w a t e r r e s e a r c h 5 9 ( 2 0 1 4 ) 2 7 1 e2 8 2

Available online at www.sciencedirect.com

ScienceDirect journal homepage: www.elsevier.com/locate/watres

How different is the composition of the fouling layer of wastewater reuse and seawater desalination RO membranes? Muhammad Tariq Khan a, Markus Busch b, Veronica Garcia Molina b, Abdul-Hamid Emwas c, Cyril Aubry a, Jean-Philippe Croue a,* a

Water Desalination and Reuse Research Center, King Abdullah University of Science and Technology, Thuwal 23955-6900, Saudi Arabia b Dow Water and Process Solutions, Spain c NMR Core Lab, King Abdullah University of Science and Technology, Thuwal 23955-6900, Saudi Arabia

article info

abstract

Article history:

To study the effect of water quality and operating parameters on membrane fouling, a

Received 6 November 2013

comparative analysis of wastewater (WW) and seawater (SW) fouled reverse osmosis (RO)

Received in revised form

membranes was conducted. Membranes were harvested from SWRO and WWRO pilot

6 April 2014

plants located in Vilaseca (East Spain), both using ultrafiltration as pretreatment. The

Accepted 9 April 2014

SWRO unit was fed with Mediterranean seawater and the WWRO unit was operated using

Available online 18 April 2014

secondary effluent collected from the municipal wastewater treatment plant. Lead and terminal SWRO and WWRO modules were autopsied after five months and three months of

Keywords:

operation, respectively. Ultrastructural, chemical, and microbiological analyses of the

Seawater desalination

fouling layers were performed. Results showed that the WWRO train had mainly bio/

Wastewater reuse

organic fouling at the lead position element and inorganic fouling at terminal position

Reverse osmosis

element, whereas SWRO train had bio/organic fouling at both end position elements. In the

Membrane fouling

case of WWRO membranes, Betaproteobacteria was the major colonizing species; while Ca, S, and P were the major present inorganic elements. The microbial population of SWRO membranes was mainly represented by Alpha and Gammaproteobacteria. Ca, Fe, and S were the main identified inorganic elements of the fouling layer of SWRO membranes. These results confirmed that the RO fouling layer composition is strongly impacted by the source water quality. ª 2014 Elsevier Ltd. All rights reserved.

1.

Introduction

To overcome the growing water scarcity problem in the planet, several water management approaches (i.e., water

conservation or improvement of distribution systems) have been implemented and different water treatment processes have been introduced (Elimelech and Phillip, 2011). Due to its versatility, Reverse Osmosis (RO) process is one of the available solutions for the purification of water that has quickly

* Corresponding author. Tel.: þ966 808 2984. E-mail addresses: [email protected] (M.T. Khan), [email protected] (J.-P. Croue). http://dx.doi.org/10.1016/j.watres.2014.04.020 0043-1354/ª 2014 Elsevier Ltd. All rights reserved.

272

w a t e r r e s e a r c h 5 9 ( 2 0 1 4 ) 2 7 1 e2 8 2

expanded worldwide. This semi-permeable membrane technology provides high removal of dissolved solids, organics, colloidal matter, and microbes from feed water. Because of this attractive contaminant rejection feature and efficiency profile, RO process is been widely used for seawater desalination (Ettouney and Wilf, 2009) and it is also finding increased application in the domain of municipal and industrial wastewater treatment (Bartels et al., 2005). Despite the wide applicability of RO, membrane fouling is still an important shortcoming limiting the efficiency of this process. RO membrane fouling is driven by the quality of inlet water and by the downstream pretreatment train (Khan et al., 2013a). Today, many seawater reverse osmosis (SWRO) desalination and advanced wastewater reverse osmosis (WWRO) treatment plants are applying microfiltration (MF) or ultrafiltration (UF) prior to RO to guarantee high removal of suspended and colloidal matter. Regardless of the tight pores of MF and UF membranes, a fraction of foulants/foulant precursors passes through these pretreatment barriers and consequently induces RO membrane fouling (Mo et al., 2008; Rapenne et al., 2007). Due to this limited fouling inhibition role of these pretreatment technologies, the source water composition remains as a major factor that governs membrane fouling phenomenon. To understand and overcome membrane fouling, many studies have focused on investigating the fouling cake layer occurring during WWRO reclamation (Ang et al., 2011; Zhu et al., 2012) and SWRO desalination processes (Flemming and Schaule, 1988). Nevertheless, due to dissimilar source water, pretreatment processes, and membrane characteristics, the composition of the fouling layers characterized in previous studies have significantly differed from each other. Therefore, the fundamental mechanisms governing fouling phenomena could not be precisely elucidated and a comparative analysis between these systems could not be established. In the present work, polyamide thin film composite RO membrane elements from SWRO and WWRO treatment trains positioned after similar ultrafiltration pretreatment process were autopsied. Although operating parameters were different to some extent, the main dissimilarity between two systems was the use of two different source waters (i.e., wastewater and seawater). The autopsy of fouled membrane modules is a commonly used technique that offers an important insight on the characteristics of foulants and the structure of fouling layers (Darton et al., 2004). Two end position modules (i.e., lead and terminal end modules) from both treatment trains were selected for characterization. The fouling layers were characterized by Fourier Transform Infrared (FTIR) spectrometry, Pyrolysis Gas Chromatography coupled with Mass Spectrometry (Pyrolysis/GCeMS), Inductively Coupled Plasma Optical Emission Spectrometry (ICP-OES), elemental analysis, and microbial phylogenetic examination procedures. The aim of this study was to investigate the effect of the drastically different microbial and chemical quality of the source water on the structural characteristics and nature of the fouling layer formed on RO membranes. The differences in microbial and chemical composition of the fouling layers of two sets of membranes were rigorously analyzed and discussed in detail.

2.

Materials and methods

2.1.

Pilot scale RO units

Membrane samples were obtained from two pilot plants located at DOW Global Water Technology Development Center in Spain. Both pilot plants were similar in principle but were operated with two different source waters, a treated urban wastewater and the Mediterranean seawater collected from an open intake.

2.2.

