See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/235259061

Comparison of ceramic and polymeric membrane permeability and fouling

using surface water

Article  in  Separation and Purification Technology · June 2011

DOI: 10.1016/j.seppur.2011.03.025

CITATIONS READS

125 3,243

5 authors, including:

Bas Hofs Julien Ogier

Evides Lanxess AG
26 PUBLICATIONS   685 CITATIONS    9 PUBLICATIONS   150 CITATIONS   

SEE PROFILE SEE PROFILE

Dirk Vries Erwin Beerendonk

KWR Water Research Institute KWR Water Research Institute
35 PUBLICATIONS   292 CITATIONS    42 PUBLICATIONS   1,688 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Optimal (experimental) input design View project

Control of infinite dimensional systems View project

All content following this page was uploaded by Bas Hofs on 07 December 2017.

The user has requested enhancement of the downloaded file.

 Separation and Purification Technology 79 (2011) 365–374

Contents lists available at ScienceDirect

Separation and Purification Technology

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

Comparison of ceramic and polymeric membrane permeability and fouling using

surface water
Bas Hofs ∗ , Julien Ogier, Dirk Vries, Erwin F. Beerendonk, Emile R. Cornelissen
KWR Watercycle Research Institute, P.O. Box 1072, 3430 BB Nieuwegein, The Netherlands

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

Article history: The trans membrane pressure (TMP) increase at a constant flux of 150 L m−2 h−1 due to membrane fouling
Received 11 January 2011 by direct filtration with lake water is investigated for four ceramic (Al2 O3 , ZrO2 , TiO2 , SiC) and a PES/PVP
Received in revised form 2 March 2011 polymeric microfiltration membrane(s). The membrane structures and compositions are characterised
Accepted 2 March 2011
with permporometry (pore size, porosity) and XPS. The TiO2 and SiC membrane have a 5 and 24 times
larger average pore size than expected based on supplier information.
Keywords:
The TMP increase rate due to fouling inversely follows the measured pore size. Reversible fouling
Microfiltration
decreases in the order of polymeric ≈ Al2 O3 ≈ ZrO2 > TiO2 > SiC, and for irreversible fouling poly-
Natural organic matter
Low pressure membranes
meric > ZrO2 > Al2 O3 > TiO2 > SiC.
Ultrafiltration Removal of non purgeable organic carbon (NPOC) is around 30% for the ceramic membranes, and
Water treatment 13–25% for the polymeric membrane. The reversible NPOC load decreases in the following order
(Al2 O3 > (ZrO2 ≈ TiO2 )) ≈ SiC > polymeric. The higher degree of fouling on the polymeric membrane is
partly due to its lower volume/area ratio, compared to ceramic membranes. Mass balance analysis shows
that 85 ± 8% of the NPOC is accounted for. Thus, the used soaking method (pH 12 NaOH, >1 h) does not
fully remove the NPOC, suggesting that a more aggressive cleaning solution should be used.
© 2011 Elsevier B.V. All rights reserved.

1. Introduction ceramic membranes is evident from applied high backwash pres-

Ceramic membranes have been hailed for their advantageous
properties when compared to polymeric membranes. The advan-
tages of ceramic membranes compared to polymeric membranes
are often stated to be: (i) a relatively narrow pore size distribution
and higher porosity, resulting in better separation characteristics
and a higher flux, (ii) a higher mechanical stability (allowing higher
pressures), (iii) a higher chemical stability resulting in longer mem-
brane lifetimes, and (iv) higher hydrophilicity resulting in high
fluxes at low pressures [1–7]. Higher hydrophilicity is also asso-
ciated with lower fouling [6]. Evidence to support all of these
advantages of ceramic membranes is, however, not so easily found.
A study in which the pore size distributions (usually measured by
fractional rejection of molecules) of a thin film composite and a
TiO2 membrane are compared, shows that they are very similar
[8]. Unfortunately, no further studies in which pore size distribu-
tions are measured for both polymeric and ceramic membranes
within a single article could be found. The pore size distribution
of ceramic membranes can be narrow [9], but the same is true
for polymeric membranes [10]. The higher mechanical stability of
∗ Corresponding author. Tel.: +31 0 306069697; fax: +31 0 306061165.
E-mail address: bas.hofs@kwrwater.nl (B. Hofs).
sures [11–13], although ceramic membranes are more prone to
breakage than polymeric membranes. The chemical stability of
ceramic membranes is higher than that of polymeric membranes,
and decreases in the order of TiO2 , ZrO2 , Al2 O3 , SiO2 and their
hydrothermal stability decreases in the order of ZrO2 , Al2 O3 , TiO2 ,
SiO2 [14,15]. The stability also depends on the membrane struc-
ture, with finer structures being less stable. The hydrophilicity of
polymeric membranes depends on the used polymers and can be
high; some polymeric membranes have a contact angle with water
of 10◦ [16]. For ceramic membranes the hydrophilicity depends
on the used materials. The contact angle of water for ceramic
membranes covers a broad range, and the same can be said of poly-
meric membranes [9,10,17], making it hard to determine whether
ceramic membranes are indeed more hydrophilic than polymeric
ones.
Differences in fouling behavior of polymeric and ceramic mem-
branes can be expected, as the surface groups of ceramic and
polymeric membranes are different. In the scientific literature,
studies comparing the fouling of ceramic and polymeric mem-
branes for direct water filtration could not be found. However, some
work has been done within the Techneau framework [18]. Mueller
et al. [19,20] compared the fouling of a ceramic membrane (TiO2 top
layer, Al2 O3 support) with that of a polymeric (polyethersulfone)
membrane, both operated under the same conditions, by analysing
the cleaning in place (CIP, at pH 11) waste from natural organic

