, Gong Y, Wang T, Chan W-L and Wu J 2016 Metal- Geim A K 2009 The electronic properties of graphene Rev. catalyst-free and controllable growth of high-quality Mod. Phys. 81 109–62 monolayer and AB-stacked bilayer graphene on silicon [930] Olesinski R W and Abbaschian G J 1984 The C–Ge (carbon– dioxide Carbon 96 203–11 germanium) system Bull. Alloy Phase Diagr. 5 484–6 [954] Xu S C, Man B Y, Jiang S Z, Chen C S, Yang C, Liu M, Gao X G, [931] Lee J-H et al 2014 Wafer-scale growth of single-crystal Sun Z C and Zhang C 2013 Direct synthesis of graphene on monolayer graphene on reusable hydrogen-terminated SiO2 substrates by chemical vapor deposition CrystEngComm germanium Science 344 286–9 15 1840–4 [932] Kiraly B, Jacobberger R M, Mannix A J, Campbell G P, [955] Zhao J, Wang G, Yang R, Lu X, Cheng M, He C, Xie G, Bedzyk M J, Arnold M S, Hersam M C and Guisinger N P Meng J, Shi D and Zhang G 2015 Tunable piezoresistivity of 2015 Electronic and mechanical properties of graphene- nanographene films for strain sensing ACS Nano 9 1622–9 germanium interfaces grown by chemical vapor deposition [956] Pang J et al 2017 Self-terminating confinement approach for Nano Lett. 15 7414–20 large-area uniform monolayer graphene directly over Si/SiOx [933] Lippert G et al 2014 Graphene grown on Ge(0 0 1) from by chemical vapor deposition ACS Nano 11 1946–56 atomic source Carbon 75 104–12 [957] Bi H, Sun S, Huang F, Xie X and Jiang M 2012 Direct growth [934] Pasternak I, Dabrowski P, Ciepielewski P, Kolkovsky V, of few-layer graphene films on SiO2 substrates and their Klusek Z, Baranowski J M and Strupinski W 2016 Large- photovoltaic applications J. Mater. Chem. 22 411–6 area high-quality graphene on Ge(0 0 1)/Si(0 0 1) substrates [958] Weber N-E, Binder A, Kettner M, Hirth S, Weitz R T and Nanoscale 8 11241–7 Tomovic Ž 2017 Metal-free synthesis of nanocrystalline [935] Pasternak I, Wesolowski M, Jozwik I, Lukosius M, Lupina G, graphene on insulating substrates by carbon dioxide-assisted Dabrowski P, Baranowski J M and Strupinski W 2016 chemical vapor deposition Carbon 112 201–7 Graphene growth on Ge(1 0 0)/Si(1 0 0) substrates by CVD [959] Iwai H 2009 Roadmap for 22 nm and beyond (invited paper) method Sci. Rep. 6 21773 Microelectron. Eng. 86 1520–8 [936] Baba H 1956 Studies on the reduction processes of [960] Zhang L, Shi Z and Wang Y 2011 Catalyst-free growth of germanium dioxide by hydrogen Bull. Chem. Soc. Japan nanographene films on various substrates Nano Res. 4 315 29 789–93 [961] Medina H, Lin Y-C and Jin C 2012 Metal-free growth of [937] Hasegawa R, Yagihash T and Kurosawa T 1972 Hydrogen nanographene on silicon oxides for transparent conducting reduction of germanium dioxide Trans. Japan Inst. Met. 13 39 applications Adv. Funct. Mater. 22 2123 [938] Sun J, Lindvall N and Cole M 2011 Large-area uniform [962] Nguyen V L and Lee Y H 2015 Towards wafer-scale graphene-like thin films grown by chemical vapor deposition monocrystalline graphene growth and characterization Small directly on silicon nitride Appl. Phys. Lett. 98 252107 11 3512 [939] Chen J, Guo Y and Wen Y 2013 Two-stage metal-catalyst-free [963] Yang W et al 2013 Epitaxial growth of single-domain growth of high-quality polycrystalline graphene films on graphene on hexagonal boron nitride Nat. Mater. 12 792–7 silicon nitride substrates Adv. Mater. 25 992 [964] Tang S, Wang H and Zhang Y 2013 Precisely aligned graphene [940] Hwang J et al 2013 van der Waals epitaxial growth of graphene grown on hexagonal boron nitride by catalyst free chemical on sapphire by chemical vapor deposition without a metal vapor deposition Sci. Rep. 3 2666 catalyst ACS Nano 7 385–95 [965] Zhang L, Shi Z and Liu D 2012 Vapour-phase graphene [941] Song H J, Son M and Park C 2012 Large scale metal-free epitaxy at low temperatures Nano Res. 5 258 synthesis of graphene on sapphire and transfer-free device [966] Bo Z, Yang Y and Chen J 2013 Plasma-enhanced chemical fabrication Nanoscale 4 3050 vapor deposition synthesis of vertically oriented graphene [942] Jerng S K, Yu D S and Kim Y S 2011 Nanocrystalline graphite nanosheets Nanoscale 5 5180 growth on sapphire by carbon molecular beam epitaxy J. [967] Buijnsters J G, Camero M and Gago R 2008 Direct Phys. Chem. C 115 4491 spectroscopic evidence of self-formed C60 inclusions in [943] Fanton M A et al 2011 Characterization of graphene films and fullerenelike hydrogenated carbon films Appl. Phys. Lett. transistors grown on sapphire by metal-free chemical vapor 92 141920 deposition ACS Nano 5 8062–9 [968] Tamor M A, Vassell W C and Carduner K R 1990 Atomic [944] Sun J, Lindvall N and Cole M T 2012 Controllable chemical constraint in hydrogenated ‘diamond-like’ carbon Appl. vapor deposition of large area uniform nanocrystalline Phys. Lett. 58 592 graphene directly on silicon dioxide J. Appl. Phys. [969] Buijnsters J G, Camero M and Vázquez L 2007 DC substrate 111 044103 bias effects on the physical properties of hydrogenated [945] Chen J et al 2011 Oxygen-aided synthesis of polycrystalline amorphous carbon films grown by plasma-assisted chemical graphene on silicon dioxide substrates J. Am. Chem. Soc. vapour deposition Vacuum 81 1412 133 17548–51 [970] Zhang L, Ni M and Liu D 2012 Competitive growth and [946] Chen J et al 2014 Near-equilibrium chemical vapor etching of epitaxial graphene J. Phys. Chem. C 116 26929 deposition of high-quality single-crystal graphene directly on [971] Yang R, Zhang L and Wang Y 2010 An anisotropic etching various dielectric substrates Adv. Mater. 26 1348 effect in the graphene basal plane Adv. Mater. 22 4014 [947] Ding X, Ding G, Xie X, Huang F and Jiang M 2011 Direct [972] Diankov G, Neumann M and Goldhaber-Gordon D 2013 growth of few layer graphene on hexagonal boron nitride by Extreme monolayer-selectivity of hydrogen-plasma reactions chemical vapor deposition Carbon 49 2522–5 with graphene ACS Nano 7 1324–32 [948] Rümmeli M H, Bachmatiuk A and Scott A 2010 Direct low- [973] Nishijima M, Yoshinobu J and Tsuda H 1987 The adsorption temperature nanographene CVD synthesis over a dielectric and thermal decomposition of acetylene on Si(1 0 0) and insulator ACS Nano 4 4206 vicinal Si(1 0 0) 9° Surf. Sci. 192 383 [949] Sun J et al 2014 Direct growth of high-quality graphene [974] Riikonen S, Krasheninnikov A V and Halonen L 2012 The role on high-κ dielectric SrTiO3 substrates J. Am. Chem. Soc. of stable and mobile carbon adspecies in copper-promoted 136 6574–7 graphene growth J. Phys. Chem. C 116 5802 263 2D Mater. 7 (2020) 022001 C Backes et al [975] Zhang J, Wang Z and Niu T 2014 Elementary process for CVD [997] Sun Z, Yan Z, Yao J, Beitler E, Zhu Y and Tour J M 2010 Growth graphene on Cu(1 1 0): size-selective carbon clusters Sci. Rep. of graphene from solid carbon sources Nature 468 549–52 4 4431 [998] Mattevi C, Kim H and Chhowalla M 2011 A review of [976] Koretsky C M, Sverjensky D A and Sahai N 1998 A model chemical vapour deposition of graphene on copper J. Mater. of surface site types on oxide and silicate minerals based on Chem. 21 3324–34 crystal chemistry; implications for site types and densities, [999] ASM International 2002 A.P. ASM Handbook vol 3 multi-site adsorption, surface infrared spectroscopy, and [1000] Yu Q, Lian J, Siriponglert S, Li H, Chen Y P and Pei S-S 2008 dissolution kinetics Am. J. Sci. 298 349–438 Graphene segregated on Ni surfaces and transferred to [977] Liu D, Yang W, Zhang L, Zhang J, Meng J, Yang R, Zhang G insulators Appl. Phys. Lett. 93 113103 and Shi D 2014 Two-step growth of graphene with separate [1001] Kaplas T, Matikainen A, Nuutinen T, Suvanto S, Vahimaa P controlling nucleation and edge growth directly on SiO2 and Svirko Y 2017 Scalable fabrication of the graphitic substrates Carbon 72 387–92 substrates for graphene-enhanced Raman spectroscopy Sci. [978] Kim Y S, Joo K, Jerng S-K, Lee J H, Yoon E and Chun S-H Rep. 7 8561 2014 Direct growth of patterned graphene on SiO2 substrates [1002] Xiong W, Zhou Y S, Jiang L J, Sarkar A, Mahjouri-Samani M, without the use of catalysts or lithography Nanoscale Xie Z Q, Gao Y, Ianno N J, Jiang L and Lu Y F 2012 Single-step 6 10100–5 formation of graphene on dielectric surfaces Adv. Mater. [979] Sun J, Schmidt M E, Muruganathan M, Chong H M H and 25 630–4 Mizuta H 2016 Large-scale nanoelectromechanical switches [1003] Reina A, Son H, Jiao L, Fan B, Dresselhaus M S, Liu Z and based on directly deposited nanocrystalline graphene on Kong J 2008 Transferring and identification of single- and insulating substrates Nanoscale 8 6659–65 few-layer graphene on arbitrary substrates J. Phys. Chem. C [980] Chugh S, Mehta R, Lu N, Dios F D, Kim M J and Chen Z 112 17741–4 2015 Comparison of graphene growth on arbitrary non- [1004] Hallam T, Berner N C, Yim C and Duesberg G S 2014 Strain, catalytic substrates using low-temperature PECVD Carbon bubbles, dirt, and folds: a study of graphene polymer- 93 393–9 assisted transfer Adv. Mater. Interfaces 1 1400115 [981] Muñoz R, Martínez L, López-Elvira E, Munuera C, Huttel Y [1005] Li X, Zhu Y, Cai W, Borysiak M, Han B, Chen D, Piner R D, and García-Hernández M 2018 Direct synthesis of graphene Colombo L and Ruoff R S 2009 Transfer of large-area on silicon oxide by low temperature plasma enhanced graphene films for high-performance transparent chemical vapor deposition Nanoscale 10 12779–87 conductive electrodes Nano Lett. 9 4359–63 [982] Elias D C et al 2009 Control of graphene’s properties by [1006] Zhang X, Beyer A and Gölzhäuser A 2011 Mechanical reversible hydrogenation: evidence for graphane Science characterization of carbon nanomembranes from self- 323 610–3 assembled monolayers Beilstein J. Nanotechnol. 2 826–33 [983] Wang Y, Xu X, Lu J, Lin M, Bao Q, Özyilmaz B and Loh K P [1007] Beyer A, Vieker H, Klett R, Meyer zu Theenhausen H, 2010 Toward high throughput interconvertible graphane-to- Angelova P and Gölzhäuser A 2015 Imaging of carbon graphene growth and patterning ACS Nano 4 6146–52 nanomembranes with helium ion microscopy Beilstein J. [984] Kaplas T, Sharma D and Svirko Y 2012 Few-layer graphene Nanotechnol. 6 1712–20 synthesis on a dielectric substrate Carbon 50 1503–9 [1008] Lee C-K et al 2014 Monatomic chemical–vapor–deposited [985] Kaplas T, Zolotukhin A and Svirko Y 2011 Thickness graphene membranes bridge a half-millimeter-scale gap determination of graphene on metal substrate by reflection ACS Nano 8 2336–44 spectroscopy Opt. Express 19 17226 [1009] Gong C et al 2013 Rapid selective etching of PMMA residues [986] Chavez K L and Hess D W 2001 A novel method of etching from transferred graphene by carbon dioxide J. Phys. Chem. copper oxide using acetic acid J. Electrochem. Soc. 148 G640 C 117 23000–8 [987] Kaplas T and Svirko Y 2014 Self-assembled graphene on [1010] Lin Y-C, Lu C-C, Yeh C-H, Jin C, Suenaga K and Chiu P-W dielectric micro- and nanostructures Carbon 70 273–8 2011 Graphene annealing: how clean can it be? Nano Lett. [988] Kosterlitz J M and Thouless D J 1972 Long range order and 12 414–9 metastability in two dimensional solids and superfluids [1011] Islam A E, Zakharov D N, Carpena-Nunez J, Hsiao M-S, (application of dislocation theory) J. Phys. C: Solid State Phys. Drummy L F, Stach E A and Maruyama B 2017 Atomic 5 L124–6 level cleaning of poly-methyl-methacrylate residues [989] Vlassiouk I, Smirnov S, Ivanov I, Fulvio P F, Dai S, Meyer H, from the graphene surface using radiolized water at high Chi M, Hensley D, Datskos P and Lavrik N V 2011 Electrical temperatures Appl. Phys. Lett. 111 103101 and thermal conductivity of low temperature CVD graphene: [1012] Zhang Z, Du J, Zhang D, Sun H, Yin L, Ma L, Chen J, Ma D, the effect of disorder Nanotechnology 22 275716 Cheng H-M and Ren W 2017 Rosin-enabled ultraclean and [990] Kumar R, Mehta B R, Bhatnagar M, Ravi S, Mahapatra S, damage-free transfer of graphene for large-area flexible Salkalachen S and Jhawar P 2014 Graphene as a transparent organic light-emitting diodes Nat. Commun. 8 14560 conducting and surface field layer in planar Si solar cells [1013] Fisichella G, Di Franco S, Roccaforte F, Ravesi S and Nanoscale Res. Lett. 9 349 Giannazzo F 2014 Microscopic mechanisms of graphene [991] Jawhari T, Roid A and Casado J 1995 Raman spectroscopic electrolytic delamination from metal substrates Appl. Phys. characterization of some commercially available carbon black Lett. 104 233105 materials Carbon 33 1561–5 [1014] Gao Y et al 2015 Elastic coupling between layers in two- [992] Kaplas T, Karvonen L, Ahmadi S, Amirsolaimani B, dimensional materials Nat. Mater. 14 714 Mehravar S, Peyghambarian N, Kieu K, Honkanen S, [1015] Liu L et al 2016 Graphene transfer: a mechanism for highly Lipsanen H and Svirko Y 2016 Optical characterization of efficient electrochemical bubbling delamination of CVD- directly deposited graphene on a dielectric substrate Opt. grown graphene from metal substrates Adv. Mater. Interfaces Express 24 2965 3 1500492 [993] Cassie A B D and Baxter S 1944 Wettability of porous surfaces [1016] Chen Z, Ge X, Zhang H, Zhang Y, Sui Y, Yu G, Jin Z and Liu X Trans. Faraday Soc. 40 546 2016 High pressure-assisted transfer of ultraclean chemical [994] Peng Z, Yan Z, Sun Z and Tour J M 2011 Direct growth of vapor deposited graphene Appl. Phys. Lett. 108 132106 bilayer graphene on SiO2 substrates by carbon diffusion [1017] Lee J, Kim Y, Shin H-J, Lee C, Lee D, Moon C-Y, Lim J and through nickel ACS Nano 5 8241–7 Chan Jun S 2013 Clean transfer of graphene and its effect on [995] Marchena M, Janner D, Chen T L, Finazzi V and Pruneri V contact resistance Appl. Phys. Lett. 103 103104 2016 Low temperature direct growth of graphene patterns on [1018] Jeong H J, Kim H Y, Jeong S Y, Han J T, Baeg K-J, Hwang J Y flexible glass substrates catalysed by a sacrificial ultrathin Ni and Lee G-W 2014 Improved transfer of chemical-vapor- film Opt. Mater. Express 6 2487 deposited graphene through modification of intermolecular [996] Shin H-J et al 2011 Transfer-free growth of few-layer graphene interactions and solubility of poly(methylmethacrylate) by self-assembled monolayers Adv. Mater. 23 4392–7 layers Carbon 66 612–8 264 2D Mater. 7 (2020) 022001 C Backes et al [1019] Koefoed L et al 2015 Facile electrochemical transfer of large- [1042] Kwon S-J et al 2018 Extremely stable graphene electrodes area single crystal epitaxial graphene from Ir(1 1 1) J. Phys. D: doped with macromolecular acid Nat. Commun. 9 2037 Appl. Phys. 48 115306 [1043] Kim H H, Chung Y, Lee E, Lee S K and Cho K 2014 Graphene: [1020] Miniussi E et al 2017 Fermi surface map of large-scale single- water-free transfer method for CVD-grown graphene and orientation graphene on SiO2 J. Phys.: Condens. Matter its application to flexible air-stable graphene transistors Adv. 29 475001 Mater. 26 3166–6 [1021] Krajewska A, Pasternak I, Sobon G, Sotor J, Przewloka A, [1044] Batrakov K, Kuzhir P, Maksimenko S, Paddubskaya A, Ciuk T, Sobieski J, Grzonka J, Abramski K M and Voronovich S, Lambin P, Kaplas T and Svirko Y 2014 Flexible Strupinski W 2017 Fabrication and applications of multi- transparent graphene/polymer multilayers for efficient layer graphene stack on transparent polymer Appl. Phys. Lett. electromagnetic field absorption Sci. Rep. 4 7191 110 041901 [1045] Kotsilkova R, Todorov P, Ivanov E, Kaplas T, Svirko Y, [1022] Pirkle A, Chan J, Venugopal A, Hinojos D, Magnuson C W, Paddubskaya A and Kuzhir P 2016 Mechanical properties McDonnell S, Colombo L, Vogel E M, Ruoff R S and investigation of bilayer graphene/poly(methyl methacrylate) Wallace R M 2011 The effect of chemical residues on thin films at macro, micro and nanoscale Carbon 100 355–66 the physical and electrical properties of chemical vapor [1046] Kaplas T, Bera A, Matikainen A, Pääkkönen P and deposited graphene transferred to SiO2 Appl. Phys. Lett. Lipsanen H 2018 Transfer and patterning of chemical vapor 99 122108 deposited graphene by a multifunctional polymer film Appl. [1023] Taubner T, Keilmann F and Hillenbrand R 2005 Nanoscale- Phys. Lett. 112 073107 resolved subsurface imaging by scattering-type near-field [1047] Niu T and Li A 2015 From two-dimensional materials to optical microscopy Opt. Express 13 8893 heterostructures Prog. Surf. Sci. 90 21–45 [1024] Hillenbrand F K A R 2008 Nano-Optics and Near-Field [1048] Schué L, Stenger I, Fossard F, Loiseau A and Barjon J 2016 Optical Microscopy (Boston, MA: Artech House Publishers) Characterization methods dedicated to nanometer-thick [1025] Hillenbrand R and Keilmann F 2002 Material-specific hBN layers 2D Mater. 4 015028 mapping of metal/semiconductor/dielectric nanosystems [1049] Kim K K et al 2011 Synthesis of monolayer hexagonal boron at 10 nm resolution by backscattering near-field optical nitride on cu foil using chemical vapor deposition Nano Lett. microscopy Appl. Phys. Lett. 80 25–7 12 161–6 [1026] Hillenbrand R and Keilmann F 2000 Complex optical [1050] Chatterjee S, Kim M J, Zakharov D N, Kim S M, Stach E A, constants on a subwavelength scale Phys. Rev. Lett. 85 3029–32 Maruyama B and Sneddon L G 2012 Syntheses of boron [1027] Taubner T, Hillenbrand R and Keilmann F 2004 Nanoscale nitride nanotubes from borazine and decaborane molecular polymer recognition by spectral signature in scattering precursors by catalytic chemical vapor deposition with a infrared near-field microscopy Appl. Phys. Lett. 85 5064–6 floating nickel catalyst Chem. Mater. 24 2872–9 [1028] Gunde M K 2000 Vibrational modes in amorphous silicon [1051] Jang S K, Youn J, Song Y J and Lee S 2016 Synthesis and dioxide Physica B 292 286–95 characterization of hexagonal boron nitride as a gate [1029] Huth F, Govyadinov A, Amarie S, Nuansing W, Keilmann F dielectric Sci. Rep. 6 30449 and Hillenbrand R 2012 Nano-FTIR absorption [1052] Gibb A, Alem N and Zettl A 2013 Low pressure spectroscopy of molecular fingerprints at 20 nm spatial chemical vapor deposition synthesis of hexagonal boron resolution Nano Lett. 12 3973–8 nitride on polycrystalline metal foils Phys. Status Solidi b [1030] Venugopal A 2012 Effect of Contacts, Graphene Type and 250 2727–31 Underlying Substrate on The Transport Properties Of [1053] Suzuki S and Hibino H 2012 Chemical vapor deposition of Graphene (The University of Texas at Dallas) hexagonal boron nitride e-J. Surf. Sci. Nanotechnol. 