Introduction

Topological insulators (TIs) have been intensely studied for the past few years since the discovery of graphene due to their interesting electrical, optical, and mechanical properties. TIs are uniquely insulating in the bulk, but support conducting edge and surface states in 2D and 3D, respectively.1,2,3,4 In particular, two-dimensional topological insulators (2D TIs), also known as quantum spin Hall (QSH) insulators, exhibit unique symmetry-protected helical metallic edge states with an insulating interior, which makes these materials especially well-suited for optoelectronic, spintronics, quantum computing, and other applications due to the robustness of their edge states against backscattering.5 Numerous theoretical studies have been focused on identifying new materials that can support 2D TI1,6,7,8,9 or topological crystalline insulator (TCI)10,11,12,13 phases. Moreover, effects of functionalization7,14,15,16 and substrates16,17,18 have been investigated, although experimental realizations are quite limited so far.19,20,21

In this connection, one of the most promising classes of 2D materials with semiconducting properties are the transition metal dichalcogenides (TMDs) with chemical formula MX2, where M is a transition metal atom and X is a chalcogen atom, which exhibit different structural phases such as 1T (octahedral) and 2H/3R (trigonal prismatic). Despite graphene’s popularity, the lack of an electronic bandgap in graphene has driven the interest in 2D TMD films as a promising alternative.22,23

Recent theoretical24,25,26,27,28 and experimental29,30,31,32,33,34,35 studies show that doping is a viable method for enhancing and tuning electronic and optical properties of TMDs for potential applications. One specific type of doping is substitutional doping, and there have been reports of successful experimental substitution of transition metal29 or chalcogen atoms.35 Most experimental research has focused on doping MoS2. For example, covalent nitrogen doping of MoS2 was obtained by remote N2 plasma exposure and it reportedly can induce compressive strain on TMDs31 (as can other experimental procedures32,33), and computational results show that structural phase transitions can occur by alloying.28

Very recently, ultrathin 2D Janus materials have started to attract attention. In fact, a number of honeycomb-like 2D Janus materials that could harbor topological phases have been predicted.16,17,36,37,38,39 In addition, in TMD materials, exemplary results have been obtained experimentally where S–Mo–Se crystal structure (also known as Janus SMoSe) was successfully synthesized by well-controlled sulfurization of monolayer MoSe235 confirming the possibility of chalcogen substitution. Although a number of electronic structure calculations have been performed, substitution studies are sparse and much less explored,40,41 indicating the need for the development of other doping strategies. Also, since substitutional doping is relatively at an early stage of research,42 exploration of other TMDs that could be doped is highly desirable. To our knowledge, one of the criteria in choosing a dopant is based on the capability to change the total number of electrons (hole doping or electron doping) in a compound.43 This suggests that, aside from substituting one chalcogen with another chalcogen, it will be interesting also to consider substitution with group VA (pnictogen)44 or group VIIA (halogen).45

Among the TMDs, groups VB and VIIB TMDs have odd number of electrons as well as weak magnetism and spin-orbit coupling, and their pristine films are metallic.25,46,47,48,49 Thus, substitutional doping of VA or VIIA or one-sided H adsorption, will result in making the system acquire an even number of electrons. Also, pristine ultrathin films of a number of these materials (TaS2,50 TaSe2,51 NbSe2,52,53 NbS2,54 VS2,55 VSe2,56 VTe2,57 ReS2,58,59 and ReSe260) have been experimentally synthesized, suggesting that the synthesis of the related Janus 2D materials based on VB and VIIB considered in this study will likely be viable.

So motivated, we have carried out a first-principles study of possible Janus 2D topological insulator phases via halogen (F, Cl, Br, or I) and pnictogen (N, P, As, Sb, or Bi) substitution, including one-sided hydrogen adsorption, of MX2 (M = V, Nb, Ta, Tc, or Re; X = S, Se, or Te) films in both 1T (octahedral) and 2H (trigonal prismatic) structures. A total of 294 compounds were examined. Structural phase (from 2H to 1T) and magnetic property (between non-magnetic and ferromagnetic states) transitions were also observed upon substitutional doping. 2D nontrivial phases were found for pnictogen substitution of MX2 (M = V, Nb, or Ta), while the nontrivial phases were found for MX2 (M = Tc or Re) with halogen substitution, and one-sided H adsorption.

