Desalination 278 (2011) 381–386

Contents lists available at ScienceDirect

Desalination

j ourna l homepage: www.e lsev ie r.com/ locate /desa l

Concentration and purification of whey proteins by ultrafiltration

C. Baldasso ⁎, T.C. Barros, I.C. TessaroLaboratory of Membrane Separation Processes Departamento de Engenharia Química, Universidade Federal do Rio Grande do Sul R. Eng. Luiz Englert, s/n. CEP: 90040-040, Porto Alegre, RS, Brazil

⁎ Corresponding author. Fax: +55 51 3308 3983.E-mail address: cbaldasso@gmail.com (C. Baldasso).

0011-9164/$ – see front matter © 2011 Elsevier B.V. Aldoi:10.1016/j.desal.2011.05.055

a b s t r a c t

a r t i c l e i n f o

Article history:Received 26 November 2010Received in revised form 5 May 2011Accepted 23 May 2011Available online 23 June 2011

Keywords:WheyProteinConcentrationPurificationUltrafiltration

Whey is a liquid by-product of the dairy industry produced during themanufacture of cheeses and casein. As araw material, it has many applications in food technology due to the functional and nutritional properties ofits proteins. Membrane technology, especially ultrafiltration (UF), has been used in the dairy industry toproduce whey-protein concentrates, because this technology allows the selective concentration of theproteins in relation to the other components. In this context, the objective of this work was to concentrate andto purify the whey proteins using UF in association with discontinuous diafiltration (DF). The two strategieswere tested by changing the volumetric-concentration factor (VCF), the DF water volume and the number ofDF steps. The results showed that the UF process is adequate for the production of protein concentrates; in thebest experimental strategy, the protein concentrate obtained was greater than 70% by weight (dry basis).

l rights reserved.

© 2011 Elsevier B.V. All rights reserved.

1. Introduction

Whey, the by-product of cheese or casein production, is of relativeimportance in the dairy industry due to the large volumes producedand the nutritional composition; the production of 1–2 kg of cheeseyields 8–9 kg of whey. Whey contains more than half of the solidspresent in the original whole milk, including whey proteins (20% ofthe total protein) and most of the lactose, minerals, water-solublevitamins and minerals [1,2].

Worldwide whey production is estimated at around 180 to190×106 ton/year; of this amount only 50% is processed. The wheycan be considered a valuable by-product with several applications inthe food and pharmaceutical industries; however, it is often treated asa dairy wastewater. The treatment of whey represents a seriousproblem due to its high organic load, which can reach a chemicaloxygen demand (COD) of 100,000 mg O2 L−1 [3,4].

Many techniques have been developed to selectively concentratewhey proteins, because thewhey is not a balanced source of nutrients,containing a high lactose content compared to the protein, and thusdoes not have the nutritional benefits of more typical protein sources.

Whey proteins have a high nutritional value, due to the highcontent of essential amino acids, especially sulfur-containing ones [5].Besides the nutritional properties, the whey proteins have functionalproperties which impart beneficial physical properties when used asingredients in food, mainly due to its high solubility, water absorption,gelatinization and emulsifying capacities [6]. Due to the higherspecificity of the product, and the excellent functional and nutritional

value, the commercial value of whey protein concentrate is from 3 to40 times greater than that of whey powder [3].

The conventional method of whey concentration is by thermalevaporation. The main disadvantages of this method are the highenergy consumption and the high content of ashs and lactose thatremains in the concentrate; additionally, the heat treatment canchange the characteristics of whey components, mainly the proteins,which are thermolabile and can lose their nutritional and functionalproperties during heating. Ultrafiltration (UF) is a very attractivealternative method, as it does not use heat and as a consequence doesnot involve a phase change, which makes the concentration processmore economical. UF is a membrane separation process (MSP)typically used to retain macromolecules, and has been used in thedairy industry in the recovery and fractionation of milk components.UF allows a variation in the ratio of concentration between the wheycomponents, due to the retention of protein and selective permeationof lactose, minerals, water and compounds of low molar mass [7].

