Halogen Transfer to Carbon Radicals by High-Valent Iron Chloride and Iron Fluoride Corroles

Abstract

High-valent iron halide corroles were examined to determine their reactivity with carbon radicals and their ability to undergo radical rebound-like processes. Beginning with Fe(Cl)(ttppc) (1) (ttppc = 5,10,15-tris(2,4,6-triphenylphenyl)corrolato³⁻), the new iron corroles Fe(OTf)(ttppc) (2), Fe(OTf)(ttppc)(AgOTf) (3), and Fe(F)(ttppc) (4) were synthesized. Complexes 3 and 4 are the first iron triflate and iron fluoride corroles to be structurally characterized by single crystal X-ray diffraction. The structure of 3 reveals an Ag^I─pyrrole (η¹ − π) interaction. The Fe(Cl)(ttppc) and Fe(F)(ttppc) complexes undergo halogen transfer to triarylmethyl radicals, and kinetic analysis of the reaction between (p-OMe-C₆H₄)₃C• and 1 gave k = 1345 ± 34 M⁻¹ s⁻¹ at 23 °C and 2.18 ± 0.17 M⁻¹ s⁻¹ at −60 °C, and ΔH^⧧ = +9.77 ± 0.29 kcal mol⁻¹; ΔS^⧧ = −14.39 ± 1.29 cal mol⁻¹ K⁻¹ through an Eyring analysis. Complex 4 is significantly more reactive, giving k = 1.158 x 10⁵ ± 0.064 M⁻¹ s⁻¹ at 23 °C. The data point to a concerted mechanism, and show the trend X = F⁻ > Cl⁻ > OH⁻ for Fe(X)(ttppc). This study provides mechanistic insights into halogen rebound for an iron porphyrinoid complex.

Graphical Abstract

Introduction.

The conversion of C-H bonds into other functional groups is of widespread importance in native biological processes and synthetic chemistry. Both heme and nonheme metalloenzymes catalyze such reactions with remarkable selectivity.¹ The heme monooxygenase Cytochrome P450 (CYP) is one of the most well-studied enzymes in this class, yet much remains unknown regarding how selectivity is tuned by different CYPs.^2-5 For example, hydroxylation by CYP,⁶ where R-H is converted to ROH, is sometimes diverted to a desaturation pathway that can result in toxic alkene metabolites,^7-10 or in the decarboxylation of fatty acids, as in the case of CYP-OleT.^11-15 This divergence occurs at the so-called “rebound” step, in which a transient, high-valent Fe^IV(OH)(porph) intermediate (protonated Compound-II (Cpd-II)) transfers •OH to the nascent carbon radical R•, or can mediate the other processes shown in Scheme 1.¹⁶

Scheme 1.

C-H Bond Activation by Cytochrome P450

Our laboratory has been interested in determining the fundamental properties of the metal center that may contribute to steering the outcome of the rebound process by studying appropriate porphyrinoid model compounds. Part of the challenge in studying this process in a direct fashion was the need to synthesize a suitable terminal Fe(OH) complex at the same redox level as Cpd-II, which was unknown at the time we began our efforts. We reported the first example of such a complex, taking advantage of a sterically encumbered corrole platform to give Fe(OH)(ttppc) (ttppc = 5,10,15-tris(2,4,6-triphenyl)phenyl corrolato³⁻).^{17, 18} We also recently described the manganese analog of this complex, Mn(OH)(ttppc), and showed that both species are capable of transferring •OH to tertiary carbon radicals.¹⁹ A detailed kinetic and thermodynamic analysis suggested that these reactions operate via a highly concerted mechanism, with C-O bond formation concomitant with M-OH bond cleavage and one-electron reduction of the metal center to the M^III oxidation state.

Given our success in studying •OH radical transfer reactions with the Fe and Mn ttppc complexes, we hypothesized that the sterically encumbered ttppc platform would allow us to examine the related halogen transfer reactions via a direct, rebound-type process. Our goals were to 1) establish that an iron corrole could mediate the direct transfer of halogens to carbon radicals, 2) determine the comparative reactivity of Fe(X) (X = Cl, F, OH) with carbon radicals, 3) define the mechanism of halogen transfer, and 4) gain insight into the general rules that guide the rebound process in heme and nonheme iron enzymes as well as related catalysts. The halogenation of C-H bonds is of considerable synthetic and biological significance. These reactions are mediated by heme and nonheme iron enzymes, including the heme enzyme chloroperoxidase, although this enzyme does not follow the C-H abstraction/rebound pathway seen for CYP.²⁰ However, the nonheme iron halogenases (e.g. SyrB2, WelO5) carry out selective halogenation of a range of C-H bonds, and are proposed to follow a rebound mechanism in which a putative Fe^III(OH)(halide) intermediate preferentially transfers the halogen atom over the hydroxyl group to the nearby carbon radical (Scheme 2).^21-31 The factors that control the selectivity for nonheme iron-catalyzed halogenation remain under debate, but efforts have recently been made to reengineer nonheme Fe halogenases with specific regio- and stereoselectivities.^32-35

Scheme 2.

C-H Bond Chlorination by αKG-Dependent Enzymes

From a synthetic perspective, the halogenation of C-H bonds has become an increasingly prominent target for catalysis.^36-40 The development of either heme^41-46 or nonheme^47-54 base metal (e.g. Fe, Mn) synthetic catalysts for selective C-H halogenation has been particularly alluring because of their anticipated environmental and economic benefits, as well as their relationship to the biological systems. One prominent example in oxidative C-H fluorination involves homogenous manganese porphyrin catalysts that are postulated to operate via transfer of fluorine from an Mn^IV(F)₂(porph) intermediate to carbon radicals.^41-44 In related work, a heterogenous MOF-based catalyst with incorporated Mn porphyrin centers was shown to catalyze the C-H halogenation of inert hydrocarbons in high yield.⁵⁵ A good example of the power of nonheme iron complexes to catalyze the tandem reactions of C-H cleavage and halogenation was reported with the use of a combination of a room-temperature stable nonheme iron(IV)-oxo complex and an iron(II)-halide complex to catalyze the selective halogenation of aliphatic substrates.⁴⁸ In all of these cases, the rates of halogen transfer compared to other pathways, such as hydroxyl transfer, are key to the efficiencies and selectivities of these systems. However, there remains much unknown about what features of a metal catalyst influence the rates of halogen (or other group) transfer to carbon radicals.

In the present work, we report the synthesis and first crystallographic structural characterization of a fluoride-ligated iron corrole, Fe(F)(ttppc).⁵⁶ Examination of the reactivity of this complex together with the previously reported Fe(Cl)(ttppc)¹⁷ has led to the first direct observations of halogen transfer (or halogen “rebound”), from an iron porphyrinoid complex to a carbon radical. Kinetic and thermodynamic analyses show that halogen transfer is significantly faster than hydroxyl transfer. To carry out these studies we also synthesized the first triflate-ligated iron corrole (Fe(OTf)(ttppc)), a versatile synthon in which the weakly bound OTf⁻ ligand can be replaced by exogenous donors. Density functional theory calculations on the complete structures for the Fe(X)(ttppc) complexes (156 atoms) helped support the spectroscopic and structural analyses.

EXPERIMENTAL SECTION

Materials.

All chemicals were purchased from commercial sources and used without further purification unless otherwise stated. Reactions involving inert atmospheres were performed under Ar by using standard Schlenk techniques or in an N₂-filled dry-box. Acetonitrile, benzonitrile, pentane and toluene were purified via a Pure-Solv solvent purification system from Innovative Technologies, Inc. Acetonitrile was further dried and purified by distilling over CaH₂. Pentane and toluene were further dried and purified by distilling over sodium and benzophenone. Deuterated solvents for NMR and ⁵⁷Fe (95.93%) were purchased from Cambridge Isotopes, Inc. Solvents were degassed by repeated cycles of freeze–pump–thaw and stored over 4 Å molecular sieves in an N₂-filled drybox prior to use. ⁵⁷FeCl₂ (Fe 95.93%) was synthesized using a previously published method.¹⁷ Tris(p-tert-butylphenyl)methyl bromide and tris(p-methoxyphenyl)methyl tetrafluoroborate were synthesized following previously reported procedure.^{17, 57} Tris(p-methoxyphenyl)methyl chloride, tris(p-methoxyphenyl)methanol, and triphenylmethyl tetrafluoroborate were obtained from Alfa-Aesar and Sigma Aldrich, and were used as received. The triphenylmethyl radical source, Gomberg’s dimer (Ph₃C=C₆H₅CPh₂), tris(p-methoxyphenyl)methyl radical and tris(p-tert-butylphenyl)methyl radical were synthesized following literature procedures.¹⁷

Instrumentation.

