Di- and Tetrairon(III) μ-Oxido Complexes of an N3S-Donor Ligand: Catalyst Precursors for Alkene Oxidations

The new di- and tetranuclear Fe(III) μ-oxido complexes [Fe4(μ-O)4(PTEBIA)4](CF3SO3)4(CH3CN)2] (1a), [Fe2(μ-O)Cl2(PTEBIA)2](CF3SO3)2 (1b), and [Fe2(μ-O)(HCOO)2(PTEBIA)2](ClO4)2 (MeOH) (2) were prepared from the sulfur-containing ligand (2-((2,4-dimethylphenyl)thio)-N,N-bis ((1-methyl-benzimidazol-2-yl)methyl)ethanamine (PTEBIA). The tetrairon complex 1a features four μ-oxido bridges, while in dinuclear 1b, the sulfur moiety of the ligand occupies one of the six coordination sites of each Fe(III) ion with a long Fe-S distance of 2.814(6) Å. In 2, two Fe(III) centers are bridged by one oxido and two formate units, the latter likely formed by methanol oxidation. Complexes 1a and 1b show broad sulfur-to-iron charge transfer bands around 400–430 nm at room temperature, consistent with mononuclear structures featuring Fe-S interactions. In contrast, acetonitrile solutions of 2 display a sulfur-to-iron charge transfer band only at low temperature (228 K) upon addition of H2O2/CH3COOH, with an absorption maximum at 410 nm. Homogeneous oxidative catalytic activity was observed for 1a and 1b using H2O2 as oxidant, but with low product selectivity. High valent iron-oxo intermediates could not be detected by UV-vis spectroscopy or ESI mass spectrometry. Rather, evidence suggest preferential ligand oxidation, in line with the relatively low selectivity and catalytic activity observed in the reactions.


