Langmuir 2008, 24, 4523-4529
4523
Spatial and Temporal in Situ Evolution of the Concentration Profile during Casein Micelle Ultrafiltration Probed by Small-Angle X-ray Scattering C. David,† F. Pignon,*,† T. Narayanan,‡ M. Sztucki,‡ G. Ge´san-Guiziou,§ and A. Magnin† Laboratoire de Rhe´ ologie, INPG, UJF Grenoble I, CNRS, UMR 5520, Boite Postale 53, F-38041 Grenoble Cedex 9, France, European Synchrotron Radiation Facility, Boite Postale 220, F-38043 Grenoble Cedex 9, France, and INRA, Agrocampus Rennes, UMR 1253, Science et Technologie du Lait et de l’Œuf, 65 rue de Saint Brieuc, F-35042 Rennes Cedex, France ReceiVed October 19, 2007. In Final Form: December 29, 2007 Understanding the mechanisms involved in structural development in the vicinity of membrane constitutes a considerable challenge in the improvement of ultrafiltration process in industrial applications. In situ small-angle X-ray scattering (SAXS) performed with custom-made ultrafiltration cell has permitted the structural arrangement to be probed and concentration profiles to be obtained in deposited layers during frontal filtration of casein micelle suspension. SAXS allowed the structure of the accumulated layers of casein micelles between 280 µm and 1 mm from the membrane surface to be followed at length scales from a few nanometers to about 100 nm. These results have been combined with hydrodynamic measurements (permeation flux) and rheological investigations. Under frontal filtration, the time dependence of concentration at different distances from the membrane surface has been obtained. This temporal evolution is marked by an exponential increase of the concentration followed by a slower growth which has been associated with a change in the rheological behavior of the suspension from Newtonian to shear thinning behavior.
1. Introduction Ultrafiltration processes are widely used in industrial applications (for example, wastewater treatment or separation in the food industry and biotechnology). Low energy cost, the ability to operate near ambient conditions, and easy use are advantages of this technique.1,2 The main obstacle limiting its performance is the accumulation of matter on the membrane surface (concentration polarization, gelation, and/or deposit buildup). Improvement of this process depends on the understanding of mechanisms involved in the formation of deposits, which then allow the overall performance to be controlled and predicted. In the past, experiments, modeling, and numerical simulations have been attempted to accurately explain and quantify the decrease of flux due to structural development in the vicinity of the membrane.3-17 Some of them give satisfactory results concerning * To whom correspondence should be addressed. E-mail: pignon@ ujf-grenoble.fr. † Laboratoire de Rhe ´ ologie. ‡ European Synchrotron Radiation Facility. § INRA. (1) Aimar, P.; Daufin, G. Tech. Ing. 2004, F3, 250. (2) Cheryan, M. Ultrafiltration handbook; Technomic Publishing Co.: Lancaster, PA, 1986. (3) Waite, T. D.; Scha¨fer, A. I.; Fane, A. G.; Heuer A. J. Colloid Interface Sci. 1999, 212 (2), 264-274. (4) Bacchin, P.; Aimar, P.; Sanchez, V. AIChE J. 1995, 41 (2), 368-376. (5) Razavi, M. A.; Mortazavi, A.; Mousavi, M. J. Membr. Sci. 2003, 220 (1-2), 47-58. (6) Bowen, W. R.; Jenner, F. Dynamic Chem. Eng. Sci. 1995, 50 (11), 17071738. (7) Le Berre, G.; Daufin, G. J. Membr. Sci. 1996, 117 (1), 261-270. (8) Ge´san-Guiziou, G.; Boyaval, E.; Daufin, G. J. Membr. Sci. 1999, 158 (1), 211-222. (9) Chen, V.; Fane, A. G.; Madaeni, S.; Wenten, I. G. J. Membr. Sci. 1997, 125 (1), 109-122. (10) Bowen, W. R.; Mongruel, A.; Williams, P. M. Chem. Eng. Sci. 1996, 51 (18), 4321-4333. (11) Song, L. F.; Elimelech, M. Theory J. Chem. Soc., Faraday Trans. 1995, 91 (19), 3389-3398. (12) Jo¨nsson, A. S.; Jo¨nsson, B. J. Colloid Interface Sci. 1996, 180 (2), 504518.
