D. Jermann, W. Pronk, R. Kägi, N. Schaarschmidt, M. Halbeisen, M.

Boller 1

Influence of particles on UF fouling by NOM - rethinking the perception

of NOM fouling
D. Jermann, W. Pronk, R. Kägi, N. Schaarschmidt, M. Halbeisen*, M. Boller

Department of Urban Water Management, Swiss Federal Institute of Aquatic Research (eawag), Ueberlandstrasse 133, CH-8600
Dübendorf.
(E-mail: doris.Jermann@eawag.ch)
* Department of Advanced Fibers, Swiss Federal Institute of Material Science (empa), Lerchenfeldstrasse 5, CH-9014 St.
Gallen

Abstract
This study focused on the mechanistic effects of molecular interactions between inorganic particles (kaolinite) and
the two main NOM fouling fractions of polysaccharides (alginate) and humics (humic acid) in ultrafiltration
fouling. Fouling effects were studied during dead-end filtration of individual and mixture compounds as well as
the subsequent filtration of individual compounds. SEM analyses were performed to further study the fouling-
layer structure. A significant synergistic effect was observed during combined particle-NOM fouling: it was
considerably greater than the sum of particle and organic fouling alone. We explain it on the basis of our
experimental results by the individual fouling mechanisms of the two NOM and specific NOM-particle
interactions. Humic acid adsorption onto kaolinite resulted in a fouling layer of stabilized humic acid-kaolinite
colloids. In the case of alginate, simultaneous pore-blocking and cake-layer formation of the NOM and kaolinite
dominated the fouling. In both cases, incorporation of the organics in the kaolinite fouling layer resulted in a
fouling cake of significantly reduced porosity compared to individual particle filtration. This fouling layer has a
large hydraulic resistance, which explains the synergistic effect of the substances involved. Moreover, the
irreversible hydraulic NOM fouling, especially in the case of humic acid, could not be prevented by kaolinite as
different functional groups were responsible for the interactions with kaolinite and the membrane. It may be
concluded that the synergy of particle-NOM fouling is determined by the various NOM-particle interactions of the
individual NOM fouling mechanisms.

Keywords
Ultrafiltration; fouling; humic acids; polysaccharides; particles; molecular interactions.

Introduction
Ultrafiltration (UF) is an attractive solution for drinking-water treatment as it can disinfect water and
remove its turbidity by expending only a low pressure. Its main drawback is membrane fouling, which
reduces the performance of the process.

Natural organic matter (NOM) is generally recognized as the main UF foulant from a qualitative and
quantitative point of view. It has been suggested that humics are its most detrimental constituent (Jucker
and Clark, 1994, Aoustin et al., 2001, Clark and Lucas, 1998, Jones and O'Melia, 2000), although recent
studies have shown that polysaccharides may in fact be more significant (Kimura et al., 2004,
Jarusutthirak et al., 2002, Ye et al., 2005).

Some studies have indicated that particles are an important component of membrane fouling. And
although the presence of individual particles caused a relatively minor effect in synthetic model solutions,
it seemed to be more relevant in natural water (Nghiem et al., 2006, Pianta et al., 1998, Tarabara et al.,
2002). Foulant analysis suggested that NOM binds on inorganic particles which are deposited on the
membrane. NOM probably impairs particle fouling by membrane adsorption and gel-layer formation,
thus reducing the fouling reversibility and influencing the structure of the fouling layer. In addition to its
direct impact on fouling, therefore, NOM appears to play a determining role in inorganic particle fouling.

In natural water, NOM and inorganic particles interact and mutually influence their properties. The type
and amount of such interactions depends on the types of NOM and inorganic particles and determines
whether particles aggregate or are stabilized by adsorbed organics (Buffle et al., 1998). They may affect
 D. Jermann, W. Pronk, R. Kägi, N. Schaarschmidt, M. Halbeisen, M. Boller 2
fouling greatly in this way. For reasons of simplicity, however, almost all mechanistic studies have
focused either on NOM or particle fouling separately or on only one NOM component combined with
particles.

