Observation of Fe ( V ) 5 O using variable-temperature mass spectrometry and its enzyme-like C – H and C 5 C oxidation reactions

Oxo-transfer chemistry mediated by iron underpins many biological processes and today is emerging as synthetically very important for the catalytic oxidation of C–H and C5C moieties that are hard to activate conventionally. Despite the vast amount of research in this area, experimental characterization of the reactive species under catalytic conditions is very limited, although a Fe(V)5O moiety was postulated. Here we show, using variable-temperature mass spectrometry, the generation of a Fe(V)5O species within a synthetic non-haem complex at 240 8C and its reaction with an olefin. Also, with isotopic labelling we were able both to follow oxygen-atom transfer from H2O2/H2O through Fe(V)5O to the products and to probe the reactivity as a function of temperature. This study pioneers the implementation of variable-temperature mass spectrometry to investigate reactive intermediates.

Oxo-transfer chemistry mediated by iron underpins many biological processes and today is emerging as synthetically very important for the catalytic oxidation of C-H and C5C moieties that are hard to activate conventionally. Despite the vast amount of research in this area, experimental characterization of the reactive species under catalytic conditions is very limited, although a Fe(V)5O moiety was postulated. Here we show, using variable-temperature mass spectrometry, the generation of a Fe(V)5O species within a synthetic non-haem complex at 240 8 8 8 8 8C and its reaction with an olefin. Also, with isotopic labelling we were able both to follow oxygen-atom transfer from H 2 O 2 /H 2 O through Fe(V)5O to the products and to probe the reactivity as a function of temperature. This study pioneers the implementation of variable-temperature mass spectrometry to investigate reactive intermediates. F or small-molecule activation processes, iron is the element of choice, selected by nature to perform a number of chemically challenging oxidative processes with high precision and reaction rates. Iron-based enzymes, such as cytochrome P450 (ref. 1) and Rieske dioxygenases 2 , use O 2 to catalyse highly selective C-H and C¼C oxidation reactions, key steps in the metabolic synthesis of metabolites, xenobiotic degradation and other crucial functions. At the heart of these transformations, it is proposed that the catalytic iron centre forms a highly oxidizing oxo-iron species (Fig. 1) 3 . In P450 the active species, formally Fe(V)¼O, is best described as an oxo-Fe(IV)-porphyrin radical cation 1 , but in the case of the Rieske dioxygenases family of enzymes, which lack the redox non-innocent porphyrin ligand, one postulation is that an oxo-iron(V) species is the reactive species 4,5 . Observation of these highly reactive intermediates remains a formidable challenge that was achieved recently for P450 (refs 6,7), but so far this has not been possible for a nonhaem enzyme 8,9 .
Functional models of non-haem iron-dependent oxygenases are currently the focus of intense research efforts because of the major challenges in modern synthetic chemistry presented by selective and environmentally benign C-H hydroxylation and olefin cisdihydroxylation reactions [10][11][12][13] . In addition, such studies aim to provide an intimate understanding of the processes that underpin the enzymatic reactions. Mechanistic studies of naphthalene 1,2dioxygenase 9,14 , a member of the Rieske dioxygenases family, and of model compounds [15][16][17][18][19][20] point strongly towards the involvement of a highly electrophilic oxo-iron(V) species, but direct evidence under catalytic conditions is lacking. Recently, the first example of a non-haem oxo-iron(V) species was characterized spectroscopically. This compound contained a tetraanionic ligand, which is likely to quench its electrophilicity, and shows neither C-H hydroxylation nor C¼C cis-dihydroxylation reactivity 21 .
In this respect, variable-temperature mass spectrometry (VT-MS) was envisioned as a very powerful technique with which to study highly reactive intermediate species at very low reagent concentrations, without the need for the large product accumulation required for most spectroscopic techniques. VT-MS could also be used to follow the emergence of the reactive species under much colder conditions than normally used in electrospray mass spectrometric experiments, and thus minimize bimolecular decomposition pathways commonly associated with highly reactive species. Therefore, we envisaged that the observation of reactive species may not only be possible by using low-temperature mass spectrometry [22][23][24][25][26] , but also that varying the temperature of the cryospray source during the experiment under catalytic conditions may give further insight and evidence of the identity of the reactive intermediate.
To the best of our knowledge, cryospray mass spectrometry has not been used before to follow reactive intermediates, and herein we present a new technique that uses cryospray technology to allow VT-MS. This technique allowed the temperature-controlled trapping and characterization of a Fe(V)¼O species that acted as a functional model of Rieske dioxygenases. Isotopic labelling studies were used to provide accurate chemical descriptions of these species and demonstrated atom transfer, via the Fe complex, from the reagent to the product, that is the cis-dihydroxylation of an olefin. The data presented in this work allowed us to elucidate the nature of the iron-based species responsible for performing alkane hydroxylation and olefin cis-dihydroxylation in a synthetic biomimetic system.
