Improved Access to ‘Butterfly’ Di-Iron Dithiolates Fe2(μ- SR)2(CO)6 and their Mono- and Bis(phosphine) Adducts

A series of Fe2(μ‐SR)2(CO)6 complexes {R=Me (1Me), Et (1Et), Pr (1Pr), iPr (1iPr), tBu (1tBu), PhCH2 (1Bn) and Ph (1Ph)} have been synthesised. Complexes 1Me, 1tBu, 1Bn and 1Ph were produced by addition of S2R2 to Fe3(CO)12, with all but 1tBu giving excellent yields. Two isomers of 1Me and 1Ph were isolated: the anti‐ and ‘open’ syn‐products. Complexes 1Et, 1Pr and 1iPr were synthesised by treatment of RSH with Fe3(CO)12; two isomers of each complex were isolated. Addition of one equivalent of PR’3 (R’=Me, Cy, Ph) yields the corresponding mono(phosphine) adducts, whilst use of two equivalents of the phosphine (under mild condition, reflux, or irradiation using a deep blue LED depending on SR group) affords the corresponding bis(phosphine) adducts in good to excellent yield. Treatment of 1Ph or 1Me with two equivalents of PMe3 gives the corresponding bis‐substituted phosphines when carried out in the absence of light but leads to oxidative cleavage to Fe(μ‐SPh)2(PMe3)2(CO)2 and Fe(μ‐SMe)2(PMe3)2(CO)2, respectively, under blue light irradiation. Treatment of 1Pr with two equivalents of PCy3 under blue light irradiation leads to reductive breakdown of the Fe−Fe bond to yield Fe(CO)3(PCy3)2, but in the dark at room temperature the desired product Fe2(μ‐SPr)2(PCy3)2(CO)4 may be isolated. Single crystal X‐ray structures were obtained for most family members of ‘butterfly’ {Fe2S2} cores. Cyclic voltammetry shows PMe3‐containing complexes undergo irreversible oxidation, whereas both PCy3 and PPh3 complexes show one (quasi)reversible oxidation, IR of in situ protonation showed COv blue shifting around 80–100 cm−1, while 31P{1H} NMR spectroscopy showed shifting to low field.


Introduction
Alternatives to classical fossil fuels continue to attract significant interest in the scientific community, with dihydrogen as one of the leading examples. For H 2 to deliver on its potential to be one of the 'clean' fuels projected for future use, efficient, scalable, and sustainable methods for generating the gas are required. Whilst platinum-based methods are potentially very efficient, they are neither scalable nor sustainable. Similarly, steam reforming fossil fuels to generate hydrogen simply moves the unsustainable chemistry from the point of use to point of generation. In contrast, the [FeFe]-hydrogenases can produce dihydrogen by proton reduction using only earthabundant metals and (in principle) without needing any fossil fuel input. [1,2] The challenge is therefore to take this biological chemistry and mimic it in small, scalable, synthetic systems. [3] Whilst accessing fully elaborated models for the core of the [FeFe]-hydrogenase requires a significant amount of imaginative chemistry, routes to Fe 2 (μ-SR) 2 (CO) 6 , particularly where there is a bridge between the sulfur atoms, are more tractable.
These hexacarbonyl complexes are not sufficiently electron-rich to undergo protonation with typical mineral or organic acids, although they can be protonated using super-acids. Introduction of electron-donating ligands, particularly phosphines, increases the reactivity of the metal centres.
To date, most research efforts have been directed toward sulfur-bridged synthetic models, primarily as they are closer to the structure of the enzyme. This area has been extensively reviewed, and the interested reader is directly to the many excellent articles covering the 'butterfly' systems. [4,5] However, there is now increasing interest in the 'open' (unbridged) analogues, which may offer unique reactivity, and which are also potential starting materials for novel chemistry in their own right. [6][7][8][9][10] Complexes of general formula Fe 2 (μ-SR) 2 (CO) 6 have been synthesised by many methods: (i) treatment of Fe 3 (CO) 12 with RSH derivatives in several hydrocarbon solvents with reflux [11] or photochemically, [12] (the mechanism is believed to involve formation of Fe 3 (CO) 9 (H)(RS) [13] ) (ii) the Hieber-Gruber reaction, [14] (iii) treatment of (C 5 H 4 SPh) 2 Fe with Fe 2 (CO) 9 under photolytic conditions, [15] (iv) reduction of Fe 2 (μ-S) 2 (CO) 6 by Na or KH followed by treatment with alkyl halide to give the corresponding derivatives, [16] (v) treatment of Fe(CO) 5 with RSH or RSSR under CO atmosphere, [17] and (vi) King's method by the reaction of RSSR with Fe 3 (CO) 12 under reflux. [18] Here, we have explored optimisations of King's method and subsequent introduction of phosphine in 'open' systems featuring aromatic and the aliphatic groups at the sulfur centres, and the effect of phosphine substituted on structural parameters, carbonyl stretching, and voltammetry. In our hands, the use of King's approach gave the highest yields of reliably pure material.

