Synthesis and characterization of trigonal bipyramidal Fe^III complexes and their solution behavior

Abstract

A series of air-stable trigonal bipyramidal Fe^III complexes supported by a redox non-innocent NNN pincer ligand, Cz^tBu(Pyr^R)₂⁻ (R = iPr, Me, or H), were synthesized, fully characterized, and utilized for the investigation of the interaction between acetone and the Fe^III center. The magnetic moments determined from the paramagnetic ¹H NMR spectra in conjunction with EPR and Mössbauer spectroscopy indicate the presence of a high-spin ferric center. Cyclic voltammetry studies feature two quasi-reversible events corresponding to a metal-centered Fe^III/II reduction around −0.40 V (vs. Fc) and a ligand-centered Cz^tBu(Pyr^R)₂/Cz^tBu(Pyr^R)₂^•+ oxidation at potentials near +0.70 V (vs. Fc). UV-Visible spectroscopy in CH₂Cl₂ showcases ligand-metal charge transfer (LMCT) bands, as well as coordination of acetone to Cz^tBu(Pyr^H)₂FeCl₂. In situ IR spectroscopy and solution conductivity (κ) measurements of Cz^tBu(Pyr^R)₂FeCl₂ with varied equivalents of acetone reveal that acetone is weakly associated with the iron center.

Introduction

Iron is the most abundant transition metal in the earth’s crust, making it a highly attractive element from which reagents and catalysts can be readily developed [1]. In particular, a great deal of effort has been focused on the development of FeCl₃ as both a stoichiometric reagent and a catalyst [2]. Often, reactions with FeCl₃ are performed in chlorinated solvents (i.e. CHCl₃, CH₂Cl₂, ClCH₂CH₂Cl (DCE), etc.). Under anhydrous conditions, FeCl₃ is largely insoluble in these solvents, and systems can require the addition of silica gel as a solid support [3], Lewis basic additives like nitromethane [4], or ligand systems [5]. Previous studies from our lab have shown that in the absence of these additives, the substrate must interact directly with heterogeneous FeCl₃, resulting in the substrate increasing the solubility of the Lewis acid via the formation of a Lewis pair [6]. However, this interaction changes dramatically when FeCl₃ is employed as a catalyst, allowing for the formation of complex Fe^III-centered aggregates. We report herein the characterization of complexes derived from FeCl₃ and NNN pincer ligands. Full ground state characterization data are provided to analyze the impact of the ligand system via electrochemical and spectroscopic analysis, as well as how the ligand perturbs the behavior of these systems in the presence of acetone in solution.

Previous efforts from our lab have focused on the examination of the impact of NNN pincer ligands on the behavior of first-row transition metals. We have examined, in depth, complexes generated with Ni^II [7], VI^II/IV [8], as well as an Fe^II complex that was able to undergo C–H activation to yield an unusual pyrazolide-bridged Fe^II complex at elevated temperatures [9]. We sought to examine the impact of our ligand system on Fe^III because of a report by Nishiyama and coworkers, who employed a structurally analogous NNN pincer ligand to facilitate the asymmetric hydrosilylation of ketones to alcohols in THF [10]. What further intrigued us about their results was that the starting material, product, and solvent are all competent Lewis bases that can compete for access to the Lewis acid. With these observations in mind, we sought to study how our NNN pincer system impacts the solubility of Fe^III chloride complexes and how it may facilitate similar intermolecular interactions.

Experimental Section

Materials and Methods.

All manipulations were performed under a nitrogen atmosphere using standard Schlenk techniques or in an M. Braun UNIlab Pro glovebox. Glassware was dried at 150 °C overnight. Diethyl ether, n-pentane, tetrahydrofuran, and toluene were purified using a Pure Process Technology solvent purification system. Before use, an aliquot of each solvent was tested with a drop of sodium benzophenone ketyl in THF solution. All reagents were purchased from commercial vendors and used as received. HCz^tBu(Pyr^iPr)₂ was prepared according to a modified literature procedure [7]. ¹H NMR data were recorded on a Varian Inova 500 MHz spectrometer at 22 °C. Resonances in the ¹H NMR spectra are referenced to residual CD₂Cl₂ at δ = 5.32 ppm or CDCl₃ at δ = 7.27 ppm. Solution magnetic susceptibilities were determined using the Evans method [11]. Continuous-wave (CW) EPR spectra were recorded at 77 K on a Bruker EMX plus X-band EPR spectrometer equipped with a liquid N₂ cold-finger Dewar flask. Cyclic voltammetry was conducted via a CH-Instruments electrochemical analyzer (model 620E), employing a 3 mm glassy carbon working electrode, a silver wire pseudo reference electrode, and a platinum coiled wire counter electrode. All measurements were performed using either CH₂Cl₂ or THF solutions containing 1 mM analyte and 0.1 M n-Bu₄NPF₆ as the supporting electrolyte. The potentials were referenced to a ferrocene/ferrocenium redox couple. Elemental analyses were conducted by Midwest Microlab, LLC (Indianapolis, IN).

