Structurally, tungsten oxide and its hydrates are two-dimensional layered oxides. One view is that the atomic hydrogen species can migrate along the surface of the WO₃ through the so-called "proton-electron coherent motion" — the hydrogen atom can donate its electron to the reducible metal cation of the oxide support, followed by diffusion in the form of protons, so tungsten oxide usually acts as a secondary active site for hydrogen overflow^[17,18][19]. Unless the chemical composition is changed or oxygen vacancies are introduced, the HER activity of tungsten oxide is almost negligible^[20,21].

When tungsten oxide acts as a secondary active site for HER, the primary active site is often a noble metal nanoparticle. The dispersion of different metals and the difference of Metal-support Interactions (MSI) will have an important impact on the behavior of hydrogen overflow^[22]. Lee et al. Prepared mesoporous WO_3-x supported Pt single atom catalyst (marked as Pt SA/m-WO_3-x）), and compared it with carbon supported Pt single atom (Pt SA/C) and mesoporous WO_3-x supported Pt nanoparticles （Pt NP/m-WO_3-x） control samples, and systematically studied the effects of different dispersion degrees of Pt and its interaction with different supports on hydrogen spillover in HER^[23]. By comparing and analyzing the cyclic voltammetry curves of the above three samples at different scan rates, it can be seen that Pt SA/m-WO_3-x shows faster hydrogen desorption/insertion kinetics than Pt SA/C and Pt NP/m-WO_3-x, which is mainly attributed to the strong interaction between highly dispersed Pt SA and m-WO_3-x, which enhances the hydrogen spillover effect on Pt SA/m-WO_3-x and shortens the H diffusion path (Fig. 3). In acidic environment, the HER activity of Pt SA/WO_3-x is comparable to that of Pt/C: when the Pt loading is 0.86µg·cm^-2, the overpotential is 47 mV（10 mA·cm^-2 current density); At an overpotential of 50 mV, the mass activity was 12.8 A·mg^-1（ in Pt) and the Tafel slope was 45 mV·dec^-1.

Although WO₃ often acts as a secondary active site in the hydrogen overflow process, there is still much room for improvement in the HER performance of the system due to its poor conductivity. Graphene aerogel has the characteristics of high conductivity, high specific surface area and high porosity, but its HER performance is not ideal, so it is often used as a catalyst support to enhance the catalytic interface synergy effect, accelerate electron transfer and mass transport^{[24⇓~26][27]}. Based on the above analysis, WO₃ and graphene aerogel were composited to obtain secondary active sites, and the hydrogen spillover effect on them would effectively improve the HER performance of the system. Liu et al. Obtained a WO₃/ graphene aerogel (WGA) support by a solvothermal method, and then loaded Pt nanoparticles onto the support by electrochemical deposition to prepare a low content Pt（0.8 wt%）/WO₃/ graphene aerogel electrocatalyst (denoted as LPWGA) with both excellent HER activity and durability^[28]. The WGA carrier contains abundant oxygen vacancies and hierarchical pores, which can not only provide channels for continuous mass transfer and electron transfer, but also easily anchor Pt nanoparticles and adjust the electronic state of Pt surface, so that the LPWGA has high activity and durability. The electrochemical measurement results show that the LPWGA exhibits high HER activity and stability even when the Pt loading is as low as 0.81μg·cm^-2: the overpotential is 42 mV (at a current density of ：10 mA·cm^-2); The performance is stable after 10 000 times of continuous cyclic voltammetry and 40 H of chronopotentiometry. When the overpotential is 50 mV, the Tafel slope is 30 mV·dec^-1 and the TOF is 29.05 s^-1, which is much higher than the HER activity of commercial Pt/C and low-content Pt/graphene aerogel control samples.

