Background

Two-dimensional (2D) transitional metal dichalcogenide monolayers have received increasing attention because of their amazing physical, chemical, electronic, and magnetic properties [18]. The transition-metal dichalcogenides have the formula of MX2 (M is a transition metal element from groups IV to VI, and X is a chalcogen element), where one M-atom layer is sandwiched between two X-atom layers [1]. These 2D monolayers have been extensively investigated for possible applications in many areas of science and technology, from nanodevices, photoelectronics, catalysts, to the bioscience [917]. Their application as catalysts for hydrogen production in water electrolysis is particularly interesting because of their features, such as low cost, easy large-scale fabrication, and rich abundance on Earth [8, 11, 1831]. Numerous studies have shown that the catalytic performance of 2D MX2 nanostructures is closely related to their conductivity and active sites at edges [1833]. For example, the metallic edges of MoS2, such as the zigzag edge, are active for hydrogen evolution reaction (HER) in water electrolysis [2832]. Metallic MX2 showed better catalytic activity than its semiconducting counterpart [21, 26]. To enhance the performance, the MoS2/graphene composite had been studied for HER because graphene may improve the conductivity and modify their morphologies [24, 27]. Recently, Pan reported that vanadium disulfide (VS2) monolayer shows the best HER performance in the considered systems and its catalytic activity depends on the hydrogen coverage during HER, which is reduced at high coverage due to the change of conductivity [19]. Kong et al. reported that doping is one of possible methods to improve their activity [28]. In this work, we investigate the effect of doping on the catalytic activity of the VS2 monolayer to improve the HER performance on the basis of first-principles calculation. A series of elements, including Ti, Nb, W, Ta, Mo, Pt, Fe, Co, and Ni, are systematically studied. We find that Ti is the best element to easily substitute V in the VS2 monolayer and improve the HER ability. We also show that the doping effect on HER strongly depends on the concentration of dopants.

Methods

The design of catalysts for water electrolysis is based on the first-principles calculation. The hydrogen evolution reduction of the VS2 monolayer with the dopant is investigated to improve its catalytic ability. The Vienna Ab initio Simulation Package (VASP) [34] incorporated with the projector augmented wave (PAW) scheme [35, 36], which is based on the density functional theory (DFT) [37] and the Perdew-Burke-Ernzerhof generalized gradient approximation (PBE-GGA) [38], is used in our calculations. Supercells with lattices larger than 10 Å are used to investigate the doping effect and hydrogen-density-dependent HER ability. A 3 × 3 × 1 grid for k-point sampling, based on the Monkhorst and Pack scheme [39], for geometry optimization of supercells, and an energy cutoff of 450 eV are consistently used in our calculations. Densities of states (DOSs) are calculated based on a k-point sampling of 5 × 5 × 1. To avoid image-image interaction between two monolayers in neighboring supercells in the vertical direction, a vacuum region of at least 20 Å is used for separation. Good convergence is obtained with these parameters, and the total energy was converged to 2.0 × 10−5 eV/atom. Both of spin-unpolarized and spin-polarized calculations are carried out.

Results and Discussion

In our calculations, a supercell with a hexagonal structure is set up on the basis of a unit cell of the VS2 monolayer with one surface fully covered by hydrogen atoms (a = 3.27 Å) [19] (Fig. 1 and Additional file 1: Figure S1), where the S–H bond length is about 1.37 Å. Two supercells with 3 × 3 × 1 and 4 × 4 × 1 unit cells (331 and 441 supercells, respectively) are used to investigate the effect of the doping concentration. Nine transition metal (TM) elements, including Ti, Nb, W, Ta, Mo, Pt, Fe, Co, and Ni, are considered as dopants in our calculations. To realize the doping, one or two V atoms in the supercells are substituted by TM atoms (Fig. 1c). The doped systems are fully relaxed to study the doping possibility and their HER performance.

