Pentanuclear iron catalysts for water oxidation: substituents provide two routes to control onset potentials

Two distinct routes to decrease the onset potential for water oxidation were provided by either control of redox potentials of the complex or change of the reaction mechanism in the pentairon catalysts. The results offer a novel strategy to design efficient molecule-based catalysts for water oxidation.


Introduction
Water oxidation (2H 2 O / O 2 + 4H + + 4e À ) is considered the main bottleneck in the production of chemical fuels from sunlight and/or electricity; 1-7 this is because the reaction requires the transfer of four electrons and the generation of an O-O bond, and is both thermodynamically and kinetically demanding. Therefore, the development of a highly active articial catalyst for the oxidation of water is of great importance. In this context, since the discovery of the rst molecular water oxidation catalyst, "Blue dimer", 8 a signicant number of molecular water oxidation catalysts have been reported. [9][10][11][12][13][14][15][16][17][18][19][20][21][22][23][24] More recently, metal complexes containing earth-abundant transition metal ions such as Mn, [25][26][27][28] Fe, [29][30][31][32][33][34][35][36] Co, [37][38][39][40][41][42][43][44][45][46] and Cu [47][48][49][50][51][52][53][54][55] ions have also been intensely studied. Nevertheless, the development of efficient metal-complex-based catalysts that consist of earth-abundant transition metals is still very challenging. 56,57 In nature, water oxidation is catalysed by the oxygen evolving complex (OEC) in photosystem II. [58][59][60] The OEC is a highly active and robust catalyst for water oxidation that can drive the reaction under mild conditions. 61 The active site of the OEC contains a multinuclear metal complex, a Mn 4 CaO 5 cluster, which has several water coordination sites. Due to the multinuclear structure of the Mn 4 CaO 5 cluster, the OEC can smoothly accumulate the oxidative equivalents required for the reaction via the formation of ve distinct redox intermediates, the S n states, where the subscript indicates the number of stored oxidative equivalents (n ¼ 0-4). Aer the formation of the S 4 state, H 2 O reacts with the S 4 state to generate O 2 and protons. 62 Recently, we demonstrated that a pentanuclear iron complex [Fe II 4 Fe III (m 3 -O)(bpp) 6 ] 3+ , [Fe 5 -H] 3+ (Scheme 1a, Hbpp ¼ bis(pyridyl)pyrazole), can serve as a highly active catalyst for electrocatalytic water oxidation. 33 [Fe 5 -H] 3+ can also accumulate four oxidative equivalents via the successive oxidation of each of the iron centres in the complex (Scheme 1b). In the catalysis mediated by [Fe 5 -H] 3+ , the four-electron-oxidized species, [ 3+ are the highest among those of iron-based water oxidation catalysts (Fe-WOCs) reported thus far. However, a relatively large onset potential is required for the catalysis because the S 4 state is only generated at high potentials. Therefore, the development of a novel strategy for designing catalysts that can drive the reaction at low onset potentials is essential.
Here, we report two approaches for decreasing the onset potential of pentairon water oxidation systems. Two approaches involving the installation of substituents onto the Hbpp ligand have been demonstrated. Two kinds of ligands, one with electron-donating and the other with electron-withdrawing groups at the 4-position of the Hbpp ligand (Me-Hbpp and Br-Hbpp in Scheme 2), have been employed, and the new pentairon complexes were constructed utilizing these ligands. The newly synthesized complexes catalysed the oxidation of water with high faradaic efficiencies, and the onset potentials of these complexes were lower than that of the parent complex. The mechanistic studies also revealed that two distinct routes exist to decrease the onset potentials for water oxidation in pentanuclear iron systems.

