Modeling the Syn-Disposition of Nitrogen Donors in Non-Heme Diiron Enzymes. Synthesis, Characterization, and Hydrogen Peroxide Reactivity of Diiron(III) Complexes with the Syn N-Donor Ligand H₂BPG₂DEV

Abstract

In order to model the syn disposition of histidine residues in carboxylate-bridged non-heme diiron enzymes, we prepared a new dinucleating ligand, H₂BPG₂DEV, that provides this geometric feature. The ligand incorporates biologically relevant carboxylate functionalities, which have not been explored as extensively as nitrogen-only analogs. Three novel oxo-bridged diiron(III) complexes [Fe₂(μ-O)(H₂O)₂-(BPG₂DEV)](ClO₄)₂ (6), [Fe₂(μ-O)(μ-O CAr^iPrO)(BPG₂DEV)](ClO₄) (7), and [Fe₂(μ-O)(μ-CO₃)(BPG₂DEV)] (8) were prepared. Single crystal X-ray structural characterization confirms that two pyridines are bound syn with respect to the Fe–Fe vector in these compounds. The carbonato-bridged complex 8 forms quantitatively from 6 in a rapid reaction with gaseous CO₂ in organic solvents. A common maroon-colored intermediate (λ_max = 490 nm; ε = 1500 M₋₁ cm⁻¹) forms in reactions of 6, 7, or 8 with H₂O₂ and NEt₃ in CH₃CN/H₂O solutions. Mass spectrometric analyses of this species, formed using ¹⁸O-labeled H₂O₂, indicate the presence of a peroxide ligand bound to the oxo-bridged diiron(III) center. The Mössbauer spectrum at 90 K of the EPR-silent intermediate exhibits a quadrupole doublet with $\underset{.}{\delta }$ = 0.58 mm/s and ΔE_Q = 0.58 mm/s. The isomer shift is typical for a peroxodiiron(III) species, but the quadrupole splitting parameter is unusually small compared to related complexes. These Mössbauer parameters are comparable to those observed for a peroxo intermediate formed in the reaction of reduced toluene/o-xylene monooxygenase hydroxylase (ToMOH) with dioxygen. Resonance Raman studies reveal an unusually low-energy O–O stretching mode in the peroxo intermediate that is consistent with a short diiron distance. Although peroxodiiron(III) intermediates generated from 6, 7, and 8 are poor O-atom transfer catalysts, they display highly efficient catalase activity, with turnover numbers up to 10,000. In contrast to hydrogen peroxide reactions of diiron(III) complexes that lack a dinucleating ligand, the intermediates generated here could be reformed in significant quantities after a second addition of H₂O₂, as observed spectroscopically and by mass spectrometry.

Introduction

Carboxylate-bridged non-heme diiron centers occur in a variety of enzymes that activate dioxygen to catalyze key reactions in nature.^1–5 Members of this class include ribonucleotide reductase (RNR-R2),^6–8 Δ⁹-desaturase (Δ⁹D),^9,10 and the hydroxylase components of bacterial multicomponent monooxygenases (BMMs).^11,12 Soluble methane monooxygenase (sMMOH),⁴ toluene/o-xylene monooxygenase (ToMOH),^13,14 and phenol hydroxylase (PHH)¹⁵ belong to the family of BMMs and function as catalysts for the selective oxidation of hydrocarbons. Remarkably, despite their diverse roles in biology, the active sites in these enzymes all share common structural features, a carboxylate-rich environment with two histidine donors that are bound in a syn fashion with respect to the diiron vector.¹⁶ Representations of the active sites of some of these enzymes in their reduced diiron(II) forms are provided in Chart 1.

Chart 1.

In these enzymes, dioxygen activation occurs following 2-electron reduction of their diiron(III) resting states to generate an O₂-reactive diiron(II) species. Following reaction of this reduced form with dioxygen, peroxodiiron(III) intermediate species are generated that share common spectroscopic features.⁴ Resonance Raman spectral studies of peroxo intermediates in RNR-R2¹⁷ and Δ⁹D¹⁰ suggest a μ-1,2-peroxo binding mode. Theoretical analysis of the peroxo intermediate in sMMOH, for which no vibrational spectroscopic data are yet available, find a μ-η²:η²-peroxo butterfly structure to be more stable,¹⁸ and pH-dependent studies of the dioxygen activation chemistry indicate that proton transfer reactions,¹⁹ most likely involving some kind of hydroperoxo intermediate, are involved. Recently, novel peroxodiiron(III) transient intermediates were observed in the ToMOH and PHH systems, having no visible or near-IR optical bands, notably different Mössbauer parameters, and possibly a different coordination mode.^12,20 Peroxodiiron(III) intermediates in these enzymes can in some cases undergo O–O bond cleavage to form high-valent species, such as the mixed-valent (μ-oxo)diiron(III,IV) intermediate X in RNR-R2⁶ and the methane-oxidizing di(μ-oxo)diiron(IV) intermediate Q in sMMO.^21,22 Elucidating the structures of these oxygenated intermediates remains an important challenge.

Significant effort has been expended to construct model complexes that mimic the active site structures and functions of non-heme diiron enzymes.^23,24 These studies have predominantly been conducted to probe the mechanism of O₂ activation, the influence of structure on reactivity, and the requirements of iron-based catalysts for hydrocarbon oxidation. Two general ligand motifs have been extensively employed in the construction of these model complexes, namely, tris(2-pyridylmethyl)amine (TPA)-based and sterically encumbering 2,6-diarylbenzoate ligands (Chart 2).^23–26 Whereas the latter group of ligands facilitates formation of diiron(II) complexes with the same composition as the enzyme active sites, TPA-based constructs afford complexes that react with dioxygen to generate species more closely resembling intermediates observed spectroscopically in the reactions of BMMs with dioxygen. A structural feature that neither of these ligand motifs can rigidly enforce, however, is the syn orientation of nitrogen donors with respect to the diiron vector. This feature may be important and it is likely that nature did not choose such a stereochemistry arbitrarily. The significance of the syn N-donor disposition of imidazoles from histidine residues with respect to the diiron vector is still unclear, although preliminary DFT calculations on intermediate Q of sMMOH suggest that a stereoelectronic effect derived from this configuration tunes its reactivity.²⁷ Apart from their inability to enforce syn N-donor character, terphenylcarboxylate- and TPA-based ligands are not pre-organized to stabilize a diiron core, which can lead to dissociation or aggregation to form monomeric or oligomeric species, respectively. To address these issues, we have been exploring N-donor ligands that are covalently tethered by a diethynylbenzene unit.^16,28–31 Selection of this moiety as the appropriate linker fixes the nitrogen donor atoms at a distance and orientation similar to that in the enzymes. Of particular interest is the 1,2-bis(pyridin-3-ylethynyl)benzene scaffold, because of convenience in functionalizing the pyridine moiety, as demonstrated in the synthesis of several ligands based on this platform.²⁸

Chart 2.

In the present article we report the synthesis of the syn N-donor ligand H₂BPG₂DEV (Chart 2) and three of its derivatives containing oxo-bridged diiron(III) cores. Two tripodal N,N'-bis(2-pyridylmethyl)-3-aminoacetate (BPG⁻) centers bridged by diethynylveratrole converge to incorporate diiron units and provide carboxylate groups that more accurately represent the aspartate and glutamate residues of non-heme diiron enzymes than earlier constructs. We describe how this dinucleating ligand stabilizes the oxo-bridged diiron(III) cores and considerably influences the chemical and physical properties of a peroxodiiron(III) intermediate generated by addition of hydrogen peroxide.

Experimental Section

Materials and Methods

Reagents were purchased from commercial sources and used as received. Acetonitrile (CH₃CN), dichloromethane (CH₂Cl₂), and tetrahydrofuran (THF) were saturated with nitrogen and purified by passing though activated alumina columns under argon. Triethylamine (Et₃N) was distilled from CaH₂. The compounds (5-bromo-pyridin-2-yl)methanol (2b),³² [(pyridin-2-ylmethyl)amino]acetic acid ethyl ester (3),³³ 4,5-diethynylveratrole (DEV),³¹ and (Et₄N)₂[Fe₂(μ-O)Cl₆]^34,35 were prepared using methods described in the literature. The compound 2,6-diisopropoxybenzoic acid (HO₂CAr^iPrO) was synthesized by using a modified literature procedure.³⁶ Caution! The perchlorate salts used in this study are potentially explosive and should be handled with care!

