You are on page 1of 9

Journal of Membrane Science 537 (2017) 407–415

Contents lists available at ScienceDirect

Journal of Membrane Science


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

A tight nanofiltration membrane with multi-charged nanofilms for high MARK


rejection to concentrated salts

Yongliang Chena,b, Fu Liua, , Yi Wanga, Haibo Lina, Lei Hanb
a
Ningbo Institute of Materials Technology and Engineering, Chinese Academy of Sciences, Ningbo 315201, China
b
Faculty of Materials Science and Chemical Engineering, Ningbo University, Ningbo 315211, China

A R T I C L E I N F O A B S T R A C T

Keywords: The fractionation or rejection of concentrated salts by nanofiltration membrane remains the biggest challenge
Tight nanofiltration membrane owning to the charge neutrality and concentration polarization. A tight nanofiltration membrane with ultrathin
Multi charged nanofilms dually charged nanofilm down to 17 ± 5 nm was synthesized to improve the rejection of highly concentrated
Ultrathin thickness salts up to 20 g/L. The ultrathin positive polyethylenimine (PEI) and negative poly(acrylic acid) (PAA)
Concentrated salts
polyelectrolyte nanofilm was assembled on the loose nanofiltration membrane via the aid of bio-glue dopamine.
Self-assembly
The physicochemical property of dependent nanofilm was characterized by the surface morphology (SEM, AFM),
element variation (XPS), surface charge (Zeta potential), molecular weight cut off and stokes radius respectively.
The influence of dopamine deposition time and PEI concentration on separation performance was investigated.
The optimum membrane exhibited outstanding rejection to diluted multivalent salts (1 g/L, 98.5% MgSO4,
98.3% Na2SO4, 97.2% MgCl2), high removal of heavy metal ions (1 g/L, 93.5% Cd2+, 95.2% Cu2+, 92.7% Pb2+)
as well as high removal of dyes (100% Congo red, 99.93% Victoria blue B, 99.91% Brilliant green, 99.82% Basic
red 2, 99.03% Neutral red). Moreover, the resultant membrane showed exceptional rejection to highly
concentrated salts (20 g/L, 93.4% Na2SO4, 92.6% MgCl2, 93.5% MgSO4), exceeding the previously reported
membranes. Both the multi-charged nanofilms and nano-scaled thickness contributed to the outstanding
nanofiltration performance.

1. Introduction (e.g., Mg2+, Ca2+, Zn2+) and certain small organic dyes. For example,
NF270 and NF200 from Dow only rejected 40–60% and 50–65% of
The shortage of fresh water and deterioration of water environ- CaCl2, respectively [14,15]. This limits the applicability of nanofiltra-
mental quality have seriously threatened the continuous development tion in a wide range of industrial processes such as water softening, the
of human being, which urgently requires suitable water treatment removal of heavy metal ions and the recovery of industrial effluent
technology to alleviate the current situation [1–3]. Among high [16–18]. In contrast, positively charged NF membranes possess high
efficient membrane separation technologies [4–6], nanofiltration (NF) rejection to multivalent cations but relatively low rejection to multi-
membrane, a pressure-driven membrane with low molecular weight valent anions [9,19,20]. An et al. developed a novel positively charged
cutoff (MWCO) ranging from 200 to 1000 Da, has attracted significant nanofiltration membrane prepared by interfacial polymerization be-
attention ascribed to its unique fractionation and separation for multi- tween PEI mixing with PDA modified multiwall carbon nanotubes
ions and mono-ions [7–11]. (MWCNTs) and TMC, and the resulted membrane showed 90% rejec-
State-of-art nanofiltration membranes mainly consist of a polyamide tion to multivalent cations [21]. Lee et al. prepared thin film composite
(PA) selective layer on polysulfone ultrafiltration membrane via inter- polyamine nanofiltration membranes by interfacial polymerization
facial polymerization [7]. Therefore most of commercial NF membranes between PEI and CC, which showed higher rejection to cations and
are negatively charged resulting from the existence of unreacted acyl lower rejection to anions. Besides, the membrane showed exceptionally
halide groups that can hydrolyze into hydroxyl or carboxylic groups strong resistance towards nucleophilic attack induced by extreme pH
[12]. The negatively charged nanofiltration membrane exhibited high conditions [22]. However, the positively charged surface tends to
rejection to multivalent anions (e.g.,SO42-, Cr2O72-) based on Donnan absorb anions (e.g.,SO42-) or negative foulants due to electrostatic
exclusion [13], while losing its efficiency to reject multivalent cations interaction, which partially weakens the electrostatic repulsion to


Corresponding author.
E-mail address: fu.liu@nimte.ac.cn (F. Liu).

http://dx.doi.org/10.1016/j.memsci.2017.05.036
Received 21 January 2017; Received in revised form 16 April 2017; Accepted 7 May 2017
Available online 10 May 2017
0376-7388/ © 2017 Elsevier B.V. All rights reserved.
Y. Chen et al. Journal of Membrane Science 537 (2017) 407–415

Scheme 1. Schematic preparation of tight NF membrane with multi-charged nanofilms PA/PDA/PEI/PAA.