the macroscopic behavior but have limited validity at the microscopic scale. Previous experimental works have probed the structural organization of deposits.18-25 Even if hydrodynamic measurements (permeation flux and applied pressure) give useful information, they are not sufficient to highlight the mechanisms involved in the filtration process. The interplay between mesoscopic and macroscopic observations has been identified as a key for the development of mechanistic filtration models. Moreover, the lack of pertinent experimental data prevents direct validation of these models. In this paper, an alternative way for directly investigating flux decline and the filtration process is proposed. Small-angle X-ray scattering (SAXS) has been applied on ultrafiltration cells to scan the interior of the concentration polarization layer (and deposit) during its growth and to obtain precise time-dependent structural changes. This technique has allowed in situ measurements to be performed from the beginning of filtration. SAXS (13) Bacchin, P.; Si-Hassen, D.; Starov, V.; Clifton, M. J.; Aimar, P. Chem. Eng. Sci. 2002, 57 (1), 77-91. (14) Agashichev, S. P. J. Membr. Sci. 2006, 285 (1-2), 96-101. (15) Chen, J. C.; Elimelech, M.; Kim, A. S. J. Membr. Sci. 2005, 255, 291305. (16) Bacchin, P.; Espinasse, B.; Bessiere, Y.; Fletcher, D. F.; Aimar P. Desalination 2006, 192 (1-3), 74-81. (17) Madeline, J. B.; Meireles, M.; Persello, J.; Martin, C.; Botet, R.; Schweins, R.; Cabane, B. Pure Appl. Chem. 2005, 77 (8), 1369-1394. (18) Pignon, F.; Magnin, A.; Piau, J. M.; Cabane, B.; Aimar, P.; Meireles, M.; Lindner, P. J. Membr. Sci. 2000, 174 (2), 189-204. (19) Su, T. J.; Lu, J. R.; Cui, Z. F.; Thomas, R. K.; Heenan, R. K. Langmuir 1998, 14, 5517-5520. (20) Stefanopoulos, K. L.; Romanos, G. E.; Mitropoulos, A. C.; Kanellopoulos, N. K.; Heenan, R. M. J. Membr. Sci. 1999, 153 (1), 1-7. (21) Antelmi, D.; Cabane, B.; Meireles, M.; Aimar, P. Langmuir 2001, 17 (22), 7137-7144. (22) Pignon, F.; Alemdar, A.; Magnin, A.; Narayanan, T. Langmuir 2003, 19 (21), 8638-8645. (23) Pignon, F.; Belina, G.; Narayanan, T.; Paubel, X.; Magnin A.; Ge´sanGuiziou, G. J. Chem. Phys. 2004, 121 (16), 8138-8146. (24) Chen, J. C.; Li, Q.; Elimelech, M. AdV. Colloid Interface Sci. 2004, 107 (2-3), 83-108. (25) Madeline, J. B.; Meireles, M.; Bourgerette, C.; Botet, R.; Schweins, R.; Cabane, B. Langmuir 2007, 23, 1645-1658.
10.1021/la703256s CCC: $40.75 © 2008 American Chemical Society Published on Web 04/29/2008
4524 Langmuir, Vol. 24, No. 9, 2008
DaVid et al.
