15214095, 2022, 27, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adma.202201502 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [14/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Perspective www.advmat.de How Reproducible are Surface Areas Calculated from the BET Equation? Johannes W. M. Osterrieth, James Rampersad, David Madden, Nakul Rampal, Luka Skoric, Bethany Connolly, Mark D. Allendorf, Vitalie Stavila, Jonathan L. Snider, Rob Ameloot, João Marreiros, Conchi Ania, Diana Azevedo, Enrique Vilarrasa-Garcia, Bianca F. Santos, Xian-He Bu, Ze Chang, Hana Bunzen, Neil R. Champness, Sarah L. Griffin, Banglin Chen, Rui-Biao Lin, Benoit Coasne, Seth Cohen, Jessica C. Moreton, Yamil J. Colón, Linjiang Chen, Rob Clowes, François-Xavier Coudert, Yong Cui, Bang Hou, Deanna M. D’Alessandro, Patrick W. Doheny, Mircea Dincă, Chenyue Sun, Christian Doonan, Michael Thomas Huxley, Jack D. Evans, Paolo Falcaro, Raffaele Ricco, Omar Farha, Karam B. Idrees, Timur Islamoglu, Pingyun Feng, Huajun Yang, Ross S. Forgan, Dominic Bara, Shuhei Furukawa, Eli Sanchez, Jorge Gascon, Selvedin Telalović, Sujit K. Ghosh, Soumya Mukherjee, Matthew R. Hill, Muhammed Munir Sadiq, Patricia Horcajada, Pablo Salcedo-Abraira, Katsumi Kaneko, Radovan Kukobat, Jeff Kenvin, Seda Keskin, Susumu Kitagawa, Ken-ichi Otake, Ryan P. Lively, Stephen J. A. DeWitt, Phillip Llewellyn, Bettina V. Lotsch, Sebastian T. Emmerling, Alexander M. Pütz, Carlos Martí-Gastaldo, Natalia M. Padial, Javier García-Martínez, Noemi Linares, Daniel Maspoch, Jose A. Suárez del Pino, Peyman Moghadam, Rama Oktavian, Russel E. Morris, Paul S. Wheatley, Jorge Navarro, Camille Petit, David Danaci, Matthew J. Rosseinsky, Alexandros P. Katsoulidis, Martin Schröder, Xue Han, Sihai Yang, Christian Serre, Georges Mouchaham, David S. Sholl, Raghuram Thyagarajan, Daniel Siderius, Randall Q. Snurr, Rebecca B. Goncalves, Shane Telfer, Seok J. Lee, Valeska P. Ting, Jemma L. Rowlandson, Takashi Uemura, Tomoya Iiyuka, Monique A. van derVeen, Davide Rega, Veronique Van Speybroeck, Sven M. J. Rogge, Aran Lamaire, Krista S. Walton, Lukas W. Bingel, Stefan Wuttke, Jacopo Andreo, Omar Yaghi, Bing Zhang, Cafer T. Yavuz, Thien S. Nguyen, Felix Zamora, Carmen Montoro, Hongcai Zhou, Angelo Kirchon, and David Fairen-Jimenez* Dedicated to the memory of Francisco Rodriguez-Reinoso and his unique work in adsorption and porous materials The ORCID identification number(s) for the author(s) of this article J. W. M. Osterrieth, J. Rampersad, D. Madden, N. Rampal, B. Connolly, can be found under https://doi.org/10.1002/adma.202201502. D. Fairen-Jimenez The Adsorption & Advanced Materials Laboratory (A2ML) © 2022 The Authors. Advanced Materials published by Wiley-VCH Department of Chemical Engineering & Biotechnology GmbH. This is an open access article under the terms of the Creative University of Cambridge Commons Attribution License, which permits use, distribution and re- Philippa Fawcett Drive, Cambridge CB3 0AS, UK production in any medium, provided the original work is properly cited. E-mail: df334@cam.ac.uk DOI: 10.1002/adma.202201502 Adv. Mater. 2022, 34, 2201502 2201502 (1 of 12) © 2022 The Authors. Advanced Materials published by Wiley-VCH GmbH 15214095, 2022, 27, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adma.202201502 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [14/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advmat.de L. Skoric D. M. D’Alessandro, P. W. Doheny Cavendish Laboratory School of Chemistry University of Cambridge The University of Sydney JJ Thomson Avenue, Cambridge CB3 0HE, UK New South Wales 2006, Australia M. D. Allendorf, V. Stavila, J. L. Snider M. Dincă, C. Sun Sandia National Laboratories Department of Chemistry 7011 East Avenue, Livermore, CA 94550, USA Massachusetts Institute of Technology R. Ameloot, J. Marreiros Cambridge, MA 02139, USA cMACS C. Doonan, M. T. Huxley Department of Microbial and Molecular Systems (M2S) Centre for Advanced Nanomaterials and Department of Chemistry KU Leuven The University of Adelaide Leuven 3001, Belgium North Terrace, Adelaide SA 5000, Australia C. Ania J. D. Evans CEMHTI Department of Inorganic Chemistry CNRS (UPR 3079) Technische Universität Dresden Université d’Orléans Bergstrasse 66, 01062 Dresden, Germany Orléans 45071, France P. Falcaro, R. Ricco D. Azevedo, E. Vilarrasa-Garcia, B. F. Santos Institute of Physical and Theoretical Chemistry LPACO2/GPSA Graz University of Technology Department of Chemical Engineering Graz 8010, Austria Federal University of Ceará O. Farha, K. B. Idrees, T. Islamoglu Fortaleza (CE) 60455-760, Brazil Department of Chemistry and International Institute of Nanotechnology X.-H. Bu, Z. Chang Northwestern University School of Materials Science and Engineering 2145 Sheridan Road, Evanston, IL 60208, USA National Institute for Advanced Materials P. Feng, H. Yang Nankai University Department of Chemistry Tianjin 300350, China University of California H. Bunzen Riverside, CA 92521, USA Chair of Solid State and Materials Chemistry R. S. Forgan, D. Bara Institute of Physics WestCHEM, School of Chemistry University of Augsburg University of Glasgow Universitaetsstrasse 1, 86159 Augsburg, Germany Glasgow G12 8QQ, UK N. R. Champness, S. L. Griffin S. Furukawa, E. Sanchez School of Chemistry Institute for Integrated Cell-Material Sciences University of Nottingham Kyoto University University Park, Nottingham NG7 2RD, UK Yoshida, Sakyo-ku, Kyoto 606-8501, Japan B. Chen, R.-B. Lin J. Gascon, S. Telalović Department of Chemistry KAUST Catalysis Center (KCC) University of Texas at San Antonio King Abdullah University of Science and Technology One UTSA Circle, San Antonio, TX 78249-0698, USA P.O. Box 4700, Thuwal-Jeddah 23955-6900, Kingdom of Saudi Arabia B. Coasne S. K. Ghosh, S. Mukherjee Univ. Grenoble Alpes Department of Chemistry CNRS Indian Institute of Science Education and Research (IISER) Pune LIPhy Dr. Homi Bhabha Road, Pashan, Pune 411008, India Grenoble 38000, France M. R. Hill, M. M. Sadiq S. Cohen, J. C. Moreton CSIRO Department of Chemistry and Biochemistry Private Bag 33, Clayton South MDC, Clayton, VIC 3169, Australia University of California M. R. Hill, M. M. Sadiq San Diego, La Jolla, CA 92093, USA Department of Chemical Engineering Y. J. Colón Monash University Department of Chemical and Biomolecular Engineering Clayton, VIC 3168, Australia University of Notre Dame P. Horcajada, P. Salcedo-Abraira Notre Dame, IN 46556, USA Advanced Porous Materials Unit (APMU) L. Chen, R. Clowes IMDEA Energy Leverhulme Research Centre for Functional Materials Design Avda. Ramón de la Sagra 3, (Móstoles) Madrid E-28935, Spain Materials Innovation Factory and Department of Chemistry K. Kaneko, R. Kukobat University of Liverpool Research Initiative for Supra-Materials Liverpool L7 3NY, UK Shinshu University F.