2.1 General methods and materials

αααα-[(NaOH₂)₂(Fe^III)₂(P₂W₁₅O₅₆)₂]^16– and αααα-[(NaOH₂)₂(Fe^III)₂(As₂W₁₅O₅₆)₂]^16– were prepared by the literature methods, and their purity was checked by infrared spectroscopy and cyclic voltammetry [10,13]. As, Fe, Na, P, W, and Zn analyses were performed by Kanti Labs (Mississauga, Canada). Infrared spectra (2% sample in KBr) were recorded on a Nicolet 510 FTIR instrument. Average magnetic susceptibilities were measured on a Johnson–Matthey Model MSB-1 magnetic susceptibility balance as powders at 24 °C; the balance was calibrated using Hg[Co(SCN)₄] as a standard. Pascal’s constants were used to obtain the final diamagnetic corrections. Differential scanning calorimetric and thermogravimetric data were collected on ISI DSC 550 and TGA 1000 instruments, respectively. ³¹P NMR spectra were recorded on a Varian INOVA 400 MHz spectrometer using 85% H₃PO₄ as an external reference. ¹⁸³W NMR spectra were recorded on a Varian UNITY 400 MHz spectrometer using 0.2 M Na₂WO₄·2 H₂O (WO₄^2– (aq), δ  = 0 ppm) as an external reference. For electrochemical studies, solutions were de-aerated with Ar for 30 min prior to measurements and kept under positive pressure at all times. Pure NO (N₂₀ grade) was purchased from Air Liquide (France). The concentration of the POM in solution was 2  ×  10^–4 M, unless otherwise stated. The source, mounting, and polishing of the glassy carbon electrodes (GC, Tokai, Japan, 3 mm diameter) have been described in previous work [14]. The counter electrode was a platinum gauze of large surface area. The electrochemical apparatus was a EG and G 273A under computer control (M270 software). All experiments were performed at ambient temperature, and potentials are quoted against a saturated calomel electrode (SCE). Controlled potential coulometry determinations were performed as described for the cyclic voltammetry experiments except that the working electrode used was a 4 cm² surface area glassy carbon plate. In addition, after deaeration, Ar was continuously bubbled through the solution to ensure a constant stirring of the mixture.

2.2 Synthesis of Na₁₄[(Zn^IIOH₂)₂(Fe^III)₂(P₂W₁₅O₅₆)₂]

A 1-g sample of Fe₂P₄ was added to a 10-ml solution of ZnCl₂ (0.035 g, 0.25 mmol) (or Zn(NO₃)₂·6 H₂O (0.074 g)) dissolved in 1 M NaCl, and the mixture was heated to 65 °C for 30 min. The resulting solution was slowly cooled to room temperature, and a microcrystalline precipitate formed after 1–3 days (0.6 g, yield 61%). The precipitate was re-dissolved in 0.5 M NaCl to give diffraction-quality crystals. IR (2% KBr pellet, 1200–500 cm^–1): 1088 (s), 1064 (m, sh), 1017 (w), 950 (m, sh), 942 (s), 923 (w), 894 (w), 826 (m), 764 (w), 598 (w), 576 (w), and 524 (m). Magnetic susceptibility: μ_eff  = 9.71 μ_B mol^–1 at 297 K. ³¹P NMR (in D₂O): δ  = –12.3 ppm, Δν_1/2  = 100 Hz. Anal Calcd for Na₁₄(Zn^IIOH₂)₂(Fe^III)₂(P₂W₁₅O₅₆)₂·48 H₂O (for a sample dried in air for 12 h): Fe, 1.23; Na, 3.56; P, 1.37; W, 61.0; Zn, 1.44. Found: Fe, 1.29; Na, 3.68; P, 1.41; W, 61.0; Zn, 1.50. [MW  = 9051 g mol^–1].

