journal of materials chemistry a - segiuniversity.edu.my · of the membrane in nf processes. 2...
TRANSCRIPT
Journal ofMaterials Chemistry A
PAPER
Publ
ishe
d on
12
Febr
uary
201
6. D
ownl
oade
d by
Nat
iona
l Uni
vers
ity o
f Si
ngap
ore
on 2
5/02
/201
6 14
:31:
33.
View Article OnlineView Journal
A practical appro
aAdvanced Membrane Technology Research
Malaysia, 81310 Skudai, Johor, Malaysia.
yahoo.com; [email protected]; Tel:bInstitute for Sustainability and Innovation
Victoria University, Werribee Campus, PO BocIndustrial Membrane Research Laboratory,
Engineering, University of Ottawa, 161 Lo
CanadadDepartment of Chemical Engineering, Lee
Science, Universiti Tunku Abdul Rahman, CeDiscipline of Chemical Engineering, Faculty
SEGi University, 47810 Petaling Jaya, Selan
† Electronic supplementary informa10.1039/c5ta09252c
Cite this: DOI: 10.1039/c5ta09252c
Received 16th November 2015Accepted 11th February 2016
DOI: 10.1039/c5ta09252c
www.rsc.org/MaterialsA
This journal is © The Royal Society of
ach to synthesize polyamide thinfilm nanocomposite (TFN) membranes withimproved separation properties for water/wastewater treatment†
G. S. Lai,a W. J. Lau,*a S. R. Gray,b T. Matsuura,ac R. Jamshidi Gohari,a
M. N. Subramanian,a S. O. Lai,d C. S. Ong,e A. F. Ismail,a D. Emazadahac
and M. Ghanbaria
Thin film nanocomposite (TFN) membranes containing 0.05 or 0.10 w/v% surface-functionalized titanate
nanotubes (TNTs) in a polyamide selective layer were synthesized via interfacial polymerization of
piperazine (PIP) and trimesoyl chloride (TMC) monomers. Nanomaterials were dispersed into the
monomer solution using two different approaches. In the first one, the functionalized TNTs were
dispersed into the amine aqueous solution, while in the second approach the same nanomaterials were
dispersed in TMC organic solution. The TFN membranes were characterized and compared with
a control thin film composite (TFC) membrane to investigate the effect of nanofiller loadings and the
fabrication approach on membrane properties. Results showed that introducing nanofillers into the
organic phase was more effective to synthesize a TFN membrane of greater separation performance as
the use of a rubber roller to remove aqueous solution from the substrate surface could cause the loss of
a significant amount of nanofillers, which further affected the polyamide layer integrity. It was also found
that the incorporation of a high nanofiller loading tended to interfere with interfacial polymerization and
weaken the bonds between monomer blocks, resulting in poor polyamide–nanotube integrity.
Compared to the TFC membrane, the TFN membrane made of 2% PIP and 0.15% TMC with 0.5%
nanofiller incorporation could achieve greater water flux (7.5 vs. 5.4 L m�2 h�1 bar�1) and Na2SO4
rejection (96.4% vs. 86%) while exhibiting higher resistance against fouling by protein and dye.
1 Introduction
Polyamide (PA) thin lm nanocomposite (TFN) membraneswere reported for the rst time by Hoek's research group in 2007as a new generation of composite membranes to overcomepermeability–selectivity trade-off of thin lm composite (TFC)membranes for reverse osmosis applications.1 By embeddinga small quantity of zeolite nanoparticles into the PA selective
Centre (AMTEC), Universiti Teknologi
E-mail: [email protected]; lau_woeijye@
+60 75535926
(ISI), College of Engineering and Science,
x 14428, Melbourne, VIC 8001, Australia
Department of Chemical and Biological
uis Pasteur St, Ottawa, ON K1N 6N5,
Kong Chian Faculty of Engineering and
heras, 43000 Kajang, Selangor, Malaysia
of Engineering and the Built Environment,
gor, Malaysia
tion (ESI) available. See DOI:
Chemistry 2016
layer of TFC membranes, it was reported that the waterpermeability of the resulting composite membrane could beincreased almost by an order of magnitude without compro-mising salt rejection. The authors attributed the promisingmembrane performance to the unique characteristics of zeolitethat offer preferential ow paths for water molecules through itssuper-hydrophilic surface and mesoporous structure.
To date, there are more than 100 relevant articles docu-mented in the literature reporting the separation performanceof PA TFN membranes for various processes. A recent reviewauthored by Lau et al.2 has summarized the effects of varioustypes of inorganic nanomaterials on the intrinsic properties andperformances of PA TFN membranes for separation processescovering not only aqueous-based media like nanoltration (NF),reverse osmosis (RO) and forward osmosis (FO), but also for thetreatment of organic solvents as well as pervaporationprocesses. Depending on the types and characteristics ofnanomaterials used, the resultant TFN membrane could showattractive features associated with bacterial, fouling and chlo-rine resistance, in addition to enhancement in water ux andsalt rejection.3–6
J. Mater. Chem. A
Tab
le1
Differentap
proac
hesusedin
preparingpolyam
idethin
film
nan
oco
mposite
(PATFN
)membranes
Mem
bran
eaNan
olle
rsus
edb(loa
ding,
w/v%)
Solution
that
nan
olle
rsare
adde
d
Solution
sfortheIP
processc
App
roachus
edto
remove
excess
aque
oussolution
onthesu
pportsu
rface
Cou
ntry/year
(referen
ce)
Aqu
eous
(w/v%)
Organ
ic(w
/v%)
RO
NaA
zeolite(0.04–0.4%
)Organ
ic2%
MPD
0.1%
TMCin
hexan
eRub
berrolle
rUSA
/200
7(ref.1
)RO
Carbo
nnan
otub
es(0.001
–0.1%)
Aqu
eous
2%MPD
0.1%
TMCin
hexan
eN/A
China/20
14(ref.5
)RO
Carbo
nnan
otub
e/grap
hen
eoxide
mixture
(0.000
2–0.03
%)
Aqu
eous
3%MPD
0.1%
TMCin
hexan
eAirkn
ife
Rep
ublicof
Korea/201
5(ref.7
)RO
Linde
type
Azeolite(0.2%)
Organ
ic2.3–3.2%
MPD
0.1–0.13
%TMCin
Isop
ar-G
Airkn
ife
USA
/200
9(ref.8
)NF
Ag(0–10%
)Aqu
eous
2%MPD
0.1%
TMCin
hexan
eRub
berrolle
rCan
ada/20
12(ref.9
)NF
SiO2(0–1.0%)
Aqu
eous
0.2–2%
PIP
0.05
–0.3%
TMCin
hexan
eAirkn
ife
China/20
12(ref.1
0)NF
Mesop
orou
sSiO2(0–0.07%
)Aqu
eous
1%PIP&0.2%
NaO
H0.15
%TMCin
hexan
eSo
rubb
errolle
rChina/20
13(ref.1
1)FO
SiO2(0.01–
0.1%
)Aqu
eous
2%MPD
0.1%
TMCin
hexan
eRub
berrolle
rIran
/201
4(ref.1
2)
aRO–reverseosmosis,N
F–nan
oltration
,andFO
–forw
ardosmosis.b
SiO2–silicondioxide,an
dAg–silver.c
MPD
–m-phen
ylen
ediamine,TMC–trim
esoylchloride
,PIP
–pipe
razine,an
dNaO
H–sodium
hyd
roxide.
