Modelling and optimization of microwave-assisted extraction of rosemary using response surface methodology (RSM)

Introduction: Microwave-assisted extraction (MAE) is one of the newest techniques for extracting essential oils from medicinal plants. Microwave waves penetrate into plant cells and affect the polar molecules, causing heat to build up inside the plant tissue, destroying the cells and releasing the active ingredients. Although in terms of essential oil extraction efficiency, MAE method is slightly higher than Clevenger, but in terms of energy consumption, process time and amount of solvent consumed has an inherent advantage over the Clevenger method. In this study, the aim is to evaluate the quantity and quality of rosemary essential oil in the developed microwave extraction system and process optimization in this system. Materials and Methods: Rosemary plant (Rosmarinus officinalis L.) with 200 g used for each experiment. In this study, a microwave-assisted essential oil extraction system was developed. The system consisted of a home microwave, a distillation condenser, a cold-water source and a pump. In this study, the response surface methodology (RSM) used in the form of a central composite design with 13 treatments. The effect of two independent variables including power in the range of 300 to 900 watts and time in the range of 5 to 35 minutes on the efficiency of essential oil as dependent variables was investigated. To evaluate the quality of the extracted essential oil and compare it with Clevenger treatment, GC-Mass analysis was done using an Agilent technology apparatus (Agilent 7990B, USA). Results and Discussion: A quadratic model was proposed for modelling of extraction efficiency and it was able to model with the values of R2, Adj R2 and Pred R2, which were 0.9521, 0.9180 and 0.574, respectively. The "Pred R2" of 0.547 is not as close to the "Adj R2" of 0.9180 as one might normally expect. The difference is more than 0.2. This may indicate a large block effect or a possible problem with the proposed model and/or data. Things to be considered for model modification are model reduction and response transformation. Results of ANOVA showed that only the linear terms of time, power and quadratic power of time are significant. The normal probability plot showed the normal distribution of errors observed in the experimental design space. The actual measured values R2 of the extracting efficiency from the experiments versus the values R2 predicted by the model indicated that this model could estimate the extracting efficiency with a correlation co