Journal of Membrane Science 421–422 (2012) 238–246

Contents lists available at SciVerse ScienceDirect

Journal of Membrane Science

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

Novel dual-layer hollow fiber membranes applied for forward

osmosis process
Laurentia Setiawan a,b, Rong Wang a,b,n, Lei Shi b, Kang Li c, Anthony G. Fane a,b
a
School of Civil and Environmental Engineering, Nanyang Technological University, 639798 Singapore, Singapore
b
Singapore Membrane Technology Centre, Nanyang Technological University, 639798 Singapore, Singapore
c
Department of Chemical Engineering, Imperial College London, South Kensington Campus, London, SW7 2AZ, UK

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

Article history: A novel dual-layer forward osmosis (FO) hollow fiber membrane has been designed and successfully
Received 9 March 2012 fabricated by using a triple orifice spinneret. The fiber consists of two layers made from polyamide-
Received in revised form imide (PAI) polymer for the outer layer and polyethersulfone (PES) polymer for the porous inner layer.
3 May 2012
Specifically, after obtaining asymmetric microporous PAI/PES dual-layer hollow fibers via non-solvent
Accepted 21 July 2012
induced phase inversion, polyethyleneimine (PEI) polyelectrolyte modification on the outer PAI layer
Available online 28 July 2012
was applied to produce a nanofiltration (NF)-like thin layer, while the PES porous inner layer remained
Keywords: intact as PES is inert to PEI. The membrane morphology, structure and surface property were carefully
Forward osmosis tailored by adjusting polymer dope composition and spinning conditions. These membranes were
Membrane fabrication
subsequently characterized by a series of standard protocols in terms of membrane structure,
Dual-layer hollow fiber membrane
permeability and salt rejection, and were utilized in FO process.
Positively-charged nanofiltration
Chemical cross-linking It was found that the resultant dual-layer NF hollow fiber membrane can achieve pure water
permeability of 15.9 l m  2 h  1 bar  1 and a high rejection to divalent cations up to 89%. In FO process,
the dual-layer hollow fiber exhibited a water flux of 27.5 l m  2 h  1 in the orientation of active layer
facing feed water by using 0.5 M MgCl2 as draw solution and de-ionized (DI) water as feed at room
temperature. The newly developed dual layer hollow fibers outperform all the single layer and dual-
layer NF hollow fibers reported in the literature for FO applications.
& 2012 Elsevier B.V. All rights reserved.

1. Introduction but it is expensive. The material cost can be reduced by co-extruding

this material with polysulfone which is much cheaper [4].
As an important membrane fabrication technology, the develop- A series of studies have been reported on the use of polymeric
ment of dual-layer hollow fiber membranes made by co-extrusion dual-layer hollow fiber membranes for specific applications, including
method has attracted increasing attention in the past two decades gas separation [4,5], pervaporation [6], membrane distillation [7,8],
since it was invented by Yanagimoto [1]. Compared with conven- protein separation [9], forward osmosis (FO) [3,10], and nanofiltration
tional single-layer hollow fiber membranes, the dual-layer hollow (NF) [2,11]. However, it is very challenging to make high performance
fibers offer several advantages due to the flexibility of using two dual-layer hollow fibers, as it involves a sophisticated spinning
different polymer solutions, which are extruded simultaneously to process in which two different phase inversion pathways occur
develop two layers from a triple orifice spinneret [2]. Each material simultaneously. The major problem in a dual-layer composite struc-
can impart desirable characteristics to the membrane, and the ture is the integrity and the adhesion of the two layers, and the
integration of the two layers makes it possible to avoid certain compositions of the two polymer dope solutions play a critical role.
weakness of individual materials. Yang et al. [3] employed a brittle Increasing the concentration of the outer layer polymer solution
but high performance polymer of polybenzimidazole (PBI) as the improves the adhesion considerably due to the high viscosity of the
selective layer supported by PES which has excellent mechanical solution that decreases water (non-solvent) diffusion. Therefore, a
properties. Matrimid has an impressive gas separation performance good interpenetration of outer and inner polymer dopes can be
achieved [2,12].
To date, most of dual-layer hollow fiber membranes were
n
made using a high polymer concentration as the outer layer for
Corresponding author at: School of Civil and Environmental Engineering,
Nanyang Technological University, 639798 Singapore, Singapore.
gas separation, pervaporation, nanofiltration, and FO applications.
Tel.: þ65 6790 5327; fax: þ 65 6791 0676. For example, Yang et al. developed dual-layer hollow fiber NF
E-mail address: rwang@ntu.edu.sg (R. Wang). membranes applied for FO process [3,10]. A polybenzimidazole

0376-7388/$ - see front matter & 2012 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.memsci.2012.07.020
 L. Setiawan et al. / Journal of Membrane Science 421–422 (2012) 238–246 239

