Nanoscale View Article Online COMMUNICATION View Journal | View Issue Rotated domains in selective area epitaxy grown This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. Cite this: Nanoscale, 2021, 13, 18441 Zn3P2: formation mechanism and functionality† Open Access Article. Published on 30 October 2021. Downloaded on 7/7/2022 4:18:39 PM. Maria Chiara Spadaro, ‡a Simon Escobar Steinvall,, ‡§b Nelson Y. Dzade, c,d Received 20th September 2021, Accepted 21st October 2021 Sara Martí-Sánchez, a Pol Torres-Vila,a Elias Z. Stutz, b Mahdi Zamani,b DOI: 10.1039/d1nr06190a Rajrupa Paul,b Jean-Baptiste Leran,b Anna Fontcuberta i Morral *b,e and Jordi Arbiol *a,f rsc.li/nanoscale Zinc phosphide (Zn3P2) is an ideal absorber candidate for solar theory simulations describe the rotated domains’ formation cells thanks to its direct bandgap, earth-abundance, and opto- mechanism and demonstrate their lattice-matched epitaxial electronic characteristics, albeit it has been insufficiently investi- relation. In addition, the energies of the shallow states predicted gated due to limitations in the fabrication of high-quality material. closely agree with transition energies observed by experimental It is possible to overcome these factors by obtaining the material studies and offer a potential origin for these defect transitions. Our as nanostructures, e.g. via the selective area epitaxy approach, study represents an important step forward in the understanding enabling additional strain relaxation mechanisms and minimizing of Zn3P2 and thus for the realisation of solar cells to respond to the the interface area. We demonstrate that Zn3P2 nanowires grow present call for sustainable photovoltaic technology. mostly defect-free when growth is oriented along the [100] and [110] of the crystal, which is obtained in nanoscale openings along the [110] and [010] on InP(100). We detect the presence of two stable rotated crystal domains that coexist in the structure. They Introduction are due to a change in the growth facet, which originates either Nanoscale approaches are increasingly used to help circum- from the island formation and merging in the initial stages of vent challenges experienced by certain materials if they were to growth or lateral overgrowth. These domains have been visualized be used in their bulk form. One such material is zinc phos- through 3D atomic models and confirmed with image simulations phide (Zn3P2), a potential photovoltaic absorber (Eg ∼1.50 eV) of the atomic scale electron micrographs. Density functional or IR-optoelectronics material which has experienced limit- ations due to its lattice parameter and coefficient of thermal expansion (CTE) not matching commonly available substrates, a Catalan Institute of Nanoscience and Nanotechnology (ICN2), CSIC and BIST, thus complicating high-quality epitaxial growth.1–6 What Campus UAB, Bellaterra, Barcelona, Catalonia, 08193, Spain. E-mail: arbiol@icrea.cat makes Zn3P2 attractive is the earth-abundance of the constitu- b Laboratory of Semiconductor Materials, Institute of Materials, Faculty of ent elements,7 its direct bandgap close to the optimum for Engineering, Ecole Polytechnique Fédérale de Lausanne, 1015 Lausanne, photovoltaic applications,8 high optical absorption in the Switzerland. E-mail: anna.fontcuberta-morral@epfl.ch visible range (104–105 cm−1),9,10 and long minority carrier c School of Chemistry, Cardiff University, Main Building, Park Place, CF10 3AT diffusion length,8 which combined hold promise to make this Cardiff, UK d Department of Energy and Mineral Engineering, Pennsylvania State University, a high-impact and sustainable photovoltaic material. University Park, PA 16802, USA Zn3P2 has a tetragonal crystal structure belonging to the e Institute of Physics, Faculty of Basic Sciences, Ecole Polytechnique Fédérale de P42/nmc space group (a = 8.089 Å, c = 11.39 Å), and a CTE of Lausanne, 1015 Lausanne, Switzerland 1.4 × 10−5 K−1, which is an order of magnitude larger than f ICREA, Pg. Lluís Companys 23, 08010 Barcelona, Catalonia, Spain commercial semiconductors (e.g. Si 2.96 × 10−6 K−1) and more † Electronic supplementary information (ESI) available: SEM image of coalesced nanowires, simulated HAADF STEM image and GPA analysis of different inter- closely matched to that of steel (1.1–1.3 × 10−5 K−1 depending face structures and orientations, GPA analysis of a nanopyramid, overview on exact composition).2,11–13 These factors complicate epitaxial images to look at the density of the domains and their points of nucleation, growth without a high number of misfit dislocations on com- additional DFT data of the simulations using different initial interface relax- mercially available substrates. However, the large size of the ation, and TEM data showing the rotated domains in vapour–liquid–solid grown unit cell is to account for the ordering of the ordered empty nanowires. See DOI: 10.1039/d1nr06190a sites in what would otherwise be a cubic zinc sublattice.14–17 ‡ These authors contributed equally. pffiffiffi § Current address: Center for Analysis and Synthesis and NanoLund, Lund Therefore, as a  2c, it is possible to create a pseudo-cubic University, Box 124, 221 00 Lund, Sweden. reconstruction of the unit cell by disregarding the ordering of This journal is © The Royal Society of Chemistry 2021 Nanoscale, 2021, 13, 18441–18450 | 18441 View Article Online Communication Nanoscale these sites with a lattice parameter of 5.72 Å, which has been imaging of atomic columns with elemental sensitivity and shown to be relevant in the interface ordering during epitaxial compatibility with a range of analytical techniques.56–63 Recent growth.16,18 It has thus been shown to grow epitaxially on GaAs advances have increased the achievable resolution, allowing and InP substrates.1,16,18,19 Growth on GaAs resulted in an for the analysis of structural and chemical information down interface GaP layer which was detrimental to device perform- to the atomic level. Furthermore, HAADF-STEM opens the ance, making InP the ideal system for epitaxial growth studies possibility to create 3D atomic models of the nanostructures of Zn3P2.20 under study, in order to get a deeper insight into the main Additional methods to facilitate growth of Zn3P2 include growth processes from an atomistic point of view, and in paral- molecular beam epitaxy (MBE) and its growth in the form of lel