2.3 Synthesis of Na₁₀H₄[(Zn^IIOH₂)₂(Fe^III)₂(As₂W₁₅O₅₆)₂]

A 1-g sample of Fe₂As₄ was added to a 10-ml solution of ZnCl₂ (0.035 g, 0.25 mmol) dissolved in 1 M NaCl, and the mixture was heated to 65 °C for 30 min. The resulting solution was slowly cooled to room temperature, and a microcrystalline precipitate formed after 1–3 days (0.5 g, yield 52%). The precipitate was re-dissolved in 0.5 M NaCl to give diffraction-quality crystals. IR (2% KBr pellet, 1200–400 cm^–1): 1058 (m), 949 (s), 869 (m), 822 (m), 700 (m), 590 (w), 557 (w), 523 (w), and 490 (w). ¹⁸³W NMR (in D₂O): δ  = –9.9 ppm, Δν_1/2  = 200 Hz (1 W); δ  = –46.9 ppm, Δν_1/2  = 33 Hz (1 W); δ  = –66.2 ppm, Δν_1/2  = 20 Hz (2 W); δ  = –184.4 ppm, Δν_1/2  = 13 Hz (2 W). The resonances of the other tungsten atoms are too close to the paramagnetic Fe(III) centers to give observable peaks. Anal Calcd for Na₁₀H₄[(Zn^IIOH₂)₂(Fe^III)₂(As₂W₁₅O₅₆)₂]·27H₂O: As, 3.42; Fe, 1.27; Na, 2.62; W, 62.9; Zn, 1.49. Found: As, 3.40; Fe, 1.41; Na, 3.01; W, 63.2; Zn, 1.47 [MW  = 8761 g mol^–1].

2.4 X-ray crystallography

Suitable crystals of Zn₂Fe₂P₄ and Zn₂Fe₂As₄ were coated with Paratone N oil, suspended on a fiber loop, and placed on a Brüker D8 SMART APEX CCD sealed tube diffractometer with graphite monochromated Mo Kα (0.71073 Å) radiation. Data collection, indexing, and initial cell refinements were carried out using SMART software (Version 5.624, Brüker AXS, Inc., Madison, WI). Frame integration and final cell refinements were carried out using SAINT software (Version 6.36A, Brüker AXS, Inc.). Final cell parameters were determined from least-squares refinement on 6903 reflections for Zn₂Fe₂P₄ and 8757 reflections for Zn₂Fe₂As₄. Absorption corrections for each data set were applied using SADABS (Version 2.10, George Sheldrick, University of Göttingen). The structures were determined using Direct Methods and difference Fourier techniques (SHELXTL, V6.12). A complete list of atoms refined anisotropically is provided in the Supporting Information. No H atoms associated with the water molecules of Zn₂Fe₂P₄ and Zn₂Fe₂As₄ were located in the difference Fourier maps. The final R1 scattering factors and anomalous dispersion corrections were taken from International Tables for X-ray Crystallography [15]. Structure solution, refinement, and generation of publication materials were performed using SHELXTL V6.12 software. Additional details are provided in Table 1, and thermal ellipsoid plots are given in the Supporting Information (Figs. S1 and S2).

3.1 Syntheses and physical properties of Zn₂Fe₂P₄ and Zn₂Fe₂As₄

The reaction of the divacant lacunary sandwich-type complexes, αααα-[(NaOH₂)₂(Fe^III)₂(X₂W₁₅O₅₆)₂]^16– (X  = P (Fe₂P₄) or As (Fe₂As₄)), with Zn(II) in 1 M NaCl yields αββα-[(Zn^IIOH₂)₂(Fe^III)₂(X₂W₁₅O₅₆)₂]^14– (X  = P (Zn₂Fe₂P₄) or As (Zn₂Fe₂As₄)) in modest yield (~50–60%) and in high purity. This procedure is somewhat different from that used for αββα-[(M^IIOH₂)₂(Fe^III)₂(P₂W₁₅O₅₆)₂]^14–, where M  = Ni(II) or Mn(II) [9,10]. The key difference is whether the transition metal salt is added to a solution of Fe₂X₄ (X  = P(V) or As(V)) (pH ~ 2) or whether Fe₂X₄ is added to a solution of the transition metal salt. In the case of Zn(II), the latter procedure is preferred to avoid the formation of small to modest amounts of αββα-[(Fe^IIIOH₂)₂(Fe^III)₂(X₂W₁₅O₅₆)₂]^12– (X  = P(V) or As(V)). This may be attributable to pH effects since the pH of the ZnCl₂ solution is 4–4.5 (versus 2 for a solution of Fe₂X₄). It may also be due in part to the general unreactive nature of d¹⁰ Zn(II) as ³¹P NMR studies suggest that the reaction of Fe₂P₄ with Zn(II) is slower than those observed for Fe₂P₄ with Ni(II) or Mn(II).