NMR spectra were recorded on a Varian 300 spectrometer in the Massachusetts Institute of Technology Department of Chemistry Instrument Facility (MIT DCIF). All spectra were recorded at ambient probe temperature, 293 K. IR spectra were taken on a Thermo Nicolet Avatar 360 spectrometer with OMNIC software. Mass spectra were recorded in electrospray ionization mode. ESI-MS data were obtained with an Agilent 1100 series LC/MSD mass spectrometer. UV-vis experiments were performed on a Cary 50 spectrophotometer.

[(5-Bromo-pyridin-2-ylmethyl)-pyridin-2-ylmethyl-amino]acetic Acid Ethyl Ester (4)

A solution of (5-bromo-pyridin-2-yl)methanol (2b) (4.69 g, 24.9 mmol) and Et₃N (4.53 mL, 32.4 mmol) in THF (95 mL) was cooled to 0 °C and treated dropwise with methanesulfonyl chloride (MsCl; 2.32 mL, 29.9 mmol). The mixture was warmed to room temperature, stirred for 2 h, and then combined with aqueous NH₄Cl (200 mL). The aqueous phase was extracted with CH₂Cl₂ (3 × 200 mL) and the organic layers were dried (Na₂SO₄), filtered, and concentrated to afford a solid (6.78 g, quant) that was used without further purification. This purple solid, K₂CO₃ (4.12 mg, 29.9 mmol), and [(pyridin-2-ylmethyl)-amino]-acetic acid ethyl ester (3) (5.80 g, 29.9 mmol) were stirred overnight in CH₃CN (100 mL). During this period the color changed from purple-red to orange. The reaction mixture was combined with CH₂Cl₂ (300 mL) and the organics were washed with aq Na₂CO₃ (3 × 100 mL). The organic layer was dried (Na₂SO₄), filtered, and evaporated to dryness. The crude product was purified by column chromatography (alumina; EtOAc/hexanes, 1:3) to give 4 as a yellow oil, which was identical to the compound synthesized previously as judged by ¹H NMR spectroscopy and ESI-MS.²⁸ Yield: 7.92 g (87%). ¹H NMR (300 MHz, CDCl₃): δ = 8.56–8.50 (m, 2H), 7.67–7.61 (dt, J = 1.8, 7.5 Hz, 1H), 7.29 (d, J = 8.1 Hz, 1H), 7.17–7.12 (m, 2H), 4.19–4.11 (q, J = 7.2 Hz, 2H), 3.96 (s, 2H), 3.94 (s, 2H), 3.44 (s, 2H), 1.27–1.22 (t, J = 7.2 Hz, 3H). LRMS (ESI) calcd for C₁₆H₁₈BrN₃O₂Na [M+Na]⁺: 386, found: 386.

Et₂BPG₂DEV·2HCl (5·2HCl) (Diethyl 2,2'-(5,5'-(4,5-dimethoxy-1,2-phenylene) bis(ethyne-2,1-diyl)bis(pyridine-5,2-diyl))bis(methylene)bis((pyridin-2-ylmethyl)-azane-diyl)diacetate) Hydrochloride

4,5-Diethynylveratrole (0.474 g, 2.55 mmol), 4 (1.82 g, 5.03 mmol), [Pd(PPh₃)₄] (0.250 g, 0.216 mmol), Et₃N (3.6 mL, 26 mmol), and THF (24 mL) were combined in a sealed tube under an inert atmosphere and stirred for 2.5 d at 60 °C. After the reaction mixture was cooled to room temperature, it was combined with EtOAc (50 mL) and washed three times with a solution of Na₂CO₃ (aq.), dried over Na₂SO₄, filtered, and evaporated to dryness. The crude material was purified by column chromatography (alumina; EtOAc/hexanes, 1:1–1:0) to give 5 as thick yellow oil. ¹H NMR spectroscopy confirmed complete conversion to the desired product, but a phosphine oxide impurity was observed. To purify the material, the hydrochloride salt of the amine, 5·2HCl, was synthesized. The oil was dissolved in ca. 100 mL of EtOAc and ca. 4 mL of a solution of HCl in Et₂O (2 M) was added dropwise with stirring until no further precipitation occurred. The yellow solid was filtered off and dried in vacuum. Yield: 1.20 g (57%). ¹H NMR (300 MHz, CD₃OD): δ = 8.92 (m, 2H); 8.89 (m, 2H); 8.54 (dt, J = 1.5; 7.8 Hz, 2H); 8.37 (dd, J = 2.1, 9.0 Hz, 2H); 8.07 (d, J = 7.8 Hz, 2H); 8.00–7.96 (m, 2H); 7.90 (d, J = 8.4 Hz, 2H); 7.25 (s, 2H); 4.58 (s, 4H); 4.56 (s, 4H); 4.18 (q, J = 7.2 Hz, 4H); 3.92 (s, 6H); 3.74 (s, 4H); 1.25 (t, J = 7.2 Hz, 6H). ¹³C NMR (125 MHz, CD₃OD): δ = 171.27, 154.21, 153.90, 151.40, 147.44, 146.96, 145.77, 142.35, 127.61, 127.31, 126.77, 122.78, 117.88, 115.47, 94.90, 86.30, 61.68, 56.97, 56.48, 56.24, 54.90, 13.87. HRMS (ESI) calcd for [M+H]⁺: 753.3397, found: 753.3397. IR (KBr, cm⁻¹): 2979 (m), 2914 (m), 2611 (m), 2214 (m, ν_C≡_C), 1731 (s), 1614 (m), 1593 (m), 1511 (s), 1464 (m), 1371 (m), 1251 (s), 1215 (s), 1155 (m), 1086 (m), 1024 (m), 987 (m), 858 (w), 772 (w). Mp: 103–105 °C.

H₂BPG₂DEV (5a), 2,2'-(5,5'-(4,5-Dimethoxy-1,2-phenylene)bis(ethyne-2,1-diyl)bis(pyridine-5,2-diyl))bis(methylene)bis((pyridin-2-ylmethyl)azanediyl)diacetic Acid

An aqueous solution of 5·2HCl (0.55 g, 0.67 mmol) and KOH (1.4 g, 25 mmol) were combined to yield a total volume of 100 mL. The resulting suspension was heated to 60 °C under a nitrogen atmosphere for ca. 4 h. After cooling down to room temperature, the reaction mixture was acidified with dilute HCl to pH 5 and the product was extracted with CH₂Cl₂. The organic phase was dried over Na₂SO₄, filtered, and reduced to dryness to yield a yellow-brown solid. Yield: 0.42 g (91%). ¹H NMR (300 MHz, CDCl₃): δ = 13.11 (s, 2H), 8.68 (m, 2H), 8.58 (m, 2H), 7.73 (m, 4H), 7.35 (m, 4H), 7.24 (m, 2H), 7.03 (s, 2H), 4.13 (s, 4H), 4.07 (s, 4H), 3.94 (s, 6H), 3.57 (s, 4H). ¹³C NMR (125 MHz, d₆-DMSO): δ = 173.44, 160.29, 160.05, 151.76, 150.64, 149.93, 139.86, 137.83, 123.93, 123.65, 123.40, 118.94, 118.40, 115.49, 92.16, 90.19, 60.32, 60.22, 57.00, 55.46. HRMS (ESI) calcd for [M+H]⁺: 697.2770, found: 697.2770. IR (KBr, cm⁻¹): 3054 (w), 3002 (w), 2912 (w), 2832 (w), 2206 (w, ν_C≡_C), 1714 (m), 1637 (w), 1593 (m), 1551 (w), 1511 (s), 1437 (m), 1401 (m), 1359 (m), 1248 (s), 1215 (s), 1149 (m), 1119 (m), 1085 (m), 1024 (m), 993 (m), 858 (w), 761 (m), 722 (m), 695 (m), 649 (w), 621 (w), 541 (m). Mp: 60–62 °C.

