- Email: [email protected]
Utilization of enzymatic detergents to clean inorganic membranes fouled by whey proteins
Separation and Purification Technology 41 (2005) 147–154
Utilization of enzymatic detergents to clean inorganic membranes fouled by whey proteins b , F.A. Rieraa,∗ , R. Alvarez a ´ ´ M.A. Arg¨uelloa , S. Alvarez b
a Department of Chemical Engineering and Environmental Technology, University of Oviedo, C/Juli´ an Claver´ıa, 8, 33006 Oviedo, Spain Department of Chemical and Nuclear Engineering, ETSII, Politechnical University of Valencia, Camino de Vera, s/n, 46022 Valencia, Spain
Received in revised form 7 May 2004; accepted 10 May 2004
Abstract In this work inorganic membranes used for whey protein fractionation were cleaned with enzymatic detergents. The inorganic membrane Carbosep® M1 (Orelis S.A., France), of 150 kg/mol molecular weight cut-off and ZrO2 filtering layer, was used and the commercial detergent P3-Ultrasil® 62 (Henkel Ib´erica S.A., Spain) was selected for the cleaning. Hydraulic and chemical methods were considered to characterize the membrane cleanliness. Cleaning efficiency was observed to be a function of the operating conditions: recycling versus non-recycling of permeate, cleaning solution pH, enzymatic agent concentration and cleaning time. The optimum conditions to perform the cleaning were related to the optimum conditions to hydrolyze whey proteins in a batch reactor. Very high cleaning efficiencies (∼100%) were reached in short operating times (20 min). However, residual matter was detected on the membrane surface after the cleaning. © 2004 Elsevier B.V. All rights reserved. Keywords: Inorganic membranes; Enzymatic cleaning; Whey protein solutions; Hydraulic cleanliness; Chemical cleanliness
1. Introduction For a long time proteases have been considered as being very useful ingredients in modern detergents [1]. The idea of using enzymes in detergents was first described by R¨ohm in 1913 [2]. At that time, only pancreatic enzymes were available and they were not granulated or otherwise protected by physical means. Therefore the storage ability, long-term efficiency and other health aspects were unsatisfactory [3]. The development of enzymes of microbiological origin started in the 1950s with the production of BIO40. This formulation contained a bacterial protease, what was a significant improvement compared to the previous pancreatic products; however the detergent had the disadvantage of a neutral pH [4]. During the 1970s a new generation of high active alkaline detergent enzymes were developed and passed through different generations of physical forms (powders, pills and ∗
Corresponding author. Tel.: +34 8 5103436; fax: +34 8 5103434. E-mail address: [email protected] (F.A. Riera).
1383-5866/$ – see front matter © 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.seppur.2004.05.005
encapsulates) [5]. The modern technologies (recombinant DNA, genetic engineering and protein engineering) were first applied in 1988 to produce LipolaseTM [4]. Although proteases are still used as one of the main enzymes in detergent formulations, ␣-amylases, cellulases and lipases are now successfully applied in household laundry detergents, industrial laundering and machine dishwashing. In membrane technology, a very important research goal is the development of non aggressive cleaning methods able to increase the lifetime of membranes and the proposal of short cleaning times [6]. The use of enzymatic cleaners is of interest because they operate in mild conditions, which is a determining factor for their application to the cleaning of membranes sensitive to chemicals, pH and/or temperature [7,8]. Enzymatic agents can improve the cleaning efficiency and reduce the amount of chemicals needed and the energy costs by working at a lower temperature. Moreover, they are biodegradable and the preparation of tailor-made cleaners is possible. Enzymatic formulations were used by several authors to clean organic membranes [9–11]. However, they reported low cleaning efficiencies and extended cleaning times.
148
M.A. Arg¨uello et al. / Separation and Purification Technology 41 (2005) 147–154
Nomenclature a Am cp ERW HD J Jwi n ntot P R Rc Rcw Rf Rif Rm Rres Rrf Ruf V
enzymatic activity (u) membrane area (m2 ) cleaning parameter (u/m) cleaning efficiency hydrolysis degree permeate flux (L/h m2 ) initial water flux (L/h m2 ) number of peptidic linkages hydrolyzed per unit of substrate mass number of peptidic linkages per unit of substrate mass transmembrane pressure (MPa) membrane hydraulic resistance (m−1 ) hydraulic resistance during the cleaning process (m−1 ) hydraulic resistance of the cleaned membrane (m−1 ) resistance due to membrane fouling (m−1 ) resistance due to irreversible fouling (m−1 ) intrinsic hydraulic resistance (m−1 ) residual hydraulic resistance (m−1 ) resistance due to reversible fouling (m−1 ) total resistance of the fouled membrane (m−1 ) volume of cleaning solution (m3 )
Greek letters µ dynamic viscosity (Pa s) ν kinematic viscosity (m2 /s) ρ density (kg/m3 ) The results reported in this paper refer to the cleaning of inorganic membranes used for whey protein fractionation. The feed material is a whey protein concentrate (WPC) with low mineral and residual fat content. Therefore proteins are expected to be the main compounds contributing to membrane fouling [12,13]. In a previous paper by the authors [14], several enzymatic and non-enzymatic commercial detergents were used to hydrolyze whey proteins in a discontinuous reactor. It was concluded that only the enzymatic detergents were able to produce a significant hydrolysis of whey proteins. The best results were achieved with P3-Ultrasil® 62 (Henkel Ib´erica, Spain) at temperatures between 48 and 52 ◦ C. Higher temperatures resulted in strong enzyme denaturation, thus causing a dramatic decrease of the enzymatic activity. The maximum hydrolysis degree (about 20%) was reached in 20 min. pH was observed to decrease during the process due to the hydrolysis of the proteins [15]. The optimum initial pH was within the range of 10.3–10.8, what led to an average pH of 9.5–10.0. This enzymatic detergent (P3-Ultrasil® 62) was selected for the membrane cleaning in this work. In this work the membrane cleaning ability of the detergent P3-Ultrasil® 62 was tested. The influence of operating
conditions on the cleaning efficiency was studied and the results were compared to those obtained in the batch reactor.
