You are on page 1of 12

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/315703852

Relating Silica Scaling in Reverse Osmosis to Membrane Surface Properties

Article  in  Environmental Science & Technology · March 2017


DOI: 10.1021/acs.est.6b06411

CITATIONS READS

10 251

5 authors, including:

Tiezheng Tong Chanhee Boo


Yale University Columbia University
29 PUBLICATIONS   452 CITATIONS    22 PUBLICATIONS   964 CITATIONS   

SEE PROFILE SEE PROFILE

Sara M Hashmi Menachem Elimelech


Yale University Yale University
48 PUBLICATIONS   792 CITATIONS    537 PUBLICATIONS   52,664 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Development of anti-scaling membranes for high recovery desalination View project

Electrocatalytic membrane fabrication and modification View project

All content following this page was uploaded by Tiezheng Tong on 19 October 2017.

The user has requested enhancement of the downloaded file.


Article

pubs.acs.org/est

Relating Silica Scaling in Reverse Osmosis to Membrane Surface


Properties
Tiezheng Tong,‡,§,∥ Song Zhao,†,‡,∥ Chanhee Boo,‡ Sara M. Hashmi,‡ and Menachem Elimelech*,‡,§

School of Chemical Engineering and Technology, Tianjin Key Laboratory of Membrane Science and Desalination Technology,
Tianjin University, Tianjin 300072, P. R. China

Department of Chemical and Environmental Engineering, Yale University, New Haven, Connecticut 06520-8286, United States
§
Nanosystems Engineering Research Center for Nanotechnology-Enabled Water Treatment (NEWT), Yale University, New Haven,
Connecticut, United States
*
S Supporting Information

ABSTRACT: We investigated the relationship between mem-


brane surface properties and silica scaling in reverse osmosis
(RO). The effects of membrane hydrophilicity, free energy for
heterogeneous nucleation, and surface charge on silica scaling
were examined by comparing thin-film composite polyamide
membranes grafted with a variety of polymers. Results show that
the rate of silica scaling was independent of both membrane
hydrophilicity and free energy for heterogeneous nucleation. In
contrast, membrane surface charge demonstrated a strong
correlation with the extent of silica scaling (R2 > 0.95, p <
0.001). Positively charged membranes significantly facilitated silica scaling, whereas a more negative membrane surface charge led
to reduced scaling. This observation suggests that deposition of negatively charged silica species on the membrane surface plays a
critical role in silica scale formation. Our findings provide fundamental insights into the mechanisms governing silica scaling in
reverse osmosis and highlight the potential of membrane surface modification as a strategy to reduce silica scaling.

■ INTRODUCTION
Silica is one of the most common inorganic scalants in
and subsequently prevent the formation of scale on the
membrane surface.12−14 However, the use of scale inhibitors
membrane desalination systems such as reverse osmosis (RO). increases the operation cost of desalination and can result in
Silica is ubiquitous in natural waters, with concentrations organic and biological fouling.15,16 Since the solubility of silica
normally in the range of 1−60 mg/L.1−3 Once feedwater silica increases with pH, high operation pH (i.e., pH > 10) has been
concentration exceeds its solubility limit (typically 100−150 used as an alternative strategy to scale inhibitor addition in RO
mg/L at near neutral pH2,4,5), insoluble silica precipitates and desalination of silica-rich feedwater at high water recov-
forms scale on the membrane surface. Silica scaling causes eries,17,18 but high operation pH leads to the precipitation of
severe flux decline and requires chemical and mechanical calcium and magnesium silicates.19 As a result, extensive
cleaning operations,2−4,6,7 which limits the efficiency and water pretreatments, such as chemical softening and cationic
recovery of RO desalination. In particular, silica scaling is a exchange, are required to remove hardness from the feed-
major barrier for efficient operation of inland brackish water water.20 Therefore, silica scaling remains a challenging problem
RO desalination where high water recovery is critical for brine facing membrane desalination, making development of new
management.4,6,8 strategies for scaling control highly desired.
The chemistry and mechanisms of silica scaling are complex Membrane surface modification may be a promising
and not well understood.4 Silica scaling involves the polymer- approach to reduce silica scaling. Previous studies have
ization of monomeric silicic acids forming Si−O−Si bonds via demonstrated that surface chemistry influences silica scaling.
dehydration.4,9 Polymerization of the weakly acid silicic acids For example, by using an organic silica precursor, Wallace et
produces various silica species ranging from dimers and trimers al.21 have shown that silica nucleation occurred on model
to polymers and particles.10 Although heterogeneous nuclea- surfaces with carboxyl or hybrid amine/carboxyl functional
tion, in which silicic acids deposit and polymerize on the groups but not on amine-terminated surfaces. When compared
membrane surface, has been proposed as the major mechanism to a cellulose acetate membrane with surface hydroxyl groups,
of silica scaling,2,11 bulk deposition of colloidal silica may also
play an important role.4,7 Received: December 17, 2016
Current strategies for silica scaling control in membrane Revised: February 20, 2017
desalination rely heavily on the use of scale inhibitors (or Accepted: March 28, 2017
antiscalants). Scale inhibitors stabilize silica species in solution Published: March 28, 2017

© 2017 American Chemical Society 4396 DOI: 10.1021/acs.est.6b06411


Environ. Sci. Technol. 2017, 51, 4396−4406
Environmental Science & Technology Article

Figure 1. Schematics of membrane modification approaches. Activators regenerated by electron transfer−atom transfer radical polymerization
(ARGET-ATRP) was employed to graft poly(acrylamide) (PAM), poly(sulfobetaine methacrylate) (PSBMA), and poly([2-(methacryloyloxy)-
ethyl]trimethylammonium chloride) (PMTAC) on the membrane surface, assisted by dopamine-bromoisobutyryl bromide (Bibb) as the initiator.
Redox radical initiation was employed to graft poly(acrylic acid) (PAA) on the membrane surface. Direct grafting, assisted by dopamine, was
employed to graft polyethylenimine (PEI) and 1H,1H,2H,2H-perfluorodecanethiol (PFDT) on the membrane surface. For PEI, two molecular
weights were used, 800 and 1300 g/mol.

