In this work, we present an angle-resolved photoemission spectroscopy study of a 1T′-WTe2 monolayer epitaxially grown on NbSe2 substrates, a prototypical quantum spin Hall insulator (QSHI)/superconductor heterojunction. Angle-resolved photoemission spectroscopy data indicate the formation of electronic states in the bulk bandgap of WTe2, which are absent in the nearly free-standing WTe2 grown on the highly oriented pyrolytic graphite substrate, where an energy gap of ∼100 meV is reported. The results are explained in terms of hybridization effects promoted by the QSHI–superconductor interaction at WTe2/NbSe2 interfaces, in line with recent scanning probe microscopy investigation and theoretical band structure calculations. Our findings highlight the important role of interlayer interaction on the electronic properties and ultimately on the engineering of topological properties of the QSHI/superconducting heterostructure.
I. INTRODUCTION
Owing to their layered structure, van der Waals materials offer a unique platform to explore new quantum states of matter. Held together only by weak van der Waals forces in the bulk, individual atomically thin monolayers (MLs) often demonstrate properties uncommon to their bulk counterparts. This is well exemplified by a ML of 1T′-WTe2, which is a two-dimensional (2D) quantum spin Hall insulator (QSHI),1 a topologically nontrivial quantum material characterized by helical edge states protected by time-reversal symmetry and the bulk (as opposed to the edge) bandgap opening due to the strong spin–orbit coupling.1,2 Remarkably, inducing superconductivity in the helical edge states of QSHI has been theoretically predicted1 to result in a topological superconductor, a highly sought-after state of matter3 and a potential host for elusive Majorana fermions.4,5 In this context, heterostructures of a QSHI with an s-wave superconductor (SC) offers a practical approach to engineering a topological superconducting state from various QSHI/SC combinations.6–8 Among these, van der Waals heterostructures stand out as material systems with an atomically sharp, transparent interface that preserves the structural integrity of 2D constituents.9 However, surface reconstruction,10,11 electric fields induced by charge transfer, as well as proximity effects may strongly alter their electronic band structure.12 Recent scanning tunneling spectroscopy (STS) studies of van der Waals QSHI/SC heterostructure, WTe2/NbSe2, showed a clear signature of superconducting bulk and edge states,7,8 suggesting that interlayer coupling and hybridization are strong across the van der Waals interface and are a promising candidate for the engineering of stable induced superconductivity at large pairing strength. Indeed, for both nanofabricated7 and MBE grown samples,8 STS indicates the formation of a residual 2D density in the nominally insulating 2D QSH bulk of WTe2/NbSe2 heterostructures.
Here, we present a detailed measurement of the electronic band structure of the WTe2/NbSe2 heterostructure by angle-resolved photoemission spectroscopy (ARPES) for the first time to gain further insight into interface coupling. New electronic states are observed in the bulk bandgap of ML WTe2 that are otherwise absent in the nearly free-standing WTe2 grown on the highly oriented pyrolytic graphite (HOPG) substrate. Theoretical analyses support the above observation, assigning the new states to the interlayer hybridization at the QSHI/SC interface that results in the formation of metallic states in the nominally gapped WTe2.
II. RESULTS AND DISCUSSION
We performed ARPES measurements on a 1T′-WTe2 grown on a NbSe2 (WTe2/NbSe2 in the following) and HOPG substrate (WTe2/HOPG) by molecular beam epitaxy. The details of the sample preparation procedure and photoemission spectroscopy equipment are described in Sec. S1 (Fig. S1) and Sec. S2 of the supplementary material, respectively, identical to what was reported in Ref. 8. On both substrates, a coverage of 0.7 ML was achieved, as indicated by the Scanning Tunneling Microscopy (STM) characterization after sample preparation [Figs. S1(a) and S1(b) of the supplementary material].