10 133–8 [1031] Batzill M 2012 The surface science of graphene: Metal [1054] Ismach A et al 2012 Toward the controlled synthesis of interfaces, CVD synthesis, nanoribbons, chemical hexagonal boron nitride films ACS Nano 6 6378–85 modifications, and defects Surf. Sci. Rep. 67 83–115 [1055] Ismach A, Chou H, Mende P, Dolocan A, Addou R, Aloni S, [1032] Simonet J and Lund H 1977 Electrochemical behaviour of Wallace R, Feenstra R, Ruoff R S and Colombo L 2017 graphite cathodes in the presence of tetraalkylammonium Carbon-assisted chemical vapor deposition of hexagonal cations J. Electroanal. Chem. Interfacial Electrochem. boron nitride 2D Mater. 4 025117 75 719–30 [1056] Sonde S, Dolocan A, Lu N, Corbet C, Kim M J, Tutuc E, [1033] Verguts K et al 2017 Controlling water intercalation is key Banerjee S K and Colombo L 2017 Ultrathin, wafer-scale to a direct graphene transfer ACS Appl. Mater. Interfaces hexagonal boron nitride on dielectric surfaces by diffusion 9 37484–92 and segregation mechanism 2D Mater. 4 025052 [1034] Gómez-Marín A M, Clavilier J and Feliu J M 2013 [1057] Cho H, Park S, Won D-I, Kang S O, Pyo S-S, Kim D-I, Sequential Pt(1 1 1) oxide formation in perchloric acid: an Kim S M, Kim H C and Kim M J 2015 Growth kinetics of electrochemical study of surface species inter-conversion J. white graphene (h-BN) on a planarised Ni foil surface Sci. Electroanal. Chem. 688 360–70 Rep. 5 11985 [1035] Palacio I et al 2018 Chemistry below graphene: Decoupling [1058] Lee Y-H et al 2012 Growth selectivity of hexagonal-boron epitaxial graphene from metals by potential-controlled nitride layers on Ni with various crystal orientations RSC electrochemical oxidation Carbon 129 837–46 Adv. 2 111–5 [1036] Feng X, Maier S and Salmeron M 2012 Water splits epitaxial [1059] Andrieux-Ledier N D A, Fossard F, Mérot J S, Prevost H, graphene and intercalates J. Am. Chem. Soc. 134 5662–8 Schué L, Plaud A, Héripré E, Barjon J and Loiseau A 2017 [1037] Alonso C, Alonso C G-Aand Escudero M L 2008 Spanish Under review Patent ES-200801041 [1060] Shi Y et al 2010 Synthesis of few-layer hexagonal boron [1038] Bae S et al 2010 2010 Roll-to-roll production of 30-inch nitride thin film by chemical vapor deposition Nano Lett. graphene films for transparent electrodes Nat. Nanotechnol. 10 4134–9 5 574–8 [1061] Gao Y, Ren W, Ma T, Liu Z, Zhang Y, Liu W-B, Ma L-P, Ma X [1039] Kang M H, Milne W I and Cole M T 2016 Temporal stability and Cheng H-M 2013 Repeated and controlled growth of of metal-chloride-doped chemical–vapour–deposited monolayer, bilayer and few-layer hexagonal boron nitride on graphene ChemPhysChem 17 2545–50 Pt foils ACS Nano 7 5199–206 [1040] Kang M H, Prieto López L O, Chen B, Teo K, Williams J A, [1062] Corso M 2004 Boron nitride nanomesh Science 303 217–20 Milne W I and Cole M T 2016 Mechanical robustness of [1063] Hemmi A et al 2014 High quality single atomic layer graphene on flexible transparent substrates ACS Appl. Mater. deposition of hexagonal boron nitride on single crystalline Interfaces 8 22506–15 Rh(1 1 1) four-inch wafers Rev. Sci. Instrum. 85 035101 [1041] Kim K K, Reina A, Shi Y, Park H, Li L-J, Lee Y H and Kong J [1064] Mertens S F L, Hemmi A, Muff S, Gröning O, De Feyter S, 2010 Enhancing the conductivity of transparent graphene Osterwalder J and Greber T 2016 Switching stiction and films via doping Nanotechnology 21 285205 adhesion of a liquid on a solid Nature 534 676–9 265 2D Mater. 7 (2020) 022001 C Backes et al [1065] Widmer R, Berner S, Gröning O, Brugger T, Osterwalder J [1085] Caretti I, Torres R, Gago R, Landa-Cánovas A R and and Greber T 2007 Electrolytic in situ STM investigation of Jiménez I 2010 Effect of carbon incorporation on the h-BN-Nanomesh Electrochem. Commun. 9 2484–8 microstructure of BCxN (x  =  0.25 1, and 4) ternary solid [1066] Cun H et al 2018 Centimeter-sized single-orientation solutions studied by transmission electron microscopy monolayer hexagonal boron nitride with or without Chem. Mater. 22 1949–51 nanovoids Nano Lett. 18 1205–12 [1086] Shim J et al 2018 Controlled crack propagation for atomic [1067] Berner S et al 2007 Boron nitride nanomesh: functionality precision handling of wafer-scale two-dimensional materials from a corrugated monolayer Angew. Chem., Int. Ed. Science 362 665–70 46 5115–9 [1087] Song L et al 2010 Large scale growth and characterization of [1068] Brugger T, Ma H, Iannuzzi M, Berner S, Winkler A, Hutter J, atomic hexagonal boron nitride layers Nano Lett. 10 3209–15 Osterwalder J and Greber T 2010 Nanotexture switching [1088] Wen Y, Shang X, Dong J, Xu K, He J and Jiang C 2015 of single-layer hexagonal boron nitride on rhodium by Ultraclean and large-area monolayer hexagonal boron intercalation of hydrogen atoms Angew. Chem., Int. Ed. nitride on Cu foil using chemical vapor deposition 49 6120–4 Nanotechnology 26 275601 [1069] Gubó R, Vári G, Kiss J, Farkas A P, Palotás K, Óvári L, Berkó A [1089] Park J-H et al 2014 Large-area monolayer hexagonal boron and Kónya Z 2018 Tailoring the hexagonal boron nitride nitride on Pt foil ACS Nano 8 8520–8 nanomesh on Rh(1 1 1) with gold Phys. Chem. Chem. Phys. [1090] Kim G, Jang A R, Jeong H Y, Lee Z, Kang D J and Shin H S 20 15473–85 2013 Growth of high-crystalline, single-layer hexagonal [1070] Stania R et al 2016 Self-assembly of nanoscale lateral boron nitride on recyclable platinum foil Nano Lett. segregation profiles, Phys. Rev. B 93 161402(R) 13 1834–9 [1071] Cho Y-J et al 2016 Hexagonal boron nitride tunnel barriers [1091] Caneva S et al 2016 Controlling catalyst bulk reservoir effects grown on graphite by high temperature molecular beam for monolayer hexagonal boron nitride CVD Nano Lett. epitaxy Sci. Rep. 6 34474 16 1250–61 [1072] Nakhaie S, Wofford J M, Schumann T, Jahn U, Ramsteiner M, [1092] Wang R et al 2016 Catalyst interface engineering for Hanke M, Lopes J M J and Riechert H 2015 Synthesis of improved 2D film lift-off and transfer ACS Appl. Mater. atomically thin hexagonal boron nitride films on nickel foils Interfaces 8 33072–82 by molecular beam epitaxy Appl. Phys. Lett. 106 213108 [1093] Gatensby R, McEvoy N, Lee K, Hallam T, Berner N C, [1073] Wofford J M, Nakhaie S, Krause T, Liu X, Ramsteiner M, Rezvani E, Winters S, O’Brien M and Duesberg G S 2014 Hanke M, Riechert H and Lopes J M J 2017 A hybrid MBE- Controlled synthesis of transition metal dichalcogenide thin based growth method for large-area synthesis of stacked films for electronic applications Appl. Surf. Sci. 297 139–46 hexagonal boron nitride/graphene heterostructures Sci. Rep. [1094] Gatensby R, Hallam T, Lee K, McEvoy N and Duesberg G S 7 43644 2016 Investigations of vapour-phase deposited transition [1074] Jang A R et al 2016 Wafer-scale and wrinkle-free epitaxial metal dichalcogenide films for future electronic applications growth of single-orientated multilayer hexagonal boron Solid-State Electron. 125 39–51 nitride on sapphire Nano Lett. 16 3360–6 [1095] Jung Y, Shen J, Liu Y, Woods J M, Sun Y and Cha J J 2014 [1075] Kobayashi Y and Akasaka T 2008 Hexagonal BN epitaxial Metal seed layer thickness-induced transition from vertical growth on (0 0 0 1) sapphire substrate by MOVPE J. Cryst. to horizontal growth of MoS2 and WS2, Nano Lett. 14 6842–9 Growth 310 5044–7 [1096] Lee K, Gatensby R, McEvoy N, Hallam T and Duesberg G S [1076] Caretti I, Jiménez I and Albella J M 2003 BCN films with 2013 High-performance sensors based on molybdenum controlled composition obtained by the interaction between disulfide thin films Adv. Mater. 25 6699–702 molecular beams of B and C with nitrogen ion beams [1097] O’Brien M, Lee K, Morrish R, Berner N C, McEvoy N, Diamond Relat. Mater. 12 1079–83 Wolden C A and Duesberg G S 2014 Plasma assisted [1077] Caretti I and Jiménez I 2011 Composition and bonding synthesis of WS2 for gas sensing applications Chem. Phys. structure of boron nitride B1−xNx thin films grown by ion- Lett. 615 6–10 beam assisted evaporation Chem. Phys. Lett. 511 235–40 [1098] O’Brien M, McEvoy N, Hallam T, Kim H-Y, Berner N C, [1078] Jiménez I, Torres R, Caretti I, Gago R and Albella J M 2012 Hanlon D, Lee K, Coleman J N and Duesberg G S 2014 A review of monolithic and multilayer coatings within Transition metal dichalcogenide growth via close proximity the boron–carbon–nitrogen system by ion-beam-assisted precursor supply Sci. Rep. 4 7374 deposition J. Mater. Res. 27 743–64 [1099] Yim C et al 2016 High-performance hybrid electronic [1079] Caretti I and Jiménez I 2011 Point defects in hexagonal BN, devices from layered PtSe2 films grown at low temperature BC3 and BCxN compounds studied by x-ray absorption ACS Nano 10 9550–8 near-edge structure J. Appl. Phys. 110 023511 [1100] Zhang S et al 2015 Direct observation of degenerate two- [1080] Torres R, Caretti I, Serin V, Brun N, Radnóczic G and photon absorption and its saturation in WS2 and MoS2 Jiménez I 2014 Reversed texture in nanometric carbon/ monolayer and few-layer films ACS Nano 9 7142–50 boron nitride multilayers Carbon 74 374–8 [1101] Yim C, McEvoy N and Duesberg G S 2013 Characterization [1081] Cascales Fernandez J 2016 Estudio de capas nanométricas of graphene-silicon Schottky barrier diodes using de h-BN y grafeno para su apilamiento en multicapas (study impedance spectroscopy Appl. Phys. Lett. 103 193106 of nanometric layers of h-BN and graphene for piling up as [1102] Yim C et al 2014 Heterojunction hybrid devices from vapor multilayers) PhD Thesis Universidad Autonoma de Madrid phase grown MoS2, Sci. Rep. 4 5458 [1082] Gago R, Abendroth B, Cerdá J I, Jiménez I and Möller W [1103] Yim C, O’Brien M, McEvoy N, Winters S, Mirza I, 2007 Detection of intrinsic stress in cubic boron nitride films Lunney J G and Duesberg G S 2014 Investigation of the by x-ray absorption near-edge structure: Stress relaxation optical properties of MoS2 thin films using spectroscopic mechanisms by simultaneous ion implantation during ellipsometry Appl. Phys. Lett. 104 103114 growth Phys. Rev. B 76 174111 [1104] Bandurin D A et al 2017 High electron mobility, quantum [1083] Gago R, Jiménez I, Albella J M and Terminello L J 2001 Hall effect and anomalous optical response in atomically Identification of ternary boron–carbon–nitrogen hexagonal thin InSe Nat. Nanotechnol. 12 223–7 phases by x-ray absorption spectroscopy Appl. Phys. Lett. [1105] Huang W, Gan L, Li H, Ma Y and Zhai T 2016 2D layered 78 3430–2 group IIIA metal chalcogenides: synthesis, properties [1084] Jiménez I, Jankowski A F, Terminello L J, Sutherland D G J, and applications in electronics and optoelectronics Carlisle J A, Doll G L, Tong W M, Shuh D K and Himpsel F J CrystEngComm 18 3968–84 1997 Core-level photoabsorption study of defects and [1106] Lei S et al 2016 Surface functionalization of two-dimensional metastable bonding configurations in boron nitride Phys. metal chalcogenides by Lewis acid-base chemistry Nat. Rev. B 55 12025–37 Nanotechnol. 11 465–71 266 2D Mater. 7 (2020) 022001 C Backes et al [1107] Lei S et al 2015 An atomically layered InSe avalanche [1129] Zhao J, Nam H, Ly T H, Yun S J, Kim S, Cho S, Yang H photodetector Nano Lett. 15 3048–55 and Lee Y H 2016 Chain vacancies in 2D crystals Small [1108] Mudd G W et al 2015 High broad-band photoresponsivity 13 1601930 of mechanically formed InSe-graphene van der Waals [1130] Dumcenco D et al 2015 Large-area epitaxial monolayer heterostructures Adv. Mater. 27 3760–6 MoS2 ACS Nano 9 4611–20 [1109] Debbichi L, Eriksson O and Lebegue S 2015 Two- [1131] Najmaei S, Liu Z, Zhou W, Zou X, Shi G, Lei S, Yakobson B I, dimensional indium selenides compounds: an ab initio Idrobo J-C, Ajayan P M and Lou J 2013 Vapour phase growth study J. Phys. Chem. Lett. 6 3098–103 and grain boundary structure of molybdenum disulphide [1110] Rybkovskiy D V, Osadchy A V and Obraztsova E D 2014 atomic layers Nat. Mater. 12 754–9 Transition from parabolic to ring-shaped valence band [1132] van der Zande A M, Huang P Y, Chenet D A, Berkelbach T C, maximum in few-layer GaS, GaSe, and InSe Phys. Rev. B You Y, Lee G-H, Heinz T F, Reichman D R, Muller D A 90 235302 and Hone J C 2013 Grains and grain boundaries in highly [1111] Sánchez-Royo J F et al 2014 Electronic structure, optical crystalline monolayer molybdenum disulphide Nat. Mater. properties, and lattice dynamics in atomically thin indium 12 554–61 selenide flakes Nano Res. 7 1556–68 [1133] Kim H, Dumcenco D, Frégnaux M, Benayad A, Chen M- [1112] Wickramaratne D, Zahid F and Lake R K 2015 Electronic and W, Kung Y-C, Kis A and Renault O 2016 Free-standing thermoelectric properties of van der Waals materials with electronic character of monolayerMoS2 in van der Waals ring-shaped valence bands J. Appl. Phys. 118 075101 epitaxy Phys. Rev. B 94 081401(R) [1113] Zólyomi V, Drummond N D and Fal’ko V I 2014 Electrons [1134] Boxhoorn G, Jesse A, Ernsting J and Oskam A 1978 Vapour and phonons in single layers of hexagonal indium pressure measurements on M (CO) 5PX3 (M  =  Cr, Mo, W; chalcogenides from ab initio calculations Phys. Rev. B X  =  F, Cl, Br) Thermochim. Acta 27 261–7 89 205416 [1135] Liu Z, Li X, Close M R, Kugler E L, Petersen J L and [1114] Han G, Chen Z G, Drennan J and Zou J 2014 Indium Dadyburjor D B 1997 Screening of alkali-promoted vapor- selenides: structural characteristics, synthesis and their phase-synthesized molybdenum sulfide catalysts for the thermoelectric performances Small 10 2747–65 production of alcohols from synthesis gas Ind. Eng. Chem. [1115] Lin M et al 2013 Controlled growth of atomically thin In2Se3 Res. 36 3085–93 flakes by van der Waals epitaxy J. Am. Chem. Soc. 135 13274– [1136] Close M R, Petersen J L and Kugler E L 1999 Synthesis and 7 characterization of nanoscale molybdenum sulfide catalysts [1116] Zhou J, Zeng Q, Lv D, Sun L, Niu L, Fu W, Liu F, Shen Z, by controlled gas phase decomposition of Mo(CO)6 and H2S Jin C and Liu Z 2015 Controlled synthesis of high-quality Inorg. Chem. 38 1535–42 monolayered alpha-In2Se3 via physical vapor deposition [1137] Shi Y et al 2012 van der Waals epitaxy of MoS2 layers using Nano Lett. 15 6400–5 graphene as growth templates Nano Lett. 12 2784–91 [1117] Balakrishnan N et al 2016 Quantum confinement and [1138] Hara K, Kojima T and Kukimoto H 1987 Epitaxial growth of photoresponsivity of β-In2Se3 nanosheets grown by physical cugas2 by metalorganic chemical vapor deposition Japan. J. vapour transport 2D Mater. 3 025030 Appl. Phys. 26 L1107–9 [1118] Zhou S, Tao X and Gu Y 2016 Thickness-dependent thermal [1139] Zhang X, Choudhury T H, Chubarov M, Xiang Y, Jariwala B, conductivity of suspended two-dimensional single-crystal Zhang F, Alem N, Wang G-C, Robinson J A and Redwing J M In2Se3 layers grown by chemical vapor deposition J. Phys. 2018 Diffusion-controlled epitaxy of large area coalesced Chem. C 120 4753–8 WSe2 monolayers on sapphire Nano Lett. 18 1049–56 [1119] Lauth J et al 2016 Solution-processed two-dimensional [1140] Kang K, Xie S, Huang L, Han Y, Huang P Y, Mak K F, Kim C-J, ultrathin InSe nanosheets Chem. Mater. 28 1728–36 Muller D and Park J 2015 High-mobility three-atom-thick [1120] Balakrishnan N, Steer E D, Smith E F, Kudrynskyi Z R, semiconducting films with wafer-scale homogeneity Nature Kovalyuk Z D, Eaves L, Patanè A and Beton P H 2018 520 656–60 Epitaxial growth of γ-InSe and α, β, and γ-In2Se3 on ε-GaSe [1141] Kranthi Kumar V, Dhar S, Choudhury T H, Shivashankar S A 2D Mater. 5 035026 and Raghavan S 2015 A predictive approach to CVD of [1121] Tao X and Gu Y 2013 Crystalline-crystalline phase crystalline layers of TMDs: the case of MoS2 Nanoscale transformation in two-dimensional In2Se3 thin layers Nano 7 7802–10 Lett. 13 3501–5 [1142] Kim H, Ovchinnikov D, Deiana D, Unuchek D and Kis A [1122] Huang Y-T, Huang C-W, Chen J-Y, Ting Y-H, Lu K-C, 2017 Suppressing nucleation in metal–organic chemical Chueh Y-L and Wu W-W 2014 Dynamic observation of vapor deposition of MoS2 monolayers by Alkali metal phase transformation behaviors in indium(III) selenide halides Nano Lett. 17 5056–63 nanowire based phase change memory ACS Nano 8 9457–62 [1143] Wei W-C J and Lo M-H 1998 Processing and properties [1123] Jacobs-Gedrim R B, Shanmugam M, Jain N, Durcan C A, of (Mo,Cr) oxycarbides from MOCVD Appl. Organomet. Murphy M T, Murray T M, Matyi R J, Moore R L and Yu B Chem. 12 201–20 2014 Extraordinary photoresponse in two-dimensional [1144] Kaplan L H and d’Heurle F M 1970 The deposition of In2Se3 nanosheets ACS Nano 8 514–21 molybdenum and tungsten films from vapor decomposition [1124] Mudd G W et al 2013 Tuning the bandgap of exfoliated InSe of carbonyls J. Electrochem. Soc. 117 693 nanosheets by quantum confinement Adv. Mater. 25 5714–8 [1145] Stubbles J R and Richardson F D 1960 Equilibria in the [1125] Kovalyuk Z D, Sydor O M, Sydor O A, Tkachenko V G, system molybdenum  +  sulphur  +  hydrogen Trans. Faraday Maksymchuk I M, Dubinko V I and Ostapchuk P M 2012 Soc. 56 1460 X-ray diffractometry and raman spectroscopy investigation [1146] Kaloidas V E and Papayannakos N G 1987 Hydrogen of irradiated layered III-VI crystals J. Mater. Sci. Eng. A production from the decomposition of hydrogen sulphide. 2 537–43 Equilibrium studies on the system H2S/H2/Si, (i  =  1,…,8) in [1126] Bakhtinov A P, Boledzyuk V B, Kovalyuk Z D, the gas phase Int. J. Hydrogen Energy 12 403–9 Kudrynskyi Z R, Lytvyn O S and Shevchenko A D 2013 [1147] Dhar S, Kranthi Kumar V, Choudhury T H, Shivashankar S A Magnetic properties and surface morphology of layered and Raghavan S 2016 Chemical vapor deposition of MoS2 In2Se3 crystals intercalated with cobalt Phys. Solid State layers from Mo-S-C-O-H system: thermodynamic modeling 55 1148–55 and validation Phys. Chem. Chem. Phys. 18 14918–26 [1127] Ke F et al 2014 Interlayer-glide-driven isosymmetric [1148] Fannin L W, Webb D W and Pearce R H 1992 Cause phase transition in compressed In2Se3 Appl. Phys. Lett. and resolution of problem of erratic and decreasing 104 212102 trimethylindium bubbler delivery rates J. Cryst. Growth [1128] Nakayama T and Hishikawa M 1997 Bonding and optical 124 307–10 anisotropy of vacancy ordered Ga2Se3 J. Phys. Soc. Japan [1149] Koma A 1999 Van der Waals epitaxy for highly lattice- 66 3887–92 mismatched systems J. Cryst. Growth 201–2 236–41 267 2D Mater. 