Contents lists available at ScienceDirect

Journal of Membrane Science

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

Growing covalent organic frameworks on porous substrates for molecule-sieving membranes with pores tunable from ultra- to nanofiltration

Xiansong Shi, Ankang Xiao, Chenxu Zhang, Yong Wang⁎

State Key Laboratory of Materials-Oriented Chemical Engineering, Jiangsu National Synergetic Innovation Center for Advanced Materials, and College of ChemicalEngineering, Nanjing Tech University, Nanjing 211816, PR China

A R T I C L E I N F O

Keywords:Covalent organic frameworksSolvothermal synthesisUltrafiltrationNanofiltrationTunable selectivity

A B S T R A C T

Crystalline imine-based covalent organic frameworks (COFs) have attracted great interest as next-generationwater treatment materials owning to their numerous merits. The construction of COF layers with tunable poresizes ranging from meso- to micropores, which can significantly broaden the membrane selectivity from ultra-(UF) to nanofiltration (NF), has remained unexplored. Herein, we demonstrate the controllable solvothermalsynthesis of COF layers with adjustable pore sizes on anodic aluminum oxide (AAO) substrates. With risingsynthesis duration the modulation on pore sizes of AAO supports can be achieved till the formation of continuousCOF layers, leading to a gradually enhanced selectivity. Importantly, the synthesized microporous COFs shrinkthe pore sizes of AAO substrates and provide additional nanochannels for water transporting, eventuating in a~2–9 times higher permeance than other reported UF membranes with comparable rejections. Once formingcontinuous COF layers, more rigid selectivities based on COF inherent nanochannels (~1.83 nm) can be realizedwith capacities of removing dyes from water or organic solvent. This work provides a facile methodology toconstruct COF-based membranes with broadly tunable separation performance, recommending the membranesfor removing targeted molecules including proteins, nanoparticles, and dyes.

1. Introduction

Membrane separation technology is now widely used in waterpurification [1,2], gas separation [3,4], and biological/chemical pro-cessing [5] benefiting from its cost-effective and environmentallyfriendly merits. Recently, increasing membrane permeance withoutsacrificing selectivity is sought to further improve the separation effi-ciency and reduce the energy consumption [6,7]. In the view of this,several effective strategies have been proposed in the membrane pre-paration to realize a high permeance with a desired selectivity, such asbuilding an ultrathin separation layer [6,8], surface modification[9,10], and incorporation of porous materials [11,