Understanding Factors that Control the Structural (Dis)Assembly of Sulphur-Bridged Bimetallic Sites

Abstract

Bimetallic structures of the general type [M₂(µ-S)₂] are omnipresent in nature, for biological function [M₂(µ-S)₂] sites interconvert between electronically distinct, but isostructural, forms. Different from structure-function relationships, the current understanding of the mechanism of formation and persistence of [M₂(µ-S)₂] sites is poorly developed. This work reports on bimetallic model compounds of nickel that interconvert between functional structures [Ni₂(µ-S)₂]^+/2+ and isomeric congeners [2{κ-S–Ni}]^2+/+, S = Aryl-S⁻, in which the nickel ions are geometrically independent. Interconversion of the two sets of structures was studied quantitatively by UV–VIS absorption spectroscopy and cyclic voltammetry. Assembly of the [Ni₂(µ-S)₂]⁺ core from [2{κ-S–Ni}]⁺ is thermodynamically and kinetically highly preferred over the disassembly of [Ni₂(µ-S)₂]²⁺ into [2{κ-S–Ni}]²⁺. Labile Ni-η^2/3-bonding to aromatic π-systems of the primary thiophenol ligand is critical for modeling (dis)assembly processes. A phosphine coligand mimics the role of anionic donors present in natural sites that saturate metal coordination. Three parameters have been identified as critical for structure formation and persistence. These are, first, the stereoelectronic properties of the metals ions, second, the steric demand of the coligand, and, third, the properties of the dative bond between nickel and coligand. The energies of transition states connecting functional and precursor forms have been found to depend on these parameters.

1. Introduction

Sulphur-bridged bimetallic structures of the general formula [M₂(µ-S)₂] present a recurring motif in natural metalloproteins that is associated in a variety of biological processes, including reversible bond making and breaking, electron transfer, and cellular regulation. The prominence of [M₂(µ-S)₂] sites in nature likely derives from the intriguing properties of the bond between metal and sulphur, be it inorganic sulfide or organic thiolate. The covalent character of an M–S bond varies sensitively with effective core charge and d-electron count of the metal. In-depth studies on Cu and Fe containing bimetallic Cu_A [1,2] and 2Fe-ferredoxins [3] disclosed that it is the degree of covalent bonding that determines the mechanism and strength of electronic coupling of the metals in these archetypal [M₂(µ-S)₂] structures [2]. As drawn in Figure 1, Cu_A and 2Fe-ferredoxin sites interconvert between S = 0 and open-shell S = 1/2 states. The dominance of antiferromagnetic over ferromagnetic coupling of the high-spin iron ions [4] has been attributed to result from the covalent bonding along the Fe–sulfido–Fe linkages, rendering the mixed-valent S = 1/2 form [Fe(III)(µ-S)₂Fe(II)]⁺ energetically favorable [2,5].

Whereas the understanding of the relationship between biological function and electronic properties of closed- and open-shell forms of Cu_A and 2Fe-ferredoxins has been widely developed, details on the mechanism of formation and of the parameters that control structural persistence of [M₂(µ-S)₂] sites have remained largely unknown. In vitro studies starting from an apoprotein concluded that Cu_A assembles rapidly in form of its open-shell state after stepwise introduction of Cu(I) and Cu(II) sites [6,7,8,9]. Different from natural Cu_A, preparative assembly of the bimetallic core [Cu₂(µ-S)₂] commonly involves dimerization of homovalent precurors [10,11,12,13]. Analogous studies on binuclear cores of iron are scare [14,15], whereas preparation of molecular model structures involves ligand exchange at a preformed [Fe₂(µ-S)₂] core [4,16,17,18,19,20,21,22]. According to recent mechanistic work, the ligand environment and the overall oxidation state determine the structural dynamics of the [Fe₂(µ-S)₂] core, for which S = S²⁻ [23].

Gaining a quantitative understanding of the (dis)assembly of [M₂(µ-S)₂] sites is of fundamental interest. Mechanistic work in this context requires a ligand platform that suits to stabilize precursor structures in which the metal sites are independent and can exist in different oxidation states. As depicted in Figure 1, alignment of two mesityl substituted thiophenol donors at a 1,4-disubstituted arene provides a ligand framework for which both types of structures have been accessed. The set of four bimetallic complexes of nickel comprises S = 0 and mixed-valent S = 1/2 forms [Ni₂(µ-S)₂]^+/2+, S = Aryl-S⁻, in addition to [2{κ-S–Ni}]^2+/+ in which the nickel fragments are structurally independent. The capability of nickel to adopt tetrahedral and T-shaped ligand fields in addition to η^2/3-binding of an aromatic π-system is critical. A phosphine coligand L completes the coordination sphere at nickel. A previous study revealed that all redox state changes are electrochemically reversible at potential sweep rates ν ≤ 2000 mV·s⁻¹ in 0.1 M nBu₄NPF₆/1,2-C₆H₄F₂ solution at 24 °C [24]. The homovalent, S = 0, form of [Ni₂(µ-S)₂]²⁺, Ni = Ni(II), is metastable and rearranges into [2{κ-S–Ni}]²⁺. The barrier for this process varies sensitively with the steric demand of the coligand L as half-lives decrease in the order t_1/2,d = hours (L = P(C₆H₁₁)₃) >> 730 s (PPh₃) > 5 s (PMe₃). In agreement with studies on the formation of Cu_A, structural assembly of the open-shell form [Ni₂(µ-S)₂]⁺ is kinetically substantially favored over dissociation of the homovalent congener. Digital simulation of cyclic voltammetry (CV) current-potential (i-E) curves established that [2{κ-S–Ni}]⁺, which consists of Ni(II) and Ni(I) fragments, converts into [Ni₂(µ-S)₂]⁺ on a time scale t_1/2,a < 0.001 s at 24 °C. The metastable character of homovalent [Ni₂(µ-S)₂]²⁺, Ni = Ni(II), is a critical property, considering that structural persistence of the [M₂(µ-S)₂] core in both S = 0 and 1/2 states is essential to biological function.