complement the study with DFT analysis for energetic cal- This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. nanostructures.1,3,6,16,17 MBE growth of Zn3P2 utilises relatively culations of the different interfaces created and a basic expla- nation of defect formation.57,64,65 Strain relaxation mechanism Open Access Article. Published on 30 October 2021. Downloaded on 7/7/2022 4:18:39 PM. low growth temperatures as compared to other commonly used techniques.21 The lower growth temperatures help miti- evaluation will follow by performing HAADF-STEM image gate the potential post-growth strain build up due to mis- simulation, from the 3D atomic models, and extended dedi- matching CTE, which could otherwise result in defect for- cated analysis such as geometrical phase analysis (GPA).66–69 mation of the grown material. Nanostructures, on the other The application of such technique to the experimental and hand, have been shown to allow for additional elastic strain simulated HAADF-STEM images ensures the possibility to relaxation mechanisms due to their reduced dimensions, as verify the growth relaxation mechanisms involved in the nano- well as supressing misfit dislocation formation due to their structure being studied, together with relative dilatation and limited interface area (efficiency being a function of lattice rotation of the atomic planes, giving a complete illustration of mismatch and interface area).22–28 One of the main the growth process characteristics. approaches for the epitaxial growth of nanostructures is selec- Herein, we investigate SAE grown Zn3P2 nanowires through tive area epitaxy (SAE), which utilises a mask to limit the AC HAADF-STEM and simulations to elucidate the epitaxial growth to certain (nanoscale) openings in specific directions relationship with the InP substrates and to gain insight into and is compatible with scalable processing.16,23–25,29–32 The defect formation in the material and the potential influence of small interface area helps to limit the formation of interface said defects on the final material performance. related defects, and the shape of the nanostructure can be con- trolled by the geometry of the openings.16,29,33,34 While this helps minimise the interface related defects, there are other Results and discussion sources of defects in Zn3P2.35–38 Various defect levels in the bandgap have been experimentally probed by various groups Horizontal Zn3P2 nanowires were grown by SAE through the previously, but very few have been able to describe the exact scheme illustrated in Fig. 1a, showing how the grown material origins.36,37,39–44 Conversely, other studies have observed is localised to the exposed nanoscale holes (more details in defects, such as rotated domains observed through electron the Methods section). By tuning the shape of the openings we microscopy, but their exact nature and influence on the can grow single-crystal nanowires of desired length at a range materials properties is yet to be ascertained.16 This creates a of different angles as shown in Fig. 1b. The material and inter- need for systematic and comprehensive studies able to corre- facial quality remained unchanged irrespective of the angle as late the different factors related to defect structure and their discussed in more detail below. We focused our attention on influence in Zn3P2 in order for it to realise its potential. two crystalline orientations of the Zn3P2 nanowires which are An additional benefit of nanostructures is their nanopho- at 45° with respect to each other. Specifically, the nanowires tonic properties, including directed emission, enhanced with their axial directions being (001) × [100]Zn3P2 on (001) × optical absorption cross-section, and light trapping to name a [110]InP (namely the 0° orientation) and (001) × [110]Zn3P2 on few.45–49 Nanophotonics make them ideal for optoelectronic (001) × [010]InP (namely the 45° orientation). However, at inter- and photovoltaic applications amongst other applications.50–55 mediate angles with respect to the <110> directions of the InP The potential of nanostructured materials has been extensively (100) substrate the nanowires exhibit microfaceting leading to researched for e.g. III–Vs, but their potential impact on for surface steps (see Fig. 2e). Furthermore, by patterning the example earth-abundant II–V semiconductors such as Zn3P2 openings into networks we were able to grow interconnected still remains to be fully explored. nanowires with regular junctions as shown in Fig. 1c, and even To comprehensively understand the nanostructured overgrow them into a textured film (Fig. S1†). material growth mechanisms with the aim to optimize its The crystal quality evaluation is summarised in Fig. 2 for structure and properties, it is imperative to collect structural horizontal nanowires at 0° (Fig. 2a–d) and 45° (Fig. 2e–h), information down to the atomic level to accurately characterise respectively. Fig. 2a and e are SEM overview images of the material and defects. Currently, one of the sole techniques nanowire. Here the [100]Zn3P2 and [110]Zn3P2 directions are par- capable of such detailed characterisation is aberration-cor- allel to [110]InP and [010]InP, respectively, and the resulting rected scanning transmission electron microscopy (AC STEM). nanowires grow epitaxially along the <001> direction. The High-angle annular dark-field (HAADF) has been the dominat- nanowires present a triangular cross-sectional shape with the ing imaging mode due to its incoherent nature allowing direct surface encompassed by the {011}Zn3P2 and {0–11}Zn3P2 family 18442 | Nanoscale, 2021, 13, 18441–18450 This journal is © The Royal Society of Chemistry 2021 View Article Online Nanoscale Communication the scenario when the [100]Zn3P2 direction is parallel to [110]InP, the (001)Zn3P2 planes present a 0.9% compression even after outgrowing the mask to accommodate the interfacial strain.11,71 Regarding the parallel planes, the (010)Zn3P2, there is a remnant tensile strain causing a 1% expansion to ensure a better plane-to-plane adaptation at the interface and prevent the formation of misfit dislocations. These findings are in good agreement with the Zn3P2 pyramids grown on InP16 along the same orientation as for the 0° case, as reported in This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. Fig. S4.