[Fe₂(μ-O)(H₂O)₂BPG₂DEV](ClO₄)₂ (6)

To a solution (2 mL; CH₃CN/H₂O, 10:1) of Fe(ClO₄)₃·9H₂O (77 mg, 150 μmol) was added a suspension of 5a (50 mg, 72 μmol) in the same solvent mixture (2 mL). The color instantly changed to deep-red and the resulting solution was stirred for ca. 5 min. After filtration, the solution was subjected to vapor diffusion of Et₂O to yield crystalline red plates that were analyzed by X-ray crystallography. The yield of crystalline material was highly dependent on the H₂O content of the solution. Yield: 60 mg (78%). X-ray diffraction quality crystals were grown from vapor diffusion of Et₂O into a solution of 6 in CH₃OH and H₂O. LRMS (ESI) calcd for [M-H]⁺: 857.1, found: 875.2. IR (KBr, cm⁻¹): 3430 (m), 3073(w), 2921 (w), 2854 (w), 2210 (w, ν_C≡_C), 1608 (s), 1590 (s), 1550 (m), 1512 (s), 1492 (m), 1462 (w), 1446 (m), 1402 (w), 1367 (m), 1348 (m), 1301 (w), 1286 (w), 1250 (s), 1212 (m), 1099 (vs), 1086 (vs), 1024 (m), 992 (w), 927 (w), 902 (w), 807 (m, ν_Fe–O–Fe), 768 (m), 731 (w), 667 (w), 651 (w), 622 (m), 570 (w), 476 (w), 423 (w). Anal. Calcd. for 6·2CH₃OH, C₄₂H₄₆N₆O₁₉Cl₂Fe₂: C, 44.98; H, 4.13; N, 7.49. Found: C, 45.18; H, 4.19; N, 7.12. UV–vis (CH₃CN:H₂O, 10:1) (λ_max, nm (ε, M⁻¹ cm⁻¹)): 480 (440). Mp: 175–180 °C (dec).

[Fe₂(μ-O)(μ-O₂CAr^iPrO)BPG₂DEV](ClO₄) (7)

Method A. A red CH₃OH/CH₂Cl₂ (1:1) solution of 6 (63 mg, 60 μmol) was allowed to react with HO₂CAr^iPrO (21 mg, 90 μmol) in the presence of NEt₃ (40 μL) to form an intense green solution. Clusters of green needles were isolated by vapor diffusion of Et₂O into this reaction mixture. Yield: 40 mg (57%). Method B. A CH₃OH/CH₂Cl₂ (1:1) solution of Fe(ClO₄)₃·9H₂O (77 mg, 150 μmol) was combined with HO₂CAr^iPrO (19 mg, 79 μmol) and NEt₃ (60 μL). To this reaction mixture, a solution of 5a (50 mg, 72 μmol) was instantly added and the resulting deep green solution was stirred for 10 min, filtered, and subjected to vapor diffusion of Et₂O. Deep green needle-shaped crystals, suitable for X-ray crystallographic analysis, were isolated. Yield: 38 mg (46%). LRMS (ESI) calcd for [M–ClO₄]⁺: 1059.2, found: 1059.5. IR (KBr, cm⁻¹): 3076 (w), 2975 (w), 2930 (w), 2211 (w, ν_C≡_C), 1645 (s), 1608 (s), 1593 (s), 1530 (s), 1513 (s), 1459 (s), 1422 (s), 1372 (m), 1356 (m), 1332 (m), 1289 (m), 1249 (s), 1087 (vs), 1024 (m), 996 (w), 923 (w), 906 (w), 848 (w), 834 (w), 771 (m, ν_Fe–O–Fe), 729 (w), 668 (w), 623 (m), 551 (w), 512 (w). Anal. Calcd for 7·0.5CH₂Cl_2, C_53.5H₅₃N₆O₁₅Cl₂Fe₂: C, 53.43; H, 4.44; N, 6.99. Found: C, 53.84; H, 4.87; N, 7.08. UV–vis (CHCl₃:CH₃OH, 1:1) (λ_max, nm (ε, M⁻¹ cm⁻¹)): 642 (130); 522 (sh, 218); 490 (730); 475 (sh, 569); 436 (sh, 1171); 413 (1577). Mp: 195–200 °C (dec).

[Fe₂(μ-O)(μ-CO₃)BPG₂DEV] (8)

A solution (6 mL, CH₂Cl₂/CH₃OH, 1:1) of 5a (100 mg, 144 μmol) and Et₃N (40 mL, 290 μmol) was added dropwise to a solution (7 mL, CH₂Cl₂/CH₃OH, 1:1) of (Et₄N)₂[Fe₂(μ-O)Cl₆] (87 mg, 144 μmol). Solid Ag₂CO₃ (159 mg, 576 μmol) was then added to the brown solution. The resulting suspension turned green and was allowed to react for ca. 45 min. The mixture was filtered and subjected to vapor diffusion of Et₂O. After 2–3 days, emerald-green needles of 8 were harvested that were suitable for X-ray crystallography. Yield: 93 mg (73%). LRMS (ESMS) calcd for (M+H)⁺: 883.1, found: 883.0; calcd for (M–CO₃+OH)⁺: 839.1, found: 839.2. IR (KBr, cm⁻¹): 3065 (w), 2956 (w), 2920 (w), 2846 (w), 2210 (w, ν_C≡_C), 1617 (s), 1610 (s), 1571 (w), 1533 (m), 1513 (s), 1495 (m), 1444 (m), 1403 (w), 1383 (w), 1358 (m), 1321 (m), 1304 (m), 1289 (m), 1272 (w), 1248 (m), 1214 (m), 1182 (m), 1158 (w), 1142 (w), 1123 (w), 1102 (w), 1087 (w), 1055 (w), 1040 (w), 1024 (w), 997 (w), 927 (w), 908 (w), 860 (w), 847 (w), 838 (w), 770 (m, ν_Fe–O–Fe), 742 (w), 695 (w), 663 (w), 564 (w), 552 (w), 528 (w), 460 (w), 437 (w), 423 (w). UV–vis (CHCl₃:CH₃OH, 1:1) (λ_max, nm (ε, M⁻¹ cm⁻¹)): 638 (124); 517 (193); 487 (567); 473 (sh, 342); 411 (1576). Anal. Calcd for 8·0.75CH₂Cl₂, C_41.75HH_35.5N₆O₁₀Cl_1.5Fe₂: C, 53.00; H, 3.78; N, 8.88. Found: C, 52.99; H, 4.27; N, 9.10. Mp: 150–155 °C (dec).

X-ray Crystallographic Studies

Intensity data were collected on a Bruker SMART APEX CCD diffractometer with graphite-monochromated Mo Kα radiation (λ = 0.71073 Å), controlled by a Pentium-based PC running the SMART software package.³⁷ Single crystals were mounted on the tips of glass fibers, coated with Paratone-N oil, and cooled to 110 K under a stream of N₂ maintained by a KRYO-FLEX low-temperature apparatus. A total of 2800 frames were collected for each compound. The structures were solved by direct methods and refined on F² by using the SHELXTL-97 software included in the SHELXTL software package.^38,39 Empirical absorption corrections were applied by using the SADABS program,⁴⁰ and the structures were checked for higher symmetry with the PLATON software.⁴¹ All non-hydrogen atoms were located and their positions refined with anisotropic thermal parameters by least-squares cycles. All hydrogen atoms were assigned to idealized positions and given thermal parameters equivalent to either 1.5 (methyl hydrogen atoms) or 1.2 (all other hydrogen atoms) times the thermal parameter of the carbon atoms to which they were attached. The hydrogen atoms of O8 and O9 of the H₂O ligands in the diiron(III) complex 6 and of O1W were located from a difference map and their bond distances to the corresponding oxygen atoms restrained using a DFIX command. Refinement details, data collection parameters, and crystal data for 6, 7, and 8 are reported in the Supporting Information.

Mössbauer Spectroscopy

Mössbauer spectra were recorded on an MSI spectrometer (WEB Research Company) with a ⁵⁷Co source in a Rh matrix maintained at room temperature. Solid samples of 6, 7, and 8 were prepared by suspension of 13 to 20 mg of the powdered solids in Apiezon M grease and placed in a nylon sample holder. Solution samples of reaction intermediates 8a and 7a were prepared by respectively treating suspensions (ca. 0.8 mL) of the oxo-bridged diiron complexes 8 (ca. 20 μmol, CH₃CN/H₂O, 1:1) and 7 (ca. 10 μmol, CH₃OH/H₂O, 4:1) with Et₃N (15 μL) followed by addition of aqueous solution of H₂O₂ (50 μL, 30%). After ca. 1 min, the red-brown reaction mixture was transferred to a sample holder and instantly frozen in liquid N₂. A sample of 6a was prepared similarly, but using a more dilute solution (ca. 2 μmol, CH₃OH/H₂O, 20:1) of ⁵⁷Fe-enriched complex 6, ⁵⁷Fe-6, which was allowed to react with Et₃N (5 μL) and excess H₂O₂ (10 μL, 30%) for ca. 40–50 s and then frozen instantly. The synthesis of ⁵⁷Fe-6 is described in Supporting Information. Data were acquired at 4.2 K and 90 K, and isomer shift (δ) values are reported relative to the room temperature Mössbauer spectrum of a metallic iron foil, which was used for calibration of the velocity scale. The spectra were fit to Lorentzian lines with the WMOSS plot and fit program.⁴²

Resonance Raman Spectroscopy

Resonance Raman (RR) spectra were obtained on a custom built McPherson 2061/207 spectrometer equipped with a Princeton Instrument liquid-N₂ cooled CCD detector (LN-1100PB). Excitation at 647, 568, and 413 nm was provided by a Kr laser (Innova 302, Coherent) and 514- and 488- nm excitations by an Ar laser (Innova 90, Coherent). The laser beam was kept at low power, between 40 and 10 mW, and was focused with a cylindrical lens onto frozen samples in NMR tubes. Long wave pass filters (RazorEdge filters, Semrock) or super-notch filters (Kaiser Inc.) were used to attenuate Rayleigh scattering. Sets of ~4 min accumulations were acquired at 4-cm⁻¹ resolution. Frequency calibrations were performed using aspirin and are accurate to ±1 cm⁻¹. The samples were subjected to continuous spinning at 110 K to prevent adverse effects from laser illumination. Regular visual inspection of the sample colors during the course of the RR experiments and comparison of consecutive RR spectra provided a reliable means to evaluate photobleaching of species of interest.

