Journal of Membrane Science 563 (2018) 743–751

Contents lists available at ScienceDirect

Journal of Membrane Science

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

Solvent and pH-stable poly(2,5-benzimidazole) (ABPBI) based UF T

membranes: Preparation and characterizations
⁎
Harshada R. Lohokare, Harshal D. Chaudhari, Ulhas.K. Kharul
Polymer Science and Engineering Division, CSIR-National Chemical Laboratory, Dr. Homi Bhabha Road, Pune 411008, India

A R T I C LE I N FO A B S T R A C T

Keywords: Poly(2,5-benzimidazole), commonly known as ABPBI, is an excellent thermo-chemically stable polymer that is
ABPBI membrane widely evaluated as a proton exchange membrane material in a fuel cell. Its niche intrinsic characteristics could
Ultrafiltration be highly useful in the membrane preparation for various separation applications, especially under harsher
Pore size distribution environments. To gain insights towards this feasibility, ABPBI based supported membranes were prepared by
Solvent stable membranes
phase inversion method. Effects of the nonwoven porous support material (polypropylene/polyester), non-sol-
vent (water/0.5 N NaOH) and polymer concentration (6 or 4 wt%) on the membrane properties (water flux,
rejection and porosity) were investigated. The stability of these membranes towards common organic solvents,
concentrated acid (25 N H2SO4), base (2.5 N NaOH) and an autoclave condition was analyzed. ABPBI membrane
showed a pore collapse after drying. In order to avoid this, the glycerol treatment was not only found to be
suitable but also repeatable, without significant deviations in the water flux.

1. Introduction
Excellent stability of polybenzimidazoles (PBIs and ABPBI) as a
proton/hydroxide/Li-ion exchange membrane material under highly
acidic/basic condition is well endorsed by their applications in fuel cells
and Li-ion battery [1–5]. The potential of PBIs is also investigated for
the separation applications such as ultrafiltration (UF) [6,7], nanofil-
tration [8–10], forward osmosis (FO) [11], reverse osmosis (RO) [12],
gas [13–15] and chemodialysis [16,17]. These reports demonstrated
the applicability of PBI-based membranes for the removal of solutes of
different molecular weights and chemical nature.
Among all polybenzimidazoles (PBI), poly(2,5-benzimidazole)
(ABPBI) could be a strong candidate as the membrane material for se-
paration applications, owing to its niche characteristics. Most sig-
nificantly, it is insoluble in common organic solvents such as chlori-
nated solvents, tetrahydrofuran (THF), dioxane, toluene, N,N-
dimethylformamide (DMF), N,N-dimethylacetamide (DMAc), N-me-
thyl-2-pyrrolidone (NMP), etc. It is known to be soluble only in strong
acids such as sulfuric, formic, trifluoroacetic, phosphoric and poly-
phosphoric acid [2]. On the other hand, conventional PBI is soluble in
the polar aprotic solvents such as DMF, DMAc and NMP [10]; which
would enable them to be converted into membranes for separation
applications by a phase inversion method. On the synthetic aspect,
ABPBI is obtained by self-condensation of 3,4-diaminobenzoic acid
(DABA); a less expensive, non-carcinogen and readily available
monomer [18]. Its synthesis is easy in comparison to other PBIs since
polycondensation of only a single monomer is involved [2]. Moreover,
the repeating unit of ABPBI contains only one benzimidazole moiety
and thus have higher N-H group density [19] and lower flexibility than
that of other PBIs. Owing to these structural properties, ABPBI has
better thermo-chemical stability than the conventional PBI [20]. Even
though ABPBI is well investigated as a membrane material for proton
exchange membrane fuel cell (PEMFCs) [1,2,18], it is not demonstrated
for the separation applications. It may be because of its poor solvent
solubility, which is an essential criterion for membrane preparation by
the phase inversion (a most suitable method of the membrane pre-
paration).
There is an increasing demand for membranes that are stable to
various solvents for niche separations in pharmaceutical, bio-
technology, industrial wastewater treatment, food engineering, etc. The
process, termed as the Solvent-Resistant Nanofiltration (SRNF) or
Organic Solvent Nanofiltration (OSN) is widely reviewed in the litera-
ture, detailing various aspects of the membranes, processes, advantages
and disadvantages [21–28]. The solvent stable membranes are gaining
wide attraction due to their ability to do molecular separations, higher
fluxes than RO, transmission of monovalent ions while retaining diva-
lent ions and organic molecules [21]. They can be used in diafiltration
mode to concentrate the solute and allow solvent recovery with an
attractive energy economy [25,26]. For such applications, the mem-
brane should be (i) stable in a wide range of organic solvents, (ii) show

⁎
Corresponding author.
E-mail address: uk.kharul@ncl.res.in (U.K. Kharul).

https://doi.org/10.1016/j.memsci.2018.06.052
Received 6 February 2018; Received in revised form 25 June 2018; Accepted 25 June 2018
Available online 27 June 2018
0376-7388/ © 2018 Elsevier B.V. All rights reserved.
H.R. Lohokare et al. Journal of Membrane Science 563 (2018) 743–751

Nomenclature ρp Density (g cm−3)

ds PEG diameter (nm)
A Membrane area (cm2) a Stoke's radius (nm)
σ Surface tension (dyne cm−1) M Molecular weight of PEG (Da)
σp Geometric standard deviation θ Polymer-water contact angle (°)
P Transmembrane pressure (bar) η Viscosity of water (poise)
R Rejection (%) rp Pore radius (cm)
Jw Water flux (l m−2 h−1) ε Dielectric constant
L Pore length (µm) ηsolvent Viscosity of solvent

high and reproducible performances on the long term and (iii) com- Table 1
bining elevated solvent permeabilities and acceptable rejections for Properties of the porous supports.
molecules in the 200–1000 g mol−1 M mass range [25]. Many chemi- Porous Material of GSMa Thickness Air Water
cally stable polymers have been used to prepare solvent stable, asym- support constructiona (g m−2) (mm) permeabilitya holding
metric membranes via immersion-precipitation viz., poly(vinylidene (l m−2 s−1) capacity
fluoride), polyimide, polyamide, poly(etherimide), polybenzimidazole, (wt%)b

