Electrochemical proton insertion modulates the hydrogen evolution reaction on tungsten oxides

Hydrogen is an important industrial element and energy-dense fuel used in the production of metals and chemicals as well as fuel cells.¹ Ten million metric tons, representing 14% of the global hydrogen, are produced in the USA. The production of hydrogen with lower carbon emissions via electrolysis using electricity produced by nuclear or renewable energy could replace existing hydrogen production methods such as natural gas reforming. There are several technical hurdles to the widespread use of electrolysis for hydrogen production, including the need for low-cost, high activity, and durable electrocatalysts. The highest performing electrocatalysts for the hydrogen evolution reaction (HER) are platinum and the platinum group metals, but their material and processing costs hinder the widespread application in electrolyzers at a scale that would influence the hydrogen economy.^2,3 As a result, there are ongoing efforts to identify high performance, low-cost, and abundant electrocatalysts for the HER.³ Transition metal oxides, sulfides, and phosphides are of interest as HER catalysts due to the tunability of their crystal structure, morphology, and electronic properties through doping or incorporation of defects.^4–6 Mechanistic understanding of the HER on such materials can inform the material selection and accelerate the development of high-performance economical electrocatalysts.⁷ Moreover, mechanistic understanding of the HER on electrode surfaces helps in understanding more complex electrocatalytic reactions, such as those involving more than two electrons.

Given that many transition metal oxides and sulfides can undergo electrochemical proton and cation insertion, it is important to determine the role of this process in electrocatalysis. Electrochemical ion insertion can change the chemical, physical, and electronic structures of electrocatalysts and thus influence the electrochemical reaction mechanism. For example, Mefford et al. reported that the electrocatalytic activity of CoO_xH_y toward the oxygen evolution reaction (OER) was dependent on hydroxide ion (de)intercalation.⁸ Hydroxide ion intercalation led to oxidation of Co and higher activity. Intercalated ions can also increase the electrocatalytic activity of the layered transition metal sulfides toward the HER.^9–12 Gao et al. reported that titanates intercalated with 3d transition metals exhibit lower overpotentials for catalyzing the HER.¹² Strong electronic interactions between the 3d transition metals and the [TiO₆] layers led to more favorable hydrogen adsorption free energies (ΔG_H,ads) at surface oxygen sites. These reports show that the presence of ions in the layered electrocatalysts can strongly influence the electrocatalytic reactivity.

Herein, we used a materials chemistry approach to understand the role of electrochemical proton insertion on the HER in a transition metal oxide by synthesizing a series of layered tungsten oxides, including a hybrid organic/inorganic material with no cation insertion ability, and characterizing their HER activity. We selected the tungsten oxide hydrate (WO₃·xH₂O; x = 1, 2) system for its ability to undergo fast and reversible proton insertion¹³ and a semiconductor-to-metallic transition¹⁴ during electrochemical cycling, making the electronic properties favorable for electrocatalysis. The interlayer water molecules in WO₃·xH₂O can be replaced by organic pillars to modulate the capacity for proton and cation insertion. Our results show that a high overpotential is required to drive the HER when proton insertion is suppressed by the presence of organic pillars in the tungsten oxide. Conversely, reversible proton insertion occurs in WO₃·xH₂O, which is correlated with a lower overpotential for the HER. We discuss three possible reasons for the dependence of the HER activity on proton insertion in tungsten oxides: (1) increase in electronic conductivity, (2) the influence of inserted protons on ΔG_H,ads, and (3) participation of inserted protons in the HER mechanism.