7 (2020) 022001 C Backes et al [1150] Koma A 1992 Van der Waals epitaxy—a new epitaxial growth precursors Mo(S-t-Bu)4 and Ti(S-t-Bu)4 Chem. Mater. method for a highly lattice-mismatched system Thin Solid 9 1847–53 Films 216 72–6 [1176] Huang C, Wu S, Sanchez A M, Peters J J P, Beanland R, [1151] Dobkin D M and Zuraw M K 2013 Principles of Chemical Ross J S, Rivera P, Yao W, Cobden D H and Xu X 2014 Vapor Deposition (Netherlands: Springer) Lateral heterojunctions within monolayer MoSe2–WSe2 [1152] Li H, Wu H, Yuan S and Qian H 2016 Synthesis and semiconductors Nat. Mater. 13 1096–101 characterization of vertically standing MoS2 nanosheets Sci. [1177] Iannaccone G, Bonaccorso F, Colombo L and Fiori G 2018 Rep. 6 21171 Quantum engineering of transistors based on 2D materials [1153] Yu Y, Li C, Liu Y, Su L, Zhang Y and Cao L 2013 Controlled heterostructures Nat. Nanotechnol. 13 183–91 scalable synthesis of uniform, high-quality monolayer and [1178] Banerjee S K, Register L F, Tutuc E, Reddy D and few-layer MoS2 films Sci. Rep. 3 1866 MacDonald A H 2009 Bilayer pseudospin field-effect [1154] Zhu D, Shu H, Jiang F, Lv D, Asokan V, Omar O, Yuan J, transistor (BiSFET): a proposed new logic device IEEE Zhang Z and Jin C 2017 Capture the growth kinetics of CVD Electron Device Lett. 30 158–60 growth of two-dimensional MoS2 NPJ 2D Mater. Appl. 1 8 [1179] Burg G W et al 2017 Coherent interlayer tunneling and [1155] Xuan W, Yong Ping Z and Zhi Qian C 2016 Effect of MoO3 negative differential resistance with high current density in constituents on the growth of MoS2 nanosheets by chemical double bilayer graphene-WSe2 heterostructures Nano Lett. vapor deposition Mater. Res. Express 3 065014 17 3919–25 [1156] Robinson V N E and Robins J L 1974 Nucleation kinetics of [1180] Burg G W, Prasad N, Kim K, Taniguchi T, Watanabe K, gold deposited onto UHV cleaved surfaces of NaCl and KBr MacDonald A H, Register L F and Tutuc E 2018 Strongly Thin Solid Films 20 155–75 enhanced tunneling at total charge neutrality in double- [1157] Pyeon J J, Kim S H, Jeong D S, Baek S-H, Kang C-Y, Kim J-S bilayer graphene-WSe2 heterostructures Phys. Rev. Lett. and Kim S K 2016 Wafer-scale growth of MoS2 thin films by 120 1777702 atomic layer deposition Nanoscale 8 10792–8 [1181] Kim K, Prasad N, Movva H C P, Burg G W, Wang Y M, [1158] Marx M et al 2017 Large-area MoS2 deposition via MOVPE Larentis S, Taniguchi T, Watanabe K, Register L F and J. Cryst. Growth 464 100–4 Tutuc E 2018 Spin-conserving resonant tunneling in twist- [1159] Jeong-Gyu S et al 2017 Catalytic chemical vapor deposition controlled WSe2-hBN-WSe2 heterostructures Nano Lett. of large-area uniform two-dimensional molybdenum 18 5967–73 disulfide using sodium chloride Nanotechnology 28 465103 [1182] Movva H C P, Kang S W, Rai A, Kim K, Fallahazad B, [1160] Zhan W et al 2017 NaCl-assisted one-step growth of MoS2– Taniguchi T, Watanabe K, Tutuc E and Banerjee S K 2016 WS2 in-plane heterostructures Nanotechnology 28 325602 Room temperature gate-tunable negative differential [1161] Reale F et al 2017 High-mobility and high-optical quality resistance in MoS2/hBN/WSe2 heterostructures Proc. atomically thin WS2 Sci. Rep. 7 14911 74th Annual Device Research Conf. (DRC) (http://doi. [1162] Li S et al 2018 Vapor–liquid–solid growth of monolayer org/10.1109/DRC.2016.7548486) MoS2 nanoribbons Nat. Mater. 17 535  −  42 [1183] Britnell L et al 2012 Field-effect tunneling transistor based [1163] Chen J et al 2017 Chemical vapor deposition of large-size on vertical graphene heterostructures Science 335 947–50 monolayer MoSe2 crystals on molten glass J. Am. Chem. Soc. [1184] Dean C R et al 2010 Boron nitride substrates for high-quality 139 1073–6 graphene electronics Nat. Nanotechnol. 5 722–6 [1164] Zhang L et al 2017 Damage-free and rapid transfer of CVD- [1185] Georgiou T et al 2013 Vertical field-effect transistor based on grown two-dimensional transition metal dichalcogenides graphene-WS2 heterostructures for flexible and transparent by dissolving sacrificial water-soluble layers Nanoscale electronics Nat. Nanotechnol. 8 100–3 9 19124–30 [1186] Endler I, Leonhardt A, König U, van den Berg H, Pitschke W [1165] Boandoh S et al 2017 A novel and facile route to synthesize and Sottke V 1999 Chemical vapour deposition of MoS2 atomic-layered MoS2 film for large-area electronics Small coatings using the precursors MoCl5 and H2S Surf. Coat. 13 1701306 Technol. 120–1 482–8 [1166] Yang P et al 2018 Batch production of 6-inch uniform [1187] Lee W Y, Besmann T M and Stott M W 1994 Preparation of monolayer molybdenum disulfide catalyzed by sodium in MoS2 thin films by chemical vapor deposition J. Mater. Res. glass Nat. Commun. 9 4905 9 1474–83 [1167] Chen J et al 2017 Homoepitaxial growth of large-scale highly [1188] Liu K-K et al 2012 Growth of large-area and highly organized transition metal dichalcogenide patterns Adv. crystalline MoS2 thin layers on insulating substrates Nano Mater. 30 1704674 Lett. 12 1538–44 [1168] Eichfeld S M et al 2015 Highly scalable, atomically thin WSe2 [1189] Zhan Y, Liu Z, Najmaei S, Ajayan P M and Lou J 2012 Large- grown via metal–organic chemical vapor deposition ACS area vapor-phase growth and characterization of MoS2 Nano 9 2080–7 atomic layers on a SiO2 substrate Small 8 966–71 [1169] Xu W, Ye Z, Zhou T, Zhao B, Zhu L and Huang J 2004 Low- [1190] Suzuki S, Molto Pallares R, Orofeo C M and Hibino H 2013 pressure MOCVD growth of p-type ZnO thin films by using Boron nitride growth on metal foil using solid sources J. Vac. NO as the dopant source J. Cryst. Growth 265 133–6 Sci. Technol. B 31 041804 [1170] Suzuki M, Nishio J, Onomura M and Hongo C 1998 Doping [1191] Suzuki S, Pallares R M and Hibino H 2012 Growth of characteristics and electrical properties of Mg-doped AlGaN atomically thin hexagonal boron nitride films by diffusion grown by atmospheric-pressure MOCVD J. Cryst. Growth through a metal film and precipitation J. Phys. D: Appl. Phys. 189–90 511–5 45 385304 [1171] Tan L K, Liu B, Teng J H, Guo S, Low H Y and Loh K P [1192] Rong Y, Fan Y, Leen Koh A, Robertson A W, He K, Wang S, 2014 Atomic layer deposition of a MoS2 film Nanoscale Tan H, Sinclair R and Warner J H 2014 Controlling sulphur 6 10584–8 precursor addition for large single crystal domains of WS2 [1172] Jin Z, Shin S, Kwon D H, Han S-J and Min Y-S 2014 Novel Nanoscale 6 12096–103 chemical route for atomic layer deposition of MoS2 thin film [1193] Fu Q, Wang W, Yang L, Huang J, Zhang J and Xiang B 2015 on SiO2/Si substrate Nanoscale 6 14453–8 Controllable synthesis of high quality monolayer WS2 on [1173] Cadot S et al 2017 A novel 2-step ALD route to ultra-thin a SiO2/Si substrate by chemical vapor deposition RSC Adv. MoS2 films on SiO2 through a surface organometallic 5 15795–9 intermediate Nanoscale 9 538–46 [1194] Giusca C E et al 2016 Excitonic effects in tungsten disulfide [1174] Lewis D J et al 2015 Thin films of molybdenum disulfide monolayers on two-layer graphene ACS Nano 10 7840–6 doped with chromium by aerosol-assisted chemical vapor [1195] Bianco G V, Losurdo M, Giangregorio M M, Sacchetti A, deposition (AACVD) Chem. Mater. 27 1367–74 Prete P, Lovergine N, Capezzuto P and Bruno G 2015 Direct [1175] Cheon J, Gozum J E and Girolami G S 1997 Chemical vapor epitaxial CVD synthesis of tungsten disulfide on epitaxial deposition of MoS2 and TiS2 films from the metal–organic and CVD graphene RSC Adv. 5 98700–8 268 2D Mater. 7 (2020) 022001 C Backes et al [1196] Okada M, Sawazaki T, Watanabe K, Taniguch T, Hibino H, [1218] Duan X et al 2014 Lateral epitaxial growth of two- Shinohara H and Kitaura R 2014 Direct chemical vapor dimensional layered semiconductor heterojunctions Nat. deposition growth of WS2 atomic layers on hexagonal boron Nanotechnol. 9 1024–30 nitride ACS Nano 8 8273–7 [1219] Li M Y et al 2015 Epitaxial growth of a monolayer WSe2– [1197] Chen K, Wan X, Wen J X, Xie W G, Kang Z W, Zeng X L, MoS2 lateral p-n junction with an atomically sharp interface Chen H J and Xu J B 2015 Electronic properties of MoS2– Science 349 524–8 WS2 heterostructures synthesized with two-step lateral [1220] Zhang Z, Chen P, Duan X, Zang K, Luo J and Duan X epitaxial strategy ACS Nano 9 9868–76 2017 Robust epitaxial growth of two-dimensional [1198] Gong Y et al 2014 Vertical and in-plane heterostructures heterostructures, multiheterostructures, and superlattices from WS2/MoS2 monolayers Nat. Mater. 13 1135–42 Science 357 788–92 [1199] Tongay S et al 2014 Tuning interlayer coupling in large- [1221] Xie S et al 2018 Coherent, atomically thin transition-metal area heterostructures with CVD-grown MoS2 and WS2 dichalcogenide superlattices with engineered strain Science monolayers Nano Lett. 14 3185–90 359 1131–6 [1200] Zhang X Q, Lin C H, Tseng Y W, Huang K H and Lee Y H [1222] Yu H, Kutana A and Yakobson B I 2016 Carrier delocalization 2015 Synthesis of lateral heterostructures of semiconducting in two-dimensional coplanar p–n junctions of graphene and atomic layers Nano Lett. 15 410–5 metal dichalcogenides Nano Lett. 16 5032–6 [1201] Nie Y F, Liang C P, Cha P R, Colombo L G, Wallace R M and [1223] Pulkin A and Yazyev O V 2016 Spin- and valley-polarized Cho K 2017 A kinetic Monte Carlo simulation method transport across line defects in monolayerMoS2 Phys. Rev. B of van der Waals epitaxy for atomistic nucleation-growth 93 041409(R) processes of transition metal dichalcogenides Sci. Rep. 7 2977 [1224] Dumcenco D O, Chen K Y, Wang Y P, Huang Y S and [1202] Yue R Y et al 2017 Nucleation and growth of WSe2: enabling Tiong K K 2010 Raman study of 2H-Mo1−xWxS2 layered large grain transition metal dichalcogenides 2D Mater. mixed crystals J. Alloys Compd. 506 940–3 4 045019 [1225] Woods J M, Jung Y, Xie Y, Liu W, Liu Y, Wang H and [1203] Tang S et al 2015 Silane-catalysed fast growth of large Cha J J 2016 One-step synthesis of MoS2/WS2 layered single-crystalline graphene on hexagonal boron nitride Nat. heterostructures and catalytic activity of defective transition Commun. 6 6499 metal dichalcogenide films ACS Nano 10 2004–9 [1204] Song X et al 2016 Seed-assisted growth of single-crystalline [1226] Xue Y et al 2015 Scalable production of a few-layer MoS2/ patterned graphene domains on hexagonal boron nitride by WS2 vertical heterojunction array and its application for chemical vapor deposition Nano Lett. 16 6109–16 photodetectors ACS Nano 10 573–80 [1205] Tang S, Ding G, Xie X, Chen J, Wang C, Ding X, Huang F, [1227] Georgakilas V, Otyepka M, Bourlinos A B, Chandra V, Lu W and Jiang M 2012 Nucleation and growth of single Kim N, Kemp K C, Hobza P, Zboril R and Kim K S 2012 crystal graphene on hexagonal boron nitride Carbon Functionalization of graphene: covalent and non-covalent 50 329–31 approaches, derivatives and applications Chem. Rev. [1206] Saddow A A 2004 Advances in Silicon Carbide Processing and 112 6156–214 Applications (Boston, MA: Artech House Publisher) [1228] Yang G-H, Bao D-D, Liu H, Zhang D-Q, Wang N and Li H-T [1207] Gao Y et al 2015 Large-area synthesis of high-quality and 2017 Functionalization of graphene and applications of the uniform monolayer WS2 on reusable Au foils Nat. Commun. derivatives J. Inorg. Organomet. Polym. Mater. 27 1129–41 6 8569 [1229] Bertolazzi S, Gobbi M, Zhao Y, Backes C and Samorì P 2018 [1208] Iqbal M W, Iqbal M Z, Khan M F, Shehzad M A, Seo Y, Molecular chemistry approaches for tuning the properties Park J H, Hwang C and Eom J 2015 High-mobility and air- of two-dimensional transition metal dichalcogenides Chem. stable single-layer WS2 field-effect transistors sandwiched Soc. Rev. 47 6845–88 between chemical vapor deposition-grown hexagonal BN [1230] Hirsch A, Englert J M and Hauke F 2013 Wet chemical films Sci. Rep. 5 10699 functionalization of graphene Acc. Chem. Res. 46 87–96 [1209] Rossi A, Büch H, Di Rienzo C, Miseikis V, Convertino D, Al- [1231] Huang P, Jing L, Zhu H and Gao X 2013 Diazonium Temimy A, Voliani V, Gemmi M, Piazza V and Coletti C 2016 functionalized graphene: microstructure, electric, and Scalable synthesis of WS2 on graphene and h-BN: an all-2D magnetic properties Acc. Chem. Res. 46 43–52 platform for light-matter transduction 2D Mater. 3 031013 [1232] Bekyarova E, Sarkar S, Wang F, Itkis M E, Kalinina I, Tian X [1210] Tan H J, Fan Y, Zhou Y Q, Chen Q, Xu W S and Warner J H and Haddon R C 2013 Effect of covalent chemistry on the 2016 Ultrathin 2D photodetectors utilizing chemical vapor electronic structure and properties of carbon nanotubes and deposition grown WS2 with graphene electrodes ACS Nano graphene Acc. Chem. Res. 46 65–76 10 7866–73 [1233] Milner E M, Skipper N T, Howard C A, Shaffer M S P, [1211] Huang J, Yang L, Liu D, Chen J J, Fu Q, Xiong Y J, Lin F and Buckley D J, Rahnejat K A, Cullen P L, Heenan R K, Xiang B 2015 Large-area synthesis of monolayer WSe2 on Lindner P and Schweins R 2012 Structure and morphology a SiO2/Si substrate and its device applications Nanoscale of charged graphene platelets in solution by small-angle 7 4193–8 neutron scattering J. Am. Chem. Soc. 134 8302–5 [1212] Liu B L, Fathi M, Chen L, Abbas A, Ma Y Q and Zhou C W [1234] Hodge S A, Buckley D J, Yau H C, Skipper N T, Howard C A 2015 Chemical vapor deposition growth of monolayer WSe2 and Shaffer M S P 2017 Chemical routes to discharging with tunable device characteristics and growth mechanism graphenides Nanoscale 9 3150–8 study ACS Nano 9 6119–27 [1235] Schäfer R A, Dasler D, Mundloch U, Hauke F and Hirsch A [1213] Nie Y F, Liang C P, Zhang K H, Zhao R, Eichfeld S M, Cha P R, 2016 Basic insights into tunable graphene hydrogenation Colombo L, Robinson J A, Wallace R M and Cho K 2016 First J. Am. Chem. Soc. 138 1647–52 principles kinetic Monte Carlo study on the growth patterns [1236] Strauss V, Schäfer R A, Hauke F, Hirsch A and Guldi D M of WSe2 monolayer 2D Mater. 3 025029 2015 Polyhydrogenated graphene: excited state dynamics in [1214] Nie Y F et al 2018 Dislocation driven spiral and non-spiral photo- and electroactive two-dimensional domains J. Am. growth in layered chalcogenides Nanoscale 10 15023–34 Chem. Soc. 137 13079–86 [1215] Lin Y C, Zhang W J, Huang J K, Liu K K, Lee Y H, Liang C T, [1237] Schäfer R A, Weber K, Kolesnik-Gray M, Hauke F, Krstic V, Chu C W and Li L J 2012 Wafer-scale MoS2 thin layers Meyer B and Hirsch A 2016 Substrate-modulated reductive prepared by MoO3 sulfurization Nanoscale 4 6637–41 graphene functionalization Angew. Chem., Int. Ed. [1216] Wang S S, Wang X C and Warner J H 2015 All chemical vapor 55 14858–62 deposition growth of MoS2:h-BN vertical van der Waals [1238] Knirsch K C, Englert J M, Dotzer C, Hauke F and Hirsch A heterostructures ACS Nano 9 5246–54 2013 Screening of the chemical reactivity of three different [1217] Forti S et al 2017 Electronic properties of single-layer graphite sources using the formation of reductively tungsten disulfide on epitaxial graphene on silicon carbide alkylated graphene as a model reaction Chem. Commun. Nanoscale 9 16412–9 49 10811–3 269 2D Mater. 7 (2020) 022001 C Backes et al [1239] Morishita T, Clancy A J and Shaffer M S P 2014 Optimised [1261] Chattopadhyay J, Chakraborty S, Mukherjee A, Wang R, exfoliation conditions enhance isolation and solubility of Engel P S and Billups W E 2007 SET mechanism in the grafted graphenes from graphite intercalation compounds functionalization of single-walled carbon nanotubes J. Phys. J. Mater. Chem. C 2 15022–8 Chem. C 111 17928–32 [1240] Hof F, Schäfer R A, Weiss C, Hauke F and Hirsch A 2014 [1262] Hioe J and Zipse H 2010 Radical stability and its role in Novel λ3-iodane-based functionalization of synthetic synthesis and catalysis Org. Biomol. Chem. 8 3609–17 carbon allotropes (SCAs)—common concepts and [1263] Xia Z et al 2016 Electrochemical functionalization of quantification of the degree of addition Chem. Eur. J. graphene at the nanoscale with self-assembling diazonium 20 16644–51 salts ACS Nano 10 7125–34 [1241] Englert J M, Vecera P, Knirsch K C, Schäfer R A, Hauke F [1264] Shih C-J, Wang Q H, Jin Z, Paulus G L C, Blankschtein D, and Hirsch A 2013 Scanning-Raman-microscopy for the Jarillo-Herrero P and Strano M S 2013 Disorder imposed statistical analysis of covalently functionalized graphene limits of mono- and bilayer graphene electronic ACS Nano 7 5472–82 modification using covalent chemistry Nano Lett. 13 809–17 [1242] Hof F, Bosch S, Eigler S, Hauke F and Hirsch A 2013 New [1265] Allongue P, Delamar M, Desbat B, Fagebaume O, Hitmi R, basic insight into reductive functionalization sequences of Pinson J and Savéant J-M 1997 Covalent modification single walled carbon nanotubes (SWCNTs) J. Am. Chem. Soc. of carbon surfaces by aryl radicals generated from the 135 18385–95 electrochemical reduction of diazonium salts J. Am. Chem. [1243] Vecera P, Chacón-Torres J C, Pichler T, Reich S, Soni H R, Soc. 119 201–7 Görling A, Edelthalhammer K, Peterlik H, Hauke F [1266] Gu S-Y, Hsieh C-T, Yuan J-Y, Hsueh J-H and Gandomi Y A and Hirsch A 2017 Precise determination of graphene 2018 Amino-functionalization of graphene nanosheets functionalisation by in situ Raman spectroscopy Nat. by electrochemical exfoliation technique Diamond Relat. Commun. 8 15192 Mater. 87 99–106 [1244] Knirsch K C, Schäfer R A, Hauke F and Hirsch A 2016 Mono- [1267] Jokar E, Shahrokhian S and Zad A I 2014 Electrochemical and ditopic bisfunctionalization of graphene Angew. Chem., functionalization of graphene nanosheets with catechol Int. Ed. 55 5861–4 derivatives as an effective method for preparation of highly [1245] Quintana M, Vazquez E and Prato M 2013 Organic performance supercapacitors Electrochim. Acta 147 136–42 functionalization of graphene in dispersions Acc. Chem. Res. [1268] Yadav A, Rodrigo Iost M, Neubert T J, Baylan S, Schmid T 46 138–48 and Balasubramanian K 2019 Selective electrochemical [1246] Paulus G L C, Wang Q H and Strano M S 2013 Covalent functionalization of the graphene edge Chem. Sci. 10 936 electron transfer chemistry of graphene with diazonium [1269] Sarkar S, Bekyarova E and Haddon R C 2012 Reversible salts Acc. Chem. Res. 46 160–70 grafting of α-naphthylmethyl radicals to epitaxial graphene [1247] Park J and Yan M 2013 Covalent functionalization of Angew. Chem. 124 4985–8 graphene with reactive intermediates Acc. Chem. Res. [1270] Dreyer D R, Todd A D and Bielawski C W 2014 Harnessing 46 181–9 the chemistry of graphene oxide Chem. Soc. Rev. 43 5288 [1248] Kozhemyakina N V, Eigler S, Dinnebier R E, Inayat A, [1271] Wu H, Shi H, Wang Y, Jia X, Tang C, Zhang J and Yang S 2014 Schwieger W and Hirsch A 2013 Effect of the structure Hyaluronic acid conjugated graphene oxide for targeted and morphology of natural, synthetic and post-processed drug delivery Carbon 69 379–89 graphites on their dispersibility and electronic properties, [1272] Eigler S, Hu Y, Ishii Y and Hirsch A 2013 Controlled fullerenes, nanotubes and carbon Nanostructures 21 804–23 functionalization of graphene oxide with sodium azide [1249] Birch A J 1944 117. Reduction by dissolving metals. Part I Nanoscale 5 12136 J. Chem. Soc. 1944 430  −  6 [1273] Lee J U, Lee W, Yi J W, Yoon S S, Lee S B, Jung B M, Kim B S [1250] Birch A J 1955 212. Reduction by dissolving metals. Part II and Byun J H 2013 Preparation of highly stacked graphene J. Chem. Soc. 1945 809  −  13 papers via site-selective functionalization of graphene oxide [1251] Birch A J 1946 119. Reduction by dissolving metals. Part III J. Mater. Chem. C 1 12893 J. Chem. Soc. 1946 593–7 [1274] Mei Q, Zhang K, Guan G, Liu B, Wang S and Zhang Z 2010 [1252] Birch A J 1947 25. Reduction by dissolving metals. Part IV Highly efficient photoluminescent graphene oxide with J. Chem. Soc. 1947 102–5 tunable surface properties Chem. Commun. 