We systematically investigated molecular interactions between inorganic particles (kaolinite) and the two
main NOM fouling fractions, namely polysaccharides (alginate) and humics (humic acid), with respect to
UF fouling. This set-up is an approach to obtain more accurate fouling predictions as a function of the
quality of the raw water.

Material and methods

Ultrafiltration experiments
Model compounds chosen for the synthetic water were humic acid (HA, International Humic Acid
Society, US) and alginate (Alginic Acid, Sigma Aldrich, CH) as NOM and kaolinite (Clay Society, US)
as inorganic particle. We prepared stock solutions of organic compounds of 200 mg C/L without
prefiltration to avoid loss by adsorption onto the filter media. The dissolved organic carbon (DOC)
content was controlled by LC-OCD (Huber and Frimmel, 1992). The NOM concentrations used were 2
mg C/L for HA and 0.2 mg C/L for alginate. In the NOM mixture, the amount of alginate was kept
constant while HA was reduced to 1.8 mg C/L in order to obtain a total concentration of 2 mg C/L. The
kaolinite concentration was 100 mg/L. The background solutions consisted of deionized water (DI), 20
mM NaCl and 1 mM NaHCO3 and were adjusted to pH 7.5 with HCl/NaOH (0.2 M).

We used two different UF methods to investigate the effect of substance interactions in the feed on UF. In
the “premixed” filtration method, the compounds were mixed and equilibrated for at least 4 hours prior to
filtration. In previous experiments it was confirmed that this time is sufficient to reach interaction
equilibrium. In the “subsequent” filtration method, the kaolinite was filtered prior to the NOM. In both
methods, three filtration cycles of 1 L (or 1 L kaolinite and then 1 L NOM solution) were performed.
Between the cycles, the membranes were back-washed with 50 mL of DI. Afterwards, 250 mL of DI was
filtrated to determine the water flux after the backwash. During these experiments, the following
permeate fractions were sampled in the first cycle: 0-100, 100-200, 200-300, 300-400, 400-500 and 500-
1000 mL, and in the second and third cycles: 0-100, 100-500, 500-1000 mL.

A fresh flat sheet of 100 kDa polyethersulfone (PES) membrane (PBHK, Biomax) was used for each
experiment. The membranes had previously been stored overnight in DI and were then filtered with 15L
DI under 0.5 bar to leach the DOC out of the membrane. The test unit was a 400 mL stirring cell with
removed stirrer (Amicon). The pressure (0.5 bar) and level in the cell were maintained constant by a gear
pump connected to a glass reservoir (5 L) and a pressure control device. The permeate was measured by a
scale connected to a PC. Details of these experiments may be found in Jermann, D., ( 2007).

Analytical methods
The HA was measured by UV spectrometry at 254 nm (Varian, CARY.100scan). The alginate was
hydrolyzed with alginate lyase (EC number 4.2.2.3, Sigma Aldrich, Switzerland) for 30 minutes at neutral
pH and 30ºC and then measured by UV spectrometry at 235 nm.

A SEM was used for the morphological analysis of the membranes fouled by alginate and kaolinite before
and after physical backwashes. The membranes were sputtered with Au for a few seconds prior to the
analysis to enhance the image contrast and reduce the charging of the membrane under the electron beam.
A short sputtering time was chosen to avoid the formation of a continuous Au layer which would have
covered the pores. The SEM was operated at a low acceleration voltage (2 and 4 kV) to avoid beam
damage of the membrane.
 D. Jermann, W. Pronk, R. Kägi, N. Schaarschmidt, M. Halbeisen, M. Boller 3
Results and discussion
Flux decrease
Flux curves of individual and premixed compounds are presented in Fig. 1. It can be observed that
kaolinite alone causes only a minor flux decrease, whereas both NOM types lead to substantial fouling. In
the mixtures of kaolinite with HA or/and alginate, the total flux decreases were larger than the sum of
those of the individually filtrated substances.

Figure 1 Flux decreases in premixed UF experiments: a) HA, b) alginate, c) HA and alginate filtrated
alone and premixed with kaolinite. The fluxes refer to the water flux deionized by the clean membrane.