A series of interlayer-functionalized tungsten oxides were used to understand the role of electrochemical proton insertion on the HER in acidic electrolytes. Acid-precipitated WO₃·2H₂O was used as a precursor for the synthesis of the interlayer-functionalized tungsten oxides.¹⁵ It was first dehydrated to obtain the layered orthorhombic WO₃·H₂O [Fig. 1(a)]. Stirring WO₃·H₂O in octylamine and heptane leads to a dissolution–reorganization reaction and the formation of the layered OA–WO₃ [Fig. 1(b)].¹⁶ Subsequent stirring of OA–WO₃ in an acidic solution leads to an exchange reaction with WO₃·2H₂O (exWO₃·2H₂O) as the end product.¹⁴ Scanning electron microscopy (SEM) images in Figs. 1(d)–1(f) show the morphology of each material. WO₃·H₂O consists of nanoscale platelets typical of the acid-precipitated synthesis route. OA–WO₃ is composed of much larger micron-scale sheets and ribbons, consistent with a dissolution–reorganization reaction mechanism. exWO₃·2H₂O maintains a similar micron-scale sheet morphology, consistent with an exchange reaction. The surface area of each material was obtained by applying the Brunauer–Emmett–Teller (BET) technique to N₂ adsorption isotherms (Fig. S1). The surface area decreased in the order of as-synthesized WO₃·H₂O (31.0 m²/g), OA–WO₃ (24.9 m²/g), and exWO₃·2H₂O (16.4 m²/g), following the morphology and particle size trend observed via SEM. The micron-scale sheet-like morphology is maintained between OA–WO₃ and exWO₃·2H₂O. This proved beneficial for the comparison of the HER activity of the two materials since the surface area and morphology can influence the electrocatalytic activity.

X-ray diffraction (XRD) patterns (Fig. 2) show the formation of a molecularly pillared, layered structure after completion of the dissolution–reorganization reaction (OA–WO₃). The OA pillars increased the interlayer spacing from 0.53 nm in WO₃·H₂O to 2.6 nm in OA–WO₃. The subsequent exchange of OA–WO₃ in 4M H₂SO₄ (exWO₃·2H₂O) leads to the formation of WO₃·2H₂O. The diffraction pattern of exWO₃·2H₂O matched that of WO₃·2H₂O prepared via acid precipitation, with both materials exhibiting an interlayer spacing of 0.69 nm.

We used Raman spectroscopy to distinguish differences in bonding between hydrated tungsten oxides (WO₃·H₂O and exWO₃·2H₂O) and OA–WO₃. Figure 3(a) shows the spectra for WO₃·H₂O, OA–WO₃, and exWO₃·2H₂O. The tungsten oxide hydrates display two characteristic peaks. The broad peak between 550 and 750 cm⁻¹ is assigned to the in-plane O–W–O stretching (ν) mode. WO₃·H₂O and exWO₃·2H₂O contain a sharp peak at ∼945 and ∼955 cm⁻¹, respectively, characteristic of the tungsten-to-terminal oxygen ν(W–O_t).¹⁷ Replacing the coordinated water in the interlayer with the more basic (i.e., more electron-donating) octylamine causes a shift of ν(W–O_t) to 885 cm⁻¹. Therefore, the resulting bond between the W center and terminal O is longer (i.e., weaker). Table S1 presents the W–O_t bond lengths in WO₃·H₂O, exWO₃·2H₂O, and OA–WO₃. OA–WO₃ also exhibits peaks at higher wavenumbers associated with octylamine. Peaks at 1430–1470 cm⁻¹ are characteristic of CH₂ scissoring and bending modes, and those between 2830 and 2970 cm⁻¹ are attributed to ν(C–H) from CH₂.