1383-5866/$ – see front matter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.seppur.2011.03.025
366 B. Hofs et al. / Separation and Purification Technology 79 (2011) 365–374
matter (NOM) loaded membranes by liquid chromatography
organic carbon detection (LC-OCD) and fluorescence excitation-
emission matrices (FEEM). LC-OCD showed that the CIP-waste of
the polymeric membrane showed a higher polysaccharide content
than the ceramic membrane [20]. FEEM showed that the CIP-waste
of the polymeric membrane had higher fluorescence for protein-
like organic compounds, and the CIP-waste from the ceramic
membrane showed higher fluorescence for humic substances [19].
As polysaccharides are thought to be a major foulant for microfiltra-
tion/ultrafiltration (MF/UF) [21,22], ceramic membranes may have
a lower potential for organic fouling [20]. However, more research
into fouling mechanisms of polymeric and ceramic membranes
is needed to prove whether polysaccharides lead to irreversible
fouling.
In general, pretreatment is often employed when low pressure
membranes are selected. Studies show that 30–40% of the low pres-
sure membranes operate without pretreatment, and that especially
coagulation and powdered activated carbon are commonly applied
as pretreatment for MF/UF [23,24]. Coagulation is the most success-
ful pretreatment for fouling control, but may increase fouling and
thus requires an optimized dosing of the used chemicals [24]. For
these reasons we decided to study the performance of the mem-
branes without coagulation, or any other form of pretreatment –
although in a future study we intend to incorporate this aspect as
well.
In this study, we investigate the performance of four different
ceramic membranes (TiO2 , ZrO2 , Al2 O3 , and SiC) and one poly-
meric membrane (polyethersulfone–polyvinylpyridine, PES–PVP).
The performance and fouling of the membranes by surface water
are compared under the same flux and backwash procedure. In
practice, in full scale application, the membranes would be used
under different operational conditions than here. The ceramic
membranes are usually operated at a higher flux than the poly-
meric, and the backwash and cleaning procedures would also be
very different for the ceramic and polymeric membranes. However,
since we want to compare the membrane properties on the level
of the membranes rather than different operational conditions, we
decided to fix the flux and backwash procedure.
Thus, this paper is on the comparison of the membrane proper-
ties with respect to trans membrane pressure increase (TMP) and
characterisation of fouling by surface water. Therefore we charac-
terise the membranes in order to verify the membrane properties
as supplied by the suppliers. We characterise the water type and
follow the TMP due to both reversible and irreversible fouling in fil-
tration experiments at a constant flux. The reversible fouling is the
TMP increase within one filtration cycle, and the irreversible fouling
is determined from the increase in the TMP directly after the back-
wash. The following membrane properties were obtained: pore
size distribution, porosity, zeta-potential, and elemental composi-
tion of top and support layer. The composition (Ca2+ concentration,
conductivity, LC-OCD, turbidity, etc.) of the feed water type was
measured, as well as the non-purgeable organic compound (NPOC)
content of the backwash water, permeate and the CIP-waste. This
allows us to make a mass balance of the NPOC and to determine
the separation characteristics of the membranes.
2. Materials and methods
2.1. Membranes and characterisation
The Al2 O3 , TiO2 and ZrO2 single channel tubular membranes
were obtained from Atech innovations GmbH (Gladbeck, Germany)
and had a support layer of Al2 O3 . The SiC membranes (a flat disc
type for the pore size, porosity, X-ray photoelectron spectroscopy
(XPS) and zeta-potential measurements, and a single channel
tubular membrane for the fouling experiments) were obtained
from Cometas (Bagsvaerd, Denmark). The polymeric filter pen
membrane, composed of a PES/PVP blend, was obtained from Norit
Filtrix (Amersfoort, The Netherlands). Pore sizes and the materials
as provided by the suppliers (for the top layer) of the membranes
are given in Tables 1 and 2. Note that due to the varying require-
ments of the characterisation devices we used different membrane
dimensions for each test. The pore size of the membranes as given
by the supplier was similar for the individual membranes. Details
about the membranes used in the filtration test are given in Table 1,
and for characterisation by porometry, XPS and zeta-potential mea-
surements are given in Table 2.
The pore size and pore size distribution of the ceramic mem-
branes was measured with a Porolux 1000 capillary flow porometer
(IB-FT GmbH, Wildau, Germany). The membranes were cut into
samples with a length of approximately 7 cm and were fitted on one
end with a PVC collar of 12 mm outer diameter and a Swagelok fer-
rule set. The other end was closed by potting with a two component
resin. Porofil (Benelux Scientific, Tiel, The Netherlands) wetting
fluid was used to wet the membrane prior to the measurements,
which were carried out in duplicate. The mean pore diameter is
calculated from the mean flow pressure, which corresponds to the
intersection of the wet curve with the calculated half-dry curve
(calculated from a dry and a wet run).
The porosity was measured on an Accupyc 1330 pycnometer
from Micromeritics instruments Corporation (Norcross, U.S.A.). For
these measurements the membranes were cut into small pieces
of about 7 mm length. A sample was sealed in a sample compart-
ment of known volume. Helium was added, and then expanded
into another calibrated volume. The pressure before and after the
cell expansion was measured to compute the sample volume. The
porosity was subsequently calculated from the density.
The elemental composition of the membranes was measured on
a Quantera SXM (scanning XPS microprobe) from Physical Electron-
ics (Chanhassen, U.S.A.). The XPS measurements were done with an
Al K␣ monochromatic (at 1486.6 eV) X-ray beam, with a beam size
of 100 ␮m (power 25 W, Ie 2.6 mA). The Quantera SXM was con-
trolled with Compass software and the data reduction was done
with Multipak v8.0. Samples of the membranes of approximately
7 mm length were used for the XPS measurements. Elemental spec-
trum scans of the both top and support layer (3 locations) were
measured (at two points per location), averaged and interpreted
with the help of standard books.
All the above measurements were carried out at the Euro-
pean Membrane Institute (EMI) at the University of Twente (The
Netherlands).
Zeta-potential measurements were performed using the zeta
meter SurPass (Anton Paar GmbH, Austria). These measurements
were performed on a flat sheet membrane of SiC and on seven chan-
nels tubular membranes of Al2 O3 , TiO2 and ZrO2 . A special adapter
was used to force the measuring electrolyte solutions (pH 6–10,
10−3 M KCl) through the porous wall that separate two individ-
ual channels. The zeta potential was calculated from the streaming
potential measurements as described elsewhere [25].
2.2. Water type and characterisation
The first set of membrane filtration tests at fixed backwash
pressure was performed with surface water from Noord-Bergum
lake (The Netherlands). A batch of 400 L was sampled (on the 20th
of November 2009 and subsequently homogenised and filtered
through a 1 ␮m cartridge filter. The water was stored at 4 ◦ C and
the water composition is given in Table 3 (1st batch). The amount
of NPOC found is nearly the same as the amount of DOC, due to
pre-filtering though a 1 ␮m cartridge filter. The second set of exper-
iments at comparable backwash flux was performed with surface
 B. Hofs et al. / Separation and Purification Technology 79 (2011) 365–374 367

Table 1
Membranes that were used for the filtration tests.

Type Compositiona Supplier Pore sizea Surface areaa Configuration

ZrO2 L = 250 mm
Al2 O3 Atech Innovations 0.1 ␮m 0.0047 m2 Dint = 6 mm
Ceramic
TiO2

SiC Cometas 0.04 ␮m 0.013 m2 L = 305 mm

Dint = 14 mm

Polymeric PES/PVP Norit Filtrix 0.14 ␮m 0.02 m2 L = 200 mm

Dint = 0.4 mm
a
Supplier’s information, L = length, Dint = internal diameter. Note that suppliers may use a different method to determine the pore size of their membranes than the method
used here.

Table 2
Membranes that were used for the characterisation (porometry, XPS, and zeta-potential).

Compositiona Membranes used for pore size, porosity and XPS measurements Membranes used for zeta-potential measurements

ZrO2 L = 250 mm L = 200 mm

Dint = 6 mm Dint = 6 mm
Al2 O3 Dext = 10 mm, Dext = 25.4 mm,
TiO2 1 channel 7 channels

SiC Flat membrane Flat membrane

D = 25 mm 55 mm × 25 mm
Thickness = 6 mm Thickness = 6 mm

Polymeric 1 channel, L = 200 mm –

Dint = 0.72 mm
Dext = 1.2 mm
a
Supplier’s information, L = length, Dint = internal diameter, Dext = external diameter.

water from Noord-Bergum lake (The Netherlands) sampled on the
2nd of September 2010, homogenised and filtered through a 1 ␮m
cartridge filter. The water was stored at 4 ◦ C and the water compo-
sition is given in Table 3 (2nd batch).
Liquid Chromatography-Organic Carbon Detection (LC-OCD)
was performed by the DOC-Labor (Germany). Sulphate was deter-
mined by Vitens Laboratory Utrecht (Utrecht, The Netherlands).
NPOC was determined in accordance with ISO 8245. Calcium,
magnesium and chloride concentrations were determined with
inductively coupled plasma mass spectrometry (ICP-MS) at the
KWR laboratory (Nieuwegein, The Netherlands) following an in-
house procedure (LAM-058). All other (Table 3) measurements
were performed at the KWR laboratory, with in-house methods or
text-book procedures (turbidity: in accordance with NEN-ISO 7027,
UV absorbance: at 254 nm, in-house method LAM-033, hydrogen
carbonate: by titration with acid, in-house method LAM-042, elec-
trical conductivity: measured with a CDM 83 conductivity meter
(Radiometer), conforming to ISO 7888, at 25 ◦ C).