Results

Figure 1 shows the perspective views of the investigated monolayer 1T and 2H MX2, MXY, and MX2H (M = transition metal, X = chalcogen, Y = halogen or pnictogen, H = Hydrogen) and the related first Brillouin Zone (BZ). The calculated stable structures of MX2, MX2H, and MXY are shown in Table 1. Table 2 gives the calculated topological invariant Z2 for the stable structural phase of MX2, MX2H, and MXY. Tables 3 and 4 show the calculated topological invariant Z2 for the stable structural phase of group VB upon pnictogen substitutional doping and group VIIB upon one hydrogen adsorption and halogen substitutional doping, respectively. In order to unfold the electronic properties of the stable materials, band structures of the TaS2-based films as representative of group VB are shown in Fig. 2, while results for the ReSe2-based films as representative of group VIIB are shown in Fig. 3. In order to accurately estimate the band gaps and better understand the mechanism underlying various splittings, hybrid-functional based band structures of the selected stable structural phases, including detailed orbital analysis and crystal field splittings are shown in Fig. 4. Illustrations of the topologically protected edge states, and possible synthesis of Janus 2D materials via lithography are presented in Fig. 5.

Fig. 1
figure 1

Perspective views of monolayer 1T and 2H MX2 and MXY films (M = transition metal, X = chalcogen, Y = halogen or pnictogen). a, d Pristine MX2. b, e Structures with one halogen, or one pnictogen, substituting one chalcogen per unit cell. c, f Structures with one adsorbed hydrogen atom on a chalcogen atom. g 2D Brillouin-zones with the specific high symmetry points labeled

Full size image
Table 1 Calculated stable structures of MX2, MX2H, and MXY
Full size table
Table 2 Calculated topological invariant Z2 for the stable structural phases of MX2, MX2H, and MXY
Full size table
Table 3 Calculated topological invariant Z2 for the stable structural phase of group VB upon pnictogen substitutional doping
Full size table
Table 4 Calculated topological invariant Z2 for the stable structural phases of group VIIB upon one hydrogen adsorption and halogen substitutional doping
Full size table
Fig. 2
figure 2

Band structures of the stable structural phases without SOC (top) and with SOC (bottom). a, e Monolayer pristine TaS2 film in the 2H structure. b, f One adsorbed hydrogen on TaS2 in the 2H structure. c, g One Cl atom substituting a S atom in the 2H structure. d, h One Bi atom substituting a S atom in the 1T structure

Full size image
Fig. 3
figure 3

Band structures of the stable structural phases without SOC (top) and with SOC (bottom). a, e Monolayer pristine ReSe2 in 2H structure. b, f One adsorbed hydrogen on ReSe2 film in the 1T structure. c, g One Br atom substituting a Se atom in the 1T structure. d, h One Bi atom substituting a Se atom in the 2H structure

Full size image
Fig. 4
figure 4

Hybrid-functional based electronic band structures without SOC (top) and with SOC (bottom) of the selected stable structural phases associated with the dyz (blue dot) and dxz (red dot) orbitals. a, d 1T VSeSb, b, e 1T NbSeBi, c, f 1T TaSBi, h, j 1T TcSeCl, i, k 1T ReSeBr. g Schematic energly-level diagram highlighting how degeneracies of crystal-field split levels are lifted via SOC

Full size image
Fig. 5
figure 5

a, b Illustrations of the topologically protected edge states, c Illustration of possible synthesis of Janus 2D materials via lithography methods

Full size image

Discussion

The structures of the investigated monolayer MX2 (M = V, Nb, Ta, Tc, or Re; X = S, Se, or Te) are assumed to be in 1T or 2H. The perspective views of the atomic structure for both pristine 1T and 2H are shown in Fig. 1a, d. Next, the structures of Janus TMDs (MXY) in which one halogen or one pnictogen substitutes one chalcogen per unit cell are shown in Fig. 1b, e. Lastly, we consider the case where one H is directly adsorbed on top of the chalcogen atom as shown in Fig. 1c, f. The surface Brillouin-zones is presented in Fig. 1g.

