Molecular dynamics simulations of diffusion coefficients for cork contaminants in supercritical carbon dioxide

• Autor(a) principal: Rios, William Quintela Pimentel
• Data de Publicação: 2023
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Repositório Científico de Acesso Aberto de Portugal (Repositórios Cientìficos)
• Texto Completo: http://hdl.handle.net/10773/39542
• topic: Cork contaminants; Diffusion coefficient; Molecular dynamics simulations; Supercritical carbon dioxide; Trichloroanisole

The phenomenon of cork taint has been largely attributed to the presence of 2,4,6-trichloroanisole (TCA) in cork, which significantly impairs the organoleptic properties of wine. Industrially, the removal of TCA and other contaminants can be performed with various separation processes, among which supercritical extraction in CO₂. Accurate knowledge of solute transport properties, particularly the diffusion coefficient (D₁₂), is essential for the design of efficient supercritical extraction units for the removal of cork contaminants. However, experimental diffusion coefficient data for these compounds are currently unavailable. This study sought to employ classical molecular dynamics simulations, a computational chemistry technique, to determine D₁₂ of cork contaminants in supercritical CO₂. To validate the computational approach, diverse molecular systems comprising of solutes chemically similar to 2,4,6- TCA were simulated under infinite dilution in supercritical CO₂. The simulation outcomes were subsequently compared against existing experimental D₁₂ data. The OPLS-AA force field was selected to model solute molecules, while three CO₂ force fields (EPM2, TraPPE, and Zhu) were evaluated regarding their ability to predict D₁₂. Isobaric-isothermal (NPT ensemble) and isochoricisothermal (NVT ensemble) simulations were conducted within temperature ranges of 308.15 K to 333.15 K and pressures of 103 bar to 300 bar, depending on the availability of experimental data. Among the CO₂ force field and ensemble combinations, the Zhu model in the NPT ensemble demonstrated superior performance, yielding a global average absolute relative deviation (AARD) of 5.3%, followed by the EPM2 model in NVT (AARD 6.1%). These evaluations encompassed simulations of nine distinct solutes in supercritical CO₂, totaling 54 data points. It was shown that molecular dynamics simulations provided more accurate predictions of D₁₂ relative to empirical equations such as Wilke-Chang or TLSM (Tracer-Liu-Silva-Macedo). After the validation procedure, simulations utilizing the Zhu force field in the NPT ensemble were performed for four cork contaminants: 2,4,6-trichloroanisole, 2,4,6-tribromoanisole, 2,4,6-trichlorophenol, and pentachloroanisole. The molecular dynamics-derived D₁₂ for the primary contaminant ranged from 0.66×10ˉ⁴ cm²/s (at 313.15 K, 150 bar) to 1.31×10ˉ⁴4 cm²/s (at 333.15 K, 300 bar). Given that the validation procedure demonstrated the accuracy of molecular dynamics in predicting D₁₂ for structurally similar solutes, these findings can inform the design and optimization of supercritical extraction processes, while the computational approach holds potential for investigating additional physical properties and exploring interactions between cork contaminants and CO₂.