Results and Discussion
Synthesis of Fe 2 (μ-SR) 2 (CO) 6 To study the effect of open aliphatic and aromatic substituted thiolate bridging of {2Fe2S} 'butterfly' systems, King's method was utilised to prepare Fe 2 (μ-SR) 2 (CO) 6 (1) (R=Me, Et, Pr, iPr, tBu, Ph, CH 2 Ph). Compounds Fe 2 (μ-SMe) 2 (CO) 6 (1 Me ) Fe 2 (μ-StBu) 2 (CO) 6 (1 tBu ), Fe 2 (μ-SBn) 2 (CO) 6 (1 Bn ) and Fe 2 (μ-SPh) 2 (CO) 6 (1 Ph ) were synthesised by reaction of Fe 3 (CO) 12 with the appropriate dialkyl disulfide under reflux (Scheme 1). In the case of 1 Me and 1 Ph , two isomers could be identified by thin layer chromatography (TLC). Separation of the two required that the reaction was carried out with a starting material ratio of 1 : 1, to avoid contamination by unreacted disulfide. This yielded anti-1 (one R 'up', one 'down') as the major isomer in a ratio of around 9 : 1 to the syn isomer (both R groups 'down') in 1 Ph , and 3 : 1 in 1 Me . IR spectroscopy could distinguish the two isomers of 1 Ph , with a diagnostic peak at 1943 cm À 1 for the syn isomer contrasting with one at 1979 cm À 1 in the anti product. However, IR shows identical CO stretches for the two isomers of 1 Me , likely due to the small size of the methyl substituent. Proton NMR spectroscopy of anti-1 Me in CDCl 3 shows a singlet at δ 1.54 ppm arising from the axial Me group, with the equatorial methyl groups exhibiting a signal at δ 2.07 ppm. The syn isomer in contrast shows a singlet at δ 2.03 ppm. The identities of both isomers were unambiguously confirmed by X-ray crystallography, in accord with previous reports of synthesis 1 Me and 1 Ph . [19,20] In contrast to earlier reports, [21] in our hands the isolated byproduct of the hexacarbonyl synthesis of 1 Me , 1 Ph and 1 Bn was the tri-iron complex 2.
The synthesis of 1 tBu was best achieved by reaction of Fe 3 (CO) 12 with two equivalents of (tBuS) 2 under reflux for 30 minutes. This led to isolation of three separate red products in addition to recovery of starting material: purification by column chromatography gave 1 tBu in 30 % yield and Fe 3 (μ-tBuS) 2 (CO) 9 (3) in 17 % yield, along with 10 % of recovered Fe 3 (CO) 12 and an unidentified brown solid material which did not elute under the conditions used (Scheme 2). Refluxing pure 3 in THF for 30 min produced 1 tBu in around 65 % yield, and a brown precipitate similar to that seen in the direct reaction, suggesting that 3 is an intermediate on the reaction pathway.
Treatment of Fe 3 (CO) 12 with two equivalents of RSH was the most effective route to the complexes 1 Et , 1 Pr , and 1 iPr , all of which were obtained in excellent yield (Scheme 3). Two isomers were observed for each of these materials, in syn : anti ratios of 3 : 1, 4 : 1 and 4 : 1, respectively. Proton NMR spectroscopy showed that the axial and equatorial alkyl group signals can be distinguished, which can clearly be attributed to the increasing of shielding due to the steric effect (see Supporting Information).

