Electrochemical Synthesis of a WO₃/MoS_x Heterostructured Bifunctional Catalyst for Efficient Overall Water Splitting

Abstract

Photo-/electrochemical water splitting can be a suitable method to produce “green” hydrogen and oxygen by utilizing renewable energy or even direct sunlight. In order to carry out photoelectrochemical (PEC) water splitting, a photoanode based on transition metal oxides, which absorbs photons and produces photoexcited electron–hole pairs, is needed. The positively charged holes can then participate in the water oxidation reaction. Meanwhile, a cathodic hydrogen evolution reaction (HER) can occur more efficiently with electrocatalytic materials that enhance the adsorption of H⁺, such as MoS₂. In this study, it was shown that WO₃/MoS_x heterostructured materials can be synthesized by an electrochemical method called plasma electrolytic oxidation (PEO). During this process, many micro-breakdowns of the oxide layer occur, causing ionization of the oxide and electrolyte. The ionized mixture then cools and solidifies, resulting in crystalline WO₃ with incorporated MoS_x. The surface and cross-sectional morphology were characterized by SEM-FIB, and the coatings could reach up to 3.48 μm thickness. Inclusion of MoS_x was confirmed by EDX as well as XPS. Synthesis conditions were found to have an influence on the band gap, with the lowest value being 2.38 eV. Scanning electrochemical microscopy was used to map the local HER activity and correlate the activity hotspots to MoS_x’s content and surface topography. The bifunctional catalyst based on a WO₃/MoS_x heterostructure was evaluated for PEC and HER water-splitting activities. As a photoanode, it could reach up to 6% photon conversion efficiency. For HER in acidic media, a Tafel slope of 42.6 mV·dec⁻¹ can be reached.

1. Introduction

In the near future, in order to drive certain industrial and technological processes, utilization of solar energy will become more widespread than ever. This is proven by economic outlook reports such as that issued by the International Energy Agency, Renewables 2022, which forecasts global solar photovoltaic capacity to triple over the period 2022–2027. Furthermore, solar energy is actively being investigated as a viable solution for such issues as CO₂ reduction and water splitting to produce hydrogen [1,2]. Hydrogen in particular is a promising fuel for the future as it has the largest gravimetric current density of all known substances (~120 kJ·g⁻¹) [3], and therefore could be used to power energy-intensive technologies. Admittedly, the use of liquid hydrogen as an energy carrier faces many fundamental problems, one of which is its production.

Nowadays, the production of hydrogen cannot be separated from the strive for climate neutrality. Although hydrogen can be produced in large quantities by steam methane reformation, it is classified as grey hydrogen due to the large carbon footprint of this process; this is in contrast to green hydrogen, which is synthesized using only renewable sources [4]. However, green hydrogen production processes (such as photoelectrochemical water splitting/electrolysis) are much less effective, thus the resulting hydrogen is more expensive and, as of yet, uneconomical [5].

The efficiency of solar light utilization in photoelectrochemical (PEC) processes is directly related to the band gap of the light absorbing material. The thermodynamic limit for the onset of water splitting is 1.23 eV, and the catalytic material should have a band gap close to this value. In reality, this is hardly achievable, but some transition metal oxides have a combination of properties that make them prime candidates for PEC catalysis. WO₃ is one of these materials, as it has a band gap of ~2.7 eV, which can be further lowered by tuning the synthesis conditions or creating heterostructures with other materials [6,7,8]. In addition, WO₃ is relatively stable under anodic conditions in acidic media, though it does undergo noticeable photo-corrosion in different electrolytes [9]. Photoactive WO₃ films have been prepared through various methods such as hydrothermal [10,11] or sol-gel synthesis [12], magnetron sputtering [13], spray/aerosol pyrolysis [14,15], as well as electrochemically on the cathode [16,17,18] or anode [19,20]. A modification of anodic synthesis is plasma electrolytic oxidation (PEO), which has been used to grow transition metal oxide coatings and heterostructures [21,22,23].

Another material that is often used for electrochemical water splitting is molybdenum sulfide (MoS₂) [24]. It is used as an electrocatalyst for hydrogen evolution reaction (HER)s in acidic media because of the favorable thermodynamics of H⁺ adsorption on the surface active sites [25]. The mechanism of HER on MoS₂ is identical to that on Pt. Thus, MoS₂ can be considered a non-precious metal electrocatalyst and an alternative to Pt [26]. It is also a semiconductor with a tunable band gap, averaging ~1.7 eV for 2H-MoS₂ [27]. MoS₂ has also been used in bifunctional catalysts to improve the HER activity of an oxygen evolution reaction (OER) catalyst [28]. Various methods have been developed to synthesize photoactive MoS₂ structures [29], the more prominent being chemical vapor deposition (CVD) [30,31,32], wet chemical synthesis [33,34], and electrochemical deposition [35,36,37,38].