Table 3
Feed water composition, after 1 ␮m cartridge filter, as determined before the experiments.

1st batch 2nd batch Unit

pH 7.9 7.7 –
Conductivity 690 640 ␮S/cm
Calcium 49 42 mg/L
Magnesium 14 12
Chloride 110 110
Iron 0.59 0.28
Hydrogen carbonate 125 115
Sulphate 55 55
Turbidity 9.7 3.6 FNU
NPOC 15 14 mg/L
UV 41 32 E/m
DOCa 15.1 – mg/L
HOCa nq –
POCa nq –
CDOCa 15.1 –
Humic acidsa 10.3 –
Building blocksa 2.0 –
Acidsa nq –
Neutralsa 2.0 –
Biopolymers 0.76 –
SAC 0.01 – m−1
SUVA 3.14 – L mg−1 m−1

HOC = hydrophobic organic carbon, CDOC = chromatographic DOC.nq = not quantifiable (<1 ppb). NPOC and turbidity of the first batch of water were determined after the
first set of experiments as well; NPOC was the same, but turbidity was slightly lower (9.2 FNU).
a
Data from LC-OCD measurement.
368 B. Hofs et al. / Separation and Purification Technology 79 (2011) 365–374
permeate
cooler
pressured air
Membrane
module
Feed Backwash
tank tank
P
4-way
valve
pump
Sewage
Fig. 1. Scheme of lab-scale pilot in filtration mode. In backwash mode the 4-way
valve is rotated around its central point by 90◦ .
2.3. Experimental filtration setup and procedure
Two series of experiments were performed. In the first set
the backwash pressure was kept constant as described below. In
the second set the backwash pressure or surface area was mod-
ified in order to get comparable backwash fluxes, as described
below.
2.3.1. Tests at constant backwash pressure
Tests were performed with a lab-scale pilot, see Fig. 1. A TSE
540270 dual syringes pump (TSE systems GmbH, Bad homburg,
Germany) provided an accurate constant feed flow during all the
experiments. The surface areas of the membranes were not iden-
tical, so the feed flow was adjusted to 11.8, 11.8, 11.8, 33.5, and
50 mL/min, for the ZrO2 , Al2 O3 , TiO2 , SiC and polymeric membrane,
respectively, in order to reach a constant feed flux of 150 L m−2 h−1 .
A pressure sensor in the feed of the set-up (Fig. 1) measured the
pressure in front of the membrane, which equals TMP which was
recorded on a computer every 10 s. A vessel containing fresh MilliQ
water, pressurised at 3 bars was used for the backwash of mem-
branes. The 1 min backwash occurred automatically every 20 min.
All the experiments were carried out in duplicate and started
by MilliQ filtration (150 L m−2 h−1 , with backwash as described) to
remove air from the system, until a constant TMP was reached to
obtain a reference TMP value for each test. Then the water was
switched from MilliQ to Noord-Bergum surface water. The subse-
quent filtration test lasted about 5 h. In the tests with the polymeric
membrane the duration of the filtration was shorter (2.5 h) and
could not be continued due to the fact that the maximum pump
pressure (about 1.3 bar) was reached. All tests were performed at
20 ◦ C and the temperature was controlled by a Julabo FC1200 cooler
system (JULABO Labortechnik GmbH, Seelbach, Germany).
The last backwash sample collected after 5 h of filtration was
stored, as well as the permeate sample of the 20 min filtration
cycle before this backwash. Directly after the final backwash, the
membrane was removed from its housing and soaked in sodium
hydroxide solution (pH 12) for at least 1 h. Those three samples,
i.e. last backwash water after 5 h, permeate before this backwash
and soaking solutions, were weighed and NPOC, UV and turbid-
ity were measured. The NPOC mass balance was based on these
samples. The polymeric membrane cannot withstand pH 12, so no
soaking sample was obtained for the polymeric membrane exper-
iment. We assume that the NPOC collected from the backwash
is part of the reversible fouling, whilst the NPOC collected from
soaking at pH 12 is part of the irreversible fouling. The concentra-
tion of NPOC, the turbidity and the UV values of the feed water,
backwash water, permeate and CIP waste were measured, as well
as the volume (by measuring the mass) of the collected samples.
The NPOC mass balance is based on a single backwash and thus
given by (Vb × cb ) + 1/Nb ((Vs × cs ) + (Vp × cp )) = 1/Nb (Vf × cf ), where
Nb is the number of backwash cycles in one experiment, V is the
volume, and c is the concentration of the NPOC of the stream
denoted by the subscripts – b, s, p, and f, which represent back-
wash, soaking solution, permeate and feed, respectively. The
concentration of NPOC and turbidity and volume of the back-
wash was corrected for the amount of feed water and permeate
that inevitably was mixed with the backwash water (backwash
with permeate). The percentage found by the mass balance is
((Vb × cb ) + 1/Nb ((Vs × cs ) + (Vp × cp )0/(1/Nb (Vf × cf )) × 100. A similar
mass balance for particles cannot be made based on the turbidity
measurements performed in this study.
2.3.2. Test at comparable backwash flux
Additional tests with the TiO2 and the polymeric membrane
were done to elucidate the effect of the variation in the backwash
flux. The tests were performed as described above with the fol-
lowing differences. A new batch of Noord Bergum lake water was
obtained (2nd batch, Table 3), and a new membrane was used for
each test. A number of the channels in the polymeric membrane
were blocked with two-component resin, decreasing the surface
area of the membrane to 0.0045 m2 . The feed flow was lowered to
11.3 mL/min in order to maintain the same flux in all experiments.
The duration of the filtration tests with the polymeric membrane
was about 5 h. For the test with the TiO2 membrane, the backwash
pressure was lowered by decreasing the air pressure. The cumu-
lative backwash production and cumulative permeate production
were stored and analysed for the additional tests with the TiO2
and polymeric membrane. For the calculation of the mass balances
Nb = 1 (see the formulas in Section 2.3.1).
2.3.3. Data handling
Reversible and irreversible fouling was determined from oper-
ational data. For the reversible fouling the decrease in the TMP by
a backwash was used, averaged over the total number of backwash
cycles. The TMP at time t (TMPt ) minus the TMP at t = 0 (TMP0 ) was
plotted versus the permeate volume to determine the irreversible
fouling, which was obtained by the slope of a linear fit to the TMPs
just after each of the backwashes (see Fig. 2B).
2.3.4. Membrane cleaning procedure
After a filtration test, the ceramic membrane was cleaned by
soaking; first in citric acid (1%) for a minimum of 2 h and then
in sodium hypochlorite (3000 ppm) for a minimum of 2 h. After-
wards, the membrane was rinsed with MilliQ water and stored
in MilliQ water until the next experiment. Before starting a new
test the ceramic membrane was characterised with MilliQ water
at 150 L m−2 h−1 to verify TMP0 . A new measurement was only
started if the pressure had not changed. The polymeric membrane
was not cleaned; a new module was used for each test. The addi-
tional tests at different backwash pressures were performed with
new membranes.
3. Results and discussion
3.1. Membrane characterisation
The used ceramic membranes were characterised by
permporometry, XPS and zeta-potential measurements (Table 4).
For the polymeric membrane the measured pore size is the same
as that given by the supplier. For the Al2 O3 , and ZrO2 ceramic
membranes the measured mean flow pore size deviates a factor
2.4 and 1.6, respectively, from the pore size in the supplier infor-
mation. For TiO2 and especially SiC the deviation is larger; a factor
of 5 and 24, respectively. The SiC membrane is very hydrophilic
(resulting in a water up-take when in contact with water) which
 B. Hofs et al. / Separation and Purification Technology 79 (2011) 365–374 369

