Chemometrics for analytical data mining in separation process design for recovery of artemisinin from Artemisia annua

摘要

In this work, to understand the separation behavior of the flash column chromatography during purification of artemisinin from the crude extract of Artemisia annua, multivariate data analysis technique PARAFAC (Parallel factorization) is used to mine the relevant chemical information from analytical chromatograms of 9 artemisinin containing fractions. The size of three way dataset obtained from chromatogram measurements in sample, retention time, and spectral mode is 9 × 1981 × 82. Prior to the application of PARAFAC, the dataset is preprocessed to remove baseline drift and peak misalignment caused by retention time shifts due to matrix effects. Due to the complicated nature of chromatograms, the preprocessed HPLC data were divided into intervals containing analytical signals and then PARAFAC modeling was performed on individual intervals. Loadings from the PARAFAC analysis provided pure elution profiles and pure UV spectra even for co-eluting peaks, thus enabling the identification of chromatographically unresolved components. Also, loadings were used to determine the number of components and their relative concentrations in the fractions containing artemisinin which are the most important information of the flash column performance.