Separation and Purification Technology 146 (2015) 8–14

Contents lists available at ScienceDirect

Separation and Purification Technology

journal homepage: www.elsevier .com/locate /seppur

Submerged membrane – (GAC) adsorption hybrid system in reverseosmosis concentrate treatment

http://dx.doi.org/10.1016/j.seppur.2015.03.0171383-5866/� 2015 Elsevier B.V. All rights reserved.

⇑ Corresponding author. Tel.: +61 2 9514 2641; fax: +61 2 9514 2633.E-mail address: s.vigneswaran@uts.edu.au (S. Vigneswaran).

Sukanyah Shanmuganathan, Tien Vinh Nguyen, Sanghyun Jeong, Jaya Kandasamy,Saravanamuthu Vigneswaran ⇑Faculty of Engineering, University of Technology Sydney (UTS), P.O. Box 123, Broadway, NSW 2007, Australia

a r t i c l e i n f o a b s t r a c t

Article history:Received 28 July 2014Received in revised form 12 March 2015Accepted 13 March 2015Available online 23 March 2015

Keywords:Reverse osmosis concentrateOrganicsGACHybrid system

Wastewater reclamation plants using reverse osmosis as the final polishing treatment produce reverseosmosis concentrate (ROC), which consists of high salinity, nutrients and (recalcitrant) organics. TheROC collected from the water reclamation plant in Sydney was treated with a micro filtration(MF)–GAC hybrid system that removed natural and synthetic organics prior to its discharge into theenvironment. The MF–GAC hybrid system’s performance was studied in terms of trans-membranepressure (TMP) development, and organics removal. These features were measured using liquidchromatography–organic carbon detection (LC–OCD), Fluorescence Excitation-Emission matrix(F-EEM), and Liquid chromatography with tandem mass spectroscopy (LC–MS). Adding GAC into themembrane reactor reduced the TMP by reducing membrane fouling both through mechanical scouringand pre-adsorption of organics. F-EEM confirms the removal of humics-like and fulvic-like compoundsby GAC from ROC. Pharmaceuticals and personal care products (PPCPs) were also removed by theMF–GAC hybrid system.

� 2015 Elsevier B.V. All rights reserved.

1. Introduction

RO is used widely in wastewater reclamation plants as the finaltreatment stage [1–5]. Hundreds of RO based reclamation plantsare in operation in Australia, Asia, Europe, Africa, and Americawhich includes Orange County plant in California, USA (265MLD) and three plants in Singapore (Bedok, Kranji and Changi)[6]. In Sydney alone two major water reclamation plants are usingRO after micro filtration (Homebush bay plant) [7] and ultrafiltra-tion (St. Marys water recycling plant) of biologically treatedwastewater. The resulting product from both plants is then usedfor irrigation and replenishing of Napean River, respectively.Though these RO plants lead to high quality reusable water, theyalso produce large volumes of RO concentrate (ROC) that are richin dissolved organics, pharmaceuticals, persistent organic pollu-tants (PPCPs), pesticides, inorganics, etc. The direct disposal ofROC into water bodies can pose a severe eco-toxicological risk,threaten aquatic organisms and cause serious environmental prob-lems. Consequently, proper treatment, sustainable managementand safe disposal of ROC are mandatory requirements [8,9].

Forward osmosis, membrane distillation, advanced oxidationprocesses [9–11] have been applied to treat ROC; however, the

costs associated with these technologies limit their wider applica-tion. In this context, GAC adsorption is recommended as a simple,cost effective option for removing organics from water and sea-water [12,13]. GAC was found to remove humics, building blocksand LMWs efficiently from water as these components are easilyadsorbed into GAC pores [14]. Coupling of membrane-(powdered)activated carbon hybrid system was studied by Guo et al.,Vigneswaran et al. and Kim et al. [15–17] and reported to be effi-cient in terms of organics removal. Guo et al. [15] confirmed that90% of total organic carbon (TOC) was removed with 5 g/L doseof Powder Activated Carbon (PAC). Vigneswaran et al. [16] notedthe efficiency in removing TOC was 84% with 5 g/L initial doseand followed by 2.5% of daily replacement at filtration flux of12 L/m2 h. Guo et al. [15] revealed that the PAC dose of 1 g/L waseffective in MF–GAC in terms of removing organics.

An area that has not been widely investigated is the use of acti-vated carbon in granular form in membrane-hybrid systems. Kimet al. [17] studied MF–GAC hybrid systems and reported thatemploying GAC with membrane filtration reduced the trans-mem-brane pressure (TMP) development and frequency of chemicalcleaning by half. A smaller development in TMP meant that thetwo chemical cleaning procedures without GAC were reduced toone when GAC was used. Absorbents with larger particles are bet-ter than smaller ones, because they produce better membranescouring outcomes that means less fouling. This agrees with

S. Shanmuganathan et al. / Separation and Purification Technology 146 (2015) 8–14 9

Pradhan et al. [18] who found that the addition of GAC providesmechanical scouring and helps to reduce air scouring. As such,GAC is preferred instead of PAC because it ensures that TMP devel-opment of a membrane-hybrid system remains small over thelong-term while at the same time providing better organicsremoval. In this study, the performance of MF–GAC in treatingROC was assessed in terms of TMP development and removal oforganics. A detailed analysis of organics and pharmaceuticals andpersonal care products (PPCPs) was also done.