It is well known that alloy catalysts composed of two (or more) metals can exhibit higher HER activity than monometallic catalysts^[29][23]. When the secondary active sites are all tungsten oxide, the HER performance of the system will be significantly improved by optimizing the primary active sites through alloying strategy, thereby enhancing the hydrogen spillover effect^[5,30⇓~32]. Qin et al. Prepared a WO₃ nanoarray-supported hollow PtCu alloy nanoparticle catalyst （PtCu/WO₃@CF） grown on Cu foil (CF) by a two-step method of hydrothermal and electrochemical activation^[33]. Electrochemical cyclic voltammetry results showed that PtCu/WO₃@CF exhibited a stronger hydrogen adsorption peak compared with WO₃@CF, indicating that the existence of PtCu alloy nanoparticles effectively promoted the overflow of H * on WO₃. Fitting the quantitative relationship between hydrogen adsorption peak and scan rate, it was found that the hydrogen adsorption kinetic slope of PtCu/WO₃@CF, （1.2×10^-3mV·dec^-1）, was lower than that of WO₃@CF（1.6×10^-3mV·dec^-1）, indicating that the existence of hydrogen spillover effect accelerated the adsorption of PtCu/WO₃@CF on H^*[23,34]. The mass activity of the obtained PtCu/WO₃@CF is 1.35 and 10.86 A·mg^-1 at overpotentials of 20 mV and 100 mV, respectively, which is 27 and 13 times higher than that of the commercial Pt/C electrode under the same conditions.

Oxygen-rich vacancy WO_3-x has a strong proton storage capacity, which can be used as a primary active site for hydrogen evolution: it is locally enriched with a certain concentration of protons, which can be easily transferred to the hydrogen-deficient secondary active sites, significantly increasing the hydrogen coverage on the secondary active sites and effectively improving the HER performance^[35]. The oxygen vacancy content of the WO₃ was controlled by hydrothermal, impregnation and high temperature thermal reduction methods, and the optimized catalyst system with oxygen vacancy content of 34. 3% was finally prepared by Wang's group (（Ru-WO_3-x/CP, CP refers to carbon paper)^[35]. The in situ Raman spectra showed that the peak intensity of Ru-H bond in Ru-WO_3-x/CP gradually increased when the applied cathode potential was gradually increased, which was obviously due to the overflow of H^* to Ru nanoparticles (NPs) WO_3-x, resulting in the increase of H coverage on Ru. DFT calculations further show that the WO_3-x surface has a lower energy barrier for H₂O dissociation than the Ru site, which is more favorable for H₂O dissociation. Therefore, it is reasonable to infer that in the Ru-WO_3-x/CP catalyzed HER process, H₂O is first adsorbed on WO_3-x, followed by dissociation to give H^*, which further overflows to the Ru site, and finally recombination desorption occurs to form hydrogen. Since the hydrogen flooding in traditional thermal catalysis is usually from the metal site to the oxide, while the directionality is just opposite in Ru-WO_3-x/CP catalysts, the above hydrogen flooding phenomenon in catalytic systems is called "reverse or retrograde hydrogen flooding". Based on this, the Ru-WO_3-x/CP catalyst showed excellent neutral hydrogen evolution performance. In 1.0 M phosphate-buffered saline solution (PBS solution), the Ru-WO_3-x/CP overpotential was 19 mV at a current density of 10 mA·cm^-2, which was 24 times higher than that of commercial Ru/C (86 mV) under the same conditions.

It is found that the electronic structure of the system can be regulated by introducing dopants to change the chemical composition of the tungsten oxide surface or constructing a two-phase heterointerface, which can promote the occurrence of hydrogen overflow and effectively improve the HER activity^[21,36]. Yoo et al. First treated W with H₂S/Ar plasma to obtain 1T-WS₂ nanocrystals^[37]; Then, after O₂ plasma treatment, amorphous tungsten oxide （a-WO₃） domains were introduced onto the above 1T-WS₂ to prepare a patched-structure 1T-WS₂/a-WO₃（ (WSO). The construction of this amorphous heterointerface is beneficial to shorten the proton diffusion path, so that hydrogen species can be quickly transferred to 1T-WS₂ through the channel provided by a-WO₃, that is, WSO shows a significant hydrogen overflow effect (Fig. 4), and then effectively enhances the HER activity. The optimized hybrid catalyst has an overpotential of 212 mV and a Tafel slope of 102.2 mV·dec^-1 at a current density of 100 mA·cm^-2 in an acidic environment.