SWRO pilot unit

The seawater project was conducted on the Mediterranean Sea, and the feed water was captured in the port of Tarragona. The port of Tarragona is relatively shallow with about 5 m water depth; its water quality is impacted by ship traffic and occasionally (during rain events) by the river that ends in the port. The source water was captured, shock-chlorinated (every 4 h with 2e5 ppm NaOCl), and then transported via a 10 km pipe through a 250 mm screen size ring filter (Arkal, Israel), to the feed tank of the pilot unit at the DOW Water & Process Solutions Global Water Technology Center in Tarragona. Source water was characterized by: pH of 8.1, conductivity of 57.4 ms/cm, Total Organic Carbon (TOC) of 2 mg/L, turbidity of 2 NTU, Total Suspended Solids (TSS) of 10 mg/L, and a Total Dissolved Solids (TDS) of 40,000 mg/L. Occasionally, peaks of 3 mg/L, 15 NTU, and 50 mg/L were observed for TOC, turbidity, and TSS, respectively. The temperature ranged from 13 to 30  C during operation. The pilot unit consisted of a feed tank supplying two ultrafiltration lines equipped with DOW SFP-2660 modules (30 nm nominal pore diameter, hydrophilized PVDF outside-in hollow fibers, pressurized, vertical). The SWRO line fed by the UF filtrate was positioned after a 5 mm cartridge filter. The RO line used 6 elements (4-inch diameter FILMTEC SW30XLE4040) in series configuration operated at 60 bar feed water pressure, 17 L/m2h average permeate flux, and at 45% recovery rate. Moreover, feed water flow rate for RO system was 1.5 m3/ h with cross flow velocity values of 0.12 and 0.06 m/s at the lead and terminal position elements, respectively (Fig. 1A). Modules of 1st (SM1, lead element) and 6th position (SM6, terminal element) of the RO line 1 were autopsied after approximately 5 months of operation.

2.3.

WWRO pilot unit

The wastewater UFRO project was carried out at the Vilaseca wastewater treatment plant, which treats wastewater from the cities of Vilaseca, La Pineda, and Salou. This urban wastewater (low industrial discharge) showed strong variations in quality due to touristic influence. The pretreatment included: coarse screening (1 mm), sand filtration followed by primary sedimentation, aerobic sludge treatment (designed for partial nitrification), and secondary sedimentation. The treated wastewater collected after secondary sedimentation was fed to the UFRO pilot unit (Fig. 1B) with the following physicochemical characteristics: pH of 7.0, conductivity of

w a t e r r e s e a r c h 5 9 ( 2 0 1 4 ) 2 7 1 e2 8 2

273

Fig. 1 e Schematic representation of seawater (A) and wastewater (B) treatment pilot units.

1.5 ms/cm, TOC of 6 mg/L, turbidity of 4 NTU with occasional increases to 100 NTU or higher (occurring approximately once a week), average Chemical Oxygen Demand (COD) of 15 mg/L, and average Biological Oxygen Demand (BOD) of 7 mg/L. TDS highly varied and averaged 600e1500 mg/L, while the temperature ranged from 23 to 31  C. The wastewater pilot unit (Fig. 1B) was similar to the seawater pilot unit (Fig. 1A), except for the following modifications:

This unit was operated at an average feed pressure of 7 bar, an average flux of 21 L/m2h, and a 70% recovery rate. Moreover, feed water flow rate for RO system was 1.3 m3/h with cross flow velocity values of 0.10 and 0.03 m/s at the lead and terminal position elements, respectively. Modules of 1st (WM1, lead element of 1st stage) and 6th position (WM6, terminal element of 1st stage) of the RO line 1 were autopsied after approximately 3 months of operation.

2.4.  A TEQUATIC Plus (DOW) fine particle filtration followed by a 50 mm screening filter (Amiad, Israel) were used upfront of the UF feed tank.  Two DOW Ultrafiltration SFP-2880 modules (30 nm nominal pore diameter, hydrophilized PVDF outside-in hollow fibers, pressurized, vertical) were used.  The RO train used only one high pressure pump and consisted of two stages in order to reach high recovery levels. The first stage had two lines in parallel; each line with two vessels in series, and each vessel with three RO elements in 4040 format (40 inches in length and 4 inches in diameter). The second stage also consisted of two parallel lines, each line with two vessels in series. In this case, each vessel hosts three RO elements with smaller diameter, i.e., 2.5 inches. In total, the system consisted of 12 RO elements in series. The use of smaller elements in the second stage was justified by the reduced feed flow circulating at this stage (part of the original feed flow left the system in the first stage as permeate). This setup was used to simulate a typical RO configuration with concentrate staging for higher recoveries.

Analytical methods

Membranes were subjected to ultrastructural, chemical, and microbiological analyses in triplicates (except for pyrosequencing, ICP, and LOI analyses). Pyrosequencing analysis was performed on specimens extracted from the middle region of the membrane sheets where the fouling layer was more uniform. LOI and ICP analyses were performed on freeze-dried and well-mixed samples of foulant material exhibiting uniform composition.

2.4.1.

Fouling load analysis

Foulant material was collected from the fouled membrane sheets of measured surface area (approximately 8 m2) by physical scrapping. Dried foulant material was obtained through lyophilisation process. Recovered foulant material was weighed and mass per unit surface area of the fouled membrane (fouling load) was calculated.

2.4.2.

Ultrastructural analyses

In order to explore the structural and compositional details of fouling layers at a microscopic level, freeze drying scanning

274

w a t e r r e s e a r c h 5 9 ( 2 0 1 4 ) 2 7 1 e2 8 2

electron microscopic (SEM) and energy dispersive X-ray (EDX) spectroscopic analyses were performed following the protocol guidelines by Khan et al. (2013a).

characterization (i.e., CHN, FTIR, and analyses).

2.4.4. 2.4.3.

Chemical analyses

Chemical characterization of fouling layer was carried out by the following analytical procedures.

2.4.3.1. Loss on ignition (LOI) test. Lyophilized foulant material samples were dehydrated overnight at 105  C. Accurately weighed quantities (1030 mg) were introduced in fused quartz crucibles and subjected to ignition at 550  C for 4 h in a muffle furnace. % Loss on ignition was calculated as follows (Heiri et al., 2001):   %LOI ¼ WeightðDryÞ  Weightð@550  CÞ =WeightðDryÞ  100 2.4.3.2. Elemental analysis. Accurately weighed amount of samples (1520 mg) were digested with 1 ml of trace metal analytical grade HNO3 (Trace SELECT Ultra, SigmaeAldrich) and then diluted in 20 ml of ultrapure water. Clear solution obtained after digestion and dilution were analyzed by ICPOES for inorganic element concentrations (Mondamert et al., 2009). Flash 2000 e Thermo Scientific CHNS/O Analyzer was used for CHNS analysis of pre-dried and well-powdered foulant material following the USEPA 440.0 analytical method (Zimmermann et al., 1997). 2.4.3.3. FT-IR spectroscopy. A Spectrum 100 FTIR (PerkinElmer, USA) was used for IR Spectrometry. Potassium Bromide (KBr) pellet method involving the mixing and grinding of approximately 200e300 mg of dried sample with 100 mg of KBr salt, was used. Transparent and thin pellets (of approximately 0.5 mm in thickness) were formed by submitting the mixture to a press of 10 tons/inch2. Each of the recorded spectra was the result of 20 scans performed at a resolution of 1 cm1 with a scanning range from 4000 to 400 cm1. 2.4.3.4. Pyrolysis/GCeMS. Pyrolysis/GCeMS analysis of dried foulant material was conducted according to the method described by Bruchet et al. (1990). 2.4.3.5. Solid state 13C nuclear magnetic resonance (NMR) spectroscopy. 13C NMR analysis of solid foulant material was performed following the method adapted from Leenheer (2009) and described by Khan et al. (2013a).