FTIR spectroscopy [Fig. 3(b)] corroborated the Raman results and provided further insight into the interlayer water and OA. The ν(O–W–O) and ν(W–O_t) vibrational modes were observed for WO₃·H₂O, OA–WO₃, and exWO₃·2H₂O. ν(O–H) and H₂O deformation/bending (δ) modes located at ∼3400 and 1620 cm⁻¹, respectively, were observed in WO₃·H₂O and exWO₃·2H₂O. OA–WO₃ did not contain interlayer water; therefore, corresponding O–H and H₂O peaks were not present. The presence of OA in OA–WO₃ was observed by the vibrational modes of NH₃⁺ and CH₂.^18–20 Broad peaks at 3000–3250 and 2500–3000 cm⁻¹ were assigned to ν(N–H) and asymmetrical and symmetrical ν(NH₃⁺), respectively. The δ(NH₃⁺) was at 1606 cm⁻¹ and the weak peak at ∼1150 cm⁻¹ was assigned to the ν(C–N). ν(C–H) from CH₂ in the alkyl chain was at 2957, 2925, and 2855 cm⁻¹. The peak at ∼1504 cm⁻¹ was attributed to CH₂ scissoring, and δ(CH₂) was observed at 1468 cm⁻¹.

XPS measurements of WO₃·H₂O, OA–WO₃, and exWO₃·xH₂O [Figs. (S2a)–(S2e)] were used to determine the surface chemical states of tungsten, oxygen, water, and OA. The survey spectrum of WO₃·H₂O shows the presence of W and O as well as adventitious carbon (Fig. S2a). Based on XPS peak fitting of the W 4f_5/2 and W 4f_7/2 peaks, tungsten is in the 6+ oxidation state.^21,22 The O 1s core spectrum indicates the presence of O–W and O–H₂ bonds, consistent with the hydrated structure. The survey spectrum of OA–WO₃ shows the presence of W, O, C, and N (Fig. S2a). Peak fitting of the W 4f core spectrum showed that W is also in the 6+ oxidation state in OA–WO₃. The O 1s core spectrum for OA–WO₃ shows one peak at 530.8 eV that corresponds to O–W binding. Two peaks were present in the N 1s core spectrum. The peak at 399.7 eV was attributed to OA in the form of an alkylamine (R–NH₂), and the peak at 401.6 eV was attributed to OA in the form of an alkylammonium (R–NH₃⁺).²³ Thermogravimetric analysis (TGA) (Fig. S3) shows that there are ∼2.6 mol of OA to 1 mol of WO₃ in OA–WO₃. Given that XPS shows tungsten is in the 6+ oxidations state, we hypothesize that most of the OA pillars are alkylammonium cations to compensate for the negative charge of the WO₄²⁻ inorganic layer. The survey scan of exWO₃·2H₂O showed that it contains W and O (Fig. S2a), and peak fitting of the W 4f and O 1s core spectra revealed the same chemical states as in WO₃·H₂O.

We first investigated the role of interlayer pillaring molecules on electrochemical cation insertion into tungsten oxides. We performed cyclic voltammetry in three different electrolytes to determine the insertion behavior of three cations: 1M H₂SO₄ (H⁺), 1M LiClO₄ in PC (Li⁺), and 0.1M TBAP in PC (TBA⁺). Electrochemical insertion of small monovalent cations (A = H⁺, Li⁺) into WO₃ and its hydrates occurs up to n = 1,¹³ 

Figure 4(a) shows the cyclic voltammograms (CVs; cycle 3) of WO₃·H₂O and OA–WO₃ in 1M H₂SO₄ at 10 mV s⁻¹. Proton insertion into WO₃·H₂O is reversible, with a nearly symmetrical current response and an anodic capacity of 20 mA h g⁻¹ (0.19 e⁻/W). Reversibility here refers to electrochemical proton insertion during cathodic polarization and proton de-insertion during anodic polarization (e.g., the Coulombic efficiency). Conversely, proton insertion is largely suppressed in OA–WO₃, with a capacity of only 1 mA h g⁻¹ (0.02 e⁻/W). Our prior results showed that proton insertion in WO₃·H₂O is favored at bridging oxygen sites within the corner-sharing WO₅(OH₂) network.¹³ The lack of proton insertion in OA–WO₃ suggests that access to these bridging sites is blocked by the OA.

