Identifying optimal solvents for reactions using quantum mechanics and computer-aided molecular design

A new iterative hybrid methodology, incorporating quantum mechanics (QM) calculations and a computationally inexpensive computer-aided molecular design (CAMD) methodology, QM-CAMD, for identification of optimal solvents for reactions is presented. The methodology has been applied to a Menschutkin reaction, where pyridine and phenacyl bromide are the reactants. The QM calculations take on the form of density functional theory calculations with a given solvent treated using continuum solvation models. The accuracy of the solvent QM calculations is assessed by computing free energies of solvation for different solvation models; the IEF-PCM, SM8 and SMD models are studied and SMD is identified as the best model. Rate constants kQM, determined from QM calculations, are calculated based on conventional transition state theory (Eyring 1935, Evans & Polanyi 1935). By using the SMD solvation model and a statistical mechanics derivation of kQM, rate constant predictions within an order of magnitude are achieved. For a small set of solvents investigated by QM, selected solvent properties are predicted using group contribution (GC) methods. 38 structural groups are considered in this approach. The QM-computed rate constants and solvent properties determined by GC are used to obtain a computationally inexpensive reaction model, based on an empirical linear free energy relationship, which is used to predict reaction rate constants. This predictive reaction model is incorporated into an optimisation-based CAMD methodology. With an objective function of maximising the reaction rate constant subject to molecular and reaction condition constraints, optimal solvent candidates are identified. By considering a design space of over 1000 solvent molecules, solvent candidates containing nitro-groups are predicted to be optimal for the Menschutkin reaction. This outcome supports experimental results for a related reaction available in the literature (Lassau & Jungers 1968). For verification purposes, Ganase et al. (2011) have measured (based on 1H NMR data and kinetic analysis) the rate constant for the reaction of interest in a number of solvents and report a significant increase in the rate constant with nitromethane as the solvent