Such a synergistic effect indicates a significantly different structure of the fouling layer compared to the
absence of the organics. The possible underlying mechanisms are either based on previous chemical
interactions between the NOM and kaolinite or are caused by mutual effects occurring during the build-up
of the fouling layer, such as size exclusion and entrapment of the substances leading to hindered
transmission and back-diffusion. In order to gain more insight into the mechanisms, experiments were
carried out with subsequent filtration of kaolinite and NOM. The results are shown in Fig. 2.

Figure 2 Flux decrease of premixed and subsequent UF: a) kaolinite-HA, b) kaolinite-alginate. The
subsequent cases of kaolinite before HA and HA after kaolinite are shown. The data for NOM alone are
also given.

In the case of HA, a significantly smaller flux decrease can be observed during subsequent filtration than
for premixed filtration. An analysis shows that the flux decrease in subsequent filtration is the sum of the
flux decreases in the individually filtrated humic acid and kaolinite. This leads us to conclude that the
synergistic fouling effect of the substances during premixed kaolinite-HA filtration was caused by
premixing of the substances and hence depended on the time for the chemical interactions between the
substances in the feed water to take place. These mechanisms will be discussed in more detail in § 3.5.
 D. Jermann, W. Pronk, R. Kägi, N. Schaarschmidt, M. Halbeisen, M. Boller 4

In the kaolinite-alginate system, subsequent filtration of the substances causes a flux decrease at the end
of the cycles similar to that of premixed filtration, which thus also exceeds the sum of the flux decreases
of the substances filtrated alone. So it is not necessary to premix the substances to obtain the synergistic
effect of alginate and kaolinite fouling, which is due to the mutual influences of the substances during the
formation of the fouling layer. However, the course of the fouling is different for premixed and
subsequent filtration. The latter shows a faster initial flux decrease, which is typical for alginate when
filtrated alone. The possible underlying mechanisms will be further discussed in § 3.5.

Moreover, the flux decrease of kaolinite during the second and third liters is lower in the subsequent
filtration than for kaolinite filtrated alone, as can be seen by comparing Figs. 1a and 1b with Figs 2a and
2b. This indicates that both NOM substances still had an effect on the kaolinite-fouling layer after the
hydraulic backwash and are consequently still partly present on the membrane surface.

Flux decrease reversibility

The reversibility of the flux decrease was assessed on the basis of the absolute and relative reversibility
using equations 1 and 2 and subscripts ab and bb for after and before the backwashes respectively. While
the first equation gives the extent of the reversibility compared to the initial membrane flux, the second
one relates it to the total flux decrease.
Absolute reversible flux = Jab (Equation 1)
Relative reversible flux = (Jab-Jbb)/(1-Jbb) (Equation 2)

Table 1 Flux decrease reversibilities

Reversibility Normalized Absolute Normalized Absolute Normalized Absolute
Cycle number 1 2 3
Kaolinite (100 mg/L) 1.00 1.00 1.00 1.00 0.96 1.00
HA 0.58 0.88 0.46 0.84 0.51 0.83
HA-Kaolinite 0.77 0.86 0.74 0.83 0.68 0.81
HA after kaolinite 0.82 0.92 0.73 0.88 0.74 0.88
Alginate 0.94 0.98 0.94 0.99 0.75 0.93
Alginate-kaolinite 0.97 0.98 0.85 0.91 0.74 0.84
Alginate after kaolinite 0.84 0.93 0.76 0.87 0.72 0.86
HA-alginate 0.71 0.8 0.63 0.75 0.59 0.72
HA-alginate-kaolinite 0.90 0.91 0.83 0.85 0.83 0.85

Table 1 shows the calculated reversibilities: the value is low for HA alone but high for kaolinite filtration.
The relative, but not the absolute reversibility of HA filtration is increased in the mixture. HA thus causes
the same degree of irreversible fouling in the presence of kaolinite. In contrast, filtration of HA after
kaolinite in the subsequent filtration experiment leads to slightly less irreversible HA fouling. This can be
explained by the kaolinite previously deposited on the membrane partially blocking the HA-membrane
interaction sites on the membrane surface.