Diafiltration (DF) is used for the production of whey-proteinconcentrate (WPC) with a high protein content. DF is used for proteinpurification to eliminate problems association with high concentra-tions in the retained product, generating high purification, whileretaining good performance process [8]. Also, it should be pointed outthat the addition of small DF volumes several times is more effectivethan a big volume at one time only.

The operability studies provide important information about thecapabilities and limitations of the whey ultrafiltration process. Thewhey ultrafiltration process is well designed to deliver the desiredtotal solids and protein concentrations for the production of wheyprotein concentrates. However, the process becomes less capable ofdelivering the desired product specifications after long hours ofoperation when long-term fouling is more significant [9,10].

382 C. Baldasso et al. / Desalination 278 (2011) 381–386

Further studies involving ultrafiltration operated diafiltrationmode are needed to reveal diafiltration volumes that are effective.More than that, which may take into account what level and howmany cycles diafiltration are needed, one can optimize the work. Thedesired total solids and protein concentrations cannot be achieved ifthe volume concentration ratio alone is used to control the flowrateand composition of the product stream.

The limitations of the whey ultrafiltration process highlight thenecessity for direct monitoring of the total solids and proteinconcentration in the product stream during the production of wheyprotein concentrates. Monitoring of total solids and protein concen-tration also helps determine how the amounts of diafiltration waterand recycled permeate need to be adjusted to achieve the desiredproduct flowrate and composition from the ultrafiltration plant. Suchadjustments are necessary either after deviations from the desiredspecifications are observed or after changes in the desired specifica-tions are made.

Furthermore, the development of technologies capable of solvingthe problem of whey utilization can bring economic and environ-mental benefits as the whey is a good source of protein for humanconsumption, thus justifying and studying the possibility of using itcommercially. In this context, the aims of the present work were theconcentration and purification of whey proteins by UF associated withDF. In this work, different strategies were tested, including varying thevolume of water added to the concentrate and the number of DF stepsand modifying the volumetric concentration factor (VCF; the ratio ofthe initial volume of the whey solution and volume retained). Thus,the main objective is to obtain a large protein purification, with theminimum DF solution (in this case water).

2. Materials and methods

2.1. Whey

The sweet whey powder used in this study was supplied by ElevaAlimentos (Teutonia, RS), from the manufacture of mozzarella cheese,

Fig. 1. Schematic of the membrane unit used for the experiments. (1) tank, (

with a total solids content of approximately 6%. Liquid whey wasreconstituted by manually dissolving the whey powder in distilledwater at neutral pH and a temperature of 50 °C. The initial volume ofwhey for ultrafiltration was approximately 30 L (29.6 L water and1.86 kg whey powder). The average initial contents of lactose, proteinand ashs were 42 kg m−3 (72.4% - w/w), 9 kg m−3 (15.6%) and7 kg m−3 (12%), respectively; the amount of fat was considerednegligible, since it is removed before the whey was dried in a spraydryer.

2.2. Membrane

The UF membrane was UF-6001, made of polyethersulfone, in aspiral module manufactured by Koch Membrane Systems. The molarweight cut-off was 10 kDa, the feed spacer was 80 mils (thousandthsof inch) and permeation area was 0.28 m2.

2.3. UF equipment

Experiments were performed in a pilot plant, WGM-KOCHPROTOSEP IV, shown schematically in Fig. 1.

The pilot plant comprises the following equipments:

feed tank (1), stainless steel, with a volume of 75 L, manufacturedby SULINOX. The tank has an agitator and a temperature-controlsystem that operates in the range of 25 to 150 °C;pneumatic pump (2), diaphragm type, model Versamatic VM50,operated with compressed air through a system comprising an FLRkit (filter, air regulator and lubricator);pre-filter (3), manufactured by CUNO, consisting of a PVC housingand a polypropylene filter element with a nominal pore size of1 μm;housing for module spiral membrane (5), 30 cm in length and5.8 cm in diameter, of 316 stainless steel;manometers (4) and (6), 316 stainless steel, scale from 0 to10.5 bar.

2) pump, (3) pre-filter, (4) and (6) manometers, (5) membrane module.

Fig. 2. Water and whey flux vs transmembrane pressure. Membrane UF-6001, T=50 °C,feed flowrate=840 L h−1. Legend: (♦) water, (●) whey. *coefficient of variation: ±0.5%.