cations and accelerates the membrane fouling [23,24]. Wang et al. however, the pressure was increased up to 30 bar [12]. In this case, the
achieved high divalent cations removal using a positively charged pore size sieving and diffusivity of ions dominated the nanofiltration
hollow fiber membrane prepared by inner surface layer-by-layer (LBL) process. Both the density and thickness of the selective layer would
polyelectrolyte deposition but found later that the rejection was determine the tradeoff between rejection to multi cations and anions
drastically deteriorated when there was SO42- in the feed water. The and permeability.
further crosslinking of polyelectrolyte layer can tighten the membrane Herein, we aim to construct a tight nanofiltration membrane with
to achieve the water softening with the presence of SO42- and TDS up to comprehensive rejection to highly concentrated salts containing cations
10.0 g/L [24,25]. and anions up to 20 g/L. The membrane contained ultrathin multi-
Thus, it is necessary to develop a novel nanofiltration membrane charged nanofilms. The pristine loose polyamide membrane was
with comprehensive rejection performance for both multivalent cations tailored by dopamine self-polymerization and then able to immobilize
and anions. Despite the numerous publications on nanofiltration positively charged PEI layer and negatively charged PAA layer. The top
membrane, most of which are focusing on the fabrication strategy PEI/PAA polyelectrolyte nanofilm is only 17 ± 5 nm. The multilayer
variation for loose nanofliltration membrane. Enormous efforts have morphology, element contents and surface charge was analyzed by
sought to create graphene oxide or carbon nanotubes related mem- SEM, AFM, XPS and zeta potential. The influence of dopamine
branes to improve the permeability [21,26–28], however all these deposition time and PEI concentration on the membrane performance
membranes exhibited inferior rejection to multivalent ions even in was investigated. The separation performance of the resultant tight
diluted salts. The so produced loose nanofiltration membrane usually nanofiltration membranes to both diluted and concentrated salts, heavy
deals with diluted salts (~1 g/L). The highly concentrated salts may metal ions and dyes were evaluated respectively.
diminish the Donnan effect and neutralize the charged surface, and also
the concentration polarization may also promote the rapid decline of 2. Experimental
rejection to multivalent ions [29]. And also the graphene oxide
involved membrane is far from feasible application considering the 2.1. Materials
high cost.
To tune the surface charge on electrostatic repulsion and improve The commercial polyamide based nanofiltration membrane
the fouling resistance, Wu et al. purposely designed a dually charged (VM101), designated as M0, was provided by Shanghai Fanwei
nanofiltration membrane via amination interfacial polymerization [30]. environmental engineering Co. Ltd., China. Dopamine hydrochloride,
The membrane was thought to consist of a positive interlayer and Tris(hydroxymethyl)aminomethane, Poly(ethyleneimine) (PEI,
negative top layer. The resultant membrane exhibited high multivalent- Mw=10 kDa, 99%), polyethylene glycol 600 and 1000 (PEG-600,
salt rejection (e.g. 99.0% MgCl2, 98.5% CaCl2, 96.0% Na2SO4) with PEG-1000) and other inorganic salts were purchased from Shanghai
reasonable high flux (8.7 L m−2 h−1 bar−1) while the salt concentra- Aladdin Chemistry Co. Ltd., China. Hydrochloric acid, Copper(II)
tion is as low as 1 g/L. Ba et al. prepared a composite NF membrane by sulfate pentahydrate, hydrogen peroxide, ethylene glycol, diethylene
coating a sulfonated poly (ether ether ketone) layer on a positively glycol, polyethylene glycol 200 and 400 (PEG-200, PEG-400) and
charged NF membrane, which can selectively remove multivalent ions Neutral red were purchased from Sinopharm Chemical Reagent Co.
(94% MgCl2, 95% MgSO4, 86% Na2SO4) when the salt concentration is Ltd., China. Poly (acrylic acid) partial sodium salt solution (PAA,
0.5 g/L [31]. Actually, these two strategies followed the similar average Mw=5000 Da, 50 wt% in water) was provided by Sigma
procedure: coating or polymerizing the negatively charged layer onto Aldrich. Victoria Blue B and Brilliant Green were purchased from
the positively charged inner layer. However, the above reported J & K Scientific Ltd., China. Congo Red and Basic Red 2 were supplied
membranes remains unknown for the rejection to highly concentrated by Tokyo Chemical Industry Co. Ltd.
salts. Besides, Benes et al. employed interfacial polymerization of
cyanuric chloride and monomeric amines to fabricate a nearly neutrally 2.2. Membrane fabrication
charged NF membrane, which can reject 90% CaCl2 and 80% Na2SO4
when the salt concentration was increased up to 10 g/L at neutral pH, The fabrication process of the composite membranes was shown in

408
Y. Chen et al. Journal of Membrane Science 537 (2017) 407–415

Fig. 1. Surface SEM morphology of (a) M0, (b) M1, (c) M2 and (d) M3.