(PBI) dope solution with a PBI concentration of 22.6 wt% was Table 1
employed as the outer selective layer while polyethersulfone Composition of polymer solutions.
(PES) was used to form the inner support layer. The membranes
Code Compositiona Flow rate ratio
showed a water flux of 33.8 l m  2 h  1 using 5 M MgCl2 as the outer:inner
draw solution facing the outer active layer and de-ionized (DI) Outer dope Inner dope (wt/wt)
water as the feed facing the inner support layer. Considering that
the draw solution concentration used was quite high, obviously, A PAI/NMP 14/86 PES/LiCl/NMP 16/6/78 01:02.6
B PAI/LiCl/NMP 14/3.8/82.2 PES/NMP 16/84 01:03.0
the water flux needs to be further improved. In the FO process, C PAI/LiCl/NMP 14/3.8/82.2 PES/PEG400/NMP 16/10/74 01:03.2
the low permeate flux is mainly due to internal concentration D PAI/LiCl/NMP 14/3.8/82.2 PES/PEG400/NMP 16/10/74 01:02.4
polarization (ICP) [13,14]. The ICP can be mitigated by fabricating E PAI/LiCl/NMP 14/3.8/82.2 PES/LiCl/NMP 16/6/78 01:02.4
membrane with a porous substrate [15–17]. a
Composition is based on wt%.
Recently, single-layer hollow fiber membrane with a NF-like
selective layer was successfully fabricated by utilizing asymmetric
microporous hollow fibers made of Torlons polyamide-imide (PAI)
material as the porous substrate followed by polyelectrolyte post- dope solutions. The detailed of the polymer dope compositions
treatment using polyethyleneimine (PEI), which imparted positive are shown in Table 1. A desired amount of particular polymer and
charges to the membrane due to amine groups in PEI and densified non-solvent additive was dissolved in NMP in a jacket flask
the outer skin of the membrane [18]. However, despite the fact equipped with a mechanical stirrer at a controlled temperature
that the membrane exhibited a high water flux and high salt of 70 1C. Inorganic salt of LiCl and organic macromolecules of PEG
rejection in the FO process, the chemical cross-linking still resulted have been chosen as the non-solvent additive. Both LiCl and PEG
in a denser substrate which adversely affected the water flux. To are highly soluble in water and NMP, and able to promote pore
overcome the drawback of cross-linking modification on entire PAI formation in membrane fabrication [19,20]. Once a homogeneous
membrane, a dual-layer hollow fiber with PAI polymer as the outer solution was formed, the solutions were cooled down to room
layer, supported by an inner layer made of other material inert to temperature, filtered by using a 15 mm stainless steel filter and
PEI polyelectrolyte, may be an option due to the selective cross- subsequently degassed under vacuum at ambient temperature
linking of the PAI layer. over night prior to spinning.
The main objective of present study is to develop composite PAI- The viscosity of polymer dope solutions was measured by a
PES dual-layer hollow fiber membrane applied for FO process. Physica MCR 101 rheometer (Anton Paar). The measurements
Microporous PAI-PES dual-layer hollow fibers were fabricated by were performed using a 25 mm cone plate (CP25-1) in the range
one-step co-extrusion, followed by PEI polyelectrolyte post-treat- of 0.01 to 100 s  1 shear rate at 25 1C. The viscosity of the dope
ment, resulting in a positively charged NF-like thin selective layer solutions was taken at the shear rate of 10 s  1. Measurements
supported by a porous PES inner layer. To the best of our knowledge, were repeated four times.
the cross-linking modification on microporous dual-layer hollow
fiber membranes has not been explored previously as an alternative 2.3. Fabrication of ultrafiltration PAI-PES dual-layer hollow fiber
strategy to fabricate NF membranes applied for FO application. substrates

PAI-PES UF dual-layer hollow fibers were fabricated by a dry–jet

2. Materials and methods wet spinning technique. The inner polymer solution, connected to
a high pressure nitrogen gas cylinder, was extruded through the
2.1. Materials spinneret at a specific flow rate using a Zenith gear pump, while
the outer polymer solution and the bore fluid were extruded using
Torlons 4000T (copolymer of amide and imide) (PAI, Solvay two syringe pumps, respectively. The nascent fibers went through
Advanced Polymers, Alpharetta GA) and GafoneTM polyethersulfone an air gap before immersing into an external coagulation bath at a
(PES, Solvay Advanced Polymers, Gujarat) were used as selective controlled temperature, and then collected by a roller at a free fall
layer and support layer materials, respectively. N-methyl-2-pyrroli- take-up speed. Different air gap distances ranging from 1 to 10 cm
done (NMP, 499.5%, CAS#872-50-4, Merck Chemicals, Singapore) and different ratio of inner dope flow rate to internal bore fluid
was used as solvent. Lithium chloride (LiCl, anhydrous, CAS#7447- flow rate were applied to investigate the optimum spinning
41-8, Merck) and polyethylene glycol (PEG, MW 400, Merck) and conditions of desirable dual layer hollow fiber membranes. The
were used as additives/pore formers. Purified water by a Milli-Q codes of resultant membranes are designated as composition (A–E,
system (18 M O cm) was used as the internal coagulant and solvent see Table 1)—ratio of inner dope flow rate to bore fluid flow rate by
for the preparation of aqueous solutions. Dextrans with different weight (1:1 or 2:3) – air gap (1, 5, or 10 cm) – external coagulant
molecular weights (from 6000 to 500,000 Da, CAS#9004-54-0, bath temperature (25 1C as default or 40 1C). For instance, the
Sigma) were used to characterize the molecular weight cut-off membrane substrate B-11-1 refers to the membrane made of
(MWCO) of hollow fiber membranes. Polyethyleneimine (PEI) ethy- composition B with 1:1 (wt/wt) ratio of inner dope flow rate to
lenediamine end-capped (molecular weight (Mw) 600, Sigma bore fluid flow rate and an air gap of 1 cm at 25 1C; the C-23-5-T40
Aldrich) and PEI branched (Mw 50–100 kDa, Polysciences) were means that the membrane was made of composition C with 2:3
used to perform chemical post-treatment of the microporous hollow (wt/wt) ratio of inner dope flow rate to bore fluid flow rate and an
fibers. For filtration experiments, magnesium chloride (MgCl2, air gap of 5 cm at 40 1C. Tap water and Milli-Q water were used as
hexahydrate), purchased from Merck, was employed. All the external and internal coagulants, respectively, at ambient tem-
reagents were used as received. perature. The details of the spinning conditions used for experi-
ments are shown in Table 1 and Table 2. The resultant hollow fiber
2.2. Preparation of polymer dope solutions and dope viscosity membranes were stored in a water bath for approximately two day
measurement at room temperature to ensure that the residual solvent has been
removed completely.
PAI and PES were dried in a 50 1C vacuum oven for at least A post-treatment was performed in order to minimize the
1 day to remove the moisture prior to the preparation of polymer membrane shrinkage during drying process for storage purpose.
240 L. Setiawan et al. / Journal of Membrane Science 421–422 (2012) 238–246