Filtration with alginate alone leads to a slightly worse reversible flux than with kaolinite, but both are
relatively easily reversible if filtrated alone. This agrees with the suggestion that both substances mainly
foul the membrane by cake formation and only little membrane-alginate adsorption takes place (Jermann
et al, 2007). This makes it difficult to identify the effect of kaolinite on decreasing the reversibility of the
alginate flux. However, Table 1 shows that the presence of kaolinite leads to slightly reduced reversibility
of the alginate fouling. This can be explained by the hindered back-diffusion of alginate due to
entrapment by kaolinite, which intensifies the alginate-pore blocking and the membrane attachments by
adsorption.

Transmission of DOC
Figure 3 presents the transmission of DOC during individual, premixed and subsequent filtration.
Considering DOC adsorption onto kaolinite, the transmission can be evaluated either before adsorption –
 D. Jermann, W. Pronk, R. Kägi, N. Schaarschmidt, M. Halbeisen, M. Boller 5
referred to as total DOC - or after adsorption – referred to as free DOC. The humic acid adsorption onto
kaolinite is known from the literature and has been assessed in previous experiments to be about 20% at a
low humic acid concentration of 2 mg C/L. We therefore used this value for our calculations. No
calculations were performed for alginate as no alginate-kaolinite adsorption values have yet been
published.

Figure 3 Transmission of: a) HA (± 5%) and b) alginate filtrated alone (±10%), premixed with and
subsequently after kaolinite (± 30% as values after the first 100 mL are within the detection limit). The
total and free HA for the premixed UF are shown in a). Vertical lines indicate backwashes.

Figure 3a shows that the HA transmission is lower when it is filtrated in combination with kaolinite than
alone. This indicates the incorporation of HA in the kaolinite fouling layer and can be explained by the
assumed adsorption. More detailed analysis reveals that total and free HA transmission is lower than
when HA is filtrated alone. This indicates a further effect of the kaolinite fouling layer on HA retention,
namely either size exclusion by the cake or trapping during cake formation. During subsequent filtration,
the HA transmission is initially significantly lower than during the premixed filtration but it is higher and
similar to when HA is filtrated alone. This strengthens the likelihood of the suggested HA adsorption onto
kaolinite, as it is capacity-limited and can be fast. The insignificant effect of kaolinite after the initial
phase implies no further physical retention by the kaolinite.

Moreover, HA transmission increases in all experiments over the whole filtrated volume. This indicates
HA adsorption onto the membrane and is in line with our flux reversibility results. During premixed
filtration, the transmission increase of free HA is similar to that of HA when filtrated alone. This implies
that the same amount of HA is adsorbed onto the membrane and hence that this amount is independent of
the amount adsorbed onto kaolinite.

Contrary to HA, alginate transmission decreases initially before levelling off, as can be seen in Figure 3b.
This indicates that membrane-pore blockage is the dominant fouling process, which seems partly
irreversible. Moreover, a lower alginate transmission in the presence of kaolinite is observed. This can be
caused either by adsorption of alginate onto kaolinite or alginate entrapment by the kaolinite cake.
Transmission values for premixed and subsequent filtration decline after the initial phase into
concentration ranges too low to draw significant conclusions about their difference.

SEM images
Figure 4 shows the SEM images of the membrane fouled by kaolinite with alginate (premixed) before and
after backwashes. It reveals patches of the fouling layer remaining on the membrane after the backwash,
which indicates that the backwashing does not take place homogeneously, but that some patches of
fouling material are released from the membrane while others remain attached. This corresponds to the
results of the flux decrease reversibility showing that some fouling is irreversible by backwashes if
kaolinite is filtrated with alginate rather than alone. Moreover, some fibril shapes are visible (Fig. 4-d):
these can be assumed to be alginate and strengthen our suggestion that the fouling layers are a matrix of
organic and inorganic constituents. The alginate identified in the images seems relatively small, but it has
to be considered that dehydration was necessary for the analyses. This is known to shrink organic matter
 D. Jermann, W. Pronk, R. Kägi, N. Schaarschmidt, M. Halbeisen, M. Boller 6
by a factor of 100-1000, as it is hydrated (especially polymers) while inorganic material stays unaltered. It
is therefore likely that alginate fills the interstices of the kaolinite cake to a more substantial extent than
can be seen.

