Reactions of aluminium(i) with transition metal carbonyls: scope, mechanism and selectivity of CO homologation

Over the past few decades, numerous model systems have been discovered that create carbon–carbon bonds from CO. These reactions are of potential relevance to the Fischer–Tropsch process, a technology that converts syngas (H2/CO) into mixtures of hydrocarbons. In this paper, a homogeneous model system that constructs carbon chains from CO is reported. The system exploits the cooperative effect of a transition metal complex and main group reductant. An entire reaction sequence from C1 → C2 → C3 → C4 has been synthetically verified. The scope of reactivity is broad and includes a variety of transition metals (M = Cr, Mo, W, Mn, Re, Co), including those found in industrial heterogeneous Fischer–Tropsch catalysts. Variation of the transition metal fragment impacts the relative rate of the steps of chain growth, allowing isolation and structural characterisation of a rare C2 intermediate. The selectivity of carbon chain growth is also impacted by this variable; two distinct isomers of the C3 carbon chain were observed to form in different ratios with different transition metal reagents. Based on a combination of experiments (isotope labelling studies, study of intermediates) and calculations (DFT, NBO, ETS-NOCV) we propose a complete mechanism for chain growth that involves defined reactivity at both transition metal and main group centres.


Introduction
Chemical reactions that allow the formation of useful ne chemicals from C 1 building blocks such as CO or CO 2 are of contemporary interest. 1 These reactions hold promise for sustainable chemical manufacturing methods, as CO can be obtained from renewable sources, including biomass. 2,3 One approach that has a long history is the Fischer-Tropsch reaction. Implemented on large scales and mediated by heterogeneous transition metal catalysts (typically cobalt, iron, ruthenium), the Fischer-Tropsch process (eqn (1)) allows access to mixtures of hydrocarbons with an Anderson-Schulz-Flory distribution from CO and H 2 . 4 (n)CO + (2n)H 2 / C n H 2n + (n)H 2 O Homogeneous transition metal complexes have been studied as models of the active sites of Fischer-Tropsch catalysis. 5,6 While numerous systems have been developed, there is a growing focus on systems in which CO units are reductively combined to make oxygenated chains, {C n O n } mÀ (n ¼ 2-6).  For example, Cloke and coworkers have shown that uranium(III) compounds can be used to construct carbon chains by combining two (ethynediolate), three (deltate) or four (squarate) CO units in a reductive coupling. [34][35][36] Related reactions of low-oxidation state magnesium(I) complexes have been reported by Jones and coworkers. In the absence of a catalyst, three units of CO combine to form a cyclic deltate anion, 37 while in the presence of 10 mol% of [Mo(CO) 6 ] reductive hexamerisation to form a benzenehexolate derived from six units of CO occurred. 38 Although the direct relevance of these systems to Fischer-Tropsch catalysis remains a point of debate, their formation has captured the imagination of the community. Not least because these systems, and the mechanistic information gained from them, could act as a foundation for the development of homogeneous Fischer-Tropsch catalysts for the selective construction of oxygenated hydrocarbons (e.g. polyols) from CO and H 2 . 39 Despite these advances a clear limitation of the systems reported to date can be identied: in the majority of cases reactions result in the formation of isolable products in which {C n O n } mÀ chains coordinate to metal complexes through thermodynamically stable M-O bonds (e.g. M ¼ U, Mg). In contrast, cases in which reactive M-C bonds are formed during carbon chain growth from CO are far rarer. 6 Such systems may offer an opportunity to study the (oen opaque) individual steps of carbon chain growth. In addition, although it is known that variation of the transition metal can inuence product distributions in heterogeneous Fischer-Tropsch catalysis, 40 there are few dened examples of organometallic reactions in which multiple carbon chain topologies are accessible from a single reaction.
Recently we reported the reactions of the aluminium(I) reagent [{(ArNCMe) 2 CH}Al] (Ar ¼ 2,6-di-iso-propylphenyl, [Al]) with [W(CO) 6 ]/CO mixtures. 41 We documented a system in which {C n O n } 4À (n ¼ 3,4) carbon chains could be synthesised through the cooperative action of the metal complexes. Remarkably these reactions not only proceeded from a dened transition metal carbonyl starting material, but they also led to the isolation of intermediates and products bearing reactive Al-C bonds allowing us, for the rst time, to elucidate the mechanism of chain growth from C 1 / C 3 / C 4 species. Herein, we show that the reaction scope can be expanded to a range of transition metal carbonyl precursors. We shed additional light on the mechanism of chain growth through isolation and characterisation of the C 2 intermediate and demonstrate that systematic variation of the transition metal impacts both the apparent rate and selectivity of the carbon-carbon bond forming steps.