Table 2 represents the pressure difference between the feed side and the
Spinning parameters. permeation side of the membrane (bar).
Substrate code Composition Flow rate ratio Water bath Air gap,
inner dope:bore temp, (1C) (cm) 2.5.3. Molecular weight cut off (MWCO) and pore size distribution
fluid (wt/wt) MWCO and pore size distribution of dual layer hollow fiber
substrates were assessed by using a 2000 ppm dextran aqueous
A-23-1 A 2:3 25 1
solution containing a mixture of several different molecular
A-23-5 A 2:3 25 5
B-11-1 B 1:1 25 1 weights from 6000 to 500,000 Da. The dextran solution was
B-11-5 B 1:1 25 5 circulated through the shell side of the hollow fiber module and
B-11-10 B 1:1 25 10 the permeate was collected for analysis by gel permeation
B-23-1 B 2:3 25 1
chromatography (GPC) on a Polymer Laboratories-GPC 50 plus
B-23-5 B 2:3 25 5
B-23-10 B 2:3 25 10
system (double PL aquagel-OHMixed-M 8m columns). The details
C-23-1 C 2:3 25 1 of the pore size distribution calculation can be found elsewhere
C-23-5 C 2:3 25 5 [22]. The molecular weight of dextran that gave 90% rejection was
C-23-5-T40 C 2:3 40 5 recorded as the MWCO.
D-23-5-T40 D 2:3 40 5
E-23-1 E 2:3 25 1
E-23-5 E 2:3 25 5 2.5.4. Salt rejection measurement of modified dual
layer hollow fibers
The salt rejection of chemically modified PAI-PES dual layer
hollow fibers was conducted in a bench scale cross-flow filtration
The membranes were immersed in a glycerol/water mixture (1:1 unit. The hydraulic pressure of 1 bar was applied on the shell side
by volume) for 24 h. This process allowed the glycerol to stay of the hollow fiber membrane module. The salt rejection experi-
inside the membrane pores as pore supporter to alleviate the pore ment was carried out using a 500 ppm MgCl2 solution based on
collapse during membrane drying process. Removing the glycerol conductivity measurement (Ultrameter II, Myron L Company,
can be done by immersing the membranes in DI water for 24 h Carlsbad, CA) of permeate and feed solutions.
prior to use.
2.6. Performance in FO process
2.4. Chemical cross-linking by using PEI polyelectrolyte
The schematic diagram of a lab-scale cross-flow FO unit used
The chemical post-treatment was conducted by immersing the in this study can be found elsewhere [23]. The same modules used
hollow fiber substrate into a 500 ml of 1% (wt/wt) PEI aqueous for PWP and salt rejection measurements were used for FO
solution at the temperature of 70 1C for 75 min. Next, the mem- experiment. One variable-speed gear pump was used to supply
branes were rinsed using purified water and stored for further the draw solution and two variable speed peristaltic pumps were
characterization. Details procedure can be found elsewhere [18]. used to supply the feed and dosing solution, respectively. The
volumetric flow rates of the shell side and lumen side were 1250
2.5. Characterization and analysis and 500 ml min  1, respectively, to ensure a similar Reynolds
number (around 2600) of the liquid flowing both in the module
2.5.1. Membrane morphology observation and mechanical property shell and fiber lumen. FO experiments were performed in two
measurement configurations: (1) draw solution flowed in shell side or active
The dimension of dual-layer hollow fiber membranes was layer facing draw solution (AL-DS), or known as PRO mode, and
measured by a Keyence VHX 500F Digital Microscope. Four (2) draw solution flowed in lumen side or active layer facing feed
different fibers were taken and a mean value was calculated for water (AL-FW), or known as FO mode. The volumetric water flux,
each sample measurement. The structure and morphology of Jv, was determined at a certain time interval by measuring the
resultant membranes were examined by a Zeiss EVO 50 scanning weight changes of the feed tank with a digital mass balance
electron microscope (SEM). Dry membrane samples were frozen connected to a data logging system.
in liquid nitrogen and subsequently cracked in order to obtain the
cross sections. The samples were then carefully mounted on the
SEM stubs and dried overnight in a 50 1C vacuum oven. An 3. Results and discussions
Emitech SC7620 gold sputter coater was used to deposit a layer
of gold on the samples under argon environment. 3.1. Fabrication of dual-layer hollow fibers substrates

2.5.2. Pure water permeability (PWP) of dual-layer hollow fibers 3.1.1. Effect of the air gap and the ratio of bore fluid to inner dope
membranes before and after cross-linking modification flow rate on the morphology of dual layer hollow fiber membranes
Four pieces of dual-layer hollow fibers were potted into a module The effect of the air gap and ratio of bore fluid to inner dope
and sealed to prepare a lab-scale module with an effective length of flow rate were observed by simultaneous spinning of PAI/LiCl/
21 cm. PWP experiments were performed by using two to three NMP (14/3.8/82.2 wt%) as the outer polymer dope solution and
modules from the same batch of the membrane spinning process. PES/NMP (16/84 wt%) as the inner polymer dope solution. Fig. 1
Milli-Q ultra pure water was circulated through the shell side of the shows the cross-section morphology of dual layer hollow fiber
membrane module under a pressure of 1 bar for 90 min to compact samples and illustrates how the air gap affects the inner contour
the membrane prior to PWP measurement [21]. The PWP of the morphology at 1:1 (wt/wt) ratio of inner dope to bore fluid flow
membranes (l m  2 h  1 bar  1) was calculated by: rate. It is clearly observed from Fig. 1a that the fiber spun at 1 cm
air gap has irregular inner contour resulting in uneven wall
V thickness. There are two possible reasons associated with this
PWP ¼ ð1Þ
tADP phenomenon. At a lower air gap, the polymer molecules have a
where V is the volume of permeate taken (l) per determined time, t shorter time to relax and release the stress after extruding out of
(h); A is the filtration area of the dual-layer membrane (m2); and DP the spinneret. The instantaneous demixing and solidification at
 L. Setiawan et al. / Journal of Membrane Science 421–422 (2012) 238–246 241

