Water oxidation catalysed by iron complex of N,N′-dimethyl-2,11-diaza[3,3](2,6)pyridinophane. Spectroscopy of iron–oxo intermediates and density functional theory calculations

Abstract

The macrocyclic [Fe^III(L1)Cl₂]⁺ (1, L1 = N,N′-dimethyl-2,11-diaza[3,3](2,6)pyridinophane) complex is an active catalyst for the oxidation of water to oxygen using [NH₄]₂[Ce^IV(NO₃)₆] (CAN), NaIO₄, or Oxone as the oxidant. The mechanism of 1-catalysed water oxidation was examined by spectroscopic methods and by ¹⁸O-labelling experiments, revealing that Fe^IV=O and/or Fe^V=O species are likely to be involved in the reaction. The redox behaviour of 1 and these high-valent Fe=O species of L1 has been examined by both cyclic voltammetry and density functional theory (DFT) calculations. In aqueous solutions, the cyclic voltammograms of 1 at different pH show a pH-dependent reversible couple (E_1/2 = +0.46 V vs. SCE at pH 1) and an irreversible anodic wave (E_pa = +1.18 V vs. SCE at pH 1) assigned to the Fe^III/Fe^II couple and the Fe^III to Fe^IV oxidation, respectively. DFT calculations showed that the E value of the half reaction involving [Fe^V(L1)(O)(OH)]²⁺/[Fe^IV(L1)(O)(OH₂)]²⁺ is +1.42 V vs. SCE at pH 1. Using CAN as the oxidant at pH 1, the formation of an Fe^IV=O reaction intermediate was suggested by ESI-MS and UV-vis absorption spectroscopic measurements, and the rate of oxygen evolution was linearly dependent on the concentrations of both 1 and CAN. Using NaIO₄ or Oxone as the oxidant at pH 1, the rate of oxygen evolution was linearly dependent on the concentration of 1, and a reactive Fe^V=O species with formula [Fe^V(L1)(O)₂]⁺ generated by oxidation with NaIO₄ or Oxone was suggested by ESI-MS measurements. DFT calculations revealed that [Fe^V(L1)(O)₂]⁺ is capable of oxidizing water to oxygen with a reaction barrier of 15.7 kcal mol⁻¹.

---

Introduction

Water oxidation is an energetically uphill reaction (E = +1.23 V vs. NHE at pH 0) involving the simultaneous removal of four electrons and four protons from two water molecules to give one oxygen molecule.¹ Due to its importance, there have been tremendous efforts dedicated to the design of metal catalysts for water oxidation over the past decades.² It is envisioned that mechanistic insights into the fundamental steps of this important reaction can be obtained by using structurally defined molecular catalysts.

In the literature, many examples of molecular catalysts, particularly polypyridyl ruthenium(II) and organometallic iridium(III) complexes, and also complexes of 1^st-row transition metals such as manganese, iron and cobalt bearing chelating N and/or O ligands, have been reported for water oxidation.² Mechanistic studies revealed that the availability of labile coordination site(s) for the formation of metal–oxo species is crucial for a metal complex to be used to catalyse water oxidation.² To support the formation of reactive/oxidizing metal–oxo species, ligands that bind strongly to metal ions and are resistant to oxidation are highly desirable. In this regard, macrocyclic N-donor ligands are appealing, as highly oxidizing metal–oxo complexes of both ruthenium and iron have been isolated and structurally characterized by using macrocyclic tertiary amine ligands such as 14-TMC (1,4,8,11-tetramethyl-1,4,8,11-tetraazacyclotetradecane).³

Because of the high earth abundance of iron and the recent impressive advances made in the isolation and characterization of non-heme iron–oxo complexes, notably by Nam, Que, and co-workers,^4,5 there has been increasing interest in the development of new iron-based oxidation chemistry, including the oxidation of water.² Since the first report by Bernhard, Collins, and co-workers in 2010,^6a several types of water oxidation reactions catalysed by mononuclear iron complexes have been reported, including chemical oxidation with CAN (cerium ammonium nitrate [NH₄]₂[Ce^IV(NO₃)₆]) or NaIO₄,⁶ photochemical oxidation,^6c,e,7 electrocatalytic oxidation,⁸ and photoelectrochemical oxidation.⁹ Dinuclear iron complexes that can catalyse water oxidation with CAN or NaIO₄,¹⁰ or can oxidize water to the hydroxyl radical,¹¹ are also documented.

Mononuclear iron catalysts for water oxidation are supported by tetraanionic tetraamide ligands (TAML)^6a,7,8b or monoanionic pentadentate N₅ ligands,^8a or bear neutral chelating N ligands,^6b–g,8c such as mcp, ^Me2Pytacn, tpa, and bqen (Fig. 1), which allow the formation of mononuclear metal complexes with two cis labile sites (the abovementioned 14-TMC ligand forms trans complexes such as trans-[Fe(14-TMC)(OTf)₂] which was found to be unreactive for catalytic water oxidation with CAN or NaIO₄ (ref. 6b)). Notably, the ‘[Fe^II(mcp)(OTf)₂] + NaIO₄’ system reported by Costas, Lloret-Fillol, and co-workers formed oxygen with a turnover number (TON) of up to >1000 at pH 2.^6b Oxoiron(IV) (Fe^IV=O) intermediates supported by ^Me2Pytacn^6b,f and bqen,^6e generated from reaction of the corresponding iron catalysts with CAN in aqueous solution, were detected by UV-vis spectroscopy and electrospray-ionization mass spectrometry (ESI-MS); it was proposed that [Fe^IV(bqen)(O)(OH₂)]²⁺ may be involved in O–O bond formation,^6e and density functional theory (DFT) calculations on the Fe^II–^Me2Pytacn system favoured Fe^IV=O reactive intermediates.^12b A recent report demonstrated the involvement of O=Fe^IV–O–Ce^IV species in the water oxidation with CAN catalysed by [Fe^II(mcp)(OTf)₂].¹³ In most cases, oxoiron(V) (Fe^V=O) species are proposed to be the active intermediates directly responsible for O–O bond formation in water oxidation reactions,^6a,b,f,7,8a as revealed by DFT calculations on water oxidation with CAN catalysed by the Fe^III–TAML,^12a,d Fe^II–^Me2Pytacn,^12e,f and Fe^II–mep^12f systems. Several non-heme Fe^V=O species, which were generated by oxidation with peracids, H₂O₂ or ^tBuOOH in organic solvents, have been reported in the literature.¹⁴ Notable examples include [Fe^V(TAML)(O)]⁻,^14a,j [Fe^V(^Me2Pytacn)(O)(OH)]²⁺,^14d and [Fe^V(L)(O)(S)]³⁺ (L = tpa, mep; S = H₂O or MeCN).^14c Recently, [Fe^V(TAML)(O)]⁻ (I, Fig. 1) generated by the photochemical reaction of [Fe^III(TAML)(H₂O)]⁻ with ‘[Ru^II(bipy)₃]²⁺ + Na₂S₂O₈’ in 50% MeCN–borate buffer mixture, was reported to be an active intermediate in the [Fe^III(TAML)(H₂O)]⁻-catalysed photochemical water oxidation.⁷ The proposed Fe^V=O intermediates with neutral chelating N ligands (II, Fig. 1) remain elusive for water oxidation reactions.

Fig. 1 Examples of N ligands in iron complexes used as water oxidation catalysts. The corresponding proposed Fe^V=O species are indicated. X in II stands for solvent, OH⁻ or another ligand.

N,N′-Dimethyl-2,11-diaza[3,3](2,6)pyridinophane (L1, Fig. 1),^15a a neutral macrocyclic N₄ ligand, is well known to form iron complexes in the cis-configuration, including [Fe^III(L1)Cl₂]⁺ (1).^15b In 2010, we reported that [Fe^III(L1)Cl₂][FeCl₄] (1·FeCl₄) is an efficient catalyst for the cis-dihydroxylation of alkenes using Oxone (potassium peroxymonosulfate, 2KHSO₅·KHSO₄·K₂SO₄) as the oxidant, and the generation of an [Fe^V(L1)(O)₂]⁺ intermediate in the reaction was inferred from high resolution ESI-MS analysis and DFT calculations.^16a This prompted us to examine the catalytic activity of 1 toward water oxidation. It is noted that ligand L1 has also been employed for developing the oxidation chemistry of other transition metal complexes, including [Os^III(L1)(OH)(OH₂)]²⁺, which can catalyse the cis-dihydroxylation of alkenes by H₂O₂via a reactive [Os^V(L1)(O)(OH)]²⁺ intermediate,^16c and [Pd^II(L1)(Me)₂], which reacts with oxygen to generate [Pd^III(L1)(Me)₂(OO˙)] followed by protonation to give [Pd^IV(L1)(Me)₂(OOH)]⁺.^16b Hydrogen peroxide disproportionation catalysed by [Mn(L1)(H₂O)₂]²⁺ and electrochemical oxidation of water catalysed by [Mn(L1′)(H₂O)₂]²⁺ (L1′ = the N-^tBu counterpart of L1) have been reported as well.¹⁷

In the present work, we report the use of [Fe^III(L1)Cl₂]⁺ (1) for water oxidation with Oxone, as well as CAN and NaIO₄, under mild conditions, together with mechanistic studies by means of high-resolution ESI-MS, UV-vis absorption spectroscopy, ¹⁸O-labelling experiments, kinetic studies, cyclic voltammetry, EPR analysis, and DFT calculations. During the course of this study, Sun and co-workers communicated their findings on water oxidation with CAN catalysed by [Fe^II(L1)(MeCN)₂]²⁺.^6g Detailed mechanistic studies on water oxidation catalysed by Fe–L1 systems have not been reported in the literature. Also, Oxone has not been used as an oxidant in previously reported water oxidation reactions catalysed by iron complexes,^6,10,11,13 despite literature reports on the manganese-catalysed oxidation of water with Oxone.^2c,d The experimental studies and DFT calculations in the present work point to the generation of an Fe^V=O species responsible for O–O bond formation in water oxidation catalysed by Fe–L1 systems.