etc., [25]. Recently, Organic-inorganic hybrids have been used as OSN FO2470 Polypropylene 60 0.12 200 34.14
membrane materials [27,28]. MOF-Based membranes, with the in- (PP)
troduction of MOFs in the Polyamide layer, have demonstrated super- H3160 Polyester (PET) 60 0.07 30 23.08
iority in enhancing the OSN performance, compared to integrally- 3324 Polyester (PET) 99.9 0.13 72 29.12
3265 Polyester (PET) 81.2 0.13 160 28.26
skinned asymmetric membranes [28]. Although ceramic membranes
can be a better alternative, in terms of their thermo-mechanical and a
Data specified by suppliers.
chemical stability, no compaction/swelling and easy cleanability; their b
Data generated by using the weight difference method [29].
up-scaling is difficult, they are expensive and brittle [25–27].
Recyclability of solvents, separation of temperature sensitive so- without further purification. The non-woven porous support fabrics,
lutes, lowering exhausts, low energy consumption although are major viz., FO2470 and Viledon - H3160 were procured from Freudenberg
advantages of SRNF/OSN process, some important industrial solvents, (Germany); while Hollytex (grades: 3324 and 3265) were purchased
such as THF, DMF, (dimethylsulfoxide) DMSO, NMP and (di- from Ahlstrom (USA). The properties of these supports obtained from
chloromethane) DCM, are still critical for most currently available suppliers are given in Table 1.
SRNF membranes [25]. It seems that no commercial membrane can
withstand extreme pH conditions [21]. Realizing the potential of
ABPBI, viz., (i) solvent stability in harsh solvents such as DMF, DMAc, 2.2. Synthesis and characterization of ABPBI
NMP, DMSO, chlorinated solvents (ii) pH stability (except in con-
centrated inorganic acids) and (iii) thermo-chemical stability; proper- ABPBI was synthesized by the polycondensation of DABA in PPA
ties of the porous ABPBI membranes is worth investigating. It could be [2,19] as shown in Scheme 1 below.
anticipated that ABPBI based porous membranes would be applicable in A three-necked, 1-liter round-bottomed flask equipped with an
harsher environments of organic solvents, acids or bases; and more overhead stirrer, an N2 inlet and a CaCl2 guard tube was charged with
importantly, without a need of cross-linking, unlike in the conventional PPA (500 g) and heated up to 170 °C. A 25 g of DABA (0.1643 mol) was
PBI based diafiltration/ nanofiltration membranes. With this motive, added while stirring and the heating was continued for an hour until its
present work reports, for the first time, preparation of the supported dissolution. The temperature was raised to 200 °C and stirred further for
ABPBI based porous membranes by the conventional phase inversion 30 min. The reaction mixture was precipitated into the stirred water.
method. Investigations towards fundamental membrane properties The obtained polymer was crushed in a blender and thoroughly washed
(flux, rejection and pore size) and their stability towards various sol- with water until the filtrate was neutral to pH. The obtained polymer
vents, autoclave condition and acid/base environments were performed was suspended in 1% aqueous NaOH solution for 24 h while stirring.
as an initial study to unfold potentials of porous ABPBI based mem- This was necessary in order to extract the acid that is bound to the N-H
branes. The membrane preparation parameters (viz., porous support, group of ABPBI. The polymer was finally washed with water, until the
non-solvent and dope solution concentration) were varied to analyze filtrate was neutral to pH. The collected polymer was soaked in acetone
their effects on the membrane performance. A capability of glycerol for 15 h in order to extract water from the polymer matrix and then
treatment was investigated towards prevention of the pore collapse dried in a vacuum oven at 80 °C for a week. It was preserved in the
observed even under ambient conditions. desiccator until use.
The inherent viscosity of the synthesized ABPBI was determined by
using Ubbelhode viscometer (locally made) with 0.2 g dL−1 solution in
2. Experimental conc. H2SO4 at 35 °C. The FT-IR spectrum of ABPBI powder was per-
formed on Perkin Elmer Spectrum One spectrometer (United Kingdom)
2.1. Materials (Fig. S1). The thermogravimetric analysis (TGA) of the polymer was
performed by using Perkin Elmer TGA-7 (United Kingdom) under the
A 3,4-diaminobenzoic acid (DABA) and polyethylene glycols (PEG)
of different molecular weight were obtained from Aldrich Chemicals
(USA). The polyphosphoric acid (PPA, P2O5, ca. 84% as phosphorus
pentoxide) was procured from Alfa Aesar (USA). Other solvents and
reagents, viz.; H2SO4 (98%), methanesulfonic acid (MSA), HCl, gly-
cerol, N,N-dimethyl formamide (DMF), N,N-dimethyl acetamide
(DMAc), toluene, chloroform (CHCl3), hexane, isopropyl alcohol (IPA)
and isobutyl alcohol (IBA) were procured from S. D. Fine Chemicals
(India). All these chemicals were of AR/GR grade and were used Scheme 1. Synthesis of ABPBI.

744
H.R. Lohokare et al. Journal of Membrane Science 563 (2018) 743–751

N2 atmosphere at a heating rate of 10 °C min−1 and the thermogram is

Surface porosity

4.58 (0.0055)

0.38 (0.0004)

0.27 (0.0003)

1.05 (0.0012)
as given in Fig. S2. The wide-angle X-ray diffraction (WAXD) spectrum

9.87 (0.011)
of ABPBI in the powder form was recorded using Rigaku X-ray dif-
fractometer (D-max 2500, Japan) with Cu-K∝ radiation in the 2θ range

(%)
of 5–40 ° and was as given in Fig. S3.

–
a
2.3. Water holding capacity of the porous supports

By liquid-liquid displacement method

diameter (nm)
Mean pore
The samples of the porous non-woven support fabric (5 × 5 cm2)

13.45

12.80

14.85
were immersed initially in IPA, followed by water for 24 h. After taking

12.7

13.2
–
out from the water, the samples were wiped out with the tissue paper to

2.41 × 109 (2.80 ×

4.65 × 109 (5.39 ×

4.73 × 109 (5.50 ×

remove water from the surface of the sample. They were kept in a va-

3.37 × 1010 (3.92

6.23 × 1010 (7.47

Total no. of pores
per unit surfacea
cuum oven at 60 °C for a day and weighed again. The water holding
capacity was calculated by following Eq. (1) [29]. This analysis was
performed for three different samples of each type of support and the

× 107)

× 107)
data averaged as given in Table 1.

106)

106)

106)
–
% water holding capacity
wt. of the wet sample − wt. of the dry sample

Geometric standard
=
wt. of the wet sample (1)

deviation, σp
2.4. Membrane preparation

1.51

1.68

1.78

1.95

1.34
1.37
By solute rejection method
The dope solutions concentration of 4% and 6% (w/w) were pre-

Membrane porosity
pared by dissolving ABPBI in MSA at 70 °C, while stirring under a dry

diameter, µp (nm)
atmosphere for 48 h. The formed solution was cooled and then degassed
in order to remove entrapped gases. The polymer concentration higher

Mean pore
than 6 wt% led to a highly viscous solution and thus could not be

3.21

5.66

5.33

4.22

3.39
3.38
pursued further. The dope solution viscosity was measured using the
Brookfield digital viscometer (Model DV-I, USA) at 5.0 RPM at an
ambient. An unsupported, as well as the supported membranes, were
PEG6k rejection

prepared using ‘Sheen Automatic Film Applicator-1132′ (United

93.3 ± 3.5

37.9 ± 4.1

38.0 ± 5.2

51.9 ± 3.3

83.8 ± 4.3
90.8 ± 1.7
Kingdom). The unsupported membranes were cast on the glass plate. In
the cases of supported membrane preparation, the support fabric was
(%)

taped to a glass plate, on which casting was done. A knife movement

was set to 15 cm s−1 transverse speed and the gap was set to 250 µm.
After casting, the air exposure time was 14 s at ambient (temperature of
Water flux, Jw, at
1 bar (l m-2 h-1)

~ 27 °C and relative humidity of ~ 48%), before dipping it into the non-

solvent bath; either water or 0.5 N aq. NaOH kept at ambient tem-
15.1 ± 1.5
8.2 ± 1.2

1.5 ± 0.3

1.6 ± 0.1

2.7 ± 0.4

8.7 ± 0.3
perature. The supported membranes peeled off within minutes from the
glass plate. The membranes were kept in the nonsolvent bath for
~ 20 min for complete gelation. It was then removed and kept in an-
Values in the parenthesis are determined using 0.1 μm as the pore length.
Nonsolvent used

0.5 N aq. NaOH

other bath with plenty of water. The water was replaced time to time,
until the wash was neutral to pH and then stored at 4 °C until its use.
Water flux, PEG-rejection and pore size analysis of present membranes.