-X. Coudert Nagano 380-8553, Japan Chimie ParisTech J. Kenvin PSL University Micromeritics Instrument Corporation CNRS Norcross, GA 30093, USA Institut de Recherche de Chimie Paris Paris 75005, France S. Keskin Department of Chemical and Biological Engineering Y. Cui, B. Hou Koc University School of Chemistry and Chemical Engineering Rumelifeneri Yolu Sariyer, Istanbul 34450, Turkey Shanghai Jiaotong University 800 Dongchuan Road, Minhang District, Shanghai 200240, China Adv. Mater. 2022, 34, 2201502 2201502 (2 of 12) © 2022 The Authors. Advanced Materials published by Wiley-VCH GmbH 15214095, 2022, 27, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adma.202201502 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [14/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advmat.de S. Kitagawa, K.-i. Otake D. Siderius Institute for Integrated Cell-Material Sciences (WPI-iCeMS) Chemical Sciences Division Kyoto University Institute for Advanced Study (KUIAS) National Institute of Standards and Technology Kyoto University Gaithersburg, MD 20899-8320, USA Yoshida Ushinomiya-cho, Sakyo-ku, Kyoto 606-8501, Japan R. Q. Snurr R. P. Lively, S. J. A. DeWitt, D. S. Sholl, R. Thyagarajan, K. S. Walton, Department of Chemical & Biological Engineering L. W. Bingel Northwestern University School of Chemical & Biomolecular Engineering 2145 Sheridan Road, Evanston, IL 60208, USA Georgia Institute of Technology R. B. Goncalves Atlanta, GA 30332, USA Department of Chemistry P. Llewellyn Northwestern University CNRS / Aix-Marseille Univ./Total 2145 Sheridan Road, Evanston, IL 60208, USA Marseille 64018, France S. Telfer, S. J. Lee B. V. Lotsch, S. T. Emmerling, A. M. Pütz MacDiarmid Institute for Advanced Materials and Nanotechnology Max Planck Institute for Solid State Research Institute of Fundamental Sciences Heisenbergstrasse 1, 70569 Stuttgart, Germany Massey University B. V. Lotsch, S. T. Emmerling, A. M. Pütz Palmerston North 4442, New Zealand Department of Chemistry V. P. Ting, J. L. Rowlandson University of Munich (LMU) Department of Mechanical Engineering Butenandtstrasse 5-13, 81377 Munich, Germany University of Bristol C. Martí-Gastaldo, N. M. Padial Bristol BS8 1TR, UK Instituto de Ciencia Molecular (ICMol) T. Uemura, T. Iiyuka Universitat de València Department of Advanced Materials Science Paterna, València 46980, Spain Graduate School of Frontier Sciences J. García-Martínez, N. Linares The University of Tokyo Laboratorio de Nanotecnología Molecular 5-1-5 Kashiwanoha, Kashiwa, Chiba 277-8561, Japan Departamento de Química Inorgánica M. A. van der Veen, D. Rega Universidad de Alicante Department of Chemical Engineering Ctra. San Vicente-Alicante s/n, San Vicente del Raspeig E-03690, Spain Delft University of Technology D. Maspoch van der Maasweg 9, Delft 2629HZ, The Netherlands ICREA V. Van Speybroeck, S. M. J. Rogge, A. Lamaire Pg. Lluís Companys 23, Barcelona 08010, Spain Center for Molecular Modeling (CMM) D. Maspoch, J. A. Suárez del Pino Ghent University Catalan Institute of Nanoscience and Nanotechnology (ICN2) Technologiepark 46, Zwijnaarde B-9052, Belgium CSIC and the Barcelona Institute of Science and Technology S. Wuttke, J. Andreo Campus UAB, Bellaterra, Barcelona 08193, Spain BCMaterials P. Moghadam, R. Oktavian Basque Center for Materials Department of Chemical and Biological Engineering Applications and Nanostructures The University of Sheffield UPV/EHU Science Park, Leioa 48940, Spain Sheffield S10 2TN, UK S. Wuttke, J. Andreo R. E. Morris, P. S. Wheatley IKERBASQUE School of Chemistry Basque Foundation for Science University of St Andrews Bilbao 48009, Spain North Haugh, St Andrews KY16 9ST, UK O. Yaghi, B. Zhang J. Navarro Department of Chemistry Departamento de Química Inorgánica University of California – Berkeley Universidad de Granada Kavli Energy Nanoscience Institute at UC Berkeley Granada 18071, Spain Berkeley, CA 94720, USA C. Petit, D. Danaci O. Yaghi Barrer Centre Berkeley Global Science Institute Department of Chemical Engineering Berkeley, CA 94720, USA Imperial College London C. T. Yavuz, T. S. Nguyen London SW7 2AZ, UK Department of Chemical and Biomolecular Engineering M. J. Rosseinsky, A. P. Katsoulidis Korea Advanced Institute of Science and Technology (KAIST) Materials Innovation Factory Yuseong-gu, Daejeon 34141, South Korea Department of Chemistry F. Zamora, C. Montoro University of Liverpool Departamento de Química Inorgánica Liverpool L7 3NY, UK Universidad Autónoma de Madrid M. Schröder, X. Han, S. Yang Madrid 28049, Spain School of Chemistry H. Zhou, A. Kirchon The University of Manchester Department of Chemistry Manchester M13 9PL, UK Texas A&M University C. Serre, G. Mouchaham College Station, TX 77843, USA Institut des Matériaux Poreux de Paris Ecole Normale Supérieure ESPCI Paris CNRS PSL University Paris 75005, France Adv. Mater. 2022, 34, 2201502 2201502 (3 of 12) © 2022 The Authors. Advanced Materials published by Wiley-VCH GmbH 15214095, 2022, 27, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adma.202201502 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [14/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advmat.de IUPAC warns users to apply “extreme Porosity and surface area analysis play a prominent role in modern materials caution [when using the BET equation] science. At the heart of this sits the Brunauer–Emmett–Teller (BET) theory, in the presence of micropores. (…) [The which has been a remarkably successful contribution to the field of materials BET area] represents an apparent surface science. The BET method was developed in the 1930s for open surfaces but is area, which may be regarded as a useful adsorbent ‘fingerprint’.”[19] This more now the most widely used metric for the estimation of surface areas of micro- nuanced understanding of the BET area and mesoporous materials. Despite its widespread use, the calculation of is mirrored in the writing of Rouquerol BET surface areas causes a spread in reported areas, resulting in reproduci- et al., “the meaning of the BET surface is bility problems in both academia and industry. To prove this, for this analysis, (…) that it embraces the major part of the 18 already-measured raw adsorption isotherms were provided to sixty-one amount of adsorptive in energetic inter- action with the surface.”[22] Despite these labs, who were asked to calculate the corresponding BET areas. This round- cautionary words, the BET area remains robin exercise resulted in a wide range of values. Here, the reproducibility a deeply engrained metric in the fields of of BET area determination from identical isotherms is demonstrated to be a physical chemistry and materials science. largely ignored issue, raising critical concerns over the reliability of reported Given the broad use of the BET equation, BET areas. To solve this major issue, a new computational approach to accu- it is not surprising to see that much has rately and systematically determine the BET area of nanoporous materials is been written on the applicability and the accuracy of the BET theory—that is, its developed. The software, called “BET surface identification” (BETSI), expands model of the adsorption process—and on on the well-known Rouquerol criteria and makes an unambiguous BET area the reproducibility of the raw data, i.e., the assignment possible. adsorption isotherm.