Boosting Z-scheme water splitting via increasing electron transport by manipulating multiple redox-active sites and potentials in metal hexacyanoferrate modifiers

Abstract

Metal hexacyanoferrates (MHCFs) are attractive cocatalysts for Z-scheme water splitting owing to their tunable redox properties at C-coordinated Fe^III/Fe^II sites, enabling efficient electron transfer from shuttle redox mediators. MHCFs can incorporate multiple transition metal centers through ambidentate cyanide ligands; however, rational design principles for utilizing redox-active metals at N-coordinated sites remain unclear. Herein, we show that photocatalytic H₂ evolution is governed by the relative alignment of redox potentials of MHCF modifiers and aqueous electron donors. A series of MHCFs comprising redox-active metals (i.e., Mn, Fe, Co, and Cu) and redox-inactive metals (i.e., In and Ni) were loaded onto Rh–Cr mixed-oxide-modified TaON, and their electrochemical properties were correlated with photocatalytic H₂ evolution using electron donors with different electron-donating abilities. MHCFs containing redox-active metal species with redox potential significantly more negative than those of electron donors (e.g., the N-coordinated Fe^III/II in FeHCF and Cu^II/I in CuHCF) suppressed photocatalytic activity due to backward electron transfer (reduction via photoexcited electrons), whereas MnHCF, with both Mn^III/II and Fe^III/II potentials more positive than donors, exhibited the highest activity. Furthermore, the incorporation of multiple metals into a Mn-based high-entropy MHCF (K₂Mn_0.4Fe_0.15Co_0.15Ni_0.15Cu_0.15[Fe(CN)₆]) improved durability while maintaining appropriate redox potentials, yielding a higher amount of gas evolution compared with MnHCF in Z-scheme water splitting. These findings provide a design strategy for multi-redox-active MHCF cocatalysts, highlighting that the optimal choice and tuning of metal sites can achieve both high efficiency and durability in water splitting systems.

---

Broader context

Water and sunlight are abundant and globally available resources, making solar water splitting an attractive route to sustainable hydrogen production. Among the proposed approaches, particulate Z-scheme systems are particularly appealing because they can enable scalable solar H₂ production using dispersed semiconductor materials and simple reactor configurations. However, their performance is often limited not only by the semiconductor photocatalysts themselves, but also by inefficient interfacial electron transfer between aqueous redox mediators and photocatalyst surfaces. Nevertheless, the design of cocatalysts for redox-mediator oxidation has received far less attention than that for water reduction or oxidation. In this work, we show that metal hexacyanoferrates can serve as tunable mediator-oxidation cocatalysts and that their redox potentials critically govern forward charge extraction and backward electron transfer. We further demonstrate that high-entropy compositional design improves durability while retaining function. Together, these findings establish a redox-potential-based design principle for mediator oxidation and provide a general framework for engineering interfacial redox selectivity and durability in solar-to-fuel conversion systems.

Introduction

Photocatalytic water splitting using particulate semiconductors is attracting attention for implementation in large-scale H₂ production.¹ A Z-scheme-type water splitting (Zs-WS) system can be promising from the perspective of effective utilization of visible light, which accounts for the majority of solar energy.² Zs-WS systems basically consist of three components: a H₂-evolving photocatalyst (HEP), an O₂-evolving photocatalyst (OEP), and a shuttle redox couple. The shuttle redox couple is expected to transfer excited electrons from the OEP to the HEP, as illustrated in Fig. 1.^2,3 Improving the overall efficiency of Zs-WS requires not only improvements in semiconductor bulk properties^4–6 but also promotion of four forward reactions (i.e., (1) reduction of water (or H⁺), (2) oxidation of reductant (Red) on the HEP, (3) reduction of oxidant (Ox), and (4) oxidation of water (or OH⁻) on the OEP), as well as suppression of two backward reactions (i.e., reduction of Ox on the HEP and oxidation of Red on the OEP). Among the four forward reactions, the promotion of H₂ evolution (1) and O₂ evolution (4) has been extensively studied and achieved by loading effective cocatalysts (e.g., Pt, CoO_x, and IrO₂) on the photocatalyst surfaces; most have been developed based on the knowledge accumulated from electrolysis.^7–9 Also, facet-selective deposition of such cocatalysts spatially separates reduction and oxidation sites, thereby promoting forward interfacial charge transfer while suppressing charge recombination.^9,10 Conversely, the promotion of the other two forward reactions with the redox couple (2) and (3) has not been sufficiently addressed, despite their importance in activating and/or boosting redox reactions on each photocatalyst and resulting Zs-WS.^9,11–17 Although previous studies revealed that cocatalysts such as PtO_x and RuO₂ (or their analogues) effectively promoted the reduction of Ox (e.g., IO₃⁻ and Fe³⁺), the effective promotion of oxidation of their Red (e.g., I⁻ and Fe²⁺) has seldom been reported.^12,17,18

Fig. 1 Schematic of Z-scheme water splitting (Zs-WS).

We have previously demonstrated that some metal hexacyanoferrates (MHCFs) function as effective cocatalysts for boosting the oxidation of such reductants and eventually promote H₂ evolution on various HEPs (e.g., CdS, ZnIn₂S₄, and TaON).^19–22 MHCFs comprise a three-dimensional cubic framework in which Fe_C and transition metals (M_N) are octahedrally coordinated to the carbon and nitrogen, respectively, of the cyanide ligand (Fig. 2(a)).^23,24 MHCFs offer structural tunability, enabling the modulation of various important physicochemical properties as a cocatalyst, such as the redox potential of Fe_C by varying M_N species.^24–27 The Fe^II_C species contained in MHCFs (e.g., InHCF and NiHCF) are oxidized by photogenerated holes, generating Fe^III_C species that effectively oxidize Red (i.e., electron donors) such as [Fe^II(CN)₆]⁴⁻ and Fe²⁺.^19,20,28 The driving force for the oxidation of electron donors can be tuned depending on the redox potential of Fe^III/II_C by varying the M_N.²⁸ This redox-based function of MHCFs promotes H₂ evolution and expands the choice of HEPs available for Zs-WS.¹⁵ Furthermore, introducing redox-active M_N (e.g., Mn^III/II, Fe^III/II, Co^III/II, and Cu^II/I), instead of inactive species (e.g., In^III and Ni^II), enables such MHCFs to exhibit multiple redox behaviors based on both Fe_C and M_N sites.^29–31 In addition, the design of MHCFs to incorporate multiple metal species into the M_N site generates the so-called high-entropy MHCF.^32–34 However, the attractive potential of such MHCFs with multiple redox species for photocatalytic water splitting has not been explored to date.

