Separation and Purification Technology 195 (2018) 322–331
Contents lists available at ScienceDirect
Separation and Purification Technology journal homepage: www.elsevier.com/locate/seppur
Influence of processing temperature on flux decline during skim milk ultrafiltration Kenneth S.Y. Ng, Dave E. Dunstan, Gregory J.O. Martin
T
⁎
Department of Chemical & Biomolecular Engineering, University of Melbourne, Parkville, Victoria 3010, Australia
A R T I C L E I N F O
A B S T R A C T
Keywords: Skim milk Filtration Fouling Temperature Whey proteins
The flux decline behaviour during skim milk ultrafiltration (UF) was investigated at 10 °C, 30 °C and 50 °C. Despite higher fluxes, UF processing at higher temperatures resulted in higher magnitudes and rates of irreversible fouling. Fouling in this temperature range was primarily proteinaceous, consisting of mainly peptides and alpha-lactalbumin (α-LA), with minor amounts of beta-lactoglobulin (β-LG) present only at 50 °C. The increase in fouling resistance with processing temperature was concluded to be mainly due to increased pore fouling by α-LA deposition, and in part to β-LG deposition at 50 °C. These are attributed respectively to the thermal expansion of membrane pores and reversible conformational changes of β-LG.
1. Introduction Skim milk UF is widely used in dairy processing as means of concentration and purification of milk proteins for the production of cheese [1–5], milk protein concentrates (MPC) [3–7], and for protein standardisation [3–6,8]. During filtration, additional permeation resistance is contributed by the inherent accumulation of retained particles at the membrane surface (concentration polarisation, or CP) and particle deposition on the membrane surface or inside membrane pores (fouling) [1,9]. As a result, filtration throughput is reduced and product quality is altered. In addition, chemical cleaning is necessary for removing fouling, incurring significant water and chemical consumption, and process downtime [10]. To further optimise filtration and cleaning, a deeper understanding of CP and fouling in skim milk UF is required, but to date these is not completely understood. In particular, the effect of processing temperature has yet to be fully investigated. Industrial skim milk UF is typically conducted at either ∼10 °C or ∼50 °C, each with its advantages and drawbacks. Operation at the higher temperatures (∼50 °C) is generally favoured due to higher fluxes resulting from lower permeate viscosity. However, the growth of thermophiles is also promoted [1,6]. Conversely, thermophile growth is hindered at ∼10 °C, at the expense of lower fluxes due to higher permeate viscosity. Processing temperature also affects the physicochemical properties of skim milk, which can in turn influence filtration behaviour. For instance, the structure and composition of casein micelles (CMs) are influenced by hydrophobic interactions and calcium phosphate solubility, both of which are temperature-sensitive [2]. At lower temperatures, calcium phosphate solubility increases and ⁎
hydrophobic interactions weaken, resulting in the solubilisation of micellar calcium phosphate and dissociation of β-casein [2,11,12]. Meanwhile, whey proteins are more prone to thermally-induced conformational changes at elevated temperatures [13]. Filtration behaviour has been observed to be affected by changes in milk physicochemical properties due to alterations in pH [14–18], ionic strength (via mineral addition) [17,19] and thermal pre-treatment [20]. However, despite the temperature-sensitive nature of milk and the common practice of skim milk UF at ∼10 °C in some parts of the world, the vast majority of the skim milk UF fouling characterisation studies have been conducted at 50 °C, which is more widely used. There are very few studies pertaining to skim milk UF at low temperatures, limited to investigations on flux [21,22], protein rejection [22], compositional changes [23–25], and microbial growth [26]. To our knowledge, the fouling behaviour during skim milk UF at 10 °C has not been properly examined, nor has any comparison or validation with what is known for UF at 50 °C been made. This also raises the question of how much of the knowledge established for UF at 50 °C is applicable to UF at ∼10 °C. Membrane fouling in the dairy industry has generally been attributed to adsorption of proteins and precipitation of calcium phosphate [1,6]. Accordingly, chemical cleaning typically involves alkaline and acid cleaning for the removal of organic and mineral fouling respectively [27]. However, it has been shown that fouling in skim milk UF performed at 50 °C is predominantly caused by proteins, with minerals only accounting for about 0.4% of the foulant material [28–30]. More recent studies investigating the effectiveness of various cleaning chemicals for rejuvenating UF membranes fouled by skim milk (at 50 °C) have also demonstrated the ineffectiveness of acid cleaning in protein
Corresponding author. E-mail address:
[email protected] (G.J.O. Martin).
https://doi.org/10.1016/j.seppur.2017.12.029 Received 19 May 2017; Received in revised form 23 October 2017; Accepted 14 December 2017 Available online 15 December 2017 1383-5866/ © 2017 Elsevier B.V. All rights reserved.
Separation and Purification Technology 195 (2018) 322–331
K.S.Y. Ng et al.
Nomenclature
LG lactoglobulin MPC milk protein concentrate MW molecular weight NF nanofiltration PES polyethersulfone PSf polysulfone SDS-PAGE sodium dodecyl sulfate polyacrylamide gel electrophoresis SEM-EDS scanning electron microscopy-energy dispersive X-ray spectroscopy TMP transmembrane pressure UF ultrafiltration
Abbreviations ATR-FTIR attenuated total reflection Fourier transform infrared spectroscopy BSA bovine serum albumin CFV cross-flow velocity CM casein micelle CN casein CP concentration polarisation ICP-OES inductively coupled plasma optical emission spectrometry LA lactalbumin
this study. The membrane sheets were cut from a Koch HFK-131 spiral wound module (Koch Membrane Systems, Massachusetts, USA) and stored in 1% Ultrasil 73 (Ecolab, NSW, Australia) at 6 °C to prevent bacterial contamination. Feed and permeate spacers used were also cut to size from the same membrane module.
removal [27,31–33]. The omission of acid cleaning has been suggested as this can potentially reduce chemical, water and chemical consumption as well as cleaning time, especially if the alkaline cleaning formulation is optimised [34]. The applicability of these findings to skim milk UF at 10 °C is subject to the composition of the fouling layer formed at 10 °C, but this is yet to be reported. It is not known whether changes in the equilibrium of calcium and casein between the micelles and serum affect the contribution of minerals or protein to the fouling. In this study, the influence of processing temperature on flux decline behaviour during skim milk UF was investigated in detail using a custom-built filtration rig. Respective resistance contributions of CP and fouling were evaluated using the resistance-in-series approach [35]. The temperature-dependent fouling behaviour was studied, including a comprehensive analysis of the composition of the fouling layers. An explanation of the underlying mechanisms for the observed differences was also provided.