46 7319 [1253] Englert J M, Dotzer C, Yang G, Schmid M, Papp C, [1275] Neumann C et al 2015 Raman spectroscopy as probe Gottfried J M, Steinrück H-P, Spiecker E, Hauke F and of nanometre-scale strain variations in graphene Nat. Hirsch A 2011 Covalent bulk functionalization of graphene Commun. 6 8429 Nat. Chem. 3 279–86 [1276] Thomas H R, Marsden A J, Walker M, Wilson N R and [1254] Dresselhaus M S and Dresselhaus G 1981 Intercalation Rourke J P 2014 Sulfur-functionalized graphene oxide by compounds of graphite Adv. Phys. 30 139–326 epoxide ring-opening Angew. Chem., Int. Ed. 53 7613–8 [1255] Knirsch K C, Hof F, Lloret V, Mundloch U, Hauke F and [1277] Wang S et al 2008 Band-like transport in surface- Hirsch A 2016 Topology-driven reductive silylation of functionalized highly solution-processable graphene synthetic carbon allotropes J. Am. Chem. Soc. 138 15642–7 nanosheets Adv. Mater. 20 3440–6 [1256] Clancy A J, Melbourne J and Shaffer M S P 2015 A one-step [1278] Yang H, Shan C, Li F, Han D, Zhang Q andNiu L 2009 route to solubilised, purified or functionalised single-walled Covalent functionalization of polydisperse chemically- carbon nanotubes J. Mater. Chem. C 3 16708–15 converted graphene sheets with amine-terminated ionic [1257] Dasler D, Schäfer R A, Minameyer M B, Hitzenberger J F, liquid Chem. Commun. 2009 3880 Hauke F, Drewello T and Hirsch A 2017 Direct covalent [1279] Vacchi I A, Spinato C, Raya J, Bianco A and Ménard- coupling of porphyrins to graphene J. Am. Chem. Soc. Moyon C 2016 Chemical reactivity of graphene oxide 139 11760–5 towards amines elucidated by solid-state NMR Nanoscale [1258] Chan C K, Beechem T E, Ohta T, Brumbach M T, 8 13714–21 Wheeler D R and Stevenson K J 2013 Electrochemically [1280] Stankovich S, Dikin D A, Piner R D, Kohlhaas K A, driven covalent functionalization of graphene from Kleinhammes A, Jia Y, Wu Y, Nguyen S T and Ruoff R S fluorinated aryl iodonium salts J. Phys. Chem. C 2007 Synthesis of graphene-based nanosheets via chemical 117 12038–44 reduction of exfoliated graphite oxide Carbon 45 1558–65 [1259] Hof F, Hauke F and Hirsch A 2014 Brominated single walled [1281] Jiang J-X et al 2008 Conjugated microporous carbon nanotubes as versatile precursors for covalent poly(aryleneethynylene) networks Angew. Chem. Int., Ed. sidewall functionalization Chem. Commun. 50 6582–4 47 1167–7 [1260] Dyke C A and Tour J M 2013 Unbundled and highly [1282] Peigney A, Laurent C, Flahaut E, Bacsa R R and Rousset A functionalized carbon nanotubes from aqueous reactions 2001 Specific surface area of carbon nanotubes and bundles Nano Lett. 3 1215–8 of carbon nanotubes Carbon 39 507–14 270 2D Mater. 7 (2020) 022001 C Backes et al [1283] Stoller M D, Park S J, Zhu Y W, An J H and Ruoff R S 2008 [1304] Vaisman L, Wagner H D and Marom G 2006 The role of Graphene-based ultracapacitors Nano Lett. 8 3498–502 surfactants in dispersion of carbon nanotubes Adv. Colloid [1284] Zhu Y W, Murali S, Cai W W, Li X S, Suk J W, Potts J R and Interface Sci. 128–30 37–46 Ruoff R S 2010 Graphene and graphene oxide: synthesis, [1305] Verdejo R, Bernal M M, Romasanta L J and Lopez- properties, and applications Adv. Mater. 22 3906–24 Manchado M A 2011 Graphene filled polymer [1285] Cao X H, Yin Z Y and Zhang H 2014 Three-dimensional nanocomposites J. Mater. Chem. 21 3301–10 graphene materials: preparation, structures and application [1306] Li Z, Wang R, Young R J, Deng L, Yang F, Hao L, Jiao W in supercapacitors Energy Environ. Sci. 7 1850–65 and Liu W 2013 Control of the functionality of graphene [1286] Dai L M, Chang D W, Baek J B and Lu W 2012 Carbon oxide for its application in epoxy nanocomposites Polymer nanomaterials for advanced energy conversion and storage 54 6437–46 Small 8 1130–66 [1307] Wajid A S, Das S, Irin F, Ahmed H S T, Shelburne J L, [1287] Sudeep P M et al 2013 Covalently interconnected Parviz D, Fullerton R J, Jankowski A F, Hedden R C and three-dimensional graphene oxide solids ACS Nano Green M J 2012 Polymer-stabilized graphene dispersions 7 7034–40 at high concentrations in organic solvents for composite [1288] Burress J W, Gadipelli S, Ford J, Simmons J M, Zhou W and production Carbon 50 526–34 Yildirim T 2010 Graphene oxide framework materials: [1308] Wan Y-J, Gong L-X, Tang L-C, Wu L-B and Jiang J-X 2014 theoretical predictions and experimental results Angew. Mechanical properties of epoxy composites filled with Chem., Int. Ed. 49 8902–4 silane-functionalized graphene oxide Composites A 64 79–89 [1289] Tsoufis T, Tuci G, Caporali S, Gournis D and Giambastiani G [1309] Hou S, Su S, Kasner M L, Shah P, Patel K and Madarang C J 2013 p-Xylylenediamine intercalation of graphene oxide for 2010 Formation of highly stable dispersions of silane- the production of stitched nanostructures with a tailored functionalized reduced graphene oxide Chem. Phys. Lett. interlayer spacing Carbon 59 100–8 501 68–74 [1290] Zhang X Y, Ciesielski A, Richard F, Chen P K, Prasetyanto E A, [1310] Dhakal K P, Duong D L, Lee J, Nam H, Kim M, Kan M, De Cola L and Samorì P 2016 Modular graphene-based Lee Y H and Kim J, Confocal absorption spectral imaging 3D covalent networks: functional architectures for energy of MoS2: optical transitions depending on the atomic applications Small 12 1044–52 thickness of intrinsic and chemically doped MoS2 Nanoscale [1291] Zhuang X, Gehrig D, Forler N, Liang H, Wagner M, 6 13028–35 Hansen M R, Laquai F, Zhang F and Feng X 2015 Conjugated [1311] Yuan B, Sheng H, Mu X, Song L, Tai Q, Shi Y, Liew K M and microporous polymers with dimensionality-controlled Hu Y 2015 Enhanced flame retardancy of polypropylene by heterostructures for green energy devices Adv. Mater. melamine-modified graphene oxide J. Mater. Sci. 50 5389– 27 3789–96 401 [1292] Zhuang X, Zhang F, Wu D, Forler N, Liang H, Wagner M, [1312] Xie M, Lei H, Zhang Y, Xu Y, Shen S, Ge Y, Li H and Xie J Gehrig D, Hansen M R, Laquai F and Feng X 2013 Two- 2016 Non-covalent modification of graphene oxide dimensional sandwich-type, graphene-based conjugated nanocomposites with chitosan/dextran and its application microporous polymers Angew. Chem., Int. Ed. 52 9668–72 in drug delivery RSC Adv. 6 9328–37 [1293] Jasim D A, Ménard-Moyon C, Bégin D, Bianco A and [1313] Mao Y, Wen S, Chen Y, Zhang F, Panine P, Chan T W, Zhang L, Kostarelos K 2015 Tissue distribution and urinary excretion Liang Y and Liu L 2013 High performance graphene oxide of intravenously administered chemically functionalized based rubber composites Sci. Rep. 3 2508 graphene oxide sheets Chem. Sci. 6 3952–64 [1314] Vladár A E Strategies For Scanning Electron Microscopy [1294] Wan Y-J, Tang L-C, Yan D, Zhao L, Li Y-B, Wu L-B, Jiang J-X Sample Preparation and Characterization of Multiwall and Lai G-Q 2013 Improved dispersion and interface in the Carbon Nanotube Polymer Composites (National Institute of graphene/epoxy composites via a facile surfactant-assisted Standards and Technology) (https://doi.org/10.6028/NIST. process Compos. Sci. Technol. 82 60–8 SP.1200-17) [1295] Bai H, Xu Y, Zhao L, Li C and Shi G 2009 Non-covalent [1315] Rabe J P and Buchholz S 1991 Commensurability and functionalization of graphene sheets by sulfonated mobility in 2-dimensional molecular-patterns on graphite polyaniline Chem. Commun. 1667 (https://doi.org/10.1039/ Science 253 424–7 b821805f) [1316] Grande M, Bianco G V, Vincenti M A, de Ceglia D, [1296] Atif R and Inam F 2016 Reasons and remedies for the Capezzuto P, Scalora M, D’Orazio A and Bruno G 2015 agglomeration of multilayered graphene and carbon Optically transparent microwave polarizer based on quasi- nanotubes in polymers Beilstein J. Nanotechnol. 7 1174–96 metallic graphene Sci. Rep. 5 17083 [1297] Bhattacharya M 2016 Polymer nanocomposites—a [1317] Sachs B, Wehling T O, Lichtenstein A I and Katsnelson M I comparison between carbon nanotubes, graphene, and clay 2011 Theory of doping: monovalent adsorbates Physics and as nanofillers Materials 9 262 Applications of Graphene—Theory (Rijeka: InTech) [1298] Wei J, Vo T and Inam F 2015 Epoxy/graphene [1318] Karlický F, Kumara Ramanatha Datta K, Otyepka M and nanocomposites—processing and properties: a review RSC Zboril R 2013 Halogenated graphenes: rapidly growing Adv. 5 73510–24 family of graphene derivatives ACS Nano 7 6434–64 [1299] Salavagione H J, Marti´nez G and Ellis G 2011 Graphene- [1319] Poh H L, Šimek P, Sofer Z and Pumera M 2013 Halogenation based polymer nanocomposites Physics and Applications of graphene with chlorine, bromine, or iodine by exfoliation of Graphene—Experiments (Rijeka: Intech) ch 9 DOI: in a halogen atmosphere Chem. Eur. J. 19 2655–62 10.5772/14665 [1320] Mansour A E, Dey S, Amassian A and Tanielian M H 2015 [1300] Ramanathan T et al 2008 Functionalized graphene sheets for Bromination of graphene: a new route to making high polymer nanocomposites Nat. Nanotechnol. 3 327 performance transparent conducting electrodes with low [1301] Yang H, Li F, Shan C, Han D, Zhang Q, Niu L and Ivaska A optical losses ACS Appl. Mater. Interfaces 7 17692–9 2009 Covalent functionalization of chemically converted [1321] Bruna M, Ott A K, Ijäs M, Yoon D, Sassi U and Ferrari A C graphene sheets via silane and its reinforcement J. Mater. 2014 Doping dependence of the raman spectrum of defected Chem. 19 4632–8 graphene ACS Nano 8 7432–41 [1302] Alam A, Wan C and McNally T 2017 Surface amination [1322] Cançado L G et al 2011 Quantifying defects in graphene via of carbon nanoparticles for modification of epoxy resins: raman spectroscopy at different excitation energies Nano plasma-treatment versus wet-chemistry approach Eur. Lett. 11 3190–6 Polym. J. 87 422–48 [1323] Berner N C, Winters S, Backes C, Yim C, Dumbgen K C, [1303] Ji X, Xu Y, Zhang W, Cui L and Liu J 2016 Kaminska I, Mackowski S, Cafolla A A, Hirsch A and Review of functionalization, structure and properties Duesberg G S 2015 Understanding and optimising the of graphene/polymer composite fibers Composites A packing density of perylene bisimide layers on CVD-grown 87 29–45 graphene Nanoscale 7 16337–42 271 2D Mater. 7 (2020) 022001 C Backes et al [1324] Tian X Q, Xu J B and Wang X M 2010 Self-assembly of of epitaxial graphene chemically modified with aryl moieties PTCDA ultrathin films on graphene: structural phase J. Am. Chem. Soc. 132 15399–403 transition and charge transfer saturation J. Phys. Chem. C [1344] Niyogi S et al 2010 Spectroscopy of covalently functionalized 114 20917–24 graphene Nano Lett. 10 4061–6 [1325] Huang H, Chen S, Gao X, Chen W and Wee A T S 2009 [1345] Gass M H, Bangert U, Bleloch A L, Wang P, Nair R R and Structural and electronic properties of PTCDA thin films on Geim A K 2008 Free-standing graphene at atomic resolution epitaxial graphene ACS Nano 3 3431–6 Nat. Nanotechnol. 3 676–81 [1326] Wang Q H and Hersam M C 2009 Room-temperature [1346] Meyer J C, Kisielowski C, Erni R, Rossell M D, Crommie M F molecular-resolution characterization of self-assembled and Zettl A 2008 Direct imaging of lattice atoms and organic monolayers on epitaxial graphene Nat. Chem. 1 206 topological defects in graphene membranes Nano Lett. [1327] Winters S, Berner N C, Mishra R, Dumbgen K C, Backes C, 8 3582–6 Hegner M, Hirsch A and Duesberg G S 2015 On-surface [1347] Rodrigo L, Pou P and Pérez R 2016 Graphene derivatisation of aromatic molecules on graphene: the monovacancies: electronic and mechanical properties from importance of packing density Chem. Commun. 51 16778– large scale ab initio simulations Carbon 103 200–8 81 [1348] Ugeda M M, Fernández-Torre D, Brihuega I, Pou P, [1328] Woszczyna M, Winter A, Grothe M, Willunat A, Wundrack S, Martínez-Galera A J, Pérez R and Gómez-Rodríguez J M Stosch R, Weimann T, Ahlers F and Turchanin A 2014 2011 Point defects on graphene on metals Phys. Rev. Lett. All-carbon vertical van der Waals heterostructures: non- 107 116803 destructive functionalization of graphene for electronic [1349] Kelly K F, Sarkar D, Hale G D, Oldenburg S J and Halas N J applications Adv. Mater. 26 4831–7 1996 Threefold electron scattering on graphite observed [1329] Chua C K and Pumera M 2013 Covalent chemistry on with C60-adsorbed STM tips Science 273 1371–3 graphene Chem. Soc. Rev. 42 3222–33 [1350] Lehtinen O, Kotakoski J, Krasheninnikov A V, Tolvanen A [1330] Sun Z, Kohama S-I, Zhang Z, Lomeda J R and Tour J M 2010 and Nordlund K, Keinonen J 2010 Effects of ion Soluble graphene through edge-selective functionalization bombardment on a two-dimensional target: atomistic Nano Res. 3 117–25 simulations of graphene irradiation Phys. Rev. B 81 153401 [1331] Bellunato A, Arjmandi Tash H, Cesa Y and Schneider G F [1351] Li Q, Mahmood N, Zhu J, Hou Y and Sun S 2014 Graphene 2016 Chemistry at the edge of graphene ChemPhysChem and its composites with nanoparticles for electrochemical 17 785–801 energy applications Nano Today 9 668–83 [1332] Koehler F M, Jacobsen A, Ensslin K, Stampfer C and [1352] Choi H C, Shim M, Bangsaruntip S and Dai H 2002 Stark W J 2010 Selective chemical modification of graphene Spontaneous reduction of metal ions on the sidewalls of surfaces: distinction between single- and bilayer graphene carbon nanotubes J. Am. Chem. Soc. 124 9058–9 Small 6 1125–30 [1353] Fogden S, Howard C A, Heenan R K, Skipper N T and [1333] Lomeda J R, Doyle C D, Kosynkin D V, Hwang W-F and Shaffer M S P 2012 Scalable method for the reductive Tour J M 2008 Diazonium functionalization of surfactant- dissolution, purification, and separation of single-walled wrapped chemically converted graphene sheets J. Am. Chem. carbon nanotubes ACS Nano 6 54–62 Soc. 130 16201–6 [1354] Howard C A, Thompson H, Wasse J C and Skipper N T 2004 [1334] Quintana M, Montellano A, del Rio Castillo A E, Formation of giant solvation shells around fulleride anions Tendeloo G V, Bittencourt C and Prato M 2011 Selective in liquid ammonia J. Am. Chem. Soc. 126 13228–9 organic functionalization of graphene bulk or graphene [1355] Catheline A, Vallés C, Drummond C, Ortolani L, Morandi V, edges Chem. Commun. 47 9330–2 Marcaccio M, Iurlo M, Paolucci F and Pénicaud A 2011 [1335] Wang A et al 2016 Covalent functionalization of reduced Graphene solutions Chem. Commun. 47 5470 graphene oxide with porphyrin by means of diazonium [1356] Pénicaud A, Poulin P, Derré A, Anglaret E and Petit P 2005 chemistry for nonlinear optical performance Sci. Rep. Spontaneous dissolution of a single-wall carbon nanotube 6 23325 salt J. Am. Chem. Soc. 127 8–9 [1336] Niyogi S, Bekyarova E, Hong J, Khizroev S, Berger C, de [1357] Hodge S A, Fogden S, Howard C A, Skipper N T and Heer W and Haddon R C 2011 Covalent chemistry for Shaffer M S P 2013 Electrochemical processing of discrete graphene electronics J. Phys. Chem. Lett. 2 2487–98 single-walled carbon nanotube anions ACS Nano 7 1769–78 [1337] Bernal M M, Di Pierro A, Novara C, Giorgis F, Mortazavi B, [1358] Bayazit M K, Suri A and Coleman K S 2010 Formylation of Saracco G and Fina A 2018 Edge-grafted molecular junctions single-walled carbon nanotubes Carbon 48 3412–9 between graphene nanoplatelets: applied chemistry to [1359] Chen K, Ge C and Li J 1998 Phase formation and enhance heat transfer in nanomaterials Adv. Funct. Mater. thermodynamic analysis of self-propagating high- 28 1706954 temperature synthesis Al–Zr–N system composites J. Mater. [1338] Farmer D B, Golizadeh-Mojarad R, Perebeinos V, Lin Y-M, Res. 13 2610–3 Tulevski G S, Tsang J C and Avouris P 2009 Chemical doping [1360] Graupner R, Abraham J, Wunderlich D, Vencelová A, and electron–hole conduction asymmetry in graphene Lauffer P, Röhrl J, Hundhausen M, Ley L and Hirsch A 2006 devices Nano Lett. 9 388–92 Nucleophilic–alkylation–reoxidation: a functionalization [1339] Bouša D, Jankovský O, Sedmidubský D, Luxa J, Šturala J, sequence for single-wall carbon nanotubes J. Am. Chem. Soc. Pumera M and Sofer Z 2015 Mesomeric effects of graphene 128 6683–9 modified with diazonium salts: substituent type and [1361] Mukherjee A, Combs R, Chattopadhyay J, Abmayr D W, position influence its properties Chem. Eur. J. 21 17728–38 Engel P S and Billups W E 2008 Attachment of nitrogen and [1340] Bueno R A et al 2017 Highly selective covalent organic oxygen centered radicals to single-walled carbon nanotube functionalization of epitaxial graphene Nat. Commun. salts Chem. Mater. 20 7339–43 8 15306 [1362] Wunderlich D, Hauke F and Hirsch A 2008 Preferred [1341] Castelaín M, Martínez G, Marco C, Ellis G and functionalization of metallic and small-diameter single- Salavagione H J 2013 Effect of click-chemistry approaches walled carbon nanotubes by nucleophilic addition of for graphene modification on the electrical, thermal, organolithium and-magnesium compounds followed by and mechanical properties of polyethylene/graphene reoxidation Chem. Eur. J. 14 1607–14 nanocomposites Macromolecules 46 8980–7 [1363] Howard C A, Dean M P M and Withers F 2011 Phonons in [1342] Criado A, Melchionna M, Marchesan S and Prato M 2015 potassium-doped graphene: the effects of electron-phonon The covalent functionalization of graphene on substrates interactions, dimensionality, and adatom ordering Phys. Rev. Angew. Chem., Int. Ed. 54 10734–50 B 84 241404(R) [1343] Hossain M Z, Walsh M A and Hersam M C 2010 Scanning [1364] Bard A J and Faulkner L R 2001 Electrochemical Methods 2nd tunneling microscopy, spectroscopy, and nanolithography edn (New York: Wiley) 272 2D Mater. 