complexity of ultrafiltration membrane fouling caused by macromolecular dissolved organic compounds...
TRANSCRIPT
ARTICLE IN PRESS
Available at www.sciencedirect.com
WAT E R R E S E A R C H 4 2 ( 2 0 0 8 ) 3 1 5 3 – 3 1 6 1
0043-1354/$ - see frodoi:10.1016/j.watres
�Corresponding auE-mail addresses
uta.boeckelmann@t
journal homepage: www.elsevier.com/locate/watres
Complexity of ultrafiltration membrane fouling caused bymacromolecular dissolved organic compounds insecondary effluents
Jens Haberkampa,�, Mathias Ernstb, Uta Bockelmannc, Ulrich Szewzykc, Martin Jekela
aTechnische Universitat Berlin, Chair of Water Quality Control, Sekr. KF 4, Str. des 17. Juni 135, 10623 Berlin, GermanybTechnische Universitat Berlin, Centre for Water in Urban Areas, Sekr. KF 4, Str. des 17. Juni 135, 10623 Berlin, GermanycTechnische Universitat Berlin, Chair of Environmental Microbiology, Sekr. FR 1-2, Franklinstr. 29, 10587 Berlin, Germany
a r t i c l e i n f o
Article history:
Received 14 January 2008
Received in revised form
12 March 2008
Accepted 13 March 2008
Available online 26 March 2008
Keywords:
Cross-flow ultrafiltration
EPS extraction
Extracellular polymeric substances
Membrane fouling
Secondary effluent
Size exclusion chromatography
nt matter & 2008 Elsevie.2008.03.007
thor. Tel.: +49 30 314 25367: [email protected] (U. Bockelma
a b s t r a c t
Recent investigations indicate the relevance of extracellular polymeric substances (EPS) in
terms of fouling of low-pressure membranes in advanced wastewater treatment. In this
study, the high impact of the macromolecular fraction of effluent organic matter on fouling
was confirmed in cross-flow ultrafiltration experiments using secondary effluent with and
without autochthonous biopolymers. A method for the extraction of a natural mixture of
EPS derived from the bacterium Sinorhizobium sp. is presented. Ultrafiltration of solutions of
this bacterial EPS extract revealed a correlation between the concentration of EPS and the
loss of permeate flux. However, in ultrafiltration tests using extracted bacterial EPS in a
model solution as well as in secondary effluent without autochthonous biopolymers, the
extent of membrane fouling was not identical with the fouling provoked by secondary
effluent organic matter, although the biopolymer concentrations were comparable. The
differences in the fouling behaviour of the extracted bacterial EPS and effluent organic
matter are considered to be due to different compositions of the biopolymer fraction in
terms of proteins, polysaccharides, and other organic colloids, indicating a particular
impact of proteins on ultrafiltration membrane fouling.
& 2008 Elsevier Ltd. All rights reserved.
1. Introduction
Wastewater treatment is currently facing increasing demands
regarding the enhanced protection of receiving water bodies
and the reuse of secondary effluents as resource for drinking
water production in scarcity areas. Low-pressure membrane
filtration provides a potential alternative for advanced treat-
ment of municipal sewage, e.g., in membrane bioreactors
(MBRs) or tertiary treatment of secondary effluent. Their
application in such systems has significantly increased with-
in the last decade. However, membrane fouling is still a
r Ltd. All rights reserved.
; fax: +49 30 314 23850.rlin.de (J. Haberkamp), mnn), ulrich.szewzyk@tu-b
fundamental drawback, necessitating periodical chemical
cleanings and eventually forcing the replacement of irrever-
sibly fouled membranes. Membrane fouling can be caused by
particles, dissolved or colloidal organic and inorganic sub-
stances, as well as by the attachment of microorganisms onto
the membrane surface. While the formation of a filter cake
due to the deposition of particulate matter is controllable by
appropriate hydrodynamic conditions and backwashing, the
character and size of fouling-causing substances being
smaller than 0.45mm are not yet completely identified. te
Poele (2005) indicates the significance of colloids of the size
[email protected] (M. Ernst),erlin.de (U. Szewzyk), [email protected] (M. Jekel).
