A. MARTÍNEZ-HERMOSILLA, G.J. HULBERT, W.C. LIAO ABSTRACT:

Transcription

1 JFS: Effect of Cottage Cheese Pretreatment and 2-phase Crossflow Microfiltration/Ultrafiltration on Permeate Flux and Composition A. MARTÍNEZ-HERMOSILLA, G.J. HULBERT, W.C. LIAO ABSTRACT: The effect of chemical pretreatment with calcium chloride and ethylene diamine tetraacetic acid (EDTA) in combination with 2-phase crossflow microfiltration and ultrafiltration was studied. Solutions of untreated and chemically treated whey were microfiltered at 3 flow rates, with 0%, 10%, and 20% air injection. Significant increases (p 0.05) in flux were found for untreated and EDTA treated whey at 10 L/min when 20% air was applied. Microfiltration permeates also had a higher (p 0.05) protein concentration at this flow rate when air was used. Ultrafiltration of the microfiltration permeates resulted in no difference (p 0.05) between the flux of untreated permeate ultrafiltered with 20% air and calcium chloride treated whey permeate with 0% air injection. This suggests that 2-phase flow with air injection could be used to increase the efficiency of both the microfiltration and ultrafiltration of whey and chemical pretreatment could be minimized. Key Words: cheese whey, microfiltration, ultrafiltration, 2-phase flow, pretreatment Introduction THE ADVANTAGES OF PRETREATING CHEESE WHEY TO IMPROVE ultrafiltration (UF) performance during whey protein concentrate (WPC) manufacture and to produce WPC with improved functional and compositional properties is well documented (Kim and others 1989; Rinn and others 1990; Harper 1991; Pearce and others 1991; Daufin and others 1994; Karleskind and others 1995). Pretreatment usually includes ph adjustment of the whey, modification of the mineral content, thermocalcic aggregation of lipoprotein to remove residual whey lipids, and/or microfiltration (MF) prior to ultrafiltration (Lee and Merson 1976a; Lee and Merson 1976b; Kuo and Cheyran 1983; Maubois and others 1987; Morr 1987; Patocka and Jelen 1987; Maubois 1988; Knopp 1992). The presence of calcium and residual whey lipids in the whey not only affects whey product functionality, but also contributes to fouling of ultrafiltration membranes, leading to higher cleaning and operating costs (Merin and Cheyran 1980; Tong and others 1988; Labbé and others 1990; Dauffin and others 1991; Dauffin and others 1993; Mehra and Donnelly 1993). Due to differences in whey composition and operating parameters it has been difficult to optimize pretreatment procedures for the dairy industry. A method to improve UF performance by increasing permeate flux without the use of chemicals would be very beneficial, especially if a functional product can be obtained with lower energy requirements. Cui and Wright (1996) and Bellara and others (1996) reported that gas-liquid 2-phase crossflow reduces the problem of concentration polarization during ultrafiltration of protein solutions. They obtained flux enhancements of 10% to 60% for albumin solutions and concluded that air bubbles increase the mass transfer rate due to a higher velocity in the membrane tubules. The shear force at the membrane wall is increased, reducing fouling, and therefore minimizing the rate of flux decline. Using microfiltration prior to ultrafiltration to remove residual whey lipids, the main objective of this research was to improve permeate flux of cottage cheese whey during these processes by applying chemical pretreatment and/or using air injection to create a 2-phase flow. A secondary objective was to detect differences in the protein permeation of microfiltration permeates for the untreated and chemically treated whey when air was injected. Results and Discussion Effect of pretreatment, flow rate, and 2-phase flow on microfiltration membrane flux rate. There was a difference (p 0.05) among LSmeans for raw untreated whey (RW) at 10 L/min flow rate with no air injection and with the application of either 10% or 20% air (Table 1). Differences (p 0.05) were also found between LSmeans for ethylene diamine tetraacetic acid (EDTA)-treated whey (EW) flux (no air) and 20% air. However, air percentage did not affect permeate flux for CaCl 2 -treated CW. In addition, a significant difference was not found between the injection of 10% and 20% air for RW. The highest mean flux was L/m 2 hr from CW with 20% air. As expected, the lowest mean flux was from RW at 5 L/min