Contents lists available at ScienceDirect

Journal of Membrane Science

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

Nanocomposite organic solvent nanofiltration membranes by a highly-efficient mussel-inspired co-deposition strategy

Yan Chao Xua, Yu Pan Tangb, Li Fen Liuc,d,⁎⁎, Zhan Hu Guoe,⁎⁎, Lu Shaoa,⁎

a MIIT Key Laboratory of Critical Materials Technology for New Energy Conversion and Storage, State Key Laboratory of Urban Water Resource andEnvironment, School of Chemistry and Chemical Engineering, Harbin Institute of Technology, Harbin 150001, Chinab Department of Chemical and Biomolecular Engineering, National University of Singapore, 4 Engineering Drive 4, Singapore 117585, Singaporec Center for Membrane and Water Science and Technology Ocean College, Zhejiang University of Technology, Hangzhou 3100014, Chinad Collaborative Innovation Center of Membrane Separation and Water Treatment of Zhejiang Province, Hangzhou 310014, Chinae Integrated Composites Laboratory (ICL), Department of Chemical & Biomolecular Engineering, University of Tennessee, Knoxville, TN 37996, USA

A R T I C L E I N F O

Keywords:Organic solvent nanofiltration (OSN)Nanocomposite membranePolyhedral oligomeric silsesquioxane (POSS)CatecholCo-deposition

A B S T R A C T

Herein, a novel nanocomposite organic solvent nanofiltration (OSN) membrane has been facilely fabricated by ahighly-efficient one-step co-deposition of mussel-inspired catechol and octaammonium polyhedral oligomericsilsesquioxane (POSS-NH3

+Cl-) onto supports. The basic properties and morphologies of the co-depositednanocomposite membranes were investigated with various physicochemical characterizations in detail.Attenuated total reflectance Fourier transform infrared (ATR-FTIR) spectra proved the present of POSSnanoparticles on membrane surface. X-ray photoelectron (XPS) results suggested the optimal ratio of POSS-NH3

+Cl- and catechol for co-deposition. Scanning electron microscopy (SEM) and atomic force microscopy(AFM) images demonstrated the formation of a layer on support surface. The optimized nanocompositemembrane exhibited an ethanol (EtOH) permeance of 1.26 L m−2 h−1 bar−1 with a rejection of 99% to RoseBengal (RB). The novel membrane also exhibited remarkable separation performance for dyes removal from awide range of solvents including challenging polar aprotic and strongly swelling solvents. Particularly, thenanocomposite membrane demonstrated stable performances during a two-day long term test in DMF for RBconcentration. In addition to providing a highly-efficient way to high-performance OSN membrane, this workmay stimulate the bio-inspired design of advanced nanocomposite membranes for environmental applications.

1. Introduction

Organic solvent nanofiltration (OSN) is a relatively new technologythat allows efficient and sustainable separation of molecules withmolecular weights between 200–1000 g mol−1 from organic solventsas well as solvent recovery by simply applying a pressure gradientacross a membrane [1]. It has the merits of low energy consumption,environmental friendliness and safety comparing to the conventionalseparation technologies such as crystallisation, extraction and chro-matography [2,3]. The development of OSN membranes with highstability in a wide range of solvents and good separation performance isstill a challenge and has attracted increasing attention. In recent years,nanocomposite membranes with nanoparticles incorporated in poly-mer matrix have shown great promise in OSN application. Thesemembranes combine the advantages of polymer membranes (proces-

sability, flexibility, inexpensive) with the unique features of inorganicnanoparticles (porosity, hydrophobicity/hydrophilicity). Comparedwith their reference materials, the nanocomposite membranes usuallyshow higher permeance and/or selectivity [4,5].

So far, most of the reported nanocomposite OSN membranes arethin film nanocomposite (TFN) membranes, in which nanoparticleshave been embedded into the active layer on top of a support toimprove membrane performance. A diversity of nanoparticles includ-ing MOFs [6], silicalite [7], TiO2[8], SiO2[8] and SWCNTs [9] etc. havebeen deployed for TFN OSN membrane fabrication via interfacialpolymerization or direct coating method. Despite of the improvedmembrane performance, these methodologies still suffer some draw-backs. Interfacial polymerization generally involves three steps besidespretreatment and posttreatment. First, the selective surface of supportis contacted with an aqueous solution consisting of amine monomer.

http://dx.doi.org/10.1016/j.memsci.2016.12.026Received 3 October 2016; Received in revised form 11 December 2016; Accepted 12 December 2016

⁎ Corresponding author at: MIIT Key Laboratory of Critical Materials Technology for New Energy Conversion and Storage, State Key Laboratory of Urban Water Resource andEnvironment, School of Chemistry and Chemical Engineering, Harbin Institute of Technology, Harbin 150001, China.

⁎⁎ Corresponding author at: MIIT Key Laboratory of Critical Materials Technology for New Energy Conversion and Storage, State Key Laboratory of Urban Water Resource andEnvironment, School of Chemistry and Chemical Engineering, Harbin Institute of Technology, Harbin 150001, China.

E-mail addresses: lifenliu@zjut.edu.cn (L.F. Liu), zguo10@utk.edu (Z.H. Guo), shaolu@hit.edu.cn (L. Shao).

Journal of Membrane Science 526 (2017) 32–42

Available online 14 December 20160376-7388/ © 2016 Elsevier B.V. All rights reserved.

MARK

Second, the excess of aqueous solution on surface is moved. Third, thesurface is contacted with second organic solution containing trimesoylchoride [10]. The operation time of each step ranges from seconds tominutes, coupled with the poor dispersion of nanoparticles in aqueousor organic phase, makes it hard to obtain a repeatable membraneperformance. For direct coating approach, the use of viscous coatingsolution typically yields separation layer with thickness up to tens ofmicrometers, which subsequently results in the great decline of solventpermeance [11].

The mussel-inspired polydopamine coating method has attractedconsiderable interest these years. Under a weak alkaline aqueoussolution, dopamine can spontaneously oxidize and self-polymerize toform polydopamine [12]. This self-polymerization reaction concur-rently occurs both at any pre-immersed substrate surface to formdeposition and in solution to form aggregation [13,14]. It has beenwidely used as a universal and versatile tool in material surfaceengineering for various applications, although the exact polymerizationand interaction mechanism of the adhesive polydopamine layer wasstill unknown so far [15]. Specially, polydopamine modified mem-branes have been used as substrates, while dopamine have beenemployed as aqueous phase monomer in interfacial polymerization toprepare OSN membranes [16,17]. It is commonly acknowledged thatthe catechol and amine groups in dopamine structure are criticalfactors during the process of dopamine self-polymerization, and asystem containing catechol and amine groups could replace dopamineand co-deposited on various substrates by simulating a similar poly-merization mechanism [12]. This could potentially broaden the mole-cular diversity of dopamine and its derivatives. Co-deposition ofcatechol and well-designed functional molecules could provide greatpotential for target functional materials and coatings. Nevertheless,currently there are still few research works on this issue [18,19]. Veryrecently, our group has reported a composite NF membrane via the co-deposition of catechol and branched polyethylenimine for textilewastewater treatment [20]. To the best of our knowledge, the co-deposition of catechol with nanoparticle has not been reported.

