Hydrogen Tungsten Bronzes (HxWO3): Key Insights

• Hydrogen tungsten bronzes (HxWO3) are phases formed by hydrogen intercalation into WO3, resulting in mixed valence states, lattice distortions, and tunable electronic behavior.
• Catalytic, electrolytic, and mechanochemical hydrogenation methods enable controlled phase transitions and property modulation impacting optical coloration and conductivity.
• HxWO3 exhibits chemical-driven strain actuation and mixed ionic–electronic conduction, making it valuable for applications in electrochromic devices, sensors, and energy systems.

Searching arXiv for the cited HxWO3 and WO3 hydrogenation papers to ground the article in current preprint metadata. {"query":"(Manca et al., 2021) hydrogen tungsten bronze WO3 hydrogenation WO3 single-crystal microresonators"} Hydrogen tungsten bronzes, $\mathrm{H}_x\mathrm{WO}_3$, are hydrogen-inserted phases of tungsten trioxide in which hydrogen incorporation is accompanied by electron donation to the W–O framework, partial reduction of $\mathrm{W}^{6+}$ to $\mathrm{W}^{5+}$, and coupled structural, optical, electronic, and transport changes. In contemporary work, $\mathrm{H}_x\mathrm{WO}_3$ appears not as a single narrowly defined compound but as a compositionally and structurally tunable family spanning electrochromic and gasochromic behavior, polaronic and metallic conduction, mechanically actuated thin-film states, and mixed ionic–electronic conduction at intermediate temperatures (Billeter et al., 2020, Leng et al., 2017, Matsuo et al., 9 Jul 2025).

1. Composition, crystal chemistry, and phase space

Stoichiometric $\mathrm{WO}_3$ is a $d^0$ transition-metal oxide with distorted ReO$_3$-type structures, including monoclinic $\gamma$-WO$_3$ at room temperature (Matsuo et al., 9 Jul 2025). Hydrogen tungsten bronzes form when hydrogen is inserted into the $\mathrm{WO}_3$ lattice chemically or electrochemically; hydrogen typically occupies interstitial sites or forms O–H bonds within the $\mathrm{W}^{6+}$0 octahedral framework, and the composition is written $\mathrm{W}^{6+}$1, where $\mathrm{W}^{6+}$2 is the number of hydrogen atoms per formula unit (Matsuo et al., 9 Jul 2025).

Hydrogen content is structurally consequential. One study states that $\mathrm{W}^{6+}$3 can dissolve hydrogen up to $\mathrm{W}^{6+}$4, with monoclinic symmetry for $\mathrm{W}^{6+}$5, orthorhombic symmetry for $\mathrm{W}^{6+}$6, and tetragonal symmetry for $\mathrm{W}^{6+}$7 (Matsuo et al., 9 Jul 2025). A room-temperature mechanochemical route likewise identifies tetragonal $\mathrm{W}^{6+}$8 in the range $\mathrm{W}^{6+}$9 after milling monoclinic $\mathrm{W}^{5+}$0 with polypropylene (Kato et al., 2021). In a thin-film epitaxial geometry on SrTiO$\mathrm{W}^{5+}$1, pristine $\mathrm{W}^{5+}$2 shows an out-of-plane lattice constant $\mathrm{W}^{5+}$3, whereas a strongly hydrogenated state reaches $\mathrm{W}^{5+}$4, corresponding to $\mathrm{W}^{5+}$5 (Manca et al., 2021).

At the local scale, a proton-centered picture has been developed from membrane XPS and DFT. In that description, hydrogen sits near oxygen at about $\mathrm{W}^{5+}$6–$\mathrm{W}^{5+}$7, forming an OH-like bond, while nearby tungsten sites become locally distinguishable as $\mathrm{W}^{5+}$8 and $\mathrm{W}^{5+}$9 environments (Billeter et al., 2020). This places hydrogen bronze formation simultaneously in the categories of intercalation chemistry, mixed valence, and lattice-distortion physics.

