The Origin of High Activity of Amorphous MoS2 in the Hydrogen Evolution Reaction.

Molybdenum disulfide (MoS2 ) and related transition metal chalcogenides can replace expensive precious metal catalysts such as Pt for the hydrogen evolution reaction (HER). The relations between the nanoscale properties and HER activity of well-controlled 2H and Li-promoted 1T phases of MoS2 , as well as an amorphous MoS2 phase, have been investigated and a detailed comparison is made on Mo-S and Mo-Mo bond analysis under operando HER conditions, which reveals a similar bond structure in 1T and amorphous MoS2 phases as a key feature in explaining their increased HER activity. Whereas the distinct bond structure in 1T phase MoS2 is caused by Li+ intercalation and disappears under harsh HER conditions, amorphous MoS2 maintains its intrinsic short Mo-Mo bond feature and, with that, its high HER activity. Quantum-chemical calculations indicate similar electronic structures of small MoS2 clusters serving as models for amorphous MoS2 and the 1T phase MoS2 , showing similar Gibbs free energies for hydrogen adsorption (ΔGH* ) and metallic character.

Scalable electrochemical proton reduction (hydrogen evolution reaction, HER) is crucial for realizing large‐scale storage of renewable energy. Water splitting requires efficient and robust catalysts, which are composed of earth‐abundant elements. The most active metal catalyst for HER is Pt, which is scarce and thus makes scale‐up of water electrolysis to terawatt (TW) scale too costly.1 Molybdenum disulfide (MoS2), one of the most studied transition metal chalcogenides (TMCs), has received substantial attention because of its unique physiochemical properties, such as a tunable band gap,2 high catalytic activity,3 and high electron mobility.4 These properties allow it to be exploited in transistors,5 metalloenzymes,6 and, at a practical scale, as the active phase in industrial catalysts for hydrotreatment of oil fractions.7 The ability to activate hydrogen reversibly also explains its promise for catalyzing hydrogen evolution in the context of electrochemical water splitting.8 Not surprisingly, the edge sites of MoS2 nanocatalysts have been identified as active HER sites by Jaramillo and co‐workers.9 Since then, tremendous efforts have been devoted to engineering the surface structure of MoS2 to preferentially expose these edge sites to improve HER performance.10

Different polymorphs of MoS2 exist in the form of 2H (trigonally coordinated), 1T (octahedrally coordinated), and 3R phases (rhombohedral).3, 11 2H‐MoS2 is the thermodynamically stable two‐dimensional (2D) phase with semiconductor properties (band gap≈1.9 eV for monolayer, 1.2 eV for bulk),12 a low electron mobility, and a limited number of HER‐active (edge) sites. These properties render this phase less attractive for electrocatalytic applications2, 13 than the octahedral 1T phase, which is metallic and six orders of magnitude more conductive.13a The improved charge transfer kinetics and the affinity for binding H atoms on 1T‐MoS2 are reported to be responsible for the substantially enhanced HER activity compared to the 2H phase. However, the underlying mechanism of the high HER activity of the 1T phase has yet to be elucidated.14 As the 1T phase can be formed from 2H‐MoS2 by intercalation of cations (e.g., Li+, Na+),15 it is usually characterized by distorted structural domains.16 Aside from 2H and 1T phases, amorphous MoS (x=2–3) has also been extensively investigated in the past as a hydrotreatment catalyst and as a cathode material in lithium‐ion batteries.17 Furthermore, it was recently reported by Hu and other groups that this form of MoS2 is a highly active electrocatalyst for HER.8b, 18 Although Mo edge sites of 2H‐MoS2 have been experimentally identified as the active HER sites, the question of what causes the superior catalytic performances of 1T and amorphous phase MoS2 remains unclear, which adds to the challenge of unraveling the HER mechanism in amorphous MoS2.11, 17b, 17c, 19

Here, we show how the structure and surface properties of 2H, 1T, and amorphous MoS2 influence the HER activity and stability by a combined theory as well as ex situ and operando X‐ray spectroscopy approach. In comparison to 2H‐MoS2, shorter Mo−S and Mo−Mo bonds were observed in both 1T and amorphous MoS2 thin film electrodes. Besides, both core level Mo 3d and valence band photoemission spectra indicate that 1T and amorphous phase MoS2 exhibit a similar electronic structure. The short Mo−Mo bond in 1T phase MoS2 is caused by lithium intercalation and gradually changes back to the 2H phase accompanied by a decrease in HER activity at high overpotentials. By contrast, amorphous MoS2 (Am‐MoS2) retains its intrinsic (short) Mo−S and Mo−Mo bond structure as well as high HER activity after 24 h electrochemical testing under the same conditions. Electrochemical operando X‐ray absorption spectroscopy was performed to probe the local bond and electronic structure of MoS2 under HER conditions. The results show that the observed short Mo−Mo bonds play a key role in determining the activity of both 1T and amorphous phase MoS2 electrocatalysts for HER.

