Halogen-Doped Chevrel Phase Janus Monolayers for Photocatalytic Water Splitting

Chevrel non-van der Waals crystals are promising candidates for the fabrication of novel 2D materials due to their versatile crystal structure formed by covalently bonded (Mo6X8) clusters (X–chalcogen atom). Here, we present a comprehensive theoretical study of the stability and properties of Mo-based Janus 2D structures with Chevrel structures consisting of chalcogen and halogen atoms via density functional theory calculations. Based on the analysis performed, we determined that the S2Mo3I2 monolayer is the most promising structure for overall photocatalytic water-splitting application due to its appropriate band alignment and its ability to absorb visible light. The modulated Raman spectra for the representative structures can serve as a blueprint for future experimental verification of the proposed structures.


Introduction
For the most part, the investigation of two-dimensional structures is focused on the materials obtained from layered crystals such as graphite, hexagonal boron nitride (h-BN), or transition metal dichalcogenides (TMDs). The main feature of these crystals is a weak van der Waals interaction between the layers, making it possible to easily cleave them into monolayers [1]. Reducing the dimension of the material leads to the appearance of unique new properties. For example, 2D nanomaterials are usually characterized by higher values of carrier mobility and conductivity in comparison with bulk crystals due to changes in electronic properties caused by the quantum confinement effect. Therefore, the search, investigation, and fabrication of new 2D nanomaterials from non-van der Waals crystals has attracted enormous attention [2]. Bulk materials, such as iron pyrite [3] and α-germanium [3], have relatively weak covalent bonds and can be exfoliated. Moreover, the achievements in liquid-phase exfoliation techniques have led to the rapid development of a second approach to 2D materials' fabrication, which is the cleaving of non-van der Waals bulk materials into individual monolayers [2]. The successful application of this approach has led to the expansion of the 2D nanomaterials family and the fabrication of stable and easily processable nanomaterials [2].
In the 2D form, the TMDs are widely represented by MX 2 stoichiometry, where M = Mo, V, W and X = S, Se, Te. In this stoichiometry, multiple phases have been experimentally observed-the hexagonal H phase and the tetragonal T and T' (or Td) phases,    Chevrel monolayers can be potentially fabricated by the cleavage of corresponding bulk ternary molybdenum chalcogenides along the (110) direction for non-Janus compounds. To consider this possibility, the exfoliation energy (E ex f ) was estimated through the equation proposed in [66]: where E ML is the total energy of a monolayer, E bulk is the total energy of bulk material consisting of m monolayers (in our case m = 1), and A is the in-plane area according to the relaxed bulk unit cell.    We investigated the electronic properties of dynamically stable Ch 1 2Mo3Ch 2 (Ch 1 /Ch 2 = O, S, Se, Te) and Ch2Mo3Hal2 (Ch = O, S, Se, Te and Hal = F, Cl, Br, I) mono layers. Non-Janus Mo3S4 Chevrel monolayers exhibit semiconducting properties with a band gap value of ~0.2 eV, while Mo3Te4 monolayers are metals (see Figure S2). Electroni band structures for the considered Janus monolayers are presented in Figures 4 and 5 The substitution of chalcogen atoms on one side of the Chevrel non-Janus monolaye leads to the opening/increase of the band gap, and all considered monolayers exhibi semiconducting electronic properties (see Table 3). The halogen atoms' influence on th  The substitution of chalcogen atoms on one side of the Chevrel non-Janus monolayer leads to the opening/increase of the band gap, and all considered monolayers exhibit semiconducting electronic properties (see Table 3). The halogen atoms' influence on the band gap increasing is more significant than chalcogen and the band gap values range from near-zero values in the case of Se 2 Mo 3 Te 2 to 1.17 eV in the case of S 2 Mo 3 F 2 . All Janus Chevrel monolayers except for O 2 Mo 3 F 2 have an indirect band gap, while O 2 Mo 3 F 2 is a direct band gap semiconductor. In all cases, the main contribution to the valence band maximum (VBM) and conduction band minima (CBM) comes from the molybdenum atoms. The features of the monolayer type of the Chevrel phase are that the molybdenum atoms are in a square pyramidal environment making their d-orbitals available for interaction with adsorbate molecules, which is useful for catalytic applications. In this case, the replacement of chalcogen atoms by halogen atoms leads to electron density redistribution among the surface Mo atoms since the electronegativity difference makes these Mo atoms