† For the [110]Zn3P2 direction parallel to [010]InP, the Open Access Article. Published on 30 October 2021. Downloaded on 7/7/2022 4:18:39 PM. (001)Zn3P2 planes exhibit almost complete relaxation, while the (1–10)Zn3P2 planes are slightly expanded (tensile strain). Therefore, for both investigated orientations, there is a small component of tensile strain along the <1–10>Zn3P2 direc- tion that results in a defect-free epitaxial relationship between the Zn3P2 nanowire’s lattice with respect to the InP substrate. In fact, the remnant strains for the <1–10>Zn3P2 nanowires are actually lower than those of the <100>Zn3P2 nanowires. From this analysis we can conclude that the SAE approach is suc- cessful in terms of high crystal and interface quality. The resulting Zn3P2 is minimally strained with respect to its bulk counterpart, preventing the formation of misfit dislocations and strain build-up that could cause crack formation, both which are detrimental for the final material’s application per- formance. Further details on the GPA analyses have been Fig. 1 (a) Schematic of the SAE growth process for Zn3P2. (b) Overview reported in the ESI.† SEM image of Zn3P2 nanowires grown at different angles by SAE. (c) SEM In Fig. 3 we report some examples of another type of defect image of a Zn3P2 nanowire network grown by SAE. in the Zn3P2 nanowires that may be more difficult to control during growth, which is the presence of different rotated domains within the length of the nanowire. Fig. 3a and b show of planes. Considering the [100]Zn3P2 direction parallel to a general overview of the obtained FIB lamellae for both the [110]InP (orientation 0°), we can observe that the interface cross-section (Fig. 3a) and longitudinal (Fig. 3b) cuts. It is between the substrate and the nanowire is sharp, as evidenced possible to recognize the presence of more than one domain in Fig. 2b and c. The substrate presents some faceting inside already from the HAADF-STEM image contrast (as in the insets the mask openings along the {1–13} planes, probably induced of Fig. 3c and g) and/or corresponding power spectrum ana- during the etching process. On the contrary, when [110]Zn3P2 lysis. The different Zn3P2 crystal domains observed in a nano- direction is parallel to [010]InP (orientation 45°), the surface is wire with a <010>Zn3P2 axial direction are highlighted in Fig. 3d rough as a consequence of microfaceting, as observed in in the frequency filtered map, obtained from the Zn3P2 inside Fig. 2f and g. the SiO2 mask, as indicated with the yellow square in Fig. 3c. From the GPA analysis we could investigate plane relative In the inset of Fig. 3d we show the power spectra in the region distortions, in terms of dilatation and rotation, for the Zn3P2 of interest where the colour of the spectra corresponds to the nanowire with respect to the InP substrate, as reported in regions in the filtered map on the same zone axis. In the Fig. 2d and h. The residual strain is calculated considering the orange marked area, the Zn3P2 is observed along its [−111] corresponding bulk material as in Fig. S2 and S3† and the zone axis, while in the violet area the Zn3P2 is found rotated to results are summarised in Tables 1 and 2. In particular, we its [010] zone axis. As can be seen from the simulated diffrac- focused on the parallel and perpendicular directions with tion patterns, the main reflections arising from [−111] zone respect to the (001)InP surface. When the [100]Zn3P2 direction is axis perfectly overlap with some of the reflections arising from parallel to [110]InP, the (001)Zn3P2 ( parallel) and (010)Zn3P2 ( per- the [010] zone axis. Therefore, while it is straightforward to pendicular) planes are not significantly rotated with respect to filter and show the [010] oriented domains thanks to their the substrate, while in the 45° orientation there is a slight extra diffraction spots or frequencies, it is challenging to separ- rotation towards positive angles following the white arrow in ate the [−111] oriented domain as every diffraction spot here Fig. 2h, as indicated in the line profile in the inset, which pre- has a similar frequency for the other orientation. Furthermore, viously has been attributed to strain relaxation mechanisms in in both domains the atomic arrangement is similar, therefore SAE nanowires.70 just residual strain and no presence of dislocations have been The material accommodation during the epitaxial growth is detected via GPA analysis. Zn3P2 tends to form these two reflected in the overall plane dilatation in the nanowire. For stable rotated domains already in the growth region confined This journal is © The Royal Society of Chemistry 2021 Nanoscale, 2021, 13, 18441–18450 | 18443 View Article Online Communication Nanoscale This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. Open Access Article. Published on 30 October 2021. Downloaded on 7/7/2022 4:18:39 PM. Fig. 2 (a) SEM overview image showing the Zn3P2 nanowire morphology grown along <110> with respect to the InP substrate (0°). (b) HAADF-STEM overview image of the nanowire cross-section obtained from the nanowire in (a). (c) HAADF-STEM image of the detail in the nanowire-substrate interface area. The indexed power spectrum is reported in the inset to evaluate the mutual orientation of the two systems (scale bar 5 1 nm−1). (d) GPA rotational and dilatation maps in the interface area for the plane directions that are parallel and perpendicular to the interface. (e) SEM overview image showing the Zn3P2 nanowire morphology grown along <100> with respect to the InP substrate (45°). (f ) HAADF-STEM overview image of the nanowire cross section obtained from the nanowire in (e). (g) HAADF-STEM image of the detail in the nanowire-substrate interface area. The indexed power spectrum is reported in the inset to evaluate the mutual orientation of the two systems (scale bar 5 1 nm−1). (h) GPA rotational and dilatation maps in the interface area for the plane directions that are parallel and perpendicular to the interface. Table 1 Summary of the plane mismatches for the 0° nanowire, Table 2 Summary of the plane mismatches for the 45° nanowire, specifically (001) × [100]Zn3P2 on (001) × [110]InP specifically (001) × [110]Zn3P2 on (001) × [010]InP Plane mismatch Plane Mismatch Direction (001)Zn3P2//(001)InP (010)Zn3P2//(1–10)InP Direction (001)Zn3P2//(001)InP (1–10)Zn3P2//(100)InP Measured (GPA) −3.7% −1.6% Measured (GPA) −2.5% −2.1% Theory (bulk relaxed material) −2.8% −2.7% Theory (bulk relaxed material) −2.8% −2.7% Residual strain (nanowire) −0.9% +1.1% Residual strain (nanowire) +0.3% +0.6% Compressive Tensile Tensile Tensile by the SiO2 mask apertures, and they extend even further (see domains is sharp and parallel to the zone axis, as highlighted more images in Fig. S5†). Nonetheless, in all the multidomain in the representative model reported in Fig. 3e. The crystallo- defects observed for the 0° case, the interface between the two graphic directions for the two domains are reported, together 18444 | Nanoscale, 2021, 13, 18441–18450 This journal is © The Royal Society of Chemistry 2021 View Article Online Nanoscale Communication This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. Open Access Article. Published on 30 October 2021. Downloaded on 7/7/2022 4:18:39 PM. Fig. 3 (a) HAADF-STEM overview image of the cross-sectional FIB lamella. (b) HAADF-STEM overview image of the longitudinal FIB lamella. (c) HAADF-STEM image of the detail in the nanowire-substrate interface area for 0° orientation. In the inset the high-magnification image is reported (scale bar 1 nm). (d) Frequency filtered map of the inset in (c) showing the presence of two Zn3P2 main domains with different colour: [010]Zn3P2 in violet and [111]Zn3P2 in orange. In the inset, the power spectra for the two domains are reported with different reflections corresponding to each domain. 3D atomic model of the different domain interface from front and top view are shown in (e) and (f ), respectively. (g) HAADF-STEM image of the nanowire-substrate interface area for 45° orientation. In the inset a high-magnification image is included (scale bar 1 nm). (h) Frequency filtered map of the inset image in (c) highlighting the presence of two Zn3P2 rotated domains with different colour: [001]Zn3P2 in orange and [110]Zn3P2 in violet. In the inset the power spectra for the two domains are reported indicating different reflections corresponding to each domain (scale bar 2 1 nm−1). 3D atomic models of the domain interface from the front (i) and top (k). ( j) Shows a 3D atomic model of the (101) interface of two domains rotated 120° with respect to the interface. with the top view (in Fig. 3f ) highlighting the presence of a to {101}Zn3P2 surfaces, so for a complete picture we also sharp boundary along (010) × [100]Zn3P2 (main) and (1–12) × included this combination of rotated {101}Zn3P2 facets in [111]Zn3P2 (rotated), interfacing the different domain Fig. 3j. Although this combination remains a small proportion orientations. as it is only related to the nucleation site, and other interfaces Delving deeper into the 45° case, the observed domains are are formed during growth (see Fig. 4b). similarly highlighted in the frequency filtered map in Fig. 3h, It is possible to observe that for 0° and 45° orientations, the obtained from the Zn3P2 inside the SiO2 mask as indicated in majority of the Zn3P2 domains exhibit a <001>Zn3P2 direction the orange square in Fig. 3g. As previously shown, both the fre- perpendicular to the substrate, while the rotated domain quency filtered map and the power spectra show clearly the (<110>Zn3P2 direction perpendicular to the substrate) appears presence of rotated domains which are specifically oriented with a lower density all along the volume of the nanowire (see along [110]Zn3P2 (orange region) and [001]Zn3P2 (violet region) Fig. S5†). Animated movies of the rotational domain formation zone axes. In this case, the boundaries that separate the two for different nanowire orientations can be found in the follow- domains are not parallel to the zone axis, as for the 0° nano- ing: movie 1 (0° NW)90 and movie 2 (45° NW).91 In order to wires. Taking into account both, the image contrast and power further investigate and understand this, we reproduced all the spectrum analysis, here we propose that the domain boundary observed domains with 3D atomic models, with special atten- is transversal (45°) with respect to the zone axis, generating a tion on the Zn3P2–InP interface. By performing image simu- discontinuity in the structure that forms the two domains with lation and GPA analysis we could observe that out of the two orientations equivalent to [110] × (001) and [001] × (110), as it observed orientations, the domains with [001] as growth direc- is possible to observe in the model and corresponding orien- tion are the most favorable in terms of mismatch (as reported tation in Fig. 3i. Fig. 3k shows a top view of the domain bound- in Fig. S6†). ary, highlighting its orientation. The exact origin of these A previous study observing these domains in Zn3P2 nano- rotated domains is discussed in more detail below, but is pyramids argued that the formation of the rotated domains believed to be connected to the switching of the growth front was related to the growth plane switching from (001) to (101), This journal is © The Royal Society of Chemistry 2021 Nanoscale, 2021, 13, 18441–18450 | 18445 View Article Online Communication Nanoscale This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. Open Access Article. Published on 30 October 2021. Downloaded on 7/7/2022 4:18:39 PM. Fig. 4 (a) SEM image showing the formation of (101) faceted domains within the holes. (b) Diagram showing the proposed change in growth direc- tion when the growth occurs on (001) vs. (101) facets. (c) Atomic positions and electron distribution as used for the DFT calculations of two domains joined with a (101) plane and the upper domain rotated 120°, resulting in a stable interface without dangling bonds. (d) Density of states at the inter- face showing an overall decrease of the bandgap from 1.50 eV to 1.31–1.33 eV while no interband states are observed. (e) Atomic positions and elec- tron distribution as used for the DFT calculations of two domains rotated 120° around the (101) plane and joined by (100) and (1–12) facets. (f ) Density of states at the interface showing an overall decrease of the bandgap from 1.50 eV to 1.02 eV and low density of interband states at energies of 0.10 eV, 0.21 eV, and 0.38 eV above the valence band. as it was then only observed when it switched from axial to using non-relaxed surfaces, we observed slight changes in the lateral growth.16 However, we observe them within the nano- configuration, however, we did not observe any dangling wires themselves. We