A 0.8
may have influenced the used method, as in permporometry a
liquid is displaced from the membrane. However, in the literature
no references were found reporting on shifts to larger pore size as
0.6
determined with porometry in connection with hydrophilicity of
TMP (bar)

the membrane.
0.4 The porosity of the membranes is comparable, 72–78%, except
for the SiC membrane which has a much lower porosity of 40%
0.2 (Table 4). A clean water flux obtained with milliQ at 150 L m−2 h−1
resulted in an initial TMP, of 0.042, 0.073, 0.10, 0.14, and 0.19 bar,
0.0 for the TiO2 , SiC, polymeric, Al2 O3 , and ZrO2 membrane, respec-
0 5000 10000 15000 20000 tively. Initially, we also selected a SiO2 membrane for this study.
Time (s) During the clean water flux tests however, a steady decrease in
B TMP was observed for the SiO2 membrane, probably as a result of
0.8
a slow dissolution of the membrane layer of the SiO2 membrane
[26]. Therefore the investigation with this ceramic membrane
TMPt - TMP0 (bar)

0.6 was aborted. This problem was not encountered with the other
membranes.
0.4 For the SiC membrane the iso-electric point was determined
from zeta-potential measurements and found to be in accor-
0.2 dance with the literature values (Table 4). Unfortunately, repeated
y = 1.60*10-4x + 0.071 attempts to determine the iso-electric point of the other ceramic
2
R = 0.965 membranes were unsuccessful. We assume that the values found
0.0
in literature for particles and/or surfaces made of the same material
0 200 400 600 800
Produced permeate (L/m2) as the ceramic membranes are representative.
The membrane composition was determined with XPS analysis,
Fig. 2. (A) Typical fouling experiment, from which reversible fouling was deter- to be able to compare the investigated membrane composition to
mined. (B) Determination of irreversible fouling for the Al2 O3 membrane. The literature data and supplier information. Furthermore, the elemen-
straight line in Fig. 2B is a linear fit to the first point of each filtration cycle (after
tal composition of the membranes facilitates analysis of variations
the backwash), omitting the starting point of the first filtration cycle. Flux was
150 L m−2 h−1 . in different membrane batches. The presence of the expected
elements was confirmed in the different layers, but there were
significant deviations from expectations based on supplier infor-
mation (Table 5). For the ZrO2 and Al2 O3 membranes there were
some small deviations. The presence of C for instance, is probably

Table 4
Membrane properties – comparison of measurements and information from supplier and literature.

Type Pore size (␮m) Iso-electric point Volume

porositya (%)

Measureda Supplier Measured Literature

j
Al2 O3 0.24 0.1 8–9.4b , c , d 73
j
TiO2 0.51 0.1 5.1–6.4b , c 72
j
ZrO2 0.16 0.1 6.3–7.1b , e , f 74
SiC 0.96 0.04 2.6 2.5–3.5d , g , h 40
Polymeric 0.14 0.14 – 4–5i 78
a
Mean flow pore size and porosity were determined by porometry as described in the Section 2 (Section 2.1).
b
[27].
c
[28].
d
[29].
e
[30].
f
[31].
g
[32].
h
[33].
i
According to supplier.
j
Could not be determined.

Table 5
Membrane composition in % as determined by XPS.

Location of analysis Analysis C O Al Si Ti Zr

ZrO2 Top layer 13.9 55 9 16.7

Support layer 15.6 53.2 21.4 3.6 1.7 0.2
TiO2 Top layer 20.9 51.1 7.4 0.9 11.5
Support layer 14.4 52.7 22.7 4 2
Al2 O3 Top layer 8.8 58.1 26.1 2.7 0.12
Support layer 10.5 54.1 26.2 4.4 2
SiC Top layer 24.2 45.6 3.5 26.7
Support layer 20.0 53.0 2.1 24.9

The expected composition follows from the formula given in de first column (e.g. for ZrO2 , Zr is 33 and O is 66). Some other elements (Na, Mg, Ca) were also found, but in
low amounts (generally <2%, at maximum about 5%).
370 B. Hofs et al. / Separation and Purification Technology 79 (2011) 365–374

due to organic contaminations from air [34]. The found percentages
(9–24%) are in agreement with other XPS measurements on ceramic
membranes [35]. For the ZrO2 top layer we find 16.7% Zr and 9%
Al. This gives an expected O-content of 16.7 × 2 + 9 × 1.5 = 46.9%;
whereas 55% was measured. The relatively high value for O in the
top layer can be explained by the presence of contaminants as indi-
cated by the amount of C found; it is likely that the contaminants
consist partly of C and O. Another contribution to a higher O-content
than expected is the fact that the surface of ZrO2 has a higher O-
content than the bulk. The XPS shows that support layer of the
ZrO2 membrane is made up predominantly of O and Al, in accor-
dance with supplier information. However, contaminations (high
C-content, higher O-content than expected, and some Si) are also
present on/in the support layer. Also, it is clear that the support
layer is already visible in the analysis of the top layer.
The top layers of the TiO2 and Al2 O3 membrane also show a
relatively high O-content. For the TiO2 and Al2 O3 membranes we
would expect 36.0 and 44.6% O, respectively – from the presence of
Al, Ti and Si – where 51 and 58% are found, respectively. The support
layers also show a higher O-content than expected. Again, these
higher values are probably due to the presence of contaminants on
the surface and support layer of the membrane. The high level of O-
content found for the SiC membrane can again be due to the above
mentioned contamination. Alternatively, the membrane could have
adsorbed water molecules, as the SiC membrane is very hydrophilic
(a dry membrane takes up water when put into contact with it), or
some SiO2 -groups.
3.2. Fouling experiments
3.2.1. TMP increase
An increase in the feed pressure due to reversible and irre-
versible fouling of an Al2 O3 membrane due to direct filtration with
the Noord-Bergum lake water was observed (Fig. 2A). Reversible
fouling was determined from Fig. 2A, by calculating the average
increase in TMP within all filtration cycles. The irreversible fouling
in bar m2 L−1 was determined from the slope of the fouling curve
as shown graphically in Fig. 2B. Reproducibility of the fouling mea-
surements with the different membranes, expressed as reversible
and irreversible fouling and based on a duplicate measurement was
usually within 10%, except for the ZrO2 membrane. We were unable
to clean the ZrO2 membrane completely with the described clean-
ing protocol. The initial TMP of the ZrO2 membrane with milliQ was
0.19 bar with the virgin membrane, this increased to 0.25 bar after
the first fouling test and cleaning. The reversible and irreversible
fouling of the ZrO2 membrane increased in further tests; we only
show results for the first test in this paper.
The fouling experiments were performed with the same water
type for all the experiments. The TMP0 was 0.042, 0.082, 0.073,
0.10, 0.085, 0.14, and 0.19 bar, for the TiO2 , TiO2 *, SiC, polymeric,
Table 6
Volume/Area ratio of the membranes and used backwash fluxes.
Membrane V/A (m) Backwash flux (×103 L m−2 h−1 )
Al2 O3 1.9 1.2
TiO2 1.9 1.5 or 2.8
ZrO2 1.9 0.57
SiC 3.6 1.0
Polymer 0.28 0.33 or 1.3
had a large variation in the backwash flux however, it may have
played a significant role in influencing the fouling, especially for
the polymeric and TiO2 membrane as the backwash fluxes were
relatively small and large (0.33 and 2.8 × 103 L m−2 h−1 ), respec-
tively. Therefore we performed another series of experiments with
these membranes with more comparable backwash fluxes, nar-
rowing the range of backwash flux to 4–10 times (Table 6). It is
found that lowering the backwash flux for the TiO2 membrane
from about 2800–1500 L m−2 h−1 decreases reversible fouling from
1.2 ± 0.01 to 0.08 ± 0.02 bar – note that this is barely significant –
and does not affect irreversible fouling significantly (4.3 ± 0.5 vs.
4.3 ± 1.4 × 10−5 bar m2 L−1 ). For the polymeric membrane, increas-
ing the backwash flux from about 330–1300 L m−2 h−1 has no
significant effect on reversible fouling, but seems to decrease irre-
versible fouling. Furthermore the filtration run lasted about 5 h with
the higher backwash flux whereas with the lower flux it had to be
stopped after 2.5 h due to reaching the maximum reachable pres-
sure (pump). However, the observed differences with variation in
backwash flux are small compared to the differences between the
different types of ceramic and polymeric membranes. Apart from
this, the minor differences could be caused by the lower turbidity
of the second batch of water compared to the first batch.
Regardless of the backwash flux, the reversible fouling depends
on membrane type and follows the following series: poly-
meric ≈ Al2 O3 ≈ ZrO2 > TiO2 > SiC (Fig. 3). The irreversible fouling
also depends on the membrane type and follows the following
series: polymeric > ZrO2 > Al2 O3 > TiO2 > SiC. There seems to be a
relation between the measured pore size of the membranes and
the increase in TMP due to reversible and irreversible fouling. The
membranes with the highest measured pore size show the low-
est increase in TMP due to fouling. The SiC membrane, which has
the lowest iso-electric point and highest hydrophilicity and the
highest measured pore size, has a TMP increase due to irreversible
fouling of only (2 ± 8) × 10−6 bar m2 L−1 (average ± standard devia-
tion). This could be seen as a confirmation of the hypothesis that the
membrane with the highest negative charge or hydrophilicity fouls
the least [10,37,38], even though other studies show that this is a
0.003
reversible
Reversible fouling (bar)

