Improving the QM/MM description of chemical processes:: A dual level strategy to explore the potential energy surface in very large systems

Potential energy surfaces are fundamental tools for the analysis of reaction mechanisms. The accuracy of these surfaces for reactions in very large systems is often limited by the size of the system even if hybrid quantum mechanics/molecular mechanics (QM/MM) strategies are employed. The large number of degrees of freedom of the system requires hundreds or even thousands of optimization steps to reach convergence. Reactions in condensed media (such as enzymes or solutions) are thus usually restricted to be analyzed using low level quantum mechanical methods, thus introducing a source of error in the description of the QM region. In this paper, an alternative method is proposed, coupled to the use of a micro/ macroiteration algorithm during the optimization. In these algorithms, the number of microsteps involved in the QM region optimization is usually much smaller than the number of macrosteps required to optimize the MM region. Thus, we define a new potential energy surface in which the gas-phase energy of the QM subsystem and the interaction energy with the MM subsystem are calculated at different computational levels. The high computational level is restricted to the gas-phase energy, which is only requested during the microsteps. The dual level strategy is tested for two reactions in solution (the Menshutkin and the oxy-Cope reactions) and an enzymatic one (the nucleophilic substitution of 1,2-dichloroethane in DNA). The performance of the proposed computational scheme seems to be quite promising for future applications in other systems.