APPlICATIONs OF MEMbRANE TEChNOlOGIEs FOR ThE TREATMENT OF whEY

Whey is the coproduct of the cheese- and casein-producing industries. Its composi-tion varies according to the process from which it comes, but it can be character-ized roughly as milk from which 90% to 95% of the casein and the fat have been removed. Consequently, whey still contains a lot of components of high nutritional value but some (proteins and lactose) in an unbalanced ratio for human nutrition, and others (minor proteins, growth factors, etc.) in very low concentration. Classically, two types of whey (sweet and acid) are distinguished by the dairy industry accord-ing to their pH, >6.4 and from 6.4 to 4.6, respectively. Until the end of the 1960s, whey was mainly used for animal (pig) feeding or even was spread over the fields or directed into the sewage system, with adverse environmental consequences: With a BOD₅ of 30 to 50 g.L^–1, 1000 L of whey has the same polluting power as 400 people (Marshall et al., 1968).

Thanks to the membrane technologies, a new whey industry has emerged, and it represents one of the best demonstrations of what can be done with well-thought-out uses of separation techniques and biotechnologies. In Figure 3.3, the current state of the art is summarized. The use of 0.1 mm MF for treating cheese milk has originated a new category of whey named ideal whey by Fauquant et al. (1988). This ideal whey or MMF (milk microfiltrate) is sterile, its eventual virus count is reduced by at least

Classical

FIGuRE 3.3 Membrane technologies and whey.

Applications of Membrane Technologies in the Dairy Industry 47

2.8 log (Gautier et al., 1994), it has no fat and no k-GMP, and if milk has not been heated, in industrial conditions, its protein and salt contents are those of the aque-ous phase of raw milk. When treated by UF, MMF leads with the introduction of a diafiltration step to either a WPC (whey protein concentrate) or a WPI (whey protein isolate), according to the used VCF, having, respectively, protein/total solids ratios of 0.77 and 0.975 (Maubois et al., 2001) and showing high nutritional and technico-functional qualities (foaming, gelling, solubility) that are better than those obtained for WPC and WPI made from classical whey (Bacher and Konigsfeldt, 2000).

As previously mentioned, whey processing was one of the first applications of membrane technologies in the dairy industry. Use of RO instead of vacuum evapora-tion for preconcentraevapora-tion of whey has allowed for a large saving in energy. Energy consumption is 9 kWh per ton of removed water for RO versus consumption between 90 and 150 kWh for vacuum evaporation (Daufin et al., 1998a). A large part of the RO membrane area carrying out this concentration on sweet whey is replaced by NF membrane in order to simultaneously perform concentration (until a total sol-ids content of 22% to 25%) and partial demineralization (removal of 25% to 50%

of the mineral salts, mainly the monovalent species). Moreover, this double effect obtained by the use of NF leads to a saving of energy compared to RO (the used transmembrane pressure is reduced to 30 bar or less), a reduction of effluents, and a significant improvement of the spray-drying of whey because of a better crystalliza-tion of lactose. On the other hand, use of NF has provided to the dairy industry new possibilities to commercialize, with a reasonable added value, components of acid whey which were previously difficult to adapt for animal feeding because of its high mineral content and were also the source of adverse effects on the environment.

Use of UF was extensively applied to whey to allow the development of a broad array of WPCs with a protein/total solids ratio ranging from 35% to 80%

FIGuRE 3.4 Ultrafiltration (UF) equipment for treating pH 4.6 milk.

48 Engineering Aspects of Milk and Dairy Products

(Pearce, 1992). Numerous studies related to the manufacture of WPCs (Hobman, 1992) and to their properties (Mangino, 1992) have been published. In summary, the manufacture of high-quality WPCs required particular care of the technological treatments applied to the milk used for making cheese: Heat treatments have a cumu-lating effect on the thermosensibility of whey proteins. The presence of proteolytic enzymes issued from psychotrophic or thermoduric bacteria causes casein degrada-tion and increases the NPN (nonprotein nitrogen) content. Regarding the whey, care-ful control of bacteria growth is essential because during UF, initial bacteria count can be concentrated by up to 30 to 50 times. Lastly, removal of casein fines and of globular residual fat has to be done. To our knowledge, no study has been concerned with the potential detrimental effect of the somatic cell content of whey, in spite of the fact that about 15% of the cells of the cheese milk are going in the whey and then are concentrated as bacteria 30 to 50 times, which means a count in the WPC ranging from 1.5 × 10⁶ to 3 × 10⁶ cells per mL or even more according to the initial count. The preferred temperature for the ultrafiltration of whey is generally 50°C. At this temperature, acceptable fluxes are achieved, and thermal denaturation of protein is minimized. However, most of the manufacturers of WPC preferred to operate at a lower temperature (10°C to 12°C) in spite of a lower flux: half of the flux was achieved at 50°C (Nielsen, 1988) because of a much lower growth of thermoduric bacteria in the spiral-wound membrane equipment and the increase of solubility of calcium salts at this temperature which slows fouling.