[23–26] Since the development of the first porous materials, there has been a sharp 1. Main rise in the design of highly ordered and structured porous materials (Figure S1, Supporting Information).[27,28] The advent The Brunauer–Emmett–Teller (BET) equation is arguably one of materials with more complex pore networks and dynamic of the most used equations in physical chemistry and porosim- frameworks through material design strategies such as retic- etry. Since its conception in the 1930s[1] to estimate open sur- ular chemistry has given rise to reported BET areas in excess faces whilst working with non-microporous adsorbents of the of 8000 m2 g−1.[10,29–34] Often, these modern materials have time such as Fe/Cu catalysts, silica gel, and charcoal, it found complex adsorption isotherms which are more problematic or widespread use in the characterization of synthetic zeolites.[2] ambiguous to fit to the BET model, e.g. several steps can occur Furthermore, it has gained considerable momentum fol- due to different pore types and/or flexibility being present in lowing the discovery of more complex porous materials such as the material.[35] In response, a new generation of porosimetry mesoporous silicas,[3–5] porous coordination polymers (PCPs),[6] equipment with pressure transducers capable of recording metal–organic frameworks (MOFs)[7] and covalent organic high-resolution gas adsorption isotherms at ultralow pres- frameworks (COFs).[8] Novel porous materials are of significant sures (<10−5 mmHg) has been developed. However, reliance academic and industrial interest due to their applications in gas on manual calculations of surface areas using the BET method storage and separation,[9–12] catalysis,[13] sensing,[14,15] and drug remains commonplace. In this context, “manual” refers to delivery.[16] To assess their adsorptive properties, Langmuir was the judicious selection of a pressure range by a scientist, be it the first to relate gas adsorption isotherms to surface areas, through a self-developed spreadsheet or commercial software. assuming only monolayer adsorption.[17] This was in contrast to This raises the question of the reproducibility of BET calcula- Dubinin’s proposition of micropore volumes for microporous tions from the same isotherm. An adsorption isotherm with materials.[18] Langmuir’s adsorption theory was later extended 150 points has more than 10 000 consecutive combinations of to multilayer adsorption, resulting in the titular BET model. points, all of which are potentially correct fitting ranges and Even though the BET theory was not developed for describing will return different BET areas. The answer to the question of adsorption in the microporosity, the BET area is now the de which is the optimal fitting region is far from obvious, and the facto standard for the characterization of any porous material. consequence of any irreproducibility or different interpreta- Indeed, it has been recognized by the International Union of tions are serious. Consider two groups synthesizing the same Pure and Applied Chemistry (IUPAC) as “the most widely used compound and reporting two different BET areas; Sample A procedure for evaluating the surface area of porous and finely is reported to have a BET area of 1500 m2 g−1 and Sample B’s divided materials.”[19,20] Furthermore, it has been an Interna- reported BET area is 2000 m2 g−1. Unless there is a common tional Organization for Standardization (ISO) standard for standard and protocol for calculating BET areas, we cannot surface area determination since 1995.[21] This makes it, argu- say for certain that the quality and adsorption performance of ably, the most important figure of merit for porous materials, Sample A is lower than that of Sample B. Indeed, the lack of including microporous ones. Looking at the literature, it is reproducibility of MOF syntheses and adsorption performance, clear that the idea of monolayer coverage or even the concept by comparing reported BET areas, has been highlighted already, of surface area are necessarily idealized and therefore could be but the natural spread of BET calculations was not included in inaccurate descriptions for microporous materials.[22] Indeed, the analysis.[36] Adv. Mater. 2022, 34, 2201502 2201502 (4 of 12) © 2022 The Authors. Advanced Materials published by Wiley-VCH GmbH 15214095, 2022, 27, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adma.202201502 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [14/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advmat.de Figure 1. Round-robin results of BET area calculation. a) Distribution of BET areas from identical isotherms as calculated by 61 laboratories with an expertise in adsorption science and synthesis of porous materials. Superimposed are normalized probability distribution functions obtained by kernel density estimation (sample size for each material, n = 61). Inset shows the coefficient of variation (relative standard deviation) of results for each material. b) Exemplary isotherms for materials shown in (a). The large spread of BET areas reported for NU-1104 is due to the unusual shape of its adsorption isotherm, making manual BET fits difficult. The eponymously named Rouquerol criteria (Section S2, and synthesis of porous materials. In this round-robin exercise, Supporting Information) aim to ensure good practice in iden- we asked the researchers to calculate the BET areas in the way tifying a valid fitting range, and, as such, they have found wide- they saw most fit. More details about the specific materials and spread acceptance in the literature and have been adopted in the adsorption isotherms, sampled both from our laboratory both IUPAC and ISO standards.