Scheme 1. Firstly, dopamine hydrochloride (2 g/L) was dissolved in Tris The zeta potential of the membranes (M0, M2, M3) was evaluated using
buffer solution (pH =8.5) with CuSO4/H2O2 to accelerated the self- a streaming potential method by the electrokinetic analyzer (SurPASS
polymerization and shortened the deposition time [32]. Then the Anton Paar, GmbH, Austria) with 1 mM KCl solution as electrolyte
pristine membrane (M0) was immediately immersed into the above solution [9,21].
solution for different durations at static state and room temperature to
form the adhesive bio-glue layer. Subsequently, the dopamine modified
2.4. Nanofiltration performances
membrane, designated as M1 with polydopamine layer, was taken out
and rinsed with deionized water for 5 min to remove weakly adsorbed
The molecular weight cutoff (MWCO) of nanofiltration membrane
dopamine/polydopamine. Secondly, M1 was soaked in aqueous PEI
was determined by rejection experiments of various neutral organic
solution with different concentration (adding more PEI to the same
solutes (0.2 g/L) with molecular weights ranging from 60 to 1000 Da at
amount of liquid) at 60 ℃ for 2 h to achieve the positive layer via
5 bar. The molecular weight of the solute with 90% rejection was
Michael addition and/or Schiff-base formation. The membrane was
defined as the MWCO. A total organic carbon analyzer (TOC, multi N/
fetched out and washed thoroughly with deionized water and denoted
C2100, analytikjena, Germany) was utilized to measure the organic
as M2 with positive charge. Finally, M2 was dipped into aqueous PAA
solutes concentrations of the feed and permeate solution.
(0.1 wt%) solutions containing NaCl (0.5 M) at 40 °C for 15 min for the
For neutral molecules, the relationship between stokes radius (rs)
electrostatic self-assembly. The resultant membrane, designated as M3
and molecular weight (MW) was estimated by the following empirical
with negative layer, was taken out and washed thoroughly with
equation [12]:
deionized water for further characterization.
log(rs ) = −1. 3363 + 0. 395 log(MW ) (1)
2.3. Membrane physicochemical characterization The separation performance of the NF membranes (M0, M2, M3)
was measured by a laboratory scaled cross-flow apparatus. The effective
Membrane samples were dried thoroughly in a vacuum oven (30 °C, area of membranes was 6.16 cm2. The cross-flow velocity was 18 L/h.
24 h) prior to characterization. The chemical components of membrane The samples were pre-filtrated at 5 bar for 40–60 min to reach a steady
surfaces (M0, M1, M2 and M3) were analyzed by X-ray photoelectron state before data collection. Pure water flux (F), the rejection of diluted
spectrometer (XPS, AXIS UTLTRADLD, Japan), which Al-Kα was as concentration inorganic salts (1.0 g/L) and dyes (0.1 g/L) were mea-
radiation resource. The membrane morphology of the top surface and sured at 5 bar, but concentrated inorganic salts (20.0 g/L) were tested
cross section was observed using a scanning electron microscopy (SEM, at 15 bar. The average value of three times measurements was reported
S-4800, Hitachi, Japan). The cross section samples were fractured in for every sample. The flux (F) was calculated by the following
liquid nitrogen, and all SEM samples were coated with gold by equations:
sputtering for 120 s before test. Scanning Probe Microscope (SPM,
Vecco, USA) was utilized to probe the roughness and 3D morphology of V
F =
the membrane surfaces in tapping mode in a scan size of 5 µm ×5 µm. A×t (2)

409
Y. Chen et al. Journal of Membrane Science 537 (2017) 407–415

where V, A and t represent the permeate volume (L), the effective Table 1
membrane area (m2) and the filtration time (h), respectively. Surface elemental composition of M0, M1, M2 and M3.
The rejection (R) was calculated by the following equations:
Sample XPS Atomic Concentration (%)
⎛ Cp ⎞
R = ⎜1− ⎟ × 100% O N C S
⎝ Cf ⎠ (3)
M0 15.42 11.41 72.63 0.54
M1 18.79 8.67 72.09 0.44
where Cf and Cp are the solute concentration in feed and permeate,
M2 12.19 17.74 69.67 0.4
respectively. The concentration of different inorganic salt was obtained M3 20.71 12.34 66.65 0.29
via detecting the conductivity of the salt solution with an electrical
conductivity meter (DDSJ-308F, Shanghai leici instrument, China). The
concentration of different dye was determined using an ultraviolet–vi- assembly via electrostatic interaction, M3 exhibited a quite smooth
sible spectrophotometer (Lambda 950, Perkin Elmer, US). surface and the nano-papillae micro-structure disappeared. The sharp
peaks turned to blunt and Ra decreased to 11.6 nm as shown in
Fig. 2(d). There were two possible reasons for the result. One possible
3. Results and discussion
explanation was that the covering linear chain of poly (acrylic acid)
layer was rather flexible. The other was that the salt broke some of the
3.1. Surface morphologies
anion-cation bonds in self-assembly and its removal by washing in pure
water leaded to the rearrangement of assembled polymers in a more
Compared to the pristine membrane M0, there were a large number
equilibrated conformation of the polymer chains [37]. The fine
of nano-papillae appearing on the surface of the dopamine modified
morphology evolution reflected formation process of PDA/PEI/PAA
membrane M1, as illustrated in Fig. 1(a) and (b). The nano-papillae
nanofilm on the pristine membrane.
were the aggregates of polydopamine (PDA) via covalent and non-
covalent interactions [33]. Through Fig. 2(a) and (b), it was also found
that bright high peaks increased and became even much sharper, 3.2. Chemical composition and surface properties of membranes
meanwhile, the surface mean roughness (Ra) of M1 increased from
9.29 nm to 14.0 nm after PDA immobilization. The result was similar to To further confirm the formation of multi-charged nanofilms, we
that reported in previous literatures [33–35]. As for M2, the surface investigated the surface oxygen, nitrogen, carbon and sulfur content of
nano-papillae microstructure partially disappeared as revealed in M0, M1, M2 and M3 by XPS and the corresponding element atomic
Fig. 1(c), indicating the incorporation of PEI layer through the Schiff- concentration was listed in Table 1. The oxygen content of M1
base and/or Michael addition with PDA [9,36]. After further PAA increased from 15.42% to 18.79% but the nitrogen content decreased

Fig. 2. AFM surface morphology of (a) M0, (b) M1, (c) M2 and (d) M3.