Synthesis of HCz^tBu(Pyr^H)₂.

10.0 g (23 mmol) of 1,8-dibromo-3,6-di-tert-butyl-9H-carbazole, 7.8 g (114 mmol) of 1H-pyrazole, 12.8 g (114 mmol) of potassium tert-butoxide (KO^tBu), and 2.3 g (20 mmol) of N,N,N,N-tetra-methyl-ethylenediamine (TMEDA) were dissolved in 300 mL dimethylformamide (DMF). The yellow DMF slurry was degassed by three freeze-pump-thaw cycles. 2.9 g (20 mmol) of copper(I) oxide (Cu₂O) was added and the reaction was refluxed at ~150 °C for 2 days under inert N₂ atmosphere. The resulting brown slurry was dissolved in 200 mL dichloromethane, washed with 1 M HCl (3×200 mL), and then vigorously stirred with 200 mL 1 M HCl overnight. The organic layer was again washed with 3×200 mL of 1M HCl, 3×200 mL of 1 M NH₄OH, and 2×200 mL of 1 M NH₄Cl solutions. The dichloromethane layer was collected, dried using MgSO₄, and solvent was removed by rotary evaporation. The resulting light brown solids were crystallized from a concentrated n-hexane solution at −20 °C to yield the product as off-white solids (4.9 g, 52%). ¹H NMR (500 MHz, CDCl₃, δ): 1.54 (C(CH₃)₃, s, 9H), 6.57 (ArH, s, 1H), 7.61 (ArH, s, 1H), 7.93 (ArH, s, 1H), 8.09 (ArH, s, 1H), 8.17 (ArH, s, 1H), 11.31 (NH, s, 1H). Anal. Calcd for molecular formula C₂₆H₂₉N₅: C 75.88, H 7.10, N 17.02. Found: C 75.93, H 7.06, N 16.85.

Synthesis of HCz^tBu(Pyr^Me)₂.

15.0 g (34 mmol) of 1,8-dibromo-3,6-di-tert-butyl-9H-carbazole, 8.44 g (103 mmol) of 3-methylpyrazole, 11.54 g (103 mmol) of KO^tBu, and 3.50 g (30 mmol) of TMEDA were dissolved in 370 mL of DMF. The yellow DMF slurry was degassed by 3 freeze-pump-thaw cycles. 5.39 g (38 mmol) of Cu₂O was added, and the reaction was refluxed at ~150 °C for 3 days under inert N₂ atmosphere. The resulting red/brown slurry was dissolved in 500 mL diethyl ether (Et₂O) and washed with 3×200 mL of 1 M HCl, 3×200 mL of 1 M NH₄OH, and 2×200 mL of 1M NH₄Cl solutions. The organic layer was dried with MgSO₄, and volatiles were removed in vacuo. The resulting solids were crystallized from a hot n-hexane solution to yield fluffy off-white solids (9.0 g, 60%). ¹H NMR (500 MHz, CDCl₃, δ): 1.50 (C(CH₃)₃, s, 9H), 2.54 (CH₃, s, 3H), 6.33 (ArH, d, J = 2.4, 1H), 7.52 (ArH, d, J = 1.5, 1H), 8.00 (ArH, d, J = 1.0, 1H), 8.04 (ArH, d, J = 2.4, 1H), 11.56 (NH, s, 1H). Anal. Calcd for molecular formula C₂₈H₃₃N₅: C 76.50, H 7.57, N 15.93. Found: C 76.46, H 7.51, N 16.02.

Synthesis of Cz^tBu(Pyr^iPr)₂FeCl₂ (1).