2.3. Cross-flow filtration rig Filtration experiments were carried out on a custom built filtration rig (Fig. 1). The filtration cell consisted of feed and permeate plates constructed out of grade 316 stainless steel, with a 0.046 in. feed spacer and permeate carrier inserted along the feed and permeate channels respectively. The filtration surface dimensions were 0.21 m × 0.060 m, giving a filtration area of 0.0126 m2. The assembled cell plates were secured with stainless steel screws. A gear pump (Micropump, USA) was used to deliver the feed solution to the filtration cell through a set of cooling coils immersed in a water bath. The water bath temperature was controlled by a thermoregulator connected to a chiller. Filtration was conducted under constant pressure, and the desired flow rate and pressure were obtained by adjusting the backpressure valve (Swagelok, Victoria, Australia) and pump speed. Permeate was collected in a beaker placed on a mass balance (Mettler Toledo, Switzerland) to measure the mass of permeate collected at regular time intervals. The mass balance and variable speed drive were connected to a computer via an I/O interface (National Instruments, USA). NI Labview was used to adjust the pump speed and to record permeate mass every 30 s. Pressure gauges (Swagelok, Victoria, Australia) and digital thermometers (Comark Instruments, USA) were placed in-line near the feed-side inlet and outlet of the filtration cell to monitor the feed and retentate streams respectively. 3/8 in. high-pressure nylon tubing and
2. Materials and methods 2.1. Filtration feed fluids and water Fresh pasteurised (72 °C/15 s) skim milk and whole milk were purchased from the local supermarket and stored at 6 °C. The skim milk contained 37 g/L protein and 1 g/L fat, while the whole milk contained 35 g/L protein and 34 g/L fat, as per manufacturer specifications. Double distilled water (ddH2O) and RO water were used for solution preparation and membrane rinsing respectively. 2.2. Membranes 10 kDa polyethersulfone (PES) flat sheet membranes were used in
Backpressure regulator Bypass line Feed pump
P
Thermoregulator
Feed tank
T
T
Cooling coil in refrigerated water bath
P
Filtration cell
Permeate balance T: Digital thermometer P: Pressure gauge
PC
323
Fig. 1. Schematic of the cross-flow filtration rig.
Separation and Purification Technology 195 (2018) 322–331
K.S.Y. Ng et al.
Swagelok fittings were used for all connections except for the permeate line, where 1/4 in. high-pressure nylon tubing was used. A bypass line was used for feed heating/cooling and flushing the lines. The total holdup volume of the pump, filtration cell and tubing was approximately 0.5 L.
measured using a capillary viscometer (type 531–10, Schott Gerate, Germany) submerged in a temperature-regulated water bath. μUFP values at 10 °C, 30 °C and 50 °C were measured to be 1.489, 0.888 and 0.627 centipoise respectively. 2.6. Analysis of fouling composition
2.4. Filtration procedure 2.6.1. Extraction of membrane foulants Fouled membranes were removed from the filtration cell after resistance measurements and cut into approximately 1 cm × 1 cm pieces. The cut membranes were immersed in 15 mL 1% SDS solution (SigmaAldrich, Australia) and gently shaken overnight at ambient temperature to allow desorption of the foulant material. Membranes were removed from the sample container the following day, and the extracts (1% SDS solution containing desorbed materials) were stored frozen until required for analysis. Surfaces of membranes treated in this manner were also analysed by ATR-FTIR (Varian 7000, Agilent, USA). Membrane pieces were water rinsed and vacuum-dried prior to analysis. The spectra observed were almost identical to that of the virgin membrane, confirming complete extraction of organic material (see Supplementary Material).
Fresh membranes were used for each run that were pre-conditioned to remove residual preservatives. Pre-conditioning consisted of a water flush, recirculation of 2 L Ultrasil 10 (Ecolab, NSW, Australia) at pH 11.0–11.5 at 50 °C for 20 min, followed by another water flush. Filtration experiments were performed to obtain data to evaluate the filtration resistance components, as explained in Section 2.5. Prior to each filtration run, water was recirculated through the rig at a transmembrane pressure (TMP) of 3 bar and cross-flow velocity (CFV) of 0.5 m/s. Flux was allowed to stabilise, and the pure water flux (J0) over 5 min was recorded. After pure water flux measurements, the feed tank was emptied then filled with 3 L of skim milk feed. Skim milk was brought to the filtration experimental temperature (10 °C, 30 °C or 50 °C ± 0.5 °C) by recirculating through the bypass (initial water effluent and some milk feed were purged to prevent feed dilution). Skim milk UF was then carried out at TMP = 3 bar and CFV = 0.5 m/s in quasi-constant concentration mode with permeate returned to the feed tank every 15 min. Skim milk UF (in which Jskim milk was recorded as a function of time) was ceased after 5, 60 or 240 min, after which a water flush was conducted and the water flux was measured (Jwater flush). Following water flux measurements, the filtration cell was disassembled and the fouled membranes were removed for further analysis. The filtration cell was thoroughly cleaned with labware detergent (Thermo Fisher, NSW, Australia), rinsed with water and reassembled. 2 L of Ultrasil 10 at pH 11.0–11.5 with 25 ppm sodium hypochlorite (Thermo Fisher, NSW, Australia) was then recirculated through the filtration rig at 50 °C for 20 min, and finally flushed with water. Water flushes, pre-conditioning and membrane cleaning were conducted at TMP = 1 bar and CFV = ∼0.65 m/s. Water flushes were conducted at ambient temperature.
2.6.2. Determination of protein composition and mineral content of extracted foulants Protein composition of the extracted foulants was determined by SDS-PAGE (the surfactants present in the extracts prevented the use of HPLC). SDS-PAGE was performed by first adding 20 μL samples of the stored extract solutions with 20 μL Laemmli buffer (Biorad Laboratories, NSW, Australia) containing 5% 2-mercaptoethanol (Biorad Laboratories, NSW, Australia). The sample mixtures were then heated in a 95 °C water bath for 5 min. Aliquots (20 μL) of these treated samples and a protein standard solution (Precision Plus Unstained Protein Standards, Biorad, NSW, Australia) were loaded into 8–16% linear gradient Tris-HCl Criterion 18 well gels (Biorad Laboratories, NSW, Australia) and run at 100 V for 130 min. Gels were stained with Biosafe Coomassie Blue G-250 (Biorad Laboratories, NSW, Australia) and protein bands quantitatively compared by densitometric analysis using Biorad Gel Doc XR + Imager (Biorad Laboratories, California, USA). The calcium content of the extracted foulants was determined by measuring the adsorption of the samples at 317.933 nm using inductively coupled plasma optical emission spectrometry (ICP-OES, Varian 720-ES, Varian Inc., Walnut Creek, California, USA). Calcium content of the 1% SDS solution was used as the zero measurement.
2.5. Evaluation of filtration resistance components Contributions of individual filtration resistance components (membrane, CP and fouling) were evaluated using the resistance-in-series model [35] using fluxes measured at different stages of the filtration run:
2.6.3. Microanalysis of fouled membranes by scanning electron microscopyenergy dispersive X-ray spectroscopy (SEM-EDS) Fouled membranes were vacuum-dried prior to microanalysis using an FEI Quanta SEM equipped with an Oxford INcA detector (FEI, Oregon, USA). For each fouled membrane, three surface sections each of 30,000 μm2 were analysed.
1. The pure water flux measured prior to filtration (J0) is due to the resistance of the clean membrane (Rm) only:
Rm =
TMP μ w Jo
(1)
2. The skim milk UF flux (Jskim milk) is due to resistances contributed by CP, fouling and the membrane (total resistance):
R cp + Rf + Rm =
TMP μUFP Jskim milk
3. Results 3.1. Evaluation of fluxes and resistances
(2)
Fluxes and resistances for skim milk UF conducted at 10 °C, 30 °C and 50 °C under similar hydrodynamic conditions (TMP = 3 bar, CFV = 0.5 m/s) over a period of 240 min were evaluated. Overall fluxes were observed to increase by ∼250% from 10 to 50 °C (Fig. 2a). This is expected because of reduced permeate viscosities at higher temperatures [1]. However, it needs to be emphasised here that the overall filtration resistances also increased with processing temperature (Fig. 2b). This means that the higher fluxes observed could not be solely accounted for by viscosity differences. In addition, the resistance at the commencement of filtration was similar at 30 °C and 50 °C (ca
3. The water flux measured following the water flush after skim milk UF (Jwater flush) is due to the combined resistances of the membrane and fouling layers:
Rf + Rm =
TMP μ w Jwater flush
(3)
where μw and μUFP are viscosities of water and skim milk permeate respectively. μw values were taken from [36,37], while μUFP was 324
Separation and Purification Technology 195 (2018) 322–331
K.S.Y. Ng et al.