2. Materials and methods

2.1. Materials

a. Wastewater

The ROC samples collected from a full scale MF/RO water recla-mation plant located in Sydney, Australia were used as feed water.The plant treats a combination of storm water and biologically trea-ted sewage effluent. This plant produces around 300 kL of ROC/dayand is discharged directly into a sewer system [7]. The detailedcharacteristics of ROC used in this study are summarized in Table 1.

b. Membrane

A hollow fibre membrane made of hydrophilic modified poly-acrylonitrile (PAN) (MANN + HUMMEL ULTRA-FLO PTE LTD,Singapore) was selected for use in the MF–GAC hybrid system.The effective membrane surface area was 0.2 m2 with a nominalpore size of 0.1 lm. The inner and outer diameters of the hollowfibres were 1.1 and 2.1 mm, respectively.

c. Granular Activated Carbon (GAC)

Coal-based premium grade GAC (MDW4050CB) was suppliedby James Cumming & Sons Pty Ltd. and used as an adsorbent.The particle size of GAC used ranged between 425 and 600 lm.The average pore diameter was 30 Å and surface area was1000 m2/g.

2.2. Experimental methods

a. Batch adsorption studies

A batch adsorption equilibrium experiment was conducted todetermine the optimum dose of GAC to treat ROC at equilibriumconditions. Different doses of GAC (0–1.6 g/L) were placed in con-tact with 200 mL of ROC in different beakers and equilibrated for

Table 1Physico-chemical characteristics of ROC.

Characteristics Value

Conductivity (lS/cm) 2350pH 7.5DOC (mg/L) 32–35TDS (mg/L) 2250

Anions (mg/L) Fluoride 3.0–4.0Chloride 400–650Nitrite 1.3–1.5Bromide 1.0–1.5Nitrate 23–26Phosphate 8–9Sulphate 220–250

Cations (mg/l) Na 330–360K 55–63Ca 80–93Mg 65–72

24 h at 25 �C with continuous shaking by a shaker (RatelPlatform Mixer) at 110 rpm. Upon completion of the experiments,samples were filtered through a 0.45 lm filter for DOC analysis.

An adsorption kinetics experiment was conducted to estimatethe adsorption rate of organics from ROC. A fixed dose of GAC(2 g/L) was added to the known quantity (200 mL) of ROC and itunderwent continuous shaking at 110 rpm. Samples were then col-lected at different times (5–420 min) and they were filteredthrough a 0.45 lm filter prior to DOC analysis.

b. MF–GAC hybrid system

The schematic diagram of MF–GAC hybrid system is depicted inFig. 1. A hollow fibre membrane module was immersed in a 3 Linfluent tank containing the ROC to be treated. The membranereactor tank was continuously fed with ROC at a constant rate.GAC was added into the membrane reactor only at the beginningof the experiment at a concentration of 5–20 g/L. The filtration fluxwas maintained at 36 L/m2 h by Peristaltic pump. Both inflow andoutflow were maintained at constant rates. Air flow was providedat a rate of 1.5 m3/m2 membrane area.h to produce scouring on themembrane surface as well as to keep the GAC particles in suspen-sion in the reactor. The residence time of water in the membranereactor was calculated to be 25 min.

Unlike other studies where adsorption is generally carried outas a pre-treatment, here the GAC is directly added into the mem-brane reactor. Consequently, adsorption and membrane separationtake place simultaneously in a single influent tank. The adsorbentsin the suspension reduce membrane fouling by adsorbing potentialorganic foulants before reaching the membrane surface as well asvia mechanical scouring. This agrees with Johir et al. [19] whoreported the addition of GAC helped to reduce fouling by 50%because the GAC particles also provide a scouring effect to themembrane surface.

c. Analysis

The DOC was measured after filtering through a 0.45 lm filterusing a Multi N/C 2000 analyser (Analytik Jena AG). FluorescenceExcitation-Emission matrices (FEEMs) were obtained using aVarian Eclipse Fluorescence Spectrophotometer. The 3D-EEM tech-nique is a rapid, selective and sensitive one that generates informa-tion regarding the fluorescence characteristics of organiccompounds by simultaneously changing the excitation and emis-sion wavelength. The fluorescence in different spectral regions isassociated with various types of functional groups. The fluores-cence signals are basically attributed to protein-like fluorophores,fulvic-like fluorophores and humic-like fluorophores, andcharacterized dissolved organic matter in water using fluorescencespectroscopy [20]. EEMs were recorded using scanning emissionwavelengths from 250 to 500 nm repeatedly at excitation wave-lengths scanned from 220 to 400 nm by 5 nm increments. Theexcitation and emission bandwidths were both set at 5 nm. Thefluorometer was set at a speed of 3000 nm/min, a PMT voltage of700 V and a response time of 2 s. Liquid Chromatography–Organic Carbon Detection (LC–OCD) Model 8 developed by Huberet al. [21] was equipped with a TSK HW 50-(S) column that mea-sured the hydrophilic and hydrophobic fractions of organic materi-als and provided qualitative information on organics’ molecularsize distribution.

