wat e r r e s e a r c h 4 5 ( 2 0 1 1 ) 2 1 1 1e2 1 2 1

Avai lab le a t www.sc iencedi rec t .com

journa l homepage : www.e lsev ie r . com/ loca te /wat res

Characterization of organic membrane foulants ina submerged membrane bioreactor with pre-ozonationusing three-dimensional excitationeemission matrixfluorescence spectroscopy

Ting Liu, Zhong-lin Chen*, Wen-zheng Yu, Shi-jie You

State Key Laboratory of Urban Water Resources and Environments (SKLUWRE), School of Municipal and Environmental Engineering,

Harbin Institute of Technology, No. 73 Huanghe Road, Nangang District, Harbin 150090, PR China

a r t i c l e i n f o

Article history:

Received 19 July 2010

Received in revised form

25 October 2010

Accepted 22 December 2010

Available online 1 January 2011

Keywords:

Organic membrane foulants

Fluorescence spectroscopy

Pre-ozonation

Membrane bioreactor

* Corresponding author. Tel./fax: þ86 451 86E-mail address: zhonglinchen@163.com (

0043-1354/$ e see front matter ª 2010 Elsevdoi:10.1016/j.watres.2010.12.023

a b s t r a c t

This study focuses onorganicmembrane foulants ina submergedmembranebioreactor (MBR)

process with pre-ozonation compared to an individual MBR using three-dimensional excita-

tioneemission matrix (EEM) fluorescence spectroscopy. While the influent was continuously

ozonated at a normal dosage, preferable organic matter removal was achieved in subsequent

MBR, and trans-membranepressure increased at amuch lower rate than that of the individual

MBR. EEM fluorescence spectroscopy was employed to characterize the dissolved organic

matter (DOM) samples, extracellular polymeric substance (EPS) samples and membrane fou-

lants. Four main peaks could be identified from the EEM fluorescence spectra of the DOM

samples in bothMBRs. Two peakswere associatedwith the protein-like fluorophores, and the

other ones were related to the humic-like fluorophores. The results indicated that pre-ozon-

ation decreased fluorescence intensities of all peaks in the EEM spectra of influent DOM

especially for protein-like substances and caused red shifts of all fluorescence peaks to

different extents. The peak intensities of the protein-like substances represented by Peak T1

and T2 in EPS spectra were obviously decreased as a result of pre-ozonation. Both external and

internal fouling could be effectively mitigated by the pre-ozonation. The most primary

component of external foulants was humic acid-like substance (Peak C ) in the MBR with pre-

ozonationandprotein-like substance (PeakT1) in the individualMBR, respectively.Thecontent

decrease of protein-like substances and structural change of humic-like substances were

observed inexternal foulants fromEEMfluorescence spectra due topre-ozonation.However, it

could be seen that ozonation resulted in significant reduction of intensities but little location

shift of all peaks in EEM fluorescence spectra of internal foulants.

ª 2010 Elsevier Ltd. All rights reserved.

1. Introduction Gray et al., 2008). As the filtration process continues,

Low-pressure, hollow-fiber membrane filtration has been

generally accepted as the most promising technology for

surface water purification in recent years (Huang et al., 2007;

283028.Z.-l. Chen).ier Ltd. All rights reserved

a submerged membrane filtration system becomes

a membrane bioreactor (MBR) system because of accumula-

tion of microorganism and organic substances in raw water.

Thus, a submerged MBR system, which combines membrane

.

wat e r r e s e a r c h 4 5 ( 2 0 1 1 ) 2 1 1 1e2 1 2 12112

rejection with microorganism biodegradation in a single tank,

was introduced for treatment of drinking water by Li and Chu

(2003). Afterward, much work has been done on the use of

MBR or MBR-coupled technologies for drinking water treat-

ment (Tian et al., 2008; Sagbo et al., 2008).

Recently, an increasing attention has been paid to MBR

process with pre-ozonation used for treating surface water. It

was reported that MBR, especially containing powdered acti-

vated carbon, could be effectively used to remove total alde-

hydes, assimilable organic carbon (AOC) and biodegradable

dissolved organic carbon (BDOC) from pre-ozonated water

(Williams and Pirbazari, 2007; Treguer et al., 2010). Although

the removal of contaminants can be improved by ozone

oxidation, a better understanding of the influence of pre-

ozonation on membrane fouling process is still needed. The

relative contribution of dissolved organic matter (DOM) to

membrane fouling has been proved in the range of 26e52% in

MBRs (Bouhabila et al., 2001; Lee et al., 2003). Moreover,

extracellular polymeric substances (EPS), which are produced

by bacteria, are reported as one of the factors causing

membrane fouling in MBRs (Chang et al., 2001; Drews et al.,

2006). Therefore, an insight into the impact of pre-ozonation

on DOM and EPS is helpful to develop efficient strategies for

membrane fouling control.

Three-dimensional excitationeemission matrix (EEM)

fluorescence spectroscopy has been successfully utilized to

obtain the structural information of organic substances at

a relatively low concentration (Chen et al., 2003; �Swietlik and

Sikorska, 2004b; Gone et al., 2009). It can be used as a simple

and sensitive technique to capture specific fluorescence

features which correspond to humic- and protein-like mate-

rials in a single matrix in terms of fluorescence intensities

(Hudson et al., 2007). Therefore, EEM fluorescence spectros-

copy was employed to investigate the componential differ-

ences of DOM and EPS between the anoxic and oxic phases of

an MBR process (Wang et al., 2009). EEM fluorescence spec-

troscopy was also applied to monitor the performance of pre-

treatment stages of membrane systems (Peiris et al., 2010b)

and identify the oxidation-induced structural changes of DOM

fractions of a filtered river water (Zhang et al., 2008).

The great potential of EEM fluorescence spectroscopy is

noticed for analysis of membrane foulants. Wang et al. (2009)

found that the dominant fluorescence substances in gel layer

(mainly caused by soluble microbial byproduct, colloids,

solutes, etc.) on membrane surface of MBR were protein-like

substances that might be due to the retention of proteins by

the fine pores of the membrane. In addition, Kimura et al.

