Resolving Few-Layer Antimonene/Graphene Heterostructures

Two-dimensional (2D) antimony (Sb, antimonene) recently attracted interest due to its peculiar electronic properties and its suitability as anode material in next generation batteries. Sb however exhibits a large polymorphic/allotropic structural diversity, which is also influenced by the Sb's support. Thus understanding Sb heterostructure formation is key in 2D Sb integration. Particularly 2D Sb/graphene interfaces are of prime importance as contacts in electronics and electrodes in batteries. We thus study here few-layered 2D Sb/graphene heterostructures by atomic-resolution (scanning) transmission electron microscopy. We find the co-existence of two Sb morphologies: First is a 2D growth morphology of layered beta-Sb with beta-Sb(001)||graphene(001) texture. Second are one-dimensional (1D) Sb nanowires which can be matched to beta-Sb with beta-Sb[2-21] perpendicular to graphene(001) texture and are structurally also closely related to thermodynamically non-preferred cubic Sb(001)||graphene(001). Importantly, both Sb morphologies show rotational van-der-Waals epitaxy with the graphene support. Both Sb morphologies are well resilient against environmental bulk oxidation, although superficial Sb-oxide layer formation merits consideration, including formation of novel epitaxial Sb2O3(111)/beta-Sb(001) heterostructures. Exact Sb growth behavior is sensitive on employed processing and substrate properties including, notably, the nature of the support underneath the direct graphene support. This introduces the substrate underneath a direct 2D support as a key parameter in 2D Sb heterostructure formation. Our work provides insights into the rich phase and epitaxy landscape in 2D Sb and 2D Sb/graphene heterostructures.


Results and Discussions
Morphology and Structure. We first characterize in Fig. 1 the morphology and structure of the few layer Sb on graphene model system, which is prepared by physical vapor deposition (PVD) of Sb onto chemical vapor deposited (CVD) monolayer graphene. We first focus on optimized Sb deposition conditions towards high Sb crystallinity, with the wider parameter space of the Sb PVD on graphene being discussed further below. The nominal 10 nm thick Sb deposits in Figs. 1-4 were thermally evaporated onto monolayered CVD graphene films. 61,62 During Sb PVD (base pressure ~10 -5 mbar), the graphene substrates were held at room temperature (RT) and also at controlled temperatures of 150 °C and 250 °C. The graphene either remained on its Cu CVD catalyst foils 61,62 during Sb PVD (Fig. 1a,b, Fig. 2) or was additionally also transferred prior to Sb PVD to be suspended as a freestanding monolayer membrane across holey TEM grids 63 ((i.e. no Cu foils underneath, Fig. 1c-g, Figs. [3][4]. Nominal deposited Sb thickness was measured via a co-exposed (non-heated) quartz crystal microbalance. After Sb deposition, samples were stored in ambient air. Further details on methods can be found in the Supporting Information.
The scanning electron microscopy (SEM) image of the 250 °C deposition in Fig. 1a reveals that under our optimized PVD conditions, the Sb deposits on the graphene form isolated islands with two distinctly different base shapes: First are flat 2D Sb deposits with (truncated) hexagonal or (truncated) triangular base shapes. Second are rod-like 1D Sb deposits with rectangular bases. Lateral extents of all Sb deposits are in the range of tens to hundreds of nm. While such lateral sizes are too small for device fabrication in 2D electronics, they are compatible with the requirements for 2D Sb catalysis and energy applications. 22 Importantly, such feature sizes also provide a large enough Sb/graphene heterostructure model system for convenient elucidation of Sb phases and interfacing with high resolution (S)TEM. Notably, as shown in Fig. 1a, the edges of both the triangular-/hexagonal-shaped and the rod-shaped Sb deposits show a high degree of visually apparent directional alignment amongst each phase type, respectively. This is a first indication of potential epitaxy effects between our Sb deposits and their graphene support and will be further examined below.