provides information about the average size and structure of particles (in the range of a few nanometers to about 100 nm) over a small scattering volume (100 µm in thickness). The spatial and temporal resolutions of the measurements enabled access of colloid concentration profiles between 280 µm and 1 mm from the membrane surface. Profiles were obtained during the first minutes of ultrafiltration on a polysulfone membrane (100 kg/mol) of aqueous casein micelle suspension with an initial concentration, C0, corresponding to the casein micelle content of skim milk. The results obtained by the use of SAXS have been compared to hydrodynamic measurements (permeation flux and applied pressure). Rheometric measurements have also been performed to characterize the evolution of viscosity versus shear flow properties for several casein micelle suspensions with concentration corresponding to those attained in the deposited layers. 2. Materials and Methods 2.1. Casein Micelle Suspensions. The casein micelle suspensions were obtained by dissolution of a standard commercial “high protein content powder” (Promilk 852B, Ingre´dia, 62 Arras, France) in deionized water. The dry powder contained about 86% (w/w) total protein, 92% (w/w) of which was casein micelles. To avoid bacterial development, sodium azide (NaN3; Sigma-Aldrich, 38 Saint Quentin Fallavier, France) was added to obtain a final concentration of 0.2 g/L of solution. The standard concentration of the powder, Cp0, used for the experiments has been chosen in such a way that the casein micelle content of the suspension corresponds to that of an average skim milk C0 ) 0.0278 g/g = 29 g/L. The equivalent mass of the powder (about 35.2 g) was dispersed in water and stirred to obtain a 1 kg protein suspension. The dissolution was done at 45 °C for 1 h. Several suspensions with higher and lower protein contents (6C0 to 0.01C0) have been prepared by addition of the powder or dilution in water, respectively. The mass fraction, φm, and the volume fraction, φv, of casein micelles are calculated for the corresponding casein micelle concentration, C, investigated. The volume fraction reflects the closeness between the casein micelles. The so-called voluminosity, qc, corresponds to the hydrodynamic volume of a single casein micelle. This voluminosity or specific volume of casein micelles expressed in volume per mass unit varies from 3 to 4 mL/g.26,27 Owing to the relatively high concentration range studied in this work, it is reasonable to assume a lower limit of qc ) 3 mL/g. When a mass, mp, of Promilk 852B powder is dispersed in water, it occupies a volume of mpqp. qp is the specific volume of the hydrated powder, and its experimental value was evaluated to be 0.64 mL/g. The approximate calculated volume fraction, φv, of casein micelles is then mcqc ) mpqp + meqe volume occupied by casein micelles (1) volume occupied by hydrated powder + added water volume
φv )
where mp is the mass of the powder, mc is the mass of casein (mc ) 0.86 × 0.92mp ) 0.79mp) for Promilk 852 B), me is the mass of water, qp is the specific volume of the powder (0.64 mL/g), qc is the specific volume of casein (3 mL/g), and qe is the specific volume of water (1 mL/g). The powder concentration, Cp, is defined as follows: Cp )
mp mp + m e
(2)
The casein micelle concentration C(w/w) or mass fraction φm is defined as follows: mc mc C(w/w) ) φm ) ) C m p + me m p p
(3)
Figure 1. Schematic view of the SAXS filtration cell used here in frontal mode filtration (cross-flow 0). Table 1. Concentration, Mass Fraction, and Corresponding Volume Fraction of Casein Micelle Suspensions (Promilk 852B Powder Dispersed in Water) C/C0
φm
φv
C/C0
φm
φv
1 2 3 4
0.028 0.056 0.084 0.111
0.085 0.171 0.261 0.352
5 6 7
0.139 0.167 0.195
0.446 0.543 0.642
Combining formulas 1-3, the volume fraction of the casein micelles can be calculated as follows: φv )
φmqc mp φ (q - qe) + qe mc m p
(4)
Table 1 presents the relationship between concentrations and volume fractions for the entire range studied in this paper. 2.2. Experimental Setup and Procedure. Filtration experiments have been realized using a custom-made “SAXS filtration cell” (Figure 1) developed and machined at the Laboratoire de Rhe´ologie, Grenoble, France. This cell allowed both hydrodynamic measurements (macroscopic) and SAXS measurements (nanoscale) to be performed. It was made of transparent polycarbonate and contained a flat polysulfone ultrafiltration membrane (100 kg/mol, Ple´iade Rayflow Novasep, 01 Miribel, France). The membrane was fastened and sealed between the permeate recovery and retentate canals. The small wall thickness (0.5 mm) reduced parasitic scattering and beam attenuation. This cell operated similarly to the cell presented elsewhere22 and dedicated to magnetic field experiments. The SAXS filtration cell (Figure 1) could be used for cross-flow and frontal mode filtration. In this paper, only results concerning frontal mode filtration are presented. The retentate canal was 100 mm long, 5 mm high, and 1 mm wide. The membrane section was then 100 mm long and 1 mm wide. Permeate has been continuously recovered in a recipient, and its mass was automatically registered every 10 s with an accuracy of 0.001 g (Precisa 400M, Precisa France S.A., 78 Poissy, France). The permeation flux was calculated afterward. The experiments were done at room temperature (T ) 23 °C). Pressure was applied via purified compressed air and controlled by a pressure gauge (FP 110, FGP Sensors & Instrument, 78 Les Clayes sous bois, France). A temperature-controlled flow-through capillary cell (diameter of 1.85 mm and wall thickness of 10 µm) was used to measure the absolute scattering intensity of casein micelle suspensions versus concentration necessary for the calibration of the concentration profile in the SAXS filtration cell. This flow-through capillary setup is very well adapted for investigating low-scattering liquid samples as sample and solvent scattering can be measured in the same place, allowing very reliable background subtraction. 2.3. Small-Angle X-ray Scattering. All X-ray scattering measurements have been performed at the High-Brilliance beamline (ID2) at the European Synchrotron Radiation Facility (ESRF), Grenoble, France,28 using a wavelength of the incident X-ray beam of λ ) 0.995 Å. Two different sample-detector distances (2 and 10 m) were used, covering a scattering vector range of 1.5 × 10-2 (26) Holt, C. AdV. Protein Chem. 1992, 43, 63-151. (27) Walstra, P. J. Dairy Res. 1979, 46, 317. (28) Narayanan, T.; Diat, O.; Bosecke, P. Nucl. Instrum. Methods Phys. Res., Sect. A 2001, 467-468 (2), 1005-1009.
