Redox and Structural Properties of Mixed-Valence Models for the Active Site of the [FeFe]-Hydrogenase: Progress and Challenges

Abstract

The one-electron oxidations of a series of diiron(I) dithiolato carbonyls were examined to evaluate the factors that affect the oxidation state assignments, structures, and reactivity of these low-molecular weight models for the H_ox state of the [FeFe]-hydrogenases. The propanedithiolates Fe₂(S₂C₃H₆)(CO)₃(L)(dppv) (L = CO, PMe₃, Pi-Pr₃) oxidize at potentials ~180 mV milder than the related ethanedithiolates (Angew. Chem. Int. Ed. 2007, 46, 6152). The steric clash between the central methylene of the propanedithiolate and the phosphine favors the rotated structure, which forms upon oxidation. EPR spectra for the mixed-valence cations indicate that the unpaired electron is localized on the Fe(CO)(dppv) center in both [Fe₂(S₂C₃H₆)(CO)₄(dppv)]BF₄ and [Fe₂(S₂C₃H₆)(CO)₃(PMe₃)(dppv)]BF₄, as seen previously for the ethanedithiolate [Fe₂(S₂C₂H₄)(CO)₃(PMe₃)(dppv)]BF₄. For [Fe₂(S₂C_nH_2n)(CO)₃(Pi-Pr₃)(dppv)]BF₄, however, the spin is localized on the Fe(CO)₂(Pi-Pr₃) center, although the Fe(CO)(dppv) site is rotated in the crystalline state. IR and EPR spectra, as well as redox potentials and DFT-calculations, suggest, however, that the Fe(CO)₂(Pi-Pr₃) site is rotated in solution, driven by steric factors. Analysis of the DFT-computed partial atomic charges for the mixed-valence species shows that the Fe atom featuring a vacant apical coordination position is an electrophilic Fe(I) center. One-electron oxidation of [Fe₂(S₂C₂H₄)(CN)(CO)₃(dppv)]⁻ resulted in 2e oxidation of 0.5 equiv to give the μ-cyano derivative [Fe^I₂(S₂C₂H₄)(CO)₃(dppv)](μ-CN)[Fe^II₂(S₂C₂H₄)(μ-CO)(CO)₂(CN)(dppv)], which was characterized spectroscopically.

Introduction

Modeling of the diiron active site of the [FeFe]-hydrogenases has provided a rich and tractable set of challenges to organometallic chemistry.¹ Motivated by discoveries in biophysical chemistry,^2,3 the chemistry of the Fe₂(SR)₂(CO)_6−xL_x system, previously well examined, has been rejuvenated. Ongoing work in this and other laboratories has shown that these complexes are versatile with respect to redox, substitution, and functionalizability in ways that are pertinent to the enzyme’s active site. A significant breakthrough in the modeling efforts entailed the preparation of the mixed-valence derivatives of the formula the [Fe^IFe^II(SR)₂(CO)_4−x(PMe₃)L_x]⁺ (L = carbene or diphosphine).^4,5 These species exhibit the “rotated” stereochemistry for one of the two Fe centers (Scheme 1). This stereochemistry exposes a vacant coordination site on one Fe center, as had been anticipated by numerous biophysical studies.² This geometry was rare in the otherwise extensive chemistry of diiron dithiolato carbonyls.⁶ Now that the rotated geometry has been demonstrated to exist in the absence of the protein, efforts can be expected to focus on the factors that govern this geometry and its reactivity.^7,8 In this work, we probe the effects of both the dithiolate and especially the terminal ligands on the electronic and molecular structure of the mixed valence state.

Scheme 1.

Structure of the diiron active site of the H_ox state of the [FeFe]-hydrogenases (X is CH₂, NH, or O).

Modeling of the [FeFe]-hydrogenases started with the preparation of [Fe₂(SR)₂(CN)₂(CO)₄]²⁻,⁹ this anion being notable because it featured five of six known ligands in the active site. The redox and protonation of this dicyanide has been developed only to a limited extent.¹⁰ Such studies are complicated by the reactivity inherent in the cyanide ligand, which is known to bridge to other metals and to protonate readily.¹¹ The modeling efforts have advanced significantly by using tertiary phosphine ligands in the place of the cyanides. In this paper, we return briefly to the redox chemistry of the diiron centers supported by a cyanide ligand.

Results

[Fe₂(S₂C₃H₆)(CO)₄(dppv)]⁺

The cyclic voltammetry of the recently reported complexes¹² Fe₂(S₂CnH_2n)(CO)₄(dppv) established that the propanedithiolate (2pdt) undergoes oxidation at ca. 190 mV more cathodic than the ethanedithiolato analogue (2edt) (Figure 1). The redox potentials for these couples were unchanged when the oxidations were conducted under argon, dihydrogen, and dinitrogen; traces of water also had no effect.

Figure 1.

Redox properties of diiron(I) dithiolates (CH₂Cl₂ soln).^a

^aUnless indicated otherwise, couples are reversible as judged from the finding that i_pc/i_pa ~1.

From the CV results, the complex Fe₂(S₂C₃H₆)(CO)₄(dppv), 2pdt, was shown to undergo a one-electron oxidation at a potential less than the Fc^0/+ couple (0.5 V in CH₂Cl₂ vs Ag/AgCl).¹³ This conversion was confirmed on a preparative scale by treating a CH₂Cl₂ solution of 2pdt with FcBF₄ at −45 °C. The IR spectrum of the product (ν_CO = 2076, 2019, 1943 cm⁻¹) proved to be similar to that for the recently reported⁵ [Fe₂(S₂C₂H₄)(CO)₃(PMe₃)(dppv)]BF₄ but shifted to higher energy by ~50 cm⁻¹, reflecting the effect of replacing PMe₃ by CO (Figure 2). Solutions of [2pdt]⁺ were found to react with NO to yield [Fe₂(S₂C₃H₆)(CO)₃(dppv)(NO)]⁺, with the NO bound on the Fe(dppv) side of the molecule, consistent with an open coordination site residing on the Fe(dppv) (eq 1).¹⁴

Figure 2.

IR spectra (CH₂Cl₂ solns) of [Fe₂(S₂C₃H₄)(CO)₄(dppv)]BF₄ ([2pdt]BF₄, A) and [Fe₂(S₂C₃H₆)(CO)₃(PMe₃)(dppv)]BF₄ ([3pdt]BF₄, B).

