betaFeedback

WorksPhysics and Astronomy

High Proton Conductivity of HxWO3 at Intermediate Temperatures: Unlocking Its Application as a Mixed Ionic-Electronic Conductor

Abstract

Hydrogen tungsten bronzes (HxWO3), known for their mixed protonic-electronic conduction near room temperature, are extensively studied for electrochromic and gasochromic applications. However, their proton transport properties at elevated temperatures, particularly in the intermediate-temperature range (200-500 °C), remain unexplored. This study revealed the proton transport behavior of HxWO3, focusing on its potential as a proton-conducting mixed ionic-electronic conductor (MIEC) for intermediate-temperature electrochemical applications. By employing a proton-conducting phosphate glass as an electron-blocking electrode, we selectively measured the partial proton conductivity of sintered HxWO3. Hydrogen incorporation into the sintered WO3 pellet was found to occur preferentially near the surface, forming an approximately 500-micrometer-thick hydrogen-rich region. This region reached a composition of x = 0.24 and exhibited proton conductivity exceeding 10^-1 S/cm at 275 °C, well above those of state-of-the-art perovskite proton conductors. Impedance spectroscopy revealed distinct features of proton transport, including an isotope effect. The proton diffusion coefficient was 100-1000 times greater than that of H0.0001TiO2, which exhibits mixed protonic-electronic conduction via hydrogen dissolution. The larger proton diffusion coefficient of H0.24WO3 suggests that large polaron formation enhances proton mobility.These findings unlock new functionality of HxWO3 as a MIEC in the intermediate-temperature range, paving the way for the development of next-generation hydrogen energy conversion systems.

This manuscript is a preprint and is intended for publication elsewhere.

Introduction

Tungsten trioxide (WO3\mathrm{WO}_{3}) is a well-known transition metal oxide that exhibits pronounced optical and electronic changes upon hydrogen incorporation1–10. When exposed to hydrogen gas at elevated temperatures or subjected to electrochemical hydrogen injection using acidic electrolytes, WO3\mathrm{WO}_{3} forms hydrogen tungsten bronzes (HxWO3\mathrm{H}_{x}\mathrm{WO}_{3}). WO3\mathrm{WO}_{3} can accommodate a large amount of hydrogen (up to x∼0.5x \sim0.5)11, corresponding to proton density (nHn_{\mathrm{H}}) on the order of 1021 cm−310^{21}\ \mathrm{cm}^{-3}. Upon hydrogen dissolution, hydrogen atoms are incorporated into the lattice and ionize into protons and electrons, as described by Equation (1) in Kröger–Vink notation12:

12H2→Hi×→Hi∙+e′(1)\frac{1}{2}\mathrm{H}_{2} \rightarrow\mathrm{H}_{i}^{\times} \rightarrow\mathrm{H}_{i}^{\bullet} + e^{\prime} \tag*{(1)}

The electrons introduced along with protons localize on tungsten atoms, leading to the reduction of W6+\mathrm{W}^{6+} ions to W5+\mathrm{W}^{5+}. These changes result in a characteristic blue coloration and an increase in electronic conductivity. These coupled optical and electronic changes have been widely utilized in electrochromic13–15 and gasochromic devices16,17, hydrogen sensors18–20, and resistive switching devices that operate via ion injection21,22.

Since the optical and electronic responses in HxWO3\mathrm{H}_{x}\mathrm{WO}_{3} are associated with proton diffusion processes, HxWO3\mathrm{H}_{x}\mathrm{WO}_{3} is characterized as a proton-conducting mixed ionic–electronic conductor (MIEC), and the dynamics of proton transport have attracted significant research interest. The proton diffusion process has been studied using various techniques such as proton nuclear magnetic resonance (1H{}^{1}\mathrm{H} NMR)23–27 and electrochemical measurements28–30. Changes in optical absorption31–33 have been also utilized as experimentally accessible indicators of proton migration. These previous studies have revealed that the proton diffusion coefficient in HxWO3\mathrm{H}_{x}\mathrm{WO}_{3} strongly depends on crystallinity, hydrogen content, and measurement techniques and have consistently reported that proton diffusion in HxWO3\mathrm{H}_{x}\mathrm{WO}_{3} is significantly faster than that in many other oxides34–36. Despite these attractive properties, the practical applications of HxWO3\mathrm{H}_{x}\mathrm{WO}_{3} are limited to gasochromic and electrochromic devices operating near room temperature. Only a few reports have explored other uses of its mixed conduction37,38.

Owing to its intrinsic mixed protonic–electronic conductivity, HxWO3\mathrm{H}_{x}\mathrm{WO}_{3} is a promising candidate for electrochemical devices operating at elevated temperatures, such as protonic ceramic fuel cells and steam electrolysis systems39. In such devices, mixed protonic–electronic conductors are particularly advantageous because they allow electrochemical reactions to extend beyond narrow triple-phase boundaries (gas–electrolyte–electrode) to broader two-phase boundaries (gas–electrode)40. This can reduce electrode polarization resistance and improve the overall efficiency, which is especially beneficial for operation in the intermediate-temperature range (200–500 °C), where sluggish electrode kinetics often limit performance. Thus, demonstrating high proton conductivity of HxWO3\mathrm{H}_{x}\mathrm{WO}_{3} in the intermediate-temperature range would unlock the potential use of HxWO3\mathrm{H}_{x}\mathrm{WO}_{3} as a MIEC, contributing to the development of next-generation energy devices such as hydrogen permeable membranes and fuel cells.

