Identifying crystal structures and chemical reactions at the interface of stanene on Bi₂Te₃

The strong interfacial interactions and confinement effects that arise when monolayers or heterostructures of layered materials are synthesized provide a platform to access a wide range of novel physical phenomena.^1–13 With the high degree of control in contemporary synthesis methods, the field of monolayer materials is rapidly expanding and has produced a number of important developments, such as high-mobility semiconductors,^1–3 enhanced T_c superconductors,^4–9 and switchable topological insulators and quantum spin Hall materials.^10–13 There are even more possibilities in engineering interfaces of materials from multiple classes to potentially build topological transistors^14–16 or observe Majorana fermions in topological insulator–superconductor heterostructures.¹⁷ 

There has been a surge of interest in heterostructures based on chalcogenides or pnictides due to the many novel behaviors in these classes of materials.^3,5–18 With the expanding chemical landscape, an important consideration, in addition to crystalline compatibility (i.e., crystal orientation and lattice parameter), is the chemical compatibility between the substrate and deposited materials when designing a system for a particular functionality.

Here, we characterize the growth process for stanene on Bi₂Te₃ substrates. Stanene, a monolayer of buckled hexagonal tin, similar to graphene [Fig. 1(a)], is predicted to have a wider bandgap than other topological materials and host topological edge states.^19–21 The properties of stanene are also predicted to be tunable by strain, substrate bonding, and decoration with halides or hydroxide, making stanene a promising system for topological devices.^19,20,22

Multiple theoretical and experimental attempts to understand the conditions necessary to achieve stanene have revealed the difficulty in synthesizing a continuous stanene monolayer.^23–28 A few substrates have been shown to host uniform and topological stanene monolayers, most notably on Ag(111), where stanene sits atop a reacted Ag₂Sn interface,²⁹ and on PbTe/Bi₂Te₃/Si(111) heterostructures, where bi- and tri-layer stanene have been observed to be superconducting, making possible the observation of Majorana particles.^30,31 These examples highlight the need to understand the crystal structure, chemical conditions, and growth process for stanene deposited on substrates selected for different applications.

Bi₂Te₃ is hexagonal with a stable (0001) surface and an in-plane lattice parameter (4.38 Å) closely matched to stanene (4.66 Å). The unit cell of Bi₂Te₃ consists of hexagonal quintuple layers (QLs), each bonded by van der Waals forces along the ⟨0001⟩ direction. In this work, thick Bi₂Te₃ films are grown by molecular beam epitaxy (MBE) on Si(111)−7 × 7 and Ge(111)-c(2 × 8) reconstructions, the most stable atomic reconstructions for the Si and Ge (111) surfaces.^32–36 The reconstructions can be observed by reflection high-energy electron diffraction (RHEED), seen in Fig. S1 of the supplementary material.

Once the reconstruction is achieved in RHEED, 20 QL Bi₂Te₃ films are grown by co-depositing molecular Bi and Te, with a ∼10× overpressure of Te. RHEED confirms high-quality crystal growth [Figs. 1(b) and S1 in the supplementary material] of strain-relaxed Bi₂Te₃ on both Ge (4.00 Å in-plane lattice parameter, 9.5% mismatch with Bi₂Te₃) and Si (3.86 Å lattice parameter, 13.5% mismatch) substrates. Following deposition, Bi₂Te₃ is capped with an amorphous layer of Te or Sn at a substrate temperature below 50 °C to protect the Bi₂Te₃ from oxidation while transferring to external chambers for stanene growth and measurement.

The procedure for stanene growth is based on that used previously²³ and optimized in the same MBE deposition chamber used for Bi₂Te₃ growth. We explore the growth process for substrate temperatures between 25 and 200 °C and find that RHEED patterns of stanene films grown with the substrate at 100 °C [Fig. 1(c)] produce the highest intensity diffraction. All the data reported here are for Sn grown on Bi₂Te₃ at a substrate temperature of 100 °C.