Oxidation of Fe₂(S₂C₃H₆)(CO)₃(PMe₃)(dppv)

In a preliminary report, we had described the one-electron oxidation of the ethanedithiolate Fe₂(S₂C₂H₄)(CO)₃(PMe₃)(dppv) (3edt).5 Oxidation of the related propanedithiolate (3pdt) using FcBF₄ yielded the corresponding derivative [Fe₂(S₂C₃H₆)(CO)₃(PMe₃)(dppv)]BF₄ ([3pdt]BF₄) but at a potential that is milder by ~180 mV. We verified the electrochemical results with a preparative scale reaction. When a 1:1 mixture of 3pdt and 3edt was treated with one equiv of FcBF₄ (0.5 equiv per diiron), we observed complete consumption of 3pdt as indicated by the loss of the high energy ν_CO band, uniquely assigned to the S₂C₃H₆ derivative. Only upon the addition of a second equiv of FcBF₄ did 3edt undergo oxidation. The structure of [3pdt]BF₄ was confirmed crystallographically (see below). The IR spectrum (Figure 2) of [3pdt]BF₄ salt was very similar to that for [2edt]BF₄, the greatest difference being for the ν_μ-CO band which occurred at 7 cm⁻¹ higher energy for the S₂C₃H₆ derivative.

A solution of [3pdt]⁺ was found to react with CO to yield the adduct [3pdtCO]BF₄. Intermediates were not observed. The rate of carbonylation was slower (t_1/2 ~ 600 s at 1 atm., −45 °C) than the carbonylation of [2edt]⁺ (t_1/2 ~ 30 s at 1 atm., −45 °C). The slowness of the carbonylation is attributed to the required, sluggish relocation of the PMe₃ ligand to a basal position, apparently driven by avoidance of a steric clash with the propanedithiolate strap. For diiron dithiolates bearing seven terminal ligands, it is known that the propanedithiolates (vs ethanedithiolate) tend to be less stable, which is attributable to steric congestion.^8,15 In contrast, mixed-valence complexes with a propanedithiolate are more stable than the ethanedithiolate analogues.

The stereochemistry of [3pdtCO]BF₄ can be assigned on the basis of its solution EPR spectrum, wherein we observed 1:4:4:1 quartet resulting from three smaller ³¹P hyperfine couplings associated with three basal phosphine ligands. This pattern is consistent with the previously published analysis of H_ox^CO models (see below).⁸ In addition to its slow formation, [3pdtCO]⁺ differs from [3edtCO]⁺ with respect to _νμ̃CO, which is 1791 cm⁻¹ in the propanedithiolate and 1783 cm⁻¹ in the ethanedithiolate. The terminal ν_CO bands are also shifted for [3pdtCO]⁺ but in the opposite fashion, reflecting a compensating relationship between ν_μ̃CO and ν_□-CO. The binding of CO is a reversible process: an N₂ purge slowly regenerated [3pdt]⁺. The EPR spectrum for [3edtCO]⁺ (see below) differed from that for [3pdtCO]⁺, being consistent with apical PMe₃ and dibasal dppv (Scheme 2).

Scheme 2.

Pathways for the carbonylations of [3edt]⁺ and [3pdt]⁺.

[Fe₂(S₂C₂H₄)(CO)₃(PCy₃)(dppv)]^0/+

Unlike the analogous PMe₃ derivative, the tricyclohexylphosphine complex Fe₂(S₂C₂H₄)(CO)₃(PCy₃)(dppv) (4edt) could not be prepared by thermally induced substitution of 2edt. Photochemical substitution proved efficient however. The spectroscopy for 4edt was unremarkable, and the position of the ν_CO bands (ν_CO(avg) ~ 1927 cm⁻¹) indicated that the complex resembles 3edt. Unexpectedly, the CV data (+140 mV vs Ag/AgCl) showed that 4edt was more difficult to oxidize than the analogous PMe₃ derivative, which provided the first clue that the bulk of the trialkylphosphine ligands might strongly affect the electronic structure of these diiron compounds.

Oxidation of 4edt was effected in the usual way using FcBF₄ at −45 °C in CH₂Cl₂. IR measurements showed that the oxidation shifts ν_CO by an average of ~50 cm⁻¹, as seen for the corresponding oxidations of 2pdt, 3edt, and 3pdt. The average of the terminal ν_CO bands was, however, lower than that observed for [3edt]⁺, and the pattern was distinctly different, indicating that the structures of [3edt]⁺ and [4edt]⁺ differ. These cations also differ in terms of their stabilities: solutions of [3edt]⁺ are stable for a few seconds at room temperature, whereas the decomposition of [4edt]⁺ is already apparent near −20 °C. For reasons of preparative convenience, further studies on bulky phosphines were conducted with Pi-Pr₃ instead of PCy₃ (next section).

[Fe₂(S₂CnH_2n)(CO)₃(Pi-Pr₃)(dppv)]^0/+

Using to the methods to obtain [4edt]⁺, the complexes Fe₂(xdt)(CO)₃(Pi-Pr₃)(dppv) were prepared for both xdt = edt and pdt. The IR spectra for the trisubstituted diiron(I) complexes (4edt, 3pdt, 5edt, and 5pdt) were very similar. The ³¹P DNMR properties for this series differed, however. The DNMR properties of 3edt and 3pdt have been previously discussed.¹² At 20 °C, the ³¹P NMR spectrum of 5edt featured both a broad and a sharp singlet, assigned respectively to the dppv ligand and Pi-Pr₃. At −40 °C, the dppv signal decoalesced into a pair of doublets, whereas the singlet assigned to Pi-Pr₃ remained sharp. The low barrier for dynamics at the Fe(CO)(dppv) site is consistent, indirectly, with the location of the Pi-Pr₃ in a basal site, which would destabilize the apical-basal spanning dppv due to a steric interaction across the two basal sites. For 5pdt, the 20 °C ³¹P NMR spectrum consisted of two singlets in a 2:1 ratio, again assigned to the dppv and Pi-Pr₃ ligands, respectively. The structures of 5edt and 5pdt, which wasverified crystallographically, see Supporting Information) are as follows:

The cyclic voltammetry of CH₂Cl₂ solutions of 5edt and 5pdt showed that both undergo one-electron oxidations. The separation, Δ;E_pdt−E_edt (160 mV), is comparable to the difference observed for 3edt vs 3pdt, but the absolute values of the redox couples are strikingly different. Complex 5edt was found to oxidize at +170 mV and 5pdt at +10 mV (vs Ag/AgCl). These potentials are ca. 150 mV more anodic than seen for related 3edt and 3pdt (Figure 1). One-electron oxidation of 5edt with FcBF₄ gave the mixed-valence salt [Fe₂(S₂C₂H₄)(CO)₃(Pi-Pr₃)(dppv)]BF₄, [5edt]BF₄. IR spectra in the ν_CO region are virtually identical for [5edt]⁺ and the PCy₃-containing cation [4edt]⁺. One-electron oxidation of 5pdt at −45 °C using FcBF₄ afforded [5pdt]BF₄. Values for ν_t-CO for [5pdt]⁺ and [5edt]⁺ match closely (Figure 3). A significant difference is, however, observed for ν_μ–CO, which is 1830 cm⁻¹ for [5pdt]⁺ vs 1870 cm⁻¹ for [5edt]⁺, consistent with a steric interaction between the propanedithiolate and the bulky Pi-Pr₃ ligand, which displaces the bridging CO toward a more symmetrically bridging position, as anticipated by the DFT calculations (see below, Scheme 3, see below for explanation).

