Molecular Structure of Copper and μ-Oxodiiron Octafluorocorrole Derivatives: Insights into Ligand Noninnocence

Abstract

Single-crystal X-ray structures were obtained for the copper and μ-oxodiiron complexes of 2,3,7,8,12,13,17,18-octafluoro-5,10,15-triphenylcorrole, hereafter denoted as Cu[F₈TPC] and {Fe[F₈TPC]}₂O. A comparison with the crystal structures of other undecasubstituted Cu corroles, including those with H, Ar, Br, I, and CF₃ as β-substituents, showed that the degree of saddling increases in the order: H ≲ F < Ar ≲ Br ≲ I < CF₃. In other words, Cu[F₈TPC] is marginally more saddled than β-unsubstituted Cu triarylcorroles, but substantially less saddled than Cu undecaarylcorroles, β-octabromo-meso-triarylcorroles, and β-octaiodo-meso-triarylcorroles, and far less saddled than Cu β-octakis(trifluoromethyl)-meso-triarylcorroles. As for {Fe[F₈TPC]}₂O, the moderate quality of the structure did not allow us to draw firm conclusions in regard to bond length alternations in the corrole skeleton and hence also the question of ligand noninnocence. The Fe–O bond distances, 1.712(8) and 1.724(8), however, are essentially identical to those observed for {Fe[TPFPC]}₂O, where TPFPC^3– is the trianion of 5,10,15-tris(pentafluorophenyl)corrole, suggesting that a partially noninnocent electronic structural description may be applicable for both compounds.

Introduction

Well over a half-century ago, the Danish chemist C. K. Jørgensen distinguished ligands as “innocent” and “suspect,” a suspect—or “noninnocent,” to use the modern term—ligand being one that leaves the oxidation state of the central metal uncertain or debatable.¹ In a typical situation, an electron or a hole cannot be approximated as belonging to either the metal or the ligand, but must be regarded as delocalized over the two.^2,3 Noninnocent ligands thus pose major challenges for inorganic spectroscopists and theoreticians.⁴ Such ligands are also of considerable practical importance: by acting as reservoirs of electrons or holes, noninnocent ligands facilitate a variety of reactions that innocent ligands do not.⁵ First-row transition metal corroles, as it happens, afford some of the best-characterized examples of noninnocent ligands.⁶⁻⁹

A variety of tools, spectroscopic and otherwise, have been used to characterize ligand noninnocence. Of these, NMR and EPR¹⁰ spectroscopy are arguably the most direct because under favorable circumstances they provide a direct probe of the unpaired electron density. Paramagnetic FeCl and Fe–aryl corroles provide some of the most elegant examples of the use of ¹H NMR spectroscopy to elucidate the spin density distributions of noninnocent systems.¹¹⁻¹⁷ These methods, however, are useless for a considerable number of systems in which intramolecular spin couplings result in a diamagnetic ground state. Key examples of such cryptically noninnocent systems include copper,¹⁸⁻²¹ μ-oxodiiron,^11,22 and iron–nitrosyl²³⁻²⁵ corroles. For such systems, more indirect means, such as X-ray structure determination and electrochemistry, can be used to glean electronic structure information. It is against this backdrop that we report single-crystal X-ray structures of the copper and μ-oxodiiron complexes of 2,3,7,8,12,13,17,18-octafluoro-5,10,15-triphenylcorrole, hereafter denoted Cu[F₈TPC] and {Fe[F₈TPC]}₂O, respectively.

β-Octafluorinated porphyrins and corroles are uncommon ligands, primarily because of the relative inaccessibility of 3,4-difluoropyrrole.^14,26,27 However, they are of great interest because their metal complexes are expected to exhibit significantly perturbed reactivity as well as unusual solubility, conductivity, and optical properties. Thus, β-octafluorinated cobalt(III) corroles have been found to be highly active catalysts for water splitting, for both the oxygen²⁸ and hydrogen evolution reactions.²⁹ As strongly perturbed analogues of simple corroles, β-octafluorocorroles also afford fascinating subjects for studies of ligand noninnocence.³⁰ Such studies, however, have been hobbled by an almost³¹ complete lack of X-ray crystal structures for β-octafluorocorrole derivatives, a gap in our knowledge that we have finally been able to close via this study.