2.4.3.6. Removal of free salts from foulant samples. Foulant materials were subjected to dialysis to remove free salts remaining on the membrane surface from the feed water. Foulant sample (1001000 mg) was immersed in Milli-Q water-containing dialysis bags (1K MWCO; Spectra/Por 7 by Spectrum labs, USA). The dialysis bags were immersed in a glass beaker containing 5 L of Milli-Q water, and then mildly stirred with a magnetic stirrer at room temperature for 24 h in the dark. Dried foulant material was obtained through lyophilisation. In addition to Milli-Q dialysis, the foulant material of WM6 was also subjected to acid (1.0 N HCl) dialysis to remove inorganic components interfering with the organic

13

C NMR spectroscopy

Microbiological analyses

Adenosine-50 -triphosphate (ATP) contents of fouled membranes were measured following the method described by Khan et al. (2013b). The analysis of microbial diversity in the fouling layer of membranes was conducted by the pyrosequencing of the 16S ribosomal RNA gene. DNA extraction of membrane samples was performed according to Manes et al. (2011). Pyrosequencing of the total genomic DNA extracted from membrane samples was performed in the Research and Testing Laboratory (RTL, Texas, USA) using a Roche 454 FLX Genome Sequencer, as described by Dowd et al. (2008). Bacterial 16S rDNA was sequenced using a 27F primer (50 AGAGTTTGATCMTGGCTCAG30 ) and a 519R reverse primer (50 GTNTTACNGCGGCKGCTG) targeting the V2eV3 variable regions (Dowd et al., 2008). Sequence dataset was preprocessed at the RTL facilities in order to reduce the noise and sequencing artifacts including searching for chimeras and clustering, as described by Dowd et al. (2008). Bacterial diversity of the remaining sequence data set (average >3000 sequences per sample) was queried against a database of high quality 16S sequences derived from NCBI (current version) and then statistically analyzed with the PAST software (Hammer et al., 2001). The beta diversity of the sequence dataset was analyzed using Bray Curtis algorithm in a dendrogram and multi dimension scaling (MDS), taking into account operational taxonomic unit (OTU) abundances and distribution.

3.

Results and discussion

3.1.

Morphology of fouling layers

Ultrastructural analysis showed that the foulant accumulated on the WWRO membranes covered a larger surface area compared to what could be observed on SWRO membranes (Supplementary Fig. 1). The surface morphology of the foulant layers was also different for the two types of membranes. Microbes observed either lying on the membrane surface or embedded in the extracellular polymeric substances (EPS) on SWRO and WWRO lead elements (i.e., more pronounced on WM1 than SM1) were a morphological evidence of biofouling prevailing in these systems.

3.2.

Fouling load and nature of foulant material

Despite the shorter operation time, i.e., 3 months, the fouling loads of WWRO modules (i.e., 1.25  0.10 g/m2 for WM1 and 0.25  0.05 g/m2 for WM6) were higher than those of SWRO modules, which were operated for 5 months (i.e., 0.77  0.09 g/ m2 for SM1 and 0.01  0.004 g/m2 for SM6). This finding was in accordance with microscopic observation of membrane surface morphology. Similar fouling trend was observed across both RO trains, i.e., fouling load was relatively significant at the 1st position module and it remarkably decreased on the 6th position (terminal) module (Fig. 2A).

w a t e r r e s e a r c h 5 9 ( 2 0 1 4 ) 2 7 1 e2 8 2

275

Fig. 2 e Fouling magnitude and nature of foulant material. A) Analysis of fouling load and active biomass of membranes. Error bars represent standard deviation. B) Estimation of organic and inorganic fractions of foulant materials isolated from membranes.

ATP, an energy-rich biomolecule present in all active microorganisms, is an indicator of active biomass (Holm-Hansen and Booth, 1966). The applicability of ATP contents as a parameter for the assessment of active biomass present in the RO membrane fouling layer has been previously reported (Khan et al., 2013a,b). Measurements of ATP contents revealed that the lead modules have more active biomass than the terminal modules (Fig. 2A). However, the ATP content of WM1 (i.e., 208  3.28 pg/cm2) was more than twice of SM1 membrane (i.e., 88  0.54 pg/cm2). This result indicated the presence of a relatively more developed/mature biofilm on WM1. A larger abundance of organic nutrients in the wastewater (poorly removed by UF (Rapenne et al., 2007)) might have been utilized by the bacteria passing through the UF barrier, resulting in increased biofilm formation rate on the WWRO lead position element. On the other side, despite the significantly lower fouling load, the ATP content measured per unit surface area of SM6 (i.e., 32  0.30 pg/cm2) was higher than that of WM6 (i.e., 24  0.27 pg/cm2). LOI test, discussed in detail in the following paragraph, revealed that the foulant material of WM6 was mainly inorganic in nature. This result clearly indicated that the higher fouling load of WM6 was mainly attributed to the accumulation of inorganic foulants rather than the development of biofilm and/or accumulation of organic foulants on its surface. Due to this reason, ATP contents present per unit surface area of WM6 were lower than that of SM6. The analysis of organic/inorganic fractions by LOI measurements revealed that the fouling material of SWRO membranes was mainly composed of organic matter, i.e., 89.17% and 75% for SM1 and SM6, respectively (Fig. 2B). The increase in inorganic fouling at the terminal position of SWRO train can be attributed to increased salt contents in the feed stream

resulting in enhanced concentration polarization (CP) phenomenon. Inorganic elements might have precipitated due to limited solubility of their salts (i.e., scaling) and/or attached/ bound to negatively-charged organic constituents of the fouling layer (i.e., complexation). Due to low recovery rates, i.e., 50%, it is usual to mainly find bio/organic foulants instead of inorganic foulants on SWRO membranes, even on terminal backend module (Al-Amoudi and Farooque, 2002). The LOI results of WWRO membrane foulants led to the same observation (i.e., decrease in relative abundance of the organic fraction from lead to terminal position module), however, the decrease was more prominent. In accordance with results reported in the literature (Gerringer, 2009; Li et al., 2011), the terminal module of WWRO train (WM6) was mainly fouled by inorganic foulants (i.e., 88.89%), whereas the lead position module (i.e., WM1) was mainly fouled by organics (i.e., 96.58%). As proposed by Hoek et al. (2008), the flux and recovery rates might be increased at terminal modules due to organic fouling occurring in lead modules. These conditions may favor inorganic fouling phenomenon at the backend module. Higher inorganic fouling at the terminal position of the WWRO element might also be attributed to higher recovery rates of WWRO system (i.e., 70% for WWRO and 45% for SWRO unit). The salinity of seawater is typically higher than wastewater. However, in wastewater the concentration of salt species forming precipitates (e.g., calcium and phosphate) might be comparable or even higher than that of seawater, which mainly contains NaCl. The higher recovery rate of wastewater RO system could have remarkably increased in the concentration of these relatively low solubility salts species, which eventually promoted their precipitation on the membrane surface.