7 (2020) 022001 C Backes et al [1365] Wei W, Wang G, Yang S, Feng X and Müllen K 2015 Efficient [1385] Yürüm Y, Taralp A and Veziroglu T N 2009 Storage of coupling of nanoparticles to electrochemically exfoliated hydrogen in nanostructured carbon materials Int. J. Hydrog. graphene J. Am. Chem. Soc. 137 5576–81 Energy 34 3784–98 [1366] Huttel Y 2017 Gas-Phase Synthesis of Nanoparticles (New [1386] Zhang S, Gross A F, Van Atta S L, Lopez M, Liu P, Ahn C C, York: Wiley) Vajo J J and Jensen C M 2009 The synthesis and hydrogen [1367] Frank F, Schulze W, Tesche B, Urban J and Winter B 1985 storage properties of a MgH2 incorporated carbon aerogel Formation of metal clusters and molecules by means of scaffold Nanotechnology 20 204027 the gas aggregation technique and characterisation of size [1387] Liu W, Setijadi E, Crema L, Bartali R, Laidani N, Aguey- distribution Surf. Sci. 156 90–9 Zinsou K F and Speranza G 2018 Carbon nanostructures/ [1368] Llamosa D, Ruano M, Martínez L, Mayoral A, Roman E, Mg hybrid materials for hydrogen storage Diamond Relat. García-Hernández M and Huttel Y 2014 The ultimate step Mater. 82 19–24 towards a tailored engineering of core@shell and core@ [1388] Aguey-Zinsou K-F and Ares-Fernández J-R 2008 Synthesis shell@shell nanoparticles Nanoscale 6 13483–6 of colloidal magnesium: a near room temperature store for [1369] Martínez L, Díaz M, Román E, Ruano M, Llamosa P D and hydrogen Chem. Mater. 20 376–8 Huttel Y 2012 Generation of nanoparticles with adjustable [1389] Lu Z, Schechter A, Moshkovich M and Aurbach D 1999 size and controlled stoichiometry: recent advances Langmuir On the electrochemical behavior of magnesium electrodes 28 11241–9 in polar aprotic electrolyte solutions J. Electroanal. Chem. [1370] Martínez L, Mayoral A, Espiñeira M, Roman E, Palomares F J 466 203–17 and Huttel Y 2017 Core@shell, Au@TiOx nanoparticles by [1390] Setijadi E J, Boyer C and Aguey-Zinsou K-F 2012 gas phase synthesis Nanoscale 9 6463–70 Remarkable hydrogen storage properties for nanocrystalline [1371] Ferrando R, Jellinek J and Johnston R L 2008 Nanoalloys: MgH2 synthesised by the hydrogenolysis of Grignard from theory to applications of alloy clusters and reagents Phys. Chem. Chem. Phys. 14 11386 nanoparticles Chem. Rev. 108 845–910 [1391] Setijadi E J, Boyer C and Aguey-Zinsou K-F 2014 Switching [1372] Ghosh Chaudhuri R and Paria S 2012 Core/shell the thermodynamics of MgH2nanoparticles through nanoparticles: classes, properties, synthesis mechanisms, polystyrene stabilisation and oxidation RSC Adv. 4 39934 characterization, and applications Chem. Rev. 112 2373–433 [1392] Liu W and Aguey-Zinsou K-F 2014 Size effects and hydrogen [1373] Pohl D, Surrey A, Schultz L and Rellinghaus B 2012 The storage properties of Mg nanoparticles synthesised by an impact of oxygen on the morphology of gas-phase prepared electroless reduction method J. Mater. Chem. C 2 9718 Au nanoparticles Appl. Phys. Lett. 101 263105 [1393] Cho E S, Ruminski A M, Aloni S, Liu Y-S, Guo J and Urban J J [1374] Garcia M A, Fernandez Pinel E, de la Venta J, Quesada A, 2016 Graphene oxide/metal nanocrystal multilaminates as Bouzas V, Fernández J F, Romero J J, Martín González M S the atomic limit for safe and selective hydrogen storage Nat. and Costa-Krämer J L 2009 Sources of experimental errors Commun. 7 10804 in the observation of nanoscale magnetism J. Appl. Phys. [1394] Liu G, Wang Y, Jiao L and Yuan H 2014 Understanding the 105 013925 role of few-layer graphene nanosheets in enhancing the [1375] Ruano M, Díaz M, Martínez L, Navarro E, Román E, García- hydrogen sorption kinetics of magnesium hydride ACS Appl. Hernandez M, Espinosa A, Ballesteros C, Fermento R and Mater. Interfaces 6 11038–46 Huttel Y 2013 Matrix and interaction effects on the magnetic [1395] Tachikawa H, Iyama T and Kawabata H 2009 MD simulation properties of Co nanoparticles embedded in gold and of the interaction of magnesium with graphene Thin Solid vanadium Phys. Chem. Chem. Phys. 15 316–29 Films 518 877–9 [1376] Plant S R, Cao L and Palmer R E 2014 Atomic structure [1396] Liu R, Zhao Y and Chu T 2015 Theoretical exploration of control of size-selected gold nanoclusters during formation MgH2 and graphene nano-flakes in cyclohexane: proposing J. Am. Chem. Soc. 136 7559–62 a new perspective toward functional hydrogen storage [1377] Krajewska A, Oberda K, Azpeitia J, Gutierrez A, Pasternak I, material Chem. Commun. 51 2429–32 López M F, Mierczyk Z, Munuera C and Strupinski W 2016 [1397] Tozzini V and Pellegrini V 2013 Prospects for hydrogen Influence of Au doping on electrical properties of CVD storage in graphene Phys. Chem. Chem. Phys. 15 80–9 graphene Carbon 100 625–31 [1398] Barkhordarian G, Klassen T and Bormann R 2006 Kinetic [1378] Azpeitia J, Garcia-Hernandez M, Martinez L, Huttel Y, investigation of the effect of milling time on the hydrogen Lopez Fagundez M F and Munuera C 2019 submitted sorption reaction of magnesium catalyzed with different [1379] Jordá-Beneyto M, Suárez-García F, Lozano-Castelló D, Nb2O5 contents J. Alloys Compd. 407 249–55 Cazorla-Amorós D and Linares-Solano A 2007 Hydrogen [1399] Bogdanovic B, Bohmhammel K, Christ B, Reiser A, storage on chemically activated carbons and carbon Schlichte K, Vehlen R and Wolf U 1999 Thermodynamic nanomaterials at high pressures Carbon 45 293–303 investigation of the magnesium–hydrogen system J. Alloys [1380] Züttel A, Nützenadel C, Sudan P, Mauron P, Emmenegger C, Compd. 282 84–92 Rentsch S, Schlapbach L, Weidenkaff A and Kiyobayashi T [1400] Aguey-Zinsou K-F and Ares-Fernández J-R 2010 Hydrogen 2002 Hydrogen sorption by carbon nanotubes and other in magnesium: new perspectives toward functional stores carbon nanostructures J. Alloys Compd. 330–32 676–82 Energy Environ. Sci. 3 526 [1381] Shen C and Aguey-Zinsou K-F 2017 Nanosized magnesium [1401] Graetz J 2009 New approaches to hydrogen storage Chem. electrochemically deposited on a carbon nanotubes Soc. Rev. 38 73–82 suspension: synthesis and hydrogen storage Frontiers Energy [1402] Kim K C, Dai B, Karl Johnson J and Sholl D S 2009 Assessing Res. 5 27 nanoparticle size effects on metal hydride thermodynamics [1382] Hwang S-J and Chuang Y-S 2016 Enhanced hydrogen using the Wulff construction Nanotechnology 20 204001 storage properties of MgH2 co-catalyzed with zirconium [1403] Makarova M, Okawa Y and Aono M 2012 Selective oxide and single-walled carbon nanotubes J. Alloys Compd. adsorption of thiol molecules at sulfur vacancies on 664 284–90 MoS2(0 0 0 1), followed by vacancy repair via S–C [1383] Liu Y, Zou J, Zeng X, Wu X, Tian H, Ding W, Wang J and dissociation J. Phys. Chem. C 116 22411–6 Walter A 2013 Study on hydrogen storage properties of Mg [1404] Chou S S, De M, Kim J, Byun S, Dykstra C, Yu J, Huang J and nanoparticles confined in carbon aerogels Int. J. Hydrog. Dravid V P 2013 Ligand conjugation of chemically exfoliated Energy 38 5302–8 MoS2 J. Am. Chem. Soc. 135 4584–7 [1384] Popilevsky L, Skripnyuk V M, Beregovsky M, Sezen M, [1405] Zhou L, He B, Yang Y and He Y 2014 Facile approach to Amouyal Y and Rabkin E 2016 Hydrogen storage and surface functionalized MoS2 nanosheets RSC Adv. 4 32570–8 thermal transport properties of pelletized porous Mg-2 [1406] Chhowalla M, Shin H S, Eda G, Li L-J, Loh K P and Zhang H wt.% multiwall carbon nanotubes and Mg-2 wt.% graphite 2013 The chemistry of two-dimensional layered transition composites Int. J. Hydrog. Energy 41 14461–74 metal dichalcogenide nanosheets Nat. Chem. 5 263 273 2D Mater. 7 (2020) 022001 C Backes et al [1407] Fan X, Xu P, Li Y C, Zhou D, Sun Y, Nguyen M A T, [1426] Abergel D, Russell A and Fal’ko V I 2007 Visibility of Terrones M and Mallouk T E 2016 Controlled exfoliation graphene flakes on a dielectric substrate Appl. Phys. Lett. of MoS2 crystals into trilayer nanosheets J. Am. Chem. Soc. 91 063125 138 5143–9 [1427] Castellanos Gómez A 2011 Propiedades electrónicas, [1408] Yuan K, Zhuang X, Fu H, Brunklaus G, Forster M, Chen Y, mecánicas y ópticas de cristales bidimensionales de espesor Feng X and Scherf U 2016 Two-dimensional core-shelled atómico PhD Thesis Universidad Autónoma de Madrid porous hybrids as highly efficient catalysts for the oxygen [1428] Zhang H, Ma Y, Wan Y, Rong X, Xie Z, Wang W and Dai L reduction reaction Angew. Chem., Int. Ed. 55 6858–63 2015 Measuring the refractive index of highly crystalline [1409] Backes C, Berner N C, Chen X, Lafargue P, LaPlace P, monolayer MoS2 with high confidence Sci. Rep. 5 18357 Freeley M, Duesberg G S, Coleman J N and McDonald A R [1429] Ghasemi F, Frisenda R, Dumcenco D, Kis A, Perez de 2015 Functionalization of liquid-exfoliated two- Lara D and Castellanos-Gomez A 2017 High throughput dimensional 2H-MoS2 Angew. Chem., Int. Ed. 54 2638–42 characterization of epitaxially grown single-layer MoS2 [1410] Chen X and McDonald A R 2016 Functionalization of two- Electronics 6 28 dimensional transition-metal dichalcogenides Adv. Mater. [1430] Gant P, Ghasemi F, Maeso D, Munuera C, López-Elvira E, 28 5738–46 Frisenda R, De Lara D P, Rubio-Bollinger G, Garcia- [1411] Hirsch A and Hauke F Post graphene 2D chemistry: Hernandez M and Castellanos-Gomez A 2017 Optical the emerging field of molybdenum disulfide and black contrast and refractive index of natural van der Waals phosphorus functionalization Angew. Chem., Int. Ed. heterostructure nanosheets of franckeite Beilstein J. 57 4338−54 (https://doi.org/10.1002/anie.201708211) Nanotechnol. 8 2357 [1412] Canton-Vitoria R, Sayed-Ahmad-Baraza Y, Pelaez- [1431] Castellanos-Gomez A, Quereda J, van der Meulen H P, Fernandez M, Arenal R, Bittencourt C, Ewels C P and Agraït N and Rubio-Bollinger G 2016 Spatially resolved Tagmatarchis N 2017 Functionalization of MoS2 with optical absorption spectroscopy of single-and few-layer 1,2-dithiolanes: toward donor-acceptor nanohybrids for MoS2 by hyperspectral imaging Nanotechnology 27 115705 energy conversion NPJ 2D Mater. Appl. 1 13 [1432] Rubio-Bollinger G, Guerrero R, de Lara D P, Quereda J, [1413] Vallan L, Canton-Vitoria R, Gobeze H B, Jang Y, Arenal R, Vaquero-Garzon L, Agraït N, Bratschitsch R and Benito A M, Maser W K, D’Souza F and Tagmatarchis N Castellanos-Gomez A 2015 Enhanced visibility of MoS2, 2018 Interfacing transition metal dichalcogenides with MoSe2, WSe2 and black-phosphorus: making optical carbon nanodots for managing photoinduced energy and identification of 2D semiconductors easier Electronics charge-transfer processes J. Am. Chem. Soc. 140 13488–96 4 847–56 [1414] Wirtz C, Hallam T, Cullen C P, Berner N C, O’Brien M, [1433] Zhao W, Ghorannevis Z, Chu L, Toh M, Kloc C, Tan P-H and Marcia M, Hirsch A and Duesberg G S 2015 Atomic layer Eda G 2012 Evolution of electronic structure in atomically deposition on 2D transition metal chalcogenides: layer thin sheets of WS2 and WSe2 ACS Nano 7 791–7 dependent reactivity and seeding with organic ad-layers [1434] Frisenda R, Niu Y, Gant P, Molina-Mendoza A J, Schmidt R, Chem. Commun. 51 16553–6 Bratschitsch R, Liu J, Fu L, Dumcenco D and Kis A 2017 [1415] Wang Y Y, Gao R X, Ni Z H, He H, Guo S P, Yang H P, Micro-reflectance and transmittance spectroscopy: a Cong C X and Yu T 2012 Thickness identification of two- versatile and powerful tool to characterize 2D materials J. dimensional materials by optical imaging Nanotechnology Phys. D: Appl. Phys. 50 074002 23 495713 [1435] Quereda J, San-Jose P, Parente V, Vaquero-Garzon L, [1416] Castellanos-Gomez A, Navarro-Moratalla E, Mokry G, Molina-Mendoza A J, Agraït N, Rubio-Bollinger G, Quereda J, Pinilla-Cienfuegos E, Agraït N, van der Zant H S, Guinea F, Roldán R and Castellanos-Gomez A 2016 Strong Coronado E, Steele G A and Rubio-Bollinger G 2013 Fast modulation of optical properties in black phosphorus and reliable identification of atomically thin layers of TaSe2 through strain-engineered rippling Nano Lett. 16 2931–7 crystals Nano Res. 6 191–9 [1436] Niu Y et al 2018 Thickness-dependent optical properties of [1417] Zhang H, Ran F, Shi X, Fang X, Wu S, Liu Y, Zheng X, Yang P, MoS2, MoSe2, WS2 and WSe2 Nanomaterials 8 725 Liu Y and Wang L 2017 Optical thickness identification of [1437] Kouroupis-Agalou K, Liscio A, Treossi E, Ortolani L, transition metal dichalcogenide nanosheets on transparent Morandi V, Pugno N M and Palermo V 2014 Fragmentation substrates Nanotechnology 28 164001 and exfoliation of 2-dimensional materials: a statistical [1418] Blake P, Hill E, Castro Neto A, Novoselov K, Jiang D, Yang R, approach Nanoscale 6 5926–33 Booth T and Geim A 2007 Making graphene visible Appl. [1438] Ridings C, Warr G G and Andersson G G 2012 Composition Phys. Lett. 91 063124 of the outermost layer and concentration depth profiles of [1419] Casiraghi C, Hartschuh A, Lidorikis E, Qian H, ammonium nitrate ionic liquid surfaces Phys. Chem. Chem. Harutyunyan H, Gokus T, Novoselov K S and Ferrari A C Phys. 14 16088–95 2007 Rayleigh imaging of graphene and graphene layers [1439] Nemes-Incze P, Osváth Z, Kamarás K and Biró L P 2008 Nano Lett. 7 2711–7 Anomalies in thickness measurements of graphene and [1420] Roddaro S, Pingue P, Piazza V, Pellegrini V and Beltram F few layer graphite crystals by tapping mode atomic force 2007 The optical visibility of graphene: interference colors of microscopy Carbon 46 1435–42 ultrathin graphite on SiO2 Nano Lett. 7 2707–10 [1440] Rubinstein M and Colby R H 2003 Polymer Physics (Oxford: [1421] Castellanos-Gomez A, Agraït N and Rubio-Bollinger G 2010 Oxford University Press) Optical identification of atomically thin dichalcogenide [1441] Mativetsky J M, Treossi E, Orgiu E, Melucci M, Veronese G P, crystals Appl. Phys. Lett. 96 213116 Samori` P and Palermo V 2010 Local current mapping and [1422] Benameur M, Radisavljevic B, Heron J, Sahoo S, Berger H patterning of reduced graphene oxide J. Am. Chem. Soc. and Kis A 2011 Visibility of dichalcogenide nanolayers 132 14130–6 Nanotechnology 22 125706 [1442] Song X et al 2017 Enhanced piezoelectric effect at the edges [1423] Castellanos-Gomez A, Wojtaszek M, Tombros N, Agraït N, of stepped molybdenum disulfide nanosheets Nanoscale van Wees B J and Rubio-Bollinger G 2011 Atomically thin 9 6237–45 mica flakes and their application as ultrathin insulating [1443] Sadewasser S and Lux-Steiner M C 2003 Correct height substrates for graphene Small 7 2491–7 measurement in noncontact atomic force microscopy Phys. [1424] Ares P, Zamora F and Gomez-Herrero J 2017 Optical Rev. Lett. 91 266101 identification of few-layer antimonene crystals ACS [1444] Yang K M, Chung J Y, Hsieh M F, Ferng S S, Lin D S and Photonics 4 600–5 Chiang T C 2016 Systematic variations in apparent [1425] Brotons-Gisbert M, Andres-Penares D, Martínez-Pastor J, topographic height as measured by noncontact atomic force Cros A and Sánchez-Royo J 2017 Optical contrast of 2D InSe microscopy Phys. Rev. B 74 193313 on SiO2/Si and transparent substrates using bandpass filters [1445] Novoselov K S 2004 Electric field effect in atomically thin Nanotechnology 28 115706 carbon films Science 306 666–9 274 2D Mater. 7 (2020) 022001 C Backes et al [1446] Tamayo J and García R 1997 Effects of elastic and inelastic microscopy: mapping dynamic electric phenomena in real interactions on phase contrast images in tapping-mode space Sci. Rep. 6 30557 scanning force microscopy Appl. Phys. Lett. 71 2394–6 [1470] O’Neill A, Khan U, Nirmalraj P N, Boland J and Coleman J N [1447] Paradinas M, Munuera C, Silien C, Buck M and Ocal C 2013 2011 Graphene dispersion and exfoliation in low boiling Heterogeneous nanotribological response of polymorphic point solvents J. Phys. Chem. C 115 5422–8 self-assembled monolayers arising from domain and phase [1471] Scherzer O D 1970 Die Strahlenschädigung der Objekte dependent friction Phys. Chem. Chem. Phys. 15 1302–9 als Grenze für die hochauflösende Elektronenmikroskopie [1448] García R 2002 Dynamic atomic force microscopy methods Berichte der Bunsengesellschaft für physikalische Chemie Surf. Sci. Rep. 47 197–301 74 1154–67 [1449] Wiesendanger R 1994 Scanning Probe Microscopy and [1472] Egerton R F, Li P and Malac M 2004 Radiation damage in the Spectroscopy (Cambridge: Cambridge University Press) TEM and SEM Micron 35 399–409 [1450] Tamayo J and García R 1996 Deformation, contact time, and [1473] Kaiser U et al 2011 Transmission electron microscopy phase contrast in tapping mode scanning force microscopy at 20kV for imaging and spectroscopy Ultramicroscopy Langmuir 12 4430–5 111 1239–46 [1451] García R, Magerle R and Perez R 2007 Nanoscale [1474] Eder F R, Kotakoski J, Kaiser U and Meyer J C 2014 A journey compositional mapping with gentle forces Nat. Mater. from order to disorder—atom by atom transformation from 6 405–11 graphene to a 2D carbon glass Sci. Rep. 4 4060 [1452] Obraztsov A N, Obraztsova E A, Tyurnina A V and [1475] Banhart F 1999 Irradiation effects in carbon nanostructures Zolotukhin A A 2007 Chemical vapor deposition of thin Rep. Progr. Phys. 62 1181–221 graphite films of nanometer thickness Carbon 45 2017–21 [1476] Cherns D, Minter F J and Nelson R S 1976 Sputtering in the [1453] Deng S and Berry V 2016 Wrinkled, rippled and crumpled high voltage electron microscope Nucl. Instrum. Methods graphene: an overview of formation mechanism, electronic 132 369–76 properties, and applications Mater. Today 19 197–212 [1477] Egerton R F, McLeod R, Wang F and Malac M 2010 Basic [1454] Filleter T, McChesney J L, Bostwick A, Rotenberg E, questions related to electron-induced sputtering in the TEM Emtsev K V, Seyller T, Horn K and Bennewitz R 2009 Friction Ultramicroscopy 110 991–7 and dissipation in epitaxial graphene films Phys. Rev. Lett. [1478] Suenaga K, Iizumi Y and Okazaki T 2011 Single atom 102 086102 spectroscopy with reduced delocalization effect using a 30 [1455] Vasic B, Matkovic A, Gajic R and Stankovic I 2016 Wear kV-STEM Eur. Phys. J. Appl. Phys. 54 33508 properties of graphene edges probed by atomic force [1479] Ricolleau C, Nelayah J, Oikawa T, Kohno Y, Braidy N, microscopy based lateral manipulation Carbon 107 723–32 Wang G, Hue F, Florea