2. Formation pathways and the mechanism of hydrogen incorporation

Several experimentally distinct routes lead to $\mathrm{H}_x\mathrm{WO}_3$0. In catalytic hydrogenation, ultrathin Pt on $\mathrm{H}_x\mathrm{WO}_3$1 dissociates H$\mathrm{H}_x\mathrm{WO}_3$2 into H atoms that diffuse into the oxide lattice, donate electrons, induce metallization, and cause large lattice expansion. In freestanding single-crystal microresonator devices, the Pt thickness is $\mathrm{H}_x\mathrm{WO}_3$3 equivalent, hydrogenation can be driven at room temperature, and the hydrogen incorporation rate is explicitly noted to be tunable via Pt amount (Manca et al., 2021). Under one set of structural conditions, $\mathrm{H}_x\mathrm{WO}_3$4 bar of $\mathrm{H}_x\mathrm{WO}_3$5 H$\mathrm{H}_x\mathrm{WO}_3$6/Ar for $\mathrm{H}_x\mathrm{WO}_3$7 h yields a saturated hydrogen-doped state; under a nanomechanical protocol, $\mathrm{H}_x\mathrm{WO}_3$8 mbar of $\mathrm{H}_x\mathrm{WO}_3$9 H$\mathrm{WO}_3$0/Ar produces multi-hour frequency evolution and eventual saturation (Manca et al., 2021).

Electrolyte gating provides a second route, and it has been used to separate hydrogen intercalation from competing explanations. In a staircase thin-film experiment, the saturated sheet conductance scales linearly with thickness from $\mathrm{WO}_3$1 to $\mathrm{WO}_3$2 nm, showing that the whole film thickness is doped rather than only an interfacial screening layer. The critical sheet resistance at the metal-insulator transition is reported as $\mathrm{WO}_3$3, far below the two-dimensional quantum resistance scale, and charging behavior is identical in vacuum and air; together with pure-PEG control experiments and lateral electromigration over about $\mathrm{WO}_3$4, these observations were used to rule out both purely electrostatic accumulation and oxygen-vacancy mechanisms, and to identify H$\mathrm{WO}_3$5 intercalation from trace water as the dominant process (Leng et al., 2017). The electrochemical reaction was written as

$\mathrm{WO}_3$6

A third route is mechanochemically induced hydrogen spillover. Milling monoclinic $\mathrm{WO}_3$7 with polypropylene in Ar at $\mathrm{WO}_3$8 rpm, typically for $\mathrm{WO}_3$9 h and with a WO$d^0$0:PP volume ratio of $d^0$1, yields tetragonal $d^0$2 nanoparticles and graphitic nanocarbon without noble-metal catalyst, solvent, or external heating (Kato et al., 2021). The proposed pathway is a mechanically activated cascade in which milling generates oxygen-vacancy-rich, catalytically active $d^0$3 surfaces; polypropylene depolymerizes into gaseous hydrocarbon fragments; and these fragments supply active hydrogen that spills over into the oxide framework. Time-resolved XRD shows a mixed monoclinic/tetragonal state after about $d^0$4 h and an essentially complete transformation after $d^0$5–$d^0$6 h (Kato et al., 2021).

Taken together, these routes establish that hydrogen tungsten bronze formation is accessible by gas-phase catalysis, electrochemical protonation, and dry mechanochemistry, with the controlling variables distributed among hydrogen chemical potential, catalyst or electrolyte chemistry, temperature, and time.

