The oxygen evolution reaction (OER) is the key half-reaction for many renewable energy conversion and storage technologies, including electrocatalytic water splitting and CO₂ reduction. The sluggish kinetics of the OER have long been the bottleneck hindering the widespread use of such technologies [1-3]. Ru- and Ir-based materials are the state-of-the-art OER catalysts, but they suffer from disadvantages such as scarcity, high cost, and low intrinsic activity [4, 5]. In the past decades, the development of all-inorganic and earth-abundant bi-/tri-metallic hydr(oxy)oxides as OER catalysts for alkaline media (metal oxides [6, 7], hydr(oxy)oxides [8, 9], cobalt-phosphate composites [10, 11], etc.) has been successful, and these materials exhibit excellent activity. Unfortunately, the lack of in-depth knowledge about the relationship between their atomically precise structures and the origin of their enhanced performance has hindered further design of active, stable, and cost-effective OER electrocatalysts. Metal-organic frameworks (MOFs), which have a well-defined atomic architecture, could per se be used to reveal the structure-property relationship at the molecular level [12, 13]. However, MOFs are composed of metal ions (or clusters) and organic linkers, and therefore, are less suitable as model systems for hydr(oxy)oxides.

Polyoxometalates (POMs) constitute a class of well-defined metal-oxo clusters. They feature well-defined and unique structural motifs, and can be seen as an intermediate between molecular complexes and extended oxides [14]. In addition, all-inorganic, oxidatively-resistant, and earth-abundant POMs with excellent OER activity have been produced [15-26]. Therefore, POMs are ideal candidates for a molecular model of bi-/tri-metallic hydr(oxy)oxides to explore the origin of the improved catalytic activity in the OER and to gain a clearer insight into the structure-property correlation.

For the OER, acidic media are preferred because of numerous advantages over alkaline media, such as high energy efficiency, low Ohmic losses, and easy product separation or carbonation [23, 27]. Unfortunately, earth-abundant OER electrocatalysts are unstable toward dissolution in acidic media and are typically inferior to noble-metal-based catalysts [1, 23, 28]. Developing highly active and stable earth-abundant OER catalysts in acidic media remains a great challenge in this area. Herein we show that by substitution of a parent POM cluster, {Co₄(OH)₃(PO₄)}₄(SiW₉O₃₄)₄]^32- (1), with Fe into two isostructurally analogous POM clusters [{Fe₂Co₂(OH)₃PO₄}₄(SiW₉O₃₄)₄]^24- (2) and [{FeCo₃(OH)₃PO₄}₄(SiW₉O₃₄)₄]^28- (3), we can achieve significantly enhanced OER performance. The insoluble barium salt of 3 (Ba[3]) is the best catalyst, with an overpotential of 385 mV at 10 mA cm^-2 in 0.5 M H₂SO₄; this overpotential is 66 mV lower than that of the barium salt of 1 (Ba[1]) and comparable to that of the commercial IrO₂ catalyst under similar conditions (393 mV). These catalysts exhibit negligible activity decay and retain their structural integrity even after 2000 cyclic voltammetry (CV) cycles and 24 h of long-term electrolysis.

The synthetic routes to 1, 2, and 3 and the structures of these compounds are shown in Fig. 1. Single-crystal X-ray diffraction (SCXRD) analysis confirms their isostructural characteristics [18, 29]. The water-insoluble barium salts of Ba₁₆[{Co₄(OH)₃(PO₄)}₄(SiW₉O₃₄)₄] (Ba[1]), Ba₁₂[{Fe₂Co₂(OH)₃ PO₄}₄(SiW₉O₃₄)₄] (Ba[2]), and Ba₁₄[{FeCo₃(OH)₃PO₄}₄ (SiW₉O₃₄)₄] (Ba[3]) were prepared by simple metathesis of the corresponding countercations Na⁺ or K⁺ into Ba²⁺. The FT-IR bands provide a reliable fingerprint for the POM structures [23, 30, 31]. The presence of 1 in Ba[1], 2 in Ba[2], and 3 in Ba[3] was confirmed by FT-IR spectroscopy (Fig. S1 in SI). Besides, the corresponding signature FT-IR spectra of Fe-substituted Ba[2] and Ba[3] were consistent with that of the parent Ba[1], due to the isostructural nature of 1, 2, and 3.

The OER activities of Ba[1], Ba[2], and Ba[3] were measured using a typical three-electrode system in O₂-saturated 0.5 mol L^-1 H₂SO₄ solution at 25 ℃. Linear sweep voltammetry (LSV) experiments were performed at a scan rate of 5 mV s^-1 after the system reached equilibrium. The electrodes were prepared by mixing Ba[1], Ba[2], or Ba[3] with commercial carbon paste oil (CPO) at 40% content by weight [23]. As shown in Figure 2a, both Ba[2]/CPO and Ba[3]/CPO electrodes showed significantly enhanced electrocatalytic activity as compared with the Fe-free Ba[1]/CPO electrode. Ba[2]/CPO showed an overpotential of 406 mV at 10 mA cm^-2 (Table 1), which was 45 mV smaller than that of Ba[1]/CPO (451 mV). With a decrease in the Fe substitution ratio, the corresponding overpotential decreased to 385 mV for Ba[3]/CPO (66 mV smaller than that of Ba[1]/CPO). Because of the isostructural nature of 1, 2, and 3, the enhanced OER catalysis was attributed solely to the Fe substitution, which gives us a well-defined model for substitution effects on the catalytic enhancement. It is also noteworthy that excess Fe, as in the case of Ba[2], can lead to an increase in the OER overpotential. These results are consistent with those of previous studies using W₂₃Co_wFe_xO_y(OH)_z transition-metal clusters (TMCs) [29], where the lower Fe content favored low adsorption energies (ΔG_O – ΔG_OH > 1.6 eV) for the OER intermediates, while a larger Fe content was associated with large adsorption energies (ΔG_O – ΔG_OH < 1.6 eV). In this case, an optimum catalyst would exhibit a free-energy difference (ΔG_O – ΔG_OH) of around 1.6 eV, resulting in the well-known volcano-like behavior [32, 33].