Electrospray Ionization Mass Spectrometry of the Peroxo Intermediate

Products 6a, 7a, and 8a, which result from reaction of H₂O₂ and NEt₃ with 6, 7, or 8, respectively, were identified by ESI-MS using a Finnigan LTQ ion trap mass spectrometer maintained at the Department of Chemistry, Tufts University. The samples were prepared by addition of a 50–100 fold excess of a 2% aqueous H₂O₂ solution containing 5 equiv of NEt₃ to solutions of the diiron(III) complexes (ca. 1–2 mM) in solvent mixtures of CH₃CN/H₂O or CH₃OH/CH₂Cl₂. The peaks of interest were confirmed by comparison of the observed and calculated m/z values including isotope patterns.

Reaction of 6, 7, and 8 with H₂O₂ and Quantification of O₂ Evolution

In order to quantify the amount of dioxygen formed during hydrogen peroxide decomposition, 2 mL of a 0.4 mM solution of the diiron complex were placed in a 25-mL round-bottom flask equipped with a stir bar. The flask was then tightly sealed with a rubber septum and electrical tape. The solution was treated with ca. 5 μL of NEt₃ and cooled to 0 °C before H₂O₂ was added through the septum with a syringe. The concentration of diiron complex was kept constant by adjusting the final volume of the reaction mixture. The reaction flask was connected by a cannula to an inverted graduated cylinder, filled with H₂O, and the amount of O₂ generated was determined applying the ideal gas law. In the case of 6, the solution was thoroughly purged with nitrogen to remove any traces of CO₂ that can potentially react with 6 under basic conditions to form the carbonato-bridged complex 8.

Substrate Oxidation Studies

Oxidation reactions were carried out under an inert atmosphere of N₂ (g). In each experiment, cis-cyclooctene (0.26 mL, 2.0 mmol) and NEt₃ (5 μL) were added to 2.0 mL of a 1.0 M solution of 6 or 7 in CH₃CN/H₂O (95:5) at 25 °C. To this mixture, H₂O₂ (97 μL, 10.3 M; diluted with CH₃CN to a total volume of 400 μL) was added by syringe pump over a period of one h. After stirring for an additional 2h, the products were determined by GC analysis. The structures of the products were confirmed by GC-MS spectrometry, and by comparison with authentic samples.

Results and Discussion

Synthesis of H₂BPG₂DEV (5a)

Previously we reported an efficient route to a new family of diethynyltriptycene-linked dipyridyl ligands, including Et₂BPG₂DET (DET = 2,3-diethynyltriptycene), an analog of Et₂BPG₂DEV (5).²⁸ Our initial route started from commercially available 2,5-dibromopyridine (1). Following a known reaction, we synthesized 2-formyl-5-bromopyridine (2a) by treatment of 1 with n-BuLi, followed by quenching with DMF.³² With aldehyde 2a in hand, we prepared the pyridylbromide coupling partner 4 by reductive amination of 2a with the N-pyridylmethylglycine ethyl ester 3, which is available in one step from commercially available materials and can be isolated easily by distillation on a multi-gram scale.³³ Although the reaction sequence used to prepare 4 proceeded with moderate yields of less than a gram, we observed that isolated yields of 2a and 4 diminished by roughly 30% for each step upon scale up. We therefore devised an optimized route to the pyridylbromide 4 from 1, which is compared to the previous method in Scheme 1. In this procedure we prepared alcohol 2b by reduction of 2a with NaBH₄ in situ.³² Isolated yields for 2b ranged between 54–60% starting with 10 g of 1. In addition to the higher yield, 2b could be readily purified by crystallization, in contrast to the aldehyde 2a, which was difficult to separate from undesired byproducts. Mesylation of 2b with MsCl and Et₃N afforded a mesylate that was allowed to react with 3 in the presence of K₂CO₃ to furnish 4 in 87% yield over two steps. Using this optimized route, we were able to prepare multi-gram quantities of 4 from 1 in a three step synthetic sequence that required only a single alumina column chromatography purification procedure.

Scheme 1.

Preparation of pyridylbromide 4.

With sufficient amounts of the pyridylbromide 4 in hand, we were able to synthesize large quantities of Et₂BPG₂DEV (5) via a cross-coupling reaction between 4 and 4,5-diethynylveratrole (DEV)³¹ using 10 mol% of [Pd(PPh₃)₄] catalyst in a mixture of Et₃N and THF at 60°C (Scheme 2). We chose the DEV over the DET backbone due to its greater solubility in organic solvents and more facile synthesis. The Sonogashira cross-coupling reaction worked remarkably well upon scale up and the conversion to the reaction product was greater than 90% as determined by ¹H NMR spectroscopy. In order to remove residual phosphine impurities that were present after purification of the crude product by alumina column chromatography, the HCl salt of the amine was isolated by treatment of an EtOAc solution of the purified material with an etheral hydrogen chloride solution. This purification sequence lowered the isolated yield of the corresponding dihydrochloride salt to 57%. Saponification of the esters with KOH in H₂O proceeded cleanly, affording H₂BPG₂DEV in nearly quantitative yield. Although the salt of the ester is stable for months under ambient atmospheric conditions, the dicarboxylic acid decays to form a complex mixture of products over time as evidenced by ¹H NMR spectroscopy. Therefore, we stored the ligand as the hydrochloride salt of the diester and prepared the acid freshly for the following metal complexation studies.

Scheme 2.

Synthesis of the Oxo-Bridged Diiron(III) Compounds [Fe₂(μ-O)(H₂O)BPG₂DEV](ClO₄)₂ (6), [Fe₂(μ-O)(μ-O₂CAr ^iPrO)BPG₂DEV](ClO₄) (7), [Fe₂(μ-O)(μ-CO₃)BPG₂DEV] (8)

Compound 6 crystallizes in good yield as deep red-colored blocks by vapor diffusion of Et₂O into a reaction mixture containing H₂BPG₂DEV and two equiv of Fe(ClO₄)₃·9H₂O in CH₃CN/H₂O (10:1). The yields and the crystal quality depended strongly on the ratio of CH₃CN and H₂O. Only a few examples of oxo-bridged diiron(III) compounds with aqua ligands have been reported because H₂O is prone to substitution and tends to form oxo-bridged polyiron(III) complexes.⁴³ When excess base, such as NEt₃, was added under aerobic conditions in the synthesis of 6, the color changed from deep red to intense green. The resulting complex 8 contained a carbonato- and oxo-bridged diiron(III) center that forms upon reaction with CO₂ from the air. This reaction was complete in less than a few seconds when CO₂ gas was introduced directly into the reaction mixture. With the aforementioned procedure, isolated yields of 8 varied widely. Therefore, an optimized protocol for the synthesis of 8 was developed. This procedure involved treating H₂BPG₂DEV with one equiv of the pre-formed (μ-oxo)diiron(III) complex (Et₄N)₂[Fe₂(μ-O)Cl₆] dissolved in CH₃OH/CH₂Cl₂, producing a yellow-brown solution. Addition of excess Ag₂CO₃ resulted in nearly instantaneous formation of a green suspension that was stirred for 45 min, filtered, and subjected to vapor diffusion of Et₂O. Emerald-green needles were isolated in 73% yield after several days and analyzed by X-ray crystallography. Compound 8 belongs to a rather rare group of carbonato-bridged diiron complexes.^44–50

As with the synthesis of the carbonate complex 8, the carboxylato-bridged complex 7 could be prepared in a variety of ways in moderate yields, either in a reaction between the ligand H₂BPG₂DEV and two equivalents of Fe(ClO₄)₃·9H₂O in the presence of base and excess HO₂CAr^iPrO, or by addition of the carboxylic acid to a CH₃CN solution of 6 in the presence of base. Scheme 3 summarizes the pathways for the synthesis of compounds 6, 7, and 8.

