betaFeedback

WorksPhysics and Astronomy

Charge order with unusual star-of-David lattice in monolayer NbTe2

Abstract

Interplay between fermiology and electron correlation is crucial for realizing exotic quantum phases. Transition-metal dichalcogenide (TMD) 1T-TaS2 has sparked a tremendous attention owing to its unique Mott-insulating phase coexisting with the charge-density wave (CDW). However, how the fermiology and electron correlation are associated with such properties has yet to be claried. Here we demonstrate that monolayer 1T-NbTe2 is a new class of two-dimensional TMD which has the star-of-David lattice similarly to bulk TaS2 and isostructural monolayer NbSe2, but exhibits a metallic ground state with an unusual lattice periodicity root19xroot19 characterized by the sparsely occupied star-of-David lattice. By using angle-resolved photoemission and scanning-tunneling spectroscopies in combination with first-principles band-structure calculations, we found that the hidden Fermi-surface nesting and associated CDW formation are a primary cause to realize this unique correlated metallic state with no signature of Mott gap. The present result points to a vital role of underlying fermiology to characterize the Mott phase of TMDs.

One of key challenges in materials science is to find outstanding two-dimensional (2D) materials by reducing the dimensionality from bulk (3D) to 2D, as highlighted by the discovery of room temperature quantum Hall effect in graphene[1]. Transition-metal dichalcogenides (TMDs) offer a fertile platform to explore exotic 2D materials, since two-dimensionalization of bulk TMDs often creates fundamentally different physical properties such as the Ising superconductivity associated with the space-inversion-symmetry breaking in monolayer 1H1H-NbSe2_2 [2] and the quantum spin Hall insulator phase in monolayer 1T′1T'-WTe2_2 and 1T′1T'-WSe2_2 [3, 4]. When electron correlation is introduced into 2D systems, even more exotic quantum states would emerge, as exemplified by the discovery of superconductivity on the verge of Mott-insulating phase in twisted bilayer (BL) graphene, where the enhanced electron correlation due to the band narrowing by moiré potential plays a key role[5]. A 2D Mott-insulating state that coexists with charge-density wave (CDW) in monolayer 1T1T-NbSe2_2 and TaSe2_2 has been also discussed to be triggered by the enhanced electron correlation due to the CDW-induced band narrowing[6, 7, 8].

CDW-Mott phase in TMDs is still far from being well understood[6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19], owing to the existence of complex energy bands in the CDW phase. It is still unknown how electron correlation, CDW, and Mott phases are interrelated. It is thus highly important to pin down a key ingredient to realize the Mott phase by exploring ultrathin 2D TMDs. In this regard, monolayer 1T1T-NbTe2_2 is a promising target because it is isostructural to Mott-insulating monolayer 1T1T-NbSe2_2. Bulk NbTe2_2 crystallizes in the 1T1T structure at high temperatures. Below 530 K, it undergoes a structural transition to the monoclinic 1T′1T' phase with the 3×13 \times1 periodic lattice distortion[20, 21, 22] associated with the CDW triggered by the Fermi-surface (FS) nesting[23]. Superconductivity was observed below TC=0.5T_{\mathrm{C}} = 0.5 K[ref.24. Despite such intensive studies of bulk properties, basic properties such as crystal structure and ground-state characteristics of monolayer NbTe2_2 remain unexplored. It is also important to clarify the electronic states of isostructural Se- and Te-based TMDs to understand how the difference in the band character between Se and Te manifests itself in the exotic physical properties such as Mott insulating properties and CDW.

A central player for such unique CDW-Mott phase is the star-of-David (soD) cluster of transition-metal atoms, where the corner atom is slightly displaced from the original position towards the central atom [see Figure 3(d)]. In this soD cluster, the half-filling condition to realize the Mott-insulating phase is satisfied, because, as exemplified in the case of bulk 1T1T-TaS2_2, twelve electrons at the displaced twelve Ta atoms form 6 fully occupied bands and remaining one electron at the central Ta atom forms a half-filled metallic band[9, 10, 11, 12]. However, the mechanism of

Schematic and measurements of monolayer NbTe₂ and NbSe₂

Figure 3. (a) Schematics of monolayer NbTe2_2 islands on BL graphene. (b) STM image in the surface area of 10×1010 \times10 nm2^2 (Vs=−1.5V_{\mathrm{s}} = -1.5 V and It=0.8I_{\mathrm{t}} = 0.8 nA). Inset shows Fourier-transform image of (b). (c) Calculated Fermi surface for monolayer 1T-NbTe2_2 (red curves), overlaid on the ARPES-intensity plot. The 19×19R23.4∘\sqrt{19} \times\sqrt{19} R 23.4^\circ nesting vector is shown by green arrows. (d) STM image magnified in the area shown by red square in (b). VsV_{\mathrm{s}} and ItI_{\mathrm{t}} were set to be -1.0 V and 0.8 nA, respectively. Inset shows the illustration of soD cluster. (e),(f) Comparison of the schematic soD lattice between monolayer NbTe2_2 and NbSe2_2. (g),(h) Near-EFE_{\mathrm{F}} ARPES intensity along the ΓM\Gamma M cut for monolayer NbTe2_2 and NbSe2_2, respectively, measured at T=30T = 30 K. (i) EDC at the Γ\Gamma point for monolayer NbTe2_2 and NbSe2_2. (j) Schematics on the relationship between the band dispersion and the soD lattice for monolayer NbTe2_2 and NbSe2_2.

In this paper, we report angle-resolved photoemission spectroscopy (ARPES) and scanning tunneling microscopy (STM) studies on the electronic state of monolayer 1T1T-NbTe2_2 fabricated on BL graphene. We uncovered the metallic 19×19\sqrt{19} \times\sqrt{19} CDW state coexisting with an unusual soD lattice distortion in 1T1T-NbTe2_2 in stark contrast to the gapped Mott-insulating nature of monolayer 1T1T-NbSe2_2. We discuss implications of the present results to address the origin of unique CDW-Mott phase in 2D TMDs.

Monolayer NbTe2 films were grown on BL graphene by the molecular-beam-epitaxy (MBE) method. ARPES measurements were carried out using a MBS-A1 electron-energy analyzer (MB Scientific AB) at Tohoku University and a DA-30 electron-energy analyzer (Omicron-Scienta) at beamline BL28 in Photon Factory, KEK. STM measurements were carried out using a custom-made ultrahigh vacuum STM system[25]. First-principles band-structure calculations were carried out by using the Quantum Espresso code package[26] For details, see Appendix A.

