Cycloaddition of alkynes to diimino Mo3S4 cubane-type clusters: a combined experimental and theoretical approach

A heterocyclic ligand 4,40-di-tert-butyl-2,20-bipyridine (dbbpy) has been coordinated to the Mo3S4 cluster unit affording the complex [Mo3S4Cl3(dbbpy)3] + ([1]) in a one-step ligand-exchange protocol from [Mo3S4(tu)8(H2O)]Cl4 4H2O (tu = thiourea). The new cluster was isolated as [1]PF6 and [1]Cl salts in high yields and the crystal structure of the latter determined by X-ray analysis. The synthetic procedure was extended to tungsten to afford [W3S4Cl3(dbbpy)3] + ([2]). Kinetic and NMR studies show that [1] reacts with several alkynes to form dithiolene species via concerted [3+2] cycloaddition reactions whereas [2] remains inert under similar conditions. The different rates for the reactions of [1] are rationalised by computational (DFT) calculations, which show that the more electron-withdrawing the substituents of the alkyne the faster the reaction. The inertness of [2] is due to the endergonicity of its reactions, which feature DGr values systematically 5–7 kcal mol 1 more positive than for those of [1].


Introduction
Molybdenum dithiolene complexes have been extensively studied since the 1960s due to their interesting properties and applications. 1Broadly, compounds with five-membered MoS 2 C 2 (R 2 ) chelate rings can be synthesised by reaction of a molybdenum compound with a preformed alkenedithiolate (or related species), i.e.Mo + S 2 C 2 (R 2 ); or result from the addition of alkynes to molybdenum sulfides, i.e.MoS 2 + C 2 (R 2 ).1a The latter method, which implies the formation of two C-S bonds, has been used in the past to generate mono-, 2 di-, 3 and trinuclear [4][5][6][7] dithiolene complexes (see Scheme 1).][10] Our results show that it can be viewed as a concerted [3+2] cycloaddition reaction between a Mo(m-S) 2 moiety of the cluster and the two sp-hybridised carbon atoms of the alkyne; however, differences appear upon modification of the solvent, alkyne substituents, metal coordination environment, or the nature of the metal itself.0] Interestingly, no reaction is observed upon replacement of molybdenum by tungsten in the case of the acac-py cluster. 10In contrast, the isothiocyanate tungsten complex [W 3 S 4 (SCN) 9 ] 5À reacts with acetylene to afford a final product containing two acetylene groups with different coordination modes [W 3 (m 3 -S)(m 3 -SCHQCHS)(m-SCHQCH 2 )(NCS) 9 ] 5À . 5n order to shed more light on the mechanism of these processes, we decided to investigate the reaction of the trinuclear [M 3 S 4 Cl 3 (dbbpy) 3 ] + complexes (M = Mo ([1] + ) and W ([2] + )) with various alkynes.By doing so we are able to analyse the influence of the nature of the metal and the outer ligands in the cycloaddition process.[1] + and [2] + have been synthesised adapting a procedure recently published by some of us 11,12 on similar systems and the crystal structure of [1] + has been determined by single-crystal X-ray diffraction methods.For the reaction of [1] + and [2] + with alkynes, unsaturated compounds with different substituents have been chosen.All other reagents were obtained from commercial sources and used as received.Organic solvents were dried by standard methods before use.

