RESUMO
A simple, fast, and efficient ultrasonic-assisted supramolecular solvent microextraction combined with high performance liquid chromatography method was developed for the determination of coumarins in Cortex fraxini, including esculin, esculetin and fraxetin. In this study, a novel supramolecular solvent was prepared with 1-octanol, tetrahydrofuran and water for the first time, and its composition, viscosity, density, structure, and micromorphology were characterized. The prepared supramolecular solvent exhibited vesicular structures and had the characteristics of low viscosity. Through single-factor experiments, response surface methodology and artificial neural network-genetic algorithm, the optimal extraction conditions were obtained as follows: NaCl concentration of 1 mol mL-1, pH value of 10, solid-liquid ratio of 10:1, vortex time of 30 s, ultrasonic power of 100 W, ultrasonic temperature of 60 °C, ultrasonic time of 15 min, centrifugation speed of 5000 rpm, and centrifugation time of 1 min. The results demonstrated that the artificial neural network model exhibited maximum R-values of 0.98703, 0.97440, 0.99836, and 0.95447 for training, testing, validation, and all dataset, respectively. The minimum mean square errors were 0.75, 10.15, 1.99, and 2.63, respectively. This indicated that the predicted values were almost consistent with the actual values. Under the optimal conditions, the total extraction yields of target analytes reached 2.80 %. The calibration curves for each analyte exhibited excellent linearity within the linear range (r > 0.9993). The limits of detection and quantification ranged from 4.87 to 6.55 ng mL-1 and 16.24 to 21.84 ng mL-1, respectively. The recoveries ranged from 98.71 % to 111.01 % with relative standard deviations of less than 3.6 %. The present method had the advantages of short extraction time (15 min) and less solvent consumption (0.5 mL). The prepared supramolecular solvent was proved to have great potential in extracting coumarins from medicinal plants.