To 1.00 g (0.21 mmol) of HCz^tBu(Pyr^iPr)₂ dissolved in 5 mL of THF at room temperature under inert N₂ atmosphere was added 0.23 g (0.21 mmol) of lithium diisopropylamide (LDA) in 5 mL of THF. The resulting fluorescent yellow mixture was stirred for 1 h. The fluorescent mixture was then added to 0.36 g (0.22 mmol) of FeCl₃ in 50 mL of THF and stirred overnight at ambient temperature. Volatiles were removed by rotary evaporation to afford dark green solids, which were washed with 30 mL of n-hexane and collected by vacuum filtration over a celite pad. The remaining green solids were dissolved in 100 mL of CH₂Cl₂ and dried to yield the final dark green product. Crystals suitable for X-ray diffraction were grown from a concentrated toluene solution at room temperature. Yield: 1.2 g, 98%. ¹H NMR (500 MHz, CDCl₃, δ): 6.23 (C(CH₃)₃), 16.71 (CH(CH₃)₂), 36.87 (ArH), 69.24 (ArH), 75.16 (ArH), 89.05 (ArH). μ_eff (CDCl₃) = 5.5(3) μB. Anal. Calcd for molecular formula C₃₂H₄₀Cl₂FeN₅: C 61.85, H 6.49, N 11.27. Found: C 62.14, H 6.44, N 11.37. UV–Vis (CH₂Cl₂) λ_max, nm (ε): 356 (12390), 427 (sh, 2230), 743 (3450). UV–Vis (acetone) λ_max, nm (ε): 356 (10480), 420 (sh, 2120), 728 (2590).

Synthesis of Cz^tBu(Pyr^Me)₂FeCl₂ (2).

The procedure was adapted from the synthesis of 1 using HCz^tBu(Pyr^Me)₂ (1.00 g, 0.23 mmol), LDA (0.26 g, 0.24 mmol), and FeCl₃ (0.41 g, 0.25 mmol)_. Crystals suitable for X-ray diffraction were grown from a concentrated toluene solution at room temperature. Yield: 1.2 g, 92%. ¹H NMR (500 MHz, CDCl₃, δ): 6.30 (C(CH₃)₃), 37.39 (ArH), 68.47 (ArH), 78.41(ArH), 88.71 (ArH). μ_eff (CDCl₃) = 5.4(2) μB. Anal. Calcd for molecular formula C₂₈H₃₂Cl₂FeN₅: C 59.49, H 5.71, N 12.39. Found: C 59.71, H 5.50, N 12.19. UV–Vis (CH₂Cl₂) λ_max, nm (ε): 357 (14060), 422 (sh, 2590), 739 (3790). UV–Vis (acetone) λ_max, nm (ε): 351 (13590), 427 (sh, 2650), 714 (3890).

Synthesis of Cz^tBu(Pyr^H)₂FeCl₂ (3).

The procedure was adapted from the synthesis of 1, but a more dilute concentration is required. To 1.00 g (0.24 mmol) of HCz^tBu(Pyr^H)₂ dissolved in 10 mL of THF at room temperature under an inert N₂ atmosphere was added 0.27 g (0.25 mmol) of LDA in 10 mL of THF. The resulting fluorescent yellow mixture was stirred for 1 hour. The fluorescent mixture was then added to 0.43 g (0.26 mmol) of FeCl₃ in 100 mL of THF and stirred overnight at ambient temperature. Volatiles were removed by rotary evaporation to afford a dark green solid. The dried solids were washed with 30 mL of n-hexane and collected by vacuum filtration over a celite pad. The remaining green solids were dissolved in 150 mL of CH₂Cl₂ and dried to yield the final dark green product. Crystals suitable for X-ray diffraction were grown from a concentrated toluene solution at room temperature. Yield: 1.3 g, 99%. ¹H NMR (500 MHz, CD₂Cl₂, δ): 5.53 (C(CH₃)₃), 32.37 (ArH), 66.33 (ArH), 80.65 (ArH), 90.81 (ArH). μ_eff (CD₂Cl₂) = 5.3(2) μB. Anal. Calcd for molecular formula C₂₆H₂₈Cl₂FeN₅: C 58.12, H 5.25, N 13.03. Found: C 58.17, H 5.17, N 12.99. UV–Vis (CH₂Cl₂) λ_max, nm (ε): 349 (11260), 374 (sh, 8670), 405 (sh, 3490), 680 (3480). UV–Vis (acetone) λ_max, nm (ε): 359 (7470), 486 (sh, 790), 931 (1650).