Herein, for the first time, we reported the facile one-step co-deposition of catechol and octaammonium polyhedral oligomericsilsesquioxane ([SiO1.5(C3H6NH3

+Cl-)]8, referred to as POSS-NH3

+Cl-) onto cross-linked polyimide supports to fabrication nano-composite OSN membranes, as shown in Fig. 1. Polyhedral oligomericsilsesquioxane (POSS), which possesses a siloxane cubic core frame-work with a 0.53 nm pore size and a spherical diameter of 1–3 nm

[21], has previously been incorporated into membranes for gasseparation [22] and water treatment [23]. POSS-NH3

+Cl- was used inthis study because the amine groups at arms of POSS-NH3

+Cl- canensure high cross-linking density and endows the resulting nanocom-posite membrane with good solvent resistance. The possible polymer-ization mechanism behind co-deposition has been explored. Themorphologies and physicochemical properties of the as-preparedcomposite membrane were regulated by the mass ratio of POSS-NH3

+Cl-/catechol. The separation performance toward dyes in a widelyrange of solvents, long term operation performances and mechanicalproperties of the composite membrane were investigated in detail.

2. Experimental

2.1. Materials

A polyimide (PI) polymer, P84®, was provided by Granulat SG STD.Tris (hydroxymethyl) aminomethane (Tris), 1,6-hexanediamine(HDA), catechol, Methyl Orange (MO), Rose Bengal (RB), CrystalViolet (CV), Orange G (OG), Acid Fuchsin (AF), Methyl Blue (MB) andSolvent Blue II (SB II) were obtained from Aladdin Industrial Co., Ltd.Octaammonium POSS (POSS-NH3

+Cl-) was purchased from HybridPlastics Inc. All organic solvents were supplied by Xilong ChemicalIndustrial Co., Ltd. All used water was deionized.

2.2. Membrane preparations

2.2.1. Preparation of PI supportA P84 UF membrane was fabricated by a traditional non-solvent

induced phase separation process. Prior to use, the polymer was firstdried in a vacuum oven at 120 °C overnight to remove moisture. Apolymer solution was prepared by dissolving 15 wt% of P84 in NMPand stirring until complete dissolution. The solution was allowed tostand for a further 12 h at room temperature to remove any trapped airbubble and then cast on a glass plate using a casting knife set to athickness of 200 µm. The polymer membranes were then precipitatedfrom solution via immersion in water. The membranes were thenimmersed in IPA to remove water from the polymer matrix, and thentransferred to a 20 g L−1 solution of HDA in IPA overnight for cross-linking. Next, the membranes were washed with IPA to remove anyresidual HDA and then stored in fresh IPA before use.

Fig. 1. Schematic illustration of the fabrication of novel nanocomposite NF membranes via the mussel-inspired co-deposition of catechol and POSS-NH3+Cl- for organic solvent

nanofiltration.

Y.C. Xu et al. Journal of Membrane Science 526 (2017) 32–42

33

2.2.2. Preparation of co-deposited nanocomposite membranesThe co-deposited membranes were facilely fabricated by a one-step

co-deposition of catechol and POSS-NH3+Cl- onto the PI substrates.

First, catechol (100 mg) and POSS-NH3+Cl- with determined mass

ratios were dissolved in a Tris-HCl buffer solution (100 mL, pH 8.5).Subsequently, the PI substrates (5 cm×5 cm) were immersed in theabove solutions and shaked at 30 °C for 0.5–3.5 h. Then, the modifiedmembrane was rinsed with DI water for several times to removeunreacted residuals. The clean nanocomposite membrane was stored inDI water for further use.

2.3. Membrane characterizations

The ultraviolet-visible (UV–vis) absorption of the centrifuged clearcoating solution was measured with an ultraviolet spectrophotometerUV 2450 (SHIMADZU, Japan). The chemical structure of the functionlayer was characterized by Fourier transform infrared spectroscopy(FT-IR, Perkin-Elmer, USA) and X-ray photoelectron spectra (XPS,AXIS UltraDLD, SHIMADZU, Japan). The water contact angles ofmembranes were measured by a contact angle measuring system (G10Kruss, Germany). At least five values at different sites were attained tocalculate an average. The morphologies of the membranes werecharacterized by scanning electron microscopy (SEM, HitachiSU8000). Element composition of the functional layer was concur-rently analyzed by Energy-dispersive X-ray spectroscopy (EDX) on thesame equipment. Surface roughness of the membranes was observedby atomic force microscopy (AFM, Nanoscope IIIA) in air atmosphere.Mechanical strength of the membranes (50 mm×15 mm×0.13 mm)was evaluated on an AG-1 Universal Tester (SHIMADZU, Japan) at astrain rate of 5 mm min−1, and five membranes of each type weretested to calculate the average value.

2.4. Membrane performance evaluation

All filtration experiments were carried out at 5 bar using a self-made dead-end stirred-cell filtration apparatus (Fig. A1). The effectivemembrane area was 21.2 cm2 and at least three independentlyprepared membranes were tested. The membranes were conditionedfor 2 h at 5 bar to reach a steady value before permeance and rejectionmeasurements. Solutions of dyes (35 μmol L−1) in a variety of solventswere used as feeds. The feed was stirred at 700 rpm in the cell tominimize concentration polarization. The membrane permeance wasmeasured as given in Eq. (1):

VA t ΔP

P =× × (1)

where P is the permeance (L m−2 h−1 bar−1), V (L) is the volume ofpermeate in an operating time t (h), A is the effective area (m2) and ΔPis the applied pressure across the membrane (bar).

The rejection of dyes was calculated from Eq. (2):

⎛⎝⎜⎜

⎞⎠⎟⎟R

CC

= 1 − × 100%p

f (2)

where Cp and Cf were the dye concentrations in the permeate and thefeed, respectively. Dye concentrations were measured by a UV–visCintra20-GBC apparatus.

