Computational chemistry - research interests

Head of research group: Jan M.L. Martin (a.k.a. Gershom)

Benchmark ab initio calculations: the right result for the right reason?

A long-standing research interest of our group consists in trying to push ab initio methods to their absolute limit in terms of accuracy. This involves such issues as: basis set convergence and extrapolation to the 1-particle basis set limit,  correction for higher-order correlation effects, effect of inner-shell correlation, scalar relativistic effects, and (for spectroscopic properties) the study of rotational-vibrational anharmonicity. We are presently able, for small polyatomic molecules, to predict molecular binding energies to an acccuracy of 1 kJ/mol (0.24 kcal/mol) or better, geometries to within less than 0.001 Å, and vibrational band origins to within about 1 cm-1.

Our development efforts have culminated in two more-or-less "black box" computational thermochemistry protocols knows as W1 and W2 theory. (Click here for more details.)

Aside from applications which require this level of accuracy in their own right (e.g. determination of fundamental thermochemical properties), computed data of this quality are very helpful in assessing the accuracy of less CPU-intensive methods and/or to assist in the parametrization of methods that involve empirical parameters (e.g. a number of recent DFT methods).

Some representative recent publications on this topic:

Computational modeling of homogenous catalysis: methodology and applications

In collaboration with the experimental group of Prof. David Milstein, we are involved in the computational modeling of homogenous catalysis by late transition metal complexes. We apply density functional techniques to elucidate the mechanism of catalytic reactions, compare barrier heights of competing reaction pathways, characterize reaction intermediates, compare relative stabilities of various produtcs, predict spectra in order to facilitate experimental identification, and otherwise assist the experimental effort.

Examples of specific subjects include the mechanism of selective C-C vs. C-H activation (and conversely) by means of `pincer ligand' catalysts, activation of strong CN and NN bonds by same, the mechanism of the Heck coupling, and cooperativity in hydrogenation of acetone by catalytic monolayers.

The prediction of reaction barrier heights is a known weakness of present-day density functional methods. Specialized 'kinetics' functionals yield more reliable barrier heights, at the expense of severely degraded performance for all other properties. We developed a novel hybrid meta-GGA functional termed BMK (Boese-Martin for Kinetics), which has hybrid GGA performance overall but yields reliable barrier heights as well.

As a side-project, we have become involved (with Dr. Milko E. van der Boom) in computational studies of an intriguing class of molecules known as metallabenzenes.

Some representative recent publications on this topic:

Other subjects

As by-products of the two above principal subjects, we are involved in some ancillary research topics such as:

Some representative recent publications on these topics:

A full publication list is available on our website.

 |Back to Computational Chemistry home page|