2. Materials and methods 2.1. Feed solutions Feed solutions were prepared from the powdered WPC Protarmor PS90 (Armor Prot´eines, Saint Brice en Cogles, France). Protarmor PS90 WPC was produced from acid whey by successive ultrafiltration (UF) and diafiltration (DF) operations. Its composition is shown in Table 1. The rest of the solids are mainly other proteins and peptides, and lactose. According to the manufacturer, the amount of proteins and peptides in this preparation is about 90% (w/w) on dry matter basis. The amount of low molecular weight compounds, such as salts, in this WPC was very low due to the intensive UF and DF steps. The preparation of the solutions from this concentrate was carried out as described in Arg¨uello et al. [14]. 2.2. Cleaning solutions As feed solutions are mainly composed of proteins, a commercial proteolytic detergent was considered as cleaning agent: P3-Ultrasil® 62 (Henkel Ib´erica S.A., Barcelona, Spain). This is a commercial formulation that is a mixture of proteases and anionic tensioactive agents. The molecular weight of the proteolytic enzyme contained in this formulation is approximately 25 kg/mol. In some experiments NaOH was added to the cleaning solutions to adjust the pH. 2.3. Experimental set-up Experiments were performed in a standard ultrafiltration device. The cleaning tests were carried out on a Carbosep® M1 membrane (Orelis, S.A., Saint Maurice de Beynost, France), which is being used for the fractionation of whey proteins. This membrane consisted of a 6 mm internal diameter tube with a ZrO2 filtering layer on a carbon support. Membranes of 25, 60 and 120 cm length were used. Carbosep® M1 membranes were reported to have a molecular weight cut-off of 150 kg/mol. Table 1 Composition of the powdered WPC Protarmor PS90 Compound
Concentration (g/kg)
-Lactoglobulin ␣-Lactalbumin BSA Immunoglobulin G Fat Salts
500 100 16 18 <20 <30
M.A. Arg¨uello et al. / Separation and Purification Technology 41 (2005) 147–154
2.4. Operating conditions The fouling–cleaning experiments consisted of several steps: membrane conditioning-disinfection, rinsing, water flux measurement, membrane fouling with WPC solutions, rinsing, water flux measurement, cleaning, rinsing and water flux measurement. The membrane conditioning was performed before the first use of the membrane. After this step the initial water flux, Jwi , was recorded to calculate the intrinsic hydraulic resistance, Rm , which served as a reference for the cleaning procedure efficiency. All the cleaning cycles were performed at 50 ◦ C as the enzymes showed the highest activity at this temperature [14]. The values of pressure and feed flow were selected so that the shear stress was not very high in order to avoid enzyme denaturation. However, flow rate should not be low as it could induce a large boundary layer. Higher flow rates increase the sweeping of substances deposited on the membrane surface. Some authors recommend to use the same pressure and the same or higher feed flow than those used to foul the membrane [16]. Tap water filtered through a series of 10, 5, 2, 1 and 0.2 m microfilters (final fouling index <3) was used to rinse the membranes and prepare the solutions. Operating conditions are summarized in Table 2. 2.5. Analytical methods The amount of whey proteins and the molecular weight of the peptidic fragments formed during the hydrolysis were determined by chromatography using a reverse phase (RP) PLRP-S 300 column of 150 × 7.5 mm (Polymer Laboratories Inc., USA), and a size exclusion (SE) TSK-GEL 2000 SW column of 300 mm × 7.5 mm (Tosoh Corporation, Japan), respectively, and a Hewlett Packard (USA) HP 1050 HPLC chromatograph, following the procedure described in refs. [14,17]. The chromatographic conditions and chemicals used are shown in Arg¨uello et al. [17]. Reverse phase analyses were carried out using a modification of the Resmini’s method [18]. In order to determine the activity of the enzyme used, an arbitrary unit (u) was considered. This activity unit was defined as the amount of product (in mL) necessary to achieve 10% hydrolysis degree in 5 min at a pH of 9.5, a
149
temperature of 50 ◦ C and for a concentration of WPC solution of 1 g/L. Therefore, the number of activity units per volume of enzyme preparation is a measure of the activity of the preparation. The activity was determined in a batch reactor as described in Arg¨uello et al. [14], and using the operating conditions indicated. The activity units experimentally measured for the commercial detergent selected in this work were of 1.34 u/mL. Protein hydrolysis degree (HD) was defined as the relationship between the number of peptidic linkages hydrolyzed per unit of substrate mass (n) and the total number of peptidic linkages per unit of substrate mass (ntot ): n HD = × 100 (1) ntot The hydrolysis degree was determined by means of the ophtaldialdehyde method [19] as described in Arg¨uello et al. [14], using a PU 8700 UV–vis spectrophotometer (Philips Scientific, Oxford, UK). Permeate flux was measured gravimetrically and kinematic viscosity (ν) of each cleaning product was determined using a Cannon–Fenske capillary viscosimeter (Afora, Barcelona, Spain). Dynamic viscosity (µ) was calculated from density (ρ) at 50 ◦ C, which was measured using an Abbe densimeter (Sibuya Optical, Tokyo, Japan). IR spectra were registered on Perkin-Elmer (Shelton, CN, USA) spectrometers, dispersive (783) and Fourier transform (1710) models. Different types of evaluations were carried out: (a) upper layers: 1–10 m from the upper part of the deposits were taken; and (b) lower layers: the material closest to the membrane support was considered. The quantification of proteins in both upper and lower layers was carried out as described in Daufin et al. [20]. XPS supplies information from very thin surface layers (no more than 10 nm thickness). Measurements were taken through a VSW HA 100 hemisphere analyzer fixed on an ultrahigh vacuum bell. The surface under study was exposed to unmonochromatized X-rays from a thin Mg/Al source (Mg K␣ line, 1250 eV; Al K␣ line, 1487 eV). The relative atomic composition was recorded and results expressed by means of the relationships C/Zr, N/Zr and O/Zr, which represent the ratio between the relative atomic composition of the corresponding atoms.
Table 2 Operating conditions for the fouling–cleaning cycles Feed stream
v (m/s)
P (bar)
t (min)
T (◦ C)
Operation
NaOCl, pH 11 Water WPC Protarmor PS90, 50 g/L, 0.2 M NaCl, VCR 1 Water P3-Ultrasil® 62 Water NaOCl, pH 11 Water
2 2 1 2 2 2 2 2
1 1 2 1 1 1 1 1
20 10 60 20 30–150 10 20 10
50 50 50 50 50 50 50 50
Conditioning Rinsing Fouling Rinsing Cleaning Rinsing Disinfection Rinsing
150
M.A. Arg¨uello et al. / Separation and Purification Technology 41 (2005) 147–154
Hydraulic and/or chemical methods can be considered to characterize the cleaning efficiency [20].
cleaning efficiency (ERW ) was evaluated in this study as the relationship between the hydraulic resistance removed during the cleaning (Rif − Rres ) and the hydraulic resistance that should have been removed to reach 100% efficiency (Rif ):
3.1. Hydraulic characterization
ERW =
Permeate flux data was used to evaluate the membrane hydraulic resistance (R), according to Darcy’s law:
3.2. Chemical characterization
3. Cleaning efficiency
R=
P , and µJ
(2)
Ruf = Rm + Rf = Rm + Rif + Rrf
(3)
where P is the transmembrane pressure, J is the permeate flux and Ruf , Rm and Rf , are, respectively, the total resistance of the fouled membrane, the intrinsic hydraulic resistance of the membrane and the resistance due to membrane fouling, which combines reversible (Rrf ) and irreversible (Rif ) phenomena. The error in this evaluation was estimated by summing the relative errors as described in Daufin et al. [20], and resulted to be lower than 6.7% in this work. Fig. 1 shows a distribution of the hydraulic resistances during the ultrafiltration, rinsing and cleaning steps. Several authors [20,21] proposed the comparison between the hydraulic resistance of the cleaned membrane, Rcw , and the intrinsic hydraulic resistance of the membrane to evaluate the cleaning efficiency. Thus the membrane can be considered clean when (Rcw − Rm )/Rm ≤ 0.067. The difference between Rcw and Rm was defined as the residual resistance of the membrane after cleaning, according to Eq. (4). Therefore, the membrane can be assumed to be clean when Rres /Rm ≤ 0.067. Rres = Rcw − Rm
(4)
The difficulties to clean the membrane depend on its initial degree of fouling. In order to take this effect into account, the
Rif − Rres × 100 Rif
(5)
From the chemical point of view the membrane is considered to be clean when there are no substances attached to its surface that do not belong to the membrane material itself. IR and XPS measurements provide information on the composition of the membrane surface. New and fouled-cleaned membranes were characterized and the results compared in order to check the presence of any residual matter after the cleaning cycle.
4. Results and discussion As already indicated in the introduction, in a previous work by the authors [14], the ability of the enzyme preparation P3-Ultrasil® 62 to hydrolyze whey proteins was tested in a batch reactor. The optimal conditions to carry out the hydrolysis were observed to be a temperature between 48 and 52 ◦ C, and an average pH within the range of 9.5–10.0. In this work the membrane cleaning ability of this detergent was tested. The operating conditions that were studied are the following. 4.1. Permeate recycling during cleaning Different approaches can be considered to carry out the cleaning process: (a) recycling of both retentate and permeate
Fig. 1. Distribution of hydraulic resistances during ultrafiltration, rinsing and cleaning steps.
M.A. Arg¨uello et al. / Separation and Purification Technology 41 (2005) 147–154
to the feed tank; (b) retentate recycling to the feed tank while permeate is removed from the system; and (c) removal of both permeate and retentate from the system. The last approach involves the consumption of very large amounts of cleaning solutions so that it is not very often employed and was not considered in this research. The transmission of enzymes through UF and MF membranes is usually high [22]. The molecular weight of the enzyme contained in the enzymatic detergent (25 kg/mol) is lower than the molecular weight of the membrane (150 kg/mol), so that a high transmission could be expected. If permeate is not recycled, the overall amount of enzyme available for the cleaning can decrease at the same time as the volume of the cleaning solution. Its concentration in the feed tank would increase unless its transmission through the membrane is 100%. As a result of the degradation of the deposits by the enzymes, the permeate can contain species that can foul the membrane, so that the recycling of the permeate could produce a reduction of the cleaning efficiency. To study the influence of these factors in the cleaning efficiency, a set of experiments were carried out with and without permeate recycling. The results obtained are shown in Fig. 2, where the evolution of the hydraulic resistance during the cleaning process (Rc ) is plotted. It can first be observed that the evolution of the resistance with time does not present a minimum, but it reaches a stationary value. Therefore, it can be considered that there is no important redeposition of protein fragments on the membrane, or, if there is, it does not significantly affect the hydraulic resistance. It can also be observed that the evolution of the resistance with time is practically the same in both cases, independently that permeate is or not recirculated. Moreover, after the cleaning the intrinsic hydraulic resistance of the membrane was recovered (∼100% cleaning efficiency) so that it can be concluded that there is no significant redeposition of the hydrolyzed species on the membrane (or its redeposition does not affect the hydraulic resistance).