To understand the growth process of stanene on Bi₂Te₃, Sn growth is performed in a custom x-ray photoelectron spectroscopy (XPS) chamber that allows in situ measurements during Sn deposition. The deposition rate is calculated by depositing Sn on a quartz disk and measuring absorption of a laser (λ = 633 nm) before and after growth. The coverage of Sn on Bi₂Te₃ corresponding to one monolayer is 1.2 × 10¹⁵ atom/cm². Bi₂Te₃ films used for XPS measurements are capped with Te and subsequently decapped in the XPS chamber by annealing to 350 °C. To verify that the cap has been removed, XPS is measured during heating and the sample is cooled once the Bi and Te core level intensities stop changing. XPS spectra before and after anneal (Fig. S2 in the supplementary material) show the emergence of Bi³⁺ peaks following the anneal.

To extract the crystal structure of Sn deposited on Bi₂Te₃, the same growth process is performed in an MBE growth chamber mounted for in situ synchrotron x-ray diffraction (XRD) at the 33ID-E beamline at the Advanced Photon Source, Argonne National Laboratory. The deposition rate is measured using a quartz crystal microbalance. Bi₂Te₃ substrates prepared for XRD measurements are capped with Sn and then decapped by argon sputtering in the XRD chamber. Following sputtering, Bi₂Te₃ is annealed at 250 °C to recover an atomically smooth surface, indicated by a strong RHEED pattern [Fig. 1(b)]. Sn, which has a lower vapor pressure than Te, is chosen as a cap for XRD measurements due to the extended time necessary to transport the samples to the beamline.

We grow Sn at 0.5 monolayers/min and measure XPS continuously during growth so that spectra of all constituent elements are measured at each 0.4 monolayers. In addition, we pause growth at 1.6, 3.2, and 16 monolayers to perform longer measurements to improve the signal to noise ratio. The energy of each scan is calibrated to the valence band edge. The spectra for bare Bi₂Te₃ and 1.6 monolayers deposited Sn are shown in Fig. 2(a), with a clear appearance of Sn peaks following growth.

Beginning at 0.4 monolayer coverage, the spectra for Bi and Sn show multiple peaks [Bi 4f and Sn 3d transitions at 0.8 monolayer shown in Figs. 2(b) and 2(c)], which is a sign of multiple valences in the Bi and Sn atoms. We identify Bi³⁺ 4f_7/2 and Sn⁰ 3d_5/2, at 157.3 and 485.1 eV, respectively, which correspond to the binding energies expected for Bi₂Te₃ and elemental Sn.^37,38 We also identify binding energies corresponding to Bi⁰ 4f_7/2 and Sn²⁺ 3d_5/2, at 156.3 and 486.4 eV, respectively, which suggests the reaction of Bi₂Te₃ and stanene to Bi and SnTe.^38,39 Other recent work to recreate stanene on Bi₂Te₃ showed a similar evidence of this reaction in XPS.⁴⁰ 

To understand the chemical reaction during growth, we extract the integrated intensity of each transition as a function of film thickness (Fig. 3). Immediately upon depositing Sn, 3d_5/2 peaks corresponding to both Sn⁰ and Sn²⁺ appear, with the Sn⁰ peak becoming increasingly dominant as growth continues [Fig. 3(a)]. The ratio of Sn²⁺/Sn⁰, plotted in the inset of Fig. 3(a), shows that as Sn is initially deposited, approximately half as much Sn²⁺ forms as Sn⁰. This suggests that half as much SnTe forms as elemental Sn at the beginning of growth. This chemical reaction is reflected in the intensity of Bi as well [Fig. 3(b)]. Initially, Bi³⁺ from the bulk Bi₂Te₃ dominates the Bi 4f_7/2 signal, and as growth proceeds, Bi⁰ grows in intensity as Sn and Bi₂Te₃ react into SnTe and Bi.