Journal of Materials Chemistry A Paper
Publ
ishe
d on
12
Febr
uary
201
6. D
ownl
oade
d by
Nat
iona
l Uni
vers
ity o
f Si
ngap
ore
on 2
5/02
/201
6 14
:31:
33.
View Article Online
Although the potential of PA TFN membranes for watertreatment processes in particular has been demonstrated bymany researchers over the past 8–9 years, there remain severalconcerns related to TFN membrane making. One of them is thedifferent approaches used by researchers to introduce nano-materials during interfacial polymerization processes. As can beseen from Table 1, inorganic nanomaterials could be addedeither into amine aqueous or acyl chloride organic solution inthe process of PA synthesis.1,5,7–12 However, owing to thehydrophilic nature of nanomaterials, most research studieshave preferred to introduce nanomaterials into amine aqueoussolution. This is because of the difficulties in producinga homogenous mixture in a non-polar organic phase. Sinceremoving excess amine solution from the substrate surface(using a rubber roller or air knife) is a necessary step duringinterfacial polymerization processes, a large amount of nano-materials could be removed together with the amine solution(in the case of nanomaterial–aqueous mixture), leaving onlya small amount of nanomaterials in the substrate. As a result,the positive features of nanomaterials such as hydrophilicity,anti-fouling properties and surface charge density are likely tobe reduced in the synthesized PA layer. Many researchers,however, did not take into consideration the inefficiency of thisapproach in making TFN membranes.
In this work, we have reported on synthesis methods for PATFN membranes by adding inorganic functionalized nano-materials into aqueous or organic solution, respectively. Themain objective of the work was to compare the efficiency ofapproaches used for preparing PA TFN membranes for water/wastewater treatment processes. Titanate nanotubes (TNTs)were selected as the nanollers owing to their hydrophilicproperties and large surface area, as well as the existence ofnarrow channels that might promote water transport throughthemembrane.13,14 To improve the dispersion quality of TNTs inorganic solution, a silane agent was used to modify the surfaceof TNTs. It was also an objective of this work to study howdifferent concentrations of piperazine (PIP) and trimesoylchloride (TMC) monomers affect the properties of the PAnanocomposite layer, which will in turn alter the performanceof the membrane in NF processes.
2 Experimental2.1 Materials
Polysulfone (PSF, Udel® P-3500) in pellet form, purchased fromSolvay Specialty Polymers, USA, was used to fabricate a micro-porous substrate for TFC/TFN membranes. Poly-vinylpyrrolidone (PVP) K29-32 (MW: 58 000 g mol�1) used asa pore forming agent during substrate fabrication waspurchased from Acros Organics. Titanium dioxide (TiO2)Degussa P25 nanoparticles supplied by Evonik Industries wereused as received to synthesize TNTs. 1-(2-Amino-ethyl)-3-ami-nopropyl trimethoxysilane (AAPTS) from Merck was used tomodify the TNT surface to produce amino-functionalized tita-nate nanotubes (NH2-TNTs). Piperazine (PIP) and trimesoylchloride (TMC) purchased from Acros Organics and Merck,respectively, were used to form a PA layer on top of the PSF
J. Mater. Chem. A This journal is © The Royal Society of Chemistry 2016
Table 2 Properties of aqueous and organic solutions used for TFC andTFN membrane preparation
Membrane
NH2-TNTs (wt/v%)
PIP (wt/v%) TMC (wt/v%)In aqueous In cyclohexane
TFC (control) — — 2 0.15TFNaq-0.05 0.05 —TFNaq-0.10 0.10 —TFNcyclo-0.05 — 0.05TFNcyclo-0.10 — 0.10
Table 3 Composition of monomer solutions used for preparing TFNmembranes
MembraneaNH2-TNTs incyclohexane (wt/v%) PIP (wt/v%) TMC (wt/v%)
TFNPIP-1.5 0.05 1.5 0.15TFNPIP-2.5 2.5 0.15TFNTMC-0.1 2.0 0.1TFNTMC-0.2 2.0 0.2
a TFNcyclo-0.05 shown in Table 2 can also be named as either TFNPIP-2.0 orTFNTMC-0.15.
Paper Journal of Materials Chemistry A
Publ
ishe
d on
12
Febr
uary
201
6. D
ownl
oade
d by
Nat
iona
l Uni
vers
ity o
f Si
ngap
ore
on 2
5/02
/201
6 14
:31:
33.
View Article Online
substrate. N-Methyl-2-pyrrolidone (purity > 99.5%) and cyclo-hexane supplied by Acros Organics and Merck, respectively,were used without further purication. Sodium sulfate (Na2SO4,Riedel-de-Haen), sodium chloride (NaCl, Merck), bovine serumalbumin (BSA, Sigma-Aldrich) and Reactive Black 5 (RB5, MW ¼991 g mol�1, Sigma-Aldrich) were used as test solutes formembrane ux and rejection determination. The feed solutioncontaining the respective test solute was prepared by dissolvingthe solute in Milli-Q® water. Other chemicals used in this workwere of analytical grade and were used as received withoutpurication.