Fig. 2 Structure of (a) MHCF (K₂M^II[Fe^II(CN)₆]) and (b) V-POMs.

This study investigated how the redox properties of MHCF cocatalysts vary depending on the M_N species, with particular emphasis on the influence of redox-active M_N species on the promotion of oxidation, coupled with V-substituted polyoxometalates (V-POMs) as potential-tunable redox couples.^{19–22,35–38} Each MHCF was loaded onto an HEP (i.e., Rh–Cr mixed oxide (RCO)/TaON), and photocatalytic H₂ evolution was evaluated using two types of V-POMs with different redox potentials, [SiV^IVW₁₁O₄₀]⁶⁻ (SiV^IV) and [BV^IVW₁₁O₄₀]⁷⁻ (BV^IV), as an electron donor (Fig. 2(b)). The impact of MHCF modifiers on H₂ evolution, their stability during redox cycling, and the resultant Zs-WS were thoroughly examined.

Experimental

Details of the reagents are provided in the SI.

Synthesis of MHCFs

MHCFs were synthesized using a simple coprecipitation method.^19,28 An aqueous solution of metal chlorides (InCl₃·4H₂O, MnCl₂·4H₂O, FeCl₂·4H₂O, CoCl₂·6H₂O, NiCl₂·6H₂O, or CuCl₂·4H₂O; 1 M, 10 mL) was added to an aqueous solution of either K₄[Fe(CN)₆]·3H₂O or K₃[Fe(CN)₆] (0.5 M, 20 mL). The mixture was stirred for 5 min. The resulting precipitates were collected by centrifugation, washed thrice with Milli-Q water and once with methanol, and then dried under vacuum overnight at 308 K. Mixed MHCFs, such as K₂Mn_0.4Fe_0.15Co_0.15Ni_0.15Cu_0.15[Fe(CN)₆], were synthesized using the same procedure with a mixed metal chloride solution (1 M total, 10 mL).³³

Preparation of RCO/TaON

TaON powder was synthesized according to a previously reported method as follows.³⁹ A mixture of ZrO(NO₃)₂·2H₂O (0.6 mmol) and Ta₂O₅ (3.0 mmol) was calcined in air at 1073 K for 2 h, followed by calcination at 1123 K for 15 h under an NH₃ flow (12 mL min⁻¹). Characterization is provided in the SI (Fig. S1: XRD and S2: UV-vis diffuse reflectance spectrum). Rh–Cr mixed oxide particles (RCO) were loaded onto TaON by a photodeposition method.^19,40 TaON particles were first suspended in an aqueous methanol solution (20 vol%) containing the required amount of Na₃RhCl₆·nH₂O and K₂CrO₄ (1.0 and 1.5 wt%, calculated as Rh and Cr with respect to TaON, respectively). The suspension was irradiated with visible light (λ > 400 nm) from a 300 W Xe lamp (LX-300F, Cermax) equipped with a cutoff filter (HOYA, L-42), while Ar gas was purged at 50 mL min⁻¹ for 6 h. The resulting powder was collected by filtration and then dried under vacuum at 308 K overnight. The sample was denoted as RCO/TaON.

Modification of MHCFs on RCO/TaON

MHCFs were loaded onto RCO/TaON using the impregnation method.^19,28 MHCFs were dispersed in Milli-Q water (0.25 mmol/5 mL) and sonicated for 10 min, followed by stirring overnight to ensure complete dispersion. Thereafter, RCO/TaON particles were dispersed in the MHCF dispersion (Fe_C/Ta molar ratio = 0.1), and the mixture was stirred with a glass rod in an evaporation dish on a steam bath. The dried powder was heated at 373 K under an Ar flow (20 mL min⁻¹) for 1 h. The sample was denoted as MHCF/RCO/TaON.

Synthesis of (Fe, Ru)O_x/Bi₄TaO₈Cl

Bi₄TaO₈Cl was synthesized by a flux method.^4,5 A precursor mixture of Bi₂O₃, BiOCl, and Ta₂O₅ (molar ratio BiOCl : Bi₂O₃ : Ta₂O₅ = 1 : 1.5 : 0.5) was combined with a NaCl–CsCl mixed-flux (molar ratio NaCl : CsCl = 65 : 35 and flux : precursor = 5 : 95). The mixture was heated at 20 K h⁻¹ to 1023 K and maintained for 15 h and then cooled naturally. The resulting yellow solid was washed with 500 mL of Milli-Q water, filtered by vacuum filtration, and then washed thrice on the filter with Milli-Q water (100 mL), followed by drying for 2 days (Fig. S1: XRD and S2: UV-vis diffuse reflectance spectrum). FeO_x and RuO_x species were loaded onto Bi₄TaO₈Cl by the impregnation method.⁴¹ Aqueous solutions of FeCl₃·6H₂O (250 mM) and RuCl₃·nH₂O (50 mM) were added to Bi₄TaO₈Cl particles (10 mol% of each relative to Bi₄TaO₈Cl) with 2 mL of Milli-Q water. The suspension was evaporated on a steam bath, and the obtained dark-green powder was subsequently calcined at 573 K for 1 h under an Ar flow (20 mL min⁻¹). The resulting sample is denoted as (Fe, Ru)O_x/Bi₄TaO₈Cl.

Synthesis of V-substituted polyoxometalates

K₆[α-SiV^IVW₁₁O₄₀]·11H₂O was synthesized via K₈[α-SiW₁₁O₃₉]·13H₂O according to previously reported methods.^38,42 K₇[α-BV^IVW₁₁O₄₀]·7H₂O was synthesized according to a previously reported method.⁴³ These POMs were denoted as SiV^IV and BV^IV, respectively. The hydration numbers were determined by TG-DTA measurement. See the SI for characterization (Fig. S3: FT-IR and S4: cyclic voltammogram).