(2009) demonstrated that EEM fluorescence spectra could be

an effective analytical tool for the investigation of physically

irreversible foulants in MBRs under different solid retention

times. Peiris et al. (2010a) combined principal component

analysis and fluorescence EEM measurements to characterize

three membrane foulant fractions in the loosely attached

foulants and chemically extracted foulants during UF of

natural riverwater. According to a review byMeng et al. (2010),

membrane fouling mainly results from the accumulation of

retained substances on the membrane surface (i.e., external

fouling) and the deposition of substances in membrane pores

(i.e., internal fouling). Some studies reported that external

fouling or foulant layer formation is the major cause of

membrane fouling in MBRs (Lee et al., 2001; Meng et al., 2007).

On the other hand, the internal fouling or pore-blocking can

lead to the formation of irreversible fouling, which is harmful

for the long-term operation of MBRs (Meng et al., 2010). Thus,

if external and internal foulants are both taken into consid-

eration and analyzed using EEM fluorescence spectroscopy, it

will be of great significance to give an insight into the fouling

behavior in membrane-based water treatment processes.

The aim of the present work is to characterize the organic

membrane foulants in a submerged MBR with pre-ozonation

compared to an individual MBR by EEM fluorescence spec-

troscope. The DOM and EPS samples, which are closely related

with membrane fouling, were also analyzed by the EEM fluo-

rescence technology. The external and internal foulants in

both MBRs were identified and the comparison between them

was conducted to contribute to a better understanding of

membrane fouling in MBR processes.

2. Materials and methods

2.1. Experimental set-up

An individual MBR process without pre-ozonation (denoted as

MBR-A) and an identical MBR process with pre-ozonation

(denoted as MBR-B) were operated in parallel in this study. A

schematic illustration of MBR-B is shown in Fig. 1. The hollow-

fiber UF membrane modules (Litree, China) were made of

polyvinyl chloride (PVC) with a nominal pore size of 0.01 mm

and a total surface area of 0.1 m2. The raw water was fed into

a constant-level tank to manipulate the water head for the

subsequent units. Ozone gas generated from an ozone gener-

ator (DHX-1, Jiujiu, China) was continuously bubbled into the

water through a porous glass plate in anozone contact column.

A gas-phase ozone monitor was connected to a side stream

from the generator to measure the ozone concentration. Pre-

ozonated feedwater was then supplied to MBR-B from a reten-

tion column for further reaction of residual ozone to prevent its

impact on microorganism in MBR (Li et al., 2006). The effluent

was collected directly from themembranemodule by a suction

pump, and a manometer was fixed between the membrane

module and the suction pump tomonitor the trans-membrane

pressure (TMP). To supply the oxygen for microbial respiration

and turbulence for membrane surface cleaning, continuous

aeration was provided at the bottom of bioreactor. The exper-

imental set-up ofMBR-Awas the same as that of MBR-B except

for the absence of ozonation unit.

2.2. Simulated raw water supply

The raw water was prepared in a way similar to that used by

Tian et al. (2008). Domestic sewage was added to the local tap

water (Harbin, China) of a volumetric ratio of 1:30 to simulate

the surface water supply slightly polluted by sewage

discharge. 1 mg/L of humic acid (Jufeng, Shanghai, China) was

also added to the raw water. The synthesized raw water was

then stabilized for 24 h at room temperature before use.

During the experiment, the raw water was kept at a tempera-

ture in the range of 15.5e18.3 �C and the pH in the range of

7.1e7.4; other water quality parameters are listed in Table 1.

water line

air line

ozone gas line

Fig. 1 e Schematic diagram of MBR-B (1-feed pump; 2-high level tank; 3-constant level tank; 4-ozone contact column;

5-retention column; 6-bioreactor; 7-membrane module; 8-manometer; 9-effluent pump; 10-backwash pump; 11-ozone

generator; 12-gas phase ozone analyzer; 13-ozone gas flowmeter; 14-ozone destruction unit; 15-air blower; 16-air

flowmeter; 17-air diffuser; 18-sludge discharge valve).

wat e r r e s e a r c h 4 5 ( 2 0 1 1 ) 2 1 1 1e2 1 2 1 2113

2.3. Operating conditions

MBRs with each effective volume of 1 L were conducted in

a dead-end filtration mode at constant flux. Membrane flux

was predetermined at a relatively low value of 10 L/(m2h),

which corresponded to a hydraulic retention time (HRT) of

1.0 h. A 30-min operational cycle for suction followed by 1-min

backwashing with the effluent was controlled by a timer. The

ratio of air to influent was kept at 20:1 (V:V), and sludge

retention time (SRT) was maintained at 40 d. The two MBRs

fedwith rawwater had been operated stably for threemonths,

so that the biomass was accumulated in the reactors for

acclimation to the water. Then the mixed liquid of the two

bioreactors was mixed and shared between them to have the

same initial condition. The MBRs were operated continuously

again with new membranes and pre-ozonation process was

applied to MBR-B at ozone dosage of 1.5 mg/L-raw water. The

HRT of the ozone contact column and the retention column

was 15 min and 20 min respectively. The experiments were

carried out under normal operating conditions (e.g., ozone

dosage and reaction time) commonly adopted in water plants.

Table 1 e Pollutants removal efficiencies of MBR-A and MBR-B

Water qualityindexs

Raw water Afterpre-ozonation

Effl

Turbidity (NTU) 2.31 � 1.00 2.09 � 0.60 0.07

CODMn (mg/L) 4.25 � 0.27 3.15 � 0.22 2.30

DOC (mg/L) 6.020 � 0.784 5.497 � 0.572 4.336

UV254 (cm�1) 0.077 � 0.003 0.046 � 0.002 0.059

2.4. Extraction of EPS from the mixed liquid

EPS were extracted from the mixed liquid in the MBR

according to the thermal treatment method (Chang and Lee,

1998). The mixed liquid was centrifuged for 30 min at

3200 rpm to remove the bulk solution. After the supernatant

was discarded, the remaining pellet was washed and resus-

pended with saline water (0.9% NaCl solution). The mixed

liquid was then subjected to heat treatment (100 �C, 1 h) and

centrifuged again under the same operating conditions. The

centrifuged supernatant was EPS solution, which was filtered

through a 0.45 mm acetate fiber membrane and used for EEM

fluorescence analysis.