This study investigates the proton conductivity of HxWO3\mathrm{H}_{x}\mathrm{WO}_{3} in the intermediate-temperature range using an electron-blocking method. A proton-conducting phosphate glass with a proton transport number of unity41 is employed as an electron-blocking electrode, enabling selective detection of protonic conduction in HxWO3\mathrm{H}_{x}\mathrm{WO}_{3} while suppressing electronic contributions. The partial proton conductivity of sintered HxWO3\mathrm{H}_{x}\mathrm{WO}_{3} is above 10−2 S cm−110^{-2}\ \mathrm{S\,cm^{-1}} at 250 °C, which significantly exceeds that of well-known perovskite-type proton conducting oxides. The large proton diffusion coefficient responsible for this high proton conductivity is discussed in terms of the thermodynamics of diffusion, particularly the contributions of activation enthalpy and activation entropy.

EXPERIMENTAL

Sample Preparation

The sintered WO3\mathrm{WO}_{3} samples were prepared by spark plasma sintering (SPS; SPS-511S, Fuji Electronic Industrial Co., Ltd., Japan). The raw yellowish green WO3\mathrm{WO}_{3} powder (99.9%, Kojundo Chemical Laboratory Co., Ltd., Japan) was ground using a mortar and pestle made of stabilized ZrO2\mathrm{ZrO}_{2}, put into a carbon mold with a diameter of 20 mm, and then sintered at 800 °C under 50 MPa for 10 min in vacuum. All sintered pellets were polished to a uniform thickness of 2.3 mm. The as-sintered sample had a black appearance, owing to oxygen deficiency caused by annealing under an atmosphere with low oxygen chemical potential. After subsequent annealing in pure oxygen at 600 °C for 40 h, the sinters turned back yellowish green, indicating that oxygen vacancies were filled with oxygen. The apparent density of the sinters, determined by the Archimedes method, was approximately 99% of the theoretical density. To promote hydrogen incorporation into the sintered WO3\mathrm{WO}_3 samples, a 100-nm-thick Pd film was deposited on both faces by magnetron sputtering (JFC-1600, JEOL, Japan), followed by annealing in a hydrogen atmosphere at 300 °C for up to 144 h. After hydrogen annealing, the surface of the sinter turned dark blue, indicating the formation of W5+\mathrm{W}^{5+} due to hydrogen incorporation, i.e., the formation of HxWO3\mathrm{H}_x\mathrm{WO}_3. A cross-sectional scanning electron microscopy image (JSM-7800F, JEOL, Japan) of HxWO3\mathrm{H}_x\mathrm{WO}_3 revealed a dense structure with no visible cracks or voids (Figure S1). Grain boundaries were too fine to be observed, suggesting a dense and homogeneous microstructure.

Cross-sectional SEM image of H$_x$WO$_3$

Figure S1. Cross-sectional SEM image of Hx_xWO3_3. The cross-sectional surface was prepared using a cross-section polisher after cutting the hydrogenated WO3_3 pellet. Needle-like features were observed on the cross-sectional surface as a result of ion-beam polishing. Grain boundaries were too fine to be resolved, suggesting a compact and homogeneous microstructure composed of very fine crystallites.

Evaluation of Hydrogen Incorporation into Sintered WO3\mathrm{WO}_3

The crystalline phase in the sintered HxWO3\mathrm{H}_x\mathrm{WO}_3 was identified by X-ray diffraction (XRD; SmartLab, Rigaku, Japan) using Cu-Kα\alpha radiation. Prior to measurement, the sintered pellet was mechanically polished to obtain a flat surface for accurate diffraction analysis. To evaluate the depth profile of hydrogen, a two-dimensional hydrogen concentration map was acquired by using time-of-flight secondary ion mass spectrometry (TOF-SIMS; TOF-SIMS5, ION-TOF GmbH, Germany) over a 500-μ\mum-wide region from the outer surface of the cross-sectional face of the HxWO3\mathrm{H}_x\mathrm{WO}_3 pellet. The amount of hydrogen dissolved in the sintered HxWO3\mathrm{H}_x\mathrm{WO}_3 was quantified by thermal desorption spectrometry (TDS) using a quadrupole mass spectrometer (QMASS; M-200QA, ANELVA, Japan). A small piece of the sintered sample (approximately 0.1 g), either oxygen- or hydrogen-annealed, was placed in a SiO2\mathrm{SiO}_2-glass crucible, which was then set in a closed-end alumina tube. The tube was placed in a furnace and connected to an analyzing chamber equipped with a QMASS. After the experimental apparatus, comprising the sample and analyzing chambers, was evacuated by using a turbomolecular pump (TMU-261, Pfeiffer Vacuum, Germany) down to <1×10−7 Pa<1 \times10^{-7}\ \mathrm{Pa} at room temperature, the furnace was heated up to 1000 ∘C1000\ ^\circ\mathrm{C} at a rate of 2.5 ∘C min−12.5\ ^\circ\mathrm{C}\ \mathrm{min}^{-1}. H2O\mathrm{H_2O} (m/z=18m/z = 18) and H2\mathrm{H_2} (m/z=2m/z = 2) gases released from the samples were detected by the QMASS and quantified using La(OH)3\mathrm{La(OH)_3} and TiH2\mathrm{TiH_2} as standard samples. Details of the TDS apparatus and experimental procedure have been described elsewhere 42.