The identification of membranes is given in Table 2.

Water

Water

Water

Water

Water

2.5. Membrane characterizations

Support used

2.5.1. Water flux, rejection and pore size

FO2470

FO2470
FO2470
H3160

The water flux and rejection of PEGs of different molecular weights

3324

3265

(viz.; 1000, 2000, 4000, 6000, 8000, 10,000, 20,000 or 35,000 Da)
with a feed concentration of 0.1% were analyzed using a dead end,
Membrane code

stirred cell assembly possessing an active area of 14.5 cm2. The con-
centration of PEG in the feed and permeate was analyzed by following
‘Simps and Snape's method’ [30] using a double beam UV spectro-
M-1

M-2

M-3

M-4

M-5
M-6

photometer (Chemito, Spectrascan UV 2700, India). The rejection (R)

was calculated by using Eq. (2). The water flux analysis was done at
Polymer concentration in dope

least for six membrane samples prepared under identical conditions and
data was averaged (Table 2). The PEG rejection analysis was performed
using 5 membranes samples and the data is averaged, as given in
Table 2.
solution (wt. %)

⎡ Cp ⎤
R (%) = ⎢1− ⎜⎛ ⎟⎞ ⎥ × 100
Table 2

C (2)
⎣ ⎝ f ⎠⎦
a
6

where, Cp is a concentration of the permeate, while Cf is the feed

745
H.R. Lohokare et al. Journal of Membrane Science 563 (2018) 743–751

concentration. the case of water-immiscible solvents, a sequential solvent exchange

The mean pore diameter was determined by the solute rejection
method [31,32]. An average rejection of PEG was plotted versus the
diameter of that particular PEG. A straight line was obtained from a log-
normal plot [33]. The mean solute size (µs) was calculated as the PEG
diameter (ds) corresponding to 50% PEG rejection. The geometric
standard deviation about the mean diameter (σp) was determined from
the ratio of ds at 84.13% and 50%. The Stokes radii of PEGs were ob-
tained from their molecular weights by using Eq. (3) [33].
a = 16.73 × 10−10M 0.557
where ‘a’ is the Stokes radius and ‘M’ is the molecular weight of PEG.
For the pore size and its distribution analysis, water and water-sa-
turated isobutanol (surface tension, σ = 1.7 dyne cm−1) were used as
the solvent pair [34]. The membranes were first kept in the distilled
water for overnight. The water present in membrane pores was then
displaced by water-saturated isobutanol while applying pressure with a
sequential increment of 0.2 bar. The pore size was calculated using
Cantor's Eq. (4), while the pore-density was calculated using Hagen-
Poiseuille's Eq. (5) [34].
2σcosθ
rPi =
Pi
J P 8ηl
ni = ⎛Ji− i − 1 i ⎞
⎜ ⎟
protocol was followed. The membrane was initially dipped in IPA for
24 h and then in the respective solvent for 24 h. The solvent flux was
measured and the membrane was again immersed in IPA for 24 h in
order to replace the solvent by IPA. This was followed by dipping the
membrane in water and finally, the water flux was measured. The
change in the membrane thickness before and after the solvent treat-
ment was also measured. These analyses were repeated at least with
three different membrane samples prepared under identical conditions
(3) and the data was averaged as given in Table 3. The same membrane
samples were used for the PEG4k rejection analysis.
2.7. Glycerol treatment
An aqueous glycerol treatment was used for assessing its capability
to prevent the pore collapse of the present membrane (which occurs
even under air drying of the membrane at the ambient). The membrane
samples with a known water flux were dipped in the incremental
concentrations (wt% in water) of glycerol (10% for 8 h, 20% for 16 h
and finally in 50% glycerol solution for 24 h). After the glycerol treat-
ment, membranes were kept in an oven at 60 °C for 24 h. These dried
membranes were then dipped into a large quantity of water for 18 h, in
(4)
order to remove the glycerol. This was followed by repeated water-
wash and then measuring the water flux. Effects of repeated glycerol
treatment (pore filling by glycerol treatment, membrane drying and

⎝ Pi − 1 ⎠ πPi rp4i (5)

where, rpi is the radius of a pore, θ is the polymer-water contact angle,
ni is the number of pores per unit area, η is the viscosity of an effluent
fluid, l is the pore length (usually assumed to be 0.1 µm), Ji correspond
to the flux measured at the ith increment, where the applied pressure is
Pi. The surface porosity was also calculated using the pore length as
86.2 µm. The explanation for using this length is given in the Section
3.2.1.
2.5.2. Scanning Electron Microscopy (SEM)
The membrane specimen was prepared by fracturing at the liquid
nitrogen temperature and then dried in a vacuum oven at 40 °C for 24 h.
The cross-sections of as-cast membranes were analyzed by SEM, Leica,
Stereoscan, 440. The membrane was also treated with 10%, 20% and
50% aqueous glycerol (wt/wt), sequentially (for 8 h, 16 h and 24 h,
respectively) and then used for imaging. Its cross-section was analyzed
in a low vacuum SEM mode by Quanta 200 3D SEM using the large field
detector (LFD). The sample chamber pressure was 200 Pa.
2.6. Membrane stability in organic solvents, acid and base solution
The stability of present membranes was analyzed towards organic
solvents (DMF, DMAc, IPA, CHCl3, toluene, hexane), 25 N H2SO4, 2.5 N
NaOH and autoclave condition (15 psi and 121 °C for 20 min). Before
measuring the solvent flux, membranes with the known water flux were
dipped in a solvent (those are water miscible) for 24 h at an ambient. In
glycerol removal) on water flux of the membrane was analyzed.
3. Results and discussion
3.1. ABPBI synthesis and characterizations
The self-polycondensation of DABA (Scheme 1) in PPA, which acts
as a solvent as well as dehydrating agent [35] offered a viscous solu-
tion. The inherent viscosity of the obtained polymer was 1.72 dL g−1.
The formation of benzimidazole moiety was confirmed by the FT-IR and
the thermogravimetric analysis (TGA). The FT-IR (Fig. S1) exhibited the
characteristic IR bands at 1630 and 1556 cm−1. They can be ascribed to
the C˭C and C˭N stretching of the benzimidazole groups of ABPBI. The
typical broader N-H stretching vibration bands were seen at 3423 and
3189 cm−1 of ABPBI (Fig. S1) The TGA (Fig. S2) showed initial weight
loss up to ~355 °C, which could be attributed to the sluggish loss of
absorbed water form densely packed ABPBI matrix [19]. Such a loss of
water in cases of PBIs based on 3,3′-diaminobenzidine (DAB) and dif-
ferent dicarboxylic acids was observed up to ~ 190 °C and was attrib-
uted to the chain packing and the water holding capacity of that PBI
[13]. In the present case of ABPBI, the average d-spacing estimated
from the amorphous the WAXD spectra (Fig. S3) was 3.40 Å, which is
smaller than that of other PBIs [13]. ABPBI, synthesized using a single
monomer, has a higher N-H group density and thus possesses strong H-
bonding in its repeat unit [19]. These N-H groups would not only help
to hold the water strongly but also would slow down its diffusion