0.4
polymeric*, Al2 O3 , and ZrO2 membrane, respectively. The linear fits
Irreversible fouling

irreversible
to determine the irreversible fouling had a R2 of 0.95 ± 0.03 (aver-
(bar*m /L)

0.3 0.002
age ± standard deviation) for all membranes, except the SiC. For
2

the SiC membrane the TMP increase due to irreversible fouling was
limited, resulting in a low R2 (about 0.3). 0.2
Initially experiments were performed with a fixed back- 0.001
wash pressure, but additional experiments were done to obtain 0.1
a set of experiments with comparable backwash flux (about
600–1500 L m−2 h−1 , see Table 6). Criteria for an efficient backwash 0 0
er
C
3

*
2*

procedure include; sufficiently high pressure (about 3–4 times fil-

er
O

Si

m
O

m
Zr
Ti
2

ly
Ti
Al

ly

tration pressure) [36] and sufficient volume to carry the fouling

Po

Po

from the membrane to the backwash vessel. Usually a backwash
flux of about 2–3 times the filtration flux is used. We meet the
criteria for efficient backwash as the backwash flux used varied
from 2 to 20 times the filtration flux and the backwash volume was
at least 4 times the volume of the tubing + membrane. Since we
Membrane type
Fig. 3. TMP increase due to reversible and irreversible fouling. TiO2 *, Poly-
mer* = results with these membranes of tests performed with second batch of water
at comparable; backwash flux (see Table 6). ZrO2 results are based on the first
non-reproducible test with the virgin membrane.
 B. Hofs et al. / Separation and Purification Technology 79 (2011) 365–374
simplification [39,40] and that other factors like surface rough-
ness and membrane structure (average pore size and porosity) and
geometry play an important role as well [41–43].
The irreversible fouling found in this work is comparable to val-
ues in literature; irreversible fouling of a ceramic membrane by
a synthetic water after coagulation was found to vary from about
10−5 to about 10−6 bar m2 L−1 [44], and irreversible fouling of a
PES/PVP UF membrane by surface water was found to be about
10−4 bar m2 L−1 [45]. These two papers [44,45] use different meth-
ods to calculate the irreversible fouling. We basically followed the
method by Meyn et al. [44]. Recalculating the irreversible fouling
obtained in this study to conform to the method used by Abrahamse
et al. [45], gives values ranging from 2 × 10−6 to 2 × 10−4 bar m2 L−1
(for SiC to polymeric, respectively). The latter method thus gives
smaller values for irreversible fouling than the method employed
in this study.
Another factor that plays a role in fouling of membranes is the
geometry of the membranes [43]. The volume over area ratio, V/A
(Table 6), for the Al2 O3 , TiO2 and ZrO2 membranes is the same,
1.9, since the membranes have the same geometry. When V/A is
relatively high, the concentration of accumulated foulants during a
filtration cycle in the membrane (and possibly tubing in front of the
membrane) remains relatively low. Fouling might than be expected
to be lower for membranes with a higher V/A and vice versa. The
polymeric membrane has the lowest V/A, 0.28, and should therefore
be expected to have the highest fouling, which is indeed observed
(Fig. 3). The V/A of the polymeric membrane is a factor 7 lower than
that of the Al2 O3 , TiO2 and ZrO2 membranes, whilst the difference
in fouling is a factor of 1–4 for reversible fouling and 10–50 for
irreversible fouling. We hypothesize that the differences found here
in fouling between the polymeric and Al2 O3 or ZrO2 membranes
can mostly be attributed to the differences in V/A.
The SiC membrane has the highest V/A of the used membranes,
3.6, and shows the lowest fouling (Fig. 3). The difference between
V/A for the SiC and the Al2 O3 , TiO2 and ZrO2 membranes is a factor of
2, whilst the difference in fouling is a factor of 10–30 for reversible
fouling and 20–100 for irreversible fouling. It seems unlikely that
the difference V/A alone can explain the observed difference in
fouling for the SiC membrane.
3.2.2. Removal efficiency
Removal of particles based on turbidity is more than 95% for
all membranes except for the SiC membrane, which shows 92%
removal. This suggests that the pore size of the SiC membrane is
larger than the pore size of the other membranes, as shown by the
measurement of the pore size (Section 3.1).
Removal of NPOC is around 30% for the ceramic membranes, and
13–25% for the polymeric membrane (Fig. 4). The removal of 13%
for a polymeric membrane is comparable with results published for
membranes with similar pore sizes [46–48], and so is the removal of
30% for the ceramic membranes [2,11]. The low removal of 13% was
found for the 2.5 h long experiment with the polymeric membrane,
whereas a higher value of 25% was found when the test lasted the
full 5 h (which was the duration of the test with the ceramic mem-
branes). Apparently the removal increases with the duration of the
test, indicating that the increase of accumulated irreversible fouling
increases the NPOC removal efficiency. UV removal shows simi-
lar trends as the NPOC removal. The lower UV and NPOC removal
for the polymeric membrane compared to the ceramic membranes
could on one hand be expected, as the polymeric membrane has a
slightly larger pore size compared to ceramic membranes, accord-
ing to the suppliers. On the other hand, it is unexpected as the pore
size measurements give a very large pore size for the SiC membrane
compared to the value given by the supplier (0.9 ␮m and 0.04 ␮m,
respectively). As the SiC membrane shows the same removal as the
other membranes, this indicates that the pore size determination
371
100
75
Removal (%)
NPOC
50 Turbidity
UV
25
0
er
C
3
*
2*
er
O
O
Si
ym
O
m
Zr
Ti
2
Ti
Al
ly
l
Po
Po
Membrane type
Fig. 4. Removal of NPOC, turbidity and UV by the different membranes. TiO2 *, Poly-
mer* = results with these membranes of tests performed with second batch of water.
Shown are average and standard deviation (error bar). Where no error bar is shown,
the duplicate measurement resulted in the same values (except for ZrO2 , where no
duplicate measurement could be performed).
using porometry is not well suited for this membrane. This is in
contradiction with the results obtained for the TMP increase due to
fouling, which correlate rather well with the measured pore size.
In principle, the rejection of negatively charged NPOC may be
increased by the presence of negative charges on the membrane.
The SiC membrane is negatively charged as it has an iso-electric
point of 2.6. Fouling probably decreases the charge on the SiC mem-
brane, but it probably remains negatively charged. Based on data
in Table 3 and assuming the missing cations necessary for elec-
troneutrality are Na+ , we estimate the ionic strength of the water is
about 9 mM. This gives a Debye screening length [49] (the distance
over which the electrostatic potential drops to 1/e of the origi-
nal value) of about 3 nm. If the pore size of the SiC membrane is
larger than that of the other membranes, for example 0.2 ␮m, the
pore diameter would be only slightly decreased by electrostatic
screening (about 10%; in the case described above, 10 nm from the
edge of the pore the electrostatic potential has dropped to about
5% of its value at the edge). The electrostatic effect therefore is not
expected to decrease the effective pore size significantly in these
experiments and is expected to play only a minor role in removal
of the negatively charged fraction of NOM for this membrane.
Besides the properties of the membrane, also the properties
of the fouling layer are expected to influence the removal effi-
ciencies (see Section 3.2.3). Since here we analysed the removal
efficiency at the end of the tests, the charge on the membranes is
influenced by the fouling. We did not measure to which extent the
fouling changes the charge size and/or sign. If one wants to esti-
mate the effect of charge on the removal efficiency of NPOC, these
data (obtainable by e.g. streaming potential measurements on the
membrane) would be needed.
3.2.3. Fouling load
The NPOC load of the membranes was measured for reversible
fouling by collecting the backwash water of the last filtration cycle,
and for irreversible fouling by analysing CIP soaking water (see Sec-
tion 2.3). For the ZrO2 membrane we were unable to measure a
duplicate and only the results of the first test are given as an indi-
cation. The reversible fouling load decreases in the following order;
(Al2 O3 > ZrO2 ≈ TiO2 ) ≈ SiC > polymeric (Fig. 5).
The irreversible fouling load after 5 h of operation seems highest
for Al2 O3 and ZrO2 and lowest for TiO2 and SiC, but the error bars
for the latter two are rather large. Differences between the NPOC
load of the different membranes are within a factor of 1.5, whereas
the differences in the TMP increase due to fouling are much more
pronounced (Fig. 3).
372 B. Hofs et al. / Separation and Purification Technology 79 (2011) 365–374