The IR spectra of Zn₂Fe₂P₄ and Zn₂Fe₂As₄ are compared with their divacant lacunary precursors Fe₂P₄ and Fe₂As₄ in Fig. 2. The W = O (W–O) stretching and W–O–W bending vibrations are quite similar in all four compounds suggesting that the incorporation of Zn(II) into Fe₂X₄ (X  = P(V) or As(V)) has mimimally perturbed the framework of the α-[X₂W₁₅O₅₆]^12– moieties. The splitting of the ν₃ vibrational mode of the central [PO₄]^3– unit observed in Fe₂P₄ is removed upon incorporation of the two Zn(II) cations. This suggests that the T_d symmetry of the [PO₄]^3– unit is nearly restored upon binding Zn(II) [16–18].

Differential scanning calorimetry data of Zn₂Fe₂X₄ and Fe₂X₄ are consistent with their molecular formulas, analytical data, and their crystal structures (Fig. 3). A series of endotherms are observed between 25 and 250 °C in all four complexes that are attributable to the loss of solvent water molecules and coordinated aqua ligands. These data are corroborated by thermogravimetric analyses over the same temperature range. In addition, a series of exotherms are observed between 450 and 600 °C which may be attributable to the crystallization of tungsten oxide and/or the decomposition of the tungsten–oxygen framework of the α-[X₂W₁₅O₅₆]^12– moieties. The infrared spectra of the complexes show significant changes in the W = O (W–O) stretching and W–O–W bending modes upon heating the samples between 450 and 600 °C. The P analogues (Fe₂P₄ and Zn₂Fe₂P₄) show a single exothermic process (Fig. 3A) whereas the As analogues (Fe₂As₄ and Zn₂Fe₂As₄) have a series of stepwise exotherms (Fig. 3B). This is most likely due to heteroatom effects (P versus As) since structurally the two pairs of complexes (Fe₂X₄ and Zn₂Fe₂X₄) are virtually identical. Interestingly, the divacant lacunary complex, Fe₂P₄, is more stable in the solid state than the product Zn₂Fe₂P₄. This is surprising given the fact that Fe₂P₄ has a higher negative charge than Zn₂Fe₂P₄ (16^– vs. 14^–, respectively). However, Fe₂P₄ contains four α-junctions whereas Zn₂Fe₂P₄ contains two α-junctions and two β-junctions. Given the fact that α-junctions were previously determined to be 4 kcal mol^–1 more stable than β-junctions for dehydrated, solid samples of (NH₄)₆[P₂W₁₈O₆₂], these results suggest that the type of junctions present in the sandwich-type structures plays a key role in their stability [19]. In addition, the steric repulsion present between the external M^II-OH₂ units and the bridging oxygen atoms of the neighboring belt W(VI) atoms (most prominent in α-junctions) would no longer be a factor since the aqua ligands are lost at temperatures higher than 250 °C.

3.2 Crystallographic studies

The X-ray crystal structure of Zn₂Fe₂P₄ reveals that a tetranuclear iron-zinc core, [Fe^III₂Zn^II₂O₁₄(H₂O)₂]^18–, is sandwiched between two trivacant Wells–Dawson α-[P₂W₁₅O₅₆]^12– moieties. In addition, two Zn(II) atoms have replaced the two Na⁺ atoms in the central unit of Fe₂P₄ with simultaneous 60° rotation of the structure to the conventional inter-POM-unit connectivity known as β (i.e. β-junctions). Each Zn(II) atom is ligated by three oxygen atoms from one of the two α-[P₂W₁₅O₅₆]^12– units, two oxygen atoms from the other α-[P₂W₁₅O₅₆]^12–, and a terminal aqua ligand. Bond valence sum calculations from the X-ray crystal structure of Zn₂Fe₂P₄ give an average oxidation state of 2.92  ± 0.03 and 2.09 ± 0.03 for the two Fe and the two Zn atoms, respectively [20].

