O−O Bond Formation and Liberation of Dioxygen Mediated by N5‐Coordinate Non‐Heme Iron(IV) Complexes

Abstract Formation of the O−O bond is considered the critical step in oxidative water cleavage to produce dioxygen. High‐valent metal complexes with terminal oxo (oxido) ligands are commonly regarded as instrumental for oxygen evolution, but direct experimental evidence is lacking. Herein, we describe the formation of the O−O bond in solution, from non‐heme, N5‐coordinate oxoiron(IV) species. Oxygen evolution from oxoiron(IV) is instantaneous once meta‐chloroperbenzoic acid is administered in excess. Oxygen‐isotope labeling reveals two sources of dioxygen, pointing to mechanistic branching between HAT (hydrogen atom transfer)‐initiated free‐radical pathways of the peroxides, which are typical of catalase‐like reactivity, and iron‐borne O−O coupling, which is unprecedented for non‐heme/peroxide systems. Interpretation in terms of [FeIV(O)] and [FeV(O)] being the resting and active principles of the O−O coupling, respectively, concurs with fundamental mechanistic ideas of (electro‐) chemical O−O coupling in water oxidation catalysis (WOC), indicating that central mechanistic motifs of WOC can be mimicked in a catalase/peroxidase setting.

critical step in oxidative water cleavage to produce dioxygen. High-valent metal complexes with terminal oxo( oxido) ligands are commonly regarded as instrumental for oxygen evolution, but direct experimental evidence is lacking. Herein, we describe the formation of the OÀOb ond in solution, from non-heme,N 5 -coordinate oxoiron(IV) species.O xygen evolution from oxoiron(IV) is instantaneous once meta-chloroperbenzoic acid is administered in excess.Oxygen-isotope labeling reveals two sources of dioxygen, pointing to mechanistic branching between HAT( hydrogen atom transfer)-initiated free-radical pathways of the peroxides,w hich are typical of catalase-like reactivity,and iron-borne OÀOcoupling,which is unprecedented for non-heme/peroxide systems.I nterpretation in terms of [Fe IV (O)] and [Fe V (O)] being the resting and active principles of the O À Oc oupling, respectively,c oncurs with fundamental mechanistic ideas of (electro-) chemical O À O coupling in water oxidation catalysis (WOC), indicating that central mechanistic motifs of WOCc an be mimicked in ac atalase/peroxidase setting.
Efficient water oxidation catalysis (WOC) is one of the major challenges in the context of future-oriented energy management schemes.Catalytic water oxidation is ademanding task, owing to its energetic uphill character and the requirement for ac oupled multielectron/multiproton shuttle (4 H + /4 e À )t op revent the formation of hazardous reactiveoxygen species (ROS). Tw ot ypes of reagent hold particular promise here:m etal-oxide-based heterogeneous (electro)catalysts [1][2][3] and low-molecular-weight transition-metal complexes (typically of Ru;C o, Fe ;I r), which operate in homogeneous solution. [4][5][6][7][8][9] As for the latter, an umber of functional models are now known for the enzyme-complex-appended {Mn 4 Ca} cluster, which is the active site of the oxygen-evolving complex (OEC) in biological photosystem II. Models based on ruthenium are the most numerous;t hey show robust and efficient oxidative water turnover, have large turnover numbers TON,and use positive electrode potentials or highly oxidizing additives (e.g., cerium(IV) ammonium nitrate). [10,11] Less numerous to date are models based on 3d metals (Co, Fe). [12] This is bound to change,however;iron in particular is readily available (Febeing the second most abundant metal in the earthscrust), and there are few,ifany,concerns in terms of element toxicity.
Aside from these advantages,current interest in dioxygenrelated iron coordination chemistry has been further fuelled by the following:While metal-mediated oxygen-oxygen bond formation is generally agreed to be the critical step in both biological photosynthesis and model complex-based WOC, examples for iron-mediated O À Ob ond formation are still rare. [10,[13][14][15][16] High-valent oxo-iron complexes are invoked as critical intermediates en route to O 2 liberation-with oxoiron(IV) as the "resting state" and oxoiron(V) as the "active state" of water oxidation, respectively. [17][18][19] As of yet, however, few details are known regarding the chemical nature of the O À Ob ond coupling step,a nd the molecular species involved.
