Experimental

SiO₂ (300 nm)/Si wafers (1–10 Ω-cm, <100> plane, DASOMRMS Co., Ltd) were cleaned with a standard piranha solution (3:1 volume ratio of H₂SO₄, 95% grade and H₂O_2, 30% grade) with conventional cleaning procedures followed by ultrasonication in acetone, isopropyl alcohol, and deionized (DI) water. The DI water was purified by water purification system (Human Power, Human Co., Ltd., Seoul, Republic of Korea, 18.3 MΩ-cm). The tungsten trioxide (WO₃, 99.995% purity, Sigma-Aldrich, 204781) thin films with various thicknesses (5, 10, and 20 nm) were deposited by using thermal evaporator (KVE-T2000, Korea Vacuum Tech Co., Ltd., Gimpo, Gyeonggi-do, Republic of Korea,). In order to synthesize the WS₂ thin films, we used the simple sulfurization process as following. The N₂ and H₂ (99.9999%, high-purity) gases were used for the sulfurization process in a thermal CVD system. First, the CVD furnace temperature was raised to 700 °C and was maintained at this temperature for 50 min under flow of H₂ and N₂ at 1 Torr. The flow rate of H₂ and N₂ was set at 40 and 200 cm³/min, respectively, by using mass flow controllers (MKS Co., Ltd., Seoul, Republic of Korea, Type 1479A). Then, sublimation of sulfur power (99.998%, Sigma-Aldrich, 213292) precursor was initiated in the other heating zone, whose temperature was set at 300 °C. This process was kept for 1 h for the successful sulfurization of WO₃ thin films. Once the sulfurization process was completed, the [poly(methyl methacrylate)] (PMMA, GPC 10000, Sigma-Aldrich, 81497) dissolved in chlorobenzene (99.5% purity, Sigma-Aldrich, 319996) with concentration of 2 g/50 mL was spin-coated onto the CVD-grown thin films. The PMMA/thin films/SiO₂/Si substrates were immersed in a buffered oxide etchant to separate the thin films from the substrate by etching away the SiO₂ layers. The obtained thin films were washed with DI water seven to nine times to remove the residual etchants and were transferred to the arbitrary substrates. After the PMMA/thin film membranes had perfectly adhered onto the substrates, the PMMA layers were removed using hot acetone bath at 50 °C for 1 h.

The PEC measurements (Ivium Technologies, Eindhoven, Netherlands, Model: Nstat) were carried out with a three-electrode system using a saturated calomel electrode as the reference electrode and a graphite rod as the counter electrode in a 0.5 M H₂SO₄ standard electrolyte solution. We used a graphite rod counter electrode (WonATech Co., Ltd., Seoul, Republic of Korea) to avoid the deposition of Pt on the working electrode. The three-electrode system was set up inside a quartz vessel, which protected the samples from UV absorption. The conversion equation of potentials versus RHE is expressed as:

(1) $$\eqalign{E_{{\rm RHE}} & = E_{{\rm SCE}} + E_{{\rm SCE}}^{\rm ^\circ} + 0.059 \times {\rm pH}\;{\rm (}E_{{\rm SCE}}^{\rm ^\circ} \cr & = 0.{\rm 242}\;{\rm versus}\;{\rm SHE)}}$$

where the E_SCE is potential of saturated calomel electrode, the E°_SCE is potential of the saturated calomel electrode with respect to the standard hydrogen potential (SHE), and the pH is 0.5 M H₂SO₄ is 0.27. A Xe arc lamp (Abet Technologies, Milford, CT, USA, LS150) was used and calibrated to an output of 100 mW/cm² (AM 1.5 G condition). A scan rate of 10 mV/s was used for the linear sweep with cathodic direction. The incident-photon-to-current conversion efficiency (ICPE) was measured with a light source and a monochromator (DONGWOO OPTRON, MAC150) and the efficiency was calculated by following equation:

(2) $$\eqalign{{\rm IPCE}\;{\rm (}\lambda {\rm )} & = EQE\; (\lambda ) \cr & = \displaystyle{{ \vert j_{{\rm ph}}({\rm mA/c}{\rm m}^{\rm 2}) \vert \times 1239.8\; ({\rm V} \times {\rm nm})} \over {P_{{\rm mono}}\;({\rm mW}/{\rm c}{\rm m}^{\rm 2}) \times \lambda \;({\rm nm})}}}$$

where the j _ph is photocurrent of the sample, the λ is wavelength, and the P _mono is the power of monochromator. The external quantum efficiency was measured with a standard Si photodiode. EIS was conducted by applying a constant potential of 0.27 V versus RHE near the open-circuit potential. The sweeping frequency was varied from 250 kHz to 1 Hz using a 10 mV AC dither.