[19–22,24,25,37] Despite this safe- and from the NIST/ARPA-E database,[38] are included in the guard, we herein propose that current BET area calculations are Supporting Information, (Section S1). To avoid any recognition irreproducible for two reasons: first, the Rouquerol criteria are bias, all isotherms were anonymized and scaled-off arbitrarily. indeterminate in identifying the correct fitting region, as they Figure 1a shows the large spread of results obtained from apply to multiple regions simultaneously. Second, even if they manual calculations of BET areas in the round-robin experi- were determinate, they are too cumbersome and lengthy to ment, the full details can be found anonymized in the Sup- implement and are therefore often neglected in practice. This porting Information: tabulated in Section S3, and represented dilemma is reminiscent of the Skeptic’s Argument from Gor- graphically in Section S4. Most groups (90%) reported using gias, here paraphrased: i) the BET area does not exist (e.g., for the Rouquerol criteria in their manual calculation, 23% used microporous materials); ii) even if it exists, it cannot be system- a commercial software package, and 6% used a self-developed atically and unequivocally calculated (i.e., determined by the code. Details on the methods for each group can be found in Rouquerol criteria). Section S11 of the Supporting Information. Bar a few excep- To prove our hypothesis and to assess the current spread tions, virtually no two groups of experts reported identical BET of BET calculation results, we have shared a dataset of 18 iso- areas for any given isotherm. We observed a spread of at least therms (reported as relative pressure versus amount adsorbed), 300 m2 g−1 for each isotherm; however, that number was sig- already measured, and representing four classes of micro- and nificantly higher for some individual isotherms. For NU-1104, a mesoporous materials (zeolites, mesoporous silicas, MOFs, and modern MOF with substantial porosity (isotherm Figure 1b),[32] COFs) with 61 laboratories with expertise in adsorption science the highest estimate of 9341 m2 g−1 and the lowest estimate of Adv. Mater. 2022, 34, 2201502 2201502 (5 of 12) © 2022 The Authors. Advanced Materials published by Wiley-VCH GmbH 15214095, 2022, 27, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adma.202201502 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [14/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advmat.de Figure 2. Working principle for the BETSI algorithm. a) The isotherm is shown with a particular fitting region highlighted in red. The linear BET equation is applied, and an ordinary least-squares regression is applied to the fitting region. b) Subsequent checks against the Rouquerol criteria[24] are performed (insets) and c) valid fits are passed. The analysis shown in (a) is repeated for all consecutive combinations of points on the isotherm. A results matrix with n × n dimensionality stores all acceptable and rejected fits. d) All acceptable BET areas are output and plotted against the percentage error under the 4th Rouquerol criterion ((a), top inset). All BET areas ending on the highest permissible point under the 1st Rouquerol criterion ((a), bottom inset, maximum in the N(1 − (P/P0)) function) are labelled as the isotherm knee and shown in blue. The BETSI Optimal BET area (yellow) belongs to the isotherm knee group and has the lowest percentage error under the 4th Rouquerol criterion. 1757 m2 g−1 differed by an astonishing 7584 m2 g−1, making the only systematically selects an optimal fitting region but does so highest estimate more than five times higher than the lowest by eliminating all other hypothetical fitting regions. With thou- estimate. The spread of values for frequently reproduced MOFs sands of consecutive combinations of points, the large number such as HKUST-1, MOF-5, and ZIF-8 was slightly smaller than of potential fittings is impossible to carry out manually. that of literature cited values.[36] While this observation affirms To solve the problem of manual BET fitting, we developed a the natural assumption that material synthesis and isotherm computational tool for BET analysis, BET surface identification measurement play a more important role in determining the (BETSI). This tool makes an unambiguous calculation of the BET area than the calculation, we can nevertheless attribute a BET area based on the original Rouquerol criteria but modified significant portion of this spread to current BET fittings. The to prevent manual interaction, requiring only the adsorption results of this social study demonstrate that there is significant isotherm as input data. As such, the results obtained from the variation in BET area calculations from the same isotherm, as round-robin evaluation were compared with the BETSI calcu- it is extremely unlikely for two researchers to select identical fit- lations to assess the inter-rater reliability of manual BET cal- ting regions. At this point, we propose a novel method that not culations. Figure 2 shows the working principle of the BETSI Adv. Mater. 2022, 34, 2201502 2201502 (6 of 12) © 2022 The Authors. Advanced Materials published by Wiley-VCH GmbH 15214095, 2022, 27, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adma.202201502 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [14/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advmat.de Table 1. Results of BETSI analysis and round-robin evaluation for the isotherms used in the study. Material, isotherm of material under investiga- tion; BETSI, optimal BET area predicted by BETSI; BETSI range, full spread of BET areas that pass under BETSI; BETSI variation coefficient, relative standard deviation of BET areas that pass under BETSI; pass rate, number of BET areas that pass under BETSI expressed as a fraction of all hypothet- ical fittings; round-robin average, mean of BET areas calculated in round-robin evaluation (sample size for each material, n = 61); round-robin range, full spread of BET areas determined in round-robin evaluation; round-robin variation coefficient, relative standard deviation of BET areas calculated in round-robin evaluation; hit rate, fraction of BET areas calculated in the round-robin evaluation that lie within the BETSI range. Material BETSI BETSI range BETSI variation Pass rate Round-robin average Round-robin range Round-robin variation Hit rate [m2 g−1] [m2 g−1] coefficient [%] [%] [m2 g−1] [m2 g−1] coefficient [%] [%] HKUST-1 1556 8 0.090 2.419 1520 583 7.451 52 Zeolite13X 833 4 0.140 0.538 813 356 7.405 35 Mg-MOF74 1010 5 0.114 2.300 990 459 7.101 48 Al-Fumarate 1007 14 0.398 1.736 989 478 6.740 60 MCM-41 1001 60 1.573 3.329 994 1186 15.090 85 DMOF-1 1924 4 0.074 0.107 1860 795 8.500 15 MOF-5 3255 20 0.250 0.071 3170 1382 7.203 13 UiO-66 1145 91 1.901 0.870 1120 796 12.045 65 UiO-66-NH2 1424 285 4.710 1.722 1388 750 8.727 48 NU-1000 2068 160 1.619 4.218 2014 1486 7.752 80 ZIF-8 1709 188 3.718 0.861 1672 2085 14.396 58 MIL-101 2446 680 8.353 3.738 2429 2404 14.816 78 TPB-DMTP-COF 2875 711 7.298 5.375 2787 5031 21.472 80 MIL-100 2199 616 7.611 12.111 1964 1554 13.042 78 NU-1102 4931 204 1.139 0.862 4770 2915 8.541 38 NU-1104 5684 235 1.327 0.024 5553 7584 31.047 5 NU-1105 3635 0 0.000 0.011 3585 3974 16.991 0 PCN-777 2079 483 5.624 6.960 1946 2168 15.814 87 algorithm on a simplified N2 adsorption isotherm at 77 K for Beyond the isotherm knee, adsorptive activity decreases rapidly ZIF-8 (full details can be found in the Supporting Information, as the pores are mostly filled and the internal surfaces are sat- Section S5). First, the linearized BET equation is fitted to a par- urated. Within this subset of BET areas, the