2.2 Synthesis and modication of TNTs
TNTs were synthesized using the alkaline hydrothermalmethod. First, 3 g of TiO2 nanoparticles was added into 100 mLof 10 M NaOH aqueous solution. Aer vigorous stirring for 12 h,the milk-like mixture was moved into a 150 mL Teon-linedhydrothermal reactor and heated at 180 �C in an oven for 24 h.Aer the heating process, the reactor was allowed to cool toambient temperature naturally before the white powder wascollected by a centrifugation process. The white product wasfurther washed with 0.5 M HCl solution followed by pure waterrinsing until the pH of washing solution became close to 7. Thewhite powder of TNTs was then dried overnight at 100 �C fol-lowed by grinding into ne powder. Surface modication ofTNTs was then performed on the ne powder according to thedescribed procedure as follows. 1.0 g of TNTs was added into 50mL toluene containing 1.0 g of AAPTS. The mixture was stirredat 95 �C overnight followed by ltration and subsequentwashing using toluene, ethanol, ethanol/water (1 : 1 v/v) andwater to produce NH2-TNTs. Finally, the modied TNTs weredried at 100 �C in an oven to completely remove the residualwater. To compare the dispersion quality of TNTs before andaer surface modication in the organic phase, 0.5% (w/v) TNTsand NH2-TNTs were added respectively into cyclohexane, andphotos were taken to examine the settling behavior of nano-materials for up to 90 min.
2.3 Preparation of TFC and TFN membranes
All the composite membranes were prepared using the in-housemade PSF microporous substrate. The substrate prepared viathe phase inversion method possessed a molecular weight cut-off (MWCO) of 47.5 kDa (see Fig. S1†) with a pure water ux andsurface contact angle recorded to be �150 L m�2 h�1 bar�1 and78.2�, respectively. The MWCO of the substrate prepared was inthe range of 30–50 kDa as reported in the literature.15,16 Thesynthesis of the PA layer was performed via the interfacialpolymerization technique that used PIP and TMC asmonomers.Table 2 shows the amounts of NH2-TNTs present in the solu-tions during interfacial polymerization together with theconcentration of monomers used in aqueous and organicsolutions. A TFC membrane (without NH2-TNTs) was alsoprepared for comparison purpose. The interfacial polymeriza-tion process was initiated by pouring 2% (w/v) PIP aqueoussolution on the top surface of the substrate. The aqueoussolution was then held horizontally on the substrate surface for
This journal is © The Royal Society of Chemistry 2016
2 min to ensure the penetration of aqueous solution into thesubstrate pores. The excess aqueous solution was drained fromthe substrate surface by using a so rubber roller. Secondaryorganic solution that contained 0.15% (w/v) TMC was pouredonto the same substrate surface followed by draining off aer10 s of contact time. The unreacted monomers were removedfrom themembrane surface by rinsing the membrane with purewater. The resultant TFCmembrane was then post-treated in anoven at 60 �C for 8 min. Finally, the membrane was stored ina pure water container until use.
TFN membranes were synthesized exactly as describedabove, except that the NH2-TNTs (0.05 or 0.10%, w/v) weredispersed either in aqueous or organic solution (see Table 2).Prior to the interfacial polymerization reaction, the solutionscontaining nanollers were ultrasonicated for 30 min and usedimmediately aer the ultrasonication process.17,18 Based on thequantity of the used NH2-TNTs and their presence either inaqueous or organic solution, the prepared compositemembranes are labeled as TFC (control), TFNaq-0.05, TFNaq-0.10,TFNcyclo-0.05 and TFNcyclo-0.10 membranes. To further study theeffect of the monomer concentration on the characteristics ofPA nanocomposites, the conditions used to prepare the TFNcyclo-0.05
membrane were further varied by changing the concentration ofPIP and TMC without altering the quantity of NH2-TNTs (seeTable 3).
2.4 Filtration experiments
The ltration experiments for TFC and TFN membranes wereperformed using a commercial stirred dead-end permeation cell(HP4750, Sterlitech Corp.). Prior to any measurement, themembranes were washed thoroughly for 30 min under
J. Mater. Chem. A
Journal of Materials Chemistry A Paper
Publ
ishe
d on
12
Febr
uary
201
6. D
ownl
oade
d by
Nat
iona
l Uni
vers
ity o
f Si
ngap
ore
on 2
5/02
/201
6 14
:31:
33.
View Article Online
a pressure of 7 bar, until the membranes were fully compactedand the ux reached a steady state. The performance ofmembranes was then evaluated for water permeability as well asrejection against Na2SO4 (1000 mg L�1), NaCl (1000 mg L�1),BSA (1000 mg L�1) and RB5 (1000 mg L�1) at an operatingpressure of 6 bar. The volume of permeate (mL) collected overa specied time divided by the membrane area (14.6 cm2) andtime to collect the permeate (min) produced the reported valuefor permeate ux (L m�2 h�1). The observed solute rejections, R(%) ¼ ((1 � Cf/Cp) � 100), were determined from the differencebetween feed (Cf) and permeate (Cp) solute concentrations. Abench conductivity meter (4520, Jenway) was used to measurethe conductivity value in the feed and permeate of salt solutionswhile a UV-vis spectrophotometer (DR5000, Hach) was used todetermine the concentration of BSA and RB5 in solutions. Toevaluate the membrane fouling resistance, TFC and TFNmembranes were subjected to a 3 h permeation test using feedaqueous solution containing either BSA or RB5 as a modelfoulant. The ux decline determined from the initial permeateux, Ji, and the nal permeate ux, Jf, was then used to study theextend of membrane fouling.
A prolonged ltration test using pure water (up to 480 min)was also conducted to determine if there was any leaching of thesurface-modied TNTs from the PA nanocomposite layer to feedor permeate samples. For each set of feed and permeatesamples collected at specic time intervals for Ti analysis, thepure water in the permeation cell was withdrawn and the cellwas relled again with pure water. The presence of nanollersin the solutions was detected by using an inductively coupledplasma mass spectrometer (ICP-MS) (ELAN6100, Perkin Elmer)with respect to the titanium (Ti) element at the mg L�1 level(equivalent to parts per billion, ppb).
