Embedded mean-field theory: chemical simulation in complex environments

Density functional theory (DFT) is now widely used in many branches of chemistry and related disciplines, both in industry and academia. It provides a good level of accuracy at low computational cost, enabling researchers to optimize structures, study mechanism and compute semiquantitative energetics for a huge range of processes. Realistic modelling of more complex systems - such as catalysis on nanoparticle surfaces, electrolyte decomposition in batteries, or reactivity in biological systems - demands a combination of accuracy and extensive thermodynamic sampling of nuclear configurations. Current methods do not deliver this.

Multiscale modelling has produced huge gains in our ability to model complex systems. Most notably the QM/MM approach - which combines a quantum method in one region with molecular mechanics in the environment - has been widely celebrated (2013 Nobel Prize for Chemistry) and is very widely used.

But there are two primary reasons why it is essential to move beyond the QM/MM paradigm. First the interface with a nonpolarizable, point-charge model can give rise to spurious effects that can only typically be mitigated by increasing the size of the active subsystem. Second, there is no quantum mechanical interaction between subsystems, so for example there are no number fluctuations between the subsystems, and this is critical for processes in electrochemistry, or on metal or nanoparticle surfaces.

We will develop a quantum embedding scheme in which a complex system is described using the highly efficient density functional tight binding method, with a small, important subsystem described by more accurate DFT treatments. The coupling between subsystems will be treated quantum mechanically, with a mixed quantum state in each subsystem, allowing, for example, for electrons to flow between subsystems.

This project has been conceived with industrial impact as a key motivation, so we will liaise with project partners in Toyota and BASF to ensure that this method is efficiently transferred to industrial settings, maximizing impact from the project.

The proposal contains a new idea crafted specifically for impact across a range of both industrial and academic areas, and will form part of our effort to develop a new software code for simulation of chemistry in complex environments. The implications for UK research, including for UK industry, could be profound.

Industries in a wide range of sectors employ computational modelling, including at an atomistic level using quantum methods, to develop new products and new processes, and to help tighten the focus of more expensive experimental research. In consultation with researchers from two highly successful, but very different, industries (Toyota Central R&D and BASF) we have developed a research programme which aims to address a specific challenge, namely improving the statistical sampling for finite temperature processes in complex systems, whilst maintaining the accuracy of KS-DFT. This consultation process is on-going (for example, Ryoji Asahi, Department Manager for Sustainable Energy and Environment for Toyota will visit our lab in August this year), and will be broadened to include representatives of other sectors.

Our plans for maximizing impact focus on:

(1) Impact through development of sustainable software
(2) Impact through proper management of intellectual property and revenue generation from software
(3) Impact through end-user outreach

Details of each strand can be found in the case for support.