The formation of [Ni₂(µ-S)₂]⁺ from [2{κ-S–Ni}]⁺ is thermodynamically highly favored on account of a difference of ~1 V between respective formal redox potentials [24]. However, the markedly higher rate of formation of [Ni₂(µ-S)₂]⁺ appears to result from a kinetic effect. The proposal of a kinetic contribution considers that the spatial alignment of the two metal fragments in the transition state should benefit from a higher degree of freedom that arises from divergent stereoelectronic preferences of the metal ions. The lower effective core charge of Ni(I) renders changes of its coordination geometry and number energetically less demanding than for Ni(II), which should manifest in respective transition state energies. We herein report circumstantial evidence in support of such a kinetic contribution. Substitution of previously employed coligands L = PMe₃ and PPh₃ for PMe₂Ph is reported to result in an increase of the rate of disassembly of [Ni₂(µ-S)₂]²⁺, Ni = Ni(II), at low temperatures as well as of the rate of formation of [Ni₂(µ-S)₂]⁺, Ni = Ni(I)/Ni(II), from [2{κ-S–Ni}]⁺. Quantitative analysis of kinetic and electrochemical data indicate that the conversion of metastable [Ni₂(µ-S)₂]²⁺ into [2{κ-S–Ni}]²⁺ involves the weakening of the dative bond between Ni and coligand L, and that the steric properties of L affect the barrier to formation of the mixed-valent sulphur-bridged structure.

2. Results

2.1. Preparation and Characterization of Bimetallic Compounds

As shown in Figure 2, compound [Ni₂(µ-S)₂]⁺ has been prepared in form of its N(SO₂CF₃)₂, NTf₂, salt following a two-step procedure that starts from mononuclear complex Ni–2S. Complex Ni–2S is isostructural to the PMe₃ ligated congener [24] in the solid state and in solution. ¹H and ¹³C{¹H} NMR spectra in CDCl₃ solution display N-line splitting patterns for the carbon nuclei at methyl and the quaternary position of the phenyl group of the coligand, which corroborates trans-disposition of PMe₂Ph at nickel. The reaction of Ni–2S and Ni(cod)₂, cod = 1,5-cyclooctadiene, in toluene solution at ambient temperature affords the bimetallic complex 2Ni–2S almost quantitatively. Product 2Ni–2S can be isolated from saturated pentane solutions as a dark yellow microcrystalline solid in moderate yield of about 50%. To avoid the loss of product owing to high solubility, 2Ni–2S is oxidized in situ using [Fc]⁺, Fc = ferrocene, affording [Ni₂(µ-S)₂]⁺ in 80% crystalline yield.

Single crystal X-ray diffraction (XRD) analysis confirmed the skeletal connectivity of [Ni₂(µ-S)₂]⁺ as drawn in Figure 2. Despite moderate resolution of diffraction data impaired structure refinement to a high level of accuracy, PMe₂Ph ligated [Ni₂(µ-S)₂]⁺ is isostructural to its PMe₃ congener [24]; pertinent data are provided in the Supplementary Material. Short Ni–C distances of ~2.01 Å point to a η²:η²-coordination of the 1,4-disubstituted π-system to the nickel atoms. Coordination of two thiophenolates in µ-fashion in addition to the PMe₂Ph coligand completes a distorted tetrahedral ligand environment at each nickel atom. Acute Ni–(µ-S)–Ni bond angles of ~63° render the Ni–Ni distance short at 2.366(3) Å. Tetrahedral metal coordination implies the presence of an S = 1 Ni(II) and an S = 1/2 Ni(I) site. X-Band cw-EPR spectra recorded from a 2-methyltetrahydrofuran solution at 293 and 77 K corroborate a total S = 1/2 system that results from formal antiferromagnetic coupling of S = 1 and S = 1/2 configured nickel ions. As depicted in Figure 2, the spectrum of a frozen sample is of rhombic appearance. The g-factor at highest field, g₃ = 2.08, deviates substantially from the electron free-spin g-factor, g_e ~ 2.00, but g-anisotropy, Δg = 0.06, is small. Considering that hyperfine coupling to any of the two ³¹P nuclei is not resolved, EPR data point to a predominant localization of the unpaired electron at nickel, sulphur, and likely the 1,4-disubstituted π-system. A mixed-valent description [Ni(I)(µ-S)₂Ni(II)]⁺ is preferred but the homovalent form [Ni(1.5)₂(µ-S)₂]⁺ cannot be excluded. The orange appearance of diluted solutions of [Ni₂(µ-S)₂]⁺ in CH₂Cl₂ results from moderately intense transitions at λ(ε/M⁻¹·cm⁻¹) = 567 (4200) and 773 nm (2500).

Bulk oxidation of either 2Ni–2S or [Ni₂(µ-S)₂]⁺ by a 1:2 equivalents mixture of benzoquinone and HN(SO₂CF₃)₂ in 1,2-C₆H₄F₂ solution affords the NTf₂ salt of [2{κ-S–Ni}]²⁺ that was isolated as an orange powder in 60–70% yield. Monitoring the reaction of [Ni₂(µ-S)₂]⁺ and the 1,1′-diacetylferrocene cation, [Fc’’]⁺, by NMR spectroscopy in CD₂Cl₂ solution at T ≤ −60 °C did not allow detecting [Ni₂(µ-S)₂]²⁺ as the initial product of oxidation but [2{κ-S–Ni}]²⁺ was observed exclusively in a 1:1 molar ratio to Fc’’ coproduct. The diamagnetism of [2{κ-S–Ni}]²⁺ points to a T-shaped ligand field at Ni(II), comprising κ-thiophenolate, cis-PMe₂Ph and the π-system of the flanking mesityl group as donors. ¹³C{¹H} NMR solution data are in agreement with η³-coordination of Ni(II) to the mesityl group in trans-position to PMe₂Ph by the presence of doublet resonances for mesityl ipso-¹³C (J_C,P = 8 Hz; δ(C_ipso) = 121 ppm) and equivalent ortho-¹³C nuclei (J_C,P = 2 Hz). The presence of structurally independent Ni(II) sites is augmented by a low intensity ligand field transition at λ(ε/M⁻¹·cm⁻¹) = 872 (200) nm in UV–VIS spectra, which replaces the transition of starting material [Ni₂(µ-S)₂]⁺ at 773 nm (2500 M⁻¹·cm⁻¹). NMR spectroscopic and UV–VIS absorption data of [2{κ-S–Ni}]²⁺ match those of the mononuclear congener [κ-S–Ni]⁺, for which λ(ε/M⁻¹·cm⁻¹) = 915 (100) nm and δ(C_ipso) = 123 ppm (doublet with J_C,P = 8 Hz). The molecular structure of a single-crystalline sample of [κ-S–Ni]⁺ confirmed pertinent structural properties as depicted in Figure 3. A property of [κ-S–Ni]⁺ that is notable in regard to the rearrangement of [Ni₂(µ-S)₂]²⁺ into [2{κ-S–Ni}]²⁺ is the migration of the Ni(II)-L fragment between the two mesityl π-systems that occurs slowly on the NMR time scale at T = 26 °C according to ¹H-¹H-ROESY (Rotating frame Overhauser Enhancement SpectroscopY) data.

