Beyond Classical Molecular Dynamics: Developing DL_POLY

DL_POLY is a CCP5 flagship UK code, one of a handful of international codes to carry out classical molecular dynamics simulations. These are a powerful form of computational microscope, a real Laplace's demon, in which, given a specified set of forces between atoms, one follows their changing positions and velocities by Newtonian mechanics. One can thus analyse and understand the behaviour of condensed matter at the atomic level. However, classical molecular dynamics has two major limitations which dominate the majority of community requests for new functionality. First, the use of classical force fields limits the accuracy of the model and restricts its applicability -- quantum effects, bond breaking and making and charge transfer are, for example, not included. Secondly, the timescales (the time over which the system can be followed) accessible via molecular dynamics are too short to study properly many phenomena and processes in chemistry, physics, materials science and biology where events occur only rarely. We aim to address both these methodological limitations. We shall implement in DL_POLY: (a) density functional tight binding (DFTB) to address the accuracy and quantum effects problem and (b) forward flux sampling (FFS) to enable simulations of rare events and thus address long timescale phenomena.

We shall involve members of the community both as advisers and early users right from the start of the project. A number of test applications are planned to demonstrate the power of the new software: i) examination of the effects of nanostructuring on the electronic structure of potential thermoelectrics and the link between local structure and conductivity in highly disordered bismuth oxides which are among the very best superionic conductors (DFTB) ii) heterogeneous nucleations, such as those promoted by nanoparticles and exploited in enhanced drug delivery by nanobubbles in cancer treatment (FFS). The new functionality will open up a myriad of new applications in areas such as functional materials (an EPSRC priority) -- e.g., energy materials, battery materials, nanomaterials, ferroelectrics, superconductors, thermoelectrics for use in information and communications technology, energy generation storage, transport, healthcare, defence and consumer goods. FFS implementation will enable users to tackle problems as diverse as crystal nucleation, protein folding, pore formation, protein translocation through pores, rare switching in magnetic nanostructures, isomerization, wetting of rough surfaces, droplet coalescence and flipping of genetic switches. The new software produced will be freely available to academics worldwide. The new features will assure that DL_POLY remains internationally competitive. Comprehensive manuals, documentation and training will be made available to both academia and industry.

Molecular dynamics is one of the most important simulation methods in modern science. We propose a major upgrade to the UK national molecular dynamics code DL_POLY - one of a very small number of such codes worldwide. Much enhanced functionality, proposed on the basis of the most common user requests, combined with excellent scaling, usability and analysis tools, will impact on all users of this general purpose program both in academia and in industry - chemists, condensed matter physicists, mineralogists, material scientists, engineers. There are approximately 3700 current licence holders of DL_POLY_4. These and new users will benefit from being able to tackle problems involving quantum effects in possibly highly disordered systems. Many restrictions of classical molecular dynamics will be removed - classical simulations are limited to small subsets of materials for which accurate parameterised force fields exist. Studies involving bond breaking, bond formation, charge transfer, and redox processes will be enabled - vital in problems involving functional materials in areas as diverse as information and communications technology, energy generation and storage, transport, healthcare, defence and consumer goods. Implementation of new methods for simulation of rare events will impact on a wide user base concerned with processes such as crystal nucleation, protein folding, pore formation, protein translocation through pores, switching in magnetic nanostructures, isomerization, wetting and droplet coalescence.

None of our developments are focussed on a small section of the community. There will be wide general interest in the new code and outputs; we will engage with the broader community (e.g., CCP_BioSim, the Materials Chemistry Consortium). DL_POLY is installed on a number of HPC facilities (ARCHER, Hartree Centre platforms, SCAR, NSCCS), so many users will be able to benefit quickly from the new DL_POLY.

Industrial users will benefit from our new "Simulation for the Experimentalist & Industrialist" course, and the Hartree Centre's business outreach events. Daresbury has established partnerships with IBM, Unilever, GSK and Syngenta in microemulsion dynamics , an area which will benefit considerably from the new code. Impact will be enhanced due to increasing interest by industry in direct coupling of computational fluid dynamics with atomistic molecular dynamics.

The materials industry contributes £200 billion to the UK p.a. 70% of technological innovations come directly from advances in materials - enabled and accelerated by simulation. Understanding condensed matter at the atomic level offers crucial insights into many products and processes. DL_POLY is vital in enabling this research by connecting UK academics and industry partners, and keeping industrial researchers in touch with the latest software and methodologies. Many internationally competitive companies (e.g. Sony, Samsung, Johnson Matthey, Syngenta) either use DL_POLY or collaborate with academic groups who do. Industry will benefit from personnel trained as scientific software developers.

UK industry increasingly sees computer simulation as a viable alternative to costly experimental screening and in prediction of new improved materials. Speed-to-market is critical, and product innovation key to success. CCP5 attacks head-on technical challenges to develop predictive models that are simple and affordable yet with potential to enable novel product design. Long-term societal benefits are through subsequent development of new materials that are cheaper, more environmentally friendly, and enhance well-being.

Outreach activities using simulation movies maximise impact for schools and the lay public who will be able to appreciate how the atomic scale determine the function of so many of the materials they use and aid understanding of key scientific issues underlying major societal challenges.