polyamide membranes with native carboxyl groups showed


more irreversible silica scaling in both RO and forward osmosis
■ MATERIALS AND METHODS
Materials and Chemicals. Acrylamide (>99.0%), [2-
operation.2 (methacryloyloxy)ethyl]dimethyl-(3-sulfopropyl)ammonium
To date, membrane surface modification has been used hydroxide (also known as sulfobetaine methacrylate, SBMA),
extensively for reducing organic and biological fouling in [2-(methacryloyloxy)ethyl]trimethylammonium chloride solu-
membrane desalination systems.22−28 In contrast, very few tion (MTAC, 80 wt % in H2O), polyethylenimine (PEI,
studies have focused on surface modification for inorganic branched, Mw of ∼800 g/mol and ∼1300 g/mol),
scaling control. For example, a hydrophilic brush layer of 1H,1H,2H,2H-perfluorodecanethiol (PFDT, 97%), acrylic acid
poly(methacrylic acid) or poly(acrylamide) was shown to (99%), N,N-dimethylformamide (DMF), triethylamine (TEA,
mitigate gypsum scaling on polyamide RO membranes.29,30 ≥99%), α-bromoisobutyryl bromide (BiBB, 98%), dopamine
These brush layers provided effective screening of the hydrochloride, copper(II) chloride (99%), copper(II) bromide
underlying membrane surface, and their partial local motility (99%), tris(2-pyridylmethyl)amine (TPMA), L-ascorbic acid,
reduced the attachment rate of gypsum nuclei and/or potassium persulfate (K2S2O8, ≥99%), sodium metabisulfite
crystallites.29,30 To the best of our knowledge, however, (Na2S2O5, ≥99%), and sodium metasilicate pentahydrate
applications of membrane surface modification in reducing (Na2SiO3·5H2O, >95.0%) were purchased from Sigma-Aldrich.
silica scaling have not been reported in the literature. Calcium chloride dihydrate (CaCl2·2H2O) and magnesium
Elucidating how membrane surface properties influence silica chloride hexahydrate (MgCl2·6H2O) were purchased from Alfa
scaling is the prerequisite for developing scaling-resistant Aesar and J.T. Baker, respectively. Commercial TFC RO
membrane surfaces. membranes (SW30 XLE) were provided by Dow Chemical.
In this work, we investigated the relationship between Deionized (DI) water was obtained from a Milli-Q ultrapure
membrane surface properties and silica scaling in reverse water purification system (Millipore).
Membrane Surface Modification Approaches. Com-
osmosis. A commercial thin-film composite (TFC) polyamide
mercial RO membranes were immersed in 25% isopropyl
membrane was modified with diverse polymer coatings, which
alcohol for 30 min, after which the membranes were washed
provided different surface hydrophilicity, free energy for thoroughly with DI water and stored at 4 °C until use. Three
heterogeneous nucleation, and surface charge. The performance modification approaches were employed to tailor the surface
of the modified membranes with a silica-saturated feed solution chemistry of RO membranes with various polymer coatings: (i)
was tested in a bench-scale cross-flow RO system. The water activators regenerated by electron transfer−atom transfer
flux decline rate obtained from the scaling experiments was radical polymerization (ARGET-ATRP), (ii) redox radical
used to examine the effect of membrane surface properties on initiation, and (iii) dopamine-assisted direct grafting (Figure 1).
the extent of silica scaling. We demonstrate that membrane The selected polymers (as described below) have different
surface chemistry significantly influences silica scaling, with functional groups, creating diverse membrane surface proper-
membrane surface charge identified as the primary regulating ties in terms of surface hydrophilicity, free energy for
factor. Our findings promote a mechanistic understanding of heterogeneous nucleation, and surface charge.
silica scaling of RO membranes, which may guide the design ARGET-ATRP is a robust and versatile approach to produce
and development of effective scaling-resistant membranes. polymer brushes with narrow polydispersity and controllable
4397 DOI: 10.1021/acs.est.6b06411
Environ. Sci. Technol. 2017, 51, 4396−4406
Environmental Science & Technology Article

thickness and architecture.31,32 This approach, performed PEI aqueous solution or 0.05% (v/v) PFDT ethanol:DI water
following literature protocols with slight modification,31,33 was solution (1:1, v/v) for 4 h at room temperature. After the
used for grafting poly(acrylamide) (PAM), poly(sulfobetaine reaction, the solution was removed, and the PEI- or PFDT-
methacrylate) (PSBMA), and poly([2-(methacryloyloxy)ethyl]- modified membrane was rinsed thoroughly with DI water and
trimethylammonium chloride) (PMTAC) on the membrane stored at 4 °C until use.
surface. Membrane Surface Characterization. Membrane surface
Briefly, dopamine hydrochloride (400 mg, ∼2.10 mmol) was morphology was investigated by scanning electron microscopy
dissolved in 20 mL of DMF in an amber bottle with a PTFE/ (SEM, Hitachi SU-70). Before taking the images, membrane
red rubber septum, followed by adding TEA (0.15 mL, 1.05 samples were dried and sputter-coated with a thin layer of
mmol) and Bibb (0.13 mL, 1.05 mmol). After 3 h of stirring chromium or iridium. Membrane surface roughness was
under N2 at room temperature, the mixture containing evaluated by atomic force microscopy (AFM, Bruker
dopamine-Bibb was added to 100 mL of aqueous tris- Dimension Fastscan) in tapping mode with a silicon nitride
(hydroxymethyl)aminomethane buffer (pH 8.5), which was probe (ScanAsyst-air, Burker). Micrographs were captured
then immediately poured onto the membrane active layer. from six different locations with an area of 10 μm × 10 μm.
Dopamine-Bibb self-polymerized and formed poly(dopamine- Attenuated total reflectance-Fourier transform infrared (ATR-
Bibb) (PDA-Bibb) on the membrane surface, which served as FTIR) spectra were collected using a Thermo Nicolet 6700
the initiator for polymer growth. After 3 h, the PDA-Bibb spectrometer with 32 scans for each sample. Membrane surface
deposited membrane was thoroughly rinsed with DI water. hydrophilicity was analyzed by measuring the water contact
Acrylamide (14.2 g, ∼0.2 mol), SBMA (15.64 g, ∼56 mmol), or angle using the sessile drop method.38 The zeta potential of the
MTAC (14.52 g, ∼56 mmol) monomers were dissolved in 200 membrane surface was determined using a streaming potential
mL of 1:1 isopropyl alcohol:DI water mixture (v/v) in a bottle analyzer with an asymmetric clamping cell (EKA, Brookhaven
with a septa lid (covered with aluminum foil). After degassing Instruments). The measurements were conducted with a
with N2 for 10 min, the PDA-Bibb deposited membrane was solution containing 1 mM KCl and 0.1 mM KHCO3. Details
placed into the bottle. After another 10 min degassing with N2, on the procedure used to calculate the zeta potential from the
a solution of copper(II) salt and TPMA in an 8 mL 1:1 measured streaming potential are described elsewhere.39
isopropyl alcohol:DI water mixture (v/v) was injected into the Free Energy for Heterogeneous Silica Nucleation. The
bottle. Cu(II) bromide (0.008 g, ∼35.8 μmol) and TPMA thermodynamic barrier to silica nucleation is determined by the
(0.065 g, ∼0.225 mmol) were used for grafting PAM, whereas free energy of forming a silica nucleus of a critical size
Cu(II) chloride (0.004 g, ∼29.8 μmol) and TPMA (0.056 g, (ΔGcri).40 Because the membrane surface promotes nucleation,
∼0.193 mmol) were used for grafting PSBMA and PMTAC. the critical free energy for heterogeneous nucleation (ΔGhet
cri ) is
After an additional 10 min degassing with N2, 8 mL of ascorbic smaller than that for homogeneous nucleation (ΔGhom cri ). The
acid (0.8 g, ∼4.5 mmol) in a 1:1 isopropyl alcohol:DI water critical free energy for heterogeneous nucleation can be
mixture (v/v) was injected into the bottle to initiate the calculated from21,40,41
polymerization. The reaction lasted for 1 h, 3 d, and 7 d for het hom
PSBMA, PMTAC, and PAM, respectively. The longer ΔGcri = f (θ )ΔGcri (1)
polymerization duration for PMTAC and PAM was due to
their slower polymerization rate than SBMA. Finally, the bottle (2 + cos θ )(1 − cos θ )2
was opened to air to terminate the reaction. The modified f (θ ) =
membranes were washed thoroughly with DI water and stored 4 (2)
at 4 °C until use. The correction factor f(θ), or the wetting function, describes
Redox radical initiation was used to create poly(acrylic acid) the geometry of the nucleus-surface interaction (Figure S1).21
(PAA)-modified membranes, following the protocol reported The value of f(θ) ranges from zero to one and is a function of
by Belfer et al.34 In this approach, oxygen-centered radicals the contact angle (θ) between the silica nucleus and the
formed by the action of redox initiators (i.e., K2S2O8 and membrane surface. θ is calculated using Young’s equation that
Na 2 S 2O 5) are effective to graft vinyl monomers with involves the interfacial free energy among silica, water, and the
subsequent polymerization.34 In brief, 1 M solution of acrylic membrane surface41
acid was prepared, followed by adding 0.01 M K2S2O8 and 0.01
M Na2S2O5 to the monomer solution. The mixture was then γmem − wat − γmem − SiO
2
immediately poured onto the membrane active layer. After 1 h, cos θ =
γSiO − wat (3)
the PAA-modified membrane was rinsed thoroughly with DI 2

water and stored at 4 °C until use.


