Sc 3+ -promoted O–O bond cleavage of a ( μ -1,2-peroxo)diiron(III) species formed from an iron(II) precursor and O 2 to generate a complex with an Fe IV2 ( μ -O) 2 core

: Soluble methane monooxygenase (sMMO) carries out methane oxidation at 4 °C and under ambient pressure in a catalytic cycle involving the formation of a peroxodiiron(III) intermediate ( P ) from the oxygenation of the diiron(II) enzyme and its subsequent conversion to Q , the diiron(IV) oxidant that hydroxylates methane. Synthetic diiron(IV) complexes that can serve as models for Q are rare and have not been generated by a reaction sequence analogous to that of sMMO. In this work, we show that [Fe II (Me 3 NTB)(CH 3 CN)](CF 3 SO 3 ) 2 (Me 3 NTB = tris((1-methyl-1 H -benzo[d]imidazol-2-yl)methyl)amine) ( 1 ) reacts with O 2 in the presence of base, generating a ( µ -1,2-peroxo)diiron(III) adduct with a low O–O stretching frequency of 825 cm -1 and a short Fe•••Fe distance of 3.07 Å. Even more interesting is the observation that the peroxodiiron(III) complex undergoes O–O bond cleavage upon treatment with the Lewis acid Sc 3+ and transforms into a bis( μ -oxo)diiron(IV) complex, thus providing a synthetic precedent for the analogous conversion of P to Q in the catalytic cycle of sMMO.

Mössbauer spectroscopy of 2 measured at 4.2 K and zero applied field reveals a single quadrupole doublet with an isomer shift (δ) of 0.49 mm/s and a quadrupole splitting (ΔEQ) of 1.06 mm/s ( Figure 1, right panel, top), representing the two equivalent high-spin iron(III) sites of the (μoxo)(μ-1,2-peroxo)diiron(III) complex. Notably, the quadrupole splitting of 2 is among the smallest of the values previously reported for peroxodiiron(III) complexes (Table 1). 2 High-field measurements show that the doublet originates from a diamagnetic species that corresponds to 75-80% of the iron in the sample ( Figure S1).  (Table 1). Two other vibrations at 454 and 518 cm -1 also downshift by about 24 cm -1 upon 18 O2 incorporation, leading to their respective assignments as the νsym(Fe−    [c] Represents 70% of the 2+H + sample. The remaining 30% is another quadrupole doublet with d = 0.45 mm s -1 and DEQ = 0.69 mm s -1 ; its low-field Mössbauer spectrum is shown in Figure S2. Table 1.

Scheme 2. Ligand structures used in
The conclusions derived from the Raman-based correlations have been confirmed by X-ray absorption spectroscopy. The XANES spectrum of 2 has an Fe K-edge energy (E0) of 7126.3 eV ( Figure S6a), consistent with values for high-spin iron(III) centers. The pre-edge region of 2 can be fit with one peak centered at 7114.5 eV with an area of 18.2 units (Figure S6b and Table S2), values comparable with  those reported for other complexes with Fe III   2(μ-O)(μ-1,2-O2) cores. 30,32 The best fit of the EXAFS data obtained for 2 ( Figure 4)  complexes reported so far (Table 1). This observation is tied to the fact that 2 has one of the lowest observed ν (  A comparison of the spectroscopic properties of the (μoxo)(μ-1,2-peroxo)diiron(III) complexes listed in Table 1 shows that 2 has properties that most closely resemble those of A, [Fe III , where BPPE provides 6 pyridines to support a [Fe III 2(μ-O)(μ-1,2-O2)(μ-OAc) core. Complexes 2 and A are distinct from the others on the list in having blue-shifted peroxo(πv*)-toiron(III) (dπ orbital) charge transfer bands in the visible region, lower ν(O-O) values, and shorter Fe•••Fe distances. Complex 2 is the only complex on the list with benzimidazole donors, which are more basic than the pyridine and quinoline donors found on the other complexes 42 and thus expected to decrease the Lewis acidity of the iron(III) centers in 2 and give rise to the blue shift of its absorption maximum. In the case of A, the addition of a carboxylate bridge serves to lower the Lewis acidity of the metal centers. More importantly, these two complexes resemble each other in having the lowest ν(O-O) values and the shortest diiron distances in the series (Table 1). For A, the bridging acetate and the ethylene linker of the dinucleating ligand bring the two iron atoms closer to each other and presumably give rise to the lower ν(O-O) observed, but it is quite remarkable that 2 has the same features without similar structural constraints.
