100 M.J. Lewis diffuse nature of the membrane, this is not so. This approach ignores molecular shape face and within the membrane itself, A nponents in the feed, and at the membrane sur- vertheless it is useful for a preliminary(initial) selection of a suitable membrane. However, it tells you nothing about the rejection value of a component below the molecular weight cut-off, say 500 or 1000. In fact, it rather implies that such components would be freely permeating. In reality, this is not the case as most membranes are diffuse in their separation ability. The concept of a sharp and diffuse membrane is useful in this respect(see data from Table 4.2) Figure 4. I shows the rejection characteristics of two such membranes. The sharp membrane is an ideal situation, offering the perfect separation. Real membranes offe quite diffuse rejection characteristics, requiring a molecular weight difference of about tenfold to provide an effective separation. Therefore they would give a poor separation of components with slight differences in molecular weights, even components with differences up to two times would not necessarily be well separated. For example it ould not be easy to fractionate the proteins in cheese whey or to separate mono saccharides from disaccharides. McGregor (1986)has undertaken some interesting xperiments, using electrophoresis to examine the sharpness of separations performed on mixtures of protein of different molecular weights. His results showed considerable differences in the sharpness of the separation between different membranes with the same nominal molecular weight cut-off value. Gekas et al.(1990)found that experimental flux and rejection data correlated better with porosimetric data(pore size and pore size distri bution as measured by bubble pressure and solvent permeabilities)than molecular weight cut-off value These types of observation illustrate that although some physicochemical measure- ments might be useful, the selection of the best membrane is best done experimentally, by measuring the rejection characteristics of the components to be separated at the selected There is also evidence that the rejection value for most components increases during the course of an ultrafiltration process. Some of the experimental work on rejection measurement and practical problems involved are described in Section 4.2.4 Fig. 4. 1. Characteristics of a sharp and diffuse membrane: Ij, ideal, 10 000 molecular weight cut- off: 12. ideal, 100 000 molecular weight cut-off: S, sharp membrane: D, diffuse mem1.0 c 0 8 0.5 '6 a: .- c 0 - - -4' I