Scheme 3.

Structural Characterization

An ORTEP diagram of 6 is displayed in Fig. 1 and selected bond lengths and angles are reported in Table S2. Compound 6 is a dication and crystallizes with two perchlorate anions in the asymmetric unit and numerous water molecules in the lattice, one of which bridges the two methoxy groups of the veratrole backbone via hydrogen bonds. A packing diagram (Fig. S1 and Table S3, Supporting Information) reveals strong hydrogen-bonding interactions between the H₂O ligands and carboxylate groups of neighboring molecules as well as stacking of the diethynyl-veratrole units. The two iron atoms are separated by 3.4837(9) Å with an Fe-O-Fe angle of 155.37(16)°. The two H₂O molecules bind in a syn fashion with respect to the diiron vector, similar to stoichiometrically analogous complexes that contain either two unlinked BPG^- or two BPP^- (bis(2-pyridylmethyl)aminopropionate) ligands.^51,52 In contrast to these two compounds, the pyridine nitrogen atoms of the BPG^- moieties in 6 are bound syn instead of anti to each other, which is a consequence of the linker. Furthermore, the values for the Fe-Fe distance and the Fe-O-Fe angle in 6 are significantly smaller in comparison to these two complexes (Table 1). In a similar complex with the ligand 6-HPA, where the two neutral TPA units are linked to one another by an ethylene bridge, the distances between the two iron atoms are even longer than 3.56 Å and the Fe-O-Fe angles close to linear.^51–53 The shorter distance and angle in 6 are most likely a consequence of more rigid DEV linker, which draws the iron atoms closer together.

Figure 1.

ORTEP diagram of 6 displaying thermal ellipsoids (40%). For clarity, the ClO₄⁻ anions are not shown.

Table 1.

Comparison of Fe-Fe Distances and Fe-O-Fe Angles (deg) of Diiron Compounds with a [Fe₂(μ-O)(H₂O)₂] Core.

Compound	Fe-Fe (Å)	Fe-O-Fe (deg)	Ref.
6	3.4837(9)	155.37(16)	a
[Fe2(μ-O)(H2O)2BPG2]2+	3.56	169.8(2)	51
[Fe2(μ-O)(H2O)2BPP2]2+	3.564(2)	168.8(3)	52
[Fe2(μ-O)(H2O)2(6-HPA)2]4+	3.607(3)	179.2	53

^aThis work.

Abbreviations: BPP = bis(2-pyridylmethyl)aminopropionate; 6-HPA = 1,2-bis[2-{bis(2-pyridylmethyl)aminomethyl}-6-pyridyl]ethane.

Compounds 7 and 8 have very similar structural parameters. These complexes differ in the chemical nature of the bridging ligand, a carboxylate in the case of 7 and a carbonate ion in 8. Accordingly, compound 7 is positively charged and 8 is neutral. Both diiron cores are displayed in Fig. 2, and crystallographic information and selected bond distances and angles are summarized in Table S2. The BPG₂DEV^2- ligand again stabilizes the dinuclear structures by bridging the diiron centers. The carbonato and carboxylato bridges in 7 and 8 lie approximately in the plane defined by the Fe-O-Fe unit, a feature generally observed in these types of compounds.^47,54

Figure 2.

ORTEP diagram of 7 (left) and 8 (right) displaying thermal ellipsoids (50%). For clarity, the diethynylveratrole backbones, the Ar^iPrO group, and the ClO₄⁻ anion are not shown.

For compound 7 there are two crystallographically independent molecules in the asymmetric unit and geometric comparisons are reported for averaged values. In 7, the Fe-Fe distance is 3.218 Å and the Fe-O-Fe angle 128.5°. Compound 8 contains a (μ-oxo)(μ-carbonato)diiron(III) core in which the two iron atoms are separated by 3.1142(18) Å and related by a pseudo-C₂ axis. The Fe-O-Fe angle is 125.97(15)° and the distance to the bridging oxygen atom is 1.740(3) Å, which is consistent with values for other oxo-bridged diiron(III) complexes.^55,56

Despite an extensive literature on diiron complexes having tripodal amine ligands, especially TPA derivatives, little has been reported for more carboxylate-rich systems.^24,25 In a CSD search, only three crystal structures with the simple BPG⁻ ligand bound to iron are listed.⁵⁷ The first has a structure similar to that of 7, but with a bridging benzoate.⁵⁴ No reactivity studies of this compound were reported. The second is a hydroxo-bridged diiron compound⁵⁸ and the third structure contains two H₂O ligands bound to an oxo-bridged diiron core (vide infra).⁵¹ Although we obtained evidence for a hydroxo-bridged BPG₂DEV^2– diiron(III) complex by mass spectrometry (vide infra), we were not able to isolate such a species.

Spectroscopic Characterization of 6, 7, and 8

The UV-vis spectrum of 6, which contains only oxo as a bridging ligand, differs dramatically from those of the oxo/carboxylato- and oxo/carbonato-bridged compounds 7 and 8. Fig. 3 reveals that 6 has a maximum in its optical absorption spectrum around 490 nm, whereas more bands appear in the spectra of 7 and 8, which are similar to one another. Despite this spectral similarity, the solubilities of the two latter complexes differ significantly. Compound 7 has excellent solubility in CH₃CN and other organic solvents, presumably owing to the lipophilic carboxylate ligand, but 8 is only appreciably soluble in mixtures of CH₃OH/CH₂Cl₂ or CH₃CN/H₂O.

Figure 3.

UV-vis absorption spectra of 1 mM solutions of 7 (black) and 8 (green) in CH₃OH/CHCl₃ (1:1) and of 6 (red) in CH₃CN/H₂O (10:1).

Zero-field Mössbauer spectra of solid samples of 6, 7, and 8 were acquired at 4.2 K and/or 90 K. Table 2 summarizes the derived Mössbauer parameters from these experiments and the spectra are displayed in Fig. 4. The isomer shifts acquired at 90 K lie between δ = 0.45 and 0.47 mm/s and quadrupole splitting parameters between ΔE_Q = 1.35 and 1.59 mm/s for the three compounds. These values are characteristic of 6-coordinate high-spin (μ-oxo)diiron(III) complexes, which generally have quadrupole splitting values of > 1 mm/s.⁵⁵

Table 2.

Zero-Field Mössbauer Parameters of Solid 6, 7, and 8, Acquired at 90 K.

Compound	δ (mm/s)	ΔEQ (mm/s)	Γ (mm/s)
6	0.47(2)	1.59(2)	0.29
7	0.46(2)	1.47(2)	0.31
8	0.47(2)a	1.35(2)a	0.28a
	0.45(2)	1.36(2)	0.31

^aAcquired at 4.2 K.

Figure 4.

Zero-field Mössbauer spectrum of 8 (A), 7 (B), and 6 (C) acquired at 90 K [experimental data (❶) and calculated fits (—)].

Fixation of CO₂ to Form the Carbonato-Bridged Complex 8

Complex 6 reacts with atmospheric CO₂ in organic solvents containing triethylamine to form the carbonato-bridged diiron(III) complex 8. When this reaction was monitored by UV-vis spectroscopy, rapid and quantitative formation of a species having the spectral properties of 8 was apparent (Fig. 5). Even at low temperatures (−78 °C), the reaction occurred in a few seconds following introduction of excess CO₂ (g). Fixation of CO₂ by hydroxo-bridged diiron(II) complexes has been observed previously,⁵⁹ but only three examples exist in which oxo-bridged diiron(III) complexes react with CO₂ to form carbonato-bridged species.^46,47,60

Figure 5.

UV-vis spectra of 6 before (- - -) and after reaction with CO₂ (—) in a basic CH₃OH/CH₂Cl₂ (1:1) solution.

We propose that 8 forms in stepwise manner via formation of a metal hydroxo species. An analogous mechanism was established for the enzyme carbonic anhydrase, which catalyzes the physiologically essential hydration of CO₂ to form bicarbonate.^61,62 Deprotonation of a water ligand in 6 could produce a bridging hydroxo ion, which is a reasonable supposition because a diiron species with this composition is observed in the electrospray mass spectrum of 6 under basic conditions (vide infra) and because base is required to promote the reaction in organic solvents. In a second step, nucleophilic attack of the hydroxo ligand on the carbon atom of CO₂ would form bicarbonate. The latter is readily deprotonated to produce the observed carbonato-bridged diiron(III) complex.