Synthesis of mono-substituted phosphine derivatives
Replacement of the CO groups in Fe 2 (μ-SR) 2 (CO) 6 is of broad interest as it allows control of the electronic properties of the iron centres, in particular in mimicking the active side of [FeFe]hydrogenase enzymes. [1,2,20,21] Mono-substituted phosphine {2Fe2S} complexes are normally synthesised by using Me 3 NO · 2H 2 O as decarbonylation reagent; [22][23][24] these reactions are typically carried out at room temperature. However, reaction of 1 proceeded without the need for active decarbonylation, irrespective of the alkyl substituent involved. Addition of one equivalent of phosphine to solutions of the hexacarbonyls proceeded smoothly either at room temperature (PMe 3 ) or under reflux/blue (455 nm) light irradiation (PPh 3 , PCy 3 ) (Scheme 4). Yields in these reactions were moderate to high and did not appear to be compromised by the omission of Me 3 NO · 2H 2 O. By surveying the conditions for these reactions, we have established reliable and (moderately) high-yielding access to the full family of these complexes.
Treatment of 1 Bn , 1 Ph and 1 tBu with one equivalent of PMe 3 at room temperature cleanly produce the corresponding substitution phosphine, confirmed by the presence of low-field singlets in the 31 P{ 1 H} NMR spectra (δ 13.2 ppm, 25.6 ppm and 34.3 ppm, respectively, for 4a Bn , 4a Ph and 4a tBu ). 1 H NMR spectroscopy shows that 4a Bn exhibits two isomers in solution, in a 6 : 1 ratio in CDCl 3 and approximately 1 : 1 in CD 2 Cl 2 , driven by ready interchange of isomers and indicating a 'flat' energy surface in which small solvent interactions can tip the energy balance.
Moving to the synthesis of mono-PCy 3 substituted derivatives, treatment of 1 Me , 1 Et , 1 iPr , 1 tBu , 1 Ph and 1 Bn with one equivalent of PCy 3 under reflux give the corresponding monosubstituted complexes Fe 2 (μ-SR) 2 (PCy 3 )(CO) 5 (4 b), 31 P{ 1 H} NMR spectra in CDCl 3 showed singlet signals for the bound phosphine in the range 55-66 ppm. The equivalent mono-PPh 3 complexes 4 c could also be accessed by refluxing the starting materials in a 1 : 1 ratio. NMR spectroscopy showed that 4c Me , 4c Et , 4c Pr and 4c iPr exists in two isomers in CDCl 3 solution.
Infrared spectroscopy for all the phosphine complexes confirmed a blue shift of the CO bands (Table S1). As anticipated, the largest shifts compared with the parent hexacarbonyl complexes were seen in the PCy 3 derivatives (65-45 cm À 1 ), with the smallest shifts for the PPh 3 systems (40-15 cm À 1 ), reflecting the electron-donating abilities of the phosphine ligands. The magnitudes of these changes are in line with those reported for other mono-phosphine substituted {2Fe2S} systems.