Results

Water oxidation catalysed by complex 1

At the outset, a series of iron complexes bearing N₃, N₄ and N₅ ligands (e.g. 1,4,7-trimethyl-1,4,7-triazacyclononane, 2,2′:6′:2′′-terpyridine, L1, and 2,2′:6′,2′′:6′′,2′′′:6′′′,2′′′′-quinquepyridine) were screened for catalytic water oxidation with CAN as oxidant under different reaction conditions (Table S1 in the ESI†). We found that [Fe^III(L1)Cl₂][FeCl₄] (1·FeCl₄) exhibited the best performance, affording oxygen with TONs of up to 41 and 32 in 0.1 M HNO₃ and in pure water, respectively, after 30 minutes with CAN (840 equiv.) as oxidant (Table 1, entries 1d, 1e). With NaIO₄ as oxidant, the reaction in 0.1 M HNO₃ afforded oxygen with TON of 12, whereas oxygen evolved with TON of only 3 from the reaction in pure water (Table 1, entries 2a and 2b). With Oxone as oxidant, under the conditions of 100 μM 1 and 84 mM Oxone in 0.1 M HNO₃, oxygen was produced with TON of 89 in 30 minutes (Table 1, entry 3a); lowering the concentration of 1 to 12.5 μM increased the TON to 113 (Table 1, entry 3b), a value higher than those obtained using CAN or NaIO₄ as oxidant (TONs of 93 and 44 respectively, entries 1f and 2c in Table 1). Control experiments using [Et₄N][FeCl₄] instead of 1·FeCl₄ as the catalyst did not produce detectable amounts of oxygen for the reactions with CAN and NaIO₄, and afforded oxygen with TON of only 1.2 for the reaction with Oxone, revealing that 1 is the key species accounting for the catalytic activity of 1·FeCl₄. With 1·ClO₄ as catalyst, similar oxygen evolution to that catalysed by 1·FeCl₄ was observed (e.g. TON of O₂: 40 vs. 41 with 84 mM CAN as oxidant, 11 vs. 12 with 84 mM NaIO₄ as oxidant, and 91 vs. 89 with 84 mM Oxone as oxidant for reactions using 100 μM catalyst in 0.1 M HNO₃).

Table 1 Catalytic water oxidation by 1 under different reaction conditions^a

Entry	Oxidant	[Catalyst] (μM)	[Oxidant] (mM)	Medium	TON of O2^b	OE^c (%)
a All reactions were performed under argon atmosphere and in 0.1 M HNO3 or water at room temperature for 30 min. b Determined by GC integration. c Based on the stoichiometric ratio (1 : 4) of O2 formed per oxidant in 4e^− oxidation; oxidant efficiency (OE) = [no. of moles of O2 formed/(no. of moles of oxidant used/n)] × 100%, where n = 4 for CAN and n = 2 for NaIO4 and Oxone.
1a	CAN	100	10	0.1 M HNO3	17	68
1b	CAN	100	20	0.1 M HNO3	19	38
1c	CAN	100	42	0.1 M HNO3	31	30
1d	CAN	100	84	0.1 M HNO3	41	20
1e	CAN	100	84	H2O	32	15
1f	CAN	12.5	125	0.1 M HNO3	93	4
1g	CAN	12.5	125	H2O	81	3
2a	NaIO4	100	84	0.1 M HNO3	12	3
2b	NaIO4	100	84	H2O	3	1
2c	NaIO4	12.5	125	0.1 M HNO3	44	1
2d	NaIO4	12.5	125	H2O	3	0.1
3a	Oxone	100	84	0.1 M HNO3	89	21
3b	Oxone	12.5	125	0.1 M HNO3	113	2

---

The origin of the oxygen gas produced from the reaction mixture of 1-catalysed water oxidation was investigated by using a mixture of H₂¹⁶O and H₂¹⁸O (v/v = 1 : 1; the H₂¹⁸O used had 97 atom% ¹⁸O) as the medium and by analysing the gaseous product(s) using GC-MS. With CAN as oxidant, the gaseous products obtained in the first 5 minutes were a mixture of ¹⁶O₂/¹⁶O¹⁸O/¹⁸O₂ with a ratio of 30.0 : 49.6 : 20.4, which is close to the theoretical value of 26.5 : 50.0 : 23.5 (ref. 6b) expected for oxygen gas derived from water. In the case of NaIO₄ as oxidant, similar ¹⁸O-labelling studies using H₂¹⁸O were not performed, because the oxygen atoms of IO₄⁻ undergo rapid exchange with those of water.¹⁸ However, the exchange of oxygen atoms between Oxone and water is sufficiently slow for useful ¹⁸O-labelling studies.¹⁹ Water oxidation with Oxone (84 mM) in a 1 : 1 mixture of H₂¹⁸O and 0.2 M HNO₃ gave oxygen with a ¹⁶O₂/¹⁶O¹⁸O/¹⁸O₂ ratio of 68.9 : 29.9 : 1.2 in the first 5 minutes. This ratio falls within the range of (49–91.9) : (7.6–39) : (0.51–12) reported for water oxidation with Oxone catalysed by a Mn^III(O)₂Mn^IV complex.¹⁹

Kinetic studies

The time courses of 1-catalysed water oxidation with CAN in 0.1 M HNO₃ using different concentrations of CAN (12.5–125 mM) and 1 (6.25–25 μM) were examined; the plots of oxygen evolution (measured by GC) over time are depicted in Fig. S1 and S2 in the ESI.† A linear dependence of the initial rate of oxygen evolution on both concentration of CAN and concentration of 1 was observed, as depicted in Fig. 2.

Fig. 2 (A) Plot of initial rate of O₂ evolution against different [CAN] (12.5–125 mM) at fixed [1] (12.5 μM) in 0.1 M HNO₃. (B) Plot of initial rate of O₂ evolution against different [1] (6.25–25.0 μM) at fixed [CAN] (125 mM) in 0.1 M HNO₃.

For 1-catalysed water oxidation with NaIO₄, the time course (0–60 min) plots of oxygen evolution at different concentrations of NaIO₄ (12.5–125 mM) and 1 (6.25–25 μM) in 0.1 M HNO₃ are shown in Fig. S3 and S4,† respectively. In this case, the initial rate of oxygen evolution showed a linear dependence on the concentration of 1 but not on the concentration of NaIO₄ (see Fig. S5†).

We then examined the time courses of oxygen evolution during water oxidation with Oxone (37.5–125 mM) catalysed by 1 (6.25–25 μM) in 0.1 M HNO₃ (Fig. S6 and S7†). Again, a linear correlation between the initial rate of oxygen evolution and the concentration of 1 was observed, whereas the initial rate was not significantly dependent on Oxone concentration (Fig. S8†), which is analogous to the findings obtained with NaIO₄ as oxidant.

High-resolution ESI-MS analysis

In our previous work, analysis of a reaction mixture of [Fe(L1)Br₂]⁺ and Oxone (8 equiv.) in MeCN–H₂O (5 : 1 v/v) by high-resolution ESI-MS revealed a cluster peak at m/z 356.0981, attributed to [Fe^V(L1)(O)₂]⁺.^16a In this work, in order to detect the reaction intermediates in the oxidation reactions of 1 with CAN and NaIO₄ in aqueous solution, we performed ESI-MS measurements on the corresponding reaction mixtures.

CAN as oxidant. Before examination of the reaction intermediates in 1-catalysed water oxidation with CAN, high resolution ESI-MS was employed to analyse an aqueous solution of 1, and revealed a major cluster peak at m/z 358.0906, attributed to [Fe^III(L1)(OH)₂]⁺ (calcd m/z 358.1092) based on the isotopic distribution, collision-induced dissociation, and ¹⁸O-labelling (Fig. S9–S12†). We then performed high-resolution ESI-MS analysis of a reaction mixture of 1 in H₂O after addition of CAN (200 equiv.) for 30 seconds; the spectrum showed three new cluster peaks at m/z 357.0992, 402.0854 and 448.0815 (Fig. S13B†), assignable to [Fe^IV(L1)(O)(OH)]⁺, [Fe^IV(L1)(O)(NO₃)]⁺, and [Fe^III(L1)(NO₃)₂]⁺, respectively. The ESI-MS spectrum for the reaction in 0.1 M HNO₃ (Fig. S13C†) resembles that obtained for the reaction in H₂O. The isotopic distributions of the three new cluster peaks are depicted in Fig. 3 and in Fig. S14† (for their collision-induced dissociation, see Fig. S15 and S16†). The new cluster peaks at m/z 357.0992 and 402.0854 shifted to m/z 361.1107 and 404.0927, respectively, when the reaction of 1 with CAN (200 equiv.) was conducted in H₂¹⁸O (instead of H₂¹⁶O); the corresponding isotopic patterns and collision-induced dissociation spectrum obtained in the ¹⁸O-labelling study are shown in Fig. S17–S19.†

Fig. 3 Cluster peaks at m/z 357.0992 and 402.0854 observed during high-resolution ESI-MS analysis of a reaction mixture of 1 and CAN (200 equiv.) in H₂O. Simulated isotopic patterns for (A) [Fe^IV(L1)(O)(OH)]⁺ and (B) [Fe^IV(L1)(O)(NO₃)]⁺ are shown.

