- Email: [email protected]
Cleaning of skim milk PES ultrafiltration membrane: On the real effect of nitric acid step
Journal of Membrane Science 428 (2013) 275–280
Contents lists available at SciVerse ScienceDirect
Journal of Membrane Science journal homepage: www.elsevier.com/locate/memsci
Cleaning of skim milk PES ultrafiltration membrane: On the real effect of nitric acid step Lydie Paugam a,b,n, David Delaunay a,1, Nde´ye Wemsy Diagne a,b, Murielle Rabiller-Baudry a,b a b
Universite´ Rennes 1, UMR ‘‘Institut des Sciences Chimiques de Rennes’’—CNRS, CS 74205, case 1011, 35042 Rennes Cedex, France Universite´ Europe´enne de Bretagne, France
a r t i c l e i n f o
abstract
Article history: Received 13 July 2012 Received in revised form 10 October 2012 Accepted 11 October 2012 Available online 23 October 2012
Cleaning of PES membranes after ultrafiltration of skim milk is still often performed at industrial scale with a formulated alkaline first step followed by a nitric acid step. This cleaning mode gives sufficiently satisfactory results in terms of water flux recovery but cannot prevent a progressive decline of production flux over time. The analysis of the deposit on the membrane by FTIR-ATR highlights the nature of the irreversible fouling, exclusively made of proteins, and the real cleaning efficiency. Sodium hydroxides alone allows up to 24% of protein removal which is in good agreement with the water flux recovery after this treatment. Nevertheless, the amount of protein remains exactly the same after nitric acid treatment whereas the flux significantly grows. This flux increase without any protein removal is strongly dependent on the proteins amount on the membrane at the start of nitric acid step. This phenomenon is not observed with HCl, but it seems not totally specific from HNO3 as the use of H3PO4 in particular leads also to similar results. The specific adsorption of oxoanions on proteins at acid pH that will change the hydrophobicity of the deposit is suspected. The nitric acid step can be suppressed as it is useless and misleads on the real efficiency of the overall cleaning. & 2012 Elsevier B.V. All rights reserved.
Keywords: Ultrafiltration Polyethersulfone Skim milk Cleaning Nitric acid
1. Introduction Ultrafiltration is the process used to standardise the skim milk protein content in dairy industry for the consumption or before cheese making. The high fouling resulting from this application limits the productivity and needs twice daily cleaning operations (2–3 h for 6–8 h of production) to recover an acceptable flux. At industrial scale, cleaning is generally carried out with various empirical sequences performed with more or less complex cleaning solutions including an alkaline (typically sodium hydroxide NaOH with surfactant at pH 11.5) to remove organic matter and a nitric acid step (pH 1.6) to remove the mineral fouling (calcium deposit). An alternative to this chemical, also today more widely used industrially is the enzymatic cleaning [1] that is followed by a deactivation step with nitric acid as recommended by the supplier. This chemical or enzymatic cleaning is followed
n Corresponding author at: Universite´ Rennes 1, UMR ‘‘Institut des Sciences Chimiques de Rennes’’—CNRS, CS 74205, case 1011, 35042 Rennes Cedex, France. Tel.: þ33 223235172; fax: þ33 223235765. E-mail address: [email protected] (L. Paugam). 1 Present adress: Equipe de recherche´ en physico-chimie et biotechnologie, Universite´ de Caen Basse Normandie, Bvd mare´chal Juin 14032 Caen Cedex, France.
0376-7388/$ - see front matter & 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.memsci.2012.10.013
by disinfection with NaOH and sodium hypochorite (150– 200 ppm). This cleaning mode is justified by quite acceptable results in terms of flux recovery just after the treatment. Nevertheless, it can neither prevent a decrease of the production flux over several months scale nor sometimes an alteration of the membrane selectivity as it is often observed at industrial [2,3]. The cleaning step is clearly identified as a bottleneck of membrane separations due to a lack of knowledge about its fundamental mechanisms that makes its optimisation difficult. Some works focused on the cleaning of polyethersulfone ultrafiltration membranes used to concentrate whey proteins [4], of polysulfone membrane used to ultrafiltrate pasteurised milk [5,6], reconstituted skim milk [2] whey proteins from whey protein concentrate [7–10], cheese whey [11] and on the cleaning of inorganic membranes [12,13]. But the efficiency of the cleaning sequence is strongly dependent of the couple membrane/ fluid, particularly for complex dairy fluids. For the ultrafiltration of skim milk, the PES membrane HFK131 (5–10 kg mol 1) is a worldwide used membrane that represents 70% of the market. A previous study focused on the nature of the fouling on this PES membrane for this application. After skim milk ultrafiltration and rinsing, the analysis of the fouling revealed by SEM–EDX that there are here no minerals. Proteins are then the exclusive target of the cleaning. The nitric acid is
276
L. Paugam et al. / Journal of Membrane Science 428 (2013) 275–280
used after the alkaline cleaning to remove the mineral part (calcium phosphate) of the fouling by industrials that justify its use by the observed flux increase after the treatment with this product. The question that arises is if there any oxidative action on the residual proteins or any change in the structure of the deposit due to the pH change [4,15]. In this work, a cascade of NaOH followed by HNO3 is first studied to better understand the action of nitric acid during the cleaning in this given application. The first step with NaOH is certainly not the more efficient to remove the fouling, nevertheless it will highlights the real effect of nitric acid avoiding the impact of formulated alkaline detergent on the membrane hydrophobicity. Indeed surfactants contained in alkaline cleaning products are known to adsorb on the membrane and to increase their flux [16]. The effect of NaOH on the irreversible protein fouling is then first presented and is followed by the study of the second step of the cleaning sequence i.e. the nitric acid. Other solutions are studied to highlights the action of this reagent: sodium nitrate, hydrochloric acid, phosphoric acid, citric acid and sulphuric acid. The efficiency of both cleaning agents is evaluated in terms of water flux recovery (as in industry) and real protein removal (quantified by FTIR-ATR) in plate and frame and spiral configuration. The effect of nitric acid is finally presented in a biological cleaning where it used to inactivate the enzyme at the end of the cleaning process.
