Journal of Membrane Science 428 (2013) 225–231

Contents lists available at SciVerse ScienceDirect

Journal of Membrane Science

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

Preparation of thin film composite membranes with polyamide film

on hydrophilic supports
Inger Lise Alsvik n, May-Britt Hägg
Department of Chemical Engineering, Norwegian University of Science and Technology, N-7491, Norway

a r t i c l e i n f o abstract

Article history: The interest in osmotically driven processes such as pressure retarded osmosis (PRO) has increased
Received 30 March 2012 during the past decade. The synthesis of new membranes specifically designed for the process is
Received in revised form essential for the development of these technologies. Conventional TFC have a relatively hydrophobic
6 October 2012
support layer e.g., polysulfone. However, a more hydrophilic support layer is desirable in osmotically
Accepted 9 October 2012
Available online 24 October 2012
driven processes as such supports would give better water flux, less internal concentration polarization
(ICP) and less fouling. In the current project it was demonstrated that it is possible to coat a hydrophilic
Keywords: support membrane given there are enough functional/reactive groups on the surface of the support
Pressure retarded osmosis membrane. This is achieved by reacting the support with polyfunctional acid chlorides in a reaction
Thin film composite
step prior to interfacial polymerization (IP). In this way, covalent bonds between the support and the
Membrane preparation
active layer are formed and the composite will be stabilized. Hydrolysed cellulose acetate CA
Hydrophilic support
Cellulose acetate membranes have been coated with a polyamide (PA) layer by this modified IP method. These
membranes displayed a salt rejection up to 97%, and water fluxes from 7.6  10  7 to 4.7  10  6 m3/m2/s
at 1.3  106 Pa differential pressures.
& 2012 Elsevier B.V. All rights reserved.

1. Introduction to potential mechanical energy by the use of a semi-permeable

membrane. In a fresh water/salt water system, the water flows
1.1. General from the fresh water side to the salt water side. A pressure builds
up at the salt water side, and this can be used to run a turbine to
The global energy demand is increasing as the world’s popula- generate electric power. The optimal operating pressure is 11–
tion is growing and developing countries are becoming more 15 bars, enabling the generation of 1 MW/m3 freshwater. The
industrialized. A significant portion of this energy is supplied total global potential for osmotic power has been calculated to be
through the use of fossil fuels such as coal, gas and oil. Anthro- 1655 TW h/year [5]. The concept of osmotic power was first
pogenic climate change is a serious threat to modern society and introduced by Sidney Loeb in the 1970s [6,7]. Due to low prizes
there is an urgent need to search for alternative emission free on electricity and poor membranes at the time, there were little
energy sources [1–3]. The current global energy supply is still or no commercial interests in developing PRO as an alternative
dominated by fossil fuels. Fossil fuels emit greenhouse gases and energy source. However, in the late 1990s the Norwegian com-
there is a common consensus among scientists that the emission pany Statkraft invested in a PRO project, resulting in the opening
of greenhouse gases is causing global warming. As an example, in of the world’s first PRO pilot plant November 2009 [8]. One of the
2008, 79% of the European gross inland energy consumption came major obstacles to the completion of a full scale osmotic power
from fossil fuels. However, the leaders of the European Union (EU) plant (OPP) has been the lack of a membrane designed specifically
have agreed on legally binding national targets for increasing the for the process. The membrane power performance, often referred
share of renewable energy. The goal is to achieve a 20% share for to as the power density (W/m2), is dependent on the water flux
renewable energy for the entire EU by 2020, compared with an across the membrane. Therefore, a high water flux is important
8.5% share in 2005 [4]. Pressure retarded osmosis (PRO) is a for an efficient PRO process. Most of the commercial available
potential sustainable, non-polluting and renewable energy osmotic membranes today are reverse osmosis (RO) membranes,
source. In a PRO process the free energy of mixing is converted designed for pressure driven processes. Conventional seawater RO
membranes are not suitable for use in osmotically driven mem-
brane processes; the water flux obtained is too low. This is due to
n
Corresponding author. Tel.: þ47 92452120. their hydrophobic and thick support layer and the hydrophobic
E-mail address: [email protected] (I.L. Alsvik). interface between the active layer and the support. Membranes