BETSI optimum ticular region of the isotherm using an ordinary least-squares is chosen as the one with the smallest percentage error under (OLS) regression (Figure 2a). The top panel shows the isotherm the 4th Rouquerol criterion, thus making the BET assignment with a fitting region highlighted in red, and the OLS regression unambiguous. is shown below. The plot insets show the checks against the Next, we ran BETSI on the isotherms distributed in the Rouquerol criteria (Figure 2b). If all criteria are met, the fitting round-robin experiment. In all cases, the spread of potential is passed. This calculation is looped over all data intervals of BETSI results (i.e., those in agreement with the Rouquerol cri- at least 10 points on the isotherm. The resulting BET fits are teria) was considerably narrower than that obtained by manual stored in a large n × n matrix, where the (j,i)-matrix element calculation (Table 1). Figure 3a shows the individual results corresponds to a fitting region starting at the jth-point and from the social experiment and the comparison with the BETSI ending on the ith-point (Figure 2c). All valid fitting results are results; the corresponding variation coefficients are shown in output and plotted against the percentage error under the 4th Section S6 (Supporting Information) and an alternative repre- Rouquerol criterion (Figure 2d). Alongside, BETSI outputs all sentation normalized to the BETSI range is shown in Section other BET parameters, such as monolayer capacity and the C S7 (Supporting Information). Since most groups reported using constant, as well as full regression diagnostics (Section S5, Sup- the Rouquerol criteria to calculate their BET areas, this substan- porting Information). tiates our second hypothesis—that the manual implementation Since multiple fittings comply with the Rouquerol criteria of the Rouquerol criteria is cumbersome and difficult to carry (Figure 2c,d), BETSI demonstrates that an unambiguous out in practice. For instance, in the case of NU-1104, the range assignment of the BET area is impossible under the Rouquerol of estimates decreases from 7500 m2 g−1 in the social study to criteria alone. This proves our hypothesis that the criteria in 235 m2 g−1 under BETSI. Interestingly, some isotherms gave their current form are indeterminate. For the prototypical ZIF-8 much larger spreads of results than others, suggesting that isotherm, a flexible MOF with narrow windows,[35] valid BET the BET model does not describe them as naturally and thus areas fall within a range of 1550 and 1750 m2 g−1 (Figure 2c,d). they are more susceptible to problems associated with the BETSI assigns special relevance to fitting ranges that end on Rouquerol criteria. Unsurprisingly, we also observed this trend the highest permissible point, which are usually dictated by the in the round-robin evaluation. To further investigate the good- 1st Rouquerol criterion, and labels these as the isotherm knee. ness of the isotherm fittings, we define the BETSI variation Adv. Mater. 2022, 34, 2201502 2201502 (7 of 12) © 2022 The Authors. Advanced Materials published by Wiley-VCH GmbH 15214095, 2022, 27, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adma.202201502 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [14/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advmat.de Figure 3. Social study results versus BETSI results. a) Distribution of BET areas for identical isotherms from the social study (red) and BETSI (blue) obtained by kernel density estimation (sample size for each material, n = 61). Superimposed is the BETSI optimum (yellow). Note that the distributions of values obtained by BETSI are considerably narrower in all cases than those in the social study. b) Plot of the BETSI variation coefficient (relative standard deviation of BETSI results) against the pass rate (fraction of valid fits against all hypothetical ones). Bubble size scales with the hit rate, the fraction of results from the social study that lie within the BETSI range. The red symbols have a hit rate of zero. Note the positive correlation between all three parameters. c) Isotherm fit classifications. Type A fits have a relatively wide fitting window, within which multiple fits are possible, but return a relatively narrow spread of BET results. Type B fits have a narrow fitting window and concomitantly return a narrow set of spread of results. Type C fits have wide fitting windows, which translates to multiple passable fits and a wide spread of permissible BET areas. coefficient as the relative standard deviation of BETSI results, the shape of these isotherms, we still offer some discussion of and the pass rate as the number of BET fits that pass under common features. Type A isotherms fit the BET model “best”. the Rouquerol criteria as a fraction of all potential fits. Further, Under BETSI, they have a relatively high pass rate and return the hit rate expresses the fractional number of BET areas cal- a fairly narrow spread of results. Examples include materials culated in the round-robin exercise that lie within the BETSI such as Al-fumarate, NU-1000, Zeolite-13X, and MCM-41. Many range. Figure 3b demonstrates the correlation between the pass of these isotherms do not have strongly pronounced isotherm rate, the BETSI variation coefficient, and the hit rate. Simply knees and some have mesoporous steps. Hit rates greater than put, the more BET fits are valid, the greater the spread of pos- 70% are generally observed for these materials, suggesting that sible BET areas is, and the more likely researchers are to sat- the majority of researchers did not struggle with the fittings. isfy the Rouquerol criteria in manual calculations. To account Type B isotherms only fit the BET model over a very limited for the non-equal spacing of points on all different isotherms, range. These have extremely low pass rates, meaning that only the pressure-adjusted pass rate expresses the total sum of pres- few BET fits are valid, which in turn will be spread narrowly. sure intervals that fit Rouquerol criteria as a fraction of the Examples include MOF-5, DMOF-1, NU-1104, HKUST-1, and sum of all pressure intervals of the hypothetical fitting ranges NU-1105. For the latter, out of 9409 hypothetical 10-point fits, (Section S8, Supporting Information). From Figure 3b, we clas- only one is permissible under the Rouquerol criteria. Such sify adsorption isotherms into three broad categories, types A, prohibitively low pass rates make the correct BET assignment B, and C (Figure 3c). While it is difficult to generalize about by hand virtually impossible and demonstrate the need for Adv. Mater. 