We had reported on the coordination chemistry and spin state preferences of the pentadentate ligand L [21] (its improved synthesis,which gives faster access to L in higher yield, is detailed in the Supporting Information, along with additional XRD data of [Fe II (L)(OTf)](OTf)·(0.5 Et 2 O);F igure S1). From the distorted octahedral iron(II) precursor [Fe II (L)(MeCN)] 2+ (triflate salt), the oxoiron(IV) complex [Fe IV (L)(O)] 2+ is accessible in moderate yields (ca. 30-40 %) by reaction with an equimolar amount of mCPBAi nM eCN solution, [21] but forms close to quantitatively with mCPBA present in excess (we find an optimum for ar atio [Fe II (L)-(MeCN)] 2+ /mCPBA = 1:5; see Figure S2;o ptimum yield > 85 %). Similar observations have been reported by Que et al. in at opologically related system. [22] [Fe IV (L)(O)] 2+ is identified through its prominent peak in the ESI mass spectrum, which responds in the expected manner to 16 O/ 18 Oisotope exchange,upon treatment of the reaction mixture with H 2 18 O( Figure S3a). TheV is/NIR spectroscopic properties of [Fe II (L)(O)] 2+ (l max = 730 nm; e 730nm = 260 m À1 cm À1 ; Figure S3c) in dilute solution are in the range typical of oxoiron(IV) complexes with Fe in at etragonal coordination environment. [23] [Fe II (Bn-TPEN)(O)] 2+ is synthesized in MeCN solution in high yield from [Fe II (Bn-TPEN)(OTf)]-(OTf) according to published procedures. [24,25] Similar to other non-heme oxoiron(IV) species, [26] [Fe IV (L)(O)] 2+ exhibits moderate reactivity towards hydrogen-atom donors (see Figure S4), as well as the oxygen-atom acceptor PPh 3 (see Figure S5 with all components,c arried out in order to exclude potential apparatus leakage,a sw ell as the direct formation of O 2 from mCPBAi nt he absence of the iron complex, proved all negative.Using the fiber-optic sensor, [27] which is operated discontinuously,significant O 2 evolution is traceable after the addition of mCPBA. Reaction of [Fe II (L)-  2+ in MeCN,ormCPBAinMeCN,show no such behavior, but detect even traces of dioxygen if these are purposely admitted at alater stage.Reasonably assuming the solution phase to be near-saturated with dioxygen ([O 2 ] max % 11 mm [28] ), oxygen formation amounts to ca. 160 mmol;this renders its formation super-stoichiometric with respect to the iron content (n(O 2 )/n(Fe) % 1.5:1).
Continuous monitoring of oxygen evolution in solution was performed with aClark-type oxygen electrode system [7,29] (water/MeCN 4:1; [[Fe II (L)(MeCN)] 2+ ] = 2mm ;a na queous solvent is required for electrode function). After addition of mCPBA( 10 equiv) to the solution of [Fe II (L)(MeCN)] 2+ ,a n instantaneous but gradually diminishing increase of the dioxygen concentration is detected over 30 min (Figure 1a, blue curve). It is emphasized that astable plateau signal does not indicate ceased O 2 evolution, but as teady state of electrochemical consumption and sustained iron-dependent production. Ab lank test with only mCPBAi nt he solution showed avery slight, if any,increase in the oxygen signal. The initial rate of O 2 evolution via [Fe IV (L)(O)] 2+ is estimated to be 0.2 mmol min À1 ,t ranslating into an (apparent [30] )i nitial turnover frequency TOF 0 % 2.8 h À1 in the presence of 10 equiv mCPBA. Both the initial slope and the step height grow in proportion with the amount of mCPBAa dded. Importantly, aged solutions can be re-activated by iterative administration of mCPBAa liquots ( Figure S7). Recovery of the initial Scheme 1. Top: Structures of the pentadentate N 5 podands Bn-TPEN and L and the iron(II) complex, which has dissociable MeCN at the sixth coordination site (X). Bottom:P henomenology of oxoiron(IV) formation and decay as described here.  2+ is derived from PhIO and subsequently reacted with t-BuOOH. Thus,t he formation of both O 2 and CO 2 is triggered by mCPBA. Carbon dioxide formation implies formation of significant amounts of elusive RCO 2 C (with R = 3-chlorophenyl);s uch aromatic carboxyl radicals are known to undergo rapid and selective decarboxylation, RCO 2 C ! RC + CO 2 . [31,32] They may derive from parent RCO 3 Hv ia aHAT-initiated bimolecular sequence or formal loss of ahydroxyl radical OH C (Scheme 2a,b) or through OÀO bond homolysis of iron(III) acylperoxido species (i.e., [Fe III - . [22] In neat MeCN the measured ratio of dioxygen isotopomer ion currents i 32 /i 34 % 200:1 matches the isotope distribution expected from natural abundance (32-O 2 in Figure 2, left). Insertion of ap re-equilibration step in the presence of 18 Olabeled water (purity:9 7% 18 O [33] )i nt he above reaction sequence induces massive shifts in the product ratio.I sotopomer ratios of i 32 /i 34 % 3:2( from three iterations;F igure 2, middle) and i 32 /i 34 % 1:1.1 (from