Optimization of the atomic structures and lattice constants was done using the methods discussed in the Methods section below. Relevant information regarding the optimized structures is summarized in Tables S1–S10 in the Supplementary Materials, and includes details of lattice constants, stable phases, as well as the energetics of the two structures using different magnetic configurations (non-magnetic and ferromagnetic).

Regarding structural stability, Table 1 shows that our result that the structurally stable phase for all the investigated pristine MX2 films is 2H, which is in agreement with other theoretical and experimental findings,25,46,47,48 . Subsequently, we performed the substitution of chalcogen atoms with pnictogen atoms (group VA) or halogen atoms (group VIIA) on one side, as well as one-hydrogen-adsorption on the chalcogen. Hydrogen adsorption is assumed to have a similar effect to halogen substitution in terms of changing the total number of valence electrons in pristine MX2. Consequently, after the aforementioned substitution or hydrogen adsorption, the total number of valence electrons would become even due to the initial odd number in pristine MX2. Moreover, numerous studies have shown that the structural phase transition may occur after doping.25,28,32,33 Our findings for group VB indeed indicate a structural phase transition from 2H to 1T for all the MXYpnic films, whereas the structures after halogen substitution or hydrogen adsorption are found to stay 2H. On the other hand, our findings for group VIIB indicate a structural phase transition from 2H to 1T for all the MXYhal or MX2H films, while for pnictogen substitution, the structures stay 2H. All these results are summarized in Table 1.

To corroborate the structural stability of the selected stable MXY films, the corresponding phonon spectra were calculated. A total of nine phonon bands were found in each selected case, with three acoustic and six optical branches, as shown in Figs. S1–S4. The group VIIB MXY films exhibit negative frequency (Fig. S4a and b) and are thus, structurally unstable. In contrast, these negative frequency modes did not occur in the group VB MXY films as shown in Figs. S1–S3, indicating their structural stability and viability of experimental synthesis. Together with the phonon spectra, the formation energies were also calculated to further check their stability. The formation energies and energy differences shown in Table S11 are consistent with the results of phonon spectra. Group VB MXY films exhibit lower energies compared to the corresponding pristine phases with a positive energy difference, indicating structural stability. On the other hand, group VIIB MXY have higher energy compared to their pristine counterparts with a negative energy difference, indicating structural instability. We have also carried out first-principles molecular dynamics (MD) simulations using the NVT ensemble for a 3 × 3 superlattice at 500 K as shown in Figs. S5–S9. Group VB MXY structures are found to be dynamically stable even at elevated temperature, while this is not the case for the group VII MXY structures. The MD results are thus also consistent with the results of phonon spectra.

To further investigate the band topologies of the computed stable compounds, we follow the method of Refs. 61,62 for calculating the Z2 topological invariant in non-magnetic cases or Chern number in ferromagnetic cases. We have computed the Z2 or the Chern number of all the systems we investigated for completeness, including the unstable phase, see Tables S1–S5 under Supplementary Materials.

The topologies of the stable MX2, MX2H, and MXY films are summarized in Tables 2–4. As shown in Table 2, the pristine MX2 films exhibit a metallic state due to their half-filled band. Further shown in Table 2 are the Z2 invariants of group VB MXYhal and MX2H as well as group VIIB MXYpnic films, which are found to be zero corresponding to the trivial phase. These trivial metallic phases were also found in several group VB MXYpnic films. But, importantly, we found a total of nineteen compounds belonging to the VXY, NbXY, and TaXY film clasees, with 7, 7, and 5 nontrivial phases, Z2 = 1, respectively. Notably, the nontrivial phases we have found are mostly generated through the substitution by relatively heavy atoms such as Sb and Bi. The Table 3 summarizes our results. Moving to group VIIB, several metallic states were only found in halogen substituted TcXY films. A total of 18 compounds in TcXY, TcX2H, ReXY, and ReX2H films exhibited non-trivial phases, see Table 4. We can see that upon pnictogen substitution of group VB, several nontrivial phases were induced, whereas the occurrence of nontrivial phases in group VIIB arose upon halogen substitution or one-sided H adsorption. These results support the notion that the effects of halogen substitution and one-sided H adsorption are similar in that both of theses dopings induce the same structural phase transition between the trivial or nontrivial phases.