The X-ray crystal structure of Zn₂Fe₂As₄ reveals that it is nearly isostructural to Zn₂Fe₂P₄, except that the As–O bond lengths are longer than those observed for the analogous P–O bond lengths in Zn₂Fe₂P₄. Overall, this has not caused significant perturbation of the tungsten-oxygen framework of the α-[X₂W₁₅O₅₆]^12– (X  = P(V) or As(V)) moieties or the central [Fe^III₂Zn^II₂O₁₄(H₂O)₂]^18– unit as revealed by a comparison of selected average bond lengths and angles summarized in Table 2. Bond valence sum calculations from the X-ray crystal structure of Zn₂Fe₂As₄ give an average oxidation state of 3.03  ± 0.02 and 2.03  ± 0.02 for the two Fe and the two Zn atoms, respectively [20]. There is rotational disorder in both of the polar W₃O₁₃ caps of Zn₂Fe₂As₄ with two different conformations present (αα and ββ). This type of cap disorder is very common in Wells–Dawson sandwich-type structures. Interestingly, cap disorder is also observed in solution by ¹⁸³W NMR. These studies show that two pairs of peaks (δ  = –9.9 ppm (1 W) and –184.4 ppm (2 W) for one cap configuration and δ  = –46.9 ppm (1 W) and –66.1 ppm (2 W) for the other configuration. The usual disorder of cations and solvent molecules makes it impossible to account for their total numbers in Zn₂Fe₂As₄ by X-ray crystallography alone. As a result, elemental and thermogravimetric analyses were used to determine the total number of Na⁺ cations and water molecules, respectively.

3.3 Electrochemical studies

Among the mixed-metal complexes characterized to date, αββα-[(M^IIOH₂)₂(Fe^III)₂(X₂W₁₅O₅₆)₂]^14–, where M  = Ni, Mn, or Co and X  = P(V) or As(V) reported recently by us [9,10] and by others [11], we have shown that a number of interesting features are displayed by the Mn(II)-containing complexes [10,21]. First, the complexes Mn₂Fe₂As₄ and Mn₂Fe₂P₄ display remarkable stability from pH 0 to 7 in their oxidized forms. Second, electrochemical data indicate that the Fe(II) state within the complexes is stabilized by the presence of the two Mn(II) centers, in contrast with the instability observed for Mn₂Fe₂P₄ synthesized by a different method [11]. Third, the divacant parent complexes Fe₂As₄ and Fe₂P₄ were shown to be quite unstable upon electrolysis at pH  < 4, in the absence of Mn(II) [22]. Finally, striking cooperativity effects were observed between the Fe and Mn centers in the electrocatalytic reduction of nitrite and nitric oxide [21]. Therefore, we wondered whether substitution of two Zn(II) into Fe₂X₄ (X  = P or As) might impart similar properties to the resulting complexes analogous to those observed in Mn₂Fe₂As₄ and Mn₂Fe₂P₄.

The stability of Zn₂Fe₂As₄ and Zn₂Fe₂P₄ as a function of pH and time was evaluated by monitoring their UV–vis spectra between 200 and 600 nm over at least 24 h. The As-derivative, Zn₂Fe₂As₄, was found to be stable from pH 0 to 7, and Zn₂Fe₂P₄, at the very least, was stable from pH 0 to 6. Only a small amount of decomposition (ca. 12%) was observed after 24 h for Zn₂Fe₂P₄ at pH 7. The stability of the complexes was also evaluated by cyclic voltammetry and the results were found to be complementary to those obtained from the electronic spectral data.

Fig. 4A compares the cyclic voltammograms of Zn₂Fe₂As₄ and Zn₂Fe₂P₄ in a pH 3 medium (0.5 M Na₂SO₄  +  H₂SO₄). Table 3 gathers the reduction peak potentials measured from Fig. 4A. The expected overall shift of the voltammogram in the positive potential direction upon the replacement of the P heteroatoms with As in the POMs is once again observed in these studies as well. The shifts are less pronounced for the Fe-waves than they are for the W-waves. The complexes, Mn₂Fe₂As₄ and Mn₂Fe₂P₄, display analogous cyclic voltammograms, published previously [10,21] and with the corresponding peak potential values spanning the same range as those obtained here. In short, in all of the present M₂Fe₂X₄ complexes (M  = Zn(II) or Mn(II); X  = P or As), the Fe- and W-wave systems were found to be well-separated from each other, facilitating the sequential study of each of the two groups of redox centers. All of the complexes display the classical behaviors usually encountered for the W-centers in the sandwich-type Wells–Dawson complexes. Therefore, in this manuscript attention will be focused primarily on the Fe-waves that are observed at more positive potentials than the W centers. There are two specific features about the Fe-centers which we will focus on most closely: the stability of the Fe(II) state within the complexes and their electrocatalytic properties. Controlled potential coulometry at potentials just negative of the peak potential of the second Fe-wave or the combined single Fe-wave (vide infra) consumed 1.95  ±  0.07 electrons per molecule thereby establishing the oxidation state of the two Fe(III) centers in the freshly prepared complexes.