2. Experimental 2.1. Fluids The UHT skim milk (Lait de Montagne, Carrefour, France) chosen contain an average of 31.5 g L 1 of proteins and 48 g L 1 of carbohydrates as well as minerals, close to 8 g L 1. In this study several single solutions were used: sodium nitrate at 8.5 g L 1 and pH 6.6, nitric acid at pH 1.6 (65% analytical grade, Acros), hydrochloric acid at pH 1.6 (37% analytical grade, Acros), phosphoric acid at pH 1.6 (85%, Fisher), citric acid at pH 1.8 (Fisher), sulphuric acid at pH 1.6 (95–97%, Fluka), sodium hydroxide at pH 11.5 (pellets 97% Rectapur, Prolabo) and a chlorinated alkaline one with 200 mg L 1 of active Cl2 prepared from a stock NaOCl solution (bleach 371 Cl2, Lacroix) adjusted with NaOH at pH 11.5. The enzymatic cleaning was lead with P3-Ultrasil 53 (Ecolab). It was prepared at 1 wt% leading to a natural neutral pH. As the latter parameter was not regulated during the cleaning, the operating conditions were probably not perfectly the optimal. As recommended by providers, a nitric acid rinsing was done after the treatment to inactivate the enzyme activity. Nitric acid at pH 2.5 is usually suggested but HCl was also tested at the same pH. Water used for preparation of cleaning solutions was filtered (1 mm) and deionised. 2.2. Membrane The spiral UF membrane used was made of polyethersulfone (PES, HFK-131, Koch, 6.5 m2) with a MWCO of 5–10 kg mol 1. The spiral membrane was stored in sodium metabisulphite solution (5 g L 1, Acros, analytical grade) after each experiment to avoid any microorganism growth. Flat membranes (0.0127 m2) were also used as model for spiral membrane in order to quantify protein fouling by FTIR-ATR. Before use, flat membranes were first rinsed 10 min in methanol to remove the conservative (glycerol) and were then rinsed with deionised water. A new pristine membrane was used for every
new UF and was compressed by fluxing water for 1 h at a transmembrane pressure of 4 bar before the water reference flux was measured. 2.3. Ultrafiltration process 2.3.1. Spiral-wound pilot A pilot of 100 L capacity (TIA, Bolle ne, France) equipped with the spiral membrane HFK-131 [standard size 25–40, 6.5 m2, spacer of F type (2 mm)] was used. The fouling of the membrane was performed at 2 bar with a recirculation rate fixed at 10.5 m3 h 1 corresponding to a cross-flow velocity of about 0.3 m s 1 (as at industrial scale). 24 L of skim milk were ultrafiltered during 150 min at transmembrane pressure TMP ¼2 bar and 50 1C with a volume reduction ratio VRR¼1 (batch mode). Water rinsing was preceded by a total draining of the pilot and achieved at the minimal pressure allowed by this pilot (0.2 bar) at 150 L h 1 (i.e. a cross-flow velocity about 0.008 m s 1). In the first step, the module was rinsed with 70 L of filtered tap water (active carbon and 1 m filter) that was discarded towards waste. Then, in the second step 50 L of deionised water were recirculated in the loop during 20 min with the permeate discarded. Cleaning of the membrane by 25 L (3.8 L m 2), if not specified, of cleaning solution was performed 60 min at 50 1C in the same hydrodynamic conditions as the fouling. A second rinsing, as described before, was realised before a water flux measurement. 2.3.2. Plate-and-frame module The tangential plate and frame UF module (Ray-Flow X100, Orelis) presented an effective membrane surface of 0.0127 m2. Flat membranes as spacers (F type, 2 mm) were recovered from a commercial spiral module. The fouling was achieved by UF at VRR¼1 of 4 L of skim milk during 150 min at 50 1C and 2 bar. The following water rinsing was performed in 2 steps after draining the pilot, in order to minimise the water consumption. First, during 2 min rinsing was performed without recirculation of permeate and retentate and then, during 30 min, only retentate was recycled. The following cleaning was made with 4 L of solution. The pilot and the membrane were then rinsed with water until a neutral pH was reached both in the retentate and the permeate. A second rinsing was proceed before the measure of the aftercleaning water flux. The 3 steps of a set of experiments were always performed in the same hydrodynamics conditions: cross-flow velocity v¼0.3 m s 1 at 2 bar and 50 1C for fouling, water rinsing and cleaning. 2.4. Evaluation of cleaning efficiency At least two replicates of each cleaning experiments were done in spiral and in plate and frame configuration. Each time, water flux recovery and residual protein amount were evaluated. 2.4.1. Water flux recovery The water flux recovery was the ratio of the water flux after the chemical treatment and a rinsing (J) on the reference water flux (J0) that corresponded to the pristine membrane well cleaned and rinsed. The accuracy on flux measurement was better than 3% (coefficient of variation). Water flux recovery ¼ J=J 0 The hydraulic cleanliness is reached when the water flux recovery is up to 90%. The relative error on this parameter was 5%.
L. Paugam et al. / Journal of Membrane Science 428 (2013) 275–280
2.4.2. Physico-chemical characterisation: amount of residual proteins by FTIR-ATR The quantification of the amount of proteins on the flat membrane was obtained by FTIR-ATR analysis. The FTIR-ATR spectra were registered with a spectrometer Perkin–Elmer (Paragon 1000, spectrum for windows software) equipped with a ZnSe crystal with an incidence angle of 451 and 12 reflections (20 scans, 2 cm 1 resolution). Quantification in the range 0.5–350 mg cm 2 (geometric area, precision close to 1 mg cm 2 ) based on the H1539/H1240 ratio of the height of the amide II band of proteins (H1539 at 1539 cm 1) to the height of a PES band (H1240 at 1240 cm 1) was already described [16,17]. To take into account a possible variation of the deposit amount on the flat sheet membrane (due to small variations of velocity profiles), an average value was calculated from nine pieces, cut in the membrane.
3. Results and discussion The global fouling of the membrane results from a reversible and an irreversible fouling. The first one is removed by water rinsing when the second remains and need a chemical cleaning. At industrial scale the skim milk is pasteurised before UF. A previous study was lead to analyse the irreversible fouling after pasteurised skim milk UF at 50 1C [18]. It was shown that the irreversible fouling is exclusively due to proteins. There is no mineral fouling as firstly expected (industrial cleaning sequences are thought on this base). This was also shown by Rabiller-Baudry et al. [14] in the case of UHT skim milk UF. As it is more difficult for technical and practical reasons to use pasteurised skim milk at lab scale and on the base of these results, UHT skim milk was chosen here to model industrial skim milk. After the ultrafiltration of (UHT) skim milk, the membrane is first rinsed with deionised water leading to a first water flux recovery of 40%. 3.1. Alkaline step The flux recovery after a treatment with sodium hydroxide at pH 11.5 is shown on Table 1. The variation of various parameters is presented: time treatment, volume of NaOH to membrane surface ratio and configuration. For the NaOH on spiral wound configuration with 3.8 L m 2 (that corresponds to a volume of 25 L in our installation), it gives J/J0 ¼0.60. The cleaning with this solution is then not sufficient to reach the hydraulic cleanliness. Neither the increase of NaOH volume to the membrane surface ratio nor the treatment duration, the use of a fresh solution of sodium hydroxide and the change of configuration allow to increase the flux recovery. The increase of solution volume to membrane surface ratio in spiral configuration gives here the best results i.e. J/J0 ¼ 0.74. Table 1 Water flux recovery after treatment with sodium hydroxide solutions depending on treatment duration, ratio of solution volume to membrane surface (spiral and plate and frame configuration, pH 11.5, 50 1C and 60 min). J/J0 Water rinsing NaOH (spiral, 3.8 L m 2) 60 min NaOH (spiral, 3.8 L m 2) 90 min NaOH (spiral, 3.8 L m 2) 2 60 min NaOH (spiral, 7.7 L m 2) 60 min NaOH (plate and frame, 315 L m 2) 60 min
0.42 0.60 0.67 0.70 0.74 0.70
277
Table 2 FTIR-ATR quantification of residual proteins on the PES UF membrane after the cleaning (residual proteins before cleaning¼ 32 71 g cm 2) in plate and frame module.
HNO3 pH 1.6 NaOH pH 11.5
Residual proteins on the membrane after cleaning (mg cm 2)
Proteins removed with the cleaning step (%)
33 23
0 28
The quantity of residual proteins on the membrane (in plate and frame configuration) is analysed by FTIR-ATR. The results are presented in Table 2. It shows that the treatment with sodium hydroxide at pH 11.5 leads to the removal of 24% of the protein from the irreversible fouling. This protein removal is in good agreement with the increase of flux recovery. Previous results [19] have shown that sodium hydroxide alone does not hydrolyse proteins at pH 11.5 and 50 1C in 1 h. The action of this product on the protein deposit will rather be explained by the increase of protein charge imposed by the pH change [14,15,20]. The deposit becomes swollen and less rigid. This would favour the partial removal of the protein [7]. 3.2. Nitric acid step Nitric acid usually comes as a second step in the cleaning of skim milk UF membranes. It is used after the alkaline step to remove the eventual mineral part of the fouling. However, the irreversible fouling, that remains after the rinsing, was previously identified as exclusively made of proteins. In this context, no real cleaning action of this chemical reagent can really be expected. 3.2.1. Effect of nitric acid on the protein deposit The effect of nitric acid on the protein deposit is first studied when it is used alone and then in sequence with NaOH. Table 2 shows the results of the analysis of residual amount of proteins on the membrane after UF of skim milk, water rinsing and nitric acid treatment. The quantity of proteins of the irreversible fouling is 32 mg cm-2 (71 mg cm 2). The protein amount remains the same after nitric acid treatment. So there is no protein removal with this treatment, as expected from the knowledge of fundamental mechanisms of organic soiling cleaning. 3.2.2. Effect of nitric acid on the flux Fig. 1 shows the water flux recovery J/J0 of the membrane HFK131 fouled with skim milk and then just rinsed or rinsed and treated with a single step of HNO3 or NaOH or with two steps HNO3/NaOH or the classical sequence NaOH/HNO3. It appears first that the water flux recovery after nitric acid is very close (J/J0 ¼0.57) from that obtained with sodium hydroxide (J/J0 ¼0.60). The flux recovery is much higher than after a single rinsing of the membrane (J/J0 ¼0.40). The sequence that presents the better results in terms of flux recovery is the [NaOHþHNO3] (J/J0 ¼0.74) which is in good agreement with information from industrial source. The inverse one [HNO3 þ NaOH] gives present a lower performance (J/J0 ¼0.60) i.e. a result similar from NaOH or HNO3 alone. It is to note that V¨ais¨anen et al. [4] present quite different results as for a PES ultrafiltration membrane (PES 50H, Hoechst) fouled with whey proteins, the flux decreases after nitric acid treatment in a [NaOHþHNO3] sequence. How explain the increase of the flux observed here with nitric acid alone or to finish the sequence without any removal of proteins from the fouling as previously shown?