Syntheses and characterization of ligands and pentairon complexes
Me-Hbpp was prepared using the three-step synthetic route shown in Scheme S1 in the ESI. † Initially, 2-acetylpyridine and pyridine-2-carboxylic acid methyl ester were reacted in the presence of sodium ethoxide to yield 1,3-di(2-pyridyl)-1,3propanedione. The methylation of the product and further treatment with hydrazine monohydrate afforded Me-Hbpp. The total synthetic yield of Me-Hbpp was 26% (see the Experimental section for details). Br-Hbpp was synthesized in a moderate yield (65%) by the bromination of Hbpp. 63 Both ligands were characterized by 1 H and 13 C NMR spectroscopy and elemental analyses. The syntheses of two pentairon complexes with the obtained ligands were performed by reacting the corresponding ligand (6 eq.) with FeSO 4 $7H 2 O (5 eq.) in the presence of a base (NaOH, 6 eq.) in methanol at 80 C (Scheme S2 †). The reaction mixture was further treated with a saturated solution of aqueous NaBF 4 or NaPF 6 , and the obtained precipitate was collected by ltration. The precipitate was recrystallized from MeCN/Et 2 O to afford crystalline products. The electrospray ionization mass spectrometry (ESI-MS) and elemental analysis data of the obtained crystalline samples conrmed the formation of the desired pentairon complexes, [ Fig. 1 and S1, † and the crystallographic data for the newly synthesized complexes are summarized in Table S1 3+ belongs to the I 4 space group, and the asymmetric unit contains half of the cationic pentairon complex and one and half BF 4 anions. The bond distances between the iron atoms and the N atoms on R-bpp À ligands are not signicantly changed by either bromo or methyl substitution (Table S2 †). These results clearly demonstrate that the substituents do not affect the pentanuclear core structure of the complexes (Fig. S2 †).

UV-Vis absorption spectra
The UV-Vis absorption spectra of [Fe 5 3+ , reecting the electron-donating or electron-withdrawing nature of the substituents.

Electrochemical properties
The inuence of the electron-donating and electronwithdrawing groups on the redox properties of the pentairon complexes was investigated by cyclic voltammetry. redox couples, indicating that the electron transfer ability arising from the pentairon structure is preserved even aer the introduction of substituents on the ligands. Importantly, all the redox waves of [Fe 5 -Me] 3+ were shied to a more negative potential relative to those of [Fe 5 -H] 3+ , whereas the redox waves of [Fe 5 -Br] 3+ were positively shied ( Fig. 2 and Table 1). These trends are consistent with the electron-donating and electron-withdrawing properties of the methyl and bromo substituents, respectively. This result clearly demonstrates that the redox potentials of the pentairon complexes can be tuned by the introduction of    (Fig. 4b). The faradaic efficiencies of the reaction based on the 4e À process were 92 and 86% for [Fe 5 -Me] 3+ and [Fe 5 -Br] 3+ , respectively. Based on the results of CPE experiments, turnover frequencies (TOFs) and turnover numbers (TONs) for water oxidation were roughly estimated. For [Fe 5 -Me] 3+ , TOF and TON were 3 Â 10 2 s À1 and 2 Â 10 6 , respectively, and TOF and TON values of [Fe 5 -Br] 3+ were estimated to be 20 s À1 and 1 Â 10 5 , respectively (for the details of calculation see the ESI (P.S24) †). Although these values were lower than those of [Fe 5 -H] 3+ (1 Â 10 3 s À1 (TOF) and 7.5 Â 10 6 Table 1 Redox potentials (E 1/2 , V vs. Fc/Fc + ) and the onset potentials for water oxidation (E onset , V vs. Fc/Fc + ) of a series of pentairon complexes in acetonitrile solutions with TBAP (0.1 M). E 1/2 (À1), E 1/2 (1), E 1/2 (2), E 1/2 (3), and E 1/2 (4) correspond to the E 1/2 values of the Fe II  4 Fe III /  Fe II  5 , Fe II  3 Fe III  2 /Fe II  4 Fe III , Fe II  2 Fe III  3 /Fe II  3 Fe III  2 , Fe II Fe III  4 /Fe II  2 Fe III  3 , and Fe III  5 /  Fe II Fe III  4