INTRODUCTION
The interaction between iron and sulfur in metalloproteins and in biomimetic molecular systems has attracted increased attention in recent years (Beinert et al., 1997;Ohnishi, 1998;Rao and Holm, 2004;Ballmann et al., 2008a;Meyer, 2008;Lill, 2009). Among these, the majority of the Fe/S structural motifs are dominated by clusters where the iron center is in distorted tetrahedral environments (Beinert et al., 1997;Meyer, 2008); nonetheless, there are examples of molecular systems where the Fe centers display diverse coordination geometries (Ballmann et al., 2008b;Fuchs et al., 2010). In some typical molecular systems, the metal center has also been found to be involved in secondary bonding interactions with ether-O and thioether-S units (Ballmann et al., 2008b). Although in all these examples sulfur, being a soft donor, displays preference for low valent Fe(II), there are molecular systems where the sulfur atom has been found to be bonded to high valent (III and IV) iron centers with distorted octahedral geometry (Harrop and Mascharak, 2004;McDonald et al., 2010;Widger et al., 2014). Overall, the Fe-S interaction in various molecular systems having iron in different formal oxidation states is an interesting field of research.
On the other hand, the selective and environmentally benign oxidation of hydrocarbons using affordable and efficient catalysts is another important area of modern synthetic chemistry. In this regard, bio-inspired iron chemistry has received increasing attention in recent decades due to the natural iron abundance in the earth crust, and the highly selective catalytic hydrocarbon oxidations of iron-containing oxygenases such as cytochrome P450 (Hasemann et al., 1995), soluble methane monooxygenase (sMMO) (Tinberg and Lippard, 2011), and Rieske dioxygenases (Wackett, 2002). Among the (catalytic) properties of these ironbased enzymes, the selective activation of C-H bonds under mild conditions is one of the most striking aspects. Many mono-and diiron complexes with various multidentate N-based ligands have been investigated in order to mimic the structures as well as functions of non-heme iron enzymes, and to potentially develop sophisticated oxidation catalysts (Costas et al., 2000(Costas et al., , 2004Sun et al., 2011;Lindhorst et al., 2015;Gamba et al., 2017). Such studies indicate that fine tuning of the coordination environment of the iron centers plays a crucial role in the catalytic activity, which includes O 2 binding, followed by electron-transfer from Fe to O 2 to afford either iron-superoxo (Fe III -O .− 2 ), iron-peroxo (Fe III -O 2− 2 ), or iron-oxo (Fe IV/V =O) species after initial O-O bond cleavage (Kim et al., 1997;Hazell et al., 2002;Rohde et al., 2003;Ye and Neese, 2011;De Visser et al., 2013;Wang et al., 2013a;Mitra et al., 2014;Nam, 2015).
A number of dinuclear Fe(III)-µ-oxido complexes relevant to the aforementioned enzymes from both structural and functional perspectives, have been studied as oxidation catalysts (Romakh et al., 2007;Visvaganesan et al., 2009;Wang et al., 2013a). Depending on the specific ligand environment, catalytic oxidation reactions using such dinuclear µ-oxido Fe(III) complexes can proceed via a radical intermediate or a high valent metal oxo species, or a combination of both, and the nature of these intermediates/active oxidants exerts a profound influence on the product distribution (Costas et al., 2000(Costas et al., , 2004. The experimental evidence in all cases indicate that there is a strong influence of the ligand system on the catalytic activity, as well as the choice of oxidant (i.e., ultimate oxygen donor), which can also affect the product distribution (Costas et al., 2000(Costas et al., , 2004. In spite of the extensive oxidative catalytic activity studies of these dinuclear Fe(III)-µ-oxido complexes, to the best of our knowledge there are very few well-characterized complexes that have been used as oxidation catalysts where sulfur occupies one of the coordination sites (McQuilken and Goldberg, 2012). The easily oxidizable nature of sulfur argues against its use in these types of systems (Widger et al., 2014), despite the fact that studying small molecular catalysts with sulfur-containing ligands can be very useful in modeling key metal-sulfur interactions that play a significant role in nonheme enzymes. Studies on sulfur oxygenation in a number of biomimetic non-heme iron(III)-thiolate complexes indicate that a long Fe-S bond distance makes the sulfur unit susceptible to attack by O 2 in a reaction where iron maintains the +3 oxidation state (McQuilken and Goldberg, 2012). The active site of the non-heme iron enzyme cysteine dioxygenase (McCoy et al., 2006) is believed to pass through a Fe(III)-superoxo intermediate. DFT and QM/MM computational studies predict that the formation of the superoxo intermediate is followed by formation of an energetically favorable cyclic four-membered Fe-O-O-S ring structure that undergoes O-O bond cleavage to form a Fe(IV)(Oxo)-sulfinate analog. This metal oxo unit transfers the second oxygen atom to generate the cysteine-sulfinic acid product (Kumar et al., 2011;McQuilken and Goldberg, 2012).
Goldberg and coworkers have reported Fe(II) complexes of interesting pentadentate ligand systems incorporating one sulfur donor in efforts to model the reactivity of the active site of cysteine dioxygenase. While a pentadentate ligand with one thiolate donor is oxidized to the corresponding sulfoxide upon reaction with dioxygen, an analogous thioether ligand permits formation of a non-heme Fe(IV)oxo species that can perform oxygen atom transfer (Widger et al., 2014). In this context, it is worth mentioning that one of the most realistic biomimetic model systems (both in terms of structure and reactivity) for cysteine dioxygenase, [Tp Me,Ph Fe II CysOEt] (Tp Me,Ph = hydridotris(3-phenyl-5-methylpyrazol-1-yl)borate), was reported by Limberg and coworkers (Sallmann et al., 2012), who used isotope ( 16 O/ 18 O) experiments to confirm that the treatment with dioxygen mainly leads to cysteine dioxygenase activity, i.e., deoxygenation of the bound cysteine ethyl ester.