ARTICLE IN PRESS
Table 1 – Composition of R2A culture medium agar(following Reasoner and Geldreich, 1985)
Substance Amount per litre of ultra-pure water
Yeast extract 0.5 g
Proteose peptone no. 3 0.5 g
Casamino acid 0.5 g
Glucose 0.5 g
Sodium pyruvate 0.3 g
Dipotassium hydrogen
phosphate (K2HPO4)
0.3 g
Magnesium sulphate
(MgSO4 � 7H2O)
0.05 g
Agar 15 g
Tween 80 (fatty acid ester) 1 mL
WAT E R R E S E A R C H 4 2 ( 2 0 0 8 ) 3 1 5 3 – 3 1 6 13154
fraction between 0.45 and 0.1 mm in secondary effluent,
whereas Laabs et al. (2006) suggest the relevance of organic
macromolecules with sizes between 0.1 and 0.01mm regarding
the fouling of low-pressure membranes. Further studies have
pointed out the particular impact of dissolved organic
macromolecules on the loss of filtration performance in
micro- and ultrafiltration (Jarusutthirak and Amy, 2006),
especially in terms of irreversible long-term fouling (Rosen-
berger et al., 2006). The dissolved organic matter of secondary
effluent includes non-biodegradable substances deriving
from the raw wastewater, as well as compounds released
during the treatment process. The macromolecular fraction is
mainly composed of extracellular polymeric substances (EPS),
i.e., biopolymers of microbial origin. These are especially
polysaccharides, which are excreted for the adhesion of
bacteria onto surfaces (biofilm formation) or the cohesion to
other bacteria (formation of microbial aggregates), and
proteins, which possibly act as exo-enzymes, but also nucleic
acids and lipids (Flemming and Wingender, 2001).
Several studies focussing on the influence of the EPS
concentration on the extent of membrane fouling have
recently been published (Rosenberger et al., 2005; Ye et al.,
2005a; Garcia-Molina et al., 2006; Katsoufidou et al., 2007; van
de Ven et al., 2008). While te Poele and van der Graaf (2005)
indicate the relevance of proteins in ultrafiltration of second-
ary effluent, other authors point out the impact of poly-
saccharides on membrane fouling (Ye et al., 2005b;
Rosenberger et al., 2006; Fonseca et al., 2007). However, most
of the fouling studies are conducted either by means of
bench-scale tests, using model solutions of commercially
available proteins and polysaccharides with limited compar-
ability to natural wastewaters; or by observation of pilot
plants or full-scale filtration systems, which are fed by real
effluents with complex and varying composition, making it
difficult to draw explicit conclusions between the water
constituents and the flux decline in the filtration process.
The objective of this study was to investigate the impact of
dissolved organic matter (defined as substances o0.45mm) on
the extent of membrane fouling in cross-flow ultrafiltration.
Apart from ultrafiltration tests using secondary effluent, the
fouling behaviour of natural EPS in a model solution and in
secondary effluent with and without autochthonous biopoly-
mers was examined. For this purpose, a method for the
extraction of bacterial EPS was developed, allowing controlled
variations of the concentration of natural EPS in secondary
effluent.
2. Materials and methods
2.1. Extraction of bacterial EPS
Natural bacterial EPS were obtained from the bacterium
Sinorhizobium sp. This originally soil-borne, gram-negative
microorganism had previously been isolated from a slow sand
filter used for infiltration of surface water in Berlin-Marien-
felde (Germany) where it has been identified as predominant
bacterium, excreting high quantities of EPS.
For the extraction of bacterial EPS, an overnight culture of
Sinorhizobium sp. was plated onto petri dishes containing the
solid oligotrophic medium R2A (Table 1). After 72 h of
incubation at 28 1C, the bacterial cells together with the
produced EPS were scraped off from the agar plates
and resuspended in phosphate buffer solution (4 mmol L�1
NaH2PO4, 2 mmol L�1 Na3PO4, 9 mmol L�1 NaCl, 1 mmol L�1
KCl, pH ¼ 7; according to Frolund et al., 1996). In order to
detach the EPS bound to the bacterial cells, the suspension
was stirred for 2 h in contact with the cation exchange resin
Dowexs Marathons C (6 g per 0.1 L of suspension), which had
previously been equilibrated for 1 h in the phosphate buffer
solution. Thus, stabilising calcium ions were removed and
bound EPS were released into the solution (Jahn and Nielsen,
1995). Cation exchange resin and bacterial cells were subse-
quently separated from the EPS solution by centrifugation
(35 min at 3500 rpm) and filtration through 0.45mm cellulose
nitrate filters.
2.2. Analytical methods
2.2.1. Size exclusion chromatographySize exclusion chromatography with continuous UV254 nm and
organic carbon (OC) detection was used to characterise the
DOC composition of EPS solutions and secondary effluent (LC-
OCD system by DOC-Labor Dr. Huber, Karlsruhe, Germany;
SEC column: Toyopearls HW-50S by Tosoh Bioscience, Tokyo,
Japan). A characteristic LC-OCD chromatogram of secondary
effluent is presented in Fig. 1. Proteins, polysaccharides, and
further organic colloids elute within the so-called biopolymer
peak and are quantified by the calibrated infrared detector of
the LC-OCD system. The detection limit of the LC-OCD
system is 10mg L�1, the standard deviation is less than 1% of
the measured value (measurement range: 1–5 mg L�1 C;
samples containing higher DOC concentrations are diluted).
