Methylidynetricobaltnonacarbonyl

Methylidynetricobaltnonacarbonyl is an organometallic cobalt cluster with the chemical formula Co3(CO)9CH that contains a metal carbonyl core with the methylidyne ligand, first discovered in the late 1950s. A variety of substituents can be added to the methylidyne group to form derivatives of the parent compound that have unique spectroscopic properties and reactivity. This page will explore the discovery and synthesis of methylidynetricobaltnonacarbonyl, the structure and bonding of the parent compound, as well as some examples reactivity and catalysis with the cluster.

Discovery and Synthesis
Methylidynetricobaltnonacarbonyl was first discovered in the late 1950s by Markby and Wender, though the absolute structure of the molecule could not be determined. The group obtained Nuclear Magnetic Resonance (NMR), Infrared Spectroscopy (IR), and dipole moment data on a new trinuclear cobalt complex formed from reacting dicobaltoctacarbonyl with acetylenes, but were unable to determine the structure and hoped that X-ray crystallography would provide the evidence needed to determine the structure of the new compound. In 1962, independent groups from Italy and Germany reported the synthesis and properties of methylidynetricobaltnonacarbonyl as well as several derivatives. The structure has been unambiguously determined via X-ray crystallography, demonstrating idealized C3v symmetry with three equivalent six-coordinate cobalt atoms.

The synthetic procedure developed by Bor and coworkers relied on reacting the [Co(CO)4]- anion with chloroform, bromoform, or iodoform in a solution of acetone or THF to afford the desired product as dark violet crystals in ~20% yield with chloroform, ~18% yield with bromoform, and ~1-2% yield with iodoform. The chloride derivative Co3(CO)9CCl was obtained by reacting CCl4 with the [Co(CO)4]- anion to afford shiny, lilac crystals. Several other functional groups could be installed, including C6H5 and CO2CH3, all of which afforded dark violet crystals. The compounds demonstrated extraordinary stability in air and were dissolved without decomposition in organic solvents. Treatment of the parent compound with 100 atm CO at 160 °C resulted in the destruction of the complex and the formation of dicobaltoctacarbonyl: Co2(CO)8.

Structure and Bonding
The Co-Co bond lengths, as determined by X-ray crystallography are ~2.47 Å and the Co-C bond distances are ~1.88 Å. These bond lengths do not change significantly depending on the substituent on the carbon group. The bond lengths confirm the presence of C-Co and Co-Co single bonds, confirming the structure drawn above. The Co-C-Co angle is approximately 80°, indicating a significant bending of the carbon tetrahedral structure, and the ligands are arranged to provide maximum cobalt-carbon interaction. The IR spectra demonstrate four absorption bands of varying intensities in the region of terminal C-O groups: between 2111-2101, 2066-2054, 2047-2038, and 2034-2018. The electronegative character of the chlorine atom in Co3(CO)9CCl causes lower C-O frequencies than Co3(CO)9CH, indicating that electron density is being pulled from the metal atoms to the chlorine, causing electrons to shift from the carbonyl groups to the metal atoms, giving them more double-bond character and decreasing their frequency. This effect can be seen with other electron withdrawing substituents, as well as the inverse with electron-donating substituents. This suggests that the cluster behaves as an electron sink capable of interacting with the substituent on carbon via pi bonds, which is a theory that is also supported by the chemical shift of the carbon on the substituent in 13C NMR, as the chemical shift of the apical carbon increases with more electron withdrawing substituents. The mass spectra demonstrate parent molecular ions (Co3(CO)9CH .+) in good abundance, which demonstrates the high stability of the Co3(CO)9 clusters towards oxidation. Initial loss of CO, which is characteristic of most metal-carbonyls, though rupture of the metal-metal bond also often occurs. The d9-ML3 fragment Co(CO)3 fragment is isolobal with methylidyne, CH. Though there are differences in the a1 and e energies of the two fragments, there are a series of organometallic tetrahedranes of the C4, CoC3, Co2C2, Co3C, and Co4 variety, where a variety of ligands and functional groups can be installed on the cobalt and carbon groups, respectively. Hoffmann demonstrated that a substituent with an acceptor orbital of appropriate symmetry could interact with the highest energy orbital to withdraw electron density from the filled e set on the apex carbon. Seyferth demonstrated that the carbonium ion had the correct symmetry to accept electron density from the apex carbon via s-p stabilization. They explained that apex carbon stabilize the carbonium ion via s-p conjugation. Early reactivity studies on methylidynetricobaltnoncarbonyl were conducted by Cetini et al. in the early 1960s by conducting 14CO labelling studies on the kinetics of carbonyl ligand exchange. The group concluded the exchange rate of the carbonyl ligands decreases according to the order F > Cl > Br > H, which they attributed to the electronegativity of the atom bound to the methylidyne carbon. Additionally, they demonstrated that the nine carbonyl groups are not kinetically equivalent, and that only 3 CO groups exchange within the 35 – 55 °C temperature range studied, which agrees with the calculated frontier molecular orbitals discussed above. In the case with the COOCH3 group, the exchange occurs for only one CO group which they attributed to the more electron deficient nature of the complex.