In addition, the WO₃·2H₂O/WS₂ hybrid catalyst prepared by in-situ anodization of tungsten disulfide （WS₂） film also shows significant hydrogen spillover effect, in which protons and electrons can be embedded into the WO₃·2H₂O lattice through electrochromic effect under the action of an electric field.At the same time, hydrogen atoms doped in the lattice of WO₃·2H₂O can also be transported to the adjacent face defects or edge sites of WS₂ through electrochromic effect. The dynamics of H^* diffusion from the WO₃·2H₂O to the defect and edge sites of the WS₂ is theoretically studied. It is found that the overflowing hydrogen atoms are attracted to the positively and negatively charged W and S vacancies around the defect and edge sites by the charge electric field of these defects and sites.The charge generates a radiation-like field on the surface of WO₃ with the site as the center, which promotes the occurrence of hydrogen overflow effect in the HER process. The optimized WO₃·2H₂O/WS₂ hybrid catalyst exhibited excellent HER activity: an overpotential of 152 mV at a current density of 100 mA·cm^-2 and a Tafel slope of 54 mV·dec^-1[38].

In thermal catalysis, the TiO₂ is also often used as the secondary active site of hydrogen spillover. Under appropriate conditions, the "reverse charge transfer" of the TiO₂ to the metal loaded on it means that electrons are transferred from the TiO₂ to the metal active site and enriched around it, which makes the metal d-band center move down, which is conducive to the desorption of H^*, thus promoting the occurrence of hydrogen spillover effect^[39][40]. Lu Tongbu et al. Anchored Pt nanoclusters (Pt NCs) on porous TiO₂ nanosheets rich in oxygen vacancies, and its mass activity was 45.28 A·mg^-1（ in Pt) at − 0.1 V (vs RHE), which was 58.8 times higher than that of commercial Pt/C^[41]. The charge transfer from TiO₂ to Pt NCs was proved by the valence state analysis of Ti element in XPS, which proved the existence of reverse charge transfer. In situ Raman spectroscopy monitoring the structural changes of TiO₂ during HER process shows that the peak generated by the E_g（1） vibration mode in the V_O-Pt/TiO₂ rich sample has a blue shift, and the blue shift of the peak is more obvious as the HER reaction proceeds; When the reaction stops, the blue shift disappears. This peak blue shift is attributed to the overflow of H^* from Pt NCs to the TiO₂ support. In addition, the blue shift of the E_g（1） peak of the oxygen vacancy rich Pt/TiO₂ is more obvious than that of the oxygen deficient vacancy Pt/TiO₂ sample, indicating that more H^* is transferred from Pt NCs to the oxygen vacancy rich TiO₂ support. DFT calculations show that the reverse charge transfer (from TiO₂ to Pt NCs) in the oxygen-vacancy rich Pt/TiO₂ promotes the formation of electron-rich Pt NCs, effectively optimizing the d-band center. The projected density of States (PDOS) of Pt atoms near the Fermi level is further calculated, and it is found that the center of the d-band of Pt atoms in the Pt-O bridge shifts to the Fermi level by 0. 12 eV, which leads to the weakening of the adsorption strength of H^* and makes the hydrogen spillover effect from Pt NCs to TiO₂ support more significant^[5,42,43]. Electrochemical Impedance Spectroscopy (EIS) showed that the oxygen-rich vacancy Pt/TiO₂ exhibited a smaller charge transfer resistance than the oxygen-deficient vacancy Pt/TiO₂ due to the Electrochemical Impedance Spectroscopy spillover effect on the catalyst surface, which also confirmed the existence of hydrogen spillover to some extent. The application of TiO₂ as a secondary active site of hydrogen overflow to HER can significantly improve its hydrogen evolution effect.

NiO often acts as an effective primary active site for hydrogen spillover by foreign atom doping modification, which promotes the dissociation of water molecules under alkaline and neutral conditions and indirectly regulates the free energy of H^* adsorption^[44]. Song Bo et al. Prepared Mo-NiO/Ni by loading metal Mo on NiO/Ni, which weakened the adsorption strength of NiO on H^*, facilitated the migration of H^* from the primary active site Mo-NiO to the secondary active site Ni, accelerated its composite desorption, and improved the electrocatalytic performance^[45]. DFT calculations show that the strong adsorption of single NiO on H^* will hinder the transfer of NiO to active Ni sites^[41,46]; After doping Mo into NiO, the amorphous Mo-NiO plane provides more active sites for water dissociation, and it is unstable to the adsorption of H^*, which can lead to the migration of H^* to Ni, accelerate the Volmer step, and promote the occurrence of hydrogen spillover effect (shown in the lower inset of Fig. 5B)^[28,47]. The electrochemical test results show (Figure 5A) that Mo-NiO/Ni has an overpotential of 50 mV and a Tafel slope of 86 mV·dec^-1 at a current density of 10 mA·cm^-2.