Figure 4(b) shows a similar outcome for Li⁺ insertion. There is virtually no Li⁺ insertion in OA–WO₃ (0.008 e⁻/W), while WO₃·H₂O has the same capacity for Li⁺ as for H⁺ (20 mA h g⁻¹; 0.19 e⁻/W). Figure 4(c) shows that TBA⁺ insertion is restricted for both materials: the observed capacities for WO₃·H₂O and OA–WO₃ were 0.53 mA h g⁻¹ (0.005 e⁻/W) and 0.21 mA h g⁻¹ (0.004 e⁻/W), respectively. This suggests that charge storage is only due to surface adsorption processes. The restricted insertion of TBA⁺ in WO₃·H₂O is likely due to steric hindrance from the inorganic layers since its ionic diameter (0.83 nm)²⁴ is larger than the interlayer spacing of 0.53 nm. However, the interlayer spacing of OA–WO₃ (2.6 nm) should provide sufficient space for TBA⁺ insertion. An alternative explanation, in agreement with similar studies,²⁵ is that the interlayer OA pillar density is too high to allow the insertion of guest cations. Figure 4(d) summarizes the electrons stored per tungsten in each electrode when cycled in electrolytes containing three different size cations. These results indicate that the presence of interlayer OA molecules in tungsten oxides suppresses proton and cation insertion, whereas the presence of interlayer H₂O molecules does not restrict the insertion of small cations like H⁺ and Li⁺.

Continued electrochemical cycling of OA–WO₃ in 1M H₂SO₄ led to an increase in capacity with cycle number. Figure 5(a) shows CVs of OA–WO₃ cycled 250 times in 1M H₂SO₄ at 10 mV s⁻¹. The CV characteristics of the 250th cycle (symmetrical CV shape, insertion current over a broad potential range, and redox peaks centered at 0.27 V) are like those associated with proton insertion into WO₃·2H₂O. The capacity of OA–WO₃ increased from 0.02 to 0.18 e⁻/W from cycle 1 to cycle 250. Ex situ XRD [Fig. 5(b)] after 50 and 200 cyclic voltammetry cycles showed the disappearance of the (001) peak of OA–WO₃ at 3.9° 2θ and the emergence of a peak at 12.8° 2θ consistent with the (010) peak of WO₃·2H₂O. These structural changes indicate an electrochemically assisted transformation of OA–WO₃ to WO₃·2H₂O in the acidic electrolyte, similar to the chemical exchange process to obtain exWO₃·2H₂O.

Cyclic voltammetry was used to investigate the role of proton insertion on the HER activity of tungsten oxides by comparing OA–WO₃, OA–WO₃-200, exWO₃·2H₂O, WO₃·H₂O, and WO₃·2H₂O (Fig. 6). The reasons for comparing the behavior of these five materials are the following: The lack of proton insertion in OA–WO₃ and its transformation to the proton-inserting WO₃·2H₂O in acidic electrolyte provides a materials platform to understand the role of proton insertion on the HER activity of transition metal oxides. The HER activity of OA–WO₃ was compared to OA–WO₃ that underwent 200 electrochemical cycles (OA–WO₃-200) and chemically exchanged exWO₃·2H₂O. OA–WO₃-200 provides the same electrode film structure and similar surface area, with the main differences being the replacement of OA with interlayer water and the inherent decrease in interlayer spacing. exWO₃·2H₂O shares the same characteristics as OA–WO₃-200 but has better electrochemical cycling stability since the OA/H₂O exchange occurred prior to electrode fabrication. As-synthesized WO₃·H₂O and WO₃·2H₂O exhibit reversible proton insertion and have the same surface area and morphology. Finally, exWO₃·2H₂O and OA–WO₃-200 were compared to as-synthesized WO₃·H₂O and WO₃·2H₂O to determine whether hydrated materials were obtained from the exchange with OA retained the same electrochemical insertion and HER characteristics.