410
Y. Chen et al. Journal of Membrane Science 537 (2017) 407–415

Fig. 3. Cross-section SEM morphology of (a) M0, (b) M1, (c) M2 and (d) M3.

from 11.41% to 8.67% after depositing PDA layer. This was related to
the high oxygen percentage in PDA (20.92% in dopamine monomer).
After the covalent immobilization of PEI layer, the nitrogen percentage
of M2 significantly increased to 17.74%, while the oxygen percentage
declined to 12.19%. The obvious variation was ascribed to the enriched
nitrogen content of PEI. Furthermore, the adsorption of PAA had
significant impacts on the element composition of the membrane. The
oxygen content (20.71%) of M3 was apparently higher than that
(12.19%) of M2, while the nitrogen content of M3 decreased to
12.34%. In addition, the sulfur content originating from the polysulfone
support was gradually reduced after each layer deposition, which is
consistent with the increasing thickness of multi-charged layers. As
shown in Fig. 3, the pristine membrane demonstrated a nanofilm with
the thickness of 165 ± 5 nm, and the dense nanofilm of M3 increased to
220 ± 5 nm. The top dually charged PEI/PAA ultrathin nanofilm is
down to 17 ± 5 nm in thickness as the effective separating layer.
For nanofiltration membrane, surface charge played a vital role on
removing charged solutes based on the Donnan exclusion and dielectric Fig. 4. Zeta potential of M0, M2 and M3 as a function of pH.
effects [38]. The zeta potentials of the pristine membrane (M0) and the
modified membranes (M2, M3) were shown in Fig. 4. It was found that of bio-glue layer PDA, positively charged nanofilm PEI and negatively
the pristine membrane was negatively charged under test conditions charged nanofilm PAA respectively.
(pH=6.0–7.0) owning to the lower isoelectric point (3.96). In contrast,
M2 with PEI nanofilm anchored by bio-glue PDA exhibited a higher 3.3. Effect of PDA deposition time and PEI concentration on nanofiltration
isoelectric point (8.15). The similar results have been reported in the performance
previous literature [22]. Numerous primary, secondary and tertiary
amines of grafted PEI layer could be protonated and endowed the The construction of PDA and PEI spacer layer influenced the
surface with positive charges at pH=6.0–7.0. However, the isoelectric nanofiltration performance significantly. To systematically investigate
point of M3 was reduced to 4.21, which was caused by the negatively the effect of PDA deposition time on the separation performance, we
charged PAA layer through electrostatic attraction with PEI [39]. And measured the pure water flux and the rejection of Na2SO4 and MgCl2
the abundant carboxylic groups were responsible for the charge (1.0 g/L) under 5 bar of the resultant NF membranes with PDA
inversion. Combing the morphology evolution, element variation and deposition time. As shown in Fig. 5, the pure water flux gradually
surface charge inversion, we can confirm the dependent immobilization declined from 30.2 to 19.9 L m−2 h−1 as the PDA deposition time

411
Y. Chen et al. Journal of Membrane Science 537 (2017) 407–415

Table 2
Molecular weight and stokes radius of neutral solutes [12,44].

Solute Molecular weight (g/mol) rs (nm)

Ethylene glycol 62 0.24


Diethylene glycol 106 0.29
PEG-200 200 0.37
PEG-400 400 0.49
PEG-600 600 0.58
PEG-1000 1000 0.71

positively charged PEI interlay rejected Mg2- by electrostatic repulsion.


Besides, the chains of PEI and PAA can interpenetrate mutually to
densify the active layer, which contributed to the rejection through
steric hindrance. Considering the reaction efficiency, the deposition
time and the PEI concentration was determined as 30 min and 0.075 wt
% for the optimal reaction condition.

3.4. MWCO and nano-channel


Fig. 5. Effect of PDA deposition time on the separation performance. PEI and PAA
concentration were fixed at 0.075 wt% and 0.1 wt%, respectively.
By far the widely accepted nanofiltration mechanisms are the
extended from 10 to 70 min. It has been reported previously that the Donnan exclusion, dielectric effects and steric hindrance [41]. To
deposited PDA can effectively narrow the surface pore size and cause understand the nano-channels of nanofiltration membrane, the MWCOs
the surface denser [33,40]. With prolonging the deposition time, more of the pristine membrane (M0) and the resultant membrane (M3) were
PDA was adhered onto the surface and enhanced the thickness as well. measured by the rejection of various neutral organic solutes, whose
Moreover, more PDA adhered on the surface provided more reaction properties were listed in Table 2.
sites that can bind more PEI by Michael addition and/or Schiff-base, As illustrated in Fig. 7, with increasing the molecular weight, the
which further caused the nanofilm denser and thicker, which was solute rejection increased for both M0 and M3, which can be reasonably
verified by morphology evolution. These eventually led to the increase explained by the size exclusion theory. Compared with the MWCO
of mass transfer resistance toward water and decreased the water flux (approximately 400 Da) of M0, the MWCO of M3 decreased to 180 Da
accordingly. The rejection of both anionic Na2SO4 and cationic MgCl2 after the immobilization of multi charged nanofilm PDA/PEI/PAA. The
enhanced to 97 ± 2% within 30 min and became steady afterwards. pore size of M3 membrane was determined to be around 0.36 nm
Both the steric hindrance and Donnan exclusion contributed to the high according to the relationship between stokes radius and molecular
rejection to multivalent ions. weight of neutral solutes [12,42]. The anchored PDA/PEI/PAA multi-
The inorganic salts rejection is greatly related to the charge density charged nanofilm with the thickness of 55 nm was thought to tighten
of the NF membranes. The effect of PEI concentration on the positive the skin layer to form a tight nanofiltration membrane [43].
interlayer and the separation performance was studied as well. As
shown in Fig. 6, the pure water flux decreased with increasing the PEI 3.5. Separation performance of M0, M2 and M3 membranes
concentration. The rejection of MgCl2 increased to 97 ± 2% when the
PEI concentration was 0.075 wt%. With further increasing the PEI 3.5.1. Rejection of diluted salts, heavy metal ions and dyes
concentration, the membrane still kept the high rejection to both MgCl2 The removal of diluted salts (1.0 g/L) was measured to evaluate the
and Na2SO4 ascribed to the synergy effect between the intensified separation performance of M0, M2 and M3. As shown in Fig. 8, for M0,
positively charged PEI interlay and the top negatively charged PAA the removal of Na2SO4 was over 95%, while the removal of MgCl2 was
layer. The top negatively charged PAA layer rejected SO42- and the only 40%. The salt rejection followed the order of Na2SO4 > MgSO4 >