Figure 3.

IR spectra (absorbance mode, CH₂Cl₂ at −45 °C) showing the effects of edt vs pdt for the oxidation of complexes containing bulky phosphine ligands: [Fe₂(S₂C₂H₄)(CO)₃(Pi-Pr₃)(dppv)]BF₄ ([5edt]BF₄, top) and [Fe₂(S₂C₃H₆)(CO)₃(Pi-Pr₃)(dppv)]BF₄ ([5pdt]BF₄, bottom).

Scheme 3.

Proposed solution structures of [5edt]⁺ and [5pdt]⁺.

Complex [5edt]⁺ was found to form an adduct with CO, but the carbonylation (~30 min at −45 °C) is significantly slower than for [3edt]⁺ (~30 s at −45 °C). The IR and EPR spectra of [3edtCO]⁺ and [5edtCO]⁺ are very similar, indicating similar structures, despite structural differences of the precursors. The average of the two ν_tCO bands is 20 cm⁻¹ lower for [5edt]⁺ vs [3edt]⁺. This difference is consistent with the presence of a Fe^II(CO)₂(PMe₃) vs a Fe^I(CO)₂(Pi-Pr₃) site, i.e., the larger phosphine redirects the site of oxidation. A similar pattern applies also to the [3pdt]⁺/[5pdt]⁺ pair. Solutions of [5pdt]⁺ were found to be stable at 0 °C for several minutes, whereas [5edt]⁺ was found to rapidly decompose above −20 °C. Solid [5pdt]BF₄ was rapidly precipitated by transferring a CH₂Cl₂ solution at −45 °C into a large excess of hexane cooled to −78 °C. Although we were unable to isolate an analytically pure sample, the IR spectrum of finely powdered [5pdt]BF₄ (KBr pellet) matched the solution data (Δ;ν_μ–CO ~3 cm⁻¹).

Solution EPR Studies of Mixed-Valence Models

The X-band EPR spectra of frozen solutions were similar, showing axial patterns with well defined hyperfine coupling to phosphorus (Table 1). Thus, a frozen solution of [2pdt]BF₄ showed a triplets with A(³¹P) ~ 75 MHz, Figure 5. For the mononuclear Fe(I) complex trans-[Fe(CO)₃(PPh₃)₂]⁺, the hyperfine values are ca. 55 MHz.¹⁶ The magnitude and multiplicity of the hyperfine couplings indicate that the spin resides on the Fe(CO)(dppv) center wherein the two phosphine sites are equivalent. The EPR spectrum of the analogous salt [3edt]BF₄ also showed triplets with comparable g-values and hyperfine constants (Figure 4).

Table 1.

Selected EPR Parameters for [Fe₂(S₂C_XH_2x)(CO)₂L(dppv)]BF₄ (Spectra Recorded at 110 K (−163 °C), in 1:1 CH₂Cl₂:toluene).

Complex (donor ligands)	gz	Azz	gy	Ayy	gx	Axx
[2pdt]+ (dppv)	2.1490	78, 78	2.0199	71, 71	2.0086	72, 72
[3edt]+ (dppv, PMe3)	2.1384	80, 80	2.0280	74, 74	2.0102	72, 72
[3pdt]+a (dppv, PMe3)	2.1310 2.1410	82, 82 81, 81	2.0223 2.0268	76, 76 73, 73	2.0085 2.0064	74, 74 75, 75
[4edt]+ (dppv, PCy3)	2.0958	69	2.0423	67	2.0098	73
[5edt]+ (dppv, Pi-Pr3)	2.0968	70	2.0437	71	2.0105	75
[5pdt]+ (dppv, Pi-Pr3)	2.1046	80	2.0268	73	2.0149	80

^aFor [3pdt]⁺ the two species are present in at 2:1 ratio.

Figure 5.

X-Band EPR spectra (110K, 1:1 CH₂Cl₂:toluene frozen solution): [Fe₂(S₂C₃H₆)(CO)₄(dppv)]⁺ ([2pdt]⁺, a), [Fe₂(S₂C₃H₆)(CO)₃(PMe₃)(dppv)]⁺ ([3pdt]⁺, b), [Fe₂(S₂C₃H₆)(CO)₃(Pi-Pr₃)(dppv)]⁺ ([5pdt]⁺, c).

Figure 4.

X-Band EPR spectra (110K, 1:1 CH₂Cl₂: toluene frozen solution) of [Fe₂(S₂C₂H₄)(CO)₃(PMe₃)(dppv)]BF₄ ([3edt]BF₄). Top is the simulation and bottom is experimental data.

An interesting feature of the EPR spectrum of [3pdt]⁺ is the presence of pairs of separate features in a ~2:1 ratio, especially discernable in the lower field signal. Each set of axial signals has a pattern (both g and A values) similar to that for [3edt]⁺. The two isomers are proposed to differ with respect to the fold of the propanedithiolate. The sensitivity of the EPR spectrum to the orientation of the dithiolate backbone shows that the SOMO of [3pdt]⁺ is affected by the central atom of the dithiolate (eq 2). Upon warming the solution to −50 °C, only a single isotropic triplet signal is observed.

A similar equilibrium was not observed for other propanedithiolate derivatives, which we attribute to either (i) greater “steric asymmetry,” which favors the tilting of the central methylene toward the open site on the rotated iron center ([5pdt]⁺) or (ii) insensitivity of the SOMO to the asymmetry due to the presence of CO ligands in both apical positions ([3pdtCO]⁺, see below).

The stereochemistry of the CO adduct, [3pdtCO]BF₄ was indicated by EPR spectroscopy. For such H_ox^CO models, we recently showed that phosphine ligands in the apical positions give rise to large A-values (>200 MHz), whereas A-values are much smaller (<50 MHz) for phosphine ligands attached to the basal sites.⁸ For example, the EPR spectrum of [3edtCO]⁺ is characterized by one large A-value (>250 MHz) and two smaller ones (42, 28 MHz). A very different spectrum was obtained for [3pdtCO]⁺: the three A(³¹P) values were quite small (27, 39, 28, MHz; see Table 3), indicative that all three phosphine ligands occupy basal positions. Upon binding of CO, the PMe₃ undergoes a change in geometry, from apical in [3pdt]⁺ to basal in [3pdtCO]⁺ (Scheme 2). In general, reflecting steric crowding, the CO adducts of ethanedithiolate derivatives are significantly more stable than the propanedithiolate analogues.

Table 3.

Selected Structural Parameters for [Fe₂(S₂C₃H₆)(CO)₃L(dppv)]BF₄ (L = PMe₃, Pi-Pr₃).