2H and amorphous MoS2 films were prepared by plasmaenhanced atomic layer deposition (PEALD) on glassy carbon plates at 450 and 250 °C, respectively, whereas the 1T phase was synthesized by lithium intercalation of the as‐deposited 2H‐MoS2 (see the Supporting Information for details). The HER electrocatalytic activity of as‐prepared MoS2 films was assessed in 0.1 m H2SO4 in a typical three‐electrode electrochemical cell. As shown in Figure 1, both cyclic voltammetry (CV) and linear sweep voltammetry (LSV) curves present higher current densities for 1T and amorphous MoS2, as compared to the 2H phase. However, even though 1T and amorphous MoS2 have comparable current densities initially, the catalytic activity of the 1T phase gradually decreases during the stability test, whereas Am‐MoS2 maintained its higher initial activity (Figure 1 d). This particular behavior led us to investigate further the electronic and structural properties of the materials.

Figure 1

a, b) Cyclic voltammetry (CV; a) and linear sweep voltammetry (LSV; b) curves of 2H‐, 1T‐, and Am‐MoS2 films corrected by uncompensated resistance with scan rates of 50 mV s−1 for CV and 5 mV s−1 for LSV. c) Tafel slopes obtained from LSV curves in (b). d) Chronopotentiometric responses (V–t) recorded at a constant current density of 3 mA cm−2. Electrolyte: 0.1 m H2SO4.

We used X‐ray absorption spectroscopy at the Mo K‐edge to probe the electronic as well as local geometric structure of these films. Ex situ X‐ray absorption near‐edge spectra (XANES) of MoS2 films before and after HER stability tests recorded under a grazing incidence angle of 0.3° (grazing incidence X‐ray absorption spectroscopy)23 are shown in Figure 2. The suppression of features A and D in 1T (Figure 2 d) compared to 2H‐MoS2 emphasizes its distinct bond structure. Importantly, features A and D reappear for 1T‐MoS2 after 24 h HER stability testing, which implies that the 1T phase is not stable under the HER conditions and gradually changes back to 2H‐MoS2. Feature D is absent for the amorphous phase MoS2 both before and after the HER, indicating its stable bond structure (Figure 2 g), which is in contrast to 2H‐MoS2. Absorption edge features in the XANES spectra are very sensitive to the electronic properties of the atoms being probed:20 the less expressed shoulder at the edge and the shift of the white line for 1T‐MoS2 compared to 2H‐MoS2 are indicative of the structural differences. Simulations of the Mo‐K edge XANES spectra of MoS2 with hexagonal (2H phase) and monoclinic (1T phase with Li intercalation) symmetry were performed to understand these differences. The red curves (Figure 2 j, k) represent calculated spectra based on the model structure and the blue curves are calculated taking into account broadening by core‐hole lifetime effects.21 The fitted XANES spectra in both cases reproduce the experimental features of 2H‐ and 1T‐MoS2 well, which confirms their assignment. As monoclinic MoS2 shows octahedral Mo coordination with a shorter bond distance than 2H‐MoS2 upon Li intercalation, we may conclude that the as‐prepared 1T‐MoS2 in this study has a distorted bond structure. For further comparison, ex situ grazing incidence extended X‐ray fine structure (GI‐EXAFS) data of MoS2 films were recorded before and after stability measurements. The Fourier transform (FT) profiles in R‐space (Figure 2 b, c) present two main peaks at 2.40 Å and 3.16 Å (Table 1) corresponding to the nearest Mo−S and Mo−Mo bonds, respectively. The coordination number (CN) values shown in Table 1 suggest that there is no complete shell of S atoms around the central Mo at the surface of the MoS2 films, which can be due to termination by Mo edges or oxidation by emersion from the electrolyte and air exposure.22 By contrast, FT curves of 1T‐MoS2 exhibit a distinct decrease of the Mo−Mo bond length (short Mo−Mo bond) from 3.16 Å to 2.75 Å (Table 1), which corresponds to the characteristic bond length found in 1T phase MoS2.24, 25 Evidence for this feature can be also found by the larger Debye–Waller (σ 2) factor of both Mo−S and Mo−Mo bonds in 1T phase compared to 2H‐MoS2 (see the Supporting Information, Table S1) although the normal Mo−Mo bond (3.16 Å) is still present in 1T.24 Nevertheless, this shortened Mo−Mo bond disappeared after 24 h HER stability testing, which is consistent with the observations from XANES that 1T changes back to the 2H phase under these conditions. In the case of Am‐MoS2, a similarly short Mo−S and Mo−Mo bond structure was found (Figure 2 h, i). The similarities between the 1T and amorphous phases in XANES and EXAFS are also reflected in the Mo 3d core level and valence band photoemission spectra (Figure S3, Tables S2 and S3), shifting consistently to lower binding energies. Therefore, we may suggest that the short Mo−Mo bond features observed in both 1T‐MoS2 and Am‐MoS2 play a key role in enhancing the HER activity of MoS2 catalysts. Distinct from the 1T phase, the short Mo−Mo bond in Am‐MoS2 was retained after 24 h of HER stability testing. Considering the HER stability of Am‐MoS2, we may conclude that the bond structure in Am‐MoS2 is intrinsic (viz. not caused by Li intercalation), resulting in a higher stability during 24 h HER stability testing.