First, we present fabrication and characterization of monolayer NbTe2. To fabricate a NbTe2 film, we used the van der Waals epitaxy technique by using BL graphene grown on silicon carbide as a substrate [Fig. 1(a)]. Figure 1(b) shows the RHEED (reflection high energy electron diffraction) pattern of BL graphene on 6H-SiC(0001) which signifies a 1×11 \times1 streak pattern together with a weaker 63×63R30∘6\sqrt{3} \times6\sqrt{3}R30^\circ pattern originating from BL graphene and carbon-mesh layer beneath it, respectively. Corresponding spots are also visible in the LEED (low-energy electron diffraction) pattern in Figure 1(c). After co-evaporation of Nb and Te atoms onto the substrate kept at 300 ∘300\,^\circC under ultrahigh vacuum, the intensity of 63×636\sqrt{3} \times6\sqrt{3} spot is reduced and a new 1×11 \times1 pattern appears [Figs. 1(d)-(f)]. This behavior is characteristic of TMD ultrathin films, as observed in various monolayer TMDs such as NbSe2 and TaSe2 [6, 7, 8]. STM measurements revealed the formation of monolayer NbTe2 islands [Fig. 1(g)] whose height is ∼1.0\sim1.0 nm [Fig. 1(h)], in rough agreement with the distance between adjacent NbTe2 layers in bulk 1T1T-NbTe2 (∼0.7\sim0.7 nm)[28], supporting its monolayer nature (note that another 0.4-nm-height step originates from the step of the SiC substrate).

Schematic view, diffraction patterns, STM image, ARPES data, and crystal structure of monolayer NbTe2

Figure 1. (a) Schematic view of BL graphene on SiC. (b),(c) RHEED and LEED patterns of bilayer (BL) graphene. (d)-(f) Same as (a) to (c) but for NbTe2 on BL graphene. (g) Constant-current STM image at T=4.8T = 4.8 K in the surface area of 56×10056 \times100 nm2^2 (sample bias voltage Vs=−2V_s = -2 V, and set-point tunneling current It=50I_t = 50 pA). (h) Height profile along a cut indicated by red arrow in (g). (i) EDCs for monolayer NbTe2 measured at T=30T = 30 K with hν=72h\nu= 72 eV along the ΓM\Gamma M line of NbTe2 Brillouin zone (inset). (j) Plot of valence-band ARPES intensity for monolayer NbTe2, compared with the calculated band structure for monolayer 1T1T-NbTe2 (red curves) and BL graphene (blue curves). Experimental three holelike bands are labeled as α\alpha, β\beta, and γ\gamma in (i) and (j). (k) Schematic crystal structure of monolayer 1T1T-NbTe2.

We characterized the overall band structure of monolayer NbTe2 film by in-situ angle-resolved photoemission spectroscopy (ARPES). Figure 1(i) displays the energy distribution curves (EDCs) at T=30T = 30 K measured at hν=72h\nu= 72 eV along the ΓM\Gamma M cut of hexagonal NbTe2 Brillouin zone. Besides the band structure originating from BL graphene situated at the binding energy (EBE_B) higher than ∼3\sim3 eV, one can clearly identify several dispersive features originating from NbTe2 within 3 eV of the Fermi level (EFE_F). There exist three holelike bands centered at the Γ\Gamma point; one has a shallow dispersion (labeled as α\alpha) within 1 eV of EFE_F and appears to cross EFE_F in the vicinity of the Γ\Gamma point, and the other two have a wider dispersion each topped at EB=0.5E_B = 0.5 and 2.5 eV at the Γ\Gamma point (labeled as β\beta and γ\gamma, respectively). To see more clearly the dispersive features, we plot in Figure 1(j) the ARPES intensity as a function of kxk_x and EBE_B, together with the calculated band structure for monolayer 1T1T-NbTe2 (red) and BL graphene (blue). One can immediately recognize that the intense features at higher EBE_B’s well overlap with the calculated π\pi and σ\sigma bands of BL graphene. A qualitative matching can be also found for the NbTe2 bands; the calculated holelike Te 5p5p bands topped at 0.2 eV and 2.1 eV seem to have experimental counterparts although there exist quantitative differences in their energy positions. The calculated holelike bands crossing EFE_F which are attributed to the Nb 4d4d (outer) and Te 5p5p (inner) orbitals also have a good correspondence with the experimental band which crosses EFE_F around the Γ\Gamma point. Such an overall agreement between the experiment and calculation suggests that monolayer NbTe₂ on BL graphene takes the 1T1T structure [Figure 1(k)], as corroborated by our STM observation of soD clusters [Figure 3] which are known to be stabilized only in the 1T1T structure. It is also inferred from a reasonable matching in the energy position of bands around the Γ\Gamma point between experiments and calculations shown in Figure 1(j) that our film is almost stoichiometric (note that a slight disagreement at around the Γ\Gamma point in higher EBE_B’s may be explained in terms of the band renormalization due to the electron correlation). The 1T1T nature of our NbTe₂ film is also supported by the fact that the experimental band structure shows a better agreement with the calculated band structure for the 1T1T phase than that for the 1H1H phase (for details, see Appendix B). Also, the EFE_F-crossing of bands in Figure 1(i) and Figure 1(j) is distinct from the large Mott gap exceeding 0.2 eV in isostructural monolayer 1T1T-NbSe₂ and 1T1T-TaSe₂[6, 7, 8]. We will come back to this point later.

Although the experimental band structure around the Γ\Gamma point shows a reasonable agreement with the calculation for monolayer 1T1T-NbTe₂, we found a fatal disagreement of the Fermi-surface topology around the KK point. The experimental Fermi surface of monolayer 1T1T-NbTe₂ is characterized by the existence of a hole pocket only at the Γ\Gamma point. The holelike nature is directly visualized by the equi-energy contour plot at T=30T = 30 K in Figure 2(a) showing a systematic expansion of the intensity pattern on increasing EBE_B. As shown in Figure 2(b), while the calculated Fermi surface consists of a large triangular hole pocket centered at the KK point besides the hexagonal one at the Γ\Gamma point, the spectral weight corresponding to this triangular Fermi surface is missing in the ARPES intensity. To further examine this unusual behavior, we compare the ARPES intensity near EFE_F along several high-symmetry kk cuts (cuts A-C) with the corresponding calculated band structure in Figure 2(c)-Figure 2(e). Although the experimental band structure along the ΓM\Gamma M cut (cut A) shows a reasonable agreement with the calculation, the holelike Te 5p5p band along the ΓK\Gamma K cut (cut B) shows a much smaller group velocity and no shallow electron bands exist aside from the hole band. Moreover, along the MKMK cut (cut C), the spectral feature is broad and flat, and one cannot identify a predicted highly dispersive V-shaped band in the experiment. We have confirmed that the flat feature is robust against the variation in photon energy and light polarization. This suggests that it is not an artifact associated with the matrix-element effect of photoelectron intensity, but is an intrinsic nature of monolayer NbTe₂ (for details, see Appendix C). Temperature-dependent ARPES measurements at the kFk_F point along the MKMK cut shown in Figure 2(f) signify a broad peak at EB∼0.5E_B \sim0.5 eV which gradually smears out on increasing temperature but still survives at T=420T = 420 K. This peak is likely associated with the formation of CDW (for detailed discussion of its origin, see Appendix D). We have confirmed that the KK-centered pocket is absent in a wide temperature range, as highlighted by the Fermi-surface mapping [Figure 2(g)] as well as the ARPES intensity along the ΓK\Gamma K cut [Figure 2(h)] measured at T=440T = 440 K.

