Study of biofilms on PVDF membranes after chemical cleaning by sodium hypochlorite

Separation and Purification Technology 141 (2015) 314–321

Contents lists available at ScienceDirect

Separation and Purification Technology journal homepage: www.elsevier.com/locate/seppur

Study of biofilms on PVDF membranes after chemical cleaning by sodium hypochlorite Anna Piasecka a, Roy Bernstein a, Frans Ollevier b, Filip Meersman c,d, Caroline Souffreau b, Roil M. Bilad a, Karl Cottenie b, Louise Vanysacker a, Carla Denis b, Ivo Vankelecom a,⇑ a

Centre for Surface Chemistry and Catalysis, Katholieke Universiteit Leuven, Kasteelpark Arenberg 23, PO Box 2461, 3001 Heverlee, Belgium Laboratory of Aquatic Ecology, Evolution and Conservation, Katholieke Universiteit Leuven, Charles Deberiotstraat 32, 3000 Leuven, Belgium Biomolecular & Analytical Mass Spectrometry Group, University of Antwerp, Groenenborgerlaan 171, 2020 Antwerp, Belgium d Department of Chemistry, University College London, 20 Gordon Street, London, United Kingdom b c

a r t i c l e

i n f o

Article history: Received 9 July 2014 Received in revised form 1 December 2014 Accepted 3 December 2014 Available online 23 December 2014 Keywords: Biofouling Cleaning PVDF membrane Biofilm

a b s t r a c t Sodium hypochlorite (NaOCl) is widely used to remove biofouling in membrane bioreactors (MBRs) to recover the membrane performance. In this study, the effect of membrane cleaning with different NaOCl concentrations (0.01%, 0.1%, 1% and 10% of a stock solution containing 39.92 g/L of free chlorine) on biofouling was investigated in a molasses based lab-scale MBR. Study of the bacterial biofilm community re-growth after six consecutive cleanings revealed that a minimal concentration of 0.1% NaOCl diminishes the bacterial richness and cell density on the membranes. ATR-FTIR analysis of the layer on the membrane surface revealed the presence of peaks associated with proteins and carbohydrates present in the biofouling layer and their intensity decreased after treatment with NaOCl. Analysis of the membrane performance by chemical oxygen demand (COD) measurement of the permeate and retentate showed that the rejection of the membranes after NaOCl chemical treatment was still high. The data showed that since NaOCl removes the bacterial biofilm and at the same time does not affect the membrane treatment performance, NaOCl can be recommended as a cleaning agent to remove biofouling in a lab-scale molasses based MBR. Ó 2014 Published by Elsevier B.V.

1. Introduction Membrane bioreactors (MBRs) are a promising technology for wastewater treatment [1], but their widespread application is hindered by the inability to control effectively membrane fouling. Fouling is the process where solutes or particles deposit onto a membrane or into the membrane pores causing membrane obstruction and decrease of MBR performance. Fouling limits the achievable permeate flux, reduces the sustainability of operation, increases the cleaning frequency, reduces the lifetime of the membrane, etc. Membrane fouling can be classified as colloidal (clays, floc), organic (oils, humics), inorganic (mineral participates) or biofouling (caused by bacteria or fungi) [2]. Biofouling is driven by bacteria, present in the activated sludge, that adhere to the membrane surface and start to produce a biofilm by excreting extracellular polymeric substances (EPS) [3]. EPS are a complex mixture of mainly proteins and carbohydrates, but also acid polysaccharides, DNA and lipids, ⇑ Corresponding author at: Centrum voor Oppervlaktechemie en Katalyse, Dept. M2S, Faculteit Bio-ingenieurswetenschappen, KU Leuven, Belgium. E-mail address: [email protected] (I. Vankelecom). http://dx.doi.org/10.1016/j.seppur.2014.12.010 1383-5866/Ó 2014 Published by Elsevier B.V.

that form a matrix that surrounds cells in flocs and biofilms [4,5]. The initial phase of biofilm formation also promotes the aggregation of other activated sludge components, such as metal ions, resulting in a dense structure present on the membrane surface. Biofouling increases the transmembrane pressure (TMP) that is required over the membrane to maintain a constant flux operation. Therefore, when a critical threshold pressure is reached, the membrane requires chemical cleaning or even replacement [6]. Various strategies are used to control membrane biofouling and include physical cleaning, such as, back-washing [7], back-pulsing [8] and air sparging [9]. Another promising effort to alleviate biofouling is membrane modifications [10–14]. Yet other methods involve the application of biological based cleaning by using EPS degrading enzymes, such as proteases, polysaccharases and DNAses [15–17], by adding bacteriophages [18] or by inhibiting quorum sensing signals [19,20]. However, all these methods are still in their developmental phase and have not been translated into effective strategies for industrial MBRs. Therefore, in many full-scale MBRs, membrane chemical cleaning is still an essential step to maintain performance on the longer term. In most full-scale MBR installations, the most popular cleaning agent remains sodium

A. Piasecka et al. / Separation and Purification Technology 141 (2015) 314–321

hypochlorite (NaOCl) [21], often combined with citric acid [6]. Up to date, studies on NaOCl mainly focussed on its direct impact on MBR performance, investigating the effect of NaOCl cleaning with respect to water flux recovery [22] and membrane permeability recovery [23]. Also, a number of studies have investigated the effects of chemical cleaning using sodium hypochlorite on polyvinylidene fluoride (PVDF) membranes [24,25] or other membranes such as polysulfone (PSf), polyether-sulfone (PES) and cellulose acetate based ones [26–30]. Only two studies refer to microbial community characterization after chemical cleaning. In both of them, the cleaning was performed inside the MBR (chemically enhanced backwashing), thus these studies focused on microorganism activities in activated sludge as a function of NaOCl doses supplied via backwashing [31,32]. However, the initial biofilm formation with respect to microbial communities re-growing after multiple cleanings and under different regimes has not yet been examined. Fundamental information such as bacterial richness and density is essential for optimising anti-biofouling strategies. In practice, it is also important to determine optimal NaOCl concentrations that eliminate most of the attached organic material and bacteria with limited damage to the membrane. To gain insight into these aspects, a series of cleaning experiments was performed in a lab-scale MBR treating molasses wastewater and using NaOCl as a cleaning agent. The objectives of this study were to analyze the effect of the chemical cleaning with four different NaOCl concentrations on (i) bacterial communities attached to the membrane in terms of bacterial richness and cell density, (ii) chemical functional groups of initially formed biofilms, and (iii) PVDF membrane functionality.