Synthesis of bis-substituted phosphine derivatives
Bis-substitution by monodentate phosphines is usually achieved through addition of excess phosphine to the hexacarbonyl. [21][22][23][24][25] However, this can proceed poorly, particularly when using PPh 3 , which is both bulky and a poor electron donor. Irradiating a reaction mixture containing a 1 : 2 ratio of hexacarbonyl and phosphine with a 455 nm LED gave smooth reaction to produce Fe 2 (μ-SR) 2 (PR' 3 ) 2 (CO) 4 (5) in moderate to high yields for PMe 3 , PCy 3 and PPh 3 (Scheme 5). 13 C{ 1 H} NMR spectra showed shifting of CO signals toward high frequency by around 7.0-8.0 ppm comparing with the corresponding hexacarbonyl complexes, and showed that 5a Me , 5a Et and 5a Pr existed single isomers in solution which showed triplet signals (1 : 2 : 1) due to the effect of two phosphine groups and the coupling average of J PC = 3.8 Hz. In the case of 5a Ph , two polymorphs could be isolated depending on reaction temperature, a phenomenon that has been observed in related {2Fe2Se} systems isomers. [3] Infrared spectroscopy showed a blue shift in the carbonyl stretching with increasing electron density on the metal centres (Table S2). The symmetry of the bis-phosphine complexes means that in most cases only three bands were resolved in the carbonyl region: in contrast, many of the lower-symmetry mono-phosphines exhibited four or five distinct bands.
Most hexacarbonyl complexes containing bridging dithiolene ligation are resistant to breakdown even if reacted with excess phosphine at elevated temperatures. In contrast, the 'open' hexacarbonyls are more reactive, and therefore control of conditions is important to avoid cleavage of the FeÀ Fe bond. Treatment of 1 Me with two equivalents of PMe 3 with reflux or blue light irradiation led to breakdown to the {2Fe2S} core to the Fe(II) species Fe(SMe) 2 (CO) 2 (PMe 3 ) 2 (6 Me ). In the case of 1 Ph , application of the LED route gave complex Fe(SPh) 2 (CO) 2 (PMe 3 ) 2 (6 Ph ) as a second product, but in contrast to 1 Me , this was not an issue under reflux conditions. Similar decomposition has been reported recently in the reactions of {(μ-S) 2 (C 4 N 2 H 2 )}Fe 2 (CO) 6 with excess of PMe 3, [26] and treatment of 1 Ph with dppe (dppe = 1,2-diphenylphosphinethane). [27] The treatment of 1 Pr with two equivalents of PCy 3 leads to FeÀ Fe bond cleavage via reductive elimination to give Fe(0) complex of Fe(PCy 3 ) 2 (CO) 3 7 as a bipyramidal complex. Typically, this complex is accessed by reaction of PCy 3 with Fe(CO) 5 .
Notably, we have largely been able to avoid the use of excess phosphine by appropriate selection of reaction conditions. This saving of expensive starting materials, and reduction in the formation of by-products, means that our reported protocols are preferable to those reported previously: minor adjustments to conditions here make a significant difference in synthetic utility.

X-ray crystallography
Single crystal X-ray diffraction studies were carried out for most complexes reported here and were used to confirm the anticipated structures (see Supporting Information). A summary of the data sets collected, included details of those previously reported, is given in Table 1.
The structures for the hexacarbonyls are unremarkable, and as detailed in Table 1, the majority have been reported previously. The structures confirm the isolation of both syn-and anti-isomers for 1 Ph , 1 Me , 1 Et and 1 iPr ; complexes 1 Bn and 1 tBu were isolated only as the anti isomers. Structural data for syn-1 Et and syn-1 iPr has not previously been reported, whilst for anti-1 iPr , a novel polymorph was obtained. The metrical data for these previously unreported structures are comparable with the known materials, [6,20,28,29] suggesting that a relatively flat energy surface for interchange of the sulfur-substituents exists, and the preferred polymorph form is driven by temperature and concentration factors.
In contrast to the hexacarbonyls, most of the phosphinecontaining structures have not previously been reported: only the Me and Et-substituted thiolates have been extensively studied in the past. Example ORTEP representations are shown in Figure 1 (monophosphine complexes featuring the tBu thiolate) and Figure 2 (bulky bis-phosphines).
The metrical parameters for the set of structures are in accord with previous reports [7,[30][31][32][33][34] and are available as Supporting Information. The steric effect of phosphine substitution largely forces the complexes to the syn-geometry, the only exceptions being 4c Me and 4c Et . In these systems, the small size of the alkyl group combines with the π-acceptor nature of PPh 3 to enable access to the anti-isomers.