Results and discussion C 1 to C 4 chain growth with CO We have previously reported the reaction of [Al] with [W(CO) 6 ] and the isolation of the C 3 homologation product and its chain growth to a C 4 analogue. Keen to understand the role of the transition metal and expose the complete mechanism for chain growth from C 1 to C 4 , we conducted a series of reactions in which the transition metal fragment was varied. Reaction of [Al] with a series of group 6-9 metal carbonyl complexes in the presence of CO was investigated (Scheme 1 and Fig. 1). The transition metal complexes were selected based on an 18-electron conguration and low-spin electronic structure. In addition, a cis-dicarbonyl motif was considered an essential reactive component for chain growth.
These reactions allowed the isolation of a series of heterometallic products incorporating C 2 , C 3 or C 4 carbon chains derived from a 2 : 1 stoichiometry of [Al] : 1-M. The aluminium(I) complex [Al] acts as a formal 2-electron reductant in these reactions transferring a total of 4-electrons to the CO derived ligand in the chain growth process. Most notably modication of the electronics at the transition metal appears, qualitatively, to inuence the reaction rate of each chain growth step, allowing isolation of a previously unobserved C 2 intermediate. Hence while reaction of 1-Cr, 1-Mo, 1-W, or 1-Co with [Al] under CO (1 atm.) led directly to the C 3 products 3-Cr, 3-Mo, 3-W, and 3-Co at 25 C, in the case of 1-Mn conversion to the C 2 homologue 2-Mn was achieved.
2-Mn is an incredibly rare example of a C 2 homologation product that contains multiple reactive metal-carbon bonds. Although related complexes have been proposed as reaction intermediates, they are seldom isolated. 2-Mn could be isolated and shown to convert to 3-Mn under an atmosphere of CO at 25 C. A minor isomeric product tentatively assigned as 3-Mn 0 was observed alongside 3-Mn in crude reactions mixtures. 3-Mn and 3-Mn 0 form in a 9 : 1 ratio and likely co-crystallise: 3-Mn could not be separated from 3-Mn 0 by fractional crystallisation. Analogous reactions with the rhenium precursor ultimately shed light on this data. The reaction of 1-Re with CO formed a 3 : 1 mixture of two products from CO homologation, the expected product 3-Re which contained [6,4]-fused ring system, alongside 3-Re 0an isomer with a [5,5]-fused ring system (Scheme 1). 42 The ratio of 3-Re : 3-Re 0 was unaffected by heating for 18 h at 100 C. Furthermore, isolated samples of 3-Re do not convert to mixtures of 3-Re and 3-Re 0 . DFT calculations suggest that 3-Re 0 is more stable than 3-Re by 18.8 kcal mol À1 , despite forming as the minor product. These data suggest the reactions Scheme 1 Synthesis of C 2 to C 4 carbon chains by CO homologation. Reactions conducted in C 6 D 6 or C 6 H 6 .
are under kinetic control with the selectivity determining step occurring during the mechanism of C 1 to C 3 chain growth. In all other cases, [6,4]-fused ring systems were identied as the only isomer of the C 3 chain. This observation along with the variation of product ratios with different transition metal precursors, i.e. 3-Mn : 3-Mn 0 (9 : 1) and 3-Re : 3-Re 0 (3 : 1), demonstrate that the transition metal fragment impacts the selectivity of the CO homologation process.
Under more forcing conditions (1 atm., 100 C), 3-Cr, 3-Mo, 3-W, 3-Mn, and 3-Co could all be converted to the C 4 products 4-Cr, 4-Mo, 4-W, 4-Mn, and 4-Co respectively. These reactions were found to be reversible and at elevated temperatures (100-150 C) under a dinitrogen atmosphere the de-insertion of CO could be observed spectroscopically. Mixtures of 3-Re : 3-Re 0 reacted selectively to form 4-Re. Only 3-Re was consumed in this reaction suggesting that the strained [6,4] fused ring system is more labile than the [5,5] isomer. 4-Re could be separated from 3-Re 0 through fractional crystallisation. Subsequent heating of 4-Re under static vacuum to effect CO de-insertion allowed the isolation of pure samples of 3-Re.