The Raman spectrum corresponding to the 250 °C deposition in Fig. 1b displays primarily two peaks at low wavenumbers that are characteristic for elemental Sb (117 cm -1 ; 154 cm -1 ). These peaks are best matched with Eg and A1g modes of few layer β-Sb, respectively, but are also potentially consistent with α-Sb and/or pressure-induced phases of Sb. 8,31,33,46,54 We note that thicker Sb deposits may be overrepresented in Raman intensity. 42,43 Significant volume Sb-oxide formation can be excluded based on our Raman data as the signal intensity at wavenumbers corresponding to Sb-oxides is comparatively weak (e.g. for thermodynamically most stable Sb2O3 expected at ~190 cm -1 and 250 cm -1 ). 64 Raman peaks characteristic for graphene (G at 1593 cm -1 ; and 2D at 2701 cm -1 ) are also found in Fig. 1b, consistent with the high quality CVD graphene used as substrate. 61,62 The absence of a significant defect-related D-peak at ~1350 cm -1 confirms that the CVD graphene support was not degraded during Sb PVD. Thereby our Raman data also confirms that no covalent Sb-carbon bond formation has occurred and that our 2D Sb/graphene interfaces are of vdW-type, 38 consistent with theoretical predictions. 8,57,58 To assess the crystallographic structure of the Sb deposits in a localized fashion, we employ in Fig. 1c-g aberration-corrected, atomically-resolved and element-specific STEM (Nion UltraSTEM 100 at 60 kV electron acceleration voltage) in annular dark field (ADF) 60 mode to image individual Sb deposits at high resolution in top plan view. Corresponding ADF STEM and bright-field (BF) TEM data from focused-ion-beam (FIB) cross-sections in Fig. 2 provide a complementary side view of the Sb deposits. Supporting Figs. S1 and S2 provides atomic models and Fourier transform (FT)/selected area electron diffraction (SAED) simulations of all identified phases.
Comparing our results with prior literature we note that overall morphology and size of our Sb domains on carbon substrates are consistent not only with vacuum-based vapor deposition techniques as usually used in electronics 38,48,50,51 (akin to our PVD synthesis) but also with several wet-chemistry synthesis routes (incl. using SbCl3 12,14,19,20 and ballmilled and annealed Sb/carbon mixtures 15 ) as usually used in energy materials synthesis. This highlights that our here investigated 2D Sb/graphene heterostructure model system is relevant to a wide range of synthesis conditions and electronics and energy-related application profiles of Sb on carbon.
In terms of application potential, we note that trigonally deformed Sb (like simple cubic Sb) has recently been predicted to feature superior thermoelectrical performance over β-Sb. 56 Given that monolayered 2D β-Sb has been predicted to surpass all other pristine 2D materials in terms of thermoelectric performance, 27 future studies on band structure and electronic properties of the here observed β-Sb [2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20][21]/cubic Sb(001) deposits merit consideration.  Van-der-Waals Epitaxy. So far our data has shown that we have grown 2D Sb/graphene heterostructures, where the Sb deposits are comprised of two co-existing morphologies, namely few-layer 2D β-Sb(001) and 1D nanorods β-Sb [2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20][21]/cubic Sb(001). Importantly, for both these Sb morphologies Figs. 1a indicated a high degree of directional alignment of their respective domain edges on the monolayer graphene support. Given the vdW nature of the Sb/graphene interface (Fig. 1b), three mechanisms could contribute to such alignment: First is direct rotational vdW epitaxy between the growing Sb and its graphene support directly underneath. 71 For 2D β-Sb(001) direct epitaxial relationships with various substrates have been reported incl., e.g., WSe2, 39 tellurides, 32,36 mica 33 and Ge. 35 Particularly, for β-Sb(001) on graphene prior work has given a mixed picture: Some work 38 reported rotational vdW epitaxy for β-Sb/graphene via indirect measurements, while other work observed no such epitaxy. 