Dopamine-assisted direct grafting modification was used to where γmem−wat, γmem−SiO2, and γSiO2−wat denote the interfacial
create PEI- and PFDT-modified membranes. The catechol free energies associated with the membrane−water, mem-
groups of polydopamine (PDA) (after oxidation to quinone) brane−silica, and silica−water boundaries at thermodynamic
react with amine or thiol groups of the polymers through the equilibrium, respectively.
Michael-type addition or Schiff base reaction,35−37 thereby Since ΔGhom
cri for silica is constant at a fixed temperature and
grafting the polymers directly onto the membrane surface. In saturation index (eq S1 in the Supporting Information), ΔGhet cri
brief, dopamine hydrochloride (30 mg, ∼0.157 mmol) was is proportional to the value of f(θ), which is a characteristic
dissolved in 30 mL of tris(hydroxymethyl) aminomethane property of the membrane surface. The value of f(θ) was
buffer (10 mM, pH 8.5). The mixed solution was then poured calculated for each membrane, with a larger f(θ) corresponding
onto the membrane active layer. Dopamine polymerization to to a higher thermodynamic barrier (or lower membrane scaling
PDA lasted for 1 h at room temperature. After rinsing with DI propensity) to heterogeneous silica nucleation. The procedure
water, the PDA-deposited membrane was immersed in 10 g/L for calculating f(θ), γmem−wat, γmem−SiO2, and γSiO2−wat is detailed
4398 DOI: 10.1021/acs.est.6b06411
Environ. Sci. Technol. 2017, 51, 4396−4406
Environmental Science & Technology Article

Figure 2. (A) Water contact angle, (B) zeta potential, and (C) water permeability coefficient (A), salt permeability coefficient (B), and salt rejection
of the different membranes analyzed in this study. The full names of PAM, PSBMA, PMTAC, PEI, PFDT, PAA, and Bibb have been defined in
Figure 1 and in the main text. PDA and PDA-Bibb stand for dopamine and dopamine-Bibb after polymerization, respectively. The error bars
represent standard deviation and were calculated from eight (for zeta potential) and duplicate (for A, B, and salt rejection) measurements.

in the Supporting Information, following methods and flux of the membrane was recorded under 27.6 bar (400 psi) at
equations described in the literature.23,40,42 a crossflow velocity of 8.5 cm/s. Then, the silica scaling
Measurement of Membrane Transport Properties. experiment was initiated by adding the silica-saturated solution
The transport properties of the membranes were determined to the RO feed reservoir. The applied pressure (27.6 ± 2.8 bar)
using a bench-scale crossflow RO system. Details of the was adjusted to create an initial water flux of 56 ± 2 L m−2 h−1.
experimental setup have been described in our previous The water flux was continuously monitored for 1400 min at a
work.43,44 The membrane coupon (with an effective area of constant crossflow velocity of 8.5 cm/s. A recycling mode was
20.02 cm2) was compacted overnight using DI water under a employed, with the permeate recycled back to the feed solution.
pressure of 31.0 bar (450 psi). The water permeability The feed solution temperature during the silica scaling tests was
coefficient (A) was calculated from the pure water flux maintained constant at 22 ± 1 °C.
measured under 27.6 bar (400 psi) at 22 ± 1 °C. The salt After silica scaling, membrane cleaning was performed by
permeability coefficient (B) was calculated from the salt rinsing the membrane with DI water at a high crossflow velocity
rejection measured under 27.6 bar (400 psi) at 22 ± 1 °C of 21.3 cm/s for 30 min. After the cleaning step, pure water flux
with 50 mM NaCl as the feed solution and a cross-flow velocity of the membrane was measured at 27.6 bar (400 psi) and a
of 21.3 cm/s.45 crossflow velocity of 8.5 cm/s to determine the flux recovery
RO Membrane Silica Scaling Tests. Scaling tests with the ratio. The membranes before and after silica scaling were
various membranes were conducted using a bench-scale RO analyzed by SEM, energy-dispersive X-ray (EDX) spectroscopy,
system and a silica-saturated feed solution. The configuration of and ATR-FTIR spectroscopy, to examine the morphology and
the RO system was identical to that used in measuring the chemical composition of silica scale formed on the membrane
membrane transport properties.43,44 surface.
The silica-saturated feed solution was composed of 2.8 mM
Na2SiO3·5H2O, 7.0 mM CaCl2, 3.5 mM MgCl2, and 35 mM
NaCl, simulating brackish groundwater after 70−80% recov-
■ RESULTS AND DISCUSSION
Modified Membranes Exhibit Diverse Surface Proper-
ery.46 A solution containing Na2SiO3·5H2O and NaCl was ties. The membranes tested in this study were analyzed by a
prepared first with the pH of ∼11. Then, CaCl2 and MgCl2 combination of different techniques to characterize key surface
were added after the pH was adjusted to less than 7 in order to properties, namely membrane morphology, surface function-
avoid precipitation of calcium and magnesium silicate salts.13,47 ality, surface hydrophilicity, and surface charge.
The pH was further adjusted to 6.50 ± 0.05. Thermodynamic SEM was employed to observe the morphology of membrane
calculations using PHREEQC48 and the database MINTEQ surfaces (Figure S2). Except for the membrane grafted with PEI
(version 4) showed that silica was the only precipitate formed of 1300 g/mol (PEI-1300), all modified membranes exhibited a
in the feed solution, with a saturation index (defined as the ratio ridge-and-valley surface structure similar to the control
of the ion activity product to the solubility product) of 1.5 for commercial RO membrane, indicating that the polymer films
amorphous silica. formed on the membrane surfaces were ultrathin. The surface
Prior to silica scaling, membranes were compacted overnight roughness of each membrane was further quantified by AFM
using DI water under 31.0 bar (450 psi), after which pure water (Figure S3). Results show that surface modification did not
4399 DOI: 10.1021/acs.est.6b06411
Environ. Sci. Technol. 2017, 51, 4396−4406
Environmental Science & Technology Article