The lower ν(O-O) of 2 suggests that its O-O bond may be weaker than those of the other reported (μ-oxo)(μ-1,2-peroxo)diiron(III) complexes (Table 1). Additionally, the halflife of 2 (13 min at -40 o C) is also quite short when compared with related complexes in Table 1. The weakened bond and short half-life might prime it to undergo O-O bond cleavage by addition of a suitable acid and convert 2 into a highvalent diiron species. 30,32 However, addition of 1.5 eq HClO4 (or HOTf) red-shifts its lmax from 595 nm to a broader feature of comparable intensity around 640 nm (ε = 1300 M -1 cm -1 ) (Figures S10 and S11).
Surprisingly, the Fe•••Fe distance of 2+H + is essentially unchanged within experimental error from that of 2 (3.09 vs 3.07 Å, respectively) and shorter than the ~3.4-Å distances found for the other two (μ-hydroxo)(μ-1,2-peroxo)diiron(III) complexes characterized to date (G and H in Table 1 (Table 1). Thus protonation affects the stabilities of the Me3NTB and BnBQA peroxo complexes in opposite directions for reasons we do not yet fully understand.
In stark contrast, treatment of 2 with the Lewis acid Sc(OTf)3 in place of strong acid results in the cleavage of its O-O bond to afford a diiron(IV) complex 3. This outcome is manifested by the growth of an intense absorption feature at 600 nm (9000 M -1 cm -1 ) that forms over the course of an hour at -40 °C (Figure 1). Excitation of the intense visible chromophore of 3 using a 660-nm laser at -40 °C elicits resonantly enhanced Raman bands at 653 and 528 cm -1 , which downshift respectively by 30 and 17 cm -1 in a sample of 18 O2-labeled 3 ( Figure 5). The vibrational frequency of 653 cm -1 with an isotopic shift of 30 cm -1 for 3 reflects an Fe-O-Fe angle of close to 100°, characteristic of complexes with an M2(μ-O)2 'diamond core'. [37][38][39][40][41]43 Such an acute angle has thus far only been shown to be enforced by the presence of a second μ-oxo bridge. Furthermore, experiments starting with a 1:1 mixture of 16 (Figure S8). Thus, both oxygen atoms in 3 must derive from the peroxo moiety of one unique molecule of 2 ( Figure S8) and the oxo-bridged O-atom in 2 must be released from the complex, presumably upon combination with Sc(OTf)3 .   Table 2.

Scheme 3. Structures of ligands used in
The 653-cm -1 Raman band is reminiscent of the 674 cm -1 feature reported for [Fe IV , 39 which is associated with an A1 breathing mode of an Fe2(μ-O)2 core that has an Fe−O−Fe angle of 100°. 38,41 Furthermore, the observation of an 18 O-isotope shift of ~30 cm -1 confirms that this vibration is essentially an Fe-O stretching mode. Interestingly, the 653-cm -1 peak falls within error on the higher frequency line in Figure 3  , [Fe I- , shown as triangles, also fall on this line (see Table 2 for a comparison of properties for Fe2(μ-O)2 complexes). A similar assignment for the 650-700-cm -1 vibration observed in [Fe III Fe IV (μ-O)2(R-TPA)2] 3+ complexes has been made by Solomon and co-workers using normal coordinate analysis. 38 Complex 3 also exhibits a peak at 528 cm -1 with an 18 O-isotope shift of 17 cm -1 , which would arise from a different Fe2O2 mode. Such a feature has thus far not been observed for any other complex with an Fe2(μ-O)2 core, but all the valence-delocalized [Fe III Fe IV (μ-O)2(R3-TPA)2] 3+ complexes show a feature near 410 cm -1 with comparable 18 O-isotope shifts, 41 suggesting that these modes may be related, a notion corroborated by our computational studies (vide infra). Clearly, these features do not fall on the lower frequency line of the correlation shown in Figure 2  4+ (δ = -0.04 mm s -1 and ΔEQ = 2.09 mm s -1 ). 39 The diiron(IV) state in 3 is validated by field-dependent studies performed at 4.2 K with various applied magnetic fields (2.0 T, 4.0 T, 7.0 T) ( Figure S3). The other doublet has δ = 0.48 mm s -1 and ΔEQ = -1.22 mm s -1 , parameters typical of a high-spin diferric species that likely derives from 3 decay. A third component corresponds to a mononuclear high-spin Fe(III) byproduct that is most easily identified in the high-field spectra. After the high-spin ferric component is subtracted out, a cleaner spectrum with only diiron(III) and diiron(IV) components can be obtained ( Figure S4). Analysis of all the spectra from two different samples of 57 Fe-enriched 3 shows that the samples contain ~35% of the diiron(IV) complex, together with 35% of a diferric species and 25% of high-spin ferric components (see Table S1 for more details).