Considering the reaction that occurs when Sn is deposited on Bi₂Te₃, a uniform stanene layer may not form at this interface. XPS results suggest an inhomogeneous film profile, consisting of Bi₂Te₃, Sn, Bi, and SnTe. To better understand the structure that develops, we perform XPS at different takeoff angles (TOAs) from the film plane, which varies the depths probed by the electron analyzer. We measure spectra with TOA = 45° (more surface sensitive) and 90° (more bulk sensitive) and compute the ratios of relevant elemental transitions to extract their layering order (Fig. 4).

Beginning at 1.6 monolayer Sn, the ratio of Bi³⁺ to Sn⁰ is greater at 45°, suggesting that more Bi₂Te₃ is located at the surface than elemental Sn [Fig. 4(a)]. Considering also that the Bi³⁺ to Sn²⁺ ratio is approximately equal at both angles for 1.6 monolayers [Fig. 4(b)], these results suggest that Sn initially reacts with subsurface Bi₂Te₃, creating an upper layer of Bi₂Te₃, SnTe, and Bi. The Sn²⁺/Sn⁰ and Sn²⁺/Bi⁰ ratios further indicate that SnTe is located above the deposited Sn and Bi above the SnTe at 1.6 monolayers [Figs. 4(c) and 4(d)]. This profile for films ≤1.6 monolayer thick is shown in Fig. 4(e). For films ≥3.2 monolayer thick, the upper Bi₂Te₃ layer has fully reacted. The profile consists of the following layers: intermixed SnTe and Bi, elemental Sn, and the Bi₂Te₃ substrate [Figs. 4(a)–4(d), profile in Fig. 4(f)]. At an atomic level, multiple compounds intermixed in the same layer may manifest in a variety of ways—separate lateral domains or alternating layering, for example—but techniques beyond XPS and XRD are necessary to go beyond our identification of the layer order of atomic valences. This ordering is likely due to compounds of low surface energy, i.e., Bi and Bi₂Te₃, migrating toward the surface, effectively wetting the other compounds with higher surface energies. Bi is a well-known surfactant,^36,41 and Bi₂Te₃(0001) should have low surface energy because of the dominant van der Waals bonding between QLs.

While the Sn/Bi₂Te₃ interface consists of several intermixed materials, it is possible to examine stanene on the surface, as has been done with STM studies of hexagonal stanene.²³ To quantitatively study stanene on the surface, we perform in situ synchrotron x-ray diffraction at the Advanced Photon Source. Previous studies suggest that stanene is strained coherently to the Bi₂Te₃ substrate; thus, we examine stanene signals along the Bi₂Te₃ crystal trucation rods (CTRs).

We observe clear differences between Bi₂Te₃ CTRs measured before and after deposition of 1 monolayer Sn (grown at 0.2 monolayers/min), shown in Figs. 5(a) and 5(b), indicating that the deposition of Sn has created a film structure coherently strained to the Bi₂Te₃ lattice. However, from XPS, we know that SnTe and Bi are both present in addition to Sn, and both may possess hexagonal structures that may grow coherently strained to Bi₂Te₃. Any difference we observe along Bi₂Te₃ CTRs could be due to stanene, SnTe, or Bi, meaning that one cannot distinguish between deposited stanene, SnTe, and Bi with scans along Q_z alone.

To elucidate more structural information of the stanene film, we perform anomalous x-ray scattering at the Sn K-edge to extract the diffraction of Sn atoms from the entire system, allowing one to probe for only stanene. As the energy of the incoming photons crosses an absorption edge, the atomic form factor has a significant complex energy dependent term, f′(E) + if″(E), where f″ is the absorption and f′ is related to f″ by the Kramers–Kronig relation. The values of f′ and f″ for Sn, calculated by measuring the absorption of x rays through a 58 μm Sn foil, are shown in Fig. 5(c).