change the membrane surface roughness. The lack of alteration increased the membrane zeta potential dramatically, with the
in surface roughness after modification by PEI-1300 was likely membrane surface charge shifting from negative to positive at
due to the low thickness of the polymer layer relative to the almost every investigated solution pH. The positive charge was
surface roughness of the unmodified (control) membrane. derived from the quaternary ammonium of MTAC31,59 or
Membrane surface functionality was characterized by ATR- abundant amine groups of PEI molecules.60,61 In contrast, the
FTIR spectroscopy. As shown in Figure S4, ATR-FTIR spectra PAA-modified membrane exhibited a more negatively charged
of the tested membranes represented a combination of spectra surface than the control membrane, consistent with its higher
from both the polyamide active layer and the polysulfone surface density of carboxyl groups as revealed by ATR-FTIR
support layer. For example, the absorbance at ∼1670 cm−1 and spectra. The other polymers used in this study generally
∼1540 cm−1 corresponds to N−CO and C−N−H vibrations diminished the negative surface charge as compared to the
of the amide groups,34 while the peak at 1294 cm−1 originates unmodified control membrane. For example, the zeta potential
from the SO stretching in polysulfone.31 A new peak was at near neutral pH changed from ∼−12 mV for the control
observed at 1726 cm−1 for both PSBMA- and PMTAC- membrane to ∼−6 mV for the PSBMA- and PAM-modified
modified membranes, which is attributed to the carbonyl in the membranes. The grafting of these net zero charge polymers
ester group of SBMA and MTAC molecules.31 The PSBMA- covered the underlying negatively charged functional
modified membrane also showed an additional peak at 1039 groups.29,31
cm−1 that arises from the symmetric stretch of sulfonate group Membrane Transport Properties. As shown in Figure
in SBMA.31 Although native carboxyl groups are present in the 2C, most modified membranes exhibited slightly higher water
polyamide active layer due to the hydrolysis of unreacted acyl permeability coefficients (A between 2.6 ± 0.0 and 3.0 ± 0.2 L
chloride groups,49 their density (approximately 1−30 charges m−1 h−1 bar−1) than the control membrane (A of 2.5 ± 0.0 L
nm−250) was too low to be detected by ATR-FTIR for the m−1 h−1 bar−1), probably due to their improved membrane
unmodified control membrane.34 The PAA-modified mem- hydrophilicity. However, both PEI-modified membranes
brane exhibited an increased signal at ∼1720 cm−1 due to C showed reduced water permeability (2.3 ± 0.1 L m−1 h−1
O stretching,51,52 suggesting an increase of carboxyl group bar−1). We noticed that the grafting of hydrophobic PFDT
density on the membrane surface. For the other modified polymer did not decrease membrane water permeability. This
membranes, the characteristic peaks of the grafted polymers counterintuitive phenomenon was attributed to the use of
overlapped with those of the TFC polyamide membrane. ethanol to dissolve PFDT during the membrane modification
Therefore, no difference in the ATR-FTIR spectra was process. Ethanol could swell the polyamide active layer by
observed as compared to the control membrane. enhancing chain flexibility and chain−chain distance, resulting
Membrane surface hydrophilicity was determined by in a larger free volume that facilitates water transport.62 After
measuring the water contact angle of each membrane. As immersing the control membrane in 1:1 ethanol:DI water
shown in Figure 2A, the control membrane was relatively solution for 4 h, the membrane water permeability increased by
hydrophilic with a water contact angle of ∼30°, which was 40% and became significantly higher than that of the PFDT-
lower than reported values for pristine TFC polyamide modified membrane (Figure S5).
membranes (>50°38,53). We attribute this result to proprietary Salt permeability coefficients (B) of the membranes were
polymer coating on the commercial membrane surface.54 After calculated from the measured water flux and salt rejection after
modification with PAM, PSBMA, PEI, and PAA polymers, the accounting for concentration polarization (Figure 2C).45 The
membrane surface became more hydrophilic, with water average salt rejection of the control membrane was 99.0% with
contact angles lower than 20°. The reduction in water contact 50 mM NaCl in the feed solution, while the modified
angle was attributed to the introduction of polar functional membranes exhibited similar or slightly lower salt rejection of
groups (−OC−NH2, −NH2/−NH, and −COO− for PAM, 98.4%−99.1%. Except for the membrane modified with PEI-
PEI, and PAA, respectively) or zwitterionic brush layer (for 1300, the calculated salt permeability coefficients moderately
SBMA), which increased the affinity of water molecules to the increased after membrane modification, reflecting the water
membrane surface.55−57 Conversely, grafting of the PFDT permeability−salt permeability trade-off of TFC polyamide
polymer created a more hydrophobic surface (water contact membranes.63 Therefore, the intrinsic membrane transport
angle >110°) due to the abundance of fluorine groups.36 In properties were not greatly affected by surface modification in
addition, membrane modifications with PMTAC, PDA, and our study.
PDA-Bibb resulted in a slight, but statistically insignificant, Membrane Surface Modification Influences Silica
increase of membrane hydrophilicity as compared to the Scaling. The silica scaling tests were conducted with a silica-
control membrane. saturated feed solution (saturation index of 1.5) in a bench-
Zeta potential indicates the surface charge properties of the scale cross-flow RO system. As shown in Figure S6A, the water
membrane,39 which may impact the adsorption of charged flux of the control membrane decreased gradually with a total
foulants due to electrostatic interactions. The zeta potential of flux decline of ∼15% after 1400 min. No flux recovery was
each membrane was calculated from streaming potential achieved after membrane cleaning (Figure S6B), consistent
measurements as a function of pH.39 As shown in Figure 2B, with our previous finding that silica scaling of polyamide
the control membrane had negative surface charge over the membranes in RO mode was irreversible.2 SEM images
range of investigated solution pH (i.e., pH of 3−9). The revealed that the scaled membrane surface was fully covered
negative surface charge is attributed to the deprotonation of by a layer of particles (Figures S6C and S7), with the ridge-and-
carboxylic groups formed in the interfacial polymerization valley surface structure no longer visible. EDX spectra clearly
process.58 showed the energy peak of Si at 1.74 keV (Figure S6C); no
Membrane surface modification with different polymers signal of Ca or Mg was detected, thus excluding the formation
altered the membrane zeta potential, providing additional of calcium and magnesium scale. Compared to the pristine
evidence for the success of polymer grafting. PMTAC and PEI membrane, increased signals associated with Si−OH (1654
4400 DOI: 10.1021/acs.est.6b06411
Environ. Sci. Technol. 2017, 51, 4396−4406
Environmental Science & Technology Article

Figure 3. Results of silica scaling tests for the different membranes analyzed in this study (definition of each membrane was given in Figure 1). (A)
Representative water flux decline curve of each membrane. (B) Normalized water flux after scaling and after rinsing. The silica scaling tests were
conducted with an initial water flux of 56 ± 2 L m−2 h−1 and a cross-flow velocity of 8.5 cm/s for 1400 min at a temperature of 22 ± 1 °C. The
composition of feed solution included 2.8 mM Na2SiO3, 7 mM CaCl2, 3.5 mM MgCl2, 35 mM NaCl, and pH = 6.50 ± 0.05. The saturation index of
amorphous silica was 1.5 as calculated by PHREEQC and the thermodynamic database MINTEQ (Version 4). The error bars in part B represent
standard deviation and were calculated from triplicate measurements.

Figure 4. Relating the extent of silica scaling (expressed as water flux decline ratio) to (A) membrane water contact angle and (B) free energy for
heterogeneous nucleation (expressed as the wetting function f(θ)). Water flux decline ratio is defined as the percentage of flux decline as compared
to the initial water flux. The value of f(θ) was calculated as described in the Materials and Methods section and the Supporting Information. The
error bars represent standard deviation and were calculated from at least five (for water contact angle) and triplicate (for water flux decline ratio)
measurements. The calculation of f(θ) values involves the average contact angles in water, glycerol, and diiodomethane, as well as the average surface
roughness of the membranes.

cm−1)64 and Si−O−Si bonds (between 1050 and 1100 observed for all other modified membranes after membrane
cm−1)65,66 were detected in the ATR-FTIR spectra after cleaning (Figure 3B), underscoring the irreversibility of silica
membrane scaling (Figure S6D). These analyses unambigu- scaling in RO operation.
ously confirmed the formation of silica scale on the membrane Silica scaling on the membrane surface is governed by both
surface. silica−membrane and silica−silica interactions.2 The membrane
Water flux decline curves of all modified membranes were surface was directly exposed to silica scalants at the initial stage
compared with that of the control membrane (Figure 3A). RO of the scaling tests, when membrane surface properties
membranes modified with different polymers showed varied influenced silica deposition and nucleation by controlling
extents of flux decline due to silica scaling. Since other factors silica−membrane interactions. However, the influence of
potentially influencing the rate of silica scaling (e.g., permeate membrane surface properties diminished as the membrane
flux, trans-membrane pressure, cross-flow velocity, and feed-
surface was progressively covered by a silica layer, after which
water chemistry) were kept the same during the scaling tests for
further accumulation of silica scale was controlled by silica−
all membranes, the observed variation in water flux decline is
attributed to the difference in membrane surface properties. silica interactions. This transition was notably observed for the
Membranes modified with PMTAC and two PEI polymers MTAC- and PEI-modified membranes that have high scaling
experienced the most severe flux decline (∼30% after 1400 propensity. Specifically, these membranes experienced a rapid
min), significantly higher than that for the control membrane flux decline at the beginning of the scaling tests, reducing the
(∼15%, p < 0.015 in paired t test). In contrast, the PAA- water flux by more than 10% within 6 h. The flux decline rate
modified membrane showed the lowest flux decline (∼12%). was then decelerated and became comparable to the other
All other modified membranes displayed a larger water flux membranes after ∼800 min, indicating that the membrane
decline than the control membrane, suggesting that the grafted surfaces were covered by silica scale. In contrast, the PAA-
polymer layers favored silica scale formation. In addition, except modified membrane maintained 99% of the initial water flux
for the PAA-modified membrane, no recovery of water flux was during the first 6 h, suggesting that the initial deposition of
4401 DOI: 10.1021/acs.est.6b06411
Environ. Sci. Technol. 2017, 51, 4396−4406
Environmental Science & Technology Article