Table 3
Stability of the membrane M-6 towards organic solvents.
Solvent Used Dielectric Viscosity of Solubility parametera Surface tensiona Solvent flux % Reduction in % Change PEG4k rejection
constanta (ε) at solventa (ηsolvent) (δ) (cal cm−3)1/2 (σ) (dyne cm−1 at (Jsolvent) at 1 bar membrane thickness in Jw (%)
20 °C (cP) 20 °C) (l m−2 h−1)

DMF 36.7 0.82 12.1 35 14.7 8 13 66 ± 3.78

DMAc 37.8 0.92 11.0 34 6.6 15 10 45 ± 6.83
IPA 18.3 2 11.5 21.7 7.0 4 8 70 ± 2.08
Chloroform 4.8 0.57 9.3 27.16 13.6 33 − 59 85 ± 0.42
Toluene 2.38 0.59 8.9 28.5 12.7 27 − 59 55 ± 5.83
Hexaneb 1.9 0.31 6.9 18.4 – 20 – –
Autoclaving – – – – – 13 − 15 50 ± 4.12

a
From Reference [36].
b
Membrane delaminated from support.

746
H.R. Lohokare et al.
through the densely packed ABPBI matrix. This seems to be a reason for
a sluggish water loss observed at a higher temperature in the TGA
spectra. The degradation temperature of ABPBI was found to be 610 °C,
conveying its excellent thermal stability. It should be noted that the
char yield was 44.7%. Although the use of such a high temperature is
unlikely during usual separation applications of the ‘porous’ mem-
branes, this information could be useful towards applicability of ABPBI
based membranes at the high temperature.
Generally, solvents do plasticize the polymer matrix. Thus, the high
temperature stability would be an added advantage of ABPBI, in ad-
dition to its solvent and chemical stability. Although the glass transition
temperature would have been a better method of assessment of the
thermal stability, it could not be recorded even after repeated cycles of
heating and cooling. As in the present case, it is said that the Tg of
ABPBI could not be detected [37].
3.2. Membrane preparation and analysis
The ABPBI based supported UF membranes could be successfully
prepared by the phase inversion method. Effects of variations in the
dope solution concentration and the type of the porous nonwoven
support used were evaluated. Use of MSA as a solvent was preferred
over other possible acidic solvents. The viscosity of 4% and 6% (w/w)
dope solution was 3040 cPs and 17,255 cPs at 5.0 RPM, respectively.
The two different non-solvents, viz., water and 0.5 N NaOH were used
at ambient temperature. The aq. NaOH solution as a non-solvent was
anticipated to enhance the gelation process since the dope solution is
prepared an acidic solvent. Effects of these variants on the membrane
performance are discussed in the following sections.
3.2.1. Effect of properties of the porous support
The supported membranes were prepared using 6 wt% dope solu-
tion concentration on the polypropylene (PP, FO2470) and the polye-
ster (PET, H3160, 3324 and 3265) nonwoven fabrics possessing dif-
ferent properties (Table 1). The membranes prepared using these
supports are designated as M-1, M-2, M-3 and M-4; respectively. The
water flux at 1 bar transmembrane pressure for M-1 was 8.2 l m−2 h−1,
while for M-2, M-3 and M-4 it was 1.5, 1.6 and 2.7 l m−2 h−1; re-
spectively (Table 2). These values are considerably different from each
other. A correlation was observed between the water flux of the
membrane and water holding capacity of a non-woven fabric (Fig. 1). It
is known that properties of the support fabric are crucial in deciding
performance of the resulting membranes [38,39].
The membrane M-1, prepared using the PP support showed an
average rejection of PEG6k as 93.3%. On the other hand, the PET-sup-
ported membranes; viz., M-2, M-3 and M-4 offered considerably lower
rejection (Table 2) for the same PEG. It was thus decided to determine
the MWCO for all these membranes. The MWCO of the PET supported
membranes (i.e., M-2, M-3 and M-4) was higher (~ 35 kDa) than that of
the PP-supported membrane, M-1 (Fig. 2). Since M-1 possessed highest
Jw, an unanticipated observation of its lower MWCO than that of PET-
supported membranes (M-2, M-3 and M-4) suggests that the pore for-
mation with two different supports (PP and PET) is taking place dif-
ferently. The penetration of the dope solution was observed in cases of
membranes prepared with the PET supports. The dope solution of
ABPBI was prepared in highly polar methanesulfonic acid as the sol-
vent. The formed solution would be better compatible with ‘PET’ sup-
port (which is polar in nature due to ester groups present in it), rather
than that of nonpolar ‘PP support’. This could be a reason for the ob-
served difference in the penetration. The observed difference in the
water flux could be correlated to the penetration that would block the
support porosity randomly. On the other hand, the nonpolar nature of a
PP support may not allow penetration of the dope solution through its
pores.
In order to gain more insights on the above observation, information
on the pore size and their distribution is necessary. The value of the
747
Journal of Membrane Science 563 (2018) 743–751
mean pore diameter and σp was determined by the solute rejection
method by plotting the graph of PEG diameter versus its average re-
jection (Fig. 3), as given in Section 2.5.1. Although the surface porosity
can be determined by the solute rejection method, it was preferred to be
determined by the liquid-liquid displacement method. The mean pore
diameter of M-1 was lower than that of other membranes. Although
similar observation on the pore size was conveyed by the pore size
distribution (PSD) analysis (Fig. 4, Table 2), it shed light on the surface
pore density, expressed either as a total number of pores per unit sur-
face area or the surface porosity expressed in percent. These values for
M-1 were considerably higher than that of M-2, M-3 and M-4. This
analysis thus explains the observed higher flux for M-1, but its lower
MWCO, than that of other membranes. Values of the mean pore size
obtained by the solute rejection method and by the liquid-liquid dis-
placement method are different, which is a well-known observation
[32]. Among membranes prepared using the PET supports, M-4 made
using ‘3265′ as the support exhibited a maximum number of pores and
surface porosity.
The SEM analysis of M-1 is shown in Fig. 5. The dry membrane got
shrunk and delaminated from the support (Fig. 5a). This could be due to
the dewatering followed by the contraction of the membrane while
applying vacuum during SEM analysis. The air drying of this membrane
at ambient (27 °C) for ~ 15 min also showed such shrinkage. Thus, the
membrane was immersed in glycerol (concentration increased gradu-
ally up to 50%, as elaborated in Section 2.5.2). This treatment was
anticipated to preserve the membrane morphology during the SEM
analysis, as it did not show analyzable shrinkage after the membrane
drying. Although the SEM image (Fig. 5b) did not reveal a reduction in
the thickness of the top layer (as was seen in the case of Fig. 5a), it did
