Nadeesha L. Kothalawalaa,
Nipun Chandrasiri
a,
Manisha De Alwis Goonatilleke
a,
Keerthan R. Raoab,
Sujan Shrestha
c,
Udari S. Kodithuwakkua,
Irene Y. Kimd,
Chad Risko
ab,
Ambrose Seo
c,
Beth S. Guiton
a,
Fuqian Yange and
Doo Young Kim
*a
aDepartment of Chemistry, University of Kentucky, Lexington, Kentucky 40506-0055, USA
bCenter for Applied Energy Research, University of Kentucky, Lexington, Kentucky 40506, USA
cDepartment of Physics and Astronomy, University of Kentucky, Lexington, Kentucky 40506-0055, USA
dDepartment of Psychology, University of California, Los Angeles, California 90095, USA
eDepartment of Chemical and Materials Engineering, University of Kentucky, Lexington, Kentucky 40506, USA
First published on 9th June 2025
Transition metal dichalcogenide (TMD) nanostructures have emerged as promising electrocatalysts for the hydrogen evolution reaction (HER) as there is an increasing demand for cost-effective and sustainable hydrogen production. Despite significant progress, there is still a critical need for developing facile and green methods for synthesizing ultrathin TMD nanostructures. In this study, we introduce a green, top-down synthesis method to produce highly exfoliated WS2 nanosheets. The process combines the ultrasonication of bulk WS2 in a binary water–ethanol solvent with a solvothermal treatment. The resulting ultrathin WS2 nanosheets exhibit clean surfaces free of surface ligands and impurities, high crystallinity in the semiconducting hexagonal phase, and outstanding electrochemical activity for HER. Key performance metrics include a low onset potential of −0.32 V (vs. reversible hydrogen electrode (RHE)) at 10 mA cm−2 and a low Tafel slope of 160 mV dec−1 with a catalyst loading of 0.76 mg cm−2. The promising HER performance is attributed to (1) a high density of exposed edges and defects, (2) enhanced charge transport due to high crystallinity, and (3) clean surfaces enabling efficient interfacial electron transfer. Furthermore, operando Raman spectroscopy using a 3D-printed electrochemical cell identifies the catalytically active sites on WS2 nanosheets for HER. This work provides a green route to high-performance, low-dimensional electrocatalysts for sustainable hydrogen production.
The transition to clean energy and achieving net-zero carbon emissions are critical goals for modern society. Hydrogen, as a clean and versatile energy carrier, offers significant potential for energy storage, transport, distribution, and utilization. Among various production methods, electrocatalytic water splitting powered by renewable energy sources stands out as a sustainable and eco-friendly approach to hydrogen generation. While Pt-based catalysts are the most efficient for hydrogen evolution reaction (HER), their high cost and scarcity hinder a widespread use in economical hydrogen production. To address this limitation, ultrathin TMD nanosheets, such as MoS2 (ref. 11–13) and WS2,14,15 have emerged as promising alternatives, demonstrating notable HER activities when prepared via liquid phase exfoliation (LPE). In the LPE of WS2, an intercalation process transforms the semiconducting hexagonal phase into the metallic trigonal phase, enhancing electrocatalytic activity.14 Wang et al. reported that the HER activity of WS2 nanosheets originates from exposed edge sites, which promote efficient charge transfer.15
Despite extensive research, green, impurity-free, and scalable synthesis methods for TMD nanostructures have yet to be realized. Current synthetic approaches suffer from lengthy reaction times, high reaction temperatures, and labor-intensive purification steps. LPE offers the advantage of a simple and potentially scalable synthesis; however it typically relies on high-boiling-point organic solvents,16 surfactants or polymers,13,17 and reactive chemicals such as lithium intercalants.11 These practices often result in residual impurities in the final products, which can interfere with electrochemical processes18 and block catalytically active sites, limiting their performance.