An additional UV detector allows the qualitative estimation of
organic nitrogen (ON) contents of the separate fractions by
measuring the absorbance of nitrate (at l ¼ 220 nm), which is
formed by oxidation of organic compounds inside the
oxidation reactor. A method for the quantitative analysis of
the ON concentration of the different fractions is currently
being developed.
ARTICLE IN PRESS
WA T E R R E S E A R C H 4 2 ( 2 0 0 8 ) 3 1 5 3 – 3 1 6 1 3155
2.2.2. Determination of polysaccharides, proteins, and totalnitrogen contentThe polysaccharide concentration (as glucose equivalents) of
the bacterial EPS solution and secondary effluent was
determined using the photometrical method following Du-
bois et al. (1956). The protein concentration (as BSA equiva-
lents) was measured by the modified photometrical Lowry
method according to Frolund et al. (1996). The total nitrogen
concentration of the EPS solution was determined using a
Multi N/C 3100 high-temperature analyser (Analytik Jena AG,
Jena/Germany).
2.3. Experimental set-up
The fouling tests were conducted using the experimental set-
up illustrated in Fig. 2. The applied ultrafiltration flat-sheet
membrane UP 150 is made of permanently hydrophilised
polyethersulfone (PES) and has a nominal molecular weight
200
2
4
6
8
low molecularweight neutrals
low molecularweight acidshumic
substances
biopolymers(EPS)O
C s
igna
l [A
U]
elution time [min]120100806040
Fig. 1 – LC-OCD chromatogram of secondary effluent (three-
fold dilution).
flowmeter
feedtank
cross-flowtest cell
gearpump
peristalticpump
ba
feed pressuregauge and valve
Fig. 2 – Ultrafiltrati
cut-off (MWCO) of 150 kg mol�1 (supplied by Microdyn-Nadir
GmbH, Wiesbaden, Germany). For each filtration experiment,
a new membrane sheet was inserted into the Plexiglass
cross-flow test cell (effective membrane surface area: 0.02 m2)
and rinsed with 12 L of deionised water in order to remove
solvent residues originating from the production process. The
membrane was subsequently pre-compacted for 24 h using a
solution of 6 mmol L�1 NaCl and 3 mmol L�1 CaCl2, resulting in
initial permeabilities of 326740 L m�2 h�1 bar�1 at 1 bar trans-
membrane pressure (TMP). Thereafter, the 24-h fouling test
was started either by addition of the EPS concentrate to the
NaCl/CaCl2 solution, or by positioning the suction tube of the
gear pump into a feed tank containing 10 L of secondary
effluent with or without additional bacterial EPS. In the
following, only the filtration curves after 24 h of pre-compac-
tion are shown and discussed. The ultrafiltration tests were
carried out at a constant TMP of 1 bar, a cross-flow velocity of
0.2 m s�1, and T ¼ 25 1C. The membrane flux was continuously
measured by an electronic balance; data were recorded by a
computer. The experiments were conducted in recycle mode,
returning the retentate continuously and the permeate
periodically (after accumulation of 0.8 L of permeate) back
into the feed tank, resulting in a nearly constant feed
concentration throughout a filtration run.
In order to compare the results of one test series, the
absolute permeate volume was graphically related to the total
filtration resistance R, which was calculated by dividing the
TMP by the dynamic viscosity of the permeate m and the
permeate flux J:
R ¼TMPm J
2.4. Ultrafiltration test solutions
2.4.1. Secondary effluentSecondary effluent was obtained from the sewage treat-
ment plant Berlin-Ruhleben (accomplishing mechanical and
permeate
retentate
lance
retentate pressuregauge and valve
dataacquisition
on test set-up.
ARTICLE IN PRESS
WAT E R R E S E A R C H 4 2 ( 2 0 0 8 ) 3 1 5 3 – 3 1 6 13156
biological treatment with biological nitrogen and phosphorus
removal). Due to the focus of this study on the impact of
dissolved effluent organic matter on membrane fouling, the
secondary effluent was filtered through 0.45mm cellulose
nitrate filters prior to filtration tests in order to remove any
particulate matter. The DOC concentration of secondary
effluent samples was 11.170.2 mg L�1, of which the biopoly-
mer fraction comprised 0.4 mg L�1 (as measured by the
LC-OCD system). The pH value was between 7.5 and 8.0.
Polysaccharide concentrations were 572 mg L�1 (as glucose
equivalents), protein concentrations were 1575 mg L�1
(as BSA equivalents).