31,44,50,51 For the 1D nanorod β-Sb [2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20][21]/cubic Sb(001), epitaxial effects have to date not been reported. 31,51 Therefore, the question if direct vdW epitaxy is prevalent in the Sb/graphene system remains open. Secondly however, complicating elucidation of this question, also recently reported "remote" epitaxy needs consideration in which epitaxy is impressed "remotely" between a deposit and its underlying substrate through an intermediate 2D layer. 72 In the present work, this would involve interactions between Sb and the underlying Cu catalyst foils impressed through the graphene monolayer. 38 Notably, in this scenario the graphene could also be required to act as a diffusion barrier to prevent chemical reactions between Sb and Cu, 73,74 thus actually actively facilitating the remote epitaxy. Third, in contrast to the atomic-scale epitaxy, the last possibility involves macroscopic corrugations on the support (e.g. Cu surface steps) that result in alignment via preferred heterogeneous nucleation sites (e.g. at steps) and diffusion directing effects. 75 From SEM data as in Fig. 1a alone, these three possible causes of the observed Sb alignments are hard to disentangle: First, direct vdW epitaxy would be readily compatible with the observed lateral length scales of alignment in Fig. 1a as the lateral size of our CVD graphene domains is in the tens of μm range. 61,62,68 Therefore the field of view in Fig.  1a represents most likely only one single-crystalline graphene domain (although not confirmable by SEM) which could facilitate rotational alignment over the entire field of view. Second however, graphene-mediated remote epitaxy between Sb and Cu is also conceivable for Fig. 1a, since the Cu grain sizes in our Cu foils after graphene CVD are in the mm-range. 61,62 Notably, no Cu grain boundary is visible in Fig. 1a, 62 thus confirming a single Cu orientation across the field of view in Fig. 1a. However, as we show in Supporting Fig. S5, direct deposition of Sb on Cu (i.e. without graphene in between) does not show any indications of epitaxial order in the Sb deposits under our conditions. 73 Nevertheless, a graphene-mediated remote epitaxy mechanism between Sb and Cu 38 cannot be excluded based on Fig. 1a. The third possibility, i.e. surface corrugations on the Cu support, could also direct the Sb deposits, although not resolvable in Fig 1a. To disentangle these three possible influences, we investigate in Fig. 3 the relative orientation of Sb deposits at 150 °C and 250 °C directly onto suspended monolayer graphene membranes, i.e. without Cu foil underneath. In doing so, we exclude any possible indirect influence of Cu underneath the graphene on the Sb alignment (i.e. we exclude "remote" epitaxy and an influence from Cu surface corrugations). In particular, we correlate the STEM-derived orientation (via FT analysis) of the lattice of the Sb deposits (β-Sb(001): Fig. 2a-g; cubic Sb(001): Fig. 2i-n) with the underlying graphene lattice orientation measured adjacent to the Sb deposit within a few nm distance 68 for multiple Sb deposits of both morphologies. Via plotting histograms of the relative rotational (mis)orientations of the graphene [120] orientation and prominent orientations in the two respective Sb lattices (β-Sb [110]: Fig. 2g; cubic Sb [110]: Fig. 2n) we find clear peaks in the (mis)orientation distributions for both Sb phases. This unambiguously suggests direct epitaxy effects to be present between the graphene and both Sb phases. In particular 2D β-Sb shows a preferred misorientation of ~0° between the graphene [120] and the β-Sb Prior work has investigated possible epitaxy between Sb polymorphs and graphene (and graphite) with mixed results: Early work did not find evidence for epitaxy in β-Sb/graphite (but had only limited statistics measured). 51 Also recent other studies did not observe epitaxy in β-Sb/graphene. 31,44 In contrast, another recent study of β-Sb and Cu-supported graphene suggested epitaxy for β-Sb/graphene to exist based on indirect measurements, identifying two preferred orientations of (in our notation) 0° and 30° offset between β-Sb [110] and graphene [120]. 