To gain further insight, we explore the band structures of TaS2-based films as representative of group VB. As shown in Fig. 2a and e, our band structures of the pristine TaS2 films are in agreement with previous studies.47 TaS2 has an odd number of electrons, so that the Fermi level crosses the highest occupied band. By substituting one chalcogen with one halogen, one extra electron is introduced, which could also be done by adsorbing adding one hydrogen on TaS2. The effects of halogen substitution and one-sided hydrogenation on the pristine TaS2 are indeed found to be similar, both inducing a downward shift in the band structures resulting in a trivial phase as shown in Fig. 2b, c, f, and g. This is true for all group VB MXYhal and MX2H films.

Referring to Table 1, in contrast to halogen substitutions and hydrogen adsorption, we find that pnictogen substitution induces a structural transition from 2H to 1T in group VB, and for this reason, the 1T band structures are shown in Fig. 2d, and h. Contrary to halogen substitution and one-sided hydrogenation, pnictogen substitution results in the loss of one electron. In the case of TaSBi film (without SOC), we see from Fig. 2d that HOMO and LUMO levels touch at Γ. When the SOC is included, a 108 meV gap opens at the Γ point. The calculated topological invariant of 1T TaSBi film is Z2 = 1 (Table 3), so that this film harbors a nontrivial phase.

We turn now to discuss in-depth stable ReSe2 films as representative of group VIIB. Referring to Table 1, we see that halogen substitutions and hydrogen adsorptions can induce a structural phase transition from 2H to 1T in group VIIB, and here again for this reason, 1T band structures are presented in Fig. 3b, c, f, and g. In Fig. 3a and e, our band structures for the pristine ReSe2 film are in agreement with previous studies.25 Similar to the case of TaS2, the ReSe2 film has an odd number of electrons, and thus the Fermi level crosses the highest occupied band as expected. A splitting driven by the SOC is evident in Fig. 3f and g for one hydrogen adsorption and halogen substitution, respectively, and respective band gaps at Γ of 145 and 274 meV. The calculated topological invariants of 1T ReSe2H and 1T ReSeBr are non-zero Chern number and Z2 = 1, respectively, indicating a nontrivial phase. The effect of halogen substitution and one-sided hydrogenation on the pristine ReSe2 film are found to be similar in that they induce a downward shift in the band structure and result in a nontrivial phase, see Fig. 3b, c, f, and g. Similar to group VB MX2, the effects of halogen substitution and one-sided hydrogenation can be attributed to the addition of one electron in group VIIB MX2 films.

As already noted, in contrast to halogen substitution and one-sided hydrogenation, pnictogen substitution results in a loss of one electron. Here, as shown in Fig. 3d and h, in the case of 2H ReSeBr one obtains a trivial insulator with Z2 = 0.

Clearly, the effects of pnictogen substitution, and halogen substitution or one-sided H adsorption are opposite for group VB and group VIIB in terms of the structural phase transitions and band topologies. Pnictogen substitution induces nontrivial phases for group VB but not for group VIIB. On the other hand, halogen substitution and one-sided H adsorption induces nontrivial phases for group VIIB but not for group VB. Notably, these nontrivial phases were only obtained when the structures undergo a phase transition from 2H to 1T for both loss or gain of an electron.

Results for all other films considered, including their band structures, are given in Tables S1–S10 and Figs. S10–S24 in the Supplementary Materials.

In order to better understand the mechanism underlying the formation of the nontrivial phases, we carried out a detailed orbital analysis for one representative nontrivial film for each metal, as shown in Fig. S25. Since the GGA is well known to underestimate band gaps, we performed hybrid functional based calculations for nontrivial phases the selected films. We found that the key orbitas are the M-dyz and M-dxz orbitals for all selected cases as seen in the partial band projections in Fig. 4a–f, h, k. Without SOC, all selected systems are gapless in which the valence and conduction bands are degenerate at the Γ point, and the systems are all metallic. With SOC, the doubly degenerate M-dyz, M-dxz orbitals experience a gap opening at the Γ point in all these selected films. In short, the M-dyz and M-dxy orbitals remain degenerate under the crystal-field, but this degeneracy is lifted via the effects of the SOC to open a gap at the Fermi energy at Γ to drive the system into the topological phase, see Fig. 4g.