2.4.2. EPS solutionsThe fouling potential of dissolved organic matter is depend-
ing on ionic strength, calcium concentration, and pH value of
the solution due to the influence of these parameters on the
effective net charge and the spherical extension of organic
macromolecules (Kilduff et al., 2004). In order to approach the
respective conditions of secondary effluent (from the sewage
treatment plant Berlin-Ruhleben) in filtration tests, the
extracted EPS were dissolved in a model solution of
6 mmol L�1 NaCl and 3 mmol L�1 CaCl2, resulting in an ionic
strength of 15 mmol L�1. The pH value was adjusted to
approximately 7.5 using sodium hydroxide solution. Since
the biopolymer concentration of secondary effluent was
0.4 mg L�1, the DOC concentration of EPS solutions for
ultrafiltration tests was adjusted to 0.4, 0.8, and 1.6 mg L�1,
respectively, in order to obtain solutions containing a
comparable amount of EPS, but no other fouling-relevant
DOC.
3. Results and discussion
3.1. Extraction of bacterial EPS
During 72 h of incubation, the low molecular weight OC
sources of the R2A culture medium were metabolised and
transformed into macromolecular EPS (Fig. 3a). While the EPS
concentration in the liquid culture medium was relatively low
200
2
4
6
8
10
formationof EPS
liquid culture medium R2A liquid R2A after 72 h incubation
of Sinorhizobium sp.
OC
sig
nal [
AU
]
elution time [min]120100806040
Fig. 3 – (a) Formation of bacterial EPS in liquid culture medium R
bacterial EPS concentrate (50-fold dilution) after extraction from
cation exchange resin Dowexs Marathons C.
and the solution contained comparatively high amounts of
other organic substances that might have interfered in
ultrafiltration tests, the incubation of Sinorhizobium sp. on
R2A agar plates and subsequent extraction (as described
above) resulted in a concentrated, viscous, and relatively pure
EPS solution (Fig. 3b). Due to the mobilisation of bound EPS
using a cation exchange resin, the EPS yield was about 20%
higher compared to the extraction procedure without cation
exchange resin.
The EPS concentration of the extracted solutions was
11378 mg L�1 (quantification of the biopolymer peak in LC-
OCD chromatograms). Photometrical quantification of poly-
saccharides and proteins yielded 100715 mg L�1 of polysac-
charides (as glucose equivalents) and 2078 mg L�1 of proteins
(as BSA equivalents); the total nitrogen concentration was
2.670.2 mg L�1. Therefore, the bacterial EPS concentrate
consisted mainly of polysaccharides and contained only low
amounts of proteins. Flemming and Wingender (2002) also
found evidence that within the EPS produced by bacteria in
pure cultures, polysaccharides are more likely to be the
predominant fraction than in mixed cultures occurring under
environmental conditions. By contrast, the secondary effluent
used in the present study contains higher proportions of
proteins (see above).
3.2. Ultrafiltration of secondary effluent
After 24 h of preconditioning and compaction, the permeate
flux of pure NaCl/CaCl2 model solution continued to decrease,
indicating that the compaction of the membrane was not
completed (Fig. 4a), which is confirmed by the slight and
linear increase in the filtration resistance in relation to the
cumulated permeate volume (Fig. 4b). However, ultrafiltration
of secondary effluent resulted in a significant decrease of the
permeate flux, especially during the first hours of the
filtration run (Fig. 4a). This flux loss is reflected by the high
increase in the filtration resistance (Fig. 4b) and indicates the
rapid blockage of a large number of membrane pores by
molecules of a molecular size comparable to the diameter of
the membrane pores, which is likely to be accompanied by
200
2
4
6
8
10 EPS concentrate (1:50)
with ion exchange resin EPS concentrate (1:50)
without ion exchange resin
OC
sig
nal [
AU
]
elution time [min]120100806040
2A (LC-OCD chromatograms). (b) LC-OCD chromatograms of
R2A agar petri dishes with and without application of the
ARTICLE IN PRESS
01
2
3
4
5
6
7
8model solutionUF permeatesec. effluent
norm
. res
ista
nce
R/R
0 [ ]
permeate volume [L]
00.0
0.2
0.4
0.6
0.8
1.0model solutionUF permeatesec. effluent
norm
. flu
x J/
J 0 [
]
filtration time [h]2420161284 5040302010
Fig. 4 – Ultrafiltration of secondary effluent, UF permeate of secondary effluent, and NaCl/CaCl2 model solution: (a) normalised
permeate flux (J0: permeate flux after 24 h of pre-compaction); (b) normalised total filtration resistance vs. cumulated
permeate volume (R0: total filtration resistance after 24 h of pre-compaction).
200
2
4
6
8 feed permeate
OC
sig
nal [
AU
]
elution time [min]20
0
2
4
6
8 feed permeateO
C s
igna
l [A
U]
elution time [min]120100806040 120100806040
Fig. 5 – LC-OCD chromatograms of feed and permeate samples of ultrafiltration tests using: (a) secondary effluent and (b) UF
permeate of secondary effluent (samples taken after 1 h; three-fold dilution).