383C. Baldasso et al. / Desalination 278 (2011) 381–386

2.4. Analytical methods

The concentration and purification characteristics of concentrateand permeate samples were determined by analysis of the followingparameters: concentrations of protein and lactose, total solidscontent, electrical conductivity and pH.

Determination of total solids content was carried by a gravimetrictechnique in accordance with the procedures given in LANARA [11].The concentration of lactose was determined by the dinitrosalicylicacid (DNS) method by Miller [12]. Protein concentration wasdetermined by the Lowry method [13]. These methods are spectro-photometric and the absorbance readings were performed in a UV-visible spectrophotometer (Varian Cary I).

pH analysis was with a Digimed DM20 pH meter. Electricalconductivity was determined with a Digimed - DM31 equipped withelectrode model DMC-010M.

2.5. Experimental

The first experiments were performed by measuring the permeateflux at different transmembrane pressures (ΔP) with water and withwhey.

The UF permeation tests were carried out in total recirculationmode (for pressure selection) and in concentration mode. In the totalrecirculation mode, both concentrate and permeate, are recirculatedto the feed tank, to achieve homogenization of the solution andthermal equilibrium in the system. In the concentration mode onlythe concentrate is recirculated to the feed tank. The concentration-mode experiments were performed at a transmembrane pressure of2 bar and a feed flowrate of 840 L h−1; these operating conditionswere determined in previous studies [14]. The whey temperature waskept constant at 50 °C, based on the temperature output of the wheyin the cheese manufacturing process and the maximum allowabletemperature for the membrane (55 °C).

In all experiments, measurements of water flux were carried outbefore and after ultrafiltration with whey to quantify the formation offouling on the membrane.

Two experiments were conducted with different strategies to verifywhich condition would allow for greater purification of proteins. Theexperiments consisted of one step of UF in batchmode, where the initialvolume was reduced to a certain VCF. And another stage, it operatedultrafiltration in discontinuous diafiltration mode, which was to addsome incremental volumes ofwater distilled to concentrate and removethis added volume, in the permeate, to remove most of the lactose andother low molecular weight compounds, increasing the degree proteinpurification.

The experiments were made as are described below.

Experiment 1 — the inicial whey solution (30 L) was concentratedto a volume of 6 L in the feed tank (VCF=5); subsequently, five DFsteps were done: two DFs (DF1 and DF2) with 6 L of distilled watereach and three DFs (DF3, DF4 and DF5) with 3 L of distilled watereach, five cycles of DF, totaled 21 L of water, were added in theconcentrate, this order: 6 L+6 L+3 L+3 L+3 L.*Experiment 2 — the inicial whey solution (30 L) was concentratedto a volume of 5 L (VCF=6); subsequently, four DF steps weredone: DF1 and the DF2with 5 L of distilled water each, DF4 and theDF5 with 2.5 L of distilled water each. Four diafiltration cycles ofdistilledwaterwere added in the concentrate, so: 5 L+5 L+2.5 L+2.5 L, in total were added during the diafiltration 15 L of water.*

*Each DF cycle is equivalent to add the volume of distilled waterindicated in each experiment in the concentrate, and to keep constantthe volume of concentrate.

The membrane was not cleaned between the concentration andpurification stages, i.e., the process was continuous.

A chemical cleaning procedure was performed at the end of eachexperiment to restore the flux and retention characteristics of themembrane and prevent the growth of microorganisms in the system.The cleaning consisted of the following steps: a rinse with distilledwater, an alkaline cleaning, an alkaline chlorine cleaning and an acidcleaning. These steps were always within the limits of pH andtemperature tolerance of the membrane.

3. Results and discussion

Fig. 2 shows the water-permeate and whey-permeate fluxes asfunctions of transmembrane pressure; observe that the flux of waterincreased linearly with transmembrane pressure (r2=0.9982).Moreover, for same operating conditions there was a significantdifference in the whey and water-permeate fluxes. This samebehavior was found by Rektor and Vatai [1], Atra et al. [2] andButylina et al. [15] in their works, i.e., the whey permeate flux waslower than the water flux at all pressures. Possible causes for thesedifferent fluxes include lower interactions between the membraneand solution, effects of mass diffusivity and mainly the higher viscosity(μ) of the solution.