Recent efforts have focused on developing cleaner LPE processes for TMD nanostructures that eliminate the use of intercalants or surfactants.18 In these processes, the solvent's surface energy must match that of the TMD to enable efficient breakdown and exfoliation.17,19–22 Trapped solvent molecules between the TMD layers induce a steric repulsion, weakening van der Waals interactions and facilitating exfoliation.23 However, the use of high-boiling-point organic solvents in LPE, such as N-methyl-2-pyrrolidone (NMP) and dimethylformamide (DMF), is not ideal due to their environmental toxicity, the difficulty of removing solvent residues, and the risk of impurity formation in the final product.24 Conversely, low-boiling-point solvents, which are also more environmentally friendly, have shown promise in achieving highly exfoliated layered nanostructures while minimizing these issues.22,25
In this work, we present a versatile and greener exfoliation method for WS2, a less explored material than MoS2 for HER. This approach is based on a modified LPE through ultrasonication and a solvothermal process in a water–ethanol solvent mixture. The method yields ultrathin, crystalline WS2 nanosheets with diameters of a few hundred nanometers while avoiding the need for long sonication times, high temperatures, surfactants, or intercalants. Characterization techniques, including Raman spectroscopy, atomic force microscopy (AFM), X-ray diffraction (XRD), and high-resolution transmission electron microscopy (HR-TEM), confirm the successful exfoliation of WS2 into mono- to few-layered nanosheets with high crystallinity and clean surfaces. Scanning transmission electron microscopy (STEM) reveals structural defects that may act as catalytically active sites, while X-ray photoelectron spectroscopy (XPS) and HR-TEM confirm that the predominance of the semiconducting prismatic (2H) phase. The resulting WS2 nanosheets demonstrate excellent electrochemical HER performance, comparable to or exceeding previous studies. Operando Raman spectro-electrochemical measurements affirm that sulfur surface sites are the catalytically active centers for HER, as indicated by Raman bands in the low-wavenumber region corresponding to S–H bending (812 cm−1) and wagging (722 cm−1) modes, further supported by periodic density functional theory (DFT) models.
Prior to exfoliation, the mixture of ethanol and DI water (30:
70 v/v) was degassed to remove dissolved oxygen. WS2 bulk powder (0.5 gram) was added to a Schlenk flask, and degassing was performed to remove oxygen. 100 mL of the degassed solvent was added to the Schlenk flask with WS2 powder. The mixture was subjected to ultra-probe sonication (Sonics & Materials, Model VCX500) at 55% amplitude for 1.5 hours (2 seconds on and 2 seconds off). 60–70 mL of the ultra-sonicated mixture was heated at 85 °C for 6 hours under an inert environment. The mixture was then centrifuged at 4000 rpm (relative centrifugal force or g-force: 1010) for 7 minutes to separate the less-exfoliated WS2 (intermediate product) from ultrathin WS2 nanosheets (final product). The supernatant was subjected further to ultra-centrifugation at 12
000 rpm (relative centrifugal force or g-force: 1304) to separate the highly exfoliated ultrathin WS2 nanosheets. The final product was washed three times with the fresh portion of the same solvent mixture to remove any impurities. The obtained ultrathin WS2 nano-sheets were dried in a vacuum oven at 60 °C for further analysis. The typical yields of highly exfoliated ultrathin WS2 nanosheets (final product) and less-exfoliated WS2 nanosheets (intermediate product) were 3% and 76%, respectively. The low yield of highly exfoliated (mono- and few-layered) WS2 nanosheets can be addressed by upgrading the exfoliation apparatus and equipment – for example, optimizing the probe size and vibration amplitude in the probe sonicator, as well as adjusting the solvent volume in the solvothermal step.
UV-Vis absorbance measurements were performed on a Thermo-Scientific Evolution 201 UV-Visible spectrophotometer. A dispersion of nanosheets in a 30% ethanol–water solvent mixture was used to measure the absorbance in the wavelength range of 200–1000 nm with 0.1 s integration time and 1 nm bandwidth.