ARPES intensity plots and Fermi-surface mappings for monolayer 1T-NbTe₂

Figure 2. (a) ARPES-intensity plots as a function of 2D wave vectors (kxk_x and kyk_y) for representative EBE_B slices measured at T=30T = 30 K with hν=72h\nu= 72 eV. (b) Same as (a) but at EB=EFE_B = E_F, overlaid with the calculated Fermi surface (red curves). (c)-(e) ARPES intensity plots near EFE_F for monolayer NbTe₂ measured along the ΓM\Gamma M (cut A), ΓK\Gamma K (cut B), and KMKKMK (cut C) cuts shown in (b), respectively. Red curves correspond to the calculated band structure for monolayer 1T1T-NbTe₂. (f) Temperature dependence of EDC measured with hν=21.218h\nu= 21.218 eV at the kFk_F point of the calculated triangular pocket along the MKMK cut shown by a purple line in (e). (g), (h) Fermi-surface mapping and ARPES-intensity plot as a function of kyk_y and EBE_B, respectively, measured with hν=21.218h\nu= 21.218 eV at T=440T = 440 K for monolayer 1T1T-NbTe₂.

To clarify the mechanism behind the disappearance of a large triangular pocket, we have performed high-resolution STM measurements on a NbTe₂ island [Figure 3(a)]. The STM image at T=4.8T = 4.8 K in Figure 3(b) signifies individual Te atoms originating from the top layer of monolayer NbTe₂. Noticeably, the atomic image exhibits a strong intensity modulation; a dark region involving at least 7 Te atoms arranges periodically and forms a hexagonal superlattice surrounded by a brighter region with a honeycomb-shaped pattern. The Fourier-transformation image in the inset to Figure 3(b) signifies 19×19R23.4∘\sqrt{19}\times\sqrt{19}R23.4^\circ superpots (green circles) which correspond to the supercell shown by green rhombus in Figure 3(b). We found in Figure 3(c) that the 19×19R23.4∘\sqrt{19}\times\sqrt{19}R23.4^\circ superlattice vector well connects the parallel segments of calculated triangular pockets at the KK point, consistent with the calculated electronic susceptibility (see Appendix E) [note that the crystal geometry allows the existence of two types of crystal domains (R±23.4∘R\pm23.4^\circ) which equally satisfy the nesting condition, as detailed in Appendix F]. These suggest the formation of CDW triggered by the Fermi-surface nesting, which is also inferred from the observation of real-space charge reversal across the CDW gap similarly to the case of KV₃Sb₅ in the CDW phase[29] (see Appendix G). The CDW may fully gap out the triangular pocket and create many backfolded subbands. Electron correlation may further smear out the fine structure of each subband, leading to the featureless spectral intensity around the K point as seen in Figure 2(e). The CDW origin of disappearance of the triangular pocket needs to be experimentally confirmed by modulating the FS-nesting condition by carrier doping, electrical gating, or epitaxial strain. We have estimated the CDW transition temperature (TCDWT_{\mathrm{CDW}}) to be much above 420 K from the persistence of a peak in the EDC in Figure 2(f) (note that it was difficult to estimate TCDWT_{\mathrm{CDW}} from the LEED pattern because superspots were not clearly seen. Such a vague feature of LEED superspots despite the CDW formation was also recognized in some other monolayer TMDs[30, 31]).

To obtain further insights into the characteristics of CDW, we show in Figure 3(d) a magnified STM image. One may see that the dark region consists of twelve atoms which coincide well with the schematic double triangles rotated by 60∘60^\circ from each other (red broken triangles), a signature of soD cluster (inset) reported in bulk 1T-TaS2_2 [refs.11,14,15] as well as monolayer 1T-NbSe2_2 and 1T-TaSe2_2 [refs.6-8]. These soD clusters form a hexagonal lattice with the 19×19R23.4∘\sqrt{19} \times\sqrt{19} R23.4^\circ periodicity. As a consequence, residual atoms outside the soD clusters are seen as relatively bright spots intervening adjacent soD clusters. It is thus inferred that although isostructural family of monolayer NbTe2_2, NbSe2_2, and TaSe2_2 (together with bulk 1T-TaS2_2) commonly form the soD cluster, that for NbTe2_2 partially occupies the lattice [Fig. 3(e)]; 13 atoms among 19 atoms in the superstructure unit cell are involved in the soD cluster], in stark contrast to the full occupation in other TMDs [Fig. 3(f)][6, 7, 8, 11, 14, 15], demonstrating a unique characteristic of monolayer NbTe2_2. It is worthwhile to note that, although the 19×19R23.4∘\sqrt{19} \times\sqrt{19}R23.4^\circ periodicity is absent in bulk NbTe2_2, it was reported to locally emerge when the bulk sample was pulse heated by electron beam[27], and such periodicity was discussed in terms of the CDW driven by the FS nesting[28]. In this respect, it would be reasonable to infer that the soD modulation observed in monolayer is associated with the formation of CDW, whereas this point needs to be clarified by further experiments.

Now we discuss the relationship between the soD lattice and Mott characteristics. We suggest that the occupation of soD clusters is crucial for understanding the electronic properties at low temperature. Monolayer NbSe2_2 is an insulator as seen from the spectral-weight suppression around EFE_{\mathrm{F}} in the ARPES intensity [Fig. 3(h)] and the apparent energy-gap opening in the EDC at the Γ\Gamma point [Fig. 3(i)]. This insulating gap was attributed to the Mott-Hubbard gap[6], because the half-filling condition is satisfied for each soD cluster and there are no other conducting electrons because of the fully occupied nature of the soD lattice. On the other hand, in NbTe2\mathrm{NbTe}_2, half-filling condition for the Nb 4d4d orbital is globally violated because the Te 5p5p band participates in the Fermi surface. Also, electron hopping between the soD and outside-soD regions could disturb the effective half-filling nature in a single soD cluster and deteriorate the electron localization. These lead to the metallic nature as seen from the sizable spectral weight at EFE_{\mathrm{F}} in the EDC at the Γ\Gamma point [see Figure 3(i)].