Table 2 – Retention of biopolymers and humic substancesduring ultrafiltration of secondary effluent and UFpermeate of secondary effluent
Ultrafiltrationtest
Samplingtime
Retention (%)
Biopolymers Humicsubstances
Secondary
effluent
10 min 81 19
1 h 77 11
24 h 82 27
UF permeate of
secondary
effluent
10 min – 15
1 h – 3
24 h – 10
WA T E R R E S E A R C H 4 2 ( 2 0 0 8 ) 3 1 5 3 – 3 1 6 1 3157
the formation of a filter cake by substances that are retained
by the membrane. LC-OCD chromatograms of feed solution
and permeate illustrate the high retention of biopolymers and
the comparatively low retention of humic substances,
whereas smaller molecules are completely transmitted
(Fig. 5a). Since interactions between retained molecules and
the membrane are considered to be the reason for membrane
fouling, the high retention of biopolymers confirms the
relevance of this fraction in terms of fouling.
In order to obtain a solution containing nearly the same
organic matrix as secondary effluent, but only minor con-
centrations of biopolymers, secondary effluent was filtered
through the same membrane UP 150. Subsequent ultrafiltra-
tion of the UF permeate resulted in a distinctly decreased
decline of the permeate flux (Fig. 4a), accompanied by a linear
and comparatively low increase in the filtration resistance
(Fig. 4b). As would be expected due to the previous ultrafiltra-
tion step, the biopolymers remaining in this solution were
completely transmitted through the membrane during the
subsequent ultrafiltration test, and humic substances were
not retained either after 1 h of filtration (Fig. 5b). However, a
time-dependent retention of humic substances was observed
(Table 2), consisting of three phases: (1) partial retention (15%)
at the beginning, presumably due to adsorption on membrane
surface and pore walls of the initially clean membrane; (2)
nearly complete transmission after 1 h, indicating that
despite the initial adsorption phase, the pores remain wide
enough to enable the passage of macromolecules; (3) sub-
sequent increase in the retention (10% at the end), presum-
ably due to the continuous reduction of the pore diameter due
to deposition of further humic substances inside the pores,
thus slightly changing the filtration characteristics of the
ARTICLE IN PRESS
WAT E R R E S E A R C H 4 2 ( 2 0 0 8 ) 3 1 5 3 – 3 1 6 13158
membrane and resulting in a constantly increasing filtration
resistance. A comparable time-dependent retention of humic
substances was observed during ultrafiltration of secondary
effluent, but due the high retention of biopolymers and the
resulting formation of a filter cake, influencing the separation
properties of the membrane, the increase in the retention was
higher towards the end of the filtration run. In summary,
these experiments confirm that within the complex composi-
tion of dissolved organic compounds in secondary effluent,
biopolymers play the crucial role regarding the fouling of
ultrafiltration membranes, whereas humic substances are of
minor relevance.
3.3. Ultrafiltration of EPS solutions
Within 2 h after addition of extracted bacterial EPS to the
NaCl/CaCl2 model solution, the permeate flux decreased to
less than 30% of the initial value (Fig. 6a). The initial flux
decline observed during ultrafiltration of 0.8 and 1.6 mg L�1
was higher compared to the test using 0.4 mg L�1 EPS. The
retention of EPS was between 92% and 97%, indicating that
the extracted EPS were too large to pass through the
membrane pores. It is noteworthy that there was a slight
recovery of the permeate flux in the second half of the
filtration tests using 0.8 and 1.6 mg L�1 EPS. This effect
was reproducible and might be due to structural changes of
the EPS fouling layer, and thus, an increased permeability
inside it.
For a better comparison of the different filtration curves,
the normalised permeate flux was related to the amount of
EPS delivered to the membrane surface, which was calculated
as follows:
delivered EPS½mg� ¼Xt¼24 h
t¼0
ðpermeate volume ½L�
� EPS feed concentration½mg � L�1�Þ
The decline of the filtration curves obtained this way is
nearly identical (Fig. 6b), confirming that the extent of
membrane fouling is proportional to the amount of bacterial
EPS delivered to the membrane surface.
00.0
0.2
0.4
0.6
0.8
1.00.4 mg/L EPS0.8 mg/L EPS1.6 mg/L EPS
norm
. flu
x J/
J 0 [
]
filtration time [h]2420161284
Fig. 6 – Ultrafiltration of bacterial EPS solutions: (a) normalised p
vs. cumulated mass of EPS delivered to the membrane.