NaIO₄ as oxidant. ESI-MS analysis of a reaction mixture of 1 and NaIO₄ (800 equiv.) in 0.1 M HNO₃ during the first 30 seconds of the reaction revealed new species at m/z 356.0944 (Fig. 4) and 373.0940 (Fig. S20†), which can be respectively assigned to [Fe^V(L1)(O)₂]⁺ and [Fe^III(L1)(OO˙)(OH)]⁺. Collision-induced dissociation spectra of the two species are depicted in Fig. S21 and S22.† When the reaction of 1 with NaIO₄ (800 equiv.) was conducted in H₂O under similar conditions, a new peak at m/z 356.0937 assignable to [Fe^V(L1)(O)₂]⁺ was formed within 10 seconds (Fig. S23†). By carrying out the same reaction in H₂¹⁸O instead of H₂¹⁶O, a shift of the m/z 356.0937 signal to m/z 360.1042 was observed (Fig. S24†).

Fig. 4 Cluster peak at m/z 356.0944 observed during high-resolution ESI-MS analysis of the reaction mixture of 1 with NaIO₄ (800 equiv.) in 0.1 M HNO₃. Simulated isotopic patterns for (A) [Fe^III(L1)(OH)₂]⁺ and (B) [Fe^V(L1)(O)₂]⁺ are shown.

UV-vis absorption spectroscopy

The reaction of 1 (1.5 mM) with 5 equiv. of CAN in 0.1 M HNO₃ at room temperature was monitored by UV-vis absorption spectroscopy. This reaction immediately (within 10 seconds) generated a new species with λ_max 830 nm and a shoulder near 540 nm (Fig. 5). The new species decayed rapidly with a half-life of 107 seconds (Fig. 5, inset).

Fig. 5 UV-vis absorption spectral changes for the reaction of 1 (1.5 mM) with CAN (7.5 mM) in 0.1 HNO₃ at room temperature. Inset: time course of the decay of the absorption band at 830 nm.

For the reaction of 1 (1.5 mM) with 5 equiv. of NaIO₄ or Oxone in 0.1 M HNO₃, UV-vis measurements revealed different phenomena from that observed for the ‘1 + CAN’ system. Upon treatment of 1 with NaIO₄, a new species with λ_max 830 nm and a shoulder near 540 nm was gradually formed over 10 minutes (Fig. S25†) and decayed much more slowly with a half-life of ∼40 minutes (see the inset of Fig. S26†), in contrast to the CAN counterparts (Fig. 5). In the case of Oxone, the new band at ∼830 nm was barely discernible in the UV-vis spectrum of the reaction mixture, being much weaker than that observed for the reaction with NaIO₄.

EPR analysis

EPR spectroscopy was employed to examine the reaction mixture of 1-catalysed water oxidation, using NaIO₄ oxidant as an example. Complex 1 is a high-spin Fe^III complex with S = 5/2;^15b however, the X-band EPR spectrum of 1·ClO₄ in 0.1 M HNO₃ recorded at 6 K did not show an appreciable signal (Fig. 6, black line), presumably due to quick relaxation. The X-band EPR spectrum of the reaction mixture obtained by mixing 1·ClO₄ with NaIO₄ (10 equiv.) in 0.1 M HNO₃ at room temperature followed by immediate cooling to 6 K is depicted in Fig. 6 (red line). In this EPR spectrum, there are two prominent signals, one with g_avg = 5.54 and the other with g ≈ 2.

Fig. 6 X-band EPR spectra (at 6 K) of 1·ClO₄ in 0.1 M HNO₃ (black line), and a reaction mixture of 1·ClO₄ (1 mM) and NaIO₄ (10 mM) in 0.1 M HNO₃ (red line).

Electrochemistry of complex 1

In order to avoid interference arising from the electrochemical reactions of the FeCl₄⁻ anion in 1·FeCl₄, we used the ClO₄⁻ salt of 1 and examined its electrochemical properties in 0.1 M HNO₃ at pH 1 and in buffered solutions at various pH by means of cyclic voltammetry and rotating disk voltammetry using glassy carbon as the working electrode. The cyclic voltammograms of 1·ClO₄ at pH 1–6 are depicted in Fig. 7A and S27,† and show a reversible couple I at E_1/2 +0.46 V (pH 1) to +0.12 V (pH 6) vs. SCE, a small irreversible oxidation wave II at E_pa +1.18 V (pH 1) to +0.86 V (pH 6) vs. SCE, and the onset of a catalytic oxidation wave at +1.4 V (pH 1) to ca. +1.1 V (pH 6) vs. SCE. Both E_1/2 of the reversible couple I and E_pa of the irreversible wave II shift cathodically by ∼67 mV per pH unit upon increasing the pH (Fig. 7B and S28†). The magnitude of the irreversible wave II was found to be sensitive to the electrode surface, but could be reproducibly observed during both the cyclic voltammetric scans and the rotating disk voltammetric experiments. The linear scan voltammograms of 1·ClO₄ at pH 1, 3, and 5 obtained from rotating disk voltammetric measurements are depicted in Fig. 7C and S29,† and that recorded at pH 1 at various rotation rates is shown in Fig. S30.† For example, at pH 3, a small but distinct current was recorded at ca. +1.05 V vs. SCE (Fig. 7C), which could be correlated to the small irreversible wave II in the corresponding cyclic voltammogram (Fig. 7A). Notably, the catalytic oxidation wave, presumably due to water oxidation, can be observed at pH 1 to 6 (Fig. 7A and S27†). The limiting current (i_L) of the redox process corresponding to the reversible couple I increased linearly with the square root of the rotation rate (ω^1/2). A plot of i_L against ω^1/2 (Levich plot) shows a straight line with R² = 0.999 (Fig. S31†). This is indicative of a totally mass-transfer-limited condition at the electrode surface. From the slope of the Levich plot, the diffusion coefficient of 1·ClO₄ under the experimental conditions was calculated to be 0.48 × 10⁻⁵ cm² s⁻¹.

Fig. 7 (A) Solid line: cyclic voltammogram of 1·ClO₄ in 0.1 M acetate buffer at pH 3; dotted line: buffer background; working electrode: glassy carbon; scan rate: 100 mV s⁻¹. (B) Redox potentials (E_1/2 of reversible couple I) of 1·ClO₄ in solutions of various pH. (C) Linear scan voltammogram of 1·ClO₄ in 0.1 M acetate buffer at pH 3. Working electrode: rotating glassy carbon disk; rotation rate: 100 rpm; scan rate: 5 mV s⁻¹ (inset: magnified region at approximately 1.0 V).

Density functional theory calculations

Electrochemical potentials of iron–oxo complexes of L1. The electrochemical potentials of iron–oxo complexes [Fe(L1)(O)(X)]ⁿ⁺ (X = H₂O, HO⁻, O²⁻) in different oxidation states, together with that of the Fe^III/Fe^II couple of 1, were estimated using DFT calculations. To avoid direct calculation of the proton free energy and systematic errors, we computed the redox potentials through isodesmic reactions using the electrochemical proton-coupled electron transfer (PECT) reactions of the cis-(dioxo)ruthenium(VI) complex [Ru^VI(L2)(O)₂]²⁺ (L2 = N,N,N′,N′-tetramethyl-3,6-dimethyl-3,6-diazaoctane-1,8-diamine)²⁰ as references. The isodesmic reactions are based on the assumption that both the proton free energy and the systematic errors for each pair of redox reactions (Fe^V/Fe^IVvs. Ru^V/Ru^IV, Fe^IV/Fe^IIIvs. Ru^IV/Ru^III, and Fe^III/Fe^IIvs. Ru^III/Ru^II) are comparable and can be cancelled out in the calculations of ΔΔG1, ΔΔG2, and ΔΔG3 depicted in reactions (1a)–(3a) and (1b)–(3b).

From the DFT computed ΔΔG1, ΔΔG2 and ΔΔG3 values, and the experimental redox potentials of the Ru^V/Ru^IV, Ru^IV/Ru^III and Ru^III/Ru^II couples,²⁰ the redox potentials of the Fe^V/Fe^IV, Fe^IV/Fe^III and Fe^III/Fe^II couples can be calculated as:

E(Fe^V/Fe^IV) = E(Ru^V/Ru^IV) + ΔΔG1

E(Fe^IV/Fe^III) = E(Ru^IV/Ru^III) + ΔΔG2

E(Fe^III/Fe^II) = E(Ru^III/Ru^II) + ΔΔG3

where E(Ru^V/Ru^IV) = +0.72 V, E(Ru^IV/Ru^III) = +0.63 V, and E(Ru^III/Ru^II) = +0.26 V at pH 1 according to the experimental findings.²⁰

We calculated the Fe^III/Fe^II and Fe^IV/Fe^III redox potentials at different pH by using three commonly used density functionals (DFs): BPW91 (pure-GGA), B3LYP (hybrid-GGA), and M06L (meta-GGA). A correlation between the experimental redox potentials of Fe^III/Fe^II and Fe^IV/Fe^III and those calculated using these DFs is depicted in Fig. 8 (note of caution: calculated redox potentials are thermodynamic values). It is evident that BPW91 and B3LYP led to marked-to-severe (up to ca. +0.57 V) overestimation of the E_1/2 of Fe^III/Fe^II, although B3LYP showed good performance in the prediction of the electrochemical potential of Fe^IV/Fe^III. Only M06L gave good estimations of the electrochemical potentials of both Fe^III/Fe^II and Fe^IV/Fe^III at different pH with a linear fit (E(M06L) = E(expt) + 0.06, R = 0.998, SD = 0.03) between the calculated redox potentials and the experimental data. Hence, M06L was chosen for the subsequent DFT calculations in this work. In the literature, the M06L functional has been reported to show good performance in modelling water oxidation by ruthenium²¹ and iron^12a,c complexes.