Casein Micelle Ultrafiltration
Figure 2. Evolution of the transmitted X-ray signal with distance from the membrane surface (H) through the SAXS filtration cell. A schematic view of the cell containing a casein micelle suspension (the concentration C0 ) 29 g of casein/L) at rest is drawn on the left side to better understand the shape of the curve. Z1 corresponds to the permeate canal, Z2 represents the membrane, Z3 is a zone in the retentate canal where the SAXS signal is perturbed, and finally Z4 is the SAXS measurement zone.
Langmuir, Vol. 24, No. 9, 2008 4525
Figure 3. Calibration curve used to determine the concentration of casein micelles in the polarization layer and cake deposit during ultrafiltration of the casein micelle suspension. Ten suspensions of Promilk 852B powder with concentration ranging from 0.01C0 to 6C0 were used. C0 represents the casein micelle concentration for a standard skim milk.
nm-1 < q < 6.1 nm-1 (q ) (4π/λ) sin(θ/2), θ being the scattering angle). Structural information in the range from a few nanometers to about 100 nm could then be obtained. To reach a high spatial resolution in the growth direction of the casein accumulation, the X-ray beam was highly collimated and the beam profile was carefully determined by a high-resolution CCD camera. The sample area illuminated by the beam was approximately 250 µm wide and 70 µm high. Several qualitative results could be directly extracted from the 2D scattering patterns. The analysis of the symmetry of the 2D patterns indicates the anisotropy and orientation of the particles. In the case of isotropic scattering patterns, the scattering intensity, I(q), in absolute units was obtained by azimuthal averaging of the normalized 2D patterns. I(q) gives access to the particle mean size, shape, and structural arrangement.18,21-25,28,29 The absolute scattering intensity, I(q), has been used to determine the concentration of the casein micelles suspensions. For accurate background subtraction, scattering of deionized water (equivalent to the signal of the suspending solvent, results not shown) was recorded for each height in the cell. 2.4. Position Calibration in the SAXS Filtration Cell. Figure 2 shows a schematic view of the SAXS filtration cell containing a casein micelle suspension at rest and the transmitted intensity through this cell at different distances (H) from the membrane surface. Several zones have been identified as a function of the vertical beam position. The first zone (Z1) corresponds to the canal where permeate was recovered. There, the X-ray beam was completely absorbed by the polycarbonate support (20 mm in depth). The end of this zone is defined by the increase of the transmitted signal (H ) -200 µm in Figure 2). In the second zone (Z2) the intensity transmitted by the membrane (2 mm in depth) was recorded. The scattering signal in the third zone (Z3) originates partially from the first deposited layers of casein micelles. However, the recorded scattering and absorption signal is still influenced by the X-ray beam partially crossing the polycarbonate support and the membrane. This is due to a small inclination of the cell relative to the X-ray beam (on the order of ∼1° over a total depth of the cell of 20 mm) and certain unavoidable undulations in the membrane due to its mounting. These effects do not have any influence on the scattering above 280 µm from the membrane in the fourth zone (Z4). Therefore, all the studies on accumulated layers were carried out in this zone. The measured scattering intensity at a given position H in this zone corresponds to the signal of a deposited layer of about 100 µm height defined by the beam size (70 µm) and the 1° inclination.