315

modules (Fig. 1). The pressure difference created by the pump (transmembrane pressure, TMP) was monitored by a pressure gauge (accuracy ±2%), which was installed for each module individually. The operational flux was 20 L/m2 h. Each membrane had a maximum effective filtration area of 165 cm2 and a pore size of approximately 0.138 lm. The air source with constant pressure of 2 bar, was directly linked to the bottom of each module position. The MBR was operated in a fed-batch mode using sludge that was fed with a 2.2 mL/L molasses-based synthetic wastewater. The sludge seed was obtained from a pilot-scale MBR treating molasses wastewater (Waterleau, Wespelaar, Belgium). The characteristics of the wastewater and operational parameters are presented in Table 1. The MBR was run for one year for a different experiment prior to this study and therefore can be considered as fully stabilized. 2.1.2. PVDF Membrane preparation The lab-made polyvinylidene fluoride (PVDF) membranes were prepared via the phase inversion technique. Membranes were labprepared to ensure full control over the membrane chemistry and morphology. Polyvinylidene fluoride (PVDF) membranes were prepared by dissolving 12 wt% of polymer (MW  534 kDa, Aldrich, Germany) into N,N-dimethylformamide (DMF, supplied by Acros Organic). The solution was cast to form a 250 lm wet-thickness film onto a polypropylene support (Novatexx 2471, supplied by Freudenberg, Germany) at a casting speed of 2.25 cm/s and then coagulated into a demineralised water bath, acting as the non-solvent. After complete coagulation, all membranes were washed with water, dried and stored in open air.

2. Materials and methods 2.1. Experimental set-up 2.1.1. MBR system and operating conditions A lab-scale MBR (High-Throughput Membrane Systems, Leuven, Belgium) (30 L) was equipped with a holder for 20 membrane

2.1.3. Methods of PVDF membrane characterization The surface and cross-section images of the membranes were obtained using scanning electron microscopy (SEM) (Philips SEMXL30 FREG with EDX dx-4i system) and further analyzed for pore size and surface porosity with Image J (http://rsbweb.nih.gov/ij/). Before SEM analysis, the membranes were dried by immersing

Fig. 1. (A) Scheme of the membrane module arrangement and cleaning strategy. Membranes (M1 to M10) were cleaned ex situ using different NaOCl concentrations (0.01– 10%) or Milli-Q (control) during 24 h and placed back in the same position after water rinsing and one hour of water filtration to remove the residual NaOCl, (B) A scheme of single membrane module.

316

A. Piasecka et al. / Separation and Purification Technology 141 (2015) 314–321

Table 1 Main molasses wastewater characteristics and operational parameters of activated sludge. Influent

Mean concentration (mg l1)

Chemical oxygen demand (COD) Total nitrogen (TN) Total phosphorous (TP) NH4AN NO3AN PO4AP Calcium Iron Potassium Magnesium Manganese Sodium Zinc Sulfur

963 25 4.11 4.07 3.25 0.44 0.95 0.95 0.05 0.06 0.03 7.09 0.07 5.79

Activated sludge DO pH MLSS

8 7–8 6000

2.2. Bacterial community characterization in initial biofilm formation

them for 10 min in 4 different aqueous ethanol solutions containing 25%, 50%, 75% and 100% ethanol and dried afterwards. Membrane surface hydrophobicity was determined using contact angle (CA) goniometry (VCA Optima video camera system, AST Products, Billerica) based on the sessile drop method. The CA was measured at least at five different positions and the contact angle was determined immediately after the drop had reached the membrane and after 10 min. The critical flux (CF) was measured using the stepwise method suggested by Le-Clech et al. [33]. The applied initial flux, step height and step duration were 2 L/m2 h, 2 L/m2 h and 5 min, respectively. To determine the CF, the final TMP values of each step were plotted against the fluxes. Below the CF, a linear relationship exists between the TMP increment and the imposed fluxes. The CF was determined to be the minimum flux at which the linear TMP-to-flux proportionality was not present anymore [34]. The membrane characterization is presented in Table 2. 2.1.4. Cleaning protocol and sample collection During 43 days of operation, the membranes were cleaned physically and chemically or in case of control samples with Milli-Q water on a weekly basis. The membranes were taken out of the MBR and the biofilm was physically removed using a sterile cell scraper. Afterwards, the membranes were immersed into NaOCl. Sodium hypochlorite solutions were prepared based on ‘‘eau de javel’’, Lacroix, 12% of chlorine (39.92 g/L). Afterwards, the membranes were rinsed and a one hour filtration with tap water was carried out to remove the residual NaOCl. Four starting concentrations of NaOCl were tested: 0.01%, 0.1%, 1%, 10% v/v of stock solution containing 39.92 g/L free chlorine (4.03, 40.3, 403, 4030 ppm) for 24 h at room temperature. The pH values of the solutions were respectively 8.44, 9.28, 11.34 and 11.97 respectively. In total, 6 weekly cleaning routines were performed on days 7, 14, 21, 28, 35 and 42. During the cleaning, the MBR operation was temporarily stopped. After the sixth treatment (last treatment), the membranes were placed back for one additional day

Table 2 Characteristics of the lab-made PVDF membranes used in the study. Pore size (lm)

0.01

Surface porosity (%)

19

CF (L/m2 h)

18

into the MBR to obtain a sufficient amount of biofilm for further analysis. This final bacterial biofilm was characterized using TRFLP and Q-PCR. The chemical functional groups of biofouled PVDF membranes were determined using ATR-FTIR. For T-RFLP and QPCR, 1 cm2 of each membrane with one day old biofilm was sampled and used for DNA extraction. For ATR-FTIR, membranes with a one-day old biofilm were cut into 3 cm2 pieces, dried and used directly for the measurements.