without injection of air. Increasing the flow rate from 10 L/min to 20 L/min for RW without injection of air did not affect permeate flux (p 0.05). The effect of 2-phase cross flow microfiltration appears to depend on the level of fouling during processing of cottage cheese whey. Unclarified or raw whey, which contains casein fines, whey proteins, lipids, and minerals among other severe foulants, showed the highest increase in permeate flux when air was applied to the microfilter (52%), as compared to EW (49%) Table 1 Effect of pretreatment, flow rate and air percentage on permeate flux of cottage cheese whey during microfiltration (L/m 2 hr). Flux values are reported as the average of two replicates Flow Rate (L/min) Pretreatment Raw EDTA CaCl 2 Air (%) Air (%) Air (%) e de bc bc d b 23.3 ab a cd * LSmeans with different letters differ (p 0.05) 334 JOURNAL OF FOOD SCIENCE Vol. 65, No. 2, Institute of Food Technologists

2 and CW (28%). This suggests that air is more effective when fouling is the result of severe concentration polarization. The absence of differences (p 0.05) between 10% air in contrast to 20% air suggests that only a low percentage is needed to increase permeate flux of whey solutions. In other words, a small amount of air bubbles provide the necessary shear force to reduce the fouling problem. Cui and Wright (1996) have previously shown that this enhancement effect was more profound when fouling was severe and less significant with turbulent liquid flow. A flow rate of 10 L/min falls into the laminar region for the microfilter in our experiments. Effect of pretreatment and 2-phase flow on the protein permeation of microfiltration permeates Statistical analysis revealed that there were differences (p 0.05) in protein permeation among whey treatments (Table 2). In all cases, air injection improved the protein permeation. Although individual protein permeation was not determined, the values for total protein permeation support the theory that fouling during microfiltration of cottage cheese whey is probably due to irreversible concentration polarization, in which proteins form sheets that affect the microfiltration rate and their permeation. Both, -lactalbumin and -lactoglobulin are much smaller than the 0.1 m pore size of the MF membrane. Nevertheless, these proteins did not permeate freely. Many factors could have contributed to the relatively low protein permeation values. The fouling phenomena was greatest during MF of untreated cottage cheese whey, resulting in the lowest protein permeation value. In addition, it is known that flux rates can be improved not only by pretreatment modifications, but also by processing at higher temperatures, allowing higher permeation rates for whey proteins (Knopp 1992). The processing temperature was maintained at 20 C for all experiments to eliminate temperature as a variable and to reduce energy requirements for the overall process. Therefore, lower values than those reported in the literature were expected. There could also be an effect of ph and ionic strength on protein permeation. Protein permeation was improved when processing at higher ph. Higher values for protein permeation were obtained for CW, where the pretreatment resulted in a final ph of 7.1. According to Marshall and others (1993), maximum deposition of proteins occurs around the isoelectric point. Most of the whey proteins have their isoelectric point near 5.0, where they are least soluble and more susceptible to aggregation. Since both, the membrane and the state of the protein affect the degree of protein deposition, an increase of the ph away from the isoelectric point probably resulted in less deposition. This pretreatment also resulted in the removal of lipoproteins. As expected, the removal of lipoproteins resulted in an increased relative permeation percentage of the remaining proteins. Researchers have observed that the presence of large molecules in a solution can increase the retention of smaller molecules. Large molecules traveling through the narrow, tortuous confines of the porous membrane matrix are slowed down by friction with the pore wall and hinder the transport of smaller molecules. (Marshall and others 1993). Bovine serum albumin (BSA) and -lactoglobulin are sheet forming proteins that hinder permeation. Studies have shown that BSA accumulates not only at the membrane surface but also within the pores of membranes (Dauffin and others 1991). These researchers reported a retention of 100% for BSA and Ig, 50% for lactoferrin and lactoperoxidase, and approximately 30% for -lactoglobulin and -lactalbumin during MF of untreated whey obtained from rennet casein production when using an inorganic membrane. BSA is also known to interact with other whey proteins, such as lactoferrin, lysozyme, or -lactalbumin, forming Table 2 Effect of pretreatment and air injection on the protein permeation of whey proteins. Permeation ratios are reported as the average of 2 replicates Air % Pretreatment Raw EDTA CaCl f e d c b a *LSmeans with different letters differ (p 0.05) complexes that could be larger than the molecular weight cut off of the membrane under study. Also, -lactoglobulin has a tendency to polymerize below ph 8.0. All these factors contribute to the low permeation ratios during microfiltration. Lee and Merson (1976a) previously stated that permeation could be improved if these proteins were maintained in a disperse state without allowing them to deposit on the membrane. Previous studies showed that air sparging disrupts the concentration polarization boundary layer during UF, enhancing permeate flux but reducing the sieving coefficient of BSA solutions at concentrations of 2 g/l (Bellara and others 1996). Our results indicate that despite the low permeation rates during the MF of cottage cheese whey, application of air enhanced permeate flux by decreasing membrane fouling, leading to significantly higher protein permeation values. Effect of pretreatment and 2-phase flow on ultrafiltration membrane flux rate Injection of 20% air increased the permeate flux for each of the feeds used in the experiment (Table 3). In addition, applying 20% air while processing microfiltration permeate of RW is as effective as processing microfiltration permeate of CW without air. Therefore, flux enhancement could result in a reduced processing time for the 2-phase cross-flow mode of operation without chemical pretreatment. These results apply to the conditions of this experiment, in which the operating parameters were kept the same for all runs and replicates. As expected, flux enhancement was more significant when processing microfiltration permeate from RW (34.3%) as compared to CW (10%). As mentioned before, air seems to disrupt the concentration polarization layer formed by protein solutions, which seems to be more pronounced when processing raw whey. The air provides shear force to prevent and disrupt the fouling. Some air also passes through as permeate and thus may help to inhibit clogging of pores. There was no difference between the UF membrane fluxes obtained when processing RW and EW microfiltration permeates without injection of air. Therefore, the EDTA pretreatment modification did not have an enhancing effect during ultrafiltration. As stated before, no significant difference was observed between the flux obtained for RW microfiltration permeate (with 20% air) and CW microfiltration permeate (with no air), with values of L/m 2 hr and L/m 2 hr, respectively. During microfiltration experiments, these 2 permeates reflected similar values for protein permeation, with RW MF permeates (20% air) being significantly higher than CW MF permeates (no air). These results are important and suggest that the calcium chloride pretreatment modification could be replaced in the future by the sole application of air during processing of cottage cheese whey. One important factor to be considered would be to monitor the individual protein permeation obtained for these two MF processes to ensure that -lactoglobulin and -lactalbumin are present in similar proportions to ensure functional properties of WPCs. MF permeates should be analyzed by size exclusion high performance liquid chromatography (SE-HPLC) since they owe Vol. 65, No. 2, 2000 JOURNAL OF FOOD SCIENCE 335