Fig. 8 Correlation between experimental and calculated redox potentials. The E(expt) values for the Fe^III/Fe^II and Fe^IV/Fe^III couples were taken from Fig. 7 and S28† (see Discussion section). Note of caution: E(expt) values for Fe^IV/Fe^III are the E_pa values of the irreversible wave II.

For the calculation of the pK_a values, as an example, the pK_a of [Fe^V(L1)(O)(OH)]²⁺ was calculated based on the following pair of isodesmic reactions (reactions (4a) and (4b)).

pK_a([Fe^V(L1)(O)(OH)]²⁺) = pK_a([Ru^V(L2)(O)(OH)]²⁺) − ΔΔG4/2.303RT

where pK_a([Ru^V(L2)(O)(OH)]²⁺) = 1.8 according to the experimental study.²⁰

The calculated redox potentials for various Fe^V/Fe^IV, Fe^IV/Fe^III and Fe^III/Fe^II couples, including those of [Fe(L1)(O)(X)]ⁿ⁺ (X = H₂O, HO⁻, O²⁻), along with the calculated pK_a values, are depicted in Scheme 1, revealing the following features: (1) the pK_a of the [Fe^V(L1)(O)(OH)]²⁺/[Fe^V(L1)(O)₂]⁺ equilibrium is 3.0; (2) the calculated E_1/2 of [Fe^V(L1)(O)(OH)]²⁺/[Fe^IV(L1)(O)(OH₂)]²⁺ (+1.42 V vs. SCE at pH 1) is comparable to the E_1/2 of Ce^IV/Ce^III (+1.38 V vs. SCE) and close to the calculated E_1/2 values of [Fe^V(L)(O)(OH)]²⁺/[Fe^IV(L)(O)(OH₂)]²⁺ with L = ^Me2Pytacn or mep (+1.70 to +1.73 V vs. SHE,^12f i.e. +1.46 to +1.49 V vs. SCE); (3) the calculated E_1/2 of [Fe^IV(L1)(O)(OH₂)]²⁺/[Fe^III(L1)(OH)(OH₂)]²⁺ (+1.25 V vs. SCE at pH 1) matches the experimental E_pa value of Fe^IV/Fe^III (E_pa = 1.18 V vs. SCE at pH 1, see Discussion section); (4) the calculated E_1/2 of the Fe^III/Fe^II couple is +0.49 V at pH 1, which matches the experimental E_1/2 value for Fe^III/Fe^II (+0.46 V vs. SCE at pH 1, see Discussion section) well; (5) the calculated E_1/2 of the Fe^V/Fe^III couple (E_1/2 = +1.34 V vs. SCE at pH 1) coincides with the catalytic oxidation wave at +1.4 V observed in the cyclic voltammogram of 1 at pH 1. It is noted that there is just 170 mV difference in the calculated E_1/2 values of [Fe^V(L1)(O)(OH)]²⁺/[Fe^IV(L1)(O)(OH₂)]²⁺ and [Fe^IV(L1)(O)(OH₂)]²⁺/[Fe^III(L1)(OH)(OH₂)]²⁺ at pH 1.

Scheme 1 Calculated reduction potentials for Fe^V/Fe^IV, Fe^IV/Fe^III and Fe^III/Fe^II, together with calculated pK_a values, based on the M06L functional.

Electronic properties of [Fe^V(L1)(O)₂]⁺. The ground state of [Fe^V(L1)(O)₂]⁺ is a quartet state with three unpaired electrons; the lowest-lying excited state is a sextet state which is 2.8 kcal mol⁻¹ higher in energy than the ground state. A quartet ground state for an Fe^V=O species has previously been predicted by DFT calculations for [Fe^V(^Me2Pytacn)(O)(OH)]²⁺.^12f,14d The computed Fe=O distance in [Fe^V(L1)(O)₂]⁺ in the quartet state (⁴Fe^V) is 1.614 Å, which is shorter than that in the sextet state (⁶Fe^V, 1.695 Å) due to the antibonding character of the d1 and d2 orbitals (Fig. 9, right). Similar Fe=O distances were reported for [Fe^V(L1)(O)₂]⁺ (1.613–1.638 Å), computed at the B3LYP/6-31(G) (lanl2dz) level in our previous work, and for [Fe^V(^Me2Pytacn)(O)(OH)]²⁺ (1.63 Å)^12f and low-spin [Fe^V(TAML)(O)]⁻ (1.60 Å).^14a The spin density plot for [Fe^V(L1)(O)₂]⁺ calculated in this work is shown in Fig. 9 (left). Apart from the spin density on Fe of 2.3, each oxo group has considerable spin density of 0.27. The MO diagram of [Fe^V(L1)(O)₂]⁺ (Fig. 9, right) reveals a d³ configuration in accordance with the assignment of the Fe(V) oxidation state.

Fig. 9 Spin density plot (left) and MO diagram (right) of [Fe^V(L1)(O)₂]⁺.

Electronic properties of [Fe^IV(L1)(O)(OH)]⁺. The ground state of [Fe^IV(L1)(O)(OH)]⁺ is a quintet state with four unpaired electrons, and the triplet state is 16.0 kcal mol⁻¹ higher in energy. The computed Fe=O distance of [Fe^IV(L1)(O)(OH)]⁺ (⁵Fe^IV) in the ground state is 1.632 Å, very close to the experimental Fe=O distance in [Fe^IV(N4Py)(O)]²⁺ (1.639(5) Å).^5aFig. 10 (left) shows the spin density plot calculated in this work for [Fe^IV(L1)(O)(OH)]⁺. Fe has the largest spin density of 3.1; the oxo and hydroxyl groups have spin densities of 0.59 and 0.18, respectively. The MO diagram of [Fe^IV(L1)(O)(OH)]⁺ (Fig. 10, right) reveals a d⁴ configuration, which is consistent with the assignment of the Fe(IV) oxidation state.

Fig. 10 Spin density plot (left) and MO diagram (right) of [Fe^IV(L1)(O)(OH)]⁺.

Mechanism of water oxidation based on [Fe^V(L1)(O)₂]⁺. Both the quartet (S = 3/2, ⁴Fe^V) and sextet (S = 5/2, ⁶Fe^V) states of [Fe^V(L1)(O)₂]⁺ were considered in the mechanism studies. The doublet state (S = 1/2) of [Fe^V(L1)(O)₂]⁺ was found to be the most energetically unfavourable (16.4 kcal mol⁻¹ higher in energy than the quartet ground state) and was not considered in this work. Fig. 11 depicts the computed potential energy surfaces and key structure parameters of the stationary points. The oxidation of water starts from the reactant complex ⁴RC1/⁶RC1, formed between ⁴Fe^V/⁶Fe^V and the substrate (water molecule), which proceeds to the transition state ⁴TS1/⁶TS1 in which an O⋯O bond and an O⋯H bond are formed. The ground state is ⁴RC1, which is 2.8 kcal mol⁻¹ lower in energy than ⁶RC1. The free energy barrier (ΔG^‡) for the quartet state is 15.7 kcal mol⁻¹, lower than that for the sextet state (18.9 kcal mol⁻¹). Both activation barriers are comparable to those for the similar oxidation of water by [Fe^V(TAML)(O)]⁻ (20.0 kcal mol⁻¹, estimated from experimental data)^12a and [Fe^V(^Me2Pytacn)(O)(OH)]²⁺ (18.8 kcal mol⁻¹ obtained from DFT calculations).^12f As the reactions proceed through the transition state, there is a reversal of the energy levels of the sextet and quartet states, with ⁶INT1 being 6.4 kcal mol⁻¹ more stable than the quartet state ⁴INT1. The location of the minimum energy crossing point (MECP) between the sextet and quartet states was found using the code developed by Harvey and co-workers.²² The energy of the MECP is only 0.9 kcal mol⁻¹ higher than that of ⁴INT1, suggesting that spin state reversion can easily occur, leading to the more stable complex ⁶INT1.

Fig. 11 Potential energy surfaces (in kcal mol⁻¹), key structural parameters (in Å and degrees) and relative stabilities (in kcal mol⁻¹) of the stationary points. The energy of the MECP point (13.4 kcal mol⁻¹) is the average energy of the quartet and sextet states in water.

