Support for the UK Car-Parrinello Consortium

This grant enabled a large network of researchers in 19 different physics, chemistry and materials science departments to do computational research using the national academic supercomputer.

As PI, Probert had overall management of the consortium and was also involved in several specific projects, including "high pressure crystal phases" and "first principles material design". Both of these objectives were met, with high quality publications in both areas. The Genetic Algorithm for "first principles material design" was developed further, to improve its robustness and remove a problem that had been identified, that of 'stagnation' where there is a loss of genetic diversity and all the search candidates converge on a common solution. The "high pressure crystal phases" was developed by applying the structure search ideas to the high pressure phases of iron, with relevance to Earth-type planetary interiors. Using a combination of techniques, we confirmed earlier predictions that there would be a sequence of phase transitions with increasing pressure, going from BCC -> HCP -> FCC -> HCP -> BCC at the highest pressures studied.

Dr Phil Hasnip (as PDRA) was employed for the duration of the grant. One of the stated aims for the PDRA was that as well as supporting the other nodes of UKCP and doing general code developments, they were to develop their own research program. This has worked very well, with Hasnip forming a very productive collaboration with an experimental physics group in York to study the atomic processes in the controlled growth of various technologically important materials (e.g. MgO as highlighted in one of the papers in the 'key publications' section). Hasnip has also contributed significantly to improving the widely-used CASTEP code, with new algorithms for finding the ground state electronic structure of a material, which are faster and scale better to more processor cores on HECToR. Probert and Hasnip have also contributed to a number of other CASTEP improvements, including new spectroscopic capabilities, which has also resulted in a highly cited publication (also in 'key publications' section).

The network has also underwritten the cost of 4 different workshops and 2 conferences, as highlighted in the 'other research outputs' section. This enabled the conference organizers to provide bursaries and speaker expenses, which have enabled a wide variety of students and early-career researchers to participate. The workshops have focused on spreading best practice in using computational modelling to different communities, whilst the conferences have focused on the application of computational modelling to the semiconductor physics community. There was also industrial sponsorship of both conferences, and industrial attendees were welcome at all events.

Probert has also developed a fruitful collaboration with Michaelides which extended the ab initio path integral molecular dynamics developments of Probert from the previous network grant, and applies it to the study of water on metal surfaces showing how subtle quantum nuclear effects change the process of wetting.

This network has recently been renewed for another 4 years, and continues to play in important part in keeping the UK at the forefront of materials modelling.