L, Pierron Bohnes V and Alloyeau D [1456] Martínez N F and García R 2006 Measuring phase shifts and 2012 Performances of an 80–200 kV microscope employing energy dissipation with amplitude modulation atomic force a cold-FEG and an aberration-corrected objective lens microscopy Nanotechnology 17 S167–72 Microscopy 62 283–93 [1457] Palacios-Lidón E, Munuera C, Ocal C and Colchero J 2010 [1480] Krivanek O L, Ursin J P, Bacon N J, Corbin G J, Dellby N, Contrast inversion in non-contact dynamic scanning force Hrncirik P, Murfitt M F, Own C S and Szilagyi Z S 2009 High- microscopy: what is high and what is low? Ultramicroscopy energy-resolution monochromator for aberration-corrected 110 789–800 scanning transmission electron microscopy/electron [1458] Carpick R W and Salmeron M 1997 Scratching the surface: energy-loss spectroscopy Phil. Trans. R. Soc. A 367 3683–97 fundamental investigations of tribology with atomic force [1481] Morishita S, Mukai M, Suenaga K and Sawada H 2016 microscopy Chem. Rev. 97 1163–94 Resolution enhancement in transmission electron [1459] Berman D, Erdemir A and Sumant A V 3014 Graphene: a microscopy with 60 kV monochromated electron source new emerging lubricant Mater. Today 17 31–42 Appl. Phys. Lett. 108 013107 [1460] Butt H-J, Cappella B and Kappl M 2005 Force measurements [1482] Linck M et al 2016 Chromatic aberration correction for with the atomic force microscope: technique, interpretation atomic resolution TEM imaging from 20 to 80 kV Phys. Rev. and applications Surf. Sci. Rep. 59 1–152 Lett. 117 076101 [1461] Cappella B and Dietler G 1999 Force-distance curves by [1483] Lee Z, Meyer J C, Rose H and Kaiser U 2012 Optimum atomic force microscopy Surf. Sci. Rep. 34 1–104 HRTEM image contrast at 20 kV and 80 kV—exemplified by [1462] Kazakova O, Panchal V and Burnett T 2013 Epitaxial graphene Ultramicroscopy 112 39–46 graphene and graphene–based devices studied by electrical [1484] Egerton R F 2014 Choice of operating voltage for a scanning probe microscopy Crystals 3 191–233 transmission electron microscope Ultramicroscopy [1463] Hormeño S, Penedo M, Manzano C V and Luna M 2013 145 85–93 Gold nanoparticle coated silicon tips for Kelvin probe force [1485] Algara-Siller G, Lehtinen O, Turchanin A and Kaiser U 2014 microscopy in air Nanotechnology 24 395701 Dry-cleaning of graphene Appl. Phys. Lett. 104 153115 [1464] Filleter T, Emtsev K V, Seyller T and Bennewitz R 2008 Local [1486] Algara-Siller G, Kurasch S, Sedighi M, Lehtinen O and work function measurements of epitaxial graphene Appl. Kaiser U 2013 The pristine atomic structure of MoS2 Phys. Lett. 93 133117 monolayer protected from electron radiation damage by [1465] Ziegler D, Gava P, Güttinger J, Molitor F, Wirtz L, Lazzeri M, Appl. Phys. Lett. 103 203107 Saitta A M, Stemmer A, Mauri F and Stampfer C 2011 [1487] Chamberlain T W, Biskupek J, Skowron S T, Bayliss P A, Variations in the work function of doped single- and few- Bichoutskaia E, Kaiser U and Khlobystov A N 2015 layer graphene assessed by Kelvin probe force microscopy Transmission electron microscopy: isotope substitution and density functional theory Phys. Rev. B 83 235434 extends the lifetime of organic molecules in transmission [1466] Eriksson J, Pearce R, Iakimov T, Virojanadara C, Gogova D, electron microscopy (Small 5/2015) Small 11 510 Andersson M, Syväjärvi M, Lloyd Spetz A and Yakimova R [1488] Wiesendanger R Related scanning probe methods Scanning 2012 The influence of substrate morphology on thickness Probe Microscopy and Spectroscopy (Cambridge: Cambridge uniformity and unintentional doping of epitaxial graphene University Press) pp 265–88 on SiC Appl. Phys. Lett. 100 241607 [1489] Wiesendanger R Applications of scanning probe microscopy [1467] Nonnenmacher M, O’Boyle M P and Wickramasinghe H K and spectroscopy Scanning Probe Microscopy and Spectroscopy 1991 Kelvin probe force microscopy Appl. Phys. Lett. (Cambridge: Cambridge University Press) pp 289–90 58 2921–3 [1490] Merino P, Rodrigo L, Pinardi A L, Méndez J, López M F, [1468] Mohn F, Gross L, Moll N and Meyer G 2012 Imaging Pou P, Pérez R and Martín Gago J A 2014 Sublattice localized the charge distribution within a single molecule Nat. electronic states in atomically resolved graphene-Pt(1 1 1) Nanotechnol. 7 227–31 edge-boundaries ACS Nano 8 3590–6 [1469] Collins L, Belianinov A, Somnath S, Balke N, Kalinin S V [1491] Merino P, Santos H, Pinardi A L, Chico L and Martin- and Jesse S 2016 Full data acquisition in Kelvin probe force Gago J A 2017 Atomically-resolved edge states on surface- 275 2D Mater. 7 (2020) 022001 C Backes et al nanotemplated graphene explored at room temperature [1516] Glebov A L, Mokhun O, Rapaport A, Vergnole S, Smirnov V Nanoscale 9 3905–11 and Glebov L B 2012 Volume Bragg gratings as ultra-narrow [1492] Telychko M et al 2015 Electronic and chemical properties of and multiband optical filters Proc. SPIE 8428 84280C donor, acceptor centers in graphene ACS Nano 9 9180–7 [1517] Faugeras C, Nerrière A, Potemski M, Mahmood A, [1493] Lauffer P, Emtsev K V, Graupner R, Seyller T, Ley L, Dujardin E, Berger C and de Heer W A 2008 Few-layer Reshanov S A and Weber H B 2008 Atomic and electronic graphene on SiC, pyrolitic graphite, and graphene: a Raman structure of few-layer graphene on SiC(0 0 0 1) studied with scattering study Appl. Phys. Lett. 92 011914 scanning tunneling microscopy and spectroscopy Phys. Rev. [1518] Das A et al 2008 Monitoring dopants by Raman scattering B 77 155426 in an electrochemically top-gated graphene transistor Nat. [1494] Varchon F, Mallet P, Magaud L and Veuillen J-Y 2008 Nanotechnol. 3 210–5 Rotational disorder in few-layer graphene films on [1519] Basko D M, Piscanec S and Ferrari A C 2009 Electron- 6H–SiC(0 0 0   −  1): a scanning tunneling microscopy study electron interactions and doping dependence of the two- Phys. Rev. B 77 165415 phonon Raman intensity in graphene Phys. Rev. B 80 165413 [1495] Nika D L and Balandin A A 2016 Thermal transport in [1520] Piscanec S, Lazzeri M, Mauri F, Ferrari A C and Robertson J graphene, few-layer graphene and graphene nanoribbons 2004 Kohn anomalies and electron–phonon interactions in Thermal Transport in Low Dimensions (Berlin: Springer) graphite Phys. Rev. Lett. 93 185503 pp 339–63 [1521] Das A, Chakraborty B, Piscanec S, Pisana S, Sood A K and [1496] Xu X et al 2014 Length-dependent thermal conductivity in Ferrari A C 2009 Phonon renormalization in doped bilayer suspended single-layer graphene Nat. Commun. 5 3689 graphene Phys. Rev. B 79 155417 [1497] Sadeghi M M, Pettes M T and Shi L 2012 Thermal transport [1522] Pisana S, Lazzeri M, Casiraghi C, Novoselov K S, Geim A K, in graphene Solid State Commun. 152 1321–30 Ferrari A C and Mauri F 2007 Breakdown of the adiabatic [1498] Balandin A A 2011 Thermal properties of graphene and Born–Oppenheimer approximation in graphene Nat. Mater. nanostructured carbon materials Nat. Mater. 10 569–81 6 198–201 [1499] Ho C Y, Powell R W and Liley P E 1972 Thermal conductivity [1523] Yan J, Zhang Y, Kim P and Pinczuk A 2007 Electric field effect of the elements J. Phys. Chem. Ref. Data 1 279–21 tuning of electron–phonon coupling in graphene Phys. Rev. [1500] Nika D L and Balandin A A 2017 Phonons and thermal Lett. 98 166802 transport in graphene and graphene-based materials Rep. [1524] Ferrari A C and Robertson J 2000 Interpretation of Raman Prog. Phys. 80 036502 spectra of disordered and amorphous carbon Phys. Rev. B [1501] Seol J H et al 2010 Two-dimensional phonon transport in 61 14095–107 supported graphene Science 328 213–6 [1525] Casiraghi C 2007 Probing disorder and charged impurities [1502] Balandin A A, Ghosh S, Bao W, Calizo I, Teweldebrhan D, in graphene by Raman spectroscopy Phys. Status Solidi Miao F and Lau C N 2008 Superior thermal conductivity of 3 175–77 single-layer graphene Nano Lett. 8 902–7 [1526] Cançado L G, Pimenta M A, Neves B R A, Dantas M S S and [1503] Majumdar A 1999 Scanning thermal microscopy Annu. Rev. Jorio A 2004 Influence of the atomic structure on the Raman Mater. Sci. 29 505–85 spectra of graphite edges Phys. Rev. Lett. 93 247401 [1504] Gomès S, Assy A and Chapuis P-O 2015 Scanning thermal [1527] Casiraghi C, Pisana S, Novoselov K S, Geim A K and microscopy: a review Phys. Status Solidi a 212 477–94 Ferrari A C 2007 Raman fingerprint of charged impurities in [1505] Yoon K, Hwang G, Chung J, Kim H G, Kwon O, Kihm K D graphene Appl. Phys. Lett. 91 233108 and Lee J S 2014 Measuring the thermal conductivity of [1528] Zhao W, Tan P H, Liu J and Ferrari A C 2011 Intercalation of residue-free suspended graphene bridge using null point few-layer graphite flakes with FeCl3: Raman determination scanning thermal microscopy Carbon 76 77–83 of fermi level, layer by layer decoupling, and stability J. Am. [1506] Pumarol M E, Rosamond M C, Tovee P, Petty M C, Zeze D A, Chem. Soc. 133 5941–6 Falko V and Kolosov O V 2012 Direct nanoscale imaging [1529] Mohiuddin T M G et al 2009 Uniaxial strain in graphene of ballistic and diffusive thermal transport in graphene by Raman spectroscopy:Gpeak splitting, Grüneisen nanostructures Nano Lett. 12 2906–11 parameters, and sample orientation, Phys. Rev. B 79 205433 [1507] Martinek J, Klapetek P and Campbell A C 2015 Methods [1530] Proctor J E, Gregoryanz E, Novoselov K S, Lotya M, for topography artifacts compensation in scanning thermal Coleman J N and Halsall M P 2009 High-pressure Raman microscopy Ultramicroscopy 155 55–61 spectroscopy of graphene Phys. Rev. B 80 073408 [1508] Yovanovich M M, Culham J R and Teertstra P 1998 [1531] Zabel J, Nair R R, Ott A, Georgiou T, Geim A K, Analytical modeling of spreading resistance in flux tubes, Novoselov K S and Casiraghi C 2012 Raman spectroscopy half spaces, and compound disks IEEE Trans. Compon. of graphene and bilayer under biaxial strain: bubbles and Packag. Manuf. Technol. A 21 168–76 balloons Nano Lett. 12 617–21 [1509] Menges F, Riel H, Stemmer A, Dimitrakopoulos C and [1532] Huang M, Yan H, Heinz T F and Hone J 2010 Probing strain- Gotsmann B 2013 Thermal transport into graphene through induced electronic structure change in graphene by Raman nanoscopic contacts Phys. Rev. Lett. 111 205901 spectroscopy Nano Lett. 10 4074–9 [1510] Tuinstra F and Koenig J L 1970 Raman spectrum of graphite [1533] Mohr M, Maultzsch J and Thomsen C 2010 Splitting of the J. Chem. Phys. 53 1126–30 Raman 2D band of graphene subjected to strain Phys. Rev. B [1511] Ferrari A C and Robertson J 2004 Raman spectroscopy of 82 209905 amorphous, nanostructured, diamond-like carbon, and [1534] Yoon D, Son Y-W and Cheong H 2011 Strain-dependent nanodiamond Phil. Trans. R. Soc. A 362 2477–512 splitting of the double-resonance Raman scattering band in [1512] Saito R, Nugraha A R T, Hasdeo E H, Siregar S, Guo H and graphene Phys. Rev. Lett. 106 155502 Yang T 2015 Ultraviolet Raman spectroscopy of graphene [1535] Ni Z H, Yu T, Lu Y H, Wang Y Y, Feng Y P and Shen Z X 2008 and transition-metal dichalcogenides Phys. Status Solidi b Uniaxial strain on graphene: Raman spectroscopy study and 252 2363–74 band-gap opening ACS Nano 2 2301–5 [1513] Tyborski C, Herziger F, Gillen R and Maultzsch J 2015 [1536] Huang M, Yan H, Chen C, Song D, Heinz T F and Hone J Beyond double-resonant Raman scattering: Ultraviolet 2009 Phonon softening and crystallographic orientation of Raman spectroscopy on graphene, graphite, and carbon strained graphene studied by Raman spectroscopy Proc. Natl nanotubes Phys. Rev. B 92 041401 Acad. Sci. 106 7304–8 [1514] Tan P H et al 2012 The shear mode of multilayer graphene [1537] Kuzuba T, Era K, Ishii T and Sato T 1978 A low frequency Nat. Mater. 11 294–300 Raman-active vibration of hexagonal boron nitride Solid [1515] Zhang X, Han W P, Wu J B, Milana S, Lu Y, Li Q Q, Ferrari A C State Commun. 25 863–5 and Tan P H, 2012 Raman spectroscopy of shear and layer [1538] Stenger I, Schué L, Boukhicha M, Berini B, Plaçais B, breathing modes in multilayer MoS2 Phys. Rev. B 87 115413 Loiseau A and Barjon J 2017 Low frequency Raman 276 2D Mater. 7 (2020) 022001 C Backes et al spectroscopy of few-atomic-layer thick hBN crystals 2D [1560] Cao X K, Clubine B, Edgar J H, Lin J Y and Jiang H X 2013 Mater. 4 031003 Two-dimensional excitons in three-dimensional hexagonal [1539] Arenal R, Ferrari A C, Reich S, Wirtz L, Mevellec J Y, boron nitride Appl. Phys. Lett. 103 191106 Lefrant S, Rubio A and Loiseau A 2006 Raman [1561] Cassabois G, Valvin P and Gil B 2016 Hexagonal boron spectroscopy of single-wall boron nitride nanotubes Nano nitride is an indirect bandgap semiconductor Nat. Photon. Lett. 6 1812–6 10 262–6 [1540] Nemanich R J, Solin S A and Martin R M 1981 Light [1562] Watanabe K, Taniguchi T, Miya K, Sato Y, Nakamura K, scattering study of boron nitride microcrystals Phys. Rev. B Niiyama T and Taniguchi M 2011 Hexagonal boron 23 6348–56 nitride as a new ultraviolet luminescent material and its [1541] Reich S, Ferrari A C, Arenal R, Loiseau A, Bello I and application—Fluorescence properties of hBN single-crystal Robertson J 2005 Resonant Raman scattering in cubic and powder Diamond Relat. Mater. 20 849–52 hexagonal boron nitride Phys. Rev. B 71 205201 [1563] Pierret A, Loayza J, Berini B, Betz A, Plaçais B, Ducastelle F, [1542] Chakraborty B, Bera A, Muthu D V S, Bhowmick S, Barjon J and Loiseau A 2014 Excitonic recombinations Waghmare U V and Sood A K 2012 Symmetry-dependent inh  −  BN: from bulk to exfoliated layers Phys. Rev. B phonon renormalization in monolayer MoS2 transistor Phys. 89 035414 Rev. B 85 161403(R) [1564] Museur L and Kanaev A 2008 Near band-gap [1543] Ribeiro-Soares J, Almeida R M, Barros E B, Araujo P T, photoluminescence properties of hexagonal boron nitride J. Dresselhaus M S, Cançado L G and Jorio A 2014 Group Appl. Phys. 103 103520 theory analysis of phonons in two-dimensional transition [1565] Silly M G, Jaffrennou P, Barjon J, Lauret J S, Ducastelle F, metal dichalcogenides Phys. Rev. B 90 115438 Loiseau A, Obraztsova E, Attal-Tretout B and Rosencher E [1544] Scheuschner N, Gillen R, Staiger M and Maultzsch J 2015 2007 Luminescence properties of hexagonal boron Interlayer resonant Raman modes in few-layer MoS2 Phys. nitride: Cathodoluminescence and photoluminescence Rev. B 91 235409 spectroscopy measurements Phys. Rev. B 75 085205 [1545] Mignuzzi S, Pollard A J, Bonini N, Brennan B, Gilmore I S, [1566] Vuong T Q P, Cassabois G, Valvin P, Ouerghi A, Pimenta M A, Richards D and Roy D 2015 Effect of disorder Chassagneux Y, Voisin C and Gil B 2016 Phonon–photon on Raman scattering of single-layerMoS2 Phys. Rev. B mapping in a color center in hexagonal boron nitride Phys. 91 195411 Rev. Lett. 117 097402 [1546] Conley H J, Wang B, Ziegler J I, Haglund R F, Pantelides S T [1567] Bourrellier R, Meuret S, Tararan A, Stéphan O, Kociak M, and Bolotin K I 2013 Bandgap engineering of strained Tizei L H G and Zobelli A 2016 Bright UV Single Photon monolayer and bilayer MoS2 Nano Lett. 13 3626–30 Emission at Point Defects inh-BN, Nano Lett. 16 4317–21 [1547] Wilson J A and Yoffe A D 1969 Transition metal [1568] Tran T T, Elbadawi C, Totonjian D, Lobo C J, Grosso G, dichalcogenides. Discussion and interpretation of the Moon H, Englund D R, Ford M J, Aharonovich I and Toth M observed optical, electrical, and structural properties Adv. 2016 Robust multicolor single photon emission from point Phys. 18 193–335 defects in hexagonal boron nitride ACS Nano 10 7331–8 [1548] Paton K R and Coleman J N 2016 Relating the optical [1569] Castle J E 1894 Practical surface analysis by Auger and absorption coefficient of nanosheet dispersions to the x-ray photoelectron spectroscopy. D. Briggs and M. P. Seah intrinsic monolayer absorption Carbon 107 733–8 (Editors). John Wiley and Sons Ltd, Chichester 1983 533 pp., [1549] Yadgarov L, Choi C L, Sedova A, Cohen A, Rosentsveig R, £44.50 Surf. Interface Anal. 6 302 Bar-Elli O, Oron D, Dai H and Tenne R 2014 Dependence [1570] Crist B V 2005 Handbooks of Monochromatic XPS Spectra of the absorption and optical surface plasmon scattering (Mountain View, CA: XPS International LLC) of MoS2 nanoparticles on aspect ratio, size, and media ACS [1571] Preobrajenski A B, Ng M L, Vinogradov A S and Nano 8 3575–83 Mårtensson N 2008 Controlling graphene corrugation on [1550] Mak K F, Lee C, Hone J, Shan J and Heinz T F 2010 lattice-mismatched substrates Phys. Rev. B 78 073401 Atomically thin MoS2: a new direct-gap semiconductor Phys. [1572] IkeoY I N, Nimura N, Sigematsu M, Tazawa T, Matsumoto S, Rev. Lett. 105 136805 Kojima K and Nagasawa Y 1991 Handbook of x-ray [1551] Splendiani A, Sun L, Zhang Y, Li T, Kim J, Chim C-Y, Photoelectron Spectroscopy (JEOL) Galli G and Wang F 2010 Emerging photoluminescence in [1573] Bianco G V, Losurdo M, Giangregorio M M, Capezzuto P monolayer MoS2 Nano Lett. 10 1271–5 and Bruno G 2014 Exploring and rationalising effective [1552] Watanabe K, Taniguchi T and Kanda H 2004 Direct-bandgap n-doping of large area CVD-graphene by NH3 Phys. Chem. properties and evidence for ultraviolet lasing of hexagonal Chem. Phys. 16 3632 boron nitride single crystal Nat. Mater. 3 404–9 [1574] Blume R et al 2014 The influence of intercalated oxygen [1553] Jaffrennou P, Barjon J, Lauret J S, Attal-Trétout B, on the properties of graphene on polycrystalline Cu under Ducastelle F and Loiseau A 2007 Origin of the excitonic various environmental conditions Phys. Chem. Chem. Phys. recombinations in hexagonal boron nitride by spatially 16 25989–6003 resolved cathodoluminescence spectroscopy J. Appl. Phys. [1575] Vinogradov N A, Schulte K, Ng M L, Mikkelsen A, 102 116102 Lundgren E, Mårtensson N and Preobrajenski A B 2011 [1554] Taniguchi T and Watanabe K 2007 Synthesis of high-purity Impact of atomic oxygen on the structure of graphene boron nitride single crystals under high pressure by using formed on Ir(1 1 1) and Pt(1 1 1) J. Phys. Chem. C 115 9568– Ba–BN solvent J. Cryst. Growth 303 525–9 77 [1555] Schué L, Berini B, Betz A C, Plaçais B, Ducastelle F, [1576] Barinov A, Malcioǧu O B, Fabris S, Sun T, Gregoratti L, Barjon J and Loiseau A 2016 Dimensionality effects on the Dalmiglio M and Kiskinova M 2009 Initial stages of luminescence properties of hBN Nanoscale 8 6986–93 oxidation on graphitic surfaces: photoemission study and [1556] Galvani T, Paleari F, Miranda H P C, Molina-Sánchez A, density functional theory calculations J. Phys. Chem. C Wirtz L, Latil S, Amara H and Ducastelle F 2016 Excitons in 113 9009–13 boron nitride single layer Phys. Rev. B 94 125303 [1577] Palacio I et al 2017 Spectroscopic characterization of [1557] Arnaud B, Lebègue S, Rabiller P and Alouani M 2008 the on-surface induced (cyclo)dehydrogenation of a Arnaud, lebègue, rabiller, and alouani reply Phys. Rev. Lett. N-heteroaromatic compound on noble metal surfaces Phys. 100 189702 Chem. Chem. Phys. 19 22454–61 [1558] Watanabe K and Taniguchi T 2009 Jahn–Teller effect on [1578] Casero E et al 2012 Comparative response of biosensing exciton states in hexagonal boron nitride single crystal Phys. platforms based on synthesized graphene oxide and Rev. B 79 180405 electrochemically reduced graphene Electroanalysis [1559] Museur L et al 2011 Exciton optical transitions in a 25 154–65 hexagonal boron nitride single crystal Phys. Status Solidi [1579] Starke U 1997 Atomic structure of hexagonal SiC surfaces 5 214–6 Phys. Status Solidi b 202 475–99 277 2D Mater. 