XAS studies also support the assignment of 3 as having an Fe IV 2(μ-O)2 core. The XANES region in Figure S6a shows an increase in the K-edge energy of more than one eV from 7126.3 eV for 2 to 7127.5 eV for 3 ( Figure S6b and Table S2), consistent with an increase in the average iron oxidation state in the latter sample. The K-edge energy value for 3 is not as high as that for [Fe  (Tables 2 and  3). The n value for this Fe scatterer has been constrained to 0.4 to reflect the fraction of 3 in the XAS sample, leading to a very small Debye-Waller factor ( , 39 which reflects the rigidity of the diamond core motif. From the EXAFS fit of 3, an Fe-O-Fe angle of 97° can be calculated. As a further test of our EXAFS analysis approach, the χ(k) data were multiplied by k 5 in order to magnify the contributions of heavier-atom scatterers. 45 These contributions become particularly apparent in the Fourier-transformed data (Figure 4, bottom panel), where features in the second coordination sphere increase in intensity in the order of 2+H + , 2 and 3, with the shortening of the Fe•••Fe distance and the increased rigidity of the diiron unit. Similar UV-vis spectral changes are seen when 2 is treated with Al(OTf)3 as the Lewis acid ( Figure S13). However, the addition of Yb(OTf)3, Y(OTf)3, Zn(OTf)2, Ca(OTf)2, or Ba(OTf)2 does not result in the formation of 3, suggesting that the other Lewis acids are not powerful enough to convert 2 into 3. 46 Figure S14), and 2+H + is formed instead. This effect resembles the effect of treating 2 with HClO4 or HOTf (Figures S10 and S11) and suggests that Sc(OTf)3 hydrolyzes under these conditions to produce protons in solution to give rise to 2+H + .

Scheme 4. Ligand structures and acronyms used in
A mechanism for the conversion of 2 to 3 is proposed in Scheme 5 that accounts for the incorporation of both peroxo oxygen atoms from a unique molecule of 2 into the product 3 ( Figure S8). Intermediate 2 has been shown to have a Fe III 2(μ-O)(μ-1,2-O2) core by an array of spectroscopic techniques. The fact that the peroxo oxygen atoms in 2 are completely retained in the resultant Fe IV 2(μ-O)2 core of 3 requires the water-derived oxo bridge of 2 to be lost prior to the formation of 3, a disposal function that presumably can be assigned to the highly Lewis acidic Sc 3+ (or Al 3+ ) ion. Upon loss of the oxo bridge in 2, both iron coordination spheres become coordinately unsaturated and the peroxo O-atoms remaining on the μ-oxo-depleted 2 are then poised to isomerize from μ-1,2 binding to a μη 2 :η 2 coordination mode that would set the stage for the subsequent O-O bond cleavage step to generate the bis(μoxo)diiron(IV) product 3. Similar conversions have been well established in dicopper model systems since 1996. 60,61 Treatment with Sc 3+ (or Al 3+ ) is found to promote O-O bond cleavage in 2, which is likely to occur homolytically based on the similarity of the activation parameters for the generation of 3 to those of other O-O cleaving reactions (Table 4). Further insight into the mechanism is provided by DFT calculations presented in the next section.

Scheme 5.
Proposed mechanism for the conversion of 2 to 3. The water-derived O-atom shown in blue is lost during the course of the reaction, possibly by binding to Sc 3+ .