New materials based on WO₃ and MoS₂ have been synthesized by methods such as sol-gel spin coating, probe sonication, or wet chemical synthesis, and were tested for electrochromic devices [39], as sensors [40] and photocatalysts [41].

In this study, the possibility of depositing WO₃ and MoS₂ through a one-stage process, namely PEO, is investigated. Although this method has been known for decades, it is regaining attention in the scientific community [42,43]. PEO is an excellent method to produce transition metal oxide films with unique properties. It also provides the possibility to incorporate ions into the film. Therefore, a new bifunctional WO₃/MoS_x heterostructured catalyst was synthesized and comprehensively evaluated. The perspectives of this novel heterostructure, in which WO₃ can act as a photoanode and the MoS₂ as a lower band gap semiconductor, and the HER electrocatalyst were also assessed.

2. Materials and Methods

2.1. Plasma Electrolytic Oxidation Procedure

Tungsten plates (99.5%, Alfa Aesar, Ward Hill, MA, USA) were used as working electrodes with a geometrical area of 2 cm². They were connected as the anode in a two-electrode cell, with a stainless steel coil as the counter electrode. The WO₃/MoS_x heterostructures were obtained from the following electrolytes: 1 M Na₂SO₄ (≥99%, Roth, Karlsruhe, Germany), 75 mM NaF (≥99%, Roth), 0.1 M NaH₂PO₂ (≥99%, Roth), and 25 mM of MoS₄²⁻. The last component was synthesized from (NH₄)₆Mo₇O₂₄·4H₂O (≥99%, Roth) and 60% Na₂S (Roth). WO₃ films were also obtained from the same bath, but without MoS₄²⁻ for comparison. The syntheses were carried out at room temperature (22 ± 3 °C) under galvanostatic conditions (Consort EV245, Cleaver Scientific, Rugby, Warwickshire, UK) power supply) by applying current of 50 mA. The solution was agitated with a magnetic stirrer. In order to evaluate the influence of the targeted properties, the duration of synthesis was 2, 5 and 30 min. The tags 2-min, 5-min, and 30-min WO₃ or WO₃/MoS_x were assigned, respectively.

2.2. Structural Characterization

An FEI Helios Nanolab 650 (Hillsboro, OR, USA) dual beam system with an energy dispersive X-ray (EDX) spectrometer INCA Energy 350 (Oxford Instruments, Abingdon, Oxfordshire, UK) and X-Max 20 mm² (Oxford Instruments, Abingdon, Oxfordshire, UK) detector was used to observe the surface morphology by scanning electron microscopy and EDX mapping. A Ga⁺ ion beam was used for focused ion beam (FIB) analysis. The XPS analyses were carried out with a Kratos Axis Supra spectrometer (Kratos Analytical Limited, Manchester, UK) using a monochromatic Al K(alpha) source (25 mA, 15 kV). Survey scan analyses were carried out on the area of 300 μm × 700 μm at a pass energy of 160 eV. High-resolution analyses were carried out also on the area of 300 μm × 700 μm, but at a pass energy of 20 eV. The XPS signal due to adventitious carbon located at 284.8 eV was used as a binding energy (BE) reference. An X-ray diffractometer (D2 Phaser, Bruker, Billerica, MA, USA, λ = 1.5418 Å/Cu Kα) was used to characterize the crystal structure of the obtained films.

2.3. Optical Properties

Diffuse reflectance spectra were measured in the range of 250–900 nm by means of an Ocean Optics HR4000 spectrometer (Ocean Insight, Orlando, FL, USA), Ocean Optics reflectance probe (angle 45°) and deuterium–halogen light source (DH-2000-BAL, Ocean Insight, Orlando, FL, USA). Diffuse reflectance R was converted to absorbance F using the following known Kubelka–Munc equation:

$$F=\frac{{\left(1-R\right)}^2}{2·R}$$

(1)

Band gap E_g was graphically calculated from the following equation:

$${\left(F·hv\right)}^{1/2}=A\left(hv-E_g\right)$$

(2)

2.4. Photo-/Electrochemical Characterization and Electrochemical Treatment

All photo- and electrochemical measurements were carried out by means of an Autolab 302 N (Metrohm, Utrecht, The Netherlands) potentiostat connected to an Autolab optical bench system. A 365 nm LED (Thorlabs, Newton, NJ, USA) was used for photoexcitation, with the light intensity calibrated to 25 mW cm⁻². The photoelectrochemical (PEC) properties of synthesized films were evaluated in a quartz three-electrode cell with a Ag/AgCl reference electrode and a stainless steel counter electrode. The HER characterization was carried out in a regular 3-electrode glass cell. An electrochemical reduction-activation treatment was carried out to prove the possibility of further improving the PEC properties of the heterostructures. Specific experimental steps are summarized in

Table 1. Outline of experiments used to characterize the PEC and HER properties of synthesized catalysts. Potentials were recalculated vs. RHE.

