Statistical mixture design--principal component optimization for selective compound extraction from plant material

A simplex centroid design is used to optimize solvent mixtures for selective extraction of compounds from Erythrina speciosa Andrews leaves. Three main groups of compounds characterized by chromatographic retention times of 1.7, 3.1, and 6.6 min were extracted. The chromatographic peak heights registered at 1407 equally spaced times for each design extract were converted into principal components (PCs). Three PCs account for 89.1% of the data variance and have important loading values at the above retention times. Quadratic mixture models were found to adequately describe the response surfaces of their PC score values. Maximizing the first PC scores while minimizing the second and third ones results in an optimum 83% dichloromethane-17% acetone mixture that selectively extracts the group of compounds with retention times around 3.1 min. Maximizing the second PC scores and minimizing the third PC scores leads to a 78% hexane-22% acetone mixture adequate for extracting the 6.6 min retention time compounds as well as other compounds with high retention times. The 1.7 min group of compounds is most efficiently extracted on maximizing the third PC that results in a 50% ethanol-50% acetone mixture. The Derringer-Suich algorithm confirmed the determination of these optimized mixtures.