Characterization

The prepared samples were characterized by powder X-ray diffraction (XRD; Mini Flex II, Rigaku, X-ray source; Cu Kα) measurement, UV-vis diffuse reflectance spectroscopy (V-650, JASCO), scanning electron microscopy (SEM; NVision 40, Carl Zeiss-SIINT), and scanning transmission electron microscopy coupled with energy-dispersive X-ray spectroscopy (STEM-EDS; JEOL, JEM-ARM200F). The samples were also characterized by ATR-FTIR (ATR; ATR Pro One, JASCO, FT-IR; FT-4200, JASCO) using a diamond prism. ATR-FTIR spectra were recorded against air as a background. For the POM samples, pellets were prepared by mixing the sample with KBr, and measurements were performed using the transmission mode. The absorption spectra of the solutions containing each POM were measured using a UV-vis spectrometer (UV-1800, Shimadzu). The X-ray absorption fine structure (XAFS) measurements were performed at the BL12C beamlines of the Photon Factory (High Energy Accelerator Research Organization, Tsukuba, Japan). The X-ray absorption spectra were measured in transmission or fluorescence mode at room temperature using a Si(111) two-crystal monochromator.

Electrochemical measurement

The working electrode was prepared by coating the MHCF sample onto an FTO substrate. The sample powder was dispersed in a small amount of Milli-Q water or acetonitrile, and the dispersion was drop-cast onto the FTO substrate and wiped. This procedure was repeated once to ensure uniform coating. A three-electrode cell was employed, using an Ag/AgCl (3 M NaCl, BAS) as the reference electrode and a Pt coil as the counter electrode. Prior to measurements, the electrolyte solution was purged with Ar gas (100 mL min⁻¹). Potentials were controlled with a potentiostat (VersaSTAT 4, Princeton Applied Research).

Photocatalytic reaction

The photocatalytic reaction was evaluated in a Pyrex side-illuminated reaction vessel connected to a closed gas-circulation system. Visible light irradiation was provided by a 300 W Xe lamp equipped with an L-42 cutoff filter (λ > 400 nm). The reaction temperature was maintained at 288 K using a water bath. For the H₂ evolution half reaction, MHCF-modified or unmodified RCO/TaON powder (25 mg) was dispersed in aqueous KH₂PO₄ solution (0.05 M, 100 mL) containing SiV^IV or BV^IV as an electron donor (0.5 mM, 50 µmol). The suspension was thoroughly degassed and purged with Ar prior to irradiation. For the Zs-WS reaction, (Fe, Ru)O_x/Bi₄TaO₈Cl powder (50 mg) was added under the same conditions. The evolved H₂ and O₂ were quantified using a gas chromatograph (GC-8A, Shimadzu, TCD detector, MS 5A column, Ar carrier) directly connected to the closed gas-circulation system.

Results and discussion

Physicochemical properties of MHCFs

All prepared MHCFs (Fe_C=Fe^II and M_N = Mn^II, Fe^II, Co^II, Ni^II, Cu^II, and In^III), including Mn-based high-entropy hexacyanoferrate (HEMHCF), were confirmed to be of single-phase by XRD patterns (Fig. 3(a)). Most samples had a typical cubic structure, whereas MnHCF had a monoclinic phase that was slightly distorted from the cubic structure.^23,44,45 SEM images showed particle sizes of several tens of nanometers (Fig. S5), which aggregated into secondary particles with diameters of 100–200 nm. STEM-EDS mapping indicated that, after MHCF loading onto RCO/TaON, the MHCF was deposited on the RCO/TaON surface as a layer with a thickness of several tens of nanometers (Fig. S6). Fig. 3(b) shows ATR-FTIR spectra of reduced (Fe_C=Fe^II) and oxidized (Fe_C=Fe^III) forms. The observed IR absorption could be assigned to the C≡N stretching vibration of each MHCF. Apart from FeHCF, the oxidized forms of the MHCF exhibited absorption peaks at higher wavenumbers than their reduced counterparts, consistent with increased back-donation upon oxidation. Some MHCFs (M_N = Mn^II, Ni^II, and In^III) exhibited only weak light absorption in the visible-light region (Fig. S7), whereas those incorporating Fe^II, Cu^II, Co^II, and HEM exhibited strong absorption. Such absorption probably interfered with the absorption of the TaON photocatalyst (absorption edge at approximately 500 nm, Fig. S2) when MHCFs were loaded onto its surface.

Fig. 3 (a) XRD patterns of MHCFs (Fe_C=Fe^II) and (b) ATR-FTIR spectra of MHCFs (black lines: Fe_C=Fe^II and red lines: Fe_C=Fe^III). The XRD pattern was indexed with reference to the ICSD database.⁴⁶

Electrochemical properties of MHCFs

Fig. 4 shows the cyclic voltammograms (CVs) of InHCF and FeHCF as examples of MHCFs with redox-inactive M_N and redox-active M_N, respectively. CVs of other MHCFs are provided in Fig. S8. InHCF and NiHCF exhibited simple CVs with a single redox pair originating from the Fe^III/II_C (Fig. 4(a) and S8(b)). The simple CV behavior of InHCF and NiHCF was reasonable considering the stable, essentially redox-inactive valence states of In^III and Ni^II.^30,47 Conversely, the CVs of FeHCF (Prussian blue), CoHCF, and CuHCF exhibited two redox couples (Fig. 4(b), S8(c–f) and S9). These multiple redox behaviors were assigned to Fe^III/II_C and redox-active M_N^n+1/n+: in FeHCF, for example, the positive and negative redox couples correspond to Fe^III/II coordinated to nitrogen (Fe^III/II_N) and carbon (Fe^III/II_C), respectively.^48,49

Fig. 4 CVs of representative MHCFs with redox-inactive M_N ((a) InHCF) and redox-active M_N ((b) FeHCF) in 0.05 M KH₂PO₄ aqueous solution. Initial potential was set to the open rest potential, and scanning began in the anodic direction (50 mV s⁻¹).