3.4. Impact of biopolymers in secondary effluent andbacterial extract on fouling
Although the biopolymer concentration (related to carbon)
was comparable in EPS model solution and secondary
effluent, the flux decline observed during ultrafiltration of
the latter was more severe, indicating that the predominant
fouling mechanisms during ultrafiltration of these solutions
were not identical (Fig. 7a). In order to examine whether the
reduced extent of fouling induced by bacterial EPS was due to
the lack of background DOC (e.g., humic substances) in the
model solution, extracted EPS were added to UF permeate of
secondary effluent. Ultrafiltration of UF permeate spiked with
EPS increased the flux decline significantly over that pre-
viously observed in ultrafiltration of UF permeate (cf. Fig. 4).
However, the results in Fig. 7a also show that ultrafiltration of
the spiked UF permeate yielded less fouling than secondary
effluent, although the biopolymer concentration and the
composition of the background DOC were comparable in
both solutions. Even an increased EPS concentration by
further addition of bacterial EPS to UF permeate (0.8 and
1.6 mg L�1; data not shown) could not provoke a flux decline of
comparable intensity as observed during ultrafiltration of
secondary effluent, confirming the differences between the
fouling mechanisms induced by extracted bacterial EPS and
effluent organic matter, respectively.
Detailed analysis of the LC-OCD chromatograms indicates
differences between secondary effluent and EPS-spiked solu-
tions regarding the biopolymer peak (Fig. 7b). The nominal
upper size exclusion limits of the applied SEC column are
20,000 g mol�1 for polysaccharides and 80,000 g mol�1 for
globular proteins, respectively (manufacturer information by
Tosoh Bioscience, Tokyo, Japan). The steep increase and
subsequent distinct decrease of the biopolymer peak in the
LC-OCD chromatograms of EPS model solution and EPS-
spiked UF permeate indicate a higher amount of large
polysaccharides eluting with the void volume. By contrast,
the slighter increase and broader shape of the biopolymer
peak in the chromatogram of secondary effluent indicate
an increased proportion of more compact biopolymers
(e.g., proteins). The differences in the composition of the
00.0
0.2
0.4
0.6
0.8
1.00.4 mg/L EPS0.8 mg/L EPS1.6 mg/L EPS
norm
. flu
x J/
J 0 [
]
delivered EPS [mg C]2420161284
ermeate flux vs. filtration time; (b) normalised permeate flux
ARTICLE IN PRESS
00.0
0.2
0.4
0.6
0.8
1.0secondary effluent0.4 mg/L EPS in UF permeate0.4 mg/L EPS in model solution
norm
. flu
x J/
J 0 [
]
filtration time [h]
350.0
0.3
0.6
0.9 secondary effluent 0.4 mg/L EPS in UF permeate 0.4 mg/L EPS in model solution
ON
sig
nal [
AU
]
elution time [min]35
0.0
0.2
0.4
0.6 secondary effluent 0.4 mg/L EPS in UF permeate 0.4 mg/L EPS in model solution
UV
254
nm s
igna
l [A
U]
elution time [min]
200
2
4
6
8
10 secondary effluent 0.4 mg/L EPS in UF permeate 0.4 mg/L EPS in model solution
OC
sig
nal [
AU
]
elution time [min]2420161284 120100806040
504540 504540
Fig. 7 – Ultrafiltration of secondary effluent and bacterial EPS in UF permeate of secondary effluent and in NaCl/CaCl2 model
solution: (a) normalised permeate flux vs. filtration time; (b) LC-OCD chromatograms of the feed solutions (three-fold
dilution); (c) UV254 nm signal of the LC-OCD chromatograms (detail); (d) organic nitrogen (ON) signal of the LC-OCD
chromatograms (detail).
WA T E R R E S E A R C H 4 2 ( 2 0 0 8 ) 3 1 5 3 – 3 1 6 1 3159
biopolymer fraction are confirmed by detailed examination of
the UV254 nm and ON signals of the LC-OCD chromatograms
(Figs. 7c and d). In contrast to proteins, polysaccharide
molecules do neither contain UV-active components nor
nitrogen. Therefore, the significantly elevated UV254 nm and
ON signals are a qualitative evidence of higher proportions of
proteins in the secondary effluent. These chromatographic
results are in accordance with the photometrical determina-
tion of proteins and polysaccharides, revealing a significantly
increased polysaccharide concentration in the bacterial EPS
extract and a higher proportion of proteins in secondary
effluent (see above).
Apart from polysaccharides and proteins, bacterial cell
fragments represent another fraction of organic colloids in
secondary effluent eluting within the biopolymer peak (Laabs
et al., 2004) and containing nitrogen. Therefore, cell frag-
ments might also contribute to the ON signal of the
biopolymer peak of secondary effluent. Considering the
comparatively mild procedure applied for the extraction of
bacterial EPS, the content of cell fragments in the EPS extract
due to cell lysis is likely to be relatively low (Frolund et al.,
1996), and they are not contained in the UF permeate of
secondary effluent either due to their previous removal by
ultrafiltration. Since the proteins cannot be distinguished
from cell fragments using the LC-OCD method applied, the
latter have to be taken into consideration as a further fraction
of biopolymers with potential relevance in terms of ultra-
filtration membrane fouling.