Contact angle 5s

10 min

73 ± 7

49 ± 2

2.2.1. Bacterial community composition and richness by TerminalRestriction Fragment Length Polymorphism (T-RFLP) analysis Microbial DNA was extracted using the MoBio UltracleanTM Soil [35,36] DNA kit (MoBio Laboratories, USA) according to the manufacturer’s instructions. DNA quality and quantity was determined with a NanoDrop spectrophotometer (NanoDrop Technologies, ND-1000). For T-RFLP analysis, a universal primer pair was used to amplify a fragment of 900 bp from the 16S rRNA gene. The bacterial 8–27 f-primer (50 -AGAGTTTGATCCTGGCTCAG-30 ) was labelled at the 50 -end with carboxyfluorescein (6-FAM) and the 907–926r (50 -CCGTCAATTCCTTTTAGTTT-30 ) was non-labelled [37]. Each PCR mix (25 lL) contained 2.5 lL 10  PCR buffer (Eurogentec), 1 lL MgCl2 (50 mM), 2.5 lL dNTP (2 mM), 0.5 lL of each primer (0.5 pmol), 0.25 lL of Silverstar Taq DNA polymerase (2.5 U) (Eurogentec). Also 50 ng of template DNA was added to the PCR mix. Thermal cycling conditions were 94 °C for 2 min, followed by 30 cycles of 94 °C for 1 min, 60 °C for 1 min, 72 °C for 1 min, and final extension at 72 °C for 5 min. PCR products were checked by agarose gel electrophoresis and UV visualization using GelRedTM (Biotium) stain. Purified PCR-products (200 ng) were digested with 20 U HhaI (Fermentas) according to Smalla et al. [37]. The enzymatic reaction was subjected to the following conditions: 10 lL PCR product, 2 lL enzyme, 2 lL Buffer Tango (10) and 6 ll MQ water. The samples were incubated at 37 °C for 4 h. Digested PCR samples (5 lL each) were checked on 1.5% agarose gel after GelRedTM staining (Biotium). 1.2 lL of digested PCR products were mixed with 0.3 lL of internal size standard (GeneScanTM 1200 LizÒ size standard, Applied Biosystems) and 8.5 lL formamide. Terminal fragments were separated by capillary electrophoresis on an ABI Prism 3130-Avant Genetic Analyser (Applied Biosystems) using POP 7 polymer. A pilot reproducibility test of the T-RFLP protocol showed that the overall error rate was negligible for our analysis. More than 95% of the bands were reproducible in a subset of samples that was analyzed three times. T-RFLP profiles were scored using GeneMapper (version 4.0, Applied Biosystems). Profiles of low T-RFLP quality having peak height <100 fluorescence units (FUs) or cumulative peak height <1000 FUs were considered unreliable and removed. T-RFLP profiles were then manually aligned by visual inspection of the internal size standard. The presence or absence of peaks at loci was scored using 100 FUs as a threshold. The samples were combinations of terminal restriction fragments (T-RFs) of different sizes (i.e., base pair lengths of the fragments) and different relative abundance (height of the fluorescence peak). Two profiles per membrane sample were obtained. The bacterial richness was defined as the sum of T-RFs as detected by the TRFLP analysis and was statistically compared between treatments using one-way ANOVA and post hoc honest significant difference test. This analysis was performed in Statistica 10 (Statsoft. Inc). 2.2.2. Cell density by real-time quantitative polymerase chain reaction PCR (Q-PCR) The positive control DNA used for the standard curve for 16S rRNA gene was based on extracted DNA of known amounts of

317

A. Piasecka et al. / Separation and Purification Technology 141 (2015) 314–321

Escherichia coli cells that were cultured on Brain Heart Infusion Agar (BHIA) for 24 h. The number of genes coding for 16S rRNA can vary from 1 to 14 which normally required a correction factor on the cell number [38,39]. However, as E. coli has 7 copies of its rRNA operon (which is a good approximation for the average) no corrections were made for the number of genes and E. coli density was determined by taking the average of conventional plate counting in triplicate and re-calculations of the total amount of DNA (as determined by Thermo Scientific™ NanoDrop Lite Spectrophotometer) on the average amount of 8.4 fg (1015 g) DNA per bacterial cell [40]. DNA amplification and PCR products detection were performed using ABI PrismÒ 7000 Sequence Detection System (Applied Biosystems). The mastermix contained 12.5 lL SYBR Green, 0.75 lL of both primers (10 mmol), MQ water 6.75 lL, and 5 lL template DNA. The standardized Q-PCR contained the following steps: 2 min at 50 °C, 10 min at 95 °C, followed by 50 cycles of 15 s at 95 °C and 1 min at 60 °C. The E. coli standard curve was automatically generated by the ABI Prism system by plotting the cycle threshold (Ct) versus the logarithmic concentration of the positive control DNA. The Ct is defined as the number of cycles required for the fluorescent signal of the target DNA to cross the threshold (i.e., exceeds background signal). Ct levels are inversely proportional to the amount of target nucleic acid in the sample. The PCR inhibition for this type of sludge was previously analyzed by Vanysacker et al. [41]. No inhibition was detected. Q-PCR data were further analyzed using SDS 1.2.3 software (Applied Biosystems). Bacterial cell densities (cells/cm2 membrane) of the five different treatments were statistically compared using one-way ANOVA and post hoc Tukey HSD (Honest Significant Difference) test in Statistica 11 (StatSoft. Inc).

2.3. Chemical functional groups in the initially formed biofilms Attenuated total reflectance fourier transform infrared (ATR FTIR) spectroscopy was performed to analyze the changes in chemical functional groups of the biofouled and chemically cleaned membranes. This analysis is aimed at observing the residual fouling, which could not be removed via the applied treatment and to observe the impact of each treatment to the membrane material. The dried membranes were placed directly onto a germanium crystal and infrared spectra were recorded in a N2 flushed cell with a Bruker IFS66 Fourier transform infrared (FTIR) spectrophotometer equipped with a liquid nitrogen cooled broad band mercurycadmium-telluride solid-state detector. Two-hundred fifty scans were collected at a resolution of 2 cm1.

2.4. Membrane performance TMP was monitored by an electronic pressure gauge (WIKA Benelux, accuracy ±0.5%) each time before and after the cleaning procedure for each membrane separately. The average from two membranes was taken into account. Standard methods were used to measure the chemical oxygen demand (COD). It was measured in influent and in permeates of each membrane to have an indication of membrane rejection capability. COD was measured at day 7 and day 43 (after the experiment) using a Hach COD analysis kit and the cuvets tests were measured with Hach(R) DR2800 spectrofotometer. To test statistically the effect of treatment on the COD, a repeated ANOVA measurement was performed in Statistica 11 (StatSoft. Inc).

