Abstrakt
We present a systematic approach for predicting the best solvents for selective extraction of components with unknown structure from complex mixtures (e.g., natural products) - a tool promising dramatic simplification of extraction process optimization. Its key advantage is that identification of the component(s) is unnecessary - prediction is based on a small set of experimental distribution coefficients (obtained using a combination of shake-flask extraction and chromatographic analysis) rather than structure-based descriptors.
The methodology is suitable for the very common situations in practice where the desired compound needs to be separated from unknown impurities (i.e., selectively extracted from the mixture), as well as for large-scale and high-throughput work. The proof-of-concept methodology was developed and evaluated using an extensive set of experimental distribution data of lignin-related compounds obtained in this work.