Partial Proton Conductivity Measurements

The partial proton conductivity was measured by the electron-blocking method using the proton-conducting phosphate glass (36HO1/236\mathrm{HO}_{1/2}–4NbO5/24\mathrm{NbO}_{5/2}–2BaO2\mathrm{BaO}–4LaO3/24\mathrm{LaO}_{3/2}–4GeO24\mathrm{GeO_2}–1BO3/21\mathrm{BO}_{3/2}–49PO5/249\mathrm{PO}_{5/2} glass) with a thickness of 2020–30 μm30\ \mu\mathrm{m} and a proton transport number of unity 41 as the electron-blocking electrode. The junction between the sintered HxWO3\mathrm{H_xWO_3} and the glass was fabricated through hot-pressing, as previously reported 43,44. To reduce interfacial resistance across the junction between the sintered HxWO3\mathrm{H_xWO_3} and the glass, a 100-nm-thick Pd buffer layer deposited by rf-magnetron sputtering (1″ sputter cathode, Kenix, Japan) was inserted in the junction. Next, 100-nm-thick Pd reversible electrodes for protons and electrons were deposited on both ends of the cell by rf-magnetron sputtering. The conductivity was measured by using both DC and AC-impedance methods (VersaSTAT-3F, Princeton Applied Research, USA) at 200200–275 ∘C275\ ^\circ\mathrm{C} under H2\mathrm{H_2} or deuterium (D2\mathrm{D_2}) atmosphere. The applied voltage for the DC measurement was 100 mV. For AC-impedance measurements, the frequency range varied from 100 mHz to 1 MHz with an amplitude of 10 mV.

Results and Discussion

Hydrogen Incorporation into Sintered WO3\mathrm{WO_3}

Figure 1(a) presents the XRD patterns obtained from the surface and core regions of the WO3\mathrm{WO_3} pellet annealed in H2\mathrm{H_2} for 144 h. The inset shows a cross-sectional image of the pellet, with the measured regions corresponding to the XRD data. The cross-section reveals a sharp transition from the dark blue surface to a pale blue core. The diffraction peaks from the surface and core are indexed to the tetragonal and monoclinic phases of WO3\mathrm{WO}_3, respectively45,46. The blue coloration deepens as the hydrogen content increases4, and its crystal structure varies accordingly—monoclinic for x≲0.1x \lesssim0.1, orthorhombic for 0.1≲x≲0.150.1 \lesssim x \lesssim0.15, and tetragonal for 0.15≲x≲0.50.15 \lesssim x \lesssim0.547. These results suggest that the outer surface of the pellet is hydrogen-rich, while the inner core is hydrogen-poor.

XRD patterns and depth profile of hydrogen-related ions

Figure 1. (a) XRD patterns of the surface (red) and core regions (blue) of the HxWO3\mathrm{H}_{x}\mathrm{WO}_{3} pellet. The inset shows a cross-sectional optical microscope image of the pellet, revealing a sharp color transition from the dark blue surface to the pale blue core. (b) Depth profile of hydrogen-related ions (H−\mathrm{H}^{-} and OH−\mathrm{OH}^{-}) obtained by TOF-SIMS near the phase boundary in the cross section of HxWO3\mathrm{H}_{x}\mathrm{WO}_{3}.

Figure 1(b) displays the hydrogen depth profile of the sinter annealed in H2\mathrm{H}_2 for 144 h, obtained by lateral averaging a two-dimensional hydrogen distribution map measured by using TOF-SIMS. The hydrogen signal is less detectable at the edge of the pellet, but a hydrogen-rich region is clearly observed near the surface. The concentration gradually begins to decrease around a depth of 500 μ\mum, corresponding to the color boundary, and reaches a lower constant level toward the core. The hydrogen concentration is nearly constant within both the hydrogen-rich surface and the hydrogen-poor core, with no apparent internal gradients observed in the respective regions. Further details on the TOF-SIMS measurement are provided in Figure S2.

Depth profiles and lateral SIMS images of H⁻, OH⁻, and W⁻

Figure S2. (a) Depth profiles of H−^{-} and OH−^{-} signals normalized to the W−^{-} signal for reference WO3_3 and Hx_xWO3_3 pellets. (b) Lateral SIMS images of H−^{-}, OH−^{-}, and W−^{-} in the vicinity of the phase boundary in Hx_xWO3_3.

The hydrogen-rich surface and hydrogen-poor core regions were separated through mechanical grinding, and the hydrogen content in each region was determined by TDS. The hydrogen content was determined to be x=0.24x = 0.24 in the hydrogen-rich surface and x=0.0048x = 0.0048 in the hydrogen-poor core (Figure S3). The presence of two regions with distinct hydrogen concentrations, separated by an intermediate region exhibiting a concentration gradient in the depth direction, suggests that the proton diffusion coefficient varies across regions with different hydrogen concentrations. According to Fick’s law (Equation (2)), steeper concentration gradients correspond to lower local diffusion coefficients:

TDS curves of $m/z=2$ and $m/z=18$ for surface, core, and WO$_3$

Figure S3. TDS curves of m/z=2m/z=2 (H2_2) and 18 (H2_2O) for the hydrogen-rich surface (red) and hydrogen-poor core (blue) regions of Hx_xWO3_3 along with WO3_3 pellet annealed in oxygen at 800 °C as a reference (green). The release of hydrogen from Hx_xWO3_3 in the form of not only H2_2 but also H2_2O is likely due to heating under vacuum conditions.

J=−D(c)dc(x)dx(2)J = -D(c)\frac{dc(x)}{dx} \tag*{(2)}

where JJ is the proton flux, D(c)D(c) is the proton diffusion coefficient as a function of the local proton concentration cc, and dc(x)/dxdc(x)/dx represents the hydrogen concentration gradient along the depth coordinate xx. Similar behavior has also been reported in studies on the gasochromic properties of HxWO3\mathrm{H}_x\mathrm{WO}_3 thin films, with a drop in the diffusion coefficient at the monoclinic/orthorhombic–tetragonal phase boundary33. The present study results and previous observations regarding HxWO3\mathrm{H}_{x}\mathrm{WO}_{3} thin films suggest that structural changes induced by hydrogen incorporation suppress the inward extension of the hydrogen-rich region. As shown in Figure S4, the hydrogen-rich region gradually extended inward with increasing annealing time. Even after 144 h of hydrogen annealing, the hydrogen-rich region extended only to approximately 500 μm\mu\mathrm{m} from the pellet surface. Although the detailed mechanism of this behavior is beyond the scope of this study, it is worth noting that, at the phase boundary, proton diffusion must be accompanied by the tilt of WO6\mathrm{WO}_{6} octahedra necessary for the monoclinic/orthorhombic-to-tetragonal phase transition46. Such coupling between proton migration and lattice deformation may result in reduced proton mobility, which explains the lower diffusion coefficient in the phase boundary48.