2.5 Characterization
Specic surface measurements of TNTs were performed usinga Micromeritics® ASAP 2010 with N2 at 77.3 K. The specicsurface area was calculated according to the Brunauer–Emmett–Teller (BET) method. Powder XRD analysis was made usinga D-Max Rigaku diffractometer with 2q ranging from 10� to 80�.Transmission electron microscopy (TEM) analysis was per-formed using a Hitachi HT7700 to study the morphologicalstructure of TNTs in the nm scale. Fourier transform infrared(FTIR) spectra of TNTs and membranes were recorded ona Thermo Scientic Nicolet 5700. The spectra were collected inthe attenuated total reection (ATR) mode. The spectra wererecorded in the 600–4000 cm�1 wavenumber range with anaverage of 16 scans and at a resolution of 4 cm�1. The samecharacterization was performed to analyze the interactionbetween NH2-TNTs and TMC monomers in the absence of PIPmonomers. To conrm the chemical reaction, NH2-TNT–TMCcompounds were obtained by directly adding NH2-TNTs intoTMC/cyclohexane solution followed by washing and drying.
Surface images of the composite membranes together withelemental analysis were obtained using a Hitachi SU8000 eldemission scanning electronmicroscope (FESEM) equippedwith anenergy-dispersive X-ray (EDX) spectrometer. Thermogravimetric
J. Mater. Chem. A
analysis (TGA) was performed on a Mettler Toledo instrument ata rate of 10 �Cmin�1 under a nitrogen atmosphere from 100 �C to700 �C. 3D membrane surface images and roughness values wereobtained using a Park System XE-100 atomic force microscope(AFM) with a scan size of 10 mm� 10 mm. The membrane surfacehydrophilicity was characterized based on a sessile drop techniqueusing a Dataphysics OCA15plus contact angle goniometer.
3 Results and discussion3.1 Characteristics of TNTs
Fig. 1 shows the properties of TNTs characterized by usingdifferent instruments. The morphological structure of TNTs canbe evidenced from the TEM image shown in Fig. 1(a). It is foundthat the synthesized nanotubes are of tubular-shape with open-ended lumens. The inner diameter and length of these nano-tubes are in the ranges of 9–14 nm and 0.35–2.25 mm, respec-tively, with a BET surface area reported to be about 200 m2 g�1.The XRD peaks of 10.85�, 24.5� and 48.5� as shown in Fig. 1(b)correspond to the (200), (110) and (020) planes, respectivelyindexed to TNTs. Meanwhile, the broad FTIR band between3000 and 3500 cm�1 shown in Fig. 1(c) indicates the presence ofabundant –OH groups on the surface of TNTs. The peak at 1630cm�1 could be attributed to the characteristic of stretching andbending vibrations of water molecules that adsorbed on thedried TNTs.
3.2 Comparison between TFNaq and TFNcyclo membranes
3.2.1 FTIR. FTIR spectra of three composite membranesurfaces are shown in Fig. S2.† Overall, there is no differencebetween the spectra of the TFC and TFN membranes prepared.The results suggest that the incorporation of small quantities ofinorganic nanollers do not alter the spectra of the organic PAstructure, indicating that the dominant functional groups donot alter. A similar observation was also reported in the work ofBarona et al.19 in which aluminosilicate single wall nanotubeswere incorporated into the PA layer of the compositemembrane. In general, the peaks originating from the inter-facially polymerized layer could be found at 3445 cm�1, 1630cm�1 and 1540 cm�1. These peaks correspond to the –OHstretching of carboxylic acid, C]O stretching vibration of theamide group and C–N stretching, respectively.
Further FTIR analysis of the interaction between aminofunctionalized TNTs and TMC monomers (Fig. S3†) in theabsence of PIP monomers shows no peak arising from theamide II (N–H) of functionalized TNTs at 1550 cm�1. Thissuggests that the signal of the amino group attached to the TNTsurface is weak. The minimum amount of amino groupsattached to the TNT surface can be benecial to interfacialpolymerization, avoiding competition with PIP for interactionwith TMC monomers and increasing the PA cross-linkingdegree.
However, in order to prove that the surface of TNTs has beenmodied, Fig. 2 compares the dispersion quality of TNTs withand without AAPTSmodication in cyclohexane. It is found thatthe amino-functionalized TNTs are clearly dispersed better in
This journal is © The Royal Society of Chemistry 2016
Fig. 1 Properties of synthesized TNTs, (a) TEM image, (b) XRD spectra and (c) FTIR spectra.
Fig. 2 Dispersion quality of (a) TNTs and (b) NH2-TNTs in cyclohexane at different times, (i) 5 s, (ii) 2 min, (iii) 4 min, (iv) 10 min and (v) 90 min.
Paper Journal of Materials Chemistry A
Publ
ishe
d on
12
Febr
uary
201
6. D
ownl
oade
d by
Nat
iona
l Uni
vers
ity o
f Si
ngap
ore
on 2
5/02
/201
6 14
:31:
33.
View Article Online
the non-polar solvent in comparison to the unmodied TNTs.Owing to the improved surface properties, modied TNTs takea longer time to settle. Similar surface modications wereconducted by Namvar-Mahboub et al.20 and Rajaeian et al.21 onUZM-5 zeolite and TiO2 nanoparticles, respectively, but bothresearch studies did not demonstrate the dispersion quality ofamino-functionalized nanomaterials in the organic solvent.
3.2.2 FESEM-EDX-AFM. Fig. 3 shows the FESEM image,contact angle and EDX mapping on the PA surface of threedifferent types of composite membranes. In the FESEM imagesof the top surface, the white parts represent the peaks while the
This journal is © The Royal Society of Chemistry 2016
dark areas correspond to the valleys. The formation of ridge andvalley structures conrms that the PA active layer is successfullyformed over the PSF substrate. However, as can be clearly seen,the addition of NH2-TNTs into either aqueous or organic solu-tion has altered the structure of PA leading to more peaks beingformed for the TFNaq-0.05 and TFNcyclo-0.05 membrane. Particu-larly for the TFNaq-0.05 membrane, obvious irregular nodules areformed which affect the integrity of the PA layer. The observa-tion can be explained by the approach used in removing excessaqueous solution containing NH2-TNTs. The use of a rubberroller to remove aqueous solution from the substrate surface
J. Mater. Chem. A
Fig. 3 SEM images, contact angle and EDX results of the membrane top surface, (a) TFC, (b) TFNaq-0.05 and (c) TFNcyclo-0.05. The red arrowsindicate the presence of TNTs in the membrane.
Journal of Materials Chemistry A Paper
Publ
ishe
d on
12
Febr
uary
201
6. D
ownl
oade
d by
Nat
iona
l Uni
vers
ity o
f Si
ngap
ore
on 2
5/02
/201
6 14
:31:
33.