A modified electrode of 40% IrO₂/CPO was prepared for comparison, and its OER performance was examined under the same conditions. At 10 mA cm^-2, the IrO₂/CPO electrode produced an overpotential of 393 mV (Fig. 2b and Table 1), which was 8 mV higher than that of Ba[3]/CPO. In 1.0 M H₂SO₄, the Ba[3]/CPO electrode generated an overpotential of 398 mV at 10 mA cm^-2 (Table 1 and Fig. S2 in SI), which is comparable to the remarkable example of the Ba[Co-POM] catalyst [23]. To the best of our knowledge, Ba[3] is also one of the best non-noble metal-based OER electrocatalysts in acidic media (Table 1).

To gain better insight into the superior performance of Ba[3]/CPO, we estimated the available active sites in the modified electrodes by evaluating the total amounts of Fe and Co, assuming that both are active during the OER. This approach imposes an upper limit on the available active sites. In comparison, the IrO₂/CPO mixture contains almost 3.4 times more active sites than does the corresponding Ba[3]/CPO mixture (Table S1 in SI). For comparison, a modified electrode of 12% IrO₂/CPO with a similar mass of active sites as 40% Ba[3]/CPO was prepared, and its OER performance was examined under the same conditions. As shown in Table S1 and Fig. S3, 12% IrO₂/CPO exhibits an overpotential of 528 mV at 10 mA cm^-2, which is greater than that of Ba[1]/CPO (385 mV) by 143 mV. These results suggest that Ba[3] may show better intrinsic activity than the state-of-the-art IrO₂ at 10 mA cm^-2 in acidic media, but its activity might be limited by other parameters such as electron transfer, as indicated by the magnitude of the Tafel slope. The Tafel slopes of all the catalysts are depicted in Fig. S4 in SI. Ba[1] (194 mV dec^-1) and Ba[2] (189 mV dec^-1) exhibit a larger Tafel slope than does Ba[3], suggesting that the OER is kinetically favored on Ba[3] (181 mV dec^-1). Similar large Tafel slopes have been reported for other POM-based OER electrocatalysts [23-25], possibly due to the competition between a chemical step and an electron-transfer limiting step [23]. In the case of IrO₂, a smaller Tafel slope was observed (96 mV dec^-1).

The operational stability of the Ba[1], Ba[2], and Ba[3] catalysts was evaluated by a long-term cycling test in 0.5 M aqueous H₂SO₄ solution. Satisfactorily, the LSV curves after 2000 CV cycles practically overlapped with the original ones, displaying a very small change in overpotential (< 8 mV) at 10 mA cm^–2 (Fig. 3a and Figs. S5 and S6 in SI). Chronoamperometric response (i-t) tests further showed that all the three catalysts maintained their activity for at least 24 h of electrolysis (Fig. 3b and Figs. S7 and S8 in SI). In particular, the LSV curves of Ba[3]/CPO showed only a minimal overpotential change (< 25 mV, Fig. 3c) at 10 mA cm^–2 after 24 h of electrolysis.

To evaluate the structural integrity under operation, the FT-IR spectra of the POM catalysts after the stability test were measured [21, 22, 36]. The FT-IR spectra of Ba[1] (Fig. S9 in SI), Ba[2] (Fig. S10 in SI), and Ba[3] (Fig. 3d) after 2000 CV cycles were largely identical to those of the as-synthesized catalysts. Inductively coupled plasma-mass spectrometry (ICP-MS) analysis was carried out to quantify the metal (Ba, Co, Fe, or W) content in the 0.5 mol L^-1 H₂SO₄ mother liquor after 2000 CV cycles. The ICP-MS data demonstrated that quite a few amount (around 2%) of metal may leach out of the prepared catalysts (Table S2 in SI). These results collectively suggested the superior stability of these three bi-/trimetallic earth-abundant electrocatalysts.

In summary, the OER performance of an earth-abundant POM-based cobalt-phosphate cluster is largely improved by isostructural Fe substitution. The best catalyst Ba[3] exhibits an overpotential of 385 mV at 10 mA cm^-2 in 0.5 mol L^-1 H₂SO₄ solution, which is comparable to that observed for the IrO₂ catalyst. Multiple experiments collectively confirmed the good operational stability of Ba[3]. Our "in situ isostructural substitution" approach affords a synthetic route to well-defined multimetallic catalysts with enhanced performance. This also offers the ideal model for future experimental or theoretical work to study the structure-property correlation in detail, with the aim of developing more efficient molecular-based heterogeneous catalysts beyond OERs and understanding the related catalytic mechanisms.

Supporting Information. Electrocatalytic experiments, ICP-MS analysis and IR spectra.