Table 1. Interfacial Free Energies and Wetting Function of the Different Membranes Analyzed in This Study
membrane control PAM PSBMA PMTAC PEI-800 PEI-1300 PFDT PAA PDA PDA-Bibb
γmem−wata −1.73 −6.64 −4.70 −2.60 −5.74 −6.71 31.96 −3.59 0.15 −0.88
γmem−SiO2a 1.75 1.23 1.20 1.20 1.44 0.96 24.56 1.29 0.44 0.15
γSiO2−wata −8.84 −8.84 −8.84 −8.84 −8.84 −8.84 −8.84 −8.84 −8.84 −8.84
f(θ)b 0.22 0.009 0.074 0.197 0.025 0.012 0.98 0.128 0.47 0.33
a
γmem−wat, γmem−SiO2, and γSiO2−wat represent the interfacial free energies between membrane surface and water, membrane surface and silica, and silica
and water, respectively (in mJ/m2). bf(θ) is the wetting function that is proportional to the free energy for heterogeneous nucleation. The procedure
of calculating these parameters is detailed in the Materials and Methods section as well as the Supporting Information.

silica species on the membrane surface was effectively reduced The calculated f(θ) values, however, were not correlated to
with PAA polymers. the water flux decline caused by silica scaling (R2 = 0.07, p =
Silica Scaling Is Independent of Membrane Hydro- 0.46, Figure 4B). For example, silica nucleation proceeds
philicity and Free Energy for Heterogeneous Nuclea- against a higher energy barrier on the PMTAC-modified
tion. In order to delineate the relationship between membrane membrane (f(θ) = 0.197) than the PAM- and PSBMA-
hydrophilicity and silica scaling, the water flux decline ratio of modified membranes ( f(θ) < 0.1). Although silica nucleation
each membrane was plotted against the corresponding water was less likely to occur on the PMTAC-modified membrane, a
contact angle (Figure 4A). No correlation was observed (R2 = more severe water flux decline was observed on that membrane
0.03, p = 0.61), indicating that silica scaling is independent of than on the other two membranes (Figure 3). The PAA-
membrane hydrophilicity. For example, although membranes modified membrane, which showed the least extent of silica
modified with PFDT, PDA, and PSBMA exhibited distinct scaling, had a lower f(θ) value than the control membrane. In
water contact angles (119 ± 7°, 26 ± 3°, and <10°, respectively, addition, the PFDT-modified membrane with the highest
Figure 2A), they displayed comparable water flux decline ratios thermodynamic barrier to silica nucleation experienced a larger
(∼20%) after 1400 min scaling tests. flux decline as compared to the control and PAA-modified
An increase in membrane hydrophilicity generally improves membranes.
membrane resistance to organic and biological fouling.22,67 A Our results suggest that although heterogeneous nucleation
strong hydration layer formed on hydrophilic surfaces plays a significant role in silica scaling of TFC polyamide
effectively reduces adsorption of both organic and biological membranes,2,11 the thermodynamic barrier to heterogeneous
foulants (e.g., organic macromolecules,23,38 proteins,23,31,68 and silica nucleation does not predict the rate of silica scaling on
bacterial cells31). However, this strategy was ineffective in RO membranes. It is likely that heterogeneous nucleation is
reducing silica scaling in our study. In fact, modified influenced by both thermodynamic and kinetic factors. A study
membranes with improved hydrophilicity experienced a larger by Wallace et al.21 reported that kinetic drivers were the
water flux decline than the unmodified control membrane in dominating factor in regulating silica nucleation on self-
most cases (except for the PAA-modified membrane). Unlike assembled monolayers. The kinetic variability of silica
organic and biological foulants, the small silicic acid molecules nucleation could arise from varied rates of silicic acid
can diffuse through the hydration layer and grafted polymer attachment and different surface concentrations of ionized
films. Due to their abundant silanol groups, silicic acid silicic acid species.21 As a result, the rate of silica scaling can be
molecules or other silica species can form strong hydrogen significantly enhanced without reducing the energy barrier to
bonding with the grafted polymer and polyamide surface, heterogeneous nucleation. The impact of membrane surface
subsequently initiating the formation of silica nuclei. Our chemistry on the kinetics of silica nucleation will be discussed
hypothesis is further supported by a recent study in which in more detail in the following subsection.
hydrogen bonding between silanol groups and hydrophilic Deposition of colloidal silica from bulk solution (homoge-
polyethylene glycol polymers was observed in aqueous neous nucleation) could also contribute to silica scaling.
solution.69 Therefore, despite the presence of a bound Dynamic light scattering (DLS) was performed to investigate
hydration layer, silicic acids could still reach the membrane particle formation in the feedwater for the silica scaling tests
surface, polymerizing and forming silica scale on the hydro- (detailed procedure of DLS measurements is provided in the
philic polymer brushes and/or the underlying polyamide active Supporting Information). Although a negligible amount of silica
layer. particles was detectable in the feedwater within 24 h, formation
The wetting function, f(θ), was calculated as a proxy to of colloidal silica particles was clearly evident when considering
estimate the energy barrier to heterogeneous silica nucleation concentration polarization (Figure S8). The presence of silica
on each membrane surface.21,40 As shown in Table 1, the particles near the membrane surface could cause water flux
membranes tested in this study showed different values of f(θ), decline by direct deposition, causing the scaling results to
which correspond to varied membrane susceptibility to silica deviate from those predicted by heterogeneous nucleation
nucleation. The highly hydrophilic membranes (e.g., the PAM-, theory.
PSBMA-, and PEI-modified membranes) exhibited lower f(θ) Silica Scaling Is Strongly Dependent on Membrane
values than the unmodified control membrane, suggesting that Surface Charge. As shown in Figure 5, silica scaling exhibited
silica nucleation is more likely to occur on the hydrophilic a strong and statistically significant correlation with membrane
membranes. In contrast, the hydrophobic PFDT-modified surface charge at near neutral pH (R2 = 0.958, p < 0.001).
membrane showed the largest f(θ) value (close to 1), for Membranes with a more positive surface charge exhibited a
which heterogeneous silica nucleation is thermodynamically higher water flux decline due to silica scaling, whereas an
unfavorable. increase of negative surface charge reduced silica scaling. The
4402 DOI: 10.1021/acs.est.6b06411
Environ. Sci. Technol. 2017, 51, 4396−4406
Environmental Science & Technology Article

due to the lower pKa value of silica oligomers (pKa of ∼6.8)


compared to monomeric silicic acids (pKa of ∼9.8).75 Similar to
ionized silicic acid, the resultant ionized silica species are very
reactive, and other silica species react preferentially with those
negatively charged species.75
The negatively charged ionized silica species play a critical
role in the kinetics of silica polymerization, which explains our
observed RO membrane scaling results. As illustrated in Figure
6, a positively charged membrane surface (e.g., the PEI- and
MTAC-modified membranes) attracts negatively charged
ionized silica species via electrostatic interactions. The resultant
high concentration of ionized silica species near the membrane
Figure 5. Relating the extent of silica scaling expressed as water flux surface led to a faster rate of silica polymerization and facilitated
decline ratio to membrane surface charge characteristics quantified by silica scale formation. The importance of positively charged
zeta potential at pH 6. The data were fitted by linear regression, which functional groups in enhancing silica polymerization has been
showed a strong and statistically significant correlation (R2 > 0.95, p < demonstrated in biogenic silica formation. Coradin et al.76
0.001) between membrane surface charge and water flux decline ratio. reported that electrostatic interactions between positively
The error bars represent standard deviation and were calculated from charged peptides and negatively charged silica species
eight (for zeta potential) and triplicate (for water flux decline ratio) promoted silicic acid polymerization. Conversely, more
measurements. Similar correlation was observed when using zeta negatively charged membranes (e.g., the PAA-modified
potential at pH 7 (Figure S9 in the Supporting Information). Note membrane) repel ionized silica species. Thus, the concentration
that our silica scaling experiments were carried out at pH 6.5.
of active ionized silica species was relatively lower near the
membrane surface, resulting in reduced silica scaling.
PEI- and MTAC-modified membranes, which have positive As discussed earlier in the paper, colloidal silica particles
zeta potentials at near neutral pH (Figure 2B), experienced the could form by homogeneous nucleation near the membrane
largest water flux decline in the scaling test (Figure 3). In surface; hence, their role in silica scaling cannot be excluded in
contrast, the PAA-modified membrane with the most negatively the current study. Silica has an isoelectric point at pH ∼ 2,77
charged surface showed the lowest susceptibility to silica rendering the silica particles negatively charged at the near
scaling. The other modified membranes exhibited a higher neutral pH used in our RO scaling experiments (Figure S10).
extent of silica scaling than the control membrane, which was in Thus, colloidal silica particles are attracted by a positively
accordance with their less negative surface charge. charged membrane surface and repelled by a negatively charged
Proposed Mechanism for Silica Scaling. Polymerization surface, which could also support our proposed mechanism for
of silicic acids on the membrane surface, which has been the impact of membrane surface charge on silica scaling.
considered the major mechanism of silica scaling on TFC Implications. We have demonstrated that membrane
polyamide membranes,2,11 is mainly driven by the reaction surface charge, rather than membrane hydrophilicity and free
between ionized and neutral silicic acid molecules.9,70−72 energy of heterogeneous silica nucleation, governed RO silica
Compared with their neutral counterparts, the ionized silicic scaling. Table S2 in the Supporting Information further
acids (with negative charges) react more readily with other illustrates the interplay between membrane surface properties
silicic acid molecules.73 The high reactivity of ionized silicic and the propensity for RO silica scaling. RO membranes with a
acids is probably due to the formation of Si−O bonds between high density of negative surface charge were found to have a
the deprotonated silanol groups and the Si center of silicic acid lesser extent of scaling. However, negatively charged carboxyl
molecules.74 As polymerization of silicic acids proceeds, silica functional groups enhance organic fouling by complexation
species tend to have a higher density of ionized silanol groups, with calcium ions that form intermolecular bridges among