thus dwelled in the early stages of growth bonds and Ead for this scenario was −0.167 eV (Fig. S8†). The for a further understanding how the nanowire form. Fig. 4a final scenario we simulated for comparison was the joining of shows a representative SEM micrograph of a sample grown for two non-rotated domains. As expected, there were no dangling 15 minutes. We observe that small domains with what appears bonds formed and it is interesting to highlight that the calculated to be (101) facets have started to form. These domains are ran- Ead is comparable to the other scenarios, namely −0.175 eV. In domly distributed within the localized growth area, but they all fact, all these values are within the approach’s error of each other. seem to be at least partially touching the edge of the holes This has two consequences: (i) differences due to the use of (Fig. S7†). These domains thus provide the (101) facets for sim- relaxed or non-relaxed surfaces in this case are negligible, and (ii) ultaneous growth of regions with different rotation. This mecha- there is no significant penalty for forming a rotated interface and nism is illustrated in Fig. 4b. We propose that the domains of they exhibit a similar stability to a non-rotated interface. In order different orientation will compete during the selective area to further elucidate the exact conditions needed for the rotation growth, till they are completely blended, forming a nanofaceted to occur one would need to include the kinetics and growth rates, nanowire structure with the buried interfaces. through for example phase-field modelling. This could potentially In the hypothesis that the rotation would be strictly related show how growth conditions and parameters could be optimised to growth on (101) facets, one should also observe how they to minimise these domains, but this is beyond the scope of this rotate again eventually. However, this is not observed. In order study. As the domains grow they will also form extended to gain a deeper insight into the stability of these domains we (100)Zn3P2//(112)Zn3P2 interfaces between these domains. This com- turned to DFT calculations and investigated the difference in bination also has a very low Ead (−0.203 eV) as calculated by DFT, energetics and electronic structure when joining two Zn3P2 indicative of a high stability. domains at (101) facets as a function of the rotation (0° or Another interesting observation from the DFT results is the 120°), as shown in Fig. 4c, d and Fig. S8.† Fig. 4c shows the change in electronic structure around the interfaces of the configuration observed when two rotated domains (120°) with rotated domains as shown in Fig. 4d and Fig. S8.† The first relaxed (101) surfaces are brought into contact. Interestingly, thing to note is that there are no states within the bandgap for we observe that there should not be any dangling bonds at the the (101) interface, however, the (100)Zn3P2/(112)Zn3P2 case did interface. Furthermore, the calculated interface adhesive show a low density of shallow states at energies of 0.10 eV, 0.21 energy for this scenario (Ead) is found to be −0.179 eV. When eV and 0.38 eV above the valence band. Moreover, we do 18446 | Nanoscale, 2021, 13, 18441–18450 This journal is © The Royal Society of Chemistry 2021 View Article Online Nanoscale Communication observe a decrease in the overall bandgap. Compared to the faces. Finally, through the DFT calculations we could deter- bulk material where a DFT+U approach predicted the band mine that no dangling bonds or mid-gap states are formed at gap at 1.44 eV, a slight underestimation relative the HSE06 the rotated (101) interface, albeit the (100)Zn3P2//(112)Zn3P2 inter- functional (1.51 eV), the DFT+U estimated bandgap was 1.33 faces do exhibit shallow states at 0.10 eV, 0.21 eV, and 0.38 eV eV when relaxed surfaces were joined (1.31 eV for non-relaxed). according to the DFT calculations. In addition, we observe a A less significant shrinkage (1.37 eV) was observed for the shrinkage of the bandgap to approximately 1.31–1.33 eV for non-rotated domains. In the (100)Zn3P2//(112)Zn3P2 case the the (101) interface and 1.02 eV for the (100)Zn3P2//(112)Zn3P2 shrinkage was even greater, resulting in a bandgap of 1.02 eV. case. While there is still uncertainty associated with the energy Due to large number of atoms (640 in total) in the interface values calculated through DFT, this could potentially explain This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. system, the HSE06 functional could not be used. While the the transitions observed in this energy range by previous Open Access Article. Published on 30 October 2021. Downloaded on 7/7/2022 4:18:39 PM. accuracy of the DFT+U approach is not sufficient to establish studies. This study is a significant step towards the utilisation the exact energies, the overall trend still stands and is of great of selective area epitaxy to grow complex lattice-mismatched interest. Previous experimental studies of Zn3P2, including SAE structures outside of III–V materials and highlighting the capa- grown Zn3P2,16 have shown transitions or defect levels in the bilities of combining AC HAADF-STEM and DFT to fully under- range of 0.14–0.20 eV, 0.25–0.29 eV, and 0.36 eV below the stand the defect formation and impact in novel materials. bandgap, which puts them in the potential range of these rotated interfaces.6,16–18,36,37,39–44,72 While there may be other origins of transitions in this range acting in parallel, these Methods rotated domains are a potential source. With this in mind, we inspected the surface of zigzag nanowires grown by a vapour– The 30 nm thick SiO2 mask layer was deposited using plasma- liquid–solid (VLS) growth mechanism (as opposed to the enhanced chemical vapour deposition (PECVD, Oxford vapour–solid (VS) one of SAE) that are terminated with (101) Plasmalab System100) on InP(100) epi-ready substrates. Next, facets, but still exhibit sub-bandgap emission in a similar the pattern was created using electron-beam lithography range as to those calculated here.14,15,17,72 The VLS grown (Raith EBPG5000+) and etching using a combination of dry nanowires did contain small, rotated domains on the surface (fluorine plasma, SPTS APS) and a wet etch (BHF final dip for in crevices with an increased portion of radial VS growth 10s). The samples were then transferred to a Veeco GENxplor (Fig. S9†). No domains were observed in the VLS grown cores. molecular beam epitaxy system, where the substrates were fist These domains can therefore be attributed as a potential degassed for 1.5 h at 150 °C and 300 °C for 2 h before being source of the sub-bandgap emission observed from these introduced into the growth chamber. Additional growth details structures, however, the spatial resolution used in these can be found in ref. 16. studies was not sufficient to accurately attribute this emission TEM samples were prepared using focused ion beam (FIB) solely to these surface domains.17,72 processing. Before the samples were processed, a 20 nm thick SiO2 film was deposited using PECVD. A Focused Ion Beam HELIOS 600 FIB system was then used to create electron trans- Conclusions parent lamellae of longitudinal and transverse sections of the nanowires growing along the [110] (0°) and [010] (45°) of the InP Selective area epitaxy has been shown herein as a promising substrate. Aberration corrected HAADF-STEM was carried out approach to produce high-quality horizontal nanowires from using an FEI Titan probe corrected transmission electron micro- the earth-abundant compound semiconductor Zn3P2 in scope. Electron Energy Loss Spectroscopy in STEM mode various complex configurations. To properly assess these struc- (EELS-STEM) compositional maps have been obtained in the in tures, we turned to AC HAADF-STEM, combined with 3D a Tecnai F20 microscope by using a GATAN QUANTUM filter. atomic modelling and image simulation, which showed that 3D atomic models have been created by using Rhodius73,74 and there were no misfit dislocations forming at the interface, and STEM_CELL softwares,75–77 the latter was also used for STEM that all strain was accommodated elastically. The vertical image simulation. Geometrical phase analysis (GPA) has been growth direction was determined to be [001], which is the performed on simulated and experimental images by using the ideal growth direction to minimize interfacial strain. In licenced GPA plug-in available in Gatan Digital Micrograph. addition, we were able to observe domains rotated 120° within The density functional theory (DFT) calculations were per- the entire nanowires. These domains were formed when formed within periodic boundary conditions using the Vienna growth switched from (001)Zn3P2 facets to (101)Zn3P2 ones. Ab Initio Simulation Package (VASP).78–80 The projected aug- Further growth of the rotated domains resulted in extended mented wave (PAW) method81 was used to describe the inter- (100)Zn3P2//(112)Zn3P2 interfaces. DFT calculations showed that actions between the valence and cores electrons. The Perdew– the domain boundaries associated with this rotation are very Burke–Ernzerhof (PBE) generalized gradient approximation stable. Furthermore, we have been unable to observe a rotation (GGA) functional was used to calculate electronic exchange– back to the original orientation, indicating that the initial correlation potential.82 A plane wave cutoff energy of 600 eV rotation is a consequence of a change in growth kinetics and a Gamma-centred k-point mesh of 5 × 5 × 3 were found to related to the formation or disappearance of competing sur- converge the total energy of Zn3P2 to within 10−6 eV and the This journal is © The Royal Society of Chemistry 2021 Nanoscale, 2021, 13, 18441–18450 | 18447 View Article Online Communication Nanoscale residual Hellmann–Feynman forces on all relaxed atoms selective area growth and the substrate patterning reached 0.01 eV Å−1. A full unit cell relaxation yielded a strain- procedures. M. C. S., S. M. S., P. T. V. and J. A. acknowledge free Zn3P2 with lattice parameters a = b = 8.029 Å, c = 11.336 Å, funding from Generalitat de Catalunya 2017 SGR 327. ICN2 is which compares closely with available experimental data. supported by the Severo Ochoa programme from Spanish Given the fact that the 3d-electrons of transition metals are MINECO (Grant No. SEV-2017-0706) and is funded by the strongly correlated, methods beyond standard density func- CERCA Programme/Generalitat de Catalunya. M. C. Spadaro tional theory (DFT) are needed to correctly describe transition has received funding from the European Union’s Horizon 2020 metal-based materials in terms of electronic and magnetic research and innovation programme under the Marie properties.83–85 To accurately predict the electronic bandgap of Skłodowska-Curie grant agreement No. 754510 (PROBIST) and This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. Zn3P2, we have used both hybrid HSE06 functional with 25% the Severo Ochoa programme. This project has received Hartree–Fock exchange86 and Hubbard correction (DFT+U)87 Open Access Article. Published on 30 October 2021. Downloaded on 7/7/2022 4:18:39 PM. funding from the European Union’s Horizon 2020 research with Ueff value of 20 eV found for Zn ions. The hybrid HSE06 and innovation programme under grant agreement No 823717 functional gave a bandgap of 1.51 eV good agreement with – ESTEEM3. Authors acknowledge the use of instrumentation experiment whiles the DFT+U approach slightly underesti- provided by the National Facility ELECMI ICTS, node mates the bandgap at 1.44 eV. The DFT+U correction approach “Laboratorio de Microscopías Avanzadas” at University of adds an on-site Coulomb repulsion to the DFT Hamiltonian, Zaragoza. S. E. S., E. Z. S., M. Z., R. P., J. B. L. and A. F. i M. providing a better treatment of the strong electron correlation were supported by SNSF Consolidator grant BSCGI0-157705 in the localized Zn d-orbitals. Thus it predicts the electronic and the Max-Planck-EPFL-Center for Molecular Nanoscience band gap of Zn3P2 in closer agreement with the hybrid HSE06 and Technology. N. Y. D. acknowledges the UK Engineering functional compared to the standard DFT-PBE functional, and Physical Sciences Research Council (EPSRC) for funding which significantly underestimates the band gap at 0.67 eV. (Grant No. EP/S001395/1). The DFT calculations were per- The METADISE code,88 was used to create the (101), (100), and formed using the computational facilities of the Advanced (112) surfaces from the fully relaxed bulk material. To evaluate Research Computing @ Cardiff(ARCCA) Division, Cardiff the interface energetic and properties of domains formed with University. the (101) facets joined together such the upper domain is rotated 120°, a supercell approach was employed. For the (101) interface, the bottom domain was constructed with (3 × 2)– References (101) units, which is large enough to accommodate the growth of the 120° rotated top domain constructed with (2 × 1)–(101) 1 J. P. Bosco, G. M. Kimball, N. S. Lewis and H. A. Atwater, units. For the (100)Zn3P2/(112)Zn3P2 interface, the bottom J. Cryst. Growth, 2013, 363, 205–210. ((100)Zn3P2) and top ((112)Zn3P2) domains were constructed with 2 J. Long, J. Electrochem. Soc., 1983, 130, 725–728. (3 × 2) and (2 × 2) unit, respectively. Due to large unit cell size 3 T. Suda, K. Kakishita, H. Sato and K. Sasaki, Appl. Phys. and number of atoms in the interface system (640 for the (101) Lett., 1996, 69, 2426–2428. interface and 1200 for the (100)Zn3P2/(112)Zn3P2 interface), only a 4 M. Bhushan and A. Catalano, Appl. Phys. Lett., 1981, 38, 39– 1 × 1 × 1 mesh of k-point was use for geometry optimisation. 