The hydrated materials (WO₃·2H₂O, WO₃·H₂O, and exWO₃·2H₂O) show similar CV shapes, with proton insertion over a broad potential range before the onset of the HER. WO₃·2H₂O and exWO₃·2H₂O have two similar redox peaks centered at −0.2 and 0.25 V vs reversible hydrogen electrode (RHE). WO₃·H₂O has one distinct set of redox peaks at −0.1 V as well as broad symmetric peaks between 0 and 0.25 V. These peaks are associated with structural transformations due to proton insertion.¹³ Ex situ Raman spectroscopy (Fig. S4) shows that all three of the hydrated materials return to their original structure after cycling, indicating reversibility of the proton insertion process. OA–WO₃ cycled 200 times between −0.2 and 0.8 V showed anodic peaks similar to the tungsten oxide hydrates. This is expected based on its transformation to the WO₃·2H₂O structure, corroborated by ex situ Raman spectra in Fig. S4. The O–W–O and W–O_t stretching modes aligned with the positions expected for WO₃·2H₂O. The anodic capacity of ∼0.13 e⁻/W is lower than expected; however, this may be underestimated due to possible loss of active material and electrical contact from volume change upon exchange of OA with water. OA–WO₃ undergoes an irreversible reduction event at approximately −0.45 V vs RHE, close to the onset of the HER. Ex situ Raman spectroscopy (Fig. S4) after five cyclic voltammetry cycles revealed a new peak at ∼960 cm⁻¹ (shifted to higher wavenumbers compared to the expected peaks for WO₃·xH₂O). The shifted W–O_t peak and lack of a broad peak associated with O–W–O stretching modes show that OA–WO₃ does not transform to a tungsten oxide hydrate. Furthermore, ex situ XRD showed no long-range order for OA–WO₃ cycled five times to an average cutoff potential of 0.764 V vs RHE (Fig. S5). The peak around 3.9° 2θ disappeared, and there was no evidence of crystalline hydrates. This result confirms that the reduction event at −0.45 V vs RHE causes an irreversible structural transformation in OA–WO₃. Although the structure is distorted during cycling, the electrocatalytic activity remains consistent and distinct from the hydrates. These results demonstrate that for all five materials, the HER does not occur without proton insertion. In particular, blocking of proton insertion (OA–WO₃) leads to the lowest HER activity.

To better highlight the different HER activities of the five materials, we subtracted the capacitive and proton insertion current from the CVs (method described in the supplementary material, Fig. S6); the residual current, attributed to the HER, is shown in Fig. 7(a). We compared the HER activity by determining the onset of the HER, defined as the overpotential (η) at 5 mA cm⁻² based on the geometric surface area. WO₃·H₂O had the lowest η value, 470 mV. WO₃·2H₂O and exWO₃·2H₂O showed slightly higher η of 502 and 485 mV, respectively. OA–WO₃ cycled 200 times between −0.2 and 0.8 V showed an η of 478 mV, similar to the tungsten oxide hydrates. OA–WO₃ shows the lowest HER activity with an η of 682 mV.

The decrease of the HER activity with OA–WO₃ compared to the tungsten oxide hydrates could arise from different mechanisms: (1) low electrochemically active surface area, (2) blocking of the electrochemical interface by OA molecules, or (3) the lack of proton insertion in OA–WO₃ which is necessary to “activate” the material for the HER. The fact that OA–WO₃ has a surface area and morphology similar to exWO₃·2H₂O and OA–WO₃-200 but has a lower HER activity makes mechanism (1) unlikely. With respect to mechanism (2), Malkhandi et al. reported that alkyl chains attached to the surface of an Fe electrode suppressed the HER by hindering electron transfer and decreasing the accessible surface area.²⁶ Although the outer surfaces of OA–WO₃ are likely coated with OA, we observed that long-term soaking or stirring of OA–WO₃ in acidic conditions leads to WO₃·2H₂O. We performed in situ Raman spectroscopy to determine how quickly this reaction occurs to expose electrochemically active surface sites in an acidic electrolyte (Fig. S7). Within 2.5 min, a peak emerged at ∼955 cm⁻¹ characteristic of WO₃·2H₂O. This rapid transformation of the near-surface structure indicates that surface and near-surface interlayer OA molecules do not hinder the accessible surface area or electron transfer. Furthermore, the lack of polarization observed in the CVs in Fig. 5(a) along with some proton insertion into the structure suggests that electron transport in the hybrid material is not limiting the electrocatalytic activity. Therefore, we expect that OA–WO₃ has sufficient electronic conductivity and a suitable surface composition in acidic conditions to catalyze the HER making mechanism (2) an unlikely reason for its low activity compared to the tungsten oxide hydrates.