38 This is in good agreement with our findings in Fig. 3g which are based on direct observations of the β-Sb/graphene interface. For the β-Sb[2-21]/cubic Sb(001)/graphite system no evidence for epitaxy has been reported prior. 31,51 In contrast, we here find strong evidence also for rotational vdW epitaxy in the β-Sb [2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20][21]/cubic Sb(001)/graphene system. Combined, our observations show that vdW epitaxy can be enforced on 2D and 1D Sb deposits on graphene. Oxidation Susceptibility. After having identified the nature of our Sb deposits and their relation to the Graphene support, we turn to the oxidation susceptibility of our Sb deposits. Oxidation susceptibility is of significant importance in terms of processing and applications. Additionally, 2D Sb-oxides are beginning to attract research interest in their own right. 64,76,77 While Raman spectroscopy in Fig. 1b did not suggest significant Sb-oxide presence in our samples, close inspection of the β-Sb(001) FT in Fig. 1d reveals another, weaker intensity set of spots of six-fold symmetry at lower k-vectors (indexed "*" at ~0.4 nm) than the six-fold (110) β-Sb(001) spot family which is indexed in the FT. These weak inner spots may be identified with the presence of cubic Sb2O3 viewed along the [111] zone axis 64 (i.e. cubic Sb2O3(111), see Supporting Fig. S1 and S2). This poses the question whether our β-Sb deposits are partly and/or superficially oxidized during sample storage in ambient air. Some prior work has reported stability of antimonene against oxidation in the ambient conditions 33,35,36,42,43,46,78 but other work has suggested thin antimony oxide present around Sb structures to be also prevalent. 22,23,26,38,[79][80][81] Notably, for the β-Sb [2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20][21]/cubic Sb(001) deposits no signs of additional crystalline oxides are found in (S)TEM or FT data.
To investigate possible oxidation effects for our Sb deposits in a localized fashion we use chemical identification via electron energy loss spectroscopy (EELS). In EELS of Sb/Sboxide mixtures, compositional analysis based on the commonly used EELS core loss regions is however difficult since the core loss Sb M4,5 edge at ~528 eV (which follows a delayed maxima fashion) is very close/partially overlapping the O K-edge at ~532 eV (Supporting Fig. S6). 82,83 An alternative approach is investigating the valence EELS (VEELS) low loss region in which a sharp bulk plasmon peak at ~16.8 eV is related to metallic Sb, 84 while the plasmon peak shifts for Sb-oxides to a distinctly higher energy of ~22eV. 82 For VEELS, however, also the graphene support (and, if present, amorphous carbon TEM grid membrane) has to be considered with plasmon signatures at ~27 eV. 85 In Fig. 4a a typical VEELS spectrum acquired from a flat 2D β-Sb crystal (inset) is shown. We note that this particular Sb deposit was characterized by VEELS after ~8 months of ambient air storage, thus allowing us to probe long-term resilience against oxidation. We find that a sharp metallic Sb VEELS peak is dominating the fitted VEELS spectrum with only a small contribution of the Sb-oxide component present even after the long term air exposure. VEELS data for β-Sb [2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20][21]/cubic Sb(001) deposits shows similar results. The VEELS findings are thereby in agreement with the Raman data in Fig. 1b that suggested metallic Sb to be dominant in our deposits. The VEELS results however suggest the possibility of a very thin superficial oxide layer (which might be below the detection limit for Raman). This further implies for the β-Sb(001) that the inner reflections (labelled "*") in the FT in Fig. 1d may be indeed related to a very thin crystalline cubic Sb2O3(111) overlayer on the elemental 2D β-Sb(001) crystal (where reflection "*" in Fig. 1d corresponds to the (2-20) reflection in Sb2O3, see also Supporting Figs. S1 and S2). This suggests the possibility of intrinsic Sb-oxide/Sb/graphene heterostructure