151
In order to check the transmission of the enzyme through the membrane, several aliquots of the permeate were used to catalyze the hydrolysis of small amounts of WPC solutions in the batch reactor. Enzymatic activity was detected in the permeate stream, but it was lower than that determined in the concentrate and than that determined in a previous work [17] when Alcalse® (molecular weight of 27.5 kg/mol) was used to clean Carbosep® M6 membranes (molecular weight cut-off of 350 kg/mol). The lower enzyme transmission observed for Carbosep® M1 membranes can be explained as a result of their smaller pore size. The enzymatic activity per volume unit in the retentate at the beginning and at the end of the experiments with and without permeate recycling was also measured in the batch reactor. It was observed that the activity at the end was lower than that at the beginning of the process. The reduction of the enzymatic activity during the cleaning can be due to different reasons, such as the denaturation of the enzyme caused by the shear stress in the equipment; the adsorption of the enzyme on the equipment walls and/or on the membrane; the inhibition of the enzyme activity by compounds present in the system; and/or proteases digestion by another proteases. Nevertheless, to avoid this last drawback, manufacturers usually look for the broadest possible substrate specificity. It was also observed that the activity per unit volume in the retentate was slightly higher when the permeate was not recycled due to the increase of enzyme concentration. However, the difference between the activity in both retentates was very low, thus explaining the almost identical results obtained with both operating modes. Protein fragments were detected in the permeate stream due to protein hydrolysis. However, as the cleaning efficiency reached when permeate is recirculated is very high, it can be considered that these fragments do not significantly foul the membranes. Permeate recirculation is more advantageous from the economical point of view, because lower amounts of cleaning solutions are required. As almost the same cleaning efficiency was reached with both operating modes, permeate recycling was selected to carry out the rest of the runs. 4.2. pH of the cleaning solutions
Fig. 2. Evolution of the hydraulic resistance during the cleaning of Carbosep® M1 membranes with P3-Ultrasil® 62 (1.34 u/L, 9.5 initial pH).
In the experiments carried out in the batch reactor [14] it was established that the optimum pH for proteins hydrolysis using P3-Ultrasil® 62 was within the range of 10.3–10.8. However, pH decreased during the process, being the average value of 9.5–10.0. Similar variation was observed during membrane cleaning. However, the reduction of pH with time was smaller than that observed in the batch reactor, probably due to a lower amount of substrate. The optimum pH for membrane cleaning can be affected by several factors, and not only by the optimum pH for proteins hydrolysis. One factor that can be important is the ionic exchanger character of ZrO2 membranes, which can
152
M.A. Arg¨uello et al. / Separation and Purification Technology 41 (2005) 147–154
Fig. 4. Evolution of the cleaning efficiency with the cleaning parameter (cp). Fig. 3. Influence of the initial pH on the cleaning efficiency during the cleaning of Carbosep® M1 membranes with P3-Ultrasil® 62 (1.34 u/L).
exchange cations or anions with the solution depending on the value of the pH [23]. Other factor can be the contribution to the cleaning of the NaOH added to adjust the pH. The optimum pH value was considered to be the one that produces the highest cleaning efficiency at the same operating conditions (temperature, flow rate, transmembrane pressure, enzyme concentration, time and hydraulic resistance at the beginning of the cleaning process). To study the effect of the cleaning solution pH on the cleaning efficiency, the pH of the original cleaning solution (about 8) was modified by adding small amounts of 3N NaOH up to the desired value. Fig. 3 shows the influence of the cleaning solution initial pH on the efficiency. The highest efficiency was observed at initial pH values between 10.0 and 10.3. As previously indicated, pH decreased due to protein hydrolysis. The average pH during the enzymatic cleaning was within the range of 9.6–10.2. The optimum average pH is very similar to that obtained in the batch reactor (9.5–10.0), thus indicating that enzymatic hydrolysis of proteins is the main factor that produces membrane cleaning. To check the contribution of the NaOH added to the cleaning, several membranes were cleaned with diluted NaOH solutions at the optimum pH. The efficiency was observed to be lower than 15%, thus demonstrating that the high cleaning efficiency achieved is mainly due to the enzymatic hydrolysis of the deposited proteins. 4.3. Enzymatic agent concentration From the economic point of view the determination of the optimum amount of enzyme required to clean the membrane is very important. Lower amounts of enzyme can result in low cleaning efficiencies or high cleaning times, and higher amounts of enzyme can increase operation costs and even result in further membrane fouling. The amount of cleaning agent depends on a lot of factors, such as the degree of membrane fouling, operating conditions, volume of cleaning solution (V), and membrane area (Am ). To take into account some of these factors the cleaning parameter (cp) was defined in a previous work [17] as the relationship between the initial
enzymatic activity (a), the hydraulic resistance that should be removed during the cleaning (Rif ) and membrane area: cp =
a Rif Am
(6)
The optimum concentration of enzymatic agent was considered the one that produces the highest cleaning efficiency for a certain membrane area and membrane fouling (Rif ). Fig. 4 shows the cleaning efficiency as a function of the cleaning parameter. The optimum value for cp was of about 35 × 10−9 u/m, which was very similar to the optimum cp values reported in previous works [17,24], where different proteolytic enzymes had been used to clean inorganic membranes (optimum cp values within the range of 30–35 × 10−9 u/m). Moreover, 100% cleaning efficiency was reached at the optimum cp value. Higher amounts of enzyme resulted in a decrease of the cleaning efficiency. However, when the hydrolysis was carried out in the batch reactor, the hydrolysis degree was observed to increase with the ratio enzyme concentration/substrate concentration up to a steady value that depended on the selectivity of the enzyme. A maximum value of the hydrolysis degree was not observed. Therefore, the decrease of the cleaning efficiency with cp can probably be due to membrane fouling caused by the own cleaning agent and/or by the redeposition of solutes on the membrane surface. If the hydrolysed proteins caused membrane fouling, recycling of permeate would not be recommended, the same as long cleaning times. However, recycling of permeate and long cleaning times did not result in a decrease of the hydraulic cleaning efficiency. 4.4. Cleaning time Cleaning time was defined as the time required to reach a steady value of the hydraulic resistance and it was observed to be about 20 min, independently on the value of cp, as can be noted in Fig. 5. According to the results obtained in the batch reactor, the maximum hydrolysis degree was achieved in 20 min. Therefore, the cleaning time can be related to the time required to hydrolyze whey proteins. Some authors [9–11] reported much higher cleaning
M.A. Arg¨uello et al. / Separation and Purification Technology 41 (2005) 147–154
153
4.5. Chemical characterization of the membrane surface
Fig. 5. Evolution of the hydraulic resistance with time for different values of the cleaning parameter (cp): () 15.2 × 10−9 u/m; () 23.9 × 10−9 u/m; () 36. 9 × 10−9 u/m; ( ) 44.1 × 10−9 u/m; ( ) 50.1 × 10−9 u/m; (䊉) 65.5 × 10−9 u/m; ( ) 148.1 × 10−9 u/m.
times when organic membranes were cleaned with enzymatic detergents. However, the cleaning ability of several enzymes (Alcalase® 0.6 L, Novo Nordisk A/S, Denmark, and Maxatase® XL, Genencor International, The Netherlands) was tested in a previous work by the authors with different inorganic membranes [17,24]. High hydraulic cleaning efficiency was achieved in 20 min as well. Inorganic membranes fouled by whey proteins were also chemically cleaned using NaOH and NaOCl [25]. The cleaning time ranged within 20 and 30 min and the best results in terms of hydraulic cleaning efficiency were achieved with NaOCl (98% versus 90% for NaOH). In Fig. 6 the cleaning efficiency is plotted against cp for different cleaning times. The most favorable couple of values is the one that combines the lowest value of cp and time (cheapest and shortest cleaning cycle) to reach the desired efficiency, which for dairy industries should be as high a possible.
Fig. 6. Evolution of the cleaning efficiency with the cleaning parameter (cp) for different cleaning times: () 1 min; (䊉) 3 min; () 5 min; ( ) 10 min; () 15 min; ( ) 30 min.
