Epitaxial growth and characterization of SnSe phases on Au(111)

Two-dimensional (2D) layered group IV–VI semiconductors attract great interest due to their potential applications in nanoelectronics. Depending on the dimensionality, different phases of the same material can present completely different electronic and optical properties, expanding its applications. Here, we present a combined experimental and theoretical study of the atomic structure and electronic properties of epitaxial SnSe structures grown on a metallic Au(111) substrate, forming almost defect-free 2D layers. We describe a coverage-dependent transition from a metallic β-SnSe to a semiconducting α-SnSe phase. The combination of scanning tunneling microscopy/spectroscopy, non-contact atomic force microscopy, x-ray photoelectron spectroscopy/diffraction and angle-resolved photoemission spectroscopy, complemented by density functional theory, provides a comprehensive study of the geometric and electronic structure of both phases. Our work demonstrates the possibility to grow two distinct SnSe phases on Au(111) with high quality and on a large scale. The strong interaction with the substrate allows the stabilization of the previously experimentally unreported β-SnSe, while the ultra-thin films of orthorhombic α-SnSe are structurally and electronically equivalent to bulk SnSe.

Two-dimensional (2D) layered group IV-VI semiconductors attract great interest due to their potential applications in nanoelectronics. Depending on the dimensionality, different phases of the same material can present completely different electronic and optical properties, expanding its applications. Here, we present a combined experimental and theoretical study of the atomic structure and electronic properties of epitaxial SnSe structures grown on a metallic Au(111) substrate, forming almost defect-free 2D layers. We describe a coverage-dependent transition from a metallic β-SnSe to a semiconducting α-SnSe phase. The combination of scanning tunneling microscopy/spectroscopy, non-contact atomic force microscopy, x-ray photoelectron spectroscopy/diffraction and angle-resolved photoemission spectroscopy, complemented by density functional theory, provides a comprehensive study of the geometric and electronic structure of both phases. Our work demonstrates the possibility to grow two distinct SnSe phases on Au(111) with high quality and on a large scale. The strong interaction with the substrate allows the stabilization of the previously experimentally unreported β-SnSe, while the ultra-thin films of orthorhombic α-SnSe are structurally and electronically equivalent to bulk SnSe. * Authors to whom any correspondence should be addressed.
Original content from this work may be used under the terms of the Creative Commons Attribution 4.0 licence. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.

Introduction
IV-VI metal monochalcogenides are versatile materials with a wide range of applications due to their distinctive anisotropic crystal structure leading to directionally dependent electronic and optical properties. During last years, they have gained great attention in view of their promising applications in sustainable electronics as an alternative to conventional semiconductors [1][2][3]. Their versatility arises not only from the combination of distinct elements of groups IV-VI, but also from the structural phase transitions that can be induced in these materials, allowing to tailor their electronic properties. Among binary IV-VI chalcogenides, the SnSe is a layered semiconductor that presents particular interest as its band gap of ≈1.4 eV and high carrier mobility make it a good candidate for optoelectronic and photovoltaics applications [4][5][6]. In the last years, there was a quickly growing interest in the understanding of the properties of this material due to its exceptional thermoelectric properties [7,8]. Furthermore, it is made from earth abundant elements and does not deteriorate under ambient conditions [9][10][11]. At room temperature, SnSe crystallizes in an orthorhombic structure (space group Pnma), equivalent to a distorted rock-salt structure, while at T > 750 K a Cmcm phase is favored [7,12]. Additionally, a distinct phase of SnSe (single layer β-SnSe) has been theoretically reported regarding its application for photocatalytic water splitting and gas sensing [13][14][15][16][17]. Even more interestingly, it is predicted to have better thermoelectric properties than any other SnSe phase [16].
The epitaxial growth of 2D materials allows the design of high quality defect-free samples with a precise control over the thickness. Furthermore, an adequate choice of the substrate permits the stabilization of certain phases that are thermodynamically unfavored, modifying their electronic properties [18]. The possibility of controlling the structure of these layered materials is important since different phases of the same compound can have drastically different properties. Recently, the epitaxial growth of SnSe in its rock-salt structure on a Bi 2 Se 3 substrate has been demonstrated, revealing its topological crystalline insulator phase [19]. The successful growth of several group-IV monochalcogenide ultra-thin films on different substrates has been reported, such as SnS on mica, SnTe and SnSe on graphene, and lead monochalcogenide compounds on alkali halides [20][21][22][23]. Such works confirm remarkable ferroelectric, optical and electronic properties of IV-VI semiconductors and encourage further investigations to expand their properties and overcome chemical stability challenges.
