Desalination Technology for Management Water Issue in California. 15 pages

Chemical Engineering Journal 262 (2015) 946–957

Contents lists available at ScienceDirect

Chemical Engineering Journal

j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / c e j

Interaction of humic substances on fouling in membrane distillation for seawater desalination

http://dx.doi.org/10.1016/j.cej.2014.10.060 1385-8947/� 2014 Elsevier B.V. All rights reserved.

Abbreviations: AOC, assimilable organic carbon; BB, building blocks; DCMD, direct contact membrane distillation; DI, deionized; DOC, dissolved organic carb energy dispersive spectrometer; FE-SEM, field emission scanning electron microscope; HA, humic acid; LC-OCD, liquid chromatography-organic carbon detector; L molecular weight; MD, membrane distillation; MQ, MiliQ; MW, molecular weight; NF, nanofiltration; NOM, natural organic matter; PTFE, polytetrafluoroethyl reverse osmosis; SEC, size exclusion chromatography; SW, seawater; TDS, total dissolved solids; UV, ultraviolet; VCF, volume concentration factor. ⇑ Corresponding author. Tel.: +61 2 9514 2641; fax: +61 2 9514 2633.

E-mail address: [email protected] (S. Vigneswaran).

Gayathri Naidu, Sanghyun Jeong, Saravanamuthu Vigneswaran ⇑ Faculty of Engineering, University of Technology Sydney, P.O. Box 123, Broadway, NSW 2007, Australia

h i g h l i g h t s

� Effect of salt and CaSO4 on humic substances fouling in seawater was tested with DCMD. � Humic substances disaggregated to

LMW–humics in the presence of salt and under thermal condition. � CaSO4 reduced the disaggregation of

humic substances due to the binding effect. � Irreversible fouling occurred with

seawater and LMW–humics caused membrane pore penetration. � LMW–humics increased biofouling

possibility in seawater treatment with DCMD.

g r a p h i c a l a b s t r a c t

a r t i c l e i n f o

Article history: Received 27 June 2014 Received in revised form 17 October 2014 Accepted 17 October 2014 Available online 24 October 2014

Keywords: Biofouling potential Desalination Direct contact membrane distillation Humic substances Organic fouling Seawater

a b s t r a c t

The interaction of humic substances on organic and biofouling during the treatment of seawater using direct contact membrane distillation (DCMD) was studied. Organic fouling was investigated at different feed temperatures in terms of distillate flux pattern and detailed organic characterization using liquid chromatography-organic carbon detection. The penetration of organics through the membrane pores was observed with SEM cross-section analysis. Also, chemical cleaning analysis showed seawater organic fouling was irreversible in DCMD. Humic substances and low molecular weight (LMW) organics are dom- inant organic contents in seawater. In the tested DCMD system, humic substances were thermally disag- gregated to LMW–humic organics. This phenomenon was more significant in the presence of salt (NaCl), while inorganic scalant (CaSO4) reduced the disaggregation of humic substances due to the binding effect. In this study, the assimilable organic carbon (AOC) test was used to estimate the biological fouling poten- tial during the MD process. AOC concentration is closely associated with biofilm growth in water and on the membrane (biofouling). The AOC concentration increased as the concentration of LMW–humic organ- ics increased. This indicated the possibility of biofouling increasing in the feed and on the membrane in the MD process.

� 2014 Elsevier B.V. All rights reserved.

on; EDS, MW, low ene; RO,

G. Naidu et al. / Chemical Engineering Journal 262 (2015) 946–957 947

1. Introduction studies have observed the disaggregation of humic substances to

Membrane distillation (MD) is a thermal membrane process that uses vapor pressure from temperature difference as the driving force across a hydrophobic membrane [1]. MD is especially advan- tageous as a standalone desalination process for producing drinking water from high saline water sources in remote inland areas [1,2]. This is because the vapor pressure-driven MD operation is not lim- ited by high salinity, in comparison to hydraulic pressure-driven membrane processes such as reverse osmosis (RO) [2]. Despite these advantages, MD’s commercial application is still limited. This is because the energy consumption and its associated cost in MD is still high compared to RO process [2]. For instance, a comparison was made of the water production costs in MD and RO systems by Zuo et al. [3]. The study found that the estimated total water produc- tion cost of MD was $1.55/m3 which was much higher than RO at $0.50/m3. On the other hand, some studies have highlighted that the MD process is competitive with the RO process in that it uses alternative energy such as waste heat or if the water source is a chal- lenge to be treated with the RO system [2,3]. Nonetheless, similar to pressure-driven membrane processes, MD faces the challenge of fouling development. Previous MD studies have acknowledged that fouling occurs with seawater based on declining distillate flux and increased distillate conductivity [4–6]. In MD, fouling development by inorganic salts such as gypsum (CaSO4) and calcium carbonate (CaCO3) present in natural water source has been analyzed in detail, highlighting the influence of thermal operation and hydrodynamic condition on crystal size formation, wetting phenomena and scaling reversibility [7,8]. However, a detailed investigation on the contri- bution of organic compounds to membrane fouling (organic fouling) in MD operation is still lacking.

MD fouling studies have been mostly conducted using individ- ual model organic compounds, which may not reflect the fouling performance with natural water condition [9–11]. This is because fouling in natural water sources is attributed to mixed organic compounds that are referred to as natural organic matter (NOM) [12–14]. At the same time, in natural water sources such as seawa- ter, the presence of salinity and inorganic salts will also contribute to complex membrane fouling development [15,16]. For instance, Shirazi et al. [6] compared the performance of DCMD system with real seawater and synthetic feed sample. Their study observed that the distillate flux reduction was more considerable for the real sea- water feed compared to the synthesized feed sample. This was attributed to the prevalent formation of membrane fouling by mixed constituents in the seawater. Similarly, in a nanofiltration study (NF), it was demonstrated that NOM fouling increased signif- icantly at higher electrolyte (NaCl) concentration and with the addition of divalent cations attributed to electrostatic charges [15].

With regards to the mixed NOM combination, it has been acknowledged that humic substances and low molecular weight organics (LMW) are the main organic compounds present in natu- ral water sources such as seawater [13]. Further, humic substances have been identified as one of the major foulant in membrane pro- cesses [13,14,17]. In this regard, it is worth highlighting that humic substances absorb more favorably onto hydrophobic membrane surfaces [17]. Khayet et al. [10] analyzed fouling of humic sub- stances using two commercial MD membranes with DCMD. A greater amount of fouling effect was observed on the more hydro- phobic membrane [10]. At the same time, it was demonstrated that membrane cleaning with deionized (DI) water flushing after the humic substances fouling experiment was not sufficient to recover the initial distillate flux. This suggested that fouling related to humic substances could be irreversible in the MD process. It is also evident that thermal heating in MD may also influence the charac- teristics of humic substances. Under thermal condition, several

lower molecular sized humics and other organics [18,19]. This aspect has not been evaluated in detail in MD organic fouling studies.

Meanwhile, LMW organics in natural water sources have been regarded as a biofoulant precursor. Previous RO studies have high- lighted the relationship between the presences of bio-available DOC fraction, which was represented as assimilable organic carbon (AOC). This causes biofouling development [20]. A high level of AOC is directly linked to the rapid formation of a biofilm formation [21,22]. In MD, the role of organics as a biofouling precursor has not been explored.

It is worth highlighting that most MD studies represented membrane fouling in terms of distillate flux decline, which may not reflect the full extent of organic fouling [4,5,9–11]. A more sen- sitive analytical method such as liquid chromatography-organic carbon detector (LC-OCD), would provide a valuable guide on the organic compounds and their behavior on the membrane foulant, and in both feed and distillate solution. The LC-OCD analysis have been used to study the detailed membrane fouling trend using nat- ural water sources such as seawater and lake water by other mem- brane processes [12,23–25]. Furthermore, the LC-OCD analysis uses the size exclusion principle to fractionate the organic com- pounds based on molecular weight (MW). Here, the high MW bio- polymer peak (>20,000 Da) elutes first, followed by peaks of humic substances (800–1000 Da), building blocks (BB) (350–600 Da) and LMW organics (<350 Da) [23,26]. In the LC-OCD analysis, BB com- pounds were considered to be the degraded products of humic substances [23]. Meanwhile, smaller disaggregated humic sub- stances (LMW–humic organics) co-elute as LMW organics accord- ing to the LC-OCD analysis.