0376-7388/$ - see front matter & 2012 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.memsci.2012.10.011
226 I.L. Alsvik, M.-B. Hägg / Journal of Membrane Science 428 (2013) 225–231

used in RO are either thin film composite (TFC) membranes or The water flux from the feed solution dilutes the draw solution in
asymmetric cellulose based membranes like CA or cellulose the porous support, resulting in a decrease in the transmembrane
triacetate (CTA). TFC membranes are generally considered to be osmotic pressure (Fig. 1) [10,13].
the most promising membranes for osmotically driven processes. The structure parameter (S) describes the inherent physical
However, the synthesis of new membranes specifically designed properties of an osmotic membrane. S is a function the thickness
for the process is essential for the commercialization of PRO. In (x), the tortuosity (t) and the porosity (j) of the membrane:
order to establish PRO as a profitable renewable energy source, xt
the power density of the membranes used must be in the range of S¼ ð1Þ
j
4–6 W/m2 [5].
A fully wetted hydrophilic support might increase effective
porosity and therefore reduce the S value [14]. The desired value
1.2. Challenges related to PRO versus RO membranes for the structure parameter in a PRO membrane is lower than
1500 mm. Standard RO membranes have a structure parameter
Conventional RO TFC membranes are designed to withstand  10,000 mm [5,15].
high pressure, and therefore have a thick support layer. This Because the water passes through the membrane from the
support layer is also relatively hydrophobic; a property that is not support side in PRO, fouling is also a more important issue in this
desirable in PRO. In RO the water first permeates the active layer process than in RO. In a real life PRO process, the feed from
by a solution-diffusion mechanism and then simply percolates natural surface water will be of different quality than the feed
through the pores of the support layer by hydraulic pressure. This used in laboratory research. Consequently, natural organic matter
implies that the porous support does not need to be fully wetted (NOM), calcium carbonate (CaCO3) and microbial growth on the
to ensure adequate water flux. However, in an osmotically driven membrane will cause considerable fouling on the membranes
membrane process the support layer must be fully wetted to used. The membrane materials and its properties are important
ensure adequate water flux. The surface that the water meets factors for the membrane fouling characteristics. Polysulfone (PSf)
when entering the active layer should preferably be hydrophilic. membranes is a hydrophobic material and have a tendency to
Vapour or air trapped in the pores might block the passage of adsorptive fouling, while cellulose acetate (CA) membranes dis-
water through the support layer. Reduced continuity of the water play little adsorptive fouling and are much easier to clean [11].
within the layer may exacerbate internal concentration polariza- Several experiments have been performed at Sunndalsøra in
tion (ICP) and reduce the effective porosity [9,10]. A more Norway in order to investigate the fouling potential in PRO using
hydrophilic support would most likely reduce or prevent this typical Norwegian ocean water and surface water [16]. The
effect. In addition, hydrophilic CA membranes give less adsorptive membranes used were an asymmetric CA membrane manufac-
fouling than other more hydrophobic membranes [11]. tured by GKSS, Germany. Chemical cleaning of NOM and CaCO3
Most RO membranes display a flux decline during the initial were able to restore the initial flux for the membranes at 93% or
period of a membrane separation process. The primary reason for higher. However, the effect of membrane cleaning was observed
this flux decline is external concentration polarization (ECP) on to decrease when repeating the cleaning. Periodical disinfection
the salt solution side from where water is forced through the with 5 ppm free chlorine was observed to be sufficient in order to
membrane. ECP is, however, considered to be reversible and can prevent microbial growth on the membranes during long term
be controlled by promoting turbulence of the flow across the experiments with natural water sources [16]. Based on the
membrane, adjusting the shear rate, by the use of ultrasound or findings referred here, and realizing the challenges related to
by the means of an electric field [12]. ICP is closely related to ECP, make a high performance PRO membrane, the focus of the current
but it occurs exclusively in osmotically driven membrane pro- work is to use a sufficiently, thin and hydrophilic support
cesses. The concentration polarization occurs within the porous membrane and coat it with a very thin film of an active polyamide
support layer and cannot be reduced by altering the shear rate (PA) layer. Interfacial polymerization of PA on hydrophilic support is
and/or turbulence of flow across the membrane. Concentrative not trivial, as the active layer easily delaminates from the support,
ICP occurs when the active layer of the membrane is facing the but was successfully achieved by Nilsen and Alsvik [17,18]. The most
draw solution (Fig. 1). Salt diffusion through the active layer likely explanation for the delamination is that the layers in a
results in a higher CICP than CF, resulting in a decrease in the conventional PA-TFC membrane are held together by adhesion
transmembrane osmotic pressure. Dilutive ICP occurs when forces. When the support becomes too hydrophilic, the adhesion
the active layer of the membrane is facing the feed solution. forces are too weak to keep the layers together. Unfavorable

