RESUMO
Selective extraction is a great concern in the field of natural products. The interest is to apply specific conditions favouring the solubility of targeted secondary metabolites and avoiding the simultaneous extraction of unwanted ones. Different ways exist to reach selective extractions with suited conditions. These conditions can be determined from experimental studies through experimental design, but a full experimental design takes time, energy, and uses plant samples. Prediction from varied solubility models can also be applied allowing a better understanding of the final selected conditions and eventually less experiments. The aim of this work was to develop and use a chromatographic model to determine optimal extraction conditions without the need for numerous extraction experiments. This model would be applied on the selective extraction of the desired antioxidant compounds in rosemary leaves (rosmarinic and carnosic acids) vs chlorophyll pigments to limit the green colour in extracts. This model was achieved with Supercritical Fluid Chromatography (SFC) and then applied to Supercritical Fluid Extraction (SFE) and Pressurised Liquid Extraction (PLE) assays. SFC models predicted low solubility of chlorophylls for low (5%) and high (100%) percentage of solvent in carbon dioxide. Also, low solubility was predicted with acetonitrile solvent compared to methanol or ethanol. This was confirmed with different extractions performed using SFE with different percentages of solvent (5, 30, and 70%) and with the three solvents used in the SFC models (acetonitrile, methanol and ethanol). Also extractions using PLE were carried out using the same neat solvents in order to confirm the SFC models obtained for 100% of solvent. Globally, extractions validated the SFC models. Only some differences were observed between ethanol and methanol showing the complexity of plant extraction due to matrix effect. For all these extracts, the content of carnosic acid and rosmarinic acid was also monitored and selective extraction conditions of bioactive compounds could be determined.