The depth of the scattering volume is defined by the canal width of 1 mm. Also two polycarbonate windows, each of 0.5 mm thickness, contributed a constant scattering which was subtracted together with the deionized water background. 2.5. Rheometric Measurements. The rheometric measurements were carried out on several casein suspensions prepared at different concentrations as explained in section 2.1. These tests were conducted with a stress-controlled rheometer (ARG2, TA Instrument, 78 Guyancourt, France) with a stainless steel cone and plate geometry (diameter 60 mm, angle 1°, gap 29 µm). The temperature of the samples was controlled (23 °C). Any slip at the wall was detected, and the atmosphere around the sample was controlled to avoid evaporation during the tests.30
(29) Marchin, S.; Putaux, J. L.; Pignon, F.; Le´onil, J. J. Chem. Phys. 2007, 126, 045101.
(30) Magnin, A.; Piau, J. M. J. Non-Newtonian Fluid Mech. 1990, 36, 85108.
3. Results and Discussion 3.1. Static Measurements of Casein Micelle Suspensions. 3.1.1. Concentration Calibration. To calculate the particle concentration in layers of matter accumulated on the membrane surface, a calibration curve has been determined (Figure 3). The measurements were performed in the capillary flow-through cell at 23 °C. The measured SAXS profiles were normalized to an absolute scale using the standard procedure.28 Several scattering curves with suspensions of various concentrations were obtained (results not shown). The intensity at q ) 1 nm-1 has been used to determine the calibration curve (Figure 3). At this scattering vector, interactions between casein micelles do not influence the intensity level. Increasing the amount of matter in the suspension results in a proportional increase of the scattered intensity level. With the obtained calibration curve (Figure 3), the particle concentration of any casein micelle suspension could be determined by introducing the corresponding absolute scattered intensity at a scattering vector q ) 1 nm-1 on the graph. This method already employed elsewhere22,23 has been used to determine the evolution of the concentration profile in the SAXS filtration cells. 3.1.2. Equilibrium Structure of Casein Micelle Suspensions. Some complementary static measurements were made on the casein micelle suspension with different casein micelle contents (Figure 4). These measurements were performed on the wide scattering range (1.5 × 10-2 < q < 6.1 nm-1) to access complete
4526 Langmuir, Vol. 24, No. 9, 2008
DaVid et al.
Figure 5. Correspondence between the results obtained with the casein micelle suspension (Promilk 852B) and another phosphocaseinate suspension (PPCN).23 C0 represents the casein micelle concentration for a standard skim milk.
Figure 4. (i) Evolution of scattering patterns with increasing casein micelle concentration. (ii) Superposition of the curves. C0 represents the casein micelle concentration for a standard skim milk. T ) 15 °C.
nanometric structural information. The results presented in Figure 4 were carried out with the standard commercial milk (Promilk 852B). Data are well in accordance with precedent investigations (Figure 5) carried out with other phosphocaseinate dry model powders23 (PPCN and Low Heat) as well as with fresh cow milk.29 In the first approach,23 the analysis of the scattering intensity of these casein micelles has been performed using the unified fit function given by Beaucage.31 The scattering intensity, I(q), over the low q branch (corresponding to the larger structure) was fitted by a Debye-Bu¨che function32 decaying by a q-4 power law. This clearly demonstrates that the larger units consist of globular micelles. The high q region showed a significantly different behavior with I(q) decreasing like q-2.6. If the scattering intensity originates only from an entangled chainlike conformation of proteins, the scattering intensity should follow a q-2 power law decay. The observed power law exponent could be attributed to the scattering due to nanometric calcium phosphate particles reticulated in the protein matrix. Consequently, the results of these SAXS experiments suit well a model which describes casein micelles as a relatively uniform matrix containing a disordered micelle calcium phosphate.33 Scattering curves displayed in Figure 4 show that no structural evolution was observed for suspensions with concentration (31) Beaucage, G.; Kammler, H. K.; Pratsinis, S. E. J. Appl. Crystallogr. 2004, 37 (4), 523-535. (32) Brumberger, H. Modern Aspects of Small-Angle Scattering; Kluwer: Dordrecht, The Netherlands, 1995.