Distances (Å)	L = Pi-Pr3	L = PMe3	Angles (deg.)	L = Pi-Pr3	L = PMe3
Fe(1)–Fe(2)	2.5824 (10)	2.5768 (8)	Fe(1)–S(1)–Fe(2)	69.53 (3)	69.53(3)
Fe(1)–S(1)	2.2745 (10)	2.2730 (8)	Fe(1)–Fe(2)–S(1)	55.60 (3)	55.73 (2)
Fe(2)–S(1)	2.2544 (11)	2.2460 (8)	Fe(2)–Fe(1)–S(1)	54.87 (3)	54.74 (2)
Fe(1)–C(1)	1.785 (5)	1.786 (4)	Fe(2)–Fe(1)–C(1)	78.55 (16)	73.20 (13)
Fe(2)–C(1)	2.833	2.678	Fe(2)–Fe(1)–P(1)	136.39 (3)	135.44 (2)
Fe(1)–P(1)	2.2387 (11)	2.2369 (8)	P(1)–Fe(1)–C(1)	92.15 (12)	93.86 (10)
Fe(2)–C(2)	1.774 (4)	1.787 (3)	Fe(1)–Fe(2)–C(2)	103.25 (13)	108.92 (10)
Fe(2)–P(2)	2.2830 (15)	2.2273 (15)	Fe(1)–Fe(2)–P(2)	154.09 (5)	147.79 (5)
C(1)–O(1)	1.147 (6)	1.148 (5)	P(2)–Fe(2)–C(2)	94.63 (13)	94.37 (18)
C(2)–O(2)	1.139 (5)	1.142(4)	C(2) –Fe(2)–C(2A)	91.9 (3)	93.0 (2)
			Fe(1)–C(1)–O(1)	173.0 (5)	171.2 (4)
			Fe(2) –C(2)–O(2)	175.8 (4)	177.8 (3)

Frozen solution EPR spectra of the mixed-valence Pi-Pr₃ and PCy₃ complexes exhibit patterns with doublet features indicating that the SOMO is localized on the Fe(CO)₂(PR₃) center, in contrast to the behavior of the PMe₃ complexes (Figure 5). The EPR data for [4edt]⁺, [5edt]⁺, and [5pdt]⁺ are very similar with respect to g and A values (Table 1). The A(³¹P) values for these doublets are similar to the previously observed A values for the triplets seen for [3edt]⁺ and [3pdt]⁺. Upon warming a sample of [5pdt]⁺ to −40 °C, whereupon the solvent is melts, the signal merges to give an isotropic doublet, consistent with the SOMO being localized on the Fe(CO)₂(Pi-Pr₃) center.

The carbonylation of [5edt]⁺ (1 atm, −45 °C) proceeded more slowly (~30 min) than for [3edt]⁺, which implicates a geometric rearrangement of the phosphine ligands. The EPR spectrum of the adduct [5edtCO]⁺ matched that observed for [3edtCO]⁺, i.e., one large and two smaller A(³¹P) values (Table 2).

Table 2.

Selected EPR Parameters for [Fe₂(S₂C_xH_2x)(μ-CO)(CO)₃L(dppv)]BF₄ (Spectra Recorded at 110K (−163 °C), in 1:1 CH₂Cl₂:toluene)

Complex	gz	Azz	gy	Ayy	gx	Axx
[3edtCO]+	1.9943	323, 39, 42	1.9740	263, 28, 30	2.0073	258, 30, 33
[3pdtCO]+	1.9976	30, 40, 36	1.9814	24, 30, 26	2.0131	24, 31, 26
[5edtCO]+	1.9939	313, 36, 33	1.9750	254, 30, 28	2.0075	255, 27, 27

Crystallographic Characterization of [Fe₂(S₂C₃H₆)(CO)₃(PR₃)(dppv)]BF₄ (R = Me, i-Pr)

Crystals of these salts were grown at low temperatures. The [3pdt]⁺ salt is unsolvated while [5pdt]⁺ crystallized with a CH₂Cl₂ in the lattice; for both complexes the BF₄⁻ is well separated from the cationic diiron centers (Figure 6, Table 3). The cations adopt the expected structure wherein the terminal ligands on the quasi-octahedral Fe centers are staggered but that the Fe(CO)(dppv) adopts the “rotated” geometry. The CH---Fe(1) distance is 2.49 Å in [3pdt]BF₄, respectively beyond the limit recognized for an agostic interaction.¹⁷ The greater steric bulk of the S₂C₃H₆ vs S₂C₂H₄ is manifested in the rotation of the pair of phenyl groups that surround the vacant coordination site. That angle increases as follows: 4° ([3edt]BF₄), 14° ([3pdt]BF₄), and 33° ([5pdt]BF₄), reflecting a slightly increased tilting of the dithiolate toward the rotated Fe (the angle of the axial R₃P-Fe-S, [3pdt]⁺ = 105.11 (15)° and [5pdt]⁺ = 107.93°), although the CH---Fe(1) distance does not contract further.

Figure 6.

Structure of the cations in [3pdt]BF₄ and [5pdt]BF₄with thermal ellipsoids set at 35% level. Phenyl ellipsoids, hydrogen atoms, and the BF₄⁻ were omitted for clarity.

Small but significant differences are observed between the Fe-P distances for [3pdt]BF₄ and [5pdt]BF₄. Specifically, the Fe-Pi-Pr₃ distance is 2.2830(15) Å, which is 0.06 Å longer than the Fe-PMe₃ distance in [3pdt]BF₄. The Fe-P(dppv) distances are the same in both [3pdt]BF₄ and [5pdt]BF₄. This elongated Fe-P distance is consistent with steric conjestion at this site.

Since the EPR and the crystallographic data for [5pdt]BF₄ do not agree, we considered the possibility that the solution structure differs from that observed in the single crystals. Indeed, an IR spectrum of a solid sample (KBr) prepared from single crystals differ distinctly from spectra for [5pdt]⁺, both as a solution and as a rapidly precipitated solid (Figure 7). Significantly, the IR band at 1958 cm⁻¹ for the rapidly precipitated solid is absent in the spectrum of the crystals. Furthermore, the ν_μ-CO band at 1825 cm⁻¹ in the rapidly precipitated material is not present in the crystalline material. When the single crystals were redissolved in CH₂Cl₂ (−45 °C), the solution IR spectrum matched that for a solution of [5pdt]⁺. As a control, we demonstrated that the similarity of the IR spectra for solid (KBr) and solutions of [3pdt]BF₄, ν_Δ-CO = −3 cm⁻¹ (see Supporting Information). It was not practical to compare the spectra of solution vs solid [5edt]BF₄, which is far less stable than [5pdt]BF₄.

Figure 7.

IR spectra (KBr) of [5pdt]BF₄ that was rapidly precipitated (top) and single crystals grown at −20 °C (bottom).