The ⁶INT1 complex can subsequently be oxidized by [Fe^V(L1)(O)₂]⁺ (⁴Fe^V) to give the superoxo complex [Fe^III(L1)(OO˙)(OH)]⁺ (⁷INT2/⁵INT2). Meanwhile, ⁴Fe^V is reduced to [Fe^IV(L1)(O)(OH)]⁺ (⁵Fe^IV). The subsequent reaction steps include exchange of an oxygen molecule by a water molecule and the release of ³O₂/¹O₂ (reaction (5)). The search for the transition state has not been successful. A barrier of 8.8 kcal mol⁻¹ for the release of O₂ from [TAML–Fe^IV–OO˙]⁻ was reported by Crammer and co-workers.^12a After release of O₂, complex [Fe^II(L1)(OH₂)(OH)]⁺ (⁵Fe^II) with a quintet state ground state is formed; this complex can further comproportionate with [Fe^IV(L1)(O)(OH)]⁺ (⁵Fe^IV) to regenerate two [Fe^III(L1)(OH)₂]⁺ (⁶Fe^III) molecules.

[Fe^III(L1)(OO˙)(OH)]⁺ + OH₂ → [Fe^II(L1)(OH₂)(OH)]⁺ + O₂ (5)

Discussion

The main objective of this work is to gain insight into the mechanism of water oxidation by iron complexes of tetradentate macrocyclic ligands with a cis non-planar coordination geometry. The macrocyclic N₄ ligand L1 (which is well known to form iron complexes in the cis configuration^15b) was chosen, as our previous work had suggested the involvement of in situ generated high-valent Fe=O species of L1, including [Fe^V(L1)(O)₂]⁺, in the cis-dihydroxylation of alkenes with Oxone catalysed by [Fe^III(L1)Cl₂]⁺ (1),^16a and Fe^V=O species were proposed to be the reactive intermediates in iron-catalysed water oxidation reactions reported in the literature.^6a,b Also, it is envisaged that the strong donor strength of L1 would stabilize iron–oxo species against demetallation in solution.

During optimization of the reaction conditions, it was noted that although the turnover number of oxygen increased with oxidant concentration, the oxidant efficiency (defined in note c of Table 1) decreased from 68% to 20% (Table 1, entries 1a–1d). Such a significant decrease in oxidant efficiency is attributed to partial decomposition/deactivation of the catalyst under strongly oxidizing reaction conditions.²³ Similar findings were observed when NaIO₄ or Oxone was used as oxidant.

Unlike the ‘[Fe(bqen)(OTf)₂] + CAN’ system (which can oxidize water to O₂ in non-buffered aqueous solution but not in 0.1 M HNO₃ solution),^6e the ‘1 + CAN’ system produced oxygen with comparable TONs in water and in 0.1 M HNO₃ (entries 1d and 1e in Table 1), consistent with the similar ESI mass spectra obtained under the two conditions (Fig. S13†). This reflects the remarkable stability of 1 toward ligand dissociation under acidic conditions, like other iron complexes such as [Fe^II(mcp)(OTf)₂]^6b and [Fe^II(L1)(MeCN)₂]²⁺ (ref. 6g) used as water oxidation catalysts at pH 1. The 1-catalysed water oxidation was sensitive to the oxidant, with the TON of oxygen evolved under the same conditions (in 0.1 M HNO₃) following the order Oxone > CAN > NaIO₄ (113 vs. 93 vs. 44, see Table 1, entries 1f, 2c and 3b). According to the ¹⁶O₂/¹⁶O¹⁸O/¹⁸O₂ ratio (68.9 : 29.9 : 1.2) obtained in the ¹⁸O-labelling studies, the oxygen evolved in the ‘1 + Oxone’ system originated not only from Oxone but also from water (see the possible pathways of oxygen evolution, along with the corresponding calculated ¹⁶O₂/¹⁶O¹⁸O/¹⁸O₂ ratios, depicted in Scheme S1†), indicating the involvement of water oxidation in the reaction.

Considering the excellent catalytic ability of [Fe^II(mcp)(OTf)₂] for water oxidation, as reported by Costas, Lloret-Fillol and co-workers,^6b we compared the activity of catalysts [Fe^II(mcp)Cl₂]^6b and [Fe^III(L1)Cl₂]⁺ (1) for water oxidation. Under our reaction conditions with catalyst/oxidant = 1 : 840, the amounts of oxygen formed in [Fe^II(mcp)Cl₂]-catalysed water oxidation in 0.1 M HNO₃ with CAN and NaIO₄ as oxidants were approximately 1.4 times (TON = 56) and 2.9 times (TON = 35) the amounts formed in the 1-catalysed reactions, respectively. Changing the catalyst/oxidant ratio from 1 : 840 to 1 : 10 000 led to an increase in the TON for oxygen formation from 41 to 93 for ‘1 + CAN’, from 56 to 177 for ‘[Fe^II(mcp)Cl₂] + CAN’, from 12 to 44 for ‘1 + NaIO₄’, and from 35 to 114 for ‘[Fe^II(mcp)Cl₂] + NaIO₄’. These results showed that [Fe^II(mcp)Cl₂] is more active than 1 in catalysing water oxidation.

Iron oxide nanoparticles, which can be generated from simple iron salts or iron complexes at high pH,^6c have been shown to be efficient catalysts for water oxidation.^6c,24 For the 1-catalysed water oxidation with CAN, NaIO₄, or Oxone in 0.1 M HNO₃ (pH 1), the possibility that the observed catalytic activity of 1 is due to Fe₂O₃ nanoparticles generated in situ could be excluded, as in acidic solution either formation of Fe₂O₃ nanoparticles would be prevented or any Fe₂O₃ nanoparticles formed would be converted to Fe³⁺(aq) ions. It has been reported that Fe³⁺(aq) ions are not active in catalysing water oxidation with CAN at low pH.^6c

Electrochemical studies of ruthenium-based molecular water oxidation catalysts often revealed strong catalytic currents due to water oxidation occurring at potentials corresponding to the generation of oxidizing Ru–oxo complexes.²⁵ In general, electrochemical oxidation of Ru–OH₂ to Ru=O species would be kinetically slow as a result of deprotonation of Ru–OH₂ prior to oxidation. Hence the working electrode is well documented to play a crucial role in the electrochemical reversibility of the Ru=O/Ru–OH₂ couple(s). Glassy carbon or pyrolytic carbon is commonly used as a working electrode for electrochemical studies of Ru=O complexes in aqueous solutions. While there have been numerous reports on the electrochemistry of Ru=O/Ru–OH₂ couples in aqueous solution, to the best of our knowledge, related studies on the electrochemical oxidation of Fe–OH₂ to Fe=O species in aqueous solution are sparse.²⁶

The cyclic voltammetric and rotating disk voltammetric measurements of 1 at various pH provide useful information on the electrochemical potentials of the iron–oxo species of L1. As seen in Fig. 7B and S28,† the plots of the redox potentials against pH (from 1 to 6) for both couple I and wave II show negative slopes, and the redox potential decreases by ∼67 mV per pH unit, lending support to the two oxidation processes which are due to one electron and one proton transfer reactions. These findings, together with the experimental data obtained from rotating disk voltammetry (Fig. 7C and S29†) and the results of the DFT calculations (Fig. 8 and Scheme 1), support the assignment of couple I to the [Fe^III(L1)(OH)(OH₂)]²⁺/[Fe^II(L1)(OH₂)₂]²⁺ redox couple and the assignment of wave II to the oxidation of [Fe^III(L1)(OH)(OH₂)]²⁺ to [Fe^IV(L1)(O)(OH₂)]²⁺. Despite the low scan rate of 5 mV s⁻¹ for the rotating disk voltammetry, the magnitude of the current recorded for the [Fe^IV(L1)(O)(OH₂)]²⁺/[Fe^III(L1)(OH)(OH₂)]²⁺ oxidation wave is not the same as that recorded for the [Fe^III(L1)(OH)(OH₂)]²⁺/[Fe^II(L1)(OH₂)₂]²⁺ redox process. As the oxidation of Fe^III–OH to an Fe^IV=O species involves the deprotonation of an Fe–OH unit, the electrochemical oxidation would be kinetically slow and highly sensitive to the electrode surface. Indeed, the electrochemical oxidation of Ru–OH₂/Ru–OH to a Ru=O species is well documented to be significantly affected by the nature/surface of the working electrode.²⁷ At pH 1 to 6, the difference between the observed redox potentials for wave II (Fe^IV/Fe^III) and couple I (Fe^III/Fe^II) is about 700 mV. For a ruthenium complex (also with cis vacant sites) supported by the mcp ligand,²⁸ the difference between the redox potentials of the [Ru^IV(mcp)(O)(OH₂)]²⁺/[Ru^III(mcp)(OH)(OH₂)]²⁺ and [Ru^III(mcp)(OH)(OH₂)]²⁺/[Ru^II(mcp)(OH₂)₂]²⁺ couples is 550 mV, which is 150 mV smaller than that between the iron analogues and can be attributed to the stronger π-bond formed in the Ru^IV=O moiety. There is a previous report on the observation of a pH-dependent reversible couple assigned as a proton-coupled electron transfer Fe^IV/Fe^III couple for [Fe^IV(N4Py)(O)]²⁺ (N4Py = N,N-bis(2-pyridylmethyl)-bis(2-pyridyl)methylamine) in buffered aqueous solution (pH 1.5–4) with an E_1/2 value of +0.41 V vs. SCE at pH 4 (the E_1/2 value anodically shifted upon decreasing the pH with a shift of 55 mV per pH unit).²⁹ However, the Fe^III/Fe^II couple was not reported for [Fe^IV(N4py)(O)]²⁺. The E_1/2 value of +0.41 V for the [Fe^IV(N4py)(O)]²⁺/[Fe^III(N4py)(OH)]²⁺ couple in aqueous solution at pH 4 is unexpectedly low. Should this be the case, the corresponding Fe^III/Fe^II couple would have to occur at a potential less than 0.0 V vs. SCE based on the DFT calculations in this work. Our DFT calculations using the M06L functional gave redox potentials of +1.21 V vs. SCE for the [Fe^IV(N4py)(O)]²⁺/[Fe^III(N4py)(OH)]²⁺ couple and +0.3 V vs. SCE for the [Fe^III(N4py)(OH)]²⁺/[Fe^II(N4py)(OH₂)]²⁺ couple at pH 1. The observation of a catalytic wave for 1 (Fig. 7A and S27†) beyond the irreversible wave II (assigned to Fe^IV/Fe^III) at, for example, E_pa +1.18 V vs. SCE (pH 1), which is comparable to the DFT-calculated potential of +1.25 V vs. SCE (pH 1) for the [Fe^IV(L1)(O)(OH₂)]²⁺/[Fe^III(L1)(OH)(OH₂)]²⁺ couple, corroborates the finding that 1 can catalyse water oxidation through iron–oxo species at oxidation states beyond Fe^IV.