3. Results and discussion 3.1. Bacterial biofilm community characterization in initial biofilm formation In the present study, four different NaOCl concentrations (0.01%, 0.1%, 1% and 10%, corresponding to 4.03, 40.3, 403, 4030 ppm of free chlorine and Milli-Q water (for control samples) were selected for the treatment (24 h immersion). In individual applications, the optimal range of NaOCl concentrations and cleaning conditions for membranes have not been standardized so far. Moreover, it is also common that MBR suppliers propose their own chemical cleaning recipes, which differ mainly in terms of concentration and method [42]. Also in studies published so far, there is a large variation in NaOCl concentrations that have been studied for membrane cleaning. Table 3 summarizes some recently published work on NaOCl cleaning of different membrane types and shows that the concentration of NaOCl applied by other researchers is usually around 100–1000 ppm. Although it is difficult to compare between cleaning experimental set-ups due to different conditions, the highest concentration in this experiment (10% NaOCl for 24 h) is probably too high to be considered of practical use due to high costs and the possible faster membrane degradation. However, in order to have a broad variation in the NaOCl concentrations, both very low and very high NaOCl concentrations were selected and analyzed in this study. Furthermore, the wide range of NaOCl concentrations used and the high frequency of cleaning (weekly) accelerated the membrane aging and allowed to observe in a short experimental time the possible impact on re-growing biofilms. The bacterial richness on the membrane surfaces was analyzed based on the number of T-RFs peaks obtained from the T-RFLP profile of each membrane treated with a different NaOCl concentration. As shown in Fig. 2, the bacterial richness was significantly lower for the membranes cleaned with higher NaOCl concentrations (0.1%, 1% and 10%) compared to membranes cleaned with 0.01% NaOCl and the control condition. However, no further significant decrease in richness is apparent between the NaOCl concentrations of 0.1%, 1% and 10% (p < 0.05). This shows that a NaOCl concentration of 0.1% already has the capacity to diminish the bacterial richness to the same level as higher concentrations. The bacterial cell densities in the re-growing biofilms over the different treatments are shown in Fig. 3. In general, the cell densities depend on the applied NaOCl concentration, as there is a statistical difference between each treatment, except between the treatments involving 0.1% and 1% NaOCl. However, similar to bacterial richness, some reduction in cell density was observed for the concentrations of 0.1% and higher (from ca. 40  106 bacteria/cm2 for the control and 37  106 bacteria/cm2 for the 0.01% NaOCl to ca. 10  106 bacteria/cm2 for the 0.1% NaOCl). It must be acknowledged that the effect is quite limited, which is probably related to the technique used in this study. Even though DNA based real-time quantitative polymerase chain reaction (Q-PCR) is the most widely applied technology for direct quantification of cells

Table 3 Published studies on NaOCl cleaning of different membranes. NaOCl (ppm)

Type of membrane

Reference

100–500 100–5000 100 400 700 100 750, 3600, 22,200 and 44.300

Not mentioned PVDF PSf/PVP PSf/PVP/PEG PES PVDF PVDF

[6] [25] [41] [27] [42] [43] [24]

318

A. Piasecka et al. / Separation and Purification Technology 141 (2015) 314–321

Fig. 2. Average bacterial richness (as number of T-RFs) in the control and NaOCl treatments 1 day after the sixth cleaning. Error bars represent one standard deviation (N = 2). Different letters between treatments represent significant differences (p < 0.05).

Fig. 3. Q-PCR approximating the density of the bacteria on the membranes surfaces after different treatment and control. Error bars represent one standard deviation (N = 2). Different letters between treatments represent significant differences (p < 0.05).

in mixed samples and is increasingly used for direct detection and quantification of bacteria in foods and environmental and clinical samples [43], the major disadvantage is its inability to differentiate between viable and nonviable cells [44]. Intact DNA can be present although the organisms are dead [43,45–47]. This is particularly relevant for pathogens subjected to killing treatments by using disinfectants as in the current experiment. The lack of viable/dead differentiation can present a limitation for the implementation of DNA diagnostics in routine cleaning application. Recently, ethidium monoaizide bromide (EMA) combined with (Q-PCR) was shown to be a promising approach for qualitative DNA-based viable/dead differentiation in samples with mixed bacterial population [43,48] and should be considered in follow-up studies on chemical cleaning by NaOCl. Nevertheless, the present study demonstrates that there is a reduction in species density upon NaOCl treatment, in agreement with other studies demonstrating that relatively high concentrations (P0.1%) of NaOCl are necessary to effectively kill the bacteria in different medical sources. For instance, Siqueira et al. [49] used 4% NaOCl to eliminate Enterococcus faecalis from root canals after incubation for 4 days. Sassone et al. [50] showed that 1% and 5% NaOCl combined with 0.5% chlorhexidine solution had antimicrobial properties against bacteria such as Staphylococcus aureus, E. faecalis, E. coli, Porphyromonas gingivalis and Fusobacterium nucleatum. Radcliffe et al. [51] and Vianna et al. [52] found that 2.5% NaOCl inhibited the growth of Candida albicans and E. faecalis when incubated for 5 and 10 min, respectively. Therefore, it can be assumed that 10% NaOCl completely removes bacteria from the