Figure S4. Thickness variation of the hydrogen-rich region as a function of annealing time. The green curve represents a fitting result. The sintered WO3\mathrm{WO}_{3} samples were annealed in a hydrogen atmosphere at 300 °C for varying durations. The thickness of the hydrogen-rich region, measured from the Pd-deposited surface, gradually increases with annealing time.

Proton Conductivity of HxWO3\mathrm{H}_x\mathrm{WO}_3 in Hydrogen Atmosphere

Figure 2(a) shows a schematic illustration of the electron-blocking cell, together with a top-view photograph of the sample. The phosphate glass was tightly laminated onto the surface of the sintered HxWO3\mathrm{H}_x\mathrm{WO}_3, forming a well-defined interface that allows reliable characterization of proton conduction under electron-blocking conditions (Figure S5). Figure 2(b) shows the current–time profile measured under the application of a constant DC voltage of 100 mV as the atmosphere was switched between H2\mathrm{H}_2 and D2\mathrm{D}_2. The increase and decrease in current upon switching between H2\mathrm{H}_2 and D2\mathrm{D}_2 clearly demonstrate an isotope effect, suggesting that protons contribute to the current and that electronic conduction is effectively suppressed by the phosphate glass.

Schematic image of the electron-blocking cell and time evolution of the DC current

Figure 2. (a) Schematic image of the electron-blocking cell along with a top-view photograph. (b) Time evolution of the DC current (under a 100-mV applied bias) during the sequential switching of the atmosphere from H2\mathrm{H}_2 to D2\mathrm{D}_2 and back to H2\mathrm{H}_2 at 250 °C.

Cross-sectional SEM image and corresponding elemental maps

Figure S5. Cross-sectional SEM image of the electron-blocking cell and the corresponding elemental maps for P Kα\alpha and W Mα\alpha obtained by energy-dispersive X-ray spectroscopy (EDX). The elemental distributions confirm the presence of the phosphate glass electrolyte and the Hx_xWO3_3 layer. No delamination or gap is observed at the interface, indicating that the phosphate glass and Hx_xWO3_3 are well adhered.

Figure 3(a) shows the decay of the transient current immediately after the application of a DC voltage of 100 mV, indicating that the electron-blocking cell contains not only a resistive component associated with proton transport, but also a capacitive component. To examine the details of this behavior, the frequency-dependent responses of the cell were measured by AC impedance spectroscopy. Figure 3(b) shows the overall Nyquist plot of the electron-blocking cell, revealing the following distinct components: a high-frequency intercept resistance (R1R_{1}), mid-frequency arcs, and a low-frequency inclined straight-line response. Among these, the straight line observed in the low-frequency region has a slope close to 1, indicative of Warburg impedance (RWR_{\mathrm{W}})49. As shown in the enlarged view in Figure 3(c), two distinct arcs are observed in the mid-frequency region. These are denoted as R2R_{2} and R3R_{3} and are visually highlighted in red and blue, respectively, for clarity.

Current decay, Nyquist plot, and enlarged mid-frequency impedance plot with equivalent circuit

Figure 3. (a) Current decay after applying a 100-mV DC bias under a hydrogen atmosphere at 250 °C. (b) Nyquist plot of the AC impedance measured under the same conditions as in (a). (c) Enlarged view of the mid-frequency region in (b) along with the simplified equivalent circuit (top) used for fitting and the fitting result (green line).

The equivalent circuit of the asymmetric electron-blocking cell can be described by a complex transmission line model that represents mass and charge transport, as shown in Figure S650. The model includes elements related to charge transport at interfacial regions, leading to a complex impedance response. However, this complex impedance response makes it difficult to readily identify the resistive component associated with proton transport in HxWO3\mathrm{H}_{x}\mathrm{WO}_{3}. Although the transmission line model provides a physically meaningful representation of the system, previous studies have shown that fitting with a series of Randles circuits yields nearly equivalent results44,51. Therefore, while the transmission line model is more appropriate for physical interpretation, we herein discuss the fitting results based on the simplified equivalent circuit shown in the upper part of Figure 3(c).

Equivalent circuit for the asymmetric electron-blocking cell

Figure S6. Illustration of the equivalent circuit for the asymmetric electron-blocking cell based on the transmission line model. The model includes ionic and electronic conduction pathways in the Hx_xWO3_3 pellet, as well as charge transfer processes at the glass blocking electrode interface and reactions at the gas–solid interface. The effects of hydrogen distribution inhomogeneity or the presence of multiple phases in Hx_xWO3_3 are not considered.

As a first step, the contribution from proton transport in the glass electrode can be readily separated, as its relaxation is the fastest and appears as a distinct high-frequency resistance. As shown in Figure S7, the temperature dependence of R1R_1 agrees well with the reported value of the proton-conducting glass41, allowing it to be assigned to proton transport within the glass. The remaining impedance components were identified through the subsequent measurements.

Temperature dependence plot showing resistance versus temperature, with measured $R_1$ points and a calculated glass-resistance curve labelled Yamaguchi et al.