Figure 2

a, d, g) Mo K‐edge XANES spectra of 2H‐MoS2 (a), 1T‐MoS2 (d), and Am‐MoS2 (g) before (solid line) and after (dash line) stability test. b, e, h) Mo K‐edge Fourier transform EXAFS (k 3‐weighted) of 2H‐MoS2 (b), 1T‐MoS2 (e), and Am‐MoS2 (h) before stability test. c, f, i) Mo K‐edge Fourier transform EXAFS (k 3‐weighted) of 2H‐MoS2 (c), 1T‐MoS2 (f), and Am‐MoS2 (i) after stability test. j, k) Mo‐K edge XANES spectra of experimental data (black curve) and calculated simulation based on hexagonal (j, inset) and monoclinic (k, inset) structure model (purple, yellow, and green balls corresponds to Mo, S, and Li atoms, respectively); red curves represent simulated spectra whereas blue curves represent simulated spectra convoluted with the Mo 1s core‐hole lifetime.

Table 1

Summary of the ex situ grazing incidence Mo K‐edge EXAFS spectroscopic features obtained for MoS2 films under grazing incidence reflecting information about the top ≈3 nm of the material.[a]

Sample	Shell	Fresh		Spent
		CN	R [Å]	CN	R [Å]
2H‐MoS2	Mo−S	4.25	2.402	4.80	2.405
	Mo−Mo	2.26	3.155	2.98	3.158
1T‐MoS2	Mo−S	3.06	2.419	5.86	2.365
	Mo−S (short)	1.78	2.019	–	–
	Mo−Mo	1.70	3.148	2.26	3.145
	Mo−Mo (short)	0.96	2.748	–	–
Am‐MoS2	Mo−S	5.29	2.430	3.50	2.368
	Mo−S (short)	0.60	1.767	0.56	1.802
	Mo−Mo (short)	1.08	2.778	1.57	2.824

[a] Detailed fitting parameters can be found in Table S1.

–

–

–

–

Several structural models for amorphous MoS2 or MoS3 have been proposed in previous reports by for example, Hibble et al.17b and Weber et al.11 However, based on our experimental observations, we cannot conclusively assign a structure to Am‐MoS2. Nonetheless, the disorder in amorphous MoS2 reported in this work is consistent with earlier reports.17a, 17d It is worth noting that even though the HER activity of 1T‐MoS2 decreases gradually (Figure 1 d), the corresponding overpotential is still much lower than that of 2H‐MoS2, which we have recently attributed to the presence of remaining Li adsorbed on the 1T‐MoS2 even after loss of intercalated Li.18c Inductively coupled plasma optical emission spectroscopy (ICP‐OES) analysis of MoS2 films after stability tests confirms the presence of adsorbed Li on 1T‐MoS2 (Table S4), and the adsorption of Li on MoS2 was observed to promote the activity of MoS2‐catalyzed hydrogen evolution reaction in our recent work.18c