The cleanliness of the membrane from the chemical point of view was assessed using IR and XPS spectroscopy. The IR spectra of a Carbosep® M1 membrane after a fouling–rinsing cycle showed amide I and II bands, which are typical of proteins. These bands present a maximum absorption at 1650 and 1540 cm−1 , respectively. Amino and carboxylic groups were also detected. Amino groups typically show an absorption band in the region between 3300 and 3500 cm−1 , whereas carboxylic groups present absorption bands in the wavenumber regions of 1700–1725 cm−1 (as COOH) and 1550–1630 (as COO− ). After membrane cleaning the increase in the relative amount of carboxylic and amoino groups due to protein hydrolysis was observed. Table 3 shows the average results obtained from the XPS and IR characterization of new, conditioned, fouled, rinsed and cleaned membranes. The carbon (C 1s) and nitrogen (N 1s) structures are typical of organic fouling, while the oxygen (O 1s) comes from the membrane itself as well as from the fouling. The disappearance of the zirconium (Zr 3p3/2 ) structure can be related to the fouling layer depth. It can be observed from the table that the ratio C/Zr and N/Zr increases 10 times after membrane conditioning and the protein content increases by 15% in the upper layers as well as in the lower layers. The reason for these observations could be a possible contamination of the conditioning solution by dust or something similar. Another reason could be the adsorption of proteins from previous fouling–cleaning experiments on the equipment walls. These proteins could desorb during the conditioning process and foul the membrane. Nevertheless, this “fouling” did not lead to an increase in the hydraulic resistance. The comparison between the results obtained from a fouled membrane and those obtained from a rinsed membrane show how important the rinsing step is. The hydraulic resistance decreased by about 78% after the rinsing. The amount of proteins decreased by about 20% after this step, while XPS data showed that the fouling layer thickness decreased as the intensity of the Zr 3p3/2 peak increased. The amount of matter deposited on the membrane decreased during the cleaning. However, after this cycle residual matter was detected on the membrane surface even though the hydraulic cleaning efficiency was 100%. Other authors [26] reported the same discrepancy between hydraulic and chemical characterization of the membrane cleanliness using non-enzymatic cleaning agents. It can also be observed that after the cleaning the amount of proteins on the upper membrane layers is higher than that on the lower layers. This residual deposit could favorably affect the subsequent ultrafiltration operation, as at the beginning of the process protein–protein interactions would be expected instead of protein–membrane interactions. Moreover it is possible that certain amount of enzyme could be part of the residual deposit. The adsorption of proteolytic enzymes on ZrO2 membranes was demonstrated in a previous work by the authors [24] using different types of enzyme formulations and inorganic membranes. The adsorption of
154
M.A. Arg¨uello et al. / Separation and Purification Technology 41 (2005) 147–154
Table 3 Average results obtained from the chemical characterization of Carbosep® M1 membranes Sample
R (1012 m−1 )
C/Zra
N/Zrb
O/Zrc
Proteinsd upper layers
Proteinsd lower layers
New Conditioned Fouled Fouled and rinsed Cleaned and rinsed
4.0 4.0: Rm 40.0: Ruf 12.0: Rif 4.0: Rcw
0.31 3.00 1066.0 58.0 21.3
0.08 0.81 206.0 10.4 5.2
2.80 4.8 419.0 35.8 10.5
0.02 0.32 12.90 10.5 6.71
0.02 0.32 11.00 8.77 3.3
a b c d
C/Zr: ratio between C 1s and Zr 3p3/2 relative compositions. N/Zr: ratio between N 1s and Zr 3p3/2 relative compositions. O/Zr: ratio between O 1s and Zr 3p3/2 relative compositions. wt.%.
the enzyme on the membrane was reported to enhance the whey ultrafiltration process, probably due to a self-cleaning mechanism, as described by several authors [27]. Similar favorable results are expected when P3-Ultrasil® 62 is used to clean inorganic Carbosep® M1 membranes.
de Spectroscopie du Solide” (Universit´e de Rennes I) and “Laboratoire de’Etudes de la Corrosion” (Univerist´e Pierre and Marie Curie, Paris) where the XPS and IR measurements were carried out.
References 5. Conclusions The commercial enzymatic detergent P3-Ultrasil®
62 was able to clean the inorganic membrane Carbosep® M1 fouled by whey proteins. Very high hydraulic cleaning efficiencies (about 100%) were reached in short operating time (20 min). The cleaning efficiency depended on the operating conditions. Similar cleaning efficiency was reached with both, permeate recycling and non-recycling operating modes. The cleaning solution pH was observed to decrease during the cleaning process due to protein hydrolysis. The optimum average pH for proteins hydrolysis (9.6–10.2) was the same as the optimum pH for proteins hydrolysis obtained in the batch reactor. The optimum value of the cleaning parameter (cp) defined in this work was observed to be 35 × 10−9 u/m. The highest hydraulic cleaning efficiency was reached in 20 min and can be related to the time required for proteins hydrolysis. However, chemical cleanliness was not achieved. The presence of residual matter on the membrane surface was observed even when the hydraulic cleaning efficiency was 100%.
[1] [2] [3] [4] [5] [6]
[7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19]
Acknowledgements [20]
This work was supported by the EU (European Union) through the AIR project CT93-1207 and the “Human Capital and Mobility” program, contract No. ERBCHRX-0393, and by the Spanish foundations CICYT and FICYT. The authors also wish to thank the company Henkel-Ib´erica, S.A. for the technical contribution, the French companies Armor Prot´eines and Orelis, S. A. for the WPC and membranes supplied, respectively, and the French laboratories “Laboratoire
[21] [22] [23] [24] [25] [26] [27]
J.H. Van Ee, Chimicaoggi 4 (1991) 30. O. R¨ohm, German Patent DRP 283923 (1913). G. Jensen, Tenside Surf. Det. 27 (1990) 30. M. Malmos, Chem. Ind. 3 (1990) 183. C.A. Starace, J. Am. Oil Chem. Soc. 58 (1981) 165A. R.W. Baker, E.L. Cussier, W. Eykamp, W.J. Koros, R.L. Riley, H. Strathmann (Eds.), Membrane Separation Systems: Recent Developments and Future Directions, Noyes Data Corporation, New Jersey, 1991 (Chapter 5). K.E. Smith, R.L. Bradley, J. Dairy Sci. 70 (1987) 243. G. Tr¨ag˚ardh, Desalination 71 (1989) 325. S. Bragulla, K. Lintner, Alimenta 25 (1986) 114. K.E. Smith, R.L. Bradley, J. Food Protection 51 (1988) 89. M.J. Mu˜noz-Aguado, D.E. Wiley, A.G. Fane, J. Membr. Sci. 117 (1996) 175. K. Mehira, W.J. Donnelly, J. Dairy Res. 60 (1993) 89. D. Marshall, P.A. Munro, G. Tr¨ag˚ardh, Desalination 91 (1993) 65. ´ M.A. Arg¨uello, D. Elzo, F.A. Riera, R. Alvarez, Can. J. Chem. Eng. 76 (1998) 276. A. Margot, E. Flaschem, A. Renken, Proc. Biochem. 29 (1994) 257. C.R. Thomas, A.W. Nienow, P. Dunnill, Biotechnol. Bioeng. 21 (1979) 2263. ´ ´ M.A. Arg¨uello, S. Alvarez, F.A. Riera, R. Alvarez, J. Agric. Food Chem. 50 (2002) 1951. P. Resmini, L. Pellegrino, R. Andreini, F. Prati, Technica Latteriocasearia 40 (1993) 7. F.C. Church, H.E. Swaisgood, D.H. Porter, G.L. Catignani, J. Dairy Sci. 66 (1988) 1219. G. Daufin, U. Merin, F.L. Kerh´erve, J.P. Labb´e, A. Quemerais, C. Bousser, J. Dairy Res. 59 (1992) 29. C. Taddei, P. Aimar, G. Daufin, V. Sanchez, Le Lait. 68 (1988) 157. W.R. Bowen, N.J. Hall, Biotech. Bioeng. 46 (1995) 28. S. Dumon, H. Barnier, J. Membr. Sci. 74 (1992) 289. ´ ´ M.A. Arg¨uello, S. Alvarez, F.A. Riera, R. Alvarez, J. Membr. Sci. 216 (2003) 121. ´ P. Matzinos, R. Alvarez, J. Membr. Sci. 208 (2002) 23. G. Daufin, U. Merin, J.P. Labb´e, A. Quemerais, F.L. Kerh´erve, Biotechnol. Bioeng. 38 (1991) 82. ¨ Velicangil, J. Appl. Polym. Sci. 27 (1982) 21. J.A. Howell, O.