2.5. Extraction of foulants

At the end of operation, the fouled membrane modules were

taken out from the reactors when the TMP exceeded 35 kPa.

The external foulants on membrane surface were carefully

scraped off with a plastic sheet (Deli, China) and simulta-

neously flushed with deionized (DI) water. The collected

.

MBR-A MBR-B

uent Totalremoval (%)

Effluent Totalremoval(%)

� 0.01 97.0 � 1.3 0.07 � 0.01 97.0 � 1.2

� 0.28 45.9 � 3.7 1.60 � 0.20 62.4 � 3.3

� 0.603 28.0 � 6.3 3.542 � 0.568 41.2 � 6.2

� 0.002 23.4 � 2.8 0.036 � 0.002 53.2 � 3.0

40

45

50

10

12

wat e r r e s e a r c h 4 5 ( 2 0 1 1 ) 2 1 1 1e2 1 2 12114

sample was fully mixed using a magnetic blender (JB-2, Leici,

China) at 200 rpm for 1 h, it was then filtered through a 0.45 mm

acetate fiber membrane for EEM fluorescence analysis (Wang

et al., 2009). After the membrane surface was wiped with

a sponge, 0.01 mol/L NaOH was used for extraction of internal

foulants and the fibers were soaked for 24 h at 20 �C in the

alkaline solution according to the method described by

Kimura et al. (2009).

2.6. Analytical methods

2.6.1. Water quality analysisWater quality analysis was conducted following the standard

methods (APHA, 1998). Turbidity was monitored by a turbi-

dimeter (Turbo550, WTW, Germany). The dissolved organic

carbon (DOC) was measured by a total organic carbon (TOC)

analyzer (TOC-VCPH, Shimadzu, Japan). CODMn was analyzed

by the potassium permanganate oxidation methods. UV

absorbance at 254 nm (UV254) was determined by using

a spectrometer (T6, Puxi, China).

2.6.2. Three-dimensional excitationeemission matrix (EEM)fluorescence spectroscopyFluorescence measurements were conducted using a spectro-

fluorometer (FP-6500, Jasco, Japan) equipped with a 150 W

xenon lamp at ambient temperature of 24 �C. A 1-cm quartz

cuvette with four optical windows was used for the analyses.

Emission scans were performed from 220 to 550 nm at 5 nm

steps, with excitationwavelengths from 220 to 450 nmat 5 nm

intervals. The detector was set to high sensitivity, and the

scanning speed wasmaintained at 2000 nm/min in this study;

the slit widths for excitation and emission were 5 nm and

3 nm respectively. Under the same conditions, fluorescence

spectra for Milli-Q water were subtracted from all the spectra

to eliminate water Raman scattering and to reduce other

background noise. During the course of fluorescence analysis,

the Raman scattering peak intensity for Milli-Q water (exci-

tation at 350 nm, emission at 400 nm) was recorded as

a standard to verify the instrument stability. Mean intensity of

the Raman peak was 36.10 units and the differences were less

than 2%, confirming that there were no significant fluctua-

tions in the performance of the spectrofluorometer

throughout the experimental period. The EEM spectra are

plotted as the elliptical shape of contours. The X-axis repre-

sents the emission spectra while the Y-axis represents the

excitation wavelength, and the third dimension, i.e., the

contour line, is given to express the fluorescence intensity.

0 10 20 30 40 500

5

10

15

20

25

30

35

Time (day)

TMP

(kpa

)

0

2

4

6

8 TMP (MBR-A) TMP (MBR-B) Flux (MBR-A) Flux (MBR-B)

Flux (L/(m2h))

Fig. 2 e Comparison of TMP and membrane flux variations

in MBR-A and MBR-B.

3. Results and discussion

3.1. Process performance

As shown in Table 1, the turbidity was reduced from

2.31� 1.00 NTU to a level as low as 0.07� 0.01 NTU for the two

MBRs, which indicated an excellent performance of particle

removal. The organic matter removal efficiencies in terms of

CODMn, DOC and UV254 in both MBR-A and MBR-B processes

are summarized in Table 1. The results showed that the

remarkably improved performance in organic matter removal

was ascribed to pre-ozonation implementation in MBR-B.

Especially, an approximately 30% higher UV254 decrease was

achieved in MBR-B (53.2 � 3.0%) compared to that in MBR-A.

Three functions were identified for the contributions to

removal of organic contaminants in MBR-B: partial degrada-

tion or complete mineralization by ozone oxidation, physical

retention by UF membrane, and biodegradation by active

biomasswithin the reactor. It should be noted that the organic

matter removalwas attributed to the synergetic effect of these

three functions.

It can be seen from Fig. 2 that TMP increased with opera-

tion time while the membrane flux was maintained constant

at about 10 L/(m2h) during the experiment before TMP

exceeded 30 kpa. As a two-step fouling phenomenon, the TMP

variations exhibited a slow increase followed by a rapid

increase. The TMP gradually increased with time from the

initial 5 kPa for both systems with a similar trend within the

beginning phase (0e8 days). However, there was a distinct gap

of 1.5 kPa between them on Day 16. The permeate flux obvi-

ously declined when the TMP of MBR-A reached 35 kPa, and

this operation process came to an end for a further analysis of

membrane foulants. The final TMP of MBR-B only increased

to 23.5 kPa, which was 12.5 kPa lower than that of MBR-A

(36 kPa). It was thus believed that ozone pre-oxidation was an

effective pre-treatment strategy to reduce the increasing rate

of TMP to lower energy requirement for the membrane

filtration process at a constant flux. In order to identify the

proportions of external and internal fouling resistances, the

membranes were taken out from the reactors at the end of

experiment and the external foulants were removed from

membrane surface. The membrane modules were reinstalled

into the bioreactors and internal fouling resistances were

evaluated. The TMP inMBR-A andMBR-Bwere then decreased

to 12 kPa and 9 kPa respectively. Compared to those of 77.4%

and 22.6% inMBR-A, the contributions of external and internal

fouling resistances to TMP development were 78.4% and 21.6%

in MBR-B. Therefore, it can be concluded that the proportions

of external and internal fouling resistances were similar in

both MBRs, suggesting that ozone pre-oxidation was able to

alleviate both of the two kinds of membrane fouling.

wat e r r e s e a r c h 4 5 ( 2 0 1 1 ) 2 1 1 1e2 1 2 1 2115

3.2. EEM fluorescence spectra of DOM in the two systems

Three-dimensional EEM fluorescence spectroscopy has been

successfully utilized to identify the chemical composition of

DOM because of its ability to distinguish among certain

classes of organic matter in natural waters (Saadi et al., 2006).