With respect to mechanism (3), Figs. 4(a) and 6 show that the electrochemical response due to proton insertion differs significantly between OA–WO₃ and the tungsten oxide hydrates. Each material displays at least one reduction reaction associated with proton insertion; however, the reduction peak positions and proton storage capacities differ considerably between the materials. Figure 7(b) shows the relationship between η and the number of electrons/protons inserted per tungsten (calculated from the anodic current response) for OA–WO₃ and the tungsten oxide hydrates. More proton insertion decreases the overpotential for the HER. The relationship between insertion capacity and overpotential demonstrates that the HER is dependent on the electrochemical insertion characteristics in transition metal oxides.

There are several hypotheses for why proton insertion in WO₃·xH₂O contributes to improved HER activity: (1) proton insertion changes the electronic band structure of WO₃·xH₂O, (2) the presence of bulk protons can influence ΔG_H,ads at the surface sites, and (3) inserted protons may participate in the HER mechanism on WO₃·xH₂O. With respect to mechanism (1), proton insertion in tungsten oxides causes a semiconductor-to-metallic transition.^14,27 Electrons filling the conduction band states increase the electrical conductivity and may facilitate proton-coupled electron transfer during the HER. Additionally, WO₃·H₂O accommodates 0.24 more protons/electrons per W than WO₃·2H₂O and η is 32 mV lower [Fig. 7(b)]. This suggests that the higher proton insertion capacity and higher filling of conduction band states enhance the HER activity. Mechanism (2) must also be considered as a possibility, as ΔG_H,ads is one of the most common descriptors of a material’s activity toward the HER. Protons generally prefer to bind to more basic sites, and Miu et al. recently reported that hydrogen adsorption and proton insertion potentials are dependent on the proton concentration in H_xWO₃.²⁸ In this work, the material with a higher proton insertion capacity shows the lowest overpotential for the HER. Given that the HER is most active on surfaces with a ΔG_H,ads close to 0 eV, one could hypothesize that as the proton concentration in H_xWO₃·xH₂O increases, ΔG_H,ads approaches 0 eV and reduces the overpotential required to catalyze the HER. Mechanism (3), whereby adsorption and insertion of protons are coupled, suggests that inserted protons can also migrate to the surface to participate in the HER. Proton insertion into WO₃·xH₂O begins at 0.5 V vs RHE; thus, protons are present in the material prior to reaching the thermodynamic driving force required for the HER. We hypothesize that the high mobility of protons in WO₃·xH₂O could enable transport to the surface to participate in the HER. Based on this mechanism, the HER activity could be partially dependent on the transport kinetics of the inserted protons to electrocatalytically active sites, in addition to other factors such as electronic conductivity and ΔG_H,ads.