Step	PEC (0.1 M Na2SO4, pH 2)	HER (0.5 M H2SO4)
1	Settling of OCP for 60 s.	Settling of OCP for 60 s.
2	LSV scan from 0.72 V to 2.3 V at 2 mV·s−1, with 10 s ON/OFF pulses of 25 mW·cm−2 365 nm light.	LSV at 2 mV·s−1 from 0 V to a cut-off current density −100 mA·cm−2.
3	Potentiostatic illumination pulse at 1.53 V.	EIS measurements from −0.1 V to −0.5 V. Spectra obtained from 10 kHz to 100 mHz, perturbation amplitude ±10 mV.
4	Electrochemical reductive treatment of 30-min-WO3/MoSx	Fitting data with Zview software.
5	Repeat of steps 1–3 for r-30-min-WO3/MoSx

, and have been described in detail recently [44,45]. The potential values were recalculated vs. RHE, and are presented as such throughout the study, unless it is stated otherwise.

The relationship between incident light intensity (I₀) and photon conversion efficiency was also evaluated by measuring potentiostatic light pulses at 1.53 V, but under different I₀, from 5 mW cm⁻² to 30 mW cm⁻². Obtained photocurrent values were recalculated to incident photon conversion efficiencies (IPCE) by Equation (3):

$$IPCE (\%)=\frac{j_{ph}\left(\mathrm{m}\mathrm{A}{ \mathrm{cm}}^{-2}\right)\times 1240\left(\mathrm{V} \mathrm{nm}\right)}{I_0\left(\mathrm{m}\mathrm{W}{ \mathrm{cm}}^{-2}\right)\times \mathsf{\lambda }\left(\mathrm{nm}\right)}$$

(3)

2.5. Scanning Electrochemical Microscopy (SECM)

SECM measurements were carried out using a Versa SCAN (Ametek, Berwyn, PA, USA) workstation connected with Versta STAT 3 and Versa STAT 3F bipotentiostat setup (Princeton Applied Research, Oak Ridge, TN, USA). The mode of operation was substrate generation/tip collection (SG/TC), wherein the substrate (WO₃/MoS_x heterostructures) was a generator, and the tip (25 μm Pt) was a collector. The substrate was immersed in 0.5 M H₂SO₄ solution. Before measurement, the substrates were subjected to an equilibration treatment by applying a −0.3 mA·cm⁻² current for 20 min until a stable overpotential related to HER is reached. After suitable conditions were reached, the probe was lowered to ~15–25 μm of the substrate by a typical approach curve method. Then, the probe was polarized to 0 V (vs. Ag/AgCl), which had been determined to give a good hydrogen oxidation signal, and a 2D map was obtained by scanning the probe at a speed of 100 µm s⁻¹ and measuring data points at 25 μm increments.

3. Results and Discussion

3.1. Synthesis and Structural Characterization of Heterostructures

PEO can be characterized by measuring the change of voltage with time under galvanostatic conditions. Similarly to TiO₂ [46,47], the WO₃ formation process (Figure 1) undergoes three consecutive stages. At the beginning of oxidation, voltage increases almost linearly with time, and the conventional anodic oxidation of W occurs. The growth of the anodic WO₃ can be represented by the three-step process (Equations (4)–(6)), wherein metallic W is electrochemically dissolved and interacts with water from the solution to form WO₃ film [48]:

$$\mathrm{W}+2{\mathrm{H}}_2\mathrm{O}\to {\mathrm{WO}}_2^{2+}+4{\mathrm{H}}^{+}+6{\mathrm{e}}^{-}$$

(4)

$${\mathrm{WO}}_2^{2+}+2{\mathrm{H}}_2\mathrm{O}\to {\mathrm{W}}_2{\mathrm{O}}_5+2{\mathrm{H}}^{+}$$

(5)

$${\mathrm{W}}_2{\mathrm{O}}_5+{\mathrm{H}}_2\mathrm{O}\to 2{\mathrm{WO}}_3+2{\mathrm{H}}^{+}+2{\mathrm{e}}^{-}$$

(6)

In this stage, the voltage increases rapidly to maintain the applied current as the oxide layer grows, although some current is consumed for oxygen gas evolution. Notably, the obtained slopes of the V-t curves are very steep for both WO₃ and WO₃/MoS_x, and the value of ~65 V is reached within 0.5 s (~130 V s⁻¹). For comparison, the slopes observed for TiO₂ in various electrolytes were 24 to 35 V s⁻¹ [47]. It has been proposed that the voltage at the end of stage I is called the “breakdown voltage” (V_b), which in this case is approximately 65 V for both WO₃ and the heterostructure. This is much lower than that for TiO₂, but it is expected because WO₃ is a less dielectric material.