The observed intriguing spectral difference between monolayer NbTe2\mathrm{NbTe}_2 and NbSe2\mathrm{NbSe}_2 is also explained in terms of fermiology. As shown in Figure 3(c), the calculated Fermi surface in the normal state for monolayer NbTe2\mathrm{NbTe}_2 consists of two hole pockets at the Γ\Gamma point associated with the EFE_{\mathrm{F}}-crossing of the topmost Nb 4d4d and Te 5p5p bands [schematically shown in Figure 3(j)], besides a large triangular pocket centered at the KK point due to the Nb 4d4d band. On the other hand, these small pockets at the Γ\Gamma point are absent in the calculation for monolayer NbSe2\mathrm{NbSe}_2 [32] because the Se bands are pulled downward and fully occupied [Figure 3(j)], as in the case of other chalcogenides[7, 33]. Since the total volume of Fermi surface must be identical between NbTe2\mathrm{NbTe}_2 and NbSe2\mathrm{NbSe}_2 according to the Luttinger theorem, such a difference causes the change in the volume of Nb 4d4d pocket and the resultant modification in the Fermi-surface topology[32]. This argument is also supported by estimating the carrier number in the observed Fermi surface. The result indicates 1.1 electrons/unit-cell for the Nb-4d4d triangular pocket in NbTe2\mathrm{NbTe}_2 while it is 1.0 electrons/unit-cell for NbSe2\mathrm{NbSe}_2. The deviation from the half-filling condition in NbTe2\mathrm{NbTe}_2 is ascribed to the extra Te 5p5p hole carriers. Such a difference in the fermiology leads to the difference in the Fermi-surface-nesting vector, i.e. 19×19R23.4∘\sqrt{19}\times\sqrt{19}R23.4^\circ for NbTe2\mathrm{NbTe}_2 and 13×13R13.9∘\sqrt{13}\times\sqrt{13}R13.9^\circ for NbSe2\mathrm{NbSe}_2. Since the small pockets at the Γ\Gamma point in NbTe2\mathrm{NbTe}_2 are not well connected with each other by the nesting vector, a gapless kk region remains in the CDW phase, contributing to the absence of Mott characteristics of NbTe2\mathrm{NbTe}_2 in contrast to NbSe2\mathrm{NbSe}_2 where the small pocket is intrinsically absent. In this regard, it is very important to systematically fabricate monolayer NbSe2−xTex\mathrm{NbSe}_{2-x}\mathrm{Te}_x films and investigate the Mott-transition characteristics as a function of xx.

One may wonder how the above interpretation based on the real space (occupation of soD) is reconciled with that of the momentum-space (fermiology). The small hole pocket at the Γ\Gamma point seen by ARPES may be attributed to the outside-soD region observed by STM. However, this argument is too simplistic because the electron localization in the real space is incompatible with the formation of energy band that requires non-local nature in the real space. This in return suggests that the soD and outside-soD regions are not phase separated and electrons can coherently hop between these two regions. This scenario seems consistent with the dI/dVdI/dV curves in the STM data which smoothly evolve across the boundary of two regions (see Appendix § 12). In this context, it is inferred that a small but finite atomic displacement takes place even in the outside-soD region. A sophisticated diffraction study is necessary to clarify this point.

Now we discuss the difference between the metallic NbTe2\mathrm{NbTe}_2 and Mott-insulating NbSe2\mathrm{NbSe}_2 in terms of the effective electron correlation. Since the on-site Coulomb energy UU for Nb 4d4d electrons is expected to be similar between NbSe2\mathrm{NbSe}_2 and NbTe2\mathrm{NbTe}_2, a crucial parameter to determine the effective coulomb energy U/WU/W is the bandwidth WW in the CDW phase. Since the reconstructed Brillouin zone for NbTe2\mathrm{NbTe}_2 is smaller than that for NbSe2\mathrm{NbSe}_2 due to the larger superlattice unit cell (19×19\sqrt{19}\times\sqrt{19} vs 13×13\sqrt{13}\times\sqrt{13}), the reconstructed subbands are expected to be flatter in NbTe2\mathrm{NbTe}_2 than in NbSe2\mathrm{NbSe}_2 when one assumes that the strength of CDW potential is similar. This leads to smaller WW and resultantly larger U/WU/W in NbTe2\mathrm{NbTe}_2 suggestive of the stronger “Mottness” in NbTe2\mathrm{NbTe}_2. However, this is opposite to our observation, suggesting a less important role of electron correlation to account for the difference between NbTe2\mathrm{NbTe}_2 and NbSe2\mathrm{NbSe}_2. It is noted though that this conjecture is not fully supported by the current experiment because each subband is hard to resolve due to the strong spectral broadening. Another alternative explanation to account for the difference between the metallic NbTe2\mathrm{NbTe}_2 and insulating NbSe2\mathrm{NbSe}_2 is that the Te 5p5p band with wide bandwidth (which is even wider than that of bulk 1T1T-TaS2\mathrm{TaS}_2 in the nearly commensurate CDW phase[13]) crosses EFE_{\mathrm{F}} and reduces the effective electron correlation to prevent this system from the Mott transition.

In conclusion, we have performed ARPES and STM combined with first-principles band calculations to study the electronic structure of monolayer 1T1T-NbTe2\mathrm{NbTe}_2 which is characterized by a partially occupied soD lattice possibly associated with the CDW formation. The present observation demonstrates that the formation of soD lattice is not a sufficient condition to realize the Mott-insulating phase which coexists with CDW, while it has been discussed as a necessary condition in bulk 1T1T-TaS2\mathrm{TaS}_2 and monolayer TMDs. We also found that the underlying Fermi-surface topology and associated hidden Fermi-surface nesting play a crucial role to control the key electronic properties such as periodicity of superlattice, occupation of soD lattice, and Mott-insulating vs metallic properties. The present result opens a pathway toward switching and controlling the Mott-insulating and CDW phases via fermiology engineering in ultrathin TMDs.

Acknowledgments

We thank Y. Nakata, M. Kitamura, K. Horiba, and H. Kumigashira for their help in the ARPES experiments. This work was supported by JST-CREST (no. JPMJCR18T1), JST-PRESTO (no. JPMJPR20A8), Grant-in-Aid for Scientific Research on Innovative Areas “Topological Materials Science” (JSPS KAKENHI Grant numbers JP15H05853, and JP15K21717), Grant-in-Aid for Scientific Research on Innovative Areas “Discrete Geometric Analysis for Materials Design” (JSPS KAKENHI Grant numbers 20H04624), Grant-in-Aid for Scientific Research (JSPS KAKENHI Grant numbers JP18H01821, and JP21H01757), Science research projects from Shimadzu Science Foundation, and World Premier International Research Center, Advanced Institute for Materials Research. T. Kawakami and T. Kato acknowledges support from GP-Spin at Tohoku University.

Sample fabrication, ARPES measurements, and band calculations

High-quality monolayer NbTe2_2 films were grown on bilayer (BL) graphene by the molecular-beam epitaxy (MBE) method in an ultrahigh vacuum of 5×10−105 \times10^{-10} Torr. BL graphene was grown by annealing an nn-type Si-rich 6H-SiC(0001) single-crystal wafer[6, 7] by resistive heating at 1100 °C for 30 min in an ultrahigh vacuum better than 1.0×10−91.0 \times10^{-9} Torr. Monolayer NbTe₂ film was grown by co-evaporating Nb and Te on the BL graphene substrate. The substrate temperature was kept at 300 °C during the epitaxy. The as-grown film was annealed at 300 °C for 30 min to improve the crystallinity, and then transferred to the ARPES-measurement chamber without breaking vacuum. The growth process was monitored in real-time by reflection high-energy electron diffraction (RHEED).