Crystallography.

Single crystal data were collected on either a Mo Kα wavelength (λ = 0.71073 Å) Bruker Quest diffractometer with a fixed chi angle, a sealed tube fine focus X-ray tube, single crystal curved graphite incident beam monochromator, and a Photon100 CMOS area detector (complex 1) or on a Cu Kα wavelength (λ = 1.54178 Å) Bruker Quest diffractometer with kappa geometry, a I-μ-S microsource X-ray tube, laterally graded multilayer (Goebel) mirror for monochromatization, a Photon2 CMOS area detector (complex 3). Both instruments were equipped with Oxford Cryosystems low temperature devices and examination and data collection were performed at 100 K (for 1) 150 K (for 3). Data were collected, reflections were indexed and processed, and the files scaled and corrected for absorption using APEX3 [12] and SADABS [13] or TWINABS [13]. Space groups were assigned, and the structures were solved by direct methods using XPREP within the SHELXTL suite of programs [14] and refined by full matrix least squares against F2 with all reflections using Shelxl2014 [15] or Shelxl2018 [14] using the graphical interface Shelxle [16]. Additional data collection and refinement details, including hydrogen atom treatment, description of disorder, twinning and pseudosymmetry (where present) can be found in the Supporting Information. Complete crystallographic data, in CIF format, have been deposited with the Cambridge Crystallographic Data Centre. CCDC 1996921–1996922 contain the supplementary crystallographic data for this paper. These data can be obtained free of charge from The Cambridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_request/cif.

Computational Details.

Geometry optimization was carried out using Spartan ‘16 [17]. The DFT modelling method, using the hybrid B3LYP functional with the 6–31G* basis set, was used to calculate the model complexes 1’−3’ and [3’-acetone]⁺, where the tBu groups were truncated to an H atom. Geometry optimization was carried out until global minima were achieved. The results of the optimized structures are reported in Tables S4–S7 and are comparable with X-ray diffraction results. The atomic contributions to frontier molecular orbitals and simulated electronic spectra were calculated using the Chemissian program [18].

Results and Discussion

Synthesis and Characterization of Complexes 1–3

Dark green complexes 1–3 can be prepared in high yield (>90%) by deprotonation of HCz^tBu(Pyr^R)₂ (R = iPr (1), Me (2), H (3)) using LDA followed by transmetallation with FeCl₃ at ambient temperature in a N₂ atmosphere (Scheme 1). Unlike 1 and 2, formation of 3 is dependent upon concentration during synthesis based on 1H NMR spectroscopy; since the coordination site of 3 is more exposed than in 1 or 2, the use of dilute concentrations prevents the formation of undesired byproducts, e.g. [{Cz^tBu(Pyr^H)₂}₂Fe]Cl. ¹H NMR spectroscopy of 1–3 shows broad, paramagnetically shifted resonances ranging from 5.53 to 90.81 ppm in various deuterated solvents. Note that one resonance is missing for each complex, which could be due to the fast relaxation of protons that are close to the Fe^III center.

Scheme 1.

Synthesis of 1–3.

Solution magnetic measurements in CDCl₃ (for 1 and 2) and CD₂Cl₂ (for 3) at room temperature (298 K) using Evans method gave μ_eff = 5.3(2), 5.4(2), and 5.5(3) μB for 1–3, respectively, all of which are slightly lower than the expected value for high-spin FeIII systems and may be caused by mixing of another spin state. Complexes 1–3 are highly soluble in various organic solvents and 1 is even soluble in n-pentane. In the solid state, 1–3 are all dark green powders. Interestingly, 1 and 2 exhibit the same color when dissolved in both coordinating (acetone) and non-coordinating (CH₂Cl₂) solvents; whereas, we observed dark green and teal solutions when dissolving 3 in CH₂Cl₂ and acetone, respectively (vide infra). Finally, complexes 1–3 are air-stable indefinitely and resistant to decomposition upon heating up to 80 °C.