Although previous reports revealed Mn in MnHCF to be redox-active and exhibit two distinct redox couples assigned to Mn^III/II and Fe^III/II_C in organic solution,^48,50 the CV of MnHCF measured in aqueous media (Fig. S8(c)) did not display two distinct redox couples. Instead, its voltammetric profile more closely resembled those of MnO₂ and MnCl₂ in water (Fig. S10(a)) than that of MnHCF in acetonitrile (Fig. S10(b) and (c)). These results suggest that the redox properties of MnHCF in an aqueous medium are irreversible due to disproportionation of Mn^III species generated during the anodic scan: 2Mn^III + 2H₂O → Mn^II + MnO₂ + 4H⁺.^51–54 Therefore, redox cycling of MnHCF has intrinsic instability when used in aqueous media; this phenomenon will be further discussed later. The redox potentials obtained by the CV in acetonitrile solution will be used hereafter, after appropriate conversion of potential to the SHE scale (Fig. S11).^30,55

HEMHCF (i.e., K₂Mn_0.4Fe_0.15Co_0.15Ni_0.15Cu_0.15[Fe(CN)₆]) displayed a single, well-defined redox couple (Fig. S12), despite the presence of multiple redox-active M_N species. This convergence of redox features is commonly attributed to a so-called cocktail effect, in which the redox processes at the Fe_C and the M_N species proceed concurrently.^33,34,56

Table 1 summarizes the half-wave potentials (E_1/2) of redox species in these MHCFs measured in 0.05 M KH₂PO₄ aqueous solution, except for that of MnHCF measured in 0.1 M KPF₆ acetonitrile solution. The assignment of redox couples for each MHCF was determined based on previous reports and additional experimental data provided in the SI.^48,57–60 Notably, all MHCFs displayed an irreversible reduction peak at approximately +0.1 V (vs. SHE, pH 4.5) to a greater or lesser extent (Fig. S8). These peaks likely arise from the reduction of dissolved [Fe^III(CN)₆]³⁻ as some of the oxidized MHCFs (e.g., Fe_C=Fe^III) are known to be partially soluble in water depending on the M_N species, which will be discussed in a later section.⁶¹

Table 1 Half-wave potentials (E_1/2) of MHCFs

MN	E1/2/V vs. SHE
	Fe^III/IIC	MN^n+1/n+
a Measured in 0.05 M KH2PO4 aqueous solution (pH 4.5).b Measured in 0.1 M KPF6 acetonitrile.
In^a	+0.87	Inactive
Mn^b	+0.86	+1.12
Fe^a	+1.03	+0.37
Co^a	+0.84	+0.62
Ni^a	+0.75	Inactive
Cu^a	+0.88	+0.24
HEM^a	+0.86	+0.86

---

Dependence of M_N species on the photocatalytic half reaction of H₂ evolution

Fig. 5(a) shows the time courses of H₂ evolution over various MHCF-modified RCO/TaON samples (10 mol% loading with respect to Fe_C relative to TaON), along with the unmodified RCO/TaON, in the presence of SiV^IV as an electron donor under visible-light irradiation. Except CuHCF, all MHCF modifications enhanced H₂ evolution relative to the unmodified RCO/TaON, on which the H₂ evolution rate gradually decreased, likely due to self-oxidation of TaON by photogenerated holes (2N³⁻ + 6h⁺ → N₂).¹⁵ For M_N = Mn, Co, In, and HEM, higher H₂ evolution rates were observed in the initial stage (from 0.5 to 1.5 h), attributable to efficient hole consumption via Fe^III/II_C and/or M_N^n+1/n+ redox cycles. Subsequently, H₂ evolution proceeded at a nearly steady rate, reaching approximately 25 µmol or higher, indicating that almost all the SiV^IV was oxidized to SiV^V. Specifically, the saturated amounts of H₂ on Mn and Co samples were obviously larger than 25 µmol, most probably due to the accumulation of oxidized species (i.e., Fe^III_C and/or M_Nⁿ⁺¹) within the MHCF by photogenerated holes, which provided excess electrons. This occurrence is supported by H₂ generation under similar conditions but in the absence of POMs (i.e., in pure water), as shown in Fig. 7(a), and will be discussed below. For MnHCF-modified RCO/TaON, the relationship between MnHCF loading amount and activity was investigated, and the rates showed little dependence on the loading amount over the range of 1–15 mol% (Fig. S13). Although Fe- and Ni-modified samples initially displayed H₂ evolution rates comparable to that of unmodified RCO/TaON, their performance clearly improved after 1 h, indicating partial consumption of photogenerated holes through the oxidation of Fe^II_C and/or M_Nⁿ⁺, which suppressed self-oxidation. By contrast, CuHCF modification produced a H₂ evolution profile similar to that of unmodified RCO/TaON.

Fig. 5 Time courses of H₂ evolution under visible-light irradiation over MHCF-modified or unmodified RCO/TaON in the presence of (a) 0.5 mM SiV^IV or (b) 0.5 mM BV^IV in aqueous KH₂PO₄ solution (50 mM KH₂PO₄, pH 4.5, 100 mL). Dotted lines show the amount of H₂ evolution corresponding to one electron donation by SiV^IV or BV^IV. (c) Rates of the H₂ evolution half reaction for each combination. (d) Schematic illustration of the half reaction of H₂ evolution over MHCF-modified RCO/TaON using POM as an electron donor and potential relationship between V-POMs and MHCFs.

Fig. 5(b) shows H₂ evolution in the presence of BV^IV as an electron donor. All MHCF-modified samples, including FeHCF and NiHCF systems, exhibited enhancement in H₂ evolution from the initial stage of irradiation. CuHCF suppressed the rate decrease observed for unmodified RCO/TaON, unlike the SiV^IV system, indicating a functional Fe^III/II_C redox cycle in this case, although the overall H₂ evolution enhancement remained limited.

Fig. 5(c) summarizes the maximum H₂ evolution rates obtained with both POMs. All MHCF-modified samples exhibited higher rates with BV^IV than with SiV^IV, except for CuHCF, consistent with the more negative potential of BV^V/IV (+0.40 V) compared with that of SiV^V/IV (+0.69 V), which afforded a larger driving force for electron transfer from the POM to the MHCF.²⁸

Relationship between the H₂ evolution rate and redox properties of MHCFs depending on M_N

Fig. 5(d) summarizes the redox potentials of MHCFs and POMs. Given the valence band maximum (VBM) of TaON (approximately +1.9 V vs. SHE at pH 4.5), the photogenerated holes could oxidize all the redox-active species in the MHCFs (Fe^II_C and M_N). For MHCFs containing redox-inactive M_N species (i.e., In and Ni), the observed H₂ evolution could be rationalized by the potential difference between Fe^III/II_C and POMs, with larger differences corresponding to higher H₂ evolution rates.²⁸ However, for MHCFs with redox-active M_N species (Mn, Co, Fe, and Cu), the H₂ evolution activity can be classified into three distinct types.