Scanning tunneling microscopy (STM) measurements were carried out using a custom-made ultrahigh vacuum STM system[25]. Te capping (∼10\sim10 nm thick) for surface protection of NbTe₂ film was removed in the STM chamber by Ar⁺ ion sputtering for 30 min and annealing at 250 °C for 30 min. STM measurements were carried out with an electrochemically etched W tip at 4.8 K under UHV below 2.0×10−102.0 \times10^{-10} Torr. All STM images were obtained in constant current mode. ARPES measurements were carried out using a MBS-A1 electron-energy analyzer with a high-flux helium discharge lamp at Tohoku University and a DA-30 electron energy analyzer with synchrotron radiation at the beamline BL-28A at Photon Factory (KEK). The He Iα resonance line (hν=21.218h\nu= 21.218 eV) and linearly polarized light of hν=72h\nu= 72 eV were used to excite photoelectrons. The energy and angular resolutions were set to be 16–30 meV and 0.2–0.3 °C, respectively. The Fermi level (EFE_{\mathrm{F}}) of samples was referenced to that of a gold film deposited onto the sample substrate.

First-principles band-structure calculations for freestanding monolayer 1T- and 1H-NbTe₂ were carried out by using the Quantum-Espresso code[26]. Spin-orbit interactions were included in the calculation. The plane-wave cutoff energy and the kk-point mesh were set to be 70 Ry and 12×12×112 \times12 \times1, respectively. Thickness of the inserted vacuum layer for monolayer was set to be ∼10\sim10 Å.

Comparison between ARPES intensity and band calculations for monolayer 1T/1H-NbTe₂

To clarify the possibility of different crystal structures, we have carried out band-structure calculations for monolayer NbTe₂ for 1T and 1H phases. Figure 4(a) and Figure 4(b) show the ARPES intensity along the ΓM cut overlaid by the calculated band structure for monolayer 1T- and 1H-NbTe₂, respectively [note that the calculation does not take into account the formation of star-of-David (soD) lattice]. One can see an overall agreement between experiment and calculation for the 1T case, such as a weakly dispersive holelike band at EB=EF−0.1E_{\mathrm{B}} = E_{\mathrm{F}} - 0.1 eV and a highly dispersive holelike band at EB=0.3E_{\mathrm{B}} = 0.3–1.5 eV, whereas the matching between experiment and calculation is relatively poor for the 1H case [Fig. 4(b)]. This supports the 1T nature of our NbTe₂ film.

ARPES intensities of monolayer NbTe₂ compared with calculated band dispersions for monolayer 1T-NbTe₂ and 1H-NbTe₂

Figure 4. (a), (b) ARPES intensities of monolayer NbTe₂ compared with the calculated band dispersion for monolayer 1T-NbTe₂ (red curves) and monolayer 1H-NbTe₂ (black curves), respectively.

Photon-energy and light-polarization dependences of band structure in monolayer NbTe₂

We performed ARPES measurements along the ΓK and KMK cuts with different sample-light geometry and photon energy (hνh\nu) to clarify the influence from the matrix-element effect of photoelectron intensity to the ARPES intensity. Figure 5(a) and Figure 5(b) show the ARPES intensity along the ΓK cut at hν=21.218h\nu= 21.218 eV with linear horizontal polarization and at hν=75h\nu= 75 eV with right circular polarization, respectively. At hν=21.218h\nu= 21.218 eV [Figure 5(a)], one can recognize a broad nondispersive feature around the K point at EBE_{\mathrm{B}} of ∼0.5\sim0.5 eV, in addition to a highly dispersive holelike band centered at the Γ point. This flat feature is likely associated with the formation of CDW. As shown in Figure 5(b), a similar flat feature is also observed at hν=75h\nu=75 eV around the KK point (indicated by white arrow), whereas its intensity is largely suppressed. As shown in Figure 5(c), such a weak and flat feature also exists along another ΓK\Gamma K cut rotated by 60∘60^\circ with respect to the ΓK\Gamma K cut in Figure 5(b). We find that the flat feature is also recognized along the MKMK cut with different hνh\nu’s and light polarizations, as shown in Figure 5(d)–Figure 5(f). These results suggest that the existence of observed flat feature (i.e. the absence of a triangular pocket predicted by the DFT calculation) is robust against the variation in photon energy and light polarization. This suggests that the flat feature is not an experimental artifact associated with the matrix-element effect of photoelectron intensity, but is an intrinsic feature of monolayer 1T1T-NbTe2_2.

ARPES intensities along the ΓK and MK cuts

Figure 5. (a), (b) ARPES intensities along the ΓK cut obtained at hν=21.218h\nu= 21.218 eV with linear horizontal light and at hν=75h\nu= 75 eV with right circular polarized light, respectively. (c) Same as (b) but obtained along another ΓK cut rotated by 60° with respect to the ΓK cut in (b). (d)–(f), Same as (a)–(c) cut, but measured along the MK cut. White arrows in (b) and (c) indicate the location of flat feature.

Intensity plot of the calculated imaginary part of electronic susceptibility

Figure 6. Intensity plot of the calculated imaginary part of electronic susceptibility Im⁡χ(q)\operatorname{Im}\chi(\mathbf{q}) for monolayer 1T1T-NbTe2_2. Arrows correspond to q=1/19 G\mathbf{q}=1/\sqrt{19}\,\mathbf{G} nesting vectors.

Origin of a flat band around the KK point for monolayer NbTe2_2

We discuss the origin of a flat feature at EB∼0.5E_B\sim0.5 eV around the KK point observed in Figure 2 and Figure 5. We have considered some possible mechanisms unrelated to the CDW, such as (i) the band gap, (ii) the Anderson gap, and (iii) the Mott-Hubbard gap. The possibility (i) is ruled out because the DFT calculation of monolayer NbTe2_2 fails to reproduce the flat band and a finite energy gap. The possibility (ii) is also unlikely because our NbTe2_2 film is close to stoichiometry and the ARPES spectrum around the KK point (Figure 2 of the main text) does not show a power-low behavior against EBE_B expected from the disorder-induced Coulomb gap. The possibility (iii) is also ruled out because the gap size Δ\Delta of ∼0.5\sim0.5 eV is too large compared to the Mott-Hubbard gap of isostructural 1T1T-NbSe2_2 (Δ∼0.1\Delta\sim0.1 eV) despite the expectation of a similar on-site Coulomb energy. Taking into account these points, we think that the flat band is associated with the CDW gap formation.