When V_b is reached, the second oxidation stage begins, in which the slope of V-t decreases and tiny electric sparks generate the micro-breakdowns in the film. Now, the total current is governed by two components: the ionic current and the electric current. Due to this reason, a lower voltage is needed to maintain the applied current. The voltage may still increase until V_c (critical voltage), where V_c = 79 V for WO₃ and V_c = 74 V for the WO₃/MoS_x heterostructure.

In stage III, the voltage values are almost constant and approach the final voltage (V_f). This is the case for WO₃, and it means that the film has attained a constant resistance. However, for WO₃/MoS_x, the voltage drops from a maximum V_c = 74 V to V_f = 63 V. Because a lower voltage is needed to maintain the applied current, the films become more conductive when MoS₂ is incorporated. Notably, ions from the electrolyte can be incorporated into the film during PEO because of complex chemical, electrochemical, and physical processes that occur during this process [49]. To generalize, electron avalanches during micro-breakdowns ionize both the oxide and electrolyte, and a mixture is formed and then solidifies again. However, in this particular case of MoS₂, the anodic formation of MoS₃ can also occur by disproportionation of MoS₄²⁻ [50,51], by reaction 7, as proposed in [52]:

$${\mathrm{MoS}}_4^{2-}\to {\mathrm{MoS}}_3+\frac{1}{8}{\mathrm{S}}_8+2{\mathrm{e}}^{-}$$

(7)

For further characterization, the heterostructured films were synthesized by capping PEO time to match the different stages: 2 min to reach V_b, 5 min when V_b is passed and value of V_c is reached, and 30 min for V_f. The surface and bulk morphologies of the obtained materials are shown in Figure 2, and are similar to those obtained for the plain WO₃ that had been formed by a similar method [45] The films synthesized for 2 and 5 min are disorderedly nanoporous, with surface pore diameters apparently ranging up to several hundred nanometers (Figure 2a,b). Such pores are also sometimes called “discharge channels”. In addition, it is evident from the cross-sections of the respective samples (Figure 2d,e) that the channels do not propagate vertically from the substrate, as would be expected in a typical anodic tungsten oxide structure. On the contrary, the cavities within the growing material seem to preferentially propagate horizontally. This is the result of the PEO mechanism of oxide layer growth. At certain locations, film breakdown occurs and the electron avalanche flowing between the metallic W substrate and the film–electrolyte interface ionizes the oxide with some amount of electrolyte and causes a vapor blowout, leaving behind a cavity and ionized plasma mixture. The plasma is then cooled by the electrolyte, hardening and forming a pore. It is also worth distinguishing the film, obtained after 30 min (Figure 2c,f), the morphology of which is much rougher, although it was shaped by the same fundamental processes, only for a much longer duration. The average thickness of the films is linearly dependent on synthesis time, and ranges from 0.56 μm (±0.13 μm) to 3.48 μm (±1.1 μm) as the synthesis duration is increased from 2 to 30 min.

The crystalline nature of the films was revealed by XRD. Typically, non-annealed WO₃ films that had been obtained by anodization are amorphous [53], but in this case, the diffractograms show that the peaks correspond to a monoclinic WO₃ structure, wherein the (002), (020), and (200) peak intensities have similar proportions for all coatings (Figure 3). The relative intensities of monoclinic WO₃ peaks are larger for thicker films, and distinct W substrate peaks are also detected for all samples. This is due to the X-ray penetration depth, because a stronger substrate signal and weaker oxide signal are seen for thinner films. All of the films, regardless of synthesis time, have a monoclinic structure. This is yet another effect of the PEO mechanism of oxide film growth; although no external heat is applied, the plasma caused by electric discharges produces enough energy to electrodeposit a crystalline material. Small peaks at the 2θ range of 50° to 60° that could not be attributed to WO₃ appear on the diffractogram of the thickest film. Peaks at such 2θ values have been observed for hexagonal and synthesized MoS₂ [54,55]. The weak signal is most likely related to the small amount of incorporated MoS_x and the amorphous-like structure of it.

In the structural characterization of these heterostructured materials, it is important to investigate how MoS₂ is dispersed throughout the WO₃ matrix. For this purpose, the films of a WO₃/MoS_x sample deposited for 30 min were examined by an EDX mapping technique, which was carried out for areas on an FIB channel cross-section (Figure 4a) and top-down surface (Figure 4b). A strong tungsten signal was obtained, and the quantity of W correlates with O, suggesting the existence of a tungsten trioxide. The presence of Mo and S was also confirmed, but the excitation energies for these materials overlap in EDX analysis, so no definite quantitative conclusions could be drawn. However, the maps show that the MoS₂ material is detected on the surface of the electrode in larger chunks of micrometer-order diameters. Cross-sectional EDX maps were also obtained for the thinner films (synthesized for 2 and 5 min), but no discernible Mo or S signal could be obtained.