60 50 Flux (LMH)
flux decline observed earlier.
(a)
3.2. Analysis of foulant composition
40
3.2.1. Proteinaceous vs. mineral fouling To identify the reasons for the observed differences in fouling behaviour, the compositions of the foulant material irreversibly adsorbed to the membrane during filtration were first analysed. For this, fouled membranes were immersed in 1% SDS solution and gently shaken to desorb the foulant material. Since protein and calcium phosphate are generally the main concerns in the membrane fouling of dairy streams [1,6], the extract solutions were analysed for proteins and calcium content. Proteinaceous and calcium content in the extracted foulants obtained were determined by SDS-PAGE densitometry and ICP-OES respectively. The results (Table 1) show that the foulant material obtained at 10 °C and 50 °C were predominantly proteinaceous, which is consistent with previous reports of fouling compositions obtained from skim milk UF at 50 °C [28,29]. Fouled membranes were also directly examined with SEM-EDS to check that any minerals present were also fully extracted by the 1% SDS solution. Minor nitrogen peaks were detected on the fouled membranes, indicating the presence of proteinaceous material, but no calcium or other minerals could be detected (see Supplementary Material). These results are consistent with that obtained from the analysis of the extracted foulants, confirming that skim milk UF fouling within 10–50 °C is mainly due to proteinaceous material.
30 20 10 0
0
60
120
180
240
180
240
Time (min) Total Resistance (×1012 m-1)
45
(b)
40 35 30 25
0
60
120 Time (min)
Normalised Resistance, R/Rt=0
1.3
(c)
3.2.2. Composition of proteinaceous fouling The SDS-PAGE gel (Fig. 4) of the extracted foulants was further analysed to determine the proteinaceous composition of the fouling layer. Substantial amounts of low MW peptides were evident in the foulants at all three UF temperatures investigated (lanes 3–11). α-lactalbumin (α-LA) was also present at all three temperatures, and constituted the majority of the protein fraction at 30 °C and 50 °C. Intensities of α-LA and peptides bands also increased with temperature. In comparison, β-lactoglobulin (β-LG) and casein (mainly κ-CN) were only detected at 50 °C and in minor amounts. The dominant presence of α-LA over β-LG and caseins in the fouling layer at 50 °C is consistent with SDS-PAGE observations by Tong et al. [38] on UF membranes fouled by whole milk. This was also verified with an experiment conducted with whole milk, which showed a foulant composition identical to that of skim milk (lane 12). Some αs-CN was also detected at 10 °C. However the presence of αs-CN was not consistent – in the 30 °C filtration runs, αs-CN was detected in foulants obtained at 60 min (lane 7) but not at 240 min (lane 8). This may be due to incomplete removal by water rinsing. The variations of peptide, α-LA and β-LG content in the extracted foulants with temperature and time are more clearly illustrated by quantitative estimation provided by densitometric analysis (Fig. 5). Total proteinaceous content was observed to increase with temperature and filtration duration, and the rate of increase was also greater at 50 °C than at 10 °C and 30 °C. A similar trend is seen with α-LA content. On the other hand, peptide content showed a significant increase from 10 °C to 30 °C, and only a very minor increase from 30 °C to 50 °C. Again, β-LG was only observed at 50 °C.
1.2
1.1
1.0
0.9
0
60
120
180
240
Time (min) Fig. 2. Overall fluxes (a), resistances (b) and normalised resistances (c) for skim milk UF runs 10 °C (blue diamonds), 30 °C (red squares) and 50 °C (green triangles) (TMP = 3 bar, CFV = 0.5 m/s). Error bars correspond to standard deviation of 2–4 experiments. For clarity, error bars for fluxes (a) and resistances (b) are presented for every third data point, but were consistent over the 240 min period. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
34 × 1012 m−1), both higher than at 10 °C (ca 32 × 1012 m−1). Conversely, the relative increase in resistance during filtration was approximately equal at 10 °C and 30 °C, but significantly higher at 50 °C (Fig. 2b and c). Total filtration resistances at the commencement of filtration were also of an order of magnitude larger than the intrinsic membrane resistance (Rm = 1.93–2.79 × 1012 m−1), meaning that significant flux decline had already occurred at the beginning of filtration. It should be noted that despite the variability in intrinsic membrane resistances, there were no statistical differences in concentration polarisation and fouling resistances measured amongst runs conducted under the same set of experimental conditions. To determine the relative contributions of CP and fouling, a breakdown of resistances was evaluated using the resistance-in-series approach as outlined in Section 2.5. The results (Fig. 3) show no clear variations in CP resistance as a function of temperature and time. In contrast, fouling resistances were observed to increase with processing temperature and filtration duration, and the rate of resistance increase at 50 °C is greater than their counterparts at lower temperatures. Changes in irreversible fouling behaviour are therefore accountable for both the temperature-dependent differences in magnitude and rate of
4. Discussion 4.1. Minor presence of mineral fouling The minor presence of mineral fouling in skim milk UF at both 10 and 50 °C contrasts with the severe mineral fouling that is often reported in the filtration of other dairy streams, such as whey, UF permeate and simulated milk ultrafiltrate [1,6,39–41]. Mineral fouling by dairy streams is attributed to the precipitation of calcium phosphate as it is sparingly soluble (solubility of CaHPO4 and Ca3(PO4)2 in milk 325
Separation and Purification Technology 195 (2018) 322–331
K.S.Y. Ng et al.
Fig. 3. Resistance breakdown for skim milk UF runs at 10 °C, 30 °C and 50 °C (TMP = 3 bar, CFV = 0.5 m/ s). Rf = internal fouling resistance, Rcp = concentration polarisation resistance. For reference, inwas trinsic membrane resistance (Rm) 1.93–2.79 × 1012 m−1. Due to flux variations between data sets obtained at a given processing temperature, Rf and Rcp values determined for 5 and 60 min runs were scaled according to total resistances at their corresponding times in the 240 min run minus the membrane resistance (Rt-Rm). Error bars correspond to standard deviation of 2–4 experiments. Single runs were conducted for the 5 min runs and the 30 °C 60 min run.
Component Resistance (×1012 m-1)
35
Rf
Rcp
30 25 20 15 10 5
10 30 50
350 ± 120 561 ± 110 1058 ± 8
3.1 ± 0.2 n.d. 6.3 ± 0.2
0.0089 n.d. 0.0060
are 1.8 mmol/L and 0.06 mmol/L respectively) [2]. Calcium phosphate also has a reverse solubility, i.e. its solubility decreases with increasing temperature [1]. Correspondingly, mineral fouling is reduced upon calcium demineralisation [31,42] or operation at lower temperature and/or pH [39,40], where calcium phosphate is more soluble.