Multinuclear NMR and infrared characterisation of 2-4
In C 6 D 6 solution, 2-Mn displays 13 C NMR resonances for the carbonyl and isocarbonyl ligands at d ¼ 236.8 and 189.5 ppm respectively, along with diagnostic resonances for the C 2 ligand fragment at d ¼ 249.2 and 167.2 ppm. The most deshielded resonance is assigned to the C 2 position and is consistent with the location of this atomic site adjacent to the transition metal. In the solid-state, the IR spectrum of 2-Mn shows an absorption at 1921 cm À1 assigned to the n(CO) vibrations of the carbonyl ligand. The vibration(s) associated with the isocarbonyl ligand could not be assigned clearly as they overlap with stretches associated with the ligand and are shied into the ngerprint region. The C 3 homologation products, 3-Cr, 3-W, 3-Mn, 3-Re and 3-Co are characterised by diagnostic 13 C resonances ranging from d ¼ 265-315 ppm and 167-172 ppm for the C 1 and C 2 positions respectively. The largest variation in these data across the series is seen for the C 1 position as it is connected directly to the transition metal and inuenced by the nature of M]C 1 bond. These chemical shis are consistent with the assignment as a metallocarbene ligand. [43][44][45] Chemical shis of the C 2 position are insulated from the transition metal fragment and vary little across the series. In nearly all cases the quadrupolar broadening associated with the 27 Al nucleus prevents identication of the C 3 resonance. Despite the different topologies, chemical shis of the C 1 and C 2 position of 3-Re and 3-Re 0 are remarkably similar. Characterisation of 13 C enriched isotopomers of 3-W allows aspects of the J coupling in the carbon chain to be resolved and the C 3 resonance to be unambiguously identied. Hence, 13 CO enriched 3-W shows the C 3 resonance at d ¼ 176.5 ppm. The coupling constants in the chain were determined as 1 J C2-C3 ¼ 41.8 Hz, 1 J C1-C2 ¼ 37.1 Hz, along with 1 J W-C1 ¼ 102 Hz determined from natural abundance satellites of 183 W.
The C 4 homologation products, 4-W, 4-Mn and 4-Re are characterised by diagnostic 13 C resonances for the C 1 , C 2 , and C 3 positions which range from d ¼ 281-349 ppm, 161-169 ppm, and 135-145 ppm respectively. 13 C enriched samples of 4-W provide 13 C-13 C coupling constants of the carbon chain with 1 J C1-C2 ¼ 35.8 Hz, 1 J C3-C4 ¼ 27.4 Hz, and 2 J C2-C4 ¼ 9.1 Hz. The coupling from W to the C 1 position of 4-W is 1 J W-C1 ¼ 109 Hz and is similar to the value observed in 3-W, suggesting little change in the nature of the carbene ligand on elongation of the carbon chain from C 3 to C 4 . 41

Solid-state characterisation of 2-4
The X-ray structure of 2-Mn incorporates a [7,4] ortho-fused ring system. Bond lengths and angles at the main group fragments support the assignment of an aluminium +3 oxidation state. 46 The 7-membered ring system includes both Mn and Al centres along with the C 2 fragment and isocarbonyl ligand. The C 1 /C 2 distance is $2.6 A and the system is pre-organised for carboncarbon bond formation through ring contraction. The C 1 -O 1 bond length of 1.209(6) A of the isocarbonyl ligand is $0.1 A longer than the terminal carbonyl of 1.097 (8) A. The C 2 fragment is bound to Mn through a short Mn-C 2 bond of 1.971 (5) A. The C 2 -C 3 distance of 1.481 (7) A is beyond that expected for a C]C bond. Similarly, the C 2 -O 2 bond of 1.500 (6) 47 While neither isolated nor spectroscopically characterised, these species have been inferred by DFT studies. The C 2 fragment of 2-Mn is related to these proposed zig-zag intermediates, however the C 2 ligand in 2-Mn shows considerable asymmetry and a unique elongated C-O bond that strongly suggests an unusual electronic structure (Fig. 2).