Pharmaceuticals and personal care products (PPCPs) wereextracted using solid phase extraction (SPE) and analysed byLiquid Chromatograph with tandem mass spectroscopy. Analyteswere extracted using 5 mL, 500 mg hydrophilic/lipophilic balance(HLB) cartridges (Waters, Milford, MA, USA). These analytes wereseparated using an Agilent (Palo Alto, CA, USA) 1200 series high

Fig. 1. Schematic diagram of the submerged membrane – PuroliteA502PS/GAC hybrid system.

10 S. Shanmuganathan et al. / Separation and Purification Technology 146 (2015) 8–14

performance liquid chromatography (HPLC) system equipped witha 150 � 4.6 mm, 5 lm particle size, Luna C18 (2) column(Phenomenex, Torrance, CA, USA). Mass spectrometry was exe-cuted using an API 4000 triple quadrupole mass spectrometer(Applied Biosystems, Foster City, CA, USA) equipped with aturbo-V ion source employed in both positive and negative elec-tro-spray modes. All calibration curves had a correlation coefficientof 0.99 or better.

3. Results and discussion

3.1. Batch adsorption equilibrium and kinetics

The results of ROC equilibrium studies conducted with GAC atdifferent doses and contact times are given in Fig. 2. As per batchequilibrium, the dose 1.5 g/L achieved up to 80% of DOC removal(Fig. 2a). The kinetics results presented in Fig. 2b show as muchas 80% was removed using a 2 g/L dose. Batch kinetic experimentswith a GAC dose of 2 g/L indicated that DOC removal increasedwith time up to 6–8 h.

3.2. MF–GAC hybrid system

3.2.1. Selection of GAC dosageTwo different doses of GAC (5 g/L and 20 g/L) of membrane

reactor volume were selected for the MF–GAC hybrid system sothat the MF–GAC hybrid system’s ability to treat ROC could beevaluated. These doses of GAC were chosen based on previousstudies and our batch adsorption equilibrium and kinetics results.Vigneswaran et al. [16] commented that increasing the PAC dosefrom 2 g/L to 10 g/L in a MF–GAC in turn increased DOC removalfrom 83.4 % to 87.5% in synthetic wastewater (DOC concentrationswere between 3.8 and 4.2 mg/L). Furthermore this reduced TMPdevelopment from 19.5 to 12.8 kPa. Dialynas et al. [4] concludedthat a GAC dose of 5 g/L removed the most dissolved organic mat-ter (91.3%) from ROC of membrane bioreactor effluent.

Our batch studies indicated approximately 1.5 g/L of GAC wasrequired to reach 80% removal efficiency over a 24 h period. In thisstudy the MF–GAC hybrid system was operated only for a shorttime, i.e. 6 h where 7–8 litres of ROC were treated. This volumeof water required a total of 10.5–12 g of GAC based on a batchadsorption study where the most suitable dose was 1.5 g/L. This

explained why 5–20 g/L of GAC was used in the membrane adsorp-tion hybrid system.

3.2.2. Transmembrane pressure (TMP) developmentThe TMP development of MF–GAC is presented in Fig. 3. The

results show that the TMP of submerged MF membrane systems(without adsorbent addition) increased noticeably from 10.2 to27.4 kPa over 6 h. The addition of 5 g/L of GAC at the beginningof the experiment helped to reduce TMP development by 10 kPa.It should be noted that GAC was added only at the start of theexperiment and no further addition was made. The small develop-ment of TMP observed with 5 g/L of GAC in MF–GAC could be dueto the pre-adsorption of organics onto GAC prior to their contactwith the membrane surface. Also, extra mechanical scouring wasprovided by GAC to the membrane surface due to the circulationof GAC in the reactor. The additional scouring provided by GACparticles on the membrane surface could help prevent a build-upof the cake layer on the membrane surface. This phenomenonagrees with previous studies. Vigneswaran et al. [16] observed thata increase in PAC dose from 2 g/L to 10 g/L reduced TMP develop-ment from 19.5 kPa to 12.8 kPa at a filtration flux 48 L/m2.h.Pradhan et al. [18] observed an 85% reduction in TMP developmentwhen GAC was added to the membrane hybrid system.