Figure 6. Schematic of proposed mechanism for the role of membrane surface charge in silica scaling. For a positively charged membrane surface, the
negatively charged silica species are attracted to the membrane surface, causing severe decline in water flux. In contrast, a negatively charged
membrane surface repels silica species, resulting in reduced water flux decline during silica scaling. Note that the negatively charged silica species not
only represent monomers and dimmers but also include other silica species such as trimers, oligomers, and polymers.

4403 DOI: 10.1021/acs.est.6b06411


Environ. Sci. Technol. 2017, 51, 4396−4406
Environmental Science & Technology Article

organic foulant molecules.78−80 Accordingly, membranes with technical assistance and gratefully acknowledge the use of SEM
improved silica scaling resistance may be potentially con- and AFM facilities supported by the Yale Institute for
strained by their high organic fouling propensity when treating Nanoscience and Quantum Engineering (YINQE). DLS and
feedwaters with high content of both organic foulants and silica. zeta potential measurements were performed in the Facility for
On the other hand, despite their potential to increase silica Light Scattering (FLS) at Yale University.


scaling, positively charged polymers, such as PEI and MTAC,
have been applied to increase membrane resistance to organic REFERENCES
and biological fouling, respectively.31,60,81 As a result, a trade-off
between antiscaling and antifouling membrane properties may (1) Koo, T.; Lee, Y. J.; Sheikholeslami, R. Silica fouling and cleaning
of reverse osmosis membranes. Desalination 2001, 139 (1−3), 43−56.
exist when designing high-performance RO membranes. When (2) Mi, B. X.; Elimelech, M. Silica scaling and scaling reversibility in
RO membranes are exposed to a mixture of silica and other forward osmosis. Desalination 2013, 312, 75−81.
foulants, their performance is determined by the combined (3) Sheikholeslami, R.; Zhou, S. Performance of RO membranes in
effect of silica−membrane, silica−foulant, and foulant− silica bearing waters. Desalination 2000, 132 (1−3), 337−344.
membrane interactions. Although this trilateral process is (4) Milne, N. A.; O’Reilly, T.; Sanciolo, P.; Ostarcevic, E.; Beighton,
complex and beyond the scope of the current study, it should M.; Taylor, K.; Mullett, M.; Tarquin, A. J.; Gray, S. R. Chemistry of
be considered when designing fouling- and scaling-resistant RO silica scale mitigation for RO desalination with particular reference to
membranes. remote operations. Water Res. 2014, 65, 107−133.


(5) Sahachaiyunta, P.; Koo, T.; Sheikholeslami, R. Effect of several
ASSOCIATED CONTENT inorganic species on silica fouling in RO membranes. Desalination
2002, 144 (1−3), 373−378.
*
S Supporting Information (6) Lisitsin, D.; Hasson, D.; Semiat, R. Critical flux detection in a
The Supporting Information is available free of charge on the silica scaling RO system. Desalination 2005, 186 (1−3), 311−318.
ACS Publications website at DOI: 10.1021/acs.est.6b06411. (7) Semiat, R.; Sutzkover, I.; Hasson, D. Scaling of RO membranes
Details on the calculation of the wetting function ( f(θ)) from silica supersaturated solutions. Desalination 2003, 157 (1−3),
169−191.
to estimate the free energy for heterogeneous nucleation,
(8) Den, W.; Wang, C. J. Removal of silica from brackish water by
DLS measurement of the feedwater used in the silica electrocoagulation pretreatment to prevent fouling of reverse osmosis
scaling test, schematic of heterogeneous silica nucleation membranes. Sep. Purif. Technol. 2008, 59 (3), 318−325.
on membrane surfaces (Figure S1), SEM micrographs of (9) Makrides, A. C.; Turner, M.; Slaughter, J. Condensation of silica
the control and modified membranes (Figure S2), AFM from supersaturated silicic-acid solutions. J. Colloid Interface Sci. 1980,
analysis of the membranes (Figure S3), ATR-FTIR 73 (2), 345−367.
spectra of the membranes (Figure S4), effects of ethanol (10) Iler, R. K. The Chemistry of Silica Silica: Solubility, Polymerization,
treatment on transport properties of the membranes Colloid and Surface Properties, and Biochemistry; John Wiley and Sons:
(Figure S5), silica scaling results of the control New York, 1979.
membrane (Figure S6), SEM micrographs of membranes (11) Xie, M.; Gray, S. R. Silica scaling in forward osmosis: From
solution to membrane interface. Water Res. 2017, 108, 232−229.
after silica scaling (Figure S7), DLS results (Figure S8), (12) Kempter, A.; Gaedt, T.; Boyko, V.; Nied, S.; Hirsch, K. New
correlation of the extent of silica scaling to membrane insights into silica scaling on RO-membranes. Desalin. Water Treat.
surface charge quantified by zeta potential at pH 7 2013, 51 (4−6), 899−907.
(Figure S9), zeta potential of silica particles formed in (13) Neofotistou, E.; Demadis, K. D. Use of antiscalants for
the RO feed solution at the membrane surface after mitigation of silica (SiO2) fouling and deposition: Fundamentals and
considering the effect of concentration polarization applications in desalination systems. Desalination 2004, 167 (1−3),
(Figure S10), contact angles and surface energy data of 257−272.
the membranes (Table S1), membrane surface properties (14) Ning, R. Y.; Troyer, T. L.; Tominello, R. S. Antiscalants for near
and silica scaling propensity of modified membranes as complete recovery of water with tandem RO process. Desalin. Water
compared to the control membrane (Table S2) (PDF) Treat. 2009, 9 (1−3), 92−95.


(15) Turek, M.; Mitko, K.; Piotrowski, K.; Dydo, P.; Laskowska, E.;
Jakóbik-Kolon, A. Prospects for high water recovery membrane
AUTHOR INFORMATION desalination. Desalination 2017, 401, 180−189.
Corresponding Author (16) Sweity, A.; Zere, T. R.; David, I.; Bason, S.; Oren, Y.; Ronen, Z.;
*Phone: +1 (203) 432-2789. E-mail: menachem.elimelech@ Herzberg, M. Side effects of antiscalants on biofouling of reverse
osmosis membranes in brackish water desalination. J. Membr. Sci.
yale.edu.
2015, 481, 172−187.
ORCID (17) Subramani, A.; Jacangelo, J. G. Treatment technologies for
Menachem Elimelech: 0000-0003-4186-1563 reverse osmosis concentrate volume minimization: A review. Sep. Purif.
Author Contributions Technol. 2014, 122, 472−489.
∥ (18) Mukhopadhyay, D. Method and apparatus for high efficiency
T.T. and S.Z. contributed equally. reverse osmosis operation. US Patent 6537456 B2, Mar 25, 2003.
Notes (19) Gill, J. S. Inhibition of silica-silicate deposit in industrial waters.
The authors declare no competing financial interest. Colloids Surf., A 1993, 74 (1), 101−106.