0.10 0.30 0.25 0.30

reversible load

Irreversible load (g/m )

Reversible load (g/m )

Irreversible load (FNU/m )

Reversible load (FNU/m )
2
2

2
2
0.08 irreversible load
0.20
0.20
0.06 0.20
0.15
0.04
0.10 0.10
0.02 0.10

0.05
0.00 0.00 reversible load
irreversible load

er
C
3

*
2*

er
O

Si 0.00 0.00

m
O

m
Ti

Zr
2

ly
Ti
Al

ly
Po

Po

er
C
3

*
2*

er
O

Si

m
O

m
Zr
Ti
2

ly
Membrane type

Ti
Al

ly
Po

Po
Membrane type
Fig. 5. NPOC load of the membranes. Note that the NPOC load of the polymeric
membrane is after 2.5 h of filtration, whereas for the other membranes (including
Fig. 6. Turbidity load of the membranes. Note that the turbidity load of the poly-
polymer*) it is after 5 h. TiO2 *, Polymer* = results with these membranes of tests
meric membrane is after 2.5 h of filtration, whereas for the other membranes
performed with second batch of water.
(including polymer*) it is after 5 h. TiO2 *, Polymer* = results with these membranes
of tests performed with second batch of water.

Mass balance analysis (see Section 2.3) shows that 85 ± 8% (aver-

age + standard deviation) of the total NPOC in the feed water is
accounted for in the backwash water, soaking solution and perme-
ate. Mass balance analysis shows that 88, 87, 90, 68, and 82% of the
expected NPOC was accounted for, for the Al2 O3 , TiO2 , TiO2 *, ZrO2 ,
and SiC membrane, respectively. This shows that a significant part
of the fouling NPOC is probably still attached to the membranes
after soaking in 0.01 M NaOH. Especially the ZrO2 membrane is not
cleaned well. Additionally, for the ZrO2 membrane the CIP removed
only a part of the irreversible NPOC load, which is shown by the
observed non-reproducibility of the fouling experiment. The low
mass balance percentage of NPOC accounted for can also be due to
the fact that we based our analysis on the last filtration cycle. It is
likely that NPOC removal increased with increasing filtration time
due to the build up of a fouling layer that acts as an extra barrier.
However, for the experiments with comparable backwash flux we
collected all the backwash water and the results of the mass bal-
ance were similar. Thus, it seems that not all NPOC was removed
from the membranes by soaking in a NaOH solution (pH 12), even
though this method is used for cleaning of membranes [50]. This
suggests that for ceramic membranes, a more extensive or aggres-
sive treatment is necessary to remove all the NPOC, for example a
more concentrated NaOH solution (e.g. pH 13) at 50 ◦ C [51].
No clear relation seems to exist between the NPOC load of foul-
ing and the TMP increase due to reversible or irreversible fouling,
although the membranes with the lowest irreversible NPOC load
(the SiC and TiO2 membranes), are also the ones that show the
least irreversible fouling characteristics (compare Figs. 3 and 5).
The fact that no clear relation seems to exist between either NPOC
load or turbidity load and fouling shows that the combined effect
of NOM and particles determine the TMP increase due to fouling
[52].
The turbidity load of the membranes is shown in Fig. 6.
The reversible turbidity load is highest for the TiO2 membrane
(although the error bar shows this might not be very signifi-
cant), which is advantageous since this turbidity fraction is easily
removed with a backwash and does not contribute to long term
irreversible fouling. One might expect a higher reversible turbidity
load for the SiC membrane, but this membrane did not remove as
much turbidity as the other membranes (see Fig. 4). The level of
reversible turbidity load is apparently sufficiently high to prevent
any pressure build-up due to irreversible fouling. Furthermore, less
turbidity will build up because of the potentially more open struc-
ture of the SiC membrane as shown by the pore size measurements.
Unfortunately, a relation between the turbidity load and the
TMP increase could not be identified. In this respect it is noted
that turbidity is a very indirect measure of particle load. A more
advanced method to determine the particle load, which can distin-
guish the particles’ sizes and particle size distribution is needed
to assess the composition of the particles in the reversible and
irreversible fouling. With this information, a TMP increase may
possibly be related to a particle load parameter and membrane
type.
For a thorough comparison of membrane fouling using different
types of membranes, the following membrane properties should
preferably be measured or known: membrane chemistry, mem-
brane geometry, membrane permeability or resistance, pore size
distribution, porosity, contact angle, zeta-potential of the support
and the top layer, thickness of the top layer, and surface roughness.
The composition (concentrations of various ions, ionic strength,
DOC, LC-OCD, turbidity, zeta-potential) of the feed water type
should also be measured, as well as the composition of the back-
wash water and the fouling layer. We did not determine the surface
roughness, the thickness of the top layer and the contact angles of
the membranes, nor did we perform an extensive analysis of the
reversible and irreversible fouling composition. In order to under-
stand the differences in fouling a more detailed analysis is needed.
However, this is beyond the scope of the investigation described
here.
4. Conclusions
We compared the permeability and fouling of four different
ceramic membranes and one polymeric membrane using surface
water.
Pore size measurements by permporometry showed that the
pore size of the Al2 O3 and ZrO2 were similar to the values given
by the supplier. The measured pore size for the TiO2 and SiC mem-
brane deviated by a factor of 5 and 24 from those given by the
supplier, respectively. The properties of the membranes may cause
deviations in the used method; the SiC membrane is exceptionally
hydrophilic.
XPS measurements confirmed the membrane compositions as
documented by the suppliers, although the measurements showed
that all membrane samples were contaminated with carbon and
oxygen.
The TiO2 and especially the SiC membrane showed a low TMP
increase due to low reversible and irreversible fouling, compared