Hence, a detailed characteristic of organic compound behavior in MD would provide a better representation of MD for desalina- tion application. For this purpose, organic fouling behavior in a DCMD system was carried out with actual seawater (SW). Organic characterization of the SW feed, distillate, and membrane foulant was carried out with LC-OCD. The morphology and composition of the deposit layer formed on the membrane surface was exam- ined using scanning electron microscopy – energy dispersive spec- trometer (SEM-EDS). Membrane cleaning and hydrophobicity measurements were conducted to determine the SW fouling reversibility. In this study, the emphasis was placed on humic sub- stances and LMW organics as the main organic compounds in SW. Therefore, the fouling pattern of synthetic humic acid (HA) feed solution was evaluated under different conditions. Firstly, the ther- mal effect on humic substances was evaluated by varying the oper- ating temperatures. At the same time, the influence of the physico- chemical conditions of the feed water sources was analyzed with saline HA feed and inorganic divalent HA feed solution. Further- more, the role of organics as a biofoulant precursor was evaluated with LMW organics and HA compound using AOC measurement.

2. Materials and methods

2.1. DCMD set-up

A bench-scale DCMD system was used in this study as shown in Fig. 1. Photo of the DCMD experimental set-up can be found in Fig. S1. The feed and distillate tanks were placed on electronic bal- ances to monitor the feed reduction and distillate production over time. A mass balance (CAS model NT-501A, Republic of Korea) was used with an accuracy limit of 0.01 kg. A hydrophobic polytetra- fluoroethylene (PTFE) flat sheet membrane (General Electric, US) was used for this study. The effective membrane area was 0.0168 m2. The membrane channel dimensions were 21.0 cm

Fig. 1. DCMD experimental setup.

Table 2 Synthetic HA feed solution combination and operating conditions.

No Model Foulant Feed Solution Ft (�C)

I HA HA (10 mg-C L�1) 50 and 70

II Saline HA HA (10 mg-C L�1) with NaCl (1 M) 70 III Inorganic

scalant HA HA (10 mg-C L�1) with CaSO4 (60 mM NaSO4 + 20 mM CaCl2)

70

948 G. Naidu et al. / Chemical Engineering Journal 262 (2015) 946–957

(length), 8.0 cm (width), and 0.4 cm (height). The porosity, average pore size and membrane thickness provided by the supplier were 70–80%, 0.2 lm, and 179 lm, respectively [27]. The membrane support layer is made of polypropylene. A hydrophobic PTFE mem- brane was used in this experimental study because PTFE repre- sents an ideal material for MD membrane. It exhibits one of the highest hydrophobic membrane characteristics among polymers and also one of the best chemical resistance and thermal stability [28]. Shirazi et al. [29], observed that PTFE membranes with pore sizes below 0.5 lm achieved the highest salt rejection efficiency (99.9%) compared to PTFE membranes with higher pore sizes.

2.2. Feed solution

2.2.1. Seawater (SW) The first set of experiments was conducted with seawater.

Actual SW was collected from the Sydney Institute of Marine Sci- ence, Chowder Bay, Australia. It was collected from 1 m below the sea surface level and filtered through 140 lm centrifuge filtra- tion system to remove all large particles. The detailed characteris- tics of the SW are shown in Table 1.

2.2.2. HA To study the fouling behavior of HA (one of the major organic

components in seawater) on MD, a synthetic HA feed solution, HA compound (Sigma–Aldrich, St. Louis, MO) was used at a con- centration of 10 mg-C L�1 (feed I). The HA used in this study was prepared from a stock solution by dissolving 250 mg-C L�1 HA compound (available in powder form) into deionized (DI) water, followed by filtration through a 0.45 lm filter to exclude initial deposition of bulk suspended solids on the membrane. The initial concentration of the organic substances (10 mg-C L�1) was

Table 1 Characteristics of seawater used in this study.

Analysis category Measurement value

pH 8.2 Conductivity (mS cm�1) 51.8–55.5 TSS (mg L�1) 3.6 Salinity (g L�1) 35.5 Turbidity (NTU) 0.41 UV254 (cm

�1) 0.026 DOC (mg L�1) 1.29 Ca (mg L�1) 411 Mg (mg L�1) 1,400 SO4 (mg L�1) 2,680 Na (mg L�1) 11,300 Cl (mg L�1) 19,225

determined in terms of total organic carbon concentration. For HA feed II and HA feed III, 1.0 M of NaCl and CaSO4 (0.06 M NaSO4 + 0.02 M CaCl2) were added respectively into feed I (Table 2).

2.3. Fouling experimental procedure

For the fouling experiments, 2 L of feed solution was heated to a temperature of 70 �C with an electric blanket. Experiments in this study were conducted at high feed temperatures of 70 �C to obtain high distillate flux output, based on our DI water baseline study. The DCMD system achieved a high distillate flux of 35.7 L m�2 h�1

at Ft = 70 �C, which reduced considerably to 16.0 L m �2 h�1 at

Ft = 50 �C. Previous MD studies have similarly recommended high feed temperature settings between 60 �C and 80 �C for optimum distillate flux output [2,28,30]. Meanwhile, 2 L of cooling solution (DI water) was set constantly at an average temperature of 25 �C. The temperatures were maintained at ±2 �C for all experiments. A temperature transmitter (T158, Wise Sensor Inc, Republic of Korea) was placed at the top in the feed tank to detect the bulk feed tank temperature while the temperature gradient along the cool- ing inlet and outlet as well as the feed outlet was measured by the respective thermocouples as shown in Fig. 1. A feed and distil- late flow velocity of 1.1 m s�1 was used to channel the feed and cooling solution to both sides of the membrane with a gear pump (Cole Parmer, model 75211-15, United States). At feed and distil- late flow velocity of 1.1 m s�1, the respective applied pressure was 103 kPa and 7 kPa, respectively. The experiments were contin- ued till a feed volume concentration factor (VCF) of 4.5–5.0 times was obtained whereby the initial 2.0 L feed volume was reduced to a feed volume of 0.4 L. The experiment’s duration varied between 3.5 and 7.0 h to achieve the respective VCF.

2.3.1. DCMD fouling test with seawater A DCMD fouling test was carried out with actual SW at a Ft of

70 �C (SW-70 �C) using the procedure described above. In order to study the thermal effect of DCMD on SW fouling pattern, a lower Ft of 50 �C was also used (SW-50 �C). Further, a chemical cleaning was carried out at the end of the SW-70 �C experiment to observe the reversibility of the fouled membrane. In this reversibility test, 0.1 M sodium hydroxide (NaOH) solution was used as a chemical cleaning solution targeting organic foulant removal, based on another MD fouling study [10]. The reversibility test was carried out using the following procedure: (i) recirculation of MiliQ (MQ) water for 3 h, (ii) recirculation of NaOH cleaning solution for 30 min, (iii) rinsing the membrane module several times with MQ to eliminate NaOH residues, and (iv) drying the membrane for 24 h before the following experiment. To complement the reversibility test, membrane hydrophobicity was measured with contact angle, based on the procedure explained below.

2.3.2. HA fouling test The HA fouling test was carried out with HA feed I at Ft of 70 �C

using the procedure described above. In order to study the thermal effect on HA organic compound, a lower Ft of 50 �C was also used. In addition, HA feed II was used to study the effect of salinity on

G. Naidu et al. / Chemical Engineering Journal 262 (2015) 946–957 949

organic fouling while HA feed III was used to analyze the presence of inorganic scalant on organic fouling.

2.3.3. Biofouling test The role of HA compound as a biofoulant was measured with

feed II and III. This represented the HA compound in the presence of salinity and inorganic factors in an organic feed solution. The biofouling test conducted with SW feed and membrane foulant was compared to the individual organic results.