UV-vis measurements were performed on a Hewlett- Packard Agilent 8453 diode-array spectrophotometer with a 3.5 mL quartz cuvette (path length = 1 cm) fitted with a septum. For all UV-vis spectroscopy measurements acquired below 23 °C, data collections were performed on a Varian Cary 60 Bio spectrophotometer. For reactions with total reaction time of <10 s, stopped-flow experiments were carried out using a HiTech SHU-61SX2 (TgK scientific Ltd.) stopped-flow spectrophotometer with a xenon light source and Kinetic Studio software. Electrospray ionization mass spectra (ESI-MS) were acquired using a Finnigan LCQ Duo ion-trap mass spectrometer equipped with an electrospray ionization source (Thermo Finnigan, San Jose, CA) in positive ion mode. Samples were introduced into the instrument at a rate of 15 μL/min using a syringe pump via a silica capillary line. The heated capillary temperature was 110 °C and the spray voltage was 4 kV. Gas Chromatography (GC-FID) was carried out on an Agilent 6850 gas chromatograph fitted with a DB–5 5% phenylmethyl siloxane capillary column (30 m x 0.32 mm x 0.25 μm) and equipped with a flame–ionization detector. Gas chromatography mass spectrometry (GC-MS) was performed using a Shimadzu GC17A/QP5050A GC/MS combination (Shimadzu Instruments, Columbia, MD). The GC17A is equipped with a low polarity (5% phenyl-, 95% methyl- siloxane) capillary column (30 m length, 0.25 mm ID, 0.25 μm film thickness, 10 m length guard column). Samples were dissolved in toluene and were injected into the instrument using an autosampler. The QP5050A is an EI quadrupole based mass spectrometer with a maximum scan range of 900 amu and an ionizing electron energy of 70 eV. ¹H and ¹⁹F {¹H} NMR spectra were recorded on a Bruker Advance 300 and 400 MHz NMR spectrometer. Electron paramagnetic resonance (EPR) spectra were recorded with a Bruker EMX spectrometer equipped with a Bruker ER 041 X G microwave bridge and a continuous-flow liquid helium cryostat (ESR900) coupled to an Oxford Instruments TC503 temperature controller for low temperature data collection. Mössbauer spectra were recorded on a spectrometer from SEE Co. (Edina, MN) operating in the constant acceleration mode in a transmission geometry. The sample was kept in an SVT-400 cryostat from Janis Research Co. (Wilmington, MA), using liquid N₂ as a cryogen for 80 K measurements. Isomer shifts were determined relative to the centroid of the spectrum of a metallic foil of α-Fe collected at room temperature. Data analysis was performed using version F of the program WMOSS (www.wmoss.org), and quadrupole doublets were fit to Lorentzian lineshapes. Cyclic voltammetry was performed on an EG&G Princeton Applied Research potentiostat/galvanostat model 263A with a three-electrode system consisting of a glassy carbon working electrode, an Ag wire reference electrode with 0.25 M Bu₄N⁺PF₆⁻ in C₆H₅CN, and a platinum wire counter electrode. Potentials were referenced using a ferrocene standard. Scans were run under an Ar atmosphere at 23 °C using Bu₄N⁺PF₆⁻ (0.25 M) as the supporting electrolyte.

Fe(OSO₂CF₃)(ttppc) (2).

An amount of Fe(Cl)(ttppc)¹⁷ (15 mg, 11.5 μmol) was dissolved in toluene (5 mL) to give a dark red solution, and AgOTf (3.0 mg, 12 μmol, 1.0 equiv) was added, forming a dark red suspension. The reaction mixture was stirred in the dark at 23 °C for 1.4 h, without any noticeable color change. An amount of pentane (10 mL) was added, and the mixture was stored in the freezer at −35 °C for 12 h, during which time a fine, off-white precipitate formed which was presumed to be silver salts. The reaction mixture was then filtered, and the supernatant was dried under vacuum to give a red solid (15 mg, 93% yield). UV-vis (toluene): λ_max, nm (ε x 10⁻⁴ M⁻¹ cm⁻¹): 421 (3.40), 532 (0.86).¹H NMR (400 MHz, C₆D₅CD₃) δ (ppm): 8.52 (d), 8.28 (d), 8.21 (t), 8.09 (t), 6.91 (t), 6.82 (t), 3.92 (br), −0.10, −7.38, −9.34, −10.91, −14.10, −52.00. Anal. Calcd for C₉₉H₆₇N₄FeO₃SF₃ (2 • C₇H₈): C, 78.98; H, 4.49; N, 3.72. Found C, 79.40; H, 4.21; N, 3.96.

[Fe(OSO₂CF₃)(ttppc)(AgOTf)] (3).

An amount of Fe(Cl)(ttppc)¹⁷ (15 mg, 11.5 μmol) was dissolved in toluene (5 mL) to give a dark red solution, and AgOTf (6.5 mg, 25 μmol, 2.2 equiv) was added, forming a dark red suspension. The reaction mixture was stirred in the dark at 23 °C for 3 h, without any noticeable color change. The crude mixture was stored in the freezer at −35 °C for 12 h, during which time a fine, off-white precipitate formed which was presumed to be silver salts. The solution was then filtered, and the supernatant volume was reduced under vacuum to 1 mL. Crystals of 2 suitable for XRD were obtained by slow vapor diffusion of pentane into toluene over 3 weeks, giving dark red blocks (16 mg, 85% yield). UV-vis (toluene): λ_max, nm: 358, 419, 533. ESI-MS (m/z): isotopic cluster at 1521.2, ([M – OTf]⁺) (calcd 1521.26). ¹H NMR (400 MHz, C₆D₅CD₃) δ (ppm): 8.50 (d), 8.28 (d), 8.20 (t), 8.09 (t), 6.92 (t), 6.81 (t), 4.07 (s,br), −0.10, −6.68, −8.11, −9.94, −11.20, −14.09, −52.20. Anal. Calcd for C₉₃H₅₉N₄FeO₆S₂F₆Ag: C, 66.87; H, 3.56; N, 3.35. Found C, 65.84; H, 4.14; N, 3.23.

Fe(F)(ttppc) (4).

An amount of Fe(Cl)(ttppc)¹⁷ (25 mg, 19.2 μmol) was dissolved in toluene (6 mL) to give a dark red solution, and an amount of AgOTf (0.77 mL, 19.2 μmol, from a 25 mM stock solution in CH₃CN) was added, forming a dark red suspension. The reaction mixture was stirred in the dark at 23 °C for 2 h, without any noticeable color change. After which time, the resulting solution was passed through a Celite^® plug to remove silver salts, and CsF (3.0 mg, 19.7 μmol) was added and the reaction mixture was stirred for 1 h at 23 °C, forming a dark red suspension. An amount of hexane (8 mL) was added, and stored in the freezer at −35 °C for 12 h, during which time a fine, white precipitate formed which was presumed to be cesium salts. The reaction mixture was then filtered, and the volume of supernatant was reduced under vacuum to give a red solid (21 mg, 87%). A crystal of 4 suitable for XRD was obtained by reducing the volume of supernatant to 1 mL, and layering hexane over 2 weeks, giving a few, very small dark red blocks (10 – 20 microns per side). However, attempts to scale-up this crystallization led to a significant amount of precipitation, and thus the routine preparation of 4 did not involve recrystallization. Isolation of 4 as a red solid, as described above, gave material which yielded highly reproducible UV-vis, ¹H/¹⁹F NMR, and Mössbauer spectra. The Mössbauer spectra indicated that 4 is one species (99% of total Fe). UV-vis (toluene): λ_max, nm (ε x 10⁻⁴ M⁻¹ cm⁻¹): 348 (2.18), 413 (3.33), 488 (1.62), 634 (0.26). ESI-MS (m/z): isotopic cluster at 1283.8 (M⁺) (calcd 1283.4). ¹H NMR (400 MHz, C₆D₅CD₃) δ (ppm): 20.01, 8.44 (d), 8.34 (d), 8.04 (m), 3.99 (br), −2.96, −6.27, −6.99, −8.63, −29.45.