3 Effect of Cottage Cheese Pretreatment... Table 3 Effect of pretreatment and air injection on the ultrafiltration permeate flux (L/m 2 hr). Flux is reported as average of two replicates Air (%) Pretreatment Raw EDTA CaCl d d b b c a *LSmeans with different letters differ (p 0.05) most of their functional properties to these proteins, followed by BSA. In addition, SE-HPLC could be used to monitor and compare the removal of phospholipoproteins during microfiltration of untreated and calcium chloride treated whey. These residual whey lipids continue to be the cause of membrane fouling during whey processing and also have a detrimental effect on WPCs, since they greatly affect their functional properties (Joseph and Mangino 1988). Rinn and others (1990) had previously studied several pretreatment modifications for cheese whey and found that WPCs made from calcium chloride treated whey and processed with a 0.6-mm pore size microfilter had the best foaming and gelation properties when compared to nine other pretreatments. For future research, it would be advisable to study functional properties of WPCs produced with untreated whey and the 2-phase cross-flow mode of operation, since some foaming problems were encountered during our research. As stated by Harper (1991), multiple functional properties are usually necessary for most food systems, however foaming is not required for all applications. If foaming during processing reduced the foaming capacity and stability of the most important whey proteins ( -lactoglobulin and -lactalbumin), specific applications should be determined for these WPCs. However, a study made by Phillips and others (1995), revealed that the unfolding of -lactoglobulin during foaming is reversible after this protein has been subjected to shear and heat. According to Bellara and others (1996), current studies have shown that gas sparging does not deteriorate enzyme activity either. Nevertheless, further study is suggested in relation to the whey protein configuration and the use of air. From a processing perspective, there should be a gas liquid separator when processing with air, as suggested by Bellara and others (1996). Proteins in the foam could be recovered in the separator, since the foam collapses. Effect of air and pretreatment on individual whey proteins. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) was used to analyze protein composition as affected by the different pretreatments and under the influence of air. Individual protein bands were identified from MF/UF permeates and retentates of treated and untreated whey taken after one hour of filtration (Fig. 1). A 12% polyacrylamide gel, showed weaker bands for all microfiltration permeates when compared to the original whey, in which the bands of the major whey proteins were easily identified. In addition, no higher molecular weight proteins, such as BSA and Ig were present after the microfiltration process. This confirms the results obtained by Dauffin and others (1991), who observed a 100% retention of BSA during microfiltration. This is consistent with the low protein permeation ratios. There are no bands present for UF permeates, because the proteins are retained and concentrated in the retentate during this process. The same proteins are present for the cases in which air was applied to the membranes. It appears that a 2-phase cross-flow method of operation does not have any effect on the whey protein s molecular weight, but a positive effect in a-lactalbumin and b-lactoglobulin permeation. Fig. 1 SDS-PAGE of MF and UF whey solutions. (A) RW. (B) Standards. (C) MF permeate from RW. (D) Supernatant from CW pretreatment. (E) MF permeate from CW. (F) UF permeate from CW (air). (G) UF permeate from CW. (H) MF permeate from EW. (I) MF permeate from EW (air). (J) UF permeate from RW. (K) UF retentate from RW. (L) RW. (M) MF permeate from RW. Conclusions THE PROTEIN LOSSES DURING MF EXPERIMENTS REFLECTED AS low protein permeation values would affect the final protein concentration of WPCs. It would be necessary to find potential uses for MF retentate fractions with high levels of proteins such as BSA and Ig, as revealed by SDS-PAGE. At the same time, further study is needed to optimize the operating parameters for the microfiltration process, in particular, temperature, flow rate and pressure, as well as MF membrane pore size and type to reduce protein losses to those caused by the removal of lipoproteins. Materials and Methods Unclarified cottage cheese whey was obtained from Purity Dairy (Nashville, Tenn., U.S.A.), immediately after cheese manufacture, and transported to the Department of Food Science and Technology at the University of Tennessee. The whey was transferred to clean buckets filled to approximately 10 to 15 L, which were stored in a freezer at 20 C until ready to process. The average composition of the whey is shown in Table 4. Chemical pretreatment The whey was thawed at 20 to 25 C the night before processing. A portion of the whey was chemically pretreated with EDTA (Sigma Chemicals, St. Louis, Mo., U.S.A.) until it 336 JOURNAL OF FOOD SCIENCE Vol. 65, No. 2, 2000