Figure 6. Influence of the applied shear rate on the viscosity of the casein micelle suspension with different concentrations. C0 represents the casein micelle concentration for a standard skim milk. T ) 23 °C.
ranging from 0.1C0 to C0. When the casein micelle concentration is increased to 6C0, the development of a structure factor due to intermicelle interactions can be detected at small scattering vectors (q < 5 × 10-2 nm-1). In reality, 6C0 corresponds to a volume fraction of about 0.54 (which has been calculated using the known voluminosity of casein micelles of 3 mL/g).26,27 Despite this very high volume fraction, the structure factor peak is rather weak due to large polydispersity of casein micelles. This large polydispersity of the globular micelles complicates a quantitative analysis of this structure factor, and it goes beyond the scope of the present investigation. Moreover, for deducing the concentration profile, the form factor part of the scattering curve is utilized. 3.1.3. Rheological BehaVior. In Figure 6, the viscosity is plotted as a function of shear rates for different casein micelle concentrations. For concentrations below 4C0 (corresponding to a volume fraction of about 0.35), the suspensions exhibit a Newtonian behavior and the viscosity is the same, independent (33) Holt, C.; de Kruif, C. G.; Tuinier, R.; Timmins, P. A. Colloids Surf., A 2003, 213 (2-3), 275-284.
Casein Micelle Ultrafiltration
Figure 7. Temporal evolution of the concentration profile at different distances from the membrane surface (H) along filtration of the casein micelle suspension. The applied pressure is 1.2 bar, and the initial concentration of the suspension C0 ) 29 g of casein/L. Dashed lines are only given as guides for the eyes.
Figure 8. Evolution of the casein concentration with filtration time at different distances from the membrane surface under 1.2 bar. The initial concentration of the suspension C0 ) 29 g of casein/L. Dashed lines are only given as guides for the eyes.
of the shear rate applied. Above a particular concentration next to 4C0, the viscosity decreases at increasing shear rate (see Figure 6). This reveals a shear thinning behavior. The observed evolution of the rheological behavior from Newtonian to shear thinning behavior was attributed to the apparition of excluded volume interaction as well as weak electrostatic interactions between the casein micelles and reinforces the qualitative remark made with respect to scattering measurements above. 3.2. In Situ Measurements in the Frontal Filtration Cell. For short time frontal mode experiments (6 h), a standard pro-
Langmuir, Vol. 24, No. 9, 2008 4527
Figure 9. Evolution of d[ln(C/C0)]/dH with C/C0 at different filtration times. The initial suspension concentration C0 ) 29 g of casein/L. Dashed lines are only given as guides for the eyes.
tein suspension (C ) C0) was filtered under constant pressure (1.2 bar) and at ambient temperature (23 °C). Twelve scans were obtained along the filtration time. A scan consisted in acquiring several scattering patterns each made at a different distance from the membrane surface. Using the method previously presented, the concentration of casein micelles was determined in the accumulated layers for each time and height. Several concentration profiles were obtained showing the temporal evolution of the accumulated layer of matter at the membrane surface (Figure 7). The results are deduced from azimuthally averaged absolute scattered intensities and represent the mean values of the concentration in a layer of (50 µm around the measured distance (H). It is to be noted that, for the first time, measurements were made in layers of accumulated casein micelles less than 1 mm high. The use of the SAXS technique offered the possibility to visualize the concentration profile inside the accumulation layer for nanoscale particles. As measurements were made in situ, the dynamics of the layer growth can be obtained. Using the same data, the evolution of the casein micelle concentration as a function of filtration time has been plotted at different distances from the membrane surface (Figure 8). This graph displays a continuous increase in concentration with two steps characteristic of different kinetics. In the first step when the casein micelle concentration is weak, the time dependence is exponential. At higher concentrations, the concentrating phenomenon slowed down. To illustrate this change visible in both Figures 7 and 8, the curves d[ln(C/C0)]/dH have been plotted in Figure 9. These derivatives are proportional to the diffusion coefficient. The observed minimum as a function of concentration in Figure 9 is interpreted as a different mechanism involving a transition from simple overcrowding of colloids on the membrane surface to a more constrained structure with increasing mutual casein micelle interactions. This transition is coherent with the change in macroscopic rheometric measurements which exhibit an evolution from a Newtonian to a shear thinning behavior around this concentration.
4528 Langmuir, Vol. 24, No. 9, 2008
DaVid et al.
