New Phosphorous-Based [FeFe]-Hydrogenase Models

Abstract

[FeFe]-hydrogenases have attracted research for more than twenty years as paragons for the design of new catalysts for the hydrogen evolution reaction (HER). The bridging dithiolate comprising a secondary amine as bridgehead is the key element for the reactivity of native [FeFe]-hydrogenases and was therefore the midpoint of hundreds of biomimetic hydrogenase models. However, within those mimics, phosphorous is barely seen as a central element in the azadithiolato bridge despite being the direct heavier homologue of nitrogen. We herein synthesized three new phosphorous based [FeFe]-hydrogenase models by reacting dithiols (HSCH₂)₂P(O)R (R = Me, OEt, OPh) with Fe₃(CO)₁₂. All synthesized mimics show catalytic reactivity regarding HER and change their mechanisms depending on the strength of the used acid. In all presented mimics, the oxide is the center of reactivity, independent of the nature of the bridgehead. However, the phosphorous atom might be reduced by the methods we present herein to alter the reactivity of the model compounds towards protons and oxygen.

1. Introduction

[FeFe]-hydrogenases and model compounds thereof have been extensively studied over the past few years and are also investigated in regards to element homologue replacement (EHR). The active site of [FeFe]-hydrogenases, the so-called “H-cluster”, consists of a cubane-like [4Fe-4S] cluster and a [2Fe-2S] cluster, which bears additional CO and CN ligands and is bridged by an azadithiolate (adt) (Figure 1a) [1,2,3,4]. Since the groups of Happe, Lubitz and Fontecave found that [2Fe]_H can be replaced by synthetic cofactors, much effort has been made to synthesize a variety of different cofactors that can be placed into the enzyme [5,6,7]. However, only mimics that comprise an amine as the bridgehead, i.e., ADT (native activity), ADSe (100%) and a monocyanide of ADT (50%), result in an active enzyme after maturation [7,8]. Besides variation of the bridging dithiol to pdt, odt, edt and sdt, there are two more models that pick up the principle EHR. One example is Ru-ADT, in which both iron atoms are exchanged to ruthenium [9]. Although being inactive for H₂ oxidation and evolution, this variant has the advantage to trap the Hhyd state. The other example is the formerly mentioned ADSe, in which the dithiolate bridge is replaced by a diselenolate bridge [8]. This hybrid enzyme is active for proton reduction and H₂ oxidation but is more biased towards the former. The chalcogenide exchange further results in a dramatically reduced oxygen stability and, consequently, an even more sensitive H-cluster than the native cofactor.

The third moiety that can be altered by an EHR is the secondary amine of the bridging dithiolate. A phosphine bridgehead would be the next heavier homologue of this amine and most likely result in a model compound with changed oxygen tolerance, reactivity/state accessibility and proton transport activity. Nitrogen and phosphorus show very different chemical behavior, although both are non-metal elements from the fifth main row. In particular, the high oxygen affinity of phosphorous compared to the inertness of nitrogen describes these differences perfectly. While more than 99% of the worldwide nitrogen abundance stems from inert N₂ (oxidation state ±0), phosphorous occurs natively in derivatives of H₃PO₄, in which P is commonly in the oxidation state +V (Figure 2), illustrating the different reactivity. However, phosphorus in its oxidation state +III can replace nitrogen in organic compounds, showing their similar structural features in this oxidation state as well (

Table 1. Overview of some organo phosphorous (P^+III) compounds and structural related amine compounds.

	P-Type	P—Compounds	N-Type	N—Compounds
Aliphatics	Phosphines		Amines
	Phosphaalkenes		Imines
	Phosphaalkynes		Nitriles
	Diphosphenes		Azo compounds
Aromatics/ Heterocycles	Phosphinines		Pyridines
	Phospholes		Pyrroles

). However, although they are structurally equivalent to their lighter homologues, organo phosphorous compounds (P^III) show a different reactivity. These phosphines tend to readily oxidize to form the more stable phosphine oxides, [10] whereas amines are hard to oxidize and such an oxidation requires the use of strong oxidants, e.g., H₂O₂ [11]. Furthermore, phosphorous compounds with double- and triple bonds between P and C show a reactivity more comparable to alkenes and alkynes than to imines and nitriles due to the smaller difference in electronegativity (ΔE_N) of phosphorous (2.19) and carbon (2.55) compared to nitrogen (3.04) and carbon [12]. The small ΔE_N^C-P is also responsible for the aromatic character of phosphinines, although they are air- and moisture sensitive, which is not the case for pyridine. Opposite to this, phospholes do not show any aromaticity, since the lone pair of phosphorous is not delocalized, which is not the case for pyrroles that show aromatic behavior. These examples show how element homologue replacement (EHR), i.e., N to P, can affect the reactivity, oxygen sensitivity and electronic structure of a molecular system.

However, there is only one model known that comprises a phosphorous bridge. In 2010, Song and coworkers attempted to synthesize [Fe₂(CO)₆][(μ-SCH₂)₂PR] (R = Ph, η⁵-C₅H₄CH₂FeCp) using the tertiary phosphine RP(CH₂OH)₂ and (µ-HS)₂Fe₂(CO)₆ but found that the phosphine replaces one CO ligand to afford A (Fe₂S₂(CO)₅PR₃, Figure 3) due to the high nucleophilicity of the phosphine ligand [13].

Subsequently, Weigand and coworkers took up this result and synthesized PhP(O)(CH₂Cl)₂ as the possible bridge and received B ([Fe₂(CO)₆][(μ-SCH₂)₂P(O)Ph], Figure 3) after reacting it with (µ-LiS)₂Fe₂(CO)₆ [14]. This complex is the hitherto only described [FeFe]-hydrogenase model comprising any phosphorous atom as the bridgehead atom. However, due to the bulky P(O)Ph moiety and the hydrophobic phenyl substituent, B unlikely fits into the binding pocket of the [FeFe]-hydrogenase, which renders this model compound unsuitable for functional hybrid enzyme modifications. Furthermore, the basicity of the bridgehead, which is a prerequisite for an efficient proton transfer from the proton transfer path inside the protein to the iron center, where the proton is reduced to a hydride, was found to be a key aspect for a fully active enzyme.

The native cofactor ADT exhibits a pK_a of approximately eight [15,16], which is the pK_a one should aim for in the model compounds. The mimic presented by Weigand and coworkers, however, must be protonated by HBF₄·OEt₂, a very strong acid, indicating a low pK_a value of the P(O)Ph moiety [14]. We herein present a set of novel [FeFe]-hydrogenase model compounds that comprise phosphine oxide, phosphinic acid and phosphinates within the bridge. In addition, we herein describe our attempts to form the hitherto unprecedented complexes D–F (Figure 4). Notably, following literature reported basicities for various phosphine derivatives, these complexes possess a broad range of pK_a values (Figure 4). Thus, we anticipate that the suggested model compounds are interesting replacements for the native cofactor regarding stability and reactivity.