Ph₃CF.

An amount of [(C₆H₅)₃C]⁺(BF₄)⁻ (50 mg) was dissolved in acetonitrile (3 mL) and potassium fluoride (26 mg, 0.45 mmol, 3 equiv) was added. The reaction mixture was stirred at 23 °C for 2 h. A color change of an orange to colorless solution was observed. A white suspension was formed after the addition of benzene (1 mL) which was presumed to be the formation of potassium salts. The supernatant was passed through Celite® plug to give a colorless solution. The filtrate was dried under vacuum to give a white solid (36 mg, 91%). White crystals of product were obtained by layering pentane into dichloromethane. GC-MS (C₆H₅CH₃): m/z 262 (M⁺) (calcd 262.4). ¹H NMR (400 MHz, C₆D₆) δ (ppm): 7.01 – 7.12 (m, 10H), 7.18 – 7.12 (m, 1H), 7.26 – 7.32 (m, 2H). ¹⁹F {¹H} NMR (300 MHz, C₆D₆) δ (ppm): −125.6 (s) (δ = −125.4 ppm⁵⁸ in CDCl3).

(p-MeO-C₆H₄)₃CF.

An amount of [(p-MeO-C₆H₄)₃C]⁺(BF₄)⁻ (32 mg) was dissolved in acetonitrile (3 mL) and potassium fluoride (13 mg, 0.23 mmol, 3 equiv) was added. The reaction mixture was stirred at 23 °C for 2 h. A color change of a dark red to colorless solution was observed. A white suspension was formed by addition of benzene (1 mL) which was presumed to be the formation of potassium salts. The supernatant was passed through Celite® plug to give a colorless solution. The filtrate was dried under vacuum to give a white solid (25 mg, 93%). ¹H NMR (400 MHz, C₆D₆) δ (ppm): 7.06 – 7.12 (m, 6H), 6.65 – 6.72 (m, 6H), 3.46 (s, 9H). ¹⁹F {¹H} NMR (300 MHz, C₆D₆) δ (ppm): −119.8 (s).

Reaction of Fe(Cl)(ttppc) (1) with Triphenylmethyl Radical (Ph₃C•). ¹H NMR Spectroscopy.

An amount of 1 (20 mg, 15 μmol) was dissolved in C₆D₆ (0.75 mL, 26.9 mM) and Ph₃C=C₆H₅CPh₂ (58 mg, 0.12 mmol, 8 equiv) dissolved in C₆D₆ (1.75 mL) was added. The reaction was stirred at 50 °C for 5 h. An aliquot of the reaction mixture (0.75 mL) was loaded into an NMR tube and a ¹H NMR spectrum was collected. The aromatic region of the spectrum is dominated by peaks from the excess Ph₃C=C₆H₅CPh₂ starting material, but a multiplet at 7.34 – 7.39 ppm is well-separated and indicates formation of Ph₃CCl based on comparison with an independent sample. The reaction mixture was quenched with EtOH (1 mL), and then the volatiles were removed under vacuum to give a dark red solid. Analysis by TLC of the crude solid dissolved in ethyl acetate (eluent: hexanes/ethyl acetate 6/1 v/v) showed multiple spots, but a major spot (r_f = 0.34) matched Ph₃CCl + Ph₃COH, which ran together as seen by independent standards. Separation of the major spot by silica gel chromatography (hexanes/ethyl acetate 6/1 v/v) gave a resulting white solid which was analyzed by ¹H NMR spectroscopy. Major peaks for both Ph₃CCl and Ph₃COH were observed, and integration against a standard (hexamethyldisiloxane) gave an average yield for the Ph₃CCl of 36 ± 1% (average of 3 runs). The production of Ph₃COH can be assigned to the reaction of excess Ph₃C• with dioxygen in the air.

Reaction of Fe(F)(ttppc) (4) with Triphenylmethyl Radical (Ph₃C•). ¹H & ¹⁹F {¹H} NMR Spectroscopy.

Complex 4 was made fresh, stored at −35 °C, and used within 2 – 3 days for all reactivity studies. The complex was not crystallized prior to routine use, but the purity of each batch was assessed by ¹H and ¹⁹F NMR spectroscopy. The ¹H NMR spectrum was consistent with 4 as the only major corrole species in solution, and the ¹⁹F {¹H} NMR spectrum showed no peaks corresponding to F⁻ impurities. An amount of 4 (0.55 mL of a 4.25 mM solution in C₆D₆, 2.34 μmol) was combined with Ph₃C=C₆H₅CPh₂ (0.11 mL of a 62 mM solution in C₆D₆, 6.8 μmol) in an NMR tube. The reaction mixture was manually mixed for 30 min at 23 °C, after which time a ¹⁹F {¹H} NMR spectrum was recorded. A single peak at −125 ppm was observed, indicating the formation of Ph₃CF based on comparison with an independent sample. Integration against an internal standard (CF₃C(O)OC₂H₅, −74.85 ppm) gave an average yield of Ph₃CF = 82% ± 3% (average of 2 runs).

Kinetic Study of Fe(Cl)(ttppc) (1) and Tris-(para-Methoxyphenyl)methyl Radical ((p-MeO-C₆H₄)₃C•) at 23 °C.

Under light-limiting conditions, (p-MeO-C₆H₄)₃C• was freshly prepared in toluene according to a literature procedure.¹⁷ The concentration of radical (0.86 −2.54 mM) was determined by UV-vis with the known extinction coefficient (ε = 570 M⁻¹ cm⁻¹ in toluene)¹⁷. Varying amounts of radical (27 – 79 equiv) were combined with 1 (64 μM, 0.2 mL) in toluene in a single-mix stopped-flow apparatus at 23 °C. The spectral changes showed isosbestic conversion of 1 to Fe^III(ttppc) (λmax = 573 nm) over 1 – 2 s. The pseudo first-order rate constants, k_obs, were obtained by nonlinear least-squares fitting of a plot of absorbance at 573 nm corresponding to the production of Fe^III(ttppc). The (Abs_t) versus time (t) plots were fit according to the equation Abs_t = Abs_f + (Abs₀ − Abs_f)exp(−k_obst), where Abs₀ and Abs_f are the initial and final absorbance values, respectively. The second-order rate constant (k₂) was obtained from a slope of the best-fit line of a plot of k_obs versus radical concentration.

Kinetic Study of Fe(Cl)(ttppc) (1) and Tris-(para-Methoxyphenyl)methyl Radical ((p-MeO-C₆H₄)₃C•) at −60 °C.

Under light-limiting conditions, (p-MeO-C₆H₄)₃C• was freshly prepared in toluene according to a literature procedure.¹⁷ The concentration of radical was determined by UV-vis with the known extinction coefficient (ε = 570 M⁻¹ cm⁻¹ in toluene)¹⁷. Varying amounts of radical (19 – 33 equiv) were combined with 1 (27 – 30 μM, 2.15 mL – 1.97 mL) in toluene at −60 °C to start the reaction. The spectral changes showed isosbestic conversion of 1 to Fe^III(ttppc) (λmax = 573 nm) over 30 – 60 min. The pseudo first-order rate constants, k_obs, were obtained by nonlinear least-squares fitting of a plot of absorbance at 573 nm corresponding to the production of Fe^III(ttppc). The (Abs_t) versus time (t) plots were fit according to the equation Abs_t = Abs_f + (Abs₀ − Abs_f)exp(−k_obst), where Abs₀ and Abs_f are the initial and final absorbance values, respectively. The second-order rate constant (k₂) was obtained from a slope of the best-fit line of a plot of k_obs versus radical concentration.

Reaction of Fe(Cl)(ttppc) (1) with Tris-(para-Methoxyphenyl)methyl Radical ((p-MeO-C₆H₄)₃C•). GC-FID.

