Sc³⁺-triggered oxoiron(IV) formation from O₂ and its nonheme iron(II) precursor via a Sc³⁺–peroxo-Fe³⁺ intermediate

Abstract

We report that redox-inactive Sc³⁺ can trigger O2 activation by the Fe^II(TMC) center (TMC = tetramethylcyclam) to generate the corresponding oxoiron(IV) complex in the presence of BPh_4⁻ as an electron donor. To model a possible intermediate in the above reaction, we generated an unprecedented Sc³⁺-adduct of [Fe^III(η²-O₂)(TMC)]⁺, which was characterized to have an Fe^III-(μ-η²:η²-peroxo)-Sc³⁺ core and found to convert to the oxoiron(IV) complex. These results have important implications for the role a Lewis acid can play in facilitating O–O bond cleavage during the course of O₂ activation at nonheme iron centers.

There is much current interest in investigating the ability of redox-inactive metal ions to modulate redox reactions by virtue of their Lewis acidity, particularly with respect to their possible roles in O₂ evolution¹ and activation.^2,3 For example, the oxygen-evolving complex of Photosystem II requires a redox-inactive Ca²⁺ ion to produce O₂.¹ In addition, redox-inactive ions have been found to affect the stability and reactivities of high-valent metal-oxo complexes in biomimetic systems² as well as to accelerate O₂ activation by Fe^II and Mn^II complexes.³ For the latter, heterobimetallic O₂ adducts and high-valent metal-oxo species are presumably involved but have not been observed. We previously demonstrated that [Fe^II(TMC) (NCCH₃)]²⁺ (1) (TMC = 1,4,8,11-tetramethylcyclam) reacts with O₂ in CH₃CN in the presence of stoichiometric H⁺ and BPh_4⁻ to form [Fe^IVO(TMC)(NCCH₃)]²⁺ (4).⁴ Herein, we report that a redox-inactive Sc³⁺ ion can replace the strong acid in this reaction to trigger the formation of 4. An unprecedented Sc³⁺ adduct (3) of [Fe^III(η²-O₂)(TMC)]⁺ (2) was trapped, spectroscopically characterized in detail, and found to convert to 4 (Scheme 1).

Scheme 1.

Proposed mechanism for the formation of 4 from 1 and O₂

Complex 1 is air stable in acetonitrile solution for days. However, the addition of 1 equiv. Sc(OTf)₃ together with 1 equiv. NaBPh₄ to an aerobic solution of 1 resulted in the formation of 4 in >70% yield over the course of ~1 h at 0 °C, as indicated by its signature near-IR band at 820 nm (Figure 1A).⁵ Electrospray ionization mass spectra of the solution revealed the evolution of a prominent peak at m/Z = 477.0, assigned to the {[Fe^IV(O)(TMC)](OTf)}⁺ ion based on its position and isotope distribution pattern (Figure S1). When the reaction was carried out with ¹⁸O₂, the m/Z = 477 peak showed an upshift of 2 units (Figure S2), confirming that the oxo moiety of 4 derived from dioxygen and that O–O bond cleavage must occur for the formation of 4 from 1 and O₂.

Figure 1.

Reaction of 0.96 mM 1 with NaBPh4 and Sc(OTf)3 in aerobic CH3CN at 0 °C. (A) UV-visible spectral changes observed with 1 equiv. NaBPh4 and 1 equiv. Sc(OTf)3. Inset: TMC ligand. (B) Plot of the yield of 4 vs. equiv. BPh_4⁻ in the presence of 1 equiv. Sc³⁺. Inset: plot of the yield of 4 vs. equiv. of Sc³⁺ with 1 equiv. BPh_4⁻.

Further investigation demonstrated the requirement for both Sc³⁺ and BPh_4⁻ for the formation of 4 from 1, as addition of either BPh_4⁻ or Sc³⁺ alone to 1 in air-saturated CH₃CN solution did not elicit any detectable change in the UV-visible spectra. In addition, the yield of 4 correlated linearly with the amount of BPh_4⁻ added, plateauing at 1.0 equiv. BPh_4⁻ (Figure 1B). ¹H-NMR studies of the final solution showed that BPh_4⁻ had decomposed to 1,1′-biphenyl (Figure S3) with a stoichiometry of 0.95±0.15 equiv. relative to 1, demonstrating that BPh_4⁻ provides the two electrons needed to convert 1 and O₂ into 4. On the other hand, a sub-stoichiometric amount of Sc³⁺ was sufficient for the maximal formation of 4 (Figure 1B inset), suggesting that Sc³⁺ can act somewhat ‘catalytically’.