Fig. 1. Cross-section morphology of dual layer hollow fiber membrane spun by using dope formula B (PAI/LiCl/NMP 14/3.8/82.2 wt% and PES/NMP 16/84 wt%) at different
air gaps: (a) 1 cm; (b) 5 cm; (c) 10 cm with inner dope flow rate to bore fluid flow rate ratio of 1:1 (wt/wt).

the outer side of the fiber caused by the strong external coagulant 3.1.2. Effects of coagulation bath temperature and dope flow rate
tended to freeze the polymer molecules, forcing the stress held by Coagulation bath temperature has an effect on the structure
the macromolecules to release toward lumen radial direction and morphology of the resultant hollow fiber membranes. Thus
[24]. The second is that, when the pressure inside the fiber lumen the membrane structure can be tailored simply by varying the
is low due to insufficient supply of bore fluid, there will be a system temperature [30]. Fig. 3 shows the morphologies of dual
compression of the inner wall because of a rapid formation of the layer hollow fiber membranes spun at 25 and 40 1C coagulation
lumen skin when a strong bore fluid is used [25]. bath temperatures. A significant change of the cross-section
Two approaches were carried out to confirm the above morphology has been observed when the coagulation bath
hypotheses. (1) Fig. 1b and c show the SEM pictures of fiber cross temperature increased from 25 to 40 1C. The twisted finger-like
section morphology spun at 5 and 10 cm air gap, respectively. It macrovoids formed at 25 1C spun (Fig. 3a) disappeared and
can be seen that the irregularity of the inner contour of the fibers became straight at 40 1C (Fig. 3b). The increase in coagulation
became less severe with an increase in the air gap. When the air bath temperature enhanced the thermodynamic stability of the
gap was increased, the stress accumulated within the polymer polymer solutions, leading to a delayed demixing that seems to
chains in the spinneret can be reduced gradually, as the outer part allow the polymer molecules to re-arrange [19].
of the nascent fiber was still soft allowing the polymer chains to However, the increase in water bath temperature made the
re-arrange and relax in the period of passing through the air gap boundary between the outer layer and inner layer visible as
[24,26]. (2) Fig. 2a shows the cross section morphology of hollow pointed by the arrow on Fig. 3b. The major reason might be due
fiber membrane spun at 1 cm air gap and inner dope to bore fluid to the different solidification rates between the outer and the
ratio of 2:3 (wt/wt). By increasing the bore fluid flow rate, the inner polymer dope solutions, as the bore fluid temperature was
irregularity of the inner contour can be prevented. Santoso et al. at ambient temperature (25 1C). The inner layer experienced
[25] also reported that increasing the flow rate of bore fluid can instantaneous demixing due to the thermodynamic instability
maintain a uniform inner layer structure. while the delayed demixing occurred on the outer layer. The
However, it seems that increasing the bore fluid flow rate in fibers that spun based on the ratio of outer to inner dope flow rate
combination with an increase in air gap can lead to the formation of 1:3.2 (wt/wt) has an outer layer with a thickness of around
of big macrovoids (Fig. 2b) and cracks on the fiber cross-section 25 mm. As the ratio changes to 1:2.4 (wt/wt), the outer layer
(Fig. 2c) though it can avoid the irregularity of the inner contour. became thicker which is around 40 mm (Fig. 3c).
The formation mechanism of macrovoids in the phase separation
has been studied extensively in the literature [27–29]. Since
water was also used as the bore fluid that was in contact with 3.1.3. Effect of non-solvent additive on the morphology of the dual-
the inner polymer dope first, resulting in an earlier phase layer hollow fiber membranes
inversion than the outer polymer dope, the macrovoids tended The effect of the addition of an additive in the outer dope of PAI/
to develop from the inner side toward the outer layer. Big NMP system is shown in Figs. 4 and 5. Fig. 4 shows the cross section
macrovoids are not generally favorable because it can reduce morphology of the dual layer hollow fiber membrane obtained by
the mechanical strength of the hollow fiber membrane. simultaneous spinning of PAI/NMP (14/86 in wt%) as the outer layer
242 L. Setiawan et al. / Journal of Membrane Science 421–422 (2012) 238–246

Fig. 2. Cross-section morphology of dual layer hollow fiber membrane spun by using dope formula B (PAI/LiCl/NMP 14/3.8/82.2 wt% and PES/NMP 16/84 wt%) at different
air gaps: (a) 1 cm; (b) 5 cm; (c) 10 cm with inner dope flow rate to bore fluid flow rate ratio of 2:3 (wt/wt).

Fig. 3. Cross-section morphology of dual layer hollow fiber membrane spun at coagulation bath temperature of 25 1C (a) and 40 1C (b) and (c) at air gap of 5 cm with PEG
400 as the additive of inner polymer dope. Ratio of outer layer to inner layer thickness: (b) 1/4 and (c) 1/3.

and PES/LiCl/NMP (16/6/78 in wt%) as the inner layer (composition significantly to 8.2 Pa s when 3.8% LiCl was added. The high dope
A). A separation of two layers formed by each polymer dope solution viscosity delays the diffusion of the non-solvent to the interface.
can be clearly observed, suggesting the interpenetration delay of the Similar result has been reported elsewhere [2]. Further investigation
inner and outer dopes. Fig. 5 shows the cross section morphology of is needed to achieve a better understanding for this delamination
the dual layer hollow fiber membrane obtained by simultaneous issue.
spinning of PAI/LiCl/NMP (14/3.8/82.2 in wt%) as the outer layer and In order to fabricate a dual layer membrane with a porous
PES/LiCl/NMP (16/6/78 in wt%) as the inner layer (composition E). No inner layer, two types of additives, PEG400 and LiCl, were added
delamination can be observed as the outer dope viscosity increased into the PES solution as the inner dope individually. As shown in
 L. Setiawan et al. / Journal of Membrane Science 421–422 (2012) 238–246 243

Fig. 4. Cross-section morphology of dual layer hollow fiber membrane spun by using dope formula A (PAI/NMP 14/86 wt% and PES/LiCl/NMP 16/6/78 wt%) at different air
gaps: (a) 1 cm; (b) 5 cm; with inner dope flow rate to bore fluid flow rate ratio of 2:3 (wt/wt).