41. This was, however, increased to 3 × 3 × 1 for accurate determi- 5 J. Misiewicz, L. Bryja, K. Jezierski, J. Szatkowski, nation of the electronic structure where the DFT+U approach N. Mirowska, Z. Gumienny and E. Placzekpopko, was employed. The thermodynamic stability of the domain Microelectron. J., 1994, 25, R23–R28. interface was evaluated via interfacial adhesion energy, calcu- 6 R. Paul, N. Humblot, S. Escobar Steinvall, E. Z. Stutz, lated as: Ead = (Edomain1/domain2 − (Edomain1 + Edomain2)/S, where S. S. Joglekar, J.-B. Leran, M. Zamani, C. Cayron, R. Logé, Edomain1/domain2 is the total relaxed energies of the bottom/top A. G. del Aguila, Q. Xiong and A. Fontcuberta i Morral, domain interface structure with interface surface area S.89 Cryst. Growth Des., 2020, 20, 3816–3825. Edomain1 and Edomain2 represent the individual ground state 7 C. Wadia, A. P. Alivisatos and D. M. Kammen, Environ. Sci. relaxed total energy of the bottom (domain1) and top Technol., 2009, 43, 2072–2077. (domain2), respectively. 8 G. M. Kimball, A. M. Mueller, N. S. Lewis and H. A. Atwater, Appl. Phys. Lett., 2009, 95, 112103. 9 E. A. Fagen, J. Appl. Phys., 1979, 50, 6505–6515. Conflicts of interest 10 M. Y. Swinkels, A. Campo, D. Vakulov, W. Kim, L. Gagliano, S. Escobar Steinvall, H. Detz, M. De Luca, A. Lugstein, There are no conflicts to declare. E. Bakkers, A. Fontcuberta i Morral and I. Zardo, Phys. Rev. Appl., 2020, 14, 024045. 11 C. Pistorius, G. Clark, J. Ceotzer, G. Kruger and O. Kunze, Acknowledgements High Pressure High Temp., 1977, 9, 471–482. 12 M. Ghasemi, E. Stutz, S. Escobar Steinvall, M. Zamani and We acknowledge Martin Friedl, Didem Dede, Nicholas A. Fontcuberta i Morral, Materialia, 2019, 6, 100301. Morgan, and Wonjong Kim for helpful discussions regarding 13 U. Keppler, Int. J. Mater. Res., 1988, 79, 157–158. 18448 | Nanoscale, 2021, 13, 18441–18450 This journal is © The Royal Society of Chemistry 2021 View Article Online Nanoscale Communication 14 H. S. Im, K. Park, D. M. Jang, C. S. Jung, J. Park, S. J. Yoo 35 F. Briones, F. Wang and R. Bube, Appl. Phys. Lett., 1981, 39, and J.-G. Kim, Nano Lett., 2015, 15, 990–997. 805–807. 15 H. S. Kim, Y. Myung, Y. J. Cho, D. M. Jang, C. S. Jung, 36 T. Suda and R. Bube, Appl. Phys. Lett., 1984, 45, 775–777. J. Park and J.-P. Ahn, Nano Lett., 2010, 10, 1682–1691. 37 T. Suda and A. Kuroyanagi, Jpn. J. Appl. Phys., 1986, 25, 16 S. Escobar Steinvall, E. Z. Stutz, R. Paul, M. Zamani, L993–L995. N. Y. Dzade, V. Piazza, M. Friedl, V. de Mestral, J.-B. Leran, 38 E. Z. Stutz, S. Escobar Steinvall, A. P. Litvinchuk, R. R. Zamani and A. Fontcuberta i Morral, Nanoscale Adv., J.-B. Leran, M. Zamani, R. Paul, A. Fontcuberta i Morral 2021, 3, 326–332. and M. Dimitrievska, Nanotechnology, 2020, 32, 085704. 17 S. Escobar Steinvall, N. Tappy, M. Ghasemi, R. R. Zamani, 39 N. Mirowska and J. Misiewicz, Semicond. Sci. Technol., This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. T. LaGrange, E. Z. Stutz, J.-B. Leran, M. Zamani, R. Paul 1992, 7, 1332–1336. and A. Fontcuberta i Morral, Nanoscale Horiz., 2020, 5, 40 J. Misiewicz, J. Phys. Chem. Solids, 1989, 50, 1013–1022. Open Access Article. Published on 30 October 2021. Downloaded on 7/7/2022 4:18:39 PM. 274–282. 41 J. Misiewicz, B. Sujak-Cyrul and A. Bartczak, Solid State 18 M. Zamani, E. Z. Stutz, S. Escobar Steinvall, R. R. Zamani, Commun., 1986, 58, 677–679. R. Paul, J.-B. Leran, M. Dimitrievska and A. Fontcuberta i 42 A. Möller, U. Elrod, P. Munz, J. Hönigschmid, C. Clemen Morral, J. Phys. Energy, 2021, 3(3), 034011. and E. Bucher, Inst. Phys. Conf. Ser., 1979, 43, 825–828. 19 T. Suda and K. Kakishita, J. Appl. Phys., 1992, 71, 3039– 43 F.-C. Wang, A. L. Fahrenbruch and R. H. Bube, J. Electron. 3041. Mater., 1982, 11, 75–88. 20 S. Jeon, J. P. Bosco, S. S. Wilson, S. J. Rozeveld, H. Kim and 44 K. Sierańki and J. Szatkowski, Phys. Status Solidi A, 1986, H. A. Atwater, J. Phys. Chem. C, 2014, 118, 12717–12726. 94, K133–K135. 21 R. Katsube, K. Kazumi, T. Tadokoro and Y. Nose, ACS Appl. 45 S. A. Mann, R. R. Grote, R. M. Osgood Jr., A. Alu and Mater. Interfaces, 2018, 10, 36102–36107. E. C. Garneet, ACS Nano, 2016, 10, 8620–8631. 22 F. Glas, Phys. Rev. B: Condens. Matter Mater. Phys., 2006, 74, 46 P. Krogstrup, H. I. Jørgensen, M. Heiss, O. Demichel, 121302. J. V. Holm, M. Aagesen, J. Nygard and A. Fontcuberta i 23 T. Særkjær, T. C. Thann, T. Stankevič, S. Schuwalow and Morral, Nat. Photonics, 2013, 7, 306–310. P. Krogstrup, https://arxiv.org/abs/2102.03587v1. 47 R. Frederiksen, G. Tutuncuoglu, F. Matteini, K. L. Martinez, 24 C.-Y. Chi, C.-C. Chang, S. Hu, T.-W. Yeh, S. B. Cronin and A. Fontcuberta i Morral and E. Alarcon-Llado, ACS P. D. Dapkus, Nano Lett., 2013, 13, 2506–2515. Photonics, 2017, 4, 2235–2241. 25 G. Tutuncuoglu, M. de la Mata, D. Deiana, H. Potts, 48 M. Heiss, E. Russo-Averchi, A. Dalmau-Mallorquí, F. Matteini, J. Arbiol and A. Fontcuberta i Morral, G. Tütüncüoğlu, F. Matteini, D. Rüffer, S. Conesa-Boj, Nanoscale, 2015, 7, 19453–19460. O. Demichel, E. Alarcon-Lladó and A. Fontcuberta i Morral, 26 G. E. Cirlin, V. G. Dubrovskii, I. P. Soshnikov, N. V. Sibirev, Nanotechnology, 2013, 25, 014015. Yu. B. Samsonenko, A. D. Bouravleuv, J. C. Harmand and 49 E. C. Garnett, B. Ehrler, A. Polman and E. Alarcon-Llado, F. Glas, Phys. Status Solidi RRL, 2009, 3, 112–114. ACS Photonics, 2021, 8, 61–70. 27 L. Güniat, P. Caroff and A. Fontcuberta i Morral, Chem. 50 N. Guan, X. Dai, A. V. Babichev, F. H. Julien and Rev., 2019, 119, 8958–8971. M. Tchernycheva, Chem. Sci., 2017, 8, 7904–7911. 