As shown in Figure 1A, no intermediates are evident in the UV-Vis spectra during the conversion of 1 to 4.⁶ To account for the role of Sc³⁺ in this transformation, we propose the formation of a Sc³⁺-peroxo-Fe³⁺ adduct that is reminiscent of the Fe^III–OOH species proposed in the H⁺-and-BPh_4⁻-promoted generation of 4 from O₂ and 1.^4,7 To test this hypothesis, Sc(OTf)₃ was added to a solution of the blue Fe^III(η-O₂) complex 2 (purified via precipitation as its BPh₄ salt; see SI for details), which resulted in the immediate generation of a magenta intermediate (3) and its subsequent conversion to 4 in ~70% yield over the course of ~1 h at −10 °C (Figure 2A).

Figure 2.

(A) UV-visible spectral changes upon addition of 3 equiv. Sc³⁺ to 1.5 mM purified 2 (ε835 = 650 M⁻¹cm⁻¹) in CH₃CN at −10 °C instantly generating 3 (ε520 = 780 M⁻¹cm⁻¹), which in turn decayed to 4. (B) UV-visible changes upon titration of 1.5 mM 2 in CH3CN at −40 °C with Sc³⁺ (0, 0.5, 1.0, 1.5, 2.0, 9.0 equiv, respectively).

What is the identity of 3? Complex 3 exhibits a λ_max of 520 nm (ε₅₂₀ = 780 M⁻¹cm⁻¹) established from its UV-visible spectrum (Figure 2A) and Mössbauer analysis. The large blue shift observed for the peroxo-to-Fe(III) charge transfer band of 2 (λ_max 835 nm) is reminisent of that seen upon protonation of 2 to form [Fe^III(TMC)(η¹-OOH)]²⁺ in CH₃CN (5),^7a indicating partial neutralization of the negative charge of the peroxo ligand. Titration of 2 with Sc(OTf)₃ showed that 1 equiv Sc(OTf)₃ was nearly sufficient to cause the 835-nm band of 2 to disappear, suggesting a 1:1 stoichiometry for the Sc³⁺-adduct of 2 (Figure 2B). The EPR spectrum of 3 shows features at g = 9.1, 5.1, 3.6 and ~2, consistent with an S = 5/2 iron(III) center with an E/D of 0.18 (Figure 3 left), compared to E/D values of 0.28 and 0.097 for 2 and 5,^7a respectively. The Mössbauer spectra of 3 (Figure 3 right) are typical of high-spin iron(III); their analysis is described in the SI and Mössbauer parameters are listed in Table 1 and Figure 3 caption. A comparison of the spectroscopic properties in Table 1 shows that 3 is quite different from 2 and 5, indicating that Sc³⁺ significantly affects the properties of the peroxoiron(III) unit.

Figure 3.

Left panel: EPR spectra of 2 (blue line, top)^7a and 3 (red line, bottom) at 2 K and 0.2 mW microwave power. Right panel: 4.2 K Mössbauer spectra of 3 in MeCN recorded in parallel applied fields of 0.5 T (A) and 8.0 T (B). The red lines in (A) and (B) are theoretical curves based on eq 1 of the SI, using the following parameters: D = +1.3 cm⁻¹, E/D = 0.18, g0 = 2.00, Ax/gnβn = −20.0T, Ax/gnβn = −20.6 T Ax/gnβn = −19.9 T , ΔEQ = 0.50 mm/s, η = −0.5, δ = 0.47 mm/s. The Mössbauer sample contained 90% 3⁸ and 10% Fe^IV=O species (blue line).

Table 1.