2.2. Quantitative UV–VIS Spectroscopic Study of Oxidation-Induced Disassembly of [Ni₂(µ-S)₂]⁺

Probing the reaction of [Ni₂(µ-S)₂]⁺ and [Fc’’]⁺ by NMR spectroscopy revealed that [Ni₂(µ-S)₂]²⁺ is short-lived even at T ≤ −60 °C. Rapid formation of [2{κ-S–Ni}]²⁺ under these conditions contrasts the reactivity of the PMe₃ ligated congener that reacts noticeably at T ≥ −40 °C [24]. For this reason, the oxidation reaction was monitored by UV–VIS spectroscopy at T = −90 °C in CH₂Cl₂ solution. The first absorption spectrum recorded after oxidant addition displayed an intense electronic transition centered at λ(ε/M⁻¹·cm⁻¹) = 715 nm (9000). The appearance of an intense low-energy band is a characteristic feature of the [Ni₂(µ-S)₂]²⁺ core, and the position of the PMe₂Ph ligand isomer fits the order of transition energies of PMe₃ and PPh₃ ligated derivatives at 695 (7000) and 765 nm (6000), respectively [24]. Even at T = −90 °C, the spectral profile of PMe₂Ph ligated [Ni₂(µ-S)₂]²⁺ converts into that of [2{κ-S–Ni}]²⁺. A semi-logarithmic plot of normalized concentrations of [Ni₂(µ-S)₂]²⁺ vs. time is linear, pointing to a 1st order dependence of the reaction rate on the concentration of [Ni₂(µ-S)₂]²⁺. Rate constants k_d(T) were determined at −90 ≤ T ≤ −50 °C, and half-live times decreased correspondingly in the range 944 ≤ t_1/2,d ≤ 19 s. Linear regression of a plot of ln(k_d(T)/T) vs. 1/T yielded ΔH^‡_d = 34 ± 2 kJ·mol⁻¹ and ΔS^‡_d = −(120 ± 25) J·mol⁻¹·K⁻¹. Pertinent kinetic data are provided in the Supplementary Material. Based on the activation parameters obtained, extrapolation provides t_1/2,d(25 °C) ≈ 0.14 s or k_d(25 °C) ≈ 5 s⁻¹, which equals an approximately three times higher reaction rate than for the PMe₃ ligated congener in CH₂Cl₂ solution; t_1/2,d(25 °C, L = PMe₃) ≈ 0.35 s or k_d(25 °C, L = PMe₃) ≈ 2 s⁻¹, cf. supplementary Table S3 of [24]. The decrease of ΔH^‡_d by ≈ 50 kJ·mol⁻¹ by substituting L = PMe₃ for PMe₂Ph is substantial but is compensated by the change of ΔS^‡_d ≈ −100 J·mol⁻¹·K⁻¹ that becomes dominant with increasing temperature. The lower contribution of ΔH^‡_d to the free energy of activation indicates that a change of the properties of the dative bond between coligand L and Ni(II) determines the energy of the transition state that connects structures [Ni₂(µ-S)₂]²⁺ and [2{κ-S–Ni}]²⁺. On the one hand, weakening of the L→Ni(II) bond in preparation of the transition state manifests in a higher ΔH^‡_d but, on the other hand, reduces steric interaction between the aryl-S–Ni–L fragments during structural rearrangement, rendering ΔS^‡_d more favorable (less negative) as observed.

Any attempt to detect [2{κ-S–Ni}]⁺ as the initial product from the reduction of [2{κ-S–Ni}]²⁺ by (η⁵-C₅H₅)₂Co, and monitor its subsequent conversion into [Ni₂(µ-S)₂]⁺, were unsuccessful. The skeletal rearrangement of [2{κ-S–Ni}]⁺ is subject to a small kinetic barrier, reflecting that it is facile even at T = −95 °C. For an independent estimation of k_a and k_d, (dis)assembly processes have been studied by cyclic voltammetry starting from [Ni₂(µ-S)₂]⁺.