not convey any information on the membrane morphology. It is re-
ported that in the case of an asymmetric PBI based membrane, top
20 nm layer had no resolvable pores [12]. In the present case, the
membrane might have spongy nature but with a small pore size (as
ascertained by its lower MWCO of 6 kDa), distributed all over the width
of the membrane. To ascertain this, an unsupported membrane was
prepared while keeping all the casting parameters (polymer con-
centration, air exposure time and non-solvent) same as that of the
supported membrane. The SEM image of this membrane (Fig. 5c) also
did not reveal any morphological features and was similar to that of
Fig. 5b, an image of the glycerol treated supported membrane.
An absence of the distinct pores in the SEM of a membrane that
possessed the lower MWCO but, the higher water flux indicates the
spongy nature of the membrane. The spongy nature is not as observed
in the case of usual microfiltration membranes. The present membrane
exhibited small pore size as indicated by small MWCO. This nature is
Fig. 1. A correlation between the water flux of the membrane and the water
holding capacity of the support used.
H.R. Lohokare et al.
Fig. 2. PEG rejection (R) for membranes prepared using different supports and
6 wt% ABPBI in the dope solution.
Fig. 3. Effect of support on the PEG rejection and PEG diameter for membranes
prepared using 6 wt% dope solution concentration on different supports.
Fig. 4. Pore size distribution analysis of ABPBI membranes.
same all over the width of the membrane (Fig., 5). There is no figure-
like porosity as known for conventional UF membranes. The formation
of such porosity can be explained based on the acid-base interaction of
ABPBI with highly acidic solvent, MSA. The gelation is a crucial stage
governing the porosity during the membrane formation by the phase
748
Journal of Membrane Science 563 (2018) 743–751
inversion method. The acid-PBI gel membranes are known to be highly
stable and are used as the proton exchange membrane in fuel cell
[40,41]. One of these demonstrated membranes was prepared by the
gelation method [40], a widely used process for the membrane pre-
paration for PEMFC; which is anticipated to lead to the symmetric
membrane as an application requirement. In the present case, although
MSA may not form a stable gel as in the ABPBI-phosphoric acid case
used in PEMFC, it would be more stable than the gel formed for other
common polymers (PSF, PES, PAN, etc.) used for the membrane for-
mation by a phase inversion process. The longer-lived gel formation is
ascribable to the acid-base complex formation, which may not be
readily broken by the water used as a non-solvent (since water is not a
stronger base than PBI). The demixing/gelation of ABPBI though is
sluggish leading to a sponge-like pore structure; the pore size remained
small (MWCO of 6 kDa) due to the high interactions among the polymer
chains. The sponge-like structure of the top layer of PBI membrane is
known [42]. Conversely, in the cases of conventional polymers, as
stated above, the sluggish/delayed gelation is known to lead to a larger
pore size [43]. This difference can be attributed to the high acid-base
interactions between the basic nature of ABPBI and the acidic solvent,
MSA.
The thickness of a sponge-like structure as measured from the SEM
cross-section of the membrane was found to be 86.2 µm. Thus, for the
calculation of surface porosity, the skin layer thickness of 86.2 µm was
used. It provided an acceptable range of surface porosity for the
membrane as given in Table 2, which would offer high water flux de-
spite its low porosity. If the porosity calculations are done using the
pore length as 0.1 µm, as usually is the case for UF membranes prepared
from common polymers; it leads to a tiny surface porosity of 0.0055%.
Such a small porosity is unlikely to offer high water flux, as observed.
3.2.2. Effect of the non-solvent
The dope solution of basic ABPBI was prepared in an acidic solvent,
MSA. During the phase inversion process, an aqueous NaOH to be used
as a non-solvent, instead of water would serve following purposes. It is
possible that the phase inversion kinetics can be enhanced since the
solvent is acid. Moreover, the acid sorbed by an imidazole group of
ABPBI can be abstracted by NaOH, since later is a stronger base than the
−NH functionality of ABPBI. The membrane M-1 (water as the non-
solvent) and M-5 (0.5 N NaOH solution as the non-solvent) offered si-
milar water flux, i.e., 8.2 l m−2 h−1 and 8.7 l m−2 h−1, respectively
(Table 2). Both membranes, M-1 and M-5 showed > 90% rejection for
PEG8k, indicating their MWCO to be less than 8000 Da. In the case of M-
1, PEG6k rejection was 93.3% while M-5 exhibited the average rejection
of 83.8% for the same PEG (Fig. 6). The lower rejection for the later
membrane could be because of its slightly higher mean pore diameter
as determined by the solute rejection method than that of earlier
(Table 2). These membranes showed almost similar rejection for PEGs
of low molecular weight (Fig. 6).
This indicates that aq. NaOH is not serving as a better non-solvent
than the water. It could be because ABPBI itself possesses interchain
interactions via H-bonding within the polymer chains as well as with
the surrounding water and thus controls the gel formation by itself. This
gel formation, since is controlled by the nature of ABPBI, is rarely af-
fected by the non-solvent used for the phase inversion process. This may
be a possible reason for the observed similar flux and the rejection
properties of membranes prepared with water and aqueous NaOH as
the non-solvent.
3.2.3. Effect of the polymer concentration
In order to study effects of the dope solution concentration, mem-
branes prepared with 6% (M-1) and 4% (M-6) dope solution con-
centration and water as the non-solvent are considered. The membranes
M-1 and M-6 exhibited water flux of 8.2 and 15.1 l m−2 h−1 at 1 bar
transmembrane pressure, respectively. This variation is in agreement
with the effect of dope solution concentration, which is a well-
H.R. Lohokare et al. Journal of Membrane Science 563 (2018) 743–751

Fig. 5. SEM cross-section images of (a) as casted and dried M-1, (b) glycerol treated and dried M-1 and (c) unsupported glycerol treated and dried membrane.

Table 4
Change in the membrane thickness and water flux after treatment of acid and
base.
Membrane code 25 N H2SO4 treatment 2.5 N NaOH treatment

% Reduction in % Change % Reduction in % Change

membrane in Jw membrane in Jw
thickness thickness

M− 1 22.9 − 23.7 12.7 − 7.3

M− 6 26.0 − 36.1 11.5 − 1.5

Fig. 6. PEG rejection (R) for ABPBI membranes prepared by varying the dope
solution concentration (filled symbol: water as nonsolvent, unfilled symbol:
0.5 N aqueous NaOH as the nonsolvent).