3. Results and discussion

3.1. Reactive behaviour in coating solution and possible mechanism

Optical images of coating solutions with different POSS-NH3+Cl-/

catechol mass ratio after 2.5 h was shown in Fig. 2(A). The coatingsolution with sole catechol was light-colored, while it gradually turneddark brown as a function of POSS-NH3

+Cl-/catechol mass ratio from

1:1 to 10:1. The reactions between the two parties were revealed byUV–vis spectroscopy. As a control, the curves of the sole catechol (atneutral pH) and POSS-NH3

+Cl- (at pH 8.5) solutions were alsopresented. Fig. 2(B) shows that the solution with sole catechol atneutral pH exhibits one significant absorption peak at 275 nm. Thepresence of a new absorption peak at 330 nm at pH 8.5 may be due tothe oxidation of catechol into o-quinone. This reaction is similar to theoxidation of dopamine (270 nm peak) into dopaquinone (350 nm peak)during dopamine self-polymerization [24]. The intensity of this peakwas magnified with the increase of POSS-NH3

+Cl-/catechol mass ratio,indicating that amine group of POSS particle could accelerate theoxidation of catechol [25]. Meanwhile, the peak at 275 nm broadened,which is probably an evidence of cross-linking of quinone into phenol-coupled oligomers [26]. It should be noted that an absorption peak atabout 480 nm is generally reported for dopamine coating solution andrelated to the intramolecular cyclization and subsequent oxidizationreactions during polymerization [24]. However, this peak was notobserved in Fig. 2(B), which is probably due to the fact that theintrinsic isolation nature between catechol and amine group has ruledout the possibility of intramolecular cyclization in our system [18].

Quite interestingly, some aggregates can be observed in the coatingsolution with a POSS-NH3

+Cl-/catechol mass ratio of 5:1, whereas novisible aggregates can be found in solutions with 0:1, 1:1 and 10:1(Fig. 2(A)). Comparing to sole catechol coating (0:1), the prime aminegroup in co-deposited solution can facilitate the formation of aggre-gates by contributing to the extra cation-π interactions (e.g. NH3

+-quinone/benzene/phenol) [27]. At a relatively lower POSS-NH3

+Cl-/catechol mass ratio, such as 1:1, the aggregation rate in solution couldat a low level due to the large mean free path between POSS-catecholintermediates. With increasing the POSS-NH3

+Cl-/catechol mass ratio,the concentration of POSS-catechol intermediates increased and themean free path between them reduced, which substantially promotedthe aggregation rate in solution. Thus, some visible aggregates pre-cipitated when the POSS-NH3

+Cl-/catechol mass ratio increased to 5:1.However, further increase in mass ratio of POSS-NH3

+Cl-/catechol to10:1 would result in the excess of amine groups in solution, whichwould disturb the non-covalent interactions between POSS-catecholintermediates and inhibit the formation of aggregates [28,29].

The aggregates were collected, washed with DI water and char-acterized by FT-IR and XPS in order to identify their chemicalstructure. As shown in Fig. 3(A), in the curve of catechol, the peaksat 3450 and 3325 cm−1 are assigned to -OH, and the multiple peaks inthe range of 1630–1200 cm−1 are related to aromatic ring framevibration. In the spectrum of POSS-NH3

+Cl-, the peaks located at2950 and 1110 cm−1 are related to ammonium salt stretching vibrationand Si-O-Si asymmetric vibration in POSS cage, respectively [30,31].Compared with POSS-NH3

+Cl-, the spectrum of aggregates show notonly the peak assigned to Si-O-Si group in POSS cage but also aromaticring characteristic peaks related to catechol. In addition, the peaksrelated to -CH2- and -OH/-NH- can also be observed in the curve ofaggregates. No discernable peak assigned to carbonyl group can beobserved, which could be due to the absorbance was obscured by thebroad peak centered at 1580 cm−1[32]. Therefore, the FT-IR resultsdemonstrate that the cross-linking reaction between POSS-NH3

+Cl-

and catechol.The XPS spectrum of the aggregates in Fig. 3(B) exhibits C, N, O

and Si peaks simultaneously, further confirming the formation ofhybrid aggregations. It should be pointed out that the sole POSS-NH3

+Cl- did not lead to any aggregate. The N 1 s spectra of aggregateswere fitted with two peaks assigned to C-NH3

+ at 401.3 eV and -C-NHat 399.2 eV (Fig. 3(C)). The major contribution to N 1 s came from -C-NH, inferring the Michael addition reaction during polymerization. Theabsence of Cl peak in the spectra suggests ammonium chloride salt didnot existed in the aggregates. The possible production of C-NH3

+ couldbe ascribed to spontaneous proton transfer from the acid (catechol, C-OH) to the amine (C-NH2) [14]. The ratio of the intensities of the two N

Y.C. Xu et al. Journal of Membrane Science 526 (2017) 32–42

34

1 s peaks indicates that about 6 of the 8 functional groups of the POSSmolecule are involved in the Michael addition reaction. It is intriguingto find that the peak at a lower binding energy, which was generallyreported in N 1 s spectra of polydopamine due to the intramolecularcyclization and subsequent oxidization reactions [14], was not beenobserved in this study, further confirming the conclusion from UV–vismeasurements. In addition, analyses of C 1 s, O 1 s and Si 2 s (Fig. A2)

also proved the existence of quinone and the formation of catechol-POSS complex.

As a result, we proposed a possible polymerization mechanism ofPOSS-NH3

+Cl- and catechol, as shown in Fig. 4. In an alkaline solution,catechol was oxidized to quinoid form [24]. Meanwhile, the ammoniumchloride groups in POSS-NH3

+Cl- were deprotonated to form primaryamines, which subsequently reacted with quinoid via Michael addition

Fig. 2. (A) Optical images and (B) UV–vis spectra of typical coating solutions.

Fig. 3. (A) FTIR spectra of catechol, aggregates and POSS-NH3+Cl-, (B) XPS spectra and (C) N 1 s spectra of aggregates.

Y.C. Xu et al. Journal of Membrane Science 526 (2017) 32–42

35

reaction to yield an enedione structure. Subsequently, the cross-linkingbetween quinoid and enedione via a reverse dismutation reactionresulted in phenol-coupled oligomers [26]. The covalent bonding inoligomers and the noncovalent bonding, such as π-π stacking, hydro-gen bonding and charge transfer interactions, among oligomers con-tributed to the aggregate in solution or co-deposition on substrate [33].Additionally, the intramolecular cyclization reaction did do not happenin our system based on XPS and UV–vis analyses.