Reactivity
MIT’s Dietmar Seyferth explored the reactivity of methylidynetricobaltnonacarbonyls beginning the early 1970s, and his work uncovered many of the interesting properties of the metal carbonyl cluster. In 1970, his group demonstrated that a variety of organomercury compounds of the R2Hg and RHgX type are capable of alkylating methylidynetricobaltnonacarbonyl with good yields and facile synthetic procedures. The group observed the side product Hg[Co(CO)4]2 was forming during the reaction and hypothesized that if the decomposition of the starting material was due to the reversible loss of CO ligand, then conducting these reactions under a CO atmosphere could improve yields. This was observed, and yields of the reaction increased to >90%. In 1973, Seyferth and coworkers demonstrated that adding azobisisobutryonitrile (AIBN), a common reagent for the initiation of radical reactions, to a solution of Co3(CO)9CH and allyl acetate afforded a red crystalline solid that they believed formed via a radical mechanism. Around the same time, Czájlik and coworkers were exploring how metal carbonyls initiate radical polymerization, as they hoped that the tricobalt complex would be more active than Co2(CO)8 and more stable than Mo(CO)6. The derivatives of Co3(CO)9CX demonstrated different activity depending on the structure of X, which aligns with their hypothesis that the first step of initiation would be loss of a CO ligand. The order of initiator activity was found to be X = Cl > H > Br > Ph > F > i-Pr > C2F5, which aligns with the trend seen for CO ligand exchange studies. In 1974, Seyferth and coworkers published their mechanistic insights into the reaction described above. The group highlighted that though the reaction may proceed via the homolytic decomposition of some Co3(CO)9CH molecules, the presence of a radical catalyst or a radical inhibitor did not change the rate of the reaction. They also proposed that the electrophilic cleavage of the mercury-carbon bond could be initiated by Co3(CO)9CH, but the group couldn’t provide evidence for this theory either, so no conclusions could be drawn about the absolute mechanism for this reaction.

In 1976, Reed and coworkers sought to explore the electrochemical properties of these cobalt carbonyl clusters to discover how the nature of the X group in Co3(CO)9CX changes the anodic and cathodic properties of the cluster. The group observed a reversible, one-electron reduction in the range of -0.7 to -0.9V versus the saturated calomel electrode. They observed that clusters with more electronegative substituents are reduced around -0.8V while substituents with more electron donating groups are reduced around -0.9V, which is almost identical to the pattern observed for the reduction of ferrocene with the same substituents, demonstrating that substituents have the same effect on the HOMO and LUMO of ferrocene and the cobalt cluster. Subsequent oxidation of the reduced species occurred irreversibly around +1.5V, which was then irreversibly reduced at -1.0V, leading to another decomposition product that oxidized irreversibly at 0.4V.

Several years later, in 1998, Sugihara, et al. demonstrated that the Pauson-Khand reaction could be catalyzed by the methylidynetricobaltnonacarbonyl cluster. The Pauson-Khand reaction allows for the cyclic cotrimerization of an alkyne, an alkene, and carbon monoxide via a [2+2+1] cycloaddition and has been widely utilized for the synthesis of cyclopentenones for application in natural products. Methylidynetricobaltnonacarbonyl is more air-stable than the parent dicobalt octacarbonyl, making it a more attractive catalyst. The clusters demonstrated no need for additives such as trimethylphosphite, as is necessary with the dicobalt octacarbonyl, and the best results were obtained with the parent methylidyne cluster. More recently, Nordlander et al. have been exploring using the methylidynetricobaltnonacarbonyl derivatives as a precatalyst for the asymmetric intramolecular Pauson-Khand reaction. The group treated the parent Co3(CO)9CH with the chiral Josiphos diphosphines to form the desired asymmetric precatalyst. The group determined that the clusters acted as pre-catalysts only as no clusteres were recovered from the reaction mixture after catalysis. Despite the unique approach, the clusters were found to be only moderately effective for this reaction, and are inferior compared to other metal clusters with NORPHOS or Me-DuPHOS ligands, which gave higher yields and fewer side products.