Crystal Structures of 1 and 3

We obtained single crystals suitable for X-ray structure determination for complexes 1 and 3 (Fig. 1) [19]. The solid-state structures confirm complexes 1 and 3 as mononuclear iron complexes with one NNN pincer ligand in association with two Cl atoms. Both 1 and 3 have a slightly distorted trigonal bipyramidal geometry (τ₅ values are 0.86 and 0.83, respectively) with two neutral pyrazole-nitrogen atoms (N_pyr) in the axial positions and three anionic atoms in the equatorial plane [20]. The structural parameters of 1 and 3 agree well with those of closely related carbazolide-based pincer-supported Fe^III chloride complexes, except for slightly shorter iron to carbazolide nitrogen (N_cz) distances [21]. This could be attributed to the stronger electron-donating tert-butyl groups, compared to previously reported methyl and phenyl groups on the carbazolide backbone. Different from complex 3, the carbazolide and two flanking pyrazole rings in 1 are not coplanar. This may be caused by steric hindrance resulting from the bulkier iso-propyl substituents or close contact of the iPr C−H groups to the Cl ligands [22].

Fig. 1.

Molecular structure of 1 with thermal ellipsoids at the 50% probability level. Hydrogen atoms and solvent molecules are omitted for clarity. Color key: orange = Fe, blue = N, gray = C, green = Cl [19].

The dihedral angles between the pyrazole groups and the carbazolide ring are 28.7(9)° and 0° for 1 and 3, respectively (Fig. 2). Surprisingly, we observed a lesser distortion (17.7(2)°) for the V^III congener, Cz^tBu(Pyr^iPr)₂VCl₂,[8] despite comparable ionic radii for both Fe^III and V^III [23].

Fig. 2.

Overlaid structure of complexes 1 (blue) and 3 (red). Spheres represent Fe atoms. Hydrogen atoms and tert-butyl groups are omitted for clarity [19].

Electrochemical Studies

Cyclic voltammetry analysis of the Fe^III complexes carried out in CH₂Cl₂ and acetone solutions using 0.1 M n-Bu₄NPF₆, a platinum wire auxiliary electrode, glassy carbon working electrode, and Ag wire reference electrode gave insight into the electrochemical behavior. We varied scan rates from 100 mV s⁻¹ to 500 mV s⁻¹ and referenced to [FeCp₂]/[FeCp₂]⁺. In CH₂Cl₂, two distinct quasi-reversible events represent a diffusion-controlled process for complexes 1−3 (Fig. 3): the peak current is proportional to the square root of the scan rate (Fig. 4) [19].

Fig. 3.

Cyclic voltammograms of 0.1 mM of 1 (blue), 2 (green), and 3 (red) in CH₂Cl₂ (0.1 M n-Bu₄PF₆) at scan rates of 100 mV s⁻¹ [19]. The arrow indicates the initial scan direction.

Fig. 4.

Cyclic voltammograms of 0.1 mM 3 in CH₂Cl₂ (0.1 M n-Bu₄PF₆) at scan rates of 100, 200, 300, 400, and 500 mV s⁻¹. Inset: Plot of anodic, I_pa, and cathodic, I_pc, peak current, versus square root of scan rate for the first (•, blue; R² = 0.9992) and second oxidation (◆, red; R² = 0.9650) [19]. The arrow indicates the initial scan direction.

Complex 3 exhibited E_1/2 values of −0.41 and +0.72 V, which are assigned as metal-centered (Fe^III/II) and ligand-centered Cz^tBu(Pyr^H)₂/Cz^tBu(Pyr^H)₂^•+ redox couples, respectively. The plots of v^1/2 vs. the I_pa and I_pc for both events suggest these electron transfer processes are governed by diffusion control.

Voltammograms similar to those for 3 obtained for 1 and 2 yielded less negative redox potentials (−0.38 V for 1 and −0.39 V for 2) for the metal-centered redox step, which are conversely related to the increasing electron donation from the pyrazole substituents to the Fe center. The same unusual trend shared by 1 and 2 is present for the ligand-centered oxidation, where the redox potentials of 1 (0.70 V) and 2 (0.69 V) are less than 3 (0.72 V). This contradiction could be attributed to the poor overlap between the p orbital of N_cz and d orbital of Fe, resulting from the increased out-of-plane distortion of the Fe from the carbazolide plane in 1. Complex 3 is the most reversible while 1 is the least. The decrease of reversibility of 1 is consistent with out-of-plane distortion. In acetone, complexes 1−3 exhibit more complicated redox events, which could be due to multiple species in solution, owing to the solvent-specific coordination of acetone to the Fe center.