No significant difference was observed in the reduction of theTMP when the GAC dose was increased from 5 to 20 g/L. This isbecause the pre-adsorption achieved by 5 g/L GAC dose may havebeen enough to reduce organic deposition/fouling on the mem-brane surface. This observation on TMP at different GAC doseswas made based on the short-term membrane adsorption experi-mental results. A long-term membrane experiment will confirmthat TMP reduction is better at high doses of GAC [19].

It emerged that the GAC particles did not have any adverseeffect on the membrane surface because: firstly, the clean waterflux was the same as that of a virgin membrane; and secondly,the filtered turbidity was reasonably low (less than 0.2 NTU).This finding can also be validated from those in other studies[22,23]. For example, Siembida et al. [23] used granular polypropy-lene particles (with a diameter of 2.5–3.0 mm) in a submergedmembrane bioreactor for more than 600 days. They confirmedvia scanning electron microscopy (SEM) images that no damagewas done to the membrane’s functionality. A long-term submergedmembrane adsorption bioreactor experiment conducted with

Fig. 2. Batch adsorption results (a) batch equilibrium study (b) batch kinetic (GAC dose = 2 g/L).

Fig. 3. The effect of GAC on the development of TMP in MF–GAC hybrid system(DOC of ROC = 21.6–22.9 mg/L; filtration flux 36 L/m2 h; reactor volume 3 L).

S. Shanmuganathan et al. / Separation and Purification Technology 146 (2015) 8–14 11

seawater for more than 120 days also revealed no damage hadoccurred to the membrane surface [24].

3.2.3. Detailed organic removalMF filtration alone did not remove much DOC and as shown in

Fig. 4 it was observed to be less than 10% 4. This is due to the factthat the hollow fibre MF membrane has a larger pore size of0.1 lm. An observed marginal removal of 10% could be due tothe adsorption of organics onto the membrane surface. With theaddition of GAC dose of 5 g/L, the removal of DOC rose 20–60%.

Fig. 4. Removal of organics by MF–GAC hybrid system (DOC of ROC 32 mg/L;filtration flux 36 L/m2 h; GAC dose 5–20 g/L).

The DOC removal significantly increased to 65–90% when a doseof 20 g/L of GAC was added.

The detailed organics removal can be explained by LC–OCD.Size-exclusion chromatography in combination with organic car-bon detection (SEC–OCD) is an established method that can sepa-rate the pool of natural organic matter (NOM) into majorfractions of different sizes and chemical functions. It can also quan-tify these on the basis of organic carbon [21]. The organic fractionsbefore and after treatment with different doses of GAC are pre-sented in Table 2. Based on the LC–OCD results, the ROC contains5.2 mg/L of hydrophobic-DOC and 27.6 mg/L of hydrophilic-DOCwhich comprised 15.9% and 84.1% of total DOC. The latter is com-posed of 44.3% of humics, 21.2% of building blocks, 15.8% of LMWneutrals and 3.0% of biopolymers. LMW acids were non-detectable.As this water was microfiltered before RO, and most of the highmolecular weight compounds would have been removed, biopoly-mers were detected at 1 mg/L. The majority of organics of ROCwere hydrophilic compounds (84.1%) and were humics and build-ing blocks with MW in the range of 350–500 gmol�1. Velten et al.[25] and Cheng et al. [26] found that the adsorption of theseorganic fractions predominantly occur in mesopores (2–50 nmwidth) and large micropores (1–2 nm width). The average pore sizeof GAC used in this study was 3 nm which provides favourable con-dition for the adsorption of these compounds.

Table 2 makes it clear that MF–GAC effectively removed most ofthe organic fractions present in ROC, however, the degree ofremoval was highly dependent on GAC dosages. The removal oforganics nearly doubled when the GAC dose increased from 5 g/Lto 20 g/L. The superior removal of organic fractions was due tomore micro-pore spaces on GAC being available for adsorption.Among the Hydrophilic-DOC, the adsorption of buildingblocks was high, followed by humics and LMWs. According toVelten et al. [25], the ability of organic fractions to be adsorbeddecreases with larger molecule sizes as follows: biopoly-mers < humics < building blocks < LMW organics. Further, theynoticed that effective adsorption is highly dependent on the domi-nance of particular fractions of organics. The reason for the highremoval of building blocks (69.1%) in this present study could beexplained as follows. Since the feed water is biologically treated,LMWs which are biodegradable would have been removed. Thefeed water contains humics and its derivatives (building blocks)in high concentrations compared to other fractions. As such, theremovals of building blocks and humics are observed to be largefollowing GAC treatment. Johir et al. [22] also observed theremoval of more humics when GAC was added the membranebioreactor.

Both hydrophobic-DOC and hydrophilic-DOC compounds arethe major foulants causing membrane fouling and are responsiblefor an increase in TMP. The GAC dose of 5 g/L reduced thehydrophobic-DOC and hydrophilic-DOC to 3 mg/L and 15 mg/L,

Table 2Removal of organic fractions by MF–GAC at different doses of GAC – 5 g/L and 20 g/L. The effluent level organics are in mg/L and Removal efficiencies (%) are given withinparenthesis.