Results and discussion

The WS₂ thin films used in this study were synthesized by simple sulfurization of WO₃ thin films. The WO₃ thin films with various thicknesses (5, 10, and 20 nm) were deposited on cleaned SiO₂ (300 nm)/Si substrates by using thermal evaporator. To synthesize the WS₂ thin films, the simple sulfurization of WO₃ thin films was conducted by using thermal CVD system with the sublimation of sulfur power precursors. For the characterization of the synthesized thin films and their application as PEC catalyst, they had to be transferred to substrates such as glass, p-Si, and SiO₂/Si wafers. The PMMA was used as the supporting polymer and was spin-coated onto the synthesized thin films for wet transfer. The photographic images of the synthesized thin films floated onto DI water bath and transferred on glass substrates are shown in Fig. 1(a). The yellow color of synthesized WS₂ thin films gradually changes to deep as a function of thickness of WO₃ precursor thin films. Figure 1(b) shows that the absorbance spectra of WS₂ thin films prepared with different thickness of WO₃ thin films exhibit similar peak positions for each thin film. The excitonic absorption peaks attributing from the transition of direct gap at the K point of the Brillouin zone. The corresponding peaks are located at the wavelengths of 612, 616, and 624 nm for the 8-, 23-, and 45-nm-thick WS₂ thin films, respectively.^[ Reference Mishra, Lakshmi and Huang ^{23 ]} The optical absorption peak change originates from the variation of optical band gaps of synthesized thin films. These results are in good agreement with previously reported absorption spectra of the WS₂ thin film via other CVD methods.^[ Reference Mishra, Lakshmi and Huang ^{23 ]}

Figure 1.

Photographic images and characterization of synthesized WS₂ thin films. (a) The photographic images of synthesized WS₂ thin films floated onto de-ionized water bath and transferred to glass substrates. (b) Absorbance spectra and transmittance of synthesized WS₂ thin films. The inset figure shows the transmittance of WS₂ thin films. (c) Raman spectra of synthesized WS₂ thin films. (d) XRD analysis of synthesized WS₂ thin films on glass substrates.

Raman spectroscopy is a general investigation method to detect the atomic vibrations in thin films, especially 2D layered materials. For a WS₂ thin film, the in-plane vibration (E¹ _2g) appears near 355 cm⁻¹, while the out-of-plane vibration (A_1g) appears near 418 cm⁻¹.^[ Reference Berkdemir, Gutiérrez, Botello-Méndez, Perea-López, Elías, Chia, Wang, Crespi, López-Urías, Charlier, Terrones and Terrones ^{24 ]} The Raman spectra of the synthesized WS₂ thin films from sulfurization of WO₃ thin films with different thicknesses are shown in Fig. 1(c). The difference in frequencies between those two Raman modes of WS₂ thin films depends on the number of layers and can be easily used as a thickness indicator.^[ Reference Li, Zhang, Yap, Tay, Edwin, Olivier and Baillargeat ^{25 ]} The frequency difference between E¹ _2g and A_1g modes was approximately 64 cm⁻¹, indicating that the synthesized WS₂ thin film is multilayer. The peak gap between E¹ _2g and A_lg gradually increased as a function of increase of WO₃ thin films precursor layer thickness. Figure 1(d) shows the x-ray diffraction (XRD) patterns of the WS₂ thin films prepared with different WO₃ precursor layer thicknesses. Because the XRD measurement of synthesized thin films was conducted on soda-lime glass substrate, the broad peak near 23° was obtained in all the samples. The small peak, which is WS₂ (002) plane, near 13° appeared in the synthesized WS₂ thin films, regardless of the WO₃ precursor layer thicknesses.^[ Reference Nguyen, Sohn, Oh, Jang and Kim ^{26 ]}

To evaluate the thickness and surface morphology of the synthesized thin films, atomic force microscopy was used as shown in Figs. 2(a)–2(c). The thickness of synthesized WS₂ thin films increased from 8 to 45 nm when the thickness of WO₃ precursor thin layer increased from 5 to 20 nm. The increase of film thickness in each sample is attributed to the structure modification from 3D monoclinic WO₃ (space group of P2_1/n) to 2D hexagonal WS₂ (space group P6₃/mmc). The weak van der Waals bonds in the WS₂ thin films make the films thicker than the original 3D WO₃ thin films. All the synthesized thin films were nano-granular. The serrated-surface of the synthesized thin films with different thickness provides a large surface area than the atomically flat surface of a single-crystalline WS₂ layer. The root-mean-square (RMS) roughness was 2.6 nm for the 8 nm-thick WS₂ thin films fabricated with the sulfurization from 5 nm-thick WO₃ film. The RMS roughness increased from 2.6 to 6.2 nm. These increased RMS roughness values could help enhancing the catalytic activity due to the increase of surface area of synthesized 23 nm-thick WS₂ thin films. Figures 2(d)–2(f) shows the high-resolution transmission electron microscopy images of synthesized WS₂ thin films. The periodic atomic arrangements of the WS₂ thin films are obtained and the selected area electron diffraction (SAED) patterns reveal a hexagonal structure with poly-crystalline. Furthermore, the pin-hole free WS₂ thin films, which can protect the dissolution of p-Si photocathode were successfully synthesized by using simple sulfurization method. These results indicate that the poly-crystalline WS₂ thin films were synthesized by sulfurization of WO₃ films. In order to investigate the chemical components and atomic ratios of the synthesized thin films, the x-ray photoemission spectroscopy (XPS) was conducted as shown in Figs. 2(g) and 2(h). For the WS₂ thin films with different thicknesses, the atomic ratio of W to S was constant at 65% to 35%, regardless of the WO₃ precursor thicknesses. Five peaks appear at 32.4, 34.6, 38, 162, and 163.3 eV in the synthesized WS₂ thin films, which are assigned to W 4f _7/2, W 4f _5/2, W 5p _3/2, S 2p _3/2, and S 2p _1/2, respectively. These results are well-consistent with the findings of previous reports on WS₂ crystals.^[ Reference Xu, Zhang, Lin, Zheng, Zhong, Xia, Li, Sophia, Fuhrer, Cheng and Bao ^{27 ]}

Figure 2.

(a)–(c) Atomic force microscopy images of synthesized WS₂ thin films by using sulfurization of different thickness of WO₃ thin films. (d)–(f) High-resolution transmission electron microscopy images of synthesized WS₂ thin films. The SAED pattern shows that the synthesized thin films were poly-crystalline. (g)–(h) The core level x-ray photoemission spectroscopy (W 4f and S 2p) analysis of synthesized WS₂ thin films.

The PEC measurement of the WS₂/p-Si heterostructure photocathodes was conducted using a 3-electrode cell with 0.5 M sulfuric acid (H₂SO₄) as the HER electrolyte.^[ Reference Nguyen, Choi, Jeon, Kwon, Jang and Kim ^{28 ]} We employed a saturated-calomel electrode as reference electrode and a graphite rod as counter electrode. Figure 3(a) displays linear-sweep voltammetry (LSV) measurements under a solar simulated light source (air mass 1.5 G). The dashed lines show the result of LSV measurement without light. We set the potential when the photocurrent density reaches −1 mA/cm² as the onset potential. The large negative onset potential of −0.35 V was observed for the bare p-Si photocathode. As mentioned in introduction, the bare p-Si requires an additional overpotential due to the carrier concentration and kinetics issue.^[ Reference Reece, Hamel, Sung, Jarvi, Esswein, Pijpers and Nocera ^{6 –} Reference Strandwitz, Turner-Evans, Tamboli, Chen, Atwater and Lewis ^{8 ]} The 23-nm-thick WS₂/p-Si photocathode exhibited the lowest onset potential of 0.2 V. For the 23-nm-thick WS₂/p-Si photocathode, the anodic shift of the onset potential was 0.63 V with regard to that of the bare p-Si. The PEC performances with photocurrent density at 0 V and potential at 20 mA/cm² were summarized in Table I. It is well-known that the edges of 2D TMD materials could act as catalytic active sites. The nano-serrated surface of the synthesized WS₂ thin films as displayed in Figs. 2(a)–2(c) should contain a number of active sites. The RMS roughness values of the 23- and 45-nm-thick WS₂ thin film shows the higher value (6.3 and 5.3 nm) than that of 8-nm-thick WS₂ thin film (2.6 nm), indicating that the thicker WS₂ thin films have the plenty of catalytic edge sites and a large surface area. For this reason, the photocurrent densities of the 23- and 45-nm-thick WS₂/p-Si photocathodes at 0 V (versus reversible hydrogen electrode (RHE)) showed the higher values of 8.3 and 8.2 mA/cm² than that of the 8-nm-thick WS₂/p-Si photocathode (3.2 mA/cm²). Meanwhile, the saturation photocurrent density of the 45-nm-thick WS₂ thin film/p-Si photocathode is lower than the other photocathodes due to its lower transmittance, as shown in the inset of Fig. 1(b). These results obviously display that the PEC performances of the WS₂/p-Si photocathodes rely on the thickness of the WS₂ thin film and that an optimum WS₂ thin-film thickness exists.

Figure 3.