formation from simple ambient air exposure. In particular, whenever present, the six-fold Sb2O3 (2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20) reflection family consistently has a rotational misorientation of ~30 ° with the sixfold β-Sb (110) reflection family (as in Fig. 1d). This indicates an epitaxial relation Sb2O3(111)||β-Sb(001)/Sb2O3 [2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20]||β-Sb[110] for the Sb-oxide/Sb interface. See Fig. 4b for an atomic model of the suggested Sb-oxide/Sb heterostructure. Unfortunately, the top regions in the β-Sb flakes in our cross-section (S)TEM data are all not well enough resolved (due to Pt/C protection layers) to finally fully confirm the suggested presence of this ultrathin epitaxial Sb2O3 overlayer. We note however that prior x-ray photoelectron spectroscopy measurements of 2D Sb oxidation found oxide stoichiometries consistent with the here suggested crystalline Sb2O3 phase, 80 and that our core loss EELS in Supporting Fig. S6 also is best matched with Sb2O3 stoichiometry. No such crystalline overlayers are suggested from our data for the β-Sb [2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20][21]/cubic Sb(001), albeit an amorphous Sb-oxide overlayer could be present. Combined, our microscopic and spectroscopic data shows that while our Sb deposits are overall well resilient against environmental bulk oxidation, the possibility of superficial oxidation in ambient air still requires consideration, in particular since for β-Sb(001) deposits the formation of an epitaxial Sb2O3(111) oxide overlayer is inferred from our data. Growth Parameter Space. Finally, we examine the wider parameter space of our Sb PVD. Fig. 5 compares Sb deposition results of nominally 10 nm Sb (regulated via a co-exposed (non-heated) quartz crystal microbalance) as function of substrate temperature from RT to 250 °C onto CVD graphene-covered Cu foils (Fig. 5a,c,e, SEM) as well as directly onto suspended graphene membranes (no Cu underneath, Fig. 5b,d,f, TEM). Fig. 5 shows that the Sb deposit morphology drastically changes between RT and elevated temperature (150 °C, 250 °C) depositions: For RT depositions (Fig. 5a,b) merged (truncated) (semi-)spherical features dominate. High resolution STEM in Supporting Fig. S7 shows that these RT-deposited (truncated) (semi-)spheres are amorphous. For 150 °C depositions (Fig. 5c,d) the above described triangular/hexagonal shaped 2D β-Sb and rod-shaped cubic Sb crystals along with few (semi-)spherical Sb deposits are found. Among the 2D β-Sb deposits the hexagonal base shape is more prevalent. For 250 °C depositions (Fig. 5e,f) practically only triangular/hexagonal shaped 2D β-Sb and rod-shaped cubic Sb are found, whereby now among the 2D β-Sb triangles dominate. Notably, for neither 150 °C nor 250 °C we find evidence for an underlying continuous Sb layer on neither Cu-supported nor freestanding graphene, the former in contrast to prior literature. 38 Besides deposit morphology, also coverage and retained Sb amount of the nominally 10 nm Sb deposits is strongly influenced by substrate temperature during Sb deposition and, notably, also strongly dependent on substrate-type. In particular, Sb coverage and retained Sb amount strongly decrease with increasing substrate temperature. For RT depositions a homogeneous coverage close to 100 % is achieved on both Cu-supported and freestanding Graphene in Figs. 5a,b and for RT samples good agreement between nominal 10 nm thickness and AFM-calibrated average deposit thickness was found. In comparison, the coverage for 150 °C and 250 °C depositions decreases, whereby the coverage decrease with substrate temperature is even more prominent on the freestanding Graphene (150 °C~40 %, 250 °C: <5 %) than on the Cu-supported Graphene (150 °C: ~40%, 250 °C: ~20%). For Sb deposits with an average thickness of 21 ± 14 nm at 250 °C (see above) this equates to a reduction in Sb amount deposited from RT to 250 °C of ~50 % on Cu-supported Graphene and of ~90 % on freestanding Graphene, respectively. Notably, also the size of individual deposits of Cu is significantly larger than on the suspended graphene, best seen in the 250 °C depositions (Fig. 5e,f). Combined, this suggests a key influence of temperature dependent desorption processes on Sb nanostructure growth. 