Although the signal of Mo and S was not always strong enough for mapping, EDX data obtained for top-down areas of the samples show that Mo, S, and P are detected in the film. Up to ~1.5 % of P is detected in the films, and the amount does not vary with deposition time. P is incorporated into the film from H₂PO₂⁻, as plasma discharges ionize the oxide and electrolyte and later cool down to a solid material. As for Mo and S, both materials are detected in the elemental composition of the films, but not in any expected stoichiometries such as MoS₂ or MoS₃. In fact, the apparent composition of MoS_x is severely sulfur-deficient (e.g., MoS_0.4 for 5-min-WO₃/MoS_x). This probably occurs due to molybdenum sulfide oxidation to some degree under anodic deposition conditions.

XPS was used to check whether the MoS_x material is in fact comprised from Mo-S bonds. The core-level Mo3d XPS spectrum (Figure 5a) displayed the same characteristic peaks that had been observed for MoS₂ synthesized by other methods [56,57], particularly for cathodically electrodeposited MoS₂ [44]. The Mo3d region deconvolutes into several doublets that are related to Mo⁶⁺-O bonds (3d_5/2 232.7 eV; 3d_3/2 235.7 eV), Mo⁴⁺-S bonds (3d_5/2 229.3 eV; 3d_3/2 232.1 eV), and a Mo⁵⁺ signal, which may be related to a heptavalent oxide, but also to sulfur-deficient Mo/active sites (3d_5/2 230.9 eV; 3d_3/2 234.2 eV). Interestingly, the S2p core-level spectra (Figure 5b) deconvolute into two clear doublets, the existence of which can be attributed to terminal (S-term. S2p_3/2 161.8 eV; Sp_1/2 163.0 eV) and bridging (S-br. S2p_3/2 163.2 eV; S2p_1/2 164.2 eV) S bonds within the material [56]. Unlike for cathodically deposited MoS₂, a very strong signal of the bridging sulfide bond component is seen. This observation implies a somewhat different structure in the MoS_x deposited anodically. Similar behaviors were reported in the literature [58]. This also may have a pronounced effect on the material’s use as an HER electrocatalyst; terminal sulfur bonds may break during hydrogen evolution, leaving a sulfur-deficient Mo/an active site. A molybdenum sulfide structure comprising mostly bridging bonds may be less susceptible to corrosion, but is also less electrocatalytically active. The W4f core-level spectra of this film showed that the material is almost entirely W⁶⁺ oxide (99.52%), with a very small W⁵⁺ signal (0.48%) (Figure 5c).

3.2. Properties of WO₃/MoS_x Heterostructures

3.2.1. Optical Properties

The optical properties of the heterostructured and WO₃ films were measured by UV-visible light spectroscopy. Figure 6a shows the diffuse reflectance spectra obtained at wavelengths in the range of 300 to 800 nm. As expected, the absorption maximum for all films lies in the UVA range at ~360 nm. Moreover, the heterostructures have evidently decreased reflectance in comparison to 5-min-WO₃ film, which attests to the changes in the structure of the obtained novel materials. Furthermore, an increase in the synthesis time resulted in a red shift of the absorption edge. These films also have a fairly prominent absorption tail in the visible light region. It has been proposed that for WO₃-based materials, visible light absorption can be attributed to surface W⁵⁺ sites [59,60]; however, in this study, XPS showed almost no W⁵⁺ for 5-min-WO₃/MoS_x. Thus, visible light absorption may be an effect of compositing with MoS_x, or perhaps the existence of W⁵⁺ sites deeper within the film cavities into which XPS cannot penetrate. The band gaps also depend on synthesis conditions and have a clear tendency to decrease with increased synthesis time (Figure 6b). In particular, 2.75 eV, 2.49 eV, and 2.38 eV band gaps were calculated for the 2-min, 5-min, and 30 min films, respectively. The 5-min-WO₃ has a larger band gap (2.66 eV) than the heterostructure counterpart, which indicates alteration of optical properties through incorporation of MoS_x.

3.2.2. Photoelectrochemical Properties

The PEC activity curves obtained for WO₃ and WO₃/MoS_x films are shown in Figure 7a,c,e. For all films, starting from onset at ~0.8 V, the photocurrent increases with applied potential, which is expected as the space–charge layer expands. The photocurrent of the films that had been synthesized for 2 min is relatively similar, which may be explained by recalling Figure 1 and the related discussion. During this shortest duration of synthesis, the same voltage is reached, regardless of the presence of MoS₄²⁻ in the electrolyte. Consequently, very little MoS_x is incorporated into the film. For the 5 min WO₃ and WO₃/MoS_x films, an essential difference in their PEC properties is seen, illustrating the effect of incorporation of MoS_x into WO₃. The thickest films are unique; 30-min-WO₃ has very poor PEC activity, whereas the heterostructure shows significantly higher activity. The 30-min-WO₃/MoS_x film has a strong anodic background current component that can be inferred from the “UV OFF” parts of the LSV curve. The origin of such current is electrochemical rather photoelectrochemical, and may be related to the charging current of a large surface area. However, anodic oxidation of the incorporated MoS_x material is also possible.