2022, 34, 2201502 2201502 (8 of 12) © 2022 The Authors. Advanced Materials published by Wiley-VCH GmbH 15214095, 2022, 27, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adma.202201502 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [14/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advmat.de computational support. In contrast to type A isotherms, type A2ML measurements and from the NIST Adsorption Database. B isotherms often have sharp isotherm knees following strong Arbitrary scaling factors were introduced to minimize recollec- adsorptive interactions at low relative pressures. Isotherms tion bias of the isotherms. The isotherms were sent out in .csv with more complex shapes such as NU-1104 also appear in this format. All colleagues received the same email with the same category. Type C isotherm fittings are arguably the most prob- set of instructions (Section S1, Supporting Information): to cal- lematic. They have high pass rates and, concomitantly, they culate the BET area from the data in the way they saw most fit return large spreads of BET results. Typical materials that fit and to report a rough estimate of how long it took them to cal- into this category are MIL-101, MIL-100, TPB-DMTP-COF and culate them. An anonymized one-page summary of each lab’s PCN-777. Like type A isotherms, these have rounded isotherm own account of their calculation can be found in Section S11 knees, which appear at higher relative pressures. It is for these (Supporting Information). materials that the necessity to extend the Rouquerol criteria is For easier data handling, once rescaled, all results were demonstrated and the BETSI algorithm makes an unambig- rounded to the next integer. None of the data points has been uous BET assignment possible. eliminated. The data is presented as a jitter plot for each mate- rial, with a superimposed kernel-density estimation obtained in Python. 2. Outlook BET theory is a great success story. Developed in the 1930s for 3.2. BETSI non-microporous, open surfaces, it continues to this day to be applied to modern adsorbents with complex porosity. Despite The BETSI algorithm, including executables (for Linux, Win- the advances from classical density functional theory (DFT) dows, and macOS) and a user manual are available at ref. [39]. methods, the BET area will likely continue playing a crucial The “How to Use BETSI” guide is included in the Supporting role in porosimetry for decades to come, with impacts in energy Infomation (Section S15). The program is written in Python research, transport, medical applications and climate-change and uses principally the NumPy library. It loops linear regres- mitigation. In light of these future developments, it will become sions over all consecutive combinations of at least three points, increasingly important to share critical scientific metrics reli- performs full BET analyses, and stores the fitting parameters in ably to find a common language to report both academic and n × n results matrices, where the (j,i)-matrix element denotes a industrial progress. linear regression from the jth to the ith point on the isotherm. Here, we have demonstrated the difficulties in unambigu- Binary pass/fail matrices with the same dimensionality are used ously determining BET areas from adsorption isotherms, independently to assess compliance with linearity and fitting which, in turn, affect the assessment of material quality and criteria. The “filtering” of BET areas is achieved by element- reproducibility. These problems arise from imperfect and insuf- wise matrix multiplication of the results matrices and the pass/ ficient manual calculations and can only be met using modern fail matrices. This allows independent “activation” and “deacti- computational methods. Furthermore, we propose BETSI as a vation” of the criteria and observing the effects on the results. step toward greater transparency and criticality in determining The minimum fitting requirement of ten points is coded in a BET areas. We stress here that it is neither the function nor the pass/fail matrix to allow for some minimum point flexibility, as purpose of BETSI to eliminate doubt and treat a particular BET is the cut-off value for R2 of 0.995. To avoid low-leverage non- area as “true.” Researchers should remain aware of the limita- linearity in the linear region, the first Rouquerol criterion has tions of BET theory when applied to microporous adsorbents in been extended to also require the linearized BET function to general and when BET areas are reported, the pressure range increase monotonically with P/P0, as well as N(1 − (P/P0)). The and number of points used should always be stated. We fur- third and fourth Rouquerol criteria are implemented through a ther recommend here that isotherms must be reported trans- 10 000-point Pchip interpolation of the isotherm to reconstruct parently and in detail, i.e., semi-log representation to show the the Nm (Read). As the third and fourth criteria require the Nm low-pressure regions. The “experiment” is the adsorption iso- (BET) to be a real value, i.e., they require C to be positive, the therm—not the BET area. second criterion cannot be independently deactivated from the third and the fourth. The associated logic has been written into the program. Following the BETSI filtering by multiplication 3. Methods Section of results and pass/fail matrices, the isotherm knee is identi- fied as the subset of BET areas whose fitting region end on the 3.1. Round-Robin Evaluation highest permissible pressure point. In most cases, this will be the highest permissible point under the first Rouquerol cri- N2 adsorption isotherms of 18 different materials (Section S10, terion. The optimal BETSI prediction is chosen as the fitting Supporting Information) were sent to international collabora- region with the lowest percentage error under the fourth crite- tors: HKUST-1, ZIF-8, NU-1000, MIL-101, UiO-66, Al fumarate, rion and belonging to the isotherm knee subset. Zeolite13X, Mg-MOF-74, UiO-66-NH2, MOF-5, DMOF-1, MCM- BETSI only requires the adsorption isotherm as input data 41, TPB-DMTP-COF, MIL-100, NU-1102, NU-1104, NU-1105, and returns six plots used to validate the results: the isotherm and PCN-777; they were anonymized and labelled A-R respec- itself, with the optimal linear region highlighted as well as tively. Note that this is not the order in which the isotherms the BET fit; the “Rouquerol representation” of the isotherm, appear in the paper. The isotherms were sampled from the N(1 − (P/P0)) plotted against P/P0; the linearized plot with the Adv. Mater. 