STM image and real-space and reciprocal-space configurations of NbTe2 superstructures

Figure 7. (a), STM image at T=4.8T=4.8 K in the surface area of 25×2525\times25 nm2^2 on NbTe2_2 island (Vs=−1.6V_s=-1.6 V, and It=0.1I_t=0.1 nA). (b) Real-space configuration of Nb atoms (open circles) and superlattice unit cell (red rhombus) for the 19×19R−23.4∘\sqrt{19}\times\sqrt{19}R-23.4^\circ superstructure and (c) corresponding reconstructed Brillouin zone (red) drawn on the original 1×11\times1 Brillouin zone (black). (d), (e) Same as (b) and (c), but for the R+23.4∘R+23.4^\circ superstructure.

Calculated electronic susceptibility for monolayer NbTe2_2

To clarify the origin of CDW in monolayer NbTe2_2, we calculated the electronic susceptibility χ(q)\chi(\mathbf{q}) based on the calculated Fermi surface shown in Figure 2(b). Figure 6 shows the imaginary part of calculated electronic susceptibility [Im⁡χ(q)][\operatorname{Im}\chi(\mathbf{q})] for monolayer 1T1T-NbTe2_2. One can recognize three straight lines running along the ΓM\Gamma M direction. Same lines also appear along the other two equivalent ΓM\Gamma M directions rotated by ±60∘\pm60^\circ from each other. These lines are mainly associated with the parallel segments of the calculated triangular pockets. One can see that two types of nesting vectors q\mathbf{q} for 19×19R+23.4∘\sqrt{19}\times\sqrt{19}R+23.4^\circ and 19×19R−23.4∘\sqrt{19}\times\sqrt{19}R-23.4^\circ superstructures suggested from the STM data are located near the intersection of straight lines at which Im⁡χ(q)\operatorname{Im}\chi(\mathbf{q}) value shows an overall enhancement. This supports the Fermi-surface-instability-driven CDW in monolayer NbTe2_2. It is noted that the electronic susceptibility does not show sharp peak or strong divergence at the corresponding nesting vector. Similar behavior was also recognized in other 2D TMDs, such as 1T1T-VTe2_2 and 2H2H-NbSe2_2[34, 35, 36].

Two types of crystal domains in monolayer NbTe2_2 on bilayer graphene

We found two types of domain structures associated with the 19×19R23.4∘\sqrt{19}\times\sqrt{19}R23.4^\circ star-of-David (soD) lattice by STM measurements. Figure 7(a) shows the STM image of monolayer 1T1T-NbTe2_2 island on bilayer graphene. One can identify existence of two types of superstructures. One has a 19×19R−23.4∘\sqrt{19}\times\sqrt{19}R-23.4^\circ periodicity (red rhombus) @page 8/9 starts_mid=0 ends_mid=1

Intensity plot of the calculated imaginary part of electronic susceptibility

Figure 6 (PDF p. 8). Intensity plot of the calculated imaginary part of electronic susceptibility Im⁡χ(q)\operatorname{Im}\chi(\mathbf{q}) for monolayer 1T1T-NbTe2_2. Arrows correspond to q=1/19 G\mathbf{q}=1/\sqrt{19}\,\mathbf{G} nesting vectors.

flat feature is also observed at hν=75h\nu=75 eV around the KK point (indicated by white arrow), whereas its intensity is largely suppressed. As shown in Figure 5(c), such a weak and flat feature also exists along another ΓK\Gamma K cut rotated by 60∘60^\circ with respect to the ΓK\Gamma K cut in Figure 5(b). We find that the flat feature is also recognized along the MKMK cut with different hνh\nu’s and light polarizations, as shown in Figure 5(d)–Figure 5(f). These results suggest that the existence of observed flat feature (i.e. the absence of a triangular pocket predicted by the DFT calculation) is robust against the variation in photon energy and light polarization. This suggests that the flat feature is not an experimental artifact associated with the matrix-element effect of photoelectron intensity, but is an intrinsic feature of monolayer 1T1T-NbTe2_2.

Origin of a flat band around the KK point for monolayer NbTe2_2

We discuss the origin of a flat feature at EB∼0.5E_B\sim0.5 eV around the KK point observed in Figure 2 and Figure 5. We have considered some possible mechanisms unrelated to the CDW, such as (i) the band gap, (ii) the Anderson gap, and (iii) the Mott-Hubbard gap. The possibility (i) is ruled out because the DFT calculation of monolayer NbTe2_2 fails to reproduce the flat band and a finite energy gap. The possibility (ii) is also unlikely because our NbTe2_2 film is close to stoichiometry and the ARPES spectrum around the KK point (Figure 2 of the main text) does not show a power-low behavior against EBE_B expected from the disorder-induced Coulomb gap. The possibility (iii) is also ruled out because the gap size Δ\Delta of ∼0.5\sim0.5 eV is too large compared to the Mott-Hubbard gap of isostructural 1T1T-NbSe2_2 (Δ∼0.1\Delta\sim0.1 eV) despite the expectation of a similar on-site Coulomb energy. Taking into account these points, we think that the flat band is associated with the CDW gap formation.

STM image and real-space and reciprocal-space configurations of NbTe2 superstructures

Figure 7 (PDF p. 8). (a), STM image at T=4.8T=4.8 K in the surface area of 25×2525\times25 nm2^2 on NbTe2_2 island (Vs=−1.6V_s=-1.6 V, and It=0.1I_t=0.1 nA). (b) Real-space configuration of Nb atoms (open circles) and superlattice unit cell (red rhombus) for the 19×19R−23.4∘\sqrt{19}\times\sqrt{19}R-23.4^\circ superstructure and (c) corresponding reconstructed Brillouin zone (red) drawn on the original 1×11\times1 Brillouin zone (black). (d), (e) Same as (b) and (c), but for the R+23.4∘R+23.4^\circ superstructure.

Calculated electronic susceptibility for monolayer NbTe2_2

To clarify the origin of CDW in monolayer NbTe2_2, we calculated the electronic susceptibility χ(q)\chi(\mathbf{q}) based on the calculated Fermi surface shown in Figure 2(b). Figure 6 shows the imaginary part of calculated electronic susceptibility [Im⁡χ(q)][\operatorname{Im}\chi(\mathbf{q})] for monolayer 1T1T-NbTe2_2. One can recognize three straight lines running along the ΓM\Gamma M direction. Same lines also appear along the other two equivalent ΓM\Gamma M directions rotated by ±60∘\pm60^\circ from each other. These lines are mainly associated with the parallel segments of the calculated triangular pockets. One can see that two types of nesting vectors q\mathbf{q} for 19×19R+23.4∘\sqrt{19}\times\sqrt{19}R+23.4^\circ and 19×19R−23.4∘\sqrt{19}\times\sqrt{19}R-23.4^\circ superstructures suggested from the STM data are located near the intersection of straight lines at which Im⁡χ(q)\operatorname{Im}\chi(\mathbf{q}) value shows an overall enhancement. This supports the Fermi-surface-instability-driven CDW in monolayer NbTe2_2. It is noted that the electronic susceptibility does not show sharp peak or strong divergence at the corresponding nesting vector. Similar behavior was also recognized in other 2D TMDs, such as 1T1T-VTe2_2 and 2H2H-NbSe2_2[34, 35, 36].

