Iron-sulfur carbonyls such as [(μ-S)₂Fe₂(CO)₆] [1], [(μ-SR)₂ Fe₂(CO)₆] [2] and [(μ-S)(μ-SR)Fe₂(CO)₆] [3] are very known in organometallic chemistry, and bioinorganic chemistry as well [4]. In particular, the active site of [FeFe]-H₂ase, also known as H-cluster, is composed of a {2Fe2S} core that is attached by a redoxactive [Fe₄S₄] cluster [5]. As a family of bimetallic hydrogenase, it catalyzes the interconversion of H₂, electrons and protons at remarkably fast rate and with a low over potential [6]. In {2Fe2S} core, the diiron centers are assembled by an azadithiolate ligand (SCH₂NHCH₂S^-) [7] and a CO bridging ligand, while each iron center is coordinated by one CO and one CN^- ligand [8]. Overall, the active site can be formulated as [(Cys-S-Fe₄S₄)(NC)(OC)Fe {(SCH₂)₂NH}Fe(CO)₂(CN)], and the key structural feature is the rotated Fe_d center that exhibits a vacant coordination site trans to the bridging CO for binding dihydrogen or a hydride ligand during the catalysis (Fig. 1a). Since the reports of the crystallographic structure of the active site in the late of 1990s, intensive efforts have been devoted to the synthetic models of [FeFe]-H₂ase aiming at exploring efficient iron-based catalysts for H₂ production [9]. The modeling efforts have focused on the substitution CO ligands of parent complex [(μ-SR)₂ Fe₂(CO)₆] by a variety of more basic ligands such as phosphines [10], cyanide [11], N-heterocyclic carbenes [12], providing synthetic models of [(μ-SR)₂ Fe₂(CO)_6-xL_x] type (Fig. 1b).

The phosphine ligands have been found to exhibit a range of steric and electronic effects to the diiron center in [(μ-SR)₂ Fe₂(CO)_6-x(PR₃)_x] class. For example, symmetrical bisPMe₃ compound [Fe₂(pdt)(CO)₄(PMe₃)₂] (pdt = CH₂(CH₂S^-)₂) allows its Fe(Ⅰ)-Fe(Ⅰ) bond to undergo protonation giving the bridging hydride [(μ-H)Fe₂(pdt)(CO)₄(PMe₃)₂]⁺ [13]. The cationic diferrous bridging hydrides [Fe(Ⅱ)HFe(Ⅱ)]⁺ can be reduced to the neutral mixed-valence Fe(Ⅰ)HFe(Ⅱ) hydrides, which are susceptible to protonation resulting in H₂ evolution [14]. Protonating unsymmetrically substituted compounds such as [(CO)₃Fe(pdt)-Fe (dppv)(CO)] (dppv = cis-1, 2-C₂H₂(PPh₂)₂) bearing a chelating diphosphine ligand at one of the iron centers gives terminal hydrides [(t-H)Fe(Ⅱ)Fe(Ⅱ)]⁺ at low temperatures [15]. More likely as a kinetic product, the terminal hydrides transform to the thermodynamically stable bridging hydride [(μ-H)Fe(Ⅱ)Fe(Ⅱ)]⁺. The life time of the terminal hydride species can be prolonged by the increase in steric bulk of the phosphine ligands [16], and even be stabilized by the introduction of an anionic ligand C₅Me₅^-, i.e., [(C₅Me₅)Fe(pdt)Fe(CO)(dppe)H] [17]. Synthesis of [Fe(Ⅰ)Fe(Ⅰ)] diiron compounds with the rotated structure is challenging. The possibilities have been explored by using bulky dithiolate ligands with the combination of unsymmetrical ligand substitution [18]. Models of [Fe(Ⅰ)Fe(Ⅱ)] with a rotated configuration have been reported by one-electron oxidation of the unsymmetrical substituted Fe(Ⅰ)Fe(Ⅰ) precursors such as [(IMes)(CO)₂Fe(SC₃H₆S)Fe (PMe₃)(CO)₂] (IMes = 1, 3-bis(2, 4, 6-trimethylphenyl)imidazol-2- ylidene) [19], and [(PMe₃)(CO)₂Fe(SC₂H₄S)Fe(CO)₂(dppv)] [20]. The reactivity of [Fe(Ⅰ)Fe(Ⅱ)]⁺ models is indeed interesting, because it has been found that in the presence of base oxidizing [Fe(Ⅰ)Fe(Ⅱ)]⁺ models enables H₂ splitting [21]. Especially, the CO ligands in [Fe(Ⅰ)Fe(Ⅱ)]⁺ compounds can be replaced by more basic ligands such as phosphines, providing new synthetic complexes for further investigation [22].