Except for the transition metal fragment, the solid-state structures of 3 and 4 vary only slightly within their series. Key bonding metrics for both the C 3 and C 4 carbon chains are consistent, within error, across every single crystal X-ray diffraction study reported herein (Tables 1 and 2). Although we have commented on these types of structures before, some details are included here for clarity. 3 possesses an ortho-fused [6,4] ring system derived from contraction of the [7,4]  contains a [5,5] ortho-fused ring system. In all cases, the carbon atoms of both the C 3 and C 4 carbon chains are close to idealised sp 2 hybridised. The C 1 -C 2 and C 2 -C 3 bond lengths of 3 range from 1.428(6)-1.482(9) and 1.375(4)-1.399(6) A respectively; those for 4 are similar. In combination with the C 3 -C 4 lengths of 4 which vary from 1.486 (7) (9) a As two molecules were found in the asymmetric unit, data is provided for 3-Mo(B) one of these molecules. b Data from ref. 41. length and 1 J W-C coupling constant strongly suggest that the metal-ligand interaction in these complexes is very similar. ETS-NOCV calculations are consistent with the formalisation of the M]C 1 interaction of 3-Mn as a metallocarbene. Extended Transition State Natural Orbitals for Chemical Valence (ETS-NOCV) calculations involve partitioning the molecule into fragments and interrogating the interaction energy between said fragments. From this analysis, the stabilising component from the orbital interaction can be derived and further decomposed into contributions corresponding to pairs of natural orbitals for chemical valence. The total DE orb of the carbon chain ligand and transition metal fragment of 3-Mn is À80.1 kcal mol À1 . There are two principal components Dr 1 (À41.7 kcal mol À1 ) and Dr 2 (À25.2 kcal mol À1 ) comprised of sdonation from a lled C 1 -based orbital to the metal, along with p-backdonation from a metal d-orbital to a p z -based orbital on C 1 (Fig. 3a). A similar bonding situation is found in 4-Mn.
Charge localisation on Mn, along with the large C 1 -O 1 and C 1 -C 2 WBIs can be rationalised by a series of zwitterionic resonance structures, reminiscent of those commonly invoked for Fischer carbenes (Fig. 3b). The electronic structure of 2-Mn is more complex. While variable temperature NMR data suggests that 2-Mn may undergo uxional rearrangement in solution (vide infra), there is no evidence this species is paramagnetic. The calculated singlet-triplet gap for 2-Mn is 38.0 kcal mol À1 , while such values are known to be functional dependent, the data strongly suggesting a singlet ground state. The Mn-C 2 WBI in 2-Mn (0.58) is notably lower than the Mn-C 1 values in 3-Mn and 4-Mn. In combination with the large C 2 -C 3 WBI (1.79) and increased charge accumulation on C 2 (+0.15), C 3 (À0.48), Mn (À0.32) in this complex relative to 3-Mn and 4-Mn, these calculations suggest a unique electronic structure of 2-Mn with the C 2 ligand adopting some vinyl character. ETS-NOCV calculations support this assertion and capture only a s-component to the bonding with no clear p-backdonation from the metal to C 2 (ESI †). The WBIs calculated for the isocarbonyl fragment suggest a lengthened C]O bond (1.38) and compressed M]C bond (1.52) relative to the terminal CO ligand as is typical of bridging isocarbonyl fragments.

Mechanism of C 1 to C 4 chain growth
Preliminary experiments using 13 C labelled materials [W( 13 CO) 6 ] and 13 CO showed that the rst two carbon atoms of  3-W and 4-W (C 1 and C 2 ) derive from the transition metal carbonyl, while the second two carbon atoms (C 3 and C 4 ) derive from atmospheric CO. Further, these labelling experiments reveal that an additional equivalent of atmospheric CO is incorporated into the cis-position of the transition metal complex during the reaction sequence. 41 Any plausible mechanism for chain formation must account for these isotopic labelling experiments.
The DFT calculations were expanded to consider the formation of the key intermediate 2-Mn along with the interconversion of 2-Mn / 3-Mn / 4-Mn (Fig. 4). The C 1 to C 4 chain-growth mechanism involves alternating reduction and oxidation events at the growing carbon, by [Al] and CO respectively. Stationary point geometry optimisation and frequency analysis were performed using the uB97x functional. 49 Single point energies were calculated using the M06L 50 functional with dispersion (D3) 51 and solvent (PCM, benzene) 52 corrections. The M06L functional has been previously benchmarked to provide good agreement with experimentally derived thermodynamic parameters on closely related systems. 53 The calculated pathway begins with nucleophilic attack of [Al] onto a C 2^O2 unit bound to the transition-metal centre, forming Int-1. 54,55 A single transition state connects the starting materials, 1-Mn and [Al], with Int-1. The aluminium centre is bound to both the C 2 and O 2 atoms in Int-1 to form a threemembered ring. Hence, this step could more precisely be described as a (2 + 1) cycloaddition from the perspective of [Al]. NBO analysis of Int-1 shows that 2e À have been transferred from [Al] to the CO unit; both the natural charges on C 2 (À0.11) and O 2 (À0.85) are lower than that of free CO (+0.69 and À0.52 respectively). Donation of the lone-pair of [Al] into the CO p* orbital is a key interaction in TS-1.
Subsequent insertion of exogeneous CO into the Al-C bond of Int-1 is calculated to be facile. This occurs via a single transition state, TS-2, to form the acyl intermediate Int-2. This step of the mechanism is the rst carbon-carbon bond forming event, but it constructs the C 2 -C 3 bond of the chain. The NPA charge analysis shows that insertion of C 3^O3 into the Al-C 2 bond of Int-1 is accompanied by partial oxidation of the carbon chain; both carbon atoms of Int-2 have positive NPA charges (C 2 +0.31; C 3 +0.02). There is literature precedent for the insertion of CO into three-membered aluminocyclopropanes 56 and aluminocyclopropenes. 57 Prior calculations from our group have shown that these reactions are reminiscent of migratory insertion of CO at transition metal centres but have a propensity to occur in a concerted manner. 57 The reaction of a second equivalent of [Al] with Int-2 is calculated to be thermodynamically downhill ðDG 298 K ¼ À48:2 kcal mol À1 Þ. The NPA charge of the C 2 and C 3 carbon in 2-Mn (+0.15, À0.48) are lower than those in Int-2 (+0.31, +0.02) reecting the reduction of the carbon chain in this step. Although a transition state could not be located for this intermolecular process, the step is entirely feasible and leads directly to the key intermediate 2-Mn, which has been isolated and crystallographically characterised.
The structure of 2-Mn is preorganised for the migratory insertion of the C 1^O1 isocarbonyl ligand into the Mn-C 2 bond. The C 1 -C 2 bond along with the [6,4] fused ring system of 3-Mn is formed in this reaction step. A single transition state, TS-3, connects 2-Mn with Int-3. The reaction barrier from 2-Mn to TS-3 (DG ‡ 298 K ¼ 26.0 kcal mol À1 ) is consistent with this being a slow step, and combined with the relative thermodynamic stability of 2-Mn, predicts this intermediate to be an isolable species. For comparison, the equivalent step in the pathway from 1-W is calculated to occur via a lower in energy transition state (see ESI †) and all experimental attempts to detect 2-W, the analogue of 2-Mn, failed. While it is challenging to comment on the origin of this effect, based on the broader scheme of reactivity (Scheme 1) it does not appear to be steric in origin. One possible explanation for the different reactivity of 2-Mn and 2-W is that the migratory insertion step occurs more readily when the C 2 position is more nucleophilic (see ESI, Fig. S18 †). Migratory insertion reactions at transition metal centres are commonly reversible, with trapping of the products by an exogeneous ligand providing a thermodynamic driving force for the forward reaction. 58 In line with this statement, the formation of Int-3 from 2-Mn is calculated to be mildly endergonic ðDG 298 K ¼ þ3:7 kcal mol À1 Þ and likely reversible. Coordination of CO to Int-3 results in the formation of 3-Mn and is downhill from 2-Mn ðDG 298 K ¼ À41:1 kcal mol À1 Þ. This coordination event explains the incorporation of exogenous CO into the cis-position of the transition metal fragment observed in isotopic labelling studies.