Utilization of enzymatic detergents to clean inorganic membranes fouled by whey proteins b , F.A. Rieraa,∗ , R. Alvarez a ´ ´ M.A. Arg¨uelloa , S. Alvarez b
a Department of Chemical Engineering and Environmental Technology, University of Oviedo, C/Juli´ an Claver´ıa, 8, 33006 Oviedo, Spain Department of Chemical and Nuclear Engineering, ETSII, Politechnical University of Valencia, Camino de Vera, s/n, 46022 Valencia, Spain
Received in revised form 7 May 2004; accepted 10 May 2004
Abstract In this work inorganic membranes used for whey protein fractionation were cleaned with enzymatic detergents. The inorganic membrane Carbosep® M1 (Orelis S.A., France), of 150 kg/mol molecular weight cut-off and ZrO2 filtering layer, was used and the commercial detergent P3-Ultrasil® 62 (Henkel Ib´erica S.A., Spain) was selected for the cleaning. Hydraulic and chemical methods were considered to characterize the membrane cleanliness. Cleaning efficiency was observed to be a function of the operating conditions: recycling versus non-recycling of permeate, cleaning solution pH, enzymatic agent concentration and cleaning time. The optimum conditions to perform the cleaning were related to the optimum conditions to hydrolyze whey proteins in a batch reactor. Very high cleaning efficiencies (∼100%) were reached in short operating times (20 min). However, residual matter was detected on the membrane surface after the cleaning. © 2004 Elsevier B.V. All rights reserved. Keywords: Inorganic membranes; Enzymatic cleaning; Whey protein solutions; Hydraulic cleanliness; Chemical cleanliness
1. Introduction For a long time proteases have been considered as being very useful ingredients in modern detergents [1]. The idea of using enzymes in detergents was first described by R¨ohm in 1913 [2]. At that time, only pancreatic enzymes were available and they were not granulated or otherwise protected by physical means. Therefore the storage ability, long-term efficiency and other health aspects were unsatisfactory [3]. The development of enzymes of microbiological origin started in the 1950s with the production of BIO40. This formulation contained a bacterial protease, what was a significant improvement compared to the previous pancreatic products; however the detergent had the disadvantage of a neutral pH [4]. During the 1970s a new generation of high active alkaline detergent enzymes were developed and passed through different generations of physical forms (powders, pills and ∗
Corresponding author. Tel.: +34 8 5103436; fax: +34 8 5103434. E-mail address: [email protected] (F.A. Riera).
1383-5866/$ – see front matter © 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.seppur.2004.05.005
encapsulates) [5]. The modern technologies (recombinant DNA, genetic engineering and protein engineering) were first applied in 1988 to produce LipolaseTM [4]. Although proteases are still used as one of the main enzymes in detergent formulations, ␣-amylases, cellulases and lipases are now successfully applied in household laundry detergents, industrial laundering and machine dishwashing. In membrane technology, a very important research goal is the development of non aggressive cleaning methods able to increase the lifetime of membranes and the proposal of short cleaning times [6]. The use of enzymatic cleaners is of interest because they operate in mild conditions, which is a determining factor for their application to the cleaning of membranes sensitive to chemicals, pH and/or temperature [7,8]. Enzymatic agents can improve the cleaning efficiency and reduce the amount of chemicals needed and the energy costs by working at a lower temperature. Moreover, they are biodegradable and the preparation of tailor-made cleaners is possible. Enzymatic formulations were used by several authors to clean organic membranes [9–11]. However, they reported low cleaning efficiencies and extended cleaning times.
148
M.A. Arg¨uello et al. / Separation and Purification Technology 41 (2005) 147–154
Nomenclature a Am cp ERW HD J Jwi n ntot P R Rc Rcw Rf Rif Rm Rres Rrf Ruf V
enzymatic activity (u) membrane area (m2 ) cleaning parameter (u/m) cleaning efficiency hydrolysis degree permeate flux (L/h m2 ) initial water flux (L/h m2 ) number of peptidic linkages hydrolyzed per unit of substrate mass number of peptidic linkages per unit of substrate mass transmembrane pressure (MPa) membrane hydraulic resistance (m−1 ) hydraulic resistance during the cleaning process (m−1 ) hydraulic resistance of the cleaned membrane (m−1 ) resistance due to membrane fouling (m−1 ) resistance due to irreversible fouling (m−1 ) intrinsic hydraulic resistance (m−1 ) residual hydraulic resistance (m−1 ) resistance due to reversible fouling (m−1 ) total resistance of the fouled membrane (m−1 ) volume of cleaning solution (m3 )
Greek letters µ dynamic viscosity (Pa s) ν kinematic viscosity (m2 /s) ρ density (kg/m3 ) The results reported in this paper refer to the cleaning of inorganic membranes used for whey protein fractionation. The feed material is a whey protein concentrate (WPC) with low mineral and residual fat content. Therefore proteins are expected to be the main compounds contributing to membrane fouling [12,13]. In a previous paper by the authors [14], several enzymatic and non-enzymatic commercial detergents were used to hydrolyze whey proteins in a discontinuous reactor. It was concluded that only the enzymatic detergents were able to produce a significant hydrolysis of whey proteins. The best results were achieved with P3-Ultrasil® 62 (Henkel Ib´erica, Spain) at temperatures between 48 and 52 ◦ C. Higher temperatures resulted in strong enzyme denaturation, thus causing a dramatic decrease of the enzymatic activity. The maximum hydrolysis degree (about 20%) was reached in 20 min. pH was observed to decrease during the process due to the hydrolysis of the proteins [15]. The optimum initial pH was within the range of 10.3–10.8, what led to an average pH of 9.5–10.0. This enzymatic detergent (P3-Ultrasil® 62) was selected for the membrane cleaning in this work. In this work the membrane cleaning ability of the detergent P3-Ultrasil® 62 was tested. The influence of operating
conditions on the cleaning efficiency was studied and the results were compared to those obtained in the batch reactor.