Fig. 1
figure 1

The representative structures of a fully H-covered VS2 331 supercell: a top view, b side view, and c doped VS2 in the 331 supercell. The indication in (a) shows the way to take away the hydrogen atom one by one

Basically, the HER performance of the catalyst can be characterized by free energy of adsorption of reactive intermediates on its surface based on the Sabatier principle [40]. To qualify the catalytic ability, the reaction free energy of hydrogen adsorption (ΔGH) [19, 4043] is calculated as the following equation:

$$ \Delta {\mathrm{G}}_{\mathrm{H}}=\Delta {\mathrm{E}}_{\mathrm{H}}+\Delta {\mathrm{E}}_{\mathrm{ZPE}}-\mathrm{T}\Delta {\mathrm{S}}_{\mathrm{H}} $$
(1)

where ΔEH is the hydrogen chemisorption energy defined as:

$$ \Delta {\mathrm{E}}_{\mathrm{H}}=\mathrm{E}\left({\mathrm{VS}}_2+n\mathrm{H}\right)-\mathrm{E}\left({\mathrm{VS}}_2+\left(n-1\right)\mathrm{H}\right)-\frac{1}{2}\mathrm{E}\left({\mathrm{H}}_2\right) $$
(2)

where n is the number of H atoms adsorbed on a MX2 monolayer and changed from 1 to 9 (for full hydrogen coverage on the 331 supercell) (Fig. 1a, b) or 1 to 16 (for full hydrogen coverage on the 441 supercell) to investigate the effect of hydrogen coverage on catalytic activity. The hydrogen coverage refers to \( \frac{n}{9} \) (in the 331 supercell) or \( \frac{n}{16} \) (in the 441 supercell). Full coverage refers to each S atom on one side of the VS2 monolayer that is attached with one H atom. Therefore, ΔGH as a function of the hydrogen coverage can be obtained. E(VS2 + nH), E(VS2), and E(H2) in Eq. (2) are the energies of the monolayer with hydrogen atoms (n), pure VS2 monolayer, and hydrogen molecule, respectively. ΔSH is the difference in entropy. The entropy of adsorption of 1/2 H2 is \( \Delta {\mathrm{S}}_{\mathrm{H}}\cong -1/2{\mathrm{S}}_{{\mathrm{H}}_2}^0 \), where \( {\mathrm{S}}_{{\mathrm{H}}_2}^0 \) is the entropy of H2 in the gas phase at standard conditions. ΔEZPE is the difference in zero point energy between the adsorbed and the gas phase, related to the reaction 1/2H2(g) → H *, where H* denotes a hydrogen atom adsorbed on the surface. ΔEZPE − TΔSH is about 0.24 eV [19, 4043]. So, Eq. (1) is simplified to ΔGH = ΔEH + 0.24.

To realize the partial hydrogen coverages, we start from the full hydrogen coverage on the supercell and take hydrogen atoms away one by one, as indicated in Fig. 1a. All of the systems with different hydrogen coverages are relaxed to calculate ΔGH. The relaxed structures show that their geometry are stable and hydrogen atoms keep on the tops of S atoms with the S–H bond of 1.37 Å (Additional file 1: Figure S2). We first study the 331 supercell with one V atom replaced by one TM atom, which is corresponding to a doping concentration of \( \frac{1}{9} \). The calculated Gibbs free energies for hydrogen adsorption on the 331 supercell show that the catalytic activities of doped VS2 monolayers are still dependent on the hydrogen coverage, which decrease with the increment of hydrogen density (Fig. 2). Compared with a pure VS2 monolayer, we see that doping can partially improve its catalytic activity at a certain range of hydrogen density as indicated by the reduced ΔGH (Fig. 2). For example, the Ni-doped VS2 monolayer shows better HER performance than a pure one in ranges of hydrogen density from \( \frac{3}{9}\kern0.37em \mathrm{t}\mathrm{o}\;\frac{4}{9} \) and from \( \frac{6}{9}\kern0.37em \mathrm{t}\mathrm{o}\;\frac{8}{9} \) (Fig. 2a). Pt-doping and Ti-doping improve the performance in hydrogen density ranging from \( \frac{6}{9}\kern0.37em \mathrm{t}\mathrm{o}\;\frac{8}{9} \) (Fig. 2b) and from \( \frac{1}{9}\kern0.37em \mathrm{t}\mathrm{o}\;\frac{5}{9} \) (Fig. 2c). Interestingly, we see that ΔGH for W-doped VS2 at the full hydrogen coverage is almost close to zero (0.09 eV) (Fig. 2c). In all of the considered doping elements, we see that Ni- and Ti-doping can improve the HER performance of the VS2 monolayer in a wide hydrogen coverage and W-doping can dramatically enhance its catalytic activity at a high hydrogen coverage. For comparison, we put the calculated ΔGH of Ni-, Ti-, and W-doped VS2 monolayers as a function of hydrogen density together (Fig. 3a). Clearly, Ti-doping is better than Ni-doping on the HER performance in a range of \( \frac{1}{9}\kern0.37em \mathrm{t}\mathrm{o}\;\frac{5}{9} \), while Ni-doping is better than Ti-doping in a range of \( \frac{6}{9}\kern0.37em \mathrm{t}\mathrm{o}\;\frac{8}{9} \) (Fig. 3a). W-doping is the best at the full hydrogen coverage. The relaxed structure of the W-doped VS2 monolayer with full hydrogen coverage shows that one of the H atoms moves away from the surface around the W-doping site and bonds to other H atoms nearby (inset in Fig. 3a), where the S–H and H–H bond lengths are 1.662 and 0.983 Å, respectively. We also investigate the effect of spin-polarization on the HER performance of the Ti-doped system. We find that spin-polarization may affect slightly the calculated Gibbs free energy at a lower hydrogen coverage but is negligible as the hydrogen coverage increases (Fig. 4). Therefore, spin-polarization is not considered below.