View Article Online
has negatively affected the distribution of nanollers on thesubstrate surface. Compared to the TFNcyclo-0.05 membrane, inwhich the nanollers are introduced from the organic solutionaer excess aqueous solution has been removed, the formationof ridge-and-valley like structures is more even. Besides, it wasalso experienced that a signicant amount of NH2-TNTs waslost in the TFNaq- membrane during the excess aqueous solu-tion removal process. This is further supported by TGA results(see Fig. S4†) in which the residue for the TFNaq-0.05 membraneat 700 �C is lower than that for the TFNcyclo-0.05 membrane,indicating less TNTs are embedded in the TFNaq- membrane.Compared to the polymeric materials (PA and PSF), TNTs arehigh melting point materials and do not decompose attemperatures less than 700 �C.
The EDX data, on the other hand, showed that the TFNcyclo-0.05
membrane surface exhibits signicantly higher weight percentsof Ti and O elements compared to the TFNaq-0.05 membrane,suggesting the greater amount of nanotubes embedded in thePA layer. Meanwhile, no Ti element is detected in the TFCcontrol membrane. With respect to the contact angle, it is foundthat the TFC control membrane displays lower values than
J. Mater. Chem. A
those of TFN membranes. Its contact angle of 39.0� is accept-able for a typical hydrophilic PA layer and falls within the rangeof PA made of PIP-TMC.22
A correlation between membrane hydrophilicity and thewater contact angle might not be established in this case as theincrease in the surface roughness of TFN membranes mightaffect the contact angle according to the Cassie's model. The 3DAFM images (see Fig. S5†) indicate that the TFNaq-0.05
membrane displays a higher Rz value (1074.85 nm) compared toTFNcyclo-0.05 (741.7 nm) and TFC (419.8 nm). The highest Rz
value shown by the TFNaq-0.05 membrane is in good agreementwith the SEM surface image, i.e. obvious irregular nodules aredetected in this membrane type. Compared to the TFCmembrane, the relatively higher surface roughness values of theTFNcyclo-0.05 membrane could be attributed to the existence ofNH2-TNTs within the PA layer. Previous work has also reportedthat both the contact angle and surface roughness of the PAlayer are increased upon the introduction of zeolite nano-particles.17 In order to understand if the presence of hydrophilicnanotubes do improve PA surface hydrophilicity, detaileddiscussion on the TFN membrane performance with respect to
This journal is © The Royal Society of Chemistry 2016
Paper Journal of Materials Chemistry A
Publ
ishe
d on
12
Febr
uary
201
6. D
ownl
oade
d by
Nat
iona
l Uni
vers
ity o
f Si
ngap
ore
on 2
5/02
/201
6 14
:31:
33.
View Article Online
water ux and salt rejection will be provided in the followingsection.
3.2.3 Filtration performance. Fig. 4 compares the ltrationperformance of TFN membranes made from the differentapproaches with the TFC membrane. With respect to the purewater ux, it can be observed from Fig. 4(a) that all the TFNmembranes exhibit higher water permeability than that of theTFC membrane, regardless of the nanoller loading and thefabrication approach. Particularly for TFNcyclo- membranes,they achieve 39.0–40.6% higher pure water ux in comparisonto the TFC membrane without NH2-TNT incorporation. Theresults obtained in this work are in line with many previousndings,1,5–7,16,23,24 i.e. TFN membranes always achieve greaterwater production rates compared to the TFC membrane irre-spective of the solution that is used to disperse nanollers. Theimproved water productivity as observed in the TFNaq- orTFNcyclo- membrane might be attributed to the presence ofhydroxyl groups (average 5.8 –OH per nm2) at the TNT surfaceand its narrow channels (several tens of nanometers) thatprovide a smooth transport pathway for water to passthrough.14,25 FESEM images shown in Fig. 3 reveal that some ofthe incorporated nanotubes would be exposed to the membranesurface which makes the hydrophilic nanotubes accountablefor the increased membrane hydrophilicity to some extent.However, possible changes in PA network pores and/or chainpacking density upon TNT incorporation cannot be ruled out asfactors affecting the membrane water ux as the nature of theinorganic ller–PA interaction is still not fully understood.
Based on the membrane surface characterization and waterltration results, it is proven that introducing nanollers intoorganic solution is much more effective to improve thecomposite membrane water ux, owing to the signicantquantity of nanollers present in the PA layer. As a comparison,the water ux of the TFNcyclo- membrane is 31.2% and 27.7%higher compared to the TFNaq- membrane at NH2-TNT loadingsof 0.05% and 0.10%, respectively.
Fig. 4 Properties of TFNmembranes made by different approaches, (a) p– flux: 32.3 L m�2 h�1 and rejection: 86.0%).
This journal is © The Royal Society of Chemistry 2016
Although the TFN membrane water ux is further enhancedwith increasing the NH2-TNT loading, one can observe a nega-tive impact of the use of higher loadings of nanollers on saltrejection (Fig. 4(b)). For two different types of the TFNmembranes fabricated (i.e. TFNaq- and TFNcyclo-), it is found thatthe higher the loading of nanoller used, the lower the saltrejection and vice versa. It is also found that with increasing theamount of TNTs from 0.05% to 0.10%, the rejection rate of TFNmembranes decreases to the level close to the TFC membrane.The excessive use of nanollers is believed to be able to nega-tively affect the integrity of organic PA–inorganic nanotubestructures, interfering with the interfacial polymerization reac-tion. This, as a consequence, leads to the formation of defects(bigger pores) that allow ions to pass through. Our statement iswell supported by other relevant works that incorporatednanollers into the PA thin selective layer. These include thework of Ma et al.,17 Ghanbari et al.6 and Mollahosseini andRahimpour26 in which zeolite, TiO2/halloysite nanotube nano-composites and silver nanoparticles were used, respectively, asnanollers.
Based on the ndings, it can be said that the TFNcyclo-0.05
membrane is the best performing membrane among the vecomposite membranes studied by taking into consideration itshigh water permeability and excellent salt rejection. In thefollowing section, further investigations are conducted on thecharacteristics of this TFN membrane type by varying PIP andTMC monomers.