DOC Hydrophobics HydrophilicsBio polymers Humics Building blocks LMW neutrals

Influent (ROC) (mg/L) 32.8 5.2 27.6 1.0 14.5 6.9 5.2Effluent GAC 5 g/L 18 (45.1) 3 (41.9) 15 (45.7) 0.3 (69.2) 9.1 (37.0) 2.1 (69.1) 3.4 (34.6)Effluent GAC 20 g/L 4.5 (86.4) 0.8 (84.1) 3.6 (86.9) 0.1 (91.3) 1.9 (87.2) 0.7 (89.6) 0.9 (81.7)

12 S. Shanmuganathan et al. / Separation and Purification Technology 146 (2015) 8–14

respectively. This corresponds to approximately 42% and 46%reduction which may be good enough to restrict TMP developmentin the short-term (Fig. 3). In long-term experiments, one will needa larger dose of GAC to control the TMP development. It is a well-established adsorption process for the removal of dissolved organicmatter (DOM) from water, due to its strong affinity for removinghydrophobic organic compounds even at a low concentration ofGAC 5 g/L. Therefore, GAC adsorption can be considered as for amethod for reducing membrane fouling.

3.2.4. Fluorescence excitation emission matrixThree characteristic peaks were observed in EEMs including:

humic-like substances (ex/em = 250–285/380–480 nm); fulvic-likesubstances (ex/em = 300–370/400–500 nm); and protein-like sub-stances (ex/em = 270–280/300–350 nm). This analysis is semi-quantified since the average value of fluorescence intensities inthe range of ex/em of each peak is employed in comparison tothe relative abundance of organics. The FEEM obtained for theuntreated ROC and effluent samples with 5 g/L GAC dose and20 g/L GAC dose are presented in Fig. 5.

Humic substances and building blocks, which are organic com-pounds that can be detected by LC–OCD with UV detector, weredominant DOC compounds in ROC. By contrast, biopolymers con-centration was low in ROC. Similarly, a significant reduction ofhumic-like and fulvic-like substances was observed through theFEEM analysis. Fluorescence intensity of protein-like substanceswas weak in ROC.

Fluorescence intensity of protein-like substances (53.1 a.u.) wasless than humic-like (94.3 a.u.) and fulvic-like (232.7 a.u.) sub-stances in ROC. GAC reduced protein-like organic substances to27.5 a.u. and 9.7 a.u. with 5 g/L and 20 g/L, respectively. Thisrevealed that GAC also reduced the problem of biofouling duringthe membrane process. The presence of this peak represents the bio-fouling potential since the protein-like substances detected in FEEMcontain an indole functional group. It is an essential amino acid asdemonstrated by its effect on the growth of micro-organisms [27].

As observed in LC–OCD analysis, FEEM results revealed thathumic-like and fulvic-like substances were dominant organic com-pounds in ROC. This indicated that the hydrophobic fractions

Fig. 5. FEEM intensity of untreated ROC and after GAC treatment of 5 g/L and 20 g/L.

(humic and fulvic) were rich in biologically treated effluent whilenatural waters mainly consists of hydrophobic fractions. Notably,fulvic-like peaks were much stronger than humic-like peaks indi-cating that fulvic-like substances presented a larger portion ofhumic-like substances of ROC. FEEM analysis made the detailedstudy of different fractions in humic substances possible. Fulvic-like substances consist of high molecular weight aromatic humicsubstances [28]. ROC contained the highest amount of aromaticfulvic-like materials, which can damage membranes as they aresevere organic foulants. However, GAC was effective in reducingfulvic-like substances from 232.7 a.u. to 90.9 a.u. and 15.8 a.u. with5 g/L and 20 g/L, respectively.

3.2.5. Pharmaceuticals and personal care productsMany researchers have discussed the occurrence of PPCPs in the

effluents of wastewater treatment plants (WWTPs) [29]. The elim-ination of these PPCPs through the use of conventional treatmentprocesses is not effective [30]. GAC has been found to be effectivein the removal of PPCPs [31]. The results of this study relating tothe removal of PPCPs by GAC are presented in Table 3. In this study,the removal efficiency of PPCPs by GAC dose of 20 g/L was found tobe more significant (81–100%) than that with a GAC dose of 5 g/L(65–100%). The results were comparable with previous results ofHernández-Leala et al. [32] and Yang et al. [33] in which theremoval of triclosan, diclofenac, trimethoprim, carbamazepine,and caffeine by GAC alone were 95%, 100%, 90%, 75%, and 45%respectively. The removals of PPCPs by GAC can be explained basedon hydrophobicity and charge of the PPCPs molecules and thesecan be expressed via Log Kow (octonol–water partition coefficient)and pKa (acid dissociation constant) values. The PPCPs havinghigher Log Kow values are known to have a more hydrophobic nat-ure and these are significantly adsorbed by GAC. In addition, thePPCPs having higher pKa values (>7) are highly adsorbed by GACsince they are positively charged. As the degree of the removal ofPPCPs is influenced by the combination of solute Log Kow andpKa values, the Log Kow values were corrected and the correctedvalue is expressed as Log D values. Here D is the distribution coef-ficient of the PPCPs in n-octonol to water at equilibrium.