2. Results

2.1. Synthesis of Phosphinates and Phosphine Oxides as Well as the Iron Carbonyl Complexes Thereof

Scheme 1 presents the various reaction steps for the phosphine oxides 1–10. Step A is a condensation reaction between phosphinic acid and paraformaldehyde. Concerning the following chlorination of the hydroxy groups, yielding 2, Maier and coworkers already noted that the anhydride is formed upon reacting 1 at room temperature. Upon heating the reaction mixture to reflux, the anhydride can be converted to the chloride 2 [18]. Therefore, 1 was directly added very slowly to refluxing thionyl chloride. Step C is a hydrolyzation reaction of an acid chloride that was performed by a slow addition of 2 to distilled water. The formation of compounds 3-R (step E) is an esterification that was catalyzed by NEt₃. After NEt₃·HCl was filtered off and the solvent removed, compounds 3-R were purified to afford colorless oils in good yield and high purity. Besides the phosphinic acid derivative 5 and its esters 3-R, methyl phosphine 4 was prepared via Grignard reaction with MeMgBr (step H).

We found that the phosphinate 3-Et is especially easily accessible and reacted it with Li₂[Fe₂S₂(CO)₆] to afford [Fe₂(CO)₆][(μ-SCH₂)₂P(O)OEt] (12-OEt) (Scheme 2, Method A). The purification of the [FeFe]-hydrogenase model compound was performed by column chromatography with SiO₂ as stationary phase and a 1:1 mixture of PE/EtOAc as the mobile phase to get a red crystalline solid in 5% yield after removing the solvent.

The IR spectrum of 12-OEt shows four sharp CO bands at 2079, 2044, 2004 and 1971 cm⁻¹, which are within ±10 cm⁻¹ discrepancy in the same range as those observed for [Fe₂(CO)₆][(μ-SCH₂)₂NH], the precursor of the native cofactor, showing a comparable electron density at the iron centers.

Even though we could effectively obtain a [FeFe]-hydrogenase model 12-OEt with a phosphinate bridgehead using method A, it is not the most efficient way due to low yields and the required laborious synthesis and workup of Fe₂S₂(CO)₆ (11) [19]. We therefore aimed for the synthesis of bis(mercaptomethyl)phosphine oxides and -phosphinates instead of their bis(chloromethyl) derivatives, which would produce the same models using Fe₃(CO)₁₂, a commercially available iron carbonyl, as the starting material (Scheme 2, Method B).

In the case of 3-Et and 3-Ph, nucleophilic substitution with potassium thioacetate was found to be the suitable method to introduce a protected thiol. Thus, the chlorides were dissolved in DMF and 2.5 equivalents of KSAc were added as solid. The mixture was stirred for 16 h at room temperature to afford a dark red solution that was condensed to a greasy solid. Solvation in water and extraction with DCM gave the respective thioacetates 7-Et and 7-Ph with 78% and 86% yield, respectively. For the phosphinic acid 6, the addition of thiourea was used instead. Here, chloride 5 was dissolved in n-butanol and two equivalents of thiourea were added as solid. While stirring the mixture at 110 °C, the product forms as a white solid and was filtered off and washed with fresh n-butanol after 6 h. In all cases, deprotection with sodium hydroxide and following acidification of the reaction mixture gave the free thiols 6, 8 and 10, which were then purified via bulb-to-bulb distillation and used to synthesize the [FeFe]-hydrogenase models according to method B (Scheme 2). The two additional steps to synthesize dithiols 6, 8 and 10 starting from chlorides 3–5 are far more convenient regarding the reaction protocol than the synthesis of 11 from Fe(CO)₅ [19]. Furthermore, method B poses an easy to handle synthesis for [FeFe]-hydrogenase models without the need of the additional reaction step to generate Li₂[Fe₂S₂(CO)₆] from 11 and LiBHEt₃, which is very air sensitive and can undergo unwanted side reactions. For example, we observed the formation of [Fe₄(SEt)₂(S₂)(CO)₁₂] upon reacting 11 with LiBHEt₃ and a nitro-derivative of 3-Ph. Therein, the ethyl groups of the reductant are the only ones present in this reaction mixture and thus have to be the source of the SEt moieties. The formation of this dimer was previously observed by Cheng and coworkers, who aimed to obtain the synthesis of an ADT derivative starting from 11 and LiBHEt₃ as well [20].

In the case of methyl phosphinate, we found that both the addition of potassium thioacetate or thiourea leads to an unidentified, water-soluble product after workup. Whether this compound is the phosphinic acid derivative 6, which is water soluble, or a species resulting from decomposition could not be concluded.

Following method B, compounds 12-R were synthesized by reacting dithiol 6, 8 and 10 with Fe₃(CO)₁₂ in THF at room temperature for 2 h. The purification of 12-OEt was performed as described for method A, affording the desired compound with 46% yield. However, although the same method was used to synthesize model compound 12-OPh, it readily undergoes transesterifications with stronger nucleophiles. Thus, eluting with PE/EtOAc produces 12-OEt, as indicated by NMR, while elution with DCM/MeOH produces 12-OMe in very low yields (Figure 5b). Interestingly, the lost phenyl group co-crystallizes with the methyl phosphinate-bridged compound in close proximity (2.7 Å). When changing the mobile phase to DCM/acetone (1:1) to avoid the use of methanol, the same result as for DCM/MeOH was obtained as verified by single crystal XRD. Since Fe₃(CO)₁₂ is stabilized with 5%–10% methanol, it is a reasonable assumption that, besides the solvent, the stabilizer can undergo the transesterification as well, which explains the overall low yields for this reaction. The model compound 12-Me, bearing a phosphine oxide instead of a phosphinate bridgehead, was purified by column chromatography as well. Elution with DCM/MeOH (50:1) gave a red solid with 46% yield after complete removal of the solvent. Unfortunately, mimic 12-OH from the reaction of 8 with Fe₃(CO)₁₂ decomposes readily in solution after column chromatography (MeCN/water 4:1), observed by significant amounts of precipitate. The supernatant solution was investigated by IR spectroscopy, showing the typical CO patterns of [FeFe]-hydrogenase mimics (Figure 6). However, due to the rapid decomposition of 12-OH, this model compound was not further investigated.