Fig. 1. Dilutive and concentrative internal concentration polarization in an osmotically driven process [10].
 I.L. Alsvik, M.-B. Hägg / Journal of Membrane Science 428 (2013) 225–231 227

interactions between the two layers of different hydrophilicity formation of a covalently bonded PA layer on CA: The CA
will allow water to intrude between the layers, resulting in membrane is firstly hydrolysed to increase the number of hydroxyl
delamination. Different degrees of swelling of the two polymers groups on the surface (a), the hydroxyl groups react with TMC (b),
could also explain the delamination. In the current work, the PA MDA reacts with the TMC. The latter reaction takes place in the
layer has been successfully attached to the hydrolysed CA/CTA presence of two miscible liquids (water and EGDE) and does not
supports by covalent bonds (Fig. 2). The theory behind this is form a continuous PA layer. However, polyamide will be formed
explained in the next Section 1.3 and also discussed in [17,18]. and the formation of a ‘‘pre-layer’’ might occur. The last step in the
process is an IP between a MDA aqueous solution and TMC in
hexane (d). This modified IP method results in covalent bonds
1.3. Preparation of CTA-TFC membranes between the active layer and the support, which hinder delamina-
tion and is expected to result in a more stable membrane. Due to
Hydrolysis of CA membranes results in an increased number of
functional hydroxyl groups on the membrane surface (Fig. 3). As
acid chloride is a very reactive functional group, they will form an
ester in the reaction with the hydroxyl on the membrane surface.
Trimesoyl chloride (TMC) is a trifunctional acid chloride com-
monly used in IP on membranes. The number of functional groups
and the surface charge will increase when the hydrolysed mem-
brane surface is treated with TMC. The acid chloride can bind to
the surface by one, two or three functional groups (Fig. 4). The
latter would result in a very high bond tension and is considered
highly unlikely. The surface charge of a membrane is dependent
on the number of acidic/basic functional groups on the surface
and the degree of ionization of these groups. Negatively charged
functional groups are usually the conjugated bases of organic
acids, e.g., carboxylate (–COO  ) [19]. By using a titration method
developed by Schaep et al. [20] it is possible to detect the increase
in membrane surface charge after the reaction with TMC to detect
how many free acid groups (from acid chloride) there are left on
the surface. When the highly reactive acid chloride functional
groups on the membrane surface reacts with m-phenylenediamine
(MDA) followed by a reaction with TMC, a polymer covalently Fig. 4. Illustration of the three different theoretical conformations of the acid
bonded to the hydrolyzed CA will be formed. Fig. 5 displays the chloride on the membrane surface.

Fig. 2. A comparison of a conventional TFC membrane (a) where the active layer stays on the support by adhesion forces, and the modified IP method (b) with a covalent
bond between the active layer and the support layer.

Fig. 3. (a) Hydrolysis of CA/CTA (b) Modification of hydrolyzed CA by reacting it with TMC.
228 I.L. Alsvik, M.-B. Hägg / Journal of Membrane Science 428 (2013) 225–231

Fig. 5. The illustration of formation of the active PA layer on a hydrophilic support membrane. (a) Hydrolysis/deacetylation of the support membrane, (b) reaction between
TMC and the hydroxyl groups on the membrane surface, (c) addition of MPD and (d) reaction with TMC.