Electron Paramagnetic Resonance and Mössbauer Spectroscopy

EPR spectroscopy of 1−3 was conducted in both the solid and solution state at room temperature and 77 K [19]. In the solid state, complex 1 gives spectra with g values near 4.3 and 2.0 at room temperature and 77 K. When a CH₂Cl₂ solution of 1 is measured at room temperature, a large g = 2.0 signal is accompanied by a diffuse wing stretching at g = 4.3 (Fig. 5); whereas, the intensity of the signal at g = 4.3 is larger than the g = 2.0 signal at 77 K implying the presence of a high-spin Fe^III center.

Fig. 5.

EPR spectra of 1 in CH₂Cl₂ at room temperature (top) and 77 K (bottom) [19].

Room temperature EPR measurement of complexes 2 and 3 as solids and solutions exhibit spectra with a g value near 2. The solid-state spectra are considerably broader in comparison to the solution state spectra, which may be due to multiple transitions coinciding at the same time resulting in an imperfect rhombic shaped signal. Measurement of 2 and 3 as solids and CH₂Cl₂ glasses at 77 K gave spectra with similar g ≈ 2 values with an additional small signal around g ≈ 4, which is typical of a high-spin Fe^III center [19].

The differences in EPR spectra of 1, 2, and 3 in CH₂Cl₂ can be attributed to differences in overall steric bulk and the corresponding out-of-plane distortion imparted by the -R groups on the pyrazole ring. -iPr analogue 1 has the most steric bulk and least free rotation, which decreases the zero-field splitting average. When the -iPr substituent is replaced by -CH₃ (2) or -H (3), the out-of-plane distortion decreases and allows the free rotation and zero-field splitting averaging to increase. This may account for the differences observed in the EPR spectra and consequent trends observed in cyclic voltammetry and UV-Visible spectroscopy. The range of EPR spectra for 1-3 observed in the solid state may arise from the presence of intermolecular interactions and/or an imperfect rhombic environment, causing several transitions to coincide at the same time and make precise interpretation of the solid-state spectra challenging.

Mössbauer spectroscopy of 1 at 80 K gives rise to an asymmetric spectrum with an isomer shift (δ) of 0.42 mm/s and a quadrupole splitting (ΔE_Q) of 1.18 mm/s (Fig. 6), typical of non-integer spin systems. Increasing the temperature to 294 K produces a spectrum with a slightly lower δ value (0.32 mm/s) and a slightly larger ΔE_Q (1.27 mm/s). The δ value agrees with previously reported quantum spin-admixed systems, while the ΔE_Q values are consistent with the presence of a high-spin ferric iron center [24].

Fig. 6.

Mössbauer spectrum of 1 at 80 K (top) and 294 K (bottom).

At low temperatures, complexes 1−3 can be assigned as high-spin Fe^III species, although they do possess some similarities to previously reported pincer and porphyrin-based spin-admixed (S = 3/2, 5/2) and spin crossover systems [24c, 25, 26]. Overall, the results of solution magnetic susceptibility measurements, EPR, and Mössbauer spectroscopy lead us to conclude 1–3 are most likely high-spin ferric complexes.

Electronic Absorption Spectra

The UV-Vis spectra of complexes 1–3 in CH₂Cl₂ and acetone are shown in Fig. 7. “Cz^tBu(Pyr^iPr)₂ ” pincer complexes typically have π → π* transitions below the 400 nm region, and the lowest absorption bands are best described as ligand-to-metal charge transfer bands (LMCT) [7].

Fig. 7.

Electronic spectra of 6.66 × 10⁻² mM of 1 (blue), 2 (green), and 3 (red) in CH₂Cl₂ (top) and acetone (bottom) [19].

The position of the LMCT bands shifts toward shorter wavelengths (3 (680 nm) < 2 (739 nm) < 1 (743 nm)), when there are less electron donating substituents on the pyrazole groups. Although it is well understood that the ligand field is enhanced with electron donating substituents, the same kind of reverse trend has been observed in the cyclic voltammograms that led to conclude that the extent of out-of-plane distortion of Fe dictates the degree of metal-ligand orbital overlap.