Potentiostatic light pulse measurements were carried out in order to evaluate the steady-state photocurrent, and the data obtained at 1.53 V vs. RHE are presented in Figure 7b,d,f. The largest photocurrent is generated by the 5-min-WO₃/MoS_x sample is ~0.47 mA·cm⁻², representing a 2.6-fold improvement over 0.17 mA·cm⁻² for 5-min-WO₃. Here, it is also important to examine the initial photocurrent (j_t=0) versus the steady-state photocurrent, which is assumed as the stable current at the end of the 30 s pulse (j_t=30) minus the background current (j_ph = j_t=30 − j_bg). If j_t=0 is larger than j_t=30, we know that recombination of photogenerated holes and electrons occurs at the electrode–electrolyte interface, and this limits the photoelectrochemical current. In this case, no significant recombination is seen. In addition, for the 30-min-WO₃/MoS_x, sample j_t=0 is smaller than j_t=30, which shows slower photogenerated charge transfer kinetics.

Previously, it was demonstrated that cathodic electrochemical reduction/H⁺ intercalation into WO₃ films synthesized by the same method enhances their PEC properties [45]. This is also true for obtained WO₃/MoS_x heterostructures, and the biggest improvement in PEC properties was obtained for the thickest film (30-min-WO₃/MoS_x). The chronoamperometric curve registered during reduction shows that the current density settles at ~−4 mA·cm⁻², compared to ‒0.3 mA·cm⁻² for a plain WO₃ film (Figure 8a). Then, a potential of 1.53 V was applied and the electrode was illuminated, with 2 s OFF pauses every 60 s. (Figure 8b). The photocurrent, which is the difference between the dark current and total current, plateaus after ~2000 s. Then, the PEC characterization was repeated, and it is evident that the reduced film (r-30-min-WO₃/MoS_x) exhibits superior performance to the entire measured potential range (Figure 8c) and in steady-state pulse measurements (Figure 8d), estimated to be an increase of ~30%. This improvement can be attributed to the same process that had been characterized with XPS before, i.e., the reduction of W⁶⁺ to W⁵⁺ by intercalated protons, followed by oxygen vacancy restructuring towards more optimal levels during activation [45].

In order to establish a more comprehensive characterization of the PEC properties of these materials, the photocurrent response was evaluated in relation to the light intensity (I₀). According to the simplified Gärtner–Butler equation [61], the current generated by a photoanode will be a function of two variables: the width of the space charge layer and incident photon flux. The former is related to applied/bias potential and is reflected by potential sweep experiments, whereas the incident photon flux is directly proportional to I₀. As can be inferred from the Gärtner–Butler equation, provided charge carrier photogeneration and transfer is the rate-limiting step, then j_ph~I₀.

It is indeed observed that for all measured photoanodes that the photocurrent grows linearly from 5 mW·cm⁻² to 30 mW·cm⁻² for all films (Figure 9a). The unmodified WO₃ film generates the lowest photocurrents, which is also proven by results presented in Figure 7, whereas the heterostructures show improved PEC activity at all light intensities. IPCE, on the other hand, is inversely related to I₀, which means that photon conversion efficiency drops with increasing light intensity. This trend is observed for the measured photoanodes (Figure 9b), wherein the largest IPCE values are calculated at 5 mW·cm⁻², and the lowest at 30 mW·cm⁻². The best photon conversion efficiency at 1.53 V was obtained for the 5-min-WO₃/MoS_x film, at 6%. It is also evident that after reduction, the r-30-min-WO₃/MoS_x photoanode’s PEC properties improved significantly. The decrease in IPCE with I₀ is not as sharp for this film, which is a contrasting characteristic in comparison to the non-reduced films, in which IPCE drops with I₀ at a similar rate. The cause of this behavior is uncertain, but in practical terms, it means that higher light intensities can be used with fewer potential photon conversion efficiency losses.

3.2.3. Scanning Electrochemical Microscopy of HER

Electrochemically deposited MoS₂ is typically an excellent HER catalyst in acidic media. MoS_2-x can be electrodeposited either cathodically or anodically (see Equation (7)). However, during PEO, MoS_x is incorporated into the oxide film through plasma discharges, as discussed in Section 3.1. XPS has already shown that stoichiometric MoS₂ exists on the surface of the heterostructured films. It was therefore decided to characterize their local and total HER electrocatalytic properties.

