Combinaison d'un modèle de solvatation continu et d'une équation d'état pour calculer les effets de solvatation dans des modèles cinétiques détaillés

Detailed kinetic models are valuable assets for unraveling the dynamics of complex reacting systems governed by radical-chain mechanisms. They provide a comprehensive overview of the various reaction pathways involved, helping in the identification of the key reaction steps and can be used in reactor design. Automatic kinetic generators are pivotal tools in the development of such models. These generators rely on ideal gas approximations to compute the required thermo-kinetic properties, employing group additivity methods along with predefined rate rules and reaction templates. However, these mechanisms may take place within other phases than gases in certain applications such as, e.g., (i) the liquid-phase oxidation processes to produce phenol and KA oil, important precursors in the polymer sector, (ii) liquid-phase oxidation mechanisms that result in the ageing of fuels and biofuels, and (iii) new intensified combustion technologies using supercritical fluids as a reaction medium, like in the Allam power cycle. Nonetheless, developing a detailed kinetic model for such scenarios is quite a challenge due to the extensive demand for thermodynamic and kinetic data, which not only depend on the solutes but depend also on their solvents in order to model the complex liquid or supercritical mixtures of interest. In particular, such data are nonexistent for the free radicals and transition states at the heart of chain mechanisms. In this context, this work introduces a predictive version of the Peng-Robinson equation of state (PR EoS), designed to correct ideal gas thermo-kinetic data generated by automatic kinetic models through the computation of solvation quantities. For this purpose, the PR EoS was adapted in two ways. On the one hand, we have proposed new group contribution methods to estimate EoS input parameters related to pure species, free radicals and transition states. On the other hand, we have proposed advanced predictive mixing rules based on a continuous solvation model derived from a quantum chemical approach (COSMO-RS and COSMO-SAC). Although no experimental data is needed to estimate the mixture parameters of the EoS (only predictive quantum-based calculations are used instead), our model provides reliable predictions of solvation free energies, with average errors below 0.5 kcal/mol. The proposed approach, based on an equation of state, has also proven to be a valuable tool for calculating complex fluid phase diagrams, as well as heat capacities and enthalpies, which are essential for process and product design.