attractive to ion sorption. Bader charge analysis of surface Mo atoms (see Table S1) shows similar values of electron deficiency on Mo atoms, which depend on the chalcogen or halogen type and do not depend on its combination in the Janus structure. For example, the Mo atom on the S side has an electron deficiency of 0.89 e, which makes this atom attractive for negatively charged ions. Mo atoms attractive to ion sorption. Bader charge analysis of surface Mo atoms (see Table  S1) shows similar values of electron deficiency on Mo atoms, which depend on the chalcogen or halogen type and do not depend on its combination in the Janus structure. For example, the Mo atom on the S side has an electron deficiency of 0.89 e, which makes this atom attractive for negatively charged ions.     To be applied in water-splitting reactions, the CBM of the non-Janus monolayer must be higher than the reduction level of hydrogen, while the VBM must be lower than the oxidation level of oxygen [73], resulting in restriction of the band gap value: Eg > 1.23 eV [73]. The feature of Janus monolayers is the presence of two inequivalent surfaces-(001) and (001 ̅ )-leading to the appearance of an intrinsic dipole moment, resulting in the electrostatic potential difference between the opposite sides of the monolayers. The intrinsic dipole moment in the Janus monolayer leads to the modification of the restriction on the band gap value: where ΔΦ is the difference between the vacuum levels of the opposite sides of Janus monolayers. Additionally, the necessary condition of the photocatalytic water-splitting process is the following: the VBM of the (001) surface must be lower than the oxidation potential, while the CBM of the (001 ̅ ) surface must be higher than the reduction potential [73]. It should be noted that in actuality, the redox potentials of water are pH resistant [74]. To be applied in water-splitting reactions, the CBM of the non-Janus monolayer must be higher than the reduction level of hydrogen, while the VBM must be lower than the oxidation level of oxygen [73], resulting in restriction of the band gap value: E g > 1.23 eV [73]. The feature of Janus monolayers is the presence of two inequivalent surfaces-(001) and 001 -leading to the appearance of an intrinsic dipole moment, resulting in the electrostatic potential difference between the opposite sides of the monolayers. The intrinsic dipole moment in the Janus monolayer leads to the modification of the restriction on the band gap value: where ∆Φ is the difference between the vacuum levels of the opposite sides of Janus monolayers. Additionally, the necessary condition of the photocatalytic water-splitting process is the following: the VBM of the (001) surface must be lower than the oxidation potential, while the CBM of the 001 surface must be higher than the reduction potential [73]. It should be noted that in actuality, the redox potentials of water are pH resistant [74]. The band edge positions of dynamically stable Ch 1 2 Mo 3 Ch 2 2 (Ch 1 /Ch 2 = O, S, Se, Te) and Ch 2 Mo 3 Hal 2 (Ch = O, S, Se, Te and Hal = F, Cl, Br, I) monolayers, taking into account the influence of the intrinsic dipole moment concerning the redox potentials of water at pH = 0 and pH = 7, are presented in Figure 6. According to the obtained results, Ch 1 2 Mo 3 Ch 2 2 Janus monolayers do not provide either the reaction of hydrogen production or the reaction of oxygen generation at both considered values of pH; therefore, these structures are not consistent for photocatalytic water splitting.
The band edge positions of dynamically stable Ch 1 2Mo3Ch 2 2 (Ch 1 /Ch 2 = O, S, Se, Te) and Ch2Mo3Hal2 (Ch = O, S, Se, Te and Hal = F, Cl, Br, I) monolayers, taking into account the influence of the intrinsic dipole moment concerning the redox potentials of water at pH = 0 and pH = 7, are presented in Figure 6. According to the obtained results, Ch 1 2Mo3Ch 2 2 Janus monolayers do not provide either the reaction of hydrogen production or the reaction of oxygen generation at both considered values of pH; therefore, these structures are not consistent for photocatalytic water splitting. Most of the Ch2Mo3Hal2 monolayers belong to the type-1 material in neutral and acidic media, according to [75], as this material is suitable for only the HER or the OER due to the band alignments (see Table S2).
Overall water splitting can be achieved for S2Mo3I2 in neutral media and S2Mo3Br2 in alkali media. A possible way to tune electronic properties is through a variation of halogen content on one side of the Janus monolayer. The replacement of a Br with an S atom (S2Mo3Br2) leads to CBM shifting on the halogen side, which makes this material suitable for HER in neutral and acidic media (see Figure S3). However, the reverse side of the coin is that CBM will shift upwards on the chalcogen side.