To follow the structural evolution of the different MoS2 catalysts under HER conditions, an operando electrochemical cell (Figure S5) was developed and applied for X‐ray absorption spectroscopy experiments. The operando EXAFS spectra of different MoS2 polymorphs in dry state and at set potentials of +0.3 V and −0.3 V versus RHE in 0.1 m H2SO4 are shown in Figure 3. Table 2 summarizes the EXAFS fitting results. It can be seen that, despite a small reduction in CN for the Mo−Mo shell, the Mo−Mo and Mo−S bond distances as well as the Mo‐S CN remained the same within the accuracy range, pointing at the overall structural stability of 2H‐MoS2 under HER conditions. For both 1T and Am‐MoS2, a shortened Mo−S bond could be identified as well. In contrast to the disappearance of short Mo−S and Mo−Mo bonds after the 24 h stability test (Figure 2), the operando EXAFS data of 1T‐MoS2 confirms the retention of short Mo−S bonds under various potentials. On the one hand, the operando XAS was not carried out at grazing incidence angle and therefore reflects mostly bulk film information. On the other hand, the operando XAS measurements were performed at −0.3 V vs. RHE with a current density of only −200 μA cm−2 (Figure S6), while the 24 hours stability tests were performed at −3 mA cm−2.

Figure 3

a–c) Mo K‐edge Fourier transform EXAFS (k2‐weighted) of 2H‐ (a), 1T‐ (b), and Am‐ (c) MoS2 under operando electrochemical conditions. Gray region represents R‐range for fitting. d, e) X‐ray photoemission spectra of Mo 3d (d) and S 2p (e) before and after (spent) operando XAS measurements.

Table 2

Summary of ex‐situ grazing incidence Mo K‐edge EXAFS spectroscopic features obtained for MoS2 films under operando HER conditions.[a]

Sample	Shell	Dry		+0.3 V		−0.3 V
		CN	R [Å]	CN	R [Å]	CN	R [Å]
2H‐MoS2	Mo−S	4.28	2.385	4.33	2.391	4.71	2.403
	Mo−Mo	3.06	3.154	2.70	3.142	2.11	3.165
1T‐MoS2	Mo−S	6.66	2.354	5.82	2.416	3.78	2.406
	Mo−S (short)	5.90	2.004	0.96	1.835	0.34	1.827
Am‐MoS2	Mo−S	4.60	2.443	3.04	2.401	2.30	2.407
	Mo−S (short)	0.36	1.775	0.38	1.644	0.58	1.795

[a] Detailed fitting parameters can be found in Table S5, S6, and S7.

The surface electronic structures of 2H, 1T, and amorphous MoS2 films were probed by XPS before and after operando XAS measurements (Figure 3 d, e). The Mo 3d core level component describing the Mo−S bond for pristine 1T and Am‐MoS2 was shifted negatively by around 0.9 eV from that of the 2H phase, which is a characteristic feature for both 1T and amorphous phase MoS2.18, 26 However, the core level of MoIV−S (Mo 3d5/2 228.6 eV) for 1T‐MoS2 shifted back to 229.5 eV after operando XAS tests, suggesting a transformation from 1T to 2H‐MoS2 at the surface. By contrast, the shift in Mo 3d core level spectra for Am‐MoS2 (ca. 0.7 eV) remained unchanged after electrochemical tests (Figure 3 d). Raman spectroscopy was then used to further support the presence of different MoS2 phases (Figure S18).16 There is a redshift of and A1g peaks for 1T‐MoS2 compared to those of 2H‐MoS2, which stays constant before and after operando XAS measurements. The phase stability of the bulk 1T‐MoS2 film under mild reaction conditions is consistent with the EXAFS fitting results (Table 2). Even though sulfur dimers of Am‐MoS2 have been reported to be involved in proton reduction,19c the decreased intensity of sulfur dimers (Figure 3 e) here apparently did not influence the HER activity (Figure S6).