Figures 1(a) and 1(b) show the schematic of the Surface Brillouin Zone (SBZ) of (a) NbSe2 and (b) WTe2/NbSe2. The NbSe2 and WTe2 SBZ were reconstructed from the corresponding lattice constant values as extracted by the Scanning Tunneling Microscopy (STM) image analysis of the samples prepared under the same conditions8 (see also Fig. S2 of the supplementary material), i.e., aNbSe2 = 3.4 Å, aWTe2 = 3.5 Å, and bWTe2 = 6.2 Å. The epitaxial growth condition results in the bWTe2 direction aligned along the NbSe2 sublattices, i.e., the ΓΥ direction of 1T′-WTe2-SBZ parallel to the ΓΚ direction of the NbSe2-SBZ (Fig. S2 of the supplementary material). However, due to the symmetry mismatch between the twofold rotational symmetry of the 1T′-WTe2 ML rectangular unit cell and the threefold symmetry of the NbSe2 substrate, three energetically equivalent domains rotated by 120° [indicated as D1, D2, and D3 in Fig. 1(b)] exist.
Figure 1(c) reports the Fermi surface of the NbSe2 crystal as obtained by ARPES via momentum-resolved, constant-energy intensity mapping around EB = 0 eV (±10 meV) measured at 11 K. At low temperatures, the Fermi surface contour is enhanced due to reduced energy thermal broadening.13 A hexagonal-like intensity pattern is observed reflecting the structure of the NbSe2 SBZ in Fig. 1(a) with clearly distinguishable ΓΚ and ΓΜ directions [Fig. 1(c)]. Upon WTe2 ML deposition on NbSe2 a hexagonal-like Fermi surface is also observed [Fig. 1(d)], in apparent contrast with the rectangular symmetry of the WTe2 ML-SBZ. The constant–energy map in Fig. 1(d) originates from the superimposition of the rectangular Fermi surfaces [Fig. 1(b)] of the 120°-rotated WTe2 ML single-crystal domains (∼20 nm lateral size, see Fig. S1 of the supplementary material) under the macroscopic area probed by ARPES (beam spot size ∼0.8 mm, see experimental details in Sec. S2 of the supplementary material). A similar Fermi surface contour was reported previously in the ARPES study of WTe2 ML epitaxially grown on a threefold symmetry graphene substrate.14
The in-plane multidomain orientation may, in principle, affect the band structure measurements of WTe2 on NbSe2, leading to the superimposition of the band dispersions along both the high-symmetry and non-high-symmetry directions of the SBZ [Fig. 1(b)]. In previous ARPES investigations of the 2D and 3D layered materials with similar in-plane multidomain structure, only band dispersions along the high-symmetry directions were detected.15–17 The results were attributed to the higher density of states (DOS) along the high-symmetry directions with respect to other regions of the SBZ, resulting in the enhancement of the corresponding ARPES signal.15–17 To minimize such superimposition effects, the band dispersion of WTe2 on NbSe2 was measured along the ΓΥ direction of the D1 domain, i.e., the ΓΥ(1) [Fig. 1(b)] of WTe2 parallel to the ΓΚ of the NbSe2 surface [Fig. 1(a)], where only partial superimposition with no high-symmetry directions of equivalent rotated domains D2 and D3 is present [Fig. 1(b)]. The as-obtained electronic band structure for NbSe2 and WTe2/NbSe2 at 11 K is plotted in Figs. 1(e) and 1(f). The raw Energy Distribution Curves (EDC) at different k// were normalized to the intensity of highest point within the plotted energy range, and the as obtained data are plotted in log scale to better capture the low intensity spectral features near the Fermi level (EF).
ARPES data of WTe2/HOPG at 11 K [Fig. 1(g), log-scale ARPES intensity plot] are also included for comparison purposes, where the full in-plane random orientation of WTe2 domains and the corresponding SBZs led to the observed circular-like pattern in the constant energy ARPES intensity map [Fig. 1(h)].