3.3 Effect of PIP and TMC monomers on TFN membraneproperties
Fig. 5 presents the effects of the PIP and TMC content on theperformance of TFN membranes incorporated with 0.5% nano-llers. The effects of both monomers on the TFN membranewater ux and salt rejection are found to be very similar, i.e. thecomposite membrane achieves optimum values when interme-diate PIP (2.0%) and TMC (0.15%) concentrations are employed.
ure water flux and (b) salt rejection (reference: TFC (control) membrane
J. Mater. Chem. A
Fig. 5 Effect of the monomer content on the performance of TFN membranes that contain 0.5 w/v% NH2-TNTs, (a) PIP content and (b) TMCcontent.
Journal of Materials Chemistry A Paper
Publ
ishe
d on
12
Febr
uary
201
6. D
ownl
oade
d by
Nat
iona
l Uni
vers
ity o
f Si
ngap
ore
on 2
5/02
/201
6 14
:31:
33.
View Article Online
At the lowest concentration of PIP and TMC monomer used,both membrane water ux and salt rejection are lowercompared to the intermediate concentration of PIP and TMC.The possible reason for the poor membrane performance is dueto the lower degree of polymerization and cross-linking in thePA layer that results from the fact that insufficient monomersare available for interfacial polymerization. This concept issupported by the lowest surface roughness values of the TFNPIP-
1.5 and TFNTMC-0.1 membrane as shown in Table 4. The insuf-cient amount of monomers available for interfacial polymeri-zation might also cause poor interaction between nanotubesand the PA layer, leading to the formation of PA surface defectsand/or possible nanotubes leaching out during ltration. This,as a result, affects both water ux and salt rejection asevidenced.
When the concentration of monomers is the highest, bothTFNPIP-2.5 and TFNTMC-0.2 membranes have signicantly lowerwater ux and salt rejection than the TFNcyclo-0.05 membranemade with intermediate PIP and TMC concentrations. Theexperimental results obtained can be attributed to the forma-tion of a denser and thicker PA layer. The increase in the surfaceroughness values of TFNPIP-2.5 and TFNTMC-0.2 membranes alsosuggests more ridge-and-valley like structures created by the
Table 4 Effect of the monomer concentration on the surfaceroughness value of TFN membranes
Parameter MembraneaMonomerconc. (wt/v%)
Roughness value (nm)
Ra Rq Rz
Effect of PIP TFNPIP-1.5 1.5 74.54 93.69 662.70TFNPIP-2.0 2.0 87.51 117.27 741.65TFNPIP-2.5 2.5 103.49 126.46 782.96
Effect of TMC TFNTMC-0.1 0.1 84.34 105.85 746.43TFNTMC-0.15 0.15 87.51 117.27 741.65TFNTMC-0.2 0.2 99.33 127.16 998.86
a The properties of the TFNPIP-2.0 or TFNTMC-0.15 membrane is exactly thesame as those of the TFNcyclo-0.05 membrane.
J. Mater. Chem. A
higher degree of polymerization and cross-linking. Normally,employing higher monomer concentrations tends to producethicker polymeric lms with higher molecular weights.8 Liet al.27 elucidated that a higher PIP and TMC concentrationtended to generate a thicker and more compact PA layer thatcould affect the water ux due to increased transport resistance.In the case of the PA layer incorporated with nanollers,a thicker surface layer might reduce the extent of nanotubeexposure to the top PA surface (reduced hydrophilicity) and/orcover the narrow channels of TNTs. Both effects can possiblyaffect the water ux and salt rejection behaviour.
The ndings of this work demonstrate that changes in theconditions of interfacial polymerization are also required whennanollers are added to the formulation to produce membranesof excellent performance. Addition of nanollers simply to themonomer solution, either organic or aqueous, does not neces-sarily guarantee that you are working under the optimalconditions. The above example indicates that the concentra-tions of the monomer solutions should also be properlyadjusted. It must also be pointed out that the changes in othervariables such as substrate properties, aqueous/organic solu-tion properties, reaction time and curing time during interfacialpolymerization and types of nanollers could lead to differentoutcomes.28,29
Table 5 compares the performances of the best TFNmembrane synthesized in our work with those of other TFC orTFN membranes reported in the literature for the separation ofsalts.10,11,15,18,21,26,30–33 By taking into consideration the pure waterpermeability coefficient (L m�2 h�1 bar�1) of the membrane, itis found that our in-house synthesized TFN membrane hasachieved greater water permeability (7.48 L m�2 h�1 bar�1) thanmost of the TFC/TFN membranes reported (1.5–6.98 L m�2 h�1
bar�1), except PIP–TMC–SiO2 (9.45 L m�2 h�1 bar�1) and PIP–TMC–SAPO-34 (11 L m�2 h�1 bar�1) membranes. Its Na2SO4
rejection (96.4%) recorded is also among the highest ones thatare summarized in Table 5. The good balance between waterpermeability and solute selectivity as exhibited by our TFNmembrane can be attributed to several factors. First, the use of
This journal is © The Royal Society of Chemistry 2016
Table 5 Performance comparison between the TFN membrane synthesized in this work and other TFC/TFN membranes reported for NFapplications
Flat membrane(label)a
Monomersusedb
Nanomaterialsadded (A/O)c
Testing conditions
PWPd
(L m�2 h�1 bar�1)
Rejection (%)
ReferencePressure(bar)
Salt conc.(mg L�1) Na2SO4 NaCl
TFNcyclo-0.05 PIP–TMC Modied TNTs (O) 6 1000 7.48 96.4 18.9 In this workTFC (control) PIP–TMC — 6 1000 5.38 86.0 25.3 In this workM1 (NF4) PAMAM–TMC SiO2 (A) 5 1000 2.10 96.4 50.2 30M2 (–) PIP–TMC SiO2 (A) 6 710.2 5.33 �80 — 11M3 (–) PIP–TMC SiO2 (A) 6 2000 9.45 97.3 25.6 10M4 (I1) MPD–TMC Ag NPs (A) 5 2000 �1.50 — �14.5 26M5 (A-TFN3) MPD–TMC Modied TiO2 (A) 7.5 2000 3.60 — �33 21M6 (–) PIP–TMC Modied MWCNTs (O) 10 2000 �6.98 99 44.1 31M7 (–) MPD–TMC H-OMCs (A) 10 2000 1.95 �89 �62 15M8 (TFN) PIP–TMC SAPO-34 (O) 3 N/A �11.0 �85 — 18M9 (–) TEPA–TMC — 10 1000 5.10 �80 — 32M10 (N4) 2,20-OEL–TMC — 7 2000 5.86 �76 �23 33
a The designation in the bracket is the membrane designation used in the respective work. b PIP: piperazine, TMC: trimesoyl chloride, PAMAN:poly(amidoamine), MPD: m-phenylenediamine, TEPA: tetraethylenepentamine, 2,20-OEL: 2,20-oxybis-ethylamine. c The alphabet in the bracket isreferred to the solution that inorganic nanomaterials are added. “O” stands for organic solution while “A” stands for “aqueous solution”.d PWP: pure water permeability. PWP was calculated by normalizing the water ux with the applied pressure.
