What UKCP Does

UKCP develops the methods and software to perform computer simulations of a wide range of chemical and material systems. These methods may be applied to a vast range of chemical, physical and materials-related problems, and a selection of recent studies is given below:

  • Beating the Stoner criterion using molecular interfaces
    Two common metals that are not magnetic ― copper and manganese ― can be transformed into magnets; a surprising effect that involves combining thin films of the metals with fullerene molecules. The mechanism may allow for the design of magnetic materials and interfaces using abundant, non-­‐toxic components such as organic semiconductors, with new possibilities for electronic, power or computing applications.
  • Ligand Binding and Positional Disorder in Pentlandite
    Density functional theory, in conjunction with a cluster expansion model, has been used to study the structure and stability of the positionally disordered iron-­‐nickel sulfide mineral pentlandite (Pn), (Fe, Ni)9S8, with results indicating heterogeneous nearest neighbour metal contacts are more energetically favourable than homogeneous contacts. We also addressed the binding of ethyl xanthate, water and hydroxide to the [111] Pn surface to better understand the mode of action of flotation agents for the recovery of metals by industrial mining processes. In order to model anionic ligands bound to a periodic boundary condition surface we propose a correction scheme derived from the surface work function to remove the additional charge introduced by the ligand. The modelling study was extended to other mineral surfaces and other industrial collector ligands, and a hierarchy of ligand-­‐mineral surface binding emerged, which was subsequently verified by experiment based on electrochemistry measurements.
  • Electron-­‐phonon coupling and anharmonic vibrations in materials
    Vibrations of atoms couple to all observables in condensed matter. We have developed improved vibrational self‐consistent field methods which provide an accurate treatment of anharmonic vibrations and coupling of vibrations to observable quantities such as the bandstructure of a crystalline solid.

UKCP has also regularly funded a number of Code Development Internships, to encourage students into scientific research with computer programming. Some recent examples include:

  • Optimising density mixing
    This project funded Alex Christison at the University of York to devise and implement an optimisation method for the popular "density mixing" method in CASTEP. This method avoids the possible divergent behaviour of standard iterative approaches, leading to an optimisation method which is both robust and fast.
  • The “Sample with Machine learning then Ab initio Refine Technique” (SMART) way to new energy materials
    This project funded Weronika Wiesiolek at the University of Birmingham to develop the ML-tools software package, to make it easy to train a machine learning (ML) potential with CASTEP for use in AIRSS.
  • Implementing the Iterative Hirshfeld method in CASTEP
    This project funded Sam Butler at the University of York to implement the Iterative Hirshfeld method in CASTEP which can significantly improve the quality of the Hirshfeld charge partitioning scheme, especially for polarisable systems, with consequent improvements to semi-empirical dispersion correction schemes that are built upon Hirshfeld charges.
  • Fast and Efficient Structure Factors and XRD
    This project funded Benjamin Shi at the University of Oxford to develop a new method to calculate structure factors and hence accurate x-ray diffraction results. The method was then implemented in the CASTEP code. The result is comparable in accuracy to all-electron codes, but much faster.
  • Multi-Objective Genetic Algorithm for Crystal Structure Prediction
    This project funded Scott Donaldson at the University of York to implement the MOGA approach in the CASTEP_GA code. This enables simultaneous optimization of multiple functions as part of a crystal structure search, e.g. minimum enthalpy and maximum bulk modulus, for a given set of elemental ingredients.
  • pp_hybrid mode in CASTEP
    This project funded Tamas Stenczel at the University of Cambridge to implement on-the-fly machine learning acceleration of molecular dynamics calculations in CASTEP, by linking CASTEP to the Gaussian Approximation Potential method in the QUIP library. The result is DFT accuracy but with a dramatic speed increase.