3. Electronic structure, polarons, and the insulator-to-metal crossover

The electronic structure of $d^0$7 has long been entangled with a central controversy: whether hydrogen-induced conductivity and coloration are intrinsic consequences of hydrogen intercalation or secondary consequences of oxygen deficiency. Operando membrane photoemission addressed this directly by measuring a WO$d^0$8 film supported on Pd under hydrogen pressures up to $d^0$9 mbar. Upon hydrogenation, the intensity of the $_3$0 core level and of a state close to the Fermi level both increase, and their pressure dependence follows the pressure–composition isotherm of bulk $_3$1; at the same time, the O/W ratio changes by less than $_3$2 (Billeter et al., 2020). The resulting interpretation is a proton-polaron model in which hydrogen forms OH groups while the associated electron occupies tungsten-derived states near the conduction-band edge. In this picture, the valence-band edge of pristine $_3$3 is oxygen-dominated, whereas hydrogenation produces a near-$_3$4 regime dominated by tungsten orbitals (Billeter et al., 2020).

Electrolyte-gated transport experiments reveal the magnitude of the electronic transformation. The resistivity can be tuned by more than five orders of magnitude, and a clear insulator-to-metal transition appears. At the highest gate voltages, the resistivity at $_3$5 K is about $_3$6, and the films display metallic behavior with $_3$7 over a wide temperature range; no superconducting downturn was seen down to $_3$8 K (Leng et al., 2017). Operando infrared spectroscopy shows a broad mid-IR band around $_3$9 eV, identified as a polaronic feature. From a Lorentz-oscillator analysis, the carrier density is estimated as

$\gamma$0

corresponding to approximately $\gamma$1 electrons and hydrogen per WO$\gamma$2 unit cell if one electron accompanies each proton (Leng et al., 2017). The same work describes the high-doping state as a dense polaronic gas.

Hall data reinforce the non-Drude character of the metallic state. A single-band Hall analysis gives an unphysical carrier density corresponding to about $\gamma$3 electrons per formula unit, while the nominal Hall mobility is only about $\gamma$4 (Leng et al., 2017). These values are consistent with the paper’s interpretation that the metallic phase remains strongly polaronic rather than becoming a simple free-electron metal.

Optically, hydrogen bronze formation is classically associated with blue coloration, and the mechanochemical nanoparticle study reproduces that signature together with strong visible and near-infrared absorption, quasi-free-electron absorption above $\gamma$5, and a broadened W 4f XPS envelope containing $\gamma$6 (Kato et al., 2021). In aggregate, these observations place $\gamma$7 in the unusual regime where coloration, mixed valence, and a low-mobility metallic endpoint are all manifestations of the same hydrogen-driven reconstruction of the W–O electronic manifold.

4. Lattice expansion, strain control, and nanomechanical response

Hydrogen incorporation into $\gamma$8 is not merely an electronic doping process; it is also a large lattice-strain perturbation. In epitaxial films on Ti-terminated SrTiO$\gamma$9(001), pristine $_3$0 nm single-crystal $_3$1 is coherently strained to the substrate, with in-plane axes elongated by about $_3$2 and the $_3$3 axis compressed by about $_3$4 relative to bulk pseudocubic $_3$5 (Manca et al., 2021). After hydrogenation, the out-of-plane lattice constant increases from $_3$6 to $_3$7, i.e. $_3$8 (Manca et al., 2021).

For a freestanding film, the measured out-of-plane expansion can be related to the in-plane strain change by

$_3$9

and with $\mathrm{WO}_3$0 the in-plane strain variation is estimated as $\mathrm{WO}_3$1 (Manca et al., 2021). Combining elastic modeling and mode analysis yields an initial microbridge strain $\mathrm{WO}_3$2 and a saturated hydrogenated strain of about $\mathrm{WO}_3$3, so hydrogenation sweeps the bridge from tensile to compressive stress over a range of roughly $\mathrm{WO}_3$4 (Manca et al., 2021).

The resulting mechanics are pronounced. For the first flexural mode of a doubly clamped beam,

$\mathrm{WO}_3$5

with finite-element parameters $\mathrm{WO}_3$6, $\mathrm{WO}_3$7, and $\mathrm{WO}_3$8 in the cited simulations (Manca et al., 2021). As hydrogen is intercalated, the first mode frequency drops from about $\mathrm{WO}_3$9 Hz to a minimum near $\mathrm{W}^{6+}$00 Hz, roughly one fifth of its initial value, then rises again after the beam buckles; higher modes exhibit an avoided crossing around $\mathrm{W}^{6+}$01 MHz (Manca et al., 2021). At $\mathrm{W}^{6+}$02, the center displacement reaches about $\mathrm{W}^{6+}$03, in agreement with optical imaging of the buckled state (Manca et al., 2021).