DFT Calculations

Structures, relative stabilities, and electronic properties of different mixed valence models were studied by BP86/TZVP DFT calculations. Three families of model complexes, differing in the size of phosphine ligands, have been analyzed: [Fe₂(S₂C_nH_2n)(CO)₃(PH₃)₃]⁺, [Fe₂(S₂C_nH_2n)(CO)₃(PMe₃)(dppv)]⁺(corresponding to [3xdt]⁺) and [Fe₂(S₂C_nH_2n)(CO)₃(Pi-Pr₃)(dppv)]⁺ (corresponding to [5xdt]⁺). In each family, both edt and pdt complexes have been considered. Hereafter, the Fe atoms coordinated by one and two phosphorus ligands will be referred to as Fe^cc and Fe^c, respectively.

For [Fe₂(S₂C_nH_2n)(CO)₃(PH₃)₃]⁺, the lowest energy isomer (see Supporting Information) is characterized by PH₃ ligands in basal sites, and the Fecc(unrotated)-μ-C(O) distance (2.438 Å) is long. Analogous results are obtained for the [Fe₂(S₂C_nH_2n)(CO)₃(PMe₃)(dppv)]⁺ family (see Supporting Information), indicating that the replacement of PH₃ ligands with dppv and PMe₃ does not strongly affect the properties of the binuclear cluster. In both families ((PH₃)₃ and dppv-PMe₃), the energy difference between the most stable isomer and the other isomeric forms is greater for pdt than edt, due to increased steric clashing between apical P ligands and pdt. This situation is reversed when PMe₃ is replaced by Pi-Pr₃, where the calculations suggest that rotation of Fe^cc and Fe^c are nearly isoenergetic (Δ;E = 0.8 kcal/mol, Scheme 4).

Scheme 4.

Computed relative stabilities (kcal/mol), spin densities, and NBO partial atomic charges (italics) for relevant isomers of [Fe₂(S₂C_nH_2n)(CO)₃(Pi-Pr₃)(dppv)]⁺.

In the lowest-energy [5xdt]⁺ isomer, the apical vacant coordination site is located on Fe^cc. Moreover, relative to the corresponding (PH₃)₃ and (PMe₃)(dppv) cases, the μ-CO group is more symmetrically bridging. This finding is consistent with our IR data that show that ν_μ-CO is 1830 cm⁻¹ for [5pdt]⁺ vs 1889 cm⁻¹ for [3pdt]⁺. The relative stability of the isomers is strongly influenced by the bulkiness of the monophosphine ligand. Steric clashing is minimized when Pi-Pr₃ occupies a basal position trans from the basal end of the dppv chelate ring (Scheme 4, Figure 8).

Figure 8.

DFT structure of the lowest energy isomer of [Fe₂(S₂C₃H₆)(CO)₃(Pi-Pr₃)(dppv)]⁺ ([5pdt]⁺). Atoms are color coded as follows: carbon, green; oxygen, red; sulfur, yellow; phosphorus, purple; iron, cyan.

Our analysis of the electronic properties of [Fe₂(S₂C_nH_2n)(CO)₃(PH₃)₃]⁺, [Fe₂(S₂C_nH_2n)(CO)₃(PMe₃)(dppv)]⁺, and [Fe₂(S₂C_nH_2n)(CO)₃(Pi-Pr₃)(dppv)]⁺ reveals that the unpaired electron is always localized on the iron atom featuring a vacant apical coordination position, i.e., Fe^cc in [Fe₂(S₂C_nH_2n)(CO)₃(PH₃)₃]⁺ and [Fe₂(S₂C_nH_2n)(CO)₃(PMe₃)(dppv)]⁺, and Fecc in [Fe₂(S₂C_nH_2n)(CO)₃(Pi-Pr₃)(dppv)]⁺. This picture is consistent with previous DFT studies carried out on models,⁷ as well as on the H-cluster.^18,19 Recent pulsed EPR analysis of the the H_ox state of the enzyme indicates that the unpaired electron is more localized on the non-rotated (proximal) iron.²⁰ This apparent discrepancy between theoretical/model and biophysical data may reflect subtle effects due to the H-cluster environment.¹⁸ Our oxidation state assignments agree with earlier calculations by Brunold on the entire 6Fe H-cluster¹⁹ as well as the results from Hall et al. on a related model containing carbene ligand on the rotated iron site.⁷

Notably, the analysis of computed partial atomic charges for [Fe₂(S₂C_nH_2n)(CO)₃(PH₃)₃]⁺, [Fe₂(S₂C_nH_2n)(CO)₃(PMe₃)(dppv)]⁺, and [Fe₂(S₂C_nH_2n)(CO)₃(Pi-Pr₃)(dppv)]⁺ shows that the more electrophilic metal center always corresponds to the Fe atom featuring a vacant apical coordination position (Scheme 4). Therefore, in this class of compounds, the assignment of the Fe(I) oxidation state to the iron center carrying the unpaired electron should be taken as purely formal.

Oxidation of [Fe₂(S₂C₂H₄)(CN)(CO)₃(dppv)]⁻

We examined the oxidation of the cyanide complex Bu₄N[Fe₂(S₂C₂H₄)(CN)(CO)₃(dppv)], Bu₄N[6edt].12 This anionic complex was expected to be more easily oxidized than the related PMe₃ derivative 3edt and thus a potential precursor to a mixed-valence cyanide. Cyclic voltammetry indicated that this species oxidizes at −130 mV, 150 mV more cathodic than [3edt]^0/+ couple. Unlike the voltammogram for the PMe₃ derivative, however, the oxidation of the cyanide was electrochemically irreversible, it re-reduces at −600 mV (ΔE_p ~ 500 mV). The i-V response is reminiscent of an EC process, whereby oxidation induces an irreversible chemical reaction. The CV remained unchanged even at scan rates up to 500 mV/s.

In a preparative-scale oxidation, treatment of a CH₂Cl₂ solution of Bu₄N[6edt] with one equiv of FcBF₄ afforded a thermally stable, toluene-soluble complex that proved to be diamagnetic. This species is assigned as a four-iron compound [6edt]₂. Unlike the behavior of related phosphine complexes discussed above and elsewhere,^5,8 this oxidation was unaffected by the presence of CO, indicating the absence of long-lived unsaturated species. The corresponding two-electron oxidation of Bu₄N[6edt] in MeCN solution gives C_s-symmetric [Fe₂(S₂C₂H₄)(μ-CO)(CO)₂(CN)(NCMe)(dppv)]^+.12