The whey-permeate flux being smaller than the water flux showsthat the concentration-polarization effect is very significant for wheyin the initial concentration and this effect tends to increase as thewhey is being concentrated. In this case, we can say that the increaseof flux is limited by an increased polarization layer, i.e., the increase oftransmembrane pressure is counterbalanced by the increase in totalresistance.

In the plots the start of UF is identified by UFi, and the end by UFf.Diafiltrations are identified by DF and the corresponding stage (1, 2, 3,4 or 5). In Experiment 1, the UF lasted 225 min; DF1 and DF2 about75 min each, and DF3, DF4 and DF5 about 45 min each. In Experiment2, the UF lasted 265 min, DF1 and DF2 about 65 min each, and DF3 andDF4, 35 min each. The UF step of Exp. 1 was shorter, because the finalconcentrate volume was bigger (6 L) than for Exp 2 (5 L). In the otherhand, the DFs of Exp. 1 took more time than the DFs of Exp. 2.

Fig. 3 presents a comparison between the two concentrationexperiments with respect to the percentages of protein on a dry basis(protein content/total solids content) versus the stage of the process.

The percentage of initial protein mass in both experiments wasabout 15%. The experiment with the higher VCF resulted in a slightlyhigher protein percentage at the end of the UF and this difference wasaccentuated with the DF steps, and significant for both DFs, the largervolume and the smaller volume. In experiment 2 the DFS reached thefinal protein concentration of 71% versus 62% in Experiment 1.

Fig. 3. Percentage of protein (w/w, dry basis) versus the stage of the process forExperiments1 and2.MembraneUF-6001, T=50 °C,ΔP=2 bar, feedflowrate=840 L h−1.Legend: (■) Experiment 1; (▲) Experiment 2. *coefficient of variation: ±0.05%.

Fig. 5. Permeate flux vs volumetric concentration factor (VCF) for Experiment 2. UFmembrane UF-6001, ΔP=2 bar, T=50 °C, feed flowrate=840 L h−1.

384 C. Baldasso et al. / Desalination 278 (2011) 381–386

Fig. 4 shows the percentage mass of lactose on a dry basis (lactosecontent/total solids content) in the concentrate versus the stage of theprocess.

In the early experiments the percentage by mass (dry basis) oflactose was 72%; at the end of the UF, the percentages dropped to 64and 57% for Experiments 1 and 2, respectively. During the diafiltra-tions these percentages were further reduced to 37% for Experiment 1and 29% for Experiment 2.

In relation to ash content, the initial samples initially had 12% ashon a dry basis. In both experiments the process was efficient for theremoval of ashs, with percentages close to zero at the end of the UF,showing that nearly all of the ash was removed from the proteinconcentrate.

It was found that the DFs of smaller volumes were very effectivefor the removal of ashs and lactose; in Experiment 2, for example,there was a large reduction in the concentration of other elementsrelative to the amount of protein in the last DFs of 2.5 L.

The protein-concentration results in Experiment 2 were betterthan those obtained them in Experiment 1. It is noteworthy that in

Fig. 4. Percentage of lactose (w/w, dry basis ) versus the stage of the process, forExperiments1 and2.MembraneUF-6001, T=50 °C,ΔP=2 bar, feedflowrate=840 L h−1.Legend: (■) Experiment 1; (▲) Experiment 2. *coefficient of variation: ±0.06%.

Experiment 2, the UF concentrate volume before DF was 5 L, and thisvolume decrease resulted in a higher final protein concentration and alower volume of water to the DF stage. For Experiment 1 was used21 L of water to the stage of DF, while for Experiment 2 was used 15 Lof water in DFS and protein percentage was higher than achieved.Therefore, in Experiment 2 was obtained a higher content of protein,with fewer steps of DF, with a smaller total volume of water used topurify these, this result shows that this method is promising and canbe used to obtain protein concentrates.