Despite the WTe2/HOPG multi-domain nature, a clear band structure is detectable. The experimental data are well described by the superimposition of the DFT calculated ΓΧ (dotted dark blue line) and ΓΥ (purple dashed line) band dispersion of a free-standing WTe2 ML, their relative contribution to the measured ARPES intensity depending on the DOS as well as on the photoemission cross section and symmetry of the corresponding band wavefunctions.14,17 The coexistence of a multidomain structure with measurable band dispersions is in line with previous ARPES study on 2D multidomain materials,15–17 while the agreement between experimental data and theoretical band structure in Fig. 1(g) indicates weak interaction between the WTe2 ML and HOPG, as reported in a previous ARPES study of the WTe2/graphene interface.14
The NbSe2 band structure signal [Fig. 1(e)] shows overall good agreement with the Density Functional Theory (DFT)-based prediction [white dotted lines in Fig. 1(e) (see Sec. S4 of the supplementary material for DFT calculation details) within the limit of first principles band structure calculations and ARPES signal intensity modulation reflecting the photon energy/polarization effect and emission direction.13 We find that the NbSe2 signal is strongly suppressed after WTe2 layer deposition (thickness ∼ 8 Å) due to the substrate coverage and the reduced photoelectron mean free path (∼4 Å−1) in the probed kinetic energy range (10–20 eV).18 In WTe2/NbSe2, a single, large dispersive band (width >1 eV) is instead visible [Fig. 1(f)] apparently crossing EF along ΓY(1) directions. By comparison with the ARPES data of WTe2/HOPG [Fig. 1(g)], a good correspondence with the ΓΥ-low binding energy valence band (VB) of the nearly free-standing WTe2 ML [Fig. 1(g)] can be found once an energy shift toward EF is considered. Noteworthy is the very weak signal detected from the ΓΧ band at ∼ 1 eV with respect to the HOPG case, in line with the reduced in-plane azimuthal disorder of the WTe2 ML on NbSe2 [Fig. 1(b)].
To clarify the above experimental findings, the electronic structure near EF of WTe2/NbSe2 and WTe2/HOPG was investigated in greater detail. Figure 2 reports the ARPES band mapping of [(a) and (b)] WTe2/HOPG and [(f) and (g)] WTe2/NbSe2 as measured at [(a) and (f)] 297 K and [(b) and (g)] 11 K, respectively. The data were obtained by dividing the original ARPES signal for an energy resolution-convoluted Fermi–Dirac (FD) distribution at the measurement temperature to remove the intensity cut-off at EF and highlight the hitherto-hidden valence band (VB) and conduction band (CB) features near EF.19
The ARPES data of WTe2/HOPG exhibit a clear intensity drop in the absence of quasiparticle peak structure above the VB maximum at Γ point [EB = 0.1 eV, see energy distribution curve (EDCs) in Fig. 2(c)] up to EB ∼ 0.0 eV, where the photoemission signal from the occupied and/or thermally populated CB states near EF are detected. The result agrees well with the band structure calculation [dotted lines in Figs. 2(a) and 2(b)] of free-standing WTe2 ML predicting an energy gap along the ΓΥ direction. From the analysis of ARPES momentum energy distribution curves [MDCs, see Figs. 2(d) and 2(e)] (see Sec. S5 of the supplementary material), an energy gap value of Egap = 0.10 ± 0.04 eV is estimated, which agrees well with the previously reported values for exfoliated and deposited WTe2 ML.20
The ARPES band mapping of WTe2/NbSe2 [Figs. 2(f) and 2(g)] shows good agreement with the calculated ΓΥ VB of the free-standing WTe2 ML [dotted lines in Figs. 2(f) and 2(g)] with a shift of 0.1 eV toward EF with respect to the HOPG case [see EDC in Fig. 2(h)]. However, no clear ARPES intensity reduction or lack of quasiparticle peak is observed above the VB maximum located at EB = 0 eV as confirmed by the MDC analysis at 297 and 11 K [Figs. 2(i) and 2(j)]. Remarkably, ARPES data for the WTe2/NbSe2 interface indicate a higher density of states in the bulk energy gap when compared to the free-standing ML case, represented by WTe2 on HOPG. These new states may originate from the layer–substrate electronic hybridization effects in the WTe2/NbSe2 heterostructure.
In support of these findings, tight-binding band structure calculations were conducted for the bulk states of a free-standing WTe2 ML [Fig. 2(k)] and WTe2/NbSe2 heterostructure [Fig. 2(l)] (see Sec. S6 of the supplementary material for calculation details). For a clear comparison with the experiment, the binding energy scale of theoretical curves was shifted to match the experimental VB position at the Γ point.