Spectroscopic comparison of Fe^III(TMC)-peroxo complexes (S = 5/2) in CH₃CN

	λmax, nm	ΔEQ, mm/s	δ, mm/s	D, cm−1	E/D	Pre-edge area	ref
2	835	−0.92	0.58	−0.91	0.28	17.9	7a
3	520	0.50	0.47	1.3	0.18	14.4	*
5	500	0.20	0.51	2.5	0.097	22.4	7a

^*this work.

We also carried out Fe K-edge X-ray absorption spectroscopic (XAS) studies to investigate the structural features of 3. Complex 3 exhibits an Fe K-edge at 7125.3 eV and a pre-edge feature at 7113.3 eV, which are comparable to those of 2 and 5 obtained in CH₃CN solvent (Figure S4, Table S1).^7a The pre-edge feature of 3 has an area of 14.4(6) units, compared to 17.9 for 2 and 22.4 for 5 (Table S1). As the pre-edge area reflects the extent to which the iron center deviates from centrosymmetry, the coordination environment of 3 must be closer to that of 2 with an η²-peroxo ligand than that of 5 with an η¹-OOH ligand.

Analysis of the EXAFS data of 3 provides additional structural insight. Best fits reveal 4 N scatterers at 2.18 Å and 4 C scatterers each at 3.00 and 3.15 Å; all these features arise from the TMC ligand and have distances close to those found for 2 (Figure 4, Table S2). In addition, there is an O subshell at 1.98(1) Å arising from the peroxo ligand. Notably, the Fe–O distance of 3 is significantly longer than the 1.91-Å distance found for 2,^7a implying the addition of Sc³⁺ significantly weakens the iron-peroxo interaction. This 0.07-Å lengthening is inconsistent with conversion of the η²-peroxo ligand to an η¹-isomer, as related η¹-peroxo complexes 5 and 6 both have shorter Fe–O distances (Table 2). Cu^II adducts to (η²-peroxo)heme complexes also have one short Fe–O bond of ~1.93 Å in a highly unsymmetric η²-peroxo ligand that binds to the iron.⁹ Thus, the 0.07-Å lengthening of the r(Fe–O) of 3 relative to that of 2 favors a symmetric η²-peroxo binding mode for 3. This conclusion is also supported by a comparison of fits 7 and 8 in Table S2, where the 2-O subshell in fit 7 has a σ² value of ~4, while the 1-O subshell in fit 8 has a σ² value of −0.4. A negative σ² value for the latter indicates that either a bond is more rigid than would be expected for its distance or that there are too few scatterers associated with that shell.¹⁰ A negative σ² value was also found when only one O-scatterer (instead of two) was used in fitting the EXAFS data for 2. Our EXAFS results thus demonstrate that the binding of Sc³⁺ retains the symmetric side-on binding mode of the peroxo ligand in 3 but elongates the r(Fe–O) by 0.07 Å.¹¹

Figure 4.

Fourier transform of Fe K-edge EXAFS data for 3 over a k-range of 2-14 Å⁻¹, with k³χ(k) vs k data shown in the inset. The solid black lines represent the experimental data, while the red dashed lines correspond to the best fit with 2 O @ 1.98 Å and 4 N @ 2.18 Å (fit #22 in Table S3).

Table 2.

Comparison of Structural and Raman data for S = 5/2 Fe^III-peroxo complexes.

Complexes	r(Fe–N) (Å)	r(Fe–O) (Å)	ν(O–O) (cm−1)	Ref.
3 #	2.18	1.98, 1.98	807	*
nonheme FeIII-η2-peroxo			816–827	7, 15
2 (2′)#	2.20 (2.21)	1.91, 1.91 (1.91, 1.91)	826 (825)	7a (7b)
nonheme FeIII-η1-peroxo			830–891	7, 16&
5 (5′)#	2.15 (2.16)	1.92 (1.85)	870 (868)	7a (7b)
6$	2.17	1.89		17
(heme)FeIII- (μ-η2:η1-O2)CuII	2.09	1.92, 2.09	788–808	9a, 9b
(heme)FeIII- (μ-η2:η2-O2)CuII	2.09	1.94, 2.09	747–767	9a, 9b

^#2, 3, and 5 in CH₃CN; 2′ and 5′ in a 3:1 (v:v) mixture of acetone:CF₃CH₂OH.