L. Paugam et al. / Journal of Membrane Science 428 (2013) 275–280
1.00
0.80
0.74
0.57
J/J0
0.60
0.60
0.60
0.40 0.40
0.20
NaOH +HNO3
HNO3 +NaOH
NaOH
HNO3
Skim milk UF + water rinsing
0.00
Fig. 1. Flux recovery J/J0 (where J0 is the water flux of the pristine membrane and J the water flux after skim milk ultrafiltration, water rinsing or cleaning treatment) after HNO3 pH 1.6 or NaOH pH 11.5 alone or in sequence in spiral module.
1.00
0.80 0.57
J/J0
0.60 0.40
0.43
Skim milk UF + water rinsing
HCl
0.47
0.40
0.20
0.00 HNO3
NaNO3 neutral pH
Fig. 2. Flux recovery J/J0 (where J0 is the water flux of the clean membrane and J the water flux after skim milk UF and cleaning treatment) after HCl pH 1.6, HNO3 pH 1.6 and NaNO3 at neutral pH in spiral module.
3.2.3. Effect of nitrate at acid pH on the flux Fig. 2 gives the flux recovery after the treatment of the fouled membrane with nitric acid at pH 1.6 (J/J0 ¼0.57) compared with hydrochloric acid at the same pH (J/J0 ¼0.43) and sodium nitrate at pH 6.5 (J/J0 ¼0.47). None of these last two solutions allows to reach the J/J0 ratio obtained with HNO3. Indeed, the treatment with HCl is not significantly different from a single water rinsing step (J/J0 ¼0.40) and NaNO3 alone leads only to a flux recovery of 0.47. The impact of HNO3 on the flux values seems then to result from a synergy of the presence of nitrate ions in solution and acidic pH.
3.2.4. Effect of nitrate at acid pH on the protein fouled membrane The flux of the membrane fouled with proteins increases after the nitric acid treatment (from 30 to 48 L h 1 m 2) whereas it
Protein quantity on the membrane before nitric acid treatment (µg.cm-2)
278
40
30
20
10
0 0
5 10 20 25 30 15 Flux improvement after nitric acid treatment (%)
Fig. 3. Effect of nitric acid treatment on the flux as a function of the amount of protein on the membrane (HFK-131, plate and frame configuration).
remains constant with the pristine membrane (82 L h 1 m 2). This highlights an interaction between HNO3 and proteins rather than between HNO3 and the PES material. Moreover, Fig. 3 shows the flux variation induced by the nitric acid treatment as a function of the protein amount on the membrane. It clearly highlights that the flux variation depends on this parameter as it increases strongly with the protein amount on the membrane. 3.2.5. Effect of some oxoanions at acid pH on the protein fouled membrane Fig. 4 shows the water flux recovery and the percentage of protein removal (compared with a single rinsing) after a treatment with sulphuric, phosphoric and citric acid at pH 1.6. The effect of sulphuric acid both on flux and protein deposit is very low as these data are quite similar from that obtained with a single rinsing. So its impact on the protein fouling can be considered as negligible. On the contrary, phosphoric and citric acid lead to a strong increase of water flux recovery up to reach the criteria of hydraulic cleanliness (J/J0 ¼ 0.90). Nevertheless the amount of residual proteins is still very high. There is much more proteins than after sodium hydroxide for a flux more than twice higher. A specific adsorption of oxoanions such as phosphate and citrate on proteins has already been described by Rabiller-Baudry and Chaufer [21,22]. They show with capillary electrophoresis a change of electrophoretic mobibility, that highlights ion adsorption, correlated with a change of hydrophobic properties determined by HPLC. This phenomenon can explain results obtained here at acid pH for citrate, phosphate and also probably nitrate. The ionic adsorption would decrease the overall hydrophobicity of the fouled membrane without any removal of proteins. The more hydrophilic the deposit, the higher the flux is [23,24]. Nevertheless the cleaning with sulphuric acid leads to quite different results that are not explained here. 3.2.6. Acid nitric step is useless to clean PES UF membrane fouled with skim milk This decrease of hydrophobicity would also explain the difference obtained on the global water flux recovery value between the classical industrial sequence [NaOH þHNO3] (J/J0 ¼0.74) and its reverse [HNO3 þNaOH] (J/J0 ¼0.60) (Fig.1). Indeed, according to previous results, in both sequences only the NaOH step leads to a real protein removal. In the [NaOH þHNO3] sequence, the high final flux recovery is first due to the removal of a part of the proteins responsible for the fouling with NaOH and, secondly, to
L. Paugam et al. / Journal of Membrane Science 428 (2013) 275–280
1.00 1.00
J/J0
0.70
60 0.70 50
0.60
40
0.46
0.42
35 30
0.40
22 20
0.20 0
proteins removed (%)
58
0.80
100 70
10
3
0.00
0 Skim milk NaOH UF + rinsing
HNO3 H2SO4 H3PO4
Citric acid
Increase of water flux (%) after treatment
1.00
279
80
HNO3 step (2) Previous step (1) 2
60 15 40
20
35
35
3.2.8. Acid nitric step misleads on the real efficiency of enzymatic cleaning also In the enzymatic cleaning where the nitric acid is used as an inactivation step before the treatment, the flux recovery (Fig. 6) is 30% higher after the inactivation with HNO3 than after the enzymatic treatment alone or the enzymatic treatment followed by HCl inactivation. The flux increase is not correlated to a protein removal as its amount remains constant after the HNO3 treatment. Here also, the use of nitric acid misleads on the real efficiency of the cleaning.
1.NaOH/ClO2.HNO3
1.NaOH 2. HNO3
Only HNO3
70
1.00
60 0.80 0.7 0.60
40
0.42 0.40
32
50 0.61
0.58
32
36
30 20
Proteins removed (%)
3.2.7. On the base of flux analysis only, acid nitric step misleads on the real cleaning efficiency Fig. 5 presents the increase of water flux (%) obtained in spiral module at each step of different cleaning sequences. The nitric acid is first used directly after rinsing (0% of proteins from the irreversible fouling removed), then as a second step after a first one with increasing efficiency: hydrochloric acid (0% of proteins removed with HCl), sodium hydroxide pH 11.5 (24% of proteins removed with NaOH) and hypochlorite at pH 11.5 (73% of proteins removed with ClO- at this pH). The increase of the flux due to nitric acid treatment is in each case respectively about 35, 35, 15 and 2%. This shows that the higher the efficiency of the first step in the cleaning sequence, the lower the impact of HNO3 on the water flux. At industrial scale, the worse the alkaline cleaning, the higher the over-estimation of the global cleaning efficiency. The HNO3 is useless unless to reveal the state of the membrane cleaning without any in-depth analysis: if the membrane is well cleaned, the treatment with HNO3 will not increase the final water flux.
Fig. 5. Increase of water flux (%) after nitric acid treatment (2) depending on the pre-treatment (1¼ single rinsing, HCl, NaOH or NaOH/Cl) (spiral wound configuration).
J/J0
the specific adsorption of nitrates on residual proteins that changes the hydrophobicity of the deposit. Here, the last step with HNO3 misleads about the real cleaning efficiency of this cleaning mode. In the latter sequence [HNO3 þNaOH], where HNO3 is used in first step, the final water flux is the same than that obtained with NaOH alone. It highlights the reversible interaction between HNO3 and proteins. So nitric acid cannot be used to remove proteins and its use does not lead to a favourable synergy that will increase the cleaning efficiency of the second alkaline step. This acid step is then useless in terms of cleaning.
