Theoretical estimation of solvation free energy by continuum solvation models, as a standard approach in computational chemistry, is extensively applied by a broad range of scientific disciplines. Nevertheless, the current widely accepted solvation models are either inaccurate in reproducing experimentally determined solvation free energies or require a number of macroscopic observables which are not always readily available. In the present study, we develop and introduce the Machine-Learning Polarizable Continuum solvation Model (ML-PCM) for a substantial improvement of the predictability of solvation free energy. The performance and reliability of the developed models are validated through a rigorous and demanding validation procedure. The ML-PCM models developed in the present study improve the accuracy of widely accepted continuum solvation models by almost one order of magnitude with almost no additional computational costs. A freely available software is developed and provided for a straightforward implementation of the new approach.

Flash point is one of the most widely used properties in the risk assessment and safe design of process industries. In the current work, we have presented a novel and accurate model to predict the flash points of pure organic compounds from diverse families. The proposed model is a linear correlation between the flash point, normal boiling point, and 42 predefined functional groups constituting the molecule. Evaluation of the model through a data set of 1533 pure organic compounds shows an average absolute deviation, an average absolute relative error, and a correlation coefficient of 5.83, 1.61, and 0.992, respectively. Comparing the results of the present study with other works shows that the model proposed in this work is among the most accurate and reliable ones to date.

Accurate evaluation of combustion enthalpy is of high scientific and industrial importance. Although ab-initio computation of the heat of reactions is one of the promising and well-established approaches in computational chemistry, reliable and precise computation of heat of combustion reactions by ab-initio methods is surprisingly scarce in the literature. A handful of works carried out for this purpose report significant inconsistencies between the computed and experimentally determined combustion enthalpies and suggest empirical corrections to improve the accuracy of the ab-initio predicted data. The main aim of the present study is to investigate the reasons behind those reported inconsistencies and propose guidelines for a high-accuracy estimation of heat of reactions via ab-initio computations. We show comparably accurate prediction of combustion enthalpy of 40 organic molecules based on a DSD-PBEP86 double-hybrid density functional theory approach and CCSD(T)-F12 coupled-cluster computations, with mean unsigned errors with respect to experimental data being below 0.5% for both methods.

No abstract