A representative example of the behaviors of the Fe-waves as a function of pH is shown for Zn₂Fe₂As₄ in Fig. 4B. The formerly split Fe-waves at pH 3 now merge at pH 1, indicating a clear influence of the acidity of the medium. Analogous observations were made with Zn₂Fe₂P₄, but a slightly composite wave is obtained at pH 1, with the coalescence being somewhat incomplete. The results suggest that there is a pronounced influence from the central heteroatom, P or As. A comparison among all of the present M₂Fe₂X₄ complexes (M  = Zn(II) or Mn(II); X  = P or As), leads to the general conclusion that slight differences in their voltammetric behaviors are also a result of the different cations present within the mixed-metal complexes (i.e. Zn(II) or Mn(II)). These cations can modulate the acid–base properties of the complexes and correlatively their redox behaviors.

3.4 Electrocatalytic studies (evolution and progress)

In the following studies, we are interested in evaluating the electrocatalytic processes related to the redox and/or chemical properties of the POMs. The electrocatalytic behaviors of modified surfaces in which the basic properties of the POMs themselves cannot be distinguished are not considered [23]. The NO_x family is taken as an example. Electrocatalysis of the reduction of HNO₂ and NO was demonstrated in the late 1980s independently by Toth and Anson [24] and by our group [25,26]. Since then, this reaction has been selected and is being used worldwide as a classical test of the electrocatalytic properties of the POMs. Consideration of [Fe(H₂O)SiW₁₁O₃₉]^6– (SiW₁₁Fe for short), one of the electrocatalysts proposed by Toth and Anson [24], is particularly enlightening: a Fe-nitrosyl complex is formed upon reduction of the Fe center and the actual catalytic process occurs at more negative potentials where the necessary number of charges is accumulated and delivered by the W-framework. However, a loss in selectivity and a low yield ensue [24] from electrode derivatization [23]. Recently, to avoid the shortcomings encountered with SiW₁₁Fe, we and others have proposed several alternative POMs [14,25–28]. Among all of these potentially acceptable candidates, we have obtained N₂O with a quantitative yield of 100% with molybdic and molybdo-tungstic POMs [27]. A supramolecular compound like [P₈W₄₈O₁₈₄]^40–, with only W centers, can also be used to reduce NO efficiently without derivatization of the electrode surface [29]. Finally, sandwich-type complexes have the advantage of accumulating metal centers known for their catalytic behaviors in the reduction of NO_x, and they permit a substantial positive shift of the potential because these metal centers are easier to reduce than the W centers of the polytungstic framework [21,30–32].

In this context, electrocatalytic reduction of HNO₂ and NO by the present M₂Fe₂X₄ (M  = Mn(II) or Zn(II); X  = P or As) complexes constitutes a further step in the search for efficient electrocatalysts. Fig. 5A shows the current enhancements accompanying the addition of increasing amounts of NaNO₂ to a solution of Zn₂Fe₂As₄ at pH 1¹. This pH 1 value was selected because it ensures a fairly complete merging of the two Fe-waves in this complex. The increase in the current intensity is readily observed at the reduction potential of the Fe(III) centers for modest increases in the γ values (γ is the excess parameter defined as γ = C° (NO_x)/C° (POM)). In this potential domain, no direct reduction is observed for the NO or HNO₂ present in solution [27,33]. Furthermore, it is worth noting that all of the electrocatalysts show high efficiency even though modest γ values were used. Similar observations were made with pure NO as those obtained for nitrite in reasonably acidic solutions. Therefore, the corresponding results will not be described separately [21,27,33].