35,37,[47][48][49][50][51] In particular, the balance of Sb adsorption flux (FSb,ad) from the evaporation source onto the graphene substrate and a substrate-temperaturedependent Sb desorption flux from the graphene substrate into vacuum (FSb,de(T)) is key: The observed Sb morphologies imply that at RT FSb,ad>>FSb,de(RT) resulting in strong deposition, while the low temperature hinders crystallization of the resulting Sb deposits (possibly via incomplete fragmentation of physisorbed Sb4 species which are the preferred arriving Sb vapor species 51 ). This leads to the observed fully covering amorphous Sb deposits at RT. Increasing substrate temperature leads to a strong increase in FSb, de(150-250 °C), reducing the net retained amount of Sb at higher substrate temperature. In turn the higher substrate temperatures facilitate crystallization of the retained Sb deposits (possibly via thermally activated fragmentation of surface-bound Sb species 51 and thus increased Sb re-arrangement). Thereby we grow crystalline Sb deposits with an onset temperature of crystallization of ~150 °C. The observation that this temperature dependence is more pronounced on freestanding graphene membranes as compared to Cu-foil supported graphene, we suggest to be related either to intrinsic substrate effects whereby the Cu surface states underneath the Graphene modify e.g. sticking coefficients to the Sb flux (akin to Cu supports modifying the surface properties of graphene in liquid wetting 86,87 ) or to a different local temperature profile on Cu-foil supported graphene vs. suspended graphene membranes due to the macroscopic thickness 25 μm Cu foils. In the latter scenario, the Cu foil acts as an effective heat sink for the additional energy arriving with the incoming Sb flux FSb,ad compared to the vacuum-suspended monolayer graphene membranes (and, if present, thin amorphous carbon support), thus resulting in a (slightly) lower actual local substrate temperatures on the Cu supported graphene.
The here observed temperature dependence of Sb deposit morphology, crystallization onset and retained Sb amount is in good agreement with prior literature. 35,37,[47][48][49][50][51] Beyond this, our results confirm that not only the directly supporting growth substrate (here, monolayer graphene) but also the supporting material underneath (here, Cu vs. vacuum) can strongly influence Sb nanostructure growth results. 38 This is important to consider when designing Sb 2D/2D heterostructure stacks. Finally, in Fig. 5g,h we show that the here derived understanding of the balance of adsorption, nucleation, desorption and "sub-support" can also be advantageously employed to engineer larger Sb deposits of high crystalline quality. Fig. 5g,h shows deposition of nominally ~50 nm Sb at 250 °C on Cu-supported (Fig. 5g) and freely suspended graphene (Fig. 5h). On the Cu-supported graphene increasing the deposited Sb amount led not to laterally larger Sb domains but to the onset of undesired three-dimensional Sb overgrowth (Fig. 5g). In contrast, the relatively higher desorption on the suspended graphene enabled a lower Sb nucleation density and consequently a desired larger lateral growth of remaining Sb crystallites (Fig.  5h, since presumably desorption probability decreases with increasing deposit radius). Thereby by adjusting the substrate underneath the actual graphene support, we obtain a lateral size of >400 nm diagonal for β-Sb(001) and >500 nm long axis for β-Sb [2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20][21]/cubic Sb(001) deposits, respectively. This is an improvement not only over the undesired three-dimensional Sb overgrowth from Cu/graphene supported 250 °C/50 nm but also an improvement of factor ~2 compared to the 250 °C/10 nm Cu/graphenesupported deposits. This introduces the substrate underneath the direct 2D support as an important parameter to consider in 2D Sb deposition.