2.3. Quantitative Electrochemical Study of Rate Constants k_a and k_d for (dis)Assembly of [Ni₂(µ-S)₂]⁺

Figure 4 shows a representative CV i-E curve recorded from solutions of [Ni₂(µ-S)₂]⁺ in 0.1 M nBu₄NPF₆/1,2-C₆H₄F₂ at a glassy carbon working electrode at 20 °C. The extended square scheme included in Figure 4 assigns the three redox events observed, which are the electrochemically reversible interconversions of [Ni₂(µ-S)₂]⁺/2Ni–2S at E(I) = −0.881 V and [Ni₂(µ-S)₂]⁺/[Ni₂(µ-S)₂]²⁺ at E(II) = 0.207 V, and the irreversible reduction of [2{κ-S–Ni}]²⁺ to [2{κ-S–Ni}]⁺ at E_p(III,ν = 50 mV·s⁻¹) = −0.76 V vs. Fc/[Fc]⁺. Contrasting coligands PMe₃ and PPh₃, substitution for PMe₂Ph limits the range of potential sweep rates at which electrochemically reversible behavior is observed to 50 ≤ ν ≤ 500 mV·s⁻¹. The peak current ratio i_p,red(II)/i_p,ox(II) deviates noticeably from unity at lower potential sweep rates because the formation of [2{κ-S–Ni}]²⁺ competes effectively with the reduction of electrochemically generated [Ni₂(µ-S)₂]²⁺ back to [Ni₂(µ-S)₂]⁺. Newly formed [2{κ-S–Ni}]²⁺ is reduced at E_p(III,ν) but the subsequent skeletal rearrangement of [2{κ-S–Ni}]⁺ into [Ni₂(µ-S)₂]⁺ is very rapid on the time scale of the potential sweep experiment so that no corresponding oxidation peak is observed. At ν ≥ 500 mV·s⁻¹, E_p(III,ν) merges with the reduction peak of E(I). The shift of E_p(III,ν) to cathodic (negative) potentials with increasing scan rate is characteristic of an EC type mechanism, in which a rapid and irreversible chemical process withdraws an electrochemically generated species from a Nernst equilibrium [25]. CV i-E curves recorded by scanning in cathodic potential direction are devoid of E_p(III,ν). The corresponding CV data of [2{κ-S–Ni}]⁺ do not display redox event E(II), except if the potential scan is started in cathodic direction, first traversing through E_p(III,ν) in accord with CV data for [Ni₂(µ-S)₂]⁺.

A prerequisite to estimating k_a by digital simulation of i-E curves according to the extended square scheme in Figure 4 is, however, that the formal potential E(III) is known. To eliminate the interdependence of E_p(III,ν) and k_a, reference compound [κ-S–Ni]⁺ has been analyzed that interconverts electrochemically reversible with κ-S–Ni at E(III*) = −0.814 V. Assuming that E(III) = E(III*), experimental and simulated i-E curves match reasonably well as judged by the eye, except for a deviation of i_p,ox(I) that is noticed at potential sweep rates 50 ≤ ν ≤ 1000 mV·s⁻¹. Values obtained for k_d(20 °C) varied in the range of 0.6–2.2 s⁻¹, which is in good accord with k_d(20 °C) = 2.9 s⁻¹ computed from activation parameters. Structural assembly of [Ni₂(µ-S)₂]⁺ occurs at a much higher rate than disassembly of [Ni₂(µ-S)₂]²⁺ under identical conditions and simulation provided k_a(20 °C) = (2800 ± 500) s⁻¹. The size of k_a determines the onset of E_p(III,ν) and i_p,red(I), both of which are in good to reasonable agreement with experimental CV data.

3. Discussion

Sulphur-bridged bimetallic structures of the general form [M₂(µ-S)₂] contribute to a variety of processes in Nature. Their biological function commonly involves the interconversion of states that differ in electronic but not geometrical structures, for which Cu_A and 2Fe-ferredoxins are prominent examples. The relationship between function and electronic structure seems well understood but the mechanism of assembly of [M₂(µ-S)₂] sites has remained largely elusive. In this context, the design of molecular compounds that model function and (dis)assembly from or into precursor structures is of fundamental interest. Combining π-systems as labile ligands with thiophenol(s) has enabled studying the redox dependent interconversion between sulphur-linked [Ni₂(µ-S)₂]^+/2+ and structural isomers [2{κ-S–Ni}]^2+/+ that feature geometrically independent nickel sites. The phosphine coligand at nickel not only serves to saturate metal coordination but also mimics the effect that additional donor groups have on the properties of [M₂(µ-S)₂] cores. Notably, the electronic structure of [Ni(I)(µ-S)₂(II)]⁺ is equivalent to the [Fe(II)(µ-S)₂Fe(III)]⁺ form of 2Fe-ferredoxins.

The thermodynamic driving force of the formation of [Ni₂(µ-S)₂]⁺ equals the difference of formal potentials of redox couples [Ni₂(µ-S)₂]²⁺/[Ni₂(µ-S)₂]⁺, E(II), and [2{κ-S–Ni}]²⁺/[2{κ-S–Ni}]⁺, E(III). Computing the average over ΔE = E(II) − E(III) = E(II) − E(III*) for all coligands L yet studied yields ΔE = 976 ± 36 mV, highlighting the distinct thermodynamically stability of the S = 1/2 state. The small variation of ΔE indicates that differences between activation parameters ΔH^‡_d and ΔS^‡_d as well as the ratio of rate constants k_a(T)/k_d(T) ≥ 10³ do not result from substantial differences between ground state energies but from differences between the energies of associated transition states. Studying the temperature dependence of k_d(T) has provided insight into the relation between the properties of coligand L and transition state energy, which comprises two factors. The first relates to the steric properties of L whereas the second considers the properties of the dative bond between metal and L. Increasing the steric demand of L results in a higher ΔH^‡_d in order to weaken the Ni←L bond in preparation of the transition state. A lower steric demand of L results in smaller ΔH^‡_d but at the cost of an unfavorable contribution of ΔS^‡_d that increases with increasing temperature. The second scenario seems to apply to the case of L = PMe₂Ph, for which the ratio k_a(T)/k_d(T) decreases with increasing temperature. Secondary interactions such as dispersion forces and π–π-interactions between co- and primary ligand may further contribute to stabilizing the transition states.