membrane surface. Nevertheless, as seen in Fig. 3, around 5  106 cells/cm2 were found on the membrane cleaned with 10% NaOCl. Since the biofilm on the membranes was only one day old, the bacterial fractions present on the membrane after treatment with 10% NaOCl are most probably a result of adhesion of pioneering bacterial species to the completely cleaned membrane. Earlier studies on MBR biofouling revealed that there are specific pioneer bacteria which adhere first to the membrane surface [53,54], modifying its surface properties by their presence and/or by EPS production, and subsequently enable the colonization by secondary colonizers [55,56]. On the other hand, in case of treatments at lower NaOCl concentrations such as 0.01%, 0.1% and 1%, not enough biofouling or bacterial taxa might have been removed to allow the adhesion of the pioneer species. The fraction of bacteria on the membranes after treatment with lower NaOCl concentrations (Fig. 3) might thus represent the pioneer bacteria species together with secondary colonizers after treatment. However, since the bacterial taxa and re-colonization mechanism was not analyzed in this study, the differentiation between the pioneer and secondary bacteria species is rather hypothetical and future studies are required to support this assumption. Another potential drawback of this experiment is the lack of EPS measurement as well as the number of viable cells on the membranes immediately after each cleaning. The actual impact on EPS breakdown thus could not be estimated. It might be important to know which concentration of free chlorine breaks up the EPS matrix associated with biofilms and inactivates or kills the viable cells present inside the biofilm. Nevertheless, presented density and richness data show that despite chemical cleaning the bacteria are present on the membranes already one day after cleaning. This could potentially be an effect of rapid adhesion of either pioneer or secondary bacterial species to the membranes. The re-colonization is influenced by NaOCl concentration and the NaOCl concentration of 0.1% combined with physical cleaning diminishes the bacterial richness and cell density. It is then likely that the cleaning effect by NaOCl is not linear. This observation is in agreement with previous work by Madaeni and Mansourpanah [57], where the cleaning efficiency of a wide variety of cleaning agents including acids, bases, enzymes and complexing cleaning agents was studied. It has been shown that in cleaning of reverse osmosis membranes fouled by whey, the higher concentration did not increase the cleaning efficiency significantly or even decreased it. The same effect is most probably observed for NaOCl cleaning in our lab-scale experiment. It should be mentioned that commercial ‘‘eau de javel’’ might contain certain amounts of surfactants. The synergistic effect of the free chlorine from the ‘‘eau de javel’’ and the possible surfactants might have affected the obtained results. Free chlorine may freely diffuse from the solution to destroy the biofilm bacterial cells, present on the PVDF membrane. The spreading properties of the cleaning solution may be enhanced by addition of surfactants [58]. In other fields, synergies were found to lead to improved penetration depth of cleaning solutions [59], disinfection [60–62] and better tissue dissolution [62–64]. 3.2. Analysis of the chemical functional groups present in the membrane and the biofouling layer after NaOCl treatments In order to determine the effect of NaOCl cleaning on the chemical functional groups of PVDF membranes and their biofouling layers, the membrane and the biofouling layer were characterized by means of ATR-FTIR spectroscopy after one day of filtration following the final cleaning treatment. Fig. 4 presents the ATR-FTIR spectra of the fouled membranes after the sixth treatment followed by one day of biofilm re-growth. The peaks at 1175 cm1, 1072 cm1 and 1402 cm1 are typical for PVDF membranes and are assigned to asymmetric stretching of CAF2, CAC stretching

319

A. Piasecka et al. / Separation and Purification Technology 141 (2015) 314–321

Fig. 4. ATR-FTIR spectra of membranes after the sixth cleaning treatment carried out with various NaOCl concentrations (%) or Milli-Q water (control). Peaks at different wavelengths correspond to specific chemical structures: 1175 cm1 = C–F stretching, 1402 cm1 = CH2-wagging, 1650 cm1 = amide stretching groups; 1730 cm1 = C@O stretching carbonyl groups.

and CAH2 wagging, respectively [65]. Compared to the pristine membrane, few new peaks appeared on the fouled membranes: a peak at 1650 cm1, a broad peak at 1000–1100 cm1 and a peak at 1730 cm1. These new peaks are probably the result of the biofouling. The protein fraction of the biofouling may be represented by the peak at 1650 cm1 which is assigned to an amide group, whereas the broad peak at 1000–1110 cm1 can be attributed to the polysaccharide fraction of the biofouling [66]. These additional peaks can be attributed to polysaccharides and proteins of the biofouling layer, either irreversibly accumulated during the six weeks of treatment, or as a result of the new fouling layer following the additional one day of filtration after cleaning. The relative intensity of these peaks can be used to qualitatively estimate the biofouling layer present on the membrane surfaces. The data show that the intensity of these peaks was lower for the membranes treated with NaOCl concentrations compared to the control membrane, indicating a decrease of the biofouling layer after cleaning with NaOCl concentrations. For example, the intensity of the peak at 1072 cm1 which is likely attributed to the carbohydrate fraction of the biofouling, is the highest for the control membrane and the lowest for the membrane after treatment with NaOCl 10%. Fig. 4 shows also the new peak at 1730 cm1, only present in fouled membranes and which can be attributed to a carbonyl group (C@O). The intensity of this peak increases for membranes treated with 0.1%, 1% and 10% NaOCl. This carbonyl group normally originates from possible membrane additives, such as polyvinylpyrrolidone (PVP) or hydroxalkyl acrylates or methacrylates [67–69], which are used by the membrane manufacturer in order to increase the membrane hydrophilicity. However, the lab-made PVDF membranes used in this study were not modified by any substance, and contain only PVDF. Moreover, ATR-FTIR spectra of control experiments that were carried out by soaking pristine membranes in Milli-Q water for the same periods indicate that those membranes did not contain this peak (data not shown). Therefore, the new peak is clearly linked to biofouling, resulting from a fraction that possibly contains a carboxyl group, most probably, due to oxidation of functional groups in the biofouling. For example, it was found that the C@O bond of cane molasses can be oxidized (by curing treatment) to aliphatic carboxylic groups with a peak at 1730 cm1 [70]. The fact that in our data the intensity of this peak increased only at high NaOCl doses supports this assumption. 3.3. Membrane performance To determine the effect of biofouling on membrane performance, the TMPs were measured as an indication for biofouling

accumulation on the membrane [42]. Fig. 5 shows the TMP profiles for the membranes just before and the day after each chemical and physical cleaning. As expected, the TMPs before cleaning were higher compared to the TMPs after cleaning, confirming the elimination of biofouling. The TMP level for each membrane before cleaning reached a maximum before the second and third cleaning (day 14 and day 22) and did not increase in later stages. Moreover, the TMP level for all membranes tended to decrease after the third cleaning. This is most likely due to some changes in the membrane structure which occurred already after the second or third cleaning. This observation is in agreement with COD measurements in the permeates after the first cleaning (day 7) and the last cleaning (day 42) (Table 4). The COD concentration and% COD rejection is significantly higher in the permeates of the membranes after NaOCl treatment on day 7 and 42 (p = 0.006) compared to the COD in the permeate of the control. However, there is no significant difference between the COD values of membranes (and in COD rejection%) after different NaOCl treatments (p > 0.05). For all membranes, except for the control sample, the COD values are significantly higher on day 42 than on day 7 (p < 0.05). As more chemical components are found in the permeate of membranes after chemical treatment and their amount increases with time, this indicates that the membrane properties might change due to NaOCl cleaning, causing a worse rejection compared to the control membrane. Although it is not clear which membrane modifications were involved in our study, previous research reports that NaOCl can impact the physico-chemical characteristics of the membrane, affecting not only the extent of fouling, but also the treatment performance of the membrane systems. For instance, Abdullah and Berube [24], based on FTIR and nuclear magnetic resonance

Fig. 5. TMP values (bar) before and after each cleaning for the control membrane and membranes after hypochlorite treatment (average from the two membranes). Day 7 corresponds to the first cleaning, 14 to the second, 21 to the third, 28 to the fourth, 35 to the fifth and 42 to the sixth cleaning. The TMP data represent an average of 2 membrane replicates (N = 2).