Type I (M_N = Mn, Co): Both MnHCF and CoHCF exhibit markedly enhanced H₂ evolution with both POMs, with higher rates observed for BV^IV than for SiV^IV, similar to the case of InHCF and NiHCF. In these materials, the redox-active species exhibit redox potentials more positive than those of either POM, thereby allowing electron injection from the reduced POMs to the oxidized MHCF species. Although the redox potential of Co^III/II in CoHCF (+0.62 V) is slightly more negative than that of SiV^V/IV (+0.69 V), electron transfer remains thermodynamically suitable when considering that E_1/2 values are defined under equimolar oxidized and reduced conditions. Overall, oxidation of the redox-active species (both Fe_C and M_N) by photogenerated holes, followed by electron injection from the POMs, provides a consistent explanation for the high H₂ evolution rates observed.

Type II (M_N = Fe): FeHCF markedly enhances H₂ evolution in the presence of BV^IV, whereas only a limited enhancement is observed with SiV^IV. FeHCF contains two redox-active species, Fe^III/II_C (+1.03 V) and Fe^III/II_N (+0.37 V). In the presence of BV^IV as an electron donor, both oxidized Fe species are thermodynamically reducible, resulting in a substantial enhancement of H₂ evolution. By contrast, SiV^IV enables the reduction of only Fe^III_C, whereas the Fe^III/II_N remains unreduced (Fig. 6(a)). The persistence of Fe^III_N facilitates electron trapping from photoexcited TaON, thereby promoting electron–hole recombination and limiting H₂ evolution (Fig. 6(b)). This occurs despite partial hole scavenging, which suppresses self-oxidation of the TaON.

Fig. 6 Proposed schematic of (a) the forward reaction where SiV^IV is oxidized via Fe^III/II_C in FeHCF, and (b) the backward reaction where the excited electrons and holes recombine via Fe^III/II_N in FeHCF.

Type III (M_N = Cu): In CuHCF, the Cu^II/I redox potential is more negative than that of BV^V/IV, causing photogenerated electrons in TaON to reduce Cu^II instead of producing H₂. Therefore, H₂ evolution was only slightly enhanced with either electron donor. These results for FeHCF and CuHCF highlight an important design principle for H₂ evolution using MHCF modifiers; i.e., the presence of redox species with potentials significantly more negative than that of the electron donor is detrimental.

As for the Mn-based high-entropy MHCF (HEMHCF; K₂Mn_0.4Fe_0.15Co_0.15Ni_0.15Cu_0.15[Fe(CN)₆]), the H₂ evolution rate with HEMHCF is similar to that with InHCF, which possesses a similar redox potential, in both cases with SiV^IV and BV^IV.

As mentioned above, several MHCFs (FeHCF, CoHCF, CuHCF, and HEMHCF) exhibit strong absorption in the visible region (Fig. S7), partially overlapping with that of TaON. Although the MHCFs are modified as thin layers on the RCO/TaON surface (Fig. S6), light shielding by the MHCFs may contribute to the apparent decrease in H₂ evolution activity.^19,28 However, because the observed activity trends (Fig. 5, S14 and S15) cannot be rationalized by differences in the extent of absorption, the light shielding effect is unlikely to be the dominant factor in this system.

Stability test of MHCFs during the H₂ evolution reaction

As suggested by the CVs in Fig. S10, MnHCF is partially susceptible to oxidative decomposition. To evaluate the structural stability of MHCFs, especially their oxidized forms, MHCF-modified RCO/TaON particles dispersed in Milli-Q water were irradiated with visible light. Structural changes were examined by comparing the C≡N stretching vibrations of the sample and [Fe^III(CN)₆]³⁻ formation in aqueous solution before and after irradiation. Under irradiation, all MHCFs produced 3–20 µmol of H₂ (Fig. 7(a)). For FeHCF, CoHCF, NiHCF, CuHCF, and HEMHCF modified RCO/TaON, the IR absorption peaks shifted to higher wavenumbers after irradiation (Fig. 7(b)), indicating the oxidation of Fe_C and/or M_N species by photogenerated holes (Fig. 3(b)). The decrease in the intensity of peaks corresponding to the reduced form suggests the formation of oxidized forms and partial dissolution and/or decomposition of MHCFs. For MnHCF, the disappearance of the IR absorption after irradiation indicated that the compound was completely dissolved and/or decomposed. This was also indicated by Fe K-edge X-ray absorption near edge structure (XANES) analysis (Fig. 7(c)), where the spectrum of MnHCF/RCO/TaON after irradiation differed from that of pristine MnHCF and was similar to that of Fe₂O₃. STEM-EDS elemental mapping of MnHCF/RCO/TaON after photoirradiation (Fig. 7(e)) revealed aggregation of Mn species on the RCO/TaON surface, as well as spatial separation of Mn and Fe species. This further suggests the decomposition of MnHCF. In addition, the absorption attributable to [Fe^III(CN)₆]³⁻ was observed in the solutions of all MHCF-modified samples except CuHCF; the strongest intensity was observed for MnHCF (Fig. 7(d)). Note that MnHCF/RCO/TaON evolved a larger amount of H₂ (approximately 20 µmol) than the other MHCF-modified samples (Fig. 7(a)), indicating that both Fe^II_C and Mn^II species were oxidized.

Fig. 7 (a) Time course of H₂ evolution under visible-light irradiation over MHCF-modified or unmodified RCO/TaON in Milli-Q water. (b) Normalized ATR-FTIR spectra of MHCF/RCO/TaON before and after irradiation. (c) Fe K-edge XANES of MHCF/RCO/TaON after irradiation, together with the reference samples. (d) UV-vis absorption spectra of the reaction solution after irradiation. STEM-EDS elemental mapping images of (e) MnHCF/RCO/TaON and (f) HEMHCF/RCO/TaON before and after irradiation in pure water. The distributions of Mn, Fe, and Ta are shown in red, green, and blue, respectively.