Different from the influence of coligand L, the oxidation state of the metal is a second parameter that controls the rates of (dis)assembly processes. More precisely, the energy that is required for two S–Ni–L fragments to adopt to the transition state geometry is expected to benefit from diverging stereoelectronic properties of the nickel ions. This proposal explains the marked kinetic preference of formation of sulphur-bridged [Ni₂(µ-S)₂]⁺, Ni = Ni(I)/Ni(II), from geometrically independent, heterovalent nickel sites, and augments mechanistic data reported for the skeletal assembly of Cu_A [1,6,7,8,9]. This mechanism may also apply to the self-assembly of the mixed-valent form of 2Fe-ferredoxins, for which, however, experimental data are not yet available, rendering the bimetallic model structures of nickel unique. The kinetic stability of the homovalent form [Ni₂(µ-S)₂]²⁺, Ni = Ni(II), varies sensitively with the steric properties of coligand L. Comparison to Cu_A and 2Fe-ferredoxin structures in Nature points to a stabilizing contribution of neutral or negatively charged coligands. Bonding of σ-donating groups levels the effective core charges of the metal ions and limits σ-donation from sulphur donors. The increase of covalent M–S bond character would otherwise favor κ-S over µ-S-bonding and trigger dissociation of the homovalent state. The kinetic stability of the homovalent forms of Cu and Fe containing [M₂(µ-S)₂] structures, however, is essential for their biological function.

4. Materials and Methods

All manipulations of air- and moisture-sensitive compounds were carried out under an atmosphere of dry argon using standard Schlenk or glove box techniques. Literature procedures were followed for the preparation of (Ph₃P)₂NiN(SiMe₃)₂, ferrocenium salts, and thiophenol ligands [24,26,27]. General information on the purification of solvents and other reagents is provided in Supplementary Material. ¹H, ¹³C, and ³¹P NMR chemical shifts δ are given in ppm and scalar coupling constants J are given in Hz. ¹H and ¹³C chemical shifts are referenced to residual ¹H and naturally abundant ¹³C resonances of solvents whereas ³¹P NMR chemical shifts are referenced to an external standard sample of 85% H₃PO₄ set to δ = 0. Variable temperature (VT) electronic spectra were recorded on PerkinElmer Lambda 35 and 1050 UV–VIS/NIR photospectrometers equipped with a Unisoku CoolSpeK UV USP-203-B cryostat (Nanoscore GmbH, Glashütten, Germany), using 1 cm quartz cells with septum screw caps. Electronic absorption data were processed by Spectragryph, Software for optical spectroscopy, Version 1.2.9 (Dr. Friedrich Menges, Oberstdorf, Germany). CV was carried out at 20 °C (water bath) under an argon atmosphere using an ECi 200 potentiostat (Nordic Electrochemistry ApS, Copenhagen, Denmark) in a gas-tight, full-glass, three-electrode cell setup. The potentiostat was controlled using the EC4 DAQ (version 4.1.90.1, Nordic Electrochemistry) software, and data were treated with EC4 VIEW (version 1.2.36.1, Nordic Electrochemistry). A glassy carbon disc (Metrohm Deutschland GmbH & Co. KG, Filderstadt, Germany; electro-active area = 0.088 ± 0.002 cm²) and a 1 mm coiled Pt-wire were employed as working and counter electrodes. A 0.5 mm Ag wire in a 0.01 M AgClO₄/0.1 M nBu₄NPF₆ MeCN solution was employed as reference in a Haber-Luggin capillary/dual reference-electrode system [28]. IR compensation was applied during data collection, and CV data were corrected for background currents of electrolyte solutions for different potential sweep rates. Potentials are reported relative to E(Fc/[Fc]⁺, 0.1 M nBu₄NPF₆/1,2-C₆H₄F₂, 24 °C) = 0.197 ± 0.003 V vs. Ag/Ag⁺. The electro-active area of the working electrode was calculated from Fc/[Fc]⁺ measurements in 0.1 M nBu₄NPF₆ solution in MeCN at various concentrations and potential sweep rates, using D(Fc/MeCN, 20 °C) = 2.40 × 10⁻⁵ cm²·s⁻¹ [29]. Working electrode was rinsed with acetone, polished very gently with a paste of 0.3 µm alumina (Deutsche Metrohm GmbH & Co. KG, Filderstadt, Germany) in deionized water, rinsed thoroughly with plenty of deionized water, and finally acetone after each use and stored in a desiccator over P₂O₅. Periodic Fc/[Fc]⁺ reference measurements verify the integrity of the working electrode as well as the stability of the Ag/Ag⁺ reference potential. Samples for CV analysis were prepared by weighing substrate and 0.1 M nBu₄NPF₆/1,2-C₆H₄F₂ into an 1 mL volumetric flask. Samples were kept under argon, and aliquots transferred to electrochemical cell using a gas-tight microliter syringe. DigiElch8 Professional (Build 8214, C3 Prozess-und Analysentechnik GmbH, Haar b. München, Germany) was used for digital simulation of CV data. Elemental analysis were done on a Vario Micro Cube system (Elementar Analysesysteme GmbH, Langenselbold, Germany). CCDC 1897416-1897417 contain the supplementary crystallographic data for this paper. These data can be obtained free of charge via http://www.ccdc.cam.ac.uk/conts/retrieving.html (or from the CCDC, 12 Union Road, Cambridge CB2 1EZ, UK; Fax: +44-1223-336033; E-mail: [email protected]). General numeration is as shown in Scheme 1.