redox couples, respectively
Complex    (TON)) estimated by using the same method, they were substantially higher compared to those of the reported ironcomplex-based catalysts for water oxidation. [29][30][31][32]35 In both cases, the electrolyzed solutions were treated with oxo [5,10,15,20-tetra(4-pyridyl)porphyrinato]titanium(IV) as a chemical probe 65 and the 2e À oxidized product of H 2 O (H 2 O 2 ) was not detected (for details of the experimental procedure, see the ESI (P.S22-S23) †). Aer the CPE experiment, the ITO working electrodes used in the electrolysis were gently rinsed with small amounts of water and MeCN, and then, a second round of electrolysis was performed using the solution without the catalyst. Signicantly small currents were observed in the second electrolysis compared to the rst electrolysis in both cases ( Fig. S4 and S5 †), which indicates that the species homogeneously dissolved in the solution are catalytically active. CV measurements of the solution aer the CPE experiments also claried the presence of pentanuclear complexes in the solution phase (Fig. S6 †). Additionally, the UV-Vis absorption spectra of the ITO electrodes before and aer the CPE experiments remained almost identical ( Fig. S7 and S8 †), suggesting no formation of heterogeneous deposits during the electrolysis. We also analysed the electrolyte solutions aer the electrolysis by dynamic light scattering (DLS) measurements and no formation of heterogeneous nanoparticles was detected (Fig. S9 †) (Fig. 3a). The onset potential (E onset , Table 1) for the reaction was estimated from the cross-point of two lines that were obtained by extrapolating the slopes of the catalytic current and non-faradaic current. The E onset of [Fe 5 -Me] 3+ was determined to be 1.09 V, which corresponds to the overpotential (h) of 0.65 V at pH ¼ 5.0, and was slightly larger than E 1/2 of the fourth redox couple (1.01 V). In contrast, the catalytic current for water oxidation for [Fe 5 -Br] 3+ was observed at a more negative potential than that of the fourth redox couple (1.27 V, Fig. 3b

Discussion
Reaction mechanism of [Fe 5 -Me] 3+ As described above, the E onset of [Fe 5 -Me] 3+ is located at a slightly more positive potential than the E 1/2 of the fourth redox couple (Table 1). In other words, the formation of the four-electron oxidized species (Fe III 5 , the S 4 state) triggers the reaction with a water molecule and the subsequent oxidation of water in this case. A similar trend was also observed in the previously reported electrocatalysis by [Fe 5 -H] 3+ ; the onset of the catalytic wave is coupled with the formation of the fourelectron oxidized species. Therefore, it is suggested that [Fe 5 -Me] 3+ probably promotes electrocatalytic water oxidation through a catalytic cycle similar to that of [Fe 5 -H] 3+ , which we previously proposed based on experimental and computational studies 33 (Fig. 5, see also the ESI (P.S32) †). In the catalytic cycle, the successive four-step, one-electron oxidation of the resting Fe II 3+ to conrm if the S 3 state undergoes unimolecular or bimolecular reactions. As shown in Fig. S12, † the intensity of the catalytic peak current was linearly dependent on the concentration of [Fe 5 -Br] 3+ , suggesting that the rate is rst order to the catalyst concentration, and therefore a bimolecular path requiring the association of catalysts is ruled out. Second, we hypothesized that the reaction of the S 3 state with H 2 O to form a H 2 O bound species and subsequent oxidation of the formed species is a possible pathway. To validate this hypothesis, CVs were collected at various scan rates in the presence of 5 M H 2 O by reversing the scan of potentials at 1.04 V. As shown in Fig. S13a, † the redox potentials (E 1/2 values) and the wave shapes of the rst three redox couples remained unchanged. Additionally, the reversibility of the third redox couple (Fe II Fe III 4 / Fe II 2 Fe III 3 ) was investigated by plotting the intensity of the anodic and cathodic peak currents against the square root of the scan rates (Fig. S13b †). The linearity of the obtained plot conrms that the third redox process is fully reversible and that no EC process occurs in this potential range. Therefore, the reaction of the S 3 state with H 2 O does not proceed in this potential region, and this pathway can be excluded. Third, the possibility of the S 3 state undergoing a proton-coupled electron transfer (PCET) reaction was considered because catalytic water oxidation reactions oen involve such a process. 15 However, this process hardly occurs because no dissociative proton exists in the S 3 state. Moreover, the CVs of [Fe 5 -Br] 3+ recorded under various pH conditions showed no change in the onset potential for water oxidation (Fig. S14 †). Therefore, the S 3 state undergoing a PCET process is also unlikely. Finally, an electron transfer reaction coupled with water binding to the S 3 state was investigated. CVs of [Fe 5 -Br] 3+ at various concentrations of H 2 O were acquired. As shown in Fig. 6 and S15, † the onset potentials of the electrocatalytic current gradually shied to lower potentials as the content of H 2 O increased. Note that the onset potential of water oxidation was not affected by the concentrations of H 2 O in the case of [Fe 5 -H] 3+ (Fig. S16 †). This result clearly demonstrates that the electron transfer reaction coupled with the binding of H 2 O to the S 3 state is the key step in the [Fe 5 -Br] 3+ -catalysed reaction.