Scanning electron microscopy (SEM) imaging was carried out with a field emission Hitachi S4300 microscope with an accelerating voltage of 10 kV for the ultrathin WS2 nanosheets. Transmission electron microscopy (TEM) images and energy dispersive X-ray spectroscopy (EDS) data were collected on an FEI Talos F200X TEM/STEM electron microscope at 200 kV equipped with a Super-X EDS detector. Drop casting was used to prepare the TEM samples. Briefly, small amount of ultrathin nanosheets were suspended in analytical-grade isopropyl alcohol and sonicated for about 20 minutes. About 10 μL of the suspension was drop-cast onto a 300 mesh Cu TEM grid coated with holey carbon.
A Nion Ultra STEM 100 (U100) aberration-corrected electron microscope operated at 100 kV was used to obtain atomic resolution high-angle annular dark field (HAADF) scanning TEM (STEM) images. The Ultra STEM is a 5th-order aberration-corrected microscope with a 0.1 nm maximum spatial resolution at 100 kV. The samples were prepared similarly to the TEM samples with further annealing at 160 °C for 10 hours in a high vacuum oven prior to STEM imaging.
High-resolution TEM/STEM images were processed using the fast Fourier transform (FFT). The diffraction spots generated in reciprocal space on the FFT patterns were selectively masked and transformed to real space, creating inverse FFT (IFFT) images from the masked frequencies. Image processing was conducted on Gatan Digital Micrograph® software. Crystal structural models of the prepared WS2 were generated using CIF files on CrystalMaker® software.
Electrochemical measurements were carried out with a CHI potentiostat (CH Instruments, model: 660D). Electrochemical HER tests were conducted using a RDE with the catalyst mounted as the working electrode, reversible hydrogen electrode (RHE) as the reference electrode, and a graphite rod as the counter electrode. Details of the RHE setup are provided in the ESI.† The electrolysis was carried out in 0.5 M H2SO4 at a rotating speed of 1600 rpm. Ultra-pure N2 was purged through the electrolyte for 45 minutes before electrolysis. The preconditioning of the catalyst was performed under cyclic voltammetry (CV) with 50 cycles at 0.02 V s−1 scan speed in a potential range from 0.1 to −1.0 V (v. RHE) at a rotating speed of 1600 rpm. Linear sweep voltammetry (LSV) was then performed with a scan rate of 20 mV s−1 in the same potential range. All LSV curves were iR corrected using the solution resistance (RΩ) measured by electrochemical impedance spectroscopy (EIS). HER stability measurements were performed by cycling the potential in the same range for 4 hours at a scan speed of 0.02 V s−1 at a rotating speed of 1600 rpm. Amperometry was also performed to determine the catalyst stability by applying a constant potential (−0.4 V vs. RHE) for 3 hours at a rotating speed of 1600 rpm.
The SPE was precleaned before depositing catalyst as follows. CV was carried out in a potential range from 2.5 V to −2.5 V with three segments at a scan rate of 0.1 V s−1 in 0.5 M H2SO4. This is an essential step to remove carbon cross-linking and to increase surface roughness of SPE. The electrode was then dried, and 6 μL of the 50 wt% WS2 catalyst ink (1 mg WS2 nanosheets + 1 mg carbon black + 20 μL Nafion + 1 mL DI water) was drop-casted onto the electrode. Drop casting was repeated three times to deposit 18 μL of the catalyst ink. Finally, the electrode was dried and inserted into the 3D printed operando Raman cell.
Prior to an operando Raman measurement, an external reservoir was filled with 0.5 M H2SO4, then purged with N2 for 45 minutes. Inlet and outlet tubes to the reservoir were then connected to the corresponding ones of the cell. The electrolyte was circulated through the cell with a peristaltic pump. Finally, chronoamperometric measurements were carried out at a potential of −0.4 V, −0.6 V, −0.8 V, −1.0 V and −1.2 V for 60 seconds while a Raman spectrum was recorded at each potential. Raman spectra were acquired with 532 nm laser excitation (0.8 mW power, 2 second exposure time, 30 exposures, 50 μm slit) and 10× objective.