Using anomalous scattering to extract the Sn signal of our stanene films, we move along Bi₂Te₃ CTRs and scan the beam energy across the Sn K-edge (29.3 keV). We fit the resulting spectra to an expression for the scattered intensity, taking into account the energy dependent terms f′ and f″,

where A_Sn(q)(f_Sn(q) + f′(E) + if″(E)) is the structure factor for Sn, A_BT is the structure factor for all forms of Bi and Te (Bi₂Te₃, Bi, and Te in SnTe), f_Sn is the energy-independent atomic form factor for Sn, and φ is the phase difference between the Sn and Bi₂Te₃ structure factors. We refer to A_Sn as the Sn scattering strength.

Following the deposition of 1 monolayer Sn, we observe a strong energy dependence along the (1 1 L) Bi₂Te₃ CTR [Figs. 5(d) and 5(e)], which we fit to Eq. (1) to extract the Sn scattering strength. Performing these fits along the CTR extracts diffraction from only Sn, shown in Fig. 6(a). The extracted scattering strength of Sn along the (1 1 L) CTR, which consists of two features—a broad peak centered at Q_z = 8 and a sharper, more intense peak centered at Q_z = 8.7—can be analyzed to extract the structure of the coherently strained Sn. Simulations of 1 monolayer buckled and unbuckled stanene [structure in Fig. 1(a)] do not exhibit peaks in this range of Q_z [Fig. 6(b)]. However, assuming an AB graphene-like stacking, both an unbuckled stanene bilayer with a gap between layers of 3.81 Å and a buckled stanene bilayer with a buckling distance of 0.93 Å and a layer gap of 3.51 Å, which is similar to the structure of (111) α-Sn, can reproduce the broad peak centered at Q_z = 8 [Fig. 6(b)]. This bilayer stacking is consistent with previous STM measurements of stanene, which observe an AB-stacked second stanene layer forming before achieving full coverage of a single monolayer.²³ Considering the documented favorability for buckled stanene in the previous work,^19,20,23 the buckled bilayer stanene is the likely source for the peak centered at Q_z = 8, though measurement over more of reciprocal space is needed to confirm the presence of buckled monolayer stanene. Furthermore, since the analysis of anomalous scattering extracts all Sn diffraction signals, this CTR may still contain contributions from SnTe. XRD of stanene grown on a less reactive substrate should yield a more conclusive stanene structure.

The narrow peak width of the second feature at Q_z = 8.7, which corresponds to a film feature 10 nm thick, cannot be accounted for by the deposited film of stanene or Sn reacted to SnTe, and thus another structure, in addition to the stanene bilayer, must be present. The source of this peak is intercalation into the Bi₂Te₃ substrate, which would give a similarly wide peak at Q_z = 9. Assuming a structure where Sn atoms intercalate between Bi₂Te₃ QLs, fitting the measured peak width indicates that Sn is intercalated into the upper 10 QLs. The shift in Q_z to 8.7 Bi₂Te₃ reciprocal lattice units indicates a 3% expansion along the c-axis of the Bi₂Te₃.

A random in-plane intercalation structure would give Bragg peaks every integer in Q_z; therefore, the presence of an intercalation peak near Q_z = 9 but absence near Q_z = 8 suggests a high-symmetry in-plane structure. We deduce an intercalated Sn structure consisting of a single Sn atom per unit cell between each QL at locations shown in Fig. S3 of the supplementary material similar to structures observed in previous works studying the intercalation of Sn.^42,43 Simulating this intercalation with the c-axis expansion generates a close match to the sharp peak observed in the extracted scattering strength [Fig. 6(c)].

By comparing the extracted Sn scattering strength at the intercalation peak to the intensity of strong Bi₂Te₃ Bragg peaks at (0 1 20) and (0 1 23), normalized to their respective structure factors, we estimate that the upper 10 QLs of Bi₂Te₃ contain 10%–17% of a stanene monolayer intercalated between each QL, which is more total Sn than deposited during stanene growth. This indicates that the intercalation mostly occurs as a result of the sputter removal of the Sn cap, as opposed to deposition. This process of deposition followed by sputtering and annealing may provide a novel method for intercalating Sn into thin films of Bi₂Te₃ or other materials, contrasting the solid state crystal growth or penetration through crystal steps observed in the previous work.^42,43

Our observations that Sn deposited on Bi₂Te₃ forms a mix of SnTe, Bi, and Sn instead of pure stanene reveals that Bi₂Te₃ induces inhomogeneity during stanene growth. The favorability of this reaction is consistent with a calculation of the heat of reaction, ΔH_rxn, from the bulk literature values of reactant and product heats of formation,

This sign of ΔH_rxn indicates that this reaction is exothermic. We also perform first principles calculations of the bulk formation energies that are consistent with this thermodynamic observation.