Fig. 7. The MWCOs of the pristine membrane (M0) and the nanofiltration membrane
(M3)a. Tested at 5 bar using 0.2 g/L solutions. a M3 prepared condition: PDA deposition
Fig. 6. Effect of PEI concentration on the separation performance. PDA deposition time time, PEI and PAA concentration were fixed at 30 min, 0.075 wt% and 0.1 wt%,
and PAA concentration were fixed at 30 min and 0.1 wt%, respectively. respectively.

412
Y. Chen et al. Journal of Membrane Science 537 (2017) 407–415

Fig. 8. Rejection of M0, M2a and M3b to different salts. All tests were performed at an
operating pressure of 5 bar using 1.0 g/L salts solution. a M2 prepared condition: PDA
deposition time and PEI concentration were fixed at 30 min and 0.075 wt%. b M3
prepared condition: PDA deposition time, PEI and PAA concentration were fixed at
30 min, 0.075 wt% and 0.1 wt%, respectively. Fig. 9. Rejection of M3a membrane to different dyes. All tests were performed at an
operating pressure of 5 bar using 0.1 g/L dyes solution. a M3 membrane prepared
condition: PDA deposition time, PEI and PAA concentration were fixed at 30 min,
Table 3
0.075 wt% and 0.1 wt%, respectively.
Stokes radius of various ions [21].

Ions Na+ Mg2+ Cl- SO42- Cd2+ Cu2+ Pb2+ NO3-

Stokes radius 0.184 0.347 0.121 0.230 0.341 0.325 0.283 0.129
(nm)

Table 4
The molecular weight and charge of dyes.

Dyes Molecular weight (g/mol) Charge

Congo red 696.66 –


Victoria blue B 506.09 +
Brilliant green 482.64 +
Basic red 2 350.85 +
Neutral red 288.78 +

MgCl2, which was the typical characteristics of negatively charged NF


membranes. After the immobilization of PEI nanofilm, the positively
charged membrane M2 showed an opposite rejection order (MgCl2 > Fig. 10. Rejection of M3a membrane to different salts. All tests were performed at an
MgSO4 > Na2SO4). While the rejection to Na2SO4 decreased from 97% operating pressure of 15 bar using 20.0 g/L salts solution. a M3 membrane prepared
to 70% due to the adsorption of SO42- and the charge shielding effect, condition: PDA deposition time, PEI and PAA concentration were fixed at 30 min,
the rejection to MgCl2 and MgSO4 enhanced to 95% and 92% from 40% 0.075 wt% and 0.1 wt%, respectively.
and 91%, respectively. The surface charge inversion dominated the
rejection enhancement via electrostatic exclusion. The divalent cation the chains of PEI and PAA interpenetrated mutually [37,43]. Further-
(Mg2+) was excluded more effectively than the monovalent cation more, M3 membrane had the mean effective pore radius around
(Na+) for the positively charged surface owning to stronger co-ions 0.36 nm, which was larger than the stokes radius of SO42- (0.23 nm),
repulsion, the divalent anion (SO42-) was attracted more easily than the Mg2+ (0.347 nm) and Na+ (0.184 nm) in Table 3. The high rejection to
monovalent anion (Cl-) due to stronger counter-ions attraction. MgSO4 MgSO4 and Na2SO4 implied the Donnan exclusion governed the
containing both divalent cation and divalent anion can be rejected transport mechanism. In addition, M3 membrane exhibited high
effectively by both positively charged membranes and negatively removal of heavy metal ions (93.5% Cd2+, 95.2% Cu2+, 92.7%
charged membranes respectively. M3 membrane comprising top dually Pb2+) compared to M0 and M2. M3 showed 59 ± 1% rejection to
charged PEI/PAA nanofilm exhibited super high rejection to both NaCl, indicating the tight nanofiltration membrane exhibited certain
divalent cations and divalent anions. This result obviously violated fractionation to mono- and multivalent salts.
the common Donnan effects. The dually charged layers interaction [30] Dye removal capacity of M3 membrane was further evaluated with
cooperating with the steric effects supported the new order rejection. the dye concentration of 0.1 g/L at 5 bar. Table 4 listed molecular
The anions were mainly rejected by the outermost negatively charged weight and charge of the five kinds of dyes. Since the molecular weight
PAA nanofilm via the electrostatic repulsion. Despite the cations was of all dyes is higher than the MWCO (180 Da) of M3, M3 membrane
attracted and permeated into the positively charged layer due to the exhibited high removal to all five dyes, following the order of Congo
affinity interaction, the adsorbed cations can be rejected by the red (100%) > Victoria blue B (99.93%) ≈ Brilliant green (99.91%) >
positively charged PEI interlayer and trapped in the overlap, where Basic red 2 (99.82%) > Neutral red (99.03%) despite the dye charge, as

413
Y. Chen et al. Journal of Membrane Science 537 (2017) 407–415

Table 5
The separation performance of various nanofiltration membranes.