The ³¹P NMR spectrum of [6edt]₂ consisted of a broad singlet at δ 95 and a sharp singlet at δ 74.5. The low field signal is characteristic of dppv dynamically spanning apical and basal sites on diiron(I) species. A similar structure is assumed for Fe₂(S₂C₂H₄)(CO)₄(dppv), which exhibits a single 31P NMR resonance at δ 94.7.¹² The sharp higher field signal is consistent with dppv on dibasal sites on diiron(II) derivatives, e.g. [Fe₂(S₂C₂H₄)(μ-CO)(CO)₂(PMe₃)(NCMe)(dppv)](BF₄)₂ (δ 74.6).¹² Upon cooling the solution, each of these two signals split. The lower field signal split more, consistent with the presence of both apical and basal phosphine ligands on a diiron(I) center, whereas the splitting of the higher field signal is small (2 ppm) because the electronic asymmetry is modest, consistent with this dppv being dibasal. The IR spectrum of this charge-neutral complex resembles the sum of the spectra of [Fe₂(S₂C₂H₄)(μ-CO)(CO)₂(PMe₃)(MeCN)(dppv)]²⁺ (2058 and 2016 cm⁻¹) and the starting complex (1952 and 1901 cm⁻¹). The 1952 and 1901 cm⁻¹ bands assigned to the Fe(I)-Fe(I) center occur at ~10 cm⁻¹ higher energy than in Bu₄N[6edt], consistent with the attachment of an electrophile to the FeCN. These data are consistent with the formulation [Fe^I₂(S₂C₂H₄)(CO)₃(dppv)](μ-CN)[Fe^II₂(S₂C₂H₄)(μ-CO)(CN)(CO)₂(dppv)] (Scheme 5). Further characterization of the charge-neutral product via mass spectroscopy was unsuccessful.

Scheme 5.

Redox reactivity proposed for [Fe₂(S₂C₂H₄)(CN)(CO)₃(dppv)]⁻.

Discussion

Oxidation of diiron dithiolato carbonyls produces highly unsymmetrical derivatives featuring a single “rotated” iron center and a semi-bridging CO ligand. Relief of steric strain influences the regiochemistry of the oxidation, i.e., the bulkier Fe center undergoes rotation. The rotation desymmetrizes the diiron center, localizing the mixed valency as indicated by EPR spectra. Addition of CO to the mixed valence species re-symmetrizes the diiron center, leading to a more delocalized mixed-valence system.⁸ The central feature of these models is that the iron center that undergoes the redox-induced rearrangement remains in the Fe(I) oxidation state, whereas the iron center that undergoes oxidation remains in the same geometry and becomes coordinatively saturated. The square pyramidal iron(I) center is unsaturated and electrophilic.

Effects of Coligands on Mixed-Valency

The EPR spectra of [2pdt]⁺, [3edt]⁺, and [3pdt]⁺ are very similar as are the crystal structures for the latter two. The rotated nature of the Fe(CO)(dppv) center was crystallographically verified for both [3pdt]⁺ and [3edt]⁺, and we assume that a similar structure applies to the corresponding tetracarbonyl [2pdt]⁺. In face of this pattern, the EPR results are strikingly different when PMe₃ is replaced by Pi-Pr₃ and PCy₃. Specifically, the EPR results indicate that the spin is located on the Fe(CO)₂PR₃ center in these bulkier complexes. The crystallography shows, however, that the Fe(CO)(dppv) remains rotated in [5pdt]BF₄, but we suggest that the Fe^I(CO)₂(Pi-Pr₃) site is rotated in solution. DFT calculations indicate that the bulkier phosphine slightly favors rotation at the monophosphine site. Pi-Pr₃ and PMe₃ have comparable pK_a‘s of ~9.7 (for PCy₃, which we assume is similar to Pi-Pr₃) and 8.65 (PMe₃, compare 2.73 for PPh₃).20 Pi-Pr₃ is, however, much larger than PMe₃ with a cone angle of 160º (vs 118º).21 The influence of PMe₃ vs Pi-Pr₃ on the regiochemistry of the oxidation is also supported by the IR data as well as the electrochemical results. Compound [5pdt]⁺ resembles the active site to the extent that it features two donor ligands and one CO on the “proximal” Fe center and one donor ligand and two CO’s on the “distal” (rotated) Fe center (eq 3).

For the sake of completeness, we note that the EPR results qualitatively support two distinct descriptions, L₃Fe^II(SR)₂Fe^IL₃, which we favor,5,7,22 and L₃Fe^III(SR)₂Fe⁰L₃, which has not been discussed. The 0,III description would be consistent with oxidation occurring at the more electron-rich iron center, and implies that oxidation and rotation induce a redox change at both iron centers, i.e., L₃Fe^I(SR)₂Fe^IL₃ → L₃Fe⁰(SR)₂Fe^IIIL₃. “Dative metal-metal bonds” have been invoked to describe compounds with such disparate oxidation states.²³ Precedents for ferric carbonyls include [Cp₂Fe₂(SMe)₂(CO)₂]²⁺ (ν_CO ~ 2060 cm⁻¹)²⁴ and the dithiocarbamate [(C₅Me₅)Fe(S₂CNEt₂)(CO)]⁺.25 The oxidation state assignments will be further probed through studies on the Mössbauer spectra of these compounds.

Site Isolation

For low molecular weight models, a recurring question concerns the extent to which the synthetic ligands can or need to emulate the steric shielding provided by the protein. The issue is pertinent since the H-cluster is deeply imbedded in both of the structurally characterized proteins.²⁶ The present results highlight continuing problems that hamper the preparation of unsaturated models featuring cyanide ligands. Specifically, we showed that oxidation of a diiron cyanide leads to aggregation via formation of a μ-CN bridge. This conversion, a disproportionation, is explicable in terms of an inner sphere mechanism facilitated by the ambidentate nature of cyanide. For the present generation of models, however, tertiary phosphine ligands serve as useful surrogates for cyanide, with the advantages that their electronic and steric properties can be manipulated as we showed in this paper.

Outlook

The chemistry of mixed-valence models for the diiron active site of the [FeFe]-hydrogenases has evolved such that many details can be evaluated. For example, the nature of the dithiolate as well as the basicity and size of the coligands significantly influence the mixed valency and affect the reactivity of these species toward CO. These results will guide further efforts to understand of how substituents affect the interaction of diiron dithiolato centers toward small molecules, including H₂.

Experimental Section

Materials and methods (EPR and IR spectroscopy, and cyclic voltammetry) have recently been described.⁸ [CpFe(C₅H₄Ac)]BF₄ was prepared by literature methods.13 The diiron complexes Fe₂(S₂C₂H₄)(CO)₄(dppv), Fe₂(S₂C₃H₆)(CO)₄(dppv), Fe₂(S₂C₃H₆)(CO)₃(PMe₃)(dppv), Et₄N[Fe₂(S₂C₂H₄)(CN)(CO)₃(dppv)] have been previously described.¹² NMR spectra were recorded at room temperature on a Varian Mercury 500 MHz spectrometer (202 MHz for ³¹P).

X-ray Crystallography

Crystals were mounted to a thin glass fiber using Paratone-N oil (Exxon). Data, collected at 198 K on a Siemens CCD diffractometer, were filtered to remove statistical outliers. The integration software (SAINT) was used to test for crystal decay as a bi-linear function of X-ray exposure time and sin(Θ). The data were solved using SHELXTL by Direct Methods (Table 4); atomic positions were deduced from an E map or by an unweighted difference Fourier synthesis. H atom U’s were assigned as 1.2U_eq for adjacent C atoms. Non-H atoms were refined anisotropically. Successful convergence of the full-matrix least-squares refinement of F² was indicated by the maximum shift/error for the final cycle.