Results and Discussion

Ligand noninnocence manifests itself in copper corroles via a curious phenomenon that we have called intrinsic saddling.³² Thus, copper corroles are saddled even in the absence of sterically hindering, peripheral substituents.^32,33 In the currently accepted picture, which is supported by an extensive array of structural, spectroscopic, electrochemical, and quantum chemical studies, saddling is engendered primarily by a Cu(d_x²–y²)–corrole(π) orbital interaction, which endows the molecules with substantial Cu^II–corrole^•2– character.^{18−21,32−40} Sterically hindering substituents can accentuate the saddling,³⁴⁻⁴⁰ but substituents alone, in the absence of the specific orbital interaction, do not bring about saddling in metallocorroles (the case of gold corroles being particularly instructive⁴¹⁻⁴⁶). Interestingly, the corrolazine macrocycle, which has an even more contracted N₄ core than corroles and does not adopt noninnocent electronic structures as readily, yields planar, innocent Cu(III) complexes.⁴⁷ Likewise, carbacorroles such as azulicorrole also yield planar Cu(III) complexes.⁴⁸

The X-ray structure of Cu[F₈TPC] (Figure 1, Tables 1 and 2) revealed distinctly saddled Cu–corrole macrocycles. Interestingly, a total of seven unique metallocorrole molecules were found in the asymmetric unit, providing significant insight into the plasticity of the macrocycle vis-à-vis saddling. Indeed, the saddling dihedrals defined and listed in Table 2 were found to vary over several degrees, as expected for a soft coordinate, and as for other Cu corrole derivatives, the following trend was observed: χ₁ < χ₂ ≈ χ₃. A comparison with crystal structures of other undecasubstituted Cu corroles, including those with H,^32,33 Ar,^35,36 Br,^20,34,37 I,⁴⁰ and CF₃³⁹ as β-substituents, proved instructive and showed that the degree of saddling increases in the order H ≲ F < Ar ≲ Br ≲ I < CF₃, which is essentially the order of their Charton⁴⁹ or Sterimol B1⁵⁰ steric parameters (Table 3).⁵¹ In other words, Cu[F₈TPC] is marginally more saddled than β-unsubstituted Cu triarylcorroles, but substantially less saddled than Cu undecaarylcorroles, β-octabromo-meso-triarylcorroles, and β-octaiodo-meso-triarylcorroles, and far less saddled than Cu β-octakis(trifluoromethyl)-meso-triarylcorroles. It may be worth recalling that even for the exceptionally sterically hindered octakis(trifluoromethyl)-meso-triarylcorrole ligands, the Au complex is rigorously planar because of the energetically mismatched nature of the Au(5d_x²–y²)–corrole(π) orbital interaction.⁴⁶

Figure 1.

Thermal ellipsoid plots (at 30% probability) of Cu[F₈TPC]. Left: the seven Cu corroles in the asymmetric unit, along with solvent molecules (CHCl₃). Right: Close-up views of one of the Cu[F₈TPC] molecules.

Table 1. Crystal and Refinement Data.

	Cu[F8TPC]	{Fe[F8TPC]}2O
empirical formula	C37.86H16.14Cl2.14CuF8N4O0.14	C74H30F16Fe2N8O
formula mass	820.77	1462.76
temperature [K]	100(2)	100(2)
crystal system	monoclinic	monoclinic
space group	P21/c	Cc
λ [Å]	0.7288(1)	0.7288(1)
a [Å]	23.636(2)	20.065(3)
b [Å]	34.020(3)	21.913(3)
c [Å]	30.888(3)	16.255(3)
α [°]	90	90
β [°]	102.819(3)	98.773(5)
γ [°]	90	90
Z	28	4
V [Å]	24,218(3)	7063.4(19)
density[Mg m–3]	1.576	1.376
crystal size (mm3)	0.300 × 0.080 × 0.030	0.120 × 0.020 × 0.020
θ range [°]	1.183–21.498	2.106–22.654
meas. Reflections	384,508	51,140
unique reflections	25,642	8592
parameters	3358	911
restraints	36	38
Rint	0.0881	0.0647
R1, wR2 (I > 2σ)	0.0779, 0.2073	0.0669, 0.1728
R1, wR2 (all data)	0.0861, 0.2136	0.0696, 0.1774
S (GooF) all data	1.034	1.082
max/min res. dens. [e Å–3]	1.406/–1.059	0.854/–0.396

Table 2. Copper–Nitrogen Distances (Å) and Saddling Dihedrals (°) for the Seven Unique Molecules in the Asymmetric Unit of Cu[F₈TPC].