276

w a t e r r e s e a r c h 5 9 ( 2 0 1 4 ) 2 7 1 e2 8 2

Fig. 3 e Elemental analysis of foulant material and calculated C/N and H/C atomic ratios.

3.3.

Elemental analysis of foulant material

ICP data (Fig. 3) showed calcium (Ca; 6.02 and 51.1 mg/g in SM1 and SM6 foulant material, respectively), phosphorus (P; 8.4 and 5.76 mg/g in SM1 and SM6 foulant material, respectively), and sulfur (S; 20.5 and 12.5 mg/g in SM1 and SM6 foulant material, respectively), as predominant elements on SWRO foulants. As discussed previously, the organic fraction in SM1 foulant material was higher than that of SM6 foulant material. Additionally, the relatively high concentration of P in the fouling layer of SM1 is most likely organic in nature. The presence of organic P could be present in the form of microbial cell membrane phospholipids and ATP contents which are signpost of the presence of active biofilm. Higher concentration of inorganic/organic P has been reported by Jarvie et al. (2002) in a relatively young biofilm. Moreover, significantly higher abundance of S could result from sodium metabisulfite (SMBS) dosed as dechlorinating agent in the pretreatment line and/or from the presence of sulfur containing biomolecules such as cysteine and methionine amino acids. Interestingly, the concentrations of Al, Ca, and Fe (i.e., cations) significantly increased, whereas the content of P and S (i.e., anions) decreased from lead to terminal position modules. In this situation either there was scaling/precipitation of compounds of these three cations with other anions (e.g., in the form of their carbonates, chlorides) or there was a complexation of these elements, in the form of multivalent cations, with negatively charged or electron lone pair carrying sites present in organic foulant materials. Since the foulant material of SM6 is mainly organic in nature, the latter phenomenon is more probable to have occurred. Moreover, as the foulants of both SWRO modules were mainly organic in nature, the remarkably larger occurrence of cations, especially Ca, in SM6 foulant material might correspond to the presence of organics with

relatively higher density of complexing sites. Humic-like material, known for its high affinity towards polyamide RO membranes (Monruedee et al., 2012; Tang et al., 2007), could play a significant role in the complexation of cations due to its oxygen and nitrogen containing functional groups (e.g., carboxylic, eCOOH, phenolic/enolic, eOH, quinones/hydroquinones, eRNH) (Ephraim and Allard, 1997). Non-humic/ biopolymers contents of the fouling layer (i.e., polysaccharides, proteins) might also contribute significantly to the incorporation of cations. For WM1 foulant, ICP analysis (Fig. 3) confirmed the low abundance of inorganic elements, with Ca, Fe, and P present at concentration of 14.53 mg/g, 8.48 mg/g, and 18.45 mg/g, respectively. Higher concentrations of inorganic elements were identified in WM6 foulant with Ca, Fe, P, and S (116.3 mg/ g, 9.2 mg/g, 40.41 mg/g, and 33.48 mg/g, respectively) as the predominant species. Because WM6 foulant was mainly inorganic in nature (x90%), these elevated concentrations of Ca, P, and S may correspond to precipitates of calcium, phosphate, and sulfate. Indication of membrane scaling by these compounds was also supported by the observed Ca, P, and S molar ratios (i.e., 2.9:1.2:1) approaching the theoretical molar ratios (i.e., one mole of Ca for each mole of P and S). Larger S content might also originate from the oxidation of hydrogen sulfide (i.e., typically present in wastewater) by Fe3þ, consequently producing elemental sulfur and iron sulfide (FeS) (Lahav et al., 2004). EDX elemental analysis showed very weak signals of typical inorganic elements (i.e., Al, Ca, Mg, Fe, and P) on both SWRO membranes (Supplementary Fig. 2). The high S signals, especially in the case of SWRO membranes with relatively low fouling layer thickness, were most probably originating from the polysulfone layer of the membrane structure penetrated by EDX beam. Similarly, signals of C and O might also be

w a t e r r e s e a r c h 5 9 ( 2 0 1 4 ) 2 7 1 e2 8 2

attributed to organic membrane structure. In the case of WWRO membranes, significantly higher signals for inorganic elements, i.e., Al, Ca, Fe, and P, were observed. At many locations on WM6, intensities of signals of P and Ca were even higher than that of C, O, and S (signals coming from membrane material). This result clearly indicated the presence of compounds of Ca and P in higher abundance in the fouling layer of WM6. On overall basis results of EDX and ICP-OES elemental analyses performed on both sets of membranes confirmed that the ICP analysis results were mutually coherent. C/N and H/C ratios are good indicators of the presence/ absence of nitrogenous and aromatic structures, respectively. For instance, algal biomass enriched in proteins typically has C/N atomic ratios between 4 and 10, whereas organic matter of vascular terrestrial plants, proportionally more enriched in polysaccharides, is characterized by C/N atomic ratios of 20 or higher (Meyers, 1994). Hedges et al., 1992 reported C/N and H/C atomic ratios ranging from 35 to 46 and 1.21 to 1.57, respectively, in marine dissolved humic substances extracted using XAD-8 resin. Marine plankton has an average C/N atomic ratio near 6 (Redfield et al., 1963). In our study, low C/N atomic ratios were observed for both SWRO (i.e., 6.74 and 7.91 for SM1 and SM6, respectively) and WWRO membranes foulant materials (i.e., 5.84, 8.6, and 7.18 for WM1, WM6(before acid dialysis) and WM6(after acid dialysis), respectively), and these values slightly increased from the first to the last position module (Fig. 3). The values obtained for WM6 foulant after acid dialysis (removing the interfering inorganic species) represent a more accurate CHN distribution of the organic fraction. Based on C/ N ratio values, the order of abundance of proteinaceous/ nitrogenous substances in the studied foulant materials is the following: WM1 > SM1 > WM6ðafter acid dialysisÞ > SM6 According to Bhosle (2004), during the marine biofilm formation processes, starting from conditioning film formation followed by attachment and growth of cells, the C/N atomic ratio decrease progressively. Lower C/N ratio for the foulant materials of 1st position modules (i.e., SM1 and WM1), might be associated to the biofilm development phase with relatively intense microbial activity leading to higher number of cells per unit area of membrane. This hypothesis of higher microbial growth at lead position elements is also supported by higher ATP contents/unit surface area of these membranes. Interestingly, C/N ratios of the four foulant samples follow the same order as their ATP content values. A very slight decrease in the H/C atomic ratio is observed from lead to terminal position for both sets of modules. This decreasing trend of H/C and increasing trend of C/N atomic ratio across the RO train refer to the accumulation of unsaturated compounds having less nitrogen content, e.g., humic-like or other non-nitrogenous aromatic structures on the back end membrane surface.