Fig. 8. Change in Jw after repeated glycerol treatment (aq. 50% solution) for M-
6 membrane.

membrane with particular pore size is majorly controlled by the

Fig. 7. PEG rejection versus PEG diameter for M-1, M-5 and M-6.
investigated aspect for other polymers [44,45]. Both these membranes
exhibited almost similar MWCO (6 kDa) (Fig. 6) and the mean pore
diameter estimated by the solute rejection method. If we look at the
pore size distribution obtained by the liquid-liquid displacement
method, it could be seen that M-6 has a similar mean pore diameter as
that of M-1, This observation substantiates an estimation of the pore
size by the solute rejection method (Fig. 7).
The surface porosity of M-6 is almost double than that of M-1. This
explains high water flux observed for M-6 than that of M-1; although
their pore sizes are similar. It also supports that formation of gel
polymer property than features of the non-solvent or the dope solution
concentration.
The concentrations of the dope solution used in the present study
are 4% and 6%, which are considerably lower than the dope solution
concentrations used for the preparation of UF membrane using con-
ventional polymers such as PSF, PES, PAN, etc. [44–46]. In the case of
microfiltration (MF) membrane of Nylon or PVDF, the dope solution
concentrations used was 10–32% [47,48]. By analogy to the reported
literature, the present ABPBI membranes prepared with a low dope
solution concentration should have possessed porosity as that of the MF
membranes. Conversely, current membranes possessed a lower water
flux, high rejection (MWCO of 6 kDa), no resolvable pores in the SEM
and pore size of hardly 3.2–5.6 nm. These facts further support that in
the case of ABPBI based membranes, the gel formation with smaller
pore size is originated due to the strong H-bonding interactions present
among the polymer chains and with the surrounding water. A variation
in the polymer concentration used for making the dope solution al-
though offered different water flux, it seems to have a small effect on