to the Al2 O3 , ZrO2 and polymeric membranes. For the polymeric
membrane the differences can probably be attributed to differences
in the V/A ratio between the membrane modules. The TMP increase
due to fouling increases with decreasing measured pore size.
 B. Hofs et al. / Separation and Purification Technology 79 (2011) 365–374
Removal of NOM and UV was lowest for the polymeric mem-
brane (13–25%) and higher and comparable (around 30%) for the
different ceramic membranes. The removal of turbidity was >95%
for all membranes except the SiC membrane.
The reversible NPOC fouling load decreases in the following
order: (Al2 O3 > (ZrO2 ≈ TiO2 )) ≈ SiC > polymeric and the irreversible
fouling load after 5 h of operation decreases in the order;
Al2 O3 ≈ ZrO2 > TiO2 ≈ SiC. The load of reversible and irreversible
NPOC fouling, however, was not very different for the Al2 O3 , TiO2 ,
ZrO2 , and SiC membranes, suggesting that these parameters cannot
be used as an indicator of TMP increase.
Mass balance analysis of the NPOC load on the membranes, in
the permeate and in the backwash, showed that 85 ± 8% of the
expected NPOC could be accounted for. Probably, the soaking at
pH 12 did not remove a significant amount of the fouling materials
from the membranes. This is especially true for the ZrO2 mem-
brane where only 68% of the NPOC was accounted for. This suggests
that a more aggressive CIP is necessary for the complete removal
of irreversible fouling.
Acknowledgements
This research was conducted in the framework of the Techneau
consortium (financed by the EU) and the Joint Research Program
(BTO) of the Dutch drinking water companies. We thank one
reviewer for his/her comments, which helped to greatly improve
the quality of this paper. We thank Z. Borneman and H.A. Teunis
(both from the EMI, TU Twente, The Netherlands) and T. Luxbacher
(from Anton Paar, Graz, Austria) for their efforts to characterise the
membranes.
References
[1] H.K. Oh, S. Takizawa, S. Ohgaki, H. Katayama, K. Oguma, M.J. Yu, Removal of
organics and viruses using hybrid ceramic MF system without draining PAC,
Desalination 202 (2007) 191–198.
[2] C. Moulin, M.M. Bourbigot, A. Tazipain, F. Bourdon, Design and performance of
membrane filtration installations – capacity and product quality for drinking-
water applications, Environmental Technology 12 (1991) 841–858.
[3] T. Tsuru, Inorganic porous membranes for liquid phase separation, Separation
and Purification Methods 30 (2001) 191–220.
[4] R.J. Ciora Jr., P.K.T. Liu, Ceramic membranes for environmental related applica-
tions, Fluid: Particle Separation Journal 15 (2003) 51–60.
[5] B. Hofs, S.G.J. Heijman, J.Z. Hamad, M. Kennedy, G. Amy, Ceramic microfiltra-
tion with a sub-micron PAC pre-coat for water treatment, in: J.P.T. Melin, M.
Dohmann (Eds.), Aachener Tagung Wasser und Membranen, vol. 8, Aachen,
2009, p. W12.
[6] B. Van Der Bruggen, C. Vandecasteele, T. Van Gestel, W. Doyen, R. Leysen, A
review of pressure-driven membrane processes in wastewater treatment and
drinking water production, Environmental Progress 22 (2003) 46–56.
[7] E.R. Cornelissen, B. Hofs, U. Muller, E.F. Beerendonk, S.G.J. Heijman, Direct and
hybrid ceramic microfiltration in water treatment, in: TECHNEAU: Safe Drink-
ing Water from Source to Tap, IWA Publishing, Maastricht, 2009, 83–97.
[8] S. Lee, G. Park, G. Amy, S.K. Hong, S.H. Moon, D.H. Lee, ChoF J., Determination of
membrane pore size distribution using the fractional rejection of nonionic and
charged macromolecules, Journal of Membrane Science 201 (2002) 191–201.
[9] Q. Zhang, Y. Fan, N. Xu, Effect of the surface properties on filtration performance
of Al2 O3 –TiO2 composite membrane, Separation and Purification Technology
66 (2009) 306–312.
[10] G.N.B. Barona, B.J. Cha, B. Jung, Negatively charged poly(vinylidene fluoride)
microfiltration membranes by sulfonation, Journal of Membrane Science 290
(2007) 46–54.
[11] A. Lerch, S. Panglisch, P. Buchta, Y. Tomitac, H. Yonekawa, K. Hattori, R. Gimbel,
Direct river water treatment using coagulation/ceramic membrane microfil-
tration, Desalination 179 (2005) 41–50.
[12] T. Meyn, A. Bahn, T.O. Leiknes, Significance of flocculation for NOM removal
by coagulation–ceramic microfiltration, Water Science and Technology: Water
Supply 8 (2008) 691–700.
[13] N. Shirasaki, T. Matsushita, Y. Matsui, K. Ohno, Effects of reversible and irre-
versible membrane fouling on virus removal by a coagulation–microfiltration
system, Journal of Water Supply: Research and Technology: AQUA 57 (2008)
501–506.
[14] A. Buekenhoudt, Stability of Porous Ceramic Membranes, in: R. Mallada, M.
Menendez (Eds.), Inorganic membranes; synthesis, characterization and appli-
cations, Elsevier, Amsterdam, 2008, pp. 1–31.
373
[15] T. Van Gestel, C. Vandecasteele, A. Buekenhoudt, C. Dotremont, J. Luyten,
B. Van der Bruggen, G. Maes, Corrosion properties of alumina and titania
NF membranes, Journal of Membrane Science 214 (2003), PII S0376-
7388(0302)00513-00517.
[16] N. Park, B. Kwon, I.S. Kim, J. Cho, Biofouling potential of various NF membranes
with respect to bacteria and their soluble microbial products (SMP): character-
izations, flux decline, and transport parameters, Journal of Membrane Science
258 (2005) 43–54.
[17] A. Dobrak, B. Verrecht, H. Van den Dungen, A. Buekenhoudt, I.F.J. Vankelecom,
B. Van der Bruggen, Solvent flux behavior and rejection characteris-
tics of hydrophilic and hydrophobic mesoporous and microporous TiO2
and ZrO2 membranes, Journal of Membrane Science 346 (2010) 344–
352.
[18] http://www.techneau.org.
[19] U. Mueller, R. Schaefer, C. Zawadsky, K. Boeckle, M. Boesl, I. Eberhagen,
B. Hambsch, Techneau Report D 2.3.2.7. Removal of phages and nanopar-
ticles by ceramic membranes, in: DVGW Water technology Centre (TZW),
2009.
[20] U. Mueller, R. Schaefer, M. Witte, K. Boeckle, M. Boesl, I. Eberhagen, B. Hambsch,
Techneau Report D 2.3.2.2, in: Interim Report Removal of particulate mat-
ter by ceramic membranes during surface water treatment in, DVGW Water
technology Centre (TZW), 2008.
[21] L. Fan, J. Harris, F. Roddick, N. Booker, Fouling of microfiltration membranes by
the fractional components of natural organic matter in surface water, Water