One more important property of the catalyst material in the case of the photocatalytic process is the ability of the material to absorb the energy of incident solar radiation. To characterize the considered materials, we calculated the optical properties by considering the complex dielectric function ( ) = 1 ( ) + 2 ( ), in which the real part 1 ( ) was calculated using the Kramers-Kronig relation, while the imaginary part 2 ( ) was determined by the sum of empty states [76,77]. The wavelength-dependent extinction coefficient in the perpendicular direction and parallel to the surface of the monolayer was calculated according to [78] as: The resulting dependences for the Janus monolayers suitable for a catalytic process are presented in Figure 7, whileCh 1 2Mo3Ch 2 2 and Mo3Ch are presented4 in Figures S4 and  S5, respectively. Overall, Janus Chevrel monolayers suitable for catalysis (Figure 7) are characterized by a strong peak at 250-300 nm in both the transverse direction and perpendicular to the surface, which correspond to the UV region. In the case of oxygen-containing Janus structures, there is also a separate peak at 150-160 nm (UV region) in both the transverse direction and perpendicular to the surface. In the case of Te-containing monolayers, the intensity of the extinction coefficient decreases compared with pristine monolayers (Figure S4), while in other cases, the intensity is comparable. Overall, Janus monolayers can absorb the light at ultraviolet and visible spectral ranges, Most of the Ch 2 Mo 3 Hal 2 monolayers belong to the type-1 material in neutral and acidic media, according to [75], as this material is suitable for only the HER or the OER due to the band alignments (see Table S2).
Overall water splitting can be achieved for S 2 Mo 3 I 2 in neutral media and S 2 Mo 3 Br 2 in alkali media. A possible way to tune electronic properties is through a variation of halogen content on one side of the Janus monolayer. The replacement of a Br with an S atom (S 2 Mo 3 Br 2 ) leads to CBM shifting on the halogen side, which makes this material suitable for HER in neutral and acidic media (see Figure S3). However, the reverse side of the coin is that CBM will shift upwards on the chalcogen side.
One more important property of the catalyst material in the case of the photocatalytic process is the ability of the material to absorb the energy of incident solar radiation. To characterize the considered materials, we calculated the optical properties by considering the complex dielectric function ε(λ) = ε 1 (λ) + iε 2 (λ), in which the real part ε 1 (λ) was calculated using the Kramers-Kronig relation, while the imaginary part ε 2 (λ) was determined by the sum of empty states [76,77]. The wavelength-dependent extinction coefficient in the perpendicular direction and parallel to the surface of the monolayer was calculated according to [78] as: The resulting dependences for the Janus monolayers suitable for a catalytic process are presented in Figure 7, whileCh 1 2 Mo 3 Ch 2 2 and Mo 3 Ch are presented 4 in Figures S4 and S5, respectively. Overall, Janus Chevrel monolayers suitable for catalysis (Figure 7) are characterized by a strong peak at 250-300 nm in both the transverse direction and perpendicular to the surface, which correspond to the UV region. In the case of oxygen-containing Janus structures, there is also a separate peak at 150-160 nm (UV region) in both the transverse direction and perpendicular to the surface. In the case of Te-containing monolayers, the intensity of the extinction coefficient decreases compared with pristine monolayers (Figure S4), while in other cases, the intensity is comparable. Overall, Janus monolayers can absorb the light at ultraviolet and visible spectral ranges, indicating a high potential for application in photocatalytic water-splitting reactions and as an element of optoelectronic devices.
indicating a high potential for application in photocatalytic water-splitting reactions and as an element of optoelectronic devices. To facilitate the experimental identification of Chevrel monolayers, we simulated vibrational spectra for the initial bulk Mo6S8 Chevrel crystal [79] (Figure 8a,b) and the individual Mo6S8 monolayer (Figure 8c,d). Experimental Raman spectra of bulk Mo6S8 found in [21,80] indicate the existence of two strong sharp peaks in a high-wavenumber region associated with Mo-S stretching vibrations [21]. The same strong peaks were ob- To facilitate the experimental identification of Chevrel monolayers, we simulated vibrational spectra for the initial bulk Mo 6 S 8 Chevrel crystal [79] (Figure 8a,b) and the individual Mo 6 S 8 monolayer (Figure 8c,d). Experimental Raman spectra of bulk Mo 6 S 8 found in [21,80] indicate the existence of two strong sharp peaks in a high-wavenumber region associated with Mo-S stretching vibrations [21]. The same strong peaks were observed in the case of NiMo 3 S 4 [81]; however, no such peaks were found in the case of BaMo 6 S 8 , Cu 1.8 Mo 6 S 8 , Cu 3.2 Mo 6 S 8 , PbMo 6 S 8 , and SnMo 6 S 8 crystals with a Chevrel structure [82]. In [83], the Raman spectra of Mo 6 S 3 I 6 nanowires are presented in a wide range of temperatures and the existence of such peaks is shown in the case of samples synthesized at 900 and 1000 • C. At temperatures above 1000 • C, a powder XRD pattern of Mo 6 S 3 I 6 no longer contains lines fitting with MoS 2 and the related bands disappear from the Raman spectra. In Figure 8a, it can be seen that the high-wavenumber region of the simulated spectrum of bulk Mo 6 S 8 does not contain these strong bands, but if we consider the combination of bulk Mo 6 S 8 and bulk MoS 2 spectra, the resulting figure (Figure 8a) is in good agreement with the experimental data from [21,80]. Thus, we suppose that the samples studied in [21,80,81] can have the impurities of MoS 2 . Due to the complexity of the Mo 6 S 8 structure, the Raman spectrum obtained in this work theoretically can be assigned based on the phonon density of states calculation (see Figure 8b). The contribution of sulfur ion vibrations is significantly above 250 cm −1 . A medium band at 181 cm −1 and a strong band at 231 cm −1 are associated with Mo translation. The low-wavenumber spectral range contains both Mo and S ion vibrations.