Two types of crystal domains in monolayer NbTe2_2 on bilayer graphene

We found two types of domain structures associated with the 19×19R23.4∘\sqrt{19}\times\sqrt{19}R23.4^\circ star-of-David (soD) lattice by STM measurements. Figure 7(a) shows the STM image of monolayer 1T1T-NbTe2_2 island on bilayer graphene. One can identify existence of two types of superstructures. One has a 19×19R−23.4∘\sqrt{19}\times\sqrt{19}R-23.4^\circ periodicity (red rhombus) and another is rotated by 13.2∘13.2^\circ with respect to the former one (blue rhombus). The latter domain is assigned to the superstructure with 19×19R+23.4∘\sqrt{19} \times\sqrt{19}R+23.4^\circ by taking into account the 6-fold symmetry of NbTe2_2 (i.e. 13.2∘=60∘−23.4∘−23.4∘13.2^\circ= 60^\circ- 23.4^\circ- 23.4^\circ). The existence of two types of domains is naturally expected from the real-space configuration of Nb atoms and the supercell [Figs. 7(b) and 7(d)] as well as the corresponding Brillouin zones [Figs. 7(c) and 7(e)]. At the boundary between the R+23.4∘R+23.4^\circ and R−23.4∘R-23.4^\circ domains, a domain wall is recognized [purple dashed curve of Fig. 7(a)].

STM images across EFE_{\mathrm{F}} for monolayer NbTe2_2

We performed STM measurements at negative and positive bias voltages across the CDW gap for monolayer NbTe2_2. As clearly seen in Figure 8, the high electron-density region characterized by bright-intensity spots at negative bias voltage [Figure 8(a)] turns into the low electron-density region at positive bias voltage [Figure 8(b)], similarly to the case of KV3_3Sb5_5 [ref.8]. This supports the real-space charge reversal and the CDW origin of the observed V-shaped gap for monolayer 1T1T-NbTe2_2.

STM images at negative and positive bias voltage

Figure 8. (a), (b) STM images at T=4.8T = 4.8 K in the surface area of 6×66 \times6 nm2^2 measured at (a) negative (Vs=−0.5V_{\mathrm{s}} = -0.5 V, It=4I_{\mathrm{t}} = 4 nA) and (b) positive (Vs=+0.5V_{\mathrm{s}} = +0.5 V, It=4I_{\mathrm{t}} = 4 nA) bias voltage. White rhombus corresponds to the 19×19R23.4∘\sqrt{19} \times\sqrt{19}R23.4^\circ superstructure unit cell.

Local density of states for monolayer NbTe2_2

Figure 9(a) shows the STM image at T=4.8T = 4.8 K in the surface area of 6×66 \times6 nm2^2 for monolayer NbTe2_2. Filled circles represent the measurement position of dI/dVdI/dV curves inside the soD cluster (red and black circles) and outside it (blue and green circles).

cles represent the measurement position of dI/dVdI/dV curves inside the soD cluster (red and black circles) and outside it (blue and green circles). Figure 9(b) shows the dI/dVdI/dV curves for monolayer NbTe2_2 measured at T=4.8T = 4.8 K. The color of curves shows the correspondence to the measurement position in Figure 9(a). One can recognize several

STM image and dI/dV curves for monolayer NbTe2

Figure 9. (a), STM image at T=4.8T = 4.8 K in the surface area of 6×66 \times6 nm2^2 for monolayer NbTe2_2. (b) dI/dVdI/dV curves measured at four representative positions inside the soD cluster [red and black filled circles in (a)] and outside it (blue and green ones). Correspondence between the measurement position and dI/dVdI/dV curve is shown by coloring. Inset shows an expansion in the negative bias-voltage region. dI/dVdI/dV spectra were recorded with an open feedback loop using a lock-in technique with a modulation voltage of 20 mV at 987 Hz. The setpoints of voltage and current (VstabV_{\mathrm{stab}} and IstabI_{\mathrm{stab}}, respectively) to stabilize the tip before opening feedback loop are indicated in (b).

peak features in both positive and negative bias voltages irrespective of the location of surface, and there is no abrupt change in the dI/dVdI/dV curve across the boundary between the inside- and outside-soD regions. Expanded image in the negative-bias voltage region in the inset to Figure 9(b) signifies broad humps at around -0.2 eV and -0.55 eV, which are associated with the Te 5p5p bands at the Γ\Gamma point and the flat band at ∼0.5\sim0.5 eV around the K point observed by ARPES, respectively [Figure 2(c) and Figure 2(d)]. One can also see the V-shaped density of states (DOS) at EFE_{\mathrm{F}} supportive of the absence of an insulating energy gap and appearance of a soft gap, in stark contrast to a large Mott gap in monolayer 1T1T-NbSe2_2 and 1T1T-TaSe2_2 [ref. 6,7,8]. Interestingly, the V-shaped DOS resembles that at the boundary between the metallic and Mott-insulating domains in bulk 1T1T-TaS2_2 [ref.14,15]. The soft-gap nature with residual DOS at EFE_{\mathrm{F}} seen by STM is also consistent with the ARPES data showing a finite Fermi-edge cut-off associated with the Te 5p5p band around the Γ\Gamma point seen in Figure 2.