Due to the random nature of plasma discharges in PEO, MoS_x should not coat the anode surface evenly. Previous experience has shown that scanning electrochemical microscopy (SECM) can be used to measure the local hydrogen evolution activity of MoS₂ on complex surface structures [62]. The substrate generation–tip collection (SG-TC) mode is used, wherein the substrate generates hydrogen by the Heyrovsky step (Equation (8)) and the probe oxidizes it (Equation (9)):

$$\mathrm{M}–\mathrm{H}+{\mathrm{H}}_3{\mathrm{O}}^{+}+{\mathrm{e}}^{-}\rightleftarrows {\mathrm{H}}_2+{\mathrm{H}}_2\mathrm{O}+\mathrm{M}$$

(8)

$${\mathrm{H}}_2\to 2{\mathrm{H}}^{+}+2{\mathrm{e}}^{-}$$

(9)

It was determined that mild galvanostatic conditions are ideal for substrate generation. At −0.3 mA·cm⁻², the substrate emits a constant signal, and little gaseous hydrogen is produced. Because at cathodic potentials, WO₃/MoS_x not only generates H₂ but also intercalates H⁺, the substrates were initially subjected to simple equilibration conditioning at −0.3 mA·cm⁻² for 1200 s.

Additionally, it was important to relate the current measured by the probe to the signal produced by the substrate. To accomplish this, the probe current (i_probe) was recorded for 100 s with the substrate at OCP (Figure 10). After 100 s, a current of −0.3 mA was applied on the WO₃/MoS_x substrate, as previously, which caused the potential of the substrate to drop and settle at ~−0.42 V vs. sat Ag/AgCl within 120 s. During this time, i_probe rose rapidly to ~0.4 nA, which confirms that the probe does indeed react to the signal produced by the substrate. It must also be noted that i_probe is not particularly stable. Several sharp drops are seen, which were most likely caused by the accumulation of gaseous H₂ to form bubbles and block the probe. However, these conditions were considered optimal for 2D mapping, and it was presumed that as the probe is mobile during a scan bubble, accumulation would be less likely.

The SECM mapping data are shown in Figure 11, which consists of 2D slices, as well as 2D and maps of the respective film. For 2-min-WO₃/MoS_x, no particularly features could be distinguished (Figure 11a–c). As expected from SEM and FIB observations, this seems to be the smoothest film in terms of surface roughnessand it contains the smallest amount of MoS_x. Even so, in the 2D profile, small bumps of current are observed at ~0.4–0.5 mm and 0.7 to 0.9 mm, but due to their linear continuity across the Y axis, these are more likely substrate topography-related signals. More interestingly, HER electrocatalytic activity maps are obtained for 5-min-WO₃/MoS_x (Figure 11d–f). In this case, significant inhomogeneity of HER activity is observed in the 1 × 1 mm map. Several islands of increased local activity emerge and occupy an area of ~0.2 × 0.2 mm. The 2D slices also show a more random distribution of HER activity compared to the previous film. This result certainly shows that areas of increased local HER activity exist on the surface of a WO₃/MoS_x heterostructure, but this may be related both to the presence of MoS_x as well as to the surface topography. Similar results were obtained for the 30-min-WO₃/MoS_x film (Figure 11g,h,j), showing comparable dimension spots of increased electrocatalytic activity on the surface. In this case, however, the difference between the smallest and largest measured currents is ~0.5 nA (for the 5-min-WO₃/MoS_x film it was 0.1 nA). If these results are cross-referenced with SEM and FIB observations, it may be inferred that this film, due to its larger roughness and MoS_x content, also has more inhomogeneity in local HER activity.

3.2.4. Total HER Activity

The HER electrocatalytic properties were also evaluated in terms of total electrode activity. The LSV curves shown in Figure 12a are corrected for iR-drop by the values of uncompensated resistance (often relating only to the resistance of the solution) obtained by means of EIS. These curves are typical of hydrogen’s evolution in acidic media. All plain WO₃ films showed identical HER activity regardless of synthesis duration. The same is also true of the heterostructures, except for the 2-min-WO₃/MoS_x film, which could be expected due to its comparatively low MoS_x content. Therefore, a comparison is drawn between a characteristic WO₃ and a 5-min-WO₃/MoS_x film. The WO₃ film has an η_10mA (overpotential at which 10 mA·cm⁻² HER current is reached) of −0.45 V, and a Tafel slope of 74.1 mA·dec⁻¹. The comparable WO₃/MoS_x exhibits significantly better electrocatalytic activity, having a η_10mA of −0.22 V and a Tafel slope of 42.6 mV·dec⁻¹. Both parameters are ~2 times better in terms of efficiency in comparison to the respective WO₃ films, and are typical of MoS₂-based materials in acidic media [25].

In addition, EIS was used to characterize the catalyst–solution interface at increasing overpotentials. When comparing WO₃ and heterostructured films at the same overpotential, it is apparent that the impedance of the WO₃/MoS_x film is smaller (Figure 12b) because the rate of HER is higher. Here, the spectrum in the Nyquist coordinates is represented by a low-frequency semicircle, which is related to the rate-limiting hydrogen adsorption process for HER on MoS₂ under such conditions [24].