A fresh batch of (p-MeO-C₆H₄)₃C• stock solution in toluene was prepared from (p-MeO-C₆H₄)₃CCl following a literature procedure.¹⁷ Solid 1 (3.6 mg, 2.8 μmol) was dissolved in an aliquot of the radical stock solution (18 mM, 0.5 mL, 3.34 equiv). The reaction mixture was stirred for 30 min at 23 °C, after which time eicosane (10 μL of an 8.8 mM stock solution in toluene) was added as an internal standard for GC-FID. Analysis of the reaction mixture by GC-FID indicated formation of (p-MeO-C₆H₄)₃CCl (R_T = 26 min) based on comparison with an independent sample. The radical stock solution typically contains a small amount of unreacted (p-MeO-C₆H₄)₃CCl as seen by GC-FID, and this background signal is subtracted from the signal for the product in order to obtain a yield for the reaction of 1 and (p-MeO-C₆H₄)₃C• by GC-FID. Comparison with a calibration curve based on the internal standard gave an average yield of 81 ± 5% for (p-MeO-C₆H₄)₃CCl (average of 3 runs).

Reaction of Fe(Cl)(ttppc) (1) with Tris-(para-Methoxyphenyl)methyl Radical ((p-MeO-C₆H₄)₃C•). Mössbauer Spectroscopy.

An amount of ⁵⁷Fe-1 (3 mg, 2.3 μmol) was dissolved in toluene (0.4 mL) and frozen in a Mössbauer cup in N₂(l). A Mössbauer spectrum (80 K) was recorded, showing a sharp, quadrupole doublet corresponding to ⁵⁷Fe-1 (≥ 99% of the fit). The solution was thawed and placed in a reaction vessel. An amount of freshly prepared (p-MeO-C₆H₄)₃C• (1.4 mL of a stock solution (3.5 mM) in toluene, 2.1 equiv) was added to the solution of ⁵⁷Fe-1, and the reaction mixture was stirred for 1 h at 23 °C, after which time the solvent was removed under vacuum to prepare a more concentrated solution for analysis by Mössbauer spectroscopy. The resulting red solid was dissolved in toluene (0.4 mL) and transferred to a Mössbauer sample holder, which was frozen in liquid N₂ prior to Mössbauer spectroscopy.

Reaction of Fe(F)(ttppc) (4) with Tris-(para-Methoxyphenyl)methyl Radical ((p-MeO-C₆H₄)₃C•). ¹H & ¹⁹F {¹H} NMR spectroscopy.

An amount of 4 (0.35 mL of a 4.5 mM stock solution in C₆D₅CD₃, 1.55 μmol) was combined with a freshly prepared batch of (p-MeO-C₆H₄)₃C• (0.5 mL of a 3.2 mM stock solution in C₆D₅CD₃,¹⁷ 1.04 equiv). The reaction mixture was manually mixed in an NMR tube for 30 min at 23 °C, after which time a ¹⁹F {¹H} NMR spectrum was recorded. A single peak at −120 ppm was observed, indicating the formation of (p-MeO-C₆H₄)₃CF based on comparison with an independent sample. Integration against an internal standard (CF₃C(O)OC₂H₅, −74.85 ppm) gave an average yield of (p-MeO-C₆H₄)₃CF = 93 ± 1% (average of 2 runs).

Kinetic Study of Fe(F)(ttppc) (4) and Tris-(para-Methoxyphenyl)methyl Radical ((p-MeO-C₆H₄)₃C•) at 23 °C.

Under light-limiting conditions, (p-MeO-C₆H₄)₃C• was freshly prepared in toluene according to a literature procedure.¹⁷ The concentration of radical (0.35 – 0.62 mM) was determined by UV-vis with the known extinction coefficient (ε = 570 M⁻¹ cm⁻¹ in toluene)¹⁷. Varying amounts of radical (28 – 50 equiv) were combined with 4 (25 μM, 0.2 mL) in toluene in a single-mix stopped-flow apparatus at 23 °C. The spectral changes showed isosbestic conversion of 4 to Fe^III(ttppc) (λmax = 573 nm) over 0.15 s at 23 °C. The pseudo first-order rate constants, k_obs, were obtained by nonlinear least-squares fitting of a plot of absorbance at 573 nm corresponding to the production of Fe^III(ttppc). The (Abs_t) versus time (t) plots were fit according to the equation Abs_t = Abs_f + (Abs₀ − Abs_f)exp(−k_obst), where Abs₀ and Abs_f are the initial and final absorbance values, respectively. The second-order rate constant (k₂) was obtained from a slope of the best-fit line of a plot of k_obs versus radical concentration.

Computational methods.

All calculations were performed with the ORCA 3.0.3 program package. Starting geometry optimizations for 1 and 4 were derived from XRD structures. Details of the steps taken to achieve the best, optimized geometries and frequencies are given in the Supporting Information. The ⁵⁷Fe Mössbauer parameters for 1, 3, and 4 were determined from a single point energy calculation using TPSSh or B3LYP functional with DKH2 relativistic approximation and NORI approximation at the def2-TZVP level of theory.⁵⁹ A series of grids (Grid5 and FinalGrid6) were set to a TightSCF convergence. The iron atom had an effective core potential of “CP(PPP)” and the radial integration accuracy was increased using the “SpecialGridIntAcc” command (value 7). A simulated solvent effect (COSMO) was employed for these calculations. The calculated electron density at the ⁵⁷Fe nucleus (denoted ρ₀) was transformed into an isomer shift (δ) according to the equation: δ = α(ρ₀ − C) + β where α, β, and C are calibration constants. A published calibration curve of isomer shift with α = −0.1706325, β = 0.33226751, and C = 23615 was used for B3LYP and α = −0.172806013, β = 0.323223706, and C = 23600 was used for TPSSh.⁵⁹ The calculated quadrupole splitting (ΔE_Q) was used without any calibration or correction.

Results and Discussion

Fe(OSO₂CF₃)(ttppc) (2).

To gain access to the axially ligated fluoride complex Fe(F)(ttppc), we initially followed the method reported by Gross for the generation of Fe(F)(tpfc) (tpfc = tris(pentafluorophenyl)corrolato³⁻).⁵⁶ This method involved metathesis of Cl⁻ for F⁻ by addition of AgF to the Fe(Cl)(tpfc) starting material in CH₂Cl₂. In this case, Fe(F)(tpfc) was only metastable, decaying within 30 min as seen by ¹H NMR spectroscopy. Addition of AgF to Fe(Cl)(ttppc) did not result in a noticeable change in the UV-vis spectrum, but did show new peaks by ¹H NMR. However, this species was not stable over 24 h, similar to the observed lack of stability seen for Fe(F)(tpfc). The NMR data showed the presumed Fe(F)(ttppc) was converting back to the Fe(Cl)(ttppc) starting material over this period. Thus an alternative method was sought for the preparation of Fe(F)(ttppc).

Addition of 1 equiv of AgOTf to Fe(Cl)(ttppc) (1) in toluene causes the slow precipitation of an off-white solid over 1 h which was assumed to be AgCl (Scheme 3). Analysis of the reaction mixture by UV-vis spectroscopy shows the loss of 1 and the appearance of new bands at 421 and 532 nm. This new product was moisture-sensitive, making it impractical to purify by standard chromatographic methods under aerobic conditions. After workup to remove Ag salts, a new compound was isolated as a dark red solid. This complex was identified as the desired triflate-ligated complex, Fe(OSO₂CF₃)(ttppc) (2) (vide infra). The UV-vis spectrum of 2 is compared with that of the starting material 1 in Figure 1. The Soret band for 1 at 407 nm is red-shifted by 14 nm in 2, and the Q-band at 637 nm is no longer visible in the spectrum for 2, but a shoulder is seen at 532 nm. However, the spectrum of 2 in the presence of the coordinating solvent CH₃CN (1/1 v/v CH₃CN/C₆H₅CH₃) exhibits new bands at 360, 403, 545 and 640 nm (Figure S3), suggesting that OTf⁻ is displaced by CH₃CN in this solvent system.

Scheme 3.

