2.1 Syntheses of complexes 1a and 1b

The previously reported synthesis of complexes 1a and 1b is outlined below (Fig. 1) 〚5〛. The yields obtained are respectable but the use of 〚Ni(CO)₄〛 to prepare these complexes is a major disadvantage owing to its volatility and extremely high toxicity.

A new synthesis that eliminates the use of 〚Ni(CO)₄〛 was therefore developed. When 〚NiCl₂·dme〛 (dme = 1,2-dimethoxyethane) was reacted with 〚(C₅Me₅)MgCl〛 〚8〛 at low temperature for 24–36 h, the orange suspension gave way to a dark brown colour and the thermally unstable species 〚Ni(μ-Cl)(η-C₅Me₅)〛₂ was formed 〚9〛. When this chloro-complex was treated at low temperature with 〚M(CO)₃(η-C₅H₅)〛^–, the characteristic deep blue colour of the mixed metal complexes 1a or 1b appeared. These highly air-sensitive complexes were extracted from the reaction mixture using pentane, as outlined in the experimental section, and isolated in yields ranging from 30–65%.

The advantage that this method offers is clear, no 〚Ni(CO)₄〛 is used, and the unsaturated heterobimetallic complexes are obtained directly, without a decarbonylation step (step ‘e’ in Fig. 1). This new synthetic method is summarized in Fig. 2.

2.2 Use of simple saturated heterobimetallic complexes as cluster building blocks

Simple homo- and heterobimetallic complexes have been used by others and us as cluster building blocks. Many new and otherwise inaccessible clusters can be made this way and the method is useful. However, typical yields are not always high, and often a wide variety of side-products are observed. A major reason why many side-products are obtained is the rapid metal–metal exchange occurring between L_nM–MˈLˈ_m fragments that takes place in solution. We have shown 〚10〛 that mixtures of closely related Ni–W heterobimetallic complexes readily scramble when mixed, as shown in equation (1) below. This process, which is general for many simple non-bridged bimetallic complexes, is indeed the basis for the synthesis of many such species 〚11〛.

Processes such as this and related ones occur rapidly when two bimetallic species are mixed (in less than 2 min, by ¹H NMR) and lead to crossover mixtures. This, in turn, frequently affords unintended products and results in the target cluster that is obtained in moderate to poor yield. For example, the reaction of a saturated analogue of complex 1b with 〚Co₂(CO)₈〛 was envisaged to produce the Co₂NiW cluster shown in equation (2), by initial CO loss followed by formal addition of a Co₂(CO)₆ unit to an isolobal (η-C₅H₅)Ni–W(CO)₂(η-C₅H₄Me) group. While this species was indeed obtained in moderate yield, other bimetallic complexes and mixed-metal clusters were also formed 〚12〛.

A heterobimetallic system with multiple metal-bond character could in principle anchor the two metals together and suppress metal–metal bond exchange, just as a heterobimetallic system with bridging ligands binds the two metals together more strongly and favours keeping the bimetallic system intact. In order to test this hypothesis, complexes 1a and 1b were reacted with 〚Fe₂(CO)₉〛 and the results were compared with the reaction of the saturated tetracarbonyl analogues of complex 1a with the same di-iron complex.

2.3 The synthesis of heterotrimetallic FeNiM (M = Mo, W) clusters using Ni–Mo and Ni–W complexes as building blocks

Complex 1a reacts with 〚Fe₂(CO)₉〛 according to equation (3) to afford a deep green product 3a.

This species, which is the only product (apart from 〚Fe(CO)₅〛) when the stoichiometry of 1a to the Fe₂ complex is 1:1, is a FeNiMo cluster. Its IR spectrum is complex and shows the presence of terminal, doubly bridging and even potentially triply bridging CO ligands. The ¹H NMR spectra of 3a were not very informative but they did show singlets attributed to C₅H₅ and C₅Me₅ groups in a 1:1 ratio (1:3 proton integrated ratio respectively). This suggested that the complex contained both nickel and molybdenum.

The detailed structure of 3a was established by a single crystal X-ray diffraction study. Complex 3a is the heterotrimetallic cluster 〚FeNiMo(μ₃-CO)(μ₂-CO)(CO)₅(η-C₅H₅)(η-C₅Me₅)〛. Key crystal and data collection parameters, and tables of important bond lengths and angles for 3a are collected in Tables 1–3, respectively. The structure of 3a (hydrogen atoms are not shown, for clarity), is shown in Fig. 3.