Fig. 5B compares the electrocatalytic HNO₂ reduction waves for Zn₂Fe₂X₄ (X  = P or As) with an excess parameter γ  = 10. The pattern for the As-derivative was shifted to a more positive potential domain compared to that for the P-derivative, even though the two final plateau currents are the same. Correlatively, for a selected potential value, the electrocatalytic activity of Zn₂Fe₂As₄ remains higher than that of its P-analogue up to the plateau current of the latter. The catalytic efficiency (CAT) is defined as follows: CAT = 100 × [I (POM + NO or HNO₂) – I^d(POM)]/I^d(POM), where I(POM  +  NO or HNO₂) is the current for the reduction of the POM in the presence of NO or HNO₂ and I^d (POM) is the corresponding diffusion current for the POM alone. This parameter must be considered with care. A rigorous comparison of the CAT values between two or more electrocatalysts demands that the same potential be selected to measure the currents (and eventually, the same catalytic current if an improvement in the potential is to be evaluated). Here, for γ  = 10 and at a potential of –50 mV versus SCE, CAT values are 696% for Zn₂Fe₂P₄ and 1002% for Zn₂Fe₂As₄. This result parallels that obtained for Mn₂Fe₂X₄ (X  = P or As), where the CAT value was approximately 25% higher for the As-derivative than for the P-analogue [21]. Although the attention here is focused mainly on the behaviors of these mixed-metal sandwich-type compounds, comparable catalytic efficiencies were also obtained under similar conditions for the lacunary precursors, Fe₂As₄ and Fe₂P₄ [21]. However, in contrast to the mixed-metal complexes, the stability of the Fe(II) states within Fe₂As₄ and Fe₂P₄ in acidic media is very low[22], and the data suggests that the mixed-metal complexes are better choices for long-term assays.

Experiments with NaNO₂ at pH 5 where HNO₂ and NO do not form, show that the catalytic process was relegated to the reduction potential of the first W-wave of the present mixed-metal complexes. In contrast, the catalysis starts directly in the iron-reduction potential domain when pure NO was used at pH 5. These observations are similar to those previously made with Mn₂Fe₂X₄ (X  = P or As) [21]. However, the CAT value measured at –400 mV versus SCE in the iron-reduction potential domain is about six times larger for Mn₂Fe₂As₄ than for Zn₂Fe₂As₄.

In summary, there are three features in the preceding results that are favorable for the use of these POMs in the electrocatalytic reduction of HNO₂ and NO. First, these POMs are stable from pH 0 to at least 6, a feature not shared by the otherwise efficient molybdic or molybdo-tungstic POMs. Second, these complexes show the enhanced stability of the Fe(II) state within the mixed-metal complexes relative to those observed in their lacunary precursors, Fe₂As₄ and Fe₂P₄. Finally, the third feature exhibited by these complexes which is favorable for electrocatalytic processes is the ability of the Fe-based redox waves to trigger the catalytic processes without any direct interference from the W(VI) centers.

A pH  5 medium was selected for the studies of the electrocatalytic reductions of O₂ and H₂O₂. The efficiency of the catalysis is again evaluated by an excess parameter γ, which is defined as C°(O₂)/C°(POM) for dioxygen and C°(H₂O₂)/C°(POM) for hydrogen peroxide (C  = concentration of the relevant species indicated in parentheses). The catalytic efficiency (CAT) is defined as follows: CAT =  100  ×  [I(POM  +  O₂ or H₂O₂)  –  I^d(POM)]/I^d (POM), where I (POM  +  O₂ or H₂O₂) is the current for the reduction of the POM in the presence of O₂ or H₂O₂ and I^d (POM) is the corresponding diffusion current for the POM alone. Selected values of catalytic efficiencies (CAT) for the reduction of dioxygen and hydrogen peroxide are included in Table 4. CAT values were measured at –420 mV versus SCE, in the potential domain of the Fe(III)-reduction waves.

Corresponding CAT values for Mn₂Fe₂X₄ (X  = P or As) were also measured and included for comparison. Under the same experimental conditions, the phosphorus-containing complexes, M₂Fe₂P₄ (M  = Zn(II) or Mn(II)), are more efficient than their arsenic analogues, M₂Fe₂As₄ (M  = Zn(II) or Mn(II)), for dioxygen reduction, while the opposite is true for hydrogen peroxide reduction. This observation underscores the difficulty for anticipating which electrocatalyst would be more efficient when an inner sphere pathway is operative. The electrocatalytic processes observed here remain efficient in other media, and they establish the utility of the Fe(III) centers within these mixed-metal sandwich complexes even in media of low proton availability. There is still ample room for optimization of the selected media for efficient electrocatalytic processes within these compounds.

Examination of the two electrocatalytic processes discussed here may also shed some light on possible reaction pathways. Dioxygen does not coordinate to Fe(III) but it does coordinate to Fe(II). Then, the presence of another reduced Fe center could drive the catalytic process to completion by carrying out immediate further reduction of the Fe–O₂ adduct. Also noteworthy is that the synthesis of M₂Fe₂X₄ (M  = Zn(II) or Mn(II); X  = P or As) from Fe₂X₄ (X  = P or As) diminishes the overall negative charge of the POM, rendering its reduction more facile, and ultimately, saving energy during the electrocatalytic process.