Synthesis of Complex 2

Figure 1.

UV-vis spectra of 1 (blue, dashed line) and 2 (red, solid line) in toluene at 23 °C.

The ¹H NMR spectrum of 2 in C₆D₅CD₃ reveals 6 paramagnetic peaks from δ = 0 to −60 ppm that are in the same region as peaks seen for Fe(Cl)(ttppc) and Fe(OH)(ttppc) (Figure 2), and the related Fe(Cl)(tpc) (tpc = triphenylcorrolato³⁻).^{17, 60} These resonances have been assigned to the β-CH pyrrole protons for iron triphenylcorrole complexes.^{56, 61-63} The ¹⁹F {¹H} NMR spectrum of 2 in C₆D₅CD₃ has a broad peak at δ = −76.4 ppm (full width at half maximum (FWHM) = 692 Hz), which can be compared to the sharp resonance for AgOTf in the same solvent at −76.5 ppm (FWHM = 2.82 Hz) (Figure S4). The broad line in C₆D₅CD₃ is consistent with coordination of the triflate ligand to the paramagnetic iron center, although tight binding of OTf ⁻ is expected to give a larger downfield shift for the OTf ⁻ peak.^{64, 65} In contrast, the ¹H NMR spectrum of 2 in the presence of CD₃CN (CD₃CN/C₆D₅CD₃ (1/1 v/v) exhibits paramagnetic peaks between 0 to −50 ppm that are shifted from neat C₆D₅CD₃ (Figure S5). The corresponding ¹⁹F {¹H} NMR spectrum exhibits a much sharper peak at δ = −77.73 ppm (FWHM = 127 Hz), consistent with displacement of the OTf⁻ (FWHM = 23.9 Hz) axial ligand by CH₃CN, as indicated by the UV-vis spectra in the same solvent (Figure S3). Electron paramagnetic resonance (EPR) spectroscopy on 2 (X-band, 20 K) showed only a baseline signal, as expected for an integer spin, S = 1 ground state (Figure S6), and as seen for the starting chloride complex 1 and Fe(OH)(ttppc).¹⁷ The zero-field Mössbauer spectrum of 2 in toluene gave well-defined doublets with δ = 0.19 mm s⁻¹; ∣ΔE_Q∣ = 3.02 mm s⁻¹ at 125 K and δ = 0.20 mm s⁻¹; ∣ΔE_Q∣ = 3.01 mm s⁻¹ at 80 K (Figure 3). Neutral Fe(X)(ttppc) complexes are formally in the +4 oxidation state. However, corroles can behave as non-innocent ligands, and other iron-halide corroles have been described as exhibiting electronic structures closer to Fe^III(X)(corrole^•+).^66-69 We avoid this ambiguity by leaving out formal oxidation state assignments in the formulas for the Fe(X)(ttppc) complexes.

Figure 2.

¹H NMR spectra of 2 (top) and 3 (bottom) in C₆D₅CD₃ at 23 °C. Inset: expanded region from 6 – 9 ppm. Chemical shifts labeled in red highlight the peak at −7.4 ppm in 2 which splits into two peaks in the presence of AgOTf in 3.

Figure 3.

Zero-field ⁵⁷Fe Mössbauer spectra of 2 at 125 K (top), 2 at 80 K (middle), 3 at 80 K (bottom) in toluene. Experimental data = black circles, best fit = red line.

[Fe(OSO₂CF₃)(ttppc)(AgOTf)] (3).

During efforts to synthesize the triflate-ligated complex 2, another species was isolated and structurally characterized by XRD. This complex was produced from the reaction of 1 with 2.2 equiv of AgOTf in toluene. Dark red crystals suitable for XRD were grown from slow vapor diffusion of pentane into toluene. The crystal structure of 3 is shown in Figure 4, revealing a five-coordinate, mononuclear iron complex with an axial OTf⁻ ligand, and a single Ag ion in close contact with a pyrrole ring and two peripheral phenyl groups. A second OTf⁻ anion coordinates to the Ag^I, completing its coordination sphere. The Ag − C_β distances of 2.55(2) and 2.46(2) Å are similar to what has been seen for Ag − C_β interactions (2.44, 2.48 Å) in a Ni^II porphyrin complex from XANES and DFT analysis.⁷⁰ There are also relatively close contacts (2.55(2) – 2.62(2) Å and 2.53(2) – 2.57(2) Å) between the Ag^I ion and two of the carbon atoms on each of the adjacent phenyl rings, similar to other η² − π interactions seen for the Ag^I ion and benzene derivatives.⁷¹ A number of Ag^I − π interactions involving pyrrole rings have been structurally characterized,⁷² but to our knowledge, the structure of 3 represents the first example of a crystallographically characterized porphyrin or corrole complex that exhibits Ag^I − π bonding. The Fe─O1 bond length of 1.990(19) Å is similar to Fe^III(OTf)(tpp) (tpp = triphenylporphyrinato²⁻) (Fe-O = 1.946(6) Å – 1.996(3) Å),^73-76 and the average Fe-N_pyrrole distance of 1.901(2) Å is in line with other iron (ttppc) (S = 1) structures (Table S5). The iron atom in 3 is displaced by 0.374 Å from the least-squares plane defined by the 4 N_pyrrole atoms toward the OTf⁻ ligand, the same amount of displacement as seen for Fe(OH)(ttppc).¹⁷

Figure 4.

Displacement ellipsoid plot (40% probability) of 3 at 110 K. Hydrogen atoms omitted for clarity. Inset: chemical structure of 3.

The UV-vis spectrum for 3 is similar to 2 in toluene (Figure S7). The difference between both complexes was the appearance of a shoulder at 358 nm unseen for 2. The Soret band for 3 is slightly blue-shifted by 2 nm. These changes indicate that the second coordination sphere Ag^I ion has some influence on the absorbance bands arising from the π system of the corrole. Mass spectrometry (ESI-MS) in 1/1 (v/v) acetone/toluene gives a major peak at m/z 1521.2, corresponding to [3 – OTf]⁺ (Figure S8). These data show that the structure of 3, including the peripherally bound Ag^I ion, remains intact in solution.

The ¹H NMR spectrum of 3 in C₆D₅CD₃ reveals 7 paramagnetic peaks from δ = 0 to −50 ppm. A comparison of the spectra for 2 and 3 indicates that the outer-sphere Ag^I ion in 3 causes a splitting of the peak at δ = −7.39 ppm seen in 2, resulting in two new peaks at δ = −6.68 and −8.11 ppm (Figure 2). The other peaks in the spectrum for 3 (δ = 7 – 9 ppm) do not show any significant change from those seen for 2. The ¹⁹F {¹H} NMR spectrum of 3 in C₆D₅CD₃ reveals two peaks at δ = −48.96 and −76.60 ppm (Figure S4). The broad peak at −48.96 ppm (FWHM = 750 Hz) can be assigned to the triflate bound to the iron center, while the relatively sharp peak at δ = −76.6 ppm (FWHM = 91.6 Hz) can be assigned to the outer-sphere OTf⁻ coordinated to the Ag^I ion.

The zero-field Mössbauer spectrum of 3 generated in situ in toluene at 80 K revealed an isomer shift close to 2 (δ = 0.22 mm s⁻¹ versus 0.20 mm s⁻¹) and the quadrupole splitting is slightly shifted (∣ΔE_Q∣ = 3.36 mm s⁻¹ versus 3.02 mm s⁻¹) (Figure 3). The combined data suggests the peripheral Ag^I ion does exert a small influence on electronic environment around the iron center. Density functional theory (DFT) calculations were employed to corroborate the Mössbauer spectrum. The Mössbauer parameters were computed via unrestricted single point energy calculations (B3LYP/def2-TZVP) on the fixed geometry from the XRD structure for 3 with an S = 1 spin state. The calculations gave δ = 0.22 mm s⁻¹; ∣ΔE_Q∣ = 3.27 mm s⁻¹, which were a good match to the experimental data.

Synthesis and Structural Characterization of Fe(F)(ttppc) (4).