The Log D values were calculated based on the equations (Eqs.(1) and (2)) given by de Ridder et al. [34].

Acidsðnegatively chargedÞ : log D

¼ log Kow� logð1þ 10ðpH�pKaÞÞ ð1Þ

Basesðpositively chargedÞ : log D

¼ log Kow� logð1þ 10ðpKa�pHÞÞ ð2Þ

PPCPs having higher Log D values are more thoroughly adsorbed byGAC than PPCPs having lower Log D values. However, no definedrelationship was observed between the removals of PPCPs vs. LogD values in this study. This could be due to the high dose of GACcontaining abundant available binding sites which may haveadsorbed all the PPCPs due to van der Waals forces.

Table 3Influent, effluent levels of PPCPs and subsequent removal efficiency by MF–GAC hybrid system (The doses of GAC were 5 g/L and 20 g/L; Flux = 36 L/m2 h).

Log D (pH 7) Log Kowa (pH 7) pKa ROC* GAC* 5 g GAC* 20 g Removal (%) GAC 5 g Removal (%) GAC 20 g

Amtriptyline 3.0 4.9 9.4a 44.5 <5 <5 >89 >89Atenolol �1.9 0.2 9.6b 466 114 34 76 93Caffeine �3.0 �0.1 10.4c 1410 97 36 93 97Carbamazepine �3.1 2.5 2d 2240 386 39.7 83 98DEET �3.3 2.2 2d 67.8 12.8 6.14 81 91Diclofenac 1.2 4.5 4.2e 337 117 12.4 65 96Fluoxetine 1.5 4.1 10.1e 46.7 6.25 <5 87 >89Gemfibrozil 2.0 4.8 4.7d 344 79.5 8.52 77 98Ketoprofen 0.1 3.1 4.45a 377 34.6 <5 91 >99Naproxen �1.7 3.2 4.2a 443 46.3 4.99 90 99Primidone N/A 0.9 N/A 26 4.75 <5 82 >81Simazine �3.7 2.2 1.62a 80.1 13.6 <5 83 >94Sulfamethoxazole �4.5 0.9 2.1d 144 10.7 6.76 93 95Triclocarbon �0.3 4.9 12.7f 162 18.9 15 88 91Triclosan 4.2 4.8 7.9e 211 47.2 19.3 78 91Trimethoprim 0.4 0.9 7.12e 974 149 13.3 85 99Verapamil 2.4 3.8 8.92a 82.9 5.68 <5 93 >94

* The PPCPs concentrations are in ng/L.a U.S. National library of medicine (http://chem.sis.nlm.nih.gov/chemidplus/rn/52-53-9).b Hapeshi et al. [35].c Yang et al. [33].d Westerhoff et al. [30].e Serrano et al. [36].f Loftsson [37].

S. Shanmuganathan et al. / Separation and Purification Technology 146 (2015) 8–14 13

4. Conclusions

The addition of GAC into the MF–GAC hybrid system reducedTMP development by 10 kPa which is due to the mechanical scour-ing effect provided by GAC as well as by the pre-adsorption oforganics before reaching the membrane surface. The addition ofGAC 5 g/L removed DOC by 20–60% throughout the experiment’s6 h duration whereas hydrophobic and hydrophilic portionsremovals were 42% and 46%, respectively. The increase in GAC doseto 20 g/L resulted in up to 85% of DOC being removed. FEEM resultsshow the removal of humic-like substances and fulvic-like sub-stances to be at significant levels.

The MF–GAC hybrid system’s removal of PPCPs proved to be veryeffective. In fact, 65–100% removal was observed with a GAC dose of5 g/L of membrane reactor volume, which increased to 81–100%with a GAC dose of 20 g/L. The dose of 5 g/L of membrane reactorvolume corresponds to a GAC usage of 2 g/L of treated water.Moreover, the removal of PPCPs was not affected when naturalorganics were present, as there were abundant GAC binding sitesfor incoming PPCPs and DOC. PPCPs are smaller molecules andhydrophobic in nature and can find easy access to GAC binding sites.

Finally, MF–GAC is an effective system for treating ROC toremove dissolved natural and persisting organics prior to dischargeinto the environment in a safe manner. Alternatively the effluentcan be recirculated and mixed with other feed to a RO process tomaximize water reuse.

Acknowledgement

This study was funded by CRC-CARE (project number:4.1.15.12/13).