The isotopic labelling experiments, in combination with density functional theory (DFT) computational methods, allowed us to propose confidently that a Fe(V)(O)(OH) species is responsible for the C-H and C¼C oxidation events 18,19 . Previously, such a mechanistic scenario was proposed for related non-haem iron catalysts [15][16][17]20 , and a metastable S ¼ 1/2 species assigned to Fe(V) was observed recently in catalytic epoxidation low-temperature reactions of nonhaem iron complexes with peracids 27 . However, no direct spectroscopic evidence of the putative high-valent species, with the benefit of isotopic labelling, had been obtained so far. The epoxidation reaction mediated by 1 was employed previously to gain indirect productanalysis evidence for the implication of an oxo-iron(V) species 19 , but this reaction does not produce kinetically stable epoxide-metal complex intermediates. In addition, the reaction appears to be quite sensitive to the presence of O 2 . The inability to isolate completely the cryospray instrument meant an inert atmosphere could not be achieved to probe the reaction mechanism without the presence of O 2 . As a result of these limitations, the epoxidation was not studied in the present research.
Monitoring of the reaction of 1 with H 2 O 2 by ultraviolet-visible spectroscopy in a range of temperatures from room temperature to 240 8C did not lead to the accumulation or observation of any intermediate species competent for alkane or alkene substrate oxidation. Therefore, the reaction was explored between room temperature and 240 8C using cryospray VT-MS with the aim to observe the elusive reaction intermediates that could be present at low and presumably steady-state concentrations. The VT-MS analysis of the reaction of 1 with H 2 O 2 (100 equiv.) between 20 8C and 240 8C showed the growth of a prominent peak at m/z ¼ 470.1 that was assigned to {[Fe(III)(OH)( Me,H Pytacn)](OTf)} þ (2) and a second, less-intense peak at m/z ¼ 486 on the basis of its m/z and isotopic distribution ratio. This second peak was not observed when reactions were performed at room temperature, and it disappeared rapidly as the temperature was raised from 240 8C to room temperature. This directly implied that 3 was a metastable reaction intermediate (Fig. 3).
To distinguish between the two possible formulations for 3, isotopic labelling experiments were conducted. As a result of the consistency of our experiments with catalytic reactions, and also the experimental limitations of the isotopically labelled reagents (H 2 18 O 2 is a 2% weight/weight solution in water), we studied (i) the reaction of 1 with H 2 16  O confirmed our initial formulation 18,19 , and showed a peak centred at m/z ¼ 488 Fig. 4b(iii)). Finally, 3O was generated by using H 2 18 O 2 (10 equiv.) in the presence of H 2 18 O (1,000 equiv.) (Fig. 4b(iv)). In this case, the peak at m/z ¼ 488.1 ¼ M þ 2 continued to be the major species, but a peak at m/z ¼ 490.1 ¼ M þ 4 appeared as the second most intense component of the spectrum. The decreased intensity and stability of the ion peaks associated with 3O in this spectrum ( Fig. 4b(iv)) presumably resulted from contamination by other species with m/z values that ranged from 486 to 492, with a higher percentage of species with m/z of 488 present compared to those at 490 m/z, along with contamination     (Fig. 4a) 29,30 . Isotopic labelling experiments showed that the optimum simulation of the spectra shown in Fig. 4b Although the mass spectrometry data do not provide a direct indication of the oxidation state of the iron site, we concluded that Fe(V) was the most plausible oxidation state because of the redox innocence of all the ligands, which is also consistent with DFT analysis of this species 18,19 .
With good evidence for the identity of 3O, we sought to demonstrate its reactivity with respect to an intermolecular oxidation reaction, as this would provide further proof that 3O is a reactive species, as suggested by our mass spectrometric assignment. To do this, we chose a reaction with an olefin because we envisioned that if 3O performs an olefin cis-dihydroxylation reaction it will form a kinetically stable iron-(hydrogen)glycolate species (Fig. 5a). If so, the use of 18 O labels could also be a powerful tool to demonstrate that, indeed, the transformation was mediated by the reactive Fe(V)¼O species assigned to the identity of 3O. To this end, 3O was generated and reacted with cyclooctene (100 equiv.). Mass spectrometry analysis of the reaction indicated that the cluster peak assigned to 3O disappeared and new peaks at m/z ¼ 446.2 (not shown) and m/z ¼ 596.2 (Fig. 5b(i) However, it is well known that epoxides are very poor ligands with putative epoxide-bound species that dissociate very fast, and thus are unstable kinetically. Ion peaks that correspond to 4 and 5 remained stable over time and did not show oxygen exchange with water molecules from the reaction mixture. In addition, hydroxide and oxide ligands in thermally stable Fe(III) complexes should engage in water-exchange reactions. Hence, the stability of the ion peaks associated with 4 and 5 against water exchange further discards epoxide-bound formulations. Therefore, we can conclude that epoxide-bound species were not detected in the spectra.
