Ab-initio prediction of the melting point of organic solids


The melting point is one of the most fundamental and practically important properties of a compound. For this reason molecular simulation methods have been developed aiming towards accurate computation of melting points. Knowledge of the melting point before a compound has been synthesized could significantly accelerate the design of new materials. Generally, the molecular simulation methods developed so far for the computation of melting points are not fully predictive, since they require an experimental crystal structure as input. An interesting and challenging task is the prediction of the melting point of a compound from first principles- given just the molecular diagram. In this work, the concept of predicting the melting point of a given organic compound using as an input a computationally obtained crystal structure is investigated. To ensure reliable predictions, it is essential to develop an understanding of how the level of detail of the force fields in terms of crystal structure (CSP) prediction as well in melting point prediction affects the accuracy of the calculations. To explore these requirements the proposed approach in this work combines the application of a CSP multistage methodology developed by the Molecular Systems Engineering group at Imperial College and the freeze method which was recently developed in the group. Using the proposed approach, two different force fields are employed in this study. Initially, the freeze method is applied to the well known Lennard-Jones potential. Moving on to an organic compound, the case of benzene is investigated. A CSP search is performed and the computational structure is used for the freeze method. Proper choice of force field can affect the agreement with experimental data. For this reason two different force fields are employed in this part of the study, a standard CSP force field and a version of the OPLS force field.