1.HCl 2.HNO3
0 Fig. 4. Water flux recovery (bar charts) and % of proteins removed (black squares) after treatment with NaOH at pH 11.5 and HNO3, H2SO4, H3PO4 and citric acid at pH 1.6 compared with a single rinsing (plate and frame configuration).
0.20 10 0.00
0 Skim milk UF + rinsing
P3-Ultrasil 53
P3-Ultrasil 53 + HNO3
P3-Ultrasil 53 + HCl
Fig. 6. Effect on the flux recovery (J/J0, bar charts) and on the protein removal (%, single rinsing as reference, black squares) of the treatment with the enzymatic product P3-Ultrasil 53 and of the acid chosen to neutralise enzymes (HNO3 or HCl) (plate and frame configuration).
The removal of nitric acid step from the industrial cleaning sequence can thus be envisaged. This will decrease strongly the total cleaning duration, limiting consequently the amount of energy, water and chemicals wasted, and will eliminate the nitrate rejection in effluents.
4. Conclusion The knowledge given by the in-depth characterisation of the membrane residual fouling allows us to optimise the cleaning with simple solutions for a specific application: the cleaning of a PES UF membrane fouled by skim milk.
280
L. Paugam et al. / Journal of Membrane Science 428 (2013) 275–280
No protein removal was observed with HNO3 cleaning, contrary to what the increase of flux suggests indirectly at first sight. This phenomenon was not observed when a protein fouled membrane is treated with HCl or NaNO3 and there is also no impact of HNO3 on a clean membrane. This flux increase is then specific of the interaction between nitrates and proteins at acid pH and its order of magnitude depends of the amount of proteins on the membrane. This can be attributed to a decrease of the overall hydrophobicity due to specific adsorption of nitrate on proteins. Flux measurement can then lead to a misleading estimation of the cleaning efficiency if it is not supported by physico-chemical analyse of membrane surface. This can lead to further production troubleshooting if the following disinfection is underestimated. Indeed, the hypochlorite plays then both the role of cleanser and disinfectant. If the quantity of hypochlorite used is insufficient to complete the cleaning, this can lead to trouble with subsequent skim milk production as fluxes are still worse than with an incomplete cleaning step [25]. The HNO3 step is then useless in the cleaning sequence and it can be suppressed. In a membrane installation of around 2000 m2 dedicated to the ultrafiltration of skim milk, with one cleaning a day, the stopping of this treatment step will earn up to 8000 L of water (treatment and following water rinsing). The gain in this given application will also be realised in terms of real efficiency and cleaning duration, leading finally, for the cleaning procedure itself, to a reduction of chemicals used, energy and for the global process, to a decrease of pollutant rejection (nitrates) and an increase of time production and most probably productivity.
Acknowledgements The authors want to thanks particularly the ‘‘Re´gion Bretagne’’ (Brittany), France for the financial support of David Delaunay and Nde´ye´ Wemsy Diagne Ph.D.’s work. References [1] H.B. Petrus, H. Li, V. Chen, N. Norazman, Enzymatic cleaning of ultrafiltration membranes fouled by protein mixture solutions, J. Memb. Sci. 325 (2008) 783–792. [2] A. Makardij, X.D. Chen, M.M. Farid, Microfiltration and ultrafiltration of milk: some aspects of fouling and cleaning, Food Bioprod. Process. 77 (1999) 107–113. [3] K. Yadav, K.R. Morison, Effects of hypochlorite exposure on flux through polyethersulfone ultrafiltration membranes, Food Bioprod. Process. 88 (2010) 419–424.
¨ anen, ¨ ¨ [4] P. Vais M.R. Bird, M. Nystrom, Treatment of UF membranes with simple and formulated cleaning agents, Food Bioprod. Process. 80 (2002) 98–108. [5] T. Mohammadi, S.S. Madaeni, M.K. Moghadam, Investigation of membrane fouling, Desalination 153 (2002) 155–160. [6] M. Kazemimoghadam, T. Mohammadi, Chemical cleaning of ultrafiltration membranes in the milk industry, Desalination 204 (2007) 213–218. [7] M. Bartlett, M.R. Bird, J.A. Howell, An experimental study for the development of a qualitative membrane cleaning model, J. Memb. Sci. 105 (1995) 147–157. [8] S. Muthukumaran, S. Kentish, S. Lalchandani, M. Ashokkumar, R. Mawson, G.W. Stevens, F. Grieser, The optimisation of ultrasonic cleaning procedures for dairy fouled ultrafiltration membranes, Ultrason. Sonochem. 12 (2005) 29–35. [9] N.D. Lawrence, J.M. Perera, M. Iyer, M.W. Hickey, G.W. Stevens, The use of streaming potential measurements to study the fouling and cleaning of ultrafiltration membranes, Sep. Purif. Technol. 48 (2006) 106–112. [10] M.O. Nigam, B. Bansal, X.D. Chen, Fouling and cleaning of whey protein concentrate fouled ultrafiltration, Desalination 218 (2008) 313–322. [11] H.F. Bohner, R.L. Bradley Jr, Effective cleaning and sanitizing of polysulfone ultrafiltration membrane systems, J. Dairy Sci. 75 (1992) 718–724. [12] G. Daufin, U. Merin, J.P. Labbe´, Cleaning of inorganic membranes after whey and milk ultrafiltration, Biotechnol. Bioeng. 38 (1) (1991) 82–89. ¨ [13] G. Salazar, MaA. Arguello, F.A. Riera, R. Alvarez, J.P. Labbe´, A. Que´merais, Chemical cleaning of inorganic membrane: rinsing and acid steps, Proceedings of Euromembrane’95, Bath, UK, II-87-II-90. [14] M. Rabiller-Baudry, M.n. Le Maux, B. Chaufer, L. Be´goin, Characterisation of cleaned and fouled membrane by ATR-FTIR and EDX analysis coupled with SEM: application to UF of skimmed milk with a PES membrane, Desalination 146 (2002) 123–128. [15] K.-J. Kim, P. Sun, V. Chen, D.E. Wiley, A.G. Fane, The cleaning of ultrafiltration membranes fouled by protein, J. Memb. Sci. 80 (1993) 241–249. [16] M. Rabiller-Baudry, L. Be´goin, D. Delaunay, L. Paugam, B. Chaufer, A dual approach of membrane cleaning based on physico-chemistry and hydrodynamics: application to PES membrane of dairy industry, Chem. Eng. Process. 47 (2008) 267–275. [17] M. Rabiller-Baudry, F. Gouttefangeas, L. Le Lannic, P. Rabiller, Coupling of SEM–EDX and FTIR-ATR to (quantitatively) investigate organic fouling on porous organic composite membranes, in: Me´ndez-Vilas (Ed.), Chapter in Current Microscopy Contributions to Advances in Science and Technology, Formatex, Badajoz, Spain, 2012. [18] L. Paugam, D. Delaunay, M. Rabiller-Baudry, Euromembrane, London, III-153, 2012. [19] L. Be´goin, Ph.D. Dissertation, Universite´ Rennes 1, France, 2004. ¨ [20] H. Zhu, M. Nystrom, Cleaning results characterized by flux, streaming potential and FTIR measurements, Colloids Surf. A 138 (1998) 309–321. [21] M. Rabiller-Baudry, B. Chaufer, Small molecular ion adsorption on proteins and DNAs revealed by separation techniques, J. Chromatogr. B 797 (2003) 331–345. [22] M. Rabiller-Baudry, B. Chaufer, Specificadsorption of phosphate ions on proteins evidenced by capillary electrophoresis and reversed-phase high performance liquid chromatography, J. Chromatogr. B 753 (2001) 67–77. [23] K.-J. Kim, A.G. Fane, Performance evaluation of surface hydrophilized novel ultrafiltration membranes using aqueous proteins, J. Memb. Sci. 99 (1995) 149–162. ¨ [24] A. Weis, M.R. Bird, M. Nystrom, The influence of morphology, hydrophobicity and charge upon the long-term performance of ultrafiltration membranes fouled with spent sulphite liquor, Desalination 175 (2005) 73–85. [25] L. Paugam, D. Delaunay, M. Rabiller-Baudry, Cleaning efficiency and impact on production fluxes of oxidising disinfectants on a pes ultrafiltration membrane fouled with proteins, Food Bioprod. Process. 88 (2010) 425–429.