Physical measurements
Elemental analyses were performed with a EuroEA3000 Eurovector analyser.IR spectra were recorded in the 4000-400 cm À1 range with a Perkin-Elmer System 2000 FTIR spectrometer with samples in KBr pellets.The characteristic IR frequencies for the bipyridine ligands were assigned on the basis of previous results.A triple quadrupole mass spectrometer with an orthogonal Z-spray electrospray source (Waters, Manchester) was used.The temperature of the source block was set to 100 1C, and the desolvation temperature was set to 120 1C.A capillary voltage of 3.3 kV was used in the positive scan mode, and the cone voltage was set to U c = 20 V. Mass calibration from m/z = 50 to 3000 was performed with a solution of sodium iodide in 2-propanol/water (50 : 50).Sample solutions in CH 3 CN were injected with a syringe pump directly connected to the ESI source at a flow rate of 10 mL min À1 .The observed isotopic pattern of each compound perfectly matched the theoretical isotope pattern calculated from their elemental composition by using the MassLynx  -di-t-butyl-2,2 0 -bipyridine (0.28 g, 1.0 mmol) and CH 3 CN (40 mL) was refluxed for 5 hours.After cooling to room temperature, the resulting suspension was taken to dryness.The solid was then extracted with CH 2 Cl 2 , and the excess of insoluble thiourea was removed by filtration.The complex was precipitated with an excess of hexane to give a green-brown solid.Yield: 0.42 g (96% [1]PF 6 : this compound was prepared by the same procedure than [1]Cl, but after removal of insoluble thiourea, the solution was loaded onto a silica gel column.After the column was washed with CH 2 Cl 2 , elution with a solution of KPF 6 in acetone (10 mg mL À1 ) afforded a concentrated brown solution.This solution was evaporated to dryness, the residue extracted with CH 2 Cl 2 and filtered in order to eliminate the inorganic salts.Finally the resulting solution was allowed to evaporate slowly in air to give 0.34 g (70%) of brown solid.For  l (e) 600 nm (626 M À1 cm À1 ).

Structural determination
Single crystals of [1]ClÁ5H 2 O suitable for X-ray analysis were obtained by recrystallization of [1]Cl from CH 2 Cl 2 /hexane mixture.The diffraction data were collected on a Super Nova, Dual (CuKa, l = 1.5418Å, at 200 K) by doing f and o scans of narrow (0.51) frames.The structure was solved by direct methods and refined by full-matrix least-squares treatment against |F| 2 in anisotropic approximation with SHELXL program. 13bsorption corrections were applied with SCALE3 ABSPACK scaling algorithm. 14All non-hydrogen atoms were refined anisotropically.Hydrogen atoms of water molecules were not localised directly from the experiment.To observe all amounts of highly disordered water molecules a SQUEEZE procedure 15 and a PLATON program set 16 were used.This gives 150 e per unit cell or 2 additional water molecules per formula unit (in total 5 molecules per unit).Crystallographic data and refinement details are given in Table S1 (ESI †).CCDC 1457770.

Kinetic measurements
The kinetic experiments were carried out with an Applied Photophysics SX-18MV stopped-flow spectrometer provided with a PDA1 photodiode array detector, and with a Cary 50 Bio UV-Vis spectrophotometer.All experiments were carried out at 25.0 AE 0.1 1C in acetonitrile by mixing a solution of [1] + (7.5 Â 10 À5 M) with another solution containing the alkyne in a concentration range large enough (0.01-0.50 M) to ensure pseudo-first order conditions.Preliminary experiments at two different cluster concentrations were carried out to confirm the first order dependence of the observed rate constants on this reagent.The spectral changes in the 700-1000 nm range were analysed with program SPECFIT-32. 17

NMR experiments
NMR spectra were recorded on Agilent 500 and 600 DD2 and Bruker Avance 500 spectrometers at room temperature.Chemical shifts are given in parts per million from SiMe 4 ( 1 H and 13 C{ 1 H}) or CFCl 3 ( 19 F). 1 H and 13 C{ 1 H} NMR signal assignments were confirmed by 1 H-gCOSY, g-HSQC and g-HMBC experiments.
NMR samples were prepared under Ar, although products are not particularly sensitive to air or humidity.
2.5.1.Isolation and NMR characterization of the type I product formed from [1] + and dmad.170 mg of cluster [1]PF 6 (0.115 mmol) were dissolved in 1 mL of CH 2 Cl 2 under inert atmosphere, and 30 mL of the corresponding alkyne were added.The mixture was stirred for 15 min at room temperature.Then, 10 mL of Et 2 O were added, causing the immediate precipitation of a dark brown solid, which was filtered, washed twice with Et 2 O, and dried under vacuum, yielding 136 mg (73%).