Fig. 5. Cross-section morphology of dual layer hollow fiber membrane spun by using dope formula E (PAI/LiCl/NMP 14/3.8/82.2 wt% and PES/LiCl/NMP 16/6/78 wt%) at
coagulation bath temperature of 25 1C at different air gap: (a) 1 cm; (b) 5 cm.

Fig. 3, using PEG as an additive, the resultant dual layer hollow increases Ddt to 11.9 (MPa)1/2 which enhances the thermody-
fibers exhibited large macrovoids and a big portion of sponge-like namic instability [32]. However, the addition of LiCl increased the
structure, which are unfavourable due to weak mechanical viscosity of PES solution significantly from 0.5 Pa s (polymer and
strength. In contrast, when a small molecular additive, LiCl, was solvent only) to 6.9 Pa s (with 6% LiCl), as listed in Table 4, due to
used, the large macrovoids in the inner layer was suppressed, as the strong interaction of the additive with the solvent. A high
shown in Fig. 5. These observations were believed to be asso- viscosity of a spinning dope slows down the non-solvent diffusion
ciated with the change in the thermodynamic properties and into the polymer solution leading to a delayed demixing, and
phase inversion kinetics of the PES systems without and with the thus, suppresses the formation of big macrovoids [20].
addition of an additive [20,31]. Hansen solubility parameter for
each component in PES system is summarized in Table 3, from 3.2. Substrate characteristics: pure water permeability (PWP),
which the total solubility parameter difference can be derived MWCO, and pore size distribution
(Table 4). PES systems without an additive and with 10% PEG 400
as the additive have the total solubility parameter difference (Ddt) The PWP and MWCO of hollow fiber membranes prepared
of 1.08 and 0.98 (MPa)1/2. The addition of 6% LiCl significantly with different conditions are presented in Table 5. It can be seen
244 L. Setiawan et al. / Journal of Membrane Science 421–422 (2012) 238–246

Table 3 after PEI cross-linking for PES. Hereafter the membrane was
Hansen solubility parameter. denoted as dual layer FO hollow fiber with a code of membrane
substrate—PEI molecular weight, such as C-23-5-50K. The intrin-
Component dd (MPa)1/2 dp (MPa)1/2 dh (MPa)1/2 dt (MPa)1/2n Ref.
sic properties of PAI-PES dual layer hollow fiber membranes with
PES 17.6 10.4 7.8 21.9 [33] a positively charged NF-like selective layer such as water perme-
PEG 400 19.2 3.5 3.6 19.8 [34] ability and rejection of MgCl2 are tabulated in Table 6.
LiCl – –- – 94.2 [35] It can be seen from Table 6 that when the membrane
NMP 18.0 12.3 7.2 23.0 [32]
substrates of C-23-5 and E-23-5 were cross-linked with PEI
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
n 2 2
Total solubility parameter (dt) is calculated by dt ¼ dd þ dp þ dh [32]
2 600 Da, the substrate with a higher MWCO (E-23-5) had a higher
water permeability and a lower rejection of MgCl2. This can be
explained as follows. The C-23-5, C-23-5-T40, and E-23-5

Table 4
Viscosity and solubility parameter of polymer dope solutions.
1.6
a b n
Composition Viscosity, (Pa s) Ddt ¼9dsolvent-additive
1.4
 dpolymernn9 (MPa)1/2

Pore size distribution, f(d)

1.2
PAI/NMP 14/86 1.5 –
PAI/LiCl/NMP 14/3.8/82.2 8.2 –
1.0
PES/NMP 16/84 0.5 1.08
PES/PEG400/NMP 16/10/74 0.8 0.98
0.8
PES/LiCl/NMP 16/6/78 6.9 11.9

a 0.6
Composition is based on wt%.
b
Viscosity was measured at shear rate of 10 s  1 at 25 1C.
n
0.4
dsolvent-additive was calculated based on mol fractions of total solubility
parameter of solvent and additive.
nn 0.2
dpolymer is total solubility parameter of polymer.
0.0
0 2 4 6 8 10 12 14 16
Pore diameter, nm
Table 5 Fig. 6. Pore size distributions (f(d)) of PAI-PES dual layer hollow fiber membranes
Dimension, PWP and MWCO of dual-layer hollow fiber substrates spun at different ((’) C-23-1; (&) C-23-5; (m) C-23-5-T40 ; (n) D-23-5-T40; (K) E-23-1; (J) E-23-5) .
conditions.

Membrane Dimension PWP, Outer skin

substrate code (l m  2 h  1 bar  1) MWCO, (kDa)
OD, ID,
(mm) (mm)

C-23-1 1.43 1.08 111 7 26 9

% Transmission

C-23-5 1.43 1.07 135 7 3 15.7

C-23-5-T40 1.43 1.05 189 7 27 43.3
D-23-5-T40 1.44 1.04 162 7 14 23.6
E-23-1 1.38 1.02 119 7 34 20.5
E-23-5 1.39 1.03 149 7 16 30.8