Complex 3a crystallizes with two crystallographically independent molecules in the unit cell, but the two clusters are not significantly different from each other (Fig. 3 shows molecule B). The trimetal cluster contains a tetrahedral and chiral core as the μ₃-CO ligand caps the metal triangle, but both enantiomers of each independent molecule exist in the unit cell as the space group ($P\overline{1}$) is non-chiral. The μ₃-CO ligand does not bridge the three metals symmetrically and is perhaps better described as being semi-bridging to the iron atom. Its semi-bridging character is reflected in the respective M–C(μ₃) bond lengths that are longer for the Fe–C(μ₃) distances (2.20 Å, mean) than for the Ni–C(μ₃) (1.96 Å, mean) or even the Mo–C(μ₃) distances (2.16 Å, mean).

One of the only significant differences between the two independent molecules of 3a is the length of the C–O bond for the μ₃-CO ligand 〚1.180(6) Å for molecule A; 1.226(7) Å for molecule B〛. In addition, the Fe(CO)₃ unit is slightly rotated in one molecule with respect to the other one; but otherwise the two molecules are essentially identical. The centroid of the η-C₅Me₅ group in each case lies close to the trimetal plane, but the molybdenum bound η-C₅H₅ ring lies below this plane, away from the C(μ₃) atom in each molecule. Other distances and angles are unremarkable, with all metal–metal distances lying in the usual range.

Clusters containing iron, nickel and molybdenum bonded to each other have been reported 〚13–15〛, and in addition some sulphur and selenium capped clusters with these three metals are known, including the cluster 〚FeNiMo(μ₃-S)(η-C₅H₅)(η-C₅H₄R)(CO)₅〛 〚R = H, C(O)Me, C(O)OMe〛 and other related species 〚16–19〛.

The reaction of the unsaturated nickel–tungsten complex 1b with 〚Fe₂(CO)₉〛 proceeds similarly and affords the analogous cluster 〚FeNiW(μ₃-CO)(μ₂-CO)(CO)₅(η-C₅H₅)(η-C₅Me₅)〛, 3b. While 3b was not structurally characterized, its spectroscopic data are very similar to those of 3a. In addition the mass spectrum shows a parent ion for this formulation with the correct isotopic envelope for a FeNiW species. The very similar spectroscopic data of 3a and 3b strongly suggest that the two species are isostructural. However, 3b appears to be less stable than 3a and is obtained in lower overall yields.

The reaction of the saturated analogue of 1a, the complex 〚(η-C₅Me₅)(OC)Ni–Mo(CO)₃(η-C₅H₅)〛, 2a (see Fig. 1) was also attempted with 〚Fe₂(CO)₉〛 to see whether using the unsaturated and formally double-bonded species imparted any advantages to the cluster synthesis described earlier. The reaction proceeded smoothly, with the green colour of 2a slowly giving way to an olive green solution. Complex 3a was indeed isolated as a product from this reaction, but other species including 〚Mo(CO)₃(η-C₅H₅)〛₂, a little 〚Ni(μ-CO)(η-C₅Me₅)〛₂, a green oil that probably contains a nickel–iron cluster, and other unidentified species were also obtained when the reaction mixture was subject to column chromatography. The overall isolated yields of complex 3a were much lower, clearly showing the advantages of using the formally doubly bonded complexes as building blocks for mixed-metal cluster synthesis.

3.1 General

Reactions were carried out under nitrogen, in rigourously deoxygenated solvents. The ¹H NMR spectra were recorded on a Bruker AM300 instrument. All resonances are singlets. Chemical shifts (in ppm) were referenced to residual CHCl₃ or C₆HD₅. IR spectra were recorded on a Bruker IFS66 FT instrument.