2022, 34, 2201502 2201502 (9 of 12) © 2022 The Authors. Advanced Materials published by Wiley-VCH GmbH 15214095, 2022, 27, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adma.202201502 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [14/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advmat.de OLS regression and the regression parameters; the filtered per- R.S.F. and D.B. acknowledge funding from the European Research centage error versus BET areas plot with the isotherm knee Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (SCoTMOF), ERC-2015-StG 677289. Sandia and optimal BET area highlighted; the filtered monolayer- National Laboratories is a multimission laboratory managed and loadings plot showing all permissible monolayer loadings on operated by National Technology and Engineering Solutions of Sandia, the isotherm; and the statistical distribution of permissible BET LLC., a wholly owned subsidiary of Honeywell International, Inc., for the areas with a boxplot. Additionally, BETSI returns four regres- U.S. Department of Energy’s National Nuclear Security Administration sion diagnostics plots, which can be used to assess whether the under contract DE-NA-0003525. The authors gratefully acknowledge assumptions of OLS regression have been met: the residuals funding from the U.S. Department of Energy, Office of Energy Efficiency versus fitted values plot can be used to visually inspect whether and Renewable Energy, Hydrogen and Fuel Cell Technologies Office, through the Hydrogen Storage Materials Advanced Research Consortium the residuals are normally distributed around the regres- (HyMARC). This paper describes objective technical results and analysis. sion line, and similar information could be obtained from the Any subjective views or opinions that might be expressed in the paper QQ-plot. Finally, the scale-location plot can be used to assess do not necessarily represent the views of the U.S. Department of Energy whether the distribution of studentized residuals is homosce- or the United States Government. J.D.E. acknowledges the support of datic or heteroscedatic and the residuals versus leverage plot the Alexander von Humboldt Foundation and the Center for Information can be used to identify high-leverage points that have an abnor- Services and High Performance Computing (ZIH) at TU Dresden. S.K.G. and S.M. acknowledge SERB (Project No. CRG/2019/000906), India for mally large influence on the regression line. financial support. K.K. and R.K. acknowledge Active Co. Research Grant for funding. S.K. acknowledges funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and 3.3. Comparison between Round-Robin Evaluation innovation programme (COSMOS), ERC-2017-StG 756489. N.L. and and BETSI Results J.G.M acknowledge funding from the European Commission through the H2020-MSCA-RISE-2019 program (ZEOBIOCHEM – 872102) and the Spanish MICINN and AEI/FEDER (RTI2018-099504-B-C21). N.L. Statistical analysis of the results was performed in Python. thanks the University of Alicante for funding (UATALENTO17-05). The BETSI variation coefficient and the round-robin variation ICN2 is supported by the Severo Ochoa program from the Spanish co­ efficient are standard deviations relative to the average of MINECO (Grant No. SEV-2017-0706) S.M.J.R. and A.L. wish to thank each set. The pass rate for each isotherm is the number of per- the Fund for Scientific Research Flanders (FWO), under grant nos. missible BET fits as a fraction of all consecutive combination 12T3519N and 11D2220N. L.S. was supported by the EPSRC Cambridge of points. To account for non-equal spacing of the points on NanoDTC EP/L015978/1. C.T.Y. and T.S.N. acknowledges funds from the each isotherm, the pressure-adjusted pass-rate is obtained by National Research Foundation of Korea, NRF-2017M3A7B4042140 and NRF-2017M3A7B4042235. P.F. and H. Y. acknowledge US Department integrating along the pressure axis and dividing the total sum of Energy, Office of Basic Energy Sciences, Materials Sciences and of permissive pressure intervals by the sum of all consecutive Engineering Division under Award No. DE-SC0010596 (P.F.). R.O. pressure intervals. The hit rate is the fractional number of BET would like to acknowledge funding support during his Ph.D. study from areas calculated in the round-robin evaluation that lie within Indonesian Endowment Fund for Education-LPDP with the contract No. the BETSI range. 202002220216006. Daniel Siderius: Official contribution of the National Institute of Standards and Technology (NIST), not subject to copyright in the United States of America. Daniel Siderius: Certain commercially available items may be identified in this paper. This identification 3.4. Statistical Analysis does not imply recommendation by NIST, nor does it imply that it is the best available for the purposes described. B.V.L, S.T.E and A.M.P Data was analyzed using the NumPy, SciPy, and Matplotlib acknowledge funding from the European Research Council (ERC) under libraries in Python. For the violin plots, a sample size, n = 61 the European Union’s Horizon 2020 Research and Innovation Program was used for each material, and the estimator bandwidth for (Grant agreement no. 639233, COFLeaf). the kernel density estimator (KDE) was calculated using Scott’s rule. For the bubble plots, the hit rate has been averaged over a sample size, n = 61, for each material. Conflict of Interest The authors declare no conflict of interest. Supporting Information Supporting Information is available from the Wiley Online Library or Author Contributions from the author. J.W.M.O., J.R., D.M., and N.R. contributed equally to this work. J.W.M.O. and D.F-J. designed the study. J.W.M.O. and D.M. collected and curated the dataset of isotherms shared among co-authors. J.W.M.O., J.R., N.R., L.S. and B.C. developed BETSI code. All co-authors calculated the BET Acknowledgements areas from the pre-measured isotherms. J.W.M.O., D.M., and D.F-J. co-wrote the paper. All authors discussed the results and contributed to This project has received funding from the European Research the editing of the manuscript. Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (NanoMOFdeli), ERC-2016-COG 726380, Innovate UK (104384) and EPSRC IAA (IAA/RG85685). N.R. acknowledges the support of the Cambridge International Scholarship and the Trinity- Keywords Henry Barlow Scholarship (Honorary). O.K.F. and R.Q.S. acknowledge funding from the U.S. Department of Energy (DE-FG02-08ER15967). adsorption, BET theory, porosimetry, porous materials, surface area Adv. Mater. 2022, 34, 2201502 2201502 (10 of 12) © 2022 The Authors. Advanced Materials published by Wiley-VCH GmbH 15214095, 2022, 27, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adma.202201502 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [14/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advmat.de Received: February 15, 2022 [21] International Organization for Standardization (ISO), Ref. number Revised: March 21, 2022 ISO 2010, https://doi.org/10.1007/s11367-011-0297-3. Published online: May 23, 2022 [22] J. Rouquerol, F. Rouquerol, P. Llewellyn, G. Maurin, K. S. W. Sing,Adsorption by Powders and Porous Solids: Principles, Methodology and Applications, 2nd ed., Elsevier, San Diego, CA, USA 2013. [1] S. Brunauer, P. H. Emmett, E. Teller, J. Am. Chem. Soc. 1938, 60, [23] F. Ambroz, T. J. Macdonald, V. Martis, I. P. Parkin, Small Methods 309. 