The results show that the hydrogen spillover effect can be significantly enhanced by combining the catalytic active component with strong H₂O dissociation ability with the metal with weak hydrogen binding energy (such as Cu) to form a synergistic heterointerface, thus accelerating the formation of H₂^[48]. Nickel oxides often show strong water dissociation ability, and the additional OH^- modification (i.e., the formation of NiOOH) can further relieve their oxyphilicity, promoting the desorption of the OH^- generated in the Volmer step, thus providing a large amount of H^* for HER^{[49][6,50⇓~52]}. Lee et al. Grew NiO on copper foam, and then prepared a HO_M-NiO/Cu catalyst with a heterostructure by additional OH^- modification (Figure 6A)^[53]. Both theory and experiment confirm that HO_M-NiO and Cu act as the primary and secondary active sites for HER, respectively. It is found by theoretical calculation that the oxyphilic （ΔG_(OH)） of the HO-NiO/Cu model is significantly reduced, which means that the OH^- produced by the Volmer step is easily desorbed from it and promotes the dissociation of H₂O^[54]. In Fig. 6 B, the C_φ（ hydrogen overflow pseudocapacitance) reflects the adsorption and desorption phenomenon of H^* on the catalyst surface. The C_φ of each control sample was integrated over the overpotential η, and HO_M-NiO/Cu exhibited the largest integral value, indicating that HO_M-NiO/Cu had the largest hydrogen coverage, so the hydrogen overflow from the primary active site of HO_M-NiO to the secondary active site of Cu would proceed thermodynamically and spontaneously. HO_M-NiO/Cu was experimentally determined to exhibit excellent HER activity: an overpotential of 33 mV at 10 mA·cm^-2 current density and a Tafel slope of 51 mV·dec^-1, comparable to that of Pt/C (overpotential 39 mV, Tafel slope 46 mV·dec^-1）. This study confirms the important influence of primary active site oxyphilicity on H₂O dissociation, and provides guidance for further design of multi-active site HER electrocatalysts by hydrogen spillover effect.

RuO₂ is one of the most effective anode materials for oxygen evolution reaction (OER) in alkaline water electrolysis^[55,56]. RuO₂ itself is oxyphilic and has weak adsorption affinity for H^*, so RuO₂ is rarely used as cathode materials for water electrolysis, which also limits its application in HER. The results show that the heteroatom doping modification of RuO₂ (such as adding Ce) can effectively separate the RuO₂ phase, and the proton transfer channel can be formed between the new phase (such as CeO₂）) and the separated RuO₂ phase, which can improve the combination of H^* to some extent^[57]. Wang Feng and Zhang Zhengping first doped Ce into the lattice of RuO₂ to prepare RuCeO_x^[42]. However, it is not ideal to test the HER activity of RuCeO_x electrochemically. On the other hand, Mott-Schottky heterojunction is formed at the interface between RuO₂ and CeO₂ in RuCeO_x, and the RuO₂ phase and the CeO₂ phase act as electronic conductor and semiconductor, respectively^[58⇓~60]; Photogenerated electrons on CeO₂ under illumination will tend to be enriched on RuO₂. Based on this, they further used the photoactivation (PA) reduction method to selectively deposit Pt single atoms on the electron-rich RuO₂ phase in the RuCeO_x to prepare the Pt/RuCeO_x-PA, as shown in Figure 7A. This sample is significantly different from the control sample with random distribution of Pt obtained by chemical activation (CA) or Thermal activation (TA) reduction methods (noted as Pt/RuCeO_x-CA and Pt/RuCeO_x-TA）, respectively). An in-depth analysis of the possible reaction pathways was carried out by theoretical modeling of Pt/RuCeO_x-PA and calculation of the ΔG_H* for the adsorption of H^* on the possible neighboring sites. The calculated results show that the ΔG_H* of H^* adsorbed on Pt atom in Pt-O-Ru group (site A), O atom in Pt-O-Ru group (site B), RuO₂（110） plane (site C), Ce-O-Pt group, and CeO₂（111） plane are − 0.15, − 0.014, 0.23, 0.04, and − 0.34 eV, respectively. In general, the larger the negative absolute value of the ΔG_H* is, the stronger the H^* adsorption on the catalyst is; Similarly, the larger the positive absolute value of ΔG_H*, the easier the desorption of H^* from the catalyst surface^[61]. According to the ΔG_H* values obtained above, it is inferred that the overflow channel of the H^* on the Pt/RuCeO_x-PA should be “Pt-O-Ru-RuO₂”：, as shown in Figure 7B.The H^* is first stabilized on the A site, then gradually moves across to the B site, and then enriches on the C site, and finally realizes the formation and release of H₂. In this way, the Pt-O-Ru group and the RuO₂（110） crystal face act as the primary and secondary active sites, respectively, and the hydrogen spillover effect generated by the two will promote the rapid and effective HER. As predicted by the theory, the HER activity (including mass activity and specific activity) of Pt/RuCeO_x-PA is indeed higher than that of Pt/RuCeO_x-CA and Pt/RuCeO_x-TA, and even better than that of commercial Pt/C catalyst (20 wt%). EIS showed that compared with Pt/RuCeO_x-TA and Pt/RuCeO_x-CA, Pt/RuCeO_x-PA exhibited faster proton transfer kinetics and higher local H^* concentration, further confirming the promotion effect of hydrogen overflow on HER reaction kinetics experimentally.