Figure S7. Temperature dependence of R1R_1 compared with the resistance calculated for the glass with the same geometry using the reported conductivity dataS1.

Figure 4(a) presents the Nyquist plots of the electron-blocking cell measured under H2\mathrm{H}_2 and D2\mathrm{D}_2 atmospheres. Switching the atmosphere from H2\mathrm{H}_2 to D2\mathrm{D}_2 increased not only R1R_1, reflecting the isotope effect in the glass electrode, but also R2R_2 and R3R_3. This phenomenon provides evidence that these arcs are associated with proton-related processes such as proton transport in HxWO3\mathrm{H}_x\mathrm{WO}_3 or interfacial charge transfer. The corresponding resistance values are plotted in Figure 4(b).

Nyquist plots and fitted resistance values

Figure 4. (a) Nyquist plots of the impedance spectra measured under H2_2 (top) and D2_2 (bottom) atmospheres at 250 °C. (b, left) Fitted resistance values of R1R_1, R2R_2, and R3R_3 under H2_2 and D2_2. (b, right)

To further elucidate the assignment of the R2R_2 and R3R_3 components, we performed impedance measurements on HxWO3\mathrm{H}_x\mathrm{WO}_3 pellets of different thicknesses. The pellet with an initial thickness of 2.3 mm was mechanically polished by 500 μ\mum from the side opposite to the glass layer to reduce its thickness to 1.8 mm. The corresponding Nyquist plots are shown in Figure 4(c), and the changes in the R1R_1, R2R_2, and R3R_3 components are summarized in Figure 4(d). As noted in Figure S3, the thickness of the hydrogen-rich region is approximately 500 μ\mum. Therefore, removing an area of approximately 500 μ\mum from one side of the pellet effectively reduces the thickness of the hydrogen-rich region by about half, while leaving the hydrogen-poor core nearly unchanged. This reduction primarily affected the R2R_2 component, which suggests that R2R_2 corresponds to the proton transport resistance within the hydrogen-rich region.

In contrast, the R3R_3 component exhibited only a minor change upon thickness reduction, suggesting a possible contribution from proton transport in the hydrogen-poor core, the thickness of which remained nearly unchanged. However, since the polishing was performed from the side opposite to the glass electrode, the glass-side interface remained unchanged. Thus, contributions to R3R_3 from charge transfer resistance at the Pd/glass and Pd/Hx_xWO3_3 interfaces cannot be ruled out. Therefore, the present results are insufficient to unambiguously identify the processes responsible for R3R_3 and determine the proton conductivity of the hydrogen-poor core.

Finally, RWR_\mathrm{W} exhibited negligible change upon isotope substitution or variation in pellet thickness, suggesting the presence of diffusion-limited proton transport within the cell, regardless of sample geometry or atmosphere. This result is in line with the presence of a diffusion-retarding region at the monoclinic–tetragonal phase boundary, as described in Figure 1(b).

Corresponding resistance ratios (R(D2)/R(H2)R(\mathrm{D}_2) / R(\mathrm{H}_2)) representing isotope effects for each component. (c) Nyquist plots of impedance spectra measured under H2\mathrm{H}_2 at 250 °C for samples with different thicknesses (2.3 mm and 1.8 mm, polished). (d, left) Fitted resistance values of R1R_1, R2R_2, and R3R_3 for 2.3-mm- and 1.8-mm-thick samples. (d, right) Corresponding resistance ratios (R(1.8 mm)/R(2.3 mm)R(1.8\ \mathrm{mm}) / R(2.3\ \mathrm{mm})) for each component.

Figure 5(a) displays the Arrhenius plot of the proton conductivity (σH\sigma_{\mathrm{H}}) of HxWO3\mathrm{H}_x\mathrm{WO}_3 measured in a hydrogen atmosphere. The conductivity was calculated for the hydrogen-rich H0.24WO3\mathrm{H}_{0.24}\mathrm{WO}_3 region (thickness: 1 mm) from the R2R_2 component. The corresponding electronic conductivity (σe\sigma_{\mathrm{e}}), determined by the DC four-probe method (Figure S8), is also plotted. Although the proton transport number in the measured temperature range is relatively low (<1×10−3<1 \times10^{-3}), σH\sigma_{\mathrm{H}} reaches 1×10−1 S cm−11 \times10^{-1}\ \mathrm{S\ cm}^{-1} at 275 °C. This value significantly exceeds that of perovskite-type proton-conducting electrolytes in the intermediate-temperature range52–54, indicating that HxWO3\mathrm{H}_x\mathrm{WO}_3 holds promise as an electrode material when combined with such electrolytes.

Arrhenius plots of conductivity and proton diffusion coefficient

Figure 5. (a) Arrhenius plot of conductivity. The proton conductivity (σH\sigma_{\mathrm{H}}) of H0.24WO3\mathrm{H}_{0.24}\mathrm{WO}_{3} is shown as red closed circles, and the electronic conductivity (σe\sigma_{\mathrm{e}}) as blue open circles. For comparison, σH\sigma_{\mathrm{H}} values of proton-conducting oxide electrolytes are also plotted: BaSc0.8Mo0.2O2.8\mathrm{BaSc}_{0.8}\mathrm{Mo}_{0.2}\mathrm{O}_{2.8} (BSM20; purple line)52, BaZr0.4Sc0.6O3−δ\mathrm{BaZr}_{0.4}\mathrm{Sc}_{0.6}\mathrm{O}_{3-\delta} (BZS60; green line)53, and BaZr0.85Y0.15O3−δ\mathrm{BaZr}_{0.85}\mathrm{Y}_{0.15}\mathrm{O}_{3-\delta} (BZY15; orange line)54. (b) Arrhenius plot of the proton diffusion coefficient (DHD_{\mathrm{H}}) and activation energy (EaE_{\mathrm{a}}). The red circles represent the