Synthesis and Characterization of Complexes
The PTEBIA ligand was prepared following the procedure reported by Castillo and coworkers (Castillo et al., 2012). Addition of 1 equivalent of Fe(II)(OTf) 2 (OTf = triflate, CF 3 SO − 3 ) to a tetrahydrofuran solution of PTEBIA leads to an immediate change of the color of the ligand solution  to green. Refluxing of this green solution for 4 h, followed by crystallization by slow diffusion of diethylether into an acetonitrile solution of the product leads to yellow, needle shaped crystals of 1a (Figure 2).
The solid-state structure of 1a consists of a tetranuclear unit formed by four PTEBIA ligands, four bridging oxido ligands and four Fe(III) ions. To the best of our knowledge, this represents the first example of a tetrairon, tetra-oxido cluster with a sulfur-containing ligand. All the Fe(III) units When PTEBIA contained a sub-stoichiometric amount of HCl from incomplete neutralization during the last synthetic step in the PTEBIA synthesis (as evidenced by the immediate precipitation of AgCl upon addition of AgNO 3 to a chloroform Symmetry transformation used to generate equivalent atoms: (') 1/2-x, 3/2-y, 1-z.
Cu(I/II)-S(thioether) distance observed for coordination of methionine in plastocyanin (2.9 Å) (Sahoo and Ray, 2007). The Fe(III) centers are oxido-bridged and are 3.5821(11) and 3.5809(11)Å (two crystallographically independent molecules) apart from each other with an Fe-O-Fe angle of 180 • , which is in accordance with similar dinuclear Fe(III)-µ-oxido complexes reported by McKenzie and coworkers (Vad et al., 2012) and Wang et al. (2003). Mass spectrometric measurements of 1a gave rise to two major peaks corresponding to the mononuclear complex, i.e. at 673.7 amu for [Fe(PTEBIA)(CF 3 SO 3 )] + , and at 469.8 corresponding to protonated [HPTEBIA] + (Figure S1). A peak arising from 1b was also observed at 559.7 amu, assigned to [Fe(PTEBIA)(Cl)] + ; no µ-oxido bridged species could be detected by either ESI or MALDI-TOF mass spectrometry. IR characterization revealed sharp resonances between 820 and 600 cm −1 assigned to asymmetric and symmetric Fe-O stretching modes of the Fe-O-Fe units in 1a and 1b (819, 749, 637 cm −1 in the former; 815, 780, 746, 633 cm −1 in the latter). Additional characterization of 1a and 1b was obtained by Mössbauer spectroscopy from 0.02 g of a batch of mixed crystals obtained from the reaction of Fe(OTf) 2 with PTEBIA, revealing the presence of two high spin Fe(III) centers with different coordination environments in a 1:1 ratio of 1a and 1b (Figure 4). Since more ionic coordination  environments with higher coordination numbers result in higher isomer shifts (δ), the subspectrum with δ = 0.46 mm·s −1 can be assigned to 1b, whereas the subspectrum with δ = 0.38 corresponds to 1a.
Conversion of 1a to 1b was achieved in acetonitrile solution, as evidenced by UV-vis spectroscopy: addition of two equivalents of NBu 4 Cl as a source of chloride to 0.4 mM solutions of 1a resulted in spectra that are virtually identical to those of the chlorido-containing 1b, see Figure 5 and Figure S2.
To further probe the coordination mode of the PTEBIA ligand toward Fe(III) centers, equimolar amounts of Fe(II)(ClO 4 ) 2 and PTEBIA were heated to reflux in THF solution, followed by recrystallization from a methanolic solution by slow vapor diffusion of diethylether. Electrospray mass spectrometry on methanolic solutions of the green crystals obtained from the reaction show a major peak at 262.8 amu (see Figure S3) (Norman et al., 1998).  IR characterization of 2 shows characteristic asymmetric (ν asym ) and symmetric (ν sym ) C=O stretching bands as a broad peak at 1,614 cm −1 and two very sharp peaks for ν sym at 1,490 and 1,453 cm −1 , consistent with the presence of two bridging formates (Deacon and Phillips, 1980;Kurtz, 1990;Das et al., 2014aDas et al., ,b, 2018. In addition, the presence of a very sharp band at 1,080 cm −1 confirms the presence of perchlorate anions; the bands at 744 and 620 cm −1 correspond to the asymmetric and symmetric stretching modes of the Fe-O-Fe units (Kurtz, 1990;Norman et al., 1990). Acetonitrile solutions of 2 (0.5 mM) exhibit an absorption maximum at 342 nm (ε = 4,400 M −1 cm −1 ), characteristic of the Fe-O-Fe moiety (Kurtz, 1990;Norman et al., 1990;Do et al., 2012); unlike 1a and 1b, no lower energy bands were observed as the sulfur atom is not bound to the metal centers (Figure 7). Low temperature UV-vis experiments using a 1:1 (molar equivalent) of H 2 O 2 (30 wt.% in H 2 O) and CH 3 COOH (> 99.7%) solution as oxidant and keeping overall complex:oxidant ratio at 1:2.5 reveal that the formate ligands of 2 may be protonated under these conditions, thus allowing the sulfur atom of PTEBIA to interact with the Fe(III) centers, as evidenced by the new absorption band at around 410 nm. This band is more prominent at low temperature (228 K) and appears only as a shoulder above 250 K (Figure 7) and was assigned to a S → Fe LMCT transition.

