Ligand Effects on the Regioselectivity of Rhodium-Catalyzed Hydroformylation: Density Functional Calculations Illuminate the Role of Long-Range Noncovalent Interactions

Density functional theory calculations
have been performed to gain
insight into the origin of ligand effects in rhodium (Rh)-catalyzed
hydroformylation of olefins. In particular, the olefin insertion step
of the Wilkinson catalytic cycle, which is commonly invoked as the
regioselectivity-determining step, has been examined by considering
a large variety of density functionals (e.g., B3LYP, M06-L); a range
of substrates, including simple terminal (e.g., hexene, octene), heteroatom-containing
(e.g., vinyl acetate), and aromatic-substituted (e.g., styrene) alkenes,
and different ligand structures (e.g., monodentate PPh3 ligands and bidentate ligands such as DIOP, DIPHOS). The calculations
indicate that the M06-L functional reproduces the experimental regioselectivities
with a reasonable degree of accuracy, while the commonly employed
B3LYP functional fails to do so when the equatorial–equatorial
arrangement of phosphine ligands around the Rh center is considered.
The different behavior of the two functionals is attributed to the
fact that the transition states leading to the Rh–alkyl intermediates
along the pathways to isomeric aldehydes are stabilized by the medium-range
correlation containing π–π (ligand–ligand)
and π–CH (ligand–substrate) interactions that
cannot be handled properly by the B3LYP functional due to its inability
to describe nonlocal interactions. This conclusion is further validated
using the B3LYP functional with Grimme’s empirical dispersion
correction term: i.e., B3LYP-D3. The calculations also suggest that
transition states leading to the linear Rh–alkyl intermediates
are selectively stabilized by these noncovalent interactions, which
gives rise to the high regioselectivities. In the cases of heteroatom-
or aromatic-substituted olefins, substrate electronic effects determine
the regioselectivity; however, these calculations suggest that the
π–π and π–CH interactions also make
an appreciable contribution. Overall, these computations show that
the steric crowding-induced ligand–ligand and ligand–substrate
interactions, but not intraligand interactions, influence the regioselectivity
in Rh-catalyzed hydroformylation when the phosphine ligands are present
in an equatorial–equatorial configuration in the Rh catalyst.