ESI-MS analysis of a solution of 1 in water (Fig. S9–S12†), which revealed a major cluster peak assigned to [Fe^III(L1)(OH)₂]⁺, suggests rapid exchange of the Cl⁻ ligands of 1 with solvent. The new signals, generated upon addition of CAN, at m/z 357.0992, 402.0854, and 448.0815 can be attributed to [Fe^IV(L1)(O)(OH)]⁺ (calcd m/z 357.1014, Fig. 3), [Fe^IV(L1)(O)(NO₃)]⁺ (calcd m/z 402.0865, Fig. 3) and [Fe^III(L1)(NO₃)₂]⁺ (calcd m/z 448.0794, Fig. S14†), respectively, based on their m/z values and isotopic patterns, together with the ¹⁸O-labelling experiments which revealed new signals with m/z 361.1107 and 404.0927 attributable to [Fe^IV(L1)(¹⁸O)(¹⁸OH)]⁺ (calcd m/z 361.1099, Fig. S17†) and [Fe^IV(L1)(¹⁸O)(N¹⁶O₃)]⁺ (calcd m/z 404.0907, Fig. S19†), respectively. Species [Fe^IV(L1)(¹⁸O)(¹⁶OH)]⁺ and [Fe^IV(L1)(¹⁶O)(¹⁸OH)]⁺ (calcd m/z 359.1057) were not detected in the experiment, indicating that the oxygen atoms of the hydroxide and oxo ligands in [Fe^IV(L1)(O)(OH)]⁺ come from water and not from CAN.

In the UV-vis absorption spectra of the reaction mixture of 1 with CAN (Fig. 5), the new band at λ_max 830 nm is tentatively assigned to an Fe^IV=O species of L1, with reference to the characteristic bands of Fe^IV=O species (ranging from 700–850 nm) reported in the literature.³⁰ This assignment is in line with the detection of species formulated as [Fe^IV(L1)(O)(OH)]⁺ and [Fe^IV(L1)(O)(NO₃)]⁺ by ESI-MS. We further examined the intensity of the cluster peak at m/z 357.0992, attributed to [Fe^IV(L1)(O)(OH)]⁺, at different reaction times. The ion count of this signal decreased with reaction time, as depicted in Fig. S32,† which correlates with the decay of the absorption band at λ_max 830 nm assigned to the Fe^IV=O species in the UV-vis spectra shown in Fig. 5. Fe^IV=O species of chelating N-donor ligands have been reported to be reasonably stable, with half-lives generally ranging from 2 to 60 hours.³⁰ Unlike the [Fe^IV(mcp)(O)(H₂O)]²⁺ species (generated in situ from [Fe^II(mcp)(OTf)₂] and CAN in aqueous solution), which could persist under catalytic conditions with a half-life of 2.4 hours,^6b the Fe^IV=O species of L1 decayed with a half-life of 107 seconds under these conditions (Fig. 5). Such a fast decay may suggest that either it is taking part in a further reaction, or it is decomposing rapidly.

Given the instant generation of an Fe^IV=O species upon oxidation of 1 with CAN in aqueous solution, as supported by ESI-MS and UV-vis analysis, together with the absence of an induction period for oxygen evolution, an Fe^IV=O species is suggested to be involved in water oxidation by the ‘1 + CAN’ system. Kinetic studies revealed a linear dependence of the initial rate of oxygen evolution on the concentrations of both CAN and 1 (Fig. 2; increasing [CAN] to 125 mM lowered the pH to ∼0.6, and such pH variation alone accounted for ∼18% of the rate increase depicted in Fig. 2A), which suggests that a key step of the reaction involves an iron species, presumably [Fe^IV(L1)(O)(OH)]⁺, and one equivalent of CAN. One of the possibilities is the oxidation of [Fe^IV(L1)(O)(OH)]⁺ by CAN to give Fe^V=O species, such as [Fe^V(L1)(O)(OH)]²⁺ or [Fe^V(L1)(O)₂]⁺, and another possibility is the reaction of [Fe^IV(L1)(O)(OH)]⁺ with CAN to form an O=Fe^IV–O–Ce^IV species similar to that recently reported for the ‘[Fe^II(mcp)(OTf)₂] + CAN’ system,¹³ although Fe^V=O and O=Fe^IV–O–Ce^IV species have not been clearly detected in ESI-MS analysis of the reaction mixture of 1 with CAN in H₂O.

For water oxidation by the ‘1 + NaIO₄’ system, the new signals at m/z 356.0944 (major) and 373.0940 (minor), revealed by ESI-MS analysis of a mixture of 1 and NaIO₄ in 0.1 M HNO₃ (Fig. 4 and S20–S22†), could be assigned to [Fe^V(L1)(O)₂]⁺ (calcd m/z 356.0936) and [Fe^III(L1)(OO˙)(OH)]⁺ (calcd m/z 373.0963), respectively. The assignment of [Fe^V(L1)(O)₂]⁺ is supported by ¹⁸O-labelling studies, revealing a shift of the signal at m/z 356.0944 to m/z 360.1042 attributable to [Fe^V(L1)(¹⁸O)₂]⁺ (calcd m/z 360.1021) upon changing the reaction medium to H₂¹⁸O (Fig. S24†). Since IO₄⁻ can undergo rapid oxygen exchange with water,¹⁸ the oxo ligands of [Fe^V(L1)(¹⁸O)₂]⁺ could come from H₂¹⁸O directly and/or from ¹⁸O-incorporated IO₄⁻. Changing the reaction medium from 0.1 M HNO₃ to pure water reduced the intensity of the signal assigned to [Fe^V(L1)(O)₂]⁺; this observation is in line with the smaller TON of oxygen produced in pure water than in 0.1 M HNO₃ (entries 2b vs. 2a, and 2d vs. 2c, Table 1).

The DFT calculations on [Fe^V(L1)(O)₂]⁺ in this work and on [Fe^V(^Me2Pytacn)(O)(OH)]²⁺ in the literature^12f,14d revealed that the Fe^V=O species supported by these neutral chelating N ligands all adopt a quartet ground state (S = 3/2), different from the doublet ground state (S = 1/2) of [Fe^V(TAML)(O)]⁻ complexes, which bear tetraanionic tetraamide ligands and have been detected by EPR spectroscopy.^7,14a,j To further test the validity of the DFT method used in this work, we performed DFT calculations on [Fe^V(TAML)(O)]⁻ using the M06L functional, which revealed a doublet ground state for this species, consistent with previous DFT calculations reported in the literature.^12a,14a To the best of our knowledge, no Fe^V=O species with S = 3/2 ground state has been characterized by EPR spectroscopy. The X-band EPR spectrum of the reaction mixture of 1 with NaIO₄ in 0.1 M HNO₃ (Fig. 6) is dominated by two signals corresponding to S = 5/2 and S = 1/2 states; the former could be attributed to a high-spin Fe(III) species whereas the latter probably arose from decomposed NaIO₄, as a similar S = 1/2 signal was observed in the X-band EPR spectrum of a solution of NaIO₄ in 0.1 M HNO₃ recorded at 7 K (Fig. S33†). EPR signals of S = 3/2 Fe complexes are rather close to, or considerably overlap with, those of S = 5/2 ones.³¹ The prominent broad S = 5/2 signal in Fig. 6 would therefore render it difficult to clearly confirm the presence of the [Fe^V(L1)(O)₂]⁺ species by EPR since the signal of this S = 3/2 Fe^V=O species, which is likely to have a low concentration, could easily be masked by the S = 5/2 signal.

Generation of [Fe^V(L1)(O)₂]⁺ from the oxidation of 1 with NaIO₄ resembles the generation of [Fe^V(L1)(O)₂]⁺ from Oxone;^16a both NaIO₄ and Oxone are typically two-electron oxidants and can directly oxidize Fe(III) to Fe(V). UV-vis spectroscopy revealed that the reaction of 1 with NaIO₄ or Oxone is different from the reaction of 1 with CAN, as exemplified by the immediate formation of a new band at λ_max 830 nm attributable to an Fe^IV=O species in the ‘1 + CAN’ system (Fig. 5) but gradual formation of a much weaker band at λ_max 830 nm over 10 minutes in the ‘1 + NaIO₄’ system (Fig. S25†). Also, for both the ‘1 + NaIO₄’ and ‘1 + Oxone’ systems, kinetic studies revealed that the initial rate of oxygen evolution showed a linear dependence on the concentration of 1 but was relatively insensitive to the concentration of NaIO₄ (Fig. S5†) or Oxone (Fig. S8†). This supports exclusion of the possibility of an “oxo–oxo coupling” reaction between two Fe=O species.