Paper Journal of Materials Chemistry A
Publ
ishe
d on
12
Febr
uary
201
6. D
ownl
oade
d by
Nat
iona
l Uni
vers
ity o
f Si
ngap
ore
on 2
5/02
/201
6 14
:31:
33.
View Article Online
nanollers with abundant –OH groups. Second, the introduc-tion of nanollers into organic solution. Third, themodicationof the TNT surface for better dispersion in organic solution andnally, the use of cyclohexane (instead of n-hexane) for betterdiffusivity of the PIP and TNT dispersion. Fig. 6 compares thedispersion quality of modied TNTs in cyclohexane andn-hexane aer 10minmixing. It is clearly seen that themodiedTNTs disperse better in cyclohexane compared to n-hexanedispersed nanomaterials, indicating an alicyclic hydrocarbon
Fig. 6 Comparison between the dispersion quality of 0.05% (w/v)NH2-TNTs in (a) n-hexane and (b) cyclohexane after 10 min mixing.
This journal is © The Royal Society of Chemistry 2016
that exhibits higher boiling and ash points, a better organicsolvent for nanomaterial dispersion.
3.4 Antifouling resistance against BSA and dye
The practical applications of the TFN membrane have beenfurther studied by subjecting the membrane to solutions con-taining either 1000 mg L�1 BSA or 1000 mg L�1 RB5. From Fig. 7and 8, it is found that the TFNcyclo-0.05 membrane shows a muchhigher water ux than that of the TFC control membrane whenboth membranes are tested under the same conditions. Besidesthe greater water ux, the ux decline of the TFN membrane isless than the TFC membrane within the studied period, e.g. theinitial ux of the TFNmembrane declines only 4.95% comparedto �20% of the TFC membrane within 180 min of ltration,which is ascribed to the improved antifouling resistance againstBSA macromolecules of the TFN membrane.
Fig. 7 Comparison between the performance of TFC and TFNmembranes in filtrating the feed solution containing 1000 mg L�1 BSA.The NH2-TNT loading in the tested TFN membrane was 0.05% (w/v).
J. Mater. Chem. A
Fig. 8 Comparison between the water flux of TFC and TFNmembranes in filtrating the feed solution containing 1000 mg L�1 RB5.The inserted photos show the top surface of the membrane after thedye treatment process. The NH2-TNT loading in the tested TFNmembrane was 0.05% (w/v).
Fig. 9 Ti element detected in feed and permeate samples of theTFNcyclo-0.05 membrane as a function of filtration time.
Journal of Materials Chemistry A Paper
Publ
ishe
d on
12
Febr
uary
201
6. D
ownl
oade
d by
Nat
iona
l Uni
vers
ity o
f Si
ngap
ore
on 2
5/02
/201
6 14
:31:
33.
View Article Online
In terms of resistance against dye, the water ux of the TFNmembrane declines 10.7% in comparison to 13.6% of the TFCmembrane. This result is in accordance with the visible obser-vation that the colour stained on the TFN membrane surface isobviously less than that on the TFC membrane aer the dyetreatment (see inserted photos in Fig. 8). Further analysis showsthat the ux decline of the TFN membrane by RB5 (10.7%) ismore severe than by BSA (4.95%), which can be possiblyexplained by the severe nanotube channel blockage owing to thesmall size of the RB5 dye compound.
Regarding the separation efficiency, both TFC and TFNmembranes achieve complete removal of BSA and approxi-mately 97.5% RB5 rejection. Based on the results, it can be saidthat the TFNmembrane can produce not only higher quantity oftreated water during operation but also suffers from a lowerdegree of fouling against protein and reactive dye. This, indi-rectly, might reduce the cycles of membrane cleaning andextend membrane lifetime in real eld applications such as indairy and textile industries.
3.5 Nanoller leaching test
Fig. 9 shows the concentration of the Ti element detected in thefeed and permeate samples of the TFNcyclo-0.05 membrane asa function of ltration time (up to 8 h). Although the Ticoncentration is below the detection limit of 0.01 ppb in mostcases, the feed/permeate samples at 30min and the feed sampleat 360 min are found to contain 0.022/0.017 ppb and 0.024 ppbTi, respectively. The detection of very low concentrations of theTi element in the sample solutions could indicate the possibleleaching of TNTs from the PA nanocomposite layer to either thefeed or permeate solution. TNT leaching might be the result ofpoor interaction of nanollers with the PA layer during inter-facial polymerization (especially those exposed at themembrane surface), making them unable to be embeddedrmly within the PA matrix. Further investigation on this topicis worthy as, at the present, few relevant studies have beenconducted on the leaching of nanollers from PA layers and itsimpact on long term TFN membrane performance.
J. Mater. Chem. A
4 Conclusions
Thin lm nanocomposite NF membranes were successfullyprepared via two different approaches to introduce surface-modied TNTs into the PA layer made of PIP and TMC mono-mers. Results showed that introducing nanollers into the TMCorganic phase was more effective for the synthesis of TFNmembranes with greater separation performance than theintroduction of nanollers into the PIP aqueous phase. It isbecause the use of the rubber roller to remove aqueous solutionfrom the substrate surface has caused the loss of a signicantamount of nanollers from the substrate surface, which nega-tively affected the PA integrity as proved by instrumental anal-yses. Moreover, when the nanollers are added, monomerconcentrations should be properly adjusted to produce TFNmembranes of improved performance. For example, at anoptimum nanoller loading (0.05 w/v%), manipulating the PIPand TMC concentrations to 2 and 0.15%, respectively, isrequired in order to produce a PA TFN NF membrane witha good combination of water ux (7.5 L m�2 h�1 bar�1) andNa2SO4 rejection (96.4%). The high hydrophilicity and largesurface area of TNTs coupled with their narrow channels areproven to enhance the TFN membrane resistance againstfouling by BSA and RB5. However, the possible leaching ofnanollers (at mg L�1 or ng L�1 level) from the PA nano-composite matrix needs further attention as, at present, notmany relevant studies have been conducted on the nanollerleaching and its impact on long term TFN membraneperformance.