The molecular structures of 12-OEt and 12-OMe are shown in Figure 5 and the structure of phosphine oxide 12-Me in Figure S2. The Fe₂S₂ core of all complexes show the typical butterfly-like structure of [FeFe]-hydrogenase model compounds. The phosphorous atoms of both phosphinates 12-OEt and 12-OMe and the phosphine oxide 12-Me have a slightly distorted tetrahedral geometry and causes average P-C-S bond angles of 117.02° and 115.6° for 12-OEt and 12-OMe respectively. The phosphine oxide 12-Me exhibits a respective angle of 115.9°. The P=O distance of all compounds is 1.45 ± 0.03 Å. Taking these values into account, no structural difference between model compounds 12-OEt and 12-OMe can be observed, both comprising a phosphinate bridge and 12-Me, bearing a methylphosphine oxide. Additionally, the shown angle and P=O distance are in line with the previously reported mimic B, which bears a phosphine oxide bridge as well [14].

2.2. Spectroscopic Properties of Compounds 12-R

2.2.1. Infrared Spectroscopy

The CO stretching bands of [FeFe]-hydrogenase mimics are of utmost interest, showing differences in electron density and ligand orientation. Figure 6 presents the respective spectra of the oxidized compounds 12-R. All four compounds show two sharp main bands at 2077 and 2039 cm⁻¹, as well as a broader contribution around 1990 cm⁻¹ that is more or less dissolved. The overall band positions are in line with the native cofactor mimic C and the phenylphosphine oxide model B. These models show IR patterns at 2076, 2036, 2008 and 1989 and 1979 cm⁻¹ and at 2079, 2041, 2003 and 1990 cm⁻¹, respectively (Figure 6) [14,21]. Due to the similar CO band positions, it can be thought that all P-containing models and ADT exhibit a comparable electron density at the iron atoms. Therefore, the exchange of nitrogen to its heavier homologue, phosphorous, seems to not affect the electronics of the iron atoms at a first glance.

2.2.2. Mössbauer Spectroscopy

A second tool to investigate the electronics of [FeFe]-hydrogenase models is Mössbauer spectroscopy. While the isomer shift δ is proportional to the electron density at the respective nucleus, iron in this case, the asymmetric shape of the electronic charge around the nucleus and the nucleus itself determines the quadrupole splitting ΔE_Q. The former is expressed by the electronic field gradient (EFG). For [Fe₂(CO)₆][(μ-SCH₂)₂NH], these parameters are δ = 0.05 mm/s and ΔE_Q = 0.64 mm/s, respectively, indicating a low-spin Fe(I) system (

Table 2. Mössbauer parameter of compounds 12-R compared to different known model complexes.

Entry	Compound	δ [mm/s]	ΔEQ [mm/s]	Ref.
1	[Fe2(CO)6][(μ-SCH2)2P(O)OEt] (12-OEt)	0.07	0.91	a)
2	[Fe2(CO)6][(μ-SCH2)2P(O)OMe] (12-OMe)	0.05	0.87	a)
3	[Fe2(CO)6][(μ-SCH2)2P(O)Me] (12-Me)	0.05	0.87	a)
4	[Fe2(CO)6][(μ-SCH2)2CH2] (PDT)	0.05	0.63	a)
5	[Fe2(CO)6][(μ-SeCH2)2CH2] (PDSe)	0.06	0.65	a)
6	[Fe2(CO)6][(μ-SeCH2)2NH] (ADSe)	0.07	0.65	a)
7	[Fe2(CO)6][(μ-SCH2)2NH] (ADT)	0.05	0.64	a)
8	[Fe2(CO)6][(μ-SCH2)2Ph] (BDT)	0.08	0.46	a)
9	[Fe2(CO)4(CN)2][(μ-SCH2)2CH2]2−	0.06	0.92	a)
10	[Fe2(CO)4(CN)2][(μ-SCH2)2NH]2−	0.09	0.43	a)
11	[Fe2(CO)4(CN)2][(μ-SeCH2)2CH2]2−	0.07	0.83	a)
12	HydA1adt	0.16, 0.08	0.89, 0.55	[23]
13	[Fe2(CO)6][(μ-SCH2)2SiMe2]	−0.04	0.78	[24]
14	[Fe2(CO)6][(μ-SCH2)2SiMe(CH2SH)]	−0.04	0.82	[24]
15	[Fe2(CO)6][(μ-SCH2)2O)]	−0.02	0.81	[25]
16	[Fe2(µCO)(CO)2(PMe)(dppe)][(μ-SCH2)2)]	−0.58 (PMe), 0.70 (dppe)	0.2 (PMe), 0.04 (dppe)	[22]

a) this work.

, entry 7). All [FeFe]-hydrogenase models with altered dithiol bridges do resemble these parameters relatively well, particularly in the case of all-CO mimics, indicating unchanged electronic properties of the iron atoms. (Table 2, entries 4–8). Upon changing the carbonyl ligands to phosphines, however, δ and ΔE_Q substantially change compared to all-CO and dicyanide models, as can be observed for [Fe₂₍µCO)(CO)₂(PMe)(dppe)][(μ-SCH₂)₂)]. In this case, two different isomer shifts indicate a different electron density at each of the two iron atoms, which is a result of unsymmetrical substitution of the metals (Entry 16) [22].

The Mössbauer spectra of oxidized compounds 12-R are presented in Figure 7. All models show isomer shifts of δ = 0.05 ± 0.02 mm/s and quadrupole splittings of ΔE_Q = 0.89 ± 0.02 mm/s and therefore resemble the literature known all-CO mimics closely. In the spectrum of 12-Me, a second species can be observed, which is a direct product of the degradation of 12-Me. As concluded from IR experiments as well, Mössbauer spectroscopy substantiates the finding that an exchange of the nitrogen atom to different phosphorous species within the bridging dithiol does not affect the electron density at the iron atoms.

2.3. Reactivity towards Acids

In acetonitrile, the CO bands of compounds 12-OEt and 12-Me slightly upshift to 2081, 2045 and 2006 and 2080, 2044 and 2004 cm⁻¹, respectively, compared to the ATR-IR measurements of dried films. These shifts indicate an interaction between the carbonyl ligands and the solvent (Figure 8, blue traces). Since the CO stretching frequency is very sensitive to changes of the electron density of the central atom to which the carbonyls are bound, slight shifts of the resulting bands in the IR spectra indicate, e.g., protonation of the investigated molecules. For [FeFe]-hydrogenase model compounds, the results of such experiments are used to determine potential protonation sites and pK_a values of the compounds.