In UF application, the major membrane foulant is natural

organic matter, which contains a complex mixture of humic

and fulvic acids, proteins, carbohydrates of various molecular

size and functional groups (Her et al., 2003; Saravia et al.,

2006). There are five key fluorescence peaks referred to as

fluorophores A, C, T1, T2 and B commonly observed in fresh-

water samples (Coble, 1996; Baker, 2001). Peak A and C are

related to humic-like substance derived from the breakdown

of plant material (Lee et al., 2008); protein-like fluorophores

including tryptophan-like (Peak T ) and tyrosine-like (Peak B)

materials, are usually detected at enhanced levels in water

impacted by domestic sewage (Baker et al., 2003). As shown in

Fig. 3, Peak B has relatively lower fluorescence intensity for the

DOM samples, so the other four peaks including Peak A, C, T1

and T2 which are distinctly identifiedwere investigated in this

section. The fluorescence parameters including peak loca-

tions, fluorescence intensity, and different peak intensity

Fig. 3 e EEM fluorescence spectra of (a) the influent (raw water) a

ozonation) and (d) the effluent DOM of MBR-B.

ratios were extracted from EEM fluorescence spectra and

summarized in Table 2, which could be employed for quan-

titative analysis.

Generally, intensity reduction of the fluorescence peak

between raw water and treated water is an indication for

degradation of fluorescingmaterial. It can be seen fromTable 2

that ozonation approximately decreased the fluorescence

intensities of PeakA and C by 30e40% and those of Peak T1 and

T2 by 60e70% for DOM in raw water. Consequently, the fluo-

rescence intensities of Peak A and C in EEM spectra of MBR-B

effluent were nearly the same percentage lower than those of

MBR-A effluent.Meanwhile, the intensities of Peak T1 and T2 in

EEM spectra of MBR-B effluent were 40% and 53% lower than

those of MBR-A effluent. The peak intensity ratios are shown

as ratios to Peak C, as this component is considered to be

present in a wide range of water environments (Henderson

et al., 2009). Presented in this way, the data reflect the differ-

ences in composition rather than the considerable differences

in concentration. Peak T1 and C in EEM spectra of DOM

samples, which indicate protein- and humic acid-like

substance respectively, can be referred to as biodegradable

and nonbiodegradable DOM (Reynolds, 2002; Wang et al.,

2009). Since the feedwater synthesized to simulate the

nd (b) the effluent DOM of MBR-A, (c) the influent (after pre-

Table 2 e Fluorescence spectral identifications of DOM samples in MBR-A and MBR-B.

Process Samples Peak A Peak C Peak T1 Peak T2 Peak Int. ratio

Ex/Em Int. Ex/Em Int. Ex/Em Int. Ex/Em Int. A/C T1/C C/T2

MBR-A Influent (Raw) 240/405 311.7 310/420 298.3 280/340 283.2 225/335 535.7 1.04 0.95 0.56

Effluent 245/420 304.5 315/415 199.0 270/340 119.5 230/345 206.9 1.53 0.60 0.96

MBR-B Influent (Ozonated) 250/420 230.5 325/415 170.7 280/345 115.0 230/340 154.2 1.35 0.67 1.11

Effluent 255/425 213.7 330/425 154.3 275/345 71.2 230/345 96.5 1.38 0.46 1.60

Int.: intensity.

Fig. 4 e Normalized intensities of EEM fluorescence spectra

of DOM in raw water and pre-ozonated water on the

excitation scale (emission at 420 nm).

wat e r r e s e a r c h 4 5 ( 2 0 1 1 ) 2 1 1 1e2 1 2 12116

surface water polluted by sewage discharge had a relatively

higher biodegradability, some biodegradable DOM was

mineralized during the ozonation process. The intensity ratio

of Peak T1/Peak C decreased in both bioreactors as shown in

Table 2, suggesting that the biodegradable DOM with fluo-

rescence was gradually metabolized by microorganism. The

intensity ratio of Peak T1/Peak C of MBR-B effluent was 0.46

compared to 0.60 of MBR-A effluent. MBR-B process thus dis-

played a greater capacity for biodegradable DOM removal and

this is beneficial to biological stability of treated water and

restraint of bacterial regrowth in distribution system. The

differences in peak intensity ratios of EEM fluorescence

spectra in the two MBR systems imply that ozone oxidation is

responsible for the compositional variation of the fluorescent

compounds in DOM samples.

