Ab initio Monte Carlo simulations applied to a Si5 cluster

An ab initio Monte Carlo (MC) program has been developed, where the total configurational energies at each MC step are obtained via density functional theory using localized numerical orbitals as basis sets. As an initial application we have investigated the structural properties of a silicon cluster (Si-5) for different temperatures and estimated its isomerization temperature. In particular, we study how this temperature depends on the particular choice of exchange-correlation functional. We performed calculations within the local density approximation (LDA), a generalized gradient approximation (GGA), and the Tao-Perdew-Staroverov-Scuseria metaGGA. The Monte Carlo method is very useful when complex exchange-correlation functionals are being used, since in these cases it is computationally difficult to evaluate the forces. From the simulations we have observed two distinct isomers, which were identified as trigonal bipyramid D-3h and C-2 symmetric structures. The D-3h was found to be the global minimum and the C-2 the first local minimum. We obtain that the isomerization temperature for the small Si-5 system has a trend opposite to what happens in the bulk, where the LDA functional provides the lowest melting temperature, whereas the one obtained using the metaGGA appears to be the highest one.