Mass Spectrometry – Stability of the Dinuclear Core

The coordination chemistry of complexes 6, 7, and 8 is also reflected in ESI mass spectrometry experiments. Mass spectra of the dinuclear complexes dissolved in neutral solution revealed the parent compounds, but in basic aqueous solutions, several new ion clusters are detected. Chart 3 displays the core structures of the resulting diiron species. Two new prominent peaks at m/z = 839 and 883 correspond to positively charged hydroxo- and carbonato-bridged complexes having the formula [Fe₂(μ-O)(μ-OH)(BPG₂DEV)]⁺ and {[Fe₂(μ-O)(μ-CO )(BPG₂DEV)]+H}⁺ (8), respectively. The hydroxo-bridged complex can form either by substitution of a bridging ligand by OH⁻ or by deprotonation of a ligated H₂O molecule in 6. Although no crystal structure was obtained that revealed this particular unit, it is reasonable to propose that the core can readily form, by analogy to related diiron(III) complexes with a BPG⁻ ligand and derivatives thereof.^51,58 The carbonato-bridged species most likely derives from a reaction in basic solution of the diiron complex with atmospheric CO₂, as occurs during the synthesis of 8. Since complex 6 contains labile H₂O ligands, it is not surprising that a peak corresponding to the parent ion is not observed in the mass spectrum. When excess carboxylic or carbamic acid was added to solutions of 6, 7, and 8, a dominant peak for the corresponding carboxylate- or carbamate-bridged species was observed. Interestingly, very bulky carboxylates could also be introduced this manner, suggesting that even sterically hindered bridging ligands are accommodated by the platform (data not shown). In contrast to the described systems, analogous Fe–O–Fe complexes containing non-covalently linked BPG⁻ units and derivatives thereof readily dissociate or oligomerize to give mononuclear or trinuclear complexes, respectively.^51,58

Chart 3.

Reaction of 6, 7, and 8 with H₂O₂ – Characterization of a Common Peroxo Intermediate

(a) UV-vis Spectroscopic Studies

When 6, 7, and 8 were treated with excess H₂O₂ in the presence of a small amount of NEt₃, the solutions turned to a more intense, maroon color, concomitant with evolution of a gaseous product. A test with an alkaline pyrogallol-solution confirmed the formation of dioxygen.^63,64 The intermediate products from reactions with H₂O₂, 6a, 7a, and 8a, in solutions of CH₃CN/H₂O were monitored by UV-vis spectroscopy. Buildup of a broad visible absorption band at 490 nm (ε = 1500 M⁻¹ cm⁻¹) was observed in all three cases. The corresponding spectra are presented in Fig. 6 and in Figs. S5 and S6. When the reactions were repeated without the addition of NEt₃, the solutions turned yellow and no spectra analogous to those for 6a, 7a, or 8a were observed. The formation rate of 6a, 7a, and 8a was strongly dependent on the amount of H₂O₂ used. An excess of ca. 1000 equiv of H₂O₂ was required to achieve the maximum absorption intensity at 490 nm for the carbonato-bridged complex 8 over a period of more than an hour. With compound 7, which has a bridging carboxylate ligand, the band maximized more rapidly (ca. 10–15 min) in the presence of a smaller amount of H₂O₂ (ca. 50–100 equiv, Fig. S5). In contrast, only 5–10 equiv of H₂O₂ and less than a minute of reaction time were required to form 6a (Fig. S6). These observations are consistent with loss of a H₂O ligand occurring more readily than the less labile monoanionic carboxylate or dianionic carbonate ligands. Intermediates 6a, 7a, and 8a were also observed in other solvent mixtures, including CH₃OH/H₂O and CH₂Cl₂/CH₃OH. When 6a was generated in a solution of aqueous methanol, an absorption with essentially the same features as found for 6a in CH₃CN/H₂O was observed, but the peak maximum shifted from 490 nm to 470 nm (Fig. S7).

Figure 6.

UV-vis spectra, recorded at 0 °C, of a reaction mixture of 8 (0.37 mM, green trace) in CH₃CN/H₂O (2:1), NEt₃, and H₂O₂ to form a peroxo intermediate 8a (red trace). The grey trace corresponds to the product from the reaction.

Reaction of 8 with 1000-fold excess H₂O₂ and a catalytic amount of base was monitored by UV-vis spectroscopy, as displayed in Fig. 6. Early in the time course of the reaction (ca. 90 min), the intermediate builds up but it subsequently decays to a species having a similar absorption spectrum as the starting material. After a second addition of a 1000-fold excess of H₂O₂, the band at 490 nm reappears. These experiments, together with mass spectral evidence, reveal that the oxo-bridged diiron core remains largely intact and is most likely bridged by a carbonate ligand because of the similarity of the final spectrum to that of the starting material.

(b) Mössbauer Spectroscopic Studies

Zero-field Mössbauer spectra of frozen solution samples of 6a, 7a, and 8a were acquired at 90 K. The spectrum of 6a is shown in Fig. 7, and those for 7a and 8a are displayed in Figs. S8 and S9. Mössbauer parameters for the three compounds are listed in Table 3. A sample of 8a (in CH₃CN/H₂O) was measured at 4.2 K and fit to two quadrupole doublets with an area ratio of 78:22. Because of limited solubility, the sample contains nearly 80% of starting material 8, which can be assigned to one of the doublets. The second doublet, however, belongs to the reaction intermediate 8a with an isomer shift of δ = 0.63(6) mm/s and quadrupole splitting parameter of ΔE_Q = 0.64(6) mm/s. Similarly, in the 90 K spectrum of 7a (in CH₃OH/H₂O), there is evidence for a population of the starting material 7 in the spectrum, together with another quadrupole doublet assigned to a species having δ = 0.58(2) mm/s and ΔE_Q = 0.56(2) mm/s. The best conversion to the intermediate was achieved in a reaction of a dilute (ca. 2 mM) solution of ⁵⁷Fe-enriched 6 with H₂O₂ and base. Here, 66% of 6a was generated with Mössbauer parameters of δ = 0.58(2) mm/s and ΔE_Q = 0.58(2) mm/s, identical within error limits to those for 7a. Taking into account that the spectrum of 8a was acquired at temperatures and from solvents different than those of 6a and 7a, the data strongly suggest that 6a, 7a, and 8a are the same species. The presence of a single quadrupole doublet is most consistent with two high-spin iron sites symmetrically bridged by a peroxo-ligand.

Figure 7.

Zero-field Mössbauer spectrum [experimental data (❶) and calculated fits (—)] recorded at 90 K for a frozen solution sample of the product of the reaction of 6 with excess H₂O₂ and a trace of NEt₃. The sample contained 66% of 6a (B) and 34% of a diiron(III) species (A; δ = 0.45 mm/s and ΔE_Q = 1.41 mm/s) with Mössbauer parameters similar to those of the starting material.

Table 3.

Zero-Field Mössbauer Parameters of Intermediates 6a, 7a, and 8a, Acquired at 90 K.

	δ (mm/s)	ΔEQ (mm/s)	Γ (mm/s)
6a	0.58(2)	0.58(2)	0.35
7a	0.58(2)	0.56(2)	0.29
8a	0.63(6)	0.64(6)	0.32

We were interested to learn whether the starting diiron complex reforms upon completion of reactions with H₂O₂, as implied by the UV-vis spectroscopic experiments. We therefore measured a frozen solution sample [CH₃OH/H₂O (20:1)] of the decomposition product from a reaction of 6 with ca. 50 equiv of H₂O₂ and a trace of NEt₃. The Mössbauer parameters of δ = 0.46(3) mm/s and ΔE_Q = 1.39(3) mm/s were very similar to those for the starting material 6 (δ = 0.45(3) mm/s; ΔE_Q = 1.51(3) mm/s) in the same basic solution mixture. We conclude that this species is most likely a (μ-oxo)(μ-hydroxo)diiron(III) complex and that no decomposition occurred (vide supra and Chart 3).