Fig. 1a shows the Raman spectra of ultrathin WS2 nanosheets at several spots and bulk WS2 powder. For a better comparison, normalized spectra are displayed in the fingerprint region (Fig. 1b). There are two dominant peaks, El2g and A1g. The El2g mode at 351 cm−1 corresponds to the in-plane vibrations of S–W–S. The A1g peak at 418 cm−1 corresponds to the out-of-plane vibration of S–W–S. These two peaks and their positions are in a good agreement with the previous reports.36
Compared to bulk WS2, ultrathin nanosheets of WS2 show a slight red-shift in A1g peak. This red shift is attributed to the weaker interlayer interaction in the exfoliated nanosheets. The intensity ratio of El2g and A1g serves as an important parameter to indicate the thickness of WS2 nanosheets. The ratios of El2g and A1g intensities, , measured at different spots on WS2 nanosheets are listed in Table S1.† The ratios of exfoliated WS2 nanosheets (3.0–3.3) are significantly higher than that of the bulk (1.90), because of the weakening of van der Waals forces between layers. The ratio of >2 suggests that the exfoliated nanosheets are predominantly mono-layered.36–38 El2g and A1g peaks in the Raman spectra are characteristic of mono-layered hexagonal (1H) phase, confirming that semiconducting phase of bulk WS2 powder is retained after exfoliation.39 The characteristic peaks of metallic 1T-WS2 nanosheets in the low frequency regions are 131 cm−1, 188 cm−1, 258 cm−1 and 325 cm−1 (J1, J2, Ag, and J3 modes respectively).14,39 These peaks are not observed at any spots of our exfoliated sample due to the low concentration of 1T phase formed during our synthesis.37
Fig. 1c shows the AFM image of the WS2 nanosheets. The line–scan profiles across several nanosheets are shown in the inset. The line–scan profiles indicate that the nanosheets have an average thickness of 1.6 nm. It is known that the thickness of a WS2 monolayer is ∼1 nm.14 Hence, the ultrathin WS2 nanosheets on atomically flat mica are either mono-layered or bi-layered. The thicknesses and the diameters of WS2 nanosheets determined by AFM are summarized in Table S2.†
Powder XRD patterns of WS2 nanosheets and bulk WS2 are shown in Fig. 1d. While bulk WS2 displays all characteristic peaks of 2H phase, WS2 nanosheets exhibit sharp but simpler peaks. The sharp peaks indicate that WS2 nanosheets retain the crystallinity of hexagonal phase after exfoliation.22,40 It is noteworthy that all non-vertical planes are absent in the XRD pattern of ultrathin WS2 nanosheets.40–42 This implies that the exfoliation is successful and exfoliated WS2 nanosheets are primarily mono-layered.
Fig. 1e and f show the high-resolution XPS spectra for W4f and S2p, respectively. The deconvolution of W4f XPS peak reveals that the majority of WS2 nanosheets are 1H phase also with the minor 1T phase converted from 1H phase. Two W4f XPS peaks at 31.9 eV and 34.0 eV in the low binding energy region correspond to the 1T-phase W4f7/2 and W4f5/2 respectively, whereas the peaks at 32.46 eV and 34.61 eV correspond to the 1H-phase W4f7/2 and W4f5/2, respectively. These binding energies are in a good agreement with previous XPS studies of 1H WS2 and 1T WS2 nanostructures.14,37 The peak at 35.74 eV can be assigned to the oxidized W(W6+) 4f7/2 and originates from the minor oxidation of W sites during the synthesis. The high-resolution XPS spectrum of S shows two peaks at 162.1 eV and 163.4 eV for S2p3/2 and S2p1/2, respectively, in agreement with previous studies.14,37 The XPS survey spectra of the WS2 nanosheets shown in Fig. S1b† suggests that the samples are free of impurities. In the survey spectra, carbon and oxygen peaks are due to exposure to the atmosphere and from carbon tape used for XPS measurement. Elemental compositions and the binding energies of elements are summarized in Table S3.† The atomic ratio of W and S is close to 1:
2.