Insight into how Sc 3+ may promote the conversion of 2 to 3 has been obtained from DFT calculations at the S12g/TZ2P level of theory. 62,63 For this we need to show the computed structures of 2 and 3, and be certain that these are indeed responsible for the measured (vide supra) spectroscopic fingerprints. The optimized structure of 2 ( Figure 6) shows antiferromagnetic coupling of the high-spin iron(III) atoms, leading to an Fe•••Fe distance of 3.09 Å (in excellent agreement with the EXAFS-derived value of 3.07 Å for 2 ( Table 1) and an Fe-O-Fe angle of 118°. In this structure, the Me3NTB ligands adopt a configuration designated as DU, in which one ligand point downwards with the peroxo moiety below the oxo bridge, and the other upwards. Studies of other configurations, namely down-down (DD) and up-up (UU), as well as ferromagnetically (FM) vs. antiferromagnetically (AFM) coupled iron centers for all three ligand configurations, find the AFM-DU isomer to be the most stable (with DD only slightly less stable by ca. 1 kcal·mol -1 , and UU and ferromagnetically coupled isomers less stable by 10-15 kcal·mol -1 ).  Table 1), both of which exhibit DU ligand conformations. We have thus investigated all the complexes in Table 1 by DFT and found that most of them (B -H) have a preference for the DU conformer. The structure of A is an exception as it cannot be classified as DD or DU due to constraints imposed by the ethylene tether between the two halves of the octadentate ligand. Interestingly, Figure  S17 Fig. S17), with a value of 841 cm -1 . This good correlation between observed and calculated results stems directly from the use of the S12g functional, which is able to give a good description of electronic structures for antiferromagnetically coupled high-spin diiron(III) species, unlike other functionals like BP86-D3. The correlation from Figure S17  the EXAFS analysis. The 3-DD form is less stable than 3-DU by 16.1 kcal/mol, which is consistent with the structures of related crystallographically characterized Fe2O2 diamond-core complexes reported so far. 36,37,40,64 As it is difficult to visualize how 2-DD could easily transform into 3-DU, we have assumed that both 2 and 3 are in fact in the DU form. , remain close to the values experimentally found for 2 and 3, suggesting that these features are intrinsic characteristics of the two diiron cores and not significantly affected by the nature of the supporting Me3NTB ligand. The computed Mössbauer parameters for 3 (δ = -0.046 mm·s -1 , ΔEQ = 2.12 mm·s -1 ) and vibrational frequencies (see Figure 7 and Table S5) are consistent with experiment. The 664 cm -1 peak can be assigned to n3, the fully symmetric breathing mode, while the 531 cm -1 peak can be assigned to n2, the antisymmetric stretching vibration. This latter vibration likely corresponds to the 448cm -1 frequency calculated for the valence-delocalized ferromagnetically coupled Fe(III)/Fe(IV) analogue of m3, which is in agreement with the assignment made by Solomon and co-workers 38 for the ~400-cm -1 feature found in the resonance Raman spectra of [Fe III Fe IV (μ-O)2(R3-TPA)2] 3+ complexes. 41 The normal mode corresponding to n3 consists almost entirely of movements of the oxygen atoms (87%), while those in n2 are reduced to ca. 55%, which rationalizes the larger 18 O-isotope effects found for the 664 cm -1 diamond core vibration.  Table S5 in SI for information on all 6 M2O2 modes.