Oxidation Catalysis
Jacobsen et al. (White et al., 2001) have effected efficient epoxidation of terminal long-chain alkenes using an iron  (Fujita and Que, 2004). Similarly, Stack and coworkers (Dubois et al., 2003) used an Fe(III)-O-Fe(III) complex with aqua and nitrogen donor ligands, [Fe 2 (µ-O)(OH 2 ) 2 (phen) 2 ] 4+ (phen=phenanthroline) as a catalyst/catalyst precursor for alkene epoxidation using peracetic acid as the oxidant. The combination of hydrogen peroxide with a suitable carboxylic acid (e.g., acetic acid) is believed to generate a peracid that in turn may generate a high valent metal oxo complex that functions as an active alkane/alkene oxidant (Fujita and Que, 2004), but it has also been suggested that the reaction of an Fe(II) complex with peracetic acid can lead to an Fe(III) κ 2 -peracetate complex with subsequent dissociation of the peracetate ligand to form a ferryl acyl radical species, i.e., an (OAc . )Fe(IV)=O species, that is an active oxidant (Wang et al., 2013b).
To assess the effect of the mixed N 3 S donor set of PTEBIA on the activity and selectivity of iron complexes in the oxygenation of alkenes, we investigated the oxidation of styrene and a number of prototypical cyclic alkene substrates, viz. cyclohexene, 3-ethylcyclohexene, cyclooctene, under mild conditions (Scheme 1) using a bulk sample containing an approximate 1:1 mixture of 1a and 1b as catalyst precursor(s). The oxidation conditions were optimized by using cyclohexene as the model compound ( Table 7). Based on the maximum reactivity and selectivity, acetonitrile was used as solvent with CH 3 COOH as additive and H 2 O 2 as oxidant. These optimized conditions were applied in all oxidation experiments involving the different substrates. Mass spectrometry (ESI-MS) indicated that mixtures of 1a and 1b convert to dimeric [(Fe 2 )(OAc) 2 (µ-O)(PTEBIA) 2 (CF 3 SO 3 )] + in acetonitrile solution, as evidenced by the peak detected at 1332.8 amu (Figure S4, Supplementary Material). Cyclohexene, cyclooctene and 3-ethylcyclohexene were oxidized in moderate conversions (29,44,and 15%,Table 8) to the overoxidation products of the epoxides, namely diols and diketones. Although styrene oxidation (63% yield) afforded benzylmethanol (by epoxide ring opening), the formation of benzaldehyde as the major product indicates that the main active species is likely a transient radical (single electron oxidation). In comparison to the previously reported µ-oxido di-iron(III) complexes with N and O-based ligands [e.g., 2,6-bis(N-methylbenzimidazol-2yl)pyridine (Wang et al., 2003), DPCPMPP (Das et al., 2015)], the catalytic efficiency (% of product formation and TON) of 1a and 1b is comparatively low (Romakh et al., 2007;Das et al., 2015). In addition to [(Fe 2 )(OAc) 2 (µ-O)(PTEBIA) 2 (CF 3 SO 3 )] + detected by ESI MS, the main peaks in the mass spectra, and their isotopic patterns are consistent with species formulated as  (Prat et al., 2013). Instead, it may proceed via homolysis to give transient [LFe (IV) O] and hydroxyl radical (OH) intermediates, with participation of both species in the observed oxidations (Trettenhahn et al., 2006).