2018, 2, 1800173. [2] R. Cid, R. Arriagada, F. Orellana, J. Catal. 1983, 80, 228. [24] D. A. Gómez-Gualdrón, P. Z. Moghadam, J. T. Hupp, O. K. Farha, [3] D. Zhao, J. Feng, Q. Huo, N. Melosh, G. H. Fredrickson, R. Q. Snurr, J. Am. Chem. Soc. 2016, 138, 215. B. F. Chmelka, G. D. Stucky, Science 1998, 279, 548. [25] K. S. Walton, R. Q. Snurr, J. Am. Chem. Soc. 2007, 129, 8552. [4] J. S. Beck, J. C. Vartuli, W. J. Roth, M. E. Leonowicz, C. T. Kresge, [26] J. Park, J. D. Howe, D. S. Sholl, Chem. Mater. 2017, 29, 10487. K. D. Schmitt, C. T. W. Chu, D. H. Olson, E. W. Sheppard, [27] P. Z. Moghadam, A. Li, S. B. Wiggin, A. Tao, A. G. P. Maloney, S. B. McCullen, J. B. Higgins, J. L. Schlenker, J. Am. Chem. Soc. P. A. Wood, S. C. Ward, D. Fairen-Jimenez, Chem. Mater. 2017, 29, 1992, 114, 10834. 2618. [5] A. Corma, Chem. Rev. 1997, 97, 2373. [28] M. E. Davis, Nature 2002, 417, 813. [6] S. Kitagawa, R. Kitaura, S. I. Noro, Angew. Chem., Int. Ed. 2004, 43, [29] H. Furukawa, K. E. Cordova, M. O’Keeffe, O. M. Yaghi, Science 2013, 2334. 341, 1230444. [7] H. C. Zhou, J. R. Long, O. M. Yaghi, Chem. Rev. 2012, 112, 673. [30] O. M. Yaghi, J. Am. Chem. Soc. 2016, 138, 15507. [8] C. S. Diercks, O. M. Yaghi, Science 2017, 355, eaal1585. [31] O. K. Farha, I. Eryazici, N. C. Jeong, B. G. Hauser, C. E. Wilmer, [9] J. Li, J. Sculley, H. Zhou, Chem. Rev. 2012, 112, 869. A. A. Sarjeant, R. Q. Snurr, S. T. Nguyen, A. Ö. Yazaydin, J. T. Hupp, [10] O. K. Farha, A. Ö. Yazaydin, I. Eryazici, C. D. Malliakas, B. G. Hauser, J. Am. Chem. Soc. 2012, 134, 15016. M. G. Kanatzidis, S. T. Nguyen, R. Q. Snurr, J. T. Hupp, Nat. Chem. [32] T. C. Wang, W. Bury, D. A. Gómez-Gualdrón, N. A. Vermeulen, 2010, 2, 944. J. E. Mondloch, P. Deria, K. Zhang, P. Z. Moghadam, A. A. Sarjeant, [11] B. Li, H.-M. Wen, W. Zhou, B. Chen, J. Phys. Chem. Lett. 2014, 5, R. Q. Snurr, J. F. Stoddart, J. T. Hupp, O. K. Farha, J. Am. Chem. Soc. 3468. 2015, 137, 3585. [12] P. Z. Moghadam, T. Islamoglu, S. Goswami, J. Exley, M. Fantham, [33] Z. Chen, P. Li, R. Anderson, X. Wang, X. Zhang, L. Robison, C. F. Kaminski, R. Q. Snurr, O. K. Farha, D. Fairen-Jimenez, Nat. L. R. Redfern, S. Moribe, T. Islamoglu, D. A. Gómez-Gualdrón, Commun. 2018, 9, 1378. T. Yildirim, J. F. Stoddart, O. K. Farha, Science 2020, 368, 297. [13] A. Corma, H. García, F. X. Llabrés i Xamena, Chem. Rev. 2010, 110, [34] I. M. Hönicke, I. Senkovska, V. Bon, I. A. Baburin, N. Bönisch, 4606. S. Raschke, J. D. Evans, S. Kaskel, Angew. Chem., Int. Ed. 2018, 57, [14] L. E. Kreno, K. Leong, O. K. Farha, M. Allendorf, R. P. Van Duyne, 13780. J. T. Hupp, Chem. Rev. 2012, 112, 1105. [35] D. Fairen-Jimenez, S. A. Moggach, M. T. Wharmby, P. A. Wright, [15] W. P. Lustig, S. Mukherjee, N. D. Rudd, A. V. Desai, J. Li, S. Parsons, T. Düren, J. Am. Chem. Soc. 2011, 133, 8900. S. K. Ghosh, Chem. Soc. Rev. 2017, 46, 3242. [36] M. Agrawal, R. Han, D. Herath, D. S. Sholl, Proc. Natl. Acad. Sci. [16] P. Horcajada, R. Gref, T. Baati, P. K. Allan, G. Maurin, P. Couvreur, USA 2020, 117, 877. G. Férey, R. E. Morris, C. Serre, Chem. Rev. 2012, 112, 1232. [37] J. Rouquerol, P. Llewellyn, F. Rouquerol, Stud. Surf. Sci. Catal. 2007, [17] I. Langmuir, J. Am. Chem. Soc. 1918, 40, 1361. 160, 49. [18] Y. K. Tovbin, Russ. Chem. Bull. 1998, 47, 637. [38] NIST/ARPA-E Database of Novel and Emerging Adsorbent Materials [19] M. Thommes, K. Kaneko, A. V. Neimark, J. P. Olivier, (Eds: D.W. Siderius, V.K. Shen, R.D. JohnsonIII, R. d. vanZee), F. Rodriguez-Reinoso, J. Rouquerol, K. S. W. Sing, Pure Appl. Chem. National Institute of Standards and Technology, Gaithersburg MD, 2015, 87, 1051. USA, https://doi.org/10.18434/T43882 (accessed: August 2020). [20] K. S. W. Sing, D. H. Everett, R. A. W. Haul, L. Moscou, R. A. Pierotti, [39] BET Surface Identification via Fairen Group at GitHub, https:// J. Rouquerol, T. Siemieniewska, Pure Appl. Chem. 1985, 57, 603. github.com/fairen-group/betsi-gui (accessed: May 2022). Johannes W. M. Osterrieth did his Ph.D. in the A2ML Group under the supervision of Prof. David Fairen-Jimenez in the Department of Chemical Engineering and Biotechnology, University of Cambridge. He received his M.Chem. in chemistry at Oxford University (2016) and subse- quently moved to Cambridge to join the EPSRC Doctoral Training Centre for Nanoscience and Nanotechnology (NanoDTC). His research interests lie in nanoparticle encapsulation into metal– organic frameworks and the fundamentals of adsorption models. Adv. Mater. 2022, 34, 2201502 2201502 (11 of 12) © 2022 The Authors. Advanced Materials published by Wiley-VCH GmbH 15214095, 2022, 27, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adma.202201502 by Spanish Cochrane National Provision (Ministerio de Sanidad), Wiley Online Library on [14/10/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License www.advancedsciencenews.com www.advmat.de David Madden is currently a Marie Skłodowska-Curie Fellow at the University of Limerick, Ireland. He previously worked in the A2ML Group as a Senior Research Associate under the mentorship of Prof. David Fairen- Jimenez in the Department of Chemical Engineering and Biotechnology, University of Cambridge. He completed his Ph.D. in physical chemistry in 2015 at the University of Limerick, and his research to date has focused on the use of novel porous materials for energy- efficient purifications, carbon capture and storage and inorganic materials scale-up while working with Prof. Michael Zaworotko, Prof. Gavin Walker, and Dr. Teresa Curtin at the Bernal Institute in the University of Limerick. Nakul Rampal is currently a Ph.D. student in the A2ML Group under the supervision of Prof. David Fairen-Jimenez. Prior to Cambridge, he completed his M.S. at the University of California, Berkeley where he was advised by Prof. Berend Smit. He did his undergraduate research at IIT Bombay, where he was advised by Prof. Ateeque Malani. He completed his undergraduate degree at Manipal Institute of Technology. His research deals with the development of novel algorithms using molecular simulation, high-performance computing, and machine learning, to accelerate the discovery of novel porous materials for applications such as carbon capture, energy storage, and drug delivery. David Fairen-Jimenez is a professor of molecular engineering in the Department of Chemical Engineering & Biotechnology at the University of Cambridge, where he leads the Adsorption & Advanced Materials Laboratory (A2ML). He got his Ph.D. in porous materials in chemistry from the University of Granada and worked later at the University of Edinburgh and Northwestern University. He leads an interdisciplinary lab that focuses on the study of adsorption in porous materials, combining advanced experimental techniques and computational modeling for energy and healthcare applications. Adv. Mater. 2022, 34, 2201502 2201502 (12 of 12) © 2022 The Authors. Advanced Materials published by Wiley-VCH GmbH