Stirring MHCF particles with reduced (Fe^II_C) or oxidized (Fe^III_C) species in aqueous solution in the dark for 20 min resulted in increased absorption assigned to the iron cyanide complex for all MHCFs to varying extents (Fig. S16). Most MHCFs displayed higher solubility of the oxidized forms (Fe^III_C) than that of the reduced forms (Fe^II_C),⁶¹ and this tendency was particularly pronounced for MnHCF. These results indicate that MnHCF is susceptible to oxidative dissolution and/or decomposition, consistent with the substantial formation of [Fe^III(CN)₆]³⁻ under photoirradiation (Fig. 7(d)). This limited durability is also consistent with previous reports showing that Mn^III-induced Jahn–Teller distortion and related structural phase transitions during redox cycling reduce redox reversibility.^33,55 Assuming that MnO_x species are formed through MnHCF decomposition, we examined whether these MnO_x species contribute to the H₂ evolution activity. In the presence of SiV^IV as an electron donor, MnO_x/RCO/TaON exhibited a lower H₂ evolution rate than unmodified RCO/TaON (Fig. S17), indicating that MnO_x species are not responsible for the enhanced activity of MnHCF/RCO/TaON.

Conversely, the amount of [Fe^III(CN)₆]³⁻ after irradiation was lower for HEMHCF than that for MnHCF (Fig. 7(d)), consistent with the lower solubility of oxidized HEMHCF compared with that of MnHCF (Fig. S16(b)). The IR absorption corresponding to the cyanide ligands of HEMHCF shifted to higher wavenumbers (Fig. 7(b)), indicating the occurrence of oxidation. Furthermore, the XANES spectrum of HEMHCF/RCO/TaON after photoirradiation was identical to that of K₃[Fe^III(CN)₆] (Fig. 7(c)). STEM-EDS elemental mapping (Fig. 7(f)) further showed that Mn and Fe species were closely located. These results indicate that the structure of the surface-loaded HEMHCF is largely retained after photoirradiation. The amount of H₂ evolved from HEMHCF/RCO/TaON was approximately 10 µmol (Fig. 7(a)), which reasonably corresponded to the total amount of electron-donating species in HEMHCF (Mn^II, Fe^II, and Co^II; estimated to be approximately 17 µmol in total). These results suggest that incorporation into the high-entropy structure enhances the durability of HEMHCF against decomposition. The enhanced durability of HEMHCF is consistent with previous reports, where partial substitution of Mn in Mn-based hexacyanoferrates mitigates Mn^III-induced structural changes and improves redox reversibility.³³

Zs-WS using MHCF-modified RCO/TaON as the HEP

Finally, MHCF/RCO/TaON was coupled with an OEP ((Fe,Ru)O_x/Bi₄TaO₈Cl)⁴¹ for Zs-WS using SiV^V/IV as a shuttle redox mediator (Fig. 8(a)). TaON without the MHCF rapidly deactivated, certainly due to the self-oxidation.¹⁵ By contrast, MHCF-modified RCO/TaON initially exhibited relatively steady H₂ and O₂ evolution, indicating that Fe^II_C and/or redox-active M_N species scavenged photogenerated holes and promoted SiV^IV oxidation (Fig. 8(a)). Among these, MnHCF/RCO/TaON displayed relatively higher gas evolution rates during the first 5 h than the other MHCFs (Fig. 8(b)). However, the gas evolution rates gradually decreased after the initial 5 h. A comparable decrease was observed for the other MHCF-modified samples, apart from HEMHCF, most likely due to dissolution and/or oxidative decomposition under the reaction conditions. By contrast, HEMHCF/RCO/TaON maintained its activity over a longer period (Fig. 8(c)), demonstrating that the combination of its oxidative durability and its ability to efficiently oxidize the redox mediator enhanced hole consumption and effectively suppressed photocatalyst deactivation.

Fig. 8 (a) Time courses of H₂ and O₂ evolution over 25 mg of H₂-evolving photocatalyst (MHCF/RCO/TaON or RCO/TaON) and 50 mg of O₂-evolving photocatalyst ((Fe, Ru)O_x/Bi₄TaO₈Cl) in the presence of 0.5 mM SiV^IV in aqueous KH₂PO₄ solution (50 mM, pH 4.5). Rates of H₂ evolution during the (b) first 5 h and (c) 15–20 h.

Conclusions

This study classified the effects of M_N species in MHCFs on hole extraction from a HEP (RCO/TaON) in Zs-WS. MHCFs with redox-inactive M_N species (i.e., In and Ni) exhibited a simple Fe^III/II_C-mediated behavior, in which enhancement of H₂ evolution scaled with the potential difference between the POM electron donor and the Fe_C redox couple. MHCFs with redox-active M_N species could be further classified into three types. Type I included MHCFs with M_N = Mn or Co, which comprised beneficial dual redox sites (Fe_C and M_N), both of whose potentials were more positive than those of the POM electron donors. The M_N species of this type, as well as Fe_C, could participate in hole capture, regeneration by the electron donor, and a marked enhancement of H₂ evolution. Type II included FeHCF, where enhancement in H₂ evolution was observed only with a sufficiently reducing donor (BV^IV), because both Fe_C and Fe_N sites could participate in forward redox cycling. By contrast, with a weaker donor (SiV^IV), the Fe_N site remained oxidized and induced backward electron transfer, suppressing the apparent activity. Type III included CuHCF, where the redox potential of Cu^II/I was much more negative than that of electron donors and preferentially traps photogenerated electrons, resulting in negligible improvement in H₂ evolution. Overall, redox-active M_N species could promote oxidation of the redox mediator only when the redox potential of M_N was more positive than that of the electron donor.

Among the tested MHCFs, MnHCF exhibited the highest H₂ evolution activity but suffered from oxidative dissolution and/or decomposition, causing a marked decay in activity during the Zs-WS reaction. Accelerating electron donation from the redox couple should be an effective strategy to prevent dissolution of MHCFs in their highly oxidized states. In addition, high-entropy design was shown to be a promising approach for fundamentally improving the oxidative stability. An Mn-based high-entropy MHCF incorporating multiple M_N species stabilized Mn redox cycling, leading to minimized decay in Zs-WS activity. These results demonstrate that high-entropy engineering of the cocatalyst enables the enhancement of both activity and durability for long-term Zs-WS systems.