Planar sample for electronic conductivity measurements and Arrhenius plots

Figure S8. (a) Illustration of the planar sample for electronic conductivity measurements using the four-probe method. (b) Arrhenius plots of the electronic conductivity for H0.24WO3H_{0.24}\mathrm{WO}_3 (red circle), H0.0048WO3H_{0.0048}\mathrm{WO}_3 (blue triangle), and H∼0.0001TiO2H_{\sim0.0001}\mathrm{TiO}_2 (green square) with the corresponding activation energiesS2. The electronic conductivities of H0.24WO3H_{0.24}\mathrm{WO}_3 and H0.0048WO3H_{0.0048}\mathrm{WO}_3 were measured under Ar to prevent changes in hydrogen content, while that of H∼0.0001TiO2H_{\sim0.0001}\mathrm{TiO}_2 was measured under H2H_2 atmosphere.

Figure 5(b) presents the temperature dependence of the proton diffusion coefficient (DHD_{\mathrm{H}}) in H0.24WO3\mathrm{H}_{0.24}\mathrm{WO}_3 calculated from proton conductivity and proton carrier density using the Nernst–Einstein relation:

DH=kBTσHe2nH(3)D_{\mathrm{H}} = \frac{k_{\mathrm{B}}T\sigma_{\mathrm{H}}}{e^2 n_{\mathrm{H}}} \tag*{(3)}

where kBk_{\mathrm{B}} is the Boltzmann constant, and TT is the absolute temperature, ee is the elementary charge. The value of DHD_{\mathrm{H}} extrapolated to lower temperatures agrees well with that reported in an earlier work30, indicating that the excellent proton transport properties are retained over a wide temperature range.

Herein, we compare the proton diffusion behavior of HxWO3\mathrm{H}_x\mathrm{WO}_3 with that of rutile-type HxTiO2\mathrm{H}_x\mathrm{TiO}_2, which also incorporates protons via hydrogen dissolution (Equation (1))55,56. Like WO3\mathrm{WO}_3, TiO2\mathrm{TiO}_2 undergoes a transition from a d0d^0 to a d1d^1 electronic configuration upon hydrogen incorporation, leading to the formation of Ti3+\mathrm{Ti}^{3+} species. However, the amount of hydrogen incorporated into TiO2\mathrm{TiO}_{2} is significantly smaller than that in WO3\mathrm{WO}_{3}, typically resulting in proton densities on the order of 101810^{18}–1019 cm−310^{19}\ \mathrm{cm}^{-3} corresponding to a hydrogen composition of x∼0.0001x \sim0.000136. In both HxWO3\mathrm{H}_{x}\mathrm{WO}_{3} and HxTiO2\mathrm{H}_{x}\mathrm{TiO}_{2}, equal amounts of protons and electrons are introduced by hydrogen dissolution, leading to coupled proton–electron transport, referred to as ambipolar diffusion57. Notably, the DHD_{\mathrm{H}} of H0.24WO3\mathrm{H}_{0.24}\mathrm{WO}_{3} is 100–1000 times greater than that of H∼0.0001TiO2\mathrm{H}_{\sim0.0001}\mathrm{TiO}_{2} along its one-dimensional cc-axis diffusion pathways36,58. In HxTiO2\mathrm{H}_{x}\mathrm{TiO}_{2}, the electrons tend to localize at Ti sites, forming small polarons (TiTi′\mathrm{Ti}_{\mathrm{Ti}}^{\prime}) with low mobility59,60. The negative charge of these polarons electrostatically attracts protons and acts as proton traps, thereby limiting the proton mobility. In contrast, in HxWO3\mathrm{H}_{x}\mathrm{WO}_{3}, the electrons introduced via hydrogen dissolution are likely to form polarons with a relatively larger polaron radius, as suggested by the small activation energies (EaE_{\mathrm{a}}) for electronic conduction (Figure S8)61–63. These highly mobile and delocalized electrons do not produce strong localized electrostatic fields that immobilize protons, accounting for the larger DHD_{\mathrm{H}} of H0.24WO3\mathrm{H}_{0.24}\mathrm{WO}_{3} than that of H∼0.0001TiO2\mathrm{H}_{\sim0.0001}\mathrm{TiO}_{2}.

values calculated from the σH\sigma_{\mathrm{H}} of H0.24WO3\mathrm{H}_{0.24}\mathrm{WO}_{3} measured in this study. Black open circles are the values for crystalline HxWO3\mathrm{H}_{x}\mathrm{WO}_{3} thin film (x∼0.1x \sim0.1), as reported by Randin et al30. For comparison, DHD_{\mathrm{H}} along the abab-plane and cc-axis of H∼0.0001TiO2\mathrm{H}_{\sim0.0001}\mathrm{TiO}_{2}, reported by Johnson et al., are shown as blue and green lines, respectively36.

Interestingly, H0.24WO3\mathrm{H}_{0.24}\mathrm{WO}_{3} exhibits EaE_{\mathrm{a}} for proton diffusion that is comparable to or even higher than that of H∼0.0001TiO2\mathrm{H}_{\sim0.0001}\mathrm{TiO}_{2} along the cc-axis, despite the significantly higher diffusion coefficient. The temperature dependence of diffusion coefficient can be determined using the Arrhenius equation:

DH=D0exp⁡(−EakBT)(4)D_{\mathrm{H}} = D_{0}\exp\left(-\frac{E_{\mathrm{a}}}{k_{\mathrm{B}}T}\right) \tag*{(4)}

where D0D_{0} is the pre-exponential factor. The large DHD_{\mathrm{H}} in H0.24WO3\mathrm{H}_{0.24}\mathrm{WO}_{3} associated with the formation of a large polaron can be attributed to a larger D0D_{0} compared to that of H∼0.0001TiO2\mathrm{H}_{\sim0.0001}\mathrm{TiO}_{2}, as shown in Table 1.