2. Materials and methods 2.1. Feed solutions Feed solutions were prepared from the powdered WPC Protarmor PS90 (Armor Prot´eines, Saint Brice en Cogles, France). Protarmor PS90 WPC was produced from acid whey by successive ultrafiltration (UF) and diafiltration (DF) operations. Its composition is shown in Table 1. The rest of the solids are mainly other proteins and peptides, and lactose. According to the manufacturer, the amount of proteins and peptides in this preparation is about 90% (w/w) on dry matter basis. The amount of low molecular weight compounds, such as salts, in this WPC was very low due to the intensive UF and DF steps. The preparation of the solutions from this concentrate was carried out as described in Arg¨uello et al. [14]. 2.2. Cleaning solutions As feed solutions are mainly composed of proteins, a commercial proteolytic detergent was considered as cleaning agent: P3-Ultrasil® 62 (Henkel Ib´erica S.A., Barcelona, Spain). This is a commercial formulation that is a mixture of proteases and anionic tensioactive agents. The molecular weight of the proteolytic enzyme contained in this formulation is approximately 25 kg/mol. In some experiments NaOH was added to the cleaning solutions to adjust the pH. 2.3. Experimental set-up Experiments were performed in a standard ultrafiltration device. The cleaning tests were carried out on a Carbosep® M1 membrane (Orelis, S.A., Saint Maurice de Beynost, France), which is being used for the fractionation of whey proteins. This membrane consisted of a 6 mm internal diameter tube with a ZrO2 filtering layer on a carbon support. Membranes of 25, 60 and 120 cm length were used. Carbosep® M1 membranes were reported to have a molecular weight cut-off of 150 kg/mol. Table 1 Composition of the powdered WPC Protarmor PS90 Compound
Concentration (g/kg)
-Lactoglobulin ␣-Lactalbumin BSA Immunoglobulin G Fat Salts
500 100 16 18 <20 <30
M.A. Arg¨uello et al. / Separation and Purification Technology 41 (2005) 147–154
2.4. Operating conditions The fouling–cleaning experiments consisted of several steps: membrane conditioning-disinfection, rinsing, water flux measurement, membrane fouling with WPC solutions, rinsing, water flux measurement, cleaning, rinsing and water flux measurement. The membrane conditioning was performed before the first use of the membrane. After this step the initial water flux, Jwi , was recorded to calculate the intrinsic hydraulic resistance, Rm , which served as a reference for the cleaning procedure efficiency. All the cleaning cycles were performed at 50 ◦ C as the enzymes showed the highest activity at this temperature [14]. The values of pressure and feed flow were selected so that the shear stress was not very high in order to avoid enzyme denaturation. However, flow rate should not be low as it could induce a large boundary layer. Higher flow rates increase the sweeping of substances deposited on the membrane surface. Some authors recommend to use the same pressure and the same or higher feed flow than those used to foul the membrane [16]. Tap water filtered through a series of 10, 5, 2, 1 and 0.2 m microfilters (final fouling index <3) was used to rinse the membranes and prepare the solutions. Operating conditions are summarized in Table 2. 2.5. Analytical methods The amount of whey proteins and the molecular weight of the peptidic fragments formed during the hydrolysis were determined by chromatography using a reverse phase (RP) PLRP-S 300 column of 150 × 7.5 mm (Polymer Laboratories Inc., USA), and a size exclusion (SE) TSK-GEL 2000 SW column of 300 mm × 7.5 mm (Tosoh Corporation, Japan), respectively, and a Hewlett Packard (USA) HP 1050 HPLC chromatograph, following the procedure described in refs. [14,17]. The chromatographic conditions and chemicals used are shown in Arg¨uello et al. [17]. Reverse phase analyses were carried out using a modification of the Resmini’s method [18]. In order to determine the activity of the enzyme used, an arbitrary unit (u) was considered. This activity unit was defined as the amount of product (in mL) necessary to achieve 10% hydrolysis degree in 5 min at a pH of 9.5, a
149
temperature of 50 ◦ C and for a concentration of WPC solution of 1 g/L. Therefore, the number of activity units per volume of enzyme preparation is a measure of the activity of the preparation. The activity was determined in a batch reactor as described in Arg¨uello et al. [14], and using the operating conditions indicated. The activity units experimentally measured for the commercial detergent selected in this work were of 1.34 u/mL. Protein hydrolysis degree (HD) was defined as the relationship between the number of peptidic linkages hydrolyzed per unit of substrate mass (n) and the total number of peptidic linkages per unit of substrate mass (ntot ): n HD = × 100 (1) ntot The hydrolysis degree was determined by means of the ophtaldialdehyde method [19] as described in Arg¨uello et al. [14], using a PU 8700 UV–vis spectrophotometer (Philips Scientific, Oxford, UK). Permeate flux was measured gravimetrically and kinematic viscosity (ν) of each cleaning product was determined using a Cannon–Fenske capillary viscosimeter (Afora, Barcelona, Spain). Dynamic viscosity (µ) was calculated from density (ρ) at 50 ◦ C, which was measured using an Abbe densimeter (Sibuya Optical, Tokyo, Japan). IR spectra were registered on Perkin-Elmer (Shelton, CN, USA) spectrometers, dispersive (783) and Fourier transform (1710) models. Different types of evaluations were carried out: (a) upper layers: 1–10 m from the upper part of the deposits were taken; and (b) lower layers: the material closest to the membrane support was considered. The quantification of proteins in both upper and lower layers was carried out as described in Daufin et al. [20]. XPS supplies information from very thin surface layers (no more than 10 nm thickness). Measurements were taken through a VSW HA 100 hemisphere analyzer fixed on an ultrahigh vacuum bell. The surface under study was exposed to unmonochromatized X-rays from a thin Mg/Al source (Mg K␣ line, 1250 eV; Al K␣ line, 1487 eV). The relative atomic composition was recorded and results expressed by means of the relationships C/Zr, N/Zr and O/Zr, which represent the ratio between the relative atomic composition of the corresponding atoms.
Table 2 Operating conditions for the fouling–cleaning cycles Feed stream
v (m/s)
P (bar)
t (min)
T (◦ C)
Operation
NaOCl, pH 11 Water WPC Protarmor PS90, 50 g/L, 0.2 M NaCl, VCR 1 Water P3-Ultrasil® 62 Water NaOCl, pH 11 Water
2 2 1 2 2 2 2 2
1 1 2 1 1 1 1 1
20 10 60 20 30–150 10 20 10
50 50 50 50 50 50 50 50
Conditioning Rinsing Fouling Rinsing Cleaning Rinsing Disinfection Rinsing
150
M.A. Arg¨uello et al. / Separation and Purification Technology 41 (2005) 147–154
Hydraulic and/or chemical methods can be considered to characterize the cleaning efficiency [20].
cleaning efficiency (ERW ) was evaluated in this study as the relationship between the hydraulic resistance removed during the cleaning (Rif − Rres ) and the hydraulic resistance that should have been removed to reach 100% efficiency (Rif ):
3.1. Hydraulic characterization
ERW =
Permeate flux data was used to evaluate the membrane hydraulic resistance (R), according to Darcy’s law:
3.2. Chemical characterization
3. Cleaning efficiency
R=
P , and µJ
(2)
Ruf = Rm + Rf = Rm + Rif + Rrf
(3)
where P is the transmembrane pressure, J is the permeate flux and Ruf , Rm and Rf , are, respectively, the total resistance of the fouled membrane, the intrinsic hydraulic resistance of the membrane and the resistance due to membrane fouling, which combines reversible (Rrf ) and irreversible (Rif ) phenomena. The error in this evaluation was estimated by summing the relative errors as described in Daufin et al. [20], and resulted to be lower than 6.7% in this work. Fig. 1 shows a distribution of the hydraulic resistances during the ultrafiltration, rinsing and cleaning steps. Several authors [20,21] proposed the comparison between the hydraulic resistance of the cleaned membrane, Rcw , and the intrinsic hydraulic resistance of the membrane to evaluate the cleaning efficiency. Thus the membrane can be considered clean when (Rcw − Rm )/Rm ≤ 0.067. The difference between Rcw and Rm was defined as the residual resistance of the membrane after cleaning, according to Eq. (4). Therefore, the membrane can be assumed to be clean when Rres /Rm ≤ 0.067. Rres = Rcw − Rm
(4)
The difficulties to clean the membrane depend on its initial degree of fouling. In order to take this effect into account, the
Rif − Rres × 100 Rif
(5)
From the chemical point of view the membrane is considered to be clean when there are no substances attached to its surface that do not belong to the membrane material itself. IR and XPS measurements provide information on the composition of the membrane surface. New and fouled-cleaned membranes were characterized and the results compared in order to check the presence of any residual matter after the cleaning cycle.
4. Results and discussion As already indicated in the introduction, in a previous work by the authors [14], the ability of the enzyme preparation P3-Ultrasil® 62 to hydrolyze whey proteins was tested in a batch reactor. The optimal conditions to carry out the hydrolysis were observed to be a temperature between 48 and 52 ◦ C, and an average pH within the range of 9.5–10.0. In this work the membrane cleaning ability of this detergent was tested. The operating conditions that were studied are the following. 4.1. Permeate recycling during cleaning Different approaches can be considered to carry out the cleaning process: (a) recycling of both retentate and permeate
Fig. 1. Distribution of hydraulic resistances during ultrafiltration, rinsing and cleaning steps.
M.A. Arg¨uello et al. / Separation and Purification Technology 41 (2005) 147–154
to the feed tank; (b) retentate recycling to the feed tank while permeate is removed from the system; and (c) removal of both permeate and retentate from the system. The last approach involves the consumption of very large amounts of cleaning solutions so that it is not very often employed and was not considered in this research. The transmission of enzymes through UF and MF membranes is usually high [22]. The molecular weight of the enzyme contained in the enzymatic detergent (25 kg/mol) is lower than the molecular weight of the membrane (150 kg/mol), so that a high transmission could be expected. If permeate is not recycled, the overall amount of enzyme available for the cleaning can decrease at the same time as the volume of the cleaning solution. Its concentration in the feed tank would increase unless its transmission through the membrane is 100%. As a result of the degradation of the deposits by the enzymes, the permeate can contain species that can foul the membrane, so that the recycling of the permeate could produce a reduction of the cleaning efficiency. To study the influence of these factors in the cleaning efficiency, a set of experiments were carried out with and without permeate recycling. The results obtained are shown in Fig. 2, where the evolution of the hydraulic resistance during the cleaning process (Rc ) is plotted. It can first be observed that the evolution of the resistance with time does not present a minimum, but it reaches a stationary value. Therefore, it can be considered that there is no important redeposition of protein fragments on the membrane, or, if there is, it does not significantly affect the hydraulic resistance. It can also be observed that the evolution of the resistance with time is practically the same in both cases, independently that permeate is or not recirculated. Moreover, after the cleaning the intrinsic hydraulic resistance of the membrane was recovered (∼100% cleaning efficiency) so that it can be concluded that there is no significant redeposition of the hydrolyzed species on the membrane (or its redeposition does not affect the hydraulic resistance).