two iterations; Figure 2 water oxidation catalysis studies elsewhere: [34]  Thedetection of significant amounts of 34-O 2 necessarily implies efficient coupling between 18 O-labeled iron-borne oxygen and a 16 Oo xygen atom from another source.T his source must be unlabeled mCPBA, [35] as no O 2 formation is observed in the absence of this reagent. Significant background levels of normal 32-O 2 could, in principle,b e attributed to slow or incomplete isotope exchange in the species at hand;t he residual 16 OH 2 content in "dry" MeCN batches used throughout actually reduces the labeling level of 18 Ot oc a. 80 %. [33] In keeping with this,v ariation of the equilibration time (15 min < t eq < 100 min) has no significant effect on the observed product ratios.I ndeed, our observed time range covers and exceeds the equilibration times typically necessary for complete 16 O/ 18 Oexchange in oxoiron-(IV) complexes. [36] Therefore,w ea scribe the major part of trivial 32-O 2 formed in solutions of [Fe IV (L)(O)] 2+ and mCPBA( and tert-butyl hydroperoxide, t-BuOOH) to the operation of free-radical pathways (Scheme 2d). It is well known that organic peroxyl radicals are efficient sources of dioxygen via spontaneous decay of labile polyoxide intermediates (e.g., 2t-BuOOC ! (t-BuOO) 2 ! 2 t-BuOC + O 2 ). [37,38] This pathway,w hich has been recently studied in some detail for iron complexes of ar elated pentadentate ligand by Browne,M cKenzie,a nd co-workers, [39] must be taken to be relevant in our system, as the oxoiron(IV) complex [Fe IV (L)-(O)] 2+ is competent in HATreactions ( Figure S4 and Ref. [21]). In fact, reaction of t-BuOOH and presynthesized [Fe IV (L)(O)] 2+ exclusively yields the trivial isotopomer 32-O 2 ,i rrespective of the isotope speciation of added water (Figures S16-18;d ue to the water content of commercial t-BuOOH (30 wt %), the 18 OH 2 level amounts to ca. 50 %inMeCN solution) corroborating acatalatic nonscrambling mechanism. [40,41] In agreement with the notion of the OÀHb ond in t-BuOOH being much weaker than that in mCPBA (literature data based on t-BuOOH and peracetic acid, MeCO 3 H, suggest ad ifference in bond dissociation energies DBDE(O-H) % 36 kJ mol À1 ), [42] HATfrom t-BuOOH fully outcompetes iron-complex-borne reactions; [43] as amatter of fact, the latter become competitive when mCPBAi sused.
Owing to its highly electrophilic nature,the oxo ligand in [Fe IV O] is generally assumed to be susceptible to nucleophilic attack. [44] Tw op lausible pathways of the iron-borne O À O coupling are shown in Scheme 2. Theo xoiron(V) path (a) alludes to ideas as expressed by Costas and others, [17][18][19]45,46] whereas the concerted O-atom transfer (c) adopts the mechanistic paradigm of mCPBA-driven olefin epoxidation. [47,48] Thelatter concerted pathway invokes essentially simultaneous peroxo OÀOb ond breaking and O 2 formation within ac yclic intermediate (Scheme 2c). Although it shares some similarity with the ideas put forward by Hager et al., [13] in order to rationalize the formation of dioxygen in the reaction of ferric heme-dependent chloroperoxidase with mCPBA, we favor the oxoiron(V) pathway for the following reasons:T he À . [49] Such species have been invoked previously as the active agent in iron-catalyzed electrochemical water oxidation. [50] In both types of studies,the formation of oxoiron(V) required highly oxidizing conditions,t hat is,e ither high concentrations of the strong chemical oxidant Ce IV (E 0 (Ce III/IV ) = 1.70 Vvs. NHE [51] )and otherwise harsh conditions (i.e., pH % 1), or very positive electrode potentials (E p,a = 1.58 Vv s. NHE [50] ). Oxoiron(V) being attacked by OH À formed in an outer-sphere electron transfer (or via rapid proton transfer from labeled bulk water, 16 OH À + 18 OH 2 ! 16 OH 2 + 18 OH À ,S cheme 2b)w ould indeed rationalize the This nucleophilic OÀOc oupling is the microscopic reversal of heterolytic OÀOc leavage in iron(III) hydroperoxido species;ithas been found in DFT studies on the Nmethyl analogue of [Fe III (Bn-TPEN)(OOH)] 2+ to have ahuge driving force. [52] Nevertheless,itappears unlikely that the mild oxidant mCPBAu sed in our study can efficiently drive the Fe IV ! Fe V oxidation step in an outer-sphere electron transfer reaction (but see Ref. [53]). However,concerted inner-sphere transfer of OH À and of an electron in opposite directions avoids the high energy penalties usually attending chargebuilding reactions.I ta ppears plausible to ascribe the formation of monolabeled 34-O 2 to this net inner-sphere transfer of ah ydroxyl radical; [54] it is conceptually complementary to the coupled transfer of ap roton and an electron, PCET, [55] which in the meantime has proven its omnipresence in bioinorganic research.