the reduced mobility of the MDA during IP, the ratio TMC/MDA in 2.2. Characterization
this method could be different from the ratio in conventional IP.
This might affect the degree of crosslinking and the long term Fourier transform infra red spectroscopy (FTIR) analysis was
physical stability of the membrane. However, a long term study performed using a Perkin Elmer FTIR equipped with an ATR-cell.
was not included in this project. The scope of this project was Commercial HTI membranes were analyzed before and after
screening of numerous reaction conditions (concentrations, sol- being hydrolyzed (0.7% NaOH, 30 min) and treated with TMC
vents, time and curing temperature) to investigate the formation of (3% in EGDE, 20 h). All micrographs were acquired using a Hitachi
a stable PA layer. Although the goal was to develop membranes ultra-high-resolution analytical field emission scanning electron
suitable for PRO the membranes in this study were tested in RO. It microscope (FE-SEM) SU-70. An Emitech SC7620 sputtering
was necessary to be able to test membranes with low or no salt machine was used to coat the samples with gold-platinum (45 s).
rejection (e.g., support membranes). Furthermore, because of the
very high number of membranes screened a time efficient test 2.3. Titration
method was desirable. A dead end RO test cell is the most time
efficient way of testing the integrity of the PA layer. AL support membranes were hydrolysed in an aqueous KOH
solution (10%, 10 min), rinsed thoroughly and soaked in EGDE
(12 h) Thereafter one batch of membranes were treated with TMC
in EGDE (0.8%, 3.0  10  5 mol/g solvent, 1 h) and another with
2. Experimental IPC in EGDE (0.615%, 3.0  10  5 mol/g solvent, 1 h). The mem-
branes were subsequently rinsed thoroughly in water. To deter-
2.1. Materials and setup mine the amount of negatively charged groups on the membrane,
a titration was performed using cesium chloride (CsCl) and
Isophthaloyl chloride (IPC), MDA, and TMC from Aldrich and magnesium chloride (MgCl2). An aqueous solution of CsCl
camphorsulfonic acid (CSA) and triethylamine (TEA) from Merck (50 mL, 0.1 M) were poured over the membrane and stirred at
were used. The bottles of MDA and TMC were always flushed with room temperature for 15 min. The original counter ions were then
argon after use to reduce hydrolysis or oxidation. Diethylene exchanged for cesium and all negative charged functional groups
glycol diethyl ether (EGDEZ99.5%) from Sigma-Aldrich and had cesium as counter ion. The membrane was thoroughly
Hexane ( Z98%) from Merck were used. UF grade CA (CA600PP01) washed in distilled water and then soaked in an aqueous MgCl2
from Alfa Laval (AL) was used as support in membrane AL-1 and solution (50 mL, 0.01 M, 15 min). The cesium ions (Cs þ ) on the
AL-2. A RO grade CTA membrane with an embedded polymer membrane were replaced by the magnesium ions (Mg2 þ ).The
mesh from Hydration Technology Innovations (HTI) were used as concentration of Cs þ in the solution was analysed with ICP-MS.
support in the membranes HTI-1, HTI-2 and HTI-3. As the scope of The number of negatively charged groups at the membrane
this project was screening of various reaction parameters (con- surface was equal to the number of cesium ions measured [20].
centration, solvents, drying, time, curing temperature) in order to
find combinations resulting in a stable continuous PA layer. It was 2.4. Membrane preparation
necessary to be able to test membranes with low or no salt
rejection (e.g., support membranes). Dead end RO tests are also All support membranes were hydrolysed, the AL supports in an
the most time efficient way of testing the integrity of the PA aqueous KOH solution (10%, 10 min) and the HTI supports in an
layers. Therefore, all membranes were tested in a RO dead end aqueous NaOH solution (0.7%, 30 min). Thereafter the membranes
test cell at a differential pressure of 1.3  106 Pa. The rejections, R, were rinsed thoroughly before being soaked in EGDE (with
were determined at a 0.3% NaCl concentration in the feed. The activated molecular sieves 4 Å, 12 h). The membranes were then
salinity of the solutions was tested by the use of a conductivity put in a TMC solution in EGDE, dried under argon and soaked in an
meter. The salts used in the titration were from Merck. aqueous MDA solution containing the buffer CSA/TEA (1%/0.5%).
 I.L. Alsvik, M.-B. Hägg / Journal of Membrane Science 428 (2013) 225–231 229

Table 1
The reaction parameters for IP on hydrolysed support membranes from AL (CA) and HTI (CTA). The reactions were performed at room temperature except when other
temperatures are reported.