H0.24WO3H_{0.24}WO_3H∼0.0001TiO2H_{\sim0.0001}TiO_2 (cc-axis)
D0D_0 / cm2^2 s−1^{-1}100.002
ΔSa\Delta S_a / meV K−1^{-1}0.80.04

Table 1. Comparison of diffusion pre-exponential factors and estimated activation entropies for H0.24WO3H_{0.24}WO_3 and H∼0.0001TiO2H_{\sim0.0001}TiO_2.

From the random-walk theory, D0D_{0} is described as follows:

D0=a26ν0exp⁡(ΔSakB)(5)D_{0} = \frac{a^{2}}{6}\nu_{0}\exp\left(\frac{\Delta S_{\mathrm{a}}}{k_{\mathrm{B}}}\right) \tag*{(5)}

where aa is the jump distance, ν0\nu_{0} is the attempt frequency (i.e., the number of attempted jumps per unit time), and ΔSa\Delta S_{\mathrm{a}} is the activation entropy. For proton conducting oxides, aa can be taken as the distance between the neighboring oxide ions where proton transfer occurs, and ν0\nu_{0} can be related to be the frequencies of M–O vibrational modes, such as O–M–O bending or M–O stretching modes64,65. Thus, for HxWO3\mathrm{H}_{x}\mathrm{WO}_{3} and HxTiO2\mathrm{H}_{x}\mathrm{TiO}_{2}, aa and ν0\nu_{0} are on the order of 2.7 Å and 1013 s−110^{13}\ \mathrm{s}^{-1}, respectively, allowing a rough estimation of ΔSa\Delta S_{\mathrm{a}} from the experimentally obtained D0D_{0} values. As seen in Table 1, the estimated ΔSa\Delta S_{\mathrm{a}} is larger for H0.24WO3\mathrm{H}_{0.24}\mathrm{WO}_{3} than that for H∼0.0001TiO2\mathrm{H}_{\sim0.0001}\mathrm{TiO}_{2}. While this difference may partly result from a greater number of accessible diffusion pathways of protons in H0.24WO3\mathrm{H}_{0.24}\mathrm{WO}_{3}, it may also reflect a higher statistical multiplicity of microstates associated with the absorption of multiple phonons required to overcome a migration barrier66. A detailed understanding of this behavior remains a subject for future investigation.

CONCLUSIONS

This study reported on the proton conductivity of hydrogen tungsten bronze (HxWO3H_xWO_3) in the intermediate-temperature range using an electron-blocking method with a phosphate glass electrolyte. The sample exhibited a spatially inhomogeneous hydrogen distribution, consisting of a hydrogen-rich surface region (x=0.24x = 0.24) and a hydrogen-poor core (x=0.0048x = 0.0048). The limited penetration depth of hydrogen (∼500\sim500 μm) even after prolonged annealing suggests that structural phase boundaries may hinder uniform hydrogen incorporation. The impedance component corresponding to proton conduction in H0.24WO3H_{0.24}WO_3 was separated from the overall impedance based on comparative measurements obtained using isotope substitution and pellets with different thicknesses. The resulting proton conductivity of H0.24WO3H_{0.24}WO_3 reached 10−110^{-1} S cm−1^{-1} at 275 °C under a H2H_2 atmosphere. This value significantly exceeds that of perovskite-type proton-conducting electrolytes in the intermediate-temperature range, which suggests that HxWO3H_xWO_3 is a promising electrode material for use in combination with these electrolytes. A comparison with rutile-type H∼0.0001TiO2H_{\sim0.0001}TiO_2 revealed that H0.24WO3H_{0.24}WO_3 possesses proton diffusion coefficients that are 100–1000 times larger, indicating the formation of large polarons of electrons. The delocalized electrons do not electrostatically trap protons at specific positions, thereby allowing efficient proton migration. Furthermore, H0.24WO3H_{0.24}WO_3 exhibits a larger D0D_0 than that of H∼0.0001TiO2H_{\sim0.0001}TiO_2, which is attributed to a greater ΔSa\Delta S_a. This phenomenon likely reflects the presence of a wide variety of accessible proton migration pathways or reduced vibrational constraints in the activated state in H0.24WO3\mathrm{H}_{0.24}\mathrm{WO}_{3}, contributing to its high proton diffusivity despite a comparable or higher EaE_{\mathrm{a}}. These results highlight the critical role of entropic contributions in the proton diffusion process of HxWO3\mathrm{H}_{x}\mathrm{WO}_{3}, offering a new perspective for designing proton-conducting materials with high proton conductivity.

ASSOCIATED CONTENT

Supporting Information.

Microstructure of sintered HxWO3\mathrm{H}_{x}\mathrm{WO}_{3}, hydrogen distribution in the HxWO3\mathrm{H}_{x}\mathrm{WO}_{3} measured by TOF-SIMS, Quantification of hydrogen in the sintered HxWO3\mathrm{H}_{x}\mathrm{WO}_{3} by TDS, Growth behavior of the hydrogen-rich region by hydrogen annealing, Details of the electron-blocking measurement, and electronic conductivity of HxWO3\mathrm{H}_{x}\mathrm{WO}_{3}.