The spectroscopic and mass spectrometric (vide infra) data determined for the common intermediate formed in the H₂O₂ reactions of 6, 7, and 8 are characteristic of a (μ-oxo)(μ-peroxo)diiron(III) complex. Table 4 lists some examples of transient peroxo species observed in pre-steady-state studies of non-heme diiron enzymes and of oxo-bridged diiron(III) model systems. Well-defined peroxodiiron(III) intermediates are observed in sMMOH, RNR-R2, and Δ⁹D⁴ and in the T201S variant of ToMOH²⁰ that display characteristic peroxo ligand-to-metal charge transfer (LMCT) bands between 650 to 750 nm and Mössbauer parameters of δ = 0.62–0.68 mm/s and ΔE_Q > 1.0 mm/s.⁴ In contrast, the recently characterized, putative peroxodiiron(III) intermediate of ToMOH has Mössbauer isomer shifts smaller than δ = 0.6 mm/s, quadrupole splitting parameters significantly less than 1 mm/s (ΔE_Q = 0.66 mm/s), and lack an optical band in their absorption spectra.⁶⁵ These significant differences can be rationalized by a different peroxo coordination mode and/or protonation state. Synthetically prepared peroxodiiron(III) complexes have the peroxo ligand generally bound in a μ-1,2-fashion with spectroscopic properties very similar to those of peroxo intermediates in the former group of enzymes.⁵ The UV-vis and Mössbauer spectroscopic properties of the peroxodiiron(III) intermediate reported here, however, differ significantly from these complexes. To our knowledge, of all the synthetic examples of (μ-oxo)(μ-peroxo)diiron(III) species, only the intermediates reported here, 6a, 7a, and 8a, have a Mössbauer quadrupole splitting parameter that is nearly identical to the values for colorless peroxo intermediates in ToMOH and PHH. In addition to their unique Mössbauer spectroscopic properties, intermediates 6a, 7a, and 8a display a considerably blue-shifted LMCT band, assigned by RR spectroscopy (see below), in their UV-vis spectra compared to other peroxodiiron(III) species (Table 4). A comparison with two (μ-oxo)(μ-peroxo)diiron(III) complexes providing an N₄ (6-Me₃-TPA)⁶⁶ and a closely related N₃O (6-Me₂-BPP) coordination environment on each iron atom,⁶⁷ similar to that supplied by the BPG⁻ unit, display LMCT bands at 648 nm and 577 nm, respectively, consistent with the carboxylate being a stronger donor. This observation might explain the hypsochromic shift in intermediates containing the BPG₂DEV^2– ligand.

Table 4.

Spectroscopic Parameters for Peroxo Intermediates in Non-Heme Diiron Enzymes and Synthetic Oxo-Bridged Diiron(III) Compounds.

	Optical		Mössbauer
	γmax (nm)	ε (M−1cm−1)	δ (mm/s)	ΔEQ (mm/s)	Ref.
sMMOH (M. caps. (Bath))	700	1800	0.66	1.51	a
ToMOH (Pseudomonas sp. OX1)	-	-	0.54	0.66	65
ToMOH T201S (Pseudomonas sp. OX1)	675	1500	0.67	1.51	20
RNR-R2 D84E	700	1500	0.63	1.58	a
Δ9-desaturase	700	1200	0.68	1.90	a
[Fe2(μ-O)(μ-1,2-O2)(6-Me3-TPA)2]2+	494; 648c	1100; 1200	0.54	1.68	66
[Fe2(μ-O)(μ-1.2-O2)(μ-O2CR)(hexpy)2]+	510; 605c	1300; 1310	0.53	1.67	68
[Fe2(μ-O)(μ-1,2-O2)(6-Me2-BPP)2]	577	1500	0.50	1.46	67
[Fe2(μ-O)(μ-O2)BPG2DEV]	490	1500	0.58	0.58	b

^aParameters were taken from ref. 1.

^bThis work.

^cperoxo-Fe(III) charge transfer band.

(c) EPR Spectroscopy

EPR spectroscopic analysis of 8a reveals an axial signal with g values of 2.062 and 2.004, which we attribute to a mixed-valent diiron(II,III) (S = ½) species. Spin quantitation indicated only 2.6% abundance, however. The bulk of the sample is EPR-silent, which we assign to two iron(III) centers antiferromagnetically coupled to one another. This result is consistent with a peroxo-bridged diiron(III) complex.⁵ The absence of an EPR signal centered at g = 4.3, characteristic for mononuclear high-spin iron(III) complexes, is also in agreement with these findings. The EPR spectrum is displayed as Fig. S12.

(d) Resonance Raman Spectroscopy

RR spectra of intermediate 6a, prepared with either H₂ ¹⁶O₂ (or H₂ ¹⁸O₂) and obtained with 568-nm excitation, revealed two features at 819 and 845 (772 and 797) cm⁻¹, respectively (Fig. 8). These RR signals are not observed in samples that were allowed to proceed in time beyond decay of intermediate 6a (Fig. 8B). The intensities of the RR doublets also decrease after prolonged laser exposure and photobleaching of the samples from dark brown to light yellow. Relative to solvent bands, these intensities also vary with excitation wavelength, revealing a significant decrease at 647 nm and at 413 nm, where intra-pyridine vibrations enter into resonance (Fig. 9). Because of the complexity of these RR signals and the photosensitivity of intermediate 6a, extracting excitation profiles would be an overinterpretation of the data. Nevertheless, the excitation wavelength dependence of the isotope-sensitive signals qualitatively matches expectation based on the absorption spectra of intermediate 6a (Fig. S6).

Fig. 8.

RR spectra of intermediate 6a (A) and its decomposition product (B) obtained with 568-nm excitation at 110 K. Sharp Raman bands below 400 and above 900 cm⁻¹ correspond to solvent vibrations.

Fig. 9.

RR spectra of intermediate 6a obtained with different laser excitations at 110 K. All traces are normalized to the 921-cm⁻¹ solvent band.

The observed RR frequencies and ¹⁸O-shifts (−47/−48 cm⁻¹) are consistent with metal-peroxo vibrations. The ν(O–O) frequencies in symmetrically-bridged (μ-1,2- peroxo)diiron(III) complexes range from 830 to 908 cm⁻¹ and low frequency O–O stretching bands are generally correlated with short Fe-Fe distances.^69–71 The RR bands associated with ν(O–O) in intermediate 6a are at the lower end of this range and a comparison to these systems indicates that the Fe-Fe distance in intermediate 6a may be as short as 3.1–3.2 Å. Short Fe-Fe distances require stabilization from bridging ligands in addition to the peroxo group, with a μ-oxo group providing the shortest such distance. Although Fe-O-Fe modes were not identified in the RR spectra of intermediate 6a, it is reasonable to assume that the oxo bridge, present in the starting material, remains in intermediate 6a. This assignment is strongly supported by mass spectrometric analysis (vide infra). The covalent linkage between the two iron sites supplied by the syn N-donor ligand BPG₂DEV^2- may also play an important role in promoting a short Fe-Fe distance. In a similar peroxodiiron(III) species, with a dinucleating ligand having two covalently linked tris(2-pyridyl)methane moieties, the O–O mode of the bound peroxide appears at very low value of 830 cm^–1, which is consistent with this hypothesis.^71,72

From spectroscopic studies we can conclude that the peroxo ligand bridges the two iron centers in 6a, but we cannot definitively assign its binding mode. Detailed experimental and theoretical studies on (μ-oxo)(μ-1,2-peroxo)diiron(III) intermediates having unlinked tetradentate capping ligands, related to those in BPG₂DEV^2-, confirmed the 1,2-peroxo bridging mode. Furthermore, a crystal structure of [Fe₂(μ-O)(μ-1,2-O₂)6-Me₂-BPP] (see Table 4), which has an identical ligand set to that of 6a, reveals a 1,2-bridging peroxo ligand lying approximately in the plane defined by the Fe-O-Fe unit.⁶⁷ The spectroscopic properties and the O–O stretching frequencies of the peroxodiiron(III) compound described here, however, differ significantly from those reported for these species and we cannot rule out an alternative peroxo binding mode. Because the mass spectra of 6a–8a reveal the presence of an added proton as well as the peroxo ligand (vide infra), the structure might contain a μ-1,1-hydroperoxodiiron(III) unit in which the dangling O-atom is protonated, a unit that is consistent with all of the data.

The presence of a doublet rather than a single peak for the isotope-sensitive RR bands may result either from two different conformations of the peroxodiiron complex or Fermi splitting of a fundamental O–O stretch corresponding to the doublet average of 832 cm⁻¹. Fermi coupling of the ν(O–O) with underlying non-resonant vibrations has been invoked for several peroxodiiron complexes.^10,17,66,70 On the other hand, the possible occurrence of two distinct peroxodiiron species must also be taken into consideration, perhaps reflecting different protonation states. Optical spectroscopic⁷³ and kinetic studies indicate the presence of two peroxo intermediates in the oxygenation chemistry of sMMOH and ToMOH.²⁰ In the low-frequency region, the [¹⁶O-intermediate – ¹⁸O-intermediate] RR difference spectrum isolates a positive feature at 479 cm⁻¹ that may correspond to the expected ν(Fe-¹⁶O₂) mode, even though its negative ¹⁸O-counterpart does not emerge from the broad and nearly featureless background signal that occurs in this region (Fig. 8).