4 reached a concentration of M. The initial ph of the whey was 4.4 and was adjusted to 4.1 with 6N HCl before microfiltration. A 2 nd portion of the whey was pretreated by cooling to 2 to 5 C, adding 1.2 g of CaCl 2 (Sigma Chemicals, St. Louis, Mo., U.S.A.) per liter of whey, adjusting the ph to 7.3 with 6N NaOH and heating to a hold temperature of 55 C for 8 min by partially immersing 2 5 L beakers in a water bath at 95 C (Precision Scientific, Model 25, Chicago, Ill., U.S.A.). The treated whey was cooled to 20 C and transferred to the feed tank for microfiltration. Table 4 Composition of cottage cheese whey Component Total solids (%) Moisture (%) Protein (%) (N x 6.38) Fat (%) Ash (%) Calcium (mg/g) a average from six determinations b average from four determinations Cottage Cheese 6.32 a a 0.75 b 0.02 b 0.60 a 1.36 b Processing The experimental setup is shown in Fig. 2. A variable speed peristaltic pump (MasterFlex, Model # , Cole- Parmer Instrument Co., Vernon Hills, Ill., U.S.A.) was used to pump whey and permeate solutions through the microfilter and the ultrafilter, respectively. The different elements were connected by Tygon B-44-4X tubing. Horizontal cross-flow microfiltration was carried out using a 0.1-mm pore size, tubular hollow fiber polysulfone microfilter with 0.15 m 2 membrane area (A/G Technology Corporation, Needham, Mass., U.S.A.). This process was used to replace the clarification step needed before UF, and also to remove residual whey lipids from raw untreated whey (RW), EDTA treated whey (EW), and CaCl 2 treated whey (CW). The membrane was thoroughly cleaned before each run. All runs used 10 L of whey, and the same membrane was used for each run. A permeate flux of pure water was used to check the effectiveness of cleaning. This is important to consider when comparing permeate flux after each whey pretreatment. Warm water (50 C) was used to wash the membrane for 20 min in a non-recycling mode. A 0.5 N solution of NaOH at 50 C was recycled for 40 min, followed by flushing with warm water for another 40 min. This procedure was recommended by the membrane manufacturer. All runs were performed at 20 C and 0.69 bar (10 psi) average transmembrane pressure (TMP). An air pump (Gast Model # DOA-P104-AA, Benton Harbor, Mich., U.S.A.) was used to inject air bubbles into the liquid flow and achieve a 2-phase cross flow during MF. The air was dispersed in the form of very small bubbles. Three air percentages (0%,10%, and 20%) were set by monitoring the air flow in the rotometer and the air pressure. The air pressure (P air ), air flow rate (V air ), and 2-phase flow pressure (P TPF ) were required to calculate the air percentage. The liquid flow rate was measured as V liq, and the air flow rate was calculated by applying the ideal gas assumption: the liquid flow at 0% and 20% to see the effect of 2-phase cross flow on the UF permeate flux. Cleaning procedures between runs consisted of flushing with clean water for 20 min, recycling a 0.5 N solution of NaOH for 20 min, flushing again with clean water for 15 min, circulating a solution of 100 ppm of NaOCl for 1 h, and rinsing with clean water for 20 min. This procedure was recommended by the manufacturer. The goal was to recover the water permeate flux of 250 ml/min at 25 C and 0.69 bar TMP, reported by the manufacturer. Chemical analyses Total solids were determined using an atmospheric oven drying method (AOAC, 1980). Five grams of sample were dried in a vacuum oven (Baxter, Scientific Products, Model # , Atlanta, Ga., U.S.A.) at 100 C and atmospheric pressure for 5 h. After cooling the samples in a desiccator, they were re-weighed. The ash content of cottage cheese whey was determined by a modified AOAC (1980) method. Dried samples from the total solid determination were ashed in a furnace chamber (Thermolyne Corporation, Model # F-A1730, Dubuque, Iowa, U.S.A.) at 625 C over 16 h. Calcium was determined by wet digestion, using a modified atomic absorbance spectroscopy method (Pollman 1991). samples were brought to room temperature, thoroughly mixed and pipetted into previously weighed beakers in 2-g portions. Sixty ml of HNO 3 were added, and the samples were placed on a preheated hot plate, located inside a