Preparation of Ni–2S. 1,4-Terphenyldithiophenol ligand (600 mg, 0.933 mmol), benzyl potassium (258 mg, 1.866 mmol), and 26 g of toluene are weighed out into a 100 mL Schlenk flask, and the mixture left stirring at r.t. until a homogeneous yellow solution results. A red solution of (Me₂PhP)₂NiCl₂ (378 mg, 0.931 mmol) dissolved 10 g of toluene is added by pipette, and the reaction mixture stirred overnight. A dark purple mixture resulted that was dried under dynamic vacuum, product was extracted into hexane and separated from KCl co-product by cannula filtration. The volume of combined purple hexane solutions was reduced until formation of crystals was noticed, and the mixture was stored at −30 °C overnight. Mother liquor was removed by cannula, the product was washed with methanol to remove small quantities of OPMe₂Ph, and finally dried under dynamic vacuum to afford a dark purple crystalline solid (482 mg, 0.494 mmol, 53%). Slow diffusion of hexamethyldisiloxane layered on top of a solution of Ni–2S in toluene afforded prismatic purple single crystals that were used for X-ray diffraction and elemental analysis. ¹H NMR (500 MHz, CDCl₃, 26 °C): δ = 7.62 (s, 4H, H-8,9,11,12), 7.48 (d, 2H, ⁴J_H,H = 2.4, H-5,18), 7.12 (m, 2H, para-H, Me₂PPh), 7.10 (m, 4H, ortho-H, Me₂PPh), 6.92 (m, 4H, meta-H, Me₂PPh), 6.87 (4H, H-21,23,27,29), 6.87 (2H, H-3,16), 2.34 (s, 6H, H-34,37), 2.02 (s, 12H, H-33,35,36,38), 1.35 (s, 18H, H-39,40), 1.19 (br s, 12H, Me₂PPh); ¹³C{¹H} (150 MHz, CDCl₃, 26 °C): δ = 143.94 (C-4,17), 142.73 (C-2,15), 141.37 (C-1,14), 140.92 (C-19,25), 140.56 (C-7,10), 138.92 (C-6,13), 136.61 (N-line pattern, ^|1+3|J_C,P = 38, ipso-C, Me₂PPh), 136.11 (C-20,24,26,30), 135.3 (C-22,28), 129.96 (ortho-C, Me₂PPh), 127.78 (para-C, Me₂PPh), 127.53 (C-21,23,27,29), 127.36 (meta-C, Me₂PPh), 125.98 (C-3,16), 123.52 (C-8,9,11,12), 122.12 (C-5,18), 34.16 (C-31,32), 31.52 (C-39,40), 21.17 (C-34,37), 20.61 (C-33,35,36,38), 10.95 (N-line pattern, ^|1+3|J_C,P = 28, Me₂PPh); ³¹P{¹H} (202 MHz, CDCl₃, 26 °C): δ = −23.9; elemental analysis calcd (%) for C₆₀H₇₀NiP₂S₂: C, 73.84; H, 7.23; S, 6.57; found: C, 73.69; H, 6.76; S, 6.46; unit cell parameters: a = 10.7068(3), b = 16.6627(6), c = 17.9475(9) Å, α = 117.108(2), β = 92.836(2), γ = 106.327(2)°, V = 2675.49(19) ų, Z = 2, P−1, deposition number CCDC 1897416.

Preparation of 2Ni–2S. For details see preparation of [Ni₂(µ-S)₂]⁺. Compound 2Ni–2S crystallizes from saturated dark yellow solutions in pentane stored at −30 °C but yields are moderate (50–60%) owing to high solubility. ¹H NMR (500 MHz, C₆D₆, 26 °C): δ = 11.31 (s, 4H, H-8,9,11,12), 8.01 (d, 2H, ⁴J_H,H = 2.4, H-3,16), 7.94 (d, 2H, ⁴J_H,H = 2.3, H-5,18), 7.30 (m, 4H, ortho-H, Me₂PPh), 7.07 (m, 4H, meta-H, Me₂PPh), 6.98 (4H, H-21,23,27,29), 6.85 (tt, 2H, para-H, Me₂PPh), 2.35 (s, 6H, H-34,37), 2.21 (s, 12H, H-33,35,36,38), 1.99 (s, 12H, Me₂PPh), 1.27 (s, 18H, H-39,40); ¹³C{¹H} (150 MHz, C₆D₆, 26 °C): δ = 163.24 (br s, ν_1/2 ~24 Hz, C-7,10), 160.96 (C-2,15), 151.97 (C-4,17), 148.62 (C-6,13), 142.85 (C-19,25), 141.33 (ipso-C, Me₂PPh), 137.26 (C-20,24,26,30), 135.87 (C-22,28), 133.64 (ortho-C, Me₂PPh), 129.54 (para-C, Me₂PPh), 128.3 (meta-C, Me₂PPh, obscured by C₆D₆), 128.2 (C-21,23,27,29, obscured by C₆D₆), 126.76 (C-3,16), 125.15 (C-5,18), 125.07 (C-8,9,11,12), 121.22 (br s, C-1,14), 33.54 (C-31,32), 32.69 (C-39,40), 21.85 (C-33,35,36,38), 21.17 (C-34,37), 16.63 (Me₂PPh); ³¹P{¹H} (243 MHz, C₆D₆, 26 °C): δ = −24.4.

Preparation of [Ni₂(µ-S)₂]⁺. (A) Preparation of 2Ni–2S: Ni(cod)₂ (84 mg, 0.304 mmol) and Ni–2S (300 mg, 0.307 mmol) were weighted out into a 20 mL screw cap vial and 50 mL Schlenk flask, respectively, and dissolved in a total of 7 g of toluene at r.t. The solution of Ni(cod)₂ was transferred by pipette, and the mixture stirred for 3 h at r.t. to afford a dark yellow solution. All volatiles were removed under dynamic vacuum, the dark residue was redissolved in 7 g of toluene, and the solution stirred at r.t. overnight. All volatiles were removed again under dynamic vacuum, the residue dissolved in 7 g of THF, and this solution cooled to −30 °C in an dry ice/acetone bath. (B) Oxidation of 2Ni–2S: A dark blue solution of [Fc]NTf₂ (93 mg, 0.199 mmol) was prepared in 7 g of THF in a 25 mL Schlenk flask, and the solution cannula transferred to the cold solution of 2Ni–2S, which caused an immediate color change to a maroon appearance. After stirring for 1 h at r.t. and removal of volatiles under dynamic vacuum, the residue was washed with hexane (10 mL, 4×), dried under dynamic vacuum, and the residual maroon powder dissolved in 15 mL of 1,2-C₆H₄F₂, the solution filtered, volume reduced to about 5 mL, and layered with 20 mL of pentane. Pentane diffusion was complete within four days, affording a dull red crystalline solid that isolated from the mother liquor, washed with pentane, and finally dried under dynamic vacuum. Yield: 321 mg (0.244 mmol, 80%). Single crystals for X-ray diffraction analysis were obtained by layering a concentrated CH₂Cl₂ solution with pentane but moderate resolution of diffraction data did only establish skeletal connectivity; cf. supplementary Figure S3. X-Band cw-EPR (2-methyltetrahydrofuran, 3321/400 G = center/width): 293 K, 9.8359 GHz, 4 G modulation amplitude, 0.2 mW microwave power, g_iso = 2.116, full line width at half maximum = 0.86/3.51 mT (Gaussian + Lorentzian); 77 K, 9.4008 GHz, 1 G modulation amplitude, 0.5 mW microwave power, g₁ = 2.142, g₂ = 2.118, g₃ = 2.083; full line width at half maximum = 1.42/0.13 mT (Gaussian + Lorentzian), HStrain (EasySpin parameter to account for orientation dependent line broadening) = 11, 21, 1 MHz; UV–VIS (CH₂Cl₂, 20 °C): λ(ε/M⁻¹·cm⁻¹) = 401 (14500), 480 (5400), 567 (4200), 773 (2500) nm; elemental analysis calcd (%) for C₆₂H₇₀F₆NNi₂O₄P₂S₄: C, 56.64; H, 5.37; N 1.07; S, 9.75; found: C, 56.55; H, 5.2; N 1.64; S, 9.53; unit cell parameters: a = 9.3963(8), b = 23.8046(14), c = 28.0132(17) Å, α = 90°, V = 6265.9(8) ų, Z = 4, P 21 21 21.