The location shift of fluorescence peak provides spectral

information on the chemical structure changes of DOM

samples. After ozonation, the locations of the four fluores-

cence peaks shifted toward longer wavelength (red shift) to

different extents on the emission and/or excitation scale, and

this observation is in line with Chen et al. (2002). As reported

by �Swietlik et al. (2004a), the ozonation of hydrophobic acid

(HOA) and hydrophilic neutral (HIN) produces carboxylic

acids, aldehydes and ketones, and this may cause the

formation of oxidation byproducts with double bond-con-

taining substituents. A red shift is related to the increase of

carbonyl, hydroxyl, alkoxyl, amino, and carboxyl groups in the

structures of fluorophores (Chen et al., 2002; Uyguner and

Bekbolet, 2005), while a blue shift is ascribed to the elimina-

tion of particular functional groups such as carbonyl, hydroxyl

and amine, a reduction in the degree of p-electron systems,

and the decrease in the number of aromatic rings and conju-

gated bonds in a chain structure (�Swietlik et al., 2004a). In the

two systems, the locations of Peak A and T2 in EEM spectra of

effluent DOM were all red-shifted (5e15 nm) to longer wave-

lengths than those of influent DOM, while Peak T1 of effluent

DOM showed a blue shift (5e10 nm) on the excitation scale

compared to that of influent DOM. The location of Peak C

showed different shift trends in the two systems with respect

to the emission axis. Wang et al. (2009) observed that Peak T1

and T2 respectively demonstrated a blue and red shift of the

effluent DOM compared to those of the influent DOM in a MBR

for wastewater treatment, which is in good agreement with

the results of both MBRs in this study. Furthermore, they

described that the location of Peak C of the effluent DOM was

red-shifted along the excitation axis and blue-shifted along

the emission axis. This finding agrees with the experimental

results of MBR-A but disagrees with those of MBR-B obtained

during this study. It was therefore assumed that this

difference between the two MBRs was a consequence of

structural changes of the humic-like substances in the raw

water during ozonation process. As a matter of fact, it is likely

that some of the EEM peak shifts result from the changes in

concentration of one of the several overlapping components

(Stedmon et al., 2003; Peiris et al., 2010a). For the purpose of

identifying structural changes, the EEM fluorescence intensi-

ties were normalized with respect to the highest peak inten-

sity. Fig. 4 shows the normalized intensities of EEM

fluorescence spectra of DOM in raw water and pre-ozonated

water on the excitation scale (emission at 420 nm). When the

intensity of Peak C reduced significantly due to the ozonation

process, the visual location of Peak A would have a blue shift

to a lesser degree, which was dependent on concentration

changes. Nevertheless, Peak A showed a red shift of 10 nm on

the contrary, whichmeans that the peak shift was necessarily

caused by the structural changes of this kind of humic-like

substances (Peak A). The peak locations in EEM spectra of the

effluent DOMofMBR-B also demonstrated some differences in

comparison with those of the effluent DOM of MBR-A.

3.3. EEM fluorescence spectra of EPS in mixed liquid ofthe two MBRs

As shown in Fig. 5, the intensity of Peak B in the EEM spectra of

EPS was significantly enhanced in both MBRs, suggesting that

the organic substances indicated by Peak B were closely

related withmetabolic activity of microorganism. A new peak,

a

b

Fig. 5 e EEM fluorescence spectra of EPS extracted from

(a) MBR-A and (b) MBR-B.

wat e r r e s e a r c h 4 5 ( 2 0 1 1 ) 2 1 1 1e2 1 2 1 2117

i.e., Peak D, was present at the excitation/emission wave-

lengths (Ex/Em) of 270e280/300e310 nm in EEM spectra of EPS.

According to the five regions of EEM divided by Chen et al.

(2003), Peak D indicates soluble microbial byproduct (SMP)-

like substances (Region IV). It could be seen that protein- and

SMP-like substances are dominant among fluorescent organic

matters in EPS from both MBRs. Compared with the EEM

spectra of DOM samples (Fig. 3), intensity of Peak T2 in EEM

spectra of EPS was much weaker than that of raw water,

whichmeans that the fluorescent DOM represented by Peak T2

in effluents were originated from raw water rather than EPS.

The locations of Peak A and C in EEM spectra of EPS were both

red-shifted by 15e35 nm along the two axes compared to

those in Table 2, indicating that the structure and components

of humic-like substances in EPS were different from those in

DOM samples. Sheng and Yu (2006) found three fluorescence

peaks including Peak T1, T2 and C were present in EPS spectra

from a conventional activated sludge system. Wang et al.

(2010) identified Peak T1, C, together with a new peak associ-

atedwith humic acid-like substances at the Ex/Emof 415e420/

470e475 nm in EPS spectra of an MBR. In this study, three

main peaks and three lesser peaks were observed in EPS

spectra. The organic matters indicated by Peak T1 and C are

extensively present in EPS samples extracted from various

origins. The locations of the two peaks in EPS spectra in this

study, which located at the Ex/Em of 280/350 nm and

340/435 nm, were similar to those reported by Sheng and Yu

(2006) but different from those observed by Wang et al.

(2009). The differences might be attributed to the fact that

the EPS samples were extracted from different origins and

thus the components in EPS were chemically different.

The intensity of Peak B did not show any decline in EPS

spectra of MBR-B (Fig. 5b) due to pre-ozonation compared to

that of MBR-A (Fig. 5a). It may indicate that the protein-like

substances represented by Peak B were excreted by microor-

ganism in MBRs, and pre-ozonation had little effect on the

metabolic level related to this compound. The peak intensities

of protein-like substances represented by Peak T1 and T2 in

EPS spectra of MBR-B were decreased significantly resulted

from pre-ozonation. Cho et al. (2005) established a functional

equation in which the specific cake resistance was propor-

tional to the EPS concentration. Ahmed et al. (2007) also

observed that as EPS concentration rose, the specific cake

resistance increased, and this consequently resulted in the

rise of TMP. These investigations showed that there is a close

relationship between EPS and the resistance of cake layer on

membrane surface. It can therefore be concluded that the

protein-like substances represented by Peak T1 and T2 in EPS

might contribute more to external fouling.

3.4. EEM fluorescence spectra of membrane foulants

The amount and composition of organic membrane foulants

were related to the interaction between organic substance and

membrane. The UF membrane used in this study is made of

Polyvinyl chloride (PVC), which is a hydrophobic material.

According to the manufacturer, doping technology is used to

improve its hydrophilicity for higher flux and some other

physical properties. The contact angle reflects the hydro-

phobic/hydrophilic character of membranes. The PVC

membrane has an average contact angle of 68 � 2� (provided

by the manufacturer). On the one hand, ozone oxidation is

able to increase the polarity and hydrophilicity of the

substances to make hydrophobic membranes less susceptible

to adsorptive fouling. In this study, membrane fouling was

reduced by pre-ozonation, which means that the PVC

membrane might still exhibit hydrophobic property. On the

other hand, ozonation reduced the amount of humic

substances because of their breakdown to lower molecular

weight (MW) compounds. Therefore, although the size of

some DOM molecules before or after pre-ozonation was

smaller than the nominal pore size of PVC membrane

(0.01 mm), biodegradation of these low-MW ozonation prod-

ucts caused less accumulation of foulants on membrane

surface and/or in membrane pores.