3.2 Synthesis of complexes 1a and 1b

Both complexes are prepared in a similar fashion. In a typical preparation, 〚NiCl₂·dme〛 (755 mg, 3.43 mmol) was suspended in 30 ml THF and cooled to –25 °C. A solution of (C₅Me₅)MgCl 〚8〛 (12.8 ml of a 0.295 M solution, 3.77 mmol) was added dropwise and the solution stirred for 24 h at –25 °C. A solution of fresh 〚Mo(CO)₃(η-C₅H₅)〛^–, prepared by reductive cleavage of 〚Mo(CO)₃(η-C₅H₅)〛₂ (3.43 mmol in 15 ml THF) with 1.0 M KHBEt₃, was then added dropwise to the dark suspension, and the resulting deep blue solution was allowed to warm up to ambient temperature and then pumped down to dryness under vacuum. The dark residue was transferred anaerobically to a filter stick equipped with a pressure equalized side-arm, that was in turn attached to a 50 ml two-necked round bottomed flask with a nitrogen inlet adaptor. When the transfer was complete, a reflux condenser was attached to the top of the filter stick and 30 ml of pentane were added. As the pentane was heated to reflux, the vapours condensed and dripped on the solid residue in a manner similar to a Soxhlet extraction process, dissolving pure 1a, which concentrated and then started to deposit in the round bottomed flask. After an overnight extraction, the process was complete and 1a (779 mg, 52% based on 〚NiCl₂·dme〛) was collected as dark blue-black crystals.

3.3 Synthesis and spectroscopic data for the complexes 〚FeNiMo(μ₃-CO)(μ₂-CO)(CO)₅(η-C₅H₅)(η-C₅Me₅)〛, 3a and 〚FeNiW(μ₃-CO)(μ₂-CO)(CO)₅(η-C₅H₅)(η-C₅Me₅)〛, 3b

The synthesis of 3b is representative and is given here. Complex 3b (327 mg, 0.62 mmol) was dissolved in 25 ml Et₂O and 〚Fe₂(CO)₉〛 (226 mg, 0.62 mmol) was added. The solution was stirred magnetically until all the 〚Fe₂(CO)₉〛 was consumed (ca 1 h) and then pumped down to dryness in vacuo to remove any 〚Fe(CO)₅〛. The residue was redissolved in toluene, to give a deep green solution, and this was filtered through a short (5 cm) Al₂O₃ pad. After the toluene solution was concentrated and layered with pentane, dark green crystals of 3b (353 mg, 82%) deposited at –20 °C. The product was contaminated with traces of red crystals of 〚W(CO)₃(η-C₅H₅)〛₂, which could easily be removed manually. A purer crop of 3b was obtained in somewhat reduced yield (typically 50–70%) by column chromatography on Al₂O₃: traces of 〚W(CO)₃(η-C₅H₅)〛₂ were readily separated from the only other band, cluster 3b.

3.4 Reaction of 〚(η-C₅Me₅)(OC)Ni–Mo(CO)₃(η-C₅H₅)〛 2a with 〚Fe₂(CO)₉〛

The saturated complex 2a (306 mg, 0.66 mmol) was reacted with 〚Fe₂(CO)₉〛 (240 mg, 0.66 mmol) in THF. The solution darkened as the mixture was stirred for 2 h. It was then pumped down to dryness in vacuo, dissolved in ca 4 ml toluene and subjected to chromatography on an Al₂O₃ column. Two red bands eluted with toluene (respectively 〚Mo(CO)₃(η-C₅H₅)〛₂ and 〚Ni(μ-CO)(η-C₅Me₅)〛₂), followed by a non-characterized green band, 4a. Complex 3a eluted next using a toluene:Et2O mixture followed by small quantities of a light brown band, and then a lighter red–brown band that decomposed on the column and was not eluted. Complex 3a was recrystallized from toluene:pentane mixtures and collected in 28% yield (109 mg, 0.18 mmol).

3.5 X-ray data collection

A dark green–black crystal of 3a, of dimensions 0.10 × 0.08 × 0.05 mm, that was grown from a toluene/pentane mixture, was selected and mounted on a Kappa CCD diffractometer. Data were collected using phi-scans and the structure was solved using direct methods and refined against F² using SHELX 97 software 〚20–21〛. No absorption correction was used. A total of 10 169 reflections was collected with 1.44°< θ <27.43°, of which 4799 unique reflections had intensities I > 2 σ(I). All non-hydrogen atoms were refined anisotropically with H atoms introduced as fixed contributors (d_C–H = 0.95 Å, U₁₁ = 0.04). Full data collection parameters, and structural data are available as supplementary material.

Supplementary material

The supplementary material has been sent in electronic format to the Cambridge Crystallographic Data Centre, 12 Union Road, Cambridge CB2 1EZ, UK as cif file No. CCDC 166551, and can be obtained by contacting the CCDC.