2
3
4
5
6
7
8
9
10
11
12
Molecular Weight (kDa)
1
50°C 240 min
Calcium/ Proteinaceous ratio
50°C 60 min
Calcium Content (mg/ m2)
50°C 5 min
Proteinaceous Content (mg/m2)
30°C 240 min
In contrast, skim milk is supersaturated with respect to calcium phosphate (30 mmol/L Ca2+ and 21 mmol/L inorganic phosphate), but precipitation of calcium phosphate does not occur naturally. This is mainly due to associations with casein micelles, which provide a sufficient buffering capacity for calcium phosphate [2]. About 67% of calcium and 50% of inorganic phosphate are incorporated into casein micelles (micellar phase), with the remainder present in the aqueous phase [43]. This partitioning of calcium and phosphate shifts according to the reverse solubility of calcium phosphate, for instance from the soluble aqueous phase to the casein micelles at higher temperatures [44]. In addition, mineral re-equilibration in milk has been demonstrated to occur rapidly in response to temperature changes (within the response time of a pH probe), and remains stable post-equilibration [44]. With the high buffering capacity and responsiveness of the calcium phosphate equilibria, calcium concentrations in the serum and micelle phases would have equilibrated at the operating temperature prior to skim milk UF. In addition, UF membrane pores are orders of magnitude larger than mineral ions providing no means for enrichment
Table 1 Comparison of proteinaceous and calcium content in the extracted foulants obtained from membranes fouled by skim milk at 10, 30 and 50 °C for 240 min. Concentrations of proteinaceous material and calcium were determined by SDS-PAGE densitometry and ICP-OES respectively. Errors were calculated based on the standard deviation of duplicate measurements. UF Temperature (°C)
30°C 60 min
30°C 5 min
10°C 240 min
10°C 60 min
50°C 240 min
50°C 60 min
50°C 5 min
30°C 240 min
30°C 60 min
30°C 5 min
10°C 240 min
10°C 60 min
10°C 5 min
10°C 5 min
0
CN
β-LG α-LA peptides Protein standard
Skim milk
5min 60min 240min 5min 60min 240min 5min 60min 240min 240min 50 °C UF 50 °C UF (skim) 10 °C UF (skim) (whole) 30 °C UF (skim)
326
Fig. 4. SDS-PAGE analysis of foulant material extracted with 1% SDS after 5, 60 or 240 min of filtration. Lane 1: protein standard; lane 2: skim milk (diluted 1:100); lanes 3–11: foulants obtained after 5, 60 or 240 min of skim milk UF at 10 °C (lanes 3–5), 30 °C (lanes 6–8) and 50 °C (lanes 9–11) respectively; lane 12: foulants obtained from 240 min of whole milk UF at 50 °C; CN: casein, α-LA: α-lactalbumin, β -LG: β -lactoglobulin. Original gel photo containing lanes 1–11 was edited to remove duplicate lanes. Lane 12 was obtained from a separate gel photo containing only profiles for the standards, skim milk, and foulants collected after 240 min of UF (i.e. lanes 1, 2, 5, 8, 11, 12), carefully resized and aligned to reproduce the original profile comparison.
Separation and Purification Technology 195 (2018) 322–331
K.S.Y. Ng et al.
(a) total Quantity (mg/m2)
Quantity (mg/m2)
(b) β-LG
700
1000 800 600 400 200
600 500 400 300 200 100 0
0 0
50
100
150
200
250
0
50
Time (min)
100
150
200
250
200
250
Time (min) 800
800
(c) α-LA
700
(d) peptides
700
600
Quantity (mg/m2)
Quantity (mg/m2)
Fig. 5. Densitometrically-determined quantities of (a) total proteinaceous material, (b) β -LG, (c) α-LA and (d) peptides in foulants extracted from membranes fouled by skim milk UF at 10 °C (blue diamonds), 30 °C (red squares) and 50 °C (green triangles) at 5, 60 or 240 min, expressed as mass per unit membrane surface area. Error bars correspond to standard deviation of at least two measurements. Lines are guide for the eye, extrapolated to the origin. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
800
1200
500 400 300 200
600 500 400 300 200 100
100
0
0 0
50
100
150
200
0
250
50
Fouling Resistance, Rf (×1012m-1)
Time (min)
100
contributions by the adsorbed α-LA are estimated to be ∼8% and ∼26% (of the total calcium present in the fouling layer) at 10 and 50 °C respectively based on the amount of α-LA in the fouling layer after 240 min of skim milk UF.
20 16 12 8
4.2. Relationship between protein deposition and fouling
4 0
150
Time (min)
0
200
400
600
800
1000
The temperature-dependent increase in overall filtration resistance (Fig. 2b) and the increased rate of resistance growth during filtration at 50 °C compared to 10 °C and 30 °C (Fig. 2c) were shown earlier to be exclusively due to changes in fouling resistance (Fig. 3). Fouling in this temperature range was also confirmed to be predominantly proteinaceous (Table 1). Additionally, the temperature-dependent fouling behaviour can be related to the amount of proteinaceous material in the extracted foulants (Fig. 5a). Fouling resistances and total proteinaceous foulant content both increased with temperature, and for both, the rates of increase were higher at 50 °C than at 10 °C and 30 °C. Adding onto this, there is a strong correlation between fouling resistance and the amount of proteinaceous foulant present (Fig. 6), indicating that the increased fouling is mainly due to an overall increase in proteinaceous deposition and/or adsorption. Proteinaceous fouling is also observed to have a much larger impact on Rf at low quantities (i.e. at earlier stages of filtration) compared to later stages (Fig. 6) – the first ∼50 mg/m2 deposited contributed to ∼7 × 1012 m−1 of Rf, while the subsequent ∼1000 mg/m2 contributed to roughly the same magnitude of resistance. Fouling in skim milk UF with 10 kDa PES membranes was previously observed to manifest as reduced average pore sizes and increased surface roughness [28], which is indicative of pore blocking and surface deposition. Logically, a particle will contribute to higher permeation resistance when deposited on/in a pore (pore blockage) than on non-porous areas of the membrane (surface deposition). Larger pores are also more likely to be
1200
Protein Foulant Quantity (mg/m2) Fig. 6. Fouling resistances measured at 10 °C (blue diamonds), 30 °C (red squares) or 50 °C (green triangles) at 5, 60 or 240 min (Fig. 3) plotted as a function of total proteinaceous foulant quantity at the corresponding temperature/duration (Fig. 5a). The line is a guide for the eye, extrapolated to the origin (cross). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
of mineral ions to occur, thus fouling by calcium phosphate precipitation in the bulk fluid is unlikely to occur during skim milk UF. The origin of the trace amounts of calcium present in the fouling layer is unclear at this time. Calcium phosphate may still precipitate within the pores where casein micelles and whey proteins are longer present to provide stability [45]. It has also been previously suggested that calcium is involved in the formation of salt bridges between deposited proteins [46]. Calcium ions may also be co-adsorbed during protein adsorption as means of maintaining local charge neutrality [47], as both the membrane and milk proteins involved in fouling are negatively charged at the milk pH (6.6–6.8) [48]. Some of the calcium measured could have also been contributed by the adsorbed α-LA, which is known to contain a calcium ion tightly-bound to each α-LA molecule in its native state [49]. In this case, potential calcium 327
Separation and Purification Technology 195 (2018) 322–331
K.S.Y. Ng et al.
