We report here the properties of supported fluorous liquid membranes based on porous alumina. The alumina is first rendered compatible with fluorous solvents by surface modification with an oligomeric perfluoropropylene oxide-based carboxylic acid, Krytox 157FSH. After modification, simply dipping the porous alumina membrane into a perfluorinated solvent results in a supported liquid membrane with high selectivity for fluorous compounds. Two homologous series of compounds differing in the number of -CF2- groups were investigated, namely esters of cinnamyl alcohol and the analogous naphthyl derivative with 2H,2H,3H,3H-perfluoroalkanoic acids (HOOC-(CH2)(2)-(CF2)(m-1)CF3, m = 2,4,6 and 8). Four perfluorinated membrane solvents (FC-77, PF-5080, FC-3283 and FC-43) were investigated. In FC-3283, the permeabilities, which are the products of a diffusion coefficient and a partition coefficient in the solution-diffusion model. of cinnamyl alcohol derivatives are 3.62 +/- 0.36 times greater than those of the analogous naphthyl compounds for the solutes containing the same perfluorinated chain. Permeability. R increases as the perfluorinated chain length increases in all of the perfluorinated solvents. Values of log(P) vs m are linear with a slope of 0.147 +/- 0.002 but with different intercepts for the various solvents. Independent measurements of the partition coefficients of the solutes between the source/receiving phase solvent, ethanol. and the fluorous solvents reveal that the selectivity behavior is dominated by partitioning rather than diffusion. The free energy of transfer of a -CF2- group (ethanol to perfluorinated solvents) is -1.1 kJ/mol. Despite the fact that the solvents are mixtures, not pure liquids, the partition coefficients are well correlated with values calculated based on group contributions with 'mobile order and disorder' theory. The diffusion coefficients of four solutes in four membrane solvents were also determined based on the solution-diffusion model. The Stoke-Einstein equation shows satisfactory estimation of experimental results. (C) 2009 Elsevier B.V. All rights reserved.