Contents lists available at SciVerse ScienceDirect
Journal of Membrane Science journal homepage: www.elsevier.com/locate/memsci
Cleaning of skim milk PES ultrafiltration membrane: On the real effect of nitric acid step Lydie Paugam a,b,n, David Delaunay a,1, Nde´ye Wemsy Diagne a,b, Murielle Rabiller-Baudry a,b a b
Universite´ Rennes 1, UMR ‘‘Institut des Sciences Chimiques de Rennes’’—CNRS, CS 74205, case 1011, 35042 Rennes Cedex, France Universite´ Europe´enne de Bretagne, France
a r t i c l e i n f o
abstract
Article history: Received 13 July 2012 Received in revised form 10 October 2012 Accepted 11 October 2012 Available online 23 October 2012
Cleaning of PES membranes after ultrafiltration of skim milk is still often performed at industrial scale with a formulated alkaline first step followed by a nitric acid step. This cleaning mode gives sufficiently satisfactory results in terms of water flux recovery but cannot prevent a progressive decline of production flux over time. The analysis of the deposit on the membrane by FTIR-ATR highlights the nature of the irreversible fouling, exclusively made of proteins, and the real cleaning efficiency. Sodium hydroxides alone allows up to 24% of protein removal which is in good agreement with the water flux recovery after this treatment. Nevertheless, the amount of protein remains exactly the same after nitric acid treatment whereas the flux significantly grows. This flux increase without any protein removal is strongly dependent on the proteins amount on the membrane at the start of nitric acid step. This phenomenon is not observed with HCl, but it seems not totally specific from HNO3 as the use of H3PO4 in particular leads also to similar results. The specific adsorption of oxoanions on proteins at acid pH that will change the hydrophobicity of the deposit is suspected. The nitric acid step can be suppressed as it is useless and misleads on the real efficiency of the overall cleaning. & 2012 Elsevier B.V. All rights reserved.
Keywords: Ultrafiltration Polyethersulfone Skim milk Cleaning Nitric acid
1. Introduction Ultrafiltration is the process used to standardise the skim milk protein content in dairy industry for the consumption or before cheese making. The high fouling resulting from this application limits the productivity and needs twice daily cleaning operations (2–3 h for 6–8 h of production) to recover an acceptable flux. At industrial scale, cleaning is generally carried out with various empirical sequences performed with more or less complex cleaning solutions including an alkaline (typically sodium hydroxide NaOH with surfactant at pH 11.5) to remove organic matter and a nitric acid step (pH 1.6) to remove the mineral fouling (calcium deposit). An alternative to this chemical, also today more widely used industrially is the enzymatic cleaning [1] that is followed by a deactivation step with nitric acid as recommended by the supplier. This chemical or enzymatic cleaning is followed
n Corresponding author at: Universite´ Rennes 1, UMR ‘‘Institut des Sciences Chimiques de Rennes’’—CNRS, CS 74205, case 1011, 35042 Rennes Cedex, France. Tel.: þ33 223235172; fax: þ33 223235765. E-mail address: [email protected] (L. Paugam). 1 Present adress: Equipe de recherche´ en physico-chimie et biotechnologie, Universite´ de Caen Basse Normandie, Bvd mare´chal Juin 14032 Caen Cedex, France.
0376-7388/$ - see front matter & 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.memsci.2012.10.013
by disinfection with NaOH and sodium hypochorite (150– 200 ppm). This cleaning mode is justified by quite acceptable results in terms of flux recovery just after the treatment. Nevertheless, it can neither prevent a decrease of the production flux over several months scale nor sometimes an alteration of the membrane selectivity as it is often observed at industrial [2,3]. The cleaning step is clearly identified as a bottleneck of membrane separations due to a lack of knowledge about its fundamental mechanisms that makes its optimisation difficult. Some works focused on the cleaning of polyethersulfone ultrafiltration membranes used to concentrate whey proteins [4], of polysulfone membrane used to ultrafiltrate pasteurised milk [5,6], reconstituted skim milk [2] whey proteins from whey protein concentrate [7–10], cheese whey [11] and on the cleaning of inorganic membranes [12,13]. But the efficiency of the cleaning sequence is strongly dependent of the couple membrane/ fluid, particularly for complex dairy fluids. For the ultrafiltration of skim milk, the PES membrane HFK131 (5–10 kg mol 1) is a worldwide used membrane that represents 70% of the market. A previous study focused on the nature of the fouling on this PES membrane for this application. After skim milk ultrafiltration and rinsing, the analysis of the fouling revealed by SEM–EDX that there are here no minerals. Proteins are then the exclusive target of the cleaning. The nitric acid is
276
L. Paugam et al. / Journal of Membrane Science 428 (2013) 275–280
used after the alkaline cleaning to remove the mineral part (calcium phosphate) of the fouling by industrials that justify its use by the observed flux increase after the treatment with this product. The question that arises is if there any oxidative action on the residual proteins or any change in the structure of the deposit due to the pH change [4,15]. In this work, a cascade of NaOH followed by HNO3 is first studied to better understand the action of nitric acid during the cleaning in this given application. The first step with NaOH is certainly not the more efficient to remove the fouling, nevertheless it will highlights the real effect of nitric acid avoiding the impact of formulated alkaline detergent on the membrane hydrophobicity. Indeed surfactants contained in alkaline cleaning products are known to adsorb on the membrane and to increase their flux [16]. The effect of NaOH on the irreversible protein fouling is then first presented and is followed by the study of the second step of the cleaning sequence i.e. the nitric acid. Other solutions are studied to highlights the action of this reagent: sodium nitrate, hydrochloric acid, phosphoric acid, citric acid and sulphuric acid. The efficiency of both cleaning agents is evaluated in terms of water flux recovery (as in industry) and real protein removal (quantified by FTIR-ATR) in plate and frame and spiral configuration. The effect of nitric acid is finally presented in a biological cleaning where it used to inactivate the enzyme at the end of the cleaning process.