that all fibers possessed quite large dimensions (OD/ID 1.40/

1.05 mm/mm). A large fiber lumen is favourable for the fluid flow
with less resistance. From this table, it is clear that PWP increased 1850 1650 1450 1250 1050 850 650
as the air gap increased. A similar trend is also shown for MWCO
Wave length (cm-1)
of the outer skin. This is in agreement with the change of the pore
size and pore size distribution presented in Fig. 6. As the Fig. 7. FTIR spectra of PES before (——) and after (----) PEI cross-linking.
temperature of the coagulation bath increased from 25 to 40 1C,
the PWP and the outer skin MWCO also increased for fibers spun
at the same air gap due to the increasing of the thermodynamic
stability of the system as discussed previously. Furthermore, it is Table 6
Intrinsic properties of dual layer FO hollow fiber membranes*.
depicted from Fig. 6 that the viscosity of inner dope affects the pore
size distribution of the resultant fibers. Fibers spun with inner Substrate Polyethylenimine, MW 600 Da Polyethylenimine, MW
polymer dope having viscosity of 6.9 Pa s (E-23-1 and E-23-5) have code 50–100 kDa
a narrower pore size distribution than that having a lower viscosity.
Water permeability, Rej. Water permeability, Rej.
(l m  2 h  1 bar  1) MgCl2, (l m  2 h  1 bar  1) MgCl2,
3.3. Intrinsic properties of dual layer FO hollow fiber membranes (%) (%)

The cross-linking reaction was taken place at the outer layer of C-23-5 6.5 83 14.4 80
the dual layer substrate between PAI and PEI. The details of the C-23-5-T40 – - 22.5 55
E-23-5 22.6 58 15.9 89
cross-linking reaction and charge characteristic can be found
elsewhere [18]. The PES as the inner layer remains intact as n
Condition of polyethylenimine (PEI) crosslinking: 1% PEI aqueous solution in
shown in Fig. 7. There is no change in FTIR spectra before and 70 1C water bath for 75 min.
 L. Setiawan et al. / Journal of Membrane Science 421–422 (2012) 238–246 245

substrates had a MWCO of 15.7, 43.3, and 30.8 kDa, respectively. However, in the AL-FW orientation, the salt concentration at the
When PEI 600 Da was used, the PEI molecules were able to interfaces between the outer layer and the inner support layer was
penetrate into the membrane pores and carried out the cross- lower than the bulk concentration of the draw solution due to water
linking reaction inside the membrane matrix. However, when the dilution. Therefore, the shielding effect may not be severe and the
MWCO of the substrate was too big, the PEI 600 Da was unable to co-ions can be repelled by the positively charged outer layer of the
fully seal the big pores. Therefore, the water permeability is high membrane. In addition, the facilitated transport in the AL-DS
while the rejection is low. To further understand the effect of orientation (salt flux and salt repulsion are in the same direction)
MWCO of membrane substrate and the molecular weight of and retarded transport in the AL-FW orientation (salt flux and salt
cross-linker on the water permeability and salt rejection, PEI repulsion are in the opposite direction) also played roles [18].
50–100 kDa was used to perform chemical cross-linking for C-23- Furthermore, the water flux in the orientation of AL-FW is higher
5, C-23-5-T40, and E-23-5 substrates which have similar or lower than that in AL-DS as shown in Table 7. Normally, AL-FW is known
MWCO than the molecular weight of cross-linker. It is interesting to show a lower flux due to more severe dilutive ICP. The lower than
to see that C-23-5 and E-23-5 substrate had a good water expected water flux in AL-DS might be due to the high salt flux. As a
permeability and acceptable salt rejection while C-23-5-T40 sub- result, the salt concentration at the interface of the selective layer
strate had quite poor rejection. This might be due to the and porous support increased, leading to the decrease in the
combination of pore size and pore size distribution of each effective osmotic pressure difference across the membrane.
substrate. As shown in Fig. 6, C-23-5-T40 substrate has a very Table 8 compares the FO performance of the dual layer hollow
broad pore size distribution. Thus, the big pores cannot be fully fiber membranes developed in current work with other single
sealed during the cross-linking reaction. layer and dual layer membranes reported in the literature. It
seems that the water flux of C-23-5-600, C-23-5-50K and E-23-5-
50K dual layer FO hollow fiber membranes in the AL-FW orienta-
3.4. FO performances tion are much higher than other membranes with comparable Js/Jv
values. This may be attributed to the porous inner layer, that can
The FO performance of dual layer FO hollow fiber membranes be remained intact during the chemical cross-linking treatment,
with a salt rejection over 80% was determined using a 0.5 M MgCl2 to overcome the concentrative ICP within the inner layer.
solution as the draw solution and de-ionized water as the feed water
at ambient temperature of 23 1C, and the results are listed in Table 7.
It was found that all of the dual layer FO hollow fiber membranes
have a better performance in the AL-FW orientation than in the AL- 4. Conclusions
DS orientation, similar to the results reported previously [18]. The
Js/Jv in the AL-FW for all membranes is much smaller compared with Microporous dual layer hollow fiber membranes using PAI as
their counterparts in the AL-DS orientation. This might be due to the the selective layer and PES as the inner support layer have been
shielding effect occurred at the interface between the high concen- successfully fabricated with a delamination-free structure. The
tration draw solution and the positively charged active layer in the compositions of the outer and inner polymer solutions, air gap
AL-DS orientation by the counter ions [36] and consequently, the and coagulation water bath temperature, etc. played key roles in
active layer of the membrane was unable to reject the co-ions. determining the morphology and structure of dual layer hollow
fiber membranes.
Table 7 Following a simple chemical cross-linking treatment on the
Performance of dual layer FO hollow fiber membranes applied in FO process*. microporous dual layer hollow fibers using a PEI solution, dual-
layer FO hollow fiber membranes with a positively charged NF-
Membrane code AL-facing-DS AL-facing-FW
like selective layer can be developed. The resultant novel FO
Jv, (l m  2 h  1) Js/Jv, (g l  1) Jv, (l m  2 h  1) Js/Jv, (g l  1) membranes present promising performance in the AL-FW orien-
tation for FO application. A water flux of 27.5 l m  2 h  1 was
C-23-5-600 14.5 5.9 20.6 0.3
achieved using 0.5 M MgCl2 as draw solution and DI-water as feed
C-23-5-50K - – 24.3 0.3
E-23-5-50K 15.5 5.4 27.5 0.2 at room temperature. The newly developed dual-layer hollow
fibers outperform all the single layer and dual-layer NF mem-
n
Draw solution: 0.5 M MgCl2 and feed water: DI water. branes reported in the literature for FO applications.

