Finding low energy minima of (H2O)25 and (H2O)30 with temperature basin paving Monte Carlo method with effective fragment potential: New ‘global minimum’ and graph theoretical characterization of low energy structures

Determination of low energy structures for large water clusters requires fast and efficient geometry optimization algorithm. A new implementation of the Monte Carlo temperature basin paving (MCTBP) method (J. Phys. Chem. A 2011, 115 (42), 11866) has been used to find low energy structures of (H2O)25 and (H2O)30. MCTBP algorithm uses local minimization in combination with Monte Carlo sampling similar to the basin hopping algorithm. The difference between basin hopping and MCTBP is that in the MCTBP method, Boltzmann factor used in sampling is modified based on the history of sampling favoring lesser sampled structures in subsequent steps. In the new implementation of MCTBP, visit of similar structures during the course of optimization is reduced. With this new implementation, we have found several structures for both the clusters (6 for (H2O)25 and 13 for (H2O)30) lower in energy than previously reported lowest energy structure when the energy is calculated with effective fragment potential. Detailed explanations of the low energy structures are given both in terms of their three dimensional and graph theoretical description. A statistical analysis of the optimization runs show that the low energy structures are visited repeatedly in different trajectories indicating the robustness of the algorithm.