Electronic structures of S 3 states
To clarify the origin of the unique reactivity of the S 3 state of [Fe 5 -Br] 3+ , the electronic structures of the S 3 states of a series of pentanuclear iron complexes were investigated. As we previously reported, 33 the three-step oxidation of [Fe 5 -H] 3+ affords the S 3 state as evidenced by UV-Vis absorption spectroscopy ( Fig. S17 and S18 †). In the rst step, a slight decrease in the MLCT band at 406 nm and the growth of a new broad peak at approximately 640 nm were observed. This newly observed peak is attributed to the formation of the [Fe II Fe III 2 (m 3 -O)] central core. 33 In other words, the rst step corresponds to the oxidation of the central core, which yields [{Fe II (m-bpp) 3 } 2 -Fe II Fe III 2 (m 3 -O)] 4+ (the S 1 state). In the second step, the intensity of the MLCT band at 406 nm drastically decreased, suggesting that both iron centres at the apical positions are oxidised during the second step. Therefore, in the second step, the oxidation of the complex induces an intramolecular electron transfer process, and a species with two Fe III ions at apical positions and one Fe III and two Fe II ions in the central core, [ 3+ ; a slight decrease in the MLCT band at 402 nm and the emergence of a new band at approximately 660 nm in the rst step and a drastic decrease in the MLCT band in the second step ( Fig. 8a and b). Therefore, the iron ion at the central core is oxidised in the rst step, and an oxidation-induced intramolecular electron transfer affords [{Fe III (m-Br-bpp) 3 3+ . In this step, the intensity of the band at approximately 580 nm decreased (Fig. 8c)

Reaction mechanism of [Fe 5 -Br] 3+
Based on the aforementioned experimental evidence, a plausible reaction mechanism for the water oxidation reaction catalysed by [Fe 5 -Br] 3+ was proposed. As depicted in Scheme 3, the rst step involves a sequential, stepwise three-electron oxidation of the initial S 0 state to produce the S 3 state (via the S 1 and S 2 states), which includes a two-step intramolecular electron transfer process. In the S 3 state of the complex, all iron atoms in the [Fe 3 (m 3 -O)] core are in the Fe III state. Subsequently, a concerted process involving water binding to the fully oxidized [Fe 3 (m 3 -O)] core coupled with a one-electron oxidation process gives the water-bound Fe III 5 (OH 2 ) species, intermediate A (Scheme 3, Path B). Intermediate A then generates intermediates B and C, and the release of O 2 from intermediate C regenerates the initial S 0 state and produces O 2 as a product (Fig. S19 †). Thus, the formation of the S 4 state is favorably bypassed in the catalytic cycle of [Fe 5 -Br] 3+ , whereas the redox potentials to form the S 4 states determine the onset potentials for the catalysis in the case of [Fe 5 -Me] 3+ and [Fe 5 -H] 3+ . As a result, the onset potential for water oxidation was lower for [Fe 5 -Br] 3+ compared to [Fe 5 -H] 3+ if a sufficient amount of substrate was added to the reaction mixture (Fig. 9). The result also implies that the generation of the fully oxidized [Fe 3 (m 3 -O)] core is essential for initiating catalysis.