With confidence that the computed structures are responsible for the experimental spectra, we can return to the question of what effect the Lewis acids exert. Adding Sc(OTf)3 to 2 leads to its coordination to the peroxo moiety, which is favored by ca. 47 kcal·mol -1 over the binding to the oxo bridge (see Figure S18). Scrutiny of this calculated structure shows greater accessibility of the peroxo oxygens than the oxo bridge, which is shielded by the Me3NTB ligands (see Figure 6 bottom and Figure 8). It is thus plausible that the preference of Sc 3+ for peroxo attack derives from steric interactions. Even though the Me3NTB ligands do shield the oxo side quite well, the preference for Sc 3+ binding to the peroxo bridge persists even for the simpler model with ammonia ligands (see Figure S18), where no such steric interactions are present. Indeed Sc(OTf)3 binding to the peroxo unit is favored over the oxo bridge by 12 kcal·mol -1 in [Fe III . Interestingly, the separation of the interaction between Sc(OTf)3 and the diiron species into deformation energy (or strain) and interaction energy (following the Distortion/Interaction-Activation Strain Model 65 ) shows that the intrinsic preference for the peroxo side derives only from deformation. In the process of binding Sc(OTf)3 by m2, the diiron species needs to adjust itself only slightly on the peroxo side (deformation of 6.9 kcal·mol -1 ) versus 18.5 kcal·mol -1 for the binding to the oxo side, which is 2.5 times larger. For both sides, the interaction energy of the Sc(OTf)3 with the m2 diiron species is the same (-58.1 kcal·mol -1 , see Figure S19). In the real systems, these energies are larger: for 2 to bind Sc(OTf)3 the deformation energy (11.3 kcal·mol -1 , see Figure S18) is somewhat larger than that of the model system, but this increase by 4.4 kcal·mol -1 is compensated for largely by an increase of the interaction energy of 4.5 kcal·mol -1 , leading overall to almost similar total binding energies for Sc(OTf)3 (ca. -51 kcal·mol -1 ). counterparts. This question has been explored with 1,4-cyclohexadiene (CHD; BDE = 78 kcal mol -1 ) as substrate. At 233 K CHD addition to a solution of 3 in CH3CN results in the first order decay of its 600-nm chromophore and a second order rate constant (k2) of 7 * 10 -4 M -1 s -1 ( Figure S16). This value is comparable to the k2 value of 10 -4 M -1 s -1 reported for the oxidation of 9,10-dihydroanthracene by [Fe IV 2(μ-O)2(TPA*)2] 4+ at 193 K, 66 after extrapolation to 233 K by assuming that rates double with every 10 K increase in temperature to give a value of 1.6 x 10 -3 , making the k2 for 3 at 233 K only a factor of 2 smaller than that for [Fe IV 2(μ-O)2(TPA*)2] 4+ .
In contrast, the corresponding mononuclear complex [Fe IV (O)(Me3NTB)] 2+ is one of the most reactive nonheme Fe IV (O) complexes described to date, with k2 = 9.4 * 10 2 M -1 s -1 for 1,4-CHD oxidation in CH3CN at 233 K, 28 nearly 10 7fold more reactive towards 1,4-CHD than 3. This result suggests that a terminal oxo is more reactive towards C-H bonds than a bridging oxo if other variables such as oxidation states and spin states of the iron center are kept constant. 66 For comparison, the mononuclear [Fe IV (O)(TPA*)] 2+ is 1000-fold more reactive than the corresponding dinuclear [Fe IV 2(μ-O)2(TPA*)2] 4+ complex. 66 These complexes represent the only two pairs of iron(IV)oxo complexes supported by the same ligand framework but differ in having terminal or bridging oxo units. The 10 4fold greater reactivity difference found for the Me3NTB pair of complexes is quite amazing, and a phenomenon that deserves further scrutiny.
Lastly, it needs to be emphasized here that 3 and [Fe 4+ and S = 2 for the much more reactive Q. [66][67][68] There are also differences in the diiron(IV) core structures deduced from resonance Raman and X-ray absorption spectroscopy that have yet to be resolved. 19,44 Further scrutiny of these structure-reactivity correlations is desirable.

Summary and Perspectives
In summary, 3 represents the first diiron(IV) complex to be generated by Lewis acid-assisted O−O bond cleavage of a peroxodiiron(III) complex that is derived from the reaction of O2 with a diiron(II) precursor (black/red pathway in Scheme 6). This transformation is closely related to that of the proton-assisted conversion of sMMO-P to sMMO-Q. 17,18 The actual structure of the diiron(IV) core for Q is currently not settled, where resonance Raman data support a closed-core structure 19 but recent XAS data 44 favor an open-core structure.
