Kinetic modeling and optimization of maceration and ultrasound-extraction of resinoid from the aerial parts of white lady’s bedstraw (Galium mollugo L.)

In this paper, extraction of resinoid from the aerial parts of white lady’s bedstraw (Galium mollugo L.) using an aqueous ethanol solution (50% by volume) was studied at different temperatures in the absence and the presence of ultrasound. This study indicated that ultrasound-assisted extraction was effective for extracting the resinoid and gave better resinoid yields at lower extraction temperature and in much shorter time than the maceration. A phenomenological model was developed for modeling the kinetics of the extraction process. The model successfully describes the two-step extraction consisting of washing followed by diffusion of extractable substances and shows that ultrasound influences only the first step. The extraction process was optimized using response surface methodology (RMS) and artificial neural network (ANN) models. For the former modeling, the second-order polynomial equation was applied, while the second one was performed by an ANN-GA combination. The high coefficient of determination and the low MRPD between the ANN prediction and the corresponding experimental data proved that modeling the extraction process in the absence and the presence of ultrasound using ANN was more accurate than RSM modeling. The optimum extraction temperature was determined to be 80 and 40 °C, respectively for the maceration and the ultrasound-assisted extraction, ensuring the highest resinoid yield of 22.0 g/100 g in 4 h and 25.1 g/100 g in 30 min, which agreed with the yields obtained experimentally for the same time (21.7 and 25.3 g/100 g, respectively).

► Resinoid recovery from Galium mollugo by maceration and ultrasound-assisted extraction.
► Ultrasonic extraction gave better resinoid yield in shorter time than maceration.
► Phenomenological model describing the process kinetics in the absence and presence of ultrasound.
► Optimization of the resinoid extraction using RSM and ANN-GA models.
► ANN model was more accurate than RSM model.