Figure 4 SEM images of membranes fouled by premixed kaolinite and alginate: a) and b) before, c) and
d) after backwashing.

Kaolinite-NOM interactions and membrane fouling

The flux decrease results show that interactions between kaolinite and NOM lead to a synergy of the
effects of the individual substances on membrane fouling. On the basis of the combined experimental
results, the fouling mechanisms were synthesized as explained below.

The fouling mechanism with kaolinite alone can be assumed to be due to cake formation, because the
kaolinite particles are about 100 times larger (350 nm) than the membrane pores (100 kDa ≈ 10 nm). As
shown in Fig. 1 (kaolinite alone), the resistance of the kaolinite cake layer is very low. This can be
explained by the surface heterogeneity of kaolinite, which is characterized by negatively charged basal
faces but conditionally charged edges. Edge-to-face aggregates can therefore be formed at neutral pH
values, leading to a polyhedral face-edge structure (Tombacz et al., 2004, Tombacz et al., 2006). Such a
structure results in a relatively large porosity of the cake layer with a relatively low hydraulic resistance.

As shown in Fig. 1, kaolinite premixed with HA resulted in a stronger flux decrease than the sum of the
individual compounds. HA adsorption onto the particles is known from the literature and has been
reported to stabilize the particles, decrease their size and smooth out their surface heterogeneity (Tombacz
et al., 2004, Balcke et al., 2002). It can be assumed that the changed particle characteristics prevented
face-to-edge aggregation of the kaolinite and that the particles were more likely to show base-to-base
deposition. The relatively porous polyhedral structure cannot be formed in this case, leading to a more
packed structure with a higher resistance. As indicated in the Carman-Kozeny equation (Mulder, 1996),
the hydraulic resistance of a fouling cake shows a direct correlation with the cake porosity.
This mechanism can be complemented by HA entrapment in the fouling layer during cake formation.
This results in filling of fouling-layer interstices, which obviously also affects the reduction of the cake
porosity. Finally, HA is adsorbed to the membrane similarly to filtration of HA alone.
As shown in Fig. 1 and Table 1, the presence of kaolinite does not reduce the irreversibility of HA
fouling. HA adsorbs onto kaolinite by ionic binding between negative carboxylic groups of HA and
positively charged edges of kaolinite (Tombacz et al., 2004). In contrast, it is supposed that HA adsorbs
onto the membrane by hydrophobic interactions between its phenolic groups and the membrane (Jucker
and Clark, 1994). It can thus be concluded that the interaction of HA with the membrane and kaolinite is
based on different functional groups, which explains why kaolinite does not affect the irreversibility.
In summary, it can be concluded that the incorporation of HA into the kaolinite cake by previous
adsorption and trapping of HA in the cake leads to a fouling layer of significantly reduced porosity
resulting in a detrimental flux decrease compared to kaolinite filtration without HA or subsequently
filtrated kaolinite and HA. However, the incorporation of HA in the kaolinite fouling layer did not reduce
the irreversible HA membrane fouling due to different controlling mechanisms.
 D. Jermann, W. Pronk, R. Kägi, N. Schaarschmidt, M. Halbeisen, M. Boller 7