blocked first due to lower local permeation resistance, and this pore blocking effect diminishes as more pores become blocked and only the smaller pores remain. As such, the sharp increase in Rf at low quantities of proteinaceous fouling can be explained by a dominant pore blocking effect during early stages of filtration, while the slower but gradual increase in Rf with increasing proteinaceous foulant quantity is due to the diminishing effect of pore blocking combined with formation of a surface deposit (cake) as filtration progresses. This explanation is consistent with the underlying principles of combined fouling models, in which pore blocking is considered to either occur first before a cake is formed locally on the blocked area (consecutive models) or simultaneously with cake formation (concurrent models) [50,51]. The observed relationship between fouling resistance and foulant quantity is also similar to that described by the concurrent combined pore blocking and cake formation model described by Iritani et al. (Fig. 7) [50,52]. This model was considered suitable for comparison with our results (obtained with cross-flow experiments) even though this model has only been validated using dead-end filtration because its derivation and expression did not involve rate expressions specific to dead-end filtration, making it non-specific to either filtration mode. Finally, another observation is that the magnitude of the sharp increase in Rf during early stages of filtration increased with temperature (Fig. 8), indicating that fouling due to pore blocking is more severe when processing at higher temperatures. The occurrences of pore blocking and surface deposition can be attributed to different whey proteins. Streaming potential measurements of the same type of membranes used in this study (i.e. 10 kDa polyethersulfone (PES) flat sheet membranes from a Koch HFK-131 module) fouled by skim milk at 50 °C indicated that β-LG fouling was mainly located on the membrane surface (surface deposition), while α-LA was mainly present in the pores (pore blocking) [53]. This is consistent with expectations based on the physical dimensions of β-LG (3.6 nm × 3.6 nm × 6.7 nm) [54] and α-LA (2.3 nm × 2.6 nm × 4.0 nm) [49] compared to pores of 10 kDa PES UF membranes (4–6 nm) [28]. Meanwhile, peptides are smaller than α-LA and thus are likely to be involved in pore blocking as well. In the cases of skim milk UF at 10 °C and 30 °C where β-LG fouling is absent, surface deposition/adsorption of some α-LA would also be expected to occur. It should be noted that the relative contributions to fouling resistance are likely to vary between the different proteinaceous species due to differences in size, surface properties and fouling mechanism (e.g. pore blocking, surface deposition). However, these differences could not be resolved.
techniques), but would be a worthwhile subject for future study. Considering that β-LG monomers have sizes (∼4 nm) [54] comparable to that of the membrane pores (4–6 nm) [28], in-depth pore blocking by βLG is also likely. In contrast to β-LG, α-LA is conformationally stable between 10 and 50 °C, and therefore the observed increase in α-LA fouling content with temperature (Fig. 5c) cannot be explained by effects of protein denaturation. Since α-LA fouling is mainly located in the membrane pores, the temperature-dependent increase in α-LA fouling suggests an increased accessibility to the pores at higher temperatures. The increased accessibility cannot be explained by the higher fluxes obtained at higher temperatures leading to increased particle convection towards the membrane surface, as a plot of the quantity of α-LA foulant against permeated volume also showed an increase with temperature (Fig. 9). Another possible explanation for the increase in α-LA fouling is the thermal expansion of membrane pores, which would increase the likelihood of α-LA entering the pores causing pore blocking. Reduced solute rejections at higher temperatures attributed to thermal expansion of membrane pores have been reported in a few studies on nanofiltration (NF) membranes [63,64]. A previous observation by Pompei & Peri [22] where α-LA and β-LG were detected in the permeates obtained from skim milk UF (18 kDa) at 50 °C but not at 5 °C also supports this hypothesis. However, in our case, we did not detect any protein in UF permeates obtained at both 10 °C and 50 °C (using Bradford assay, see Supplementary Material). Instead, to investigate the hypothesis of thermal pore expansion, intrinsic membrane resistances over the temperature range of 10–50 °C were measured. To avoid the effects of membrane variability, flux measurements were conducted on a single flat sheet membrane by temperature-stepping. To minimise membrane compression which was randomly observed with some membrane sheets, the experiments were conducted at TMP = 1 bar. No sharp declines in pure water fluxes indicative of membrane compression were observed at this TMP. The data (Fig. 10) shows a decrease in Rm with increasing temperature, consistent with an increase in membrane pore size. The more similar Rm values (and hence pore sizes) at 30 °C and 50 °C compared to 10 °C may also explain the differences in initial fouling resistances observed earlier (Fig. 2b). Similarly, as peptides are likely to be involved in pore fouling, the increase in peptide fouling with temperature (at least from 10 °C to 30 °C) can also be explained by effects of thermal pore expansion. While thermal pore expansion can explain the differences in fouling behaviour of α-LA and peptides at 10 °C and 30 °C, it cannot fully account for the differences in fouling behaviour at 30 and 50 °C. In particular, fouling rate of α-LA is increased at 50 °C while the peptide content at both 30 and 50 °C are similar, and κ-CN is only present in the
4.3. Temperature-dependent fouling behaviour of β-LG, α-LA and peptides The appearance of β-LG in the foulant material formed at 50 °C but not at lower temperatures (Fig. 5b) can perhaps be explained by known thermally-induced conformational changes in this protein. A small fraction of β-LG exist as monomers at milk pH, and this monomer fraction increases with temperature (from ∼14% at 20 °C to ∼30% at 45 °C) [55]. Between 40 and 55 °C, the β-LG monomers also undergo a reversible conformational change known as the Tanford transition [56]. Such conformational changes in the β-LG monomer increase its hydrophobicity and therefore its tendency to adsorb onto the membrane surface. Further, the Tanford transition also reveals the thiol group that is normally embedded within the native protein, making it accessible for thiol-disulfide interactions [57]. Thiol-disulfide interactions have been identified as the cause of protein aggregation leading to fouling and significant flux decline during microfiltration of BSA and whey solutions [58–61], as well as β-LG suspensions at elevated temperatures [62]. We did not observe any protein bands indicative of aggregate formation with non-reducing SDS-PAGE (results not shown), but it is also possible that the aggregate concentrations were too low to be detected or too large to enter the gel. Aggregate formation was not investigated in further detail here (for instance using light scattering
Rt
(Rm+Rf)max Rm+Rf Rm Rc Amount of deposited particles per unit area Fig. 7. Relation between filtration resistances and net mass of particles deposited per unit membrane area evaluated based on the concurrent combined model (modified from Iritani et al. [50,52]). Rt, Rf, Rc, Rm correspond to the total resistance, fouling resistance, cake resistance and membrane resistance respectively.
328
Separation and Purification Technology 195 (2018) 322–331
K.S.Y. Ng et al.
