Introduction
Thin film composite (TFC) membranes are essential for various
membrane-based applications, such as seawater desalination, ultrapure
water production and advanced wastewater treatment1,2,
which play an important role in the sustainable development of economy
and society. The superior performance of TFC membranes could be
attributed to the ultrathin selective layer that is mainly formed via
interfacial polymerization (IP). Numerous endeavors have been devoted to
optimizing the IP process to improve the permeability and selectivity of
a TFC membrane3,4. Despite the increase in studies
focusing on fabricating TFC membranes3,5, limited
knowledge is available about the mechanisms accounting for the formation
of an IP film.
Most prior studies investigated the selective layer (i.e., the polyamide
layer) and support layer (i.e., the substrate) of a TFC membrane
separately6,7 and the focus was usually on the
optimization of the selective layer. There are a limited number of
studies that explored the effects of the substrate on the formation of a
polyamide (PA) layer. It has been recognized that both the pore-size
distribution and surface hydrophilicity of the substrate are two of the
most important factors affecting the formation of a PA
layer8. Singh et al .9explored the effects of substrate structures on the PA-layer formation
by employing two polysulfone substrates with varied pore-size
distributions. It was revealed that increasing the pore size could lead
to a relatively thin PA layer with a higher permeability. Similar
observations were obtained by Ghosh and Hoek in a later
study8. In addition to varying the pore size, the
effects of the morphology were highlighted in the work by ElSherbinyet al .10, where a novel polyethersulfone
substrate with isotropic pores (~100 nm) was developed
by combining the vapor- and nonsolvent-induced phase separation; this
novel substrate yielded a TFC membrane with a higher water permeability
compared with those fabricated using commercial substrates, while
maintaining the salt rejection. In comparison with the studies on the
effects of substrate structures, there are substantially different
opinions in the open literature interpreting the role of the
hydrophilicity of the substrate in an IP process.
The preference of a hydrophobic substrate for IP was supported by prior
studies in different ways. For example, Ghosh and
Hoek8 proposed a conceptual model suggesting that a
relatively hydrophobic substrate would result in a highly permeable TFC
membrane. Specifically, it was hypothesized that enhancing the
hydrophilicity of the substrate could impede the diffusion of
m-phenylenediamine (MPD) in the aqueous phase while promoting the
diffusion of trimesoyl chloride (TMC) in the organic phase into the
substrate pores to form a thicker PA layer during IP. In contrast, the
work by Alsvik et al .11 highlighted the
delamination phenomenon of a PA layer formed on a hydrophilic substrate
of hydrolyzed cellulose acetate (CA). In order to improve the adhesion
between the PA layer and the hydrophilic substrate, they proposed to
precoat the substrate with a TMC-containing solution taking advantage of
the hydroxyl groups on the CA network; this TMC-precoated substrate
would allow the covalent bonding with the PA layer and the hydrophilic
substrate. Delamination of the active layer was also observed by Zhanget al. 12 when using a series of polysulfone
(PSf) substrates for fabricating the TFC membranes via IP, though the
underlying mechanism remained unclear.
On the contrary, there are prior studies indicating that increasing the
hydrophilicity of the substrate could not only enable the
membrane-supported IP but also improve the performance of the resulting
TFC membrane. For example, Kim et al. 13demonstrated that a plasma treatment would allow polypropylene (PP) and
PSf membranes to be used as the substrate for IP owing to the increase
in the hydrophilicity. A similar method was adopted by Kim et
al. 14 to successfully fabricate a TFC membrane via IP
using a hydrophobic polyvinylidene fluoride (PVDF) substrate. In
addition to the plasma treatment, it was reported that the
hydrophilicity of the substrate could be enhanced by surface coating and
incorporating hydrophilic additives in the dope solutions so as to favor
the formation of a PA layer via membrane-supported IP and improve the
TFC membrane performance. For instance, Jimenez-Solomon et
al .15 successfully fabricated TFC membranes via IP
using crosslinked polyimide and poly(ether ether ketone) substrates,
whose hydrophilicity was increased by simply dipping the substrates into
a solution of polyethylene glycol before the IP; Ding et
al .16 examined the effects of substrate
hydrophilicity on the performance of the TFC membranes fabricated via IP
by varying the amount of polyvinylpyrrolidone mixed into the PSf dope
solutions, which were used to prepare the substrates via phase
inversion.
Despite the discrepancy between the studies on the role of substrate
hydrophilicity in the formation of a PA layer via membrane-supported IP,
it has been widely accepted that a hydrophilic substrate would
substantially improve the performance of a TFC membrane employed in an
osmotically driven process owing to the mitigation of the internal
concentration polymerization (ICP) and fouling17.