Author contributions

Harutaka Ninomiya: conceptualization; data curation; formal analysis; investigation; writing – original draft. Osamu Tomita: funding acquisition; investigation; writing – review and editing. Hajime Suzuki: investigation; writing – review and editing. Akinobu Nakada: investigation; writing – review and editing. Rie Haruki: data curation; writing – review and editing. Shunsuke Nozawa: data curation; writing – review and editing. Ryu Abe: funding acquisition; methodology; project administration; supervision; writing – review and editing.

Conflicts of interest

There are no conflicts to declare.

Data availability

Data are available within the article or its supplementary information (SI). Supplementary information: X-ray diffraction patterns, UV-vis diffuse reflectance spectra, FT-IR spectra, cyclic voltammograms, SEM images, STEM-EDS elemental mappings, UV-vis absorption spectra, and photocatalytic activity data. See DOI: https://doi.org/10.1039/d6el00066e.

Acknowledgements

This work was supported by “JSPS KAKENHI” (JP24K01528, JP25K01877, and JP25KJ1452), the “Tokyo Ohka Foundation for the Promotion of Science and Technology” and the “Advanced Research Infrastructure for Materials and Nanotechnology in Japan (ARIM)” of the Ministry of Education, Culture, Sports, Science and Technology (MEXT) (JPMXP1226KT0011). XAFS experiments were performed with the approval of the Photon Factory Program Advisory Committee (Proposal No. 2024G638 and 2024G147). The authors are indebted to the technical division of the Institute for Catalysis, Hokkaido University, for their help in building the experimental equipment. The crystal structures were illustrated using the VESTA program.⁶²