Therefore, acetonitrile solutions of phosphinate 12-OEt and phosphine oxide 12-Me were acidified with organic acids of various strength and their resulting IR spectra monitored regarding small shifts. In the case of acetic acid (HOAc), which has a pKa of 23.5 in acetonitrile, [26] no shift of the CO bands can be observed indicating no protonation of the cluster. The use of 10 equivalents of triflic acid (HOTf, pKa = 2.6 [27]) or aqueous HBF₄ (= H₃O⁺BF₄⁻, pK_a in MeCN = 2.3 ± 0.1 [28]) however, causes a slight up-shift of all CO bands (~4 cm⁻¹), resulting in a new CO pattern of 2084, 2048, 2010 cm⁻¹ in the case of 12-OEt, that does not further change upon adding more acid to the solution (Figure S3). This shift is even more pronounced in the case of 12-Me (~7 cm⁻¹), showing a new pattern of the CO bands at 2087, 2040 and 2014 cm⁻¹ after addition of HOTf. Interestingly, the use of aqueous HBF₄, although being the slightly stronger acid, results in a less pronounced shift of the CO bands.

Different protonation scenarios were already discussed in literature: While direct metal protonation, yielding a µH binding mode, results in an CO band up shift of up to 100 cm⁻¹, [29,30] a smaller shift of approx. +15 cm⁻¹ is observed in the case of N-protonation [30,31]. Ligand based protonation, however, e.g., in the case of a phosphaadamantane ligand results in shifts of +18–35 cm⁻¹ [32]. While CO protonation was not observed before, P=O protonation was previously observed and calculated by Weigand and coworkers and resulted in shifts of ~10 cm⁻¹ [14]. On the basis of the strong structural relation to the model of Weigand, a P=O protonation is suggested for our models as well.

2.4. DFT Calculations on the Protonation of [FeFe]-hydrogenase Model Compounds

DFT calculations on 12-OEt (Figure 9) and 12-Me (Figure S4) were conducted to probe a possible protonation at the bridging sulfide, the iron center in form of bridging or terminal protons/hydrides and the P=O moiety of the bridgehead. While the former led to more pronounced shifts of the calculated CO frequencies, protonation at the oxide resembles the experimental spectra very well (see SI). Notably, a slight blue shift of all CO bands was observed for the [FeFe]-hydrogenase HydA1 upon acidification under reducing conditions [33,34]. In this case, DFT calculations suggested a protonation at an H-cluster binding cysteine, which was later supported by synchrotron-based techniques [35].

2.5. Redox Properties of Model Compounds 12-R in Absence and Presence of Acids of Various Strength

In addition to the IR experiments in the presence of different acids, we also performed cyclic voltammetry before and after the addition of acid in which the phosphinate 12-OEt (Figure 10a), and the phosphine oxide 12-Me (Figure 10b) behave very similar. The redox behavior of [FeFe]-hydrogenase models is dictated by their first ligand sphere. In the case of the native cofactor’s precursor [Fe₂(CO)₆][(μ-SCH₂)₂NH] the potential for the reduction from a homovalent Fe^IFe^I to a bivalent Fe⁰Fe^I complex is at −1.59 V versus the ferrocene/ferrocenium couple. Although bridged by two selenides and down shifted CO bands within the IR spectrum, which suggests higher electron density at the iron centers, ADSe (Figure 1) shows the same reduction potential.

The half peak potential of the [FeFe]-hydrogenase model compounds 12-OEt and 12-Me is slightly shifted to more anodic potential, namely −1.42 V and −1.41 V vs. Fc/Fc⁺, respectively. Independent on the nature of the bridgehead, phosphinate 12-OEt und phosphine oxide 12-Me, therefore, exhibit the same reduction potential. Figure 10 presents the CVs of 12-OEt and 12-Me in presence and absence of acetic acid. As already indicated by IR, both model compounds behave inertly against the weak acid at open circuit potential. Therefore, both complexes, 12-OEt and 12-Me have to follow an EC mechanism. After the addition of 10 equivalents of acetic acid, a catalytic reduction wave can be observed at −2.2 V vs. Fc/Fc⁺ after reducing the H₂ase mimic to its Fe^IFe⁰ state that increases from 10 to 20 equivalents and is at its maximum at 30 equivalents. Notably, 12-OEt shows a slightly higher current at this potential (−190 µA) than phosphine oxide 12-Me (−150 µA). The reduction wave does not further increase upon increasing the amount of acid present in solution. This might indicate a maximum turn-over frequency of both catalysts and “free” protons get reduced at the surface of the glassy carbon electrode. Thus, the observed currents are a result of an overlay of current from a homogeneous and heterogeneous proton reduction and might explain the CV shape between -2.4 V and −2.6 V for the runs with 30+ equivalents of HOAc.

At approx. −2.6 V vs. Fc/Fc⁺, the second reduction wave occurs from the reduction to Fe⁰Fe⁰ that is not present in acid free medium. However, the protonation of the Fe^IFe⁰ complex decreases the electron density of the iron and shifts the potential of the second reduction to more positive values.

In the presence of a stronger acid such as H₃O⁺ (see above) however, both compounds change their reactivity, undergoing a CE mechanism (Figure 11 and Figure S5). The potential of the Fe^IFe^I/Fe^IFe⁰ redox couple of both compounds shifts approximately 0.26 V towards more positive potentials upon acidification of the solution, indicating a lower electron density at the iron center evoked by a protonation at the oxide. This result goes in hand with the observed blue shift of the CO bands within the IR spectra upon acidification of 12-OEt and 12-Me with H₃O⁺ in acetonitrile. A similar strong shift of +0.2 V was observed for the oxidation of model compound B upon addition of HBF₄·OEt₂ [14].

Compared with the cyclic voltammograms of 12-OEt and 12-Me in presence of HOAc, the catalytic wave is shifted to more positive potentials (−1.5 V vs. Fc/Fc⁺ for 10 equivalents of acid) after the addition of HBF₄ and starts immediately after reduction to Fe^IFe⁰ (see inlets of Figure 10). Furthermore, the catalytic current of the proton reduction outperforms the respective currents of HOAc experiments by a factor of two, resulting in a current of ~−400 µA for 20 equivalents of acid. The most striking result of the HBF₄ experiments is the further increasing catalytic current with increasing amounts of acid. This is not the case for the EC mechanism with the weaker acid HOAc. Thus, the protonation of the P=O moiety does not only shift the reduction potential of the Fe^IFe^I/Fe^IFe⁰ redox couple to more positive values, but seemingly alters the mechanism of catalytic proton reduction in a way that the compounds 12-OEt and 12-Me are not turn-over-rate-limited under the tested conditions, and solely diffusion limitations occur in all experiments. To account for the possible two different mechanisms that result from changing the proton source from HOAc to HBF₄, we performed spectroelectrochemical measurements on 12-OEt and combined this experiment with titration experiments of HOAc. The reduction of 12-OEt can be followed by tracing the CO-band signature shifts within the IR spectrum. The CO-bands shift from 2081, 2046 and 2006 cm⁻¹ at open circuit potential to 1975, 1938, 1902 and 1730 cm⁻¹ upon applying −1.1 V vs. Ag/Ag⁺, which is slightly more negative than the reductive peak potential (Figure S6). The CO-band shift of 100 cm⁻¹ is in agreement with the reduction to a Fe^IFe⁰ complex [12-OEt]⁻ [36]. Interestingly, an IR band at 1730 cm⁻¹ is observable upon reducing 12-OEt. Pickett and coworkers observed a similar behavior of [Fe₂(CO)₆][(μ-SCH₂)₂CH₂] (PDT) and accounted this signal to a bridging CO [36]. In this case, external CO binds upon the loss of one bridging thiolate, induced by the reduction of PDT and resulting in a dangling HS(CH₂)₃ chain. In our case, no external CO was added to the analyzed solution of 12-OEt. Additionally, a substantial reorganization of the CO ligands, which might occur in order to adopt a structure with µCO, seems not to be present according to the identical CO-band structure of the oxidized and reduced 12-OEt. However, the additional µCO ligand might occur from the degradation of a small amount of substance, resulting in free CO that binds to the intact mimic in a µCO fashion. This “self-cannibalization” process can be regularly found for [FeFe]-hydrogenase enzymes as well [37].