Table 4 The COD values and COD rejection % for influent wastewater and membrane permeates measured at day 7 (first cleaning) and 42 (last cleaning). The obtained values are from two membrane replicates (N = 2). Different letters in superscript between treatments represent significant differences (p < 0.05). COD (mg/L) Day 7

Control 0.01% 0.1% 1% 10%

Day 42

Rejection (%) Day 7

Day 42

Average

Wastewater (influent) 1360a 1360a 1360

0a

0a

0

Permeate 58.4b 64.5b 66.5b 60.4b 64.6b

95.7b 95.3b 95.1b 95.5b 95.3b

95.5c 92.7c 92.3c 93.8c 92.5c

95.6 94.0 93.7 94.7 93.9

60.5c 98.0c 105.0c 83.2c 102.2c

Average

59.5 81.3 85.8 71.8 83.4

320

A. Piasecka et al. / Separation and Purification Technology 141 (2015) 314–321

(NMR) analysis, reported that NaOCl exposure removed a portion of the hydrophilic additives present in PVDF membrane by oxidation. The mechanical strength, surface hydrophilicity, pore size and porosity, and membrane resistance of the membranes were negatively affected by NaOCl [26]. Other studies demonstrated the negative effects of NaOCl treatment for non-PVDF membranes on mechanical strength [26], surface hydrophobicity [26,28,29] and increasing pore size [27,30]. Nevertheless, in the present study the difference in COD between the control and the treatments was only on average around 20 mg/L. The rejection rate of membranes after NaOCl cleaning was still high, ranging from 93.7% (in the permeate of the membrane after treatment with 0.1% NaOCl) to 94.7% (in the permeate of membrane after treatment with 1%). The data regarding the TMP and COD measurement confirm that chemical cleaning by NaOCl possibly caused some membrane modification, but did not significantly affect the membrane treatment performance. 4. Conclusions Successive chemical cleaning of PVDF membranes with NaOCl influences the re-growing bacterial biofilm community. The bacterial richness and density decreased significantly after applying NaOCl concentrations of 0.1% or higher. ATR-FTIR detected the presence of peaks corresponding to protein and carbohydrate biofouling fraction on the membrane. Their intensity decreased after cleaning with NaOCl, indicating that the biofouling layer on the membrane is reduced. A study of the membrane performance by measurement of the COD concentration in the permeate of the control membrane and after NaOCl cleaning revealed that the control membrane, which was not subjected to NaOCl treatment, had a significantly higher COD rejection which did not change with time, whereas the COD rejection of the treated membranes became less. This indicates that the NaOCl treatment caused some membrane modifications. Nonetheless, the rejection of the membranes after NaOCl chemical treatment was still high, despite the rigorous cleaning. Since NaOCl caused the reduction of bacterial richness and density and at the same time did not significantly modify the membrane properties, NaOCl can be recommended as a cleaning agent to remove biofouling in a lab-scale molasses based MBR. Although most probably the current results cannot be scaled up to full-scale MBRs, the design was conceived to get a first, lab-scale idea of the impact of NaOCl cleaning on biofilm and membrane characteristics. Acknowledgements This work was supported by the IDO/06/008 and OT/11/061 projects of the KU Leuven Research Fund, the IAP FS2 and Methusalem CASAS funding from the Belgian aand Flemish government respectively. References [1] M.A. Shannon, P.W. Bohn, M. Elimelech, J.G. Georgiadis, B.J. Marinas, A.M. Mayes, Science and technology for water purification in the coming decades, Nature 452 (2008) 301. [2] J.F. Kramer, D.A. Tracey, The solution to reverse osmosis biofouling, in: Proceedings of IDA World Congress on Desalination and Water Use, AbuDhabi, Saudi Arabia 4, 1995, pp. 33. [3] L. Malaeb, P. Le-Clech, J.S. Vrouwenvelder, G.M. Ayoub, P.E. Saikaly, Do biological-based strategies hold promise to biofouling control in MBRs?, Water Res 47 (2013) 5447. [4] B. Frolund, R. Palmgren, K. Keiding, P.H. Nielsen, Extraction of extracellular polymers from activated sludge using a cation exchange resin, Water Res. 30 (1996) 1749–1758. [5] X.Y. Li, S.F. Yang, Influence of loosely bound extracellular polymeric substances (EPS) on the flocculation, sedimentation and dewaterability of activated sludge, Water Res. 41 (2007) 1022.