Computational details
DFT calculations were carried out with the Gaussian 09 (Revision D.01) suite of programs. 18Geometry optimizations were carried out with the BP86 functional in combination with the SDD pseudopotentials and associated basis sets (BS1). 19The clusters [1] + and [2] + were modelled using 2,2 0 -bipyridine instead of dbbpy ligands, i.e. as formally resulting from the substitution of t butyl groups by H atoms in the latter.Frequency calculations at the same level of theory confirmed that all the optimised structures on the potential energy surface correspond to either minima (all positive eigenvalues) or transition state structures (one imaginary eigenvalue).In addition, IRC calculations and subsequent geometry optimizations were employed to confirm the minima linked by each transition state.All reported energies are free energies in solution, computed at 298.15 K and 1 atm.They are based on single-point gas phase energies with a larger basis set system, BS2, on the previously BS1-optimised structures, subsequently corrected by including thermal corrections from energy to free energy within the rigidrotor-harmonic-oscillator, as well as single-point solvent and dispersion corrections.BS2 employs the SDD pseudopotentials and associated basis sets for Mo, W and S (with added d-orbital polarization on the latter 20 ) and the Pople-style 6-311+G(2d,2p) basis set for the remaining atoms.Solvent corrections were computed at the BP86/BS1 level and made use of the PCM approach with standard parameters for acetonitrile. 21,22Dispersion effects were included using Grimme's D3(BJ) parameter set. 23,24Structures were generated using CYLview (color code: Mo (turquoise), W (blue), N (purple), Cl (green), S (yellow), C (grey), H (white)). 25

Synthesis and characterisation
In a previous work we have shown that the high lability of the thiourea ligands in the easily accessible [Mo 3 S 4 (tu precursor can be used to prepare tri-substituted Mo 3 S 4 complexes with 2,2 0 -bipyridine (bpy) and 1,10-phenanthroline (phen) ligands. 12 1 together with a list of selected bond distances.The structure consists of an incomplete cuboidal arrangement in which the molybdenum and sulfur atoms occupy adjacent vertices with a missing metal position.Metal-metal and metal-sulfur distances within the Mo 3 S 4 cluster core follow the same tendencies observed for other trinuclear Mo 3 S 4 species. 26,27Each molybdenum atom presents a distorted octahedral coordination environment, surrounded by three sulfur ligands, one chlorine, and the two nitrogen atoms of the dbbpy ligand.[1]ClÁ5H 2 O crystallises in Pa% 3 space group due to the specific packing of the bulky cationic cluster units, which have almost spherical shape.Since [Mo 3 S 4 Cl 3 (phen) 3 ]ClÁ4H 2 O crystallises in P3 group, 12 we believe that switching from 1,10-phenantroline (phen) to bulkier dbbpy ligands makes the cluster cations more spherical, so that they can be packed in a more symmetrical space group.The crystal structure of [Mo 3 S 4 Cl 3 (phen) 3 ]ClÁ4H 2 O features weak C-HÁ Á ÁCl interactions that connect the cluster units into the sheets with 6 3 plane net topology, whereas in [1]ClÁ5H 2 O the spherical form of the cluster cations prevents such interactions.
The structural integrity of [1] + and [2] + in solution has been proved by 1 H NMR spectroscopy and electrospray-ionization (ESI) mass spectrometry.The positive mass spectrum of [1] + registered at 20 V shows the pseudomolecular peak centered at m/z = 1327 and a prominent peak at m/z = 665 attributed to the exchange of one chlorine atom by an acetonitrile solvent molecule [1 À Cl + CH 3 CN] 2+ on the basis of the m/z value and its characteristic isotope pattern (see Fig. S2, ESI †).In the case of complex [2] + , the positive mass spectrum registered at 20 V only shows a signal corresponding with the pseudomolecular peak centered at m/z = 1591. 1H NMR spectra for [1] + and [2] + shows that the symmetry found in the solid state is conserved in solution, featuring three equivalent dbbpy ligands asymmetrically bound to the metal centres.Therefore, six signals are observed for the dbbpy aromatic protons and two signals for the t Bu group protons.