Materials and Methods
Physical vapor deposition (PVD) of Sb employed a commercial thermal evaporation system (MANTIS deposition system QUBE series) with a base pressure of 4×10 -5 mbar. For PVD Sb powder (Goodfellow, 99.999% purity, average particle size 150 μm) was loaded into a W boat, which was heated resistively to sublime the Sb. Phase diagrams 1 of W and Sb were cross-checked to ensure that no undesired intermetallics are formed during evaporation. Samples were loaded upside down over the evaporation source and behind a manual shutter. The sample table was electrically heated to a desired substrate temperature, where room temperature (RT, i.e. non heated), 150 °C and 250 °C were employed in this study. The Sb evaporation flux and nominally deposited thickness were monitored in situ using a non-heated quartz micro balance (QMB). The nominal Sb thickness QMB measurement was calibrated by evaporation of selected Sb films over partially masked Si wafers ("thickness monitors") at room temperature to measure Sb film thickness over film edges by atomic force microscopy (AFM). Note that the nominal thicknesses quoted in this study refer to the measured thickness values obtained from the non-heated QMB and from these Si wafer calibration depositions at RT. As discussed in the exploration of the parameter space of Sb PVD in the main text, actual retained Sb thicknesses may strongly reduce as a function of increasing substrate temperature and also type of substrate type via desorption effects.
Substrates for Sb deposition were chemical vapor deposited (CVD, 960 °C in CH4/H2/Ar at ~12 mbar) polycrystalline (grain size tens of μm) monolayer graphene films remaining on their 25 μm thick Cu foil catalysts 2,3 as well as CVD graphene films suspended as freestanding monolayer membranes over the regular hole arrays in an amorphous S2 carbon film of a transmission electron microscopy (TEM) grid (Quantifoil) i.e. no Cu underneath. 4 For graphene-free reference also Cu foils without graphene were prepared as substrates by annealing at 960 °C in 12 mbar H2 without CH4.
The scanning electron microscopy (SEM) employed a FEI Quanta 250 FEG SEM. TEM studies incl. bright field (BF) imaging, selected area electron diffraction (SAED), (valence) electron energy loss spectroscopy ((V)EELS) and energy dispersive X-ray spectroscopy (EDX, confirming the Sb purity) were performed on a FEI TECNAI F20 at 60 kV electron acceleration voltage. Scanning transmission electron microscopy (STEM) studies were performed in an aberration-corrected NION ULTRA STEM100 at 60 kV electron acceleration voltage and in (high angle) annular dark field ((HA)ADF) mode (80 to 200 mrad). 5 Correlative AFM-SEM studies employed a GETec AFSEM module installed in a FEI Quanta 600F SEM. Conventional AFM studies employed a NT MDT NTEGRA Spectra in tapping mode. AFM analysis employed Gwyddion software. 6 Raman spectroscopy employed a Horiba LabRAM at 532 nm laser excitation wavelength. Cross-sections for TEM/STEM of Cu/graphene/Sb stacks were cut by focused ion beam (FIB) processing in a FEI Quanta 3D FEG. A protective C and Pt bilayer was deposited locally by onto the region of interest prior to FIB cutting.
Note that β-Sb (A7, R-3m, 166) is often described in literature with hexagonal axis (as here) but also with rhombohedral axis. [11][12][13] Therefore, numerical (hkl) and [uvw] values need consideration of selected hexagonal or rhombohedral axis system, when comparing between reports. Likewise, within the hexagonal axis system some literature uses a a,b base vector inner angle of 120 ° (as here) while other literature uses a,b base vector inner angle of 60°. [11][12][13][14] Again therefore comparison of numerical (hkl) and [uvw] values must consider the selected axis system. To avoid ambiguity the here used axis are typically plotted alongside the atomic models throughout the manuscript.