7 (2020) 022001 C Backes et al [1580] Owman F and Mårtensson P 1996 The [1603] Hertel S, Kisslinger F, Jobst J, Waldmann D, Krieger M and SiC(0 0 0 1)6  √  3  ×  6  √  3 reconstruction studied with STM Weber H B 2011 Current annealing and electrical breakdown and LEED Surf. Sci. 369 126–36 of epitaxial graphene Appl. Phys. Lett. 98 212109 [1581] Varchon F, Mallet P, Veuillen J Y and Magaud L 2008 Ripples [1604] Tombros N, Veligura A, Junesch J, Guimaraes M H D, Vera- in epitaxial graphene on the Si-terminated SiC(0 0 0 1) Marun I J, Jonkman H T and van Wees B J 2011 Quantized surface Phys. Rev. B 77 235412 conductance of a suspended graphene nanoconstriction [1582] de Lima L H, de Siervo A, Landers R, Viana G A, Nat. Phys. 7 697–700 Goncalves A M B, Lacerda R G and Häberle P 2013 Atomic [1605] Lin Y M, Dimitrakopoulos C, Jenkins K A, Farmer D B, surface structure of graphene and its buffer layer on Chiu H Y, Grill A and Avouris P 2010 GHz transistors from SiC(0 0 0 1): a chemical-specific photoelectron diffraction wafer-scale epitaxial graphene Science 327 662–2 approach Phys. Rev. B 87 081403 [1606] Moon J S et al 2010 Top-gated epitaxial graphene FETs on [1583] Abrami A et al 1995 Super ESCA: first beamline operating at Si-face SiC wafers with a peak transconductance of 600 mS ELETTRA Rev. Sci. Instrum. 66 1618–20 mm−1 IEEE Electron Device Lett. 31 260–2 [1584] Zampieri G, Lizzit S, Petaccia L, Goldoni A, Baraldi A, [1607] Guo Z L et al 2013 Record maximum oscillation frequency in Bremholm M, Gayone J E, Hoffmann S V and Hofmann P C-face epitaxial graphene transistors Nano Lett. 13 942–7 2005 Photoelectron diffraction study of the6H– [1608] Das Sarma S, Adam S, Hwang E H and Rossi E 2011 SiC(0 0 0 1)3  ×  3R30 reconstruction Phys. Rev. B 72 165327 Electronic transport in two-dimensional graphene Rev. Mod. [1585] Merino P, Švec M, Martinez J I, Jelinek P, Lacovig P, Phys. 83 407–66 Dalmiglio M, Lizzit S, Soukiassian P, Cernicharo J and [1609] Chen J H, Jang C, Adam S, Fuhrer M S, Williams E D and Martin-Gago J A 2014 Graphene etching on SiC grains as Ishigami M 2008 Charged-impurity scattering in graphene a path to interstellar polycyclic aromatic hydrocarbons Nat. Phys. 4 377–81 formation Nat. Commun. 5 3054 [1610] Shon N H and Ando T 1998 Quantum transport in two- [1586] Damascelli A 2004 Probing the electronic structure of dimensional graphite system J. Phys. Soc. Japan 67 2421–9 complex systems by ARPES Phys. Scr. T109 61 [1611] Dlubak B, Seneor P, Anane A, Barraud C, Deranlot C, [1587] Reinert F and Hüfner S 2005 Photoemission spectroscopy— Deneuve D, Servet B, Mattana R, Petroff F and Fert A from early days to recent applications New J. Phys. 7 97 2010 Are Al2O3 and MgO tunnel barriers suitable for spin [1588] Himpsel F J 1983 Angle-resolved measurements of the injection in graphene? Appl. Phys. Lett. 97 092502 photoemission of electrons in the study of solids Adv. Phys. [1612] Li X, Wu X, Sprinkle M, Ming F, Ruan M, Hu Y, Berger C and 32 1–51 de Heer W A 2010 Top- and side-gated epitaxial graphene [1589] Einstein A 1905 Über einen die Erzeugung und Verwandlung field effect transistors Phys. Status Solidi a 207 286–90 des Lichtes betreffenden heuristischen Gesichtspunkt Ann. [1613] Kim S, Nah J, Jo I, Shahrjerdi D, Colombo L, Yao Z, Tutuc E Phys., Lpz. 322 132–48 and Banerjee S K 2009 Realization of a high mobility dual- [1590] Huang D J, Lee J Y, Suen J S, Mulhollan G A, Andrews A B and gated graphene field-effect transistor with Al2O3 dielectric Erskine J L 1993 Adapting a compact Mott spin polarimeter Appl. Phys. Lett. 94 062107 to a large commercial electron energy analyzer for spin- [1614] Zhong H, Zhang Z Y, Xu H T, Qiu C G and Peng L M 2015 polarized electron spectroscopy Rev. Sci. Instrum. 64 3474–9 Comparison of mobility extraction methods based on field- [1591] McChesney J L, Bostwick A, Ohta T, Seyller T, Horn K, effect measurements for graphene AIP Adv. 5 057136 González J and Rotenberg E 2010 Extended van Hove [1615] Meric I, Han M Y, Young A F, Ozyilmaz B, Kim P and singularity and superconducting instability in doped Shepard K L 2008 Current saturation in zero-bandgap, graphene Phys. Rev. Lett. 104 136803 topgated graphene field-effect transistors Nat. Nanotechnol. [1592] Mazzola F, Frederiksen T, Balasubramanian T, Hofmann P, 3 654–9 Hellsing B and Wells J W 2017 Strong electron-phonon [1616] Droscher S, Roulleau P, Molitor F, Studerus P, Stampfer C, coupling in the σ band of graphene Phys. Rev. B 95 075430 Ensslin K and Ihn T 2010 Quantum capacitance and density [1593] Mazzola F et al 2013 Publisher’s Note: Kinks in the σ band of of states of graphene Appl. Phys. Lett. 96 152104 graphene induced by electron–phonon coupling (2013 Phys. [1617] Zhang Y, Tan Y-W, Stormer H L and Kim P 2005 Rev. Lett. 111 216806) Phys. Rev. Lett. 111 216806 Experimental observation of the quantum Hall effect and [1594] Pletikosic I, Kralj M, Pervan P, Brako R, Coraux J, Berry’s phase in graphene Nature 438 201 N’Diaye A T, Busse C and Michely T 2009 Dirac cones [1618] Poumirol J M, Escoffier W, Kumar A, Goiran M, Raquet B and minigaps for graphene on Ir(1 1 1) Phys. Rev. Lett. and Broto J M 2010 Electron–hole coexistence in disordered 102 056808 graphene probed by high-field magneto-transport New J. [1595] Yuan H et al 2016 Evolution of the valley position in bulk Phys. 12 83006 transition-metal chalcogenides and their monolayer limit [1619] Wiedmann S et al 2011 Coexistence of electron and hole Nano Lett. 16 4738–45 transport in graphene Phys. Rev. B 84 115314 [1596] Pierucci D et al 2016 Band alignment and minigaps in [1620] de Heer W A et al 2007 Epitaxial graphene Solid State monolayer MoS2-graphene van der Waals heterostructures Commun. 143 92–100 Nano Lett. 16 4054–61 [1621] Schwierz F 2010 Graphene transistors Nat. Nanotechnol. [1597] Dendzik M, Michiardi M, Sanders C, Bianchi M, Miwa J A, 5 487 Grønborg S S, Lauritsen J V, Bruix A, Hammer B and [1622] Dorgan V E, Bae M H and Pop E 2010 Mobility and Hofmann P 2015 Growth and electronic structure of saturation velocity in graphene on SiO2 Appl. Phys. Lett. epitaxial single-layerWS2 on Au(1 1 1) Phys. Rev. B 97 082112 92 245442 [1623] Robinson J A, LaBella M, Zhu M, Hollander M, Kasarda R, [1598] Ulstrup S et al 2017 Spin and valley control of free carriers in Hughes Z, Trumbull K, Cavalero R and Snyder D 2011 single-layer WS2 Phys. Rev. B 95 041405(R) Contacting graphene Appl. Phys. Lett. 98 053103 [1599] Katoch J et al 2018 Giant spin-splitting and gap [1624] Ferney A C, David J, Abhay A S, Wonjae K, Juha R, Harri L renormalization driven by trions in single-layer WS2/h-BN and Daniel N 2015 A physics-based model of gate-tunable heterostructures Nat. Phys. 14 355–9 metal–graphene contact resistance benchmarked against [1600] van der Pauw L J 1958 A method for measuring resistivity experimental data 2D Mater. 2 025006 and Hall effects of disks of arbitrary shape Phil. Res. Rep. [1625] Nagashio K and Toriumi A 2011 Density-of-states limited 13 1–9 contact resistance in graphene field-effect transistors Japan. [1601] Edwards D F A 1971 Electronic Measurement Techniques J. Appl. Phys. 50 070108 (Butterworth-Heinemann) [1626] Xia F N, Perebeinos V, Lin Y M, Wu Y Q and Avouris P 2011 [1602] Moser J, Barreiro A and Bachtold A 2007 Current-induced The origins and limits of metal-graphene junction resistance cleaning of graphene Appl. Phys. Lett. 91 163513 Nat. Nanotechnol. 6 179–84 278 2D Mater. 7 (2020) 022001 C Backes et al [1627] Shaygan M, Otto M, Sagade A A, Chavarin C A, Bacher G, [1649] Castellanos-Gomez A, Singh V, van der Zant H S J and Mertin W and Neumaier D 2017 Low resistive edge contacts Steele G A 2014 Mechanics of freely-suspended ultrathin to CVD-grown graphene using a CMOS compatible metal layered materials Ann. Phys., Lpz. 527 27–44 Ann. Phys., Lpz. 529 1600410 [1650] Novoselov K S 2011 Nobel lecture: graphene: materials in the [1628] Wang L et al 2013 One-dimensional electrical contact to a flatland Rev. Mod. Phys. 83 837–49 two-dimensional material Science 342 614–7 [1651] Huang M, Pascal T A, Kim H, Goddard W A and Greer J R [1629] Robinson J A, LaBella M, Trumbull K A, Weng X J, 2011 Electronic–mechanical coupling in graphene from in Cavelero R, Daniels T, Hughes Z, Hollander M, Fanton M situ nanoindentation experiments and multiscale atomistic and Snyder D 2010 Epitaxial graphene materials integration: simulations Nano Lett. 11 1241–6 effects of dielectric overlayers on structural and electronic [1652] Lee C, Wei X, Kysar J W and Hone J 2008 Measurement of properties ACS Nano 4 2667–72 the elastic properties and intrinsic strength of monolayer [1630] Wu Y Q, Farmer D B, Valdes-Garcia A, Zhu W J, graphene Science 321 385–8 Jenkins K A, Dimitrakopoulos C, Avouris P and [1653] Lee G H et al 2013 High-strength chemical-vapor-deposited Lin Y M 2011 Record high RF performance for epitaxial graphene and grain boundaries Science 340 1073–6 graphene transistors 2011 IEEE Int. Electron Devices [1654] Blakslee O L, Proctor D G, Seldin E J, Spence G B and Weng T Meeting (IEDM) 1970 Elastic constants of compression-annealed pyrolytic [1631] Cheng R, Bai J, Liao L, Zhou H, Chen Y, Liu L, Lin Y C, graphite J. Appl. Phys. 41 3373–82 Jiang S, Huang Y and Duan X 2012 High-frequency self- [1655] Castellanos-Gomez A, Poot M, Steele G A, van der Zant H S J, aligned graphene transistors with transferred gate stacks Agraït N and Rubio-Bollinger G 2012 Elastic properties of Proc. Natl Acad. Sci. 109 11588–92 freely suspended MoS2 nanosheets, Adv. Mater. 24 772–5 [1632] Mayorov A S et al 2011 Micrometer-scale ballistic transport [1656] Bertolazzi S, Brivio J and Kis A 2011 Stretching and breaking in encapsulated graphene at room temperature Nano Lett. of ultrathin MoS2, ACS Nano 5 9703–9 11 2396–9 [1657] Liu K et al 2014 Elastic properties of chemical–vapor– [1633] Deniz J P D, Hu Y, Hu Y, Turmaud J-P, Prudkovskiy V S, deposited monolayer MoS2, WS2, and their bilayer Berger C and de Heer W A 2018 Unpublished heterostructures Nano Lett. 14 5097–03 [1634] Datta S 1995 Electronic Transport in Mesoscopic Systems [1658] Georgiou T, Britnell L, Blake P, Gorbachev R V, Gholinia A, (Cambridge: Cambridge University Press) Geim A K, Casiraghi C and Novoselov K S 2011 Graphene [1635] Tomaino J L, Jameson A D, Kevek J W, Paul M J, van der bubbles with controllable curvature Appl. Phys. Lett. Zande A M, Barton R A, McEuen P L, Minot E D and Lee Y S 99 093103 2011 Terahertz imaging and spectroscopy of large-area [1659] Vlassak J J and Nix W D 1992 A new bulge test technique for single-layer graphene Opt. Express 19 141–6 the determination of Young’s modulus and Poisson’s ratio [1636] Buron J D et al 2012 Graphene conductance uniformity of thin films J. Mater. Res. 7 3242–9 mapping Nano Lett. 12 5074–81 [1660] Koenig S P, Boddeti N G, Dunn M L and Bunch J S 2011 [1637] Buron J D et al 2014 Electrically continuous graphene from Ultrastrong adhesion of graphene membranes Nat. single crystal copper verified by terahertz conductance Nanotechnol. 6 543–6 spectroscopy and micro four-point probe Nano Lett. [1661] Hencky H 1921 Die Berechnung dünner rechteckiger Platten 14 6348–55 mit verschwindender Biegungsteifigkeit ZAMM—Z. Angew. [1638] Buron J D, Mackenzie D M A, Petersen D H, Pesquera A, Math. Mech. 1 81–9 Centeno A, Boggild P, Zurutuza A and Jepsen P U 2015 [1662] Wan K-T and Mai Y-W 1995 Fracture mechanics of a new Terahertz wafer-scale mobility mapping of graphene blister test with stable crack growth Acta Metall. Mater. on insulating substrates without a gate Opt. Express 43 4109–15 23 30721–9 [1663] Yang R, Lee J, Ghosh S, Tang H, Sankaran R M, Zorman C A [1639] Buron J D, Pizzocchero F, Jepsen P U, Petersen D H, and Feng P X L 2017 Tuning optical signatures of single- Caridad J M, Jessen B S, Booth T J and Boggild P 2015 and few-layer MoS2 by blown-bubble bulge straining up to Graphene mobility mapping Sci. Rep. 5 12305 fracture Nano Lett. 17 4568–75 [1640] Mackenzie D M A, Buron J D, Whelan P R, Jessen B S, [1664] Haigh S J, Gholinia A, Jalil R, Romani S, Britnell L, Silajdzic A, Pesquera A, Centeno A, Zurutuza A, Boggild P Elias D C, Novoselov K S, Ponomarenko L A, Geim A K and Petersen D H 2015 Fabrication of CVD graphene-based and Gorbachev R 2012 Cross-sectional imaging of devices via laser ablation for wafer-scale characterization 2D individual layers and buried interfaces of graphene-based Mater. 2 045003 heterostructures and superlattices Nat. Mater. 11 764–7 [1641] Bøggild P, Mackenzie D M A, Whelan P R, Petersen D H, [1665] Khestanova E, Guinea F, Fumagalli L, Geim A K and Buron J D, Zurutuza A, Gallop J, Hao L and Jepsen P U 2017 Grigorieva I V 2016 Universal shape and pressure inside Mapping the electrical properties of large-area graphene 2D bubbles appearing in van der Waals heterostructures Nat. Mater. 4 042003 Commun. 7 12587 [1642] Mackenzie D M A, Whelan P R, Boggild P, Jepsen P U, Redo- [1666] Courtney T H 2005 Mechanical Behavior of Materials (Long Sanchez A, Etayo D, Fabricius N and Petersen D H 2018 Grove, IL: Waveland Press) ch 9, pp 433–6 Quality assessment of terahertz time-domain spectroscopy [1667] Zhang P et al 2014 Fracture toughness of graphene Nat. transmission and reflection modes for graphene Commun. 5 5059 conductivity mapping Opt. Express 26 9220–9 [1668] Hwangbo Y et al 2014 Fracture characteristics of monolayer [1643] Whelan P et al 2018 Conductivity mapping of graphene on CVD-graphene Sci. Rep. 4 4439 polymeric films by terahertz time-domain spectroscopy Opt. [1669] López-Polín G, Gómez-Herrero J and Gómez-Navarro C Express 26 17748 2015 Confining crack propagation in defective graphene [1644] Horng J et al 2011 Drude conductivity of Dirac fermions in Nano Lett. 15 2050–4 graphene Phys. Rev. B 83 165113 [1670] Griffith A A 1921 The phenomena of rupture and flow in [1645] Sensale-Rodriguez B, Yan R S, Liu L, Jena D and Xing H G solids Phil. Trans. R. Soc. A 221 163–98 2013 Graphene for reconfigurable terahertz optoelectronics [1671] Jang B, Mag-isa A E, Kim J-H, Kim B, Lee H-J, Oh C-S, Proc. IEEE 101 1705–16 Sumigawa T and Kitamura T 2016 Uniaxial fracture test of [1646] Akinwande D et al 2017 A review on mechanics and freestanding pristine graphene using in situ tensile tester mechanical properties of 2D materials—graphene and under scanning electron microscope Extreme Mech. Lett. beyond Extreme Mech. Lett. 13 42–77 14 10–5 [1647] Geim A K 2009 Graphene: status and prospects Science [1672] Timoshenko S P, Gere J M and Prager W 1962 Theory of 324 1530–4 elastic stability, second edition J. Appl. Mech. 29 220 [1648] Amorim B et al 2016 Novel effects of strains in graphene and [1673] Berinskii I E, Krivtsov A M and Kudarova A M 2014 Bending other two dimensional materials Phys. Rep. 617 1–54 stiffness of a graphene sheet Phys. Mesomech. 17 356–64 279 2D Mater. 7 (2020) 022001 C Backes et al [1674] Sfyris D, Sfyris G I and Galiotis C 2014 Curvature dependent [1696] Frank O et al 2011 Raman 2D-band splitting in graphene: surface energy for a free standing monolayer graphene: some theory and experiment ACS Nano 5 2231–9 closed form solutions of the non-linear theory Int. J. Non- [1697] Mohr M, Papagelis K, Maultzsch J and Thomsen C 2009 Linear Mech. 67 186–97 Two-dimensional electronic and vibrational band structure [1675] Wei Y, Wang B, Wu J, Yang R and Dunn M L 2013 Bending of uniaxially strained graphene from ab initio calculations rigidity and gaussian bending stiffness of single-layered Phys. Rev. B 80 205410–4 graphene Nano Lett. 13 26–30 [1698] Li Z, Kinloch I A, Young R J, Novoselov K S, [1676] Tapasztó L, Dumitrica T, Kim S J, Nemes-Incze P, Hwang C Anagnostopoulos G, Parthenios J, Galiotis C, Papagelis K, and Biró L P 2012 Breakdown of continuum mechanics Lu C-Y and Britnell L 2015 Deformation of Wrinkled for nanometre-wavelength rippling of graphene Nat. Phys. graphene ACS Nano 9 3917–25 8 739–42 [1699] Xu C, Xue T, Guo J, Qin Q, Wu S, Song H and Xie H 2015 An [1677] Poot M and van der Zant H S J 2008 Nanomechanical experimental investigation on the mechanical properties properties of few-layer graphene membranes Appl. Phys. of the interface between large-sized graphene and a flexible Lett. 92 063111 substrate J. Appl. Phys. 117 164301 [1678] Lindahl N, Midtvedt D, Svensson J, Nerushev O A, [1700] Jiang T, Huang R and Zhu Y 2013 Interfacial sliding and Lindvall N, Isacsson A and Campbell E E B 2012 buckling of monolayer graphene on a stretchable substrate Determination of the bending rigidity of graphene via Adv. Funct. Mater. 24 396–402 electrostatic actuation of buckled membranes Nano Lett. [1701] Shin Y, Lozada-Hidalgo M, Sambricio J L, Grigorieva I V, 12 3526–31 Geim A K and Casiraghi C 2016 Raman spectroscopy of [1679] Zhao J, Deng Q, Ly T H, Han G H, Sandeep G and highly pressurized graphene membranes Appl. Phys. Lett. Rümmeli M H 2015 Two-dimensional membrane as elastic 108 221907 shell with proof on the folds revealed by three-dimensional [1702] Lee J-U, Yoon D and Cheong H 2012 Estimation of Young’s atomic mapping Nat. Commun. 