perchloric acid approved fume hood. The samples were al- P air V air P TPF V TPF where V TPF the air flow in the two phase flow. The air percentage was then calculated based on the ratio of V TPF to (V TPF V liq ). Permeate fluxes were compared for the treated and untreated whey at three flow rates, with and without application of air. Microfiltration permeates (15 to 20 L) obtained from raw, EDTA, and CaCl 2 treated whey were later ultrafiltered in a 10,000 nominal molecular weight cut off (MWCO) hollow fiber polysulfone membrane with 0.28 m 2 of membrane area (A/G Technology Corporation, Needham, Mass., U.S.A.). Operating parameters were 20 C, 10 L/min flow rate and 0.69 bar TMP. The purpose was to compare permeate fluxes for the different microfiltrates, with and without air. Air was introduced to Fig. 2 Experimental Setup for Microfiltration and Ultrafiltration. 1 = inlet pressure gauge (P inlet ). 2 = outlet pressure gauge (P outlet ). 3 = cross-flow microfilter/ultrafilter. 4 = back pressure control valve. 5 = cheese whey reservoir. 6 = foam collector. 7 = permeate collector. 8 = peristaltic pump. 9 = air pump. 10 = air flow rate control valve. 11 = check valve. 12 = pressure gauge (P a ). 13 = air flow rate control valve. 14 = rotometer. Vol. 65, No. 2, 2000 JOURNAL OF FOOD SCIENCE 337

5 Effect of Cottage Cheese Pretreatment... lowed to boil until the nitric acid was almost dry. Then, they were cooled, 7 ml of perchloric acid (70%) were added, and they were brought back to boil until 1 to 2 ml of perchloric acid remained. Samples were removed from the hot plate, and cooled at room temperature. A series of dilutions were made, until the samples fit the concentration range used on the atomic absorption unit. A 0.1% lanthanum oxide solution (a phosphate inhibitor) was included in the last dilution, and the samples were analyzed on an Atomic Absorption and Atomic Emission Spectrophotometer (Allied Analytical Systems, Wathan, Mass., U.S.A.). Calcium absorption was determined at nm. A method modified from the work of Bligh and Dyer (1959) was used to determine total lipids. Prior to lipid extraction, whey samples were thawed in a water bath (25 C) and throughly mixed. A 20 ml sample was transferred into a 60 ml stainless steel container, and 30 ml of methanol were added. The mixture was homogenized at low speed for 1 min with a Virtis Model 23 homogenizer (The Virtis Company Inc., Gardiner, N.Y., U.S.A.). Subsequently, 20 ml of chloroform were added, and the mixture was homogenized at low speed for 1 min. After addition of 20 ml of an aqueous zinc acetate solution (0.115 g zinc acetate/5 ml) and homogenization for 10 seconds, the mixture was filtered through Whatman No 1 filter paper (Fisher, Pittsburg, Pa., U.S.A.). The permeate was recovered, and the flask was rinsed with 10 ml chloroform, which was added to the permeate. The homogenization and filtration steps were repeated. The homogenizer flask and filter were rinsed with 10 ml chloroform, and the mixture was poured into a 100 ml graduated cylinder after rinsing the homogenizer flask with 2 ml of methanol. The cylinder was covered, kept in a cooler at 4 to 6 C, until the 2 phases were clearly separated (24 h), and the volume of the lower layer (chloroform) was recorded. The cylinder contents were poured into a 250 ml separatory funnel, and allowed to set in a cooler (4 to 6 C) for 2 h. The bottom layer was drained into a 50 ml Erlenmeyer flask, and a 10 ml aliquot of the extract was pipetted to a dry beaker and allowed to evaporate to dryness overnight. The next day, the beaker was dried at 400 C in an oven for 30 min (Will Corporation, Rochester, N.Y., U.S.A.), cooled in a dessicator, and re-weighed. A modified AOAC (1984) Kjeldahl method was used to determine the total protein content of cottage cheese whey samples. Liquid 10 ml aliquots were pipetted into a Kjeltec digestion tube, followed by addition of 2 Keltabs (CuSO 4 ),15 ml of H 2 SO 4 and mixing. The sample was placed in a Digester DS 6/20 (Tecator Analytical Company, Höganäs, Sweden) and allowed to digest under a hood by slowly increasing the temperature to 420 C over a period of 6 h to avoid splattering. After the sample was cooled, it was placed in a Kjeltec 1026 Distilling Unit (Tecator Analytical Company, Höganäs, Sweden), and 25 ml boric acid solution were added to a receiver flask. The receiver flask was placed on the platform of