Nevertheless, our study is unique in that it involves the conversion of a peroxodiiron(III) complex to a diiron(IV) complex analogous to the diiron center in the sMMO enzyme. In prior related biomimetic examples in the nonheme literature, (µ-1,2-peroxo)diiron(III) intermediates have been trapped and found to convert into higher-valent diiron derivatives upon treatment with Bronsted acids (along the black path in Scheme 6). Specifically, [Fe III 2(µ-O)(µ-1,2-O2)(L)2] 2+ complexes of 6-Me3TPA (tris(6-methyl-2-pyridylmethyl)amine) 34 and BnBQA (N-benzyl-N,Nbis(2-quinolinylmethyl)amine) 30 generated from the reactions of diiron(II) precursors with O2 react with strong acid to generate nearly isotropic EPR signals at g = 2. In the case of the BnBQA complex, 57 Fe hyperfine splitting of the g = 2 signal is observed when the complex is 57 Fe-enriched, showing that the unpaired electron is associated with a (μoxo)diiron(III,IV) species. 30 In another example (along the pink path in Scheme 6), the reaction of stoichiometric In contrast, the formation of high-valent 3 from 2 (along the red path in Scheme 6) occurs by introducing Sc 3+ (and not a proton), representing the first instance of a Lewis acid-mediated O-O bond cleavage in a diiron system. In fact, adding a Bronsted acid like HClO4 or HOTf to 2 results in the protonation of its oxo bridge to form 2+H + . These contrasting observations, with support from parallel theoretical calculations, show that Sc 3+ preferentially attacks the peroxide bridge over the oxo bridge to promote O-O bond cleavage to generate 3. However, the unique role of Sc 3+ in this chemistry is nullified by the addition of excess water, which results in its hydrolysis to produce a proton that instead leads to the generation of 2+H + . These results contrast the recent report that adding 1 eq of either Sc(OTf)3 or HClO4 to [Fe III (β-BPMCN)(OOH)] 2+ (where β-BPMCN = cis-β-bis(pyridyl-2-methyl)-cis-1,2diaminocyclohexane) generates an oxidant capable of hydroxylating cyclohexane within seconds at -40 °C. 70 Browne and co-workers have also provided strong evidence that the Sc(OTf)3-enhanced olefin epoxidation by the combination of [Mn IV 2(µ-O)3(tmtacn)2] 2+ (where tmtacn = 1,4,7-trimethyl-1,4,7-triazacyclononane) and H2O2 results from Bronsted acid formation upon hydrolysis of Sc(OTf)3 by water present in the reaction mixture. 71 In contrast to these two examples, our system is unique in that the added acid (Bronsted or Lewis) has a choice between attacking the oxo bridge versus the peroxide bridge in a diiron model framework, resulting in significantly different outcomes from Bronsted and Lewis acids. Most importantly, our investigations into the conversion of 2 to 3 establish that both O atoms of one O2 end up becoming incorporated into the high-valent diiron intermediate 3 (see, e. g., Scheme 1), thereby reproducing a key feature in the activation of O2 by the diiron center of sMMO. Our investigation thus underscores the complexities of the chemistry involved in activating the peroxo O-O bond to generate high-valent oxidants in nonheme iron catalysts and highlights Nature's ability to deliver a key proton to a particular site in order to elicit a desired transformation.

Experimental Section
Materials and Physical Methods. Commercially available chemicals such as DBU (1,.0]undec-7-ene), Scandium triflate, Aluminium triflate, Yttrium triflate, perchloric acid, and solvents were used without further purification unless noted. [Fe II (Me3NTB)(CH3CN)](CF3SO3)2 (1) was synthesized as previously reported. 28 UV-vis absorption spectra were recorded with a HP 8453A diode array spectrophotometer equipped with a cryostat from UNISOKU Scientific Instruments, Japan. All UV-vis absorption experiments were carried out in 1-cm path length cuvettes. Resonance Raman spectra were obtained at -40 °C with excitation at 561 nm (50 mW at source, Cobolt Lasers) or 660 nm (100 mW at source, Cobolt Lasers) through the sample in a flat bottom NMR tube using a 90° scattering arrangement (parallel to the slit direction). Resonance Raman spectra on frozen samples (at 77 K) were obtained using a 135° back scattering arrangement. The collimated Raman scattering was collected using two Plano convex lenses (f = 12 cm, placed at an appropriate distance) through appropriate long pass edge filters (Semrock) into an Acton AM-506M3 monochromator equipped with a Princeton Instruments ACTON PyLON LN/CCD-1340x400 detector. The detector was cooled to -120 °C prior to the experiments. Spectral calibration was performed using the Raman spectrum of acetonitrile/toluene 50:50 (v:v). 