Addition of 0.1-2 equivalents of HOAc to [12-OEt]⁻ did not lead to changes within the IR-spectrum (final spectrum see SI Figure S7). Therefore no conclusions regarding the protonation of the reduced phosphinate species [12-OEt]⁻ and the HER mechanism can be drawn from these experiments.

3. Conclusions

Three new [FeFe]-hydrogenase model compounds comprising both phosphinate and phosphine oxide bridgeheads were synthesized. Here, bis(mercaptomethyl)derivates in combination with Fe₃(CO)₁₂ led to substantially better yields than the respective chloride compounds and avoided the synthesis of Fe₂S₂(CO)₆. All three model compounds show identical behavior towards HOAc, undergoing an EC mechanism for catalytic proton reduction. The use of a stronger acid, such es HBF₄ in aqueous media or HOTf, led to a change in reactivity. All compounds undergo a CE mechanism, starting with protonation followed by reduction to Fe^IFe⁰ and the catalytic proton reduction. Due to the preceding protonation, the reduction potential of the [FeFe]-hydrogenase model compounds shift approx. 200 mV towards a more positive potential, showing the decreased electron density at the iron centers. The lowering of the electron density was further approved by IR spectroscopy, which shows a shift of the CO band maxima of 7 cm⁻¹ towards higher wavenumbers. Our DFT analysis revealed best agreement between simulated and obtained IR spectra in the case of P=O protonation, while µS-H⁺, Fe-H⁺ and Fe-µH⁺-Fe protonation modes led to stronger CO band shifts or a different band pattern compared to our results. All herein synthesized compounds provide the opportunity to be reduced to phosphines, which should afford [FeFe]-hydrogenase mimics with pK_a values closer to the native azadithiolate bridge and therefore models with improved catalytic activity that may also fit into the binding pocket of apo-[FeFe]-hydrogenases.

4. Materials and Methods

IR spectroscopy. Standard IR-measurements were performed with a Bruker Tensor 27 equipped with a Miracle™ single reflection ATR unit (PIKE Technologies) at room temperature. Complexes 12-R were dissolved and dropped on the Ge crystal. The measurement was started after complete evaporation of the solvent to yield a thin complex film on the ATR crystal.

Electrochemistry. Electrochemical testing of the complexes was conducted in acetonitrile using a standard three-electrode setup using a PalmSens3 potentiostat. A glassy carbon (GC) electrode was used as working electrode. An Ag- and a Pt wire were used as quasi-reference and counter electrode, respectively. If not otherwise stated, 0.1 M TBAPF₆ was used as the electrolyte and the measured potential was referenced to the ferrocene/ferrocenium couple (Fc/Fc⁺). Cyclic voltammetry measurements were then performed at a scan rate of 100 mV/s.

Spectroelectrochemistry. Spectroelectrochemical measurements were performed with a Bruker Tensor 27, equipped with a Miracle™ single reflection ATR unit (PIKE Technologies) to which a SP-02 cell (Spectroelectrochemistry Partners) was mounted. Al compounds were measured in a concentration of 20 mM. 1 M TBAPF₆ was used as the electrolyte. A glassy carbon (GC) electrode was used as working electrode. An Ag- and a Pt wire were used as quasi-reference and counter electrode, respectively. For IR spectra, 128 scans were accumulated to one spectrum at room temperature, while the potential was adjusted using a PalmSens3 potentiostat.

Mössbauer Spectroscopy. Zero-field Mössbauer spectra were recorded at 80 K by using a constant acceleration spectrometer equipped with a temperature controller maintaining temperatures within ±0.1 K and a 57Co radiation source in a Rh matrix. Isomer shifts are referred to α-Fe metal at room temperature. Data were fit with a sum of Lorentzian quadrupole doublets by using a least- square routine with WMOSS program.

X-ray Data Collection and Structure Solution Refinement. Single crystals suitable for X-ray structure analysis were coated with Paratone N oil, mounted on a fiber loop, and placed in a cold stream in the diffractometer. For 12-OEt Oxford XCalibur diffractometer performing φ and ω scans at 170(2) K was used. Diffraction intensities were measured using graphite-monochromatic Mo Kα radiation (λ = 0.71073 Å). Data collection, indexing, initial cell refinements, frame integration, final cell refinements, and absorption corrections were accomplished with the program CrysAlisPro.⁵ Space groups were assigned by analysis of the metric symmetry and systematic absence (determined by XPREP of WinGX⁶) and were further checked by PLATON^{7, 8} for additional symmetry. Structures were solved by direct methods and refined against all data in the reported 2θ ranges by full-matrix least square on F² with the SHELXL⁹ program suite using the shelXle interface.^{10, 11} Crystallographic data as well as refinement parameters are presented in Table S1.

Syntheses

(HOCH₂)₂P(O)OH (1). Phosphinic acid (33 g, 0.5 mol), paraformaldehyde (30 g, 1 mol) and 38 mL HCl (37%) were combined in a 250 mL single necked flask and stirred at 40–45 °C for approx. 6 h, until the mixture turns into a colorless solution, which was stirred at reflux conditions for additional 30 h. Excess water was removed using a rotary evaporator and additional azeotropic distillation with toluene, yielding 120 g (0.5 mol, 100%) a slightly yellow viscous oil. ¹H NMR (250 MHz, D₂O, ppm): 3.85 (d, ²J_PH = 5.2 Hz, 4H, -CH₂OH). ³¹P NMR (101 MHz, D₂O, ppm): 45.4 (s). ESI-MS (QP, positive): 126.9 [M+H]⁺. Refractive index: $n_D^{21}$ = 1.502.