(e) Mass Spectrometry

ESI-MS of 6a, 7a, and 8a were acquired using solutions of CH₃CN/H₂O (3:1) or CH₂Cl₂/CH₃OH (1:1) in low- and high-resolution modes. In each high-resolution mass spectrum of 6a, 7a, and 8a there was an ion cluster having a mass and isotope pattern identical to that calculated for the M+H (peroxo)diiron(III) species {[Fe₂(μ-O)(μ-O₂)BPG₂DEV]+H}⁺ with a theoretical m/z = 855.1195. The recorded values were m/z = 855.1205, 855.1189, and 855.1161 for 6a, 7a, and 8a, respectively. A comparison between the theoretical and observed spectra is presented in Fig. S13. Isotope-labeling experiments carried out with H₂¹⁸₂O₂ (2% in H₂O) revealed a peak shift of four units from m/z = 855 to 859, demonstrating that the peroxide is bound to the diiron center. A comparison of the two mass spectra is shown in Fig. 10.

Figure 10.

ESI mass spectrum of 7a at 295 K in CH₃CN displaying the isotope patterns for ions at m/z = 855 for {[Fe₂(μ-O)(μ-¹⁶O₂)BPG₂ DEV]+H⁺} (top) and m/z = 859 for for {[Fe₂(μ-O)(μ-¹⁸O₂)BPG₂₂ DEV]+H⁺} (bottom). The peak m/z = 867 in the upper spectrum can be assigned to a formate-bridged diiron complex [Fe₂(μ-O)(μ-O CH)BPG₂DEV]⁺, which is due to the presence of residual formic acid impurities.

Singly-charged ion clusters of [Fe₂(μ-O)(μ-OH)(BPG₂DEV)]⁺ (m/z = 839) and {[Fe₂(μ-O)(μ-CO₃)(BPG₂DEV)]+H}⁺ (m/z = 883) were also observed in these mass spectra, which derive from the reaction of the diiron complex with hydroxide and CO₂ from the air (vide supra). Interestingly, two additional peaks at m/z = 841 and 885 were detected in the ¹⁸O-labeled sample. These values are shifted by two mass units from the peaks for these two diiron complexes, indicating that an ¹⁸O atom is incorporated into the hydroxo- and carbonato-bridged compounds, respectively. This result suggests that the O–O bond of the bound peroxide is indeed cleaved and, presumably, that H₂O is formed. An O–O bond scission is reasonable, because this reaction has been investigated extensively in related (μ-oxo)(μ-1,2-peroxo)diiron(III) complexes, which form high-valent diiron centers, and has been proposed for a similar compound in a dinucleating ligand system.^25,53,74 The intermediate could react with another equivalent of H₂O₂ to form H₂O and O₂. We propose that the newly formed H₂O molecule stays bound to the diiron center and can readily be deprotonated under basic conditions to form a bridging hydroxide ion. Such an intermediate could additionally react with CO₂ from air to form the carbonato-bridged species. By mass spectrometry, we were able to establish that the peroxo species is regenerated upon additional treatment with H₂O₂ (data not shown).

Catalase Activity

When H₂O₂ was added to a basic solution of the diiron compounds 6, 7, or 8, O₂ formation was observed, as confirmed by a test with an alkaline pyrogallol solution.^63,64 The catalase-like activity of the carbonato complex 8 was investigated in more detail and the O₂ formed in the reactions was quantified. To a fixed concentration of diiron(III) complex 8 and NEt₃ in a solution of CH₃CN/H₂O (2:1) at 0 °C were added various amounts of H₂O₂. The quantity of O₂ produced was measured volumetrically. The yield of O₂ depends linearly on the amount of H₂O₂, with a slope of 0.53 (R² = 0.99), which corresponds to catalase-like reaction stoichiometry (Fig. 11 and eq. 1). The diiron catalyst is very efficient and achieves ~10,000 turnovers. Experiments

$$2{\mathrm{H}}_2{\mathrm{O}}_2\to {\mathrm{O}}_2+2{\mathrm{H}}_2\mathrm{O}$$ (1)

with compounds 6 and 7 under the same conditions revealed them to be as efficient as 8 with respect to catalase function.

Figure 11.

O₂ formation from 0.078 to 7.8 mmol H₂O₂ at 0 °C as catalyzed by 8 (2 mL of 0.4 mM solution in CH₃CN/H₂O, 2:1) and 10 μL of NEt₃.

Catalase activity has been reported in reactions of a relatively small number of oxo-bridged diiron(III) compounds with H O ^75–80 In all cases, intermediates with spectroscopic properties characteristic for peroxo-bridged diiron(III) complexes were observed. Some common features of these H₂O₂-disproportionation catalysts include complex decomposition, TON (turn-over number) dependence on the exchangeability of the bridging ligand, and increasing O₂ formation rates at higher pH values. In contrast, to these previously reported systems, decomposition of the complexes 6, 7, and 8 upon treatment with H₂O₂ does not occur. The increased stability of the complexes is attributed to the BPG₂DEV^2- ligand. Remarkably, quantitative O₂ production is observed for all three complexes 6–8, with TONs significantly higher than those reported for other systems (the ratio of [Fe₂]:H₂O₂ never exceeded 1:500) and only when base was added to the reaction mixture. In samples that were lacking additional base, quantitative O₂ production was not observed.

Catalase activity is important for protecting the cell from oxidative damage by excess H₂O₂. A ubiquitous family of metalloenzymes disproportionate H₂O₂ in an exothermic reaction to form H₂O and dioxygen. One of the two most abundant classes of catalases contains an iron protoporphyrin IX cofactor with an axial tyrosinate ligand.⁸¹ The second most common catalase enzyme contains a dimanganese active site.⁸² For non-heme diiron enzymes, catalase activity has been reported for ToMOH⁸³ and toluene-2-monooxygenase from Burkholderia cepacia G4,⁸⁴. By comparison with previous studies of diiron model compounds, the present findings are important for our general understanding of catalase activity in BMMs and the decomposition pathway of H₂O₂ catalyzed by diiron centers.

Substrate Oxidation

We also investigated whether the peroxo intermediates 6a, 7a, and 8a might function as oxidases. The common intermediate quantitatively oxidizes PPh₃ to Ph₃PO, a very easy conversion, but is a poor catalyst for the epoxidation of cis-cyclooctene, unlike some related systems.^53,85,86 Quantitation of the epoxide revealed yields < 1% and TONs of ca. 8 and 3, for 6a and 7a, respectively (Table S4). The production of O₂ was strongly diminished by addition of phosphine to the reaction, but was not significantly affected when cyclooctene was present, which is consistent with competitive catalase and substrate oxidation reactions.

Summary and Perspective

This study describes the successful application of the dinucleating ligand H₂BPG₂DEV to specifically model the syn coordination of histidine residues in non-heme diiron enzymes. This scaffold is more biologically relevant than previous versions due to its greater carboxylate content. Three (μ-oxo)diiron(III) derivatives were characterized, one with a terminal water molecule on each iron atom and two with additional bridging ligands, either carbonate or carboxylate. Reactivity studies with H₂O₂ show that the dinucleating ligand conveys an inherent stability to all three oxo-bridged diiron complexes, which serve as catalases in multi-turnover reactions with large excesses of H₂O₂. These characteristics distinguish these complexes from the broader class of diiron(III) complexes that do not have covalently linked tripodal ligands. The present system belongs to a small group of non-heme diiron model complexes having catalase activity that dominates over its oxygenating properties. Similar catalase activity has been observed in ToMOH and PHH, suggesting additional relevance of the present model complexes to biological counterparts. Spectroscopic analysis of a transient (μ-oxo)(μ-peroxo)diiron(III) species generated in this chemistry reveals properties that deviate from the characteristic parameters for peroxo-bridged intermediates in most model complexes and non-heme diiron enzymes. The Mössbauer spectroscopic properties closely resemble those for a peroxo intermediate in ToMOH, with ΔE_Q values less than 1 mm/s. Raman studies reveal an O–O stretching frequency that is atypically low, suggesting a short iron-iron distance in the peroxo intermediate, presumably facilitated by the dinucleating ligand and possibly reflecting a previously unknown μ-1,1 bridging mode. These physical properties of these complexes differentiate them from compounds with mononucleating ligands. Further studies of related complexes are warranted in pursuit of information to understand the origin of the spectroscopic properties and reactivity of non-heme diiron sites in biology.