72 Each spectrum was accumulated, typically 60 times with 5 s acquisition time, resulting in a total acquisition time of 5 min per spectrum. The collected data was processed using Spekwin32, 73 and a multi-point baseline correction was performed for all spectra. Iron K-edge X-ray absorption spectra for 2 and 3 were collected on SSRL beamline 9−3 using a 100-element solidstate Ge detector (Canberra) with a SPEAR storage ring current of ∼500 mA at a power of 3.0 GeV. The incoming X-rays were unfocused using a Si(220) double crystal monochromator, which was detuned to 70% of the maximal flux to attenuate harmonic X-rays. Between 6 and 8 scans of the fluorescence excitation spectra for each sample were collected from 6882 to 8000 eV at a temperature (10 K) that was controlled by an Oxford Instruments CF1208 continuous-flow liquid helium cryostat. An iron foil was placed in the beam pathway prior to the ionization chamber I0 and scanned concomitantly for an energy calibration, with the first inflection point of the edge assigned to 7112.0 eV. A 3 μm, 6 μm or 9 μm Mn filter and a Soller slit were used to increase the signal-to-noise ratio of the spectra. Photoreduction was monitored by scanning the same spot on the sample twice and comparing the first derivative peaks associated with the edge energy during collection, but none was observed in the present study. The detector channels from the scans were examined, calibrated, averaged, and processed for EXAFS analysis using EXAFSPAK 74 to extract χ(k). Theoretical phase and amplitude parameters for a given absorber−scatterer pair were calculated using FEFF 8.40 75 and were utilized by the "opt" program of the EXAFSPAK package during curve fitting. In all analyses, the coordination number of a given shell was a fixed parameter and was varied iteratively in integer steps, while the bond lengths (R) and mean-square deviation (σ 2 ) were allowed to freely float. The amplitude reduction factor S0 was fixed at 0.9, while the edge-shift parameter E0 was allowed to float as a single value for all shells. Thus, in any given fit, the number of floating parameters was typically equal to (2 × num shells) + 1. The k range of the data is 2−14.5 Å −1 . The pre-edge analysis was performed on data normalized in the "process" program of the EXAFSPAK package, and pre-edge features were fit between 7108 and 7118 eV for all samples using the Fityk 76 program with pseudo-Voigt functions composed of 50:50 Gaussian/Lorentzian functions. Mössbauer spectra were recorded with two spectrometers using Janis Research (Wilmington, MA) SuperVaritemp dewars that allow studies in applied magnetic fields up to 8 T in the temperature range from 1.5 to 200 K. Mössbauer spectral simulations were performed using the WMOSS software package (SEE Co, Edina, MN) ), all figures were generated by SpinCount. 77 Sample preparation procedures. A 4 mM solution of 1 in CH3CN or CD3CN was prepared in a 1-cm cuvette in a nitrogen containing glovebox and 1.5 equivalents of DBU/H2O is added to generate the putative diron(II) species [1+DBU] at -40 °C. At this stage, a balloon containing moisture-free oxygen gas is used to purge the headspace of the cuvette to carry out oxygen activation to 1produce a deep-green solution of [Fe2 III (µ-O)(µ-O2)(Me3NTB)] 3+ (2). This process was monitored by UV-vis absorption spectroscopy. When the yield of 2 was maximized, pre-cooled pipettes were used to transfer the solution to a NMR tube and then frozen at 77K using liquid nitrogen for resonance raman studies. Similarly, for Mössbauer studies, a 2 mM solution of 57 Fe enriched 1 in CH3CN was prepared in a nitrogen containing glovebox and to it 1.5 equivalents of DBU was added and then oxygenated to generate 2. At this point, the solution was transferred to a Mossbauer cup and frozen at 77K using liquid nitrogen. For XAS studies, a 4 mM solution of 1 in CH3CN was prepared in a nitrogen containing glovebox and was generated in a similar fashion, except the solution was frozen in an XAS cup.
Computational details. All DFT calculations were performed with the ADF program 78 (version 2017) with the S12g/TZ2P setup 62 including COSMO solvation and ZORA scalar relativistic corrections self-consistently. Full details can be found in the Supporting Information.

Supporting Information
The Supporting Information is available free of charge on the ACS Publications website at DOI: XXXXXXXX Additional data, including Figures S1-S21 and Tables S1-S7 (PDF) National Accelerator Laboratory, is supported by the U.S. Department of Energy, Office of Science, Office of Basic Energy Sciences under Contract No. DE-AC02-76SF00515. The SSRL Structural Molecular Biology Program is supported by the DOE Office of Biological and Environmental Research, and by the National Institutes of Health, National Institute of General Medical Sciences (P41GM103393). The contents of this publication are solely the responsibility of the authors and do not necessarily represent the official views of NIGMS or NIH. We acknowledge several insightful discussions with Dr. Andrew Jasniewski that helped in developing the study presented here.