Kinetic models for extraction with supercritical carbon dioxide from pink pepper: Theoretical, empirical, and semi-empirical models and artificial neural network approach

•Kinetic models for supercritical CO2 extraction from pink pepper are reported.•Kinetics was governed by periods of falling extraction and diffusion-controlled rates.•Artificial neural network had better predictability than the other mathematical models. In this study, the extraction kinetics of pink pepper (Schinus terebinthifolius Raddi) using supercritical CO2 were investigated. S. terebinthifolius is an important fruit tree that is native to Brazil. Owing to its medicinal properties, this plant demonstrates potential for commercial use. However, to date, the extraction of compounds from pink pepper using SFE has been investigated in few studies, and no reports have been found on kinetic models for the SFE of compounds from this raw material. A design of experiments was developed to evaluate the influences of temperature and pressure on the extraction kinetics. The models of Tan and Liou, Crank, Esquível et al., and Martínez et al. were fitted to the extraction curves, and compared to models by artificial neural networks (ANNs). The influences of three databases and input neurons on modeling by a feedforward ANN with three layers were investigated. The pressure effect is pronounced, consequently increasing the extraction yield by increasing the solvent density. Temperature exhibits a complex effect on both solvent density and solute vapor pressure. The curves are governed mainly by periods of falling extraction and diffusion-controlled rates. The ANN, Esquível et al., and Crank models yield the most suitable results. However, the ANN model exhibits better predictability than that of the other models.