3.4.

Biopolymer analysis

Pyrolysis GCeMS analysis, a technique used to determine the relative abundance of biopolymers in organic matter, was conducted on the four foulant samples (Fig. 4). Foulants of lead

277

position elements of both sets of membranes exhibited almost similar biopolymer distribution with minor differences e.g., higher relative abundance of DNA (furanmethanol peak) and aminosugars (acetamide peak) in WM1 foulant. The higher proportion of these two microbial cell’s structural units along with higher fouling load value is an evidence of more developed and mature biofilm on WM1 as compared to that on SM1. As expected, the biopolymer signature of the biogenic matter, i.e., DNA, aminosugars and lipids, was found in lower relative abundance at both terminal positions (i.e., 1.8% and 1% DNA, 3.4% and 1.7% aminosugars, 18.3% and non-detectable lipids for SM6 and WM6 foulants, respectively) than at lead position modules (i.e., 3.4% and 4.2% DNA, 3.8% and 4.6% aminosugars, 26.9% and 21.9% lipids for SM1 and WM1 foulants, respectively). These results agree with the decreasing microbial activity/population from lead to terminal position membranes, both for WWRO and SWRO systems. Nevertheless, WM6 incorporated relatively higher proportion of proteins/nitrogenous molecules and low/negligible proportion of lipids as compared to SM6. In the foulant samples of the terminal position modules, the relative abundance of aromatic aminoacids, i.e., tryptophan (indole and methyl indole pyrofragments), tyrosine (phenol and p-cresol pyrofragments), and phenylalanine (toluene and styrene pyrofragments), was low. Phenolics i.e., phenol and p-cresol peaks, are yielded by both polyhydroxyaromatics (PHAs) and tyrosine. Based on the relative intensities of these two peaks, their predominant source can be identified, i.e., higher phenol signal suggest the predominance of PHAs, whereas higher p-cresol suggests predominant presence of tyrosine (Bruchet et al., 1990; Christy et al., 1998). Increase in phenol/p-cresol ratios from lead to terminal modules (0.43 and 0.94 for SM1 and SM6, respectively; 0.34 and 1.09 for WM1 and WM6, respectively) is another evidence of the lower presence of tyrosine containing proteins and higher presence of PHAs at terminal modules. The increase of the relative abundance of styrene while the relative abundance of toluene decreased especially in SM6 foulant (i.e., major pyrofragment of phenylalanine (Patterson et al., 1973)), supports this finding. Benzonitrile is a characteristic pyrofragment of nitrogen containing aromatic compounds, e.g., nitrophthalic acid (Patterson et al., 1973) and aromatic amino acids (Bandurski and Nagy, 1976). It could also be generated during the pyrolysis of humic material (Schulten and Schnitzer, 1997). Similarly, benzoic acid (i.e., known pyrofragment of humic-like material (Gillam and Wilson, 1985; Wilson et al., 1983)) might have more than one origin. Pyrolysis of phthalates-based plasticizer, present in numerous plastic materials, also produce the peak of benzoic acid (Kusch, 2012). A probable pathway of benzoic acid formation from phthalates could be the release of free phthalic acid which is decarboxylated to benzoic acid. A remarkable increase in abundance of benzonitrile and exclusive/prominent signal of benzoic acid in pyrochromatogram of SM6 foulant are signposts of the presence of either humic-like materials and/or phthalates (possibly leached from plastics tubing/piping used in the plant). This finding partly supports our previous hypothesis suggesting a more significant presence of humic-like material on this membrane.

278

w a t e r r e s e a r c h 5 9 ( 2 0 1 4 ) 2 7 1 e2 8 2

Fig. 4 e Analysis of biopolymers: Left) pyrochromatograms of foulant material (isolated from membranes), showing different peaks of pyrofragments of organic substances. Based on different origin, pyrofragment peaks are marked with labels of different colors. Right) pie charts showing distribution of biopolymers/organics, calculated by adding up the percent area of characteristic pyrofragment peaks of a biopolymer class. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

In the IR spectra of the foulants (Fig. 5), the presence of proteins was indicated by amideA (NeH stretching vibration, 3294 cm1), amideB (Fermi resonance of amideII overtone with NeH stretching vibration, 3090 cm1), amideI (mainly C]O stretching vibration, 1650 cm1), amideII (out of phase combination of NeH bending and the CeN stretching vibrations, 1555 cm1), and amideIII (in phase combination of NeH bending and the CeN stretching vibrations, 1410 cm1) signals. A major band located near 1037 cm1 (stretching vibration of CeO) indicated the presence of sugars. In addition to the amideI, amideII, and sugar bands, the presence of Nacetyl aminosugars, degradation product of bacterial cell wall peptidoglycans, was also revealed by a peak at 1384 cm1 (symmetric bending vibration of CH3 of N-acetyl group) (Coates, 2000; Leenheer, 2009). Bands at 2960 cm1, 2925 cm1, and 2850 cm1 assigned to asymmetric stretching vibration of aliphatic CH3, CH2, and symmetric stretching vibration of CH2 groups, respectively (Igisu et al., 2009). The COOH band at

1730 cm1 and CH2 band at 1454 cm1 along with the band at 29602850 cm1 are mainly attributable to lipids (Leenheer, 2009). The band at 1240 cm1 was attributable to the stretching vibration of phosphodiester backbone of nucleic acids (Naumann et al., 1996). The relative intensities of all these discussed bands, originating from biogenic matter, were somewhat similar for all the foulant samples except for WM6 foulant (especially before acid dialysis). Moreover, due to high inorganic contents, this sample exhibited very weak biopolymer bands. An intense and broad band at 1020 cm1 (stretching of PeO, stretching vibration of SeO) along with a prominent band at 560 cm1 (bending vibration of PeO) (Barnes et al., 2011; Holt et al., 1996), revealed the predominant presence of inorganic phosphates and sulfates, therefore corroborating the ICP results. A significantly intense band at 1425 cm1 might be attributed to carbonates (CO2 3 ). The WM6 foulant IR spectrum was in accordance with the elemental composition, i.e., membrane scaling by phosphate and

w a t e r r e s e a r c h 5 9 ( 2 0 1 4 ) 2 7 1 e2 8 2

279

Fig. 5 e FTIR spectroscopic analysis of foulant materials. Labeled functional groups were identified based on characteristic IR bands and probable class/type of compounds.