References

[1] K. Chon, J. Cho, H.K. Shon, K. Chon, Advanced characterization of organicfoulants of ultrafiltration and reverse osmosis from water reclamation,Desalination 301 (2012) 59–66.

[2] M. Umar, F.A. Roddick, L. Fan, O. Autin, B. Jefferson, Treatment of municipalwastewater reverse osmosis concentrate using UVC-LED/H2O2 with andwithout coagulation pre-treatment, Chem. Eng. J. 260 (2015) 649–656.

[3] K. Liu, F.A. Roddick, L. Fan, Impact of salinity and pH on the UVC/H2O2

treatment of reverse osmosis concentrate produced from municipalwastewater reclamation, Water Res. 46 (10) (2012) 3229–3239.

[4] E. Dialynas, D. Mantzavinos, E. Diamadopoulos, Advanced treatment of thereverse osmosis concentrate produced during reclamation of municipalwastewater, Water Res. 42 (18) (2008) 4603–4608.

[5] D. Dolar, M. Gros, S. Rodriguez-Mozaz, J. Moreno, J. Comas, I. Rodriguez-Roda,D. Barceló, Removal of emerging contaminants from municipal wastewaterwith an integrated membrane system, MBR-RO, J. Hazard. Mater. 239–240(2012) 64–69.

[6] Global water intelligence, Municipal Water Reuse Markets (2010): MediaAnalytical publishers Ltd., Oxford, United Kingdom. ISBN: 978-0-9547705-8-7.

[7] H. Chapman, WRAMS, sustainable water recycling, Desalination 188 (2006)105–111.

[8] Y.X. Sun, Y. Gao, H.Y. Hu, F. Tang, Z. Yang, Characterization and biotoxicityassessment of dissolved organic matter in RO concentrate from a municipalwastewater reclamation reverse osmosis system, Chemosphere 117 (2014)545–551.

[9] C. Kazner, S. Jamil, S. Phuntsho, H.K. Shon, T. Wintgens, S. Vigneswaran,Forward osmosis for the treatment of reverse osmosis concentrate from waterreclamation: process performance and fouling control, Water Sci. Technol. 69(12) (2014) 2431–2437.

[10] C.R. Martinetti, A.E. Childress, T.Y. Cath, High recovery of concentrated RObrines using forward osmosis and membrane distillation, J. Membr. Sci. 331(1–2) (2009) 31–39.

[11] T. Zhou, T.T. Lim, S.S. Chin, A.G. Fane, Treatment of organics in reverse osmosisconcentrate from a municipal wastewater reclamation plant: feasibility test ofadvanced oxidation processes with/without pretreatment, Chem. Eng. J. 166(3) (2011) 932–939.

[12] K.C. Graf, D.A. Cornwell, T.H. Boyer, Removal of dissolved organic carbon fromsurface water by anion exchange and adsorption: bench-scale testing tosimulate a two-stage countercurrent process, Sep. Purif. Technol. 122 (2014)523–532.

[13] T.V. Nguyen, S. Jeong, T.T.N. Pham, J. Kandasamy, S. Vigneswaran, Effect ofgranular activated carbon filter on the subsequent flocculation in seawatertreatment, Desalination 354 (2014) 9–16.

[14] S. Ciputra, A. Antony, R. Phillips, D. Richardson, G. Leslie, Comparison oftreatment options for removal of recalcitrant dissolved organic matter frompaper mill effluent, Chemosphere 81 (2010) 86–91.

[15] W.S. Guo, S. Vigneswaran, H.H. Ngo, H. Chapman, Experimental investigationof adsorption–flocculation–microfiltration hybrid system in wastewater reuse,J. Membr. Sci. 242 (2004) 27–35.

[16] S. Vigneswaran, W.S. Guo, P. Smith, H.H. Ngo, Submerged membraneadsorption hybrid system (SMAHS): process control and optimization ofoperating parameters, Desalination 202 (2007) 392–399.

[17] K.Y. Kim, H.S. Kim, J. Kim, J.W. Nam, J.M. Kim, S. Son, A hybrid microfiltration-granular activated carbon system for water purification and wastewaterreclamation/reuse, Desalination 243 (2009) 132–144.

[18] M. Pradhan, S. Vigneswaran, J. Kandasamy, R.B. Aim, Combined effect of air andmechanical scouring of membranes for fouling reduction in submergedmembrane reactor, Desalination 288 (2012) 58–65.

14 S. Shanmuganathan et al. / Separation and Purification Technology 146 (2015) 8–14

[19] M.A.H. Johir, R. Aryal, S. Vigneswaran, J. Kandasamy, A. Grasmick, Influence ofsupporting media in suspension on membrane fouling reduction in submergedmembrane bioreactor (SMBR), J. Membr. Sci. 374 (2011) 121–128.

[20] S. Jeong, S.J. Kim, C.M. Kim, S. Vigneswaran, T.V. Nguyen, H.K. Shon, J.Kandasamy, I.S. Kim, A detailed organic matter characterization of pretreatedseawater using low pressure microfiltration hybrid systems, J. Membr. Sci. 428(2013) 290–300.