We utilized grazing incidence X‐ray diffraction (GIXRD) to inspect the materials before and after operando XAS measurements. In the diffraction patterns of MoS2 films (Figure S15), the diffraction peaks at 2θ=14.2° (0 0 2) and 33.3° (1 0 1) indicative of 2H‐MoS2 are absent in the pristine 1T phase MoS2 sample. However, the re‐appearance of (0 0 2) and (1 0 1) reflections for the spent 1T‐MoS2 sample suggests that the material gradually changes back to 2H‐MoS2. In addition, scanning electron microscopy (SEM) images (Figure S16) reveal obvious morphology changes for the 1T phase after HER tests. By contrast, neither Raman spectroscopy (Figure S18) nor GIXRD show any peaks before and after HER testing for Am‐MoS2. So far, we may conclude that even under mild HER conditions, the surface bond structure for the 1T phase would disappear and change back into the 2H phase, whereas amorphous MoS2 retains its intrinsic short Mo−Mo bond feature and with that its high HER activity.

By using density functional theory (DFT), we compared the (electronic) structures of Mo3S9 and Mo6S17 clusters as a motif for Am‐MoS2 with those of 2H‐MoS2 and 1T‐MoS2 in order to understand differences in the Gibbs free energy of hydrogen adsorption (ΔG H*), which is considered as a relevant descriptor for HER activity.13b, 27 The Mo−Mo and Mo−S bond distances found for the two small clusters correspond to those observed in 1T‐MoS2 and are shorter than those in 2H‐MoS2 (Figure 4 and Figures S21 and S22). Together with the structural data derived from EXAFS for our samples, this provides good grounds to hypothesize that Am‐MoS2 consists of small MoS clusters with an increased S/Mo ratio and shortened Mo−Mo and Mo−S bonds, similar to what is known for crystalline 1T‐MoS2. We then explored how these structures affect the HER performance for which we computed the Gibbs free energies of hydrogen adsorption (ΔG H*, structures see Figures S23–S27). For optimum HER activity, the value of ΔG H* should be close to zero.28 For Mo3S9, a ΔG H* value of −0.06 eV was computed, which is much more favorable than values of +2.13 eV and +0.78 eV for the basal planes of 2H‐MoS2 and Li‐stabilized 1T‐MoS2, respectively. As the hydrogen activation and formation on MoS2 is known to occur at the edge terminations,9a, 29 we also computed ΔG H* for hydrogen adsorption on the Mo‐edges of 2H‐MoS2 (−0.23 eV) and 1T‐MoS2 (−0.10 eV). These data confirm that the edges of the distorted 1T‐MoS2 phase are the preferred sites for HER in comparison to the edges of 2H‐MoS2 and further suggest that small clusters also have a favorable ΔG H*. Figure 4 also gives the partial density of states (PDOS; Figure 4 d–g) of the two investigated clusters, 1T‐MoS2 and 2H‐MoS2. It can be immediately seen that, similar to 1T‐MoS2, the Mo3S9 and Mo6S17 cluster models exhibit metallic character with their Fermi level crossing the Mo 3d orbitals. In contrast, 2H‐MoS2 is a semiconductor with a band gap of 1.59 eV, which is consistent with valence band spectroscopy (Figures S12 and S13) and earlier theoretical predictions.30, 31 The adsorption of hydrogen does not induce significant changes to the electronic structures, although coupling is observed between H s orbital and Mo d and S p orbitals, consistent with weak bonding and high HER activity. The metallic nature of the Mo3S9 and Mo6S17 clusters and 1T‐MoS2 results in a higher intrinsic electronic conductivity for these materials than for the semiconducting 2H‐MoS2. Therefore, in addition to a more optimum free energy for hydrogen adsorption, the enhanced HER activity of 1T‐MoS2 and Am‐MoS2 can be further rationalized by a higher intrinsic electronic conductivity.

Figure 4

a, b) Optimized Mo3S9 (a) and Mo6S17 clusters serving as models for Am‐MoS2; c) Schematic illustration of the structural evolution between crystalline 2H and 1T MoS2 phases; normalized partial density of states (PDOS) of (d) Mo3S9, (e) Mo6S17 clusters, (f) 2H‐MoS2, and (g) 1T‐MoS2.

In summary, we have provided both experimental and theoretical evidence for the importance of the short Mo−Mo bond structures of 1T and amorphous MoS2 in comparison to crystalline 2H‐MoS2 for explaining the higher HER performance. Whereas crystalline 1T‐MoS2 stabilized by intercalated Li+ also displays high performance, Li ions were found to dissolve in the electrolyte during electrochemical testing, resulting in a slow transformation back to the 2H‐MoS2 phase and a concomitant decrease in HER activity. In contrast, amorphous MoS2 retains much of its high HER activity during prolonged operation.

Conflict of interest

The authors declare no conflict of interest.

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supplementary

Click here for additional data file.