CONCLUSION
We have synthesized and characterized a new tetrairon tetraoxo cluster (1a) with a sulfur-containing ligand. In the UV-vis spectroscopic experiments, a prominent sulfur to iron charge transfer band (390-430 nm) was observed at room temperature for 1a and the corresponding dinuclear Fe(III)-O-Fe(III) complex 1b, whereas for the related dinuclear dicarboxylatebridged Fe(III)-O-Fe(III) complex 2 it was only visible at low temperature (228 K) in the presence of H 2 O 2 /CH 3 COOH (presumably after initial protonation/dissociation of the oxo and formate groups of 2), and disappears at room temperature, likely due to loss of the Fe-S bond and/or sulfoxidation of PTEBIA. The catalytic efficiency of the µ-oxido iron(III) complex mixture of 1a and 1b for the oxidation of alkenes has been investigated, revealing that they act as moderate oxidation catalyst(s)/catalyst precursor(s). The reaction appears to proceed partially through a metal-centered process in tandem with freeradical oxidation by reactive oxygen species. Low temperature UV-vis spectroscopy and mass spectrometry were employed to gain insight into possible reactive intermediates/active oxidation catalysts, revealing that both H 2 O 2 /CH 3 COOH and PhIO preferably oxidize the thioether group of the ligand, in contrast with previous oxygenations with related copperbased PTEBIA systems, where the thioether functionality remains intact.

MATERIALS AND METHODS
The ligand PTEBIA was prepared following the procedure reported by Castillo et al. (2012). All reagents and solvents were of analytical or spectroscopic grade purchased from Sigma Aldrich, Fisher chemicals or VWR, and were used without further purification. Cyclohexene (≥ 99.0%) contained ∼0.01% of 2,6-di-tert-butyl-4-methylphenol as stabilizer and was used as received. Caution! Even though no problems were encountered in this work, caution should always be taken while using high concentrations of hydrogen peroxide (H 2 O 2 ), as well as metal perchlorates. Infrared spectra were recorded in the 4,000-400 cm −1 range on a Nicolet Avatar 360 FTIR spectrometer, as KBr pellets. Mass spectra were obtained on a JEOL JMS-SX-102A mass spectrometer at an accelerating voltage of 10 kV with a nitrobenzyl alcohol matrix and Xenon atoms at 6 keV (FAB + ), a JEOL JMS-AX505HA spectrometer (Electron Ionization), or a Bruker Daltonics Esquire 6000 spectrometer with ion trap (Electrospray). Elemental analyses were performed at the microanalytical facility of the Instituto de Química, UNAM, Mexico. Analytical achiral GC was performed on an Agilent 6850 GC with FID detector using an Agilent DB-WAX (30.0 m × 0.25 mm) column at mL/min He carrier gas flow. Chiral GC was performed on an Agilent 6850 GC with FID detector. The 1 H NMR spectra were recorded with a Varian Gemini 200 apparatus or a Varian Mercury 300 MHz spectrometer.
Substrate conversions in catalytic experiments were determined by GC-MS. The GC-MS analyses were performed with an Agilent 6890 N Network GC system equipped with a DB-1MS column (30 m × 0.25 mm) and an Agilent 5973 Network MS detector. Calibration curves were obtained from commercial products purchased from Aldrich or TCI when available or from pure isolated products obtained from a catalytic reaction using a FID-detector GC with a HP-INOWAX column (30 m × 0.25 mm) (1,2-dichlorobenzene used as an internal standard). The concentrations of each organic product were calibrated relative to that of an internal standard (1,2-dichlorobenzene) with a known concentration.

Oxidation Experiments
Each catalytic experiment was performed at least twice and the reported conversion is the average value. A general procedure for the oxidation experiments is as follows: magnetic stirring bar, catalyst complex (18 µmol), 2 mL of CH 3 CN, acetic acid (AcOH, 50 µL, 85 µmol), 230 µL of H 2 O 2 (33% in water, 2.0 equivalents with respect to the substrate) and substrate (1 mmol) were placed in a Schlenk flask. The reaction mixture was stirred under argon at 35 • C for the designated time. Sodium thiosulfate (ca. 400 mg, 2.5 mmol) was then added to the reaction mixture to quench further oxidation. 1,2-dicholorobenzene was added to the mixture followed by extraction with n-pentane and filtering through a silica gel column for analysis by GC-MS.