Fouling Resistance, Rf (×1012 m-1)
18
4.4. Comments on peptide fouling
16
The presence of significant amounts of peptides in the fouling layer (Fig. 5d) is surprising considering that peptides are only present in milk at minor concentrations (∼32 mg nitrogen/L milk according to literature [22,65]). This suggests that peptides were absorbed preferentially to the more abundant whey proteins. To note, the total quantity of peptides present on the fouled membranes after skim milk UF for 240 min was estimated to be 273.8 mg/m2 at 10 °C and 369.0 mg/m2 at 50 °C based on densitometric analysis of the PAGE results. At a peptide concentration of 32 mg N/L milk, and using a nitrogen-to-protein conversion factor of 6.25 for rough estimation [66], a sufficient volume of skim milk to produce this quantity of peptides passed through the membrane after ∼2.7 min and ∼4.1 min of filtration at 10 °C and 50 °C respectively. This means that there is an ample amount of peptides available that can participate in membrane fouling despite its low concentration in skim milk. To check whether the peptide content in skim milk changed with processing temperature and time (e.g. increased due to possible enzymatic degradation of proteins), the compositions of skim milks held at 50 °C for 0, 2 or 4 h were analysed by SDS-PAGE densitometry (skim milk at 10 °C was used as a control). Densitometric analyses showed no statistical differences in peptide content between the skim milks held at 50 °C for different durations when compared to skim milk at 10 °C (see Supplementary Material). Unlike β-LG and α-LA, peptides were not fully rejected by the membrane. This was elucidated from comparisons of Bradford and Lowry assay analyses of the UF permeates collected at different times (5, 60, 120, 180 or 240 min) during skim milk UF at 10 °C and 50 °C (see Supplementary Material). In particular, no proteins were detected in the UF permeates by the Bradford assay, while a ‘protein’ concentration of ∼0.42 mg/mL of equivalent BSA (∼67 mg N/L milk based on a nitrogen conversion factor of 6.25 for BSA [66]) was measured by the Lowry assay. Due to mechanistic differences between both assays, the Bradford assay does not register amino acids and peptides (< 3 kDa) [67,68], while the Lowry assay reflects contributions from proteins as well as peptides and certain amino acids [67,69]. The difference in concentrations of proteinaceous material measured by the two assays (∼67 mg N/L milk) can therefore be attributed to the presence of peptides and amino acids in the UF permeates. The exact peptide concentration could not be determined from the Lowry assay measurements due to unknown contributions by the amino acids. Despite this, the measured concentration value is close to the combined average concentrations of peptides and amino acids reported in literature (76 mg N/L milk) [65]. This suggests that there is little rejection of peptides (if any), and only a small fraction of the peptides adsorbed onto the membrane with each pass through the membrane. The preferential adsorption of peptides has been previously reported by Tong et al. [70] in UF of whey obtained from Cheddar cheese produced using calf rennet or Mucor pusilus protease. In their study, peptides were not detected in the original whey feed solutions by SDSPAGE, but were present in the extracted foulants. Our results are also in stark contrast with a separate study by Tong et al. [38]. They did not detect any peptides in the foulants extracted from UF membranes fouled by whole milk at 50 °C. This discrepancy between our results and that of Tong et al. [38] could be due to differences in the feeds and membranes used in the two studies. Firstly, skim milk was used in our experiments compared to whole milk employed in their study. However, experiments replicated using whole milk produced a fouling composition that was almost identical to that obtained with skim milk (Fig. 4). This means that the discrepancy was not due to the presence of fat globules in whole milk. Secondly, the membrane used in our study consisted of a polyethersulfone (PES) membrane attached to a polyester porous support/backing layer, whereas polysulfone (PSf) membranes with unwoven polyethylene backing were used in the study of Tong et al. [71]. As the foulant extraction protocols in both studies extracts all material adsorbed to the membrane and backing layer, and differing
14 12 10 8 6 4 2 0 0
500
1000
1500
2000
2500
Total Permeated Volume (x10-6 m3) Fig. 8. Fouling resistance as a function of permeated volume during skim milk UF at 10 °C (blue diamonds), 30 °C (red squares) or 50 °C (green triangles). Lines are guide for the eye. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
700
α-LA Quantity (mg/m2)
600 500 400 300 200 100 0 0
500
1000
1500
2000
2500
Total Permeated Volume (×10-6 m3) Fig. 9. Quantity of α-LA fouling present on the membrane as a function of total permeated volume at 10 °C (blue diamonds), 30 °C (red squares) or 50 °C (green triangles). Lines are guide for the eye. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
fouling layer at this temperature. Since β-LG fouling is only observed at 50 °C, there may be fouling interactions between β-LG with α-LA, κ-CN and peptides. For instance, α-LA and κ-CN fouling may be promoted by aggregation induced by the increase in β-LG surface hydrophobicity and/or thiol-disulfide interactions, while the inclusion of β-LG in the surface deposit may change the sieving properties of the deposit for peptides.
Rm / Rm(10°C)
1.2
1.0
0.8
0.6
10
20
30
40
50
T (°C) Fig. 10. Intrinsic membrane resistance (Rm) as a function of temperature relative to Rm measured at 10 °C. Each data set (diamonds or squares) corresponds to singular measurements on a single HFK-131 membrane sheet.
329
Separation and Purification Technology 195 (2018) 322–331
K.S.Y. Ng et al.
surface properties can influence fouling behaviour [72,73], the surface properties of the backing layer also need to be taken into consideration. Since these peptides are small and can permeate through the membrane pores, it is possible that majority of these peptides are adsorbed onto the backing layer. Adsorption on the porous backing layer may also have a lesser effect on UF fluxes. While these membrane and backing materials are known to be hydrophobic, there is insufficient information on the surface properties and structures of the membranes from which any further discussion can be made. Nevertheless, the fact that the peptides are part of the irreversible fouling layer means that they are a target of chemical cleaning.