Although the potential of applying osmotically driven membrane processes
to desalination remains controversial18,19, great
promise has been shown in their applications for pharmaceutical
production20, food and beverage
industry21, emergent relief22,
wastewater treatment23, and energy
generation24. Therefore, it is of great significance
to develop TFC membranes with a hydrophilic substrate for various
osmotically driven membrane processes.
A limited number of studies were reported to develop an effective method
for synthesizing TFC membranes with a hydrophilic substrate. In addition
to the method based on the TMC precoating by Alsvik et
al .11, Choi et al .25successfully fabricated a nanoscale-controlled PA layer with a typical
thickness ~15 nm by depositing the complementary
monomers (i.e., MPD and TMC) alternatingly on a polyelectrolyte-modified
hydrophilic polyacrylonitrile (PAN) substrate; Park et
al .26 employed toluene/xylene as the solvent for the
organic phase and synthesized TFC membranes with higher performance for
reverse osmosis (RO) using a hydrophilic PAN membrane as the substrate;
the method developed by Park et al .26 was
further optimized by Kwon et al .27 and the
synthesized TFC membranes were examined in a forward osmosis (FO)
process. However, most of these studies ignored the effects of substrate
hydrophilicity on the formation of the PA layer.
Instead of targeting on the fabrication of TFC membrane with superior
performance, the focus of the current study was on the role of substrate
hydrophilicity in a membrane-supported IP process. In line with this
idea, the formation of a PA layer via membrane-supported IP was
investigated when varying the hydrophilicity of the substrate.
Particularly, it was proposed to tune the surface hydrophilicity of the
substrate via the deposition of various polyelectrolytes, which have
been extensively explored for the Layer-by-Layer (LbL)
assembly28,29. The relative importance of the enhanced
surface hydrophobicity in the IP-film formation was analyzed in terms of
a series of characterization experiments. It was confirmed that the
polyelectrolyte deposition could increase the surface hydrophobicity of
a hydrolyzed PAN substrate and thereby favor the formation of an IP
layer with better integrity (i.e., minimized delamination). The
performance of the fabricated TFC membranes with the hydrolyzed PAN
substrate was evaluated in an FO process to verify the mitigation of the
ICP by the hydrophilic substrate.
Experimental Procedures
Interfacial polymerization on substrates with varied
hydrophilicity
The substrates employed for the membrane-supported IP in this study were
all fabricated using polyacrylonitrile (PAN) via phase inversion, whose
various applications were reported in our previous
publications30-32. Specifically, the dope solution was
prepared by dissolving polymer particles of PAN (weight-averaged
molecular weight Mw ~150,000,
Sigma-Aldrich) in N,N-dimethylformamide (DMF, ≥99.8 %, Alfa Aesar) with
a predetermined amount of lithium chloride (LiCl, ≥99%, Sigma-Aldrich)
as the pore former31,32. The mass ratio of PAN, LiCl,
and DMF in the dope solution was 18:2:80. The dope solution was mixed
using a magnetic stirrer (IKA, C-MAG HS7 Control, Germany) at a
temperature of ~60ºC. The film-coated glass plate
(casting height ~150 μm) was immediately immersed into a
coagulation bath of tap water at room temperature
(~20ºC) to initiate the nonsolvent-induced phase
inversion. The unmodified PAN films were denoted as PAN-O and used as
the substrates in the control tests.
The hydrophilicity of the PAN-O films was first changed by implementing
an alkaline treatment, whereby the nitrile groups would be converted
into carboxyl groups33. Specifically, the PAN-O films
were soaked into a 1.5 M NaOH (≥96%, Fuchen Chemical, China) solution
at 45ºC for 90 min31,32. The resulting PAN substrates
were designated as PAN-A. In order to isolate the effects resulting from
the pore-size change34, a comparative study was
carried out by applying a heat treatment to the PAN-O films.
Specifically, the PAN-O substrates were immersed in a bath of Milli-Q
water while the temperature and duration were kept at the same values
for the alkaline treatment (i.e., 45ºC for 90 min). The resulting PAN
substrates were denoted as PAN-H.
The membrane-supported IP was performed by impregnating the PAN
substrates with an aqueous solution containing 2.0 wt.%m -phenylenediamine (MPD, ≥99%, Sigma-Aldrich) and exposing the
impregnated PAN substrates to an organic solution containing 0.1 w/v.%
1,3,5-benzenetricarbonyl trichloride (TMC, ≥99%, Sigma-Aldrich). In
particular, the solvent of the TMC-containing organic solution was
hexane (≥98%, Shanghai Aladdin Biochemical Technology Co. Ltd., China);
the duration for soaking the PAN substrates in the MPD-containing
aqueous solution was approximately 5 min and the excessive MPD solution
was removed using an air knife of compressed
N22; the duration for exposing the
impregnated PAN substrates to the TMC-containing organic solution was
approximately 1 min and the residual TMC was removed by a rinse of pure
hexane.