3.2. Chemical properties of nanocomposite membranes

The optical images of support and the co-deposited membranesshow that the color of the membranes gradually changed from yellow todark brown as the mass ratio of POSS-NH3

+Cl-/catechol increased (Fig.A3), implying the possible formation of co-deposition on the mem-brane surfaces. The presence of POSS on the co-deposited membranewas confirmed by ATR-FTIR (Fig. 5). A new peak related to Si-O-Sigroup in POSS cage emerged at 1110 cm−1 after co-deposition,indicating that the POSS structure has been successfully depositedonto cross-linked PI supports. Some other peaks observed in aggre-gates are undiscernable in the co-deposited membrane, which could bedue to the few amount of co-deposition on membrane surface.

The XPS wide scanned spectra of the membranes are shown inFig. 6(A). All the membrane shared three major peaks of C 1 s, N 1 sand O 1 s. Si 2 s and Si 2p peaks emerged with POSS-NH3

+Cl- addition,and the Si atom concentration showed a maximum value for 5:1scenario as shown in Table 1. The element ratios of C/Si, N/Si and O/Siin each co-deposited membrane also exhibited the same trends. Inaddition, Cl 1 s peak emerged when the mass ratio of POSS-NH3

+Cl-/

catechol increases to 6:1 and 10:1. Membrane relative weight gain alsofollowed the same trend with that of the Si concentration (Fig. 6(B)),and 5:1 is probably the optimal ratio for the deposition on membranesurface, which is corresponding to the formation of aggregations incoating solutions. Moreover, the absence of aromatic N peak again inthe N1s spectra of composite membranes, as shown in Fig. 6(C),further confirms that intramolecular cyclization did not occur duringco-deposition process.

Fig. 4. Possible polymerization mechanism of POSS-NH3+Cl- and catechol.

Fig. 5. ATR-FTIR spectra of support, POSS-NH3+Cl-/catechol 5:1 solution coated

membrane and POSS-NH3+Cl-.

Y.C. Xu et al. Journal of Membrane Science 526 (2017) 32–42

36

3.3. Morphologies, surface properties and mechanical properties ofvarious membranes

Fig. 7 shows the morphologies of the PI support and the compositemembranes. Nano-sized pores can be observed on the top surface ofsupport and the one treated by sole catechol, suggesting that thedeposition amount of polyphenol is not enough to completely cover thesupport top surface and it is unlikely to construct a dense coating bycatechol alone [27]. Obviously, with the assistance of POSS-NH3

+Cl-,the nanopores vanished and a coating appeared on the membrane topsurface due to the enhancement of prime amine on co-depositionprocess by introducing the extra cation-π interactions [27]. When themass ratio of POSS-NH3

+Cl-/catechol increased to 5:1, some nano-spheres emerged on membrane top surface, implying the improved co-deposition behaviour. Analogous nanospheres have also been observedin polydopamine modified substrates [24,34,35]. Further increasingPOSS-NH3

+Cl-/catechol mass ratio to 10:1, although the co-deposition

on support still occurred, the surface became relatively smoother due tothe excessive amine groups suppressed the co-deposition process. Thecross-sectional images of modified membranes showed no significantboundaries between support and coating layer. This can be explainedby the following reasons: (1) catechol and POSS-NH3

+Cl- molecularspenetrated into the nanopores of supports and polymerized in thepores, obscuring the boundaries [29]; (2) the coating is too thin to bedetected. The blurry boundary between support and coating layer hasalso been reported in some polydopamine modified substrates[29,36,37]. No appreciable morphological changes can be observedfrom the bottom surface and cross-section of membranes (Fig. A4).Optical image of the modified membrane shows that the top surface ofmembrane is significantly darker than the bottom surface (Fig. A5).These indicate that the co-deposition mainly occurred at the topsurface of support.

AFM images show that the smooth surface of pristine supportbecame rough after coating. The roughness of coated membranesexhibited a maximum value for the one with a POSS-NH3

+Cl-/catecholmass ratio of 5:1. Furthermore, EDX mapping reveals that C, N, O andSi were distributed homogenously on the membrane surfaces (Fig. A6),indicating the homogeneity of the hybrid coatings.

Top surface SEM and AFM images of composite membranes interms of coating time are shown in Fig. 8. The POSS-NH3

+Cl-/catecholmass ratio was fixed at 5:1. It can be see that the amount nanopores onpristine support notably depressed and the surface became rough aftera coating time of 0.5 h. These indicated that the co-deposition occurrednot only at the surface of the support but also within the membranepore structure during the initial stages of the co-deposition. Similarobservations were also reported in polydopamine modified membranes[35,38–41]. With a further increase in coating time, the pores werealmost invisible and some nanospheres appeared on the membrane

Fig. 6. (A) XPS spectra and (B) relatively weight gain of typical membranes, and (C) N 1 s spectra of 5:1 co-deposited membrane.

Table 1Elemental compositions of typical membranes as determined by XPS.

Membrane Composition (at%) Atom ratio

C N O Si Cl C/Si N/Si O/Si Cl/Si

Support 76.07 9.42 14.51 0 0 – – – –

0:1 73.14 9.57 16.23 0 0 – – – –

1:1 66.98 7.66 23.96 1.40 0 48.84 5.47 17.11 04:1 64.80 9.25 21.70 4.25 0 15.25 2.18 5.11 05:1 63.17 9.12 22.37 5.34 0 11.83 1.71 4.19 06:1 63.60 9.85 22.44 3.91 0.19 16.27 2.52 5.74 0.0510:1 65.40 10.84 20.53 2.94 0.28 22.24 3.69 6.98 0.10

Y.C. Xu et al. Journal of Membrane Science 526 (2017) 32–42

37

surface, indicating the formation of a polymer layer on support[35,38,42].

Water contact angles of various membranes were demonstrated inFig. 9. The crosslinked PI support showed a water contact angle of 58°,which is consistent with the previous reported results [43,44]. Afterdeposited by catechol, the membrane water contact angle decreased to50°, which could be related to the hydrophilic -OH group on membranesurface after deposition. With the addition of POSS-NH3

+Cl-, the watercontact angle initially increased with POSS-NH3

+Cl-/catechol massratio ranging from 1:1 to 5:1, then decreased with POSS-NH3

+Cl-/catechol mass ratio ranging from 5:1 to 10:1. The initial increase ofwater contact angle could be due to the presence of hydrophobic POSSgroups on surface and the increased surface roughness, although theremaining -NH3

+ also affect the surface properties [45]. The decreaseof water contact angle with further increasing catechol/POSS massratio could be rationally explained by the decrease in both co-deposi-tion amount and surface roughness.

