Simulation of low index rutile surfaces with a transferable variable-charge Ti–O interatomic potential and comparison with ab initio results

The ability of a recently derived transferable variable-charge Ti–O interatomic potential to simulate low-coordination environments has been tested by application to the (1 1 0) and (0 0 1) rutile surfaces. Hartree–Fock and density functional theory calculations of the (1 1 0) rutile surface were also performed. Comparison of the variable-charge results with data from first principles and two other standard forcefield calculations, along with experimental data, reveals that the variable-charge model appears to be inadequate for surface simulations. The atomic relaxations predicted by all the forcefields are not only mutually inconsistent but are also in disagreement with ab initio and experimental data. The quantum mechanical treatments, on the other hand, yield relaxation data in agreement with each other, but different to experimental data on the (1 1 0) surface. The surface formation energies obtained for the relaxed (1 1 0), (1 0 0), and (0 0 1) surfaces are also not in the right order when computed with the variable-charge model. The low ionic charges predicted by the model appear to be a major reason for this failure. Inclusion of surface data in the parameterisation of the model is likely to improve the transferability to low-coordination environments.