The UV-Vis absorbance spectrum of WS2 nanosheets is presented in Fig. S1a.† There are three well-resolved absorption bands at 640 nm (A peak), 530 nm (B peak), and 475 nm (C peak), due to the excitonic transitions.40,43 Hence, the intense and well-resolved absorption bands confirm a direct band gap in WS2 nanosheets due to the reduced number of layers. Further, the direct band-gap transition at 640 nm (1.94 eV) shows the formation of mono-layered WS2, consistent with previous reports.25,44
The morphology, shape, and dimension of WS2 nanosheets were determined by SEM as shown in Fig. 2a. The exfoliated WS2 nanosheets present both hexagonal and irregular shapes. The distribution of particle sizes of a total of 200 particles (Fig. 2b) yields an average size of 195 ± 55 nm, which is further confirmed by low-resolution TEM image in Fig. 2c. The hexagonal atomic arrangement of W atoms and the microstructure of the prominent 1H phase of WS2 in Fig. 2e show an interlayer spacing of 0.27 nm corresponding to (100) layers of WS2. The selected area electron diffraction (SAED) pattern in Fig. 2f confirms the single-crystalline nature of the nanosheets with the sheet perpendicular to the [001] direction of WS2. Energy-dispersive X-ray spectroscopic (EDS) analysis was performed to confirm the elemental composition of WS2 nanosheets. The EDS elemental maps and spectra are given in Fig. 2g and h, respectively. The homogeneous distribution of W (purple) and S (green) throughout the nanosheet is clearly shown, and the spectrum gives intense peaks at ∼1.77 keV and ∼2.30 keV, corresponding to W-Mα and S-Kα peaks, and other minor peaks of tungsten. The sharp peaks at ∼0.27 keV (C-Kα) and ∼8.04 keV (Cu-Kα) originate from the Cu sample grid with C overlay, used in the TEM analysis. The presence of a small amount of O at ∼0.52 keV was also detected.
Fig. 3 shows the atomic resolution HAADF-STEM micrographs of ultrathin WS2 nanosheets. A mono-layered nanosheet is shown in Fig. 3a, where the atomic arrangement is clearly visible. An enlarged inverse Fast-Fourier-transform (IFFT) image of the nanosheet is given in Fig. 3b which overlaps with the simulated structural model of WS2 in the 1H phase along [001] axis. The W and S atoms are indicated in purple and green colors, respectively. Fig. 3c is the intensity profile along the yellow arrow in Fig. 3b, distinguishing the difference between the intensities of W and S. The intensity profile indicates that W (Z = 74) is more intensely than S (Z = 16), confirming their positions in the structure.45,46 The inter-atomic distance of W atoms is 0.54 nm which matches well the simulated distance of 0.546 nm between W atoms on the (001) plane of WS2. It is noteworthy that the ultrathin WS2 nanosheets produced from this method result in a small percentage of the 1T phase of WS2 (Fig. S2a–c†) which is also revealed in the XPS data. In monolayer WS2, two S layers sandwich a W layer resulting in a slab with a thickness of 3.14 Å. The 1H phase exhibits a trigonal prismatic geometry around W with the stacking sequence of AbA (the upper-case and lower-case letters denote S and W, respectively), therefore placing the second S layer directly above the first. In contrast, the 1T phase shows an octahedral geometry with an AbC stacking sequence, with the W occupying the octahedral holes of S.6 Due to the intensity variation caused by the different stacking sequences of the two phases, Z contrast imaging can distinguish the areas containing the two phases.47 The comparison of atomic arrangement between 1H and 1T phases along with their respective structural models are given in ESI (Fig. S2d†).
