Are Explicit Solvent Models More Accurate than Implicit Solvent Models? A Case Study on the Menschutkin Reaction

In this work, contemporary quantum mechanical (QM) implicit solvent models (SMD, SM12, and COSMO-RS) and a molecular mechanical (MM) explicit solvent model were used to predict the aqueous free energy barrier of a simple Menschutkin reaction (NH3 + CH3Cl). Surprisingly, the explicit solvent approach performed the worst, while the implicit solvent models yielded reasonably accurate values that are in accord with available experimental data. The origin of the large error in the explicit solvent model was due to the use of a fixed set of Lennard-Jones parameters during the free energy perturbation (FEP) calculations. Further analyses indicate that M06-2X/6-31+G(d,p) yielded solute solvent interaction energies that are in good agreement with bench-mark DLPNO-CCSD(T)/CBS values. When end-state MM to M06-2X/6-31+G(d,p) corrections were added using FEP, it significantly improved the accuracy of the explicit solvent MM result and demonstrated that the accuracy of these models may be systematically improved with end-state corrections based on a validated QM level of theory.