Until recently, the size of water systems that could be simulated using ab initio molecular dynamics (AIMD) methods over moderate time scales was limited by the high cost of DFT calculations to system sizes which displayed finite-size effects. The development of second-generation Car-Parrinello molecular dynamics has allowed AIMD water systems containing hundreds of molecules to be simulated over nanosecond timescales. Modern density functionals (such as the PBE XC functional) accurately reproduce a number of static and dynamical properties of water. However, they fail in accurately reproducing structural properties such as the radial distribution function. Most XC functionals predict water to be overstructured. This causes glassy dynamics in MD simulations. This overstructuring is widely attributed to the lack of accuracy in dispersion corrections to XC functionals and the exclusion of nuclear quantum effects from simulation protocols. A recent publication by Kremer et al showed that the inclusion of nuclear quantum effects served only to marginally reduce overstructuring. We apply a novel dispersion-corrected functional, developed by Narbe Mardirossian of the Martin Head-Gordon group at UC Berkeley, for chemical accuracy and second-generation Car-Parrinello dynamics for enhanced sampling to develop an accurate, computationally tractable AIMD model for liquid water.

Applying new dispersion-corrected functionals to the ab initio simulation of liquid water

248th National Meeting (2014)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.