6 8935 modulus of graphene by Raman spectroscopy Nano Lett. [1680] Blees M K et al Graphene kirigami Nature 524 204–7 12 4444–8 [1681] Nicklow R, Wakabayashi N and Smith H G 1972 Lattice [1703] Papagelis K T G, Parthenios J, Novoselov K and Galiotis C dynamics of pyrolytic graphite Phys. Rev. B 5 4951–62 2012 Axial deformation of monolayer graphene under [1682] Ferralis N 2010 Probing mechanical properties of graphene tension and compression GraphITA 2011 ed L Ottaviano and with Raman spectroscopy J. Mater. Sci. 45 5135–49 V Morandi (Berlin: Springer) pp 87–97 [1683] Bissett M A, Tsuji M and Ago H 2014 Strain engineering the [1704] Metten D, Federspiel F, Romeo M and Berciaud S 2014 properties of graphene and other two-dimensional crystals All-optical blister test of suspended graphene using micro- Phys. Chem. Chem. Phys. 16 11124 Raman spectroscopy Phys. Rev. Appl. 2 054008 [1684] Roldán R, Castellanos-Gomez A, Cappelluti E and [1705] Yoon D, Moon H, Son Y-W, Choi J S, Park B H, Cha Y H, Guinea F 2015 Strain engineering in semiconducting two- Kim Y D and Cheong H 2009 Interference effect on Raman dimensional crystals J. Phys.: Condens. Matter 27 313201 spectrum of graphene onSiO2/Si Phys. Rev. B 80 125422 [1685] Si C, Sun Z and Liu F 2016 Strain engineering of graphene: a [1706] Nicholl R J T, Conley H J, Lavrik N V, Vlassiouk I, review Nanoscale 8 3207–17 Puzyrev Y S, Sreenivas V P, Pantelides S T and Bolotin K I [1686] Gong L, Young R J, Kinloch I A, Riaz I, Jalil R and 2015 The effect of intrinsic crumpling on the mechanics of Novoselov K S 2012 Optimizing the reinforcement of free-standing graphene Nat. Commun. 6 8789 polymer-based nanocomposites by graphene ACS Nano [1707] Kitt A L, Qi Z, Rémi S, Park H S, Swan A K and Goldberg B B 6 2086–95 2013 How graphene slides: measurement and theory of [1687] Gong L, Kinloch I A, Young R J, Riaz I, Jalil R and strain-dependent frictional forces between graphene and Novoselov K S 2010 Interfacial stress transfer in a graphene SiO2 Nano Lett. 13 2605–10 monolayer nanocomposite Adv. Mater. 22 2694–7 [1708] Bhushan B, Israelachvili J N and Landman U 1995 [1688] Tsoukleri G, Parthenios J, Papagelis K, Jalil R, Ferrari A C, Nanotribology: friction, wear and lubrication at the atomic Geim A K, Novoselov K S and Galiotis C 2009 Subjecting scale Nature 374 607–16 a graphene monolayer to tension and compression Small [1709] Gao J, Luedtke W D, Gourdon D, Ruths M, Israelachvili J N 5 2397–402 and Landman U 2004 Frictional forces and Amontons’ law: [1689] Frank O, Tsoukleri G, Parthenios J, Papagelis K, Riaz I, Jalil R, from the molecular to the macroscopic scale J. Phys. Chem. B Novoselov K S and Galiotis C 2010 Compression behavior of 108 3410–25 single-layer graphenes ACS Nano 4 3131–8 [1710] Filintoglou K, Papadopoulos N, Arvanitidis J, Christofilos D, [1690] Frank O, Bouša M, Riaz I, Jalil R, Novoselov K S, Tsoukleri G, Frank O, Kalbac M, Parthenios J, Kalosakas G, Galiotis C Parthenios J, Kavan L, Papagelis K and Galiotis C 2012 and Papagelis K 2013 Raman spectroscopy of graphene Phonon and structural changes in deformed bernal stacked at high pressure: effects of the substrate and the pressure bilayer graphene Nano Lett. 12 687–93 transmitting media Phys. Rev. B 88 045418 [1691] Androulidakis C, Koukaras E N, Frank O, Tsoukleri G, [1711] Zandiatashbar A, Lee G-H, An S J, Lee S, Mathew N, Sfyris D, Parthenios J, Pugno N, Papagelis K, Novoselov K S Terrones M, Hayashi T, Picu C R, HSone J and Koratkar N and Galiotis C 2014 Failure processes in embedded 2014 Effect of defects on the intrinsic strength and stiffness monolayer graphene under axial compression Sci. Rep. 4 5271 of graphene Nat. Commun. 5 3186 [1692] Androulidakis C, Koukaras E N, Parthenios J, Kalosakas G, [1712] Son Y W et al 2006 Energy gaps in graphene nanoribbons Papagelis K and Galiotis C 2015 Graphene flakes under Phys. Rev. Lett. 97 216803 controlled biaxial deformation Sci. Rep. 5 18129 [1713] Cervantes-Sodi F et al 2008 Two-dimensional bipolar [1693] Rice C, Young R J, Zan R, Bangert U, Wolverson D, junction transistors Phys. Rev. B 77 165427 Georgiou T, Jalil R and Novoselov K S 2013 Raman- [1714] Overbeck J et al 2019 Optimized substrates and scattering measurements and first-principles calculations of measurement approaches for Raman spectroscopy of strain-induced phonon shifts in monolayer MoS2 Phys. Rev. graphene nanoribbons (arXiv:1907.01797) B 87 121304 [1715] Verzhbitskiy I et al 2016 Raman fingerprints of atomically [1694] Anagnostopoulos G, Androulidakis C, Koukaras E N, precise graphene nanoribbons Nano Lett. 16 3442 Tsoukleri G, Polyzos I, Parthenios J, Papagelis K and [1716] Guidetti G et al 2019 Photocatalytic activity of exfoliated Galiotis C 2015 Stress transfer mechanisms at the submicron graphite-TiO2 nanocomposites (arXiv:1903.05407) level for graphene/polymer systems ACS Appl. Mater. [1717] Fabbro A et al 2016 Graphene-based interfaces do not alter Interfaces 7 4216–23 target nerve cells ACS Nano 10 615 [1695] Papageorgiou D G, Kinloch I A and Young R J 2017 [1718] Romagnoli M et al 2018 Graphene-based integrated Mechanical properties of graphene and graphene-based photonics for next-generation datacom and telecom Nat. nanocomposites Prog. Mater. Sci. 90 75–127 Rev. Mater. 3 392 280 2D Mater. 7 (2020) 022001 C Backes et al [1719] Purdie D G et al 2018 Cleaning interfaces in layered materials [1741] Aguiar-Ricardo A, Duarte C M M, Nunes da Ponte M, heterostructures Nat. Commun. 9 5387 Marsh K N and Stølen S 2003 Properties of mixing [1720] De Fazio D et al 2019 High-mobility, wet-transferred Experimental Thermodynamics ed A R H Goodwin et al (New graphene grown by chemical vapor deposition ACS Nano. York: Elsevier) vol 6, pp 387–432 13 8926 [1742] Meng Y, Xu X-B, Li H, Wang Y, Ding E-X, Zhang Z-C and [1721] Mertens J et al 2013 Controlling subnanometer gaps in Geng H-Z 2014 Optimisation of carbon nanotube ink plasmonic dimers using graphene Nano Lett. 13 5033 for large-area transparent conducting films fabricated by [1722] Ferrari A C and Robertson J 2001 Resonant Raman controllable rod-coating method Carbon 70 103–10 spectroscopy of disordered, amorphous, and diamondlike [1743] Worsley K A, Kalinina I, Bekyarova E and Haddon R C carbon Phys. Rev. B 64 075414 2009 Functionalization and dissolution of nitric acid [1723] Frank O et al 2011 Development of a universal stress sensor treated single-walled carbon nanotubes J. Am. Chem. Soc. for graphene and carbon fibres Nat. Commun. 2 255 131 18153–8 [1724] Corso M et al 2018 Bottom-up fabrication of atomically [1744] Thayer A 2000 What’s that stuff? Silly putty Chem. Eng. News precise graphene nanoribbons On-Surface Synthesis II 78 27 (Cham: Springer) pp 113–52 [1745] Kimoto T, Itoh A and Matsunami H 1995 Step bunching in [1725] Barin G et al 2019 Surface-synthesized graphene chemical vapor deposition of 6H- and 4H-SiC on vicinal nanoribbons for room temperature switching devices: SiC(0 0 0 1) faces Appl. Phys. Lett. 66 3645 substrate transfer and ex situ characterization ACS App. [1746] Ciuk T et al 2018 Thermally activated double-carrier Nanomater. 2 2184–92 transport in epitaxial graphene on vanadium-compensated [1726] Corso M, Carbonell-Sanromà E and de Oteyza D G 2018 6H-SiC as revealed by Hall effect measurements Carbon Bottom-up fabrication of atomically precise graphene 139 776–81 nanoribbons On-Surface Synthesis II. Advances in Atom [1747] Ciuk T, Stanczyk B, Przyborowska K, Czolak D, and Single Molecule Machines ed D de Oteyza and C Rogero Dobrowolski A, Jagiello J, Kaszub W, Kozubal M, (Cham: Springer) Kozlowski R and Kaminski P 2019 High-temperature [1727] Durr R A et al 2018 J. Am. Chem. Soc. 140 807–13 Hall effect sensor based on epitaxial graphene on high- [1728] León V, González-Domínguez J M, Fierro J L G, Prato M purity semiinsulating 4H-SiC IEEE Trans. Electron Devices and Vázquez E 2016 Production and stability of 66 3134–8 mechanochemically exfoliated graphene in water and [1748] Ishihara M, Koga Y, Kim J, Tsugawa K and Hasegawa M culture media Nanoscale 8 14548–55 2011 Direct evidence of advantage of Cu(1 1 1) for graphene [1729] González-Domínguez J M, León V, Lucío M I, Prato M and synthesis by using Raman mapping and electron backscatter Vázquez E 2018 Nat. Protocols 13 495–506 diffraction Mater. Lett. 65 2864–7 [1730] Wang H et al 2019 Superhydrophobic and superoleophilic [1749] Murdock A T, Koos A, Ben Britton T, Houben L, Batten T, graphene aerogel for adsorption of oil pollutants from water Zhang T, Wilkinson A J, Dunin-Borkowski R E, Lekka C E RSC Adv. 9 8569 and Grobert N 2013 Controlling the orientation, edge [1731] Lyu J et al 2018 Separation and purification using GO and geometry, and thickness of chemical vapor deposition r-GO membranes RSC Adv. 8 23130 graphene ACS Nano 7 1351–9 [1732] Kagkoura A, Pelaez-Fernandez M, Arenal R and [1750] Mullins W W and Sekerka R F 1963 Morphological stability Tagmatarchis N 2019 Sulfur-doped graphene/transition of a particle growing by diffusion or heat flow J. Appl. Phys. metal dichalcogenide heterostructured hybrids with 34 323–9 electrocatalytic activity toward the hydrogen evolution [1751] Xia K L, Artyukhov V I, Sun L F, Zheng J Y, Jiao L Y, reaction Nanoscale Adv. 1 1489–96 Yakobson B I and Zhang Y Y 2016 Growth of large-area [1733] Istif E, Kagkoura A, Hernández-Ferrer J, Stergiou A, aligned pentagonal graphene domains on high-index copper Skaltsas T, Arenal R, Benito A M, Maser W K and surfaces Nano Res. 9 2182–9 Tagmatarchis N 2018 Self-assembled core–shell CdTe/ [1752] Bloom D S, Putman J W and Grant N J 1952 Melting point poly(3-hexylthiophene) nanoensembles as novel donor– and transformation of pure chromium JOM 4 626 acceptor light harvesting systems ACS Appl. Mater. Interfaces [1753] Fujita J I, Hiyama T, Hirukawa A, Kondo T, Nakamura J, 9 44695–703 Ito S I, Araki R, Ito Y, Takeguchi M and Pai W W 2017 Near [1734] Cooper A J, Wilson N R, Kinloch I A and Dryfe R A W room temperature chemical vapor deposition of graphene 2014 Single stage electrochemical exfoliation method for with diluted methane and molten gallium catalyst Sci. Rep. 7 the production of few-layer graphene via intercalation of [1754] Chaitoglou S, Giannakopoulou T, Tsoutsou D, tetraalkylammonium cations Carbon 66 340–50 Vavouliotis A, Trepans C and Dimoulas A 2019 Direct [1735] Xia Z, Arbizzani C, Ortolani L, Morandi V, Bellani V, versus reverse vertical two-dimensional Mo2C/graphene Giambastiani G, Gazzano M and Palermo V 2017 High heterostructures for enhanced hydrogen evolution reaction yield production of graphene-Fe2O3 nano-composites via electrocatalysis Nanotechnology 30 electrochemical intercalation of nitromethane and iron [1755] Huang M et al 2018 Highly oriented monolayer graphene chloride, and their application in lithium storage Flatchem grown on a Cu/Ni(1 1 1) alloy foil ACS Nano 12 6117–27 3 8–15 [1756] Ullah Z, Riaz S, Li Q, Atiq S, Saleem M, Azhar M, Naseem S [1736] Haastrup S et al 2018 2D Mater. 5 042002 and Liu L W 2018 A comparative study of graphene growth [1737] Cheon G, Duerloo K-A N, Sendek A D, Porter C, Chen Y by APCVD, LPCVD and PECVD Mater. Res. Express 5 035606 and Reed E J 2017 Data mining for new two- and [1757] Koren E, Sutter E, Bliznakov S, Ivars-Barcelo F and Sutter P one-dimensional weakly bonded solids and lattice- 2013 Isolation of high quality graphene from Ru by solution commensurate heterostructures Nano Lett. 17 1915–23 phase intercalation Appl. Phys. Lett. 103 121602 [1738] Choudhary K, Kalish I, Beams R and Tavazza F 2017 [1758] Sutter P, Albrecht P, Tong X and Sutter E 2013 Mechanical High-throughput identification and characterization of decoupling of graphene from Ru(0 0 0 1) by interfacial two-dimensional materials using density functional theory reaction with oxygen J. Phys. Chem. C 117 6320–4 Sci. Rep. 7 5179 [1759] Chandrashekar B N, Deng B, Smitha A S, Chen Y, Tan C, [1739] Zhou J et al 2019 2DMatPedia, an open computational Zhang H, Peng H and Liu Z 2015 Roll-to-roll green transfer database of two-dimensional materials from top-down and of CVD graphene onto plastic for a transparent and flexible bottom-up approaches Sci. Data 6 86 triboelectric nanogenerator Adv. Mater. 27 5210–6 [1740] Larsen P M, Pandey M, Strange M and Jacobsen K W [1760] Xu P, Kang J, Suhr J, Smith J P, Booksh K S , Wei B, Yu J, Li F, 2019 Definition of a scoring parameter to identify low- Byun J H, Oh Y and Chou T S 2015 Spatial strain variation dimensional materials components Phys. Rev. Mater. of graphene films for stretchable electrodes Carbon 3 034003 93 620–4 281 2D Mater. 7 (2020) 022001 C Backes et al [1761] Martins L G P, Song Y, Zeng T, Dresselhaus M S, Kong J, and [1780] Zhang H, Bekyarova E, Huang J-W, Zhao Z, Bao W, Araujo P T 2013 Direct transfer of graphene onto flexible Wang F, Haddon R C and Lau C N 2011 Nano Lett. substrates Proc. Natl Acad. Sci. 110 17762–7 11 4047–51 [1762] Hong J Y, Shin Y C, Zubair A, Mao Y, Palacios T, Dresselhaus [1781] Xu X, Liu C, Sun Z, Cao T, Zhang Z, Wang E, Liu Z and Liu K M S, Kim S H and Kong J 2016 A rational strategy for 2018 Interfacial engineering in graphene bandgap Chem. graphene transfer on substrates with rough features Adv. Soc. Rev. 47 3059–99 Mater. 28 2382–92 [1782] Sakintuna B and Yürüm Y 2005 Templated porous carbons: a [1763] Celebi K, Cole M T, Choi J W, Wyczisk F, Legagneux P, review article Ind. Eng. Chem. Res. 44 2893–902 Rupesinghe N, Robertson J, Teo K B K and Park H G 2013 [1783] Chatzi E G and Koenig J L 1987 Morphology and structure Evolutionary kinetics of graphene formation on copper of kevlar fibers: a review Polym.-Plast. Technol. Eng. Nano Lett. 13 967–74 26 229–70 [1764] Tschierske C and Ungar G 2016 Mirror symmetry breaking [1784] Stafströma S 1987 Electronic properties of polyaniline by chirality synchronisation in liquids and liquid crystals of Electronic Properties of Conjugated Polymers ed H Kuzmany achiral molecules ChemPhysChem 17 1–7 et al (Springer Series in Solid-State Sciences vol 76) (Berlin: [1765] Deng W, Zhang Y, Yang L, Tan Y, Ma M, and Xie Q 2015 Springer) Sulfur-doped porous carbon nanosheets as an advanced [1785] Kinloch I A, Suhr J, Lou J, Young R J and Ajayan P M 2018 electrode material for supercapacitors RSC Adv. 5 13046–51 Composites with carbon nanotubes and graphene: an [1766] Stolyarov M A, Liu G, Rumyantsev S L, Shur M and Balandin outlook Science 362 547–53 A A 2015 Suppression of 1/f noise in near-ballistic h-BN- [1786] Haider S, Khan Y, Almasry W A and Haider A 2012 graphene-h-BN heterostructure field-effect transistors Appl. Thermoplastic Nanocomposites and their Processing Phys. Lett. 107 023106 Techniques, Thermoplastic—Composite Materials (Adel Zaki [1767] Sen D, Novoselov K S, Reis P M and Buehle M J 2010 Tearing El-Sonbati: IntechOpen) graphene sheets from adhesive substrates produces tapered [1787] Pionteck J and Wypych G 2007 Handbook of Antistatics nanoribbons Small 6 1108–16 (ChemTec Publishing) [1768] Tian H, Shu Y, Cui Y L, Mi W T, Yang Y, Xie D and Ren T L [1788] Zeng Z et al 2012 An effective method for the fabrication of 2014 Scalable fabrication of high-performance and flexible few-layer-thick inorganic nanosheets Angew. Chem., Int. Ed. graphene strain sensors Nanoscale 6 699–705 Engl. 51 9052–6 [1769] Bunch J S and Dunn M L 2012 Adhesion mechanics of [1789] Keil S 2017 The Wheatstone bridge circuit Technology and graphene membranes Solid State Commun. 152 1359–64 Practical Use of Strain Gages (Wilhelm Ernst & Sohn Verlag [1770] Batrakov K, Kuzhir P, Maksimenko S, Paddubskaya A, für Architektur und technische Wissenschaften GmbH & Voronovich S, Lambin Ph, Kaplas T and Svirko Yu 2014 Co. KG) ch 4, pp 147–62 Flexible transparent graphene/polymer multilayers for [1790] Tortello M, Pasternak I, Zeranska-Chudek K, Strupinski W, efficient electromagnetic field absorption Sci. Rep. 4 7191 Gonnelli R S and Fina A 2019 Chemical-vapor-deposited [1771] Georgakilas V, Otyepka M, Bourlinos A B, Chandra V, Kim N, graphene as a thermally conducting coating ACS Appl. Nano Kemp K C, Hobza P, Zboril R and Kim K S 2012 Chem. Rev. Mater. 2 2621–33 112 6156–214 [1791] Schué L, Sponza L, Plaud A, Bensalah H, Watanabe K, [1772] Yan L, Zheng Y B, Zhao F, Li S, Gao X, Xu B, Weiss P S and Taniguchi T, Ducastelle F, Loiseau A and Barjon J 2019 Bright Zhao Y 2012 Chem. Soc. Rev. 41 97–114 luminescence from indirect and strongly bound excitons in [1773] Sun Z, James D K and Tour J M 2011 J. Phys. Chem. Lett. h-BN Phys. Rev. Lett. 122 067401 2 2425–32 [1792] Damascelli A 2004 Phys. Scr. T 109 61 [1774] Sun Z, Kohama S, Zhang Z, Lomeda J R and Tour J M 2010 [1793] Hüfner S 2003 Photoelectron Spectroscopy 3rd edn (Berlin: Soluble graphene through edge-selective functionalization Springer) Nano Res. 3 117–25 [1794] Haworth C A et al 2007 Nat. Phys. 3 52 [1775] Lim H, Lee J S, Shin H J, Shin H S and Choi H C 2010 [1795] Whelan P et al 2018 Electrical homogeneity mapping of Spatially resolved spontaneous reactivity of diazonium salt epitaxial graphene on silicon carbide ACS Appl. Mater. on edge and basal plane of graphene without surfactant and Interfaces 10 31641–7 its doping effect Langmuir 26 12278–84 [1796] Mackenzie D M A et al 2016 Contactless graphene [1776] Jiang D, Sumpter B G and Dai S 2006 J. Phys. Chem. B conductance measurements: the effect of device fabrication 110 23628–32 on terahertz time-domain spectroscopy Int. J. Nanotechnol. [1777] Bottari G, Herranz M Á, Wibmer L, Volland M, Rodríguez- 13 591–6 Pérez L, Guldi D M, Hirsch A, Martín N, D’Souza F and [1797] Whelan P et al 2019 Non-contact mobility measurements of Torres T 2017 Chemical functionalization and characterization graphene on silicon carbide Microelectron. Eng. 212 9–12 of graphene-based materials Chem. Soc. Rev. 46 4464–500 [1798] Lin H et al 2017 Contactless graphene conductivity mapping [1778] Kaplan A, Yuan Z, Benck J D, Govind Rajan A, Chu X S, on a wide range of substrates with terahertz time-domain Wang Q H and Strano M S 2017 Current and future reflection spectroscopy Sci. Rep. 7 10625 directions in electron transfer chemistry of graphene Chem. [1799] Halbig C, Martin O, Hauke F, Hirsch A and Eigler S 2018 Soc. Rev. 46 4530–71 Oxo‐Functionalized Graphene: A Versatile Precursor for [1779] Malig J, Englert J M, Hirsch A and Guldi D M 2011 Alkylated Graphene Sheets by Reductive Functionalization Electrochem. Soc. Interface 20 53–6 Chem. Eur. J. 24 13348–54 282