While this heat of reaction is similar in magnitude to those of Cu or Ni forming Cu₃Si and NiSi with Si (−11.6 kJ and −42.4 kJ, respectively), both of which occur at low temperatures when those elements are deposited on Si,^44,45 bulk thermodynamic calculations may not always predict the surface structure. There are examples of stable monolayer films on substrates where a reaction should take place, such as Sr on Si(001).⁴⁶ The layered nature of Bi₂Te₃ suggests that the surface and interface energies could prevent a bulk-like reaction, though our experiments suggest that the low surface energy of Bi may dominate. To verify this, we perform additional first principles calculations considering a variety of possible surface structures of Sn, Bi, and Te on a Bi₂Te₃ substrate. Accounting for the interfacial forces, it is energetically favorable for all structures tested to decompose into surface-bound films of Bi, SnTe, and Bi₂Te₃, which is consistent with the reaction we observed in the experiment.

The agreement between our observation of mixed growth and the unfavorable bulk heat of reaction provides an efficient method for checking the suitability of potential substrates for stanene growth with criteria that extend beyond favorable lattice constants and topological properties. To evaluate alternative substrates for stanene, we calculate the ΔH_rxn for other potential substrates that have been identified in the literature (Table I).^{23–26,29,30,40,47–55} In addition to deposition attempts on Bi₂Te₃, recent work depositing Sn on Au and Ag also presented evidence of substrates alloying with Sn, in agreement with the calculated heat of reaction.^29,53 There are a few substrates that offer a more favorable energy landscape, such as InSb and PbTe, consistent with previous results.³⁰ Table I points to Al₂O₃ or (Al,B,Ga)N as being particularly chemically stable for stanene growth. First principles calculations also confirm the stability of Al₂O₃ as a substrate for stanene.⁵⁶ Considering monolayer stanene has rarely been synthesized on scales larger than hundreds of nanometers,^{23,29,48,52,53,55} this work may provide important guidance to finding more appropriate substrates for large area, thin film deposition of this material.

In summary, we find that bulk thermodynamics promotes a reaction of Sn with Bi₂Te₃. We also find that Sn can intercalate in Bi₂Te₃ up to 10%–17% stanene monolayers per quintuple layer, which may have interesting properties in the topological insulator Bi₂Te₃. We predict other hexagonal materials that may serve as suitable, inert substrates for stanene growth.

See the supplementary material for RHEED images of 7 × 7 Si(111) and c(2 × 8) Ge(111) reconstructed substrates and Bi₂Te₃ grown on the respective substrates, XPS surveys of Te-capped Bi₂Te₃ and decapped Bi₂Te₃, and atomic profile and in-plane orientation of Sn-intercalated Bi₂Te₃.

This work was supported by the Function Accelerated nanoMaterial Engineering (FAME), Air Force Office of Scientific Research (AFOSR), No. FA9550-15-1-0472, and the Center for Research on Interface Structure and Phenomena (CRISP), a National Science Foundation (NSF) Materials Research Science and Engineering Center (MRSEC), No. DMR-1119826. S.E. acknowledges support from the NSF Graduate Research Fellowship, No. DGE1122492. The XRD data were taken at beamline 33ID-E at the Advanced Photon Source and supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences, under Contract No. DE-AC02-06CH11357.

The authors declare no competing interests.

The data that support the findings of this study are available from the corresponding author upon reasonable request.