Figure 10. Evolution of permeation flux with time during frontal filtration of the casein micelle suspension. The applied pressure is 1.2 bar, and the initial suspension concentration C0 ) 29 g of casein/ L. Square symbols represent the times when the concentration profiles were obtained. Circles correspond to the moment when the structure was analyzed in the accumulated layers.
Measurements of permeation flux have also been performed with the method previously explained (Figure 10). The permeation flux decrease with filtration time corresponds to matter accumulation. As already observed, permeation flux decreases quickly at the beginning of the experiment, whereas the process slows down at longer filtration times. Detected modifications concerning concentrating phenomena in the deposit between 180 and 240 min at 280 and 310 µm were not reflected in the permeation flux. This can be explained by the fact that flux reduction is due to the overall polarization layer buildup. In this case, local modifications in the deposit are damped. By extrapolating the observed evolution of concentration in zone Z4 to zone Z3, it can be seen that the threshold concentration for the onset of a strong decrease of flux could be reached already in the first layers within the first 30 min of filtration. In all cases, the decrease of flux only reveals the overall polarization layer buildup. Measurements of microscopic and nanoscopic structures provide a better understanding of the mechanism. The study of the rheological behavior of the accumulated colloids on the membrane surface is also necessary. It should be emphasized that with the in situ scanning SAXS technique, it was possible to detect local structure phenomena which were not elucidated by the flux measurements. This demonstrates the usefulness of our approach which could be applied to other types of systems. Complementary information was acquired about the structure of the accumulated layers at longer scale in the vicinity of the membrane. The structure of the deposited layers was analyzed at the low q region using two scans made at a 10 m sample to detector distance. These scans were made when the threshold concentration was attained in a part of the accumulated layers for filtration times of 280 and 305 min. Scattering curves and their superposition at 280 min (Figure 11) show a continuous evolution of the structure factor of interaction between casein micelles. The observed modification is identical to that obtained with static samples of concentrations C0 and 6C0 (Figure 4). This behavior can be understood in terms of the increase of the local concentration of casein micelles, and it is not directly related to the applied pressure or the hydrodynamic behavior in the deposit.
4. Conclusion Small-angle X-ray scattering combined with an ultrafiltration cell has allowed the in situ structural arrangement and concen-
Figure 11. Structure of matter accumulated on the membrane surface after 280 min of frontal filtration under 1.2 bar. The initial concentration of the suspension C0 ) 29 g of casein/L. (i) Scattering patterns obtained at different distances from the membrane surface. (ii) Superposition of the curves.
tration profiles of deposits during the ultrafiltration process of casein micelle dispersions to be accessed. The spatial (100 µm) and temporal (a fraction of a second) resolutions of SAXS enable structural information in layers of matter above 280 µm from the membrane surface at length scales ranging from a few nanometers to about 100 nm to be accessed. The combination of structural results with hydrodynamic ultrafiltration parameters has allowed pertinent information about the mechanisms involved in the frontal filtration mode to be deduced. For the first time, in frontal mode filtration of casein micelle suspensions, the time evolution of the concentration profile has been measured. The high reduction of permeation flux in the early stage of filtration was associated with an exponential increase of the concentration at the membrane surface. Above a threshold concentration, the accumulation process slows in the deposited layers. At the same time, filtration flux continues to decrease because of the growth of the layer of accumulated casein micelles. This threshold concentration also corresponds to a change of the rheological behavior from Newtonian to shear thinning of the casein micelle suspensions.
Casein Micelle Ultrafiltration
These results prove the possibility to obtain in situ information in the vicinity of membrane surfaces during ultrafiltration and offer new opportunities for better understanding the mechanisms responsible for the reduction of the efficiency of the separation process and testing the predictions of computer simulations. Acknowledgment. We thank the Conseil Re´gional Rhoˆne Alpes and the industrial partner Bressor S.A. for their financial support. This work was made within the framework of a program
Langmuir, Vol. 24, No. 9, 2008 4529
(Emergence 2005) entitled “Compre´hension et maıˆtrise des me´canismes de structuration des de´poˆts, lors de la filtration tangentielle de produits laitiers”. We sincerely acknowledge the European Synchrotron Radiation Facility for the SC 1954 beam time allocation. We thank Anne Jimenez-Lopez (Soredab S.A., France) for her help in the preparation of some of the casein micelle suspensions. LA703256S