Although the {2Fe2S} core of the active site contains a sulfur ligated Fe center (Fe^d, Fig. 1a), synthetic {2Fe3S} models with a thioether ligand have been rarely reported [23]. In this paper, we report a series of {2Fe3S} complexes bearing phosphino thioether chelating ligand (1, 2-Cy₂PC₆H₄SMe) based on [Fe₂(Me₂pdt)(1, 2-Cy₂PC₆H₄SMe)(CO)₄] (Me₂pdt = Me₂C(CH₂S^-)₂, 1). Compound 1 exhibits reversible one-electron redox even for [Fe(Ⅰ)Fe(Ⅱ)]^+/0 couple. The synthesis of tri-substituted [Fe(Ⅰ)Fe(Ⅱ)]⁺ cationic complex [Fe₂(Me₂pdt)(1, 2-Cy₂PC₆H₄SMe)(PPh₃)(CO)₃]⁺ ([2]⁺) was achieved by installing a phosphine ligand on the in-situ oxidized form of the [Fe(Ⅰ)Fe(Ⅰ)] precursors (Scheme 1). Reduction of [2]⁺ provides the neutral tri-substituted compound 2. The reactivity of 1 and 2 towards different strength of acids were investigated.

We initially attempted to install 1, 2-Cy₂PC₆H₄SMe on the parent compound [Fe₂(Me₂pdt)(CO)₆] by nucleophilic abstraction of carbonyl using Me₃NO [24]. However, this reaction resulted in a mixture of 1 and the undesired mono-substituted product [Fe₂(Me₂pdt)(1, 2-Cy₂PC₆H₄SMe)(CO)₅]. When 1, 2-Ph₂PC₆H₄-SMe was used as the ligand, the mono-substituted product [Fe₂(Me₂pdt)(1, 2-Ph₂PC₆H₄SMe)(CO)₅] was solely produced (Fig. S16 in Supporting information).

Alternatively, we synthesized 1 by irradiation (λ = 365 nm) of a mixture of [Fe₂(Me₂pdt)(CO)₆] and 1, 2-Cy₂PC₆H₄SMe (1:1 ratio) in toluene (Scheme 2). After four hours of photolysis, FI-IR spectroscopy indicated complete conversion of [Fe₂(Me₂pdt)- (CO)₆] to the product. After evaporating the solvent, the compounds were readily purified by recrystallization from CH₂Cl₂ by layering hexane and isolated as brown solids (ca. 87% yields). The IR spectrum of 1 displays its ν_CO bands at 2010, 1939, and 1894 cm^-1, which is ~10 cm^-1 lower in energy compared to [Fe₂(pdt)(dppbz)(CO)₄] (dppbz = 1, 2-bis(diphenylphosphanyl)benzene) (ν_CO = 2019, 1949 and 1906 cm^-1) [25] and [Fe₂(pdt)(dppv)(CO)₄] (ν_CO = 2021, 1950 and 1912 cm^-1) [24].

Similar to the reported diphosphine-based compounds [Fe₂(pdt)(P-P)(CO)₄] (P-P = chelating diphosphine ligand) [18b, 24], complex 1 also exhibit isomers as indicated by ³¹P NMR spectroscopic studies. At room temperature, the ³¹P NMR spectrum displays two sets of resonances as one broad peak at δ 90.7 and a singlet at δ 78.2 in a ratio of 1:4. The binding of phosphine atom at [Fe(P-SR)(CO)] subunit can either be apical or basal leading to the existence of at least three isomers as shown in Fig. 2. The broad phosphorus resonances at δ 90.7 are tentatively assigned for the apical-basal isomers, while the sharp singlet is assigned to the basal-basal isomer. The broad phosphorus resonances might arise from the rapid turnstile rotation of the phosphino thioether ligand bound at the iron center in an apicalbasal fashion (Fig. 2). To confirm the existence of such a dynamic movement, we have conducted variable-temperature ³¹P NMR experiments for 1 (Fig. 3). Upon cooling of the sample from 293 K to 213 K, the signal at δ 90.7 ppm shifted to δ 95.0 ppm, whereas the singlet retained at δ 78.2 ppm. Importantly, variable-temperature ³¹P NMR spectra showing coalescences and decoalescences suggest the rapid interconversion between a(P), b(S)-1 and a(P), b (S)-1 for the apical-basal isomers. The structure for the basal-basal isomer b, b-1 was confirmed by single crystal X-ray diffraction (Fig. 4).