Here, we report the growth of large (tens of nm) and defectfree ultra-thin films of two different phases of SnSe on a Au(111) surface under ultra-high vacuum (UHV) conditions. We characterize the structural and electronic properties of each phase by means of scanning tunneling microscopy and spectroscopy, non-contact atomic force microscopy, x-ray photoelectron spectroscopy and diffraction and angle-resolved photoemission spectroscopy (STM/STS, ncAFM, XPS, XPD and ARPES, respectively), supported by density functional theory (DFT) calculations. In particular, we report the layercontrolled growth of two different SnSe phases: α-and β-SnSe. We found that for a monolayer coverage, the only theoretically reported hexagonal single layer β-SnSe is formed [15][16][17], with a corrugated honeycomb geometry stabilized by the Au(111), being a structurally analogous to β-TaS on Au(111) [24]. At higher coverage, we observe the formation of a 2D layered material, which adopts an orthorhombic structure (α-SnSe) analogous to that in the SnSe crystal [25,26]. The structural and electronic properties of both phases reveal clear differences, tailoring the properties of SnSe by phase engineering. This work demonstrates the crucial role played by the metallic Au(111) substrate to stabilize the growth of the otherwise thermodynamically unstable β-SnSe phase. This observation enables to explore new phases of group-IV monochalcogenide ultra-thin films on an appropriate substrate.

Results and discussion
We have employed the molecular beam epitaxy (MBE) method to grow the different SnSe phases, due to the high precision intrinsic to this method. The epitaxial growth of most metal chalcogenides is carried out by co-deposition of elements from two separate precursors, a process that requires a precise calibration to control the stoichiometry. Here, in order to preserve the 1:1 stoichiometry, the growth of the sample was carried out by direct evaporation of SnSe solid powder at 550 • C using a Knudsen cell onto the Au(111) surface held at room temperature, followed by an annealing at 250 • C. The coverage dependence of the growth of ultrathin SnSe films was monitored by low energy electron diffraction ( Figure SI2 displays a comparison of a reciprocal space scheme considering six rotational domains of a rectangular lattice and the experimental LEED pattern. The STM image in figure 1(b) confirms the growth of a SnSe multilayer on Au(111). We highlight the high quality of both samples, with large (tens of nm) and defect-free terraces of SnSe on Au(111). Figure 1(c) shows the core level XPS spectra of Se 3d after 2 and 20 min of deposition with their corresponding fittings (green and blue curves, respectively). The energy range of Se 3d partially overlaps with the Au 5p 3/2 peak and its contribution has to be considered for the fitting. For high coverage (several layers of SnSe) only the contribution from the Se is observed. The XPS spectra in figure 1(d) show the core level doublets of Sn 4d after 2 min and 20 min. After 2 min of SnSe evaporation, we observe two components of Sn 4d. In addition to the dominant SnSe component, a minor component appears at lower binding energies, which we tentatively assign to SnSe clusters formed during the growth process. For lower SnSe coverage (≈0.5 monolayer), the Sn 4d XPS spectrum in figure SI3 shows equivalent contribution from the two components, confirming the origin of the peak at lower binding energies. From the presence of the cluster phase one could also expect a weak secondary component in the Se 3d level, but its low intensity on top of a large non-linear Au 5p 3/2 background at 2 min in figure 1(c) does not allow a reliable decomposition into two components via fit procedures. On the other hand, after 20 min of evaporation only one component is observed for both Sn 4d and Se 3d core levels. It is worth mentioning that Sn 4d 5/2 and Se 3d 5/2 peaks in the multilayer sample are located at binding energies of 24.55 eV and 53.50 eV respectively, in excellent agreement with previous works on SnSe crystals [27]. After 2 min of evaporation, Sn and Se peaks are both shifted by ≈0.25 eV towards higher binding energy (peak positions 24.75 eV and 53.76 eV, respectively) with respect the multilayer sample. Such parallel core level shifts can imply a charge transfer from SnSe to Au(111). For opposite charge transfer core level shifts to lower energies are expected, as observed, e.g. for Mo and S core levels in MoS 2 layers on Au(111) [28]. We describe the charge transfer between the SnSe monolayer and the Au(111) substrate in more detail below, based on our findings in DFT calculations and ARPES measurements. Regarding the stoichiometry of the SnSe phases, we note that the XPS spectra always overestimate the Sn contribution with respect Se. Therefore, to demonstrate the 1:1 stoichiometric ratio, we performed XPS measurements of a SnSe(100) crystal, showing a perfect agreement with the stoichiometric ratio of our epitaxially grown SnSe, as shown in figure SI4 and table SI1. Thus, we can conclude that the stoichiometry remains unchanged (1:1) independently of the thickness and the phase.  