We have described previously, on the basis of DFT computations, that water-assisted transformation of 3P into 3O is thermoneutral and has a small (Gibbs energy of activation (DG ‡ ) ¼ 20 kcal mol 21 ) activation barrier 18,19 , in good agreement with literature values for related non-haem iron complexes 29,30 21 (ref. 31). To substantiate the proposal that the Fe(V)(O)(OH) species is, indeed, the executor of the cis-dihydroxylation event, the reaction of 3O with trans-2-butene as a model substrate was also computed using DFT methods. A summary of the DFT results is given in Fig. 6. The dihydroxylation is strongly exergonic, with 5 being 60.96 kcal mol 21 more stable than 3O þ trans-2-butene. In addition, the reaction proceeds with a very small energy barrier. The ground state of 3O is S ¼ 3/2, and it is well separated in energy (11.50 kcal mol 21 ) with respect to the first excited state (S ¼ 1/2). The attack of the hydroxoligand over the olefin leads to the formation of the first C1-O1 bond, to form the intermediate IN, with a small barrier of DG ‡ ¼ 3.32 kcal mol 21 . IN then evolves via attack of the oxoligand over C2, with no energy barrier, which leads to the direct formation of the glycolate species 5. The concerted, yet unsymmetrical, nature of the cis-dihydroxylation event derived from our calculations bears a strong resemblance to the analysis described recently by Che and co-workers for the cis-dihydroxylation reaction mediated by a non-haem iron complex on reaction with Oxone 32 . However, on the basis of product analyses and DFT calculations, recent proposals suggest that Fe(IV)(OH) 2 species could be responsible for the cis-dihydroxylation reactions in selected iron catalysts [33][34][35] . These catalysts exhibit different reactivity patterns with respect to the [Fe( Me,H Pytacn)] system described herein. In the former cases, isotopic analysis shows that cis-dihydroxylation preferentially takes place via insertion of two oxygen atoms from a single H 2 O 2 molecule.
Further evidence for a different mechanistic scenario was provided by the observation that [Fe(IV)(O)(OH 2 )( Me,H Pytacn)] 2þ does not mediate the cis-dihydroxylation of olefins. This complex was prepared recently and spectroscopically characterized by us 36 , and computational analyses indicated that a [Fe(IV)(OH) 2 ( Me,H Pytacn)] 2þ intermediate is involved in its waterexchange reactions.
In conclusion, the mechanisms that underlie cis-dihydroxylation reactions mediated by [Fe(OTf ) 2 ( Me,H Pytacn)] are fundamentally distinct from those that operate through a Fe(II)/Fe(IV) cycle.

Conclusions
Previous computational and product-analysis experiments in catalytic C-H hydroxylation and C¼C cis-dihydroxylation reactions mediated by non-haem iron catalysts predicted Fe(V)(O)(OH) species as the final oxidant that executes very fast stereospecific C-H and C¼C oxidation reactions [15][16][17][18][19]29,30 . With this work we present the development and application of VT-MS to the investigation of the reaction between the non-haem iron catalyst 1 with H 2 O 2 at temperatures between 20 and 240 8C, which led to the identification of a metastable intermediate that could be formulated as {[Fe(V)(O)(OH)( Me,H Pytacn)](OTf )} þ (3O) on the basis of isotopic labelling experiments. Such experiments provide evidence that 3O contains an oxygen atom from H 2 O 2 and a second oxygen atom from water, which in turn constitutes strong evidence for a water-assisted path towards its generation, but direct kinetic proof is not yet available and will be the subject of further studies. Despite this limitation, isotopic labelling experiments demonstrated that 3O precedes the cis-dihydroxylation reaction of olefins. This study does not provide a full account of the reaction mechanisms that operate when 1 reacts with H 2 O 2 and organic substrates, but it does provide experimental evidence, corroborated by a full DFT analysis and isotopic labelling characterization, for the generation of a Fe(V)(O)(OH) species, a powerful oxidant, under conditions relevant to catalysis. Thus, this work provides a new fundamental framework for understanding the nature of the ironbased species responsible for performing alkane hydroxylation and olefin cis-dihydroxylation in a synthetic biomimetic system that may have enzymatic relevance. Also, we have demonstrated the potential of VT-MS in the investigation of reactive intermediates.