the distilling unit, and the distillation procedure ran automatically. Titration of the receiver flask solution was made to neutral gray endpoint with 0.1 N HCl, and the volume of acid was recorded. The percent protein was computed as %N Protein permeation was calculated as the ratio between protein concentration in the permeate (Pp) and protein concentration in the feed (Po). Sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE) was used to fractionate and characterize whey proteins from raw, pretreated whey, MF permeates, and UF retentates processed with and without the application of air. An equal volume (100 L) of whey sample and 2X sample buffer (12.5% 0.5 M Tris ph 6.8, 4% SDS, 20% glycerol, 0.02% bromophenol blue solution, 10% mercaptoethanol and 12.5% water) was mixed, heated for 10 min at 95 C and cooled to 20 to 25 C. Twenty L of sample, the blank, and a wide range molecular weight (6500 to 205,000 D) Sigma Marker Standard (Sigma Chemicals, St. Louis, Mo., U.S.A.) was loaded into a 12% polyacrylamide gel (Greaser and others 1983). A Hoefer SE 600 Series vertical slab gel electrophoresis unit (Hoefer Scientific Instruments, Model PR 70/ 75, San Francisco, Calif., U.S.A.) connected to an EC 3000 P Series 90 Programable power supply unit (EC Apparatus Corporation, St. Petersburg, Fla., U.S.A.) was used. Gels were stained overnight in 0.025% Coomassie Brilliant Blue R250, 50% methanol, and 9.2% acetic acid. Subsequently, gels were destained for 24 to 48 h in a solution containing 10% methanol and 7.5% acetic acid. A Randomized Block Design (RBD) blocked on whey with a fractional factorial treatment arrangement (from 3 types of pretreatment, 3 levels of flow rate and 3 levels of air) was used to find the best combination of flow rate, air percentage, and pretreatment method to achieve the highest permeate flux after 1 hour of operation in the microfilter. This statistical design did not include all possible treatment combinations, but only those necessary to estimate main effects. Data were analyzed with weighted analysis of variance using Proc Mixed (SAS Inc., Cary, N.C., U.S.A.). The pdmix612.sas algorithm from the SAS software package (Saxton 1998) was used to generate the letter group separation of the significant differences among treatments. A RBD with a factorial treatment arrangement (3 types of whey pretreatment by 2 levels of air) was used as the statistical design to find significant differences in protein permeation among the treatments, after 1 h of operation in the microfilter. Data were analyzed by Proc Mixed (SAS Inc., Cary, N.C., U.S.A.). The pdmix612.sas algorithm was used to generate the letter group separation of the significant differences among the treatments and interactions. A Complete Randomized Design (CRD) with a factorial treatment arrangement (3 types of whey pretreatment by 2 levels of air) was used as statistical design to detect significant differences in the permeate flux during the ultrafiltration experiments. Data were analyzed by Proc Mixed, and the letter groups of the significant differences among the treatments and for interactions were generated by pdmix612.sas. References AOAC Official Methods of Analysis. Washington, DC.: Association of Official Analytical Chemists. AOAC Official Methods of Analysis. Washington, D.C.: Association of Official Analytical Chemists. Bellara SR, Cui ZF, Pepper DS Gas sparging to enhance permeate flux in ultrafiltration using hollow fibre membranes. J. Membrane Sci. 121: Bligh EG, Dyer WJ A rapid method of total lipid extraction and purification. Can. J. Biochem. Physiol. 37: JOURNAL OF FOOD SCIENCE Vol. 65, No. 2, 2000 Cui ZF, Wright KIT Flux enhancements with gas sparging in downwards crossflow ultrafiltration: performance and mechanism. J. Membrane Sci. 117: Daufin G, Labbe J-P, Quemerais A, Michel F Fouling of an inorganic membrane during ultrafiltration of defatted whey protein concentrates. Neth. Milk Dairy J. 45: Daufin G, Labb J-P, Quemerais A, Michel F, Merin U Optimizing clarified whey ultrafiltration: Influence of ph. J. of Dairy Res. 61: Greaser ML, Yates LD, Krzywicki K, Roelke DL Electrophoretic methods for the separation and identification of muscle proteins. Proc. 36 th Ann. Rec. Meat Conf., p Harper WJ Functional properties of whey protein concentrates and their relationship to ultrafiltration. In: New Applications of membrane Processes. Special Issue No 9201.