References

  1. [1]K. S. Novoselov, A. K. Geim, S. V. Morozov, D. Jiang, M. I. Katsnelson, I. V. Grigorieva, S. V. Dubonos, and A. A. Firsov. Nature 87, 197 (2005).
  2. [2]X. Xi, Z. Wang, W. Zhao, J.-H. Park, K. T. Law, H. Berger, L. Forró, J. Shan, and K. F. Mak, Nat. Phys. 12, 139 (2016).
  3. [3]S. Tang, C. Zhang, D. Wong, Z. Pedramrazi, H.-Z. Tsai, C. Jia, B. Moritz, M. Claassen, H. Ryu, S. Kahn, J. Jiang, H. Yan, M. Hashimoto, D. Lu, R. G. Moore, C.-C. Hwang, C. Hwang, Z. Hussain, Y. Chen, M. M. Ugeda, Z. Liu, X. Xie, T. P. Devereaux, M. F. Crommie, S.-K. Mo, and Z.-X. Shen, Nat. Phys. 13, 683 (2017).
  4. [4]P. Chen, W. W. Pai, Y.-H. Chan, W.-L. Sun, C.-Z. Xu, D.-S. Lin, M. Y. Chou, A.-V. Fedorov, and T.-C. Chiang, Nat. Commun. 9, 2003 (2018).
  5. [5]Y. Cao, V. Fatemi, S. Fang, K. Watanabe, T. Taniguchi, E. Kaxiras, and P. Jarillo-Herrero, Nature 556, 43 (2018).
  6. [6]Y. Nakata, K. Sugawara, R. Shimizu, Y. Okada, P. Han, T. Hitosugi, K. Ueno, T. Sato, and T. Takahashi, NPG Asia Mater. 8, e321 (2016).
  7. [7]Y. Nakata, T. Yoshizawa, K. Sugawara, Y. Umemoto, T. Takahashi, and T. Sato, ACS Appl. Nano Mater. 1, 1456 (2018).
  8. [8]Y. Chen, W. Ruan, M. Wu, S. Tang, H. Ryu, H.-Z. Tsai, R. Lee, S. Kahn, F. Liou, C. Jia, O. R. Albertini, H. Xiong, T. Jia, Z. Liu, J. A. Sobota, A. Y. Liu, J. E. Moore, Z.-X. Shen, S. G. Louie, S.-K. Mo, and M. F. Crommie, Nat. Phys. 16, 218 (2020).
  9. [9]J. A. Wilson, F. J. DiSalvo, and S. Mahajan, Adv. Phys. 24, 117 (1975).
  10. [10]P. Fazekas and E. Tosatti, Philos. Mag. B 39, 229 (1979).
  11. [11]R. E. Thomson, B. Burk, A. Zettl and J. Clarke, Phys. Rev. B 49, 16899 (1994).
  12. [12]P. Fazekas and E. Tosatti, Physica B & C 99, 183 (1980).
  13. [13]R. Ang, Y. Tanaka, E. Leki, K. Nakayama, T. Sato, L. J. Li, W. J. Lu, Y. P. Sun, and T. Takahashi, Phys. Rev. Lett. 109, 176403 (2012).
  14. [14]L. Ma, C. Ye, Y. Yu, X. F. Lu, X. Niu, S. Kim, D. Feng, D. Tománek, Y.-W. Son, X. H. Chen, and Y. Zhang, Nat. Commun. 7, 10956 (2016).
  15. [15]D. Cho, G. Gye, J. Lee, S.-H. Lee, L. Wang, S.-W. Cheong, and H. W. Yeom, Nat. Commun. 8, 392 (2017).
  16. [16]S.-H. Lee, J. S. Goh, and D. Cho, Phys. Rev. Lett. 122, 106404 (2019).
  17. [17]Y. D. Wang, W. L. Yao, Z. M. Xin, T. T. Han, Z. G. Wang, L. Chen, C. Cai, Y. Li, and Y. Zhang, Nat. Commun. 11, 4215 (2020).
  18. [18]C. J. Butler, M. Yoshida, T. Hanaguri, and Y. Iwasa, Nat. Commun. 11, 2477 (2020).
  19. [19]L. Liu, H. Yang, Y. Huang, X. Song, Q. Zhang, Z. Huang, Y. Hou, Y. Chen, Z. Xu, T. Zhang, X. Wu, J. Sun, Y. Huang, F. Zheng, X. Li, Y. Yao, H.-J. Gao, and Y. Wang, Nat. Commun. 12, 1978 (2021).
  20. [20]K. Selte and A. Kjekshus, Acta. Chem. Scand. 19, 258 (1965).
  21. [21]J. A. Wilson and A. D. Yoffe, Adv. Phys. 18, 193 (1969).
  22. [22]B. B. Brown, Acta. Cryst. 20, 264 (1966).
  23. [23]C. Battaglia, H. Cercellier, F. Clerc, L. Despont, M. G. Garnier, C. Koitzsch, P. Aebi, H. Berger, L. Forró, and C. Ambrosch-Draxl, Phys. Rev. B 72, 195114 (2005).
  24. [25]K. Iwaya, T. Ohsawa, R. Shimizu, Y. Okada, and T. Hitosugi, Sci. Technol. Adv. Mater. 19, 282 (2018).
  25. [26]P. Giannozzi, S. Baroni, N. Bonini, M. Calandra, R. Car, C. Cavazzoni, D. Ceresoli, G. L. Chiarotti, M. Cococcioni, I. Dabo, A. D. Corso, S. de Gironcoli, S. Fabris, G. Fratesi, R. Gebauer, U. Gerstmann, C. Gougoussis, A. Kokalj, M. Lazzeri, L. Martin-Samos, N. Marzari, F. Mauri, R. Mazzarello, S. Paolini, A. Pasquarello, L. Paulatto, C. Sbraccia, S. Scandolo, G. Scauzlero, A. P. Seitsonen, A. Smogunov, P. Umari, and R. M. Wentzcovitch, J. Phys. Condens. Matter. 21, 395502 (2009).
  26. [27]J. van Landuyt, G. van Tendeloo, and S. Amerinckx Phys. Stat. Sol 26, 585 (1974).
  27. [28]J. A. Wilson, Phys. Rev. B 17, 3880 (1978).
  28. [29]Y.-X. Jiang, J.-X. Yin, M. M. Denner, and N. Shumiya, B. R. Ortiz, G. Xu, Z. Guguchia, J. He, M. S. Hossain, X. Liu, J. Ruff, L. Kautzsch, S. S. Zhang, G. Chang, I. Belopolski, Q. Zhang, T. A. Cochran, D. Multer, M. Litskevich, Z.-J. Cheng, X. P. Yang, Z. Wang, R. Thomale, T. Neupert, S. D. Wilson, and M. Z. Hasan, Nat. Mater. 20, 1353 (2021).
  29. [30]H. Ryu, Yi Chen, H. Kim,H.-Z. Tsai, S.Tang, J. Jiang, F. Liou, S. Kahn, C. Jia, A. A. Omrani, J. H. Shim, Z. Hussain, Z.-X. Shen, K. Kim, B. I. Min, C. Hwang, M. F. Crommie, and S.-K. Mo, Nano Lett. 18, 689 (2018).
  30. [31]P. M. Coelho, K. N. Cong, M. Bonilla, S. Kolekar, M.-H. Phan, J. Avila, M. C. Asensio, I. I. Oleynik, and M. Batzill, J. Phys. Chem. C 123, 14089 (2019).
  31. [32]A. Kikuchi and S. Tsuneyuki, Surf. Sci. 409, 458 (1998).
  32. [33]Y. Umemoto, K. Sugawara, Y. Nakata, T. Takahashi, and T. Sato, Nano Res. 12, 165 (2019).
  33. [34]T. Kawakami, K. Sugawara, T. Kato, T. Taguchi, S. Souma, T. Takahashi, and T. Sato, Phys. Rev. B 104, 045136 (2021).
  34. [35]A. Soumyanarayanan, M. M. Yee, J. van Wezel, D. J. Rahn, K. Rossnagel, E. W. Hudson, M. R. Norman, and J. E. Hoffman, Proc. Natl. Acad. Sci. USA, 110, 1623 (2013).
  35. [36]M. D. Johannes and I. I. Mazin, Phys. Rev. B 77, 165135 (2008).

Paper details

vlnc:W_238f296v
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 8

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents