Automated development of force fields for the calculation of thermodynamic properties: acetonitrile as a case study

Force fields for engineering applications are often parameterised using strategies based on quantum mechanical ab initio calculations and thermodynamic properties from experiment. An automated procedure for adjusting molecular model parameters to experimental thermodynamic property data is introduced. The process accelerates the development of molecular models by an efficient use of parallel computing power and an autonomous progress of the model development without any user interaction. As a case study, the procedure is applied to the parameterisation of a molecular model for acetonitrile. The resulting model reproduces vapourliquid equilibrium data of acetonitrile with an accuracy of 0.1% for the saturated liquid density, 4.9% for the vapour pressure and 3.7% for the enthalpy of vaporisation. These accuracies are superior to data obtained with previously published force fields for acetonitrile.