6 International Dairy Federation. Brussels, Belgium: IDF. Joseph MSB, Mangino ME The effects of milk fat globule membrane protein on the foaming and gelation properties of b-lactoglobulin solutions and whey protein concentrates. Aust. J. Dairy Technol. 43: Karleskind D, Laye I, Mei F-I, Morr CV Chemical pretreatment and microfiltration for making delipidized whey protein concentrate. J. Food Sci. 60: Kim S-H, Morr CV, Seo A, Surak JG Effect of whey pretreatment on composition and functional properties of whey protein concentrate. J. Food Sci. 54: Knopp TK Lipoprotein removal from cheese whey by cross-flow microfiltration. M. S. Thesis. The Ohio State University, Columbus, OH. Kuo K-P, Cheyran M Ultrafiltration of acid whey in a spiral-wound unit: Effect of operating parameters on membrane fouling. J. Food Sci. 48: Labbé J-P, Quemerais A, Michel F, Daufin G Fouling of inorganic membranes during whey ultrafiltration: Analytical methodology. J. of Membrane Sci. 51: Lee DN, Merson RL. 1976a. Chemical treatments of cottage cheese whey to reduce fouling of ultrafiltration membranes. J. Food Sci. 41: Lee DN, Merson RL. 1976b. Prefiltration of cottage cheese whey to reduce fouling of ultrafiltration membranes. J. Food Sci. 41: Marshall AD, Munro PA, Trägardh G The effect of protein fouling in microfiltration and ultrafiltration on permeate flux, protein retention and selectivity: A literature review. Desalination 91: Maubois JL, Pierre A, Fauquant J, Piot M Industrial fractionation of main whey proteins. In Bull Trends in Utilization, p Int. Dairy Fed., Brussels, Belgium. Maubois JL : Its biotechnological signification. 8 th International Biotechnology Symposium, Paris. p Mehra RK, Donnelly W Fractionation of whey protein components through a large pore size, hydrophilic, cellulosic membrane. J. of Dairy Res. 60: Merin U, Cheyran M Factors affecting the mechanism of flux decline during ultrafiltration of cottage cheese whey. J. of Food Proc. Pres. 4: Morr CV Effect of HTST pasteurization of milk, cheese whey and cheese whey retentate upon the composition, physicochemical and functional properties of whey protein concentrates. J. Food Sci. 52: Patocka J, Jelen P Calcium chelation and other pretreatments for flux improvement in ultrafiltration of cottage cheese whey. J. Food Sci. 52: Pearce RJ, Marshall SC, Dunkerley JA Reduction of lipids in whey protein concentrates by microfiltration-effect on functional properties. In: New Applications of membrane Processes. Special Issue No International Dairy Federation. Brussels, Belgium: IDF. Phillips LG, Hawks SE, German JB Structural characteristics and foaming properties of b-lactoglobulin: Effects of shear rate and temperature. J. Agric. Food Chem. 43: Pollman RM Atomic absorption spectrophotometric determination of calcium and magnesium and colorimetric determination of phosphorus in cheese: Collaborative study. J. Assoc. Off. Anal. Chem. 74: Rinn J-C, Morr CV, Seo A, Surak JG Evaluation of nine semi-pilot scale whey pretreatment modifications for producing whey protein concentrate. J. Food Sci. 55: SAS Institute Inc SAS/STAT Software: Release Cary, NC: SAS Institute. Saxton A, A macro for converting mean separation output to letter groupings in Proc Mixed. In: Proc. 23 rd SAS Users Group Intl. Cary, NC: SAS Institute. p Tong PS, Barbano DM, Jordan WK Permeate flux during ultrafiltration of whey: Influence of milk coagulant used for cheese manufacture. J. Dairy Sci. 71: MS received 6/1/99; revised 9/24/99; accepted 12/3/99. Author Martínez-Hermosilla is with the Department of Food Science and Technology, University of Tennessee, Knoxville, TN Author Hulbert is with the Department of Food Science and Technology & Biosystems and Agricultural Engineering, University of Tennessee, Knoxville, TN Author Liao is with the Department of Biosystems and Agricultural Engineering, University of Tennessee, Knoxville, TN Vol. 65, No. 2, 2000 JOURNAL OF FOOD SCIENCE 339