(ClCH₂)₂P(O)Cl (2). 1 (95 g, 0.75 mol) was added very slowly (15 g/h) to 100 mL refluxing SOCl₂, resulting in extensive formation of SO₂ and HCl. After gas evolution ceased, the reaction mixture was distilled to yield excess SOCl₂ at 90 °C, 1013 mbar and 2 at 150 °C, 0.2 mbar with 88% yield and 91% purity according to ³¹P NMR. ¹H NMR (250 MHz, CDCl₃, ppm): 4.06 (m, 4H, -CH₂Cl). ¹³C NMR (63 MHz, CDCl₃, ppm): 37.14 (d, ¹J_PC = 90 Hz). ³¹P NMR (101 MHz, CDCl₃, ppm): 51.0 (s). ESI-MS (QP, positive): 163.0 [M+H₂O −Cl⁻]⁺ (from hydrolyzation). Refractive index: $n_D^{21}$ = 1.519

(ClCH₂)₂P(O)OEt (3-Et). 2 (5g, 27.8 mmol) was dissolved in 65 mL dry THF and cooled to 0 °C. To that solution 4.62 mL (33.3 mmol, 1.2 equivalents) N(Et)₃ was added. To this now orange solution, 1.94 mL (33.3 mmol, 1.2 equivalents) ethanol was added drop wise and stirred for 4 h at room temperature. A white precipitate was filtered off and the filtrate concentrated to a red oil that was purified via distillation (head temperature 100-105 °C, 0.86 mbar) and follow up bulb-to-bulb distillation (160 °C, 1.4 mbar) to yield 3.2 g (60%) of a colorless oil. ¹H NMR (250 MHz, CDCl₃, ppm): 1.26 (td, ³J_HH = 7.1 Hz, ⁴J_PH = 2.8 Hz, 3H, OCH₂CH₃), 3.62 (dd, ²J_HH = 2.2 Hz, ²J_PH = 8.5 Hz, 4H, -CH₂P), 4.11 (m, 2H, OCH₂CH₃) ¹³C NMR (63 MHz, CDCl₃, ppm): 16.3 (d, ³J_PC = 5,7 Hz, OCH₂CH₃), 32.6 (d, ¹J_PC = 106 Hz, -CH₂-P), 62.8 (d, ²J_PC = 7 Hz, OCH₂CH₃). ³¹P NMR (101 MHz, CDCl₃, ppm): 39.4 (s). Refractive index: $n_D^{21}$ = 1.502

(ClCH₂)₂P(O)OMe (3-Me). 2 (10 g, 56 mmol) was dissolved in 100 mL dry THF and cooled to 0 °C. To that solution, 9.25 mL (66.7 mmol, 1.2 equivalents) N(Et)₃ was added. To this now orange solution, 2.14 g (66.7 mmol, 1.2 equivalents) methanol was added and stirred for 4 h at room temperature. A white precipitate was filtered off and the filtrate concentrated to an orange to red oil that was purified via bulb-to-bulb distillation (130 °C, 0.9 mbar) to yield 8.7 g (88%) of a colorless oil. ¹H NMR (400 MHz, CDCl₃, ppm): 3.69 (d, ²J_PH = 8.0 Hz, 4H, -CH₂Cl), 3.82 (d, ³J_PH = 12 Hz, 3H, -OCH₃). ¹³C NMR (100 MHz, CDCl₃, ppm): 32.3 (d, ²J_PC = 107 Hz, -OCH₃), 53.0 (d, ¹J_PC = 7.2 Hz, -CH₂Cl). ³¹P NMR (162 MHz, CDCl₃, ppm): 41 (s).

(ClCH₂)₂P(O)OPh (3-Ph). 2 (10 g, 56 mmol) was dissolved in 100 mL dry THF and cooled to 0 °C. To that solution 9.25 mL (66.7 mmol, 1.2 equivalents) N(Et)₃ was added. To this now orange solution, 6.26 g (66.7 mmol, 1.2 equivalents) phenol was added and stirred for 4 h at room temperature. A white precipitate was filtered off and the filtrate concentrated to a red oil that was purified via bulb-to-bulb distillation (130 °C, 0.9 mbar) to yield 10.8 g (80%) of a slightly yellowish oil. ¹H NMR (400 MHz, CDCl₃, ppm): 3.78 (d, ²J_PH = 8.5 Hz, 4H, -CH₂Cl), 6.83 (m, 1H, arom.), 7.1–7.5 (m, 4H, arom.). ¹³C NMR (100 MHz, CDCl₃, ppm): 32.6 (d, ¹J_PC = 106 Hz, -CH₂Cl), 115.5 (s, arom.), 120.8 (d, J_PC = 4.3 Hz, arom.), 126.1 (d, J_PC = 1.6 Hz, arom.), 130.3 ppm (d, J_PC = 1.4 Hz, arom.) ³¹P NMR (162 MHz, CDCl₃, ppm): 39.4 (s). ESI-MS (QP, positive): 238.8 [M+H]⁺.

(ClCH₂)₂P(O)CH₃ (4). Methylmagnesium bromide (27 mL, 83 mmol) was dissolved in 200 mL dry diethyl ether and cooled to 0 °C. The solution was mechanically stirred while 2 (10 g, 66 mmol), dissolved in 50 mL dry ethyl ether, was slowly added via dropping funnel. After addition, the mixture was stirred at 50 °C for 4 h and quenched with a saturated solution of sodium carbonate, resulting in strong heat development. The obtained solid was filtered off and washed with chloroform. The filtrate was extracted with a total amount of 400 mL chloroform. The organic phases were combined and all solvents removed by a rotavapor and a high vacuum oil pump, resulting in 4 g (25 mmol, 38%) of a colorless oil that crystallizes after complete loss of solvent. ¹H NMR (250 MHz, CDCl₃, ppm): 1.76 (d, ²J_PH = 13.2 Hz, 3H, -CH₃), 3.76 (m, 4H, -CH₂Cl). ¹³C NMR (100 MHz, CDCl₃, ppm): 10.0 (d, ¹J_PC = 73 Hz, -CH₃), 34.7 (d, ¹J_PC = 71 Hz, -CH₂Cl). ³¹P NMR (102 MHz, CDCl₃, ppm): 42.6 (s). ESI-MS (QP, positive): 161.0 [M]⁺. XRD: Figure S3

(ClCH₂)₂P(O)OH (5). 2 (5 g, 27.8 mmol) was slowly added to 5 mL water. After addition, the solution was filtered and excess water removed using a rotary evaporator, yielding in 4.1 g (91%, 98% purity according to ³¹P NMR) of 5 as white solid. ¹H NMR (400 MHz, D₂O, ppm): 3.18 (d, ²J_PH = 8 Hz, 4H, -CH₂Cl). ¹³C NMR (100 MHz, D₂O, ppm): 34.2 (d, ¹J_PC = 103 Hz). ³¹P NMR (162 MHz, D₂O, ppm): 33.0 (s).