749
H.R. Lohokare et al.
the ‘surface pore size.’
3.3. Stability of the membrane in organic solvents
The membrane prepared using the PP support, M-6, was used to
evaluate its stability towards organic solvents of commercial sig-
nificance and an autoclave condition. The PP support was chosen as it
has more chemical resistance than the PET [49]. The solvent flux was
measured after dipping the membrane into the respective solvent for
24 h. In cases of water-immiscible solvents, viz.; chloroform, toluene,
and hexane; membranes were submerged initially in IPA (24 h) and
then in the respective solvent for 24 h. The properties of the solvent and
their flux are given in Table 3 [36]. In the case of hexane, the mem-
brane got delaminated from the support. In the water/IPA environment,
strong H-bonding between the solvent and ABPBI would exist. After
replacing IPA from the membrane pores by a non-polar solvent, hexane;
the H-bonding between the polymer and the solvent present in the
pores is eliminated. At the same time, it would still be present within
the polymer chains. This phenomenon might be responsible for the
membrane shrinkage, followed by its delamination from the support.
There was no sound correlation between the solvent flux and its
properties (dielectric constant, viscosity, solubility parameter and sur-
face tension). Machado et al. observed that there is a relatively small
effect of the dielectric constant on the solvent flux of the membrane
[50]. In general, the solvent fluxes were lower in comparison to the
water flux. Among solvents used, the flux of DMF, chloroform and to-
luene were similar, while that of DMAc and IPA were lower (Table 3).
The lower flux for DMAc and IPA could be due to their higher viscosity.
It is known that the solvent with higher viscosity exhibited lower flux
through the polyimide-based UF membranes [51]. After the solvent flux
measurement, three membrane samples were dipped in water for 24 h
(via IPA treatment, in case of water-miscible solvents) and the water
flux was measured again. This was performed to analyze the change in
the membrane properties by a solvent-treatment. The PEG4k was se-
lected for the rejection analysis, as the untreated membrane showed
39% rejection. Thus, variation in the rejection in either direction can be
evaluated easily (rather than with PEG of a higher MW). The membrane
treated with DMF, DMAc and IPA exhibited marginal changes in the
water flux (8–13%) and the membrane thickness (4–15%), while it
showed 59% reduction in the water flux and ~ 30% reduction in the
membrane thickness after chloroform and toluene treatment (Table 3).
The increase in the rejection of PEG4k as compared to the membrane
without the solvent treatment confirmed a reduction in the pore size.
The chloroform treated membrane showed the highest rejection, while
DMAc treated membrane showed a smaller change in the rejection, in
comparison to the initial one.
In biotechnology applications, autoclaving of a membrane possesses
a niche role. Given the solvent stability of M-6, it was further subjected
to an autoclave condition. The water flux and rejection analysis were
performed after removing membrane samples from the autoclave. It
was seen that there is hardly 15% reduction in the water flux, asso-
ciated with an increase in the rejection of PEG4k from 39% to 50%
(Table 3). This indicated a small reduction in the pore size after a high
temperature autoclave condition of 121 °C. This showed the additional
advantage of this membrane towards thermal stability in the presence
of pressurized steam. In cases of membrane application in medicine,
pharmaceuticals and food industry; there is a high demand for safely
sterilizable membranes [52].
3.4. Membrane stability in concentrated acid and base
Based on the visual observation after 24 h exposure, it was found
that M-1 and M-6 were stable in 25 N H2SO4 solution (while in 30 N
H2SO4, the top layer showed a partial dissolution). Both the mem-
branes, M-1 and M-6, exhibited a marginal decrease in the water flux
after treatments with 25 N H2SO4 and 2.5 N NaOH, as shown in Table 4.
750
Journal of Membrane Science 563 (2018) 743–751
All these membranes after the solvent treatment showed a marginal
reduction in the thickness as well. On the other hand, the solvent re-
sistant membranes based on poly-(p-phenylenetere-phthalamide)
showed limited stability towards 2 N H2SO4 and NaOH at pH 12 [53].
3.5. Effect of drying and glycerol treatment
For analyzing an effect of the membrane-drying on its porosity, M-6
was dried at 60 °C in an oven for 24 h. It exhibited 98% and 34% de-
crease in the water flux and thickness, respectively, possibly due to the
shrinkage. This membrane after immersing in the water again for a
prolonged time (24 h) did not regain its thickness or flux. Such irre-
versible shrinkage after drying in the case of poly-(p-phenylenetere-
phthalamide) based membrane is known [53]. In order to avoid the
pore collapse, membranes were dipped in 10% glycerol and then dried
at 60 °C. The decrease in water flux and membrane thickness was up to
93% and 22%, respectively. This observation indicated that just 10%
glycerol treatment is not sufficient to avoid the pore collapse. Thus,
membranes were treated with 10%, 20% and 50% glycerol solution
(wt/wt), sequentially, and then dried at 60 °C. This membrane showed
21% decrease in the water flux with a small decrease in the thickness
(3.5%). In order to study an effect of repeated glycerol treatment, the
glycerol present in the pores was replaced by the water and flux was
measured again. Such replacement of glycerol by water in the pores was
repeated four times. Reduction in the flux was observed only after the
first treatment, and after that, it remained almost constant (Fig. 8).
Thus, a significant pore collapse of ABPBI membrane can be avoided by
50% glycerol treatment.
4. Conclusions
The choice of nonwoven support used for the preparation of ABPBI
based supported UF membrane was found to be crucial. The membrane
prepared using a PP-support offered higher flux as well as rejection.
This was due to lower pore size and a larger pore density than that of
membranes based on PET support. The PP-based membranes (M-1 and
M-6) showed 6 kDa MWCO, while it was 35 kDa for the PET-based
membranes (M-2, M-3 and M-4). Variation in the non-solvent (water/
0.5 N NaOH) had a negligible effect on properties of the resulting
membrane; attributable to the strong gel formation by ABPBI itself. It
was further supported by the results obtained by variation in the dope
solution concentration (4 and 6 wt%). Although these two membranes
showed an anticipated variation in the water flux, their similar rejec-
tion performance indicated that they possessed almost similar pore size;
an outcome of a typical gel formation by ABPBI polymer. Further
evaluations of the membrane prepared on the PP-support showed ex-
cellent stability towards organic solvents, autoclave condition, con-
centrated acid (25 N H2SO4) and base (2.5 N NaOH). Thus, these
membranes can be used in non-aqueous and harsh environments. The
dried membrane showed pore collapse, which was shown to be avoid-
able by aqueous glycerol (50 wt%) treatment.
Acknowledgments
We acknowledge the funding from the Council of Scientific and
Industrial Research, Govt. of India (31/11(220)/2013-EMRI) (research
fellowships to H.R. Lohokare and H. D. Chaudhari), and Science &
Engineering Research Board, Dept. of Science and Technology, Govt. of
India (Grant no. SB/S3/CE/096/2013).
Appendix A. Supplementary material
Supplementary data associated with this article can be found in the
online version at http://dx.doi.org/10.1016/j.memsci.2018.06.052.
H.R. Lohokare et al.
References
[1] S. Bose, T. Kuila, T.X.H. Nguyen, N.H. Kim, K. Lau, J.H. Lee, Polymer membranes
for high temperature proton exchange membrane fuel cell: recent advances and
challenges, Prog. Polym. Sci. 36 (2011) 813–843.
[2] J.A. Asensio, P. Gómez-Romero, Recent developments on proton conducting Poly
(2,5-benzimidazole) (ABPBI) membranes for high temperature polymer electrolyte
membrane fuel cells, Fuel Cells 5 (2005) 336–343.
[3] H. Hou, G. Sun, R. He, Z. Wu, B. Sun, Alkali doped polybenzimidazole membrane
for high performance alkaline direct ethanol fuel cell, J. Power Sources 182 (2008)
95–99.
[4] H. Luo, G. Vaivars, B. Agboola, S. Mu, M. Mathe, Anion exchange membrane based
on alkali doped poly(2,5-benzimidazole) for fuel cell, Solid State Ion. 208 (2012)
52–55.
[5] D. Li, D. Shi, Z. Yuan, K. Feng, H. Zhang, X. Li, A low cost shutdown sandwich-like
composite membrane with superior thermo-stability for lithium-ion battery, J.
Membr. Sci. 542 (2017) 1–7.