Conclusions
In conclusion, we have demonstrated the syntheses, electrochemical behaviour, and catalytic activity of a series of  To clarify the origin of the lower onset potentials of the complexes, their catalytic mechanisms were investigated. In the case of [Fe 5 -Me] 3+ , the formation of the S 4 state triggers the catalytic reaction, which is similar to the pathway seen with the parent [Fe 5 -H] 3+ complex. Therefore, the decrease in the onset overpotential for [Fe 5 -Me] 3+ is attributed to the electrondonating nature of the methyl substituents, which allows the generation of the S 4 state in the more negative potential region. In contrast, in the catalysis mediated by [Fe 5 -Br] 3+ , the threeelectron oxidized species (the S 3 state) served as a key intermediate due to its unique electronic structure, and the state undergoes a water binding reaction coupled with an electron transfer to initiate the catalytic reaction. In other words, the generation of the S 4 state was bypassed in this system, which enables the catalytic reaction to occur at a lower onset potential. Our results reveal that not only the simple tuning of the redox potentials by the introduction of an electron-donating group but also the control over the reaction mechanism by the introduction of an electron-withdrawing group can be a valuable strategy for controlling onset potentials.

General methods
NMR spectra were recorded on a JEOL JNM-LA 400 spectrometer. UV-Vis spectra were recorded on a SHIMADZU UV-2550UV-Vis spectrophotometer or a UV-Vis Agilent Cary8454 spectrophotometer with a conventional quartz cuvette (path length, l ¼ 1 cm). Spectroelectrochemical studies were performed using a BAS Inc. spectroelectrochemical quartz cell (l ¼ 1 mm) containing Pt gauze (working electrode), a Pt wire (auxiliary electrode) and Ag/Ag + (reference electrode) in conjunction with a CH Instruments potentiostat. Elemental analyses were performed on a J-SCIENCE LAB MICRO CORDER JM10 elemental analyser. ESI-TOF mass spectra were recorded on a JEOL JMS-T100LP mass spectrometer. Gas chromatography analysis of O 2 was performed using a Shimadzu GC-2014 gas chromatograph equipped with a thermal conductivity detector and tted with a molecular sieve (5Å) column, and the system was calibrated with air. Dynamic light scattering (DLS) data were measured using a Photal OTSUKA ELECTRONICS ELSZ-1000 zeta-potential and particle size analyser, equipped with a 785 nm red laser source (detection limit: 0.6 nm particle diameter).

X-ray crystallography
Data collection for [Fe 5 -Me] 3+ and [Fe 5 -Br] 3+ was performed at 123 K on a ROD, Synergy Custom system (Rigaku Oxford Diffraction) equipped with confocal monochromated Mo-Ka radiation, and data were processed using CrysAlisPro 1.171.39.43c (Rigaku Oxford Diffraction). The structures were solved by direct methods using SIR-92 (ref. 66) and rened by the full-matrix least squares techniques on F 2 (SHELXL-97). 67 All nonhydrogen atoms were rened anisotropically and rened with a riding model with U iso constrained to be 1.2 times U eq of the carrier atom. The diffused electron densities resulting from the residual solvent molecules were removed from the data set using the SQUEEZE routine of PLATON 68

Electrochemical studies
Electrochemical experiments were performed on a BAS ALS Model 650 DKMP electrochemical analyser at room temperature under Ar. Cyclic voltammetry experiments were performed using a one-compartment cell with a standard three-electrode conguration, which consisted of a glassy carbon disk (diameter 3 mm, from BAS Inc.), a Ag/Ag + couple, and a platinum wire as the working, reference and auxiliary electrodes, respectively. Between scans, the working electrode was polished with 0.05 mm alumina paste (from BAS Inc.) and washed with puried H 2 O. All the redox potentials of the samples presented in this paper were calibrated against the redox potential of the ferrocene/ferrocenium couple (Fc/Fc + ).