The fluoride complex Fe(F)(ttppc) (4) was synthesized as shown in Scheme 4, starting from Fe(OTf)(ttppc) generated in situ followed by addition of CsF. A white precipitate was observed after 1 h, and UV-vis analysis of the reaction mixture showed formation of a new species (λ_max = 348, 413, 488 and 634 nm). Isolation of the product was accomplished by precipitation with pentane followed by filtration to give 4 as a red solid (87%). Analysis of the isolated product by ESI-MS(+) revealed a dominant isotopic cluster centered at 1283.8 m/z, corresponding to [4]⁺ (Figure S9), in which the axial ligand remains intact.

Scheme 4.

Synthesis of Complex 4

The UV-vis spectrum of crystalline 4 gives λ_max = 351(sh), 412, 505(sh), and 635 nm in toluene (Figure S10), and the spectrum of 4 generated in situ is nearly the same (Figure 5). The Soret band for 4 is red-shifted by 6 nm compared to the chloride complex 1 (λ_max = 371, 407, 519, and 637 nm). The shoulder in the near-UV for 4 is blue-shifted compared to 1 (348 versus 371 nm), and the shoulder at 519 nm seen for 1 is not observed in the spectrum for 4. These differences are clear markers for the identity of the axial halide donor.

Figure 5.

UV-vis spectra of Fe(X)(ttppc) where X = Cl (blue, dashed line) or F (red, solid line) in toluene at 23 °C.

Dark red crystals of 4 suitable for X-ray structure determination were grown from layering hexane into toluene. The crystal structure of 4 is shown in Figure 6, revealing a five-coordinate, mononuclear iron complex with an axial F⁻ ligand. The structure of 4 represents the first example of a crystallographically characterized iron fluoride corrole complex. The Fe─F bond is disordered over two positions that lie on opposite sides of the corrole plane, leading to a major component (ca. 72%) with an Fe-F distance of 1.972(2) Å, and a minor component with an Fe-F distance of 1.815(5) Å. These distances can be compared to ferric fluoride porphyrin (Fe^III─F = 1.792(3) – 1.865(3) Å), or ferric fluoride 10-thiacorrole (Scor = 2,3,7,8,12,13,17,18-octaethyl-10-thiacorrolato³⁻) (Fe^III–F = 1.898(2) Å).^77-84 The elongated Fe-F distance seen for the more populated component may be due to intramolecular C-H---F interactions with two of the phenyl rings (C42/52(H)---F = 3.196(2)/3.201(2) Å).^85-90 The average Fe−N_pyrrole distance of 1.893(16) Å is in line with other iron ttppc structures (Table S5).¹⁷ The iron atom in 4 is displaced by 0.32 Å from the least-squares plane defined by the 4 N_pyrrole atoms toward the F⁻ ligand in both major and minor components, which is the same amount of displacement seen for Fe(Cl)(ttppc) (0.341 Å).¹⁷

Figure 6.

Displacement ellipsoid plot (50% probability) of 4 at 110 K. Hydrogen atoms are omitted for clarity.

Density functional theory (DFT) calculations were employed to help understand the molecular structures of 1 and 4. Geometry optimized models of 1 and 4 were tested with BP86 functional at the def2-TZVP level of theory with a ground spin state of S = 1. The BP86-optimized structure of 1 has Fe-Cl = 2.195 Å, which is 2.7% shorter than the Fe-Cl distance seen by XRD, whereas for 4 the calculated Fe-F bond length is 1.806 Å, which matches well with the minor component in the crystal structure. The calculated Fe-N, C-C and C-N distances for both complexes were a close match with XRD (Figure S12). Corresponding frequency calculations showed no imaginary frequencies, indicating that the optimized structures are at their ground state energies.

The ¹H NMR spectrum of 4 reveals six paramagnetic peaks in the upfield region between δ = 0 to −35 ppm, and two downfield peaks at 3.99 and 20.01 ppm in C₆D₅CD₃ (Figure 7). For comparison the spectrum for the Fe(Cl) complex is also shown in Figure 7, showing that the spectra are clearly sensitive to the identity of the axial ligand. Earlier NMR work on a series of selectively deuterated iron corroles with different axial halide ligands, Fe(X)(tdcc) (X = I, Br, Cl, F; tdcc = 5,10,15-tris(2,6-dichlorophenyl)corrolato³⁻), gave definitive assignments for the paramagnetic peaks in the ¹H NMR spectra.⁵⁶ Based on these assignments, the peaks at 20.01 ppm and −29.5 ppm for 4 can be attributed to the β-CH pyrrole protons on the C2 and C3 positions, respectively. The ¹⁹F NMR spectrum for 4 does not reveal any signals between −300 and +600 ppm, likely because of fast relaxation of the ¹⁹F nucleus caused by the paramagnetic iron center. The paramagnetic spin ground state for 4 was confirmed by Evans method, giving μ_eff = 2.63 μ_B, consistent with a triplet spin ground state (S = 1; μ_theor = 2.71 μ_B) (Figure S13). The EPR (X-band, 20 K) spectrum of 4 in toluene showed only a baseline signal, as expected for an integer-spin system (Figure S14).

Figure 7.

¹H NMR (400 MHz) spectrum of Fe(X)(ttppc) with X = F (top) and Cl (bottom) in C₆D₅CD₃ at 23 °C.

The zero-field Mössbauer spectra of 1 and 4 in toluene at 80 K consist of well-defined doublets with δ = 0.17 mm s⁻¹; ∣ΔE_Q∣ = 2.76 mm s⁻¹ for 1 and δ = 0.18 mm s⁻¹; ∣ΔE_Q∣ = 2.33 mm s⁻¹ for 4 (Figure 8). A previous report of 1 at 4 K gave δ = 0.18 mm s⁻¹; ∣ΔE_Q∣ = 2.86 mm s⁻¹.¹⁷ A comparison of parameters to other iron chloride corroles, Fe(Cl)(tpc) (δ = 0.19 mm s⁻¹; ∣ΔE_Q∣ = 2.93 mm s⁻¹ at 77 K) and Fe(Cl)(tpfc) (δ = 0.18 mm s⁻¹; ∣ΔE_Q∣ = 2.93 mm s⁻¹ at 80 K), shows that the tri(aryl) meso substituents have only a minor impact on the Mössbauer parameters.^{66, 91}

Figure 8.

Zero-field ⁵⁷Fe Mössbauer spectra of Fe(X)(ttppc) with X = Cl (top), and F (bottom) in toluene at 80 K. Experimental data = black circles, best fit = red line.

The calculated Mössbauer parameters shown in Table 1 were determined from unrestricted single point energy calculations. The calculated values for both the isomer shifts and quadrupole splittings for 1, 3 and 4 align well with the experimental data. Broken symmetry DFT calculations on Fe(X)(tpfc) (X = Cl, F), involving 3 α electrons on Fe and one β electron on the corrole (Fe^III(tpfc+•) configuration), gave Mössbauer parameters of δ = 0.181 mm s⁻¹; ∣ΔE_Q∣ = 2.45 mm s⁻¹ for X = Cl⁻, and δ = 0.194 mm s⁻¹; ∣ΔE_Q∣ = 2.03 mm s⁻¹ for X = F⁻.⁹¹ These data suggest that both the unrestricted and broken symmetry approaches lead to similar modeling of the Mössbauer spectra.

Table 1.

Mössbauer Spectroscopic Parameters (δ [mm s^–1] and ∣ΔE_Q∣ [mm s^–1]): Experimental Data and Calculated Values^a

Complex	δexp	δcalc	∣ΔEQ∣exp	∣ΔEQ∣calc
1 a	0.17	0.178	2.76	2.730
3 b	0.22	0.220	3.36	3.293
4 a	0.18	0.202	2.33	2.343

^aCalculated values from a BP86 geometry optimization followed by a single-point energy calculation at the B3LYP/def2TZVP level of theory. These values were confirmed with the alternate functional TPSSh (Table S6).

^bStructure from XRD.

Electrochemistry.