In terms of band gaps, surprisingly, 1T VSeSb becomes a topological insulator with a positive system-wide band gap (i.e. the minimum band gap throughout the Brillouin zone in the system around the Fermi energy) of 373 meV and a band gap of 1135 meV at the Γ point. Although many other films we considered remained semi-metallic, we could still see a significant increase in the band gap at Γ such as in the case of 1T TaSBi (with SOC) with a system-wide band gap of 360 meV, see Fig. 4f. Remarkably, the nontrivial phase is robust for selected groups of MXY films, demonstrating the likelihood of sustaining this nontrivial phase upon growing the film on a suitable substrate. It is noteworthy that the calculated band gaps of these materials are large enough for possible room temperature applications.

Figure 5a and b illustrates the topologically protected edge states in the real and reciprocal spaces of the predicted Janus 2D films in the quantum spin Hall (QSH) phases with the spin-up (blue) and spin-down (red) electrons conducting along one edge of the ribbon. Figure 5c shows a sheet of a TMD film with a region of Janus 2D to demonstrate the possibility of topological insulator phase upon substitutional doping. In our calculations, the topological insulator phase and several nontrivial phases are obtained in different compounds by directly replacing the top chalcogen with a certain dopant which has been discussed in-depth in this study.

Finally, impressive experimental results50,51,52,53,54,55,56,57,59,60 have been reported from ultrathin pristine TMD, VS2, and NbSe2 films successfully synthesized via chemical vapor deposition (CVD).53,55 Several techniques have also been tested such as powder vapor deposition50 for TaS2, e-beam evaporation54 for NbS2, and modified Bridgman method60 for ReS2 and ReSe2 films. Notably, the Janus 2D material SMoSe35 has been successfully synthesized via controlled sulfurization. And very recently, a single-layer Bismuth (SLB) film has been successfully deposited on NbSe2.52 These results suggest that viable Janus 2D NbSeBi films could be experimentally obtained via lithography or etching method. More generally, it will be interesting to explore the possibility of synthesizing the Janus 2D films from VB and VIIB elements by using the aforementioned methods and other mechanical exfoliation or bottom-up methods.

Methods

The first-principles calculations were carried out using the Vienna Ab initio Simulation Package (VASP)63,64 with the projected augmented wave (PAW)65 potentials. Spin polarization was considered in all cases for the comparison between the non-magnetic (NM) and ferromagnetic (FM) configurations. The exchange-correlation functional was treated within the Perdew–Burke–Ernzerhof (PBE) generalized gradient approximations (GGA).66,67,68,69,70 The cut-off energy used throughout the calculations was set to 400 eV. Atomic positions were optimized for each lattice constant value considered until the residual forces were no greater than 10−3 eV/Å. The criteria for energy convergence for self-consistency was set at 10−6 eV. The vacuum region along the z-direction was set to ~15 Å to prevent interactions between the repeated monolayer slabs under periodic boundary condition. A Γ-centered Monkhorst–Pack71 grid of 12 × 12 × 1 in the first Brillouin zone was used for calculating atomic structures and lattice relaxations (pristine and substitutional TMDs). However, a denser grid of 36 × 36 × 1 was used for band structure calculations in order to accurately capture band gaps and band topologies. The band topologies were further checked following the method of Refs. 61,62 for calculating the Chern numbers and Z2 topological invariants. The phonon spectra were calculated using the Phonopy code.72 A 2 × 2 × 1 supercell with a grid of 36 × 36 × 1 was constructed in order to obtain accurate force constants for calculating the phonon spectra. The molecular dynamics (MD) simulations were carried out in a 3 × 3 superlattice to determine the elevated-temperature stability for 5000 fs with a time step of 2.0 fs at T = 500 K.