Center for Computational Molecular Science and Technology Georgia Institute of Technology Center for Computational Molecular Science and Technology School of Chemistry and Biochemistry

Development and Application of High-accuracy Methods for Bond-breaking Reactions and Non-bonded Interactions

Principal Investigator: C. David Sherrill

Dr. Sherrill’s research group specializes in the development and application of high-accuracy, ab initio electronic structure models, including their use in calibrating DFT and lower-level theoretical approaches. Bond-breaking reactions and weak, dispersion-dominated noncovalent interactions have been the group’s two primary foci, and both are key topics in energy-related research. With respect to bond-breaking reactions, we have studied the behavior of a wide variety of theoretical models for unimolecular dissociation reactions. We are also examining the case of hydrogen transfer reactions in organic systems.

With respect to noncovalent interactions, we have developed a protocol employing MP2 with large correlation-consistent basis sets, together with a correlation correction computed using CCSD(T), to accurately determine the interaction energies of small van der Waals clusters such as the benzene dimer, benzene trimer, benzene-methane etc. These systems are critically important as prototypes of the kinds of noncovalent interactions which occur in organic crystals, in biopolymers, and in drug recognition. The accurate description of noncovalent interactions is crucial to capture the physics of intermolecu lar interactions on aromatic organic materials and the adsorption of small molecules on metal nanoclusters and solid oxide surfaces.

For efficient yet reliable computations on larger systems, improved approximations are needed. Using our benchmark quantum data for prototype systems, we evaluate the reliability of some of the more promising new DFT approaches meant to include London dispersion forces. Our potential energy curves are particularly valuable for this examination, because often the dispersion-corrected DFT meth-
ods are compared only against high-quality results for potential energy minima, and hence they may significantly over- or under-bind at longer intermolecular separations (summing to a large overall error in extended systems). We  also pursue our own re-parametrization of DFT and polarizable force fields against our benchmark data. We will further explore our recently-introduced spin-component-scaled CCSD method, which appears to be as accurate as CCSD(T) for noncovalent interactions.

Research highlights:
  • Computations of Noncovalent π Interactions, C. D. Sherrill, Reviews in Computational Chemistry, Vol. 26, 1-38 [2009]
  • Recent papers on noncovalent interactions:
  • Assessment of Standard Force Field Models Against High-Quality Ab Initio Potential Curves for Prototypes of π-π, CH/π, and SH/π Interactions, C. D. Sherrill, B. G. Sumpter, M. O. Sinnokrot, M. S. Marshall, E. G. Hohenstein, R. C. Walker, and I. R. Gould, Journal of Computational Chemistry, 30, 2187-2193 [2009]