■ ACKNOWLEDGMENTS
We acknowledge the support received from the National
(20) Mickley, M. Survey of High-Recovery and Zero Liquid Discharge
Technologies for Water Utilities; WRF-02-006a; WateReuse Foundation:
Alexandria, VA, 2008.
(21) Wallace, A. F.; DeYoreo, J. J.; Dove, P. M. Kinetics of silica
Science Foundation through the Engineering Research Center nucleation on carboxyl- and amine-terminated surfaces: Insights for
for Nanotechnology-Enabled Water Treatment (ERC- biomineralization. J. Am. Chem. Soc. 2009, 131 (14), 5244−5250.
1449500) and a fellowship awarded by the China Scholarship (22) Rana, D.; Matsuura, T. Surface modifications for antifouling
Council to S.Z. We also thank Dr. Michael Rooks for his membranes. Chem. Rev. 2010, 110 (4), 2448−2471.

4404 DOI: 10.1021/acs.est.6b06411


Environ. Sci. Technol. 2017, 51, 4396−4406
Environmental Science & Technology Article

(23) Tiraferri, A.; Kang, Y.; Giannelis, E. P.; Elimelech, M. Highly (42) Vanoss, C. J.; Chaudhury, M. K.; Good, R. J. Interfacial Lifshitz-
hydrophilic thin-film composite forward osmosis membranes function- Vanderwaals and polar interactions in macroscopic systems. Chem. Rev.
alized with surface-tailored nanoparticles. ACS Appl. Mater. Interfaces 1988, 88 (6), 927−941.
2012, 4 (9), 5044−5053. (43) Mi, B. X.; Elimelech, M. Gypsum scaling and cleaning in forward
(24) Kang, G. D.; Cao, Y. M. Development of antifouling reverse osmosis: Measurements and mechanisms. Environ. Sci. Technol. 2010,
osmosis membranes for water treatment: A review. Water Res. 2012, 44 (6), 2022−2028.
46 (3), 584−600. (44) Xie, M.; Lee, J.; Nghiem, L. D.; Elimelech, M. Role of pressure
(25) Ben-Sasson, M.; Lu, X. L.; Bar-Zeev, E.; Zodrow, K. R.; Nejati, in organic fouling in forward osmosis and reverse osmosis. J. Membr.
S.; Qi, G. G.; Giannelis, E. P.; Elimelech, M. In situ formation of silver Sci. 2015, 493, 748−754.
nanoparticles on thin-film composite reverse osmosis membranes for (45) Yip, N. Y.; Tiraferri, A.; Phillip, W. A.; Schiffman, J. D.;
biofouling mitigation. Water Res. 2014, 62, 260−270. Elimelech, M. High performance thin-film composite forward osmosis
(26) Ben-Sasson, M.; Lu, X. L.; Nejati, S.; Jaramillo, H.; Elimelech, membrane. Environ. Sci. Technol. 2010, 44 (10), 3812−3818.
M. In situ surface functionalization of reverse osmosis membranes with (46) Bond, R.; Veerapaneni, S. Zero Liquid Discharge for Inland
biocidal copper nanoparticles. Desalination 2016, 388, 1−8. Desalination; No. 500-01-040; Awwa Research Foundation: Denver,
(27) Ben-Sasson, M.; Zodrow, K. R.; Qi, G. G.; Kang, Y.; Giannelis, CO, 2007.
E. P.; Elimelech, M. Surface functionalization of thin-film composite (47) Sheikholeslami, R.; Bright, J. Silica and metals removal by
membranes with copper nanoparticles for antimicrobial surface pretreatment to prevent fouling of reverse osmosis membranes.
properties. Environ. Sci. Technol. 2014, 48 (1), 384−393. Desalination 2002, 143 (3), 255−267.
(28) Perreault, F.; Tousley, M. E.; Elimelech, M. Thin-film composite (48) Parkhurst, D.; Appelo, C. A. J. User’s Guide to PHREEQC
polyamide membranes functionalized with biocidal graphene oxide (version 2) - A computer program for speciation, batch-reaction, one-
nanosheets. Environ. Sci. Technol. Lett. 2014, 1 (1), 71−76. dimensional transport, and inverse geochemical calculations; U.S.
(29) Lin, N. H.; Kim, M. M.; Lewis, G. T.; Cohen, Y. Polymer surface Geological Survey Water-Resources Investigations Report 99-4259;
nano-structuring of reverse osmosis membranes for fouling resistance U.S. Geological Survey: 1999.
and improved flux performance. J. Mater. Chem. 2010, 20 (22), 4642− (49) Freger, V. Nanoscale heterogeneity of polyamide membranes
4652. formed by interfacial polymerization. Langmuir 2003, 19 (11), 4791−
(30) Kim, M. M.; Lin, N. H.; Lewis, G. T.; Cohen, Y. Surface nano- 4797.
structuring of reverse osmosis membranes via atmospheric pressure (50) Tiraferri, A.; Elimelech, M. Direct quantification of negatively
plasma-induced graft polymerization for reduction of mineral scaling charged functional groups on membrane surfaces. J. Membr. Sci. 2012,
propensity. J. Membr. Sci. 2010, 354 (1−2), 142−149. 389, 499−508.
(31) Ye, G.; Lee, J. H.; Perreault, F.; Elimelech, M. Controlled (51) Smitha, B.; Sridhar, S.; Khan, A. A. Polyelectrolyte complexes of
chitosan and poly(acrylic acid) as proton exchange membranes for fuel
architecture of dual-functional block copolymer brushes on thin-film
cells. Macromolecules 2004, 37 (6), 2233−2239.
composite membranes for integrated ″defending″ and ″attacking″
(52) Ma, S. M.; Yuan, Q.; Zhang, X. J.; Yang, S. G.; Xu, J. Solvent
strategies against biofouling. ACS Appl. Mater. Interfaces 2015, 7 (41),
effect on hydrogen-bonded thin film of poly(vinylpyrrolidone) and
23069−23079.
poly(acrylic acid) prepared by layer-by-layer assembly. Colloids Surf., A
(32) Barbey, R.; Lavanant, L.; Paripovic, D.; Schuwer, N.; Sugnaux,
2015, 471, 11−18.
C.; Tugulu, S.; Klok, H. A. Polymer brushes via surface-initiated
(53) Hu, Y. T.; Lu, K.; Yan, F.; Shi, Y. L.; Yu, P. P.; Yu, S. C.; Li, S.
controlled radical polymerization: synthesis, characterization, proper- H.; Gao, C. J. Enhancing the performance of aromatic polyamide
ties, and applications. Chem. Rev. 2009, 109 (11), 5437−5527. reverse osmosis membrane by surface modification via covalent
(33) Pour, S. N.; Ghugare, S. V.; Wiens, R.; Gough, K.; Liu, S. attachment of polyvinyl alcohol (PVA). J. Membr. Sci. 2016, 501, 209−
Controlled in situ formation of polyacrylamide hydrogel on PET 219.
surface via SI-ARGET-ATRP for wound dressings. Appl. Surf. Sci. (54) Tang, C. Y. Y.; Kwon, Y. N.; Leckie, J. O. Probing the nano- and
2015, 349, 695−704. micro-scales of reverse osmosis membranes - A comprehensive
(34) Belfer, S.; Purinson, Y.; Kedem, O. Surface modification of characterization of physiochemical properties of uncoated and coated
commercial polyamide reverse osmosis membranes by radical grafting: membranes by XPS, TEM, ATR-FTIR, and streaming potential
An ATR-FTIR study. Acta Polym. 1998, 49 (10−11), 574−582. measurements. J. Membr. Sci. 2007, 287 (1), 146−156.
(35) Tian, Y.; Cao, Y.; Wang, Y.; Yang, W.; Feng, J. Realizing (55) Yang, H.-C.; Liao, K.-J.; Huang, H.; Wu, Q.-Y.; Wan, L.-S.; Xu,
ultrahigh modulus and high strength of macroscopic graphene oxide Z.-K. Mussel-inspired modification of a polymer membrane for ultra-
papers through crosslinking of mussel-inspired polymers. Adv. Mater. high water permeability and oil-in-water emulsion separation. J. Mater.
2013, 25 (21), 2980−2983. Chem. A 2014, 2 (26), 10225−10230.
(36) Zhang, L.; Wu, J.; Wang, Y.; Long, Y.; Zhao, N.; Xu, J. (56) Laschewsky, A. Structures and synthesis of zwitterionic
Combination of bioinspiration: A general route to superhydrophobic polymers. Polymers 2014, 6 (5), 1544.
particles. J. Am. Chem. Soc. 2012, 134 (24), 9879−9881. (57) Ding, B.; Yamazaki, M.; Shiratori, S. Electrospun fibrous
(37) Yang, J.; Stuart, M. A. C.; Kamperman, M. Jack of all trades: polyacrylic acid membrane-based gas sensors. Sens. Actuators, B 2005,
Versatile catechol crosslinking mechanisms. Chem. Soc. Rev. 2014, 43 106 (1), 477−483.
(24), 8271−8298. (58) Elimelech, M.; Chen, W. H.; Waypa, J. J. Measuring the zeta
(38) Lu, X. L.; Castrillon, S. R. V.; Shaffer, D. L.; Ma, J.; Elimelech, (electrokinetic) potential of reverse-osmosis membranes by a
M. In Situ surface chemical modification of thin-film composite streaming potential analyzer. Desalination 1994, 95 (3), 269−286.
forward osmosis membranes for enhanced organic fouling resistance. (59) Sun, G. X.; Zhang, M. Z.; He, J. L.; Ni, P. H. Synthesis of
Environ. Sci. Technol. 2013, 47 (21), 12219−12228. amphiphilic cationic copolymers poly[2-(methacryloyloxy)ethyl trime-
(39) Walker, S. L.; Bhattacharjee, S.; Hoek, E. M. V.; Elimelech, M. A thylammonium chloride-co-stearyl methacrylate] and their self-
novel asymmetric clamping cell for measuring streaming potential of assembly behavior in water and water-ethanol mixtures. J. Polym. Sci.,
flat surfaces. Langmuir 2002, 18 (6), 2193−2198. Part A: Polym. Chem. 2009, 47 (18), 4670−4684.
(40) Wu, W. J.; Nancollas, G. H. Interfacial free energies and (60) Xu, J.; Wang, Z.; Wang, J. X.; Wang, S. C. Positively charged
crystallization in aqueous media. J. Colloid Interface Sci. 1996, 182 (2), aromatic polyamide reverse osmosis membrane with high anti-fouling
365−373. property prepared by polyethylenimine grafting. Desalination 2015,
(41) Förster, M.; Bohnet, M. Influence of the interfacial free energy 365, 398−406.
crystal/heat transfer surface on the induction period during fouling. (61) Zhang, R. N.; Su, Y. L.; Zhao, X. T.; Li, Y. F.; Zhao, J. J.; Jiang,
Int. J. Therm. Sci. 1999, 38, 944−954. Z. Y. A novel positively charged composite nanofiltration membrane