In acetone, complexes 1 and 2 exhibit a slight blue shift in the visible range, e.g. 743 → 725 nm and 739 → 712 nm for 1 and 2, respectively; whereas, the spectrum of 3 shows a drastic red shift from the visible to the near infrared region 680 → 939 nm. We believe that the drastic red-shift is not simply due to the polarity of solvents, but rather the effect of acetone coordination to the Fe center in 3, under forcing conditions. However, when 3 is exposed to 10 equiv acetone in CH₂Cl₂ solution, this spectral feature is not observed, consistent with a lack of acetone binding. The observed disparate spectra for complex 3 are consistent with a concentration-dependent interaction of acetone with 3.

Solution Behavior

To examine the intermolecular interactions further, we turned to spectroscopic investigation under synthetically relevant conditions. A variety of methods have been developed to measure the binding ability of Lewis acid catalysts such as NMR spectroscopy [27], UV-Vis spectroscopy [28], selectivity [29], and mass spectrometry [30]. More specifically, chemists have employed infrared (IR) spectroscopy when examining the interactions of paramagnetic Lewis acids and carbonyl-containing compounds, allowing for the determination of Lewis acidity [31]. To probe the solution interactions of Lewis acids and carbonyls, we have developed a titration-based method, utilizing in situ IR as a detector [32]. With this method, we benchmarked a range of Lewis acid/carbonyl interactions in solution [33] and were able to employ these benchmarks to gain insight into the mechanism of Fe^III-catalyzed carbonyl-olefin metathesis [6]. Because of these benchmarks, we chose to examine Lewis acidity of our Fe^III complexes via their interactions with acetone in DCE.

Complexes 1, 2, and 3 are highly soluble in DCE, and we performed titrations into homogeneous mixtures. When acetone is added to a solution of 1 from 0–4 equiv, we observe two vibrations: one at 1714 cm⁻¹, which is consistent with uncoordinated acetone, and an additional shoulder at approximately 1690 cm⁻¹ (Fig. 8). This extra vibration at lower wavenumbers, in the direction of the known C–O vibration, is suggestive of a slight decrease in the π-character of the C=O of acetone, consistent with previous observations of carbonyl coordination to a Lewis acid (Fig. 8) [6, 25]. When analogous titrations were carried out with 2 and 3, only unbound acetone at 1714 cm⁻¹ is observed in the IR spectra, consistent with no interaction [19].

Fig. 8.

Solution IR data for titration of 1 (1 mmol in 6 mL DCE) with 0–4.2 equiv acetone. Titration proceeds from black to violet with increasing amounts of acetone ([acetone] = 0 M, 0.089 M, 0.178 M, 0.243 M, 0.330 M, 0.499 M [19].

We have previously demonstrated that when ≤1 equiv acetone is added to FeCl₃, a monomeric complex forms in which one molecule of acetone is bound to one molecule of FeCl₃ and no uncoordinated acetone is observed [6]. This 1:1 acetone/FeCl₃ coordination complex exhibits a single characteristic vibration in the carbonyl region at 1633 cm⁻¹. The observation of exclusive formation of coordination complex in the absence of unbound acetone is representative of a high affinity Lewis acid/base interaction; whereas, the simultaneous observation of a coordination complex and unbound acetone is consistent with a low affinity Lewis acid/base interaction [6, 33]. When the titration of FeCl₃ proceeds beyond 1 equiv, the 1:1 acetone/FeCl₃ coordination complex is consumed and higher wavenumber vibrations are observed in the region between the initial Lewis pair and unbound acetone. This change in IR is consistent with the formation of a mixture of two highly ligated complexes, with multiple Lewis basic carbonyls coordinating to Fe [6, 33, 34]. Using the analogy of our previous observations, the simultaneous growth of uncoordinated acetone with the shoulder at 1690 cm⁻¹ is consistent with a low affinity interaction between 1 and acetone. Further, we have previously shown that this low affinity interaction can arise from a trigonal bipyramidal complex with a more distantly associated acetone (Fig. 9, left) [34]. Because of the analogous geometry and observed IR band, we propose an analogous solution interaction (Fig. 9, right).

Fig. 9.

Theoretically predicted C=O IR stretching of 4:1 FeCl₃:acetone solution aggregate (left) [34]. Hypothesized analogous solution interaction for 1 and acetone (right).

