Direct Growth of van der Waals Tin Diiodide Monolayers

Abstract Two‐dimensional (2D) van der Waals (vdW) materials have garnered considerable attention for their unique properties and potentials in a wide range of fields, which include nano‐electronics/optoelectronics, solar energy, and catalysis. Meanwhile, challenges in the approaches toward achieving high‐performance devices still inspire the search for new 2D vdW materials with precious properties. In this study, via molecular beam epitaxy, for the first time, the vdW SnI2 monolayer is successfully fabricated with a new structure. Scanning tunneling microscopy/spectroscopy characterization, as corroborated by the density functional theory calculation, indicates that this SnI2 monolayer exhibits a band gap of ≈2.9 eV in the visible purple range, and an indirect‐ to direct‐band gap transition occurs in the SnI2 bilayer. This study provides a new semiconducting 2D material that is promising as a building block in future electronics/optoelectronics.

The group IVA metal dihalide has been extensively explored regarding to their superior properties such as the visible-range band gap and the thickness-dependent band structure, [32][33][34][35] as well as the potential applications in semiconductor optical devices and perovskite solar cells. [36][37][38][39][40][41] However, on one hand, the layered 2H-PbI 2 , as one of the most widely studied materials, is unfriendly to the environment. On the other hand, the bulk SnI 2 hosts a rather different three-dimensional (3D) crystal structure, [42] and its vdW monolayer has never been experimentally realized.
In this work, for the first time, we successfully grew the vdW monolayers of SnI 2 via molecular beam epitaxy (MBE) technique. The growth takes a 2D mode, and the thickness is well controllable at the monolayer precision. Scanning tunneling microscopy/spectroscopy (STM/STS) measurements, with the aid of first principle density functional theory (DFT) calculations, confirm that the grown SnI 2 monolayer hosts a new structure of the layered 2H-PbI 2 type. The SnI 2 monolayer possesses an appreciable indirect band gap of ≈2.9 eV, and an indirect-direct band gap transition occurs as the thickness increases to above two layers. Such transition is opposite to the MoS 2 monolayers that exhibit a direct-indirect band gap transition as the thickness increases. [8] Figure 1. Epitaxial growth, structural, and electronic characterizations of vdW SnI 2 monolayers grown on Td-WTe 2 . a) Schematic illustration of the growth of hexagonal SnI 2 monolayer on Td-WTe 2 . b) RHEED patterns of epitaxial SnI 2 film and Td-WTe 2 substrate. c) Large-scale STM image of SnI 2 monolayer (size: 100 × 100 nm 2 , U = +2 V, I t = 100 pA). Inset: height profile of line scan across the step indicated by the green arrowed line. d) High-resolution STM image taken on the SnI 2 monolayer (size: 20 × 20 nm 2 , U = −700 mV, I t = 100 pA). The surface unit cells of SnI 2 monolayer and the Moiré pattern are marked in blue and black, respectively. Inset: the corresponding FFT pattern with the blue and red circles marked the reciprocal lattice peaks of SnI 2 -(1 × 1) and the Moiré pattern respectively. e) Atomic resolution of monolayer SnI 2 . The blue rhombus represents the unit cell (size: 6 × 6 nm 2 , U = −1 V, I t = 100 pA). It indicates the hexagonal lattice of a = b = ≈4.48 Å. f) Typical dI/dV spectra taken on monolayer SnI 2 (U = +2.2 V, I t = 100 pA, modulation: 40 mV). The black lines indicate the band edges.