These two components can consequently mutually influence their fouling mechanisms by sterical or
chemical interactions. This can be described as simultaneous cake-layer formation by the substances
which dominated the fouling. This mechanism is probably complemented by entrapment of the organics
in the fouling layer during cake formation, leading to partial filling of the fouling layer interstices.
As shown in Figs. 1 and 2, the fouling of premixed and subsequently filtrated kaolinite and alginate is
similar. This indicates that their synergistic effect was determined by mutual influences of the substances
during the fouling process, as their individual fouling mechanisms develop in the same place. This is in
line with the theory that alginate mainly fouls the membrane on the surface by pore blocking and cake
formation (Jermann et al., 2007), thus occurring in the same area as the kaolinite cake formation.
In the case of premixing, adsorption of alginate on kaolinite can be assumed, leading to a transformation
of the particles surface characteristics as discussed below. The alginate-kaolinite colloids and the free
alginate simultaneously blocked the pores and formed a heterogeneous fouling cake. In the case of
subsequent filtration, it can be assumed that the alginate was retained within the polyhedral porous
structure by size exclusion and some adsorption. However, we suppose that in both cases the interstices
between the particles will be filled with alginate, leading to a cake layer with the same resistance.

Interactions between alginate and kaolinite are reported in the literature. The assumed interaction forces
are electric charges as well as van der Waals and hydrogen bonds and are considered to lead to particle
bridging and destabilization in the solution (Buffle et al., 1998). In contrast, alginate and inorganic
particles as well as acidic polysaccharides and clay were observed to form stable colloids (Chen et al.,
2006, Labille et al., 2005). It is supposed that the presence of stable or instable colloids depends on the
substance ratio and the chemical conditions in the solution. However, the alginate-kaolinite interactions
are consistent. We also observed alginate adsorption onto kaolinite in previous experiments (in
preparation). But it is still unclear whether stable or unstable colloids are formed at the low alginate
concentrations we used. The fact that the substances synergistic effect was independent of premixing
might seem counterintuitive to alginate adsorption onto kaolinite. However, this merely implied that the
interactions in the solution are unnecessary for the synergistic effect, but not that they are absent.

Irreversible fouling of the alginate-kaolinite was induced by a patchwise stronger attachment of the
heterogeneous alginate-kaolinite fouling layer to the membrane, probably via alginate-membrane binding
processes. We recently showed an association between alginate and the membrane (Jermann et al., 2007).

We conclude from our experimental results that NOM incorporated in the particulate fouling layer has a
significant impact on the flux. This can be explained by reduced cake porosity due to altered particle
characteristics leading to the formation of a more compact cake and filling of cake interstices by NOM
entrapment. This can have a significant effect on permeability, as the hydraulic resistance of a fouling
cake is an exponential (-⅓) function of the cake porosity according to the Carman-Kozeny equation.

Combination of the three model substances led to the most detrimental flux decrease of all the
experiments performed. It can be assumed on the basis of our model that the synergistic effect of the
substances is due to the interplay of their individual fouling mechanisms. This is in line with literature
reporting synergistic effects of organic and colloidal fouling (Li and Elimelech, 2006, Park et al., 2006)
and no effect when in the (assumed) absence of an NOM-particle interaction (Kweon and Lawler, 2004)
as well as a synergistic effect of different NOM substances (Jermann et al., 2007).

The synergistic effects of HA, alginate and kaolinite described here show the importance of mutual
influences between inorganic particles and NOM. Such interactions can give a mechanistic explanation as
to why natural inorganic particles were reported as relevant foulants (Khatib et al., 1997, Pianta et al.,
1998, Tarabara et al., 2002, Nghiem et al., 2006), while synthetic ones were found to be of minor
importance (Schafer et al., 2000, Pianta et al., 1998, Li et al., 2002, Kweon and Lawler, 2004). They are
thus important when attempting to understand the fouling phenomena of natural waters.
 D. Jermann, W. Pronk, R. Kägi, N. Schaarschmidt, M. Halbeisen, M. Boller 8
Conclusions
Our study showed a synergistic effect of membrane fouling by different NOM and inorganic particles
during UF. The synergy is determined by the individual fouling mechanism of a NOM - hence suggesting
how it can be changed – and by specific NOM-particle interactions.