sulfates. Interestingly, after acid dialysis, the IR profile of WM6 became similar to the other IR profiles. The more pronounced shoulder at 1730 cm1, on left side of amideI band in the spectrum of SM6 foulant, is attributable to the larger presence of carboxyl group containing substances in this sample, i.e., humic-like materials, lipids, and proteins. By keeping in view the pyrolysis GCeMS data of this sample showing higher abundance of aromatic substances (i.e. PHAs), humic-like materials and/or phthalates could be considered as the most probable origin of this carboxyl group signal. A decrease in the intensity of the sharp band at 1384 cm1 (bending vibration of CH3 of N-acetyl group of aminosugars) confirmed that the biofouling was less intense at back end modules. Solid state 13C NMR spectroscopy was performed to further explore the nature of foulant materials. Due to low organic contents, 13C NMR spectrum of WM6 foulant before acid dialysis exhibited very week signals. After removal of interfering inorganic metals through acid dialysis, a well resolved spectrum was recorded (Fig. 6). 13C NMR spectra are discussed in the light of data published by Leenheer, 2009. The bands near 20 ppm (branched methyl carbon), 38 ppm (carbon atoms in alicyclic ring structures), and 43 ppm (quaternary carbon) were indicators of organic matter derived from primary algal and/or secondary bacterial production. Bands at 30 ppm and 172 ppm were derived, or at least partly derived, from alkyl chain methylene and carboxyl group carbons, respectively, of fatty acids. Proteins were indicated by NeC peak at 53 ppm. Carbohydrates were indicated by secondary alcohol (O-Alkyl) and anomeric carbon (Di-O-Alkyl) bands at 71 ppm and 101 ppm, respectively. A band at 20 ppm was attributable to methyl-C of N-acetyl group of aminosugars. Characteristic bands of biogenic material (i.e.,

proteins, aminosugars) exhibited lower relative intensities in the sample of terminal position modules. In accordance with elemental analysis and pyrolysis GC/MS results, the lower intensity of NeC band in the 13C NMR spectra of SM6 and WM6 foulant samples denies the possibility of a significant role of aromatic aminoacids in the intense aromatic carbon band. The spectrum of the SM6 foulant closely resemble the spectrum published for humic materials extracted from sediments collected from coastal area of Mediterranean Sea (De la Rosa et al., 2011). However, as discussed above, other possible origin of the aromatic carbon signal could be synthetic

Fig. 6 e Solid state 13C NMR spectroscopic analysis of foulant materials. Y-axis (not shown) is signal intensity with arbitrary units.

280

w a t e r r e s e a r c h 5 9 ( 2 0 1 4 ) 2 7 1 e2 8 2

Fig. 7 e Classification and distribution of 16S rRNA gene sequences into most abundant phyla and classes.

aromatic compounds, such as phthalates, leached from plastic material used in plant setup and accumulated in the fouling layer of terminal modules. The band at 63 ppm, exclusively found in terminal position membrane foulants (more pronounced in SM6 foulant), refers to primary alcohol (CH2O). Similarly, the bands at 120140 ppm, more intense in the terminal position membrane foulants, belong to aromatic carbons that could be present in esterified phthalic acid (used as plasticizer). The absence or minor presence of such substance in the foulant of lead position membranes might be attributed to microbial degradation or to the reduction/ masking of their relative load by higher abundance of other types of foulants, i.e., biopolymers.

3.5. Comparison of SWRO and WWRO membrane microbial communities Next generation sequencing of the 16S ribosomal RNA gene was used for the characterization of microbial diversity and the taxonomic affiliation of sequences obtained from RO membrane samples. The results revealed that microbial communities of seawater membrane samples were very different from that of wastewater membrane samples.

Despite the fact that the Proteobacteria phylum was predominant in both SWRO and WWRO samples (55% for SM1, 93% for SM6, 83% for WM1, and 72% for WM6); their class distribution and abundance were found to be quite dissimilar (Fig. 7). There was a substantial difference between microbial community of SM1 and SM6 fouling layer. While SM1 was dominated by Gammaproteobacteria (32% of the sequences), SM6 was largely represented by Alphaproteobacteria (55%) and Betaproteobacteria (26%). Both WM1 and WM6 were mainly represented by Betaproteobacteria, i.e., 65% and 43%, respectively. Ivnitsky et al., 2007, reported the dominance of Betaproteobacteria in the biofilm of NF membrane fed by tertiary effluent wastewater generated in a continuous flow membrane bioreactor. Predominance of Alphaproteobacteria and Gammaproteobacteria in the fouling layer of biofouled SWRO membranes, harvested from full scale plants, has been reported by Khan et al. (2013a), and Zhang et al. (2011). The calculated diversity index Shannon_H revealed that for both seawater and wastewater RO membranes the microbial community diversity was higher at position 6 (Fig. 7). The clustering of the sequence dataset at OTU level revealed that SWRO samples grouped below 20% similarity and that WWRO samples were more similar to each other clustering around 60% similarity (Fig. 8A). The multidimensional scaling (MDS)

Fig. 8 e Multivariate analysis of the sequence data set. A) Clustering of sequences using BrayeCurtis algorithm, paired group method. B) Non-metric multidimensional scaling of sequences visualizing the similarities among the samples.

w a t e r r e s e a r c h 5 9 ( 2 0 1 4 ) 2 7 1 e2 8 2

of the same dataset corroborate the clustering results, SWRO samples were apart from each other while WWRO samples are closely located (Fig. 8B).

4.