Batch Concentration of Time in TMC Drying time Concentration of Time in MDA Concentration of Time in TMC and Curing time in
TMC in EDGE [%] and EGDE under argon MDA in water [%] and water TMC in c-Hexane [%] c-Hexane [min] water bath
[min] [min] [min] [min]

AL-1 0.4 1 60 0.3 0.5 0.2 0.5 –

AL-2 0.8 1 120 0.4 0.5 0.1 0.5 –
AL-3 0.8 1 120 0.4 0.5 – – –
HTI-1 2 (80 1C) 720 120 0.3 (60 1C) 720 1.7 1 15 (70 1C)
HTI-2 10 1200 120 0.5 (60 1C) 120 1.5 0.5 30 (70 1C)
HTI-3 10 1200 120 0.5 (60 1C) 120 – – –

The membranes were then dried by using a laboratory dust free 0.0020
tissue and a rubber roller. Finally the membranes were soaked in
a solution of TMC in hexane and washed thoroughly in distilled

Surface charge density (mol/m2)

water. The TFC membranes with HTI support were in addition
0.0015
cured in water baths. The screening process included numerous
(around 100) concentrations and reaction times. Selected batches
from the present work which displayed salt rejections indicating a
continuous PA layer on the membranes are presented in Table 1. 0.0010

3. Results and discussion 0.0005

3.1. Surface modification and determination of surface charge

To investigate how effective the reaction between the hydro- 0.0000

lyzed AL membrane and TMC is to give free acid chlorides groups No modification TMC IPC
on the surface of the supports, a titration was performed using
Fig. 6. The surface charge density of the hydrolysed AL membranes functionalized
membranes treated with two different acid chlorides: TMC and with TMC and IPC.
IPC. TMC has three functional groups (acid chloride), IPC has two.
These two functional groups on the IPC are in meta-position,
which is similar to the position of the acid chloride groups on
TMC (Fig. 6). It was assumed that the acid chloride groups on TMC 100
and IPC would have similar reactivity with the hydroxyl group on
the membrane surface and similar steric hindrance. If one func-
tional group reacted with the hydroxyl groups on the surface, one
Transmittance [%]

would expect the TMC to have twice the surface charge density
90
(SCD) as the IPC. The SCD of IPC was 61% of the SCD of TMC
(Fig. 6), indicating that one binding site is the most favorable 1018
1714
interaction. It is more likely for the TMC to bind to the surface two
places than the IPC, and therefore result in a SCD of IPC higher
1741
than 50% of the TMC. These data show the presence of acid 80
chlorides on the surface of the supports after reaction with TMC,
indicating the presence of covalent bonds between the support Untreated
and the polyamide film. Treated with TMC 1033

4000 3500 3000 2500 2000 1500 1000

3.2. Characterization Wavenumber [cm-1]

3.2.1. FTIR Fig. 7. FTIR-spectra of an untreated CTA HTI membrane and a hydrolysed CTA HTI
membrane reacted with TMC.
FTIR-spectra of the untreated HTI membrane and a hydrolyzed
HTI membrane treated with TMC are presented in Fig. 7. The
carbonyl peak originally present at 1741 cm  1 was shifted 3.2.2. SEM
towards lower wavenumbers after the replacement of the acetate SEM micrographs of the coated AL and HTI supports prepared
groups with TMC. After the replacement aromatic carboxylic using the same method as HTI-2 are presented in Fig. 8. Both
groups are present as well, and therefore changing the carbonyl pictures clearly show a ridge and valley structure, indicating the
signal. Around 1560 cm  1 a new pattern appears after treatment presence of a PA layer. However, the two different coatings seem to
with TMC. This is attributed to aromatic C¼C bending. The area be slightly different. The PA layer on the HTI support seems to have
1100–1000 which corresponds to the C–O stretch of alcohols, larger features ( 500 nm) than the PA layer on the AL support
carboxylic acids and esters also undergo extensive changes due to (300). Several SEM micrographs were obtained from randomly
the reaction. The differences between the two FTIR-spectra are chosen parts of the membranes. The difference in membrane
consistent with the expected changes on the membrane surface morphology was consistent on both membranes, and the pictures
when treated with TMC. presented are representative for both the membranes.
230 I.L. Alsvik, M.-B. Hägg / Journal of Membrane Science 428 (2013) 225–231