Residual fat of whey affects the functionality (emulsifying, foaming, and gelling characteristics) of whey proteins, impairs the UF membrane flux during the manufac-ture of WPC, and can promote the development of off-flavors (Rosenberg, 1995). To remove residual lipids from whey, a thermocalcic aggregation process of these com-ponents was simultaneously proposed by Maubois and al. (1987) and Pearce (1987).

The optimized method is summarized in Figure 3.3. Whey is first concentrated by UF until a concentration of 4 to 5, then the pH of the retentate is adjusted to 7.5 by the addition of sodium hydroxide, the temperature is maintained at 55°C for 8 minutes, and finally, the lipoprotein-Ca aggregates as well as the small fat globules and the bacteria are separated by 0.1 mm membrane MF. The absence of fat in the resulting microfiltrate strongly reduces the fouling in subsequent UF; consequently, the UF running time is increased, and the UF flux is at least doubled (Maubois et al., 1987) despite the fact that the used UF membrane must have a low MWCO (no more than 5000 Da) for avoiding losses in small-sized whey proteins such as a-lactalbumin.

The introduction of a diafiltration step in the UF process allows us to easily obtain WPI with a protein/TS ratio higher than 80% and show high foaming and gelling properties, although slightly lower than those observed for WPI issued from “ideal whey” as aforementioned. Nevertheless, for example, in a meringue-like formulation, egg white can be totally substituted by a 10% WPI protein solution in both overrun and stability. As shown by Pearce (1987) and Maubois et al. (1987), defatted WPI are excellent starting materials for industrial production of purified b-lactoglobulin and a-lactalbumin through a process based on the property of a-lactalbumin to reversibly aggregate (Pearce, 1983) at low pH (3.8 by addition of HCl or preferably citric acid) with a moderate heat treatment (55°C for 30 min). If highly purified b-lactoglobulin is obtained through this process (Léonil et al., 1997), there are still problems, to our

Applications of Membrane Technologies in the Dairy Industry 49

knowledge, related to the purity of the industrially recovered a-lactalbumin (70%

to 75% due to presence of some denatured immunoglobulins, b-lactoglobulin, and bovine serum albumin) despite the numerous studies carried out (Bramaud, 1995).

Further work is required for better knowledge of the structural conformation of this protein and of its interactions with the other proteinaceous components present in whey, because a-lactalbumin has a great potential market due to its already shown biological properties (Maubois and Ollivier, 1997), both in nutraceutics (brain hor-mone precursors due to its high content in tryptophane, four residues per mole) and in therapeutics (apoptosis of lung carcinogen cells as shown by Hakansson et al., 1995).

On the other hand, owing to its high content of phospholipids, whey MF retentate that represents a volume of no more than 2% of the initial volume of whey (Baumy et al., 1990) has potential as an effective emulsification agent for food applications (low-fat dairy products or sausages) or cosmetics. As shown by these authors, it constitutes an excellent starting material for producing purified phospholipids (phosphatidyletha-nolamine, phosphatidylinositol, phosphatidylserine, phosphatidylcholine, sphyngo-myelin, and ceramids) with a yield of 150 g per 1000 L of whey. As it was for milk UF permeate, use of UF whey permeate will be either for animal feeding or manufacture of lactose after partial demineralization by NF. However, in some countries, produc-tion of edible ethanol from UF whey permeate through yeast fermentaproduc-tion, either for drinking (Carbery process) or for fuel is an industrial reality (Barry, 1982).

NF performs simultaneously separation of salts (mainly monovalent species Na, K, H⁺, and Cl) and concentration. Treatment of milk and UF permeates by NF leads to a demineralization rate around 35% (42% if a diafiltration step is added) and it results in the following claimed benefits (Kelly et al., 1992): reducing costs in con-densing the permeate to 62% TS before crystallization (75% of the water is removed by NF), reducing deposit in the finishing evaporator, and improving the lactose crystallization process (higher yields and less washings of the crystals to reach the wished purity). In addition to this improvement in lactose production, NF is the best solution to convert acid and salty whey to normal whey and consequently solve a disposable environmental problem (Kelly et al., 1992). Spray-drying of acid whey treated by NF showed a significant improvement in running parameters and a three times reduction in the hygroscopicity of the powder (Jeantet et al., 1996).

3.8 APPlICATIONs OF MEMbRANE TEChNOlOGIEs