Cu[F8TPC]	χ1	χ2a	χ2b	χ3	Cu–N1a	Cu–N1b	Cu–N2a	Cu–N2b
molecule 1	24.0(8)	44.7(1.5)	52.1(1.5)	52.9(1.3)	1.914(7)	1.902(7)	1.907(7)	1.909(7)
molecule 2	23.1(7)	49.0(1.2)	47.9(1.3)	56.1(1.2)	1.908(6)	1.915(7)	1.906(7)	1.904(6)
molecule 3	25.1(8)	49.8(1.5)	54.2(1.3)	62.1(1.4)	1.909(7)	1.907(7)	1.920(7)	1.908(8)
molecule 4	22.6(8)	50.8(1.5)	49.2(1.5)	52.5(1.3)	1.914(7)	1.905(8)	1.931(8)	1.904(7)
molecule 5	20.9(8)	47.7(1.5)	53.0(1.5)	54.9(1.3)	1.903(7)	1.899(7)	1.911(7)	1.903(8)
molecule 6	24.3(8)	49.4(1.6)	47.4(1.5)	48.8(1.4)	1.897(7)	1.910(8)	1.908(8)	1.920(7)
molecule 7	19.6(8)	44.3(1.4)	47.1(1.6)	49.3(1.3)	1.909(7)	1.896(7)	1.903(7)	1.904(7)

Table 3. Copper–Nitrogen Distances (Å) and Saddling Dihedrals (°) for Selected Cu Corroles.

complex	Cu–N1a	Cu–N2a	χ1	χ2a	χ3	refs
Cu[TPC]	1.891	1.891	27.5	53.4	48.7	(33)
Cu[F8TPC]	1.897–1.914	1.896–1.915	19.6–24.3	44.3–54.2	48.8–62.1	this work
Cu[(pCF3Ph)8TPC]	1.902	1.897	40.9	60.1	66.0	(35)
Cu[Br8TpMePC]	1.916	1.916	44.7	65.9	57.3	(20)
Cu[I8TpCNPC]	1.921	1.922	45.3	67.0	57.3	(40)
Cu[(CF3)8TpFPC]	1.921	1.925	57.2	86.4	84.5	(39)

^aAverage of two values for each experimental structure.

The X-ray structure of {Fe[F₈TPC]}₂O clearly revealed a linear Fe–O–Fe unit (Figures 2 and 3), as also observed for {Fe[OEC]}₂O⁵² and {Fe[TPFPC]}₂O,⁵³ where TPFPC^3– is the trianion of 5,10,15-tris(pentafluorophenyl)corrole. Our earlier work has suggested that μ-oxodiiron corroles are noninnocent with a significant contribution with the following intramolecularly spin-coupled description^11,22

One line of evidence in support of this conclusion is that the X-ray structure of {Fe[TPFPC]}₂O exhibits subtle but characteristic bond length alternations within and adjacent to the bipyrrole units of macrocycles, which are now recognized as a hallmark of noninnocent metallotriarylcorroles. Unfortunately, the moderate quality of the present structure (i.e., the e.s.d.’s in the corrole skeletal distances) does not allow us to draw a firm conclusion on this point. The Fe–O bond distances, 1.712(8) and 1.724(8), however, are essentially identical to those observed for {Fe[TPFPC]}₂O, which leads us to suggest that a partially noninnocent electronic structure description may also be apt for the present compound.

Figure 2.

Thermal ellipsoid plots (at 30% probability) of {Fe[F₈TPC]}₂O.

Figure 3.

Selected crystallographic bond distances for {Fe[F₈TPC]}₂O.

Conclusion

The X-ray structures of two β-octafluoro-meso-triphenylcorrole derivatives have afforded new insight into our growing understanding of the structural manifestations of ligand noninnocence in metallocorrole systems. Thus, Cu[F₈TPC] was found to be only slightly more saddled than β-unsubstituted Cu triarylcorroles and substantially less so than Cu β-octabromo-meso-triarylcorroles, consistent with the van der Waals radii of H, F, and Br. Accordingly, a noninnocent electronic structure, with substantial Cu^II–corrole^•2– character, appears plausible for Cu[F₈TPC]. Unfortunately, the moderate quality of the X-ray structure of {Fe[F₈TPC]}₂O did not allow a definitive evaluation of ligand noninnocence based on an examination of the corrole skeletal bond distances. The fact that {Fe[F₈TPC]}₂O exhibits a nearly identical geometry for the Fe–O–Fe moiety as {Fe[TPFPC]}₂O, however, suggests that the two compounds probably share a similar, partially noninnocent electronic structure.