(HSCH₂)₂P(O)OH (6). 5 (4 g, 24.7 mmol) and thiourea (3.76 g, 50 mmol) were mixed in 15 mL n-Butanol and stirred at 115 °C for 6 h and at room temperature overnight. A white precipitate was filtered off and washed with EtOH and dried under reduced pressure. An amount of 20 mL 6 M KOH was added, and the mixture stirred for 1 h at 100 °C. The solution was then acidified with 10 mL 12 M HCl and washed with DCM. The aqueous phase was narrowed to yield a white sticky solid that was dispersed in EtOH and filtered. The solvent was the ethanolic solution was removed to yield a slightly milky oil. ¹H NMR (400 MHz, D₂O, ppm):1.90 (2H, -SH), 2.90 (4H, -CH₂SH), 4.51 (1H, POH). ³¹P NMR (162 MHz, D₂O, ppm): 22.9(s).

(AcSCH₂)₂P(O)OEt (7-Et). 3-Et (5.7 g, 30 mmol) was dissolved in 60 mL dry DMF and potassium thioacetate (8.6 g, 75 mmol) was added. The red turning mixture was stirred at ambient temperature overnight. The solvent was removed, the residue dissolved in DCM and washed with water. Finally, the organic phase was narrowed to yield 6.4 g (23.3 mmol, 78%) of a dark red oil. ¹H NMR (400 MHz, CDCl₃, ppm): 4.12 (dq, 2H, -OCH₂-), 3.26 (dd. 4H, -PCH₂-) 2.39 (s, 6H, -SC(O)CH₃), 1.30 (t, 3H, -OCH₂CH₃). ¹³C NMR (100 MHz, CDCl₃, ppm): 193 (s, SC(O)-), 62.4 (d, ²J_PC = 7 Hz, -OCH₃-), 30.2 (s, -SC(O)CH₃), 24.1 (d, ¹J_PC = 99 Hz, -PCH₂-), 16.6 (d, ³J_PC = 5.8 Hz, -OCH₂CH₃). ³¹P NMR (162 MHz, CDCl₃, ppm): 43.5 (s).

(AcSCH₂)₂P(O)OPh (7-Ph). 3-Ph (7.2 g, 30 mmol) was dissolved in 60 mL dry DMF and potassium thioacetate (8.6 g, 75 mmol) was added. The red turning mixture was stirred at ambient temperature overnight. The solvent was removed, the residue dissolved in DCM and washed with water. Finally, the organic phase was narrowed to yield 8.2 g (25.8 mmol, 86%) of a dark red oil. ¹H NMR (400 MHz, CDCl₃, ppm): 7.39-7.20 (m, 5H, arom.), 3.4 (dd. 4H, -PCH₂-) 2.41 (s, 6H, -SC(O)CH₃). ¹³C NMR (100 MHz, CDCl₃, ppm): 192.7 (s, SC(O)-), 130.1 (d, ²J_PC = 1.2 Hz, arom.), 125.6 (d, ³J_PC = 1.6 Hz, arom.), 120.7 (d, ⁴J_PC = 4.5 Hz, arom.), 115.4 (d, ⁵J_PC = 1.4 Hz, arom.), 30.2 (s, -SC(O)CH₃), 24.2 (d, ¹J_PC = 99 Hz, -PCH₂-) ³¹P NMR (162 MHz, CDCl₃, ppm): 43.3 (s).

(AcSCH₂)₂P(O)Me (9). 4 (5 g, 31 mmol) was dissolved in 60 mL dry DMF and potassium thioacetate (9 g, 78 mmol) was added. The red turning mixture was stirred at ambient temperature overnight. The solvent was removed, the residue dissolved in DCM and washed with water. Finally, the organic phase was evaporated to yield 5.4 g (22.4 mmol, 72%) of a red solid. ¹H NMR (400 MHz, CDCl₃, ppm): 1.52 (d, 3H, ²J_PH = 12.9 Hz, -PCH₃), 2.41 (s, 6H, SC(O)CH₃), 3.28 (d, 4H ²J_PH = 9 Hz, -PCH₂-). ¹³C NMR (100 MHz, DMSO, ppm): 13.2 (d, ¹J_PC = 71.6 Hz, -PCH₃), 25.6 (d, ¹J_PC = 66.1 Hz, -PCH₂-), 30.0 (s, - SC(O)CH₃), 193.4 (d, ³J_PC = 2.8 Hz, SC(O)-). ³¹P NMR (162 MHz, CDCl₃, ppm): 43.4 (s). ESI-MS (QP, positive): 262.7 [M+Na]⁺, 240.8 [M+H]⁺, 198.9 [M –(CH₃CO) +2H]⁺, 157.3 [M -2(CH₃CO) +3H]⁺.

(HSCH₂)₂P(O)OEt (8-Et). 7-Et (2.7 g, 10 mmol) was dissolved in 50 mL MeOH and NaOH (3 M in water, 20 mL) was added. The solution was kept stirring at room temperature for 3 h, neutralized with 10 mL 6 M HCl and finally extracted with DCM. The organic phase was evaporated to yield a dark red oil that was purified by bulb-to-bulb distillation to obtain a yellow oil.

Note: Besides ethyl bis(mercaptomethyl) phosphinate, this procedure gives 4-ethoxy-1,2,4-dithiaphospholane 4-oxide, or a polymer by forming inter- instead of intramolecular disulfide bridges, as well. Those oxidized products can be specifically synthesized by the following procedure:

