Energy landscapes and basin sampling of bulk mixed oxide systems

This poster focuses on an energy landscape study of bulk oxide systems, a specific example being a mixture of calcium oxide and magnesium oxide (CaO/MgO). These oxide systems are modeled at the level of empirical potentials (taken from the literature) consisting of a Buckingham potential combined with electrostatics and a shell model for polarizability. This system was studied before by Mohn and Stolen using a genetic algorithm [1]. In this poster we present results obtained by applying the Basin Sampling (BS) methodology [2] to the CaO/MgO system. The BS method attempts to calculate the potential energy density of minima using a Wang-Landau-type sampling algorithm [3]. An approximate method is used for compensating for differences in volume of the basins of attraction of the minima, with the final aim of calculating the partition function using a thermodynamic superposition principle. Applying the BS method to the bulk CaO/MgO mixed oxide system proved challenging: the iterative Wang-Landau method proved to be time-consuming and the resulting density of minima is of a limited accuracy. Possible routes for improving the sampling will be discussed. As an alternative way for characterizing the energy landscape of the CaO/MgO system a method known as 'Fair Phyllis' was used to generate databases of minima and transition states with the ultimate aim of producing a disconnectivity graph.
[1] C.E. Mohn and S. Stolen, J. Chem. Phys. 123, 114104 (2005).
[2] T. V. Bogdan, D. J. Wales, F. Calvo, J. Chem. Phys., 124, 044102, (2006)
[3] F. Wang, D.P. Landau, Phys. Rev. Lett., 86, 2050, (2001)