Salt concentration (g/L) Salt rejection (%) PWP (Lm−2 h−1 bar−1) Refs.

MgCl2 CaCl2 MgSO4 Na2SO4 NaCl

0.5 94 95 86 70 [31]
1 91.5 90.5 76.1 45.2 33.8 15.32 [21]
1 98.2 97.4 95.2 33.5 [24]
1 78.8 70.3 95.2 98.9 39.5 [13]
2 97.1 97.6 35.3 6.9 [26]
2 91.5 78 85 [12]
10 34.6 32.4 26.9 10.7 [46]
10 89.5 77 83 [12]
1 97.2 98.5 98.3 59.0 5.5 This work
20 92.5 93.5 93.4

shown in Fig. 9. References

[1] X.Q. Cheng, L. Shao, C.H. Lau, High flux polyethylene glycol based nanofiltration
3.5.2. Rejection of highly concentrated salts membranes for water environmental remediation, J. Membr. Sci. 476 (2015)
The common loose nanofiltration membrane behaved well in 95–104.
separating diluted salts (~1.0 g/L). However, both rejection and [2] H.C. Yang, J. Luo, Y. Lv, P. Shen, Z.K. Xu, Surface engineering of polymer
membranes via mussel-inspired chemistry, J. Membr. Sci. 483 (2015) 42–59.
permeability will decline dramatically while treating highly concen-
[3] H.J. Kim, K. Choi, Y. Baek, D.G. Kim, J. Shim, J. Yoon, J.C. Lee, High-performance
trated salts [29,45,46]. The excessive salts will diffuse and adsorb on reverse osmosis CNT/polyamide nanocomposite membrane by controlled inter-
the membrane to neutralize the charged surface and shield the facial interactions, ACS Appl. Mater. Interfaces 6 (2014) 2819–2829.
[4] L. Huang, J. Chen, T. Gao, M. Zhang, Y. Li, L. Dai, L. Qu, G. Shi, Reduced graphene
electrostatic repulsion. The enhanced osmosis pressure and polarization
oxide membranes for ultrafast organic solvent nanofiltration, Adv. Mater. 28 (2016)
concentration on the boundary layer will accelerate the diffusion and 8669–8674.
transportation of ions and decrease the water permeability conse- [5] P. Marchetti, M.F. Jimenez Solomon, G. Szekely, A.G. Livingston, Molecular
quently. To conquer the challenging issue, the tight multi nanofilm separation with organic solvent nanofiltration: a critical review, Chem. Rev. 114
(2014) 10735–10806.
charged membrane was designed to reject 20.0 g/L multivalent ions. As [6] S. Karan, Z. Jiang, A.G. Livingston, Sub–10 nm polyamide nanofilms with ultrafast
shown in Fig. 10, the rejection of M3 to Na2SO4, MgCl2 and MgSO4 is as solvent transport for molecular separation, Science 348 (2015) 1347–1351.
high as 93.4%, 92.6% and 93.5%, respectively. Highly concentrated [7] Y. Gao, A.M.S. de Jubera, B.J. Mariñas, J.S. Moore, Nanofiltration membranes with
modified active layer using aromatic polyamide dendrimers, Adv. Funct. Mater. 23
multivalent cation Mg2+ and anion SO42- accumulated on the boundary (2013) 598–607.
zone will balance the charge environment ascribing to the strong [8] M.B. Wu, Y. Lv, H.C. Yang, L.F. Liu, X. Zhang, Z.K. Xu, Thin film composite
counter-ion attraction. Therefore, the rejection is slightly lower than membranes combining carbon nanotube intermediate layer and microfiltration
support for high nanofiltration performances, J. Membr. Sci. 515 (2016) 238–244.
that the diluted salts as shown in Fig. 8. However, the tight multi- [9] Y. Lv, H.C. Yang, H.Q. Liang, L.S. Wan, Z.K. Xu, Nanofiltration membranes via co-
nanofilm comprising negative PDA, positive PEI and negative PAA deposition of polydopamine/polyethylenimine followed by cross-linking, J. Membr.
ensured the high rejection to multivalent ions and permeability. Sci. 476 (2015) 50–58.
[10] F. Liu, Br Ma, D. Zhou, L.J. Zhu, Y.Y. Fu, Lx Xue, Positively charged loose
Compared to the previously reported membrane, the tight M3 mem-
nanofiltration membrane grafted by diallyl dimethyl ammonium chloride
brane showed superior rejection to highly concentrated salts as shown (DADMAC) via UV for salt and dye removal, React. Funct. Polym. 86 (2015)