In CH₂Cl₂, cyclic voltammogram of 1 shows a reversible oxidation event at -0.28 V (vs. Fc/Fc⁺) for the Fe(Ⅰ)Fe(Ⅰ)/Fe(Ⅰ)Fe(Ⅱ) couple (Fig. S18 in Supporting information). The oxidation potential is comparable to -0.34 V for [Fe₂(Me₂pdt)(PMe₃)₂(CO)₄] [26], and is slightly more negative than -0.19 V for [Fe₂(pdt)(dppv) (CO)₄] [27]. For the unsymmetrical substituted [Fe(Ⅰ)Fe(Ⅰ)] precursors, one-electron oxidation provides [Fe(Ⅰ)Fe(Ⅱ)]⁺ cationic complex featuring a bridging or semi-bridging CO ligand [19a, 23b, 26]. The oxidation of 1 was subsequently examined by using FcBF₄ oxidant. When the CH₂Cl₂ solution of 1 was treated with an equivalent of FcBF₄, the color of the solution immediately turned to purple from brown. In IR spectra, the new ν_CO bands were observed at 2079, 2021 and 1890 cm^-1. The Δν_CO is ~45 cm^-1 higher than that of 1, which is consistent with the one-electron oxidation reaction in the preparation of [Fe(Ⅰ)Fe(Ⅱ)]⁺ cationic complexes [22d, e]. Due to the semibridging CO, there is a vacant site in apical position of Fe1 (16e^-) in [Fe(Ⅱ)Fe(Ⅰ)]⁺ models [21b]. Different from the dynamic movement of 1, we conclude that the dibasal fashion of [1]⁺ would not turn to apical-basal fashion in order to keep the rotated square-pyramidal structure. However, the attempts to isolate the [Fe(Ⅰ)Fe(Ⅱ)]⁺ mixed-valent species of [1]⁺ was unsuccessful.

Notably, the [Fe(Ⅰ)Fe(Ⅱ)]⁺ specie can be stabilized by the introduction of mono-phosphines such as PPh₃. To the solution of [1]⁺ generated in situ from 1 with FcBF₄ in CH₂Cl₂, was added one equivalent of PPh₃ affording [2]⁺. The cationic salts [2]⁺ was isolated as dark-green solids in 86% yield. Three new ν_CO bands at 2020, 1989 and 1880 cm^-1 were displayed in the IR spectrum (Fig. 5), and the weak broad band at 1880 cm^-1 is corresponding to the ν_μ-CO. The ν_CO of [2]⁺ is ~55 cm^-1 higher than that of [1]⁺. Complex [2]BF₄ does not undergo further substitution in the presence of excess of PPh₃.

X-ray crystallographic analysis reveals the expected rotated structure for the cationic complex [2]⁺, which features a semibridging CO ligand (Fig. 6). The Fe(1)-C(1) distance is 1.771(7) Å, and about 0.71 Å shorter than Fe(2)-C(1) (2.480(6) Å). The phosphine ligand PPh₃ is bound at apical position of [Fe(CO)₂(PPh₃)] subunit, while the phosphino thioether ligand chelating at the dibasal position of the other iron center. The angle of Fe2-Fe1-C1 (66.86 (19)°) is much smaller than those in [Fe₂(pdt)(dppv)(CO)₃(PR₃)]⁺ (R = Me, 73.20(13)°; i-Pr, 78.55(16)°) [22e].