[21] S.A. Huber, A. Balz, M. Abert, W. Pronk, Characterisation of aquatic humic andnon-humic matter with size-exclusion chromatography–organic carbondetection–organic nitrogen detection (LC–OCD–OND), Water Res. 45 (2011)879–885.

[22] M.A. Johir, S. Shanmuganathan, S. Vigneswaran, J. Kandasamy, Performance ofsubmerged membrane bioreactor (SMBR) with and without the addition of thedifferent particle sizes of GAC as suspended medium, Bioresour. Technol. 141(2013) 13–18.

[23] B. Siembida, P. Cornel, S. Krause, B. Zimmermann, Effect of mechanicalcleaning with granular material on the permeability of submerged membranesin the MBR process, Water Res. 44 (14) (2010) 4037–4046.

[24] S. Jeong, S.A. Rice, S. Vigneswaran, Long-term effect on membrane fouling in anew membrane bioreactor as a pretreatment to seawater desalination.Bioresource Technology. Special Issue: CESE 2013 & Special Issue: ICABB2013, 165 (2014) 60–68.

[25] S. Velten, D.R.U. Knappe, J. Traber, H.P. Kaiser, U. von Gunten, M. Boller, S.B.Meylan, Characterization of natural organic matter adsorption in granularactivated carbon adsorbers, Water Res. 45 (2011) 3951–3959.

[26] W. Cheng, S.A. Dastgheib, T. Karanfil, Adsorption of dissolved natural organicmatter (DOM) by virgin and modified granular activated carbons, Water Res.39 (11) (2005) 2281–2290.

[27] S. Jeong, L.H. Kim, S.J. Kim, T.V. Nguyen, S. Vigneswaran, I.S. Kim, Biofoulingpotential reductions using a membrane hybrid system as a pre-treatment toseawater reverse osmosis, Appl. Biochem. Biotechnol. 167 (6) (2012) 1716–1727.

[28] M.M.D. Sierra, M. Giovanela, E. Parlanti, E.J. Soriano-Sierra, Fluorescencefingerprint of fulvic and humic acids from varied origins as viewed by single-

scan and excitation/emission matrix techniques, Chemosphere 58 (6) (2005)715–733.

[29] Y.L. Lin, J.H. Chiou, C.H. Lee, Effect of silica fouling on the removal ofpharmaceuticals and personal care products by nanofiltration and reverseosmosis membranes, J. Hazard. Mater. 277 (2014) 102–109.

[30] P. Westerhoff, Y. Yoon, S. Snyder, E. Wert, Fate of endocrine-disruptor,pharmaceutical, and personal care product chemicals during simulateddrinking water treatment processes, Environ. Sci. Technol. 39 (2005) 6649–6663.

[31] S.A. Snyder, S. Adham, A.M. Redding, F.S. Cannon, J. DeCarolis, J. Oppenheimer,E.C. Wert, Y. Yoon, Role of membranes and activated carbon in the removal ofendocrine disruptors and pharmaceuticals, Desalination 202 (2007)156–181.

[32] L. Hernández-Leala, H. Temmink, G. Zeeman, C.J.N. Buisman, Removal ofmicropollutants from aerobically treated grey water via ozone and activatedcarbon, Water Res. 45 (9) (2011) 2887–2896.

[33] X. Yang, R.C. Flowers, H.S. Weinberg, P.C. Singer, Occurrence and removal ofpharmaceuticals and personal care products (PPCPs) in an advancedwastewater reclamation plant, Water Res. 45 (2011) 5218–5228.

[34] D.J. de Ridder, M. McConville, A.R.D. Verliefde, L.T.J. van der Aa, S.G.J. Heijman,J.Q.J.C. Verberk, L.C. Rietveld, J.C. van Dijk, Development of a predictive modelto determine micropollutant removal using granular activated carbon, Drink.Water Eng. Sci. Discuss. 2 (2) (2009) 189–204.

[35] E. Hapeshi, A. Achilleos, M.I. Vasquez, C. Michael, N.P. Xekoukoulotakis, D.Mantzavinos, D. Kassinos, Drugs degrading photocatalytically: kinetics andmechanisms of ofloxacin and atenolol removal on titania suspensions, WaterRes. 44 (2010) 1737–1746.

[36] D. Serrano, S. Suárez, J.M. Lema, F. Omil, Removal of persistent pharmaceuticalmicropollutants from sewage by addition of PAC in a sequential membranebioreactor, Water Res. 45 (2011) 5323–5333.

[37] T. Loftsson, Í.B. Össurardóttir, T. Thorsteinsson, M. Duan, M. Másson,Cyclodextrin solubilization of the antibacterial agents triclosan andtriclocarban: effect of ionization and polymers, J. Inclusion phenomenamacrocyclic chem. 52 (1–2) (2005) 109–117.