Fig. 2
figure 2

Calculated overpotentials as a function of H-coverage for TM-doped VS2 monolayers in 331 supercells with TM equal to: a Fe, Co, and Ni; b Ta, Mo, and Pt; and c Ti, Nb, and W. For comparison, the overpotentials of the pure VS2 331 supercell are added

Fig. 3
figure 3

Calculated overpotentials as a function of H-coverage for the TM-doped VS2 (TM = Ti, W, Ni) monolayers in a 331 supercell and b 441 supercell

Fig. 4
figure 4

a Calculated overpotentials of Ti-doped VS2 in the 331 supercell with and without spin polarization. b Calculated energy difference between VS2-Ti with and without spin polarization, at various hydrogen coverages

To investigate the effect of the doping concentration on the HER performance, the increment and reduction of the doping density are achieved by replacing more V atoms in the same supercell and enlarging the size of the supercell, respectively. For example, we replace two V atoms using two TM (Ti, Ni, or W) atoms in the 331 supercell, which is equivalent to a doping density of 22.22 % \( \left(\frac{2}{9}\right) \). The calculated Gibbs free energies show that increasing the doping density makes the HER performance worse, indicating that a high doping concentration is not good in application (Additional file 1: Figure S3). Then, we use a 441 VS2 supercell with one V substituted by one TM atom (TM = Ti, Ni, and W), which equals to a doping density of about 6.25 % \( \left(\frac{1}{16}\right) \). Clearly, the TM-doping can improve the HER performance of the VS2 monolayer under high-H coverage in a range of \( \frac{8}{16}\;\mathrm{t}\mathrm{o}\;\frac{16}{16} \) (Fig. 3b). The catalytic abilities of Ti- and Ni-doped systems are enhanced by 50 % in the hydrogen coverage from \( \frac{8}{16}\;\mathrm{t}\mathrm{o}\;\frac{11}{16} \) (Fig. 3b). However, Ni-doping reduces its performance in hydrogen coverages less than \( \frac{8}{16} \). The catalytic ability of the Ti-doped VS2 monolayer at low hydrogen coverages is comparable with or slightly worse than that of the pure VS2 monolayer. Similarly, W-doping in the 441 supercell can only improve the HER performance at certain hydrogen densities, such as \( \frac{9}{16}\;\mathrm{and}\;\frac{16}{16} \) (Fig. 3b). The relaxed structure of W-doped VS2 with the full hydrogen coverage shows that hydrogen atoms around the doping site move together to form a triangle with a H–H bond length of 1.00 Å and an extended S–H length of 1.90 Å (inset in Fig. 3b). From the calculated Gibbs free energies, we see that Ti is the best candidate as a dopant to improve the HER performance of the VS2 monolayer. Comparing with other MX2 monolayers, we see that the basal plane of the Ti-doped VS2 monolayer has better catalytic performance under the same condition (ΔGH = −0.19 eV at 6.25 % coverage) than that of pure 1T-WS2 (ΔGH = 0.28 eV) [21]. It is also worth noting that after doping Ti, Ni, and W atoms individually into the VS2 monolayer, the catalytic ability of the basal plane on the doped VS2 monolayer has improved dramatically at certain hydrogen coverages, which are equivalent to or even better than that of the active edges sites of MoS2, MoSe2, WS2, and WSe2 under the same hydrogen coverages [20]. For example, the catalytic activity of the Ti-doped VS2 monolayer (its ΔGH equals −0.05 eV under a hydrogen coverage of \( \frac{2}{9} \) (22.2 %)) is comparable to or better than that at the edges of MoS2, MoSe2, WS2, and WSe2 (the optimal ΔGH is from −0.06 to 0.06 eV under a hydrogen coverage of 25 %).