York, 1999. [3] Y. Pouliot, Membrane processes in dairy technology – from a simple idea to worldwide panacea, Int. Dairy J. 18 (2008) 735–740. [4] M. Rosenberg, Current and future applications for membrane processes in the dairy industry, Trends Food Sci. Technol. 6 (1995) 12–19. [5] GEA Process Engineering, Membrane filtration in the dairy industry, (2015). http:// www.gea.com/global/en/binaries/Membrane Filtration in the Dairy Industry_ tcm11-18227.pdf (accessed December 1, 2016). [6] A.Y. Tamime, Membrane Processing: Dairy and Beverage Applications, John & Wiley Sons Ltd., West Sussex, 2013. [7] H. Patel, S. Patel, S. Agarwal, I. Technology, Milk protein concentrates : manufacturing and applications, (2014). http://www.usdairy.com/~/media/USD/ Public/MPC-Tech-Report-FINAL.pdf (accessed June 20, 2016). [8] S. Roustel, Milk standardisation and yields, (2014). http://www.dairyaustralia. com.au (accessed June 23, 2015). [9] L.J. Zeman, A.L. Zydney, Microfiltration and Ultrafiltration – Principles and Applications, Marcel Dekker Inc., New York, 1996. [10] M. Rabiller-Baudry, D. Delaunay, L. Paugam, L. Bégoin, B. Chaufer, Role of physicochemical and hydrodynamic aspects in cleaning of spiral PES ultrafiltration membranes of dairy industry, Desalination 199 (2006) 390–392. [11] L.K. Creamer, G.P. Berry, O.E. Mills, Study of the dissociation of beta casein from the bovine casein micelle at low temperature, New Zeal. J. Dairy Sci Technol (1977). [12] P. Walstra, On the stability of casein micelles, J. Dairy Sci. 73 (1990) 1965–1979. [13] J.N. de Wit, Thermal behaviour of bovine beta-lactoglobulin at temperatures up to 150 °C. a review, Trends Food Sci. Technol. 20 (2009) 27–34. [14] L. Meunier-Goddik, Milk ultrafiltration: impact of pH and heat and hold pretreatment on flux and fouling, Cornell University, 1991. [15] H. Bouzid, M. Rabiller-Baudry, L. Paugam, F. Rousseau, Z. Derriche, N.E. Bettahar, Impact of zeta potential and size of caseins as precursors of fouling deposit on limiting and critical fluxes in spiral ultrafiltration of modified skim milks, J. Memb. Sci. 314 (2008) 67–75. [16] M. Rabiller-Baudry, H. Bouzid, B. Chaufer, L. Paugam, D. Delaunay, O. Mekmene, et al., On the origin of flux dependence in pH-modified skim milk filtration, Dairy Sci. Technol. 89 (2009) 363–385. [17] M. Rabiller-Baudry, G. Gésan-Guiziou, D. Roldan-Calbo, S. Beaulieu, F. Michel, Limiting flux in skimmed milk ultrafiltration: impact of electrostatic repulsion due to casein micelles, Desalination 175 (2005) 49–59. [18] W. Kühnl, A. Piry, V. Kaufmann, T. Grein, S. Ripperger, U. Kulozik, Impact of colloidal interactions on the flux in cross-flow microfiltration of milk at different pH values: A surface energy approach, J. Memb. Sci. 352 (2010) 107–115. [19] A.J.E. Jimenez-Lopez, N. Leconte, F. Garnier-Lambrouin, A. Bouchoux, F. Rousseau, G. Gésan-Guiziou, Ionic strength dependence of skimmed milk microfiltration: Relations between filtration performance, deposit layer characteristics and colloidal properties of casein micelles, J. Memb. Sci. 369 (2011) 404–413. [20] L. Paugam, D. Delaunay, M. Rabiller-Baudry, Skim milk ultrafiltration with a PES membrane: effect of milk thermal pretreatment and concentration on the irreversible fouling, Procedia Eng. 44 (2012) 2038–2040. [21] K.F. Eckner, E.A. Zottola, Effects of temperature and pH during membrane concentration of skim milk on fouling and cleaning efficiency, Milchwissenschaft 48 (1993) 187–191. [22] C. Pompei, P. Resmini, C. Peri, Influence of temperature on permeation rate and retention, J. Food Sci. 38 (1973) 867–870. [23] D.Z. Liu, M.G. Weeks, D.E. Dunstan, G.J.O. Martin, Alterations to the composition of casein micelles and retentate serum during ultrafiltration of skim milk at 10 and 40 °C, Int. Dairy J. 35 (2014) 63–69. [24] D. Rose, H. Tessier, Composition of ultrafiltrates from milk heated at 80 to 230° F. in relation to heat stability, J. Dairy Sci. 42 (1959) 969–980. [25] D. St-Gelais, S. Haché, M. Gros-Louis, Combined effects of temperature, acidification, and diafiltration on composition of skim Milk retentate and permeate, J. Dairy Sci. 75 (1992) 1167–1172. [26] D.J. Kapsimalis, R.R. Zall, Ultrafiltration of skim milk at refrigerated temperatures, J. Dairy Sci. 64 (1981) 1945–1950. [27] L. Paugam, D. Delaunay, N.W. Diagne, M. Rabiller-Baudry, Cleaning of skim milk PES ultrafiltration membrane: on the real effect of nitric acid step, J. Memb. Sci. 428 (2013) 275–280. [28] L. Bégoin, M. Rabiller-Baudry, B. Chaufer, C. Faille, P. Blanpain-Avet, T. Bénézech, et al., Methodology of analysis of a spiral-wound module. Application to PES membrane for ultrafiltration of skimmed milk, Desalination 192 (2006) 40–53. [29] M. Koutake, Y. Uchida, T. Sato, K. Shimoda, A. Watanabe, S. Nakao, Filtration membrane fouling in ultrafiltration of skim milk, 1: Causes and cleaning, J. Agric. Chem. Soc. Japan. 61 (1987) 677–681. [30] M. Rabiller-Baudry, M. Le Maux, B. Chaufer, L. Begoin, Characterisation of cleaned and fouled membrane by ATR-FTIR and EDX analysis coupled with SEM: application to UF of skimmed milk with a PES membrane, Desalination 146 (2002) 123–128. [31] G. Daufin, U. Merin, J.P. Labbe, A. Quemerais, F.L. Kerhervé, Cleaning of inorganic membranes after whey and milk ultrafiltration, Biotechnol. Bioeng. 38 (1991) 82–89. [32] M. Kazemimoghadam, T. Mohammadi, Chemical cleaning of ultrafiltration membranes in the milk industry, Desalination 204 (2007) 213–218. [33] L. Paugam, M. Rabiller-Baudry, D. Delaunay, Physico-chemical effect of simple alkaline and acid solutions in cleaning sequences of spiral ultrafiltration membranes fouled by skim milk, Desalination 200 (2006) 192–194. [34] M. Rabiller-Baudry, L. Bégoin, D. Delaunay, L. Paugam, B. Chaufer, A dual approach of membrane cleaning based on physico-chemistry and hydrodynamics: Application
4.5. Implications of protein-dominated fouling The confirmation of protein-dominated fouling in skim milk UF at both 10 °C and 50 °C has implications on cleaning optimisation. As the current standard cleaning protocol consists of alkaline and acid cleaning to remove organic and mineral fouling respectively, the absence of mineral fouling means that acid cleaning is not in fact necessary [34]. Further, acid cleaning has been shown to be ineffective at removing proteins [27,31–33], and thus the omission the acid cleaning step has been suggested [34]. Omission of acid cleaning along with improvements in alkaline cleaning formulations can thus significantly reduce the water and chemical consumption, as well as process downtime [34]. 5. Conclusions The influence of processing temperature on the flux decline behaviour of skim milk UF has been comprehensively examined for the first time in this study. Despite higher fluxes, UF processing at higher temperatures resulted in higher magnitudes and rates of irreversible fouling. Fouling at 10 °C was shown to be primarily proteinaceous, consistent with fouling at 50 °C elucidated in previous studies. Fouling consisted of mainly peptides and α-LA, and the composition of the fouling layer varied with temperature. In particular, higher UF processing temperatures resulted in an increase in pore blocking by α-LA, as well as β-LG deposition/adsorption on the membrane surface at 50 °C. These were attributed to thermal pore expansion and reversible conformational changes respectively. This study extends our fundamental understanding of fouling in skim milk UF, and draws a link between existing knowledge established for UF at 50 °C with UF at 10 °C, for which there are very few fouling investigations despite its widespread use. Processing temperature could be reconsidered as a parameter for fouling mitigation and process optimisation. As fouling at 10 °C is protein-dominated, omission of acid cleaning and improving alkaline cleaning formulations as previously suggested by other authors [34] should be strongly considered. Acknowledgements This work was supported by Dairy Innovation Australia Limited (DIAL) and the Australian Research Council (ARC) through LP110200570. Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.seppur.2017.12.029. References [1] M. Cheryan, Ultrafiltration and Microfiltration Handbook, Technomic Publishing, Penssylvania, 1998. [2] P. Walstra, T.J. Geurts, A. Noomen, A. Jellema, M.A.J.S. van Boekel, Dairy Technology: Principles of Milk Properties and Processes, Marcel Dekker Inc., New
330
Separation and Purification Technology 195 (2018) 322–331
K.S.Y. Ng et al.