2. Experimental 2.1. Fluids The UHT skim milk (Lait de Montagne, Carrefour, France) chosen contain an average of 31.5 g L 1 of proteins and 48 g L 1 of carbohydrates as well as minerals, close to 8 g L 1. In this study several single solutions were used: sodium nitrate at 8.5 g L 1 and pH 6.6, nitric acid at pH 1.6 (65% analytical grade, Acros), hydrochloric acid at pH 1.6 (37% analytical grade, Acros), phosphoric acid at pH 1.6 (85%, Fisher), citric acid at pH 1.8 (Fisher), sulphuric acid at pH 1.6 (95–97%, Fluka), sodium hydroxide at pH 11.5 (pellets 97% Rectapur, Prolabo) and a chlorinated alkaline one with 200 mg L 1 of active Cl2 prepared from a stock NaOCl solution (bleach 371 Cl2, Lacroix) adjusted with NaOH at pH 11.5. The enzymatic cleaning was lead with P3-Ultrasil 53 (Ecolab). It was prepared at 1 wt% leading to a natural neutral pH. As the latter parameter was not regulated during the cleaning, the operating conditions were probably not perfectly the optimal. As recommended by providers, a nitric acid rinsing was done after the treatment to inactivate the enzyme activity. Nitric acid at pH 2.5 is usually suggested but HCl was also tested at the same pH. Water used for preparation of cleaning solutions was filtered (1 mm) and deionised. 2.2. Membrane The spiral UF membrane used was made of polyethersulfone (PES, HFK-131, Koch, 6.5 m2) with a MWCO of 5–10 kg mol 1. The spiral membrane was stored in sodium metabisulphite solution (5 g L 1, Acros, analytical grade) after each experiment to avoid any microorganism growth. Flat membranes (0.0127 m2) were also used as model for spiral membrane in order to quantify protein fouling by FTIR-ATR. Before use, flat membranes were first rinsed 10 min in methanol to remove the conservative (glycerol) and were then rinsed with deionised water. A new pristine membrane was used for every
new UF and was compressed by fluxing water for 1 h at a transmembrane pressure of 4 bar before the water reference flux was measured. 2.3. Ultrafiltration process 2.3.1. Spiral-wound pilot A pilot of 100 L capacity (TIA, Bolle ne, France) equipped with the spiral membrane HFK-131 [standard size 25–40, 6.5 m2, spacer of F type (2 mm)] was used. The fouling of the membrane was performed at 2 bar with a recirculation rate fixed at 10.5 m3 h 1 corresponding to a cross-flow velocity of about 0.3 m s 1 (as at industrial scale). 24 L of skim milk were ultrafiltered during 150 min at transmembrane pressure TMP ¼2 bar and 50 1C with a volume reduction ratio VRR¼1 (batch mode). Water rinsing was preceded by a total draining of the pilot and achieved at the minimal pressure allowed by this pilot (0.2 bar) at 150 L h 1 (i.e. a cross-flow velocity about 0.008 m s 1). In the first step, the module was rinsed with 70 L of filtered tap water (active carbon and 1 m filter) that was discarded towards waste. Then, in the second step 50 L of deionised water were recirculated in the loop during 20 min with the permeate discarded. Cleaning of the membrane by 25 L (3.8 L m 2), if not specified, of cleaning solution was performed 60 min at 50 1C in the same hydrodynamic conditions as the fouling. A second rinsing, as described before, was realised before a water flux measurement. 2.3.2. Plate-and-frame module The tangential plate and frame UF module (Ray-Flow X100, Orelis) presented an effective membrane surface of 0.0127 m2. Flat membranes as spacers (F type, 2 mm) were recovered from a commercial spiral module. The fouling was achieved by UF at VRR¼1 of 4 L of skim milk during 150 min at 50 1C and 2 bar. The following water rinsing was performed in 2 steps after draining the pilot, in order to minimise the water consumption. First, during 2 min rinsing was performed without recirculation of permeate and retentate and then, during 30 min, only retentate was recycled. The following cleaning was made with 4 L of solution. The pilot and the membrane were then rinsed with water until a neutral pH was reached both in the retentate and the permeate. A second rinsing was proceed before the measure of the aftercleaning water flux. The 3 steps of a set of experiments were always performed in the same hydrodynamics conditions: cross-flow velocity v¼0.3 m s 1 at 2 bar and 50 1C for fouling, water rinsing and cleaning. 2.4. Evaluation of cleaning efficiency At least two replicates of each cleaning experiments were done in spiral and in plate and frame configuration. Each time, water flux recovery and residual protein amount were evaluated. 2.4.1. Water flux recovery The water flux recovery was the ratio of the water flux after the chemical treatment and a rinsing (J) on the reference water flux (J0) that corresponded to the pristine membrane well cleaned and rinsed. The accuracy on flux measurement was better than 3% (coefficient of variation). Water flux recovery ¼ J=J 0 The hydraulic cleanliness is reached when the water flux recovery is up to 90%. The relative error on this parameter was 5%.
L. Paugam et al. / Journal of Membrane Science 428 (2013) 275–280
2.4.2. Physico-chemical characterisation: amount of residual proteins by FTIR-ATR The quantification of the amount of proteins on the flat membrane was obtained by FTIR-ATR analysis. The FTIR-ATR spectra were registered with a spectrometer Perkin–Elmer (Paragon 1000, spectrum for windows software) equipped with a ZnSe crystal with an incidence angle of 451 and 12 reflections (20 scans, 2 cm 1 resolution). Quantification in the range 0.5–350 mg cm 2 (geometric area, precision close to 1 mg cm 2 ) based on the H1539/H1240 ratio of the height of the amide II band of proteins (H1539 at 1539 cm 1) to the height of a PES band (H1240 at 1240 cm 1) was already described [16,17]. To take into account a possible variation of the deposit amount on the flat sheet membrane (due to small variations of velocity profiles), an average value was calculated from nine pieces, cut in the membrane.
3. Results and discussion The global fouling of the membrane results from a reversible and an irreversible fouling. The first one is removed by water rinsing when the second remains and need a chemical cleaning. At industrial scale the skim milk is pasteurised before UF. A previous study was lead to analyse the irreversible fouling after pasteurised skim milk UF at 50 1C [18]. It was shown that the irreversible fouling is exclusively due to proteins. There is no mineral fouling as firstly expected (industrial cleaning sequences are thought on this base). This was also shown by Rabiller-Baudry et al. [14] in the case of UHT skim milk UF. As it is more difficult for technical and practical reasons to use pasteurised skim milk at lab scale and on the base of these results, UHT skim milk was chosen here to model industrial skim milk. After the ultrafiltration of (UHT) skim milk, the membrane is first rinsed with deionised water leading to a first water flux recovery of 40%. 3.1. Alkaline step The flux recovery after a treatment with sodium hydroxide at pH 11.5 is shown on Table 1. The variation of various parameters is presented: time treatment, volume of NaOH to membrane surface ratio and configuration. For the NaOH on spiral wound configuration with 3.8 L m 2 (that corresponds to a volume of 25 L in our installation), it gives J/J0 ¼0.60. The cleaning with this solution is then not sufficient to reach the hydraulic cleanliness. Neither the increase of NaOH volume to the membrane surface ratio nor the treatment duration, the use of a fresh solution of sodium hydroxide and the change of configuration allow to increase the flux recovery. The increase of solution volume to membrane surface ratio in spiral configuration gives here the best results i.e. J/J0 ¼ 0.74. Table 1 Water flux recovery after treatment with sodium hydroxide solutions depending on treatment duration, ratio of solution volume to membrane surface (spiral and plate and frame configuration, pH 11.5, 50 1C and 60 min). J/J0 Water rinsing NaOH (spiral, 3.8 L m 2) 60 min NaOH (spiral, 3.8 L m 2) 90 min NaOH (spiral, 3.8 L m 2) 2 60 min NaOH (spiral, 7.7 L m 2) 60 min NaOH (plate and frame, 315 L m 2) 60 min
0.42 0.60 0.67 0.70 0.74 0.70
277
Table 2 FTIR-ATR quantification of residual proteins on the PES UF membrane after the cleaning (residual proteins before cleaning¼ 32 71 g cm 2) in plate and frame module.
HNO3 pH 1.6 NaOH pH 11.5
Residual proteins on the membrane after cleaning (mg cm 2)
Proteins removed with the cleaning step (%)
33 23
0 28
The quantity of residual proteins on the membrane (in plate and frame configuration) is analysed by FTIR-ATR. The results are presented in Table 2. It shows that the treatment with sodium hydroxide at pH 11.5 leads to the removal of 24% of the protein from the irreversible fouling. This protein removal is in good agreement with the increase of flux recovery. Previous results [19] have shown that sodium hydroxide alone does not hydrolyse proteins at pH 11.5 and 50 1C in 1 h. The action of this product on the protein deposit will rather be explained by the increase of protein charge imposed by the pH change [14,15,20]. The deposit becomes swollen and less rigid. This would favour the partial removal of the protein [7]. 3.2. Nitric acid step Nitric acid usually comes as a second step in the cleaning of skim milk UF membranes. It is used after the alkaline step to remove the eventual mineral part of the fouling. However, the irreversible fouling, that remains after the rinsing, was previously identified as exclusively made of proteins. In this context, no real cleaning action of this chemical reagent can really be expected. 3.2.1. Effect of nitric acid on the protein deposit The effect of nitric acid on the protein deposit is first studied when it is used alone and then in sequence with NaOH. Table 2 shows the results of the analysis of residual amount of proteins on the membrane after UF of skim milk, water rinsing and nitric acid treatment. The quantity of proteins of the irreversible fouling is 32 mg cm-2 (71 mg cm 2). The protein amount remains the same after nitric acid treatment. So there is no protein removal with this treatment, as expected from the knowledge of fundamental mechanisms of organic soiling cleaning. 3.2.2. Effect of nitric acid on the flux Fig. 1 shows the water flux recovery J/J0 of the membrane HFK131 fouled with skim milk and then just rinsed or rinsed and treated with a single step of HNO3 or NaOH or with two steps HNO3/NaOH or the classical sequence NaOH/HNO3. It appears first that the water flux recovery after nitric acid is very close (J/J0 ¼0.57) from that obtained with sodium hydroxide (J/J0 ¼0.60). The flux recovery is much higher than after a single rinsing of the membrane (J/J0 ¼0.40). The sequence that presents the better results in terms of flux recovery is the [NaOHþHNO3] (J/J0 ¼0.74) which is in good agreement with information from industrial source. The inverse one [HNO3 þ NaOH] gives present a lower performance (J/J0 ¼0.60) i.e. a result similar from NaOH or HNO3 alone. It is to note that V¨ais¨anen et al. [4] present quite different results as for a PES ultrafiltration membrane (PES 50H, Hoechst) fouled with whey proteins, the flux decreases after nitric acid treatment in a [NaOHþHNO3] sequence. How explain the increase of the flux observed here with nitric acid alone or to finish the sequence without any removal of proteins from the fouling as previously shown?