^*This work. & See also Table S4 of ref 7a; $ 6 = [Fe^III(TMCS)(η¹-O₂)].

The final key piece of evidence for the identity of 3 comes from resonance Raman spectroscopy. Laser excitation into the intense 520-nm band of 3 reveals two prominent peaks at 807 and 543 cm⁻¹ (Figure 5) that correspond to ν(O–O) and ν(Fe–O) modes, respectively. These assignments are corroborated by ¹⁸O-labeling, resulting in respective downshifts of 45 and 23 cm⁻¹ that correlate well with Hooke’s Law predictions for these modes and support the presence of an iron-bound peroxo ligand in 3.¹² The ν(O–O) of 3 is the lowest of any nonheme high-spin peroxoiron(III) complex thus far observed (Table 2). Relative to its precursor 2,^7a 3 has a ν(O–O) that is downshifted by 19 cm⁻¹ and a ν(Fe–O) that is upshifted by 50 cm⁻¹, consistent with the retention of the η² binding mode of the peroxo ligand. Taken together, the spectrosopic data lead us to propose a Fe³⁺(μ-η²:η²-O )Sc³⁺ core for 3, analogous to the Ni²⁺(μ-η²:η²-O₂)K⁺ core found in a complex characterized crystallographically by Limberg, Driess, and coworkers.^13,14 With the nature of 3 characterized, an important question that remains is whether it is involved in the conversion of 1 to 4 by O₂ activation. The requirement for both Sc³⁺ and two electrons to trigger O₂ activation of 1 suggests the likely formation of a Sc³⁺-peroxo-Fe³⁺ species like 3 as an intermediate (Scheme 1). However the fact that this species does not accumulate during O₂ activation (Figure 1A) suggests that 3 may correspond to a more stable isomer of the actual intermediate involved in the O₂ activation reaction. Nevertheless, 3 represents a rare example of a heterobimetallic complex bridged by a peroxo ligand^9,13 and the only one thus far involving a nonheme iron center.

Figure 5.

Resonance Raman spectra of 3 prepared in CH₃CN with H₂¹⁶O₂ (red, top) and H₂¹⁸O₂ (black, middle) obtained with 514.5 nm excitation, 100 mW. The ¹⁶O – ¹⁸O difference spectrum is shown in blue (bottom). S = solvent.

The spectroscopic characterization of 3 as a complex with an Fe³⁺(μ-η²:η²-O₂)Sc³⁺ core provides a plausible mechanism for a Lewis acid to promote O–O bond cleavage. This insight points to another role the second iron center can play in diiron enzymes besides serving as an electron source: functioning as a Lewis acid to facilitate formation of high-valent iron-oxo intermediates such as Q and X in the respective oxygen activating cycles of methane monooxygenase and Class 1A ribonucleotide reductases.¹⁸ This report of the Sc³⁺-peroxo-Fe³⁺ intermediate (3) also augments the recent literature focused on the effects of redox-inactive Lewis acidic metal ions on redox transformations.^1,2,3 Prominent among these are their accelerative properties in oxidations by high-valent metal-oxo complexes discovered by Fukuzumi and Nam^2a-f as well as the role of Ca²⁺ in forming an O–O bond from water during photosynthesis.¹ Relevant to the latter, Borovik recently showed that group II metal ions (M^II) can enhance the rates of O₂ activation by Fe^II and Mn^II complexes to afford well characterized M^II–(μ-OH)–(Mn^III/Fe^III) products, presumably via heterobimetallic O₂ adducts.³ Our results herein demonstrate that Sc³⁺ can turn “on” the activation of O₂ at a nonheme iron center and that a transient Sc³⁺–peroxo–Fe³⁺ species related to 3 could be a viable intermediate leading to O–O bond cleavage.

ABBREVIATIONS

EXAFS

extended X-ray absorption fine structure

TMC

1,4,8,11-tetramethylcyclam

TMCS

1-(2-mercaptoethyl)-4,8,11-trimethyl-1,4,8,11-tetraazacyclotetradecane

XAS

X-ray absorption spectroscopy.