Table 4.

Details of Data Collection and Structure Refinement for Crystallography.

Complex	[3pdt]BF4	[5pdt]BF4
Chemical formula	C35H37BF4Fe2O3P3S2	C41H49BF4Fe2O3P3S2
Temperature (K)	193 (2)	193 (2)
Crystal size (mm3)	0.28 × 0.12 × 0.06	0.31 × 0.08 × 0.08
Crystal system	Orthorhombic	Monoclinic
Space group	Pnma	P21/m
a (Å)	23.5362 (7)	9.2644 (7)
b (Å)	18.3985 (5)	16.1182 (12)
c (Å)	8.5950 (2)	16.1141 (12)
α(°)	90	90
β(°)	90	105.462 (2)
γ(°)	90	90
V (Å3)	3721.90 (17)	2319.2 (3)
Z	4	2
Density calcd (Mg m−3)	1.537	1.475
μ(Mo Kα, mm−1)	0.71073	0.71073
max./min. trans’n	0.9422/0.8116	0.9323/0.7729
reflections meas’d/Indep.	47014/3695	27790/4544
data/restraints/parameters	3695/261/339	4544/503/471
GOF on F2	1.083	1.069
Rint	0.0623	0.0431
R1 [I > 2σ] (all data)a	0.0380 (0.0552)	0.0505 (0.0642)
wR2 [I > 2σ] (all data)b	0.0904 (0.1003)	0.1328 (0.1414)
max. peak/hole (e−/Å3)	0.715/−0.322	1.036/−0.548

^aR1 = Σ|F_o|−|F_c|/Σ|F_o|

^bwR2 = {[w(|F_o|−|F_c|)²]/Σ[wF_o²]}^1/2, where w = 1/σ² (F_o)

EPR Spectroscopy

The following procedure is illustrative: a solution of 0.010 g (0.0116 mmol) of Fe₂(S₂C₃H₆)(CO)₃(Pi-Pr₃)(dppv) in 6 mL of CH₂Cl₂ (−45 °C) was treated with a solution of 0.003 g (0.0112 mmol) of FcBF₄ in 6 mL of CH₂Cl₂. The resulting solution (1 mM) was transferred via cannula to a chilled (−78 °C) EPR tube that had been purged with N₂. The EPR tube was cooled with liquid nitrogen and flame-sealed under vacuum.

Spin concentrations were determined by double integrating baseline corrected spectra. For calibration, a sample containing 1 mM CuSO₄ in a 20% glycerol solution was recorded at the same power, modulation amplitude, and temperature as that used for the diiron samples. In this way, the spin concentration of three samples, nominally 1mM, of [3pdt]BF₄ were found to be 0.60, 0.60, and 0.70 mM.

DFT Calculations

DFT calculations were carried out using the BP86 functional²⁷ and a valence triple-ζ basis set with polarization on all atoms (TZVP).²⁸

[Fe₂(S₂C₃H₆)(CO)₄(dppv)]BF₄ ([2pdt]BF₄)

A solution of 0.250 g (0.344 mmol) of Fe₂(S₂C₃H₆)(CO)₄(dppv) in 15 mL of CH₂Cl₂ at −45 °C was treated with a solution of 0.094 g (0.344 mmol) of FcBF₄ in 10 mL of CH₂Cl₂. The purple solution was rapidly transferred into 400 mL of hexane cooled to −78 °C. The supernatant solvent was siphoned away from the finely divided solid. Yield: 0.152 g (54%). IR (CH₂Cl₂): ν_CO = 2076, 2020, 1943 cm⁻¹. IR (solid KBr): ν_CO = 2075, 2012, 1934, 1906 (sh) cm⁻¹.

[Fe₂(S₂C₃H₆)(CO)₃(PMe₃)(dppv)]BF₄ ([3pdt]BF₄)

A solution of 0.200 g (0.258 mmol) of Fe₂(S₂C₃H₆)(CO)₃(PMe₃)(dppv) in 10 mL of CH₂Cl₂ at −45 °C was treated with a solution of 0.070 g (0.258 mmol) of FcBF₄ in 10 mL of CH₂Cl₂. The purple solution was transferred into 400 mL of hexane at −78 °C. The supernatant was removed by cannula. Crystals were grown from the liquid diffusion of hexane into a CH₂Cl₂ solution at −20 °C. Yield: 0.171 g (70%). IR (CH₂Cl₂): ν_CO = 2015, 1962, 1889 cm⁻¹. IR (solid KBr): ν_CO = 2007, 1950, 1886 cm⁻¹.

When CH₂Cl₂ solutions of [3pdt]BF₄ were exposed to CO (1 atm at −45 °C). The in-situ IR spectrum showed the disappearance of [3pdt]BF₄ and the growth of new peaks, attributed to [Fe₂(S₂C₃H₆)(μ-CO)(CO)₃(PMe₃)(dppv)]BF₄, [3pdtCO]BF₄. Due to the instability of [3pdtCO]BF₄ all data were collected in-situ. IR (CH₂Cl₂): ν_CO = 2018, 1984, 1791 cm⁻¹. Conversion of [3pdt]⁺ to [3pdtCO]⁺ using ~1 atm CO required ~10 min.

Fe₂(S₂C₂H₄)(CO)₃(PCy₃)(dppv) (4edt)

A solution of 0.300 g (0.421 mmol) of Fe₂(S₂C₂H₄)(CO)₄(dppv) in 40 mL of toluene was treated with 0.130 g (0.421 mmol) of PCy₃ in 10 mL of toluene. The reaction mixture was photolyzed with a 100 W UV immersion lamp, λ_max = 356 nm (Spectroline), until the conversion was complete (~36 h) as indicated by IR spectroscopy. The solvent was removed in vacuum, and the product extracted into 20 mL of CH₂Cl₂. The solution was concentrated to ~5 mL, and the red-brown product precipitated upon addition of 40 mL of hexanes. Yield: 0.352 g (86 %). ¹H NMR (CD₂Cl₂): δ 7.98 – 7.29 (m, 20H, C₆H₅), 2.01 – 0.89 (m, 70H, P(C₆H₂₂)₃, SCH₂CH₂S). 31P NMR (CD₂Cl₂): δ 93.96 (bs), 62.53 (s) at 20 °C. IR (CH₂Cl₂): ν_CO = 1953, 1901 cm⁻¹. FD-MS: m/z 964.4 ([Fe₂(S₂C₂H₄)(CO)₃(PCy₃)(dppv)]⁺). Anal. Calcd for C₄₉H₅₉Fe₂O₃P₃S₂ (Found): C, 61.00 (61.38); H, 6.16 (6.50).