The mechanical strength of OSN membrane is a key factor toevaluate their usefulness in practical application [46,47]. It has beenreported that the incorporation of POSS derivatives into polymermaterials can dramatically improve polymer mechanical properties[48]. So we are intrigued if the membrane mechanical properties havealso been improved after co-deposition. The tensile strength andbreaking elongation of various membranes was shown in Fig. 10. Itwas obvious that sole catechol deposited membrane exhibited nosignificant different mechanical properties with PI support. While,

both tensile strength and breaking elongation have increased afterPOSS-NH3

+Cl- addition, indicating the reinforcement effect of POSSnanoparticles. The tensile strength has increased from 3.22 to4.77 MPa with almost 48% increase and the breaking elongation alsoimproved from 1.95–4.04% with almost 107% improvement. Theimprovement of tensile strength and breaking elongation could berelated to the covalent bond between POSS and polymer matrix, as well

Fig. 7. Top surface (top), cross-sectional (middle) SEM and AFM (bottom) images of typical membranes.

Fig. 8. Top surface SEM (top) and AFM images (bottom) of composite membrane in terms of coating time.

Fig. 9. Water contact angles of typical co-deposited membranes (coating time 2.5 h).

Y.C. Xu et al. Journal of Membrane Science 526 (2017) 32–42

38

as a certain degree of branched structure originated from POSS[49,50]. The highest breaking elongation at POSS-NH3

+Cl-/catecholmass ratio of 5:1 might be due to the highest co-deposition amount atthis ratio.

3.4. Nanofiltration performance in organic solvent system

3.4.1. Effect of catechol/POSS mass ratio and coating time onseparation performance

The organic solvent nanofiltration experiments of co-depositedmembranes were carried out with a 35 μM RB solution in EtOH toevaluate the effect of POSS-NH3

+Cl-/catechol mass ratio on separationperformance. The coating time was fixed at 2.5 h. As shown inFig. 11(A), the sole catechol coated membrane (POSS-NH3

+Cl-/cate-chol mass ratio of 0:1) exhibited no obvious change in terms of bothpermeance and RB rejection, due to the limited coating efficiency asevidenced by surface morphology in Fig. 7. With the addition of POSS-NH3

+Cl-, the RB rejection presents an up-and-down trend as a functionof the catechol/POSS mass ratio. Meanwhile, the permeance exhibiteda “trade-off” trend. The optimal membrane performance with apermeance of 3.85 L m−2 h−1 bar−1 and RB rejection of 87% wasobtained at a POSS-NH3

+Cl-/catechol mass ratio of 5:1. The initialdecrease in permeance is probably due to the fact that a continuouslayer of POSS-NH3

+Cl-/catechol coating is gradually formed on thesurface of the PI support and, which could enhance solvent permeationresistance. As a result, the EtOH permeance decreased from184 L m−2 h−1 bar−1 to 3.85 L m−2 h−1 bar−1 and RB rejection in-creased from 10–87%.

However, when further increasing POSS-NH3+Cl-/catechol mass

ratio to 10:1, the co-deposition amount lessened and membranesurface became less hydrophobic, leading to an increased permeanceand decreased RB rejection. So, the mass ratio of POSS-NH3

+Cl-/catechol was fixed at 5:1 in the following tests.

Fig. 11(B) shows the membrane performance with aspect todifferent coating time. The permeance decreased and RB rejectionincreased as prolonging the coating time. Thickness of the functionallayer may gradually increase with coating duration, which increasingpermeation resistance. The EtOH permeance decreased from30 L m−2 h−1 bar−1 to 1.26 L m−2 h−1 bar−1, while RB rejection in-creased from 9% to 99% when prolong coating time to 3 h. Withfurther increase of coating time, the effective rejection of RB wasachieved; however, the permeation resistance increased and per-meance decreased continuously. Collectively, the POSS-NH3

+Cl-/cate-chol coating was optimized in terms of coating ratio and duration. APOSS-NH3

+Cl-/catechol ratio of 5:1 and a coating time of 3 h were

employed in the following study.

3.4.2. Performance for separating various dyesThe separation of some low molecular weight dyes from EtOH is

shown in Fig. 12. The optical images of feeds and filtrates of dyesolutions were shown in Fig. A7. The chemical structures, molarweights and charges of these dyes were summarized in Fig. A8. TheEtOH permeance for different dyes are similar, suggesting the type ofdyes has no significant influence on solvent flux. Dyes with relatively

Fig. 10. Tensile strength and breaking elongation of various membranes (coating time2.5 h).

Fig. 11. (A) Separation performances of co-deposited membranes in terms of POSS-NH3

+Cl-/catechol mass ratio (coating time 2.5 h); (B) separation performances of co-deposited membranes in terms of coating time (POSS-NH3

+Cl-/catechol mass ratio of5:1).

Fig. 12. Separation performances of co-deposited membrane towards various dyes inEtOH.

Y.C. Xu et al. Journal of Membrane Science 526 (2017) 32–42

39

higher molecular weight, such as RB (MW 1071) and MB (MW 800)were completely rejected, while other dyes were partly rejected. Despiteits lower molecular weight, the CV rejection was slightly higher thanOG rejection, and this could be related to the electrostatic repulsionbetween the amine groups on membrane surface and the positivelycharged CV, although the dielectric constant of EtOH is rather low(25.3 compared to 80.1 for water) [51].

3.4.3. Separation performance in various solvents and long termperformance in DMF

Fig. 13(A) depicts that the separation performance of as-preparedmembrane in various solvents. Thanks to the cross-linked structure ofboth PI support and hybrid selective layer, the composite membraneexhibited good separation performance in a variety of solvents,including harsh solvents such as THF, DCM and DMF, which aretroublesome in OSN applications. No obvious trends between solventpermeance and dye rejection have been observed. In fact, the solventtransport mechanism through OSN membrane is more complex thanaqueous applications. The membrane performances in organic solventare related to the solvent properties (viscosity, molar volume andsurface tension), mutual interactions between solvent and membrane,as well as between solvent and solute [52,53]. However, for the alcoholhomologous series, the dependence of performance on solvent is clear.For example, the permeance of alcohol solvents are in the order of

MeOH > EtOH > IPA. The highest permeance observed for MeOHcould be due to its lowest molar volume and viscosity. On the otherhand, the lowest permeance obtained for IPA could be attributed to itshighest molar and viscosity [5]. In addition, RB rejections in thealcohol solvents showed an opposite trend with the permeance. Thesecould be related to the convective coupling between solute and solvent,and the “dragging effect” of solvent [54].

Long term performance of composited membrane was tested usingsolutions of RB in EtOH and DMF as feed. Fig. A9 shows that themembrane performances did not change over two days. A weight lossless than 0.4% was obtained in an immersion test of co-depositedmembrane in DMF for two weeks, further demonstrating the stabilityof co-deposited membrane.