The present results are in agreement with those found in theliterature; Rektor and Vatai [1], Butylina et al. [15], and Zydney [16],performed studies showing that components of lowmolecular weight(lactose and ashs) preferentially permeate the membranes of UF,which retain the protein molecules.

Fig. 5 shows the flux of whey permeate as a function of thevolumetric concentration factor for the UF stage of Experiment 2,which resulted in better protein purification.

It was observed that the permeate flux decreased as the volumetricconcentration factor was increased. The VCF reached a maximum of 6,because the concentrate was to be used in subsequent diafiltrationsand these volumes and concentrations were appropriate for thatpurpose.

The data in Fig. 5 confirm what several authors have alreadyobserved. According to Smith [17], the permeate flux generallydecreases with the increase in VCF. According Rektor and Vatai [1],Bacchin et al. [18] and Cheryan [19], the more concentrated thesolution of protein is, the lower the permeate flux is, due to the higherosmotic pressure and the greater accumulation of solute molecules inthe polarized layer, increasing its thickness and, consequently, itsresistance to permeation. According Atra et al. [2], at higher VCF therewas a deposit of the largest and most dense layer that reduced thepermeate flux until it reached a static condition.

The permeate flux in the DF steps was below the initial flux in UFdue to the fouling formed in the protein-concentration stage, and thedilution factor was not very high.

Fig. 6 shows the total solids (TS) concentrations of the concentrateand permeate samples over time for Experiment 2 in the concentra-tion and diafiltration stages. There was an increase of TS in theconcentrate during UF due to the increased concentration of protein.In the permeate the concentration of TS was approximately constantuntil the end of the UF, increasing slightly in the last hour of theexperiment. The concentrate had higher concentrations of TS in boththe UF and DF stages. During the DF a reduction of TS in theconcentrate occurred due to the removal of lactose and ashs and in thepermeate due to the dilution.

Fig. 6. Total solids concentrations of the concentrate and permeate samples (w/v) vs timefor Experiment 2. Membrane UF-6001, T=50 °C, feed flowrate=840 L h−1, ΔP=2 bar.Legend: (♦) permeate UF, (■) concentrate UF, (▲) permeate DF, (●) concentrate DF.* coefficient of variation: ±6%.

Fig. 8. Lactose concentrations of the concentrate and permeate samples (w/v) vs timefor Experiment 2. Membrane UF-6001, T=50 °C, feed flowrate=840 L h−1,ΔP=2 bar.Legend: (♦) permeate UF, (■) concentrate UF, (▲) permeate DF, (●) concentrate DF.* coefficient of variation: ±6%.

385C. Baldasso et al. / Desalination 278 (2011) 381–386

Fig. 7 shows the concentration of protein versus time for theconcentrate and permeate in the concentration and purificationstages.

The concentration of protein increased in the retentate throughoutthe UF stage of the experiment. The initial concentration of proteinwas about 9 kg.m−3 and was about 36 kg·m−3 at the end of the UF.

Note that the concentration of protein in the DF did not showmuch variation however the level of contaminants, ashs and lactosedecreased significantly in DF, thus resulting in a purification of theproteins. Similar behavior was reported by Leite et al. [20].

The lactose concentrations of the concentrate and permeatesamples versus time are shown in Fig. 8.

The curve shows a behavior similar to that found for total solids,because the lactose, the most abundant component in whey, with amolecular weight lower than themolarmass cut-off of themembrane,showed a low retention and had a decisive impact on the behavior oftotal solids over time in both permeate and concentrate. In the UFstage the permeate concentration of lactose remained around 40 kg.

Fig. 7. Protein concentrations of the concentrate and permeate samples (w/v) vs timefor Experiment 2. Membrane UF-6001, T=50 °C, feed flowrate=840 L h−1,ΔP=2 bar.Legend: (♦) permeate UF, (■) concentrate UF, (▲) permeate DF, (●) concentrate DF.* coefficient of variation: ±5%.

m−3; the retentate started with 42 kg.m−3 and reached 50 kg.m−3 atthe final stage of UF. At the start of DF the concentration of lactosedecreased significantly in both concentrate and permeate. At the endof the four DFs the concentration of lactose in the retentate reached15 kg.m−3 , while the concentration of lactose in the permeate wasaround 10 kg.m−3 , almost the same as the concentration of totalsolids, indicating that virtually all ash was removed, and the solidsremaining in the permeate were almost all lactose.