[35] [36] [37] [38]
[39] [40]
[41]
[42] [43] [44]
[45] [46] [47] [48]
[49] [50] [51] [52]
[53]
[54] N.P. Wong, Fundamentals of Dairy Chemistry, third ed., Aspen Publishers Inc., 1999. [55] C. Georges, S. Guinand, Reversible dissociation of β2-lacto-globulin, at pH > 5.5. I. Light-diffusion study, J. Chem. Phys. 57 (1960) 606–614. [56] M. Dupont, Étude d’une étape réversible dans la thermodénaturation de la β-lactoglobuline bovine a, Biochim. Biophys. Acta (BBA)-Biophysics Incl. Photosynth. 102 (1965) 500–513. [57] X.L. Qi, S. Brownlow, C. Holt, P. Sellers, Thermal denaturation of beta-lactoglobulin: Effect of protein concentration at pH 6.75 and and 8.05, Biochim. Biophys. Acta – Protein Struct. Mol. Enzymol 1248 (1995) 43–49. [58] S.T. Kelly, W.S. Opong, A.L. Zydney, The influence of protein aggregates on the fouling of microfiltration membranes during stirred cell filtration, J. Memb. Sci. 80 (1993) 175–187. [59] S.T. Kelly, A.L. Zydney, Mechanisms for BSA fouling during microfiltration, J. Memb. Sci. 107 (1995) 115–127. [60] T. Maruyama, S. Katoh, M. Nakajima, H. Nabetani, Mechanism of bovine serum albumin aggregation during ultrafiltration, Biotechnol. Bioeng. 75 (2001) 233–238. [61] D.N. Lee, R.L. Merson, Chemical treatments of cottage cheese whey to reduce fouling of ultrafiltration membranes, J. Food Sci. 41 (1976) 778–786. [62] T. Steinhauer, S. Hanély, K. Bogendörfer, U. Kulozik, Temperature dependent membrane fouling during filtration of whey and whey proteins, J. Memb. Sci. 492 (2015) 364–370. [63] H.Q. Dang, W.E. Price, L.D. Nghiem, The effects of feed solution temperature on pore size and trace organic contaminant rejection by the nanofiltration membrane NF270, Sep. Purif. Technol. 125 (2014) 43–51. [64] R.R. Sharma, R. Agrawal, S. Chellam, Temperature effects on sieving characteristics of thin-film composite nanofiltration membranes: pore size distributions and transport parameters, J. Memb. Sci. 223 (2003) 69–87. [65] R.G. Jensen, Handbook of Milk Composition, Academic Press, San Diego, 1995. [66] H. Neurath, The Proteins – Chemistry, Biological Activity, and Methods: Volume II, Part B, Elsevier, 2012. [67] A.C. Lucarini, B.V. Kilikian, Comparative study of Lowry and Bradford methods: Interfering substances, Biotechnol. Tech. 13 (1999) 149–154, http://dx.doi.org/10. 1023/A:1008995609027. [68] J.J. Sedmak, S.E. Grossberg, A rapid, sensitive, and versatile assay for protein using Coomassie brilliant blue G250, Anal. Biochem. 79 (1977) 544–552, http://dx.doi. org/10.1016/0003-2697(77)90428-6. [69] S.-C. Chou, A.A. Goldstein, Chromogenic groupings in the Lowry protein determination, Biochem. J. 75 (1960). [70] P.S. Tong, D.M. Barbano, W.K. Jordan, Characterization of proteinaceous membrane foulants from whey ultrafiltration, J. Dairy Sci. 72 (1989) 1435–1442. [71] M. Fontyn, A membrane-foulant interaction study, Landbouwuniversiteit te Wageningen, 1991. [72] C. Jönsson, A.-S. Jönsson, Influence of the membrane material on the adsorptive fouling of ultrafiltration membranes, J. Memb. Sci. 108 (1995) 79–87. [73] J.M. Sheldon, I.M. Reed, C.R. Hawes, The fine-structure of ultrafiltration membranes. II. Protein fouled membranes, J. Memb. Sci. 62 (1991) 87–102.
to PES membrane of dairy industry, Chem. Eng. Process. Process Intensif. 47 (2008) 267–275. A.-S. Jönsson, Influence of shear rate on the flux during ultrafiltration of colloidal substances, J. Memb. Sci. 79 (1993) 93–99. J. Kestin, M. Sokolov, W.A. Wakeham, Viscosity of liquid water in the range of -8 °C to 150 °C, J. Phys. Chem. 7 (1978) 1978. J.C. Crittenden, R.R. Trussell, D.W. Hand, K.J. Howe, G. Tchobanoglous, MWH’s Water Treatment: Principles and Design, third ed., John Wiley & Sons, 2012. P.S. Tong, D.M. Barbano, M.A. Rudan, Characterization of proteinaceous membrane foulants and flux decline during the early stages of whole milk ultrafiltration, J. Dairy Sci. (1988) 604–612. G. Rice, A. Barber, A. O’Connor, G. Stevens, S. Kentish, Fouling of NF membranes by dairy ultrafiltration permeates, J. Memb. Sci. 330 (2009) 117–126. G.S. Rice, S.E. Kentish, A.J. O’Connor, A.R. Barber, A. Pihlajamäki, M. Nyström, et al., Analysis of separation and fouling behaviour during nanofiltration of dairy ultrafiltration permeates, Desalination 236 (2009) 23–29. J.H. Hanemaaijer, T. Robbertsen, T. van den Boomgaard, J.W. Gunnink, Fouling of ultrafiltration membranes. The role of protein adsorption and salt precipitation, J. Memb. Sci. 40 (1989) 199–217. H.G.R. Rao, M.J. Lewis, A.S. Grandison, Effect of soluble calcium of milk on fouling of ultrafiltration membranes, J. Sci. Food Agric. 65 (1994) 249–256. F. Gaucheron, Milk salts: distribution and analysis, Encycl. Dairy Sci. second ed., Academic Press, San Diego, 2002, pp. 908–916. D.Z. Liu, M.G. Weeks, D.E. Dunstan, G.J.O. Martin, Temperature-dependent dynamics of bovine casein micelles in the range 10–40 °C, Food Chem. 141 (2013) 4081–4086. J.L. Maubois, Ultrafiltration of whey, Int. J. Dairy Technol. 33 (1980) 55–58. H.G. Kessler, C. Gernedel, K. Nakanishi, The effect of low molecular weight milk constituents on the flux in ultrafiltration, Milchwissenschaft 37 (1982) 584–587. W. Norde, My voyage of discovery to proteins in flatland. and beyond, Colloids Surfaces B Biointerfaces 61 (2008) 1–9. K.S.Y. Ng, M. Haribabu, D.J.E. Harvie, D.E. Dunstan, G.J.O. Martin, Mechanisms of flux decline in skim milk ultrafiltration: A review, J. Memb. Sci. 523 (2017) 144–162. E.A. Permyakov, alpha-Lactalbumin, Nova Science Publishers, Inc., New York, 2005. E. Iritani, A review on modeling of pore-blocking behaviors of membranes during pressurized membrane filtration, Dry. Technol. 31 (2013) 146–162. C.-C. Ho, A.L. Zydney, A combined pore blockage and cake filtration model for protein fouling during microfiltration, J. Colloid Interface Sci. 232 (2000) 389–399. E. Iritani, Y. Mukai, M. Furuta, T. Kawakami, N. Katagiri, Blocking resistance of membrane during cake filtration of dilute suspensions, AIChE J. 51 (2005) 2609–2614. M. Rabiller-Baudry, D. Delaunay, L. Paugam, A. Pihlajamäki, M. Nyström, Complementary characterisations by streaming potential and FTIR-ATR of surface of virgin and fouled PES ultrafiltration membrane: What kind of information on fouling occurrence, Surf. Electr. Phenom. Membr. Microchannels, Transw. Res. Netw. Ed. Kerala, India. (2008) 45–60.
331