detail the geometry of the first layer of SnSe on Au(111), which corresponds to the β-phase [29]. An atomic resolution ncAFM image with a CO-functionalized tip of β-SnSe reveals the hexagonal lattice of the topmost atomic layer, showing a distance between the surface atoms of 3.9 ± 0.1 Å. The mismatch between the Au(111) and β-SnSe lattice gives rise to a commensurate hexagonal moiré pattern with a periodicity of ≈11.7 Å, which corresponds to a repeating 3 × 3 cell of the SnSe, equivalent to 4 × 4 cell of the Au(111). This 3 × 3 superstructure can be observed as brighter spots in the ncAFM image in figure 2(b).
To get more insight into the atomic arrangement of the β-SnSe monolayer on Au(111) surface, we carried out total energy DFT calculations using a 4 × 4 unit cell to fit the experimental observation. The DFT model of the optimized structure of β-SnSe is shown in the left panel of figure 2(b), consisting of a corrugated honeycomb geometry formed by a Sn layer on top of the Au(111) and an upper layer of Se atoms. The calculated Se-Se distance of 3.8 Å matches very well the experimental value (d Se-Se = 3.9 ± 0.1 Å).
Interestingly, after the complete growth of the first monolayer, the following SnSe layers adopt the typical layered orthorhombic structure (α-SnSe) described in figure 2(c). Perhaps, the β-SnSe could be reducing the interaction between the α-SnSe and the Au(111) surface, favoring the growth of orthorhombic SnSe. NcAFM images of α-SnSe resolve the Sn and Se atoms (figure 2(c)), while in the STM image only Sn atoms contribute to the tunneling signal and are visible (see comparison of both images in figure SI5), in agreement with a previous STM study of a cleaved SnSe crystal [25]. From these measurements, we can determine the α-SnSe unit cell a = 4.1 ± 0.1 Å and b = 4.3 ± 0.1 Å, which is consistent with the calculated bulk values of a = 4.21 Å and b = 4.52 Å.
To confirm the structure of the β-SnSe phase, so far only theoretically reported as a free-standing layer [13,16], we carried out XPD in the forward focusing regime (i.e. photoelectrons with high kinetic energy). This technique has shown itself to be valuable to determine the atomic structure of other 2D materials and adsorbates [30][31][32][33]. Figures 3(a) and (b) show the experimental and calculated stereographically projected XPD patterns of the modulation function χ for the clean Au(111) and the β-SnSe on Au(111), respectively (see supplementary note 1 for details). The simulations were performed with the electronic diffraction in atomic cluster code [34], using as input the fully optimized atomic structure of the β-SnSe/Au(111) phase as obtained from the DFT calculations (see figure 2(b)). Figure 3(a) shows the experimental Au 4f XPD pattern (color) obtained at AlKα photon energy, together with the simulated pattern (grey). For the simulation, we considered a Au lattice parameter of 4.065 Å and an emission kinetic energy of 1402 eV. As can be seen, the simulations of the Au 4f ( figure 3(a)) and the Sn 3d ( figure 3(b)) emission patterns (in grey) reproduce the main modulations of the experimental data (orange patterns). Remarkably, the emission pattern of Sn 3d reveals the presence of 3-fold symmetric features at grazing angle and no anisotropy close to normal emission. If a similar pattern is recorded for the Se 3d core level, a featureless pattern is observed, indicating the lack of diffracting species above them, i.e. they constitute the outermost layer of the β-SnSe phase. Thus, our XPD measurements corroborate the DFT model in which the β-SnSe phase consists of a double layer with the Sn atoms in direct contact with the Au(111) surface and the Se atoms forming the uppermost layer. Furthermore, our XPD patterns allow the determination of the buckling angle of the Se atoms above the Sn atoms, resulting in θ values (measured with respect to the surface normal) of 54 • ± 1 • and 70 • ± 2 • for the first and second neighbors, respectively, in excellent agreement with the theoretical ones (54 • and 68 • , respectively). Thus, our XPD measurements not only confirm the bilayer nature of the β-SnSe phase, but also validate the proposed atomic positions. Furthermore, our model reveals slight variations in the height of Se atoms (see figure SI6), as anticipated in the ncAFM image in figure 2(b). We find that the variations in the atoms' height correlate with a lateral electronic modulation imposed by the moiré effect, as illustrated in figures 3(d) and (e), where a representative constant-height STM image acquired at 5 meV is nicely reproduced by a simulated STM image.