in Table 5. We attributed the outstanding nanofiltration performance in 191–198.
[11] X.L. Li, L.P. Zhu, Y.Y. Xu, Z. Yi, B.K. Zhu, A novel positively charged nanofiltration
selectivity and permeance to the multi-charged nanofilms and ultrathin
membrane prepared from N,N-dimethylaminoethyl methacrylate by quaternization
thickness down to 17 ± 5 nm. cross-linking, J. Membr. Sci. 374 (2011) 33–42.
[12] K.P. Lee, G. Bargeman, R. de Rooij, A.J.B. Kemperman, N.E. Benes, Interfacial
polymerization of cyanuric chloride and monomeric amines: pH resistant thin film
4. Conclusions composite polyamine nanofiltration membranes, J. Membr. Sci. 523 (2017)
487–496.
[13] W.Z. Qiu, Q.Z. Zhong, Y. Du, Y. Lv, Z.K. Xu, Enzyme-triggered coatings of tea
In this work, we successfully prepared a tight nanofiltration catechins/chitosan for nanofiltration membranes with high performance, Green
membrane with multi- charged nanofilms for exceptional rejection to Chem. 18 (2016) 6205–6208.
highly concentrated salts. The positively charged PEI and negatively [14] Y.L. Lin, P.C. Chiang, E.E. Chang, Removal of small trihalomethane precursors from
aqueous solution by nanofiltration, J. Hazard. Mater. 146 (2007) 20–29.
PAA were assembled to form the dually charged nanofilm based on bio-
[15] A.W. Mohammad, Y.H. Teow, W.L. Ang, Y.T. Chung, D.L. Oatley-Radcliffe, N. Hilal,
glue PDA. The PDA deposition time and PEI concentration was Nanofiltration membranes review: recent advances and future prospects,
determined as 30 min and 0.075 wt% to achieve the optimal M3 Desalination 356 (2015) 226–254.
[16] W. Fang, L. Shi, R. Wang, Interfacially polymerized composite nanofiltration hollow
membrane. The surface morphology, element response and charge
fiber membranes for low-pressure water softening, J. Membr. Sci. 430 (2013)
inversion confirmed the ultrathin nanofilm down to17 ± 5 nm in 129–139.
thickness. The tight membrane showed a MWCO of 180 Da and stokes [17] B.A.M. Al-Rashdi, D.J. Johnson, N. Hilal, Removal of heavy metal ions by
radius of 0.36 nm. The membrane exhibited outstanding rejection nanofiltration, Desalination 315 (2013) 2–17.
[18] T.Y. Liu, L.X. Bian, H.G. Yuan, B. Pang, Y.K. Lin, Y. Tong, B. Van der Bruggen,
performance to various diluted salts, heavy metal ions and dyes by X.L. Wang, Fabrication of a high-flux thin film composite hollow fiber nanofiltration
virtue of Donnan exclusion and steric hindrance. More importantly, the membrane for wastewater treatment, J. Membr. Sci. 478 (2015) 25–36.
tight nanofiltration membrane showed surpassing selectivity than [19] Y. Tang, B. Tang, P. Wu, Preparation of a positively charged nanofiltration
membrane based on hydrophilic–hydrophobic transformation of a poly(ionic
previously reported ones. liquid), J. Mater. Chem. A 3 (2015) 12367–12376.
[20] Y. Cui, Z.K. Yao, K. Zheng, S.Y. Du, B.K. Zhu, L.P. Zhu, C.H. Du, Positively-charged
nanofiltration membrane formed by quaternization and cross-linking of blend PVC/
Acknowledgements P(DMA-co-MMA) precursors, J. Membr. Sci. 492 (2015) 187–196.
[21] F.Y. Zhao, Y.L. Ji, X.D. Weng, Y.F. Mi, C.C. Ye, Q.F. An, C.J. Gao, High-flux
positively charged nanocomposite nanofiltration membranes filled with poly
Financial support is acknowledged from National Natural Science
(dopamine) modified multiwall carbon nanotubes, ACS Appl. Mater. Interfaces 8
Foundation of China (5161101025, 51673209), Youth Innovation (2016) 6693–6700.
Promotion Association of Chinese Academy of Sciences (2014258). [22] K.P. Lee, J. Zheng, G. Bargeman, A.J.B. Kemperman, N.E. Benes, pH stable thin film

