Characterization of a heterobimetallic nonheme Fe ( III ) – O – Cr ( III ) species formed by O 2 activation †

Oxygen activation is generally carried out by metalloenzymes with mononuclear or homodinuclear iron or copper active sites. However there are two notable exceptions to this generalization, namely the heme/copper center of cytochrome oxidase essential for mammalian respiration and the nonheme Fe–O–Mn center of Class 1c ribonucleotide reductases found in pathogenic bacteria. There has been significant progress in obtaining synthetic models for the heme/copper center of cytochrome oxidase, but less effort has been devoted to the synthesis of nonheme (m-oxo)heterobimetallic complexes. In 1992 Wieghardt described a series of carboxylate-bridged (TACN)Fe–O–M(Me3TACN) (M = Cr or Mn, TACN = 1,4,7-triazacyclononane, Me3TACN = 1,4,7-trimethyl-1,4,7-triazacyclononane) complexes obtained by hydrolysis between FeCl3(TACN) and MCl3(Me3TACN) precursors. 11

][3] However there are two notable exceptions to this generalization, namely the heme/copper center of cytochrome oxidase essential for mammalian respiration 1,4,5 and the nonheme Fe-O-Mn center of Class 1c ribonucleotide reductases found in pathogenic bacteria. 6,7There has been significant progress in obtaining synthetic models for the heme/copper center of cytochrome oxidase, [8][9][10] but less effort has been devoted to the synthesis of nonheme (m-oxo)heterobimetallic complexes.In 1992 Wieghardt described a series of carboxylate-bridged (TACN)Fe III -O-M(Me 3 TACN) (M = Cr III or Mn III , TACN = 1,4,7-triazacyclononane, Me 3 TACN = 1,4,7-trimethyl-1,4,7-triazacyclononane) complexes obtained by hydrolysis between FeCl 3 (TACN) and MCl 3 (Me 3 TACN) precursors. 11ore recently, Fukuzumi and Nam reported the crystal structure of a novel Fe III -O-Sc III complex, which was obtained from the reaction of [Fe IV (O)(TMC)(NCCH 3 )] 2+ (TMC = 1,4,8,11-tetramethylcyclam) with Sc(OTf) 3 . 12However, no synthetic nonheme Fe-O-M complex (where M is a non-iron metal) has thus far been generated by Bubbling O 2 into a solution of 1 mM Fe(TMC)(OTf) 2 and 1 mM Cr(OTf) 2 in CH 3 CN at À40 1C rapidly elicited a UV-vis spectrum with bands at 358, 398, 447 and 558 nm (Fig. 1), suggesting the formation of a new species designated as 1.This spectral pattern was not observed in the absence of either Fe(TMC)(OTf ) 2 or Cr(OTf) 2 from the reaction mixture; Fe(TMC)(OTf) 2 simply did not react with O 2 , but the reaction of Cr(OTf) 2 with O 2 gave rise to features at 358 and 445 nm (Fig. 1), distinct from those of 1. Species 1 had a half-life of 10 h at À40 1C and rapidly decayed upon warming to RT.Taken together, these observations implicate both Fe and Cr in the formation of 1.
ESI-MS analysis of the solution of 1 at À40 1C revealed dominant peaks at m/z 461.2 (positive mode) and 514.8 (negative mode) (Fig. S1, ESI †), which were not observed in oxygenated solutions lacking either Fe(TMC)(OTf) 2 or Cr(OTf) 2 .The ions observed have masses and isotope distribution patterns that correspond to [Fe(TMC)(OTf)] + and [CrO(OTf) 3 ] À , respectively.Furthermore, the [CrO(OTf) 3 ] À peak was upshifted by 2 units when 18 O 2 was used, showing the incorporation of an oxygen atom from O 2 (Fig. S2, ESI †).Based on these results, we tentatively assign 1 as a heterobimetallic [(TMC)Fe III -O-Cr III (OTf) 4 ] complex, which undergoes homolysis of the Fe-O bond to give rise to the observed mass spectral features.
To test this hypothesis, we investigated the reaction of [Fe IV (O)(TMC)(NCCH 3 )] 2+ 13 with Cr(OTf) 2 in CH 3 CN at À40 1C as a more direct means of synthesizing putative species 1.As shown in Fig. 2, the addition of Cr(OTf) 2 to [Fe IV (O)(TMC)(NCCH 3 )] 2+ in CH 3 CN solution caused the instantaneous disappearance of its characteristic 824 nm peak concomitant with the growth of bands at 398, 447, and 558 nm that are assigned to 1. Titration experiments (Fig. 2, inset) revealed that the transformation was complete upon addition of 1 eq.Cr(OTf) 2 , strongly suggesting a 1 : 1 Fe : Cr stoichiometry for 1.This solution also gave rise to ESI-MS spectra with the same dominant peaks as the complex generated by O 2 activation.A control experiment between Cr(OTf) 2 and PhIO did not elicit the same peaks as found in 1 (Fig. S3, ESI †), suggesting that [Fe IV (O)(TMC)(NCCH 3 )] 2+ acts more than just an oxygen atom donor to Cr(OTf) 2 .These results demonstrate that 1 can be generated by either O 2 activation or inner-sphere electron transfer.