Unlike conventional methods such as the ratio of carbohy-

drate to protein (C/P)which is incompetent to fully characterize

membrane foulants, EEM fluorescence analysis is able to

provide more useful information on the characteristics of

organic membrane foulants (Kimura et al., 2009). Hence, both

external and internal foulants were extracted from the fouled

membranes and analyzed by using EEM fluorescence spec-

troscopy at the end of operation.

wat e r r e s e a r c h 4 5 ( 2 0 1 1 ) 2 1 1 1e2 1 2 12118

3.4.1. External foulantsFig. 6 shows EEM fluorescence spectra measured for external

foulants extracted from the fouled membranes of the two

MBRs. Peak T1 predominated the EEM fluorescence spectra of

external foulants in MBR-A, while Peak Cwas dominant in the

EEM fluorescence spectra of external foulants in MBR-B. It was

demonstrated that the external foulants were composed of

the protein-like substances represented by Peak T1 and the

humic acid-like substances represented by Peak C as the most

primary components in MBR-A and MBR-B, respectively. The

comparison in TMP development of both MBRs showed that

membrane fouling of MBR-A was more serious than that of

MBR-B. As stated in “Section 3.1”, the contribution of external

fouling resistance to TMP development reached 75e80% in the

two systems. Therefore, it can be concluded that protein-like

substances rather than humic acid-like substances contrib-

utedmore to the external fouling resistance. This is consistent

with the findings of Hong et al. (2007) and Drews et al. (2007),

who reported that proteins could induce severe membrane

fouling as one of the major components in membrane fou-

lants. The appearance of dominant protein-like peak in

external foulants in MBR-A indicated the accumulation of

protein-like substances on the membrane surface, whereas

the intensity of protein-like peak was weakened to a signifi-

cant degree in MBR-B which could be attributed to ozonation.

a

b

Fig. 6 e EEM fluorescence spectra of external foulants

extracted from (a) MBR-A and (b) MBR-B.

Thus, it was reasonable to infer that the reduced accumula-

tion of protein-like substances into gel layer may play an

important role against TMP increase in MBR-B at constant

flux.

The locations of peak T1 (Ex/Em ¼ 280/345) and peak T2

(Ex/Em ¼ 230/335) in the EEM fluorescence spectra of external

foulants from MBR-B were similar to those of external fou-

lants from MBR-A as shown in Fig. 6. However, the location of

Peak A in the EEM fluorescence spectra of external foulants of

MBR-A was red-shifted by 10 nm along the excitation axis and

blue-shifted by as much as 25 nm along the emission axis

compared to that of external foulants of MBR-B. The location

of Peak C of MBR-A external foulants was blue-shifted to

shorter wavelengths than that of MBR-B external foulants.

These observations implied that the structures of humic-like

substances represented by Peak A and C in the external fou-

lants of the two MBRs differed from each other.

In conclusion, the content decrease of protein-like

substances and the structural change of humic-like

substances were observed in external foulants from EEM

fluorescence spectra due to pre-ozonation. The study carried

out by Schlichter et al. (2003) indicated that continuous addi-

tion of ozone caused a drastic reduction in adsorption-

induced membrane fouling during the UF of humic acid

solution. Nevertheless, Her et al. (2007) reported that nano-

filtration (NF) membrane fouling increased mainly due to the

adhesive EPS released by algae upon ozonation. It may actu-

ally be attributed to the average size of NF membrane pores

which is smaller than that of UF membrane pores. Moreover,

it may also be attributed to the presence of abundant algae in

the raw water used for their study in that season, which

indicated the unsuitability for application of pre-ozonation.

They also found that ozonation showed opposite results for

humic- and protein-like substances as for UV absorbance ratio

index (UVA210/UVA254), which provides information on the

relative proportions between UV-absorbing functional groups

and unsaturated compounds in NOM. The results of our study

coincidewith the results of their investigations in this respect.

It is possible for ozonation to reduce the number of unsatu-

rated groups and form new groups to cause the structural

changes of humic substances with large molecular size.

Ozonation may also destroy the particular functional groups

of protein-like substances to thereby result in a relatively

lower level of their characteristics.

3.4.2. Internal foulantsMembrane fibers of the two MBRs with the same quantity

were chemically treated to extract the internal pore foulants

for investigation. In order to identify the difference of DOM

characteristics of internal foulants caused by ozone pre-

oxidation, analysis of EEM fluorescence spectra of internal

foulants was carried out and the results are shown in Fig. 7.

Peak C at Ex/Em¼ 440e445/275 nm predominated in both EEM

fluorescence spectra, indicating that the humic acid-like

substances represented by Peak C were the dominant

components of the internal foulants. The characteristics of

EEM fluorescence spectra of internal foulants obviously

differed from those of external foulants. The fluorescence

intensities of the four peaks of internal foulants from MBR-B

were much weaker than those of internal foulants from

a

b

Fig. 7 e EEM fluorescence spectra of internal foulants

extracted from (a) MBR-A and (b) MBR-B.

wat e r r e s e a r c h 4 5 ( 2 0 1 1 ) 2 1 1 1e2 1 2 1 2119

MBR-A, which was in line with the tendency of internal

fouling resistances indicated by TMP increase. By comparison

of the EEM fluorescence spectra of internal foulants from the

two MBRs, it could be seen that the four main peaks were

similarly locatedwith a difference of nomore than 5 nm along

the two axes.

The results indicated that the pre-ozonation process had

the potential to effectively mineralize some NOM with small

molecular size in raw water. For the NOM with relatively

higher molecular size, pre-ozonation could cleave unsatu-

rated bonds in aromatic moieties and reduce the molecular

size/weight of the substances to make them more amenable

to microbial degradation and utilization. The less quantity of

organic matter tended to deposit or adsorb into membrane

pores which induced internal fouling, suggesting that ozona-

tion conducted well as a pre-treatment process for the

biodegradation by microorganism in MBR-B. It was likely that

the structures of organic substances in internal foulants

changed slightly, and almost the same proportion of the

organic matter content decreased as a result of ozone

oxidation.

