Journal of Supercritical Fluids, Vol.19, No.1, 25-37, 2000
Fractionation of fish oils using supercritical CO2 and CO2 plus ethanol mixtures
This work reports the countercurrent extraction and fractionation of a range of crude fish oils using supercritical carbon dioxide and carbon dioxide + ethanol mixtures. Vitamin A palmitate was extracted from model Cod liver oil/vitamin mixtures using pure CO2. The separation factor was low, due to similar solubilities of the vitamin ester and the oil. Vitamin A was also recovered from Cod liver oil ethyl esters\vitamin A mixtures. The separation factor was substantially improved over the non-esterifred oil, due to large differences in the solubilities of the esters and vitamin A in supercritical CO2. Solubilities of fish oils and squalene are reported using CO2 + ethanol mixtures at 333 K, ethanol concentrations from 0 to 12% by mass and pressures from 200-300 bar. Solubilities of all oils and squalene increased exponentially with linear increases in the ethanol concentration. The solubility of polar components increased more rapidly than non-polar components. Pilot scale removal of fatty acids from Orange Roughy oil and squalene from deep sea shark liver oil was carried out using CO2 ethanol as the solvent. The extent of fatty acid removal from Orange Roughy oil was superior to pure CO2, whereas the degree of separation of squalene from shark liver oil was inferior. Throughput was substantially increased relative to pure CO2 in both cases. Mass transfer behaviour and product purity results are also presented for the demonstration scale production of squalene and diacyglycerylether fractions from deep sea shark liver oil using pure CO2. Fractionation results are compared with previous experimental results obtained at a laboratory and pilot scale to obtain the height of a transfer unit (HTU) and packed height required as a function of scale. Modelling of flooding at supercritical conditions was carried out to enable optimal design of the packed column. A liquid-liquid flooding model gave reasonable correlation of literature flooding data, obtained at a laboratory and pilot scale. (C) 2000 Elsevier Science B.V. All rights reserved.