Finally, to investigate the electronic properties of both β-and α-SnSe phases on Au(111), we performed ARPES measurements to monitor the evolution of the band structure.  figure 4(d)). After β-SnSe growth, the sharp circular feature associated to the SS in figure 4(a) is replaced by a strongly broadened and wider circular intensity in figure 4(b), indicative of substantial hybridization effects close toΓ point. Additional broad intensity appears mainly along theM-Γ direction visible in figure 4(b). At the Fermi level, the Au(111) sp-bands atM have disappeared while close toK they remain slightly attenuated ( figure 4(e)). Our observations match DFT calculations in figures 4(g) and (h) (see methods section and supplementary note 2), where a deeper and strongly broadened V-shaped parabola appears and the sp-states are heavily disrupted along theM-Γ direction. This strong rearrangement of the band structure indicates a significative hybridization between β-SnSe and Au(111) states, as previously suggested by XPS measurements and further supported by STS measurements and DFT calculations (see figures SI7 and SI8, respectively). The calculated distance between Sn and the outermost layer of the Au(111) surface is 2.4 Å, which indicates the presence of a strong chemical bond accompanied with a significant charge transfer (see figure SI8) from the β-SnSe to the Au(111) surface. This observation agrees with the XPS measurements in figure 1, where we described a shift in energy towards higher binding energies. The dI/ dV spectrum recorded on β-SnSe does not show a clear band gap, as shown in figure SI7. This quasi-metallic behavior of the β-phase monolayer in contrast to the calculated 2.22 eV band gap for a freestanding monolayer of β-SnSe once more highlights the strong hybridization of the SnSe layer with the Au(111) [16].
In  figure SI7. For bulk SnSe, VBM are known to be located at about 0.5 Å −1 and 0.6 Å −1 along theΓ-X and thē Γ-Ȳ directions [10]. The diffuse appearance of the maxima in our data are partly due to the existence of three domains, which result in a mixture of ARPES intensities from different directions of the BZ. The dI/dV spectrum on α-SnSe clearly shows a band gap of ≈1.6 eV, consistent with the STS measurements reported for bulk SnSe [35][36][37]. From these results, we can assume that the electronic properties of the α-SnSe/Au(111) multilayer are comparable with the ones of a bulk SnSe crystal.

Conclusions
To conclude, we have demonstrated the successful epitaxial growth of two different phases of SnSe on a Au(111) substrate. Surface characterization techniques as STM, ncAFM, XPS, XPD and ARPES allowed us to carry out a comprehensive comparative study of both phases, confirming for the first time the synthesis of a single layer β-SnSe. Our results reveal the completely different geometric and electronic properties between β-SnSe and α-SnSe. In the case of α-SnSe phase, we confirm a negligible role of the substrate, with properties consistent with previous works of bulk SnSe. On the other hand, we found both experimental and theoretical evidences of the strong hybridization of the β-SnSe with the Au(111). Thus, we demonstrate that depending on the dimensionality a phase transition from β-SnSe to α-SnSe is induced. Our work reports a model case of two different phases of SnSe, contributes to the understanding of their properties and it may motivate the substrate-driven synthesis of new phases of other 2D layered semiconductor materials.

Experimental methods
SnSe ultra-thin films were prepared in situ in UHV conditions with base pressures below 5 × 10 −10 mbar. The singlecrystalline Au(111) substrate was prepared by repeated cycles of argon ion sputtering and annealing. Tin Selenide (99.999%) was purchased from Alfa Aesar and sublimed from a Ta crucible at 550 • C directly onto the Au(111) surface kept at RT. After the deposition, the sample was annealed for 20 min at ≈230 • C.
The SPM experiments were performed in a low temperature (5 K) STM/nc-AFM instrument (Specs-JT Kolibri or Createc equipped with Qplus sensor) in UHV. STM topographic images were acquired in constant-current mode (unless otherwise specified), while nc-AFM images, dI/dV maps and spectroscopies were acquired in constant-height mode and with CO functionalized tips. dI/dV spectra and constant energy maps were recorded using the lock-in technique with a modulation frequency of 923 Hz and a modulation amplitude of 20 mV.
XPS, UPS and ARPES were measured in an Omicron NanoESCA apparatus. The XPS was collected using a monochromated Al Kα source (hν = 1486.6 eV), while UPS and ARPES were carried out using a He I discharge lamp (hν = 21.2 eV). The energy calibration was done using the Au 4f 7/2 core level at 84.0 eV along with the Fermi level edge. XPS spectra were fitted with Voigt peaks on Shirley background using KolXPD software.
XPD experiments were carried out in an UHV chamber with a base pressure in the 10 −10 mbar range, equipped with an Al-Kα x-ray source (1486.6 eV) and a hemispherical electron energy analyzer (SPECS Phoibos 150 spectrometer) with a 2D delay-line detector.
Each XPD pattern was produced by measuring the area of the Au 4f, Sn 3d and Se 3p core-level peaks over a wide azimuthal sector (120 • ) for polar angles ranging from grazing emission (θ = 70 • ) to normal emission (θ = 0 • ). By selecting the high magnification lens mode and closing the iris aperture at the front of the electron spectrometer, the emission acceptance angle was limited, setting an angular resolution of 1 • . The XPD patterns were produced by scanning the polar (θ) and azimuthal (ϕ ) angles with a motorized sample manipulator over the 0 • ⩽ θ ⩽ 70 • and 0 • ⩽ ϕ ⩽ 120 • ranges, with θ being measured from the sample normal and ϕ from the [112] direction lying on the Au surface. The step size in θ was set to 2.5 • , while the step size in ϕ was changed with θ to keep a constant data density. See supplementary information for further details.

Theoretical methods
Density functional theoretical calculation details: the geometry, electron density and band structure of SnSe/Au(111) were calculated using the DFT code Fritz-Haber Institute Ab-Initio Materials Simulation (FHI-AIMS) [38]. The PBE generalized gradient functional [39] was used in all the DFT calculations. Van der Waals (vdW) correction due to Tkatchenko and Scheffler [40] was added to account for dispersion forces, except for the pair interaction between two Au atoms for which no vdW correction was considered. The default 'light' atomcentered orbital bases of the FHI-AIMS code were adopted to expand the Kohn-Sham wave functions. Scalar relativistic correction was included. During geometry optimization, all structures were relaxed until forces on individual atoms did not exceed 2 meV Å −1 . The geometry of the β-SnSe on Au(111) was calculated in a slab that consisted of four layers of Au atoms to represent an unreconstructed Au(111) surface and the SnSe overlayer. The surface area of the β-SnSe/Au(111) was such that it corresponded to 4 × 4 unreconstructed Au(111) surface unit cells or 3 × 3 unit cells of the β-SnSe layer. In total, there were 9 Sn atoms, 9 Se atoms, and 64 Au atoms in the β-SnSe/Au(111) supercell. The first Brillouin was sampled by 6 × 6 k-points. For calculating the projected band structure, a thicker slab with 10 Au atomic layers and the β-SnSe overlayer on both sides was formed, without repeating full geometry optimization. The thicker slab allowed to reduce the unwanted effect of electron state quantization along the direction perpendicular to the surface. The BZ sampling for electron self-consistency calculation in this larger supercell was reduced to 4 × 4 in order to keep the calculation feasible. The properties of the α-SnSe (bulk-like SnSe) were calculated without the Au surface, in a slab consisting of 3 SnSe bi-layers, with 12 Sn and 12 Se atoms inside the whole supercell. The Brillouin-zone sampling was 24 × 24 in this case. Calculations of the bare Au(111) surface were done in the same geometry and with the same parameters as β-SnSe/Au(111), just with the β-SnSe overlayer removed.

Data availability statement
All data that support the findings of this study are included within the article (and any supplementary files).