Characterization of [Ni₂(µ-S)₂]²⁺. Solutions of [Ni₂(µ-S)₂]⁺ and [Fc’’]NTf₂ in CH₂Cl₂ were freshly prepared for each measurement from CH₂Cl₂ kept at 20 °C using a water bath and thermometer. An aliquot of [Ni₂(µ-S)₂]⁺ was transferred by microliter syringe to a cuvette charged with 2.6 mL of neat CH₂Cl₂, and a spectrum was recorded at 20 °C to determine sample concentration. After cooling to the desired temperature, an aliquot of oxidant solution was added under stirring, and the reaction monitored by recording the absorbance, A(λ,T), at λ = 714 nm at defined time intervals. The change of sample volume at T < 20 °C was corrected by once recording spectra at all temperatures of interest, and doing a polynomial regression of a plot of A(λ,T) vs. T. (UV–VIS (CH₂Cl₂, c₀([Ni₂(µ-S)₂]⁺) = 8 × 10⁻⁵ M, −90 °C): λ(ε/M⁻¹·cm⁻¹) = 416 (15000), 502 (11000), 615 (shoulder, 5800), 714 (9000) nm.

Preparation of [2{κ-S-Ni)]²⁺. Bimetallic 2Ni–2S was prepared from Ni–2S (200 mg, 0.205 mmol) and Ni(cod)₂ (56 mg, 0.203 mmol) as described above, and finally dissolved in 10 mL of 1,2-C₆H₄F₂. Alternatively, [Ni₂(µ-S)₂]⁺ can be used as the starting material. Benzoquinone (24 mg, 0.222 mmol) and HN(SO₂CF₃)₂ (112 mg, 0.398 mmol) were weighed out into a 50 mL Schlenk flask, immediately cooled to −30 °C in a dry ice/acetone bath, and taken up in 5–10 mL of 1,2-C₆H₄F₂, affording a clear yellow solution after slow warming to r.t. The oxidant solution was cannula transferred to the −30 °C cold solution of 2Ni–2S, and the mixture stirred at r.t. for 1.5 h, yielding a bright red solution. Volatiles were removed under dynamic vacuum, the residual solid washed with diethyl ether (15 mL, 3×), and finally dried to afford a dark orange powder (200 mg, 0.125 mmol, 61% with respect to Ni–2S). ¹H NMR (500 MHz, CD₂Cl₂, −40 °C, sample prepared in situ in an NMR tube from [Ni₂(µ-S)₂]⁺ and [Fc’’]⁺ at T ≤ −78 °C): δ = 7.67 (s, 4H, H-8,9,11,12), 7.61 (m, 4H, ortho-H, Me₂PPh), 7.56 (m, 2H, para-H, Me₂PPh), 7.54 (m, 4H, meta-H, Me₂PPh), 7.24 (d, 2H, ⁴J_H,H ≈ 2, H-5,18), 7.16 (d, 2H, ⁴J_H,H ≈ 2, H-3,16), 7.02 (4H, H-21,23,27,29), 2.27 (s, 12H, H-33,35,36,38), 1.61 (d, ²J_H,P = 12, 12H, Me₂PPh), 1.30 (s, 18H, H-39,40), 1.09 (s, 6H, H-34,37); ¹³C{¹H} (150 MHz, CD₂Cl₂, 26 °C): δ = 138.47 (C-7,10), 134.76 (C-2,15), 149.17 (C-4,17), 142.03 (C-6,13), 121.36 (d, ²J_C,P = 8, C-19,25), 129.31 (d, ¹J_C,P = 48, ipso-C, Me₂PPh), 127 (d, J_C,P = 2, C-20,24,26,30), 118.33 (C-22,28), 130.3 (d, ²J_C,P = 10, ortho-C, Me₂PPh), 132.09 (d, ⁴J_C,P = 3, para-C, Me₂PPh), 129.42 (d, ³J_C,P = 10, meta-C, Me₂PPh), 117.25 (C-21,23,27,29), 123.24 (C-3,16), 127.54 (C-5,18), 129.15 (C-8,9,11,12), 142.26 (d, ³J_C,P = 10, C-1,14), 34.31 (C-31,32), 30.93 (C-39,40), 19.26 (C-33,35,36,38), 18.57 (C-34,37), 13.07 (d, ¹J_C,P = 35, Me₂PPh); ³¹P{¹H} (243 MHz, CD₂Cl₂, 26 °C): δ = −0.16; UV–VIS (CH₂Cl₂, 20 °C): λ(ε/M⁻¹·cm⁻¹) = 320 (24000), 466 (6300), 872 (200) nm; elemental analysis calcd (%) for C₆₄H₇₀F₁₂N₂Ni₂O₈P₂S₆: C, 48.2; H, 4.42; N 1.76; S, 12.06; found: C, 47.42; H, 3.88; N, 1.86; S, 11.72, the noticeably lower than calculated carbon and hydrogen contents appear to scale with the high relative fluorine content F, 14.3.