The phonon DOS for the Mo 3 S 4 Chevrel monolayer is shown in Figure 8d. Despite the large sulfur phonon density of states in the high-wavenumber region (above 300 cm −1 ), the Raman intensities in the spectrum shown in Figure 8d are weak. There are no totally symmetric vibrations of the Mo 3 S 4 Chevrel monolayer like in the case of MoS 2 [84]. For example, the vibrational mode with the highest wavenumber value (404 cm −1 ) involves only half of the sulfur ions on each monolayer surface and, at the same time, the sulfur ions remain motionless on the opposite surface (see Figure S6a). Such behavior is similar to the MoSSe Janus monolayers [85]. In this case, the vibration of only one side of the monolayer does not manifest itself as a strong, medium, or even weak Raman band. The strong band at 284 cm −1 is associated with in-plane ions' movement, as shown in Figure S6b. The strong spectral band at 210 cm −1 can be described as the stretching of a Mo 3 S 4 Chevrellike molecule, as presented in Figure S6c. The strong lines at 141 cm −1 ( Figure S6d) and 123 cm −1 are out-of-plane and in-plane vibrations, respectively. Meanwhile, the spectral lines in Figure 8c above 100 cm −1 can be described as the translations of ions, and the contribution of these ions can be determined using Figure 8d. Raman bands below 100 cm −1 are assigned as the rotations of Mo 3 S 4 Chevrel-like molecules (see Figure S6c).
As a representative of Ch 2 Mo 3 Hal 2 Chevrel monolayers, we considered the S 2 Mo 3 Br 2 monolayer and simulated its Raman spectra as dynamically stable and suitable for the photocatalytic process (Figure 8e). According to the phonon density of states shown in Figure 8f, the vibrations of S and Mo ions contribute to the main intense spectral bands of the S 2 Mo 3 Br 2 Raman spectrum (from 200 to 417 cm −1 ). Bromine vibrations are in the region below 200 cm −1 ; however, the calculated intensities of corresponding Raman bands are weak. The low-intensity bands above 350 −1 cm in the Raman spectrum of S 2 Mo 3 Br 2 are mainly associated with S vibrations, the forms of which are shown in Figure S6f,g. The strongest spectral band in Figure 8e is associated with Mo-S stretching vibrations. The group of spectral bands from 200 to 235 cm −1 can be attributed to antisymmetric movements of S ions as, for example, are shown in Figure S6h.

Conclusions
The photocatalytic water-splitting materials can most likely be found in Janus-type structures consisting of transition metals, chalcogen, and halogen atoms. Here, we theo-

Conclusions
The photocatalytic water-splitting materials can most likely be found in Janus-type structures consisting of transition metals, chalcogen, and halogen atoms. Here, we theoretically investigated a novel type of Janus structure with the Chevrel phase topology and considered the physical properties and stability. The calculated exfoliation energy from the corresponding bulk chalcogenides demonstrated the possibility of monolayer exfoliation, which can be transformed to suit Janus structures by replacing chalcogen atoms with halogen atoms from one side of the monolayer. The electronic structure calculations demonstrated that halogen doping increases the band gap compared with non-Janus structures, which makes these materials appropriate for the HER and the OER. Among all the considered 2D structures, S 2 Mo 3 I 2 is the most promising candidate for overall water splitting due to the appropriate band alignments relative to the redox potentials of water and the ability to absorb at the UV and visible regions of the light spectra. The calculated Raman spectra can play the role of characteristic blueprints, facilitating the experimental identification and verification of the considered structures. The theoretical results that were obtained open up the prospects for the experimental synthesis and investigation of new compounds based on the Chevrel monolayer phase for photocatalytic as well as for other technological applications.