3.3. Membrane performance

The water fluxes and R-values for the membranes tested in a

dead end test cell (RO-mode), are presented in Table 2. Salt
retention (R-value) of the membranes of up to 97% indicates the
presence of an osmotic active PA layer. All membranes were made
in batches of 3–4 coupons and the average of at least two is
presented here. The results were sometimes difficult to repro-
duce. However, when membranes in a batch were successfully
coated, the measurements were consistent. The membranes AL-3
and HTI-3 were prepared without the second reaction with TMC
and displayed only 9% and 8% salt retention, respectively. These
membranes also displayed a higher flux than the other mem-
branes and were more unstable. These findings indicate that an
osmotic active layer was not formed during the first reaction
between TMC and MDA. However, a PA ‘‘pre-film’’ with a
relatively open structure could have been formed here. The
solvent in the first step, EGDE, was chosen because it is a polar
nonprotic solvent capable of wetting CA. However, this solvent is
miscible in water and a conventional IP would not take place in
this step. The performance of the two membranes prepared
without the second reaction with TMC (Fig. 5d) does not indicate
the presence of a continuous PA layer (Table 2). The AL-1 and AL-2
membranes with an Alfa Laval (AL) UF-membrane as support
have on average a better water flux than the HTI-1 and HTI-2
membranes with a HTI support. The HTI-support was originally
an osmotic membrane, but hydrolysis before the reactions
‘‘opened’’ the structure and gave it UF characteristics (see
Fig. 8. (a) SEM micrograph of the AL membrane coated with PA. (b) SEM
micrograph of the HTI membrane coated with PA. Fig. 9). However, the AL support displayed a higher variation in
quality when coated. It was more difficult to reproduce the results
due to defects in the AL support, resulting in pinholes in several of
the final membranes.
100 Figs. 9 and 10 display the change in the membrane perfor-
Water flux [m3/m2s] *106 Normalized

Water flux mance from the untreated membrane to the final membrane
10 Normalized salt rejection coated with PA. The AL membrane (Fig. 9), was initially a UF-
membrane with a high flux and almost no salt rejection ( 1%).
After hydrolysis the AL membrane had an order of magnitude
salt rejection

1.00 decrease in flux, and a slightly higher salt rejection (  5%). This
behavior could be attributed to a morphology change in the
0.75 polymer network. After coating with PA, the AL-1 membrane
had a salt rejection of 97% and the flux was in the same range as
the hydrolysed AL membrane. AL-2 had a lower salt rejection
0.50
(87%), but displayed a higher flux than the AL-1. The experiments
show that osmotic PA films are formed on the hydrophilic surface
0.25 of the hydrolyzed AL membrane.

0.00 100
Water flux [m3/m2s] *106 Normalized

AL as AL AL-1 AL-2
recieved hydrolysed Water flux
10 Normalized salt rejection
Fig. 9. The membrane perfomance of the AL membrane untreated, hydrolysed and
coated (AL-1 and AL-2). Tested in a dead end test cell at 1.3  106 Pa differential
salt rejection

pressure. (Note: Logarithmic scale after break on Y-axis). 1.00

0.75
Table 2
Water flux and R-values for the membranes AL-1, AL-2, HTI-1, HTI- 0.50
2 and HTI-3. Tested in a dead end test cell (1.3  106 Pa).

Membrane Water flux at 1.3  106 Pa Salt retention 0.25

[m3/m2/s] [%]

0.00
AL-1 2.3  10  6 97
AL-2 4.7  10  6 87
HTI as HTI HTI-1 HTI-2
AL-3 9.2  10  6 9 recieved hydrolysed
HTI-1 2.3  10  6 83
HTI-2 7.6  10  7 80 Fig. 10. The membrane performance of the HTI membrane untreated, hydrolysed
HTI-3 4.5  10  5 8 and coated (HTI-1 and HTI-2). Tested in a dead end test cell at 1.3  106 Pa
differential pressure. (Note: Logarithmic scale after break on Y-axis).
 I.L. Alsvik, M.-B. Hägg / Journal of Membrane Science 428 (2013) 225–231 231