Conclusions

Despite similar pretreatment setup (UF before RO), clear differences between SWRO and WWRO foulants were observed which can be mainly related to inherent differences in water quality and RO operational parameters (e.g., trans-membrane pressure, recovery rate, etc.). Although the operating parameters of the two RO systems were also different, the fouling profile of two systems was most likely impacted, to a major extent, by the different source water quality. This considerable difference in source water quality was expected to result in quite dissimilar fouling scenarios. However, some similarities were also found in two examined fouling profiles. Based on the data discussed, the ollowing points could be concluded: 1. Lead elements of both RO systems faced biofouling; additionally, biopolymer distribution in the foulant material collected from these membranes was quite similar. 2. Although, WWRO membranes were operated for a shorter period of time, they were more heavily fouled than SWRO membranes. Particularly, the lead element of WWRO plant showed a more developed biofilm. 3. Similar to the lead position element, the terminal position element of the SWRO unit was also mainly fouled by bio/ organic matter. The fouling layer was predominantly composed of aromatic moieties, most likely of humic-like structures. Some inorganic elements, i.e., Ca and S (probably coming from bisulfite dosing) were also significantly present. As a divalent cation, Ca might be involved in complex formation mechanisms with negatively charged organics. 4. The terminal position element of the WWRO unit was mainly fouled by inorganic foulants, specifically CaH(PO4), CaSO4, and FeS. 5. The bacterial community of WWRO membranes was different from that of SWRO membranes. WWRO fouled membranes were dominated by Betaproteobacteria class (Proteobacteria phylum) while SWRO fouled membranes were dominated by Gammaproteobacteria and Alphaproteobacteria (Proteobacteria phylum) class affiliated sequences. Since the implemented pretreatment (i.e., UF) is known to reduce the microbial cell count of the water but not nutrients, relatively enhanced biofilm formation in case of WWRO unit (especially on lead position membrane) is most likely due to larger abundance of nutrients (i.e., C, N, and P) in its feed water than in that of SWRO unit. This result suggested that the removal of nutrients, especially organics representing AOC, is a key factor in controlling biofouling. The potential of AOC removal by different pretreatment strategies (e.g., biological filtration, slow sand filtration, or membrane filtration coupled with in-line pre-coagulation) have been reported in literature. However, the optimization of existing anti-biofouling pretreatment strategies or the development of new procedures targeting the removal of nutrients in the water prior to RO membrane processes is highly required.

281

Acknowledgements We thank Carmem Lara de O. Manes who was hired as consultant to interpret the pyrosequencing data. We also thank KAUST Analytical Core Lab for ICP-OES and CHNS analysis. Additionally, authors thankfully acknowledge the support received from KAUST WDRC lab staff. All the funds for this work were provided by King Abdullah University of Science and Technology (KAUST).

Appendix A. Supplementary data Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.watres.2014.04.020.

references

Al-Amoudi, Ahmed S., Farooque, M., 2002. Autopsy of membranes. In: 3rd Acquired Experience Symposium. AlJubail, Saudi Arabia. Ang, W.S., Tiraferri, A., Chen, K.L., Elimelech, M., 2011. Fouling and cleaning of ro membranes fouled by mixtures of organic foulants simulating wastewater effluent. J. Membr. Sci. 376 (1e2), 196e206. Bandurski, E.L., Nagy, B., 1976. The polymer-like organic material in the orgueil meteorite. Geochim. Cosmochim. Acta 40 (11), 1397e1406. Barnes, D.H., Jugdaosingh, R., Kiamil, S., Best, S.M., 2011. Shelf life and chemical stability of calcium phosphate coatings applied to poly carbonate urethane substrates. J. Biotechnol. Biomaterials 1 (6). Bartels, C.R., Wilf, M., Andes, K., Iong, J., 2005. Design considerations for wastewater treatment by reverse osmosis. Water Sci. Technol. 51, 473e482. Bhosle, N.B., 2004. In: Ramaiah, N. (Ed.), . Marine Microbiology: Facets and Opportunities. National Institute of Oceanography, Goa, pp. 159e168. Bruchet, A., Rousseau, C., Mallevialle, J., 1990. Pyrolysis-gc-ms for investigating high-molecular-weight THM precursors and other refractory organics. J. Am. Water Works Assoc. 82 (9), 66e74. Christy, A.A., Bruchet, A., Rybacki, D., 1998. Characterization of natural organic matter by pyrolysis/gc-ms. Environ. Int. 25 (2/ 3), 181e189. Coates, J., 2000. In: Meyers, R.A. (Ed.), Encyclopedia of Analytical Chemistry. John Wiley & Sons, Ltd, Chichester, UK, pp. 10815e10837. Darton, T., Annunziata, U., del Vigo Pisano, F., Gallego, S., 2004. Membrane autopsy helps to provide solutions to operational problems. Desalination 167, 239e245. De la Rosa, J.M., Gonza´lez-Pe´rez, J.A., Gonza´lez-Vila, F.J., Knicker, H., Arau´jo, M.F., 2011. Molecular composition of sedimentary humic acids from south west iberian peninsula: a multi-proxy approach. Org. Geochem 42 (7), 791e802. Dowd, S., Callaway, T., Wolcott, R., Sun, Y., McKeehan, T., Hagevoort, R., Edrington, T., 2008a. Evaluation of the bacterial diversity in the feces of cattle using 16s rdna bacterial tagencoded flx amplicon pyrosequencing (btefap). BMC Microbiol. 8 (1), 125. Dowd, S., Sun, Y., Secor, P., Rhoads, D., Wolcott, B., James, G., Wolcott, R., 2008b. Survey of bacterial diversity in chronic