In comparison to free-standing WTe2 ML [Fig. 2(k)], tight binding calculations for WTe2/NbSe2 [Fig. 2(l)] clearly shows the VB and CB states “smearing out” in the original bulk energy gap due to the layer–substrate interaction, thus, resulting a non-negligible density of states consistent with both our ARPES measurements and earlier scanning probe investigation.8
To better illustrate the above findings, we compared the calculated DOS of ML WTe2 and WTe2/NbSe2 with the momentum-integrated ARPES signals, which reflect the energy levels’ population in the examined heterostructures.13 Figure 3 shows the tight-binding calculated DOS [Figs. 3(a) and 3(d)] with the WTe2/HOPG and WTe2/NbSe2 momentum-integrated ARPES signal at 297 K [Figs. 3(b) and 3(e)] and 11 K [Figs. 3(c) and 3(f)] as obtained from experimental data presented in Figs. 2(a), 2(b), 2(f), and 2(g). ARPES data integration was limited to ±0.4 Å−1 around the Γ point corresponding to the energy/momentum gap region.
At both 297 and 11 K, the integrated ARPES curves of WTe2/HOPG monotonously drop above the VB maximum [vertical bar at EB ≈ 0.1 eV in Figs. 3(a)–3(c)], matching qualitatively well with the theoretical DOS of the freestanding WTe2 ML [Fig. 3(a)], and thus highlighting the gapped nature of the bulk of WTe2 ML on HOPG. For the WTe2/NbSe2 interface, however, the non-monotonous structure of VB edge and in-gap electronic states of theoretical DOS [marked by arrow in Fig. 3(d)] finds a good correspondence in the intensity peak of the momentum-integrated ARPES data at 297 K [Fig. 3(e)]. For 11 K integrated ARPES data of WTe2/NbSe2, a clear VB maximum is visible, yet the energy gap density is hidden by the signal divergence above EF, which was introduced by the FD renormalization of the low temperature experimental data. Finally, the comparison with HOPG and NbSe2 substrate data (see Fig. S7 for the corresponding ARPES raw data), also included in Figs. 3(b)–3(d), and 3(f), indicates their negligible contribution to the overall measured interface DOS even without considering the expected reduction of the substrate photoemission signal due to the WTe2 deposition.18 For the NbSe2 case, the attenuated contribution of NbSe2 () to the measured WTe2/NbSe2 ARPES signal around the Γ point can be estimated as , where C is the WTe2 coverage (expressed in ML), d is the WTe2 thickness (∼8 Å according to Ref. 8), and λ is the electron mean free path at the measured kinetic energy of photoelectrons (∼4 Å).18 For C = 0.7 ML (see Sec. S1 of the supplementary material), one has . According to the above calculations, the NbSe2 substrate contributes only to ∼30% of the total WTe2/NbSe2 ARPES signal around the Γ point [see Figs. 3(e) and 3(f)]. In this context, the NbSe2 substrate has a reduced impact on the DOS measured in the energy gap of the WTe2 on NbSe2 as well on the large difference in the gap DOS between the WTe2/NbSe2 and WTe2/HOPG system.
In principle, the observed ARPES signal in the energy gap region could also be associated with a larger energy broadening of the VB EDCs of WTe2/NbSe2, reflecting a poorer structural quality (i.e., defects, surface adsorbates, etc.) of the WTe2 layer grown on the NbSe2 substrate when compared to the growth on HOPG.21,22 Due to energy broadening, introduced by the defect-mediated modification of electronic states distribution and/or photoelectron scattering, the low energy tail of photoemission peaks may extend up to tenths of meV away from the VB maximum position, thus leading to an apparent increase in the photoemission signal inside the bulk energy gap. In this context, the 297 K- and 11 K-EDCs at the Γ point of WTe2/HOPG and WTe2/NbSe2 [Fig. 2] were analyzed by the least square method and its full width at half maximum peak width (FWHM) was determined by Gaussian peak fitting. The results are reported in Figs. 3(g) and 3(h), showing, at all measured temperatures, comparable FWHMs values for the different telluride films, i.e., ∼ 0.15 eV at 297 K and ∼0.1 eV at 11 K, attesting to the quality of our WTe2 films on both substrates and confirming that the additional 2D DOS observed in the bulk arises from layer–substrate hybridization largely absent in the WTe2/HOPG interface.
