Diffusion coefficients of bioactive compounds in sub/supercritical CO2 and CO2/ethanol mixtures from molecular dynamics simulations

A computational strategy based on molecular dynamics (MD) simulations is proposed for the prediction of tracer diffusion coefficients of bioactive compounds in CO2 (D12) and CO2 modified with ethanol (Dm,i), in both supercritical and subcritical regions. Three systems were used for its validation, namely, 8-carotene and ibuprofen in supercritical CO2, and gallic acid in CO2 modified with 16 mol.% of ethanol. The MD diffusivities were assessed with experimental data, with average absolute relative deviations (AARD) between 3.82 % and 7.36 %, and average relative deviations (ARD) between -6.18 % and 5.75 %. The dependency of the diffusion coefficient on temperature, pressure, and Stoke-Einstein coordinate (T/& mu;) was also evaluated. All data followed the expected trends apart from gallic acid in CO2 modified with 16 mol.% of ethanol, for which the computed Dm,i showed a weak linear dependency with T/& mu;.The validated MD strategy was then employed to calculate diffusivities of eucalyptol in sub/supercritical CO2, and in sub/supercritical CO2 modified with 8 mol.% and 16 mol.% of ethanol. The computational strategy was re-evaluated upon comparison with experimental diffusivities of eucalyptol in sub/supercritical mixtures of CO2 with 8 mol.% ethanol. The expected dependencies on pressure, temperature, and Stoke-Einstein coordinate were obeyed, and it was also found that the diffusion coefficient decreased with increasing ethanol content. Four phenomenological models from the literature were adopted to represent MD data, with AARD values between 1.21 % and 4.40 %, suggesting they can be used for diffusivity estimations over wider ranges of operating conditions.

Automated summary

A computational strategy based on MD simulations is proposed to predict tracer diffusion coefficients of bioactive compounds in CO2 and CO2 modified with ethanol. The strategy is validated with three different systems, and shows good agreement with experimental data. The dependence of diffusion coefficients on temperature, pressure, and Stoke-Einstein coordinate (T/μ) is evaluated.