Science and Technology: Water Supply (2002) 313–320.
[22] N. Lee, G. Amy, J.P. Croue, H. Buisson, Identification and understanding of fouling
in low-pressure membrane (MF/UF) filtration by natural organic matter (NOM),
Water Research 38 (2004) 4511–4523.
[23] J.M. Laine, D. Vial, P. Moulart, Status after 10 years of operation – overview of
UF technology today, Desalination 131 (2000) 17–25.
[24] H. Huang, K. Schwab, J.G. Jacangelo, Pretreatment for low pressure membranes
in water treatment: a review, Environmental Science and Technology 43 (2009)
3011–3019.
[25] M. Wegmann, B. Michen, T. Luxbacher, J. Fritsch, T. Graule, Modifi-
cation of ceramic microfilters with colloidal zirconia to promote the
adsorption of viruses from water, Water Research 42 (2008) 1726–
1734.
[26] D. Wolff-Boenisch, S.R. Gislason, E.H. Oelkers, C.V. Putnis, The dissolution rates
of natural glasses as a function of their composition at pH 4 and 10.6 and
temperatures from 25 to 74 ◦ C, Geochimica Et Cosmochimica Acta 68 (2004)
4843–4858.
[27] M. Kosmulski, Compilation of PZC and IEP of sparingly soluble metal oxides
and hydroxides from literature, Advances in Colloid and Interface Science 152
(2009) 14–25.
[28] M. Mullet, P. Fievet, J.C. Reggiani, J. Pagetti, Surface electrochemical properties
of mixed oxide ceramic membranes: zeta-potential and surface charge density,
Journal of Membrane Science 123 (1997) 255–265.
[29] X.H. Wang, Y. Hirata, Colloidal processing and mechanical properties of SiC with
Al2 O3 and Y2 O3 , Nippon Seramikkusu Kyokai Gakujutsu Ronbunshi/Journal of
the Ceramic Society of Japan 112 (2004) 22–28.
[30] Y.K. Leong, N. Katiforis, D.B.O. Harding, T.W. Healy, D.V. Boger, Role of rheol-
ogy in colloidal processing of ZrO sub 2, Journal of Materials Processing and
Manufacturing Science 1 (1993) 445–453.
[31] M. Mao, D. Fornasiero, J. Ralston, R.S.C. Smart, S. Sobieraj, Electrochemistry
of the zircon–water interface, Colloids and Surfaces A: Physicochemical and
Engineering Aspects 85 (1994) 37–49.
[32] S. Sano, T. Banno, M. Maeda, K. Oda, Y. Shibasaki, Slip casting and sintering of
silicon carbide (part 1) – slip preparation of silicon carbide powder produced by
Acheson method, Nippon Seramikkusu Kyokai Gakujutsu Ronbunshi/Journal of
the Ceramic Society of Japan 104 (1996) 984–988.
[33] S.H. Yeh, C.C. Wan, Codeposition of SiC powders with nickel in a Watts bath,
Journal of Applied Electrochemistry 24 (1994) 993–1000.
[34] D.R. Cousens, B.J. Wood, J.Q. Wang, A. Atrens, Implications of specimen prepara-
tion and of surface contamination for the measurement of the grain boundary
carbon concentration of steels using X-ray microanalysis in an UHV FESTEM,
Surface and Interface Analysis 29 (2000) 23–32.
[35] M.I. Vazquez, R. de Lara, J. Benavente, Characterisation of two microporous
ceramic membranes with different geometrical parameters: effect of protein
adsorption on electrochemical parameters, Journal of the European Ceramic
Society 27 (2007) 4245–4250.
[36] M. Kennedy, S.M. Kim, I. Mutenyo, L. Broens, J. Schippers, Intermittent
cross flushing of hollow fiber ultrafiltration systems, Desalination 118 (1998)
175–188.
[37] B. Kaeselev, J. Pieracci, G. Belfort, Photoinduced grafting of ultrafiltration
membranes: comparison of poly(ether sulfone) and poly(sulfone), Journal of
Membrane Science 194 (2001) 245–261.
[38] H. Huang, T. Young, J.G. Jacangelo, Novel approach for the analysis of bench-
scale, low pressure membrane fouling in water treatment, Journal of Membrane
Science 334 (2009) 1–8.
[39] M. Taniguchi, J.E. Kilduff, G. Belfort, Low fouling synthetic membranes
by UV-assisted graft polymerization: monomer selection to mitigate foul-
ing by natural organic matter, Journal of Membrane Science 222 (2003)
59–70.
[40] M. Zhou, H. Liu, J.E. Kilduff, R. Langer, D.G. Anderson, G. Belfort, High-
throughput membrane surface modification to control NOM fouling,
Environmental Science and Technology 43 (2009) 3865–3871.
374 B. Hofs et al. / Separation and Purification Technology 79 (2011) 365–374
[41] E.M. Vrijenhoek, S. Hong, M. Elimelech, Influence of membrane surface prop-
erties on initial rate of colloidal fouling of reverse osmosis and nanofiltration
membranes, Journal of Membrane Science 188 (2001) 115–128.
[42] C.R. Reiss, J.S. Taylor, C. Robert, Surface water treatment using nanofiltra-
tion – pilot testing results and design considerations, Desalination 125 (1999)
97–112.
[43] S. Panglisch, Formation and prevention of hardly removable particle layers in
inside-out capillary membranes operating in dead-end mode, Water Science
and Technology: Water Supply (2003) 117–124.
[44] T. Meyn, T.O. Leiknes, Comparison of optional process configurations and
operating conditions for ceramic membrane MF coupled with coagula-
tion/flocculation pre-treatment for the removal of NOM in drinking water
production, Journal of Water Supply: Research and Technology: AQUA 59
(2010) 81–91.
[45] A.J. Abrahamse, C. Lipreau, S. Li, S.G.J. Heijman, Removal of divalent cations
reduces fouling of ultrafiltration membranes, Journal of Membrane Science 323
(2008) 153–158.
View publication stats
[46] A.I. Schaefer, A.G. Fane, T.D. Waite, Cost factors and chemical pretreatment
effects in the membrane filtration of waters containing natural organic matter,
Water Research 35 (2001) 1509–1517.
[47] M. Siddiqui, G. Amy, J. Ryan, W. Odem, Membranes for the control of nat-
ural organic matter from surface waters, Water Research 34 (2000) 3355–
3370.
[48] N. Lee, G. Amy, J. Lozier, Understanding natural organic matter fouling in low-
pressure membrane filtration, Desalination 178 (2005) 85–93.
[49] http://en.wikipedia.org/wiki/Debye length.
[50] N. Porcelli, S. Judd, Chemical cleaning of potable water membranes: a review,
Separation and Purification Technology 71 (2010) 137–143.
[51] M. Bartlett, M.R. Bird, J.A. Howell, An experimental study for the development
of a qualitative membrane cleaning model, Journal of Membrane Science 105
(1995) 147–157.
[52] A.I. Schaefer, U. Schwicker, M.M. Fischer, A.G. Fane, T.D. Waite, Microfiltration
of colloids and natural organic matter, Journal of Membrane Science 171 (2000)
151–172.