Controlled potential electrolysis
Controlled potential electrolysis experiments were performed in a custom-designed gas-tight two-compartment cell separated by an anion-exchange membrane. 33 In the rst compartment, the ITO working electrode (1.0 cm Â 1.5 cm) and Ag/Ag + reference electrode were immersed in an electrolyte solution (0.1 M Bu 4 -NClO 4 in acetonitrile/water (10 : 1) mixed solvent) containing the catalyst (0.2 mM). In the second compartment, the platinum auxiliary electrode was immersed in the electrolyte solution. The amount of evolved oxygen in the headspace of the reaction cell was quantied by gas chromatography. Subsequently, the potential production of liquid products (e.g., H 2 O 2 ) in the reaction was analysed by treating the electrolyzed solution with oxo[5,10,15,20-tetra(4-pyridyl)porphyrinato]titanium(IV) as a chemical probe. 65 Syntheses 1,3-Bis(2-pyridyl)-propane-1,3-dione (I). To a solution of pyridine-2-carboxylic acid methyl ester (2 g, 16.5 mmol) in anhydrous toluene (40 mL) under an argon atmosphere was added freshly prepared sodium ethoxide solution (9.9 mL, 2 M, 19.8 mmol). Aer heating the reaction mixture to 55 C, a solution of 2-acetyl pyridine (2.26 g, 16.5 mmol) in anhydrous toluene (10 mL) was added. Aer stirring the resulting mixture for 2 h at 55 C, a dark yellow precipitate appeared, and the reaction was stirred overnight at room temperature. The solvent was then evaporated, and the crude product was poured onto ice and neutralized to pH 7 with acetic acid (50%). The resulting solid was collected by ltration and dried under vacuum to give compound I (1.92 g, 52%). The product was used for the next step without further purication.
1,3-Bis(2-pyridyl)-propane-2-methyl-1,3-dione (II). Under an argon atmosphere, compound I (0.5 g, 2.21 mmol) was dissolved in 30 mL of anhydrous toluene. Potassium bis(trimethylsilyl)amide (6.63 mL, 0.5 M, 3.32 mmol) was then added to the solution. The resulting suspension was heated to 80 C, and the colour of the reaction mixture changed from orange to green. Then, methyl iodide (1.57 g, 11.05 mmol) was added. Aer heating the resulting mixture at 80 C for 5 h, the reaction was stirred overnight at 50 C. Aer cooling to room temperature, the reaction was quenched by the addition of 10% NaHCO 3 (10 mL) followed by brine (10 mL) and extracted with CH 2 Cl 2 . The organic phases were dried over anhydrous Na 2 SO 4 and ltered. The ltrate was concentrated under reduced pressure. The crude oily product was puried by silica gel column chromatography using 30% EtOAc/n-hexane as the eluent to give compound II as a light yellow solid (0.43 g, 82%). 1 (25 mL), and the solution was degassed with Ar for 30 min. To this solution was added hydrazine monohydrate (0.28 g, 5.6 mmol), and the reaction mixture was reuxed at 95 C under Ar for 3 h. Aer concentrating the resulting solution by rotary evaporation, the solution was kept in a refrigerator overnight. A precipitate formed, and it was collected by ltration and washed with a small amount of cold ethanol to give Me-Hbpp as a white solid (0.19 g, 61%). 1  4-Bromo-3,5-bis(2-pyridyl)-pyrazole (Br-Hbpp). This compound was synthesized using the reported procedure. 63 3,5-Bis(2pyridy)pyrazole (0.4 g, 1.8 mmol) was dissolved in CH 2 Cl 2 (60 mL) at 0 C. A solution of bromine (0.4 mL) in aqueous Na 2 CO 3 (1 N, 25 mL) was then added dropwise, and the reaction was allowed to stir for 30 min. The reaction mixture was neutralized to pH 7 with aqueous 1 M NaOH. The aqueous phase was extracted with CH 2 Cl 2 (80 mL). The organic phase was dried over anhydrous Na 2 SO 4 and ltered. The ltrate was concentrated under reduced pressure to yield the crude product, which was puried by column chromatography on silica gel using 5% MeOH/CH 2 Cl 2 to afford Br-Hbpp as a pale-yellow solid (0.35 g, 65%). 1  . Me-Hbpp (0.040 g, 0.17 mmol) was dissolved in degassed methanol (10 mL), NaOH aq (1 M, 0.17 mL, 0.17 mmol) was added, and the mixture was stirred to dissolve the Me-Hbpp. Subsequently, FeSO 4 $7H 2 O (0.038 g, 0.14 mmol) was added to the stirred solution, and the resulting dark red solution was reuxed at 80 C for 1 h under Ar. Aer cooling the reaction mixture to room temperature, the mixture was ltered to remove the undissolved residue. An aqueous solution of NaPF 6 (excess) was added to the ltrate, and a small amount of water was added to the solution. The solution was kept in a refrigerator for 30 min to generate a red brown precipitate. The precipitate was collected by ltration, washed with water and dried under vacuum.