As shown in Figure 8 a, Pt-based multi-component HER electrocatalysts usually have problems such as lengthy synthesis process, complex phase structure, high phase interface barrier, and long hydrogen overflow path^[62]. In view of the characteristics of no phase interface and short reaction path of the monobasic single-phase catalyst, if a Pt-containing single-phase catalyst can be designed, the H^* will experience strong adsorption (ΔG < 0), thermoneutral adsorption (ΔG ≈ 0) and desorption (ΔG > 0) successively by using the shortened hydrogen overflow channel between atomic-level neighbor multiple catalytic sites (Fig. 8 B), which will greatly improve the HER performance in acidic medium. Shao Zongping et al. Prepared a hexagonal oxide La₂Sr₂PtO_7+δ with a single phase structure by simple wet ball milling and high temperature calcination, and realized the use of hydrogen overflow channels between neighboring active sites to improve the HER performance of the catalyst at the atomic scale^[63]. In 0.5 M H₂SO₄ solution, the obtained La₂Sr₂PtO_7+δ has an overpotential of 13 mV at a current density of 10 mA·cm^-2 and a Tafel slope of 22 mV·dec^-1, and exhibits higher durability than the commercially available platinum black catalyst. DFT calculation result show that that adsorption of H^* on the La-Pt bridge site (I. e. "Pt-O-La") in the La₂Sr₂PtO_7+δ is almost thermoneutral (ΔG ≈ 0), which can effectively mediate hydrogen overflow; It is theoretically predicted that the unique cooperative mechanism of multiple catalytic sites participating in HER in La₂Sr₂PtO_7+δ is that ：H^* first enriches on the O site, then overflows to the adjacent La-Pt bridge site, and finally overflows on the Pt site to form H₂ by recombination desorption of H^*. On the other hand, H/D kinetic isotope effects (KIEs) can reflect the kinetic information of hydrogen transfer in chemical reactions: experimentally determined KIEs values greater than 1.5 are considered to be strong evidence for the involvement of hydrogen transfer in elementary reactions. The measured current density of La₂Sr₂PtO_7+δ in 0.5 M D₂SO₄/D₂O solution is about 2.2 – 2.6 times lower than that in 0.5 M H₂SO₄ solution (i.e., KIEs = 2.2 – 2.6, within the measured potential range), which proves experimentally that hydrogen overflow is involved in the elementary step of HER. This work theoretically and experimentally illustrates the hydrogen spillover synergy between atomically adjacent active sites in single-phase oxides, which provides a reference for the design and synthesis of high-performance HER catalysts in acidic media.