[Fe₂(S₂C₂H₄)(CO)₃(PCy₃)(dppv)]BF₄ ([4pdt]BF₄)

A solution of 0.050 g (0.052 mmol) of Fe₂(S₂C₂H₄)(CO)₃(PCy₃)(dppv) in 5 mL of CH₂Cl₂ at −45 °C was treated with 14 mg (0.052 mmol) of FcBF₄ in 3 mL of CH₂Cl₂. IR (CH₂Cl₂): ν_CO = 1987 (m), 1961 (s), 1870 (br) cm⁻¹.

Fe₂(S₂C₂H₄)(CO)₃(Pi-Pr₃)(dppv) (5edt)

A solution of 0.300 g (0.421 mmol) of Fe₂(S₂C₂H₄)(CO)₄(dppv) in 40 mL of toluene was treated with 0.27 mL (2.105 mmol) of Pi-Pr₃. The reaction mixture was photolyzed with a 100 W UV immersion lamp, λ_max = 356 nm (Spectroline) until the conversion was complete (~24 h) as indicated by IR spectra. The solvent was removed in vacuo, and the product was extracted into ~20 mL of CH₂Cl₂. The solution was concentrated to ~5 mL, and the red-brown product was precipitated upon addition of 40 mL of hexanes to this solution. Yield: 0.268 g (75%). ¹H NMR (CD₂Cl₂): δ 7.98 – 7.27 (m, 20H, C₆H₅), 2.00 (m, 3H, P(CH)Me₂), 1.56 (m, 2H, SCH₂), 1.26 (m, 18H, PCH(CH₃)₂), 1.18 (s, 2H, SCH₂). 31P NMR (CD₂Cl₂): δ 92.96 (bs), 71.74 (s) at 20 °C. 95.9 (d, J_P-P = 25 Hz), 91.6 (d, J_P-P = 25 Hz), 71.1 (s) at −60 °C. IR (CH₂Cl₂): ν_CO = 1954, 1903 cm⁻¹. FD-MS: m/z 844.4 ([Fe₂(S₂C₂H₄)(CO)₃(Pi-Pr₃)(dppv)]⁺). Anal Calcd for C₄₀H₄₇Fe₂O₃P₃S₂ (Found): C, 56.89 (56.97); H, 5.61 (5.45).

Fe₂(S₂C₃H₆)(CO)₃(Pi-Pr₃)(dppv) (5pdt)

See preceding preparation. Yield: 71%. ¹H NMR (CD₂Cl₂): δ 8.00 – 7.30 (m, 20H, C₆H₅), 2.38 (m, 3H, PCHMe₂), 1.67 (s, 2H, SCH₂), 1.55 (s, 2H, SCH₂CH₂CH₂S), 1.37 (m, 18H, PCH(CH₃)₂), 1.25 (s, 2H, SCH₂). ³¹P NMR (CD₂Cl₂, 20 °C): δ 90.69 (s), 72.94 (s). At −80 °C, the δ 90.69 signal is broad. IR (CH₂Cl₂): ν_CO = 1954, 1901 cm⁻¹. FD-MS: m/z 858.3 ([Fe₂(S₂C₃H₆)(CO)₃(Pi-Pr₃)(dppv)]⁺). Anal Calcd for C₄₁H₄₉Fe₂O₃P₃S₂ (Found): C, 57.36 (57.01); H, 5.75 (5.69).

[Fe₂(S₂C₂H₄)(CO)₃(Pi-Pr₃)(dppv)]BF₄ ([5edt]BF₄)

A solution of 0.050 g (0.052 mmol) of Fe₂(S₂C₂H₄)(CO)₃(Pi-Pr₃)(dppv) in 5 mL of CH₂Cl₂ at −45 °C was treated with 14 mg (0.052 mmol) of FcBF₄ in 3 mL of CH₂Cl₂. Due to the instability of [5edt]BF₄, all data were collected in-situ. IR (CH₂Cl₂): 3_CO = 1987 (m), 1968 (s), 1870 (br) cm⁻¹.

When CH₂Cl₂ solutions of [5edt]BF₄ were exposed to CO (1 atm at −45 °C). The in-situ IR spectrum showed the disappearance of [5edt]BF₄ and the growth of new peaks, attributed to [Fe₂(S₂C₃H₆)(μ-CO)(CO)₃(Pi-Pr₃)(dppv)]BF₄, [5edtCO]BF₄. Due to the instability of [5edtCO]BF₄, all data were collected in-situ. IR (CH₂Cl₂): ν_CO = 2022, 1987, 1968, 1787 cm⁻¹. The reaction of CO with [5edt]⁺ to form [5edtCO]⁺ required ~30 min.

[Fe₂(S₂C₃H₆)(CO)₃(Pi-Pr₃)(dppv)]BF₄ ([5pdt]BF₄)

A solution of 0.150 g (0.175 mmol) of Fe₂(S₂C₃H₆)(CO)₃(Pi-Pr₃)(dppv) in 20 mL of CH₂Cl₂ at −45 °C was treated with 0.047 mg (0.175 mmol) of FcBF₄ in 10 mL of CH₂Cl₂. The solution was transferred via cannula to 400 mL of hexane cooled to −78 °C to precipitate the red-brown solid. Single crystals were grown by diffusion of a layer of hexane into a CH₂Cl₂ solution at −20 °C. IR (CH₂Cl₂): ν_CO = 2011 (sm), 1988 (m), 1968 (s), 1830 (br) cm⁻¹. IR of rapidly precipitated solid (KBr): ν_CO = 1998 (m), 1958 (s), 1904 (sm), 1824 (br) cm⁻¹. IR of pulverized single crystals (KBr): ν_CO = 1990 (s), 1931 (s), 1903 (m) cm⁻¹.

[Fe₂(S₂C₂H₄)(CN)(CO)₃(dppv)]₂ ([6edt]₂)

A solution of 0.050 g (0.0525 mmol) of Bu₄N[Fe₂(S₂C₂H₄)(CN)(CO)₃(dppv)] in 5 mL of CH₂Cl₂ was treated with 0.014 g (0.0525 mmol) of FcBF₄ in 5.0 mL of CH₂Cl₂. Solvent was removed in vacuo, and the product was extracted into 20 mL of toluene. The orange solution was concentrated to 5 mL, and the product precipitated upon addition of 30 mL of hexanes. ³¹P NMR (CD₂Cl₂, 20 °C): δ 92.96 (bs), 71.74 (s). ³¹P NMR (CD₂Cl₂, −60 °C): 99.1 (d, J_P-P = 21 Hz), 92.7 (d, J_P-P = 21 Hz), 75.9 (s), 74.0 (s). IR (CH₂Cl₂): ν_CN = 2136, 2116 cm⁻¹. IR (CH₂Cl₂): ν_CO = 2074 (sh), 2058 (s), 2016 (m), 1975 (m), 1952 (s), 1901 (s) cm⁻¹.