STEM characterization has been further used to identify structural defects in ultrathin nanosheets that can be catalytically active for the HER. Three different types of defects visible on the WS2 nanosheets are given in Fig. 3d–f. A loss of a W atom resulting in a point defect is clearly seen in the HAADF-STEM image of a WS2 monolayer in Fig. 3d, and the defect site is highlighted in its IFFT in Fig. 3g. Most defect sites are present at the edges of the nanosheets as the atoms at surface edges contain dangling bonds and can be easily knocked off due to low displacement energy. Such defects can be seen in Fig. 3e (and its corresponding IFFT in Fig. 3h). Another type of defects observed are S vacancy lines, as shown in Fig. 3f. Since the displacement energy of S is low compared to W, they can be easily lost from the structure and eventually form vacancy lines. The loss of S rows from the structure results in the remaining W atomic columns contracting together as highlighted in the IFFT in Fig. 3i. This is confirmed with the intensity profile (inset of Fig. 3i) along the specified yellow arrow, with the size differences of the purple and pink double arrows indicating the interlayer distances. Increased number of S vacancy defects may eventually lead to a transition from semiconducting to conducting.48 It should be noted that these defects may have been created due to knock-on damage by the high energy of the electron beam.49 A previous study by Wu et al. demonstrated that the Tafel slope decreased as S-vacancy site density increased, suggesting their contributions to HER kinetics.50
Several previous studies reported electrochemical performances of doped and nanostructured WS2 nanosheets for HER.14,16,39,41,42,51–53 Here, the ultrathin WS2 nanosheets obtained through LPE exhibit excellent HER activities. Fig. 4a presents iR-corrected LSV curves of WS2 nanosheets in comparison with carbon black and commercial Pt/C (20 wt% Pt). To enhance electrical conductivity of ultrathin 1H-phase WS2 nanosheets, catalyst samples were prepared by mixing WS2 nanosheets with carbon black powder (WS2:
carbon black 10
:
90, 50
:
50, 80
:
20 w/w). All WS2 nanosheets showed better HER performances than carbon black. Among those, 80 wt% WS2 exhibits the best HER activity with a low onset potential, −0.32 V vs. RHE (corresponding to 10 mA cm−2, η10) and highest current density. Its best performance is attributed to the largest amount of WS2 with active sites available for HER.
![]() | ||
Fig. 4 (a) Polarization curves of three WS2 catalyst mixture (10 wt%, 50 wt%, 80 wt%), only carbon black, and commercial 20 wt% Pt/C (inset shows a better representation of the onset potentials), (b) Tafel plots for WS2, carbon black, and Pt/C with the calculated Tafel slopes for HER, (c) Nyquist plots of three catalyst mixture determined by EIS measurements (inset shows an equivalent circuit), and (d) electrochemical active surface areas estimated from the cyclic voltammograms at various scan rate (10–140 mV s−1, Fig. S3†). |
Tafel plots (iR-corrected) are presented in Fig. 4b. The overpotential (η10), and Tafel slope determined from the Tafel analysis, along with the catalyst loading (mg cm−2), are summarized in Table S4.† In this study, three WS2 catalyst mixtures present the similar Tafel slope of 163–168 mV decade−1, indicating HER occurs through the same mechanism. Similar Tafel slopes were reported for exfoliated 2H-WS2.50,54 The Tafel slope of ∼160 mV decade−1, characteristic of the Volmer mechanism indicate that the atomic hydrogen adsorption is the HER limiting step.55 Based on the Tafel slope and the low defect concentration observed in the high resolution STEM images, H2 generation in exfoliated WS2 is more likely to involve one adsorbed proton and a second proton from unadsorbed water, rather than two adsorbed protons. Therefore, the overall mechanism is likely to follow Volmer–Heyrovsky pathway. Tafel slopes of Pt/C and carbon black are 65 mV dec−1 and 457 mV dec−1, respectively. Table S5 in ESI† summarizes the overpotential (η10) and Tafel slope of 2H-WS2 in this study, in comparison with various WS2 catalysts reported in recent literature. Our 2H-WS2 catalyst exhibits a Tafel slope similar to those of other 2H-WS2 catalysts, with a lower overpotential. WS2 catalysts with phase conversion to 1T-phase, phosphorus incorporation, or strain engineering demonstrate significantly reduced Tafel slopes and overpotential.