By fitting the obtained impedance spectra to an equivalent electric circuit (EEC), the values of the discrete capacitive and resistive features of the catalyst–solution interface can be calculated. The used circuit is shown in the inset of Figure 12b, where R_s is the uncompensated resistance (solution resistance), C_dl is the double layer capacitance, R_ct is the charge transfer resistance, CPE_a is the constant phase element that models hydrogen adsorption related capacitance, and R_a is the hydrogen adsorption resistance.

This model comprises two RC components: the high frequency component (C_dl-R_ct) that is related to the Heyrovsky electrochemical recombination step, and the low frequency component (CPE_a-R_a) that simulates the slower Volmer hydrogen adsorption step. The resistance parameters are inversely related to the rate of their corresponding electrochemical reactions, and therefore their relation to the applied overpotential gives information on the kinetics of the hydrogen evolution reaction. For both WO₃ and WO₃/MoS_x, R_ct values (Figure 12c) are similar in the overpotential range, where little HER occurs (−0.1 V to ~−0.2 V). However, for WO₃/MoS_x, R_ct drops sharply from ~−0.2 V, coinciding with the onset of vigorous HER. At −0.4 V, R_ct approaches the lowest value of 0.11 Ω cm². The WO₃ film does not exhibit this behavior because the HER current density at −0.4 V was only approximately 1.5 mA·cm⁻². Moreover, C_dl is caused by the charge/discharge of the double layer at the catalyst–solution interface, and is therefore proportional to the electrochemically active surface area (ECSA). It was found that C_dl values obtained at −0.1 V for the 2-min, 5 min and 30 min films were 0.73 mF·cm⁻², 1.06 mF·cm⁻² and 1.53 mF·cm⁻², respectively, showing that longer synthesis durations resulted in a larger ECSA. The same ratio was retained at each applied overpotential. For comparison, C_dl of the 5-min-WO₃ film was 1.17 mF·cm⁻², which shows that the non-heterostructured film has a slightly larger electrochemically active surface area.

The hydrogen adsorption resistance R_a is even more closely linked to the HER activity of the catalysts. Because the rate of HER is limited by the Volmer hydrogen adsorption step, the measured current density and overall reaction rate are determined by the low-frequency adsorption-related parameters, especially R_a. The adsorption resistance decreases with increasingly negative overpotentials because H⁺ is more readily adsorbed (Figure 12d). However, it is apparent that on the heterostructured material, R_a drops more swiftly until ~−0.3 V and then plateaus. For WO₃, in contrast, R_a continues to decrease at −0.4 V. Experiments showed that at large enough overpotentials, the R_a values become equal, but for the heterostructure, lower R_a values are reached at lower overpotentials, which is expected of a heterogeneous HER electrocatalyst.

4. Conclusions

In this work, a bifunctional catalyst based on WO₃ and MoS₂ was prepared by relatively low-current plasma electrolytic oxidation of W in an electrolyte containing a MoS₄²⁻ precursor. SEM observations revealed a rough surface morphology comprising disordered pores of up to 500 nm diameter, depending on synthesis time. Film thickness was also directly related to the duration of synthesis, and ranged from 0.56 µm to 3.5 µm. The presence of non-stoichiometric MoS_x was confirmed by EDX and observed by elemental mapping only for the thickest film. XPS measurements showed a relatively strong Mo⁴⁺-S signal, suggesting that stoichiometric MoS₂ also exists in the film. Although the films were not annealed, due to the nature of PEO, they exhibited a degree of crystallinity, and the WO₃ was found to be monoclinic. The band gap ranged from 2.38 eV to 2.75 eV, depending on the duration of synthesis. This crystallinity and the incorporation of a lower band gap material (MoS₂) means that even the as-deposited films performed as rather efficient photoanodes for photoelectrochemical water splitting, and could reach an IPCE of ~6%. The effect of electrochemical reduction was also discussed and used to further improve the PEC activity. Due to the presence of MoS_x these materials can also act as HER electrocatalysts, and this was confirmed by measuring their HER activity in 0.5 M H₂SO₄. The best obtained Tafel slope (42.6 mV·dec⁻¹) can be compared to that of electrocatalytic MoS₂ films. Electrochemical impedance spectroscopy confirmed that the improved electrocatalytic HER activity is due to enhanced H⁺ adsorption, which is attributed to MoS₂. Scanning electrochemical microscopy was used to evaluate the local HER activity of the heterostructured films. It was found that differences in local activity can be discerned on a millimeter-level and could be related to the synthesis duration of the respective films (and thus to morphology and composition). Overall, the WO₃/MoS_x heterostructures combine the best photo/electrochemical properties of WO₃ and MoS₂ into one versatile smart material.