Insertion of a further equivalent of exogeneous C 4^O4 into the Mn-C 3 bond of 3-Mn is calculated to occur by TS-4 and results in the construction of the C 3 -C 4 bond and formation of 4-Mn. The activation barrier and thermodynamics for this step (DG ‡ 298 K ¼ 21.2 kcal mol À1 ; DG 298 K ¼ À13:5 kcal mol À1 ) again suggest that 3-Mn should be isolable. The calculations are also consistent with the reverse reaction, deinsertion of C 4^O4 from 4-Mn to form 3-Mn, being feasible under forcing conditions (DG ‡ 298 K ¼ 34.7 kcal mol À1 ) as observed experimentally. 59 Selectivity A 9 : 1 mixture of 3-Mn : 3-Mn 0 is observed to form upon reaction of 2-Mn with CO. Hence, the pathway presented in Fig. 4 cannot be the complete mechanistic picture. Based on the experimental ndings, the mechanism towards the formation of 3-Mn 0 from 2-Mn was also investigated using DFT calculations (Fig. 5). Isomerisation of 2-Mn was considered. A lowenergy transition state TS-5 connects 2-Mn with Int-4. This transition state involves migration of the aluminium fragment that is chelated by the isocarbonyl and C 2 carbon chain from the O 2 toward the C 2 atom. The potential energy surface describing this reaction step is calculated to be reasonably at and Int-4 is only 1.2 kcal mol À1 lower in energy than its associated TS-5. The low barrier to the reverse reaction suggests that the isomerisation of 2-Mn is reversible and by itself is not selectivity determining. Variable temperature 1 H NMR spectroscopy on d 8toluene solution of 2-Mn across a À60 to +80 C range revealed a uxional process involving at least one of the b-diketiminate ligands bound to aluminium. While the complexity of the data does not allow this process to be assigned to a specic molecular reorganisation, the observation is consistent with 2-Mn being unstable toward either isomerisation or conformational changes below 20 C. Subsequent rearrangement of Int-4 to Int-5 is calculated to be non-reversible and occurs via a single transition state TS-6. Like TS-5, TS-6 involves a migration of an aluminium fragment but this time from the C 2 to the O 2 atom of the carbon chain. The overall process from 2-Mn to Int-5 reorganises the molecular topology converting a [7,4] fused ring system into a [6,5] analogue. Insertion of the C 1^O1 isocarbonyl ligand into the Mn-C 2 bond of Int-5 results in the formation of Int-6, in which the [5,5] fused ring system structure of the product is set. Subsequent addition of CO forms 3-Mn 0 .
The calculations suggest that the isolated intermediate 2-Mn provides a point of mechanistic divergence. The formation of 3-Mn and 3-Mn 0 are determined by TS-3 (DG ‡ 298 K ¼ 26.0 kcal mol À1 ) and TS-6 (DG ‡ 298 K ¼ 31.2 kcal mol À1 ) respectively. Although the relative energy barriers are consistent with the prediction of 3-Mn as the major product, closer comparison of the selectivity determining step reveals an energy difference (DDG ‡ 298 K ¼ +5.2 kcal mol À1 ) beyond that predicted for a 9 : 1 mixture of products likely highlighting the limitations in accuracy of the DFT model. 60

Conclusions
In summary, we have prepared a series of heterometallic products incorporating C 2 , C 3 or C 4 chains derived from reaction of group 6-9 metal carbonyl complexes, CO, and an aluminium(I) reductant. These reactions create rare polymetallic complexes in which a {C n O n } 4À (n ¼ 2-4) carbon chain is supported by both transition metal and main group centres. Electronic structures of a manganese series were interrogated by DFT calculations, showing that C 3 and C 4 products are best described as metallocarbene complexes of the transition metal. These calculations also suggest that the {C n O n } 4À chain contains an alternating array of single and double bonds which is bonded to aluminium through polarised and largely ionic Al-O and Al-C bonds. Synthetic studies elucidated each of the steps in the reaction sequence C 2 / C 3 / C 4 , conrming the C 2 product as intermediate in chain growth and showing the step from C 3 to C 4 chains is potentially reversible.
Modication of the electronics at the transition metal inuences not only the relative rate of each chain growth step allowing but also the selectivity. Through modication of this component, the product distribution of the reaction can be tuned allowing access to different isomers of C 3 carbon chains which contain different topologies and different reactivity. Further calculations and isotopic labelling experiments were used to gain insight into the mechanism of chain growth. These studies suggest the reaction is initiated by an intermolecular attack of the nucleophilic aluminium(I) reagent at a transition metal carbonyl ligand. Chain growth then proceeds through a series of migratory insertion steps, occurring rst at the transition metal site, then at the main group site. The selectivity that controls the chain topology, is inuenced by the relative rates of migratory insertion and ligand reorganisation within the heterometallic framework.
In totality, this study provides unprecedented insight into the subtle mechanistic features (e.g. reversibility, selectivity, cooperative effects) involved in CO homologation in the presence of transition metal and main group centres. generous funding. We thank Imperial College London for the award of a President's Scholarship (RYK). We also thank the EPSRC for project funding (EP/S036628/1).