282

w a t e r r e s e a r c h 5 9 ( 2 0 1 4 ) 2 7 1 e2 8 2

wounds using pyrosequencing, dgge, and full ribosome shotgun sequencing. BMC Microbiol. 8 (1), 43. Elimelech, M., Phillip, W.A., 2011. The future of seawater desalination: Energy, technology, and the environment. Science 333 (6043), 712e717. Ephraim, J.H., Allard, B., 1997. Metal ion binding by humic substances. In: Grenthe, I., Puigdomenech, I. (Eds.), Modelling in Aquatic Chemistry. OECD Publications, Paris, pp. 207e244. Ettouney, H., Wilf, M., 2009. In: Micale, G., Rizzuti, L., Cipollina, A. (Eds.), Seawater Desalination. Springer, Berlin Heidelberg, pp. 77e107. Flemming, H.C., Schaule, G., 1988. Biofouling on membranes e a microbiological approach. Desalination 70 (1e3), 95e119. Gerringer, F.W., 2009. Relationships between Natural Organic Matter Characteristics, Reverse Osmosis Pretreatment and Membrane Performance. University of California, Los Angeles. Gillam, A.H., Wilson, M.A., 1985. Pyrolysis-GC-MS and NMR studies of dissolved seawater humic substances and isolates of a marine diatom. Org. Geochem 8 (1), 15e25. Hammer, Ø., Harper, D., Paul, D., 2001. Past: paleontological statistics software package for education and data analysis. Palaeontol. Electron. Available from: http://folk.uio.no/ ohammer/past/. Hedges, J.I., Hatcher, P.G., Ertel, J.R., Meyers-Schulte, K.J., 1992. A comparison of dissolved humic substances from seawater with Amazon River counterparts by 13c-nmr spectrometry. Geochim. Cosmochim. Acta 56 (4), 1753e1757. Heiri, O., Andre´, F., Lotter, A.F., Lemcke, G., 2001. Loss on ignition as a method for estimating organic and carbonate content in sediments: reproducibility and comparability of results. J. Paleolimnol. 25, 101e110. Hoek, E.M.V., Allred, J., Knoell, T., Jeong, B.-H., 2008. Modeling the effects of fouling on full-scale reverse osmosis processes. J. Membr. Sci. 314 (1e2), 33e49. Holm-Hansen, O., Booth, C.R., 1966. The measurement of adenosine triphosphate in the ocean and its ecological significance. Limnol. Oceanogr. 11, 510e519. Holt, C., Wahlgren, N.M., Drakenberg, T., 1996. Ability of a betacasein phosphopeptide to modulate the precipitation of calcium phosphate by forming amorphous dicalcium phosphate nanoclusters. Biochem. J. 314 (3), 1035e1039. Igisu, M., Ueno, Y., Shimojima, M., Nakashima, S., Awramik, S.M., Ohta, H., Maruyama, S., 2009. Micro-ftir spectroscopic signatures of bacterial lipids in proterozoic microfossils. Precambrian Res. 173 (1e4), 19e26. Ivnitsky, H., Katz, I., Minz, D., Volvovic, G., Shimoni, E., Kesselman, E., Semiat, R., Dosoretz, C.G., 2007. Bacterial community composition and structure of biofilms developing on nanofiltration membranes applied to wastewater treatment. Water Res. 41 (17), 3924e3935. Jarvie, H.P., Neal, C., Warwick, A., White, J., Neal, M., Wickham, H.D., Hill, L.K., Andrews, M.C., 2002. Phosphorus uptake into algal biofilms in a lowland chalk river. Sci. Total Environ. 282e283 (0), 353e373. Khan, M.T., Manes, C.-L.d.O., Aubry, C., Croue, J.P., 2013a. Source water quality shaping different fouling scenarios in a full-scale desalination plant at the Red Sea. Water Res. 47 (2), 558e568. Khan, M.T., Manes, C.-L. de O., Aubry, C., Gutierrez, L., Croue, J.P., 2013b. Kinetic study of seawater reverse osmosis membrane fouling. Environ. Sci. Technol. 47 (19), 10884e10894. Kusch, P., 2012. Pyrolysis-gas chromatography/mass spectrometry of polymeric materials. In: Mohd, M.A. (Ed.), Advanced Gas Chromatography e Progress in Agricultural, Biomedical and Industrial Applications. InTech, Rijeka, Croatia, pp. 333e362. Available from: http://cdn.intechopen. com/pdfs/32828/InTech-Pyrolysis_gas_chromatography_ mass_spectrometry_of_polymeric_materials.pdf.

Lahav, O., Ritvo, G., Slijper, I., Hearne, G., Cochva, M., 2004. The potential of using iron-oxide-rich soils for minimizing the detrimental effects of h2s in freshwater aquaculture systems. Aquaculture 238 (1), 263e281. Leenheer, J.A., 2009. Systematic approaches to comprehensive analyses of natural organic matter. Ann. Environ. Sci. 3, 1e130. Li, X., Yan, D., An, G., Jing, D., Li, J., 2011. Fouling and cleaning of reverse osmosis membranes during municipal tap water treatment on a pilot-scale plant. J. Water Sustain. 1 (1), 139e151. Manes, C.L.d.O., West, N., Rapenne, S., Lebaron, P., 2011. Dynamic bacterial communities on reverse-osmosis membranes in a full-scale desalination plant. Biofouling 27 (1), 47e58. Meyers, P.A., 1994. Preservation of elemental and isotopic source identification of sedimentary organic matter. Chem. Geol. 114 (3e4), 289e302. Mo, H., Tay, K.G., Ng, H.Y., 2008. Fouling of reverse osmosis membrane by protein (BSA): effects of ph, calcium, magnesium, ionic strength and temperature. J. Membr. Sci. 315 (1e2), 28e35. Mondamert, L., Labanowskia, J., Berjeauda, L.M., Rapenne, S., Croue´, J.P., 2009. Autopsy of ro desalination membrane: Part 2. Chemical characterisation of the foulant. Desalin. Water Treat. 9 (9), 73e81. Monruedee, M., Sarp, S., Lee, Y.G., Kim, J.H., 2012. Time-series image analysis for investigating SWRO fouling mechanism. Desalin. Water Treat. 43 (1e3), 212e220. Naumann, D., Schultz, C., Helm, D., 1996. In: Mantsch, H.H., Chapman, D. (Eds.), Infrared Spectroscopy of Biomolecules. John Wiley & Sons, New York, pp. 279e310. Patterson, J.M., Haidar, N.F., Papadopoulos, E.P., Smith, W.T., 1973a. Pyrolysis of phenylalanine, 3,6-dibenzyl-2,5piperazinedione, and phenethylamine. J. Org. Chem. 38 (4), 663e666. Patterson, J.M., Shiue, C.-Y., Smith, W.T., 1973b. Benzonitrile formation in the pyrolysis of aromatic nitrogen compounds. J. Org. Chem. 38 (14), 2447e2450. Rapenne, S., Port, C.L., Roddy, S.J., Croue´, J.-P., 2007. Pilot plant research for Sydney’s drinking water program project, pretreatment prior to RO for seawater desalination: Sydney pilotscale study. In: IDA World Congress-Maspalomas. Redfield, A.C., Ketchum, B.H., Richards, F.A., 1963. In: Hill, M.N. (Ed.), The Sea. Wiley, New York, pp. 26e77. Schulten, H.R., Schnitzer, M., 1997. The chemistry of soil organic nitrogen: a review. Biol. Fertil.Soils 26 (1), 1e15. Tang, C.Y., Kwon, Y.-N., Leckie, J.O., 2007. Fouling of reverse osmosis and nanofiltration membranes by humic acideeffects of solution composition and hydrodynamic conditions. J. Membr. Sci. 290 (1e2), 86e94. Wilson, M.A., Philp, R.P., Gillam, A.H., Gilbert, T.D., Tate, K.R., 1983. Comparison of the structures of humic substances from aquatic and terrestrial sources by pyrolysis gas chromatography-mass spectrometry. Geochim. Cosmochim. Acta 47 (3), 497e502. Zhang, M., Jiang, S., Tanuwidjaja, D., Voutchkov, N., Hoek, E.M.V., Cai, B., 2011. Composition and variability of biofouling organisms in seawater reverse osmosis desalination plants. Appl. Environ. Microbiol. 77 (13), 4390e4398. Zhu, H., Wen, X., Huang, X., 2012. Characterization of membrane fouling in a microfiltration ceramic membrane system treating secondary effluent. Desalination 284, 324e331. Zimmermann, C.F., Keefe, C.W., Bashe, J., 1997. US-EPA Method 440.0: Determination of Carbon and Nitrogen in Sediments and Particulates of Estuarine/Coastal Waters using Elemental Analysis (Accessed from. http://nla.gov.au/nla.cat-vn3844273. on 23 DEC 2011).