For molecular aggregates, liquids, and molecules at interfaces, a subsystem formulation of DFT can be a better descriptor of the underlying physics than regular DFT of the supersystem.In recent years, the Frozen Density Embedding (FDE) formulation of subsystem DFT has been implemented in many quantum chemistry packages. However, none of the existing implementations are capable of tackling truly periodic systems, such as metal surfaces.We fill this gap by introducing a new implementation of the FDE method in the Quantum-Espresso suite of softwares.Highlights of our software include  the capability of modelling metallic subsystems, an accurate calculation of nuclear gradients to allow for geometry optimizations and molecular dynamics simulations, the constraining of charges and spin densities to allow simulation of ionic/radical subsystems, and the employment of non-local Kinetic Energy functionals as non-additive functionals.Software speed-up measures have also been devised and implemented. These include a state-of-the-art machine-learning based dynamic load balancing of the CPU distribution among the subsystems, and a subsystem-based K-point sampling.We benchmark our method against molecular dynamics simulations of bulk liquids and solvated radicals, adsorption of molecules and molecular multi-layers on metals, as well as ionic organic crystals.

Massively parallel subsystem DFT implementation for molecular and periodic systems

248th National Meeting (2014)

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