As regards HA, which causes membrane adsorption in individual cases, the fouling was determined by
HA adsorption onto kaolinite leading to stabilized colloids. For the polysaccharide alginate, that alone
leads mainly to cake formation, the dominant fouling mechanism in combination with kaolinite was a
simultaneous cake-layer build-up of the substances which did not necessitate previous substance mixing.
Both fouling synergies resulted in a heterogeneous organic-inorganic fouling layer of lower cake porosity
than that formed by an individual kaolinite.

The absolute irreversible fouling of NOM, especially HA, could not be prevented by the particles because
different functional groups were responsible for the NOM interactions with kaolinite and the membrane.

References
Aoustin, E., Schäfer, A. I., Fane, A. G. and Waite, T. D. (2001) Separation and Purification Technology,
22-23, 63-78.
Balcke, G. U., Kulikova, N. A., Hesse, S., Kopinke, F. D., Perminova, I. V. and Frimmel, F. H. (2002)
Soil Science Society of America Journal, 66, 1805-1812.
Buffle, J., Wilkinson, K. J., Stoll, S., Filella, M. and Zhang, J. W. (1998) Environmental Science &
Technology, 32, 2887-2899.
Chen, K. L., Mylon, S. E. and Elimelech, M. (2006) Environmental Science & Technology, 40, 1516-
1523.
Clark, M. M. and Lucas, P. (1998) Journal of Membrane Science, 143, 13-25.
Jarusutthirak, C., Amy, G. and Croue, J.-P. (2002) Desalination, 145, 247-255.
Jermann, D., Pronk, W., Meylan, S., Boller, M. (in press) Water Research.
Jones, K. L. and O'Melia, C. R. (2000) Journal of Membrane Science, 165, 31-46.
Jucker, C. and Clark, M. M. (1994) Journal of Membrane Science, 97, 37-52.
Khatib, K., Rose, J., Barres, O., Stone, W., Bottero, J.-Y. and Anselme, C. (1997) Journal of Membrane
Science, 130, 53-62.
Kimura, K., Hane, Y., Watanabe, Y., Amy, G. and Ohkuma, N. (2004) Water Research, 38.
Kweon, J. H. and Lawler, D. F. (2004) Water Research, 38, 4164-4172.
Labille, J., Thomas, F., Milas, M. and Vanhaverbeke, C. (2005) Journal of Colloid and Interface Science,
284, 149-156.
Li, J., Sanderson, R. D., Hallbauer, D. K. and Hallbauer-Zadorozhnaya, V. Y. (2002) Desalination, 146,
177-185.
Li, Q. and Elimelech, M. (2006) Journal of Membrane Science, 278, 72-82.
Mulder, M. (1996) Basic Principles of Membrane Technology (second edition), Kluwer Academic
Publisher., 449.
Nghiem, L. D., Oschmann, N. and Schafer, A. I. (2006) Desalination, 187, 283-290.
Park, C., Kim, H., Hong, S. and Choi, S.-I. (2006) Journal of Membrane Science, 284, 248-254.
Pianta, R., Boller, M., Janex, M.-L., Chappaz, A., Birou, B., Ponce, R. and Walther, J.-L. (1998)
Desalination, 117, 61-71.
Schafer, A. I., Schwicker, U., Fischer, M. M., Fane, A. G. and Waite, T. D. (2000) Journal of Membrane
Science, 171, 151-172.
Tarabara, V. V., Pierrisnard, F., Parron, C., Bottero, J.-Y. and Wiesner, M. R. (2002) Journal of Colloid
and Interface Science, 256, 367-377.
Tombacz, E., Libor, Z., Illes, E., Majzik, A. and Klumpp, E. (2004) Organic Geochemistry, 35, 257-267.
Tombacz, E. and Szekeres, M. (2006) Applied Clay Science, Layer Charge of Clay Minerals - Selected
papers from the Symposium on Current Knowledge on the Layer Charge of Clay Minerals, Current
knowledge on the layer charge of clay minerals, 34, 105-124.
Ye, Y., Le Clech, P., Chen, V., Fane, A. G. and Jefferson, B. (2005) Desalination, 175, 7-20.
D. Jermann, W. Pronk, R. Kägi, N. Schaarschmidt, M. Halbeisen, M. Boller 9