4405 DOI: 10.1021/acs.est.6b06411


Environ. Sci. Technol. 2017, 51, 4396−4406
Environmental Science & Technology Article

prepared by bio-inspired adhesion of polydopamine and surface


grafting of poly(ethylene imine). J. Membr. Sci. 2014, 470, 9−17.
(62) Zhang, S.; Fu, F.; Chung, T.-S. Substrate modifications and
alcohol treatment on thin film composite membranes for osmotic
power. Chem. Eng. Sci. 2013, 87, 40−50.
(63) Werber, J. R.; Deshmukh, A.; Elimelech, M. The critical need for
increased selectivity, not increased water permeability, for desalination
membranes. Environ. Sci. Technol. Lett. 2016, 3 (4), 112−120.
(64) Liu, C.; Wu, Y. N.; Yu, A. M.; Li, F. T. Cooperative fabrication
of ternary nanofibers with remarkable solvent and temperature
resistance by electrospinning. RSC Adv. 2014, 4 (59), 31400−31408.
(65) Christl, I.; Brechbuhl, Y.; Graf, M.; Kretzschmar, R. Polymer-
ization of silicate on hematite surfaces and its influence on arsenic
sorption. Environ. Sci. Technol. 2012, 46 (24), 13235−13243.
(66) Tripp, C. P.; Hair, M. L. Reaction of chloromethylsilanes with
silica - A low-frequency infrared study. Langmuir 1991, 7 (5), 923−
927.
(67) Zhang, R.; Liu, Y.; He, M.; Su, Y.; Zhao, X.; Elimelech, M.;
Jiang, Z. Antifouling membranes for sustainable water purification:
Strategies and mechanisms. Chem. Soc. Rev. 2016, 45 (21), 5888−
5924.
(68) Ostuni, E.; Chapman, R. G.; Holmlin, R. E.; Takayama, S.;
Whitesides, G. M. A survey of structure-property relationships of
surfaces that resist the adsorption of protein. Langmuir 2001, 17 (18),
5605−5620.
(69) Preari, M.; Spinde, K.; Lazic, J.; Brunner, E.; Demadis, K. D.
Bioinspired insights into silicic acid stabilization mechanisms: The
dominant role of polyethylene glycol-induced hydrogen bonding. J.
Am. Chem. Soc. 2014, 136 (11), 4236−4244.
(70) Greenberg, S. A.; Sinclair, D. The polymerization of silicic acid.
J. Phys. Chem. 1955, 59 (5), 435−440.
(71) Ashley, K. D.; Innes, W. B. Control of physical structure of
silica-alumina catalyst. Ind. Eng. Chem. 1952, 44 (12), 2857−2863.
(72) Tarutani, T. Polymerization of silicic-acid - A Review. Anal. Sci.
1989, 5 (3), 245−252.
(73) Fleming, B. A. Kinetics of reaction between silicic-acid and
amorphous silica surfaces in NaCl solutions. J. Colloid Interface Sci.
1986, 110 (1), 40−64.
(74) Garofalini, S. H.; Martin, G. Molecular simulations of the
polymerization of silicic-acid molecules and network formation. J. Phys.
Chem. 1994, 98 (4), 1311−1316.
(75) Belton, D. J.; Deschaume, O.; Perry, C. C. An overview of the
fundamentals of the chemistry of silica with relevance to
biosilicification and technological advances. FEBS J. 2012, 279 (10),
1710−1720.
(76) Coradin, T.; Durupthy, O.; Livage, J. Interactions of amino-
containing peptides with sodium silicate and colloidal silica: A
biomimetic approach of silicification. Langmuir 2002, 18 (6), 2331−
2336.
(77) Elimelech, M.; Gregory, J.; Jia, X.; Williams, R. A. Particle
Deposition and Aggregation: Measurement, Modelling and Simulation;
Butterworth Heinemann: Oxford, UK, 1995.
(78) Lee, S.; Ang, W. S.; Elimelech, M. Fouling of reverse osmosis
membranes by hydrophilic organic matter: implications for water
reuse. Desalination 2006, 187 (1−3), 313−321.
(79) Lee, S.; Elimelech, M. Relating organic fouling of reverse
osmosis membranes to intermolecular adhesion forces. Environ. Sci.
Technol. 2006, 40 (3), 980−987.
(80) Li, Q. L.; Elimelech, M. Organic fouling and chemical cleaning
of nanofiltration membranes: Measurements and mechanisms. Environ.
Sci. Technol. 2004, 38 (17), 4683−4693.
(81) Blok, A. J.; Chhasatia, R.; Dilag, J.; Ellis, A. V. Surface initiated
polydopamine grafted poly([2-(methacryoyloxy)ethyl]-
trimethylammonium chloride) coatings to produce reverse osmosis
desalination membranes with anti-biofouling properties. J. Membr. Sci.
2014, 468, 216−223.

4406 DOI: 10.1021/acs.est.6b06411


Environ. Sci. Technol. 2017, 51, 4396−4406

View publication stats

You might also like