The CTA membrane from HTI was already an osmotic mem- References
brane. The hydrolysis ‘‘opened’’ the HTI membrane, and resulted
in an increased flux (Fig. 10). Furthermore, the salt rejection [1] R.F. Service, Desalination freshens up, Science 313 (5790) (2006) 1088–1090.
decreased drastically from 89% to 5% as a result of the hydrolysis. [2] J.R. McCutcheon, R.L. McGinnis, M. Elimelech, A novel ammonia—carbon
dioxide forward (direct) osmosis desalination process, Desalination 174 (1)
After coating with PA the flux were slightly lower than the
(2005) 1–11.
untreated HTI membrane, and the salt rejection increased again [3] N. Oreskes, The scientific consensus on climate change, Science 306 (2004)
to 83%. The experiments show that a PA film is formed on the 1686.
hydrophilic surface of the hydrolyzed HTI membrane. [4] European Commission, Energy 2020, Directorate-General for Energy, Editor
2011, European Comission: /http://ec.europa.eu/energy/S.
[5] K. Gerstandt, et al., Membrane processes in energy supply for an osmotic
power plant, Desalination 224 (1–3) (2008) 64–70.
4. Conclusions [6] L. Sidney, Production of energy from concentrated brines by pressure-
retarded osmosis: Part-I. Preliminary technical and economic correlations,
IP of a PA layer on hydrophilic support membranes were J. Membr. Sci. 1 (0) (1976) 49–63.
[7] S. Loeb, F. Van Hessen, D. Shahaf, Production of energy from concentrated
successfully performed using various conditions. The support
brines by pressure-retarded osmosis: Part-II. Experimental results and
membranes used were hydrolysed CA/CTA membranes. This work projected energy costs, J. Membr. Sci. 1 (0) (1976) 249–269.
demonstrates that it is possible to coat a hydrophilic support [8] Statkraft. /www.statkraft.noS. 2011 [Cited 2011 9.11.2011].
membrane with a PA layer by a modified IP method, forming a [9] J.R. McCutcheon, M. Elimelech, Influence of membrane support layer hydro-
phobicity on water flux in osmotically driven membrane processes, J. Membr.
covalent bond between the active layer and the support material. Sci. 318 (1-2) (2008) 458–466.
The method developed could be a new tool in order to design [10] J.R. McCutcheon, M. Elimelech, Influence of concentrative and dilutive
new, customized membranes for osmotically driven processes. internal concentration polarization on flux behavior in forward osmosis,
Other hydrophilic membranes having optimized pore sizes to J. Membr. Sci. 284 (1-2) (2006) 237–247.
[11] T. Thorsen, Concentration polarisation by natural organic matter (NOM) in NF
ensure adequate water flux, could also be coated with a PA layer and UF, J. Membr. Sci. 233 (1-2) (2004) 79–91.
given the presence of enough functional/reactive groups on the [12] S.S. Sablani, et al., Concentration polarization in ultrafiltration and reverse
membrane surface for formation of covalent bonds between the osmosis: a critical review, Desalination 141 (3) (2001) 269–289.
[13] G.T. Gray, J.R. McCutcheon, M. Elimelech, Internal concentration polarization
two layers. It is crucial to investigate the synthesis of new
in forward osmosis: role of membrane orientation, Desalination 197 (1–3)
membranes specifically designed for osmotically driven pro- (2006) 1–8.
cesses, and the introduction of covalent bonds between the layers [14] S. Loeb, et al., Effect of porous support fabric on osmosis through a Loeb–
will increase the number of possible material combinations. Sourirajan type asymmetric membrane, J. Membr. Sci. 129 (2) (1997)
243–249.
[15] Statkraft, European Pat. Appl. No. 019614379.
[16] W. Thelin, Fouling in Pressure Retarded Osmosis, Ph.D. Thesis Department of
Acknowledgments Chemical Engineering 2006, Norwegian University of Science and Technol-
ogy: Trondheim. p. 181.
The authors acknowledge the Norwegian Research Council, [17] T.-N, Nilsen. and I.L. Alsvik, Interfacial Polymerization with Monomers
Dissolved in Miscible Liquids Having One of the Solutions Immobilized in a
Statkraft, Statoil and Aqualyng for financial support. The authors
Pourous Structure. To be Published, 2012.
also want to acknowledge HTI for providing membranes. The [18] T.-N, Nilsen. and I.L. Alsvik, Thin Film Composites. PCT Patent Application
authors would also like to thank Prof. Menachem Elimelech at NO2011/000162, May 31, 2011.
Department of Environmental Engineering at Yale University for [19] M. Elimelech, W.H. Chen, J.J. Waypa, Measuring the zeta (electrokinetic)
potential of reverse osmosis membranes by a streaming potential analyzer,
use of the SEM instrument, Gunnar Karlsen for performing FTIR Desalination 95 (3) (1994) 269–286.
experiments and Heléne Vrålstad for help with the interpretation [20] J. Schaep, C. Vandecasteele, Evaluating the charge of nanofiltration mem-
of the FTIR results. branes, J. Membr. Sci. 188 (1) (2001) 129–136.