Electrochemistry and protonation activity
To investigate the effect of phosphine ligands on the electrochemical properties of the diiron centres, cyclic voltammograms (CVs) of the phosphine complexes were recorded in CH 2 Cl 2 under an argon atmosphere. We have previously established that oxidation potential for 'butterfly' {2Fe2S} complexes correlates with protonation rates for these systems. [35] We therefore examined the oxidation potentials of the novel 'open' systems described here. Cyclic voltammetry showed the inverse relationship between E ox and the electron density surrounding the metal centre. The PMe 3 -substituted complexes (4a Bn , 4a Ph , 5a Bn , 5a Ph ) show only irreversible oxidation, while, with the exception of 4c Ph and 4a iPr , the PPh 3 and Pcy 3 systems show one (quasi)reversible oxidation ( Table 2, Supplementary Information). Compared with our earlier data for 'butterfly' systems containing Pme 3 ligands, these values are shifted to lower potential by up to 200 mV. We therefore anticipate that the open systems will protonate very rapidly: this will be the subject of further study. Notably, the published literature lacks comparison oxidation potentials for other 'open' {2Fe2S} centres, as the focus of electrochemistry has been almost exclusively on studies of reduction.  Protonation of several complexes was undertaken using AcOH. IR of in situ protonation of 5a Pr and 5b Ph showed CO v blue shifting from 1909 cm À 1 , 1934 cm À 1 and 1978 cm À 1 to 1996 cm À 1 , 2035 cm À 1 and 2044 cm À 1 for 5a Pr H + , and from 1904 cm À 1 , 1931 cm À 1 and 1973 cm À 1 to 2002 cm À 1 , 2043 cm À 1 and 2053 cm À 1 for 5b Ph H + , while 31 P{ 1 H} NMR spectroscopy of crude solution showed protonation of 5a Pr changing the symmetry in 5a Pr H + with shifting from δ 8.2 ppm to 17.7 ppm and 20.7 ppm, and in 5b Ph showed similarity of symmetry changing in 5b Ph H + to give two signals at δ 50.8 ppm and 62.3 ppm. These data are in accord with changes seen in 'butterfly' {2Fe2S} systems on protonation and suggest that further pursuit of these reactions may be fruitful.

Conclusions
In conclusion, we report here a facile route to pure 'open' Fe 2 (μ-SR) 2 (CO) 6 systems and improving the access of mono-and bissubstitution of PR 3 ligands without using decarbonylation reagent. Comparison of synthesis routes, the effect of symmetry on carbonyl stretching (C�O bonds), the solid-state structures, oxidative electrochemistry and the protonation behaviour with AcOH have been reported. Some of the by-products of synthesis of hexacarbonyl complexes have been isolated and  characterized, in addition to the isolation of FeÀ Fe bond decomposition products.

Experimental Section General
All reactions were carried out under argon atmosphere using standard Schlenk techniques. Reagent-grade solvents were dried using appropriate drying agents and distilled prior to use by standard methods. Starting materials were purchased from Aldrich or Alfa Aesar and did not require further purification. NMR spectra were recorded on Bruker AvanceIII 400 or a Bruker AvanceIII 500. NMR spectra were referenced using the residual solvent peak ( 1 H: 7.26 δ; 13 C 77.16 δ). IR spectra were recorded using a Bruker Vertex 80 instrument using a 4 cm À 1 resolution. Mass spectra were obtained using an Atmospheric Solids Analysis Probe (ASAP) with a Single Quadrupole Time-of-Flight instrument at Newcastle University. Elemental analysis was investigated at University of Manchester. Cyclic voltammetric measurements were carried out using an Autolab PGSTAT 30 potentiostat. All measurements were made in dry degassed CH 2 Cl 2 containing 0.1 M Bu 4 NPF 6 as the supporting electrolyte. A conventional three-electrode arrangement was employed, consisting of a vitreous carbon working electrode, a platinum wire as the auxiliary electrode and a silver wire pseudoreference electrode in conjunction with the ferrocenium/ferrocene couple.