Some authors revealed results similar to this job. Roman et al.[21,22] performed experiments of NF associated with DF fordemineralization of cottage cheese whey considering two importantfactors: the duration of the process and water consumption. Theconcentration of feeding solutionwas performedwith different valuesof the ratio between the volume of water and diafiltrante permeatevolume. The degree of demineralization of monovalent ions reached70% and 90% at 2.5 volume concentration ratio, and at diafiltration andpermeate volume ratios (α) of 0.5 and 0.75, respectively. Forα=0.75,the experiment ended after 18.9 h, and hit a VCF=2. For α=0.5, theexperiment lasted 10 h, and the VCF was 2.65. The degree ofpurification of the detained, was two times higher for α=0.75, butthis process has used three times more water than the α=0.5. Thecomparison of the two processes revealed that greater dilutionretained from beginning to end of the experiment, contributed tofurther purify but increased the process time and water consumption.These results are consistentwith thefindings of Jaffrin and Charrier [23].These authors suggested that in an ideal process a pre-concentrationwould be interesting, where a certain amount of macrosolute wasreached before the onset of DF, the DF following the completion ofcontinuing with the decrease in volume would increase the degree ofpurification, the which was also suggested by Foley [24]. The latterresearcher performed DF continued after an initial concentration step,and achieved a significant reduction in water consumption withoutincreasing the process time too.

The pH did not significantly change throughout the process;samples of concentrate and permeate pH remained between 6.2 and6.4. This behavior is a good indication that the solution was notdegraded during the time of the experiment.

The measurement of electrical conductivity remained almost thesame for the permeate and concentrate throughout the UF stage,because the membrane is not selective for ashs, which are thecompounds that contribute most to the electrical conductivity. Duringthe DF stage the electrical conductivity of the samples was decreasedafter the addition of distilled water (dilution effect).

386 C. Baldasso et al. / Desalination 278 (2011) 381–386

The results show that UF associated with DF is a technology thatcan be successfully used for demineralizing the whey, and stillconcentrate components such as proteins. And still has the advantageof not degrading its main nutrients.

4. Conclusions

Based on the results of this work, we obtained the findings listedbelow:

• For the same operating conditions, there was a significant differencein the permeate flux of the whey and of the water; the flux of wateris greater than the flux of whey permeate.

• The increase in flux was limited by an increased polarization layer,i.e., the increase of transmembrane pressure was counterbalancedby the increase in total resistance.

• The higher VCF generated the best results for protein purification.For Experiment 2, with a VCF equal to 6, after the DFs a proteinconcentration of 71% by weight (dry basis) was reached.

• The process proved to be efficient for the removal of ashs; the ashconcentration approached zero end of the UF stage, i.e., practicallyall the ash was removed from the concentrate.

• The DF was very effective when performed with small volumes anda greater number of times.

• Decreasing the volume of concentrate from 6 to 5 L before the DFsteps resulted in a higher final protein concentration and a lowervolume of water to the DF stage.

• The measurements of pH and electrical conductivity showed thatthe solution was not degraded during the experiments.

• The flux of whey permeate decreased with the increase in VCF due toseveral effects: the increase of viscosity, the fouling, the higherosmotic pressure and the greater accumulation of solute molecules inthe polarization layer, which increased the resistance to permeation.

• The permeate generated can be treated by nanofiltration for therecoveryof lactose for use in the foodandpharmaceutical industries. Itis also possible to promote the demineralization of the process waterby reverse osmosis and electrodialysis for the recovery of ashs andwater in the process, generating financial and environmental benefits.

Acknowledgments

The authors thank the CNPq (National Counsel of Technologicaland Scientific Development) and CAPES (Coordination for theImprovement of Higher Education Personnel) for financial support.

References

[1] A. Rektor, G. Vatai, Membrane filtration of Mozzarella whey, Desal. 162 (2004)279–286.