In order to obtain structural insight, Fe K-edge X-ray absorption spectroscopy studies were carried out on 1.As shown in Fig. S4 (ESI †), the Fe K-edge of 1 was found at 7124.0 eV, which is comparable to those of known Fe III (TMC) and related complexes. 14Species 1 also exhibits a pre-edge feature that is associated with 1s-to-3d transitions with an area of 11 units.The Fouriertransformed EXAFS region revealed two prominent features at R + D B 1.8 Å and 3.2 Å (Fig. 3, left).The best fit of the data (fit #8 in Table S1, ESI †) consisted of 1 O/N scatterer at 1.81 Å, 5 O/N scatterers at 2.17 Å, 4 C scatterers at 2.91 Å and a Cr scatterer at a distance of 3.65 Å.The 2.17 Å and 2.92 Å scatterers arise from the supporting TMC ligand, while the 1.81 Å scatterer has an Fe-O distance typically found for oxo bridges in Fe III -O-M III complexes. 15The 3.2 Å feature corresponds to a Cr scatterer at 3.65 Å; its intensity derives from multiple scattering pathways due to a linear Fe-O-Cr core.Indeed, the FeÁ Á ÁCr distance is typical of the metal-metal distances found for linear Fe III -O-M complexes 12,16,17 and exemplified by [(py)(TPP)Cr III -O-Fe III (TMP)] (r(FeÁ Á ÁCr) = 3.60 Å; py = pyridine; TPP = tetraphenylporphin dianion; TMP = tetramesitylporphin dianion). 18We thus propose that 1 has the structure shown in Fig. 3 right.
The proposed structure for 1 resembles that found in the crystal structures of the recently described [(TMC)Fe III -O-Sc III (OTf) 4 (L)] complex (2, L = H 2 O or NCCH 3 ). 12However, they differ in several respects.Although the FeÁ Á ÁM distances are essentially identical for 1 and 2, the respective Fe-O and M-O distances are distinct.The Fe-O distance of 1.81 Å for 1 is 0.07 Å longer than that found for 2, while the Cr-O distance of 1.84 Å (deduced from the difference between the FeÁ Á ÁCr and the Fe-O distances from the EXAFS analysis, assuming +Fe-O-Cr B 1801) is 0.07 Å shorter than the Sc-O distance of 1.91 Å observed for 2 in its crystal structures.The distinct M-O distances in 1 and 2 presumably reflect the difference between the more covalent Cr-O bond and the more ionic Sc-O bond, which also affect the corresponding Fe-O bond.Another feature distinguishing 1 from 2 is the intensity of the XAS pre-edge feature.Complex 1 exhibits a pre-edge area of 11 units, typical of a six-coordinate iron(III) center, 19,20 while 2 (L = NCCH 3 ) has a much larger preedge area of 32 units, 12 reflecting the square pyramidal geometry of its iron(III) center.Lastly, the four methyl groups of the TMC ligand are shown in Fig. 3 right as being oriented anti with respect to the oxo bridge, opposite to the orientation found crystallographically for the methyl groups in 2. 12 Although we do not have direct proof, our main argument to favor the anti orientation over the syn one is the observed immediate formation of 1 upon Cr(OTf) 2 addition to a solution of [Fe IV (O)(TMC)(NCCH 3 )] 2+ .As the TMC methyl groups are oriented anti to the oxo moiety in the precursor, 13 it seems unlikely that a change in their relative orientations could occur at À40 1C within this very short time scale.
The likelihood of a sixth ligand for the iron(III) center in 1 is supported by the change in the spectral features of 1 upon addition of NCS À or NCO À .As shown in Fig. 4, there are small shifts of the three bands, as well as increases in intensity.Titration experiments showed that only 1 eq. of NCS À or NCO À was needed to transform 1 fully into 1-NCS or 1-NCO (Fig. S5 and S6, ESI †).ESI-MS analysis of 1-NCS and 1-NCO revealed respective positive mode peaks at m/z 370 and 354, corresponding to [Fe(TMC)(NCS)] + and [Fe(TMC)(NCO)] + fragment ions (Fig. S7 and S8, ESI †), suggesting the occupation of the axial position trans to the  oxo bridge by these anions.Furthermore, 1-NCS exhibits an Fe K-edge energy of 7124.3 eV, comparable to the 7124.0eV value found for 1. 1-NCS also exhibits a pre-edge feature with an area of 9 units (Fig. S9, ESI †), which is close to the 11 units found for 1 but much smaller than the 32 units associated with 2, showing that a six-coordinate iron(III) center in 1-NCS is maintained.EXAFS analysis of 1-NCS shows the presence of a linear Fe-O-Cr core like that in 1, but with a 1.85 Å Fe-O bond and an FeÁ Á ÁCr distance of 3.67 Å (Table S2 and Fig. S10, ESI †).The observed lengthening of the Fe-m-O bond can be rationalized by the axial NCS À binding to the iron(III) center.Based on all the information above, 1 is proposed to be a heterobimetallic m-oxo species with an Fe-O-Cr core, and the Fe atom has a 6-coordinate environment with the axial position available for ligand substitution (Fig. 3, right).