Cyclic voltammograms of the halide complexes are shown in Figure 9. For the chloride complex an irreversible reduction at E_red = −0.57 V along with a small oxidation peak at −0.06 V and a reversible oxidation at E_1/2 = 0.50 V was observed with a scan rate of 50 mV s⁻¹. As the scan rate was increased by 50 mV increments, an oxidative wave grows in at −0.54 V, giving a reversible reduction at E_1/2 = −0.58 V, as shown in Figure 9 for a scan rate of 200 mV s⁻¹. We assign this reversible wave to the first reduction for [Fe(Cl)(ttppc)]⁰/[Fe(Cl)(ttppc)]⁻. The irreversibility at slower scan rates is consistent with dissociation of the chloride ligand upon reduction of the complex. The spectrum for the fluoride complex 4 reveals a quasi-reversible reduction at E_1/2 = −0.44 V and a reversible oxidation at E_1/2 = +0.48 V. There are also smaller redox events centered at −0.05 V and −0.10 V which may come from minor species such as Fe(OH)(ttppc) generated by reaction of the Fe(F) complex with residual water.¹⁷ The CV data show that the identity of the halide ligand has a significant effect on the first reduction potential, but not on the first oxidation potential. The reduction wave for the fluoride complex is shifted 140 mV positive compared to the chloride complex, consistent with the greater electronegativity of F versus Cl, and indicates that complex 4 is easier to reduce than complex 1. Interestingly, the analogous hydroxide species, Fe(OH)(ttppc), appears to exhibit a reversible reduction at −0.11 V in PhCN, which is shifted even more positive by 330 mV when compared to the fluoride complex and makes the Fe(OH) complex the strongest oxidant of the series.¹⁷

Figure 9.

Cyclic voltammograms of 1 (2 mM) with a scan rate of 50 mV/s (black) and 200 mV/s (blue) (top), and 4 (3 mM) with a scan rate of 100 mV/s (black) in benzonitrile (bottom). The solution contains Bu₄N⁺PF₆⁻ (0.25 M) supporting electrolyte.

Reaction of Fe(X)(ttppc) (X = Cl, F) with Triphenylmethyl Radical.

Initial studies began with the unsubstituted triphenylmethyl radical, delivered from the stable dimer form (Ph₃C=C₆H₅CPh₂). This dimeric starting material is in equilibrium with the monomeric radical form in solution (Ph₃C=C₆H₅CPh₂ ⇆ 2 Ph₃C• (3.5% monomer in toluene at 20 °C, λ_max, nm (ε = L mol⁻¹ cm⁻¹) = 515 (661)).^{92, 93} The addition of excess Ph₃C=C₆H₅CPh₂ to 1 or 4 in toluene at 23 °C (Scheme 5) resulted in the complete conversion of the starting Fe complexes to an Fe^III complex characterized by λ_max at 420, 575, 760 nm for 1 and 420, 573, 762 nm for 4 within 1 min by UV-vis (Figure 10).¹⁷

Scheme 5.

Reaction of Fe(X)(ttppc) (X = Cl or F) with Triphenylmethyl Radical

Figure 10.

UV-vis spectra for the reaction of Fe(X)(ttppc) where X = Cl (top) or F (bottom) with Ph₃C=C₆H₅CPh₂ in toluene at 23 °C. Spectra before (blue) and after (red) the addition of excess Ph₃C=C₆H₅CPh₂.

Analysis of the reaction mixture for the chloride complex 1 and Ph₃C• showed the formation of both Fe^III(ttppc) and Ph₃CCl by ¹H NMR spectroscopy. However, quantitation of the organic product Ph₃CCl could not be obtained by NMR integration methods because the broadened peaks for Ph₃CCl overlap with the starting (Ph₃C=C₆H₅CPh₂) dimer. However, running the reaction on a larger scale (20 mM) produced enough of the chlorinated Ph₃CCl to isolate by chromatography. Slight warming of the reaction to 50 °C on this scale was employed to drive the reaction to completion within 5 h. Conversion of 1 to Fe^III(ttppc) was confirmed by NMR spectroscopy (Figure S16), and thin-layer chromatography (TLC) revealed a major spot for the expected Ph₃CCl (r_f = 0.34). Partial purification by silica gel chromatography (eluent: 6/1 v/v hexane/ethyl acetate) led to isolation of Ph₃CCl together with an amount of Ph₃COH impurity, which co-elutes with the chlorinated product (Figure S17). A yield of 33% for Ph₃CCl was obtained (average of 3 runs) by ¹H NMR analysis and comparison with an internal standard.

The reaction of the iron fluoride complex 4 with Ph₃C• was analyzed by both ¹H and ¹⁹F {¹H} NMR spectroscopies, and showed formation of Fe^III(ttppc) and Ph₃CF products (Figure S18 - S22). In this case, quantitation of the fluorinated product could be carried out by integration of the ¹⁹F {¹H} NMR spectrum, avoiding the problem of overlap with the starting material (Ph₃C=C₆H₅CPh₂) material seen for the chloride reaction, and making it possible to obtain the yield of Ph₃CF in situ. A sharp peak for Ph₃CF at −125.6 ppm in the ¹⁹F {¹H} NMR spectrum was integrated to give a yield of 82% (average of 2 runs).

Reaction of Fe(X)(ttppc) (X = Cl, F) with Tris-(para-Methoxyphenyl)methyl Radical.

Addition of the para-methoxy-substituted radical derivative ((p-MeO-C₆H₄)₃C•) to 1 led to similar reactivity. Production of (p-MeO-C₆H₄)₃CCl was confirmed by GC-FID, and a yield for (p-MeO-C₆H₄)₃CCl of 81% (average of 3 runs) was obtained (Figure S23). Mössbauer spectroscopy on the reaction mixture confirms the formation of an iron(III) corrole (δ = 0.18 mm s⁻¹; ∣ΔE_Q∣ = 3.94 mm s⁻¹) (Figure 11).^{81, 91, 94-99}

Figure 11.

Zero-field ⁵⁷Fe Mössbauer spectrum (C₆H₅CH₃, 80 K) of 1 (top), and 1 + (p-MeO-C₆H₄)₃C• (bottom). Experimental data = black circles, best fit = red line.

The reaction of the fluoride complex 4 was also examined with (p-MeO-C₆H₄)₃C• by ¹H and ¹⁹F {¹H} NMR spectroscopy. Product analysis by ¹H NMR spectroscopy confirms the complete conversion of the starting fluoride complex to the ferric product (Figure S24). The ¹⁹F {¹H} NMR shows a sharp peak at −119.8 ppm corresponding to the fluorinated product (p-MeO-C₆H₄)₃CF. Quantitation by integration of the ¹⁹F {¹H} NMR spectrum against an internal standard gave a yield for (p-MeO-C₆H₄)₃CF of 93% (average of 2 runs) (Figure S25).

Kinetics.

The rate of the reaction for 1 and (p-MeO-C₆H₄)₃C• under pseudo-first-order conditions (excess radical) was examined. The addition of (p-MeO-C₆H₄)₃C• to 1 in toluene resulted in the rapid, isosbestic conversion of 1 to Fe^III(ttppc), as seen by UV-vis spectroscopy (Figure 12b). The reaction was monitored by measuring corresponding growth curve for Fe^III(ttppc) at λ_max = 573 nm, and the best fit led to a pseudo-first-order rate constant (k_obs) (Figure S27 - S28). Measuring k_obs at different radical concentrations (27 – 79 equiv at 23 °C and 19 – 33 equiv at −60 °C) led to a linear correlation between rate constant and [radical], which gave a second-order rate constant (k₂) from the slope of the best-fit line (Figure 12c and Figure S28), with k₂ = 1344.6 ± 34.1 M⁻¹ s⁻¹ at 23 °C and 2.183 ± 0.171 M⁻¹ s⁻¹ at −60 °C. The observed second-order rate constant is ~4 times faster than that seen for Fe(OH)(ttppc) with (p-MeO-C₆H₄)₃C• at 23 °C.¹⁷

Figure 12.

(a) Overlay of UV–vis spectra for 1 (blue, solid line), 1 + (p-MeO-C₆H₄)₃C• (red, dashed line), and (p-MeO-C₆H₄)₃C• (green, dotted line) in toluene. (b) UV-vis spectra for reaction of 1 (32 μM) with (p-MeO-C₆H₄)₃C• (62 equiv) in toluene from 0 (blue) to 2 sec (red) at 0.04 s intervals (grey) at 23 °C. Inset: Changes in absorbance versus time for the formation of Fe^III(ttppc) (black dots) fit to a single exponential expression (red line). (c) Plot of k_obs (23 °C) versus [(p-MeO-C₆H₄)₃C•] with best-fit line. (d) Eyring plot of (ln(kh/k_BT) versus 1/T (−70 °C to 23 °C) with best-fit line.