151
In order to check the transmission of the enzyme through the membrane, several aliquots of the permeate were used to catalyze the hydrolysis of small amounts of WPC solutions in the batch reactor. Enzymatic activity was detected in the permeate stream, but it was lower than that determined in the concentrate and than that determined in a previous work [17] when Alcalse® (molecular weight of 27.5 kg/mol) was used to clean Carbosep® M6 membranes (molecular weight cut-off of 350 kg/mol). The lower enzyme transmission observed for Carbosep® M1 membranes can be explained as a result of their smaller pore size. The enzymatic activity per volume unit in the retentate at the beginning and at the end of the experiments with and without permeate recycling was also measured in the batch reactor. It was observed that the activity at the end was lower than that at the beginning of the process. The reduction of the enzymatic activity during the cleaning can be due to different reasons, such as the denaturation of the enzyme caused by the shear stress in the equipment; the adsorption of the enzyme on the equipment walls and/or on the membrane; the inhibition of the enzyme activity by compounds present in the system; and/or proteases digestion by another proteases. Nevertheless, to avoid this last drawback, manufacturers usually look for the broadest possible substrate specificity. It was also observed that the activity per unit volume in the retentate was slightly higher when the permeate was not recycled due to the increase of enzyme concentration. However, the difference between the activity in both retentates was very low, thus explaining the almost identical results obtained with both operating modes. Protein fragments were detected in the permeate stream due to protein hydrolysis. However, as the cleaning efficiency reached when permeate is recirculated is very high, it can be considered that these fragments do not significantly foul the membranes. Permeate recirculation is more advantageous from the economical point of view, because lower amounts of cleaning solutions are required. As almost the same cleaning efficiency was reached with both operating modes, permeate recycling was selected to carry out the rest of the runs. 4.2. pH of the cleaning solutions
Fig. 2. Evolution of the hydraulic resistance during the cleaning of Carbosep® M1 membranes with P3-Ultrasil® 62 (1.34 u/L, 9.5 initial pH).
In the experiments carried out in the batch reactor [14] it was established that the optimum pH for proteins hydrolysis using P3-Ultrasil® 62 was within the range of 10.3–10.8. However, pH decreased during the process, being the average value of 9.5–10.0. Similar variation was observed during membrane cleaning. However, the reduction of pH with time was smaller than that observed in the batch reactor, probably due to a lower amount of substrate. The optimum pH for membrane cleaning can be affected by several factors, and not only by the optimum pH for proteins hydrolysis. One factor that can be important is the ionic exchanger character of ZrO2 membranes, which can
152
M.A. Arg¨uello et al. / Separation and Purification Technology 41 (2005) 147–154
Fig. 4. Evolution of the cleaning efficiency with the cleaning parameter (cp). Fig. 3. Influence of the initial pH on the cleaning efficiency during the cleaning of Carbosep® M1 membranes with P3-Ultrasil® 62 (1.34 u/L).
exchange cations or anions with the solution depending on the value of the pH [23]. Other factor can be the contribution to the cleaning of the NaOH added to adjust the pH. The optimum pH value was considered to be the one that produces the highest cleaning efficiency at the same operating conditions (temperature, flow rate, transmembrane pressure, enzyme concentration, time and hydraulic resistance at the beginning of the cleaning process). To study the effect of the cleaning solution pH on the cleaning efficiency, the pH of the original cleaning solution (about 8) was modified by adding small amounts of 3N NaOH up to the desired value. Fig. 3 shows the influence of the cleaning solution initial pH on the efficiency. The highest efficiency was observed at initial pH values between 10.0 and 10.3. As previously indicated, pH decreased due to protein hydrolysis. The average pH during the enzymatic cleaning was within the range of 9.6–10.2. The optimum average pH is very similar to that obtained in the batch reactor (9.5–10.0), thus indicating that enzymatic hydrolysis of proteins is the main factor that produces membrane cleaning. To check the contribution of the NaOH added to the cleaning, several membranes were cleaned with diluted NaOH solutions at the optimum pH. The efficiency was observed to be lower than 15%, thus demonstrating that the high cleaning efficiency achieved is mainly due to the enzymatic hydrolysis of the deposited proteins. 4.3. Enzymatic agent concentration From the economic point of view the determination of the optimum amount of enzyme required to clean the membrane is very important. Lower amounts of enzyme can result in low cleaning efficiencies or high cleaning times, and higher amounts of enzyme can increase operation costs and even result in further membrane fouling. The amount of cleaning agent depends on a lot of factors, such as the degree of membrane fouling, operating conditions, volume of cleaning solution (V), and membrane area (Am ). To take into account some of these factors the cleaning parameter (cp) was defined in a previous work [17] as the relationship between the initial
enzymatic activity (a), the hydraulic resistance that should be removed during the cleaning (Rif ) and membrane area: cp =
a Rif Am
(6)
The optimum concentration of enzymatic agent was considered the one that produces the highest cleaning efficiency for a certain membrane area and membrane fouling (Rif ). Fig. 4 shows the cleaning efficiency as a function of the cleaning parameter. The optimum value for cp was of about 35 × 10−9 u/m, which was very similar to the optimum cp values reported in previous works [17,24], where different proteolytic enzymes had been used to clean inorganic membranes (optimum cp values within the range of 30–35 × 10−9 u/m). Moreover, 100% cleaning efficiency was reached at the optimum cp value. Higher amounts of enzyme resulted in a decrease of the cleaning efficiency. However, when the hydrolysis was carried out in the batch reactor, the hydrolysis degree was observed to increase with the ratio enzyme concentration/substrate concentration up to a steady value that depended on the selectivity of the enzyme. A maximum value of the hydrolysis degree was not observed. Therefore, the decrease of the cleaning efficiency with cp can probably be due to membrane fouling caused by the own cleaning agent and/or by the redeposition of solutes on the membrane surface. If the hydrolysed proteins caused membrane fouling, recycling of permeate would not be recommended, the same as long cleaning times. However, recycling of permeate and long cleaning times did not result in a decrease of the hydraulic cleaning efficiency. 4.4. Cleaning time Cleaning time was defined as the time required to reach a steady value of the hydraulic resistance and it was observed to be about 20 min, independently on the value of cp, as can be noted in Fig. 5. According to the results obtained in the batch reactor, the maximum hydrolysis degree was achieved in 20 min. Therefore, the cleaning time can be related to the time required to hydrolyze whey proteins. Some authors [9–11] reported much higher cleaning
M.A. Arg¨uello et al. / Separation and Purification Technology 41 (2005) 147–154
153
4.5. Chemical characterization of the membrane surface
Fig. 5. Evolution of the hydraulic resistance with time for different values of the cleaning parameter (cp): () 15.2 × 10−9 u/m; () 23.9 × 10−9 u/m; () 36. 9 × 10−9 u/m; ( ) 44.1 × 10−9 u/m; ( ) 50.1 × 10−9 u/m; (䊉) 65.5 × 10−9 u/m; ( ) 148.1 × 10−9 u/m.
times when organic membranes were cleaned with enzymatic detergents. However, the cleaning ability of several enzymes (Alcalase® 0.6 L, Novo Nordisk A/S, Denmark, and Maxatase® XL, Genencor International, The Netherlands) was tested in a previous work by the authors with different inorganic membranes [17,24]. High hydraulic cleaning efficiency was achieved in 20 min as well. Inorganic membranes fouled by whey proteins were also chemically cleaned using NaOH and NaOCl [25]. The cleaning time ranged within 20 and 30 min and the best results in terms of hydraulic cleaning efficiency were achieved with NaOCl (98% versus 90% for NaOH). In Fig. 6 the cleaning efficiency is plotted against cp for different cleaning times. The most favorable couple of values is the one that combines the lowest value of cp and time (cheapest and shortest cleaning cycle) to reach the desired efficiency, which for dairy industries should be as high a possible.
Fig. 6. Evolution of the cleaning efficiency with the cleaning parameter (cp) for different cleaning times: () 1 min; (䊉) 3 min; () 5 min; ( ) 10 min; () 15 min; ( ) 30 min.
