The choice of solvent in which to carry out liquid-phase organic reactions often has a large
impact on reaction rates and selectivity and is thus a key decision in process design. A systematic
methodology for solvent design that does not require any experimental data on the effect of
solvents on reaction kinetics is presented. It combines quantum mechanical computations for
the reaction rate constant in various solvents with a computer-aided molecular design (CAMD)
formulation. A surrogate model is used to derive an integrated design formulation that combines
kinetics and other considerations such as phase equilibria, as predicted by group contribution
methods. The derivation of the mixed-integer nonlinear formulation is presented step-by-step.
In the application of the methodology to a classic SN2 reaction, the Menschutkin reaction,
the reaction rate is used as the key performance objective. The results highlight the tradeoffs
between different chemical and physical properties such as reaction rate constant, solvent
density and solid reactant solubility and lead to the identification of several promising solvents
to enhance reaction performance.