These results define a mechanically active bronze regime: hydrogen content becomes a direct control knob for resonance frequency, static deflection, and stress state at room temperature. The same experiments also show reversibility, with dehydrogenation in air recovering both the lattice constant and the bridge geometry on the order of one hour (Manca et al., 2021). In this sense, $\mathrm{W}^{6+}$04 functions as a chemically driven strain actuator embedded in an oxide nanomechanical device.

5. Proton transport and mixed ionic–electronic conduction at intermediate temperature

A later development extended the relevance of $\mathrm{W}^{6+}$05 from room-temperature electrochromism to intermediate-temperature electrochemistry. In dense sintered WO$\mathrm{W}^{6+}$06 pellets hydrogenated in H$\mathrm{W}^{6+}$07 at $\mathrm{W}^{6+}$08 for up to $\mathrm{W}^{6+}$09 h using Pd catalyst layers, hydrogen incorporation is strongly inhomogeneous: a hydrogen-rich region about $\mathrm{W}^{6+}$10 thick forms near the surface, while the core remains hydrogen-poor (Matsuo et al., 9 Jul 2025). Thermal desorption spectrometry gives a surface-region composition of $\mathrm{W}^{6+}$11 with total hydrogen concentration $\mathrm{W}^{6+}$12, whereas the core corresponds to $\mathrm{W}^{6+}$13 and $\mathrm{W}^{6+}$14 (Matsuo et al., 9 Jul 2025). XRD identifies the outer region as tetragonal and the core as monoclinic, consistent with the composition-dependent structural sequence of the bronze family (Matsuo et al., 9 Jul 2025).

To isolate proton transport from the large electronic conductivity, the study used a proton-conducting phosphate glass with proton transport number $\mathrm{W}^{6+}$15 as an electron-blocking electrode. Under these conditions, the hydrogen-rich $\mathrm{W}^{6+}$16 region exhibits partial proton conductivity of about $\mathrm{W}^{6+}$17 at $\mathrm{W}^{6+}$18 and greater than $\mathrm{W}^{6+}$19 at $\mathrm{W}^{6+}$20 (Matsuo et al., 9 Jul 2025). The paper states that these values are one to two orders of magnitude above those of state-of-the-art perovskite proton conductors in the same temperature range (Matsuo et al., 9 Jul 2025).

The proton diffusion coefficient was obtained using the Nernst–Einstein relation,

$\mathrm{W}^{6+}$21

and was found to be $\mathrm{W}^{6+}$22–$\mathrm{W}^{6+}$23 times larger than that of $\mathrm{W}^{6+}$24 (Matsuo et al., 9 Jul 2025). Impedance spectra show an isotope effect under H$\mathrm{W}^{6+}$25/D$\mathrm{W}^{6+}$26 exchange, confirming protonic transport. At the same time, the total bulk conductivity remains strongly electronic, with proton transport number below $\mathrm{W}^{6+}$27 in the measured range, so the material is best categorized not as a pure proton electrolyte but as a proton-conducting mixed ionic–electronic conductor (Matsuo et al., 9 Jul 2025).

The same work argues that the exceptionally large proton diffusion coefficient in $\mathrm{W}^{6+}$28 reflects large-polaron electronic character rather than small-polaron trapping. This links the intermediate-temperature transport regime back to the broader bronze physics: electronic delocalization in the W–O framework is not only relevant for metallicity and coloration, but also for the freedom of protons to move through the lattice.