AUTHOR INFORMATION

Corresponding Author

Tomoyuki Yamasaki – Institute of Multidisciplinary Research for Advanced Materials (IMRAM), Tohoku University, 2-1-1 Katahira, Aoba-ku, Sendai 980-8577, Japan; ORCID orcid.org/0000-0002-6982-7538; Email: [email protected]

Takahisa Omata – Institute of Multidisciplinary Research for Advanced Materials (IMRAM), Tohoku University, 2-1-1 Katahira, Aoba-ku, Sendai 980-8577, Japan; ORCID orcid.org/0000-0002-6034-4935; Email: [email protected]

Author Contributions

Rantaro Matsuo: Investigation, Visualization. Tomoyuki Yamasaki: Investigation, Visualization, Funding acquisition, Writing – original draft. Takahisa Omata: Conceptualization, Funding acquisition, Supervision, Writing – review & editing.

Notes

The authors declare no competing financial interest.

Acknowledgment

This work was supported in part by JSPS KAKENHI Grant Numbers 21H04607 and 24K17762.

This work was partly performed under the Cooperative Research Program of the “Network Joint Research Center for Materials and Devices” (No. 20241114) and “Dynamic Alliance for Open Innovation Bridging Human, Environment, and Materials”.

Abbreviations

MIEC, mixed ionic–electronic conductor; QMASS, quadrupole mass spectrometer; SPS, spark plasma sintering; TOF-SIMS, time-of-flight secondary ion mass spectrometry; TDS, thermal desorption spectrometry; XRD, X-ray diffraction.

References

Supporting Information

Microstructure of sintered Hx_xWO3_3

Hydrogen distribution in the sintered HxWO3\mathrm{H}_x\mathrm{WO}_3 measured by TOF-SIMS

Background measurement of hydrogen in WO3\mathrm{WO}_3

To evaluate the background hydrogen signal in the TOF-SIMS analysis and determine the necessary sputtering time required to eliminate surface contamination, depth profile measurements were conducted on a reference WO3\mathrm{WO}_3 sintered pellet (annealed in flowing oxygen at 800 °C for 10 h) as well as on the HxWO3\mathrm{H}_x\mathrm{WO}_3 pellet. The TOF-SIMS depth profiles of the secondary ions were acquired in the negative polarity mode using a Bi+\mathrm{Bi}^{+} primary ion beam (25 kV, 1.0 pA) and a Cs+\mathrm{Cs}^{+} sputter ion beam (2 kV, 145 nA). Figure S2 shows the depth profiles of H−\mathrm{H}^{-} and OH−\mathrm{OH}^{-} signals, normalized to the W−\mathrm{W}^{-} signal for reference WO3\mathrm{WO}_3 and HxWO3\mathrm{H}_x\mathrm{WO}_3 pellets. In the reference WO3\mathrm{WO}_3 sample, the signals gradually decayed toward the interior of the pellet and became nearly constant after approximately 60 min of sputtering, indicating the elimination of the contribution from surface-adsorbed species. In contrast, for the HxWO3\mathrm{H}_x\mathrm{WO}_3 sample, the intensities of H−\mathrm{H}^{-} and OH−\mathrm{OH}^{-} signals remained nearly constant throughout the sputtering process. Even after sputtering beyond the depth at which surface contributions were no longer observed in the reference WO3\mathrm{WO}_3 sample, the signal intensities in HxWO3\mathrm{H}_x\mathrm{WO}_3 remained approximately an order of magnitude higher than those in the reference. These results clearly demonstrate that hydrogen in the HxWO3\mathrm{H}_x\mathrm{WO}_3 sample can be reliably detected above the background level observed in the reference WO3\mathrm{WO}_3 sample.

SIMS area imaging of the cross-section of HxWO3\mathrm{H}_x\mathrm{WO}_3

To obtain SIMS imaging data free from surface contamination, a thin surface layer was sputtered prior to acquisition, and imaging was initiated from the depth at which the signal intensities of H−\mathrm{H}^{-} and OH−\mathrm{OH}^{-} became stable. The imaging area was selected 300 μ\mum inward from the edge of the pellet. A 500 ×\times 500 μm2\mu\mathrm{m}^{2} region was imaged to visualize the lateral distribution of hydrogen. Figure S2 displays lateral SIMS images of H−\mathrm{H}^{-}, OH−\mathrm{OH}^{-}, and W−\mathrm{W}^{-}. In these images, the left edge corresponds to a depth of 300 μ\mum from the surface, and the right edge corresponds to approximately 800 μ\mum. To convert the SIMS images into a concentration profile along the thickness direction of the sintered pellet (Figure 1 (b)), the signal intensities were integrated along the Y-direction.

Quantification of hydrogen in sintered Hx_xWO3_3 by TDS

SurfaceCore
H released as H2_2 / cm−3^{-3}1.3×10211.3 \times10^{21}3.2×10173.2 \times10^{17}
H released as H2_2O / cm−3^{-3}3.2×10213.2 \times10^{21}8.9×10198.9 \times10^{19}
Total H / cm−3^{-3}4.5×10214.5 \times10^{21}8.9×10198.9 \times10^{19}
xx in Hx_xWO3_30.240.0048

Table S1. Summary of the quantification of hydrogen released as H2_2 and H2_2O.

Growth behavior of the hydrogen-rich region by hydrogen annealing

Supporting details of the electron-blocking measurement

Electronic conductivity of HxWO3H_x\mathrm{WO}_3

References

  1. [1]S1 T. Yamaguchi, S. Tsukuda, T. Ishiyama, J. Nishii, T. Yamashita, H. Kawazoe and T. Omata, J. Mater. Chem. A, 2018, 6, 23628–23637.
  2. [2]S2 T. Shiraiwa, T. Yamasaki, K. Kushimoto, J. Kano and T. Omata, submitted to J. Am. Chem. Soc.

Paper details

vlnc:W_cvhgzw2y
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 1

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents