Global optimisation of hydroxylated silica clusters: A cascade Monte Carlo Basin Hopping approach

Abstract

We report on a global optimisation study of hydroxylated silica nanoclusters (SiO2)M·(H2O)N with sizes M=6, 8, 10 12, and for each size with a variable number of dissociatively chemisorbed water molecules (N=1, 2, 3…). Due to the high structural complexity of these systems and the associated ruggedness of the underlying potential energy landscape, we employ a “cascade” global optimisation approach. Specifically, we use Monte Carlo Basin Hopping (MCBH) where for each step we employ two energy minimisations with: (i) a lightly parameterised but computationally efficient interatomic potential (IP) which does not distinguish between H-bonded conformational isomers, and then (ii) a more sophisticated IP which accounts for polarisation and H-bonding. Final energies from the MCBH search are then refined with optimisations using density functional theory. The reliability of our approach is first established via comparison with previously reported results for the (SiO2)8·(H2O)N case, and then applied to the M=6, 10 and 12 systems. For all systems studied our results follow the trend in hydroxylation energy versus N, whereby the energy gain with hydroxylation is found to level off at a point where the average tetrahedral distortion of the SiO4 centres is minimised. This optimal hydroxylation point is further found to follow an inverse power law with increasing cluster size (M) with an exponent close to −2/3, further confirming work in previous studies for other cluster sizes.