We also examined the solution conductivity (κ) of our Fe complexes as compared to FeCl₃ [6, 33]. When each of the respective complexes are dissolved in DCE, we measured the κ for the resulting solution. When FeCl₃, 1, 2, and 3 are added to DCE, we observe values of κ equal to 1.5, 84.4, 27.4, and 827.3 μS cm⁻¹, respectively. A value of 827.3 μS cm-1 for 3 suggests that this structure is spontaneously forming a solvent-separated ion pair, which likely occurs when a chloride ion dissociates from the complex. Interestingly, we do not observe this behavior for the other two complexes. This unique behavior of 3 in DCE is consistent with the unique behavior we observed via our UV-Vis and DFT analysis (vide infra), suggesting that the origin of the spectral features we see in the UV-Vis may arise from the change in the solvation sphere of 3.

We next examined the combination of 1 and acetone via conductivity, using concentrations identical to our IR investigation. We previously demonstrated that when acetone is added to FeCl₃, a negligible conductivity results between 0–1 equiv, with a conductivity of 96 μS cm⁻¹ at the equivalence point (red, Fig. 10) [6]. However at 2 equiv acetone, κ increases to 733 μS cm⁻¹, which continues up to 1244 μS cm⁻¹ at 5 equiv acetone. Over the course of titration of 1 with acetone, κ doubles from 84.4 μS cm⁻¹ to 163.8 μS cm⁻¹. When compared with the three order of magnitude increase in κ for the addition of acetone to FeCl₃, our titration of 1 with acetone is inconsistent with acetone-facilitated displacement of one of the chloride ligands [6].

Fig. 10.

Conductivity of FeCl₃ (red, 2 mmol in 12 mL DCE) [6] and 1 (blue, 2 mmol in 12 mL DCE) with increasing equivalents of acetone. Solution conductivity of each Fe complex in DCE (inset) [19].

Theoretical Calculations

To find additional support for the shift of the LMCT band in the electronic spectra and proposed solution behavior, TD-DFT calculations were performed [19]. The shift of the LMCT band found in the spectra of complexes 1-3 and 3-acetone agrees well with the calculated absorption spectra of model complexes 1’−3’ and [3’-acetone]⁺. For example, the experimental band at 743 nm corresponds to the transition calculated at 726 nm originating from HOMO (β) → LUMO+3 (β) (96%) transition with LMCT character in complex 1 [19]. The same HOMO (β) → LUMO+3 (β) LMCT transitions were observed for complexes 2 and 3 with a blue-shift [19].

In acetone, the drastic red-shift observed in the experimental spectrum of 3 (Fig. 7, bottom) was also found in the calculated spectrum [3’-acetone]⁺. Structural models of [3’-acetone]⁺ indicate that the formation of an octahedral Fe complex resulting from coordination of the NNN pincer ligand, two chloride atoms, and one acetone ligand is too sterically congested to be considered as a possible structure. The most reasonable model of [3’-acetone]⁺ was generated when one chloride ion was displaced by acetone, retaining the original trigonal bipyramidal geometry. These results agree well solution conductivity experiments.

Conclusion

Three Fe^III complexes bearing a Cz^tBu(Pyr^R)₂⁻ (R = iPr (1), Me (2), or H (3)) ligand were synthesized and characterized in the ground state. We tentatively assign these complexes as high-spin Fe^III systems as suggested by solution magnetic susceptibility and EPR spectroscopy. Complexes 1–3 exhibit quasi-reversible redox activity at both the metal center and organic ligand scaffold. We propose that the degree of metal d orbital and N_cz p orbital overlap is directly influenced by the degree of out-of-plane movement of the Fe atom and plays a major role in the redox activity. This behavior is well illustrated by the difference in metal-center reduction potentials of 1 and 3, where the reduced metal-ligand orbital overlap of 1 may inhibit electron-donation from the ligand, therefore reducing the overall effect that the electronic environment of the ligand has on reduction potential. Solvation of 1–3 in non-coordinating (CH₂Cl₂) and coordinating (acetone) solvents produces noticeable changes within the UV-Vis spectra, corresponding to interactions between the coordinating carbonyl of acetone and the Lewis-acidic Fe atom of Cz^tBu(Pyr^R)₂FeCl₂. Indeed, in situ IR spectroscopy revealed that 1 (in DCE) possesses a low affinity interaction with added acetone indicated by concomitant observation of acetone and a new vibration at ~1690 cm⁻¹ at all concentrations examined, a significant departure from what is observed for acetone and FeCl₃ [6, 33]. As a result of these studies, we propose that complexes 1–3 can serve as geometrically constrained models for further investigation of the complex interactions between Lewis-acids, such as Fe, and various carbonyl substrates.