On the basis of the above findings, a mechanism involving Fe^IV=O and/or Fe^V=O intermediate(s) for the 1-catalysed water oxidation with CAN or NaIO₄ is proposed as depicted in Scheme 2. For the corresponding reaction with Oxone, a mechanism similar to that of the reaction with NaIO₄ could be proposed. The iron–oxo species depicted in Scheme 2 are [Fe^IV(L1)(O)(OH)]⁺ and [Fe^V(L1)(O)₂]⁺ based on ESI-MS analysis; their protonated forms [Fe^IV(L1)(O)(OH₂)]²⁺ and [Fe^V(L1)(O)(OH)]²⁺ could be involved as well (note also the computed pK_a values shown in Scheme 1), but are not included in Scheme 2. Similar to [Fe^V(L1)(O)₂]⁺, [Fe^V(L1)(O)(OH)]²⁺ also adopts a quartet ground state, with the doublet state being ∼15 kcal mol⁻¹ higher in energy, according to DFT calculations. The possible involvement of the Fe^V=O reactive species in the reaction is supported by the reasonably low reaction barrier for the oxidation of water by [Fe^V(L1)(O)₂]⁺ of 15.7 kcal mol⁻¹ obtained by the DFT calculations (Fig. 11). For [Fe^V(L1)(O)(OH)]²⁺, the reaction barrier for O–O bond formation with water was calculated to be 21.0 kcal mol⁻¹ (Fig. S34†), which is 5.3 kcal mol⁻¹ higher than that calculated for [Fe^V(L1)(O)₂]⁺. The Fe^V=O species is suggested to be attacked by water molecules, assisted by hydrogen-bond interactions, leading to the formation of an O–O bond, analogous to the O–O bond formation in water oxidation by Mn^V=O,³² but the possibility of the involvement of Fe^IV=O^6e or O=Fe^IV–O–Ce^IV species¹³ in the O–O bond formation could not be excluded, considering, for example, the small difference between the DFT-calculated redox potentials of +1.25 V for [Fe^IV(L1)(O)(OH₂)]²⁺/[Fe^III(L1)(OH)(OH₂)]²⁺ and +1.42 V for [Fe^V(L1)(O)(OH)]²⁺/[Fe^IV(L1)(O)(OH₂)]²⁺ at pH 1. The resulting iron(III)–peroxo intermediate in Scheme 2 is subsequently oxidized by the sacrificial oxidant (CAN, NaIO₄, or Oxone) to give [Fe^III(L1)(OO˙)(OH)]⁺ (DFT-calculated potential for [Fe^III(L1)(OO˙)(OH)]⁺/[Fe^III(L1)(OOH)(OH)]⁺: +0.64 V vs. SCE), the presence of which is supported by ESI-MS analysis (Fig. S20†). Extrusion of oxygen is achieved upon the substitution of [Fe^III(L1)(OO˙)(OH)]⁺ with water; the resulting [Fe^II(L1)(OH)(OH₂)]⁺ is oxidized by the sacrificial oxidant (CAN, NaIO₄, or Oxone) to regenerate the catalyst [Fe^III(L1)(OH)₂]⁺ (DFT-calculated potential for [Fe^III(L1)(OH)₂]⁺/[Fe^II(L1)(OH)(OH₂)]⁺: +0.26 V vs. SCE).

Scheme 2 Proposed pathways for 1-catalysed water oxidation with CAN or NaIO₄.

Conclusions

A mononuclear iron(III) complex bearing a macrocyclic N₄ diazapyridinophane ligand catalyses the oxidation of water to oxygen with NaIO₄ or Oxone, as well as CAN, as the oxidant in acidic and/or neutral aqueous media. Studies using kinetic measurements, high-resolution ESI-MS, UV-vis absorption spectroscopy, ¹⁸O-labelling experiments, cyclic voltammetry, and DFT calculations lend support to the possible involvement of high-valent iron(IV)–oxo species such as [Fe^IV(L1)(O)(OH)]⁺ and/or iron(V)–oxo species such as [Fe^V(L1)(O)₂]⁺, or their protonated forms [Fe^IV(L1)(O)(OH₂)]²⁺ and/or [Fe^V(L1)(O)(OH)]²⁺, in the 1-catalysed water oxidation reactions.

Acknowledgements

This work was supported by the Hong Kong Research Grants Council (HKU 700813), the National Key Basic Research Program of China (No. 2013CB834802), the University Grants Committee of the Hong Kong Special Administrative Region (AoE/P-03/08), and the CAS-Croucher Foundation Funding Scheme for Joint Laboratories. We thank Prof. Hung Kay Lee for assistance with EPR measurements.