Acknowledgements
The corresponding author wishes to acknowledge the Austra-lian Government Department of Education for providing the2015 Endeavour Research Fellowship which initiates theresearch collaboration between Universiti Teknologi Malaysia(UTM) and Victoria University (VU). Financial support for thisresearch was provided by UTM under the UTMShine agshipprogram (Grant no: Q.J130000.2446.03G38) and MalaysianMinistry of Higher Education (MoHE) under the AMTEC-HICOEproject (Grant no.: R.J090301.7846.4J174).
This journal is © The Royal Society of Chemistry 2016
Paper Journal of Materials Chemistry A
Publ
ishe
d on
12
Febr
uary
201
6. D
ownl
oade
d by
Nat
iona
l Uni
vers
ity o
f Si
ngap
ore
on 2
5/02
/201
6 14
:31:
33.
View Article Online
References
1 B.-H. Jeong, E. M. V. Hoek, Y. Yan, A. Subramani, X. Huang,G. Hurwitz, A. K. Ghosh and A. Jawor, J. Membr. Sci., 2007,294, 1–7.
2 W. J. Lau, S. Gray, T. Matsuura, D. Emadzadeh, J. Paul Chenand A. F. Ismail, Water Res., 2015, 80, 306–324.
3 H. Huang, X. Qu, X. Ji, X. Gao, L. Zhang, H. Chen and L. Hou,J. Mater. Chem. A, 2013, 1, 11343–11349.
4 D. Emadzadeh, W. J. Lau, M. Rahbari-Sisakht, A. Daneshfar,M. Ghanbari, A. Mayahi, T. Matsuura and A. F. Ismail,Desalination, 2015, 368, 106–113.
5 H. Zhao, S. Qiu, L. Wu, L. Zhang, H. Chen and C. Gao,J. Membr. Sci., 2014, 450, 249–256.
6 M. Ghanbari, D. Emadzadeh, W. J. Lau, T. Matsuura,M. Davoody and A. F. Ismail, Desalination, 2015, 371, 104–114.
7 H. J. Kim, M.-Y. Lim, K. H. Jung, D.-G. Kim and J.-C. Lee,J. Mater. Chem. A, 2015, 3, 6798–6809.
8 M. L. Lind, A. K. Ghosh, A. Jawor, X. Huang, W. Hou, Y. Yangand E. M. V. Hoek, Langmuir, 2009, 25, 10139–10145.
9 E.-S. Kim, G. Hwang, M. G. El-Din and Y. Liu, J. Membr. Sci.,2012, 394–395, 37–48.
10 D. Hu, Z.-L. Xu and C. Chen, Desalination, 2012, 301, 75–81.11 H.Wu, B. Tang and P. Wu, J. Membr. Sci., 2013, 428, 341–348.12 N. Niksefat, M. Jahanshahi and A. Rahimpour, Desalination,
2014, 343, 140–146.13 T. Kasuga, M. Hiramatsu, A. Hoson, T. Sekino and
K. Niihara, Langmuir, 1998, 14, 3160–3163.14 D. Emadzadeh, W. J. Lau, M. Rahbari-Sisakht, H. Ilbeygi,
D. Rana, T. Matsuura and A. F. Ismail, Chem. Eng. J., 2015,281, 243–251.
15 E.-S. Kim and B. Deng, J. Membr. Sci., 2011, 375, 46–54.16 D. Hu, Z.-L. Xu and C. Chen, Desalination, 2012, 301, 75–81.
This journal is © The Royal Society of Chemistry 2016
17 N. Ma, J. Wei, R. Liao and C. Y. Tang, J. Membr. Sci., 2012,405–406, 149–157.
18 T.-Y. Liu, Z.-H. Liu, R.-X. Zhang, Y. Wang, B. Van der Bruggenand X.-L. Wang, J. Membr. Sci., 2015, 488, 92–102.
19 G. N. B. Barona, J. Lim, M. Choi and B. Jung, Desalination,2013, 325, 138–147.
20 M. Namvar-Mahboub, M. Pakizah and S. Davari, J. Membr.Sci., 2014, 459, 22–32.
21 B. Rajaeian, A. Rahimpour, M. O. Tade and S. Liu,Desalination, 2013, 313, 176–188.
22 W. J. Lau, A. F. Ismail, P. S. Goh, N. Hilal and B. S. Ooi, Sep.Purif. Rev., 2015, 44, 135–156.
23 M. Ghanbari, D. Emadzadeh, W. J. Lau, T. Matsuura andA. F. Ismail, RSC Adv., 2015, 27, 21268–21276.
24 D. Emadzadeh, W. J. Lau, T. Matsuura, A. F. Ismail andM. Rahbari-Sisakht, J. Membr. Sci., 2014, 449, 74–85.
25 D. V. Bavykin and F. C. Walsh, Titanate and titaniananotubes:Synthesis, properties and applications, Royal Society ofChemistry, 2010.
26 A. Mollahosseini and A. Rahimpour, Biofouling, 2013, 29,537–548.
27 Y. Li, Y. Su, J. Li, X. Zhao, R. Zhang, X. Fan, J. Zhu, Y. Ma,Y. Liu and Z. Jiang, J. Membr. Sci., 2015, 476, 10–19.
28 W. J. Lau, A. F. Ismail, N. Misdan and M. A. Kassim,Desalination, 2012, 287, 190–199.
29 W. J. Lau and A. F. Ismail, Desalination, 2009, 245, 321–348.30 L. M. Jin, S. L. Yu, W. X. Shi, X. S. Yi, N. Sun, Y. L. Ge and
C. Ma, Polymer, 2012, 53, 5295–5303.31 J. n. Shen, C. c. Yu, H. m. Ruan, C. j. Gao and B. Van der
Bruggen, J. Membr. Sci., 2013, 442, 18–26.32 Y. Li, Y. Su, Y. Dong, X. Zhao, Z. Jiang, R. Zhang and J. Zhao,
Desalination, 2014, 333, 59–65.33 J.-b. Jin, D.-q. Liu, D.-d. Zhang, Y.-h. Yin, X.-y. Zhao and
Y.-f. Zhang, Desalination, 2015, 355, 141–146.
J. Mater. Chem. A