We emphasize that the high-quality SnI 2 monolayer can be directly grown on the substrate kept at room temperature (≈20-22°C), without further annealing. The reflection high-energy electron diffraction (RHEED) pattern was in situ measured to monitor the morphology of SnI 2 monolayers. As displayed in Figure 1b, the streaks in the RHEED pattern indicate the 2D growth mode of SnI 2 on the Td-WTe 2 substrate. Quantitatively, the streak spacing for the SnI 2 film (d) is ≈1.27 times that of the Td-WTe 2 substrate (ds). The surface morphology of ≈0.5 monolayer (ML) SnI 2 deposited on the Td-WTe 2 substrate (Figure 1c) verifies the formation of atomically flat and hexagonal-shaped SnI 2 islands. The step height of the SnI 2 islands, as obtained through the line-scan profile (inset to Figure 1c) is ≈0.7 nm high. Atomically resolved STM images (Figure 1d,e), together with the corresponding fast Fourier transformation (FFT), clearly evidence the hexagonal lattice symmetry. The lattice parameter is determined to be a = b = ≈4.48 Å, in accordance with the quantitative analysis of the RHEED data. Besides, the FFT result (inset to Figure 1d) also exhibits extra dots, which originated from the Moiré pattern formed between the SnI 2 monolayer and Td-WTe 2 sub-strate, and can also be clearly observed in the STM image in Figure 1d (marked in black). It is noteworthy that the grown SnI 2 monolayer is in a new structural phase, which is completely different from its bulk counterpart. As previously reported, [42] the bulk SnI 2 crystallizes in the 3D monoclinic structure with the space group of C2/m.
Randomly distributed defects are observed on the SnI 2 islands, particularly on the SnI 2 monolayer (Figure 1c). More STM data indicate that three different defects can be identified, among which the type A is the majority, and types B and C are occasionally observed (see Figure S5, Supporting Information). In combination with DFT simulation, we ascribe the type A defect as the extra I atom at the bottom of SnI 2 monolayer, and types B and C as the extra Sn atom at the bottom and I vacancy on the top, respectively. Figure S4a, Supporting Information, shows the typical largescale STM image of SnI 2 full monolayer, which indicates the existence of multiple domains. The average size of SnI 2 single domain is about 100-150 nm. We further took the atomically resolved STM image between two adjacent domains to elucidate the boundary structure ( Figure S4b,c, Supporting Information). Even though the two adjacent domains exhibit different lattice orientations, the growth at the boundary is seamless.
To further explore the electronic structure of the SnI 2 monolayer, differential conductance (dI/dV) spectra, which reflect the local density of states, were measured. Figure 1f presents the typical dI/dV spectrum taken on the SnI 2 monolayer away from defects (more data can be found in Figure S1, Supporting Information). The semiconducting nature of the SnI 2 monolayer is revealed, and a wide band gap of ≈2.9 eV is determined, within the range of visual purple spectrum. The valence band maximum (VBM) and conduction band minimum (CBM) are also identified at ≈−1.6 eV (below Fermi energy) and ≈+1.3 eV (above Fermi energy).
Multilayered films can also be epitaxially grown with a monolayer precision. Figure 2a shows the surface of the multilayered SnI 2 films with thicknesses varied from monolayer (1L) to three layers (3L). Atomically resolved STM images, Figure 2b-d, demonstrate that the layered hexagonal lattices are maintained in these SnI 2 multilayers, and no structural transition to the bulk phase is observed (more data can be found in Figure S2, Supporting Information). It is noteworthy that the defect concentrations in SnI 2 bilayer or multilayers, typically the extra I atoms at the bottom, are significantly lower than the monolayer, which can be attributed to the different adsorption and diffusion of I atoms at the SnI 2 interlayer from SnI 2 /WTe 2 interface. We further calculated the formation energies of the extra iodine defect at the SnI 2 /WTe 2 and SnI 2 /SnI 2 interfaces. Our calculation results indicate that the extra iodine defect formed at the SnI 2 /WTe 2 interface is more stable than at the SnI 2 /SnI 2 interface by ≈0.3 eV. Interestingly, the interlayer spacing for the multilayered SnI 2 (Figure 2e) keeps a near constant of ≈0.7 nm, the same as the step height for the monolayer, indicating no identifiable layer-dependent interlayer coupling. All of these results collectively suggest that such hexagonal vdW structure can be stabilized in the 2D limit, although it does not exist in the bulk.