To reveal the mechanism of the doping effect on HER performance, the partial density of states is calculated and shows the shift of the Fermi level as the hydrogen coverage increases (Figs. 5 and 6; Additional file 1: Figures S4–S7). We see that the densities of states of the Ti-doped VS2 monolayer under various hydrogen coverages near Fermi levels are mainly contributed to Ti-d and V-d electrons (Figs. 5 and 6). In the 331 supercell, localized defect states near the Fermi level are formed at high hydrogen coverage (Fig. 5c, d), which may lead to reduced carrier mobility and HER performance (Fig. 3a). However, the doping states near the Fermi level in the 441 supercell are connected to the valence band (Fig. 6c, d), resulting in better carrier mobility and improved catalytic activity at high hydrogen coverages (Fig. 3b).

Fig. 5
figure 5

Calculated partial density of states of the Ti-doped VS2 monolayer in the 331 supercell with a hydrogen coverage at: a 1/9, b 3/9, c 8/9, and d 9/9

Fig. 6
figure 6

Calculated partial density of states of the Ti-doped VS2 monolayer in 441 supercell with a hydrogen coverage at: a 2/16, b 5/16, c 14/16, and d 16/16

Although the calculated Gibbs free energies show that Ti-doping may improve the HER performance of the VS2 monolayer, another important issue, which is the doping ability, needs to be stated. The possibility for the dopant to substitute the V atom in the host can be investigated by calculating the formation energy as below:

$$ {\mathrm{E}}_f=\left(\mathrm{E}\left({\mathrm{V}\mathrm{S}}_2+n\mathrm{T}\mathrm{M}\right)-\mathrm{E}\left({\mathrm{V}\mathrm{S}}_2\right)-n{\mu}_{\mathrm{TM}}+n{\mu}_{\mathrm{V}}\right)/n $$
(3)

where E(VS2 + nTM) and E(VS2) are the total energies of VS2 monolayer supercells with and without dopants. μ TM and μ V are the energies of TM and V atoms, respectively. n is the number of dopants in each supercell (n = 1). Our calculations show that the formation energies for Ti-, Mo-, Nb-, and Ta-doping are negative, indicating the reactions are exothermic, while the doping of W, Fe, Co, Ni, and Pt are endothermic because of their positive formation energies (Fig. 7). Particularly, it is easy to substitute the Ti atom to the V atom in the 331 supercell of the VS2 monolayer because of its lowest formation energy (−0.83 eV) (Fig. 7a). W-doping may be achieved under suitable conditions because its endothermic energy is as low as 0.06 eV. Compared with other elements, however, Ni-doping should be difficult because large energy is required (E f  = 1.88 eV). We further see that their formation energies are reduced if the doping density decreases (Fig. 7b). In this case, the energies of Ti- and W-substitutions are reduced to −1.14 and 0.005 eV, respectively, at a doping density of 6.25 % \( \left(\frac{1}{16}\right) \) (Fig. 7b), indicating that doping at a low concentration is easier than that at a high concentration.

Fig. 7
figure 7

Calculated formation energies of the TM-doped VS2 monolayers (TM = Ti, Nb, Ta, Mo, W, Fe, Co, Ni, Pt) in a 331 supercell and b 441 supercell

Conclusions

We present a first-principles study on the effect of doping on the hydrogen evolution reaction of the VS2 monolayer. We find that the catalytic activity of the doped VS2 monolayer depends strongly on the choice of dopant and the doping concentration. The catalytic ability of the VS2 monolayer under high hydrogen coverages can be dramatically enhanced by TM-doping at a low concentration, while that under low hydrogen coverages can be improved by the doping at a moderate density. High-density doping results in reduced HER activity. We further show that Ti-doping should be the best to improve the HER ability of VS2 monolayers in our considered doping elements because of the reduced Gibbs free energy at a wide range of hydrogen coverages. By investigating the formation energy of TM substitution of the V atom, we find that the reaction of Ti-substitution of V in the VS2 monolayer is exothermic and easier than other TM elements due to its lowest formation energy. It is predicted that the Ti-doped VS2 monolayer may show better HER performance and find applications as catalysts in water electrolysis.