Optimization of Thyme Essential Oil Microcapsule Production Process: Comparison of Response Surface Method and Artificial Neutral Network

Thymus vulgaris is one of the most well-known medicinal plants which its essential oil contains various active ingredients such as thymol and carvacrol with antioxidant, antibacterial, and antifungal activities. This essential oil is very sensitive and loses its properties against different factors. Microencapsulation process is performed for protecting volatile and sensitive compounds to chemical reactions. Modeling of this process can be effective in assessing and predicting conditions affecting on qualitative properties of the product. The purposes of this study are optimization of the production conditions of Thymus vulgaris essential oil microencapsulation and performance comparison of response surface and artificial neural network methods.

In this study, oil in water emulsions, consisting of 5% (weight/weight) of thyme essential oil in an aqueous suspension containing wall materials (10, 33.13, 20, 26.66 and 30%), with different protein (sodium caseinate) to polysaccharide (modified starch and maltodextrin) ratios (0, 66.6, 20, 33.33 and 40%), were prepared by the help of ultrasonic waves (30, 45, 75, 105 and 120 seconds) -. Microcapsules were prepared from prepared emulsions using freeze-drying and the effect of the above factors on the changes in quality characteristics of the capsules, including the microencapsulation efficiency according to the amount of microencapsulated essential oil, the amount of phenolic compounds, and preserved antioxidants were investigated.

The results of process modeling showed that in both methods, qualitative properties of microcapsules were increased with increasing wall concentration, protein-to-polysaccharide , ratio, and ultrasonic duration. Interaction between protein to polysaccharide ratio and ultrasonic time also improved the preservation of total phenolic compounds and antioxidant capacity of Thymus vulgaris essential oil in moderate levels of variables. 28% of wall concentration, 16% protein to polysaccharide ratio, 111 seconds of ultrasonic time and 29% of wall concentration, 18% of protein to polysaccharide ratio and 87 seconds of ultrasonic time were proposed as optimum points of neural network and response surface method, respectively. Among the optimized samples, the optimum sample obtained from the artificial neural network model showed higher total phenolic content and higher microencapsulation efficiency (P<0.05).

The results of this study showed that the independent variables were effective in the microencapsulation process And the protein to polysaccharide ratio had the greatest effect on it. Also, neural network with optimal topology was very effective than response surface methodology in predicting the qualitative characteristics of microencapsulation of Thymus vulgaris essential oil due to its unique ability in processing information and modeling complex systems.