L. Paugam et al. / Journal of Membrane Science 428 (2013) 275–280
1.00
0.80
0.74
0.57
J/J0
0.60
0.60
0.60
0.40 0.40
0.20
NaOH +HNO3
HNO3 +NaOH
NaOH
HNO3
Skim milk UF + water rinsing
0.00
Fig. 1. Flux recovery J/J0 (where J0 is the water flux of the pristine membrane and J the water flux after skim milk ultrafiltration, water rinsing or cleaning treatment) after HNO3 pH 1.6 or NaOH pH 11.5 alone or in sequence in spiral module.
1.00
0.80 0.57
J/J0
0.60 0.40
0.43
Skim milk UF + water rinsing
HCl
0.47
0.40
0.20
0.00 HNO3
NaNO3 neutral pH
Fig. 2. Flux recovery J/J0 (where J0 is the water flux of the clean membrane and J the water flux after skim milk UF and cleaning treatment) after HCl pH 1.6, HNO3 pH 1.6 and NaNO3 at neutral pH in spiral module.
3.2.3. Effect of nitrate at acid pH on the flux Fig. 2 gives the flux recovery after the treatment of the fouled membrane with nitric acid at pH 1.6 (J/J0 ¼0.57) compared with hydrochloric acid at the same pH (J/J0 ¼0.43) and sodium nitrate at pH 6.5 (J/J0 ¼0.47). None of these last two solutions allows to reach the J/J0 ratio obtained with HNO3. Indeed, the treatment with HCl is not significantly different from a single water rinsing step (J/J0 ¼0.40) and NaNO3 alone leads only to a flux recovery of 0.47. The impact of HNO3 on the flux values seems then to result from a synergy of the presence of nitrate ions in solution and acidic pH.
3.2.4. Effect of nitrate at acid pH on the protein fouled membrane The flux of the membrane fouled with proteins increases after the nitric acid treatment (from 30 to 48 L h 1 m 2) whereas it
Protein quantity on the membrane before nitric acid treatment (µg.cm-2)
278
40
30
20
10
0 0
5 10 20 25 30 15 Flux improvement after nitric acid treatment (%)
Fig. 3. Effect of nitric acid treatment on the flux as a function of the amount of protein on the membrane (HFK-131, plate and frame configuration).
remains constant with the pristine membrane (82 L h 1 m 2). This highlights an interaction between HNO3 and proteins rather than between HNO3 and the PES material. Moreover, Fig. 3 shows the flux variation induced by the nitric acid treatment as a function of the protein amount on the membrane. It clearly highlights that the flux variation depends on this parameter as it increases strongly with the protein amount on the membrane. 3.2.5. Effect of some oxoanions at acid pH on the protein fouled membrane Fig. 4 shows the water flux recovery and the percentage of protein removal (compared with a single rinsing) after a treatment with sulphuric, phosphoric and citric acid at pH 1.6. The effect of sulphuric acid both on flux and protein deposit is very low as these data are quite similar from that obtained with a single rinsing. So its impact on the protein fouling can be considered as negligible. On the contrary, phosphoric and citric acid lead to a strong increase of water flux recovery up to reach the criteria of hydraulic cleanliness (J/J0 ¼ 0.90). Nevertheless the amount of residual proteins is still very high. There is much more proteins than after sodium hydroxide for a flux more than twice higher. A specific adsorption of oxoanions such as phosphate and citrate on proteins has already been described by Rabiller-Baudry and Chaufer [21,22]. They show with capillary electrophoresis a change of electrophoretic mobibility, that highlights ion adsorption, correlated with a change of hydrophobic properties determined by HPLC. This phenomenon can explain results obtained here at acid pH for citrate, phosphate and also probably nitrate. The ionic adsorption would decrease the overall hydrophobicity of the fouled membrane without any removal of proteins. The more hydrophilic the deposit, the higher the flux is [23,24]. Nevertheless the cleaning with sulphuric acid leads to quite different results that are not explained here. 3.2.6. Acid nitric step is useless to clean PES UF membrane fouled with skim milk This decrease of hydrophobicity would also explain the difference obtained on the global water flux recovery value between the classical industrial sequence [NaOH þHNO3] (J/J0 ¼0.74) and its reverse [HNO3 þNaOH] (J/J0 ¼0.60) (Fig.1). Indeed, according to previous results, in both sequences only the NaOH step leads to a real protein removal. In the [NaOH þHNO3] sequence, the high final flux recovery is first due to the removal of a part of the proteins responsible for the fouling with NaOH and, secondly, to
L. Paugam et al. / Journal of Membrane Science 428 (2013) 275–280
1.00 1.00
J/J0
0.70
60 0.70 50
0.60
40
0.46
0.42
35 30
0.40
22 20
0.20 0
proteins removed (%)
58
0.80
100 70
10
3
0.00
0 Skim milk NaOH UF + rinsing
HNO3 H2SO4 H3PO4
Citric acid
Increase of water flux (%) after treatment
1.00
279
80
HNO3 step (2) Previous step (1) 2
60 15 40
20
35
35
3.2.8. Acid nitric step misleads on the real efficiency of enzymatic cleaning also In the enzymatic cleaning where the nitric acid is used as an inactivation step before the treatment, the flux recovery (Fig. 6) is 30% higher after the inactivation with HNO3 than after the enzymatic treatment alone or the enzymatic treatment followed by HCl inactivation. The flux increase is not correlated to a protein removal as its amount remains constant after the HNO3 treatment. Here also, the use of nitric acid misleads on the real efficiency of the cleaning.
1.NaOH/ClO2.HNO3
1.NaOH 2. HNO3
Only HNO3
70
1.00
60 0.80 0.7 0.60
40
0.42 0.40
32
50 0.61
0.58
32
36
30 20
Proteins removed (%)
3.2.7. On the base of flux analysis only, acid nitric step misleads on the real cleaning efficiency Fig. 5 presents the increase of water flux (%) obtained in spiral module at each step of different cleaning sequences. The nitric acid is first used directly after rinsing (0% of proteins from the irreversible fouling removed), then as a second step after a first one with increasing efficiency: hydrochloric acid (0% of proteins removed with HCl), sodium hydroxide pH 11.5 (24% of proteins removed with NaOH) and hypochlorite at pH 11.5 (73% of proteins removed with ClO- at this pH). The increase of the flux due to nitric acid treatment is in each case respectively about 35, 35, 15 and 2%. This shows that the higher the efficiency of the first step in the cleaning sequence, the lower the impact of HNO3 on the water flux. At industrial scale, the worse the alkaline cleaning, the higher the over-estimation of the global cleaning efficiency. The HNO3 is useless unless to reveal the state of the membrane cleaning without any in-depth analysis: if the membrane is well cleaned, the treatment with HNO3 will not increase the final water flux.
Fig. 5. Increase of water flux (%) after nitric acid treatment (2) depending on the pre-treatment (1¼ single rinsing, HCl, NaOH or NaOH/Cl) (spiral wound configuration).
J/J0
the specific adsorption of nitrates on residual proteins that changes the hydrophobicity of the deposit. Here, the last step with HNO3 misleads about the real cleaning efficiency of this cleaning mode. In the latter sequence [HNO3 þNaOH], where HNO3 is used in first step, the final water flux is the same than that obtained with NaOH alone. It highlights the reversible interaction between HNO3 and proteins. So nitric acid cannot be used to remove proteins and its use does not lead to a favourable synergy that will increase the cleaning efficiency of the second alkaline step. This acid step is then useless in terms of cleaning.