References

1. B. A. Pinaud, J. D. Benck, L. C. Seitz, A. J. Forman, Z. Chen, T. G. Deutsch, B. D. James, K. N. Baum, G. N. Baum, S. Ardo, H. Wang, E. Miller and T. F. Jaramillo, Energy Environ. Sci., 2013, 6, 1983.
2. Y. Wang, H. Suzuki, J. Xie, O. Tomita, D. J. Martin, M. Higashi, D. Kong, R. Abe and J. Tang, Chem. Rev., 2018, 118, 5201–5241.
3. R. Abe, K. Sayama, K. Domen and H. Arakawa, Chem. Phys. Lett., 2001, 344, 339–344.
4. S. Chang, L. Shi, J. Yu, R. Wang, X. Xu and G. Liu, Appl. Catal. B Environ., 2023, 328, 122541.
5. K. Ogawa, A. Nakada, H. Suzuki, O. Tomita, M. Higashi, A. Saeki, H. Kageyama and R. Abe, ACS Appl. Mater. Interfaces, 2019, 11, 5642–5650.
6. C. Zhang, X. Liang, W. Xing, K. Liu, Z. Zheng, S. Wang, G. Zhang and X. Wang, J. Am. Chem. Soc., 2026, 148, 17272–17282.
7. J. Yang, D. Wang, H. Han and C. Li, Acc. Chem. Res., 2013, 46, 1900–1909.
8. M. G. Walter, E. L. Warren, J. R. McKone, S. W. Boettcher, Q. Mi, E. A. Santori and N. S. Lewis, Chem. Rev., 2010, 110, 6446–6473.
9. T. Takata, J. Jiang, Y. Sakata, M. Nakabayashi, N. Shibata, V. Nandal, K. Seki, T. Hisatomi and K. Domen, Nature, 2020, 581, 411–414.
10. Q. Wang, D. Zheng, Z. Pan, W. Xing, S. Wang, Y. Hou, M. Anpo and G. Zhang, Adv. Funct. Mater., 2025, 35, 2501889.
11. R. Abe, M. Higashi and K. Domen, ChemSusChem, 2011, 4, 228–237.
12. K. Murofushi, K. Ogawa, H. Suzuki, R. Sakamoto, O. Tomita, K. Kato, A. Yamakata, A. Saeki and R. Abe, J. Mater. Chem. A, 2021, 9, 11718–11725.
13. A. Iwase, H. Kato and A. Kudo, Chem. Lett., 2005, 34, 946–947.
14. M. Qureshi, T. Shinagawa, N. Tsiapis and K. Takanabe, ACS Sustainable Chem. Eng., 2017, 5, 8079–8088.
15. R. Abe, M. Higashi and K. Domen, J. Am. Chem. Soc., 2010, 132, 11828–11829.
16. A. Harriman, I. J. Pickering, J. M. Thomas and P. A. Christensen, J. Chem. Soc., Faraday Trans., 1988, 1(84), 2795.
17. Y. Iwase, O. Tomita, M. Higashi, A. Nakada and R. Abe, Sustain. Energy Fuels, 2019, 3, 1501–1508.
18. Y. Iwase, O. Tomita, M. Higashi and R. Abe, Sustain. Energy Fuels, 2017, 1, 748–754.
19. H. Matsuoka, T. Inoue, H. Suzuki, O. Tomita, S. Nozawa, A. Nakada and R. Abe, Sol. RRL, 2023, 7, 2300431.
20. T. Shirakawa, M. Higashi, O. Tomita and R. Abe, Sustain. Energy Fuels, 2017, 1, 1065–1073.
21. Y. Kageshima, Y. Gomyo, H. Matsuoka, H. Inuzuka, H. Suzuki, R. Abe, K. Teshima, K. Domen and H. Nishikiori, ACS Catal., 2021, 11, 8004–8014.
22. H. Matsuoka, M. Higashi, A. Nakada, O. Tomita and R. Abe, Chem. Lett., 2018, 47, 941–944.
23. J. F. Keggin and F. D. Miles, Nature, 1936, 137, 577–578.
24. F. Scholz and A. Dostal, Angew Chem. Int. Ed. Engl., 1996, 34, 2685–2687.
25. A. Gotoh, H. Uchida, M. Ishizaki, T. Satoh, S. Kaga, S. Okamoto, M. Ohta, M. Sakamoto, T. Kawamoto, H. Tanaka, M. Tokumoto, S. Hara, H. Shiozaki, M. Yamada, M. Miyake and M. Kurihara, Nanotechnology, 2007, 18, 345609.
26. E. Kholoud, H. Watanabe, A. Takahashi, M. M. Emara, B. A. Abd-El-Nabey, M. Kurihara, K. Tajima and T. Kawamoto, J. Mater. Chem. C, 2017, 5, 8921–8926.
27. M. Zentkova and M. Mihalik, Crystals, 2019, 9, 112.
28. H. Ninomiya, O. Tomita, H. Suzuki, A. Nakada and R. Abe, ACS Mater. Lett., 2024, 6, 2474–2478.
29. K. Hurlbutt, S. Wheeler, I. Capone and M. Pasta, Joule, 2018, 2, 1950–1960.
30. Y. Lu, L. Wang, J. Cheng and J. B. Goodenough, Chem. Commun., 2012, 48, 6544–6546.
31. L. Wang, Y. Lu, J. Liu, M. Xu, J. Cheng, D. Zhang and J. B. Goodenough, Angew. Chem., Int. Ed., 2013, 52, 1964–1967.
32. B. Liu, Q. Zhang, L. Zhang, X. Yong, L. Li and C. Wang, Energy Storage Mater., 2024, 66, 103221.
33. Y. Ma, Y. Hu, Y. Pramudya, T. Diemant, Q. Wang, D. Goonetilleke, Y. Tang, B. Zhou, H. Hahn, W. Wenzel, M. Fichtner, Y. Ma, B. Breitung and T. Brezesinski, Adv. Funct. Mater., 2022, 32, 2202372.
34. Y. Ma, Y. Ma, S. L. Dreyer, Q. Wang, K. Wang, D. Goonetilleke, A. Omar, D. Mikhailova, H. Hahn, B. Breitung and T. Brezesinski, Adv. Mater., 2021, 33, 2101342.
35. O. Tomita, H. Naito, A. Nakada, M. Higashi and R. Abe, Sustain. Energy Fuels, 2022, 6, 664–673.
36. Y. Iwase, O. Tomita, H. Naito, M. Higashi and R. Abe, J. Photochem. Photobiol., A, 2018, 356, 347–354.
37. K. Tsuji, O. Tomita, M. Higashi and R. Abe, ChemSusChem, 2016, 9, 2201–2208.
38. O. Tomita, K. Tachizawa, T. Kido, H. Suzuki, A. Nakada and R. Abe, Adv. Funct. Mater., 2025, 35, 2505908.
39. K. Maeda, H. Terashima, K. Kase, M. Higashi, M. Tabata and K. Domen, Bull. Chem. Soc. Jpn., 2008, 81, 927–937.
40. K. Maeda, D. Lu, K. Teramura and K. Domen, Energy Environ. Sci., 2010, 3, 470–477.
41. A. Nakada, H. Suzuki, J. J. M. Vequizo, K. Ogawa, M. Higashi, A. Saeki, A. Yamakata, H. Kageyama and R. Abe, ACS Appl. Mater. Interfaces, 2019, 11, 45606–45611.
42. A. Tézé, G. Hervé, R. G. Finke and D. K. Lyon, in Inorganic Syntheses, John Wiley & Sons, Ltd, 1990, pp. 85–96.
43. J. J. Altenau, M. T. Pope, R. A. Prados and H. So, Inorg. Chem., 1975, 14, 417–421.
44. L. Li, Z. Hu, Y. Lu, C. Wang, Q. Zhang, S. Zhao, J. Peng, K. Zhang, S.-L. Chou and J. Chen, Angew. Chem., Int. Ed., 2021, 60, 13050–13056.
45. M. Fiore, S. Wheeler, K. Hurlbutt, I. Capone, J. Fawdon, R. Ruffo and M. Pasta, Chem. Mater., 2020, 32, 7653–7661.
46. D. Zagorac, H. Müller, S. Ruehl, J. Zagorac and S. Rehme, J. Appl. Crystallogr., 2019, 52, 918–925.
47. L. Chen, H. Shao, X. Zhou, G. Liu, J. Jiang and Z. Liu, Nat. Commun., 2016, 7, 11982.
48. B. Wang, Y. Han, X. Wang, N. Bahlawane, H. Pan, M. Yan and Y. Jiang, iScience, 2018, 3, 110–133.
49. D. Ellis, M. Eckhoff and V. D. Neff, J. Phys. Chem., 1981, 85, 1225–1231.
50. J. Song, L. Wang, Y. Lu, J. Liu, B. Guo, P. Xiao, J.-J. Lee, X.-Q. Yang, G. Henkelman and J. B. Goodenough, J. Am. Chem. Soc., 2015, 137, 2658–2664.
51. R. Benedek, J. Phys. Chem. C, 2017, 121, 22049–22053.
52. J. C. Hunter, J. Solid State Chem., 1981, 39, 142–147.
53. H. N. Wilson and J. G. M. Bremner, Q. Rev. Chem. Soc., 1948, 2, 1–24.
54. H. G. Dion and P. J. G. Mann, J. Agric. Sci., 1946, 36, 239–245.
55. X. Bie, K. Kubota, T. Hosaka, K. Chihara and S. Komaba, J. Mater. Chem. A, 2017, 5, 4325–4330.
56. C. Li, Y. Xiao, D. Zhang, X. Yuan, J. Xiao, Y. Zhao, H. Gao and H. Liu, Batteries, 2025, 11, 328.
57. D. R. Shankaran and S. S. Narayanan, Fresenius. J. Anal. Chem., 1999, 364, 686–689.
58. O. Makowski, J. Stroka, P. J. Kulesza, M. A. Malik and Z. Galus, J. Electroanal. Chem., 2002, 532, 157–164.
59. M. Giorgetti, L. Guadagnini, D. Tonelli, M. Minicucci and G. Aquilanti, Phys. Chem. Chem. Phys., 2012, 14, 5527.
60. L. Guadagnini, D. Tonelli and M. Giorgetti, Electrochim. Acta, 2010, 55, 5036–5039.
61. X. Lamprecht, F. Speck, P. Marzak, S. Cherevko and A. S. Bandarenka, ACS Appl. Mater. Interfaces, 2022, 14, 3515–3525.
62. K. Momma and F. Izumi, J. Appl. Crystallogr., 2011, 44, 1272–1276.

---