How wide is the range of applicability of an ionic model for condensed phased structure, energetics and dynamics? An "ionic" system can be thought of as one whose properties can be reproduced by an interaction model based on discrete closed-shell ions which possess integer charges. There is no charge transfer or chemical bond formation. The central ingredient is a model for the interactions between the constituent particles. Computer simulations are simplest when the total interaction energy may be regarded as a sum of interactions between pairs of particles. Many materials of technological interest, such as MgCl₂, Al₂O₃ and TiO₂, are composed of elements of widely differing electronegativity and yet their structure and properties cannot be explained with pair potentials: many-body effects must be considered. The original question becomes practical because, if the many-body effects are contained within the ionic model, large scale simulation remain possible.

In the extended ionic model a many-body character is introduced into the potential by the inclusion of the ion's response to changes in its environment. Two examples of this many-body character are the induction of multipoles on ions (polarization) and the change in size and shape of the ion in response to changes in its local environment (deformation). The latter effect is particularly important for oxides. Both of these properties can be effectively handled in a molecular dynamics (MD) scheme using ideas borrowed from Car and Parrinello's " ab initio MD" method. The property under consideration (here the ion multipole, size or shape) is represented by additional degrees of freedom in an extended Lagrangian formalism.

Both many-body effects described above may now be examined and parameterised in ab initio electronic structure calculations which focus on the specific ion properties, using novel methods which have been developed by myself and the people I have worked with. This approach is different to fitting an assumed energy function to total energy calculations. The aim is to produce potentials that are truly transferable between different systems, stoichiometries, state points and coordination environments. The parameterisation of this type of model from specific ion properties, rather than from an assumed energy partition in a total energy calculation, means that the parameters should be transferable as long as all of the appropriate physics has been included and the ionic model is really applicable. In order to judge what has to be included for each class of system requires a self-consistent ab initio - potential model - experimental observation cycle. The ab initio calculations are used to produce a potential model. Simulations, involving long time scale simulations or large crystal unit cells, are then performed in order to allow comparison with the appropriate experimental observation. The quality of the agreement with the experiment will indicate the quality/completeness of the potential model and point to what further interactions may be necessary.

The inclusion of simple dipole polarization effects leads to the understanding of a wide range of system properties. For example, the existence of layered (pseudo-two-dimensional) crystal structures, dominant for medium -sized cation MX₂ systems (i.e. MgCl₂), can be explained in terms of a screening effect from the anion dipoles which acts to oppose the cation-cation coulombic repulsion. For even smaller cations, such as Be₂ +, even "less ionic'' structures based on charge-neutral chains of edge-sharing tetrahedra are reproduced. In the amorphous state both types of structure have more than one significant length scale. In molten ZnCl₂, for example, there is intermediate-range order on a 5-10A range in the cation subdensity. Long time scale dynamics have shown that it is this length scale which dominates the relaxation phenomena around the glass transition temperature.

Current work is focusing on long time- and length-scale phenomena which are accessible only via relatively simple models of this type. To this end, we are currently focusing on liquid ZnCl₂ and BeCl₂ as examples of a three-dimensional network (strong-ish) glass and pseudo-one-dimensional chain respectively. Both are being studied in terms of their glass forming and crystallization properties. Additionally, the epitaxial growth of both structural types from the gas phase is also under investigation. This requires that the models used be transferable enough to model the extreme ion environments in isolated clusters.

Further current research projects include:

The modelling of superionic MF₂ fluoride structure systems such as CaF₂ and PbF₂.
Studying small oxide clusters (i.e. SiO₂, TiO₂, MgO and Al₂O₃) which are important components in the dust which surrounds giant and super-giant oxygen-rich stars.
Understanding the "transition" from `Strong' to 'Fragile' glasses from three-dimension networks such as BeF₂ to chain systems such as BeCl₂, through intermediate strength material such as ZnCl 2 .
The requirement for long time and length scales is leading us to pursue the use of high performance computers (for example OSCAR) often utilizing parallelization techniques.

Selected recent publications

"Covalent" Effects in "Ionic'' Systems.'', Madden, P.A and Wilson, M. Chem. Soc. Rev ., 25 , 339, 1996.
"Polarization Effects, Network Dynamics, and the Infrared spectruM. of Amorphous SiO 2 '', Wilson, M., Madden, P.A., Hemmati, M., and Angell, C.A., Phys. Rev. Lett., 77 , 4023, 1996.
"Atomistic Simulation of the High Pressure Behavior of Al 2 O 3 .'', Wilson, M., J. AM. Ceram. Soc., 81 , 2558, 1998
"The Stability of Small MgO Nanotube Clusters: predictions of a Transferable Ionic Potential Model.'', Wilson, M., J. Phys. Chem., 101 , 4917, 1997.
"The "Polymeric" Structure of BeCl 2 from an Ionic Simulation Model.'', Wilson, M., and Madden, P.A., Mol. Phys., 92 . 197, 1997.
"Voids, Rings and Layers: Prepeaks in Network forming Glasses.'' Wilson, M. and Madden, P.A, Phys. Rev. Lett., 80 , 532, 1998."Raman scattering in a Network liquid - relationship to the vibrational density of states.'' Ribero, M. C.C., Wilson, M., and Madden, P.A., J. Chem. Phys., 110 , 4803, 1999.
" a b relaxation processes in ZnCl₂ - a computer simulation study.'' J. Phys: Condensed. Matter, 11 , A237, 1999.

This page last modified 9 August, 2010