28 P. C. McIntyre and A. Fontcuberta i Morral, Mater. Today 51 M. H. Huang, S. Mao, H. Feick, H. Yan, Y. Wu, H. Kind, Nano, 2020, 9, 100058. E. Weber, R. Russo and P. Yang, Science, 2001, 292, 1897. 29 M. Friedl, K. Cerveny, P. Weigele, G. Tütüncüoglu, S. Martí- 52 D. Saxena, S. Mokkapati, P. Parkinson, N. Jiang, Q. Gao, Sánchez, C. Huang, T. Patlatiuk, H. Potts, Z. Sun, H. H. Tan and C. Jagadish, Nat. Photonics, 2013, 7, 963–968. M. O. Hill, L. Güniat, W. Kim, M. Zamani, V. G. Dubrovskii, 53 D. van Dam, N. J. J. van Hoof, Y. Cui, P. J. van Veldhoven, J. Arbiol, L. J. Lauhon, D. M. Zumbühl and A. Fontcuberta i E. P. A. M. Bakkers, J. Gómez Rivas and J. E. M. Haverkort, Morral, Nano Lett., 2018, 18, 2666–2671. ACS Nano, 2016, 10, 11414–11419. 30 P. Aseev, A. Fursina, F. Boekhout, F. Krizek, J. E. Sestoft, 54 I. Aberg, G. Vescovi, D. Asoli, U. Naseem, J. P. Gilboy, F. Borsoi, S. Heedt, G. Wang, L. Binci, S. Martí-Sánchez, C. Sundvall, A. Dahlgren, K. E. Svensson, N. Anttu, T. Swoboda, R. Koops, E. Uccelli, J. Arbiol, P. Krogstrup, M. T. Bjork and L. Samuelson, IEEE J. Photovoltaics, 2016, L. P. Kouwenhoven and P. Caroff, Nano Lett., 2019, 19, 218–227. 6, 185–190. 31 L. Desplanque, A. Bucamp, D. Troadec, G. Patriarche and 55 M. D. Kelzenberg, S. W. Boettcher, J. A. Petykiewicz, X. Wallart, J. Cryst. Growth, 2019, 512, 6–10. D. B. Turner-Evans, M. C. Putnam, E. L. Warren, 32 X. Yuan, D. Pan, Y. Zhou, X. Zhang, K. Peng, B. Zhao, J. M. Spurgeon, R. M. Briggs, N. S. Lewis and H. A. Atwater, M. Deng, J. He, H. H. Tan and C. Jagadish, Appl. Phys. Rev., Nat. Mater., 2010, 9, 239–244. 2021, 8, 021302. 56 R. R. Zamani, F. S. Hage, S. Lehmann, Q. M. Ramasse and 33 P. Staudinger, K. E. Moselund and H. Schmid, Nano Lett., K. A. Dick, Nano Lett., 2018, 18, 1557–1563. 2020, 20, 686–693. 57 R. R. Zamani and J. Arbiol, Nanotechnology, 2019, 30, 34 J. Seidl, J. G. Gluschke, X. Yuan, S. Naureen, N. Shahid, 262001. H. H. Tan, C. Jagadish, A. P. Micolich and P. Caroff, Nano 58 V. Grillo, E. Carlino and F. Glas, Phys. Rev. B: Condens. Lett., 2019, 19, 4666–4677. Matter Mater. Phys., 2008, 77, 054103. This journal is © The Royal Society of Chemistry 2021 Nanoscale, 2021, 13, 18441–18450 | 18449 View Article Online Communication Nanoscale 59 J. M. LeBeau, S. D. Findlay, L. J. Allen and S. Stemmer, 74 J. Arbiol, A. Cirera, F. Peiró, A. Cornet, J. R. Morante, Phys. Rev. Lett., 2008, 100, 206101. J. J. Delgado and J. J. Calvino, Appl. Phys. Lett., 2002, 80, 60 D. B. Williams and C. B. Carter, Transmission electron 329–331. microscopy, Springer, 1996. 75 V. Grillo and E. Rotunno, Ultramicroscopy, 2013, 125, 97– 61 M. de la Mata, R. R. Zamani, S. Martí-Sánchez, M. Eickhoff, 111. Q. Xiong, A. Fontcuberta i Morral, P. Caroff and J. Arbiol, 76 V. Grillo and F. Rossi, Ultramicroscopy, 2013, 125, 112–129. Nano Lett., 2019, 19, 3396–3408. 77 M. C. Spadaro, S. D’Addato, P. Luches, S. Valeri, V. Grillo, 62 M. de la Mata, C. Magen, J. Gazquez, M. I. B. Utama, E. Rotunno, M. A. Roldan, S. J. Pennycook, A. M. Ferretti, M. Heiss, S. Lopatin, F. Furtmayr, C. J. Fernández-Rojas, E. Capetti and A. Ponti, Nanotechnology, 2015, 26, 405704. This article is licensed under a Creative Commons Attribution 3.0 Unported Licence. B. Peng, J. R. Morante, R. Rurali, M. Eickhoff, 78 G. Kresse and J. Hafner, Phys. Rev. B: Condens. Matter Mater. Phys., 1993, 47, 558–561. Open Access Article. Published on 30 October 2021. Downloaded on 7/7/2022 4:18:39 PM. A. Fontcuberta i Morral, Q. Xiong and J. Arbiol, Nano Lett., 2012, 12, 2579–2586. 79 G. Kresse and J. Hafner, Phys. Rev. B: Condens. Matter 63 E. Oveisi, M. C. Spadaro, E. Rotunno, V. Grillo and Mater. Phys., 1994, 49, 14251–14269. C. Hébert, Ultramicroscopy, 2019, 200, 139–148. 80 G. Kresse and D. Joubert, Phys. Rev. B: Condens. Matter 64 N. Y. Dzade, Phys. Chem. Chem. Phys., 2020, 22, 1444–1456. Mater. Phys., 1999, 59, 1758–1775. 65 J. Arbiol, S. Estradé, J. D. Prades, A. Cirera, F. Furtmayr, 81 P. E. Blöchl, Phys. Rev. B: Condens. Matter Mater. Phys., C. Stark, A. Laufer, M. Stutzmann, M. Eickhoff, M. H. Gass, 1994, 50, 17953–17979. A. L. Bleloch, F. Peiró and J. R. Morante, Nanotechnology, 82 J. P. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett., 2009, 20, 145704. 1997, 78, 1396–1396. 66 M. J. Hÿtch, E. Snoeck and R. Kilaas, Ultramicroscopy, 1998, 83 A. I. Liechtenstein, V. I. Anisimov and J. Zaanen, Phys. Rev. 74, 131–146. B: Condens. Matter Mater. Phys., 1995, 52, R5467–R5470. 67 Y. Zhu, C. Ophus, J. Ciston and H. Wang, Acta Mater., 2013, 84 L. Wu, N. Y. Dzade, L. Gao, D. O. Scanlon, Z. Öztürk, 61, 5646–5663. N. Hollingsworth, B. M. Weckhuysen, E. J. M. Hensen, 68 M. de la Mata, C. Magén, P. Caroff and J. Arbiol, Nano Lett., N. H. de Leeuw and J. P. Hofmann, Adv. Mater., 2016, 28, 2014, 14, 6614–6620. 9602–9607. 69 M. de la Mata, R. Leturcq, S. R. Plissard, C. Rolland, C. Magén, 85 N. Y. Dzade, A. Roldan and N. H. De Leeuw, Minerals, 2014, J. Arbiol and P. Caroff, Nano Lett., 2016, 16, 825–833. 4, 89–115. 70 F. Krizek, J. E. Sestoft, P. Aseev, S. Marti-Sanchez, 86 A. V. Krukau, O. A. Vydrov, A. F. Izmaylov and S. Vaitiekėnas, L. Casparis, S. A. Khan, Y. Liu, T. Stankevič, G. E. Scuseria, J. Chem. Phys., 2006, 125, 224106. A. M. Whiticar, A. Fursina, F. Boekhout, R. Koops, 87 S. L. Dudarev, G. A. Botton, S. Y. Savrasov, C. J. Humphreys E. Uccelli, L. P. Kouwenhoven, C. M. Marcus, J. Arbiol and and A. P. Sutton, Phys. Rev. B: Condens. Matter Mater. Phys., P. Krogstrup, Phys. Rev. Mater., 2018, 2, 093401. 1998, 57, 1505–1509. 71 A. Iandelli, Gazz. Chim. Ital., 1940, 70, 58–62. 88 G. W. Watson, E. T. Kelsey, N. H. de Leeuw, D. J. Harris and 72 S. Escobar Steinvall, L. Ghisalberti, R. R. Zamani, N. Tappy, S. C. Parker, J. Chem. Soc., Faraday Trans., 1996, 92, 433– F. S. Hage, E. Z. Stutz, M. Zamani, R. Paul, J.-B. Leran, 438. Q. M. Ramasse, W. C. Carter and A. Fontcuberta i Morral, 89 P. K. Baviskar, S. R. Rondiya, G. P. Patil, B. R. Sankapal, Nanoscale, 2020, 12, 22534–22540. H. M. Pathan, P. G. Chavan and N. Y. Dzade, ACS Omega, 73 S. Bernal, F. J. Botana, J. J. Calvino, C. López-Cartes, 2020, 5, 6715–6724. J. A. Pérez-Omil and J. M. Rodríguez-Izquierdo, 90 https://youtu.be/-LaLd_kdL9s. Ultramicroscopy, 1998, 72, 135–164. 91 https://youtu.be/IfiWEqu1Onw. 18450 | Nanoscale, 2021, 13, 18441–18450 This journal is © The Royal Society of Chemistry 2021