Finally, we briefly comment on the VB energy difference of the WTe2/HOPG and WTe2/NbSe2 interface [Figs. 2(a)–2(d)]. In principle, the ∼0.1 eV “p-type like” VB shift observed in the WTe2/NbSe2 can also be related to a higher defect density as recently reported in photoemission studies of 2D semiconducting materials.21 However, this is less likely on the basis of the peak fitting analysis presented above, indicating a comparable quality between the two fabricated heterostructures. Alternatively, the results can also be determined by the layer–substrate hybridization process and related gap states formation, which may alter the EF position within the energy gap of the WTe2 on NbSe2 with respect to the HOPG case. In that regard, we also note that a comparable p-type like energy shift (∼50 meV) was reported in previous STS mapping of WTe2/NbS2 and WTe2/HOPG bulk DOS (see Fig. S8) in line with the present ARPES observation. More detailed investigation on the energy level alignment mechanism at the interface is required and in progress for complete understanding of the above observation.
To conclude, we have demonstrated evidence of WTe2 and NbSe2 hybridization using ARPES, in excellent agreement with prior scanning probe investigation and theoretical calculations. Our results demonstrate the suitability of ARPES technique as a prime tool to investigate the electronic band structure of the QSHI/superconducting electronic structure, complementary to the recent scanning probe investigations results.7,8
III. CONCLUSION
In this work, we reported an ARPES investigation of the electronic properties of the 1T′-WTe2/NbSe2 heterostructure, a prototypical QSHI/superconducting interface, produced by molecular beam epitaxy. The formation of electronic states in the original bandgap of 1T′-WTe2 ML is detected as confirmed by comparison with the ARPES study of the nearly free standing 1T′-WTe2 ML on graphite. The results are explained in terms of layer–substrate hybridization at the ML/substrate interfaces, resulting in the formation of metallic states in the nominally gapped WTe2 ML. The present results show direct evidence of the perturbation of QSHI’s electronic properties via substrate interaction and provides insight on the interaction between ML WTe2 and NbSe2 and its impact on the electronic band structure. We also demonstrated ARPES as a suitable tool for increasing our understanding of the QSHI/superconductor interface beyond the nanoscale-limited information provided by scanning probe-based investigations.
SUPPLEMENTARY MATERIAL
See the supplementary material for the details on the experimental setup, sample preparation, STM and STS data, and band structure calculations.
ACKNOWLEDGMENTS
This research was supported by the National Research Foundation (NRF) Singapore, under the Competitive Research Program “Toward On-Chip Topological Quantum Devices” (Grant No. NRF-CRP21-2018-0001) F.B., H.K., I.V., and K.E.J.G. acknowledge the funding support from the Agency for Science, Technology and Research (Grant No. 21709). B.W. acknowledges support from the Singapore Ministry of Education (MOE) Academic Research Fund Tier 3 Grant (No. MOE2018-T3-1-002) and acknowledges the Singapore National Research Foundation (NRF) Fellowship (Grant No. NRF-NRFF2017-11). H.K. acknowledges the A*STAR Computational Resource Center (A*CRC) for computational resources and support.
AUTHOR DECLARATIONS
Conflict of Interest
The authors have no conflicts to disclose.
Author Contributions
Fabio Bussolotti: Data curation (lead); Formal analysis (lead); Investigation (lead); Methodology (lead); Writing – original draft (lead); Writing – review & editing (lead). Hiroyo Kawai: Methodology (equal); Writing – review & editing (equal). Ivan Verzhbitskiy: Writing – review & editing (equal). Wei Tao: Writing – review & editing (equal). Duc-Quan Ho: Writing – review & editing (equal). Anirban Das: Writing – review & editing (equal). Junxiang Jia: Writing – review & editing (equal). Shantanu Mukherjee: Writing – review & editing (equal). Bent Weber: Conceptualization (equal); Writing – review & editing (equal). Kuan Eng Johnson Goh: Conceptualization (equal); Writing – review & editing (equal).
DATA AVAILABILITY
The data that support the findings of this study are available from the corresponding authors upon reasonable request.