Temperature-dependent studies of the (p-MeO-C₆H₄)₃C• reaction with 1 (−70 to 23 °C) gave the Eyring plot shown in Figure 12d, where k = k_obs/[radical]. The plot is linear and gives activation parameters: ΔH^⧧ = +9.77 ± 0.29 kcal mol⁻¹ and ΔS^⧧ = −14.39 ± 1.29 cal mol⁻¹ K⁻¹. These values can be compared to the activation parameters found for the tBu-substituted radical (p-tert-butyl-C₆H₄)₃C• in reactions with Fe(OH)(ttppc) (ΔH^⧧ = 13.4 kcal mol⁻¹; ΔS^⧧ = −5.6 cal mol⁻¹ K⁻¹) or Mn(OH)(ttppc) (ΔH^⧧ = 10.9 kcal mol⁻¹; ΔS^⧧ = −15.5 cal mol⁻¹ K⁻¹). The faster rate of Cl• transfer versus •OH transfer can be associated with a lowering of the enthalpic contribution to the reaction barrier by 3.63 kcal mol⁻¹, which compensates for an additional entropic penalty (−TΔS^⧧(Cl) = 4.3 kcal mol⁻¹; −TΔS^⧧(OH): 1.7 kcal mol⁻¹; T = 298 K). This results suggests that the Fe-Cl bond in 1 may be more susceptible to homolytic dissociation, i.e. has a lower bond dissociation energy, than the Fe-OH bond in Fe(OH)(ttppc).

The negative ΔS^⧧ value for 1 lies in between that seen for the reactions of Fe(OH)(ttppc) with (p-tert-butyl-C₆H₄)₃C• (ΔS^⧧ = −5.6 cal mol⁻¹ K⁻¹) and (p-CN-C₆H₄)₃C• (ΔS^⧧ = −28.4 cal mol⁻¹ K⁻¹), and is consistent with a concerted mechanism, as opposed to a step-wise mechanism with outer-sphere electron-transfer (ET) as the rate-determining step.¹⁹ Early work on reactions of metalloporphyrins with sterically encumbered alkyl halides showed that a concerted, S_N2-like mechanism could be distinguished from an outer-sphere ET mechanism by the observation of a negative ΔS^⧧ values for the S_N2 pathway, as compared to a small, positive ΔS^⧧ value for an ET pathway.¹⁰⁰

To gain further insight regarding the role of electron-transfer in these reactions, the redox potential for the [(p-MeO-C₆H₄)₃C]⁺/(p-MeO-C₆H₄)₃C•] couple was examined. This couple is reported as −0.58 V in CH₃CN,¹⁰¹ but in order to make a direct comparison with 1, we re-measured it in PhCN. Cyclic voltammetry revealed a single, reversible wave at E_1/2^p-OMe = −0.60 V (Figure S30). Comparison of E_1/2^p-OMe with the reduction potential for 1 in PhCN (E_1/2 = −0.58 V) gives ΔG°_ET = −0.46 kcal mol⁻¹ for an outer-sphere electron-transfer reaction, which is slightly exergonic. In contrast, the redox couple for the p-tBu radical derivative is E_1/2^p-tBu = −0.24 V (in CH₃CN).¹⁰¹ This value predicts that ET would be largely endergonic with 1, with ΔG° = +7.84 kcal mol⁻¹. In accordance with these thermodynamic trends, no reaction is observed for the p-tBu radical derivative with 1 (Figure S31 - S32). Taken together, these observations suggest that there is a modest amount of charge-transfer associated with the concerted transition state for halogen transfer, as implicated by computational studies for Fe(OH)(ttppc).¹⁹

The addition of (p-MeO-C₆H₄)₃C• to 4 in toluene resulted in the rapid isosbestic conversion of 4 to Fe^III(ttppc), as seen for 1 (Figure 13a). Measuring k_obs at different radical concentrations led to a second-order rate constant of k₂ = 115867 ± 6407 M⁻¹ s⁻¹ at 23 °C (Figure 13b). This second-order rate constant is ~86 times faster than that seen for Fe(Cl)(ttppc), and ~313 times faster than Fe(OH)(ttppc).¹⁷ The significantly enhanced reaction rate of the fluoride complex as compared to the chloride complex correlates with their respective redox potentials, in which 4 exhibits a reduction wave that is 0.14 V more positive than 1 (Figure 9). However, comparison with the Fe(OH)(ttppc) complex does not follow this trend; the reported value of E_½ = −0.11 V for the hydroxide complex indicates it is significantly more oxidizing than either 4 or 1. It is not clear at this time as to the origins of the rate differences with the OH complex, but it suggests that the rate of hydroxyl group transfer may be influenced more by other factors, including possible entropic effects, as compared to the redox potential.

Figure 13.

(a) UV-vis spectra for reaction of 4 (12.5 μM) with (p-MeO-C₆H₄)₃C• (31 equiv) in toluene from 0 (blue) to 0.1 s (red) at 0.0015 s intervals (grey) at 23 °C. Inset: Changes in absorbance versus time for the formation of Fe^III(ttppc) (black dots) fit to a single exponential expression (red line). (b) Plot of k_obs versus [(p-MeO-C₆H₄)₃C•] with best-fit line.

Conclusions.

The synthesis and characterization of the new corrole complexes 2 – 4 were described. Complexes 2 and 4 are the first structurally characterized triflate- and fluoride-ligated iron corroles, respectively. The triflate-ligated complex represents a new synthon for accessing a wide range of iron corroles with various axial donors. The reactivity of 1 and 4 with triphenylmethyl radical derivatives shows that iron corroles are capable of mediating halogen transfer to carbon radicals via a process that mimics the rebound step seen for heme or porphyrin-catalyzed hydroxylation. The mechanism of halogen transfer appears to be in line with a concerted mechanism, as described previously for hydroxyl transfer with Fe and Mn corroles, and in which partial charge-transfer occurs in the transition state.¹⁹ From the kinetic studies, it is clear that the rate of group transfer from iron corrole to a tertiary carbon radical occurs in the following order: F > Cl > OH. These findings could be important for the future design of iron corroles, or other porphyrinoid species, as catalysts for either halogenation or hydroxylation of organic substrates. They also suggest that the redox potentials of the iron complexes are only one piece of the puzzle when predicting the reactivity of these radical transfer reactions.

Metallocorroles are promising candidates for oxidative C-H functionalization, and the sterically encumbered ttppc ligand has already shown utility in shape-selective, catalytic epoxidations.¹⁰² Our current findings suggest that base metal corroles (e.g. Fe, Mn) may be excellent candidates for the development of efficient and selective halogenation catalysts because of their ability to undergo facile, concerted transfer of an equivalent of X• (X = Cl, F) to carbon radicals. The high-valent metal-halide intermediates can be expected to exhibit relatively modest reduction potentials and good stability, helping to avoid oxidative decomposition, or the non-productive oxidation of substrates. In particular, the sterically encumbered ttppc complexes may exhibit desirable patterns of selectivity in the catalytic halogenation of C-H bonds. Future efforts in our laboratory will test these ideas.

Scheme 6.

Reaction of 4 with Tris-(para-Methoxyphenyl)methyl Radical

Synopsis:

The new triflate- and fluoride-ligated iron corroles Fe(OTf)(ttppc), Fe(OTf)(ttppc)(AgOTf), and Fe(F)(ttppc) were synthesized, and the latter two complexes were characterized by single crystal X-ray diffraction. The Fe(F)(ttppc) reacts with tertiary carbon radicals to give the fluorinated product and the one-electron reduced Fe^III(ttppc), in a process that models the rebound step in halogenases and hydroxylases. Kinetic studies provided a direct comparison of reaction rates and mechanism between Fe(X)(ttppc) (X = F, Cl, OH) species.