Kinetic studies on the reaction of [1] + and [2] + with alkynes
In analogy with our previous studies on the reactivity of [M 3 S 4 (acac) 3 (py) 3 ] + (M = Mo, W), 9,10 the reactions of [1] + with a number of symmetrical and unsymmetrical alkynes were analysed.Specifically, the former group includes acetylene dicarboxylic acid (adc), dimethyl acetylenedicarboxylate (dmad) and 2-butyn-1,4-diol (btd), whereas phenylacetylene (PhA), propargyl alcohol (PrA) and ethylpropiolate (EtPr) belong to the latter.Spectral changes in the time scale of the stopped-flow technique are observed in all cases, with all reaction products showing an intense band centered at ca. 1000 nm and a shoulder at ca. 800 nm, as exemplified in Fig. 2 for the reaction with dmad.Unfortunately, the precise positions of the bands cannot be determined because they extend beyond the detector range.In any case, detailed analysis of the stopped-flow data indicates that a kinetic model with a single exponential is sufficient in all cases for a satisfactory fit, and therefore indicative of no subsequent processes.The appearance of the near-IR band is typical of [3+2] cycloaddition products of the reaction between cuboidal clusters and alkynes, 5,6 so this spectral feature is (1) The k 1obs values obtained for [1] + under pseudo-first order conditions of alkyne excess show a linear dependence on the concentration of the latter (Fig. 3), and fitting to either eqn (1) or (2) leads to the second-order rate constants listed in Table 1.It is noteworthy that, for the reaction with btd, such linear dependence shows a nonzero intercept, indicative of a reversible process, and eqn (2) was employed accordingly.For comparison, Table 1 also includes a summary of the rate constants obtained for the first step of the reaction between [Mo 3 S 4 (acac) 3 (py) 3 ] + and the same set of alkynes.In both cases it is clear that adc and dmad react faster than the remaining alkynes.This was rationalised as resulting from the different electron-withdrawing substituents of the alkyne. 9All in all, the data in Table 1 indicate that modifications of the metal coordination sphere do not lead to substantial changes in the kinetics of reaction of these Mo clusters with a given alkyne.
An interesting experimental feature of the reaction between [Mo 3 S 4 (acac) 3 (py) 3 ] + and alkynes consists in the appearance of subsequent kinetic steps for the formation of additional compounds beyond the so-called type I product formed in the initial [3+2] cycloaddition (Scheme 2).This behaviour has been observed for a number of other cuboidal clusters, 4,5,28 and Shibahara et al. proposed that such species are formed via protonation of the type I product, 5 which leads to the cleavage of one of the C-S bonds with formation of the so-called type II product, which again features a MS 2 unit available for the subsequent [3+2] cycloaddition of the second alkyne molecule, finally leading to the type III product.Dithiolate protonation is therefore required in this model to provide the MoS 2 unit necessary for a second cycloaddition.In order to test whether [1] + can behave similarly, additional kinetic experiments for the reaction with dmad were carried out under an excess of trifluoroacetic acid. 29Similar spectral changes were nonetheless observed, and in fact the overall fit of the whole set of data available led to a k 1 value of 9.8 AE 0.6 M À1 s À1 , analogous to that derived in the absence of the acid (see Fig. S3, ESI †).On the other hand, NMR experiments in the presence of, first an excess of dmad, and then an excess of trifluoroacetic acid, showed the release of the dbbpy ligands and, consequently, the cluster decomposition.When taken together with the stoppedflow information, these results suggest that dbbpy dissociation must occur in a slower time scale than the initial cycloaddition.Thus, protonation by strong acids leads to decomposition, whereas the reaction of [1] + with alkynes in the presence of weaker carboxylic acids (such as adc) or without any acid stops after the formation of type I products.This behaviour is in contrast with that of [Mo 3 S 4 (acac) 3 (py) 3 ] + , which reacts with adc showing consecutive kinetic steps with participation of two alkyne molecules in addition to pyridine dissociation. 9iven that [1] + and [Mo 3 S 4 (acac) 3 (py) 3 ] + only differ in the ligands bound to the molybdenum centres, we believe that the results shown herein confirm our previously postulated hypothesis that the polyphasic kinetics of the reactions between [Mo 3 S 4 (acac) 3 (py) 3 ] + and alkynes involve participation of the pyridine ligands. 9,10  The reaction of [1] + with an excess of dmad was also monitored by NMR in CD 2 Cl 2 .The 1 H NMR spectrum of the reaction product shows a large number of mostly overlapping and sometimes broad signals, accompanied by those corresponding to free dmad.Given that ESI-MS experiments rule out cluster decomposition, the resulting complex NMR pattern must be due to the lower symmetry of the resulting species, where dbbpy aromatic hydrogens become non-equivalent.
Bidimensional 1 H-1 H coupling experiments (gCOSY) allowed to elucidate the pattern showed by the aromatic signals.Based on the cross peaks with the neighbouring dbbpy-H 3,3 0 , it was possible to identify six resonances corresponding to dbbpy-H 2,2 0 .Hydrogens H 3,3 0 and H 5,5 0 show a reduced number of signals due to overlapping, but their overall peak integration also correspond to six.The formation of a dithiolene unit is supported by the presence of two singlet signals at 3.41 and 3.53 ppm corresponding to the carboxylate methyl groups, which become non-equivalent upon coordination to the cluster.The observed number of signals are attributable to the fully non-equivalent dbppy ligands (i.e. three ligands with two H 2,2 0 atoms would give six resonances). 13C{ 1 H} NMR assignment was performed with the help of two-dimensional HSQC and HMBC experiments.HSQC correlates directly bonded 1 H and 13 C nuclei and it allows to identify individually the twelve dbbpy-C 2,2 0 resonances for the twelve dbbpy-H 2,2 0 signals observed in the 1 H NMR spectra (see Fig. 5).Similarly, dbbpy-C 3,3 0 and dbbpy-C 5,5 0 were assigned.The dithiolene carbon atoms are both quaternary, and therefore they had to be identified from the 13 C{ 1 H} NMR spectrum by exclusion from HSQC.Note that they are the only carbons that show no correlations in this experiment.They appear at 135.1 and 134.0 ppm, values that compare well with those reported previously (133.1 and 136.7 ppm). 9,10The remaining quaternary carbon atoms were identified by their HMBC long C-H correlation with neighboring protons.The carboxylate methyl groups in dmad adduct show a three-bond correlation with the carboxylic carbon atom, which appear at 164.1 and 164.9 ppm.