Complex 1 was further studied by EPR and resonance Raman spectroscopy.It is EPR-silent, which is as expected due to antiferromagnetic coupling mediated by the oxo bridge between the Fe(III) and the Cr(III) centers, as seen for two other Fe(III)-O-Cr(III) complexes. 11,18Excitation of 1 with a 568.2 nm laser elicits a resonance-enhanced vibration at 773 cm À1 (Fig. 5), which falls within the 700-900 cm À1 range found for the n as (Fe-O-Fe) modes of oxo-bridged diiron(III) complexes. 21his assignment is corroborated by the observed downshift of this vibration to 730 cm À1 upon 18 O-substitution into the oxo bridge.Although a 35 cm À1 downshift for a diatomic Fe-O mode is predicted by Hooke's Law, the experimentally obtained 18 O shift is 43 cm À1 .This larger than predicted shift has also been reported for corresponding vibrations of several oxo-bridged diiron(III) complexes. 21There is also a weaker feature found at 746 cm À1 that exhibits an upshift of 7 cm À1 upon 18 O-substitution; this is an unusual observation that we cannot explain.The 773 cm À1 vibration is also weakly enhanced upon 514.5 nm excitation but not observed with 457.9 or 647.1 nm irradiation, suggesting that the 558 nm absorption band can be associated with a transition of the Fe-O-Cr moiety.
With the nature of 1 reasonably well characterized, we return to an analysis of the O 2 activation reaction results.Based on the molar extinction coefficients of 1 determined from the stoichiometric conversion of [Fe IV (O)(TMC)(NCCH 3 )] 2+ to 1 by Cr(OTf) 2 , we conclude that 1 is produced in about 30% yield from the reaction of equimolar amounts of Fe(TMC)(OTf) 2 and Cr(OTf) 2 to O 2 .The yield of 1 was unchanged by increasing the Fe(TMC)(OTf) 2 /Cr(OTf) 2 ratio from 1 to 10 ([Cr(OTf) 2 ] = 1 mM) (Fig. S11, ESI †), suggesting that Cr(OTf) 2 is the limiting reagent.In contrast, the yield of 1 increased to 95% when the concentration of Fe(TMC)(OTf) 2 was fixed at 1 mM and the Cr(OTf) 2 concentration was raised from 1 mM to 10 mM (Fig. S12, ESI †).Therefore, the % yield of 1 is dependent on the amount of Cr(OTf) 2 , but not on the amount of Fe(TMC)(OTf) 2 .These results can be rationalized by the O 2 activation pathway proposed in Scheme 1, in which the four electrons needed to reduce O 2 to the oxidation level of water are provided by 1 eq.Fe(TMC)(OTf) 2 and 3 eq.Cr(OTf) 2 , not unlike the four redoxactive centers required for O 2 activation by cytochrome oxidase. 4n the present case, we postulate that O 2 initially binds to the O 2sensitive Cr(OTf) 2 to form a transient adduct (analogous to that characterized by Nam in the reaction of [Cr II (TMC)Cl] + with O 2

22
) that is then trapped by Fe(TMC)(OTf) 2 to generate a yet unobserved Fe III -O-O-Cr III peroxo-bridged intermediate.This intermediate is then reduced by another 2 eq.Cr(OTf) 2 to form 1. Thus the ca.30% yield of 1 observed under limiting Cr conditions reflects the 1 : 3 stoichiometry of Fe(TMC)(OTf) 2 /Cr(OTf) 2 needed to make 1.On the other hand, under limiting Fe conditions, enough Cr-O 2 adduct is formed to react with all the available Fe(TMC)(OTf) 2 to convert the latter almost quantitatively to 1.

Fig. 3
Fig. 3 Left: Fourier-transformed Fe K-edge EXAFS data for 1 (dotted black line) and corresponding best fit (solid red line, fit #8 in Table S1, ESI †).Inset shows unfiltered k-space data and its fit.Right: Proposed structure for 1 (L = CH 3 CN or OTf; X = CH 3 CN, NCO or NCS).