1.HCl 2.HNO3
0 Fig. 4. Water flux recovery (bar charts) and % of proteins removed (black squares) after treatment with NaOH at pH 11.5 and HNO3, H2SO4, H3PO4 and citric acid at pH 1.6 compared with a single rinsing (plate and frame configuration).
0.20 10 0.00
0 Skim milk UF + rinsing
P3-Ultrasil 53
P3-Ultrasil 53 + HNO3
P3-Ultrasil 53 + HCl
Fig. 6. Effect on the flux recovery (J/J0, bar charts) and on the protein removal (%, single rinsing as reference, black squares) of the treatment with the enzymatic product P3-Ultrasil 53 and of the acid chosen to neutralise enzymes (HNO3 or HCl) (plate and frame configuration).
The removal of nitric acid step from the industrial cleaning sequence can thus be envisaged. This will decrease strongly the total cleaning duration, limiting consequently the amount of energy, water and chemicals wasted, and will eliminate the nitrate rejection in effluents.
4. Conclusion The knowledge given by the in-depth characterisation of the membrane residual fouling allows us to optimise the cleaning with simple solutions for a specific application: the cleaning of a PES UF membrane fouled by skim milk.
280
L. Paugam et al. / Journal of Membrane Science 428 (2013) 275–280
No protein removal was observed with HNO3 cleaning, contrary to what the increase of flux suggests indirectly at first sight. This phenomenon was not observed when a protein fouled membrane is treated with HCl or NaNO3 and there is also no impact of HNO3 on a clean membrane. This flux increase is then specific of the interaction between nitrates and proteins at acid pH and its order of magnitude depends of the amount of proteins on the membrane. This can be attributed to a decrease of the overall hydrophobicity due to specific adsorption of nitrate on proteins. Flux measurement can then lead to a misleading estimation of the cleaning efficiency if it is not supported by physico-chemical analyse of membrane surface. This can lead to further production troubleshooting if the following disinfection is underestimated. Indeed, the hypochlorite plays then both the role of cleanser and disinfectant. If the quantity of hypochlorite used is insufficient to complete the cleaning, this can lead to trouble with subsequent skim milk production as fluxes are still worse than with an incomplete cleaning step [25]. The HNO3 step is then useless in the cleaning sequence and it can be suppressed. In a membrane installation of around 2000 m2 dedicated to the ultrafiltration of skim milk, with one cleaning a day, the stopping of this treatment step will earn up to 8000 L of water (treatment and following water rinsing). The gain in this given application will also be realised in terms of real efficiency and cleaning duration, leading finally, for the cleaning procedure itself, to a reduction of chemicals used, energy and for the global process, to a decrease of pollutant rejection (nitrates) and an increase of time production and most probably productivity.
Acknowledgements The authors want to thanks particularly the ‘‘Re´gion Bretagne’’ (Brittany), France for the financial support of David Delaunay and Nde´ye´ Wemsy Diagne Ph.D.’s work. References [1] H.B. Petrus, H. Li, V. Chen, N. Norazman, Enzymatic cleaning of ultrafiltration membranes fouled by protein mixture solutions, J. Memb. Sci. 325 (2008) 783–792. [2] A. Makardij, X.D. Chen, M.M. Farid, Microfiltration and ultrafiltration of milk: some aspects of fouling and cleaning, Food Bioprod. Process. 77 (1999) 107–113. [3] K. Yadav, K.R. Morison, Effects of hypochlorite exposure on flux through polyethersulfone ultrafiltration membranes, Food Bioprod. Process. 88 (2010) 419–424.
¨ anen, ¨ ¨ [4] P. Vais M.R. Bird, M. Nystrom, Treatment of UF membranes with simple and formulated cleaning agents, Food Bioprod. Process. 80 (2002) 98–108. [5] T. Mohammadi, S.S. Madaeni, M.K. Moghadam, Investigation of membrane fouling, Desalination 153 (2002) 155–160. [6] M. Kazemimoghadam, T. Mohammadi, Chemical cleaning of ultrafiltration membranes in the milk industry, Desalination 204 (2007) 213–218. [7] M. Bartlett, M.R. Bird, J.A. Howell, An experimental study for the development of a qualitative membrane cleaning model, J. Memb. Sci. 105 (1995) 147–157. [8] S. Muthukumaran, S. Kentish, S. Lalchandani, M. Ashokkumar, R. Mawson, G.W. Stevens, F. Grieser, The optimisation of ultrasonic cleaning procedures for dairy fouled ultrafiltration membranes, Ultrason. Sonochem. 12 (2005) 29–35. [9] N.D. Lawrence, J.M. Perera, M. Iyer, M.W. Hickey, G.W. Stevens, The use of streaming potential measurements to study the fouling and cleaning of ultrafiltration membranes, Sep. Purif. Technol. 48 (2006) 106–112. [10] M.O. Nigam, B. Bansal, X.D. Chen, Fouling and cleaning of whey protein concentrate fouled ultrafiltration, Desalination 218 (2008) 313–322. [11] H.F. Bohner, R.L. Bradley Jr, Effective cleaning and sanitizing of polysulfone ultrafiltration membrane systems, J. Dairy Sci. 75 (1992) 718–724. [12] G. Daufin, U. Merin, J.P. Labbe´, Cleaning of inorganic membranes after whey and milk ultrafiltration, Biotechnol. Bioeng. 38 (1) (1991) 82–89. ¨ [13] G. Salazar, MaA. Arguello, F.A. Riera, R. Alvarez, J.P. Labbe´, A. Que´merais, Chemical cleaning of inorganic membrane: rinsing and acid steps, Proceedings of Euromembrane’95, Bath, UK, II-87-II-90. [14] M. Rabiller-Baudry, M.n. Le Maux, B. Chaufer, L. Be´goin, Characterisation of cleaned and fouled membrane by ATR-FTIR and EDX analysis coupled with SEM: application to UF of skimmed milk with a PES membrane, Desalination 146 (2002) 123–128. [15] K.-J. Kim, P. Sun, V. Chen, D.E. Wiley, A.G. Fane, The cleaning of ultrafiltration membranes fouled by protein, J. Memb. Sci. 80 (1993) 241–249. [16] M. Rabiller-Baudry, L. Be´goin, D. Delaunay, L. Paugam, B. Chaufer, A dual approach of membrane cleaning based on physico-chemistry and hydrodynamics: application to PES membrane of dairy industry, Chem. Eng. Process. 47 (2008) 267–275. [17] M. Rabiller-Baudry, F. Gouttefangeas, L. Le Lannic, P. Rabiller, Coupling of SEM–EDX and FTIR-ATR to (quantitatively) investigate organic fouling on porous organic composite membranes, in: Me´ndez-Vilas (Ed.), Chapter in Current Microscopy Contributions to Advances in Science and Technology, Formatex, Badajoz, Spain, 2012. [18] L. Paugam, D. Delaunay, M. Rabiller-Baudry, Euromembrane, London, III-153, 2012. [19] L. Be´goin, Ph.D. Dissertation, Universite´ Rennes 1, France, 2004. ¨ [20] H. Zhu, M. Nystrom, Cleaning results characterized by flux, streaming potential and FTIR measurements, Colloids Surf. A 138 (1998) 309–321. [21] M. Rabiller-Baudry, B. Chaufer, Small molecular ion adsorption on proteins and DNAs revealed by separation techniques, J. Chromatogr. B 797 (2003) 331–345. [22] M. Rabiller-Baudry, B. Chaufer, Specificadsorption of phosphate ions on proteins evidenced by capillary electrophoresis and reversed-phase high performance liquid chromatography, J. Chromatogr. B 753 (2001) 67–77. [23] K.-J. Kim, A.G. Fane, Performance evaluation of surface hydrophilized novel ultrafiltration membranes using aqueous proteins, J. Memb. Sci. 99 (1995) 149–162. ¨ [24] A. Weis, M.R. Bird, M. Nystrom, The influence of morphology, hydrophobicity and charge upon the long-term performance of ultrafiltration membranes fouled with spent sulphite liquor, Desalination 175 (2005) 73–85. [25] L. Paugam, D. Delaunay, M. Rabiller-Baudry, Cleaning efficiency and impact on production fluxes of oxidising disinfectants on a pes ultrafiltration membrane fouled with proteins, Food Bioprod. Process. 88 (2010) 425–429.