Preparation of [κ-S–Ni]⁺. 4-^tBu-2,6-(Mes)₂C₆H₂SH ligand (150 mg, 0.373 mmol) and (Ph₃P)₂NiN(SiMe₃)₂ (277 mg, 0.373 mmol) were weighted out into a 20 mL screw cap vial and 50 mL Schlenk flask, respectively, and dissolved in a total of 8 g of toluene at r.t. The ligand solution was transferred by pipette, and the mixture stirred for 1.5 h at r.t. to afford a bright red solution. Volatiles were removed under dynamic vacuum, and a dark red residue was dissolved in ~15 mL of hexane. A bright red solid separated from the initially homogeneous solution at r.t., was separated from the yellow supernatant by cannula, washed with pentane (10 mL, 2×), dried under dynamic vacuum, and finally dissolved in 4 g of toluene. A purple solution of [Fc’’]NTf₂ (148 mg, 0.317 mmol) was prepared in 4 g of 1,2-C₆H₄F₂, and transferred by pipette at r.t., affording an orange solution that was stirred for another 0.5 h. Toluene was removed under dynamic vacuum, the residual orange solid washed with pentane (10 mL, 4×), separated from the supernatant by cannula filtration, dried under dynamic vacuum, and finally dissolved in 6 g of toluene. Neat PMe₂Ph (64 µL, 0.447 mmol) was added by syringe at r.t., and the resulting dark red solution stirred for 3.5 h. Removal of solvent left a mixture of the solid product and a dark red oil. Washing with toluene (5 mL, 3×), pentane (10 mL, 3×), and drying under dynamic vacuum afforded the product in form of an orange powder (125 mg, 0.142 mmol, 45% with respect to [Fc’’]NTf₂). Layering a CH₂Cl₂ solution of the product with toluene afforded dark red plate-shaped single crystals suitable for X-ray diffraction analysis. Diffusion of pentane afforded orange needles used for elemental analysis. ¹H NMR (500 MHz, CD₂Cl₂, 26 °C): δ = 7.60 and 7.55 (m each, 2H each, ortho/meta-H, Me₂PPh), 7.59 (m, 1H, para-H, Me₂PPh), 7.19 (d, 1H, ⁴J_H,H = 2, H-3), 7.05 (br s, 2H, H-9,11, η³-mesityl), 7.00 (d, 1H, ⁴J_H,H = 2, H-5), 6.92 (br s, 2H, H-15,17), 2.29 (s, 3H, H-24), 2.28 (s, 6H, H-19,22, η³-mesityl), 1.96 (s, 6H, H-23,25), 1.58 (d, ²J_H,P = 12, 6H, Me₂PPh), 1.33 (s, 9H, H-26), 1.25 (s, 3H, H-20, η³-mesityl); ¹³C{¹H} (150 MHz, CD₂Cl₂, 26 °C): δ = 150.29 (C-4), 144.53 (d, ³J_C,P = 9, C-1), 143.13 (C-6), 137.77 (C-16), 136.86 (C14,18), 135.71 (C-13), 134.7 (C-2), 132.71 (d, ⁴J_C,P = 3, para-C Me₂PPh), 130.9 and 130.15 (d each, ^{2, 3}J_C,P = 10, 11, ortho/meta-C Me₂PPh), 130.43 (d, ¹J_C,P = 49, ipso-C Me₂PPh), 128.68 (C-15,17), 127.91 (C-5), 127.66 (C8,12, η³-mesityl), 123.34 (C-3), 123.17 (d, ²J_C,P = 8, C-7, η³-mesityl), 120.27 (C-10, η³-mesityl), 117.92 (C-9,11, η³-mesityl), 35 (C-22), 31.71 (C-26), 21.42 (C-24), 20.5 (C-23,25), 19.62 (C-19,21, η³-mesityl), 19.53 (C-20, η³-mesityl), 14.03 (d, ¹J_C,P = 35, Me₂PPh); ³¹P{¹H} (203 MHz, CD₂Cl₂, 26 °C): δ = −1.39; UV–VIS (CH₂Cl₂, 25 °C): λ(ε/M⁻¹·cm⁻¹) = 332 (12000), 399 (shoulder, 3500), 464 (2700), 573 (shoulder 400), 915 (100) nm; elemental analysis calcd (%) for C₃₈H₄₄F₆NNiO₄PS₃: C, 51.95; H, 5.05; N 1.59; S, 10.95; found: C, 51.3; H, 4.91; N, 1.84; S, 10.83, the lower than calculated carbon analysis is attributed to the relatively high fluorine content F, 12.97%; unit cell parameters: a = 10.7068(3), b = 16.6627(6), c = 17.9475(9) Å, α = 117.108(2), β = 92.836(2), γ = 106.327(2)°, V = 2675.49(19) ų, Z = 2, P − 1, deposition number CCDC 1897417.

5. Conclusions

A set of four bimetallic complexes of nickel is presented that provides a functional model system for mechanistic work on the skeletal (dis)assembly of structures [M₂(µ-S)₂] of Cu and Fe that are key components in natural metalloproteins. Biological function depends on the kinetic stability of the homovalent form [M₂(µ-S)₂]⁽ⁿ⁺¹⁾⁺, whose disintegration is rendered unfavorable by increasing the energy of the transition state that leads to thermodynamically preferred but inactive structures. In marked contrast, assembly of the open-shell S = 1/2 redox-conjugate form [M₂(µ-S)₂]ⁿ⁺ is thermodynamically and kinetically substantially favored. Relevant factors are covalent bonding along M–S–M linkages, and diverging stereoelectronic preferences of the two metal sites that lower the energy of the transition state toward [M₂(µ-S)₂]ⁿ⁺ core assembly.