[2] R. Atra, G. Vatai, E. Bekassy-Molnar, A. Balint, Investigation of ultra andnanofiltration for utilization of whey protein and lactose, J. Food Eng. 67 (2005)325–332.

[3] N.S.P.S. Richards, Soro Lácteo – Perpectivas Industriais e Proteção aoMeio Ambiente,Food Ingred. 17 (2002) 20–27.

[4] M.S. Yorgun, I.A. Balcioglu, O. Saygin, Performance comparison of ultrafiltration,nanofiltration and reverse osmosis on whey treatment, Desal. 229 (2008)204–216.

[5] J.N. Wit, Nutritional and functional characteristics of whey proteins in foodproducts, J. Dairy Sci. 81 (1998) 597–608.

[6] V.C. Sgarbieri, Proteínas em alimentos protéicos: propriedades, degradações,modificações, Ed. Varela, São Paulo, 1996.

[7] G. Brans, Design of membrane systems for fractionation of particles suspensions.PhD Thesis, Wageningen University, Netherlands, 2006.

[8] M.F. Ebersold, A.L. Zydney, The effect of membrane properties on the separation ofprotein charge variants using UF, J. Memb. Sci. 243 (2004) 379–388.

[9] K.W.K. Yee, D.E.Wiley, J. Bao,Whey protein concentrate production by continuousultrafiltration: operatibility under constant operating condition, J. Memb. Sci. 209(2007) 125–137.

[10] E.G. Sprinchan, Optimization of technological regimes for obtaining protein–mineral concentrated products from secondary milk raw materials, Surf. Eng.Appl. Electrochem. 45 (2009) 63–70.

[11] LANARA-Laboratório Nacional de Referência Animal, Ministério da AgriculturaMétodos analíticos oficiais para controle de produtos de origem animal e seusingredientes – II Métodos físicos e químicos. Portaria 001/81 de 07/10/81 (1981).

[12] G.L. Miller, Use of dinitrosalicylic acid reagent for determination of reducing sugar,Anal. Chem. 31 (1959) 426–428.

[13] O.H. Lowry, N.J. Rosebrough, A.L. Farr, R.J. Randall, Protein measurements with theFolin phenol reagent, J. Biol. Chem. 193 (1951) 265–275.

[14] J.R. Boschi, Concentração e purificação das proteínas do soro de queijo por UF,Dissertação de Mestrado, Departamento de Engenharia Química, UFRGS, Brasil,2006.

[15] S. Butylina, S. Luque, M. Nystroma, Fractionation of whey-derived peptides using acombination of UF and NF, J. Memb. Sci. 280 (2006) 418–426.

[16] A.L. Zydney, Protein separations usingmembrane filtration: new opportunities forwhey fractionation, Int. Dairy J. 8 (1998) 243–250.

[17] G. Smith, Dairy Processing — Improving Quality, CRC Press, New York, 2003.[18] P. Bacchin, P. Aimar, R.W. Field, Critical and sustainable fluxes: theory,

experiments and applications, J. Memb. Sci. 281 (2006) 42–69.[19] M. Cheryan, Ultrafiltration Handbook, Lancaster, Technomic Publishing Company,

Inc. 375 p., 1986.[20] Z.T.C. Leite, D.S. Vaitsman, P.B. Dutra, Leite e alguns de seus derivados – da

antiguidade à atualidade, Quím. Nova 29 (4) (2006) 876–880.[21] A. Román, S. Popović, G. Vatai, M. Djurić, M.N. Tekić, Process duration and water

consumption in a variable volume diafiltration for partial demineralization andconcentration of acid whey, Sep. Sci. Technol. 45 (2010) 1347–1353.

[22] A. Román, J. Wang, J. Csanádi, C. Hodúr, G. Vatai, Partial demineralization andconcentration of acid whey by nanofiltration combined with diafiltration,Desalination 241 (2009) 288–295.

[23] M.Y. Jaffrin, J.P. Charrier, Optimization of ultrafiltration and diafiltration processfor albumin production, J. Memb. Sci. 97 (1994) 71–81.

[24] G. Foley, Ultrafiltration with variable volume diafiltration: a novel approach towater saving in diafiltration processes, Desalination 199 (2006) 220–221.