Notes and references

1. (a) C. Tommos and G. T. Babcock, Acc. Chem. Res., 1998, 31, 18; (b) M. H. V. Huynh and T. J. Meyer, Chem. Rev., 2007, 107, 5004; (c) R. Eisenberg and H. B. Gray, Inorg. Chem., 2008, 47, 1697.
2. (a) D. G. H. Hetterscheid and J. N. H. Reek, Angew. Chem., Int. Ed., 2012, 51, 9740; (b) D. J. Wasylenko, R. D. Palmer and C. P. Berlinguette, Chem. Commun., 2013, 49, 218; (c) A. Singh and L. Spiccia, Coord. Chem. Rev., 2013, 257, 2607; (d) M. D. Kärkäs, O. Verho, E. V. Johnston and B. Ǻkermark, Chem. Rev., 2014, 114, 11863.
3. (a) T. C. W. Mak, C.-M. Che and K.-Y. Wong, J. Chem. Soc., Chem. Commun., 1985, 986; (b) C.-M. Che, T.-F. Lai and K.-Y. Wong, Inorg. Chem., 1987, 26, 2289; (c) J.-U. Rohde, J.-H. In, M. H. Lim, W. W. Brennessel, M. R. Bukowski, A. Stubna, E. Münck, W. Nam and L. Que Jr., Science, 2003, 299, 1037.
4. (a) J. Hohenberger, K. Ray and K. Meyer, Nat. Commun., 2012, 3, 720; (b) A. R. McDonald and L. Que Jr., Coord. Chem. Rev., 2013, 257, 414; (c) K. P. Bryliakov and E. P. Talsi, Coord. Chem. Rev., 2014, 276, 73; (d) K. Ray, F. F. Pfaff, B. Wang and W. Nam, J. Am. Chem. Soc., 2014, 136, 13942; (e) W. Nam, Y.-M. Lee and S. Fukuzumi, Acc. Chem. Res., 2014, 47, 1146.
5. (a) E. J. Klinker, J. Kaizer, W. W. Brennessel, N. L. Woodrum, C. J. Cramer and L. Que Jr., Angew. Chem., Int. Ed., 2005, 44, 3690; (b) J. Yoon, S. A. Wilson, Y. K. Jang, M. S. Seo, K. Nehru, B. Hedman, K. O. Hodgson, E. Bill, E. I. Solomon and W. Nam, Angew. Chem., Int. Ed., 2009, 48, 1257; (c) J. England, Y. Guo, E. R. Farquhar, V. G. Young Jr., E. Münck and L. Que Jr., J. Am. Chem. Soc., 2010, 132, 8635; (d) D. C. Lacy, R. Gupta, K. L. Stone, J. Greaves, J. W. Ziller, M. P. Hendrich and A. S. Borovik, J. Am. Chem. Soc., 2010, 132, 12188; (e) S. Fukuzumi, Y. Morimoto, H. Kotani, P. Naumov, Y.-M. Lee and W. Nam, Nat. Chem., 2010, 2, 756; (f) H. Kotani, T. Suenobu, Y.-M. Lee, W. Nam and S. Fukuzumi, J. Am. Chem. Soc., 2011, 133, 3249; (g) S. Hong, Y.-M. Lee, K.-B. Cho, K. Sundaravel, J. Cho, M. J. Kirn, W. Shin and W. Nam, J. Am. Chem. Soc., 2011, 133, 11876; (h) S. A. Wilson, J. Chen, S. Hong, Y.-M. Lee, M. Clémancey, R. Garcia-Serres, T. Nomura, T. Ogura, J.-M. Latour, B. Hedman, K. O. Hodgson, W. Nam and E. I. Solomon, J. Am. Chem. Soc., 2012, 134, 11791; (i) Y. Nishida, Y.-M. Lee, W. Nam and S. Fukuzumi, J. Am. Chem. Soc., 2014, 136, 8042.
6. (a) W. C. Ellis, M. D. McDaniel, S. Bernhard and T. J. Collins, J. Am. Chem. Soc., 2010, 132, 10990; (b) J. Lloret Fillol, Z. Codolà, I. Garcia-Bosch, L. Gómez, J. J. Pla and M. Costas, Nat. Chem., 2011, 3, 807; (c) G. Chen, L. Chen, S.-M. Ng, W.-L. Man and T.-C. Lau, Angew. Chem., Int. Ed., 2013, 52, 1789; (d) W. A. Hoffert, M. T. Mock, A. M. Appel and J. Y. Yang, Eur. J. Inorg. Chem., 2013, 3846; (e) D. Hong, S. Mandal, Y. Yamada, Y.-M. Lee, W. Nam, A. Llobet and S. Fukuzumi, Inorg. Chem., 2013, 52, 9522; (f) Z. Codolà, I. Garcia-Bosch, F. Acuna-Parés, I. Prat, J. M. Luis, M. Costas and J. Lloret-Fillol, Chem. – Eur. J., 2013, 19, 8042; (g) B. Zhang, F. Li, F. Yu, H. Cui, X. Zhou, H. Li, Y. Wang and L. Sun, Chem. – Asian J., 2014, 9, 1515.
7. C. Panda, J. Debgupta, D. D. Díaz, K. K. Singh, S. S. Gupta and B. B. Dhar, J. Am. Chem. Soc., 2014, 136, 12273.
8. (a) M. K. Coggins, M.-T. Zhang, A. K. Vannucci, C. J. Dares and T. J. Meyer, J. Am. Chem. Soc., 2014, 136, 5531; (b) E. L. Demeter, S. L. Hilburg, N. R. Washburn, T. J. Collins and J. R. Kitchin, J. Am. Chem. Soc., 2014, 136, 5603; (c) Z.-Q. Wang, Z.-C. Wang, S. Zhan and J.-S. Ye, Appl. Catal., A, 2015, 490, 128.
9. B. M. Klepser and B. M. Bartlett, J. Am. Chem. Soc., 2014, 136, 1694.
10. (a) M. M. Najafpour, A. N. Moghaddam, D. J. Sedigh and M. Holyńska, Catal. Sci. Technol., 2014, 4, 30; (b) A. R. Parent, T. Nakazono, S. Lin, S. Utsunomiya and K. Sakai, Dalton Trans., 2014, 43, 12501.
11. D. Wang and L. Que Jr., Chem. Commun., 2013, 49, 10682.
12. (a) M. Z. Ertem, L. Gagliardi and C. J. Cramer, Chem. Sci., 2012, 3, 1293; (b) E. E. Kasapbasi and M.-H. Whangbo, Inorg. Chem., 2012, 51, 10850; (c) A. Poater, Catal. Commun., 2014, 44, 2; (d) R.-Z. Liao, X.-C. Li and P. E. M. Siegbahn, Eur. J. Inorg. Chem., 2014, 728; (e) F. Acuña-Parés, M. Costas, J. M. Luis and J. Lloret-Fillol, Inorg. Chem., 2014, 53, 5474; (f) F. Acuña-Parés, Z. Codolà, M. Costas, J. M. Luis and J. Lloret-Fillol, Chem. – Eur. J., 2014, 20, 5696.
13. Z. Codolà, L. Gómez, S. T. Kleespies, L. Que Jr., M. Costas and J. Lloret-Fillol, Nat. Commun., 2015, 6, 5865.
14. (a) F. T. de Oliveira, A. Chanda, D. Banerjee, X. Shan, S. Mondal, L. Que Jr., E. L. Bominaar, E. Münck and T. J. Collins, Science, 2007, 315, 835; (b) S. H. Lee, J. H. Han, H. Kwak, S. J. Lee, E. Y. Lee, H. J. Kim, J. H. Lee, C. Bae, S. N. Lee, Y. Kim and C. Kim, Chem. – Eur. J., 2007, 13, 9393; (c) O. Y. Lyakin, K. P. Bryliakov, G. J. P. Britovsek and E. P. Talsi, J. Am. Chem. Soc., 2009, 131, 10798; (d) I. Prat, J. S. Mathieson, M. Güell, X. Ribas, J. M. Luis, L. Cronin and M. Costas, Nat. Chem., 2011, 3, 788; (e) O. Y. Lyakin, K. P. Bryliakov and E. P. Talsi, Inorg. Chem., 2011, 50, 5526; (f) O. Y. Lyakin, I. Prat, K. P. Bryliakov, M. Costas and E. P. Talsi, Catal. Commun., 2012, 29, 105; (g) O. Y. Lyakin, R. V. Ottenbacher, K. P. Bryliakov and E. P. Talsi, ACS Catal., 2012, 2, 1196; (h) Y. Hitomi, K. Arakawa, T. Funabiki and M. Kodera, Angew. Chem., Int. Ed., 2012, 51, 3448; (i) K. M. Van Heuvelen, A. T. Fiedler, X. Shan, R. F. De Hont, K. K. Meier, E. L. Bominaar, E. Münck and L. Que Jr., Proc. Natl. Acad. Sci. U. S. A., 2012, 109, 11933; (j) M. Ghosh, K. K. Singh, C. Panda, A. Weitz, M. P. Hendrich, T. J. Collins, B. B. Dhar and S. Sen Gupta, J. Am. Chem. Soc., 2014, 136, 9524.
15. (a) F. Bottino, M. D. Grazia, P. Finocchiaro, F. R. Fronczek, A. Mamo and S. Pappalardo, J. Org. Chem., 1988, 53, 3521; (b) W. O. Koch, V. Schünemann, M. Gerdan, A. X. Trautwein and H.-J. Krüger, Chem. – Eur. J., 1998, 4, 686.
16. (a) T. W.-S. Chow, E. L.-M. Wong, Z. Guo, Y. Liu, J.-S. Huang and C.-M. Che, J. Am. Chem. Soc., 2010, 132, 13229; (b) F. Tang, Y. Zhang, N. P. Rath and L. M. Mirica, Organometallics, 2012, 31, 6690; (c) H. Sugimoto, K. Ashikari and S. Itoh, Chem. – Asian J., 2013, 8, 2154.
17. W.-T. Lee, S. B. Muñoz III, D. A. Dickie and J. M. Smith, Angew. Chem., Int. Ed., 2014, 53, 9856.
18. A. R. Parent, R. H. Crabtree and G. W. Brudvig, Chem. Soc. Rev., 2013, 42, 2247.
19. J. Limburg, J. S. Vrettos, H. Chen, J. C. de Paula, R. H. Crabtree and G. W. Brudvig, J. Am. Chem. Soc., 2001, 123, 423.
20. (a) C.-K. Li, C.-M. Che, W.-F. Tong, W.-T. Tang, K.-Y. Wong and T.-F. Lai, J. Chem. Soc., Dalton Trans., 1992, 2109; (b) S. K. W. Yau, C.-M. Che and T.-C. Lau, J. Chem. Soc., Dalton Trans., 2002, 2697.
21. (a) F. Bozoglian, S. Romain, M. Z. Ertem, T. K. Todorova, C. Sens, J. Mola, M. Rodríguez, I. Romero, J. Benet-Buchholz, X. Fontrodona, C. J. Cramer, L. Gagliardi and A. Llobet, J. Am. Chem. Soc., 2009, 131, 15176; (b) X. Guan, S. L.-F. Chan and C.-M. Che, Chem. – Asian J., 2013, 8, 2046.
22. (a) J. N. Harvey, M. Aschi, H. Schwarz and W. Koch, Theor. Chem. Acc., 1998, 99, 95; (b) J. N. Harvey and M. Aschi, Phys. Chem. Chem. Phys., 1999, 1, 5555.
23. B. Limburg, E. Bouwman and S. Bonnet, Coord. Chem. Rev., 2012, 256, 1451.
24. (a) M. J. Katz, S. C. Riha, N. C. Jeong, A. B. F. Martinson, O. K. Farha and J. T. Hupp, Coord. Chem. Rev., 2012, 256, 2521; (b) F. E. Osterloh, Chem. Soc. Rev., 2013, 42, 2294.
25. (a) J. J. Concepcion, J. W. Jurss, J. L. Templeton and T. J. Meyer, J. Am. Chem. Soc., 2008, 130, 16462; (b) J. J. Concepcion, J. W. Jurss, M. R. Norris, Z. Chen, J. L. Templeton and T. J. Meyer, Inorg. Chem., 2010, 49, 1277; (c) D. J. Wasylenko, C. Ganesamoorthy, M. A. Henderson, B. D. Koivisto, H. D. Osthoff and C. P. Berlinguette, J. Am. Chem. Soc., 2010, 132, 16094; (d) Y. M. Badiei, D. E. Polyansky, J. T. Muckerman, D. J. Szalda, R. Haberdar, R. Zong, R. P. Thummel and E. Fujita, Inorg. Chem., 2013, 52, 8845.
26. D.-L. Popescu, M. Vrabel, A. Brausam, P. Madsen, G. Lente, I. Fabian, A. D. Ryabov, R. van Eldik and T. J. Collins, Inorg. Chem., 2010, 49, 11439.
27. (a) G. E. Cabaniss, A. A. Diamantis, W. R. Murphy Jr., R. W. Linton and T. J. Meyer, J. Am. Chem. Soc., 1985, 107, 1845; (b) C.-M. Che, K.-Y. Wong and F.-C. Anson, J. Electroanal. Chem., 1987, 226, 211.
28. Y. Wing-Ping, PhD Thesis, The University of Hong Kong, 2004.
29. D. Wang, M. Zhang, P. Buhlmann and L. Que Jr., J. Am. Chem. Soc., 2010, 132, 7638.
30. L. Que Jr., Acc. Chem. Res., 2007, 40, 493.
31. Y. Ohgo, Y. Chiba, D. Hashizume, H. Uekusa, T. Ozeki and M. Nakamura, Chem. Commun., 2006, 1935.
32. Y. Gao, T. Åkermark, J. Liu, L. Sun and B. Åkermark, J. Am. Chem. Soc., 2009, 131, 8726.

---

Footnote

† Electronic supplementary information (ESI) available: Experimental section, Table S1, Fig. S1–S34, Scheme S1 and computational details. See DOI: 10.1039/c5sc01680k

---