3.4.4. Comparison of separation performance with some reportedmembranes

Table 2 summarizes the separation performance of reported TFNOSN membranes and a relevant composite membrane. The co-depos-ited membrane in this study shows a comparable MeOH permeance tothe (PA/MOFs)/PI membranes containing MIL-53 or ZIF-8, althoughthe solute rejection is a little lower [6]. Comparing with (PDMS/Zeolite)/PAN, which shows slightly higher rejections of a highermolecular weight solute, the co-deposited membrane in this studyexhibits a 6.24 times higher toluene permeance and a 4.32 times higherDCM permeance [7].

It is intriguing to find that the membrane exhibits a 1.52 timeshigher MeOH permeance and 1.50 times greater IPA permeance than arecently reported catechol and polyethyleneimine co-deposited mem-brane with better or similar rejections [20]. Furthermore, we haveconducted similar coating using dopamine or catechol/PEI to replacecatechol/ POSS-NH3

+Cl-. As shown in Table 3, the (catechol/POSS)/PIstill shows better separation performance than polydopamine/PI and(catechol/PEI)/PI. The cage pore diameter of a POSS is ca. 0.53 nm,which is in the range of typical nanofiltration [21]. One possibleexplanation is that comparing to the flexible organic matrix layer incatechol and polyethyleneimine co-deposited membrane, the rigidinorganic POSS cages could provide preferential flow path for smallsolvent molecular.

4. Conclusions

In summary, a one-step mussel-inspired co-deposition strategy hasbeen discovered to fabricate novel nanocomposite OSN membranes.According to the chemical characterizations of membranes, the co-deposition mechanism is disclosed which is similar to that of dopamine

Fig. 13. Separation performances of co-deposited membrane in various organicsolvents.

Table 2Comparison of the separation performance in this study with reported TFN OSN membranes and a relevant composite membrane.a

Membrane material Nanoparticles Solvent Permeance (L m−2 h−1 bar−1) Solute (MW) Rejection (%) Ref.

(PA/MOFs)/PI MIL-53 MeOH 2.3 Polystyrenes (236–1200) 99 (1000 g mol−1) [6]ZIF-8 MeOH 2.5 Polystyrenes (236–1200) 99 (1000 g mol−1)MIL-101 MeOH 4.2 Polystyrenes (236–1200) 99 (1000 g mol−1)

(PDMS/Zeolite)/PAN ZSM-5 toluene 0.58 Wilkinson catalyst (952) 98 [7]DCM 0.71 Wilkinson catalyst (952) 93

(PA/SiO2)/PAN SiO2 IPA 1.45 PEG (1000) 99 [8](PA/TiO2)/PAN TiO2 IPA 0.3 PEG (200) 95 [8](PA/WCNTs)/PES SWCNTs MeOH 7.39 Brilliant blue R (826) 89 [9](Catechol/PEI)/PAN – MeOH 1.8 Bromothymol blue (624) 68 [20]

IPA 0.26 BTB (624) 99

(Catechol/POSS)/PI POSS MeOH 2.23 RB (1017) 84.2 This workIPA 0.39 RB (1017) 99toluene 3.62 SB 35 (350) 86.7DCM 3.07 SB 35 (350) 80.5

a PA, PDMS, PEI, PAN are abbreviations of polyamide, polydimethylsiloxane, polyethyleneimine and polyacrylonitrile.

Y.C. Xu et al. Journal of Membrane Science 526 (2017) 32–42

40

self-polymerization but without the involvement of intramolecularcyclization reaction. The mass ratio of POSS-NH3

+Cl-/catechol iscrucial to the physicochemical and separation properties of ourdeveloped membranes. The optimal nanocomposite membrane exhib-ited an EtOH permeance of 1.26 L m−2 h−1 bar−1 with RB rejection of99%. Most importantly, the membrane can demonstrate good separa-tion performance toward dyes in a wide range of organic solventsincluding challenging solvents such as THF, DCM and DMF contrib-uted by the cross-linked structure of both PI support and hybridseparating layer compositing of unique inorganic POSS cages.

Acknowledgements

This work was supported by National Natural Science Foundationof China (21676063, U1462103), the Open Research Fund Program ofCollaborative Innovation Center of Membrane Separation and WaterTreatment of Zhejiang Province, Harbin Science and TechnologyInnovation Talent Funds (2014RFXXJ028), and HIT Environmentand Ecology Innovation Special Funds (HSCJ201619).

Appendix A. Supplementary material

Supplementary data associated with this article can be found in theonline version at doi:10.1016/j.memsci.2016.12.026.

References

[1] P. Marchetti, M.F. Jimenez Solomon, G. Szekely, A.G. Livingston, Molecularseparation with organic solvent nanofiltration: a critical review, Chem. Rev. 114(2014) 10735–10806.

[2] G. Székely, M. Gil, B. Sellergren, W. Heggie, F.C. Ferreira, Environmental andeconomic analysis for selection and engineering sustainable API degenotoxificationprocesses, Green Chem. 15 (2013) 210–225.

[3] G. Szekely, M.F. Jimenez-Solomon, P. Marchetti, J.F. Kim, A.G. Livingston,Sustainability assessment of organic solvent nanofiltration: from fabrication toapplication, Green Chem. 16 (2014) 4440–4473.

[4] N.A.A. Sani, W.J. Lau, A.F. Ismail, Polyphenylsulfone-based solvent resistantnanofiltration (SRNF) membrane incorporated with copper-1,3,5-benzenetricar-boxylate (Cu-BTC) nanoparticles for methanol separation, RSC Adv. 5 (2015)13000–13010.

[5] L. Shao, X. Cheng, Z. Wang, J. Ma, Z. Guo, Tuning the performance of polypyrrole-based solvent-resistant composite nanofiltration membranes by optimizing poly-merization conditions and incorporating graphene oxide, J. Membr. Sci. 452 (2014)82–89.

[6] S. Sorribas, P. Gorgojo, C. Tellez, J. Coronas, A.G. Livingston, High flux thin filmnanocomposite membranes based on metal-organic frameworks for organic solventnanofiltration, J. Am. Chem. Soc. 135 (2013) 15201–15208.

[7] L.E. Gevers, I.F. Vankelecom, P.A. Jacobs, Zeolite filled polydimethylsiloxane(PDMS) as an improved membrane for solvent-resistant nanofiltration (SRNF),Chem. Commun. (2005) 2500–2502.

[8] H. Zhang, H. Mao, J. Wang, R. Ding, Z. Du, J. Liu, S. Cao, Mineralization-inspiredpreparation of composite membranes with polyethyleneimine–nanoparticle hybridactive layer for solvent resistant nanofiltration, J. Membr. Sci. 470 (2014) 70–79.