Surprisingly, cyclic voltammogram of [2]⁺ indicated that the substitution of the CO ligand in 1 by PPh₃ results in an anodic shift of the Fe^ⅡFe^Ⅰ/Fe^ⅠFe^Ⅰ couple of 470mV. The reversible redox event at -0.75 V (i_pa/i_pc = 0.95, vs. Fc⁺/Fc) suggests the possibility of bulk chemical reduction and isolation of 2. Treatment of [2]⁺ with one equivalent of Cp₂Co (-1.33 V in CH₂Cl₂ vs. Fc⁺/Fc) [28] in CH₂Cl₂ solution afforded 2, which was isolated as brown solids in high yield. In IR spectrum, the ν_CO bands were shown as two broad peaks at 1954 (br), 1887 (br) cm^-1. The ³¹P NMR spectrum of 2 consists of two groups of singlets at δ 76.1 (P-SR), 54.9 (PPh₃) and δ 87.2 (P-SR), 56.9 (PPh₃) at room temperature, indicating the existence of two isomers. In addition, the formation of 2 was not observed by irradiation of 1 with PPh₃ in toluene judging from ³¹P NMR spectra.

With the addition of HBF₄·Et₂O to the CH₂Cl₂ solution of 2, the phosphorous resonances were displayed as two singlets at δ 80.5 for the phosphino thioether ligand (P-SR) and δ 52.7 for PPh₃. In the ¹H NMR spectrum, the hydride signal was observed at -11.9 ppm (t, J_PH = 20 Hz), indicating the formation of bridging hydride complex [H2]BF₄. Its structure was confirmed by X-ray crystallographic studies (Fig. 7). In [H2]⁺, the phosphino thioether ligand retains the dibasal chelating mode while the PPh₃ resides at the basal position at [Fe(CO)₂(PPh₃)] subsite. The PPh₃ binding at the basal site dues to the strong trans effect of CO to H^- in [H2]⁺ [29]. The hydride ligand was located in the difference map, and its position was refined. Judging from the bond distances of Fe1-H 1.72 Å and Fe2-H 1.47 Å, the μ-H is semi-bridging.

The most striking difference between 1 and 2 is their reactivity towards different strength of acids. Protonating the di-substituted compound 1 requires strong acids such as HBF₄·Et₂O (pK_a^MeCN = 0.2) [30] and CF₃SO₃H (pK_a^MeCN = 2.6) [31]. When [HPPh₃]BF₄ (pK_a^MeCN = 7.6) [32] was used, there was no hydride signal observed in ³¹P NMR spectrum. By contrast, the tri-substituted compound 2 reacts with CCl₂COOH (pK_a^MeCN = 13.2) [33] producing [H2]⁺. The pK_a^MeCN of the Fe-Fe bond in 2 was estimated to be in the range of 13.2–14, because it cannot be protonated by 2, 6-Me₂py-H⁺ (pK_a^MeCN = 14.2) [34]. The ΔpK_a^MeCN between 1 and 2 is about 10. It is noteworthy that the tetra-phosphine substituted diiron complex [Fe₂(adt^NH)-(CO)₂(dppv)₂] possesses the pK_a^MeCN in between 15.3 and 17.6 [34].

A series of {2Fe3S} complexes bearing phosphino thioether chelating ligand were synthesized and well characterized. As well as the unsymmetrical substituted [Fe(Ⅰ)Fe(Ⅰ)] compounds with diphosphine chelating ligands, the phosphino-thioether-based disubstituted diiron compound 1 exhibits a reversible one-electron redox event for [Fe(Ⅰ)Fe(Ⅱ)]^+/0 couple. The synthesis of trisubstituted [Fe(Ⅰ)Fe(Ⅱ)]⁺ cationic complex [2]⁺ was achieved by adding a phosphine ligand to [1]⁺ generated in situ by the reaction of 1 with FcBF₄. Reduction of [2]⁺ provides the neutral trisubstituted [Fe(Ⅰ)Fe(Ⅰ)] compound 2. The substitution of the CO in 1 ligand by PPh₃ results in an anodic shift of the Fe^ⅡFe^Ⅰ/Fe^ⅠFe^Ⅰ couple of 470 mV. Acid-base features of [Fe-Fe] model complexes are important properties [35-37]. Both 1 and 2 undergo protonation affording bridging hydride complexes. Most importantly, this substitution also leads to the Fe-Fe bonds in 1 and 2 with large Lewis basicity difference, i.e. ΔpK_a^MeCN ~10.

Acknowledgments

We gratefully acknowledge the financial support from the "1000 Youth Talents Plan", the Natural Science Foundation of China (Nos. 21402107, 91427303). We also thank Prof. Di Sun for assistance with the X-ray crystallography. S. Raje acknowledges CSIR, India, for a senior research fellow ship. R. Angamuthu thanks IIT Kanpur for funding.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, in the online version, at https://doi.org/10.1016/j.cclet.2018.03.013.