2.4. Measurements

2.4.1. Distillate flux The corresponding volume of decreased feed water (L) and pro-

duced distillate (L) with the operation time (h) and membrane area (m2) was used to calculate the experimental distillate flux in L m�2 h�1. In this study the distillate flux was expressed as a func- tion of feed volume concentrate factor (VCF). The VCF factor is defined as the ratio of initial feed volume (Lf, 0) to concentrate feed volume (Lc), representing the extent of volume concentration of feed, VCF = Lf, 0/Lc [27,31].

2.4.2. Organic characterization The detailed organic fraction of the feed solution and mem-

brane foulant was determined as dissolved organic carbon (DOC) concentration using a LC-OCD (developed by Dr. Huber) [23]. The LC-OCD system utilized a Toyopearl TSK HW50S column (TOSOH Bioscience GmbH, Stuttgart, Germany), with phosphate buffer mobile phase of pH 6.4 (2.6 g L�1 KH2PO4 and 1.5 mol L

�1 Na2- HPO4) at a flow rate of 1.1 mL min

�1. The LC-OCD analysis sepa- rates the total DOC into hydrophilic and hydrophobic fractions. Further, the size exclusion chromatography (SEC) column in the LC-OCD system separates the hydrophilic organic fraction accord- ing to their molecular size during the retention time. The separated compounds are detected by an ultraviolet (UV) detector (absorp- tion at 254 nm) and an OCD detector (after inorganic carbon purg- ing). Based on the SEC retention time, the first peak of the chromatogram is the polysaccharide/biopolymer peak. The second and third peaks (or shoulder of humic peak) are attributed to humic substances and BB (sub-units of humic substances), respec- tively. The last peaks are attributed to LMW organics (acids and neutrals). The proportion of each organic fraction was calculated by a software program (ChromCALC DOC-LABOR, Karlsruhe, Ger- many) based on LC-OCD chromatogram.

In this study, a single SEC column LC-OCD at a retention time was 120 min was used to analyze the synthetic feed solution sam- ples. Meanwhile, a dual SEC column LC-OCD at double the reten- tion time of 240 min was used to examine the seawater samples. The longer SEC analysis enables to clearly project the organics sub-fraction within the multi-constituents of the seawater. For all the seawater analysis using the dual column, the chromato- grams of all the sub-fractions were reflected within 180 min. As such, in this study, the chromatograms are projected up to 200 min only.

2.4.3. Membrane analyses 2.4.3.1. Organic foulant analysis using LC-OCD. This study analyzed the characteristics of organics (extracted) from the fouled mem- brane using LC-OCD. The extraction of foulants on the membrane was conducted after each experiment by cutting the fouled MD membrane into small parts. The cut membrane pieces were placed in a beaker with MQ water. The beaker was sonicated to extract the organic residues on the MD membrane. This sonication was carried out with an ultrasonic bath (Powersonic 420, Thermoline Scientific, 300 W) for a short time (10 min) to prevent organic matter from denaturing and any biological modification [12].

2.4.3.2. Field emission-scanning electron microscope (FE-SEM). The organic fouling was observed and quantified on the membrane sur- face and pores (membrane cross-section) using Zeiss Supra 55VP field emission scanning electron microscope (FE-SEM). The fouled membrane coupons were dried in a desiccator overnight and ana- lyzed without any further treatment for SEM observation. Both the top (surface) and the cross-section of the fouled membrane cou- pons were analyzed. It was operated at 5 kV in conjunction with Bruker XFlash silicon drift detector energy-dispersive spectroscopy (EDS) detector to obtain chemical information. EDS spot analysis was carried out using a spot diameter of about 3 nm at selected areas on the samples.

2.4.3.3. Contact angle measurement. The contact angle of the mem- brane surface was utilized to determine: firstly, the hydrophobic reduction of the membrane caused by fouling; and secondly, the hydrophobicity recovery after the SW-70 �C reversibility test. This measurement was carried out by sessile drop method using a goni- ometer (Theta Lite) with 1.8–2.0 mL of MQ water droplet on the dried membrane surface. Measurements were repeated 5 times and the average reading was reported in this study.

2.4.4. Biofouling potential measurement The biofouling potential was measured using a rapid biolumi-

nescence AOC measurement method in the experiments. The detailed AOC procedure is explained elsewhere [21,22]. The sam- ples of initial and concentrated final feed solution, as well as the extracted membrane foulant were used to measure AOC. All sam- ples were filtered with a 0.45 lm filter for particle removal, fol- lowed by a 0.22 lm filter for bacteria inactivation. The AOC concentration was expressed by lg-C glucose equivalents L�1 as glucose was used as standard for the AOC compound [21]. For the membrane foulant, AOC concentration value was calculated based on the membrane area.

3. Results and discussion

3.1. DCMD with SW

3.1.1. Flux pattern and distillate quality The performance of DCMD with SW was evaluated in terms of

distillate flux pattern. The experimental operation was expressed as a function of volume concentration factor (VCF) of feed water. The initial bulk feed solution was referred to as VCF 1.0, which was the feed solution prior to the thermal heating process.

The DCMD operated with SW-70 �C showed a 7–10% distillate flux decline at the initial stage (from VCF 1.0 to 1.4). This is attrib- uted to salinity as observed in other MD studies [2]. From VCF 1.4 to 2.6, the distillate flux remained relatively stable. Thereafter the distillate flux gradually declined and a significant distillate flux decline of 31.9% was observed by VCF 3.6 (Fig. 2).

The distillate quality, measured in terms of conductivity, remained in the low range from VCF 1.0 (5.20 ± 0.76 lS cm�1) to VCF 4.0 (6.15 ± 0.94 lS cm�1). Previous MD organic fouling studies have also reported the low TDS value in the distillate solution [9,10]. However, this study also measured DOC concentration using LC-OCD. The distillate solution DOC value, obtained from LC-OCD showed a more clear association with flux pattern than the conductivity measurement. Initially, the DOC concentration of DI water in the distillate tank was 0.13 ± 0.02 mg L�1. The final DOC concentration of distillate was 0.40 ± 0.27 mg L�1 at VCF 4.0 (Fig. 2). Initially the DOC concentration was stable but from VCF 3.0 onwards, DOC concentration in the distillate indicated a grad- ually increasing pattern. This could be attributed to DOC perme- ation through the membrane as the feed DOC concentration

Fig. 2. Distillate flux and DOC concentration of distillate and feed solution as a function of VCF for SW-70 �C.

950 G. Naidu et al. / Chemical Engineering Journal 262 (2015) 946–957

increased. The presence of organic compounds in the distillate solution could be associated to an adsorption–desorption mecha- nism as explained in a recent study [32]. This study showed humic substances adsorbed onto the membrane surface via bonding through the phenolic and carboxylic functional groups in the MD process. The humic substances that adsorbed at the edge of the pores, migrated due to hydrogen bonding between unattached car- boxylic or phenolic groups on the molecule and water vapor, which consequently desorbed to the distillate side. Similarly, in our study, the presence of organic compounds was detected in the distillate solution. However the conductivity of the distillate solution did not increase, suggesting wetting did not occur by the membrane pores. Instead, the humic substances adsorbed onto the membrane surface and desorbed to the distillate solution. In another study, the chemical and biological oxygen demand as well as TDS rejec- tion rates of biodiesel’s wash water effluent were observed to have increased significantly using an electrospun polystyrene mem- brane [33]. One of the factors attributed to this was the reduced surface roughness that reduced the membrane’s adsorption capacity.

3.1.2. Organic characterization of membrane foulant, distillate, and feed solution

The LC-OCD analysis showed that the DOC concentration of the SW (feed solution, VCF 1.0) was 1.85 mg L�1 comprising of 0.04 mg L�1 of biopolymers, 0.49 mg L�1 of humic substances, 0.005 mg L�1 of BB and 1.31 mg L�1 of LMW organics. SW mainly contained humic substances and LMW organics which is more than 70% of the total DOC as other studies have established [13]. Mean- while, the biopolymer and BB compounds in SW were minimal. Thus, for the DCMD operation with SW feed, the discussion on organic characterization did not focus on the biopolymer and BB compounds in this study.