[6] S. Judd, The status of membrane bioreactor technology, Trends Biotechnol. 26 (2008) 109. [7] J.L. Wu, P. Le-Clech, R.M. Stuetz, A.G. Fane, V. Chen, Effects of relaxation and backwashing conditions on fouling in membrane bioreactor, J. Membr. Sci. 324 (2008) 26. [8] H.M. Ma, L.F. Hakim, C.N. Bowman, R.H. Davis, Factors affecting membrane fouling reduction by surface modification and backpulsing, J. Membr. Sci. 189 (2001) 255. [9] N. Gupta, N. Jana, C.B. Majumder, Submerged membrane bioreactor system for municipal wastewater treatment process: an overview, Indian J. Chem. Technol. 15 (2008) 604. [10] A. Adout, S. Kang, A. Asatekin, A.M. Mayes, M. Elimelech, Ultrafiltration membranes incorporating amphiphilic comb copolymer additives prevent irreversible adhesion of bacteria, Environ. Sci. Technol. 44 (2010) 2406. [11] C.X. Liu, D.R. Zhang, Y. He, X.S. Zhao, R.B. Bai, Modification of membrane surface for anti-biofouling performance. Effect of anti-adhesion and antibacteria approaches, J. Membr. Sci. 346 (2010) 121. [12] J. Mansouri, S. Harrisson, V. Chen, Strategies for controlling biofouling in membrane filtration systems: challenges and opportunities, J. Mater. Chem. 20 (2010) 4567. [13] Z. Sheng, Y. Liu, Effects of silver nanoparticles on wastewater biofilms, Water Res. 45 (2011) 6039. [14] H.Y. Yu, L.Q. Liu, Z.Q. Tang, M.G. Yan, J.S. Gu, X.W. Wei, Mitigated membrane fouling in an SMBR by surface modification, J. Membr. Sci. 310 (2008) 409. [15] F.C. Petersen, L. Tao, A.A. Scheie, DNA binding-uptake system: a link between cell-to-cell communication and biofilm formation, J. Bacteriol. 187 (2005) 4392. [16] C.B. Whitchurch, T. Tolker-Nielsen, P.C. Ragas, J.S. Mattick, Extracellular DNA required for bacterial biofilm formation, Science 295 (2002) 1487. [17] Y. Xiong, Y. Liu, Biological control of microbial attachment: a promising alternative for mitigating membrane biofouling, Appl. Microbiol. Biot. 86 (2010) 825. [18] S. Goldman, J. Starosvetsky, R. Armon, Inhibition of biofilm formation on UF membrane by use of specific bacteriophages, J. Membr. Sci. 342 (2009) 145. [19] W. Jiang, S. Xia, J. Liang, Z. Zhang, S.W. Hermanowicz, Effect of quorum quenching on the reactor performance, biofouling and biomass characteristics in membrane bioreactors, Water Res. 47 (2013) 187. [20] H.S. Oh, K.M. Yeon, C.S. Yang, S.R. Kim, C.H. Lee, S.Y. Park, J.Y. Han, J.K. Lee, Control of membrane biofouling in MBR for wastewater treatment by quorum quenching bacteria encapsulated in microporous membrane, Environ. Sci. Technol. 46 (2012) 4877. [21] P. Le-Clech, Membrane bioreactors and their uses in wastewater treatments, Appl. Microbiol. Biotechnol. 88 (2010) 1253. [22] S.K.S. Razavi, J.L. Harris, F. Sherkat, Fouling and cleaning of membranes in the ultrafiltration of the aqueous extract of soy flour, J. Membr. Sci. 114 (1996) 93. [23] X.X. Yan, R. Gerards, L. Vriens, I. Vankelecom, Hollow fiber membrane fouling and cleaning in a membrane bioreactor for molasses wastewater treatment, Desalin. Water Treat. 18 (2010) 192. [24] S.Z. Abdullah, P.R. Berube, Assessing the effects of sodium hypochlorite exposure on the characteristics of PVDF based membranes, Water Res. 47 (2013) 5392. [25] V. Puspitasari, A. Granville, P. Le-Clech, V. Chen, Cleaning and ageing effect of sodium hypochlorite on polyvinylidene fluoride (PVDF) membrane, Sep. Purif. Technol. 72 (2010) 301. [26] E. Arkhangelsky, U. Goren, V. Gitis, Retention of organic matter by cellulose acetate membranes cleaned with hypochlorite, Desalination 223 (2008) 97. [27] E. Gaudichet-Maurin, F. Thominette, Ageing of polysulfone ultrafiltration membranes in contact with bleach solutions, J. Membr. Sci. 282 (2006) 198. [28] I. Levitsky, A. Duek, E. Arkhangelsky, D. Pinchev, T. Kadoshian, H. Shetrit, R. Naim, V. Gitis, Understanding the oxidative cleaning of UF membranes, J. Membr. Sci. 377 (2011) 206. [29] C. Regula, E. Carretier, Y. Wyart, M. Sergent, G. Gesan-Guiziou, D. Ferry, A. Vincent, D. Boudot, P. Moulin, Ageing of ultrafiltration membranes in contact with sodium hypochlorite and commercial oxidant: experimental designs as a new ageing protocol, Sep. Purif. Technol. 103 (2013) 119. [30] S. Rouaix, C. Causserand, P. Aimar, Experimental study of the effects of hypochlorite on polysulfone membrane properties, J. Membr. Sci. 277 (2006) 137. [31] E.J. Lee, J.S. Kwon, H.S. Park, W.H. Ji, H.S. Kim, A. Jang, Influence of sodium hypochlorite used for chemical enhanced backwashing on biophysical treatment in MBR, Desalination 316 (2013) 104. [32] B.R. Lim, K.H. Ahn, K.G. Song, J.W. Cho, Microbial community in biofilm on membrane surface of submerged MBR: effect of in-line cleaning chemical agent, Water Sci. Technol. 51 (2005) 201. [33] P. Le-Clech, B. Jefferson, I.S. Chang, S.J. Judd, Critical flux deermination by the flux-step method in a submerged membrane bioreactor, J. Membr. Sci. 227 (2003) 81. [34] A. Faraquharson, H. Zhou, Relationship of activated sludge characteristics to fouling rate and critical flux in membrane bioreactors for wastewater treatment, Chemosphere 79 (2010) 149. [35] R.I. Amann, W. Ludwig, K.H. Schleifer, Phylogenetic identification and in situ detection of individual microbial cells without cultivation, Microbiol. Rev. 59 (1995) 143. [36] D.P. Chandler, F.J. Brockman, J.K. Fredrickson, Use of 16S rDNA clone libraries to study changes in a microbial community resulting from ex situ perturbation of a subsurface sediment, FEMS Microbiol. Rev. 20 (1997) 217.