A possible deposition of inorganic colloids on the fouled
membranes was examined by energy-dispersive X-ray spec-
troscopy (LEO 1530 FE-SEM by Carl Zeiss SMT AG, Oberkochen,
Germany). Deposits of the elements silicon, calcium, alumi-
nium, and iron were either not or only in traces detectable,
regardless of the type of feed water used in the previous
ultrafiltration experiment (i.e., secondary effluent, UF perme-
ate or EPS in model solution). Therefore, the influence of
inorganic colloids on the organic membrane fouling investi-
gated in this study is considered to be negligible.
In summary, despite comparable total biopolymer concen-
trations (related to carbon), qualitative differences regarding
the composition of the macromolecular fraction are likely to
be the reason for the different fouling behaviours of
secondary effluent and solutions spiked with extracted
bacterial EPS, indicating the high relevance of proteins and
possibly other organic colloids in terms of ultrafiltration
membrane fouling.
Regarding the predominant fouling mechanisms, the flux
decline induced by the extracted bacterial EPS is considered
to be influenced by the formation of a rather loosely bound
concentration polarisation layer of large polysaccharides,
whereas the filtration of secondary effluent results in the
tighter adhesion of macromolecular substances due to the
more complex variety of biopolymers within effluent organic
matter, including higher proportions of proteins. This as-
sumption is supported by forward-flush tests using deminer-
alised water after ultrafiltration of EPS-spiked secondary
ARTICLE IN PRESS
WAT E R R E S E A R C H 4 2 ( 2 0 0 8 ) 3 1 5 3 – 3 1 6 13160
effluent and UF permeate, revealing that the recovery rate of
the permeate flux was higher after ultrafiltration of EPS-
spiked UF permeate (data not shown).
4. Conclusions
A method for the extraction of EPS derived from the
bacterium Sinorhizobium sp. has been developed. Addition of
the viscous and relatively pure natural EPS mixture to
different test solutions allows the systematic variation of
the EPS concentration in order to investigate the impact of
biopolymers on membrane fouling in tertiary sewage treat-
ment.
In cross-flow ultrafiltration experiments using model solu-
tions of the extracted bacterial EPS at concentrations which
are relevant in secondary effluents (0.4–1.6 mg L�1), the EPS
were almost completely retained by the membrane, thus
causing a severe permeate flux decline, especially in the
initial filtration phase. A correlation between the EPS
concentration in model solution and the extent of membrane
fouling was observed. However, ultrafiltration of secondary
effluent resulted in a higher flux decline than ultrafiltration
of the EPS model solutions. Selective removal of the
autochthonous biopolymers (which contributed 4% to the
total DOC) from secondary effluent by previous ultrafiltra-
tion revealed a significantly lower fouling potential of the
remaining organic compounds in subsequent ultrafiltration
tests. Therefore, biopolymers are considered to be the
predominant fouling-active fraction within the DOC of
secondary effluent, whereas humic substances and smaller
organic compounds play a minor role in ultrafiltration
membrane fouling.
Addition of extracted bacterial EPS to UF permeate of
secondary effluent (without autochthonous biopolymers)
caused significantly more fouling than ultrafiltration of
UF permeate of secondary effluent alone. However, the
bacterial EPS could not provoke the same fouling rate as
observed in ultrafiltration of secondary effluent, although the
composition of the background DOC was comparable. Thus,
extracting EPS from pure bacterial cultures is not an appro-
priate surrogate for organic foulants found in secondary
effluent and would therefore not be useful for bench-scale
membrane filtration studies aimed at finding ways to reduce
the fouling rate or to clean membranes more efficiently after
fouling. However, since the differences in the fouling
behaviour of the bacterial EPS extract and secondary effluent
are due to qualitative differences in terms of the macro-
molecular composition of the solutions, a conclusion regard-
ing the impact of the complex mixture of biopolymers on
membrane fouling can be drawn. The extracted bacterial EPS
contain significantly more polysaccharides than proteins,
whereas the secondary effluent used contains a larger
proportion of proteins, as well as other organic colloids
(i.e., fragments of bacterial cells). Therefore, the increased
flux decline observed during fouling tests using secondary
effluent indicates the relevance of proteins and possibly
further organic colloids in terms of ultrafiltration membrane
fouling.
Acknowledgement
The laboratory work of Anne Konig, Angela Wurtele, Hui
Cheng, and Daniela Pallischeck is greatly acknowledged.