Photoelectrochemical hydrogen evolution reaction performances of the WS₂ thin films on p-type Si photocathodes with different thickness. (a) Linear-sweep voltammetry curves of WS₂ thin film catalysts/p-Si heterojunction photocathodes. (b) Faradaic efficiency measurements for the 23-nm-thick WS₂/p-Si photocathode under illumination at 0 V versus RHE with 0.5 M H₂SO₄ electrolyte. The inset graph shows the amount of H₂ evolved as a function of the calculated amount of H₂ by the passed charge.

Table I.

The summarized PEC performances of WS₂/p-Si heterojunction photocathodes.

To elucidate the Faradaic efficiency for H₂ evolution of the WS₂ 23 nm/p-Si photocathode, we conducted gas chromatography measurements at constant voltage of 0 V versus RHE under continuous illumination. The H₂ evolution was monitored several times at intervals of 15 ~ 20 min. The inset graph of Fig. 3(b) displays the relation between the amount of evolved H₂ and the calculated amount of H₂ by the passed charge. Although the straight line does not exactly fit into the data point, but the Faradaic efficiency represents over 90% [Fig. 3(b)], which suggests that most of the measured current goes toward H₂ evolution. Figure 4(a) shows Nyquist plots of the EIS. The charge transfer resistances (R _ct) could be determined using a simplified equivalent circuit. The low R _ct at each interface appears in the small semicircular arc in the Nyquist plot. The semicircular arcs, which correspond to R _ct were considerably reduced as the thickness of WS₂ thin films increased from 8 to 23 nm. This decrease in the semicircular arc was affected by the enhancement of the thin film quality with the increase of WS₂ thin film thickness. The 23-nm-thick WS₂/p-Si photocathode exhibited the smallest R _ct value that is smaller than one order of magnitude compared with the other WS₂/p-Si photocathodes, as displayed in Table II. We evaluated the incident-photon-to-current conversion efficiency spectra of the WS₂/p-Si photocathodes at 0 V versus RHE as illustrated in Fig. 4(b). Among the WS₂/p-Si photocathodes, the highest efficiency over 60% was obtained by the 45-nm-thick WS₂/p-Si photocathode at long wavelengths (660–730 nm). This phenomenon originates from a decrease in the band gap of the WS₂ thin film as the thickness of the WS₂ thin film becomes thicker. Meanwhile, the areas under the IPCE curve of 23- and 45-nm-thick WS₂/p-Si photocathodes are almost similar, in accordance with the similar photocurrent densities at an applied bias of 0 V versus RHE.

Figure 4.

Investigation of photoelectrochemical hydrogen evolution reaction performances for the synthesized WS₂ thin films/p-Si heterojunction photocathodes. (a) Electrochemical impedance spectroscopy analysis of fabricated WS₂/p-Si photocathodes. (b) Incident-photon-to-current-efficiency curves of fabricated WS₂/p-Si photocathodes. (c) Chronoamperometric curves of 8-, 23-, and 45-nm-thick WS₂ thin films/p-Si photocathodes for 10 h. The 23-nm-thick WS₂ can maintain the 83% of initial photocurrent density even after 10 h.

Table II.

The summarized charge transport resistance values of various thicknesses WS₂/p-Si photocathodes.

TF, thin film; EL, electrolyte; p-Si/electrolyte.

We investigated the long-term stability of the WS₂/p-Si photocathodes using chronoamperometric measurements as shown in Fig. 4(c), to examine whether our WS₂ thin film catalysts could carry out a role as a passivation layer of p-Si. For the 23-nm-thick WS₂ thin film, the degradation of the photocurrent density at 0 V versus RHE was significantly suppressed, maintaining 83% of initial current for 10 h. Even though the 8-nm-thick WS₂ thin film catalyst shows the lowest catalytic activity among the three samples, the initial photocurrent density can be maintained at 0 V versus RHE over 10 h due to the relatively strong adhesion between the p-Si and its bottom layer. On the other hand, the severe photocurrent degradation can be observed in the 45-nm-thick WS₂ thin film catalyst. To check the surface morphologies of the as-transferred and the chronoamperometry measured samples, field-emission scanning electron microscopy was conducted for 8-, 23-, and 45-nm-thick WS₂ thin film catalysts, as shown in Fig. S1. For the as-transferred samples, the clean and flat surface without pin-holes was observed. However, the surface of the 45-nm-thick WS₂ thin film was significantly damaged by the catalytic reaction, while the 8- and 23-nm-thick WS₂ thin films maintained the flat surface without pin-holes. This result is attributed to the fact that the thicker 2D material has the weaker van der Waals force in its top layers. In this reason, the top layers of the 45-nm-thick WS₂ thin film could be easily exfoliated or damaged by the catalytic reaction.