X-Ray Structure Determination
Crystal data and details of the data collections are given in Table 9. X-ray data for complexes 1a and 1b were collected on a STOE IPDS II diffractometer (graphite monochromated Mo-Kα radiation, λ = 0.71073 Å) by use of ω scans at −140 • C. The structures were solved by direct methods (SHELXS-2014) and refined on F 2 using all reflections with SHELXL-2014 (Sheldrick, 2008). Non-hydrogen atoms were refined anisotropically. Hydrogen atoms were placed in calculated positions and assigned to an isotropic displacement parameter of 1.2 / 1.5 U eq (C). In each compound one of the two CF 3 SO − 3 ions was found to be disordered [occupancy factors: 1a = 0.929(15)/0.071(15), 1b = 0.784(4)/0.216(4)]. SAME restraints and EADP constraints were used to model the respective disorders. In 1a one 2,4-dimethylphenyl moiety involving the carbon atoms C51(A/B/C) to C58(A/B/C) was found to be disordered about three positions A, B, and C along with acetonitrile (N21, C95, C96, belonging to B) and water (O31, belonging to C). After initial refinement the occupancy factors  (7)] was refined using RIGU restraints. Faceindexed absorption corrections were performed numerically with the program X-RED (Stoe & Cie and X-RED, 2005). The crystal of 2 was immersed in cryo-oil, mounted in a MiTeGen loop, and measured at a temperature of 170 K on a Rigaku Oxford Diffraction Supernova diffractometer using Mo Kα radiation. The CrysAlisPro (Rigaku Oxford Diffraction, 2013) software was used for cell refinement and data reduction. Empirical absorption correction based on equivalent reflections was [CrysAlisPro (Rigaku Oxford Diffraction, 2013)] was applied to the intensities before structure solutions. The structure was solved by charge flipping method using the SUPERFLIP (Palatinus and Chapuis, 2007) and the structure refinement was carried out using SHELXL (Sheldrick, 2015) program. The crystal of 2 contained solvent accessible voids but no satisfactory solvent model could be found. The contribution of the missing solvent to the calculated structure factors was taken into account by using the SQUEEZE routine of PLATON (Spek, 2009). The missing solvent was not taken into account in the unit cell content. Hydrogen atoms were positioned geometrically and constrained to ride on their parent atoms, with C-H = 0.95-0.99 Å, O-H = 0.84 Å, and U iso = 1.2-1.5 U eq (parent atom). The highest peak is located 1.40 Å from atom S1 and the deepest hole is located 0.59 Å from atom S1.

Mössbauer Measurement
The Mössbauer spectrum was recorded at 80 K with a 57 Co source in a Rh matrix, using an alternating constant-acceleration Wissel Mössbauer spectrometer operated in the transmission mode and equipped with a Janis closed-cycle helium cryostat or with a Mössbauer-Spectromag cryostat. Isomer shifts (cf. caption, Figure 3) are given relative to iron metal at ambient temperature. Experimental data were simulated using the Mfit software (developed by E. Bill, Max-Planck Institute for Chemical Energy Conversion, Mülheim/Ruhr, Germany, 2008).

DATA AVAILABILITY
The raw data supporting the conclusions of this manuscript will be made available by the authors, without undue reservation, to any qualified researcher. Crystallographic data for the structures in this paper have been deposited with the Cambridge Crystallographic Data Center, CCDC, 12 Union

AUTHOR CONTRIBUTIONS
BD performed all complex syntheses, performed low temperature studies in collaboration with BS-E, participated in the oxidation experiments, and contributed to the writing of the manuscript. AA-H performed all oxidation experiments and contributed to the writing of the manuscript. BS-E and EZ synthesized the PTEBIA ligand. SebD performed all Mössbauer measurements and analysis of the Mössbauer data. SerD, SB, and MH collected X-ray data and solved the crystal structures. TR supervised the oxidation studies and analyzed the results in collaboration with AA-H and BD, and supplied funding for the work. IC led and designed the study, contributed to the writing of the manuscript and supplied funding for the work. EN led and designed the study, contributed to the writing of the manuscript and supplied funding for the work.