To understand the organic compounds responsible for the dis- tillate flux pattern, the organic compounds present in the mem- brane foulant and the distillate solution were observed. The change in organic characteristics in the feed solution was studied in detail.

At the end of the DCMD experiment with SW-70 �C (VCF 4.0), a visible brown layer appeared on the membrane. LC-OCD chromato- gram showed that the foulant deposited on the MD membrane with SW-70 �C comprised of LMW organics (Fig. 3a).

Furthermore, to quantify the organic mass on the membrane, an organic mass balance was carried out, based on the DOC value of the initial and final SW-70 �C feed as well as the distillate solution. The initial DOC value of the SW-70 �C feed solution at VCF 1.0 (2.0 L) was 1.85 mg L�1 which amounted to an organic mass of 3.70 mg. At VCF 4.0 (0.5 L), the SW-70 �C feed solution DOC value was 3.12 mg L�1, resulting in a reduced organic mass of 1.56 mg

on the membrane. At the same time, around 1.40 mg organic mass was in the distillate solution, based on the distillate DOC value of 0.40 ± 0.27 mg L�1 at VCF 4.0 (3.5 L). The remaining organic mass was assumed to have been deposited on the membrane. Hence, in the SW-70 �C experiment, 43.5 mg was deposited on the unit area of the membrane (m2). The membrane area was 0.0168 m2.

The LC-OCD chromatogram of the MD distillate solution with SW-70 �C at VCF 4.0 displayed the predominant presence of LMW organics, indicating the penetration of LWM organics into the pores of the hydrophobic membrane (Fig. 3a).The results showed that the LMW organics were not only deposited on the membrane but also penetrated through the pores of the membrane.

The SW-70 �C feed solution’s organic characteristics were ana- lyzed using LC-OCD, to represent the changes of the main organic compounds, humic substances and LMW organics from VFC 1.0 to VCF 4.0 (Fig. 3b). As mentioned previously in Section 3.1, the organic characterization discussion did not focus on the biopoly- mer while the BB compound was clustered with the LMW organics due to its minimal presence.

The concentration of humic substances revealed an increasing trend from VCF 1.0 to VCF 2.0, where the concentration of the feed solution increased by twice the rate of the initial humic substances (0.49 mg L�1). Meanwhile, the humic substances increased by only 3.5 times of the initial concentration value at VCF 4.0 (contrary to the expected 4-fold concentration increase). This suggests that a reduction in the concentration of humic substances occurred by VCF 4.0.

Meanwhile from VCF 1.0 to VCF 2.0, the LMW organics increased 2.7-fold from 1.32 mg L�1 to 3.53 mg L�1. However, from VCF 2.0 to VCF 3.0 onwards, the LMW organics demonstrated a decreasing trend. At the same time, the distillate flux from VCF 2.6 onwards also indicated a declining trend. These results sug- gested that the LMW organics could have deposited on the mem- brane between VCF 2.0 and VCF 3.0, contributing to the distillate flux decline. Also, the LC-OCD chromatograms of the membrane foulant indicated only the presence of LMW organic peaks. A plau- sible explanation for this is that in comparison to the low concen- tration of DOC in SW feed solution, the presence of high inorganic salt contents formed an initial deposit layer on the membrane. From VCF 1.0 to VCF 2.0, the LMW organic concentration increased in the feed solution since it did not adhere to the membrane. More- over, the hydrophilic content in the SW increased from VCF 1.0 to VCF 2.0. This increased the chances of interaction and formation of a fouling deposit. Thus, from VCF 2.0 onwards, LMW organics adhered on the membrane and became a secondary fouling layer. This caused a decline in the LMW organics concentration in the feed SW. The same trend was observed in another study [32].

In the meantime, while the humic substances also reduced in feed side the presence of humic substances was not detected on the membrane foulant or in the distillate solution composition. This indicated that humic substances could have disaggregated into LMW organics under the MD thermal process. This explained the increment of LMW compounds at VCF 2.0.

Overall, the MD organic characterization with SW-70 �C showed that in the feed solution, the LMW organics increased significantly more than their concentration factor. At the same time, only LMW organics were present in both the foulant and distillate solution. On the other hand, although the humic substances were reduced in the feed solution, this was not detected in the foulant or distil- late solution. Humic substances are one of the main organic com- pounds in SW apart from LMW organics. Thus, it would be important to identify the specific role of humic substances in organic fouling development in MD with SW-70 �C. For this reason organic behavior in MD was further investigated in detail using a synthetic HA feed solution (see Section 3.2).

Fig. 3. DCMD fouling pattern with SW-70 �C (a) LC-OCD chromatogram of membrane foulant and distillate at VCF 4.0 (b) LC-OCD chromatogram changes of feed solution from VCF 1.0 to VCF 4.0 (A: Biopolymer; B: Humics; C: Building Blocks; and D: LMW acids and neutrals); (VCF 1.0 is the initial feed solution prior to the start of the experiment).

Fig. 4. Membrane cleaning cycle of DCMD operation with SW-70 �C.

G. Naidu et al. / Chemical Engineering Journal 262 (2015) 946–957 951

3.1.3. Fouling reversibility The reversibility of the MD organic fouling development with

SW-70 �C was evaluated with a membrane cleaning cycle test. The MD experiment with SW-70 �C was operated till VCF 3.0. This was followed by a membrane cleaning cycle with NaOH (procedure described in Section 2.3.1). The experiment was repeated 6 times (Exp I to Exp VI) and it incorporated cleaning cycles (cleaning cycle I to VI) respectively (Fig. 4).

In our study, from Exp (I) to Exp (II), a 97.6% distillate flux was recovered. However, from Exp (III) onwards, the flux recovery rate reduced between 88.5% and 91.0%. For all experiments, the distillate quality in terms of conductivity or total dissolved solids (TDS) remained low in the range of 5.23 ± 0.82–6.02 ± 0.21 lS cm�1. Con- versely, the distillate DOC (0.30 ± 0.11 mg L�1–0.44 ± 0.08 mg L�1) from Exp (I) to Exp (III) increased gradually up to a DOC concentra- tion of 0.98 ± 0.21 mg L�1 at VCF 3.0, in Exp (V) (Fig. 4). The increase in DOC concentration in the distillate side reflected the fact that the organics penetrated onto the distillate side from Exp (IV) onwards. Another cleaning cycle followed by Exp (VI) was executed to verify the pattern of flux and distillate DOC level. The distillate flux was slightly lower than that in Exp (V), while the DOC concentration increased significantly to 1.43 ± 0.05 mg L�1.

Meanwhile, the average hydrophobicity of the membrane after cleaning cycle (VI) was 113.1 ± 7.2� (19% reduction from the virgin membrane hydrophobicity of 139.9 ± 1.2�) indicating the hydro- phobicity reduction of the membrane. The results established the

initial reversibility of the organic compounds on the membrane followed by irreversible fouling deposits that reduced the mem- brane’s hydrophobicity. A similar phenomenon was observed in another DCMD study for protein foulant removal, with an initial positive recovery of distillate flux after acid solution rinsing. This was followed by a gradual decline in the maximum distillate flux recovered over time [34]. The study associated acid solution rins- ing with wettability of membrane pores.

3.1.4. Membrane observation An analysis of the MD membrane with SW-70 �C was carried

out using SEM-EDS. The organic content across the fouled mem- brane was quantified with a line depth analysis approach using the carbon (C) and oxygen (O) element peaks to represent the organic fouling deposit [34]. This strategy was based on other membrane studies that investigated organic fouling through the carbon and oxygen element peak [35,36]. The C and O elements of the virgin membrane were used as a baseline and the elements on the fouled membrane were compared to the virgin baseline value. The O element peak had only small increment, and therefore was neglected in the subsequent analysis. The SEM-EDS line anal- ysis was carried out from point A to point C (membrane cross-sec- tion depth of 0–300 lm) (Fig. 5). The membrane surface began at point 90–95 lm (point B), which was determined from the virgin membrane and the start of the fluoride peak, which is one of the compounds in the membrane structure (Fig. 5c).