[9] S. Roy, S.A. Ntim, S. Mitra, K.K. Sirkar, Facile fabrication of superior nanofiltrationmembranes from interfacially polymerized CNT-polymer composites, J. Membr.Sci. 375 (2011) 81–87.

[10] S.P. Sun, T.S. Chung, Outer-selective pressure-retarded osmosis hollow fibermembranes from vacuum-assisted interfacial polymerization for osmotic powergeneration, Environ. Sci. Technol. 47 (2013) 13167–13174.

[11] X. Yao, L. Guo, X. Chen, J. Huang, M. Steinhart, Y. Wang, Filtration-based

synthesis of micelle-derived composite membranes for high-flux ultrafiltration, ACSAppl. Mater. Interfaces 7 (2015) 6974–6981.

[12] H. Lee, S.M. Dellatore, W.M. Miller, P.B. Messersmith, Mussel-inspired surfacechemistry for multifunctional coatings, Science 318 (2007) 426–430.

[13] F. Bernsmann, A. Ponche, C. Ringwald, J. Hemmerlé, J. Raya, B. Bechinger, J.-C. Voegel, P. Schaaf, V. Ball, Characterization of dopamine-melanin growth onsilicon oxide, J. Phys. Chem. C 113 (2009) 8234–8242.

[14] Y. Ding, L.T. Weng, M. Yang, Z. Yang, X. Lu, N. Huang, Y. Leng, Insights into theaggregation/deposition and structure of a polydopamine film, Langmuir 30 (2014)12258–12269.

[15] Y. Liu, K. Ai, L. Lu, Polydopamine and its derivative materials: synthesis andpromising applications in energy, environmental, and biomedical fields, Chem. Rev.114 (2014) 5057–5115.

[16] L. Pérez-Manríquez, A.R. Behzad, K.-V. Peinemann, Sub-6 nm thin cross-linkeddopamine films with high pressure stability for organic solvent nanofiltration,Macromol. Mater. Eng. 301 (2016) 1437–1442.

[17] J. Zhao, Y. Su, X. He, X. Zhao, Y. Li, R. Zhang, Z. Jiang, Dopamine compositenanofiltration membranes prepared by self-polymerization and interfacial poly-merization, J. Membr. Sci. 465 (2014) 41–48.

[18] H. Wang, J. Wu, C. Cai, J. Guo, H. Fan, C. Zhu, H. Dong, N. Zhao, J. Xu, Musselinspired modification of polypropylene separators by catechol/polyamine for Li-ionbatteries, ACS Appl. Mater. Interfaces 6 (2014) 5602–5608.

[19] W.-Z. Qiu, H.-C. Yang, L.-S. Wan, Z.-K. Xu, Co-deposition of catechol/polyethyle-neimine on porous membranes for efficient decolorization of dye water, J. Mater.Chem. A 3 (2015) 14438–14444.

[20] Y.C. Xu, Z.X. Wang, X.Q. Cheng, Y.C. Xiao, L. Shao, Positively charged nanofil-tration membranes via economically mussel-substance-simulated co-deposition fortextile wastewater treatment, Chem. Eng. J. 303 (2016) 555–564.

[21] E.G. Vieira, I.V. Soares, G. Pires, R.A.V. Ramos, D.R. do Carmo, N.L. Dias Filho,Study on determination and removal of metallic ions from aqueous and alcoholicsolutions using a new POSS adsorbent, Chem. Eng. J. 264 (2015) 77–88.

[22] Y. Li, T.-S. Chung, Molecular-level mixed matrix membranes comprising Pebax®and POSS for hydrogen purification via preferential CO2 removal, Int. J. Hydrog.Energy 35 (2010) 10560–10568.

[23] J.H. Moon, A.R. Katha, S. Pandian, S.M. Kolake, S. Han, Polyamide–POSS hybridmembranes for seawater desalination: effect of POSS inclusion on membraneproperties, J. Membr. Sci. 461 (2014) 89–95.

[24] Q. Wei, F. Zhang, J. Li, B. Li, C. Zhao, Oxidant-induced dopamine polymerizationfor multifunctional coatings, Polym. Chem. 1 (2010) 1430.

[25] W.-Z. Qiu, Y. Lv, Y. Du, H.-C. Yang, Z.-K. Xu, Composite nanofiltration membranesvia the co-deposition and cross-linking of catechol/polyethylenimine, RSC Adv. 6(2016) 34096–34102.

[26] M.C. van der Leeden, Are conformational changes, induced by osmotic pressurevariations, the underlying mechanism of controlling the adhesive activity of musseladhesive proteins, Langmuir 21 (2005) 11373–11379.

[27] C. Lim, J. Huang, S. Kim, H. Lee, H. Zeng, D.S. Hwang, Nanomechanics ofpoly(catecholamine) coating in aqueous solutions, Angew. Chem. Int. Ed. 55 (2016)3342–3346.

[28] H.-C. Yang, K.-J. Liao, H. Huang, Q.-Y. Wu, L.-S. Wan, Z.-K. Xu, Mussel-inspiredmodification of a polymer membrane for ultra-high water permeability and oil-in-water emulsion separation, J. Mater. Chem. A 2 (2014) 10225–10230.

[29] J. Zhao, C. Fang, Y. Zhu, G. He, F. Pan, Z. Jiang, P. Zhang, X. Cao, B. Wang,Manipulating the interfacial interactions of composite membranes via a mussel-inspired approach for enhanced separation selectivity, J. Mater. Chem. A 3 (2015)19980–19988.

[30] N. Naderi, S.Y. Madani, A. Mosahebi, A.M. Seifalian, Octa-ammonium POSS-conjugated single-walled carbon nanotubes as vehicles for targeted delivery ofpaclitaxel, Nano Rev. 6 (2015) 28297.

[31] X. Ge, L. Dong, L. Sun, Z. Song, R. Wei, L. Shi, H. Chen, New nanoplatforms basedon UCNPs linking with polyhedral oligomeric silsesquioxane (POSS) for multi-modal bioimaging, Nanoscale 7 (2015) 7206–7215.

[32] D.R. Dreyer, D.J. Miller, B.D. Freeman, D.R. Paul, C.W. Bielawski, Elucidating thestructure of poly(dopamine), Langmuir 28 (2012) 6428–6435.

[33] S. Hong, Y.S. Na, S. Choi, I.T. Song, W.Y. Kim, H. Lee, Non-Covalent Self-Assemblyand Covalent Polymerization Co-Contribute to Polydopamine Formation, Adv.Funct. Mater. 22 (2012) 4711–4717.