4. Conclusions

Two identical submergedMBRswith orwithout pre-ozonation

were comparatively operated to investigate performance of

the processes and characterize organic membrane foulants

using EEM fluorescence spectroscopy. It can be seen that

preferable organic matter removal was achieved in the MBR

process with pre-ozonation, and its TMP increased at a rate

much lower than that of the individual MBR. EEMfluorescence

spectroscopywas employed to characterize the DOM samples,

EPS samples and membrane foulants of both MBRs. The

results indicated that pre-ozonation decreased fluorescence

intensities of the four main peaks in the influent DOM spectra

especially for protein-like substances and caused red shifts of

all fluorescence peaks to different extents. The peak intensi-

ties of the protein-like substances represented by Peak T1 and

T2 in EPS spectra were obviously decreased as a result of pre-

ozonation. Both external and internal fouling could be effec-

tively mitigated by the pre-ozonation. The most primary

components of external foulants were humic acid-like

substance (Peak C ) in the MBR with pre-ozonation and

protein-like substance (Peak T1) in the individual MBR,

respectively. The content decrease of protein-like substances

and structural change of humic-like substances were

observed in external foulants from EEM fluorescence spectra

due to pre-ozonation. However, it could be seen that ozona-

tion resulted in significant reduction of intensities but little

change of locations of all peaks in EEM fluorescence spectra of

internal foulants. Further work is required to assess the

impact of pre-ozonation on other kinds of DOM, such as

polysaccharide substances, to extend the knowledge of

fouling control of MBR processes.

Acknowledgements

This research was funded by National High Technology

Research and Development Program of China (2007AA06Z339)

and State Key Laboratory of Urban Water Resource and Envi-

ronment (HIT, Grant No. 2010DX12).

r e f e r e n c e s

Ahmed, Z., Cho, J., Lim, B.-R., Song, K.-G., Ahn, K.-H., 2007. Effectsof sludge retention time on membrane fouling and microbialcommunity structure in a membrane bioreactor. Journal ofMembrane Science 287 (2), 211e218.

APHA, 1998. Standard Methods for the Examination of Water andWastewater, twentieth ed. American Public HealthAssociation/American Water Works Association/WaterEnvironment Federation, Washington DC, USA.

Baker, A., 2001. Fluorescence excitationeemission matrixcharacterization of some sewage-impacted rivers.Environmental Science and Technology 35 (5), 948e953.

Baker, A., Inverarity, R., Charlton, M., Richmond, S., 2003.Detecting river pollution using fluorescencespectrophotometry: case studies from the Ouseburn, NEEngland. Environmental Pollution 124 (1), 57e70.

wat e r r e s e a r c h 4 5 ( 2 0 1 1 ) 2 1 1 1e2 1 2 12120

Bouhabila, E.H., Aim, R.B., Buisson, H., 2001. Foulingcharacterisation in membrane bioreactors. Separation andPurification Technology 22e23 (1e3), 123e132.

Chang, I.S., Lee, C.H., 1998. Membrane filtration characteristics inmembrane-coupled activated sludge system-the effect ofphysiological states of activated sludge on membrane fouling.Desalination 120 (3), 221e233.

Chang, I.S., Bag, S.Q., Lee, C.H., 2001. Effects of membrane foulingon solute rejection during membrane filtration of activatedsludge. Process Biochemistry 36 (8e9), 855e860.

Chen, J., Gu, B., LeBoeuf, E.J., Pan, H., Dai, S., 2002. Spectroscopiccharacterization of the structural and functional properties ofnatural organic matter fractions. Chemosphere 48, 59e68.

Chen, J., LeBoeuf, E.J., Dai, S., Gu, B., 2003. Fluorescencespectroscopic studies of natural organic matter fractions.Chemosphere 50 (5), 639e647.

Cho, J., Song, K.G., Yun, H., Ahn, K.H., Kim, J.Y., Chung, T.H., 2005.Quantitative analysis of biological effect on membrane foulingin submerged membrane bioreactor. Water Science andTechnology 51 (6e7), 9e18.

Coble, P.G., 1996. Characterization of marine and terrestrial DOMin seawater using excitationeemission matrix spectroscopy.Marine Chemistry 51 (4), 325e346.

Drews, A., Lee, C.H., Kraume, M., 2006. Membrane fouling -a review on the role of EPS. Desalination 200 (1e3), 186e188.

Drews, A., Mante, J., Iversen, V., Vocks, M., Lesjean, B., Kraume, M.,2007. Impact of ambient conditions on SMP elimination andrejection in MBRs. Water Research 41 (17), 3850e3858.

Gone, D.L., Seidel, J.-L., Batiot, C., Bamory, K., Ligban, R., Biemi, J.,2009. Using fluorescence spectroscopy EEM to evaluate theefficiency of organic matter removal during coagulation-flocculation of a tropical surface water (Agbo reservoir).Journal of Hazardous Materials 172 (2e3), 693e699.

Gray, S.R., Ritchie, C.B., Tran, T., Bolto, B.A., Greenwood, P.,Busetti, F., Allpike, B., 2008. Effect of membrane character andsolution chemistry on microfiltration performance. WaterResearch 42 (3), 743e753.

Henderson, R.K., Baker, A., Murphy, K.R., Hambly, A., Stuetz, R.M.,Khan, S.J., 2009. Fluorescence as a potential monitoring toolfor recycled water systems: a review. Water Research 43 (4),863e881.

Her, N., Amy, G., McKnight, D., Sohn, J., Yoon, Y., 2003.Characterization of DOM as a function of MW by fluorescenceEEM and HPLC-SEC using UVA, DOC, and fluorescencedetection. Water Research 37 (17), 4295e4303.

Her, N., Amy, G., Plottu-Pecheux, A., Yoon, Y., 2007. Identificationof nanofiltration membrane foulants. Water Research 41 (17),3936e3947.

Hong, S.H., Lee, W.N., Oh, H.S., Yeon, K.M., Hwang, B.K., Lee, C.H.,Chang, I.S., Lee, S., 2007. The effects of intermittent aerationon the characteristics of bio-cake layers in a membranebioreactor. Environmental Science and Technology 41 (17),6270e6276.

Huang, H.O., Lee, N.H., Young, T., Gary, A., Lozier, J.C.,Jacangelo, J.G., 2007. Natural organic matter fouling of low-pressure, hollow-fiber membranes: effects of NOM source andhydrodynamic conditions. Water Research 41 (17), 3823e3832.

Hudson, N., Baker, A., Reynolds, D., 2007. Fluorescence analysis ofdissolved organic matter in natural, waste and polluted waters-a review. River Research and Applications 23 (6), 631e649.