Table 8
Comparison of various membranes used in FO process.

Membrane PWP, Rejection MgCl2, FO: AL-facing-FW Testing conditions Ref.

(l m  2 h  1 bar  1) (%)
Jv, Js/Jv, Feed Draw
(l m  2 h  1) (g l  1) solution solution

C-23-5-600 6.5 83 20.6 0.3 DI-water 0.5 M MgCl2 Current

C-23-5-50K 14.4 80 20.6 0.3 work
E-23-5-50K 15.9 89 27.5 0.2
DL-PBI-PES 1.74 87 7.5 0.1 DI-water 1.0 M MgCl2 [3]
Positively charged FO single layer PAI hollow fiber 2.25 93 8.4 0.3 DI-water 0.5 M MgCl2 [18]
Positively charged FO single layer PAI hollow fiber 2.19 91 9.7 0.3
Positively charged FO PAI flat sheet membrane 7.56 87 19.2 0.5 DI-water 0.5 M MgCl2 [37]
Neutrally charged FO single layer hollow fiber 4.3 85 13 0.2 DI-water 0.5 M Na2SO4 [38]
Double dense layer flat sheet membrane (thickness 0.17 99 10.3 70.3 0.08 DI-water 2.0 M MgCl2 [39]
35 mm)
Cellulose acetate hollow fiber NF membrane 5.56 15 5 – DI-water 2.0 M MgCl2 [40]
246 L. Setiawan et al. / Journal of Membrane Science 421–422 (2012) 238–246

Acknowledgements [19] S. Wongchitphimon, R. Wang, R. Jiraratananon, L. Shi, C.H. Loh, Effect of