Synthesis of Fe 2 (μ-SR) 2 (CO) 6 (1)
All hexacarbonyl complexes were synthesised using the same general procedure. To a solution of Fe 3 (CO) 12 (10 mmol) in THF (200 mL) was added R 2 S 2 (2 eq) or RSH (3 eq), followed by reaction at reflux until consumption of the starting material by IR spectroscopy. The solvent was removed in vacuo, then the dark red material was purified by column chromatography using hexane or a hexaneÀ CH 2 Cl 2 mixture as eluent.

General Procedure A
To a solution of 1 (0.5 mmol) in THF (30 mL) was added PR' 3 (0.5 mmol). The reaction was stirred under argon whilst illuminating with a 450 nm LED until 31 P{ 1 H} NMR spectroscopy showed consumption of free PR 3 . The solvent then removed in vacuo, and the dark red material was purified by column chromatography.

General Procedure B
To a solution of 1 (2.0 mmol) in THF (50 mL) was added PR 3 (0.5 mmol). The reaction was stirred at reflux until 31 P{ 1 H} NMR spectroscopy showed consumption of free PR 3 . The solvent then removed in vacuo, and the dark red material was purified by column chromatography.

General Procedure C
To a solution of 1 (1.0 mmol) in hexane (40 mL) was added PR' 3 (2.0 mmol), and the reaction stirred under argon atmosphere for 72 h. The concentrated solution then recrystallised at À 20°C to give the product as red blocks.

General Procedure D
To a solution of 1 (1.0 mmol) in THF (50 mL) was added PR' 3 (2.0 mmol). The solution was stirred and irradiated with a 450 nm LED until 31 P{ 1 H} NMR spectroscopy showed consumption of the free phosphine. The solvent then removed in vacuo, and the red oily material was purified by column chromatography using hexaneÀ CH 2 Cl 2 .

Synthesis of Fe 2 (μ-SPh) 2 (CO) 4 (PMe 3 ) 2 5a Ph
Two methods were used to synthesise 5a Ph . Method a: To a solution of 1 Ph (0.46 g, 1.0 mmol) in THF (50 mL) was added PMe 3 (1.0 M solution in THF, 3.0 mL, 3.0 mmol). The reaction was refluxed for 3 h to give dark red solution. The solvent was removed in vacuo, and the oily red material was extracted with hot hexane (50 mL) to give the product (0.33 g, 55 %).
Method b: To a solution of 1.0 mmol. of 1 Ph (0.46 g, 1.0 mmol) in toluene (50 mL) was added PMe 3 (1.0 M solution in THF, 2.0 mL, 2.0 mmol). The solution was stirred whilst irradiating with a 450 nm LED for 16 h to give dark red solution. The solvent was removed in vacuo, and the product was purified by column chromatography using hexaneÀ CH 2 Cl 2 (10 : 1) as eluent to give red solution. Recrystallisation by slow evaporation gave the product (0.30 g, 49 %).

X-Ray Crystallography
For each sample, crystals were suspended in oil, and one was mounted on a glass fibre and fixed in the cold nitrogen stream of the diffractometer. Data were collected using Rigaku Synergy diffractometer equipped with confocal mirrors and were processed using CrysAlisPro. [36] Structures were determined using a dual-space approach in SHELXT and refined by full-matrix least-squares methods on F 2 in SHELXL. [37] Non-hydrogen atoms were refined with anisotropic thermal parameters. Hydrogen atoms were included in idealized positions and their U iso values were set to ride on the U eq values of the parent atom.