[34] J. Jiang, L. Zhu, L. Zhu, B. Zhu, Y. Xu, Surface characteristics of a self-polymerizeddopamine coating deposited on hydrophobic polymer films, Langmuir 27 (2011)

Table 3Separation performance of membranes modified with different mussel-inspired coating.

Membrane material Coatingtime (h)

EtOH permeance(L m−2 h−1 bar−1)

RB rejection(%)

polydopamine/PIa 3 2.03 556 0.32 74

(Catechol/PEI)/PIb 3 0.58 92(Catechol/POSS)/PI 3 1.26 99

a Dopamine concentration, 2 mg mL−1[55].b Catechol concentration, 3 mg mL−1; catechol/PEI mass ratio, 1:1; PEI molecular weight, 600 [20].

Y.C. Xu et al. Journal of Membrane Science 526 (2017) 32–42

41

14180–14187.[35] Y. Li, Y. Su, X. Zhao, X. He, R. Zhang, J. Zhao, X. Fan, Z. Jiang, Antifouling, high-

flux nanofiltration membranes enabled by dual functional polydopamine, ACSAppl. Mater. Interfaces 6 (2014) 5548–5557.

[36] Y. Li, Y. Su, J. Li, X. Zhao, R. Zhang, X. Fan, J. Zhu, Y. Ma, Y. Liu, Z. Jiang,Preparation of thin film composite nanofiltration membrane with improvedstructural stability through the mediation of polydopamine, J. Membr. Sci. 476(2015) 10–19.

[37] H. Niu, S. Wang, T. Zeng, Y. Wang, X. Zhang, Z. Meng, Y. Cai, Preparation andcharacterization of layer-by-layer assembly of thiols/Ag nanoparticles/polydopa-mine on PET bottles for the enrichment of organic pollutants from water samples,J. Mater. Chem. 22 (2012) 15644–15653.

[38] G. Han, S. Zhang, X. Li, N. Widjojo, T.-S. Chung, Thin film composite forwardosmosis membranes based on polydopamine modified polysulfone substrates withenhancements in both water flux and salt rejection, Chem. Eng. Sci. 80 (2012)219–231.

[39] R. Ding, H. Zhang, Y. Li, J. Wang, B. Shi, H. Mao, J. Dang, J. Liu, Graphene oxide-embedded nanocomposite membrane for solvent resistant nanofiltration withenhanced rejection ability, Chem. Eng. Sci. 138 (2015) 227–238.

[40] B.D. McCloskey, H.B. Park, H. Ju, B.W. Rowe, D.J. Miller, B.J. Chun, K. Kin,B.D. Freeman, Influence of polydopamine deposition conditions on pure water fluxand foulant adhesion resistance of reverse osmosis, ultrafiltration, and microfil-tration membranes, Polymer 51 (2010) 3472–3485.

[41] J.-H. Jiang, L.-P. Zhu, X.-L. Li, Y.-Y. Xu, B.-K. Zhu, Surface modification of PEporous membranes based on the strong adhesion of polydopamine and covalentimmobilization of heparin, J. Membr. Sci. 364 (2010) 194–202.

[42] M. Li, J. Xu, C.-Y. Chang, C. Feng, L. Zhang, Y. Tang, C. Gao, Bioinspiredfabrication of composite nanofiltration membrane based on the formation of DA/PEI layer followed by cross-linking, J. Membr. Sci. 459 (2014) 62–71.

[43] Y.H. See Toh, F.W. Lim, A.G. Livingston, Polymeric membranes for nanofiltrationin polar aprotic solvents, J. Membr. Sci. 301 (2007) 3–10.

[44] Y.C. Xu, X.Q. Cheng, J. Long, L. Shao, A novel monoamine modification strategytoward high-performance organic solvent nanofiltration (OSN) membrane for

sustainable molecular separations, J. Membr. Sci. 497 (2016) 77–89.[45] H. Hussain, B.H. Tan, G.L. Seah, Y. Liu, C.B. He, T.P. Davis, Micelle formation and

gelation of (PEG-P(MA-POSS)) amphiphilic block copolymers via associativehydrophobic effects, Langmuir 26 (2010) 11763–11773.

[46] I. Soroko, A. Livingston, Impact of TiO2 nanoparticles on morphology andperformance of crosslinked polyimide organic solvent nanofiltration (OSN) mem-branes, J. Membr. Sci. 343 (2009) 189–198.

[47] S.K. Lim, L. Setiawan, T.-H. Bae, R. Wang, Polyamide-imide hollow fibermembranes crosslinked with amine-appended inorganic networks for applicationin solvent-resistant nanofiltration under low operating pressure, J. Membr. Sci. 501(2016) 152–160.

[48] Y. Zhao, D.A. Schiraldi, Thermal and mechanical properties of polyhedral oligo-meric silsesquioxane (POSS)/polycarbonate composites, Polymer 46 (2005)11640–11647.

[49] F.X. Perrin, D.M. Panaitescu, A.N. Frone, C. Radovici, C. Nicolae, The influence ofalkyl substituents of POSS in polyethylene nanocomposites, Polymer 54 (2013)2347–2354.

[50] A. Fina, D. Tabuani, G. Camino, Polypropylene–polysilsesquioxane blends, Eur.Polym. J. 46 (2010) 14–23.

[51] W.K. Son, J.H. Youk, T.S. Lee, W.H. Park, The effects of solution properties andpolyelectrolyte on electrospinning of ultrafine poly(ethylene oxide) fibers, Polymer45 (2004) 2959–2966.

[52] D. Bhanushali, S. Kloos, C. Kurth, D. Bhattacharyya, Performance of solventresis-tant membranes for non-aqueous systems: solvent permeation results and model-ing, J. Membr. Sci. 189 (2001) 1–21.

[53] P. Silva, S. Han, A.G. Livingston, Solvent transport in organic solvent nanofiltrationmembranes, J. Membr. Sci. 262 (2005) 49–59.

[54] X. Li, S. De Feyter, I.F.J. Vankelecom, Poly(sulfone)/sulfonated poly(etherether ketone) blend membranes: morphology study and application in the filtrationof alcohol based feeds, J. Membr. Sci. 324 (2008) 67–75.

[55] X.-L. Li, L.-P. Zhu, J.-H. Jiang, Z. Yi, B.-K. Zhu, Y.-Y. Xu, Hydrophilic nanofiltra-tion membranes with self-polymerized and strongly-adhered polydopamine asseparating layer, Chin. J. Polym. Sci. 2 (2012) 152–163.

Y.C. Xu et al. Journal of Membrane Science 526 (2017) 32–42

42