Computational (DFT) studies
The mechanism of formation of type I products between the clusters [1] + and [2] + and alkynes has been computationally analysed (see Computational details).The objective of the study is threefold: (a) to understand why [2] + does not react with alkynes whereas [1] + does; (b) to confirm that those reactions lead to type I products via concerted [3+2] cycloaddition processes, as has been described for other cuboidal clusters, [8][9][10] and (c), to analyse the effect of the substituents at the alkyne on the kinetics and thermodynamics of these cycloadditions.To do so, the reactions of the clusters [1] + and [2] + with all the alkynes employed in the experimental section were studied.
The computations began with the optimization of the structures of the reactants, transition states, and corresponding type I reaction products for these reactions.Despite multiple attempts, in line with our previous mechanistic studies on the same  reaction for other cuboidal clusters [8][9][10] it was only possible to locate transition states (TS) for the concerted formation of two C-S bonds between cluster and alkyne, i.e. no intermediates with one C-S bond were located.A summary of the activation (DG ‡ ) and reaction (DG r ) free energies for these concerted [3+2] cycloadditions is included in Table 2, whereas the structures of the computed TS and products for the reactions of both clusters with btd are depicted in Fig. 6 (see ESI † for Cartesian coordinates of all of them).The DG r values in Table 2 clearly show that substitution of Mo by W leads to crucial changes in the thermodynamics of the processes, as the reaction free energies for [2] + are systematically larger than those for [1] + by 5-7 kcal mol À1 .This explains the experimentally observed lack of reaction between [2] + and alkynes, as those thermodynamic differences most of the reactions of this species endergonic while those of [1] + are all exergonic.It is worth noting that small negative DG r values are nonetheless obtained for the reactions of [2] + with adc, dmad and EtPr, a discrepancy likely to be associated with the accuracy of the method.Regarding the free energy barriers, the values in Table 2 show that compared with [1] + , those of [2] + are only 1-2 kcal mol À1 larger and thus still typical of reactions that take place relatively fast at room temperature.All in all, these results are similar to those reported for [Mo 3 S 4 (acac) 3 (py) 3 ] + and its tungsten analogue, 10 where it was also concluded that the main effect of metal substitution consists of an increase in the reaction free energies.By using a combination of activation strain model and energy decomposition analyses that report showed that substituting Mo by W induces changes in the electronic structure of the cluster that result in weaker intraand intermolecular orbital interactions.Interestingly, the smaller the distance between cluster and alkyne, the larger the energy difference between the potential energy surfaces for the reactions of both clusters with the same alkyne.Thus, as cluster and alkyne become closer throughout the [3+2] cycloaddition reaction, the differences in DG ‡ are always smaller than those in DG r . 10n agreement with the similar ionic radii of molybdenum and tungsten, 30 substitution of Mo centers by W does not lead to significant geometrical changes.As exemplified in Fig. 6 for the reactions with btd, only small deviations in C-C and C-S bond distances are computed.Despite the concerted nature of the processes, confirmed by IRC calculations, analysis of the TS structures shows that in most cases one of the C-S bond forming distances is slightly longer than the other, i.e. for TS([1] + ,btd) these are 2.43 and 2.53 Å.Similar differences in C-S bond distances at the TS structures for [Mo 3 S 4 (acac) 3 (py) 3 ] + have been already reported, and these were explained on the basis of the different orientation of the R groups of RCRCR alkynes, or the different steric and electronic properties of R and R 0 groups in RCRCR 0 alkynes. 9In the present case, such deviations are also expected to result from the coordination environment of the metal centres, which features two ligands with different trans effect, i.e. chloride and 2,2 0 -bipyridine, trans to the sulfur centres involved in the cycloaddition.This is supported by the computation of the hypothetical transition state for the reaction of [1] + with acetylene, which yields C-S bond distances of 2.38 and 2.41 Å (see Fig. S4, ESI †).
From a kinetic point of view, the computed DG ‡ values for the reactions of [1] + with alkynes are in line with the experimental rate constants, showing that those for the reactions with adc and dmad are slightly smaller than the remaining ones (see Table 2).Changing the substituents of the alkyne also leads to variations in the DG r values, although in all cases product formation remains exergonic.In general, both kinetic and thermodynamic differences computed for the reactions of [1] + with all the alkynes are similar to those reported for the reactions of [Mo 3 S 4 (acac) 3 (py) 3 ] + , and they can be explained on the basis of the different electronwithdrawing properties of the substituents of the alkyne. 9

( 3 . 3 Scheme 2
Scheme 2 Proposed mechanism for the formation of the so-called type I, II and III products between M 3 S 4 (M = Mo, W) clusters and alkynes.For clarity the m 3 -S ligand and those completing the coordination environment of each M centre have not been included.

Fig. 5
Fig. 5 Bidimensional HSQC spectrum of the species resulting from the reaction between [1] + and dmad.Detail of the aromatic dbbpy 1 H-13 C cross-peaks.

Fig. 6
Fig. 6 DFT-optimised structures of the transition states and type I products of the [3+2] cycloaddition reaction between btd and the clusters [1] + and [2] + .For clarity, only the dbbpy and chloride ligands bound to one metal centre are drawn.Selected distances are given in Å.

Table 2
Activation (DG ‡ ) and reaction (DG r ) free energies for the formation of type I reaction products between the clusters [1] + and[2]+ with the studied alkynes.The differences in these parameters, DDG = DG([2] + ) À DG([1] + ), are also included.All values are given in kcal mol À1