Materials and Instrumentation

All reagents and solvents were generally used as purchased. Ultraviolet–visible (UV–vis) spectra were recorded on an HP 8454 spectrophotometer. ¹H (400 MHz) and ¹⁹F (376 MHz) NMR spectra were acquired on a 400 MHz Bruker AVANCE III HD spectrometer, equipped with a 5 mm BB/¹H (BB = ¹⁹F, ³¹P, and ¹⁵N) SmartProbe in CD₂Cl₂ referenced to δ = 5.32 ppm and to 2,2,2-trifluoroethanol-d₃ (δ = −77.8 ppm), respectively. Mass spectra were recorded on an LTQ Orbitrap XL spectrometer. Free-base 2,3,7,8,12,13,17,18-octafluoro-5,10,15-triphenylcorrole, H₃[F₈TPC], and Cu[F₈TPC] were prepared as described previously.¹⁴

μ-Oxo-bis[2,3,7,8,12,13,17,18-octafluoro-5,10,15-triphenylcorrolatoiron], {Fe[F₈TPC]}₂O

To a refluxing solution of H₃[F₈TPC] (16 mg, 0.0239 mmol) in MeOH (8 mL) was added FeCl₂·4H₂O (24 mg, 0.121 mmol). After 0.5 h, when the Soret band of the free-base corrole was replaced by another at 378 nm, the reddish-brown mixture was allowed to cool and then evaporated to dryness. The residue was dissolved in CH₂Cl₂ (5 mL), to which was added 2 M aqueous NaOH (20 mL), and the resulting mixture was vigorously stirred for 45 min. The organic phase was then separated, gently washed with an equal volume of water, dried with anhydrous Na₂SO₄, and filtered. The filtrate was evaporated to dryness and dissolved in a minimum volume of CHCl₃. The resulting solution was chromatographed on a column of neutral alumina (activity I) with 1:1 n-hexane/CH₂Cl₂ as an eluent. The product was obtained as the first reddish-brown band. Yield: 7.0 mg (0.0048 mmol, 40%). Brown needles of X-ray quality were obtained by diffusion of n-hexane vapor into a concentrated solution of the compound in CH₂Cl₂ within 13 d.

UV–vis (CH₂Cl₂) λ_max [nm, ε × 10^–4 (M^–1 cm^–1)]: 353 (Soret, 14.23). ¹H NMR: δ 7.62–7.54 (m, 4H), 7.54–7.44 (m, 10H), 7.44–7.37 (m, 8H), 7.25 (d, 4H, J = 7.40 Hz), 7.20 (t, 2H, J = 7.56 Hz), 6.71 (d, 2H, J = 7.32 Hz). ¹⁹F NMR: δ −145.49 (s, 4F), −146.21 (d, 4F, J = 7.22 Hz), −147.80 (d, 4F, J = 7.41 Hz), –157.84 (s, 4F). MS (negative ion mode, major isotopomer) [M]⁻: 1462.0936 (expt), 1462.0996 (calcd).

X-ray structure determination

X-ray data for Cu[F₈TPC] and {Fe[F₈TPC]}₂O were collected on the beamline 12.2.1 at the Advanced Light Source, Lawrence Berkeley National Laboratory. The crystals were mounted on a MiTeGen kapton loop and placed in a 100(2) K nitrogen cold stream provided by the Oxford Cryostream 800 Plus low-temperature apparatus on the goniometer head of a Bruker D8 diffractometer equipped with a PHOTON II CPAD detector operating in shutterless mode. Diffraction data were collected using synchrotron radiation monochromated using silicon (111) to a wavelength of 0.7288(1) Å. A full-sphere of data were collected for each crystal using a combination of phi and omega scans with scan speeds of 1.0 s per 4° for the phi scans and 1 s per degree for the omega scans at 2θ = 0 and −20, with different levels of beam attenuation, respectively. Absorption corrections were applied using SADABS.⁵⁴ The structures were solved by intrinsic phasing (SHELXT)⁵⁵ and refined by full-matrix least squares on F² (SHELXL-2014).⁵⁶ All nonhydrogen atoms were refined anisotropically. Hydrogen atoms were geometrically calculated and refined as riding atoms. For Cu[F₈TPC], definable solvent molecules were found in the electron density map as Q peaks and modeled appropriately; the peaks that could not be definitively assigned to either methanol or chloroform were treated with the solvent mask routine included in OLEX2.⁵⁷ Additionally, the crystal structure resolution was truncated at 0.99 Å because of resolution shells beyond 0.99 Å having R_int values greater than 25%. These additional higher resolution reflections have signal-to-noise ratios that are too low to contribute constructively to the structure solution and would have probably required additional restraints. For {Fe[F₈TPC]}₂O, the crystal structure resolution was truncated at 0.95 Å for the same reasons as indicated above for Cu[F₈TPC].

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.0c01035.

• ¹H and ¹⁹F NMR, mass, and UV–vis spectra (PDF)

• Crystallographic data (CIF)

• Crystallographic data (CIF)

Accession Codes

The crystal structures reported in this paper have been deposited at the Cambridge Crystallographic Data Centre and assigned the deposition numbers CCDC 1978931 and 1978935.

The authors declare no competing financial interest.