61st Annual Report on Research 2016

Since CO₂ can be activated by coordination to a transition metal, the preparation and study of isolable CO₂ complexes is of great interest. Side-on (η²) coordination is often observed for monomeric complexes of CO₂ which are most frequently prepared by substitution of a labile ligand. However, η²-CO₂ complexes can sometimes be prepared by the oxidation of a carbonyl ligand. Here we report that a range of η²-CO₂ molybdenum complexes can be prepared from Tp^RMo(NO)(L)(CO) by oxidation of the carbonyl ligand.

A carbonyl ligand can be thermally liberated from Tp^RMo(NO)(CO)₂  (Tp^R = Tp, tris(pyrazolyl)borate or Tp*, tris(3,5-dimethylpyrazolyl)borate) in the presence of a σ-donor ligand (L) to give Tp^RMo(NO)(L)(CO) (Scheme 1, step 1). Using this method, thirteen complexes featuring L = P(OMe)₃, PPh₃, PMe₃, 3-fluoropyridine (3-F-Py), pyridine (Py), 4-dimethylaminopyridine (4-DMAP), 1-methylimidazole (1-MeIm), 1,3-dimethylimidazol-2-ylidene (NHC), have been prepared and purified by column chromatography. Several of these complexes are air sensitive. In some cases, IR spectroscopy indicates that small quantities of η²-CO₂ complex are formed during heating if oxygen is not rigorously excluded. Consequently, the best results were obtained by performing thermolysis in a sealed pressure tube.

The range Tp^RMo(NO)(L)(CO) is constrained by the steric and electronic properties of L. For Tp, complexes featuring alkyl phosphines larger than PMe₃ have not been isolated, including PEt₃ and P(n-Bu)₃. For Tp*, the more sterically demanding ligand platform likely prevents isolation of PPh₃ and NHC complexes. On the other hand, the added electron-donating ability of Tp* allows a 3-fluoropyridine (3-F-Py) complex to be isolated for Tp*, while the analogous Tp version cannot be isolated. This result implies that π-acceptance may be more important that σ-donation for the 3-F-Py ligand; consequently, the more electron rich Tp* complex is isolable while the less electron rich Tp congener is not.

Oxidation of a carbonyl complex to form an η²-CO₂ complex can be accomplished using either a hydroperoxide or oxygen gas (Scheme 1, step 2). Most commonly a hydroperoxide gives the best yields of η²-CO₂ complex. Excess hydroperoxide is always required to ensure complete reaction of the carbonyl complex. Thus far, η²-CO₂ complexes are isolable for a range of complexes featuring σ-donors that are phosphines, pyridines, imidazoles, and N-heterocyclic carbenes (Table 1). Carbon dioxide complexes cannot be isolated for the more electron deficient phosphines (L= PPh₃, P(OMe)₃). For these complexes, cumene hydroperoxide does not oxidize the requisite complex. Stronger oxidants such as PhIO do react, but CO₂ complexes cannot be isolated or definitively identified in situ via spectroscopic methods. It seems likely that a CO₂ complex is formed under these conditions but that the resulting complex is not stable under such harshly oxidizing conditions. This could be due to the complex becoming too electron deficient to maintain coordination of CO₂. Alternately, the putative η²-CO₂ ligand could be further oxidized (to a carbonate, for example), or the metal complex could be oxidized upon liberation of CO₂ gas. Thus far, the least electron rich η²-CO₂ complex isolated is TpMo(NO)(PMe₃)(η²-CO₂) based on IR spectroscopy and oxidation potential as measured by cyclic voltammetry.

Figure 1 shows the IR stretching frequencies of all thirteen Tp^RMo(NO)(L)(CO) complexes displayed as a correlation of carbonyl versus nitrosyl stretching frequencies. As expected, since both stretches indicate the electron density at the metal center a linear trend is seen with a slope of one. Unsurprisingly, carbonyl and nitrosyl stretching frequencies are excellent indicators of the electron density of Tp^RMo(NO)(L)(CO) complexes. Complexes featuring ν_CO > 1893 cm^-1 and ν_NO > 1607 cm^-1 did not result in isolated η²-CO₂ complexes upon oxidation. However, all isolated Tp^RMo(NO)(L)(CO) complexes featuring ν_CO and ν_NO lower than these have led to isolable η²-CO₂ complexes upon oxidation.

The η²-CO₂ ligand is often quite labile; indeed thermal instability is one reason why carbon dioxide complexes have been difficult to isolate. Furthermore, molybdenum η²-CO₂ complexes have been shown to be prone to disproportionation. In contrast, 1-8 are thermally stable in solution at room temperature for days. For example, when TpMo(NO)(NHC)(η²-CO₂) 3, is dissolved in DMSO-d₆, heated at 50 °C and monitored by ¹H NMR spectroscopy, no decomposition is observed over 4 d.

Single-crystal X-ray crystallographic analysis has been successfully performed for a number of these new η²-CO₂ complexes.  For previously reported η²-CO₂ complexes of this type, such as TpW(NO)(PMe₃)(η²-CO₂), two types of positional disorder complicate the analysis of the diffraction data. First, the two η²-CO₂ coordination diastereomers co-crystalize making the coordinated C=O atoms of the CO₂ ligand difficult to model separately. Second, the two enantiomeric forms are also present and cause positional disorder between the uncoordinated CO of η²-CO₂ ligand and the NO ligand. Fortunately the X-ray diffraction data for the analogous molybdenum complexes that feature L = PMe₃ (7 and 8) are less problematic. Only one diastereomer, in which the uncoordinated oxygen of the η²-CO₂ ligand is proximal the PMe₃, is isolated as it is the thermodynamically-favored coordination diastereomer. Consequently, neither set of diffraction data features disorder due to coordination diastereomers. In addition, complex 7 crystalizes as a single enantiomeric form which eliminates such positional disorder from the diffraction data.

Figure 2 compares the average bond lengths and bond angle for the η²-CO₂ ligand for all three Tp^RM(NO)(PMe₃)(η²-CO₂) complexes. As expected, the C-O bonds lengthen as the complex becomes more electron rich. The more donating Tp* gives rise to longer CO₂ bonds, but the effect is not as dramatic as changing the metal from molybdenum to tungsten. The tungsten complex also shows a more acute O-C-O bond angle, a clear indication of stronger backbonding into the π* orbitals of the η²-CO₂ ligand.