The cleanliness of the membrane from the chemical point of view was assessed using IR and XPS spectroscopy. The IR spectra of a Carbosep® M1 membrane after a fouling–rinsing cycle showed amide I and II bands, which are typical of proteins. These bands present a maximum absorption at 1650 and 1540 cm−1 , respectively. Amino and carboxylic groups were also detected. Amino groups typically show an absorption band in the region between 3300 and 3500 cm−1 , whereas carboxylic groups present absorption bands in the wavenumber regions of 1700–1725 cm−1 (as COOH) and 1550–1630 (as COO− ). After membrane cleaning the increase in the relative amount of carboxylic and amoino groups due to protein hydrolysis was observed. Table 3 shows the average results obtained from the XPS and IR characterization of new, conditioned, fouled, rinsed and cleaned membranes. The carbon (C 1s) and nitrogen (N 1s) structures are typical of organic fouling, while the oxygen (O 1s) comes from the membrane itself as well as from the fouling. The disappearance of the zirconium (Zr 3p3/2 ) structure can be related to the fouling layer depth. It can be observed from the table that the ratio C/Zr and N/Zr increases 10 times after membrane conditioning and the protein content increases by 15% in the upper layers as well as in the lower layers. The reason for these observations could be a possible contamination of the conditioning solution by dust or something similar. Another reason could be the adsorption of proteins from previous fouling–cleaning experiments on the equipment walls. These proteins could desorb during the conditioning process and foul the membrane. Nevertheless, this “fouling” did not lead to an increase in the hydraulic resistance. The comparison between the results obtained from a fouled membrane and those obtained from a rinsed membrane show how important the rinsing step is. The hydraulic resistance decreased by about 78% after the rinsing. The amount of proteins decreased by about 20% after this step, while XPS data showed that the fouling layer thickness decreased as the intensity of the Zr 3p3/2 peak increased. The amount of matter deposited on the membrane decreased during the cleaning. However, after this cycle residual matter was detected on the membrane surface even though the hydraulic cleaning efficiency was 100%. Other authors [26] reported the same discrepancy between hydraulic and chemical characterization of the membrane cleanliness using non-enzymatic cleaning agents. It can also be observed that after the cleaning the amount of proteins on the upper membrane layers is higher than that on the lower layers. This residual deposit could favorably affect the subsequent ultrafiltration operation, as at the beginning of the process protein–protein interactions would be expected instead of protein–membrane interactions. Moreover it is possible that certain amount of enzyme could be part of the residual deposit. The adsorption of proteolytic enzymes on ZrO2 membranes was demonstrated in a previous work by the authors [24] using different types of enzyme formulations and inorganic membranes. The adsorption of
154
M.A. Arg¨uello et al. / Separation and Purification Technology 41 (2005) 147–154
Table 3 Average results obtained from the chemical characterization of Carbosep® M1 membranes Sample
R (1012 m−1 )
C/Zra
N/Zrb
O/Zrc
Proteinsd upper layers
Proteinsd lower layers
New Conditioned Fouled Fouled and rinsed Cleaned and rinsed
4.0 4.0: Rm 40.0: Ruf 12.0: Rif 4.0: Rcw
0.31 3.00 1066.0 58.0 21.3
0.08 0.81 206.0 10.4 5.2
2.80 4.8 419.0 35.8 10.5
0.02 0.32 12.90 10.5 6.71
0.02 0.32 11.00 8.77 3.3
a b c d
C/Zr: ratio between C 1s and Zr 3p3/2 relative compositions. N/Zr: ratio between N 1s and Zr 3p3/2 relative compositions. O/Zr: ratio between O 1s and Zr 3p3/2 relative compositions. wt.%.
the enzyme on the membrane was reported to enhance the whey ultrafiltration process, probably due to a self-cleaning mechanism, as described by several authors [27]. Similar favorable results are expected when P3-Ultrasil® 62 is used to clean inorganic Carbosep® M1 membranes.
de Spectroscopie du Solide” (Universit´e de Rennes I) and “Laboratoire de’Etudes de la Corrosion” (Univerist´e Pierre and Marie Curie, Paris) where the XPS and IR measurements were carried out.
References 5. Conclusions The commercial enzymatic detergent P3-Ultrasil®
62 was able to clean the inorganic membrane Carbosep® M1 fouled by whey proteins. Very high hydraulic cleaning efficiencies (about 100%) were reached in short operating time (20 min). The cleaning efficiency depended on the operating conditions. Similar cleaning efficiency was reached with both, permeate recycling and non-recycling operating modes. The cleaning solution pH was observed to decrease during the cleaning process due to protein hydrolysis. The optimum average pH for proteins hydrolysis (9.6–10.2) was the same as the optimum pH for proteins hydrolysis obtained in the batch reactor. The optimum value of the cleaning parameter (cp) defined in this work was observed to be 35 × 10−9 u/m. The highest hydraulic cleaning efficiency was reached in 20 min and can be related to the time required for proteins hydrolysis. However, chemical cleanliness was not achieved. The presence of residual matter on the membrane surface was observed even when the hydraulic cleaning efficiency was 100%.
[1] [2] [3] [4] [5] [6]
[7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19]
Acknowledgements [20]
This work was supported by the EU (European Union) through the AIR project CT93-1207 and the “Human Capital and Mobility” program, contract No. ERBCHRX-0393, and by the Spanish foundations CICYT and FICYT. The authors also wish to thank the company Henkel-Ib´erica, S.A. for the technical contribution, the French companies Armor Prot´eines and Orelis, S. A. for the WPC and membranes supplied, respectively, and the French laboratories “Laboratoire
[21] [22] [23] [24] [25] [26] [27]
J.H. Van Ee, Chimicaoggi 4 (1991) 30. O. R¨ohm, German Patent DRP 283923 (1913). G. Jensen, Tenside Surf. Det. 27 (1990) 30. M. Malmos, Chem. Ind. 3 (1990) 183. C.A. Starace, J. Am. Oil Chem. Soc. 58 (1981) 165A. R.W. Baker, E.L. Cussier, W. Eykamp, W.J. Koros, R.L. Riley, H. Strathmann (Eds.), Membrane Separation Systems: Recent Developments and Future Directions, Noyes Data Corporation, New Jersey, 1991 (Chapter 5). K.E. Smith, R.L. Bradley, J. Dairy Sci. 70 (1987) 243. G. Tr¨ag˚ardh, Desalination 71 (1989) 325. S. Bragulla, K. Lintner, Alimenta 25 (1986) 114. K.E. Smith, R.L. Bradley, J. Food Protection 51 (1988) 89. M.J. Mu˜noz-Aguado, D.E. Wiley, A.G. Fane, J. Membr. Sci. 117 (1996) 175. K. Mehira, W.J. Donnelly, J. Dairy Res. 60 (1993) 89. D. Marshall, P.A. Munro, G. Tr¨ag˚ardh, Desalination 91 (1993) 65. ´ M.A. Arg¨uello, D. Elzo, F.A. Riera, R. Alvarez, Can. J. Chem. Eng. 76 (1998) 276. A. Margot, E. Flaschem, A. Renken, Proc. Biochem. 29 (1994) 257. C.R. Thomas, A.W. Nienow, P. Dunnill, Biotechnol. Bioeng. 21 (1979) 2263. ´ ´ M.A. Arg¨uello, S. Alvarez, F.A. Riera, R. Alvarez, J. Agric. Food Chem. 50 (2002) 1951. P. Resmini, L. Pellegrino, R. Andreini, F. Prati, Technica Latteriocasearia 40 (1993) 7. F.C. Church, H.E. Swaisgood, D.H. Porter, G.L. Catignani, J. Dairy Sci. 66 (1988) 1219. G. Daufin, U. Merin, F.L. Kerh´erve, J.P. Labb´e, A. Quemerais, C. Bousser, J. Dairy Res. 59 (1992) 29. C. Taddei, P. Aimar, G. Daufin, V. Sanchez, Le Lait. 68 (1988) 157. W.R. Bowen, N.J. Hall, Biotech. Bioeng. 46 (1995) 28. S. Dumon, H. Barnier, J. Membr. Sci. 74 (1992) 289. ´ ´ M.A. Arg¨uello, S. Alvarez, F.A. Riera, R. Alvarez, J. Membr. Sci. 216 (2003) 121. ´ P. Matzinos, R. Alvarez, J. Membr. Sci. 208 (2002) 23. G. Daufin, U. Merin, J.P. Labb´e, A. Quemerais, F.L. Kerh´erve, Biotechnol. Bioeng. 38 (1991) 82. ¨ Velicangil, J. Appl. Polym. Sci. 27 (1982) 21. J.A. Howell, O.