In this work, we used a materials chemistry approach to understand the role of proton insertion in transition metal oxides on the HER catalysis. We synthesized tungsten oxide hydrates and a molecularly pillared tungsten oxide (OA–WO₃) that exhibit different capacities for proton and cation insertion. We observed broad proton insertion peaks across a wide range of potentials in the tungsten oxide hydrates, while proton insertion did not occur until −0.45 V in OA–WO₃. The proton insertion capacity of OA–WO₃ (∼0.12 H⁺/W) was considerably lower than that of WO₃·2H₂O (0.3 H⁺/W) and WO₃·H₂O (0.54 H⁺/W). Lower proton insertion was correlated with a higher overpotential for the HER: 682 mV for OA–WO₃, whereas WO₃·2H₂O and WO₃·H₂O only required 502 and 470 mV, respectively. These results indicate that the HER activity of a transition metal oxide is dependent on the proton insertion capacity. We propose three possible mechanisms to explain this trend: (1) proton insertion changes the electronic band structure, (2) the presence of bulk protons can influence the ΔG_H,ads at the surface sites, and (3) inserted protons may participate in the HER mechanism on WO₃·xH₂O. Future work includes determining whether this trend also occurs for anhydrous WO₃, the most likely mechanism, and whether it is a general phenomenon occurring at transition metal oxides undergoing the HER in acidic electrolytes.

WO₃·H₂O was synthesized via an acid precipitation reaction.¹⁵ Briefly, 50 ml of 1M Na₂WO₄·2H₂O (99+% Acros Organics) was added dropwise to 450 ml of 4 N HCl (Certified ACS Plus, Fisher Chemical) while stirring at 300 rpm for ∼24 h. The precipitates were collected and washed with deionized water using vacuum filtration until the rinsed solution reached pH 6–7. The filtered WO₃·2H₂O was then dried at 60 °C overnight and ground into a fine powder. To obtain WO₃·H₂O, the WO₃·2H₂O powder was heated overnight at 120 °C.

OA–WO₃ was synthesized by a dissolution–reorganization reaction using WO₃·H₂O as the precursor.¹⁶ WO₃·H₂O was added to a solution of OA (99+%, Acros Organics) in heptane (99%, Acros Organics) with a molar ratio of 15 OA to 1 WO₃·H₂O and a volume ratio of 5 heptane to 1 octylamine independent of the size of the reaction. This mixture was stirred rigorously in a round-bottom flask for six days. The reaction completion was marked by a color change from yellow to white. After completion, the powder was collected via centrifugation and washed by redispersing the powder in ethanol and centrifuging three times. The resulting material was dried at 60 °C under vacuum and ground to form a fine powder.

Exchanged WO₃·2H₂O (exWO₃·2H₂O) was obtained by stirring OA–WO₃ in 4M H₂SO₄ (Certified ACS Plus, Fisher Chemical). The sulfuric acid was refreshed daily by centrifuging the powder and redispersing in a new solution until the powder turned yellow. exWO₃·2H₂O was collected via the same centrifugation procedure as OA–WO₃. The resulting material was dried at 60 °C overnight and ground to a fine powder.

SEM was performed using a field emission FEI Verios 460L microscope. Raman spectroscopy was performed using a WiTEC alpha 300R confocal Raman spectrometer with a laser wavelength of 532 nm and a 100× objective lens. FTIR spectroscopy was performed on a Thermo Scientific Nicolet 6700 spectrometer with the chamber purged with dry air. Spectra were collected in transmission mode with 200 scans between 4000 and 400 cm⁻¹. XPS was conducted with a SPECS FlexMod XPS instrument with Mg Kα excitation (1254 eV) and a hemispherical analyzer (PHOIBIS 150). The takeoff angle was normal to the surface. Charge neutralization was used for WO₃·H₂O and exWO₃·2H₂O. Spectra were calibrated to the C 1s peak of adventitious carbon at 285.0 eV. Peak fitting of XPS spectra was performed with CasaXPS software and all spectra were fitted with a Shirley background. TGA was performed on a Seiko Exstar TG/DTA6200 instrument. XRD measurements were conducted on a PANalytical Empyrean diffractometer in the Bragg–Brentano geometry with Cu Kα radiation (λ = 1.54 Å). Ex situ XRD was completed using slurry electrodes coated on glassy carbon substrates.