Ethyl bis((acetyl)methyl) phosphinate (1.5 g, 5.6 mmol) was dissolved in 10 mL degassed, dry THF. Furthermore, sodium (255 mg, 11.2 mmol) was dissolved in 5 mL MeOH, forming NaOMe, and added dropwise to the THF solution. After the addition is complete, 100 mL THF was added for dilution before 1.41 g (6 mmol) I₂, dissolved in 50 mL THF was added very slowly. This mixture was stirred for 3 d at RT before the solvent was removed yielding a greasy residue that was extracted with DCM. The solvent of the organic phase was removed to afford a dark red highly viscous oil that did not require further purification. ¹H NMR (400 MHz, CDCl₃, ppm): Dithiol: 1.17 (t, ³J_HH = 7 Hz, 3H, -OCH₂CH₃), 1.77 (s, 2H. –SH), 2.68 (dd, ²J_HH = 5.3 Hz, ²J_PH = 10.4 Hz, 4H, -CH₂SH), 3.99 (dq, ³J_HH = 7 Hz, ³J_PH = xx Hz, 2H, -OCH₂CH₃). Dithiaphospholan: 1.41 17 (t, ³J_HH = 7 Hz, 3H, -OCH₂CH₃), 3.06 (dd, ²J_HH = 6.6 Hz, ²J_PH = 8.5 Hz, 4H, -CH₂S), 4.22 (dq, ³J_HH = 7 Hz, ³J_PH = 7.9 Hz, 2H, -OCH₂CH₃). ¹³C NMR (100 MHz, CDCl₃, ppm): Dithiol: 17.5 (d, ³J_PC = 98 Hz, -OCH₂CH₃), 29.8 (d, ¹J_PC = 93 Hz, -CH₂SH), 62.0 (d, ²J_PC = 7 Hz, -OCH₂CH₃). Dithiaphospholan: 18.9 (d, ³J_PC = 97 Hz, -OCH₂CH₃), 33.7 (d, ¹J_PC = 93 Hz, -CH₂S), 62.4 (dq, ²J_PC = 7 Hz, -OCH₂CH₃). ³¹P NMR (162 MHz, CDCl₃, ppm): Dithiol: 48 (s); Dithiaphospholane: 80 (s). ESI-MS (QP, positive): 186.6 [M+H]⁺, 159.0 [M+H₂O −OEt⁻]⁺ (from hydrolyzation).

(HSCH₂)₂P(O)OPh (8-Ph). 7-Ph (3.5 g, 11 mmol) was dissolved in 50 mL MeOH and NaOH (3 M in water, 22 mL) was added. The solution was kept stirring at room temperature for 3 h, neutralized with 11 mL 6 M HCl and finally extracted with DCM. The organic phase was evaporated to yield a dark red oil that was purified by bulb-to-bulb distillation to obtain a yellow oil.

Note: Besides phenyl bis(mercaptomethyl) phosphinate, this procedure gives 4-phenoxy-1,2,4-dithiaphospholane 4-oxide, or a polymer by forming inter- instead of intramolecular disulfide bridges, as well. Here, only ³¹P-NMR spectroscopy was able to distinguish between both products

¹H NMR (400 MHz, CDCl₃, ppm): 1.76 (dt, 2H, –SH), 2.78 (m, 8H, -CH₂SH and -CH₂S-), 6.78-7.12 (m, 5H, arom., overlapped with toluene), ¹³C NMR (100 MHz, CDCl₃, ppm): 17.0 (d, ³J_PC = 100 Hz, -CH₂SH), 115.5 (s, arom.), 120.6 (s, arom.), 129.8 (s, arom.), 155.8 (s, arom.). ³¹P NMR (162 MHz, CDCl₃, ppm): Dithiol: 52 (s); Dithiaphospholane: 84 (s). ESI-MS (QP, positive): 238.8 [M+H]⁺.

(HSCH₂)₂P(O)CH₃ (10). 9 (1 g, 4.17 mmol) was dissolved in 30 mL MeOH and NaOH (3 M in water, 8.34 mL) was added. The solution was kept stirring at room temperature for 0.5 h, neutralized with 40 mL 1 M HCl and finally extracted with chloroform. The organic phase was evaporated to yield a dark red oil that was purified by bulb-to-bulb distillation to obtain a smelly yellow oil.

Note: Besides Bis(mercatptomethyl)methyl phosphine oxide, this procedure gives 4-methyl-1,2,4-dithiaphospholane 4-oxide, or a polymer by forming inter- instead of intramolecular disulfide bridges, as well.

¹H NMR (400 MHz, CDCl₃, ppm): Dithiol: 1.68 (d, ²J_PH = 12.6 Hz, 3H, -CH₃), 1.94 (td, ³J_HH = 6.6 Hz, ³J_PH = 8.5 Hz, 2H. –SH), 2.68 (dd, ²J_HH = 5.3 Hz, ²J_PH = 10.4 Hz, 4H, -CH₂SH). ¹³C NMR (100 MHz, CDCl₃, ppm): Dithiol: 17.5 (d, ³J_PC = 98 Hz, -OCH₂CH₃), 29.8 (d, ¹J_PC = 93 Hz, -CH₂SH), 62.0 (d, ²J_PC = 7 Hz, -OCH₂CH₃). ³¹P NMR (162 MHz, CDCl₃, ppm): Dithiol: 47 (s). Dithiaphospholane: 79 (s). ESI-MS (QP, positive): 156.9 [M]⁺.

Syntheses of model compounds 12-R.Method A. Fe₂S₂(CO)₆ (100 mg, 0.29 mmol) was dissolved in 5 mL dry THF in a 50 mL single necked Schlenk flask, which was sealed with a rubber septum and equipped with a bubbler. The reaction vessel was cooled to −78 °C under a constant slight Ar flow to avoid pressure differences, before 2.2 equivalents of LiBHEt₃ (1 M in THF) was added via syringe drop wise. This addition resulted in a color change from orange over purple to dark green, indicating the formation of Li₂[Fe₂S₂(CO)₆]. After 15 min, 3-Et (75 mg, 0.4 mmol) and NEt₃ (83 µL, 0.6 mmol) were added consecutively via syringe and the reaction mixture allowed to warm to room temperature overnight. The solvent was completely removed with a liquid N₂ trap, before the residual red solid was dissolved in a mixture from PE and EtOAc (1:1), filtered and finally purified via column chromatography with PE/EtOAc (1:1) as mobile phase and SiO₂ as stationary phase.

Method B. Fe₃(CO)₁₂ (500 mg, 1 mmol) was dissolved in 20 mL dry THF in a 50 mL single necked Schlenk flask, which was sealed with a rubber septum and equipped with a bubbler. To that green solution, 1.3 equivalents of dithiol were added via syringe. The reaction mixture was then kept stirring at room- or slightly elevated temperature (<40 °C) until it became intense red and no starting material was left according to TLC (elution with PE, approx. 2–3 h). The solvent was removed by a cold trap and the residual solid dissolved in the eluent for following chromatographic purification.

12-OEt. Elution with PE/EtOAc (1:1). IR (υCO, cm⁻¹): 2077, 2038, 1999, 1975. EA: calcd. for C₁₀H₉Fe₂O₈PS₂: C, 25.89; H, 1.96; Fe, 24.07; O, 27.59; P, 6.68; S, 13.82. found: C, 25.48; H, 2.22; S, 13.65

12-OMe. Elution with DCM/MeOH (50:1). IR: (υCO, cm⁻¹): 2077, 2034, 1985.

12-Me. Elution with DCM/MeOH (25:1). IR: (υCO, cm⁻¹): 2077, 2041, 1996. EA: calcd. for C₉H₇Fe₂O₇PS₂: C, 24.91; H, 1.63; Fe, 25.74; O, 25.81; P, 7.14; S, 14.78. found: C, 24.83; H, 1.75; S, 14.65