Kimura, K., Naruse, T., Watanabe, Y., 2009. Changes incharacteristics of soluble microbial products in membranebioreactors associated with different solid retention times:relation tomembrane fouling.Water Research 43 (4), 1033e1039.

Lee, J.M., Ahn, W.Y., Lee, C.H., 2001. Comparison of the filtrationcharacteristics between attached and suspended growthmicroorganisms in submerged membrane bioreactor. WaterResearch 35 (10), 2435e2445.

Lee, W., Kang, S., Shin, H., 2003. Sludge characteristics and theircontribution to microfiltration in submerged membranebioreactors. Journal of Membrane Science 216 (1e2), 217e227.

Lee, E.K., Chen, V., Fane, A.G., 2008. Natural organic matter (NOM)fouling in low pressure membrane filtration-effect ofmembranes and operation modes. Desalination 218 (1e3),257e270.

Li, X.Y., Chu, H.P., 2003. Membrane bioreactor for drinking watertreatment of polluted surface water supplies. Water Research37 (19), 4781e4791.

Li, L.S., Zhu, W.P., Zhang, P.Y., Zhang, Q.Y., Zhang, Z.L., 2006. AC/O3-BAC processes for removing refractory and hazardouspollutants in raw water. Journal of Hazardous Materials 135(1e3), 129e133.

Meng, F.G., Zhang, H.M., Yang, F.L., Liu, L.F., 2007.Characterization of cake layer in submerged membranebioreactor. Environmental Science and Technology 41 (11),4065e4070.

Meng, F.G., Liao, B.Q., Liang, S., Yang, F.L., Zhang, H.M., Song, L.F.,2010. Morphological visualization, componentialcharacterization and microbiological identification ofmembrane fouling in membrane bioreactors (MBRs). Journalof Membrane Science 361 (1e2), 1e14.

Peiris, R.H., Budman, H., Moresoli, C., Legge, R.L., 2010a.Understanding fouling behaviour of ultrafiltration membraneprocesses and natural water using principal componentanalysis of fluorescence excitation-emission matrices. Journalof Membrane Science 357 (1e2), 62e72.

Peiris, R.H., Halle, C., Budman, H., Moresoli, C., Peldszus, S.,Huck, P.M., Legge, R.L., 2010b. Identifying fouling events ina membrane-based drinking water treatment process usingprincipal component analysis of fluorescence excitation-emission matrices. Water Research 44 (1), 185e194.

Reynolds, D.M., 2002. The differentiation of biodegradable andnon-biodegradable dissolved organic matter in wastewatersusing fluorescence spectroscopy. Journal of ChemicalTechnology and Biotechnology 77 (8), 965e972.

Saadi, I., Borisover, M., Armon, R., Laor, Y., 2006. Monitoring ofeffluent DOM biodegradation using fluorescence, UV and DOCmeasurements. Chemosphere 63 (3), 530e539.

Sagbo, O., Sun, Y.X., Hao, A.L., Gu, P., 2008. Effect of PAC additionon MBR process for drinking water treatment. Separation andPurification Technology 58 (3), 320e327.

Saravia, F., Zwiener, C., Frimmel, F.H., 2006. Interactions betweenmembrane surface, dissolved organic substances and ions insubmerged membrane filtration. Desalination 192 (1e3),280e287.

Schlichter, B., Mavrov, V., Chmiel, H., 2003. Study of a hybridprocess combining ozonation and membrane filtration-filtration of model solutions. Desalination 156 (1e3), 257e265.

Sheng, G.P., Yu, H.Q., 2006. Characterization of extracellularpolymeric substances of aerobic and anaerobic sludge usingthree-dimensional excitation and emission matrixfluorescence spectroscopy. Water Research 40 (6), 1233e1239.

Stedmon, C.A., Markager, S., Bro, R., 2003. Tracing dissolvedorganic matter in aquatic environments using a new approachto fluorescence spectroscopy. Marine Chemistry 82 (3e4),239e254.

�Swietlik, J., Sikorska, E., 2004b. Application of fluorescencespectroscopy in the studies of natural organic matter fractionsreactivity with chlorine dioxide and ozone. Water Research 38(17), 3791e3799.

�Swietlik, J., Dabrowska, A., Raczyk-Stanis1awiak, U., Nawrocki, J.,2004a. Reactivity of natural organic matter fractions withchlorine dioxide and ozone. Water Research 38 (3), 547e558.

Tian, J.Y., Liang, H., Li, X., You, S.J., Tian, S., Li, G.B., 2008.Membrane coagulation bioreactor (MCBR) for drinking watertreatment. Water Research 42 (14), 3910e3920.

wat e r r e s e a r c h 4 5 ( 2 0 1 1 ) 2 1 1 1e2 1 2 1 2121

Treguer, R., Tatin, R., Couvert, A., Wolbert, D., Tazi-Pain, A., 2010.Ozonation effect on natural organic matter adsorption andbiodegradation-application to a membrane bioreactorcontaining activated carbon for drinking water production.Water Research 44 (3), 781e788.

Uyguner, C.S., Bekbolet, M., 2005. Evaluation of humic acidphotocatalytic degradation by UV-vis and fluorescencespectroscopy. Catalysis Today 101 (3e4), 267e274.

Wang, Z.W., Wu, Z.C., Tang, S.J., 2009. Characterization ofdissolved organicmatter in a submergedmembrane bioreactorby using three-dimensional excitation and emission matrixfluorescence spectroscopy. Water Research 43 (6), 1533e1540.

Wang, Z.W., Tang, S.J., Zhu, Y.F., Wu, Z.C., Zhou, Q., Yang, D.H.,2010. Fluorescent dissolved organic matter variations ina submerged membrane bioreactor under different sludgeretention times. Journal of Membrane Science 355 (1e2),151e157.

Williams, M.D., Pirbazari, M., 2007. Membrane bioreactor processfor removing biodegradable organic matter from water. WaterResearch 41 (17), 3880e3893.

Zhang, T., Lu, J.F., Ma, J., Qiang, Z.M., 2008. Fluorescencespectroscopic characterization of DOM fractions isolated froma filtered river water after ozonation and catalytic ozonation.Chemosphere 71 (5), 911e921.