All electrodes for electrochemistry in acidic electrolytes were prepared by mixing the active material with acetylene black and a 5 wt. % solution of Nafion in ethanol. The masses of active material, acetylene black, and Nafion were controlled to obtain a 60:10:30 weight ratio, respectively. Electrodes of the tungsten oxide hydrates for non-aqueous electrochemistry were prepared by mixing WO₃·xH₂O in ethanol and casting 400 µg of material onto glassy carbon substrates. Slurries of OA–WO₃ were prepared by mixing 80 wt. % OA–WO₃, 10 wt. % acetylene black, and 10 wt. % Latex-PVDF using a 20 wt. % Latex-PVDF aqueous dispersion. Slurries were cast on glassy carbon plates or glassy carbon disk electrodes. Glassy carbon plates (2 × 1 cm²) were prepared by sonicating in detergent, rinsing with DI H₂O, sonicating in ethanol, and rinsing with DI H₂O. The substrates were dried at 60 °C and Kapton tape was used to expose only the slurry-coated area to the electrolyte. The glassy carbon substrates were plasma cleaned immediately before drop-casting (Harrick Plasma PDC-32 G). Disk electrodes were polished with 0.05 µm alumina slurries followed by rinsing and sonicating in DI H₂O and ethanol.

Electrochemical characterization was conducted in three-electrode cells using a potentiostat (Bio-Logic MPG2 and VMP3). All cyclic voltammetry measurements used a scan rate of 10 mV s⁻¹. Proton insertion and hydrogen evolution reactions were carried out in 1M H₂SO₄ (Certified ACS Plus, Fisher Chemical) with the working electrode (vide supra), a graphite rod counter electrode (Pine Research Instrumentation; unless specified otherwise), and a Ag/AgCl in 4M KCl reference electrode (Pine Research Instrumentation). See Fig. S8 of the supplementary material for electrochemical cycling of OA–WO₃ with a Pt counter electrode and comparison of Coulombic efficiency when using a graphite vs Pt counter electrode. The overpotential for HER was defined as the potential vs RHE at which the current reached 5 mA cm⁻², and all reported current densities were based on the geometric surface area. Potentials were converted to the reversible hydrogen electrode (RHE) scale using the following equation:

Non-aqueous electrochemistry was performed in an argon-filled glovebox (MBraun, <1 ppm O₂, H₂O). Li⁺ insertion was studied in a 1M LiClO₄ (electrochemical grade, Sigma-Aldrich) in the PC (Sigma-Aldrich) electrolyte with the working electrode (vide supra) and Li metal counter and reference electrodes (99.9%, Sigma-Aldrich). TBA⁺ insertion was studied in a 0.1M TBAP (Sigma-Aldrich) in the PC electrolyte with a working electrode (vide supra), a Pt coil counter electrode (99.99%, Pine Research Instrumentation), and a Ag pseudoreference electrode.

See the supplementary material for N₂ adsorption isotherms, W–O_t bond lengths, XPS and TGA results, ex situ Raman and XRD of OA–WO₃ after HER testing, discussion of irreversible transformation in OA–WO₃ during HER testing, CVs showing the capacitive and insertion current subtraction, in situ Raman spectra of OA–WO₃ in H₂SO₄, and CVs and Coulombic efficiency of OA–WO₃ cycled with graphite and Pt counter electrodes.

The authors would like to thank Dr. Fred Stevie for assisting with XPS measurements and Sarah Morgan for acquiring N₂ adsorption isotherms. This work was supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences, Early Career Research Program, under Award No. DE-SC0020234. This work was performed, in part, at the Analytical Instrumentation Facility (AIF), North Carolina State University, which is supported by the State of North Carolina and the National Science Foundation (Award No. ECCS-2025064). The AIF is a member of the North Carolina Research Triangle Nanotechnology Network (RTNN), a site in the National Nanotechnology Coordinated Infrastructure (NNCI).

The authors have no conflicts of interest to disclose.

The data that support the findings of this study are openly available in Zenodo at https://zenodo.org/record/5913859#.YgLj2OrMKUk.