6. Applications, related bronze physics, and unresolved questions

Hydrogen tungsten bronzes are already embedded in several application classes explicitly named in the literature: electrochromic devices and smart windows, gasochromic and hydrogen-sensing platforms, photocatalysis, adaptive nanomechanics, and hydrogen-energy electrochemical systems (Matsuo et al., 9 Jul 2025, Manca et al., 2021, Kato et al., 2021). In photocatalysis, the mechanochemically synthesized $\mathrm{W}^{6+}$29/nanocarbon composite degrades methyl orange under visible light with a pseudo-first-order rate constant $\mathrm{W}^{6+}$30, compared with $\mathrm{W}^{6+}$31 for milled WO$\mathrm{W}^{6+}$32 without polypropylene and less than $\mathrm{W}^{6+}$33 for raw WO$\mathrm{W}^{6+}$34 (Kato et al., 2021). Annealing this composite at $\mathrm{W}^{6+}$35 in Ar reverses the tetragonal-to-monoclinic transformation and sharply reduces the photocatalytic performance, tying the activity to the bronze state (Kato et al., 2021).

Within tungsten bronze chemistry more broadly, alkali bronzes provide a benchmark for how electron count and structure shape collective states. In tetragonal $\mathrm{W}^{6+}$36, the highest superconducting transition temperature is $\mathrm{W}^{6+}$37 K at $\mathrm{W}^{6+}$38, and $\mathrm{W}^{6+}$39 decreases with increasing K content (Haldolaarachchige et al., 2014). That system is not hydrogen bronze, and the cited study does not treat $\mathrm{W}^{6+}$40 directly. A plausible implication is that, if superconductivity is pursued in hydrogen bronzes, the decisive variables are likely to be the same ones highlighted for other tungsten bronzes: framework topology, electron count, and proximity to structural boundaries (Haldolaarachchige et al., 2014). This is consistent with the electrolyte-gating result that no superconductivity appears down to $\mathrm{W}^{6+}$41 K in hydrogenated WO$\mathrm{W}^{6+}$42 thin films at approximately $\mathrm{W}^{6+}$43 (Leng et al., 2017).

A distinct open question concerns hydrogen spatial correlations at high concentration. A study of hydrogen in metallic tungsten, not in $\mathrm{W}^{6+}$44, found spontaneous formation of hydrogen platelets larger than $\mathrm{W}^{6+}$45 atoms, a critical cluster size of about six H atoms for long-term stability, and H–H interactions extending up to about $\mathrm{W}^{6+}$46 (Cusentino et al., 2023). Direct quantitative transfer to $\mathrm{W}^{6+}$47 is not appropriate because the host lattice is metallic W rather than an oxide framework. This nevertheless suggests that models of hydrogen tungsten bronzes based solely on randomly distributed, noninteracting interstitial H may become incomplete at high local hydrogen concentration.

Several unresolved issues recur across the recent literature. One is the control of hydrogen distribution: even after prolonged annealing, the intermediate-temperature proton-conduction study reports a hydrogen-rich layer only about $\mathrm{W}^{6+}$48 thick, implying a diffusion bottleneck at phase boundaries (Matsuo et al., 9 Jul 2025). Another is the precise nature of the metallic state reached by heavy hydrogenation: the dense-polaronic-gas description, anomalous Hall response, and strong mid-IR absorption all indicate that even the metallic endpoint is not a conventional Drude metal (Leng et al., 2017). A third concerns the partition of reversible hydrogen intercalation from oxygen-vacancy chemistry; recent operando XPS and gating studies have substantially narrowed that controversy, but they do not eliminate the possibility that some experimental protocols mix both mechanisms (Billeter et al., 2020, Leng et al., 2017).

In sum, $\mathrm{W}^{6+}$49 is best understood as a hydrogen-controlled tungsten-bronze family in which proton insertion reorganizes crystal structure, strain, optical absorption, and both electron and proton transport. Its importance lies precisely in that multiphysics coupling: the same hydrogen that colors the oxide also metallizes it, buckles a microbridge, and, at higher temperature, turns it into a high-performance protonic mixed conductor.