The dI/dV spectra (Figure 2f) taken on the SnI 2 monolayers indicate that the width of gap keeps nearly the same for the thickness up to six layers (6 L) (The spatial uniformity of the STS spectra are verified by the STS line scans, as shown in Figure  S1, Supporting Information). This differs from most of the other TMD vdWs materials hosting the electronic structures sensitive to the number of layers. [24,[43][44][45][46][47][48][49][50] The locations of CBM and VBM can be extracted from the dI/dV data in the logarithmic form (see Figure S6b, Supporting Information). As the thickness is decreased to monolayer, the Fermi energy (E F ) gradually shifts upwards with respect to the band edges, which may be due to the electron doping effect from the intrinsic defects or the substrate. Such doping effect becomes prominent in the proximity of interface, similar to what was reported for GaSe/graphene heterostructures. [50] To investigate the chemical stoichiometry of the grown SnI 2 monolayers, X-ray photoelectron spectroscopy (XPS) measurements were ex Situ performed after exposure in air. Figure 3a,b displays the XPS spectra of Sn 3d and I 3d core levels, respectively. The binding energy of Sn 3d5/2 (486.4 eV), Sn 3d3/2 (494.8 eV), www.advancedsciencenews.com www.advancedscience.com  I 3d5/2 (619 eV), and I3d3/2 (630 eV) reflects the valence states of Sn (+2) and I (−1) for the SnI 2 compound. [51] The chemical stoichiometry for Sn:I is quantitatively determined to be ≈1:2, consistent with the ideal SnI 2 compound. Moreover, there is no oxidation state of SnI 2 detected within the instrumental resolu-tion. Thus, it is concluded that the as-grown SnI 2 monolayers is significantly inert to air.
The DFT optimized atomic models of the SnI 2 mono-and multi-layers are constructed. The monolayer structure is shown in Figure 4a. It adopts the 2H-PdI 2 type (space group: P3m1) www.advancedsciencenews.com www.advancedscience.com structure. Each monolayer comprises three atomic planes covalently bonded in the sequence of I-Sn-I with the in-plane lattice of a = b = 4.53 Å. In multilayers, the separation between adjacent layers governed by vdW interactions is ≈0.32 nm. Both theoretical atomic structures and lattice constants agree well with the experimental results. Moreover, the calculated band structure indicates that monolayer SnI 2 has an indirect semiconducting gap of ≈3.25 eV, as shown in Figure 4d. Even though there still exists a minor discrepancy between the calculated and experimentally observed gap value (≈2.9 eV), the DFT + GW calculation method we applied corrects the usual underestimation of energy gap by DFT method. The corresponding calculated DOS based on the proposed crystal structure (Figure 4c) agrees well with the experimental dI/dV spectra, further demonstrating the as-grown SnI 2 monolayer to be a new 2D layered semiconductor. The calculated band structure of SnI 2 (Figure 4d-f) shows that the monolayer and bilayer host the indirect-gap, and a transition to direct-gap occurs on the trilayer SnI 2 , opposite to hexagonal 2H-phase MoS 2 (2H-MoS 2 ) whose direct band gap only exists in monolayer. [43] We also considered the other possible stacking, the common structure for semiconducting TMDs, such as MoS 2 . However, the calculated energy is significantly higher, and the energy difference is ≈0.56 eV per unit cell. Please note that a nomenclature inconsistency exists between 2H-PbI 2 and TMDs. In fact, the crystal structure of 2H-PbI 2 corresponds to the 1T structure, as referred to in TMD compounds, which is different from the common structure of 2H-MoS 2 .
This substrate-assisted growth of the 2D hexagonal SnI 2 monolayer is universal, regardless of the lattice symmetry and lattice constants of the substrate. As shown in Figure S3, Supporting Information, the SnI 2 monolayer with the hexagonal lattice can also be successfully grown on the bilayer graphene (BLG)/SiC substrates at room temperature (≈20-22°C). The SnI 2 monolayer grown on the graphene/SiC substrate exhibits an irregular shape, which differs from the equilateral triangular SnI 2 grown on the Td-WTe 2 substrate. This is probably due to the pinning effect triggered by the native defects on the graphene/SiC surface (see Figure S7, Supporting Information). We also find that both its structural and electronic properties are similar to the SnI 2 monolayers on the Td-WTe 2 substrate. Particularly, their similar properties include the atomic structure and in-/out-plane lattice parameters, 2D vdW growth mode as manifested by the streaky RHEED patterns, air stability as suggested by the XPS spectra, and semiconducting nature with a comparable energy gap of ≈2.94 eV. In view of the distinct atomic structures of hexagonal graphene and orthorhombic Td-WTe 2 , these results show that the growth of semiconducting SnI 2 with layered hexagonal lattice is barely influenced by the different lattice symmetry of substrates.

Discussions
The bulk SnI 2 is a monoclinic structure with a = 14.17 Å, b = 4.535 Å, and c = 10.87 Å (space group: C2/m). [42] The DFT calculated free energy per unit volume of the bulk monoclinic SnI 2 is ≈6.7 meV lower than that of the layered hexagonal SnI 2 , suggesting that the layered phase is less stable. However, in the 2D limit, the formation of monolayer or few layers of the bulk phase requires the interlayer bonds of monoclinic SnI 2 to be broken; consequently, the energy needed will be higher than that in the 2D hexagonal phase. Therefore, the hexagonal phase of SnI 2 in the monolayer limit is more stable, which is consistent with our experimental results. We also addressed the kinetic issue for the growth of 2D hexagonal phase. We calculated the adsorption energies of the SnI 2 layer on the WTe 2 substrate and another SnI 2 layer, respectively. The calculation results show that the adsorption energy of the SnI 2 on WTe 2 is higher than that on another SnI 2 layer by ≈75 meV per unit cell. Therefore, the SnI 2 prefers to adsorb on the WTe 2 substrate, instead of another SnI 2 layer, which favors the 2D growth. The hexagonal monolayer becomes kinetically trapped in the 2D limit, owing to the lower surface energy.

Conclusion
In summary, we have successfully fabricated vdW SnI 2 monolayers. The substrates with different lattice symmetries, such as the orthorhombic Td-WTe 2 and hexagonal graphene, are both feasible to the epitaxy of vdW SnI 2 monolayer. XPS data manifested its stability in air. The grown SnI 2 monolayer is found to be a new layered vdW semiconductor with an appreciable band gap of ≈2.9 eV. The SnI 2 monolayer also exhibits thickness-dependent properties, for example, the indirect-to direct-band gap transition. The experimental realization of SnI 2 monolayers, a new 2D semiconductor with thickness-dependent properties, provides an optimal material candidate for applications in electronics and optoelectronics.

Experimental Section
MBE Growth of SnI 2 Films: The SnI 2 monolayers were grown on the Td-WTe 2 and BLG/SiC substrates using the MBE. The Td-WTe 2 substrates were obtained via in situ cleaving the Td-WTe 2 single crystal in ultrahigh vacuum (UHV), and the BLG/SiC substrates were obtained by repeatedly flashing the SiC single crystal for more than three times in UHV with a base pressure lower than 1 × 10 −10 Torr. Prior to the SnI 2 growth, the surface qualities of Td-WTe 2 and BLG/SiC substrates were checked via STM characterization. Anhydrous SnI 2 powder (Alfa Aesar, 99.999%) was loaded in a Knudsen diffusion cells as the evaporation source. The growth morphology was in situ monitored by RHEED. During the growth, the SnI 2 source was heated to ≈280°C, and the substrates were kept at room temperature (≈20-22°C).
Scanning Tunneling Microscopy Characterization: The STM measurements were in situ carried out with a commercial low-temperature UHV-STM system (Unisoku, USM1500). The base pressure was lower than 1 × 10 −11 Torr. A mechanically polished Pt-Ir tip was used for scanning under the constant-current mode. The dI/dV spectra were collected using the lock-in amplifying technique with an AC modulation of ≈10 mV at a frequency of 879 Hz.
Density Functional Theory Calculation: The DFT calculations in this work were performed using the projected augmented wave method, [52] as implemented in the Vienna ab initio simulation package. [53] The exchange correlation potential was described by the generalized gradient approximation of Perdew-Burke-Ernzerhof type. [54] An energy cutoff of 250 eV was used, which was converged in the authors' test. The Brillouin zone was sampled by a 16 × 16 × 1 k-point mesh. To accurately describe the interlayer interactions, the D2 vdW correction method presented by Grimme was adopted. [55] In all the slab models, the vacuum distances were larger than 15 Å. This distance was large enough to avoid the interaction between two nearest slabs. The atomic structures were carefully relaxed until the forces on each atom were less than 0.01 eV Å −1 . To get accurate electronic structures, the authors also performed the GW calculations in the G0W0 level. [56]

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.