[6] D.Y. Xing, S.Y. Chan, T.S. Chung, Molecular interactions between poly-
benzimidazole and [EMIM]OAc, and derived ultrafiltration membranes for protein
separation, Green. Chem. 14 (2012) 1405.
[7] D. Bhagat, B. Mule, N. Mandlekar, K. Pandare, U. Kharul, PBI-BuI and PAN-PSSALi
based UF membranes: effects of solute and membrane surface interactions on re-
jection and flux, Desalination 333 (2014) 45–51.
[8] D.Y. Xing, S.Y. Chan, T.-S. Chung, The ionic liquid [EMIM]OAc as a solvent to
fabricate stable polybenzimidazole membranes for organic solvent nanofiltration,
Green. Chem. 16 (2014) 1383.
[9] I.B. Valtcheva, S.C. Kumbharkar, J.F. Kim, Y. Bhole, A.G. Livingston, Beyond
polyimide: crosslinked polybenzimidazole membranes for organic solvent nanofil-
tration (OSN) in harsh environments, J. Membr. Sci. 457 (2014) 62–72.
[10] D. Chen, S. Yu, M. Yang, D. Li, X. Li, Solvent resistant nanofiltration membranes
based on crosslinked polybenzimidazole, RSC Adv. 6 (2016) 16925–16932.
[11] M.F. Flanagan, I.C. Escobar, Novel charged and hydrophilized polybenzimidazole
(PBI) membranes for forward osmosis, J. Membr. Sci. 434 (2013) 85–92.
[12] L.C. Sawyer, R.S. Jones, Observations on the structure of first generation poly-
benzimidazole reverse osmosis membranes, J. Membr. Sci. 20 (1984) 147–166.
[13] S.C. Kumbharkar, P.B. Karadkar, U.K. Kharul, Enhancement of gas permeation
properties of polybenzimidazoles by systematic structure architecture, J. Membr.
Sci. 286 (2006) 161–169.
[14] X. Li, R.P. Singh, K.W. Dudeck, K.A. Berchtold, B.C. Benicewicz, Influence of
polybenzimidazole main chain structure on H2/CO2 separation at elevated tem-
peratures, J. Membr. Sci. 461 (2014) 59–68.
[15] R.P. Singh, X. Li, K.W. Dudeck, B.C. Benicewicz, K.A. Berchtold, Polybenzimidazole
based random copolymers containing hexafluoroisopropylidene functional groups
for gas separations at elevated temperatures, Polymer 119 (2017) 134–141.
[16] Y.J. Chendake, U.K. Kharul, Desalination and Water Treatment Transport of in-
organic acids through polybenzimidazole (PBI) based membranes by chemo- dia-
lysis membranes by chemo-dialysis, Desalin. Water Treat. 38 (2012) 96–103.
[17] Y.J. Chendake, U.K. Kharul, Transport of organic acids through polybenzimidazole
based membranes by “Chemodialysis”, J. Membr. Sci. 451 (2014) 243–251.
[18] Q. Li, J.O. Jensen, R.F. Savinell, N.J. Bjerrum, High temperature proton exchange
membranes based on polybenzimidazoles for fuel cells, Prog. Polym. Sci. 34 (2009)
449–477.
[19] S.C. Kumbharkar, U.K. Kharul, New N-substituted ABPBI: synthesis and evaluation
of gas permeation properties, J. Membr. Sci. 360 (2010) 418–425.
[20] J.A. Asensio, S. Borrós, P. Gómez-Romero, Polymer electrolyte fuel cells based on
phosphoric acid-impregnated Poly(2,5-benzimidazole) membranes, J. Electrochem.
Soc. 151 (2004) A304.
[21] A.W. Mohammad, Y.H. Teowa, W.L. Ang, Y.T. Chung, D.L. Oatley-Radcliffe,
N. Hilal, Nanofiltration membranes review: recent advances and future prospects,
Desalination 356 (2015) 226–254.
[22] A.A. Tashvigha, L. Luoa, T.S. Chunga, M. Weberb, C. Maletzko, Performance en-
hancement in organic solvent nanofiltration by double crosslinking technique using
sulfonated polyphenylsulfone (sPPSU) and polybenzimidazole (PBI), J. Membr. Sci.
551 (2018) 204–213.
[23] N. Daems, S. Milis, R. Verbeke, A. Szymczyk, P.P. Pescarmona, I.F.J. Vankelecom,
High-performance membranes with full pH stability, RSC Adv. 8 (2018)
8813–8827.
[24] L.S. White, Development of large-scale applications in organic solvent nanofiltra-
tion and pervaporation for chemical and refining processes, J. Membr. Sci. 286
(2006) 26–35.
[25] P. Vandezande, L.E.M. Gevers, I.F.J. Vankelecom, Solvent resistant nanofiltration:
separating on a molecular level, Chem. Soc. Rev. 37 (2008) 365–405.
751
Journal of Membrane Science 563 (2018) 743–751
[26] G. Szekely, M.F. Jimenez-Solomon, P. Marchetti, J.F. Kim, A.G. Livingston,
Sustainability assessment of organic solvent nanofiltration: from fabrication to
application, Green. Chem. 16 (2014) 4440–4473.
[27] P. Marchetti, M.F.J. Solomon, G. Szekely, A.G. Livingston, Molecular separation
with organic solvent nanofiltration: a critical review, Chem. Rev. 114 (2014)
10735–10806.
[28] X. Li, Y. Liu, J. Wang, J. Gascon, J. Li, B. Van der Bruggen, Metal–organic frame-
worksbased membranes for liquid separation, Chem. Soc. Rev. 46 (2017)
7124–7144.
[29] M. Tamura, T. Uragami, M. Sugihara, Studies on syntheses and permeabilities of
special polymer membranes: 30. ultrafiltration and dialysis characteristics of cel-
lulose nitrate-poly(vinyl pyrrolidone) polymer blend membranes, Polymer 22
(1981) 829–835.
[30] G.E.C. Sims, T.J. Snape, A. Method, for the estimation of polyethylene glycol in
plasma protein fractions, J. Anal. Biochem. 107 (1980) 60–63.
[31] P. Aimar, M. Meireles, V. Sanchez, A contribution to the translation of retention
curves into pore size distributions for sieving membranes, J. Membr. Sci. 54 (1990)
321–338.
[32] K.J. Kim, A.G. Fane, R. Ben Aim, M.G. Liu, G. Jonsson, I.C. Tessaro, A.P. Broek,
D. Bargeman, A comparative study of techniques used for previous membrane
characterization: pore characterization, J. Membr. Sci. 87 (1994) 35–46.
[33] S. Singh, K.C. Khulbe, T. Matsuura, P. Ramamurthy, Membrane characterization by
solute transport and atomic force microscopy, J. Membr. Sci. 142 (1998) 111–127.
[34] G. Capannelli, F. Vigo, S. Munari, Ultrafiltration membranes - characterization
methods, J. Membr. Sci. 15 (1983) 289–313.
[35] F.W. Harris, B.H. Ahn, S.Z.D. Cheng, Synthesis of polybenzimidazoles via aromatic
nucleophilic substitution reactions of self-polymerizable (A-B) monomers, Polymer
34 (1993) 3083–3095.
[36] I. Smallwood, Handbook of Organic Solvent Properties, John Wiley & Sons Inc, New
York, NY, 1996.
[37] F.J. Nores-Pondal, M.P. Buera, H.R. Corti, Thermal properties of phosphoric acid-
doped polybenzimidazole membranes in water and methanol-water mixtures, J.
Power Sources 195 (2010) 6389–6397.
[38] S. Munari, A. Bottino, G. Camera-Roda, G. Capannelli, Preparation of ultrafiltration
membranes, Desalination 77 (1990) 85–100.
[39] H.R. Lohokare, Y.S. Bhole, U.K. Kharul, Effect of support material on ultrafiltration
membrane performance, J. Appl. Polym. Sci. 99 (2006) 3389–3395.
[40] A. Sannigrahi, S. Ghosh, S. Maity, T. Jana, Polybenzimidazole gel membrane for the
use in fuel cell, Polymer 52 (2011) 4319–4330.
[41] X. Wu, K. Scott, A. H2SO4, loaded polybenzimidazole (PBI) membrane for high
temperature PEMFC, Fuel Cells 12 (2012) 583–588.
[42] K.Y. Wang, T.S. Chung, J.J. Qinb, Polybenzimidazole (PBI) nanofiltration hollow
fiber membranes applied in forward osmosis process, J. Membr. Sci. 300 (2007)
6–12.
[43] M.H.V. Mulder, Basic Principles of Membrane Technology, Kluwer Academic
Publishers, Dordrecht, 1996.
[44] D. Paul, H. Kamusewitz, H.G. Hicke, H. Buschatz, Separation properties and surface
morphology of polyacrylonitrile membranes, Acta Polym. 43 (1992) 353–355.
[45] K. Nouzaki, M. Nagata, J. Arai, Y. Idemoto, N. Koura, H. Yanagishita, H. Negishi,
D. Kitamoto, T. Ikegami, K. Haraya, Preparation of polyacrylonitrile ultrafiltration
membranes for wastewater treatment, Desalination 144 (2002) 53–59.
[46] C. Barth, M.C. Gonçalves, A.T.N. Pires, J. Roeder, B.A. Wolf, Asymmetric poly-
sulfone and polyethersulfone membranes: effects of thermodynamic conditions
during formation on their performance, J. Membr. Sci. 169 (2000) 287–299.
[47] D. Wang, K. Li, W.K. Teo, Preparation and characterization of polyvinylidene
fluoride (PVDF) hollow fiber membranes, J. Membr. Sci. 163 (1999) 211–220.
[48] A.M.W. Bulte, B. Folkers, M.H.V. Mulder, C.A. Smolders, Membranes of
Semicrystalline aliphatic polyamide nylon 4, 6: formation by diffusion-induced
phase separation, J. Appl. Polym. Sci. 15 (1993) 13–26.
[49] R. Krstic Netravali, J. Crouse, L. Richmond, Geosynthetic Soil Reinforcement
Testing Procedure by S-C Cheng, American Society for Testing and Materials Special
Technical Publication, Philadelphia, 1993, p. 207.
[50] D.R. MacHado, D. Hasson, R. Semiat, Effect of solvent properties on permeate flow
through nanofiltration membranes. Part I: investigation of parameters affecting
solvent flux, J. Membr. Sci. 163 (1999) 93–102.
[51] A. Iwama, Y. Kazuse, New polyimide ultrafiltration membranes for organic use, J.
Membr. Sci. 11 (1982) 297–309.
[52] H.G. Hicke, I. Lehmann, G. Malsch, M. Ulbricht, M. Becker, Preparation and char-
acterization of a novel solvent-resistant/rand autoclavable polymer membrane, J.
Membr. Sci. 198 (2002) 187–196.
[53] P. Zschocke, H. Strathmann, Solvent resistant membranes from poly-(p-phenylene-
terephthalamide), Desalination 34 (1980) 69–75.