R E F E R E N C E S
Dubois, M., Gilles, K.A., Hamilton, J.K., Rebers, P.A., Smith, F., 1956.Colorimetric method for determination of sugars and relatedsubstances. Anal. Chem. 28 (3), 350–356.
Flemming, H.C., Wingender, J., 2001. Relevance of microbialextracellular polymeric substances (EPSs)—Part I: structuraland ecological aspects. Water Sci. Technol. 43 (6), 1–8.
Flemming, H.C., Wingender, J., 2002. What biofilms contain—
proteins, polysaccharides, etc. Chem. Unserer Zeit 36 (1), 30–42(in German).
Fonseca, A.C., Summers, R.S., Greenberg, A.R., Hernandez, M.T.,2007. Extra-cellular polysaccharides, soluble microbial pro-ducts, and natural organic matter impact on nanofiltrationmembranes flux decline. Environ. Sci. Technol. 41 (7),2491–2497.
Frolund, B., Palmgren, R., Keiding, K., Nielsen, P.H., 1996.Extraction of extracellular polymers from activated sludgeusing a cation exchange resin. Water Res. 30 (8),1749–1758.
Garcia-Molina, V., Lyko, S., Esplugas, S., Wintgens, T., Melin, T.,2006. Ultrafiltration of aqueous solutions containing organicpolymers. Desalination 189 (1–3), 110–118.
Jahn, A., Nielsen, P.H., 1995. Extraction of extracellular polymericsubstances (EPS) from biofilms using a cation exchange resin.Water Sci. Technol. 32 (8), 157–164.
Jarusutthirak, C., Amy, G., 2006. Role of soluble microbial products(SMP) in membrane fouling and flux decline. Environ. Sci.Technol. 40 (3), 969–974.
Katsoufidou, K., Yiantsios, S.G., Karabelas, A.J., 2007. Experimen-tal study of ultrafiltration membrane fouling by sodiumalginate and flux recovery by backwashing. J. Membr. Sci. 300(1–2), 137–146.
Kilduff, J.E., Mattaraj, S., Belfort, G., 2004. Flux decline duringnanofiltration of naturally-occurring dissolved organic matter:effects of osmotic pressure, membrane permeability, and cakeformation. J. Membr. Sci. 239 (1), 39–53.
Laabs, C., Amy, G., Jekel, M., 2004. Organic colloids and theirinfluence on low-pressure membrane filtration. Water Sci.Technol. 50 (12), 311–316.
Laabs, C., Amy, G.L., Jekel, M., 2006. Understanding the size andcharacter of fouling-causing substances from effluent organicmatter (EfOM) in low-pressure membrane filtration. Environ.Sci. Technol. 40 (14), 4495–4499.
Reasoner, D.J., Geldreich, E.E., 1985. A new medium for theenumeration and subculture of bacteria from potable water.Appl. Environ. Microbiol. 49 (1), 1–7.
Rosenberger, S., Evenblij, H., te Poele, S., Wintgens, T., Laabs, C.,2005. The importance of liquid phase analyses to understandfouling in membrane assisted activated sludge processes-sixcase studies of different European research groups. J. Membr.Sci. 263 (1–2), 113–126.
Rosenberger, S., Lesjean, B., Laabs, C., Jekel, M., Gnirss, R.,Schrotter, J.-C., 2006. Impact of colloidal and soluble organicmaterial on membrane performance in membrane bioreactorsfor municipal wastewater treatment. Water Res. 40 (4),710–720.
te Poele, S., 2005. Foulants in ultrafiltration of wwtp effluent. Ph.D.Thesis, Department of Sanitary Engineering, Delft Universityof Technology.
ARTICLE IN PRESS
WA T E R R E S E A R C H 4 2 ( 2 0 0 8 ) 3 1 5 3 – 3 1 6 1 3161
te Poele, S., van der Graaf, J., 2005. Enzymatic cleaning inultrafiltration of wastewater treatment plant effluent. Desali-nation 179 (1–3), 73–81.
van de Ven, W.J.C., van’t Sant, K., Punt, I.G.M., Zwijnenburg, A.,Kemperman, A.J.B., van der Meer, W.G.J., Wessling, M., 2008.Hollow fiber dead-end ultrafiltration: influence of ionic environ-ment on filtration of alginates. J. Membr. Sci. 308 (1–2), 218–229.
Ye, Y., Le Clech, P., Chen, V., Fane, A.G., 2005a. Evolution of foulingduring crossflow filtration of model EPS solutions. J. Membr.Sci. 264 (1–2), 190–199.
Ye, Y., Le Clech, P., Chen, V., Fane, A.G., Jefferson, B., 2005b.Fouling mechanisms of alginate solutions as modelextracellular polymeric substances. Desalination 175 (1),7–20.