Irrespective It is noted, however, that the regeneration of [Fe II (L)-(MeCN)] 2+ may also be traced to the iron(III) hydroperoxide complex implied in Scheme 2a,b via an additional oneelectron oxidation or HATreaction. [56] Whereas the nature of the oxygen-liberating iron species is unclear at present, regeneration of [Fe II (L)(MeCN)] 2+ is beyond doubt. Notably,the new near-UV band peaking at l = 398 nm, which evolves after complete decay of the oxoiron-(IV) intermediate,coincides with the spectral response of the iron(II) precursor [Fe II (L)(MeCN)] 2+ (Figure 3a). Significant absorption at l < 320 nm indicates the presence of side products,p resumably iron(III) species. [57] Am ore conclusive spectroscopic argument comes from time-dependent 1 H NMR spectroscopy (Figure 3b). After addition of mCPBA (10 equiv), the widely spread resonances ([Fe II (L)(MeCN)] 2+ in MeCN is aspin crossover system with T 1/2 % 320 K; [hs]/[ls] % 1:4a tR T [21] )o f[ Fe II (L)(MeCN)] 2+ in d 3 -MeCN are instantly quenched (NMR spectroscopic studies of oxoiron-(IV) species are generally rare) [58] but are recovered in aslow process,returning to ca. 40 %ofthe initial integrated intensity after 12 h( higher yields will likely be obtained on an extended timescale,s ee Figure S19). To the best of our knowledge,the Fe II !Fe IV !Fe II reversion sequence has only asingle precedent in related literature:The iron(II) precursor  [59] of particular note here is the fact that the cited work reports iron-dependent dioxygen formation prior to precursor recovery,p resumably via nonscrambling disproportionation. As no labeling studies have been reported, the mechanistic relatedness of the two systems cannot, however, be judged with certainty.
In the present work, we have reported an unprecedented aspect of non-heme oxoiron(IV) reactivity:F irstly,our work, which uses non-heme iron(II) complexes of pentadentate ligands,a dds two new examples to the short list of exceptions [50,60,61] from the "two open cis-sites" rule,w hich describes ap utative structural requirement for an active water oxidation catalyst or, more specifically,f or complexes which support metal-borne OÀOb ond formation. Oxoiron(IV) complexes of the two N 5 ligands studied herein do in fact spontaneously produce stoichiometric amounts of dioxygen when the O-atom-donor mCPBAispresent in excess,but are metastable in its absence.The dependence of O 2 formation on the presence of an excess of mCPBAr enders oxoiron(IV) ar esting state of dioxygen formation. Accordingly,i sotope labeling studies reveal am echanistic branching between nonproductive HAT-like reactivity and, presumably,O Hgroup transfer, with the implicit passing through an oxoiron-(V) intermediate.S econdly,t he heterocoupling between two different types of activated oxygen species,o xoiron(IV) and ap eracid, is established in the present study.W hile the speciation implied herein probably differs from WOC, the option to study OÀOc oupling in isolation is expected to be av aluable tool for the scrutiny of the OÀOc oupling step in WOC, even more so since peroxides have been previously shown to be active principles in WOC. [61] There are no peculiarities in the structure of the N 5 ligands L and Bn-TPEN with respect to donor speciation and topology,and we are confident that observations similar to ours will be made in the future with other non-heme systems involving pentacoor- dinating ligands.T he decisive requirement is O 2 -indifference of the iron(II) precursors (as O 2 is liberated with concomitant re-formation of the ferrous complex). This is ap roperty shared by the complexes studied herein, [Fe II (L)(MeCN)] 2+ and [Fe II (Bn-TPEN)(MeCN)] 2+ . [20,62] Overall, the OÀObond formation pattern observed in the present work is au nique reversal of the paradigmatic iron-mediated OÀOb ond cleavage activity, [63][64][65] which usually renders non-heme oxoiron complexes active in H-atom abstraction [26,66,67] and oxygen-atom transfer chemistry. [68]