414
Y. Chen et al. Journal of Membrane Science 537 (2017) 407–415

composite polyamine nanofiltration membranes by interfacial polymerisation, J. flux nanofiltration membranes enabled by dual functional polydopamine, ACS Appl.
Membr. Sci. 478 (2015) 75–84. Mater. Interfaces 6 (2014) 5548–5557.
[23] J. Garcia-Aleman, J.M. Dickson, Permeation of mixed-salt solutions with commer- [35] L. Shao, Z.X. Wang, Y.L. Zhang, Z.X. Jiang, Y.Y. Liu, A facile strategy to enhance
cial and pore-filled nanofiltration membranes: membrane charge inversion phe- PVDF ultrafiltration membrane performance via self-polymerized polydopamine
nomena, J. Membr. Sci. 239 (2004) 163–172. followed by hydrolysis of ammonium fluotitanate, J. Membr. Sci. 461 (2014)
[24] C. Liu, L. Shi, R. Wang, Crosslinked layer-by-layer polyelectrolyte nanofiltration 10–21.
hollow fiber membrane for low-pressure water softening with the presence of [36] M. Liu, J. Ji, X. Zhang, X. Zhang, B. Yang, F. Deng, Z. Li, K. Wang, Y. Yang, Y. Wei,
SO42− in feed water, J. Membr. Sci. 486 (2015) 169–176. Self-polymerization of dopamine and polyethyleneimine: novel fluorescent organic
[25] C. Liu, L. Shi, R. Wang, Enhanced hollow fiber membrane performance via semi- nanoprobes for biological imaging applications, J. Mater. Chem. B 3 (2015)
dynamic layer-by-layer polyelectrolyte inner surface deposition for nanofiltration 3476–3482.
and forward osmosis applications, React. Funct. Polym. 86 (2015) 154–160. [37] G. Decher, Fuzzy nanoassemblies: toward layered polymeric multicomposites,
[26] S.M. Xue, Z.L. Xu, Y.J. Tang, C.H. Ji, Polypiperazine-amide nanofiltration mem- Science 277 (1997) 1232–1237.
brane modified by different functionalized multiwalled carbon nanotubes [38] X.L. Wang, Y.Y. Fang, C.H. Tu, B. Van der Bruggen, Modelling of the separation
(MWCNTs), ACS Appl. Mater. Interfaces 8 (2016) 19135–19144. performance and electrokinetic properties of nanofiltration membranes, Int. Rev.
[27] Y. Han, Y. Jiang, C. Gao, High-flux graphene oxide nanofiltration membrane Phys. Chem. 31 (2012) 111–130.
intercalated by carbon nanotubes, ACS Appl. Mater. Interfaces 7 (2015) 8147–8155. [39] H.Y. Yu, Z.K. Xu, Q. Yang, M.X. Hu, S.Y. Wang, Improvement of the antifouling
[28] J. Zhu, M. Tian, J. Hou, J. Wang, J. Lin, Y. Zhang, J. Liu, B. Van der Bruggen, characteristics for polypropylene microporous membranes by the sequential
Surface zwitterionic functionalized graphene oxide for a novel loose nanofiltration photoinduced graft polymerization of acrylic acid, J. Membr. Sci. 281 (2006)
membrane, J. Mater. Chem. A 4 (2016) 1980–1990. 658–665.
[29] C.Z. Liang, S.P. Sun, B.W. Zhao, T.S. Chung, Integration of nanofiltration hollow [40] X.Q. Cheng, C. Zhang, Z.X. Wang, L. Shao, Tailoring nanofiltration membrane
fiber membranes with coagulation–flocculation to treat colored wastewater from a performance for highly-efficient antibiotics removal by mussel-inspired modifica-
dyestuff manufacturer: a pilot-scale study, Ind. Eng. Chem. Res. 54 (2015) tion, J. Membr. Sci. 499 (2016) 326–334.
11159–11166. [41] O. Labban, C. Liu, T.H. Chong, J.H. Lienhard V, Fundamentals of low-pressure
[30] C. Wu, S. Liu, Z. Wang, J. Zhang, X. Wang, X. Lu, Y. Jia, W.S. Hung, K.R. Lee, nanofiltration: membrane characterization, modeling, and understanding the multi-
Nanofiltration membranes with dually charged composite layer exhibiting super- ionic interactions in water softening, J. Membr. Sci. 521 (2017) 18–32.
high multivalent-salt rejection, J. Membr. Sci. 517 (2016) 64–72. [42] W.R. Bowen, A.W. Mohammad, Diafiltration by nanofiltration: prediction and
[31] S. Zhao, Y. Yao, C. Ba, W. Zheng, J. Economy, P. Wang, Enhancing the performance optimization, AIChE 44 (1998) 1799–1812.
of polyethylenimine modified nanofiltration membrane by coating a layer of [43] J.B. Schlenoff, S.T. Dubas, Mechanism of polyelectrolyte multilayer growth: charge
sulfonated poly(ether ether ketone) for removing sulfamerazine, J. Membr. Sci. 492 overcompensation and distribution, Macromolecules 34 (2001) 592–598.
(2015) 620–629. [44] Z. Thong, G. Han, Y. Cui, J. Gao, T.S. Chung, S.Y. Chan, S. Wei, Novel nanofiltration
[32] C. Zhang, Y. Ou, W.X. Lei, L.S. Wan, J. Ji, Z.K. Xu, CuSO4/H2O2-Induced rapid membranes consisting of a sulfonated pentablock copolymer rejection layer for
deposition of polydopamine coatings with high uniformity and enhanced stability, heavy metal removal, Environ. Sci. Technol. 48 (2014) 13880–13887.
Angew. Chem. Int. Ed. 55 (2016) 3054–3057. [45] J. Lin, W. Ye, H. Zeng, H. Yang, J. Shen, S. Darvishmanesh, P. Luis, A. Sotto, B. Van
[33] R. Zhang, Y. Su, X. Zhao, Y. Li, J. Zhao, Z. Jiang, A novel positively charged der Bruggen, Fractionation of direct dyes and salts in aqueous solution using loose
composite nanofiltration membrane prepared by bio-inspired adhesion of poly- nanofiltration membranes, J. Membr. Sci. 477 (2015) 183–193.
dopamine and surface grafting of poly(ethylene imine), J. Membr. Sci. 470 (2014) [46] X. Bai, Y. Zhang, H. Wang, H. Zhang, J. Liu, Study on the modification of positively
9–17. charged composite nanofiltration membrane by TiO2 nanoparticles, Desalination
[34] Y. Li, Y. Su, X. Zhao, X. He, R. Zhang, J. Zhao, X. Fan, Z. Jiang, Antifouling, high- 313 (2013) 57–65.

415

You might also like