A. Piasecka et al. / Separation and Purification Technology 141 (2015) 314–321 [37] K. Smalla, M. Oros-Sichler, A. Milling, H. Heuer, S. Baumgarte, R. Becker, G. Neuber, S. Kropf, A. Ulrich, C.C. Tebbe, Bacterial diversity of soils assessed by DGGE, T-RFLP and SSCP fingerprints of PCR-amplified 16S rRNA gene fragments: do the different methods provide similar results?, J Microbiol. Met. 69 (2007) 470. [38] V. Farrelly, F.A. Rainey, E. Stackebrandt, Effect of genome size and rrn gene copy number on PCR amplification of 16S rRNA genes from a mixture of bacterial species, Appl. Environ. Microbiol. 61 (1995) 2798. [39] J.A. Klappenbach, J.M. Dunbar, T.M. Schmidt, RRNA operon copy number reflects ecological strategies of bacteria, Appl. Environ. Microbiol. 66 (2000) 1328. [40] V.L. Torsvik, J. Goksoyr, Determination of bacterial DNA in soil, Soil Biol. Biochem. 10 (1978) 7. [41] L. Vanysacker, S.A. Declerck, B. Hellemans, L. De Meester, I. Vankelecom, P. Declerck, Bacterial community analysis of activated sludge: an evaluation of four commonly used DNA extraction methods, Appl. Microbiol. Biotechnol. 88 (2010) 299. [42] P. Le-Clech, V. Chen, T.A.G. Fane, Fouling in membrane bioreactors used in wastewater treatment, J. Membr. Sci. 284 (2006) 17. [43] K. Rudi, B. Moen, S.M. Dromtorp, A.L. Holck, Use of ethidium monoazide and PCR in combination for quantification of viable and dead cells in complex samples, Appl. Environ. Microbiol. 71 (2005) 1018. [44] S. Wang, R.E. Levin, Discrimination of viable Vibrio vulnificus cells from dead cells in real-time PCR, J. Microbiol. Meth. 64 (2006) 1. [45] E. Dupray, M.P. Caprais, A. Derrien, P. Fach, Salmonella DNA persistence in natural seawaters using PCR analysis, J. Appl. Microbiol. 82 (1997) 507. [46] K.L. Josephson, C.P. Gerba, I.L. Pepper, Polymerase chain reaction detection of nonviable bacterial pathogens, Appl. Environ. Microbiol. 59 (1993) 3513. [47] C.I. Masters, J.A. Shallcross, B.M. Mackey, Effect of stress treatments on the detection of Listeria monocytogenes and enterotoxigenic Escherichia coli by the polymerase chain reaction, J. Appl. Bacteriol. 77 (1994) 73. [48] A. Nocker, A.K. Camper, Selective removal of DNA from dead cells of mixed bacterial communities by use of ethidium monoazide, Appl. Environ. Microb. 72 (2006) 1997. [49] J.F. Siqueira, A.G. Machado, R.M. Silveira, H.P. Lopes, M. DeUzeda, Evaluation of the effectiveness of sodium hypochlorite used with three irrigation methods in the elimination of Enterococcus faecalis from the root canal in vitro, Int. Endod. J. 30 (1997) 279. [50] L.M. Sassone, R.A. Fidel, C.F. Murad, S.R. Fidel, R. Hirata Jr., Antimicrobial activity of sodium hypochlorite and chlorhexidine by two different tests, Aust. Endod. J. 34 (2008) 19. [51] C.E. Radcliffe, L. Potouridou, R. Qureshi, N. Habahbeh, A. Qualtrough, H. Worthington, D.B. Drucker, Antimicrobial activity of varying concentrations of sodium hypochlorite on the endodontic microorganisms Actinomyces israelii, A-naeslundii, Candida albicans and Enterococcus faecalis, Int. Endod. J. 37 (2004) 438. [52] M.E. Vianna, B.P. Gomes, V.B. Berber, A.A. Zaia, C.C. Ferraz, F.J. de Souza-Filho, In vitro evaluation of the antimicrobial activity of chlorhexidine and sodium hypochlorite, Oral. Surg. Oral. Med. Oral. Radiol. Endod. 97 (2004) 79. [53] A. Piasecka, C. Souffreau, K. Vandepitte, L. Vanysacker, R.M. Bilad, T.D. Bie, B. Hellemans, L.D. Meester, X. Yan, P. Declerck, I.F. Vankelecom, Analysis of the

[54]

[55]

[56] [57] [58] [59]

[60] [61]

[62]

[63]

[64]

[65] [66]

[67]

[68]

[69]

[70]

321

microbial community structure in a membrane bioreactor during initial stages of filtration, Biofouling 28 (2012) 225. K. Zhang, H. Choi, D.D. Dionysiou, G.A. Sorial, D.B. Oerther, Identifying pioneer bacterial species responsible for biofouling membrane bioreactors, Environ. Microbiol. 8 (2006) 433. H. Dang, C.R. Lovell, Bacterial primary colonization and early succession on surfaces in marine waters as determined by amplified rRNA gene restriction analysis and sequence analysis of 16S rRNA genes, Appl. Environ. Microbiol. 66 (2000) 467. K.K. Jefferson, What drives bacteria to produce a biofilm?, FEMS Microbiol Lett. 236 (2004) 163. S.S. Madaeni, Y. Mansourpanah, Chemical cleaning of reverse osmosis membranes fouled by whey, Desalination 161 (2004) 13. P.O. Glantz, L. Hansson, Wetting of dentine by some root canal medicaments, Odontol. Rev. 23 (1972) 205. M. Abou-Rass, F.J. Patonai, The effects of decreasing surface tension on the flow of irrigating solutions in narrowroot canals, Oral Surg. Oral Med. Oral Pathol. 53 (1982) 524–526. E. Berutti, R. Marini, A. Angeretti, Penetration ability of different irrigants into dentinal tubules, J. Endod. 23 (1997) 725. G. Gambarini, Shaping and cleaning the root canal system: a scanning electron microscopic evaluation of a new instrumentation and irrigation technique, J. Endod. 25 (1999) 800. S. Stojicic, S. Zivkovic, W. Qian, H. Zhang, M. Haapasalo, Tissue dissolution by sodium hypochlorite: effect of concentration, temperature, agitation, and surfactant, J. Endod. 36 (2010) 1558. J.A. Cameron, The effect of a fluorocarbon surfactant on the surface tension of the endodontic irrigant, sodium hypochlorite: a preliminary report, Aust. Dent. J. 31 (1986) 364–368. R.M. Clarkson, A.J. Moule, H. Podlich, R. Kellaway, R. Macfarlane, J. Rowell, Dissolution of porcine incisor pulps insodium hypochlorite solutions of varying compositions and concentrations, Aust. Dent. J. 51 (2006) 245. M. Kobayashi, K. Tadokoro, H. Tadokoro, Molecular vibrations of three crystal forms of poly(vinylidene fluoride), Macromolecules 8 (1975) 158. N. Her, G. Amy, H.R. Park, M. Song, Characterizing algogenic organic matter (AOM) and evaluating associated NF membrane fouling, Water Res. 38 (2004) 1427. B. Chakrabarty, A.K. Ghoshal, A.K. Purkait, Preparation, characterization and performance studies of polysulfone membranes using PVP as an additive, J. Membr. Sci. 315 (2008) 36. J. Marchese, M. Ponce, N.A. Ochoa, P. Pradanos, L. Palacio, A. Hernandez, Fouling behaviour of polyethersulfone UF membranes made with different PVP, J. Membr. Sci. 211 (2003) 1. S.H. Yoo, J.H. Kim, J.Y. Jho, J. Won, Y.S. Kang, Influence of the addition of PVP on the morphology of asymmetric polyimide phase inversion membranes: effect of PVP molecular weight, J. Membr. Sci. 236 (2004) 203. M.J. Blesa, J.L. Miranda, M.T. Izquierdo, R. Moliner, Study of the curing temperature effect on binders for smokeless briquettes by Fourier transform infrared spectroscopy, Vib. Spectrosc. 31 (2003) 81.