Nyquist plots determined from EIS measurements are shown in Fig. 4c. The 80 wt% WS2 shows the best charge transfer resistance (Rct), suggesting that this catalyst contains largest concentration of active sites. The Nyquist plots were fitted with a simple circuit model and the fitting parameters, RΩ, Cdl, and Rct, for three catalyst mixtures are summarized in the ESI, Table S6.† We also compared the electrochemical surface area (ECSA) by measuring the double layer capacitance (Cdl) in Fig. 4d. Fig. S3† showed the cyclic voltammograms (CVs) of three WS2 catalyst mixtures taken with various scan rates (10–140 mV s−1). The 80 wt% WS2 catalyst shows the largest ECSA, suggesting that this catalyst has larger active surface area and more active sites for hydrogen evolution.
We further examined the long-term stability of WS2 nanosheets under HER. The long-term behavior was monitored in two ways: (1) continuous potential cycling for 4 hours, and (2) chronoamperometry at a fixed potential (−0.5 V vs. RHE) for 3 hours. Interestingly, different behaviors were observed. After continuous potential cycling, LSV showed a slight enhancement in the current (Fig. S4a†). Conversely, a gradual decay of HER current was observed at constant potential (Fig. S4b†).
To investigate the origin of activity decay in fixed-potential chronoamperometry, we analyzed WS2 catalysts by ex situ XPS at three different stages: (i) before electrolysis (Fig. S4c and S4d†), (ii) after electrolysis (Fig. S4e and S4f†), and (iii) after immersing the electrode in the electrolyte for 5 hours without electrolysis (Fig. S4g and S4h†). The comparison of the WOx 4f7/2 peaks in Fig. S4c and S4e† indicates the oxidation of WS2 during electrolysis. A peak at 169.6 eV in the S2p XPS spectra of all three samples is predominantly from sulfonate group of the Nafion binder, but it could also partially originate from WS2 oxidation.39 Ex situ XPS analysis suggests that electrochemical oxidation of WS2 is the primary cause of activity decay in chronoamperometry, consistent with the literature.39 Additionally, catalyst detachment and H2 bubble formation may contribute to the current decay.
Notably the current slightly increases after extended potential cycling (from 0 V to −1.2 V) (Fig. S4a†), because surface oxides are removed by negative potential.56 This preconditioning can activate the catalyst. Previous studies have reported that potential cycling can induce morphological changes in TMD nanosheets, leading to enhanced activity.56
To identify active sites and intermediate species for HER, we conducted operando Raman spectroscopy using a home-built, operando Raman cell and a screen-printed electrode (Fig. 5a–c). Fig. 5d shows operando Raman spectra of ultrathin WS2 nanosheets collected at applied potentials. The peaks at 351 cm−1 and 418 cm−1 correspond to the E12g and A1g vibrational modes of WS2, respectively. The additional peaks at 996 cm−1, 1054 cm−1 are due to the asymmetric stretching of SO42− electrolyte anions. The D, G and 2D bands appearing at 1351 cm−1, 1590 cm−1 and 2700 cm−1, respectively, are attributed to carbon black that is added to WS2 catalyst.