polyethylene glycol (PEG) as an additive on the fabrication of polyvinylidene
fluoride-co-hexafluropropylene (PVDF-HFP) asymmetric microporous hollow
We thank Ms Susan Sulaiman for her kind assistance in fiber membranes, J. Membr. Sci. 369 (2011) 329–338.
conducting FO experiments. This research grant is supported by [20] L. Shi, R. Wang, Y. Cao, D.T. Liang, J.H. Tay, Effect of additives on the
the Singapore National Research Foundation under its Environ- fabrication of poly(vinylidene fluoride-co-hexafluropropylene) (PVDF-HFP)
mental & Water Technologies Strategic Research Programme and asymmetric microporous hollow fiber membranes, J. Membr. Sci. 315 (2008)
195–204.
administered by the Environment & Water Industry Programme [21] K.M. Persson, V. Gekas, G. Trägårdh, Study of membrane compaction and its
Office (EWI) of the PUB (EWI RFP 0801-IRIS-05). We also acknowl- influence on ultrafiltration water permeability, J. Membr. Sci. 100 (1995)
edge the funding support from Singapore Economic Development 155–162.
Board to Singapore Membrane Technology Centre. [22] J. Ren, R. Wang, H.-Y. Zhang, Z. Li, D.T. Liang, J.H. Tay, Effect of PVDF dope
rheology on the structure of hollow fiber membranes used for CO2 capture, J.
Membr. Sci. 281 (2006) 334–344.
References [23] S. Chou, L. Shi, R. Wang, C.Y. Tang, C. Qiu, A.G. Fane, Characteristics and
potential applications of a novel forward osmosis hollow fiber membrane,
Desalination 261 (2010) 365–372.
[1] T. Yanagimoto, Manufacture of Ultrafiltration Membranes. Japanese Patent
[24] L. Shi, R. Wang, Y. Cao, C. Feng, D.T. Liang, J.H. Tay, Fabrication of
62019205 (1987).
poly(vinylidene fluoride-co-hexafluropropylene) (PVDF-HFP) asymmetric
[2] T. He, M.H.V. Mulder, H. Strathmann, M. Wessling, Preparation of composite
hollow fiber membranes: co-extrusion of hydrophilic coatings onto porous microporous hollow fiber membranes, J. Membr. Sci. 305 (2007) 215–225.
hydrophobic support structures, J. Membr. Sci. 207 (2002) 143–156. [25] Y.E. Santoso, T.S. Chung, K.Y. Wang, M. Weber, The investigation of irregular
[3] Q. Yang, K.Y. Wang, T.-S. Chung, Dual-layer hollow fibers with enhanced flux inner skin morphology of hollow fiber membranes at high-speed spinning
as novel forward osmosis membranes for water production, Environ. Sci. and the solutions to overcome it, J. Membr. Sci. 282 (2006) 383–392.
Technol. 43 (2009) 2800–2805. [26] K.C. Khulbe, C.Y. Feng, F. Hamad, T. Matsuura, M. Khayet, Structural and
[4] X. Ding, Y. Cao, H. Zhao, L. Wang, Q. Yuan, Fabrication of high performance performance study of micro porous polyetherimide hollow fiber membranes
matrimid/polysulfone dual-layer hollow fiber membranes for O2/N2 separa- prepared at different air-gap, J. Membr. Sci. 245 (2004) 191–198.
tion, J. Membr. Sci. 323 (2008) 352–361. [27] C.A. Smolders, A.J. Reuvers, R.M. Boom, I.M. Wienk, Microstructures in phase-
[5] N. Peng, T.-S. Chung, M.L. Chng, W. Aw, Evolution of ultra-thin dense-selective inversion membranes. Part 1. Formation of macrovoids, J. Membr. Sci. 73
layer from single-layer to dual-layer hollow fibers using novel ExtemÂs (1992) 259–275.
polyetherimide for gas separation, J. Membr. Sci. 360 (2010) 48–57. [28] S.A. McKelvey, W.J. Koros, Phase separation, vitrification, and the manifesta-
[6] Y. Wang, S.H. Goh, T.S. Chung, P. Na, Polyamide-imide/polyetherimide dual- tion of macrovoids in polymeric asymmetric membranes, J. Membr. Sci. 112
layer hollow fiber membranes for pervaporation dehydration of C1–C4 (1996) 29–39.
alcohols, J. Membr. Sci. 326 (2009) 222–233. [29] J. Barzin, B. Sadatnia, Correlation between macrovoid formation and the
[7] S. Bonyadi, T.S. Chung, Flux enhancement in membrane distillation by ternary phase diagram for polyethersulfone membranes prepared from two
fabrication of dual layer hydrophilic-hydrophobic hollow fiber membranes,
nearly similar solvents, J. Membr. Sci. 325 (2008) 92–97.
J. Membr. Sci. 306 (2007) 134–146.
[30] L.-P. Cheng, Effect of temperature on the formation of microporous PVDF
[8] M.M. Teoh, T.-S. Chung, Y.S. Yeo, P.V.D. Dual-layer, /PTFE composite hollow
membranes by precipitation from 1-octanol/DMF/PVDF and water/DMF/
fibers with a thin macrovoid-free selective layer for water production via
PVDF systems, Macromolecules 32 (1999) 6668–6674.
membrane distillation, Chem. Eng. J 171 (2011) 684–691.
[9] Y. Li, T.-S. Chung, Exploration of highly sulfonated polyethersulfone (SPES) as [31] I. Ahmed, A. Idris, N.F.C. Pa, Novel method of synthesizing poly(ether sulfone)
a membrane material with the aid of dual-layer hollow fiber fabrication membranes containing two solvents and a lithium chloride additive and their
technology for protein separation, J. Membr. Sci. 309 (2008) 45–55. performance, J. Appl. Polym. Sci. 115 (2010) 1428–1437.
[10] Q. Yang, K.Y. Wang, T.-S. Chung, A novel dual-layer forward osmosis [32] C.M. Hansen, Hansen Solubility Parameters—A User’s Handbook, second ed.,
membrane for protein enrichment and concentration, Sep. Purif. Technol. CRC Press, 2007.
69 (2009) 269–274. [33] Z.-L. Xu, F.Alsalhy Qusay, Polyethersulfone (PES) hollow fiber ultrafiltration
[11] S.P. Sun, K.Y. Wang, N. Peng, T.A. Hatton, T.-S. Chung, Novel polyamide- membranes prepared by PES/non-solvent/NMP solution, J. Membr. Sci. 233
imide/cellulose acetate dual-layer hollow fiber membranes for nanofiltration, (2004) 101–111.
J. Membr. Sci. 363 (2010) 232–242. [34] K. Adamska, A. Voelkel, K. Héberger, Selection of solubility parameters for
[12] D. Li, T.-S. Chung, R. Wang, Morphological aspects and structure control of characterization of pharmaceutical excipients, J. Chromatogr. A 1171 (2007)
dual-layer asymmetric hollow fiber membranes formed by a simultaneous 90–97.
co-extrusion approach, J. Membr. Sci. 243 (2004) 155–175. [35] C.L. Yaws, Thermophysical Properties of Chemicals and Hydrocarbons,
[13] T.Y. Cath, A.E. Childress, M. Elimelech, Forward osmosis: principles, applica- William Andrew Inc., Norwich, NY, 2008.
tions, and recent developments, J. Membr. Sci. 281 (2006) 70–87. [36] J.M.M. Peeters, J.P. Boom, M.H.V. Mulder, H. Strathmann, Retention measure-
[14] S. Chou, R. Wang, L. Shi, Q. She, C. Tang, A.G. Fane, Thin-film composite ments of nanofiltration membranes with electrolyte solutions, J. Membr. Sci.
hollow fiber membranes for pressure retarded osmosis (PRO) process with 145 (1998) 199–209.
high power density, J. Membr. Sci. 389 (2012) 25–33.
[37] C. Qiu, L. Setiawan, R. Wang, C.Y. Tang, A.G. Fane, High performance flat sheet
[15] R. Wang, L. Shi, C.Y. Tang, S. Chou, C. Qiu, A.G. Fane, Characterization of novel
forward osmosis membrane with an NF-like selective layer on a woven fabric
forward osmosis hollow fiber membranes, J. Membr. Sci. 355 (2010) 158–167.
embedded substrate, Desalination 287 (2012) 266–270.
[16] W. Fang, R. Wang, S. Chou, L. Setiawan, A.G. Fane, Composite forward osmosis
[38] L. Setiawan, R. Wang, K. Li, A.G. Fane, Fabrication and characterization of
hollow fiber membranes: integration of RO- and NF-like selective layers to
enhance membrane properties of anti-scaling and anti-internal concentra- forward osmosis hollow fiber membranes with antifouling NF-like selective
tion polarization, J. Membr. Sci. 394-395 (2012) 140–150. layer, J. Membr. Sci. 394-395 (2012) 80–88.
[17] L. Shi, S.R. Chou, R. Wang, W.X. Fang, C.Y. Tang, A.G. Fane, Effect of substrate [39] S. Zhang, K.Y. Wang, T.-S. Chung, H. Chen, Y.C. Jean, G. Amy, Well-constructed
structure on the performance of thin-film composite forward osmosis hollow cellulose acetate membranes for forward osmosis: minimized internal
fiber membranes, J. Membr. Sci. 382 (2012) 116–123. concentration polarization with an ultra-thin selective layer, J. Membr. Sci.
[18] L. Setiawan, R. Wang, K. Li, A.G. Fane, Fabrication of novel poly(amide-imide) 360 (2010) 522–535.
forward osmosis hollow fiber membranes with a positively charged nanofil- [40] J. Su, Q. Yang, J.F. Teo, T.S. Chung, Cellulose acetate nanofiltration hollow fiber
tration-like selective layer, J. Membr. Sci. 369 (2011) 196–205. membranes for forward osmosis processes, J. Membr. Sci. 355 (2010) 36–44.