Tuning the hydrophilicity of substrates via polyelectrolyte
deposition
The PAN-A substrates enabled the surface modification via the deposition
of polycations owing to the negatively charged surface after the
alkaline treatment (i.e., the formation of carboxyl groups). It was
expected that the deposition of various polycations would significantly
alter the hydrophilicity of the PAN-A substrates28,35.
Particularly, three polycations were employed to modify the PAN-A
substrates, including poly(allylamine hydrochloride) (PAH,
Mw ~120,000 to 200,000, ≥99%, Alfa
Aesar), polyethyleneimine (PEI, Mw~750,000, 50%,
Sigma-Aldrich), and poly(dimethyl diallyl ammonium chloride) (PDADMAC,
Mw ~200,000 to 350,000, 20%,
Sigma-Aldrich). The deposition of the polycations on the PAN-A
substrates was implemented in a way similar to that employed by the
Layer-by-Layer assembly using polyelectrolytes31,32.
Specifically, 1 g of the polycation was dissolved in 1 L Milli-Q water
containing 0.5 M sodium chloride (NaCl, ≥99.5%, Xilong Scientific,
China); the existence of NaCl would favor the assembly of the
polyelectrolyte on the surface of the substrate36. The
PAN-A substrate was placed in a stainless steel plate with the dense
side facing up. A predetermined amount of the polycation solution was
gently poured into the plates to contact with the surface of the PAN-A
substrate. The duration of exposing the PAN-A substrate to the
polycation solution was 20 min and the deposition was followed by a 10
min rinse of Milli-Q water to remove the excess polyelectrolyte on the
substrate. The polyelectrolyte-modified substrates were denominated as
XXX-m, where XXX indicates the specific polyelectrolyte used for the
substrate modification. All the polyelectrolyte-modified substrates are
summarized in Table S-1.
Membrane characterization
The hydrophilicity of the PAN substrates was characterized in terms of
the contact angle measurement. All the PAN substrates were dried using a
freeze dryer (Ningbo Scientz Biotechnology Co. Ltd., Scientz-12N, China)
before the contact angle measurement and the sessile drop method was
employed using a Drop Shape Analyzer (KRÜSS GmbH, DSA25, Germany). The
measurement was repeated (at least 9 times using three independently
prepared membrane samples) to obtain an averaged value for a more
accurate analysis.
It has been revealed in prior studies37 that surface
morphology could play an important role in the surface wetting.
Therefore, both scanning electron microscopy (SEM) and atomic force
microscopy (AFM) were exploited to characterize the surface of the PAN
substrates. In particular, the SEM characterization was performed by
precoating the membrane samples with platinum in a vacuum electric
sputter coater (Quorum, Q150TES, UK) and then observing the precoated
samples on a SEM system (ZEISS, MerlinTM, Germany).
The surface roughness of the PAN substrates was analyzed using an AFM
system (Asylum Research, MFP-3DTM Stand Alone, USA)
associated with Asylum Research.
The TFC membranes were characterized using an osmotically driven
process, whereby the efficiency of FO was estimated to assess the effect
of the substrate on the ICP. Specifically, the water permeability
(A ) of the membrane was measured using a high pressure
crossflow-filtration setup (Fumei Filter & Membrane Technology,
FlowMem0021-HP, China). The osmotically driven process was performed by
employing a 2 M NaCl solution as the draw solution (DS) and the water
flux was measured by weighing the feed solution (i.e., pure water). The
efficiency of FO (ηFO ) then was calculated as the
ratio of the measured flux to the ideal one (i.e., the product of the
water permeability and the osmotic pressure difference). In addition,
the salt rejection (R ) of the TFC membrane was determined in
terms of the conductivity measurement with a 20 mM NaCl solution as the
feed. All the measurements were repeated at least three times to obtain
the averaged values for a more accurate analysis.
Results and discussion
Effects of substrate hydrophilicity on the IP-film
formation
The measurement results in Figure 1 indicate that the PAN-O membrane
yielded a contact angle of approximately 67º, confirming the relatively
high hydrophilicity of PAN for fabricating TFC membranes via
IP38. The negligible difference in the contact angle
between the PAN-O and PAN-H membranes indicates that the heat treatment
with a temperature of 45ºC had little impact on the hydrophilicity of
the PAN substrate. In contrast, the introduction of carboxyl groups by
the alkaline treatment markedly increased the hydrophilicity of the PAN
substrate as indicated by the decrease in the contact angle for the
PAN-A membrane in Figure 1.