Importantly, the operando Raman studies reveal three peaks in the low-frequency region at 214 cm−1, 722 cm−1, and 812 cm−1, each of which evolve with applied potential. These three Raman bands are absent at open circuit potential. The zoomed-in Raman spectra in Fig. 5e show the progressive increase of the three Raman peaks with the applied potential. The voltage-dependent increase in Raman peaks is consistent with HER currents, as shown in Fig. S5a.†
There are a few reports on the similar low-frequency Raman bands assigned to H adsorption on the surface of TMDs. Afanasiev et al. and Wright et al. performed inelastic neutron scattering experiments and theoretical calculations for the gas-phase uptake of hydrogen on MoS2 and WS2,57,58 identifying low-frequency Raman peaks (in the range from 650 cm−1 to 800 cm−1) for S–H bending modes.57,58 Based on these reports, the most probable assignment for 722 cm−1 and 812 cm−1 peaks are the wagging and bending vibrations of S–H bond connected to W. The peak at 214 cm−1 can be attributed to another S–H wagging mode. This assignment is supported by previous theoretical work on H2S and H2O adsorption on Ge surfaces,59 which reported S–H wagging mode between 200 cm−1 and 300 cm−1.
To confirm that the three Raman bands originate from the adsorption of protons on WS2 nanosheets, we conducted a control experiment with carbon black alone. As shown in Fig. S5b,† carbon black alone did not exhibit these peaks. Fig. S5d† presents Raman spectra of WS2 nanosheets before and after the operando measurement without electrolyte, showing that both spectra are very similar and lack peaks around 214 cm−1, 722 cm−1 and 812 cm−1. This suggests that negligible photochemical damage occurred during the Raman measurements. We also confirmed that the operando Raman results are reversible (Fig. S5e†) and reproducible in repeated trials (Fig. S5f†).
Periodic DFT calculations were carried out to augment the findings from operando Raman studies. H atom adsorption models on the WS2 surface reveal a strong preference for the H atoms to occupy the S top site, as depicted in Fig. S6b and S6c.† The H–S bond distance is 1.43 Å with an angle of 53° with the nearest S atoms on the surface. Attempts to converge the H-atom adsorbed on to a W atom were unsuccessful, likely due to the stronger affinity of H–S bonding. The binding energy of an H atom on the surface of WS2 was determined to be −0.50 eV. In contrast, the binding energy of an H atom with a subsurface S-atom was determined to be −0.19 eV (Fig. S6d and S6e†). Normal mode analyses resulted in no imaginary modes, confirming the ground state structures. The charge density difference plot (Fig. S7†) shows the interaction between the adsorbed H atom and the S atom. For H atoms adsorbed on surface S atoms (Fig. 5f), bending (in-plane vibrations) and wagging modes (out of-plane vibrations) were identified at 626 cm−1 and 554 cm−1, respectively, closely matching the results of our operando Raman studies. All other modes corresponded to surface vibrations and frustrated translations.60 The Raman vibration mode at 214 cm−1 could not be distinguished through the DFT calculations due to the occurrence of surface vibrations and frustrated translations in the low frequency range. The red shift of the DFT-derived vibrational modes with respect to experiment could result from the calculations not being performed in an aqueous medium and other complex experimental conditions that are not considered in this computational model.
Due to their high crystallinity, large surface area, abundant edge sites/defects, and clean surfaces, the WS2 nanosheets exhibit outstanding HER performance. Notably, potential cycling activates the surface and enhances HER activity. Operando Raman spectroscopy provides valuable insights into the catalytic active sites of WS2 nanosheets by monitoring the S–H reaction intermediate. Periodic DFT calculations further corroborate the bending and wagging vibrational modes of the S–H intermediate observed in operando Raman measurements.
In summary, this green and versatile synthesis method for highly exfoliated TMD nanosheets establishes a strong foundation for efficient and sustainable hydrogen production. Future developments, such as doping and heterostructure formation, could further enhance their catalytic performance.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d5ra00712g |
This journal is © The Royal Society of Chemistry 2025 |