952 G. Naidu et al. / Chemical Engineering Journal 262 (2015) 946–957

The SEM image displayed a visible foulant layer and particle deposits on the SW-70 �C MD membrane surface (Fig. 5b) in com- parison to the virgin membrane (Fig. 5a). The SEM-EDS line analy- sis of the membrane from MD investigation with SW-70 �C showed the C element penetration to the pores from 90 lm to 300 lm in comparison to the virgin membrane (Fig. 5d).

In summary, the MD performance with SW-70 �C showed a dis- tillate flux decline and an increase in distillate DOC concentration. Despite the initial MD distillate flux being recovered, in the subse- quent operation cycles only a partial distillate flux recovery was observed. The penetration of organics to the distillate side was observed, indicating that irreversible fouling had occurred. Further, only a small organic mass was deposited on the membrane. The results evidently suggest that MD organic fouling with SW-70 �C was mainly due to the pore penetration of organics. The detailed organic characteristics indicated the reduction of humic substances in the feed solution while only the presence of LMW organics was detected on the membrane as well as in the distillate solution. Con- sequently, the factors that contribute to the evident presence of LWM organics and the reduction of humic substances must be established. Since they constitute one of the main organic com- pounds in SW, it is important to evaluate in detail the specific role of humic substances in MD fouling development with SW-70 �C. Moreover, apart from the organic compounds, a number of factors that contribute to organic fouling must be given due consideration, mainly the influence of salinity and the presence of inorganics under high temperature MD operation. For this reason organic behavior in MD was further investigated with a synthetic HA feed solution.

3.2. DCMD with HA feed solution

The MD organic fouling by humic substances was evaluated using synthetic HA feed solution (10 mg-C L�1). The influence of (i) temperature (Ft = 70 �C and Ft = 50 �C) (with HA alone-feed I)

Fig. 5. Cross section SEM-EDS analysis of (a) virgin membrane SEM image, (b) SW-70 �C m (d) carbon element membrane cross section line analysis. (A = top point at a distance membrane at a distance of 300 lm).

(ii) salinity (feed II), and (iii) the presence of inorganic scalant (feed III) on HA interaction was studied with DCMD. In this study, the behavior of humic substances was studied individually and in the presence of high salt concentration (1 M). To clearly reflect the role of humic substances as an organic foulant, especially in the pres- ence of a high salt concentration, a larger concentration of HA (10 mg-C L�1) feed solution was used. This is almost 5 times higher value than the 2–3 mg L�1 DOC concentration in natural water sources such as seawater. At the same time, the presence of DOC in the distillate side is critical to establish the occurrence of mem- brane wetting due to organic fouling in MD. Using a higher concen- tration of HA feed solution enables to clearly detect the characteristics of organics present in the distillate side.

3.2.1. Effect of temperature The influence of temperature on fouling by humic substances

was analyzed with HA feed I at Ft = 70 �C and at Ft = 50 �C. Feed temperature is the key operating parameter for the mass transfer resulting in lower initial distillate flux at Ft = 50 �C (16.0 L m

�2 h�1) compared to the distillate flux at Ft = 70 �C (35.7 L m

�2 h�1). Due to the higher mass transfer at Ft = 70 �C, more migration of humic substances occurred on the membrane surface, resulting in higher organic adsorption onto the membrane. This resulted in more rapid flux decline at Ft = 70 �C (Fig. 6a). The HA feed I organic character- istics were compared for Ft = 50 �C and Ft = 70 �C at VCF 3.0 (Table 3). The LC-OCD chromatogram of the initial feed I (VCF 1.0) comprised of humic substances at a concentration of 4.61 mg L�1 and LMW organics at 0.74 mg L�1. At VCF 3.0, the humic substances increased by a ratio of 1.1–1.3 times of their ini- tial value with Ft = 70 �C, and Ft = 50 �C. On the other hand, the LMW organics increased by only 1.5 times of their initial value at Ft = 50 �C (Fig. 6e). Comparatively for Ft = 70 �C, the LMW organics increased by 2.0 times of its initial values. The results indicated that at a higher temperature of Ft = 70 �C, the humic substances’

embrane SEM image, and (c) fluoride element membrane cross section line analysis 0 lm; B = top of membrane surface at a distance of 90–95 lm; and C = bottom of

G. Naidu et al. / Chemical Engineering Journal 262 (2015) 946–957 953

disaggregation to LMW organics occurred at a much more rapid rate. As a result, lower peaks of LMW organics were present in the membrane foulant HA-50 �C (Fig. 6c). The results highlight two important factors; firstly higher organic compound adhesion on the membrane was influenced by the high temperature opera- tion in MD; and secondly, the adhesion of organic on the mem- brane affects the mass transfer, resulting distillate flux decline. Similarly, the organic characteristics of MD with SW at Ft of 50 �C (SW-50 �C) were compared to that of SW-70 �C at VCF 3.0. At SW-50 �C, the normalized distillate flux of J/J0, showed no signifi- cant reduction from the initial J/J0 up to VCF 3.0 (Fig. 6b). Compar- atively, for SW-70 �C, by VCF 3.0, a 15–20% distillate flux decline was observed. Also, the fouled membrane of SW-50 �C showed 35% less LMW organics than SW-70 �C (Fig. 6d and f).

Previous organic studies have also noted the disaggregation of humic substances with temperature increment [19,37]. These studies highlighted that at increased temperature, thermal agita- tion disaggregates the high molecular weight humic substances, creating smaller molecular sized humics. In a study conducted with brackish water, the particle size of humic substances (mea- sured with dynamic light scattering) was reported to reduce signif- icantly from 200 nm to 70 nm when the feed temperature was increased from 15 �C to 35 �C. In this context, as highlighted in the introduction, the LC-OCD analysis fractionates organic com-

Fig. 6. Influence of feed temperature on MD with HA feed I and SW: (a) normalized dist SW-50 �C (c and d) membrane foulant (VCF 4.0) of HA-50 �C and SW-50 �C (VCF 4.0), (e Humics; C: Building Blocks; and D: LMW acids and neutrals).

pounds based on the size exclusion method. As such, the LC-OCD analysis represents the small molecular sized humic substances as LMW–humic organics. Huber et al. [23] referred to these small molecular sized humic substances as LMW–humic organics that co-elute with LMW organic peaks in the LC-OCD chromatogram. This explained the significantly higher build-up of LMW organics concentration with HA feed I in our study, which was actually LMW–humic organics formed from the thermal disaggregation of humic substances.

Overall, the results demonstrated the tendency of humic sub- stances to disaggregate more rapidly into LMW–humic organics at higher temperature. This finding could be related to the 35% more LMW organics that appeared on the membrane foulant of SW-70 �C compared to SW-50 �C. At SW-70 �C it was most likely that the humic substances underwent a higher thermal disaggrega- tion to LMW–humic organics.

3.2.2. Effect of the presence of NaCl For HA feed solution (feed I), at the initial stages of VCF 1.0–2.0,

the distillate flux declined by 5.8%. Thereafter, the distillate flux declined by 50.4% till VCF 4.0 (Fig. 7a). At VCF 1.0 the measured DOC value of the HA feed I was 9.51 mg L�1 (20 mg HA mass com- pound). At VCF 4.0 the measured DOC value of the HA feed I was 19.18 mg L�1, amounting to a remaining mass of 9.6 mg organics

illate flux of HA-70 �C and HA-50 �C, (b) normalized distillate flux of SW-70 �C and and f) feed solution (VCF 1.0–VCF 4.0) of HA-50 �C and SW-50 �C (A: Biopolymer; B:

Table 3 LC-OCD organics characteristics of HA feed I with different feed temperatures (Ft = 50 �C and Ft = 70 �C) at VCF 3.0.

Feed solution Total DOC (mg L�1) Hydrophilic compound fraction (mg L�1)

Biopolymers Humic substances LMW organics

Initial HA (VCF 1.0) 9.51 ± 0.51 n.q 4.61 0.74 HA-70 �C (VCF 3.0) 12.23 ± 0.20 n.q 5.19 1.47 HA-50 �C (VCF 3.0) 12.08 ± 0.31 n.q 5.80 1.28

Note: n.q = non quantifiable.

954 G. Naidu et al. / Chemical Engineering Journal 262 (2015) 946–957

in the feed solution. Further, a 0.94 mg L�1 organic concentration was detected in the final distillate solution of 3.5 L. This amounted to a mass of 3.3 mg organics in the distillate solution. Hence, the organic mass deposited on the membrane with HA feed I was 423.2 mg-organic mass m�2 (Table 3).

Compared to this, HA in the presence of 1 M of NaCl (HA feed II) resulted in a distillate flux decline of 65.6% at VCF 4.0 (Fig. 7d). The HA feed II distillate flux pattern appeared to be similar to that with SW-70 �C, where an initial stable distillate flux was achieved. This suggested that the presence of NaCl (salt) influenced the distillate flux pattern of MD organic fouling with SW. Furthermore, a thick brown layer of foulant was found covering the membrane surface at the end of the experiment with HA feed II. This is similar to the observation with SW-70 �C. It indicated that salinity (presence of NaCl) influences the fouling behavior of HA. In SW the presence of inorganic salt as well as organic compounds must be given due consideration in influencing the permeate flux decline. At a feed temperature of 70 �C, inorganic salts that are inversely soluble under high temperature would have resulted in the formation of crystals in the feed solution because the concentration of feed solu- tion increases over time. This forms a scale deposit on the mem- brane surface. In the presence of organics, the interaction of salt and humics resulted in the formation of a secondary layer. This in turn resulted in a rapid flux decline from VCF 3.0. The humic substances’ interaction with salt on the membrane surface was evi- dent from the observation of a brown deposit layer forming on the membrane surface at the end of the experiment. A similar scenario was observed in a recent study [32].

The LC-OCD chromatogram of the initial feed solution showed that the presence of NaCl resulted in a more significant disaggrega- tion of humic substances to the LMW–humic organics (Fig. 7e). At VCF 1.0, the initial feed II consisted of humic substances at a con- centration of 2.1 mg L�1 with higher contents of LMW organics of 2.7 mg L�1 (Table 4). In comparison, the initial HA feed I contained about half the portion of LMW organics only at 0.74 mg L�1. Recent studies have analyzed the formation of humic substances using nuclear magnetic resonance spectroscopy and size exclusion chro- matography [38,39]. These studies highlighted that the high molecular weight humic substances are a result of small HS mole- cules aggregating, and they interact with each other through hydrophobic bonding (Van der Walls) and hydrogen bonding. These studies also observed the disruption of the weak hydrogen and hydrophobic bonding in the presence of salts. Similar to the thermal disaggregation of humic substances, small sized humic substances were formed in the presence of salt (NaCl).

Meanwhile, the membrane foulant with HA feed II (Fig. 7f) showed a 112% higher concentration of LMW organics compared to the membrane foulant with HA feed I (Fig. 7c). Based on the ini- tial DOC concentration of HA feed II at VCF 1.0 (9.84 mg L�1), and the final DOC concentration of HA feed II at VCF 4.0 (12.69 mg L�1), an organic mass reduction of 11.68 mg was estimated. As well, a 2.1 mg mass of DOC was measured on the distillate side and this resulted in an organic mass of 694.4 mg-organic mass m�2 being deposited on the membrane with HA feed II. In comparison, in HA feed I, relatively smaller organic mass of 423.2 mg-

organic mass m�2 was deposited on the membrane. This investiga- tion revealed that humic substances in the presence of NaCl con- tributed to a larger foulant deposit on the MD membrane. This is because humic substances become coiled and spherical in shape due to the electrostatic shielding between the negative charges of carboxyl function groups of humic molecules by the Na+ ions. This consequently decreases the electrostatic repulsion between the membrane surface and saline humic substances, resulting in higher foulant deposition [11,15]. As a result, higher distillate flux decline occurred with HA feed II. Similarly, a study on DCMD with saline feed solution showed that the organic compound (protein) deposition increased in the presence of salt [9].

These organic mass deposit of HA feed I and II were significantly higher to that of SW-70 �C membrane foulant (43.5 mg- organic mass m�2). This could be attributed to the higher initial organic mass with synthetic HA feed. This may also explain the higher distillate flux decline with HA feed I and II compared to SW-70 �C. Nevertheless, for both SW and the HA feed solutions, the organic compositions of the membrane foulant and distillate solution were the same with prevalent LMW organics. This sug- gests that in SW-70 �C, apart from the existing LMW organics, the humic substances in SW play a major role. Humic substances disaggregate into LMW–humic organics under thermal conditions and in the presence of salts which adhere on the membrane as LMW organics.

3.2.3. Effect of the presence of inorganic scalant In brackish and seawater, Ca2+ salts are also present in addition

to Na+, SO4 2�, and organics. Thus, in this study, CaSO4 was used as

the inorganic scalant [34]. The presence of CaSO4 (inorganic sca- lant) in HA feed solution (HA feed III) resulted in significant distil- late flux decline from the initial stages onwards (from VCF 1.0 to 2.0 – 76.7% distillate flux decline) (Fig. 7g). Similarly, in a previous DCMD study, only a minimal flux decline was observed with 30 mg L�1 HA feed solution at Ft of 70 �C. On the other hand, a higher distillate flux decline of 43–45% was observed with the addition of Ca2+ (3.78 mM CaCl2) [11]. Another previous study associated the combination of divalent ions in HA solution with the reduction of electrostatic repulsion, complexation and bridging effects resulting in a coagulated mixture [13]. In line with this, at the end of our experiment, a thick layer of deposit was observed on the membrane. However, in the actual SW, unlike HA feed III, the presence of divalent ions and humic substances in smaller con- centrations did not result in the formation of a thick layer of deposit [11].

In terms of feed solution organics, the LC-OCD chromatogram showed that in the presence of CaSO4 (HA feed III), the peak of humic substances increased by 2.1 times its initial value from VCF 1.0 to 4.0 (Fig. 7h). This increased concentration of humic sub- stances was much higher compared to HA feed I (1.3 times) (Fig. 7b) and feed II (1.6 times) (Fig. 7e). On the other hand, the dis- aggregation of humic substances to LMW organics was much lower in feed III (only 1.9 times increase from VCF 1.0 to 4.0) compared to HA feed I and HA feed II. This could be attributed to the presence of the divalent Ca2+ and its role as a binding agent of the carboxyl

Fig. 7. Fouling pattern of HA compound based on distillate flux decline, LC-OCD chromatograms of feed solution and membrane foulant for (a–c) feed I (HA) (d–f) feed II (saline HA) (g–i) feed III (inorganic scalant HA compound (CaSO4) (A: Biopolymers; B: Humics; C: Building Blocks; and D: LMW acids and neutrals). (VCF 1.0 is the initial feed solution prior to the start of the experiment).

Table 4 HA organics fouling pattern under MD operation with different physiochemical conditions.

HA solution Flux decline% Organic residue mass (mg m�2) Feed solution hydrophilic compound fraction (mg L�1)

Humic substances LMW organics

HA feed I 50.4 423.2 4.61 0.74 HA feed II 65.6 694.4 2.06 2.72 HA feed III 76.7 508.3 2.27 2.01

G. Naidu et al. / Chemical Engineering Journal 262 (2015) 946–957 955

functional groups. The Ca2+ interacted with the increased concen- tration of humic substances, forming a thick cake deposit on the membrane, as established in a previous MD study [40]. The mini- mal disaggregation of humic substances in HA feed III behaved in a similar way to the humic substances in SW. This implied that in both feed solutions (HA feed III and SW), the role of inorganic Ca2+ as a binding agent influenced the fouling pattern of humic substances. This explains the lower disaggregation of humic

substances to LMW–humic organics in SW compared to HA feed I and HA feed II.

The LC-OCD chromatogram of the fouled membrane with HA feed III showed a pattern of humic substances (20–25% of foulant) together with lower MW organics (75–80% of foulant) (Fig. 7i). In comparison, the membrane foulant chromatogram of HA feed I and HA feed II did not detect the presence of humic substances. This could be attributed to the lower disaggregation of humic

956 G. Naidu et al. / Chemical Engineering Journal 262 (2015) 946–957

substances in HA feed III. The fouling pattern of feed I to feed III is summarized in Table 4.

Overall, the presence of Ca2+ binds the humic substances, reduc- ing the disaggregation to LMW–humic organics. However, this causes a thick formation of foulant deposit, resulting in significant distillate flux decline. The results suggest that distillate flux decline was not entirely related to the presence of organic compounds only. Rather, the physico-chemical state of the feed solution strongly influenced the intensity of the distillate flux decline. Sim- ilarly, the behavior of the organic compounds (humic substances) in contributing to membrane fouling development was highly influenced by the physico-chemical state of the feed solution. For instance, in feed solution with a higher temperature and under salinity, the disaggregation of humic substances to LMW–humic organics intensified in the MD operation. This resulted in the pen- etration of LMW–humic organics across the membrane. Hence, the results of this study established the crucial role of humic sub- stances in organic fouling development in MD.

3.3. Interaction of organics and biofouling in MD

3.3.1. Biofouling potential with SW The AOC concentration in the SW was 105.7 ± 3.9 lg-C glu-

cose equivalents L�1. It was increased to 143.3 ± 11.3 lg-C glu- cose equivalents L�1 at the end of the experiment (VCF = 4.0). Meanwhile, an AOC concentration of 384.7 ± 8.4 lg-C glu- cose equivalents m�2 was detected on the membrane. The presence of AOC concentration in both the SW feed solution and membrane foulant reflected the evidence of biofouling potential when SW was treated with DCMD. In SW, humic substances are a prevalent organic compound along with LMW organics. To establish the role of humic substances in SW as a biofoulant precursor, the AOC con- centration of the synthetic HA feed solutions (containing high HA concentrations) are evaluated in the following section.

3.3.2. Biofouling potential with HA feed solution All HA feed solutions showed marginal presence of AOC in the

range of 1.7 ± 0.2–3.8 ± 1.6 lg-C glucose equivalents L�1 at initial condition (VCF 1.0). The results indicated that humic substances in a feed solution (as it is) do not contribute to AOC concentration. Meanwhile, at VCF 4.0, in final feed solution (after 240 min of MD operation at 70 �C), the AOC concentration of the HA feed solutions increased in the range of 21.4 ± 1.6–58.4 ± 0.5 lg-C glucose equiv- alents L�1. These results indicated that in the synthetic HA feed solution when the humic substances disaggregated into LMW– humic organics, AOC compounds were generated.

3.3.3. AOC concentration on membrane The AOC concentration on membrane after DCMD operation

with the synthetic HA feed solution was analyzed to assess the amount of biofoulant precursor that has been deposited. Results showed that the AOC concentration for the synthetic HA compound of feed II was 17.4 ± 4.2 � 103 lg-C glucose equivalents m�2. It implied that humic substances in the MD operation were disaggre- gated into LMW–humic organics and adhered on the membrane, which led to the AOC concentration on the membrane foulant. In feed III containing humic substances and CaSO4, the organic fouling analysis showed that much less disaggregation of humic substances with Ca2+ binding occurred, resulting in minimal adhesion of LMW– humic organics on the membrane (4.7 ± 0.1 � 103 lg- C glucose equivalents m�2).

In summary, the AOC analysis indicated that humic substances themselves do not lead to biofouling. However, in the context of MD thermal operation, the humic substances were disaggregated into LMW–humic organics and led to biofouling. Further still, the membrane foulant with the HA feed solution as well as the SW pre-

dominantly consisted of LMW organics and LMW–humic organics. This highlighted the finding that HA compounds in MD are poten- tial biofouling precursors.

4. Conclusion

DCMD organic fouling development in the context of seawater desalination was analyzed in this study. DCMD performance with SW resulted in a 31.9% distillate flux decline which is mainly attributed to LMW organics penetration through the membrane pores. A small quantity of LMW organic foulant was deposited on the membrane (43.5 mg-organic mass m�2). The MD operation with SW caused irreversible membrane fouling after two operation cycles.

Organic fouling with a synthetic HA compound was influenced by temperature, salinity (NaCl) and inorganic scalant (CaSO4). Experiments with synthetic HA feed solution showed that the ther- mal disaggregation of humic to LMW–humic organics (at Ft = 70 �C and 50 �C) led to a significant distillate flux decline of 56.2%. The LMW–humic organics on the membrane foulant were also high at 423.2 mg-organic mass m�2. Likewise, with SW, larger amounts of LMW organics on the MD membrane were observed when tem- perature increased. Under saline condition, the disaggregation of humic substances to LMW–humic organics was more evident (65.6% distillate flux decline with 694.4 mg-organic mass m�2). The presence of inorganic scalant in the HA feed solution led to a 76.7% distillate flux decline due to the binding of humic substances with Ca2+. In both SW and the HA membrane foulants, the similar presence of large amounts of LMW organics was observed. This established the important role humic substances play as organic foulants in MD when they disaggregate to LMW–humic organics. Meanwhile, humic substances thermally disaggregate to LMW– humic organics leading to higher AOC concentration. The AOC con- centration detected on the MD membrane foulant with SW high- lighted the biofouling challenge of MD with SW.

Nevertheless, it is worth highlighting that with a suitable pre- treatment the MD fouling propensity with seawater can be reduced. Our previous studies have established the effectiveness of pretreatments such as granular activated carbon (GAC) biofilter and submerged membrane adsorption bioreactor (SMABR) for sea- water organic reduction [12,22,41]. These pretreatment options are suitable for seawater MD application as they can reduce humic substances and LMW organics by adsorption and biodegradation. This is in line with the results highlighted in this study on the pre- valent role of humic substances and LMW–humic organics as organic foulants in seawater MD application. Furthermore, these pretreatments are sustainable options as chemical-free systems with low energy requirements. The performance of DCMD system with SMABR pretreated SW feed was carried out and it showed promising results on MD membrane organic fouling reduction. The DOC concentration of the SW (without pretreatment) at 1.85 mg L�1 and the DOC concentration of pretreated SW reduced to 0.90 mg L�1 by SMABR. Especially, the SMABR significantly low- ered the concentration humic substances and LMW organics to 40.2% and 63.7%, respectively than SW (without pretreatment). In terms of distillate flux, a longer stabilization period was achieved with the pretreated SW feed, prior to a 30% distillate flux decline at around VCF 3.6. Comparatively, with SW feed (without pretreat- ment), the distillate flux decline occurred from VCF 1.4 onwards. Further, at VCF 4.0, the permeate DOC concentration lowered to 45% with the pretreated SW in comparison to the SW (without pre- treatment). This indicated the reduced permeation of organics to the distillate side with the pretreated SW. The SEM-EDS line anal- ysis also found that membrane operated with the pretreated SW showed lower organic penetration (C element) compared to mem-

G. Naidu et al. / Chemical Engineering Journal 262 (2015) 946–957 957

brane operated with untreated SW. In terms of fouling reversibil- ity, physical cleaning of the pretreated SW membrane with DI water, restored its contact angle to a value 137.1 ± 1.1�, close to the MD virgin membrane (139.9 ± 1.2�). The result supports organic pretreatment as a suitable mitigation strategy for DCMD organic fouling.

Acknowledgement

The authors acknowledge the financial support of the National Centre of Excellence in Desalination Australia (Project number: 08284) which is funded by the Australian Government through the Water for the Future initiative.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.cej.2014.10.060.

References

[1] F. Banat, N. Jwaied, Economic evaluation of desalination by small-scale autonomous solar-powered membrane distillation units, Desalination 220 (2008) 566–573.

[2] S. Al-Obaidani, E. Curcio, F. Macedonio, G. Di Profio, H. Al-Hinai, E. Drioli, Potential of membrane distillation in seawater desalination: thermal efficiency, sensitivity study and cost estimation, J. Membr. Sci. 323 (2008) 85–98.

[3] G. Zuo, R. Wang, R. Field, A.G. Fane, Energy efficiency evaluation and economic analyses of direct contact membrane distillation system using Aspen Plus, Desalination 283 (2011) 237–244.

[4] S.T. Hsu, K.T. Cheng, J.S. Chiou, Seawater desalination by direct contact membrane distillation, Desalination 143 (2002) 279–287.

[5] K. He, H.J. Hwang, M.W. Woo, I.S. Moon, Production of drinking water from saline water by direct contact membrane distillation (DCMD), J. Ind. Eng. Chem. 17 (2011) 41–48.

[6] M.M.A. Shirazi, A. Kargari, D. Bastani, L. Fatehi, Production of drinking water from seawater using membrane distillation (MD) alternative: direct contact MD and sweeping gas MD approaches, Desalin. Water Treat. 52 (2014) 2372– 2381.

[7] F. He, J. Gilron, H. Lee, L. Song, K.K. Sirkar, Potential for scaling by sparingly soluble salts in crossflow DCMD, J. Membr. Sci. 311 (2008) 68–80.

[8] L.D. Nghiem, T. Cath, A scaling mitigation approach during direct contact membrane distillation, Sep. Purif. Technol. 80 (2011) 315–322.

[9] M. Gryta, M. Tomaszewska, J. Grzechulska, A.W. Morawski, Membrane distillation of NaCl solution containing natural organic matter, J. Membr. Sci. 181 (2001) 279–287.

[10] M. Khayet, A. Velázquez, J.I. Mengual, Direct contact membrane distillation of humic acid solutions, J. Membr. Sci. 240 (2004) 123–128.

[11] S. Srisurichan, R. Jiraratananon, A.G. Fane, Humic acid fouling in the membrane distillation process, Desalination 174 (2005) 63–72.

[12] S. Jeong, S. Kim, L. Hee Kim, M. Seop Shin, S. Vigneswaran, T.V. Nguyen, Foulant analysis of a reverse osmosis membrane used pretreated seawater, J. Membr. Sci. 428 (2013) 434–444.

[13] D. Jermann, W. Pronk, S. Meylan, M. Boller, Interplay of different NOM fouling mechanisms during ultrafiltration for drinking water production, Water Res. 41 (2007) 1713–1722.

[14] W. Yuan, A.L. Zydney, Humic acid fouling during microfiltration, J. Membr. Sci. 157 (1999) 1–12.

[15] S. Hong, M. Elimelech, Chemical and physical aspects of natural organic matter (NOM) fouling of nanofiltration membranes, J. Membr. Sci. 132 (1997) 159– 181.

[16] A. Seidel, M. Elimelech, Coupling between chemical and physical interactions in natural organic matter (NOM) fouling of nanofiltration membranes: implications for fouling control, J. Membr. Sci. 203 (2002) 245–255.

[17] C. Jucker, M.M. Clark, Adsorption of aquatic humic substances on hydrophobic ultrafiltration membranes, J. Membr. Sci. 97 (1994) 37–52.

[18] M. Drastik, F. Novak, J. Kucerik, Origin of heat-induced structural changes in dissolved organic matter, Chemosphere 90 (2013) 789–795.

[19] X. Jin, A. Jawor, S. Kim, E.M.V. Hoek, Effects of feed water temperature on separation performance and organic fouling of brackish water RO membranes, Desalination 239 (2009) 346–359.

[20] S. Meylan, F. Hammes, J. Traber, E. Salhi, U. von Gunten, W. Pronk, Permeability of low molecular weight organics through nanofiltration membranes, Water Res. 41 (2007) 3968–3976.

[21] S. Jeong, G. Naidu, S. Vigneswaran, C.H. Ma, S.A. Rice, A rapid bioluminescence- based test of assimilable organic carbon for seawater, Desalination 317 (2013) 160–165.

[22] G. Naidu, S. Jeong, S. Vigneswaran, S.A. Rice, Microbial activity in biofilter used as a pretreatment for seawater desalination, Desalination 309 (2013) 254–260.

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

[24] Z. Li, V. Yangali-Quintanilla, R. Valladares-Linares, Q. Li, T. Zhan, G. Amy, Flux patterns and membrane fouling propensity during desalination of seawater by forward osmosis, Water Res. 46 (2012) 195–204.

[25] M.D. Kennedy, H.K. Chun, V.A. Quintanilla Yangali, B.G.J. Heijman, J.C. Schippers, Natural organic matter (NOM) fouling of ultrafiltration membranes: fractionation of NOM in surface water and characterisation by LC-OCD, Desalination 178 (2005) 73–83.

[26] S. Velten, D.R.U. Knappe, J. Traber, H. Kaiser, U. von Gunten, M. Boller, et al., Characterization of natural organic matter adsorption in granular activated carbon adsorbers, Water Res. 45 (2011) 3951–3959.

[27] G. Naidu, S. Jeong, S. Vigneswaran, Influence of feed/permeate velocity on scaling development in a direct contact membrane distillation, Sep. Purif. Technol. 125 (2014) 291–300.

[28] M. Khayet, Membranes and theoretical modeling of membrane distillation: a review, Adv. Colloid Interface Sci. 164 (2011) 56–88.

[29] M.M.A. Shirazi, A. Kargari, M. Tabatabaei, Evaluation of commercial PTFE membranes in desalination by direct contact membrane distillation, Chem. Eng. Process. 76 (2014) 16–25.

[30] M.M.A. Shirazi, A. Kargari, M. Tabatabaei, A.F. Ismail, T. Matsuura, Concentration of glycerol from dilute glycerol wastewater using sweeping gas membrane distillation, Chem. Eng. Process. 78 (2014) 58–66.

[31] H. Oh, Y. Choung, S. Lee, J. Choi, T. Hwang, J.H. Kim, Scale formation in reverse osmosis desalination: model development, Desalination 238 (2009) 333–346.

[32] S. Meng, Y. Ye, J. Mansouri, V. Chen, Fouling and crystallisation behaviour of superhydrophobic nano-composite PVDF membranes in direct contact membrane distillation, J. Membr. Sci. 463 (2014) 102–112.

[33] M. Shirazi, A. Kargari, S. Bazgir, M. Tabatabaei, M. Shirazi, M. Abdullah, et al., Characterization of electrospun polystyrene membrane for treatment of biodiesel’s water-washing effluent using atomic force microscopy, Desalination 329 (2013) 1–8.

[34] M. Gryta, Fouling in direct contact membrane distillation process, J. Membr. Sci. 325 (2008) 383–394.

[35] K. Chon, S.J. Kim, J. Moon, J. Cho, Combined coagulation-disk filtration process as a pretreatment of ultrafiltration and reverse osmosis membrane for wastewater reclamation: an autopsy study of a pilot plant, Water Res. 46 (2012) 1803–1816.

[36] T. Tran, B. Bolto, S. Gray, M. Hoang, E. Ostarcevic, An autopsy study of a fouled reverse osmosis membrane element used in a brackish water treatment plant, Water Res. 41 (2007) 3915–3923.

[37] N.E. Palmer, R. von Wandruszka, Dynamic light scattering measurements of particle size development in aqueous humic materials, Fresenius J. Anal. Chem. 371 (2001) 951–954.

[38] A. Piccolo, The supramolecular structure of humic substances, Soil Sci. 166 (2001) 810–832.

[39] M. Baalousha, M. Motelica-Heino, P. Le Coustumer, Conformation and size of humic substances: effects of major cation concentration and type, pH, salinity, and residence time, Colloids Surf. A 272 (2006) 48–55.

[40] E. Curcio, X. Ji, G. Di Profio, A.O. Sulaiman, E. Fontananova, E. Drioli, Membrane distillation operated at high seawater concentration factors: role of the membrane on CaCO3 scaling in presence of humic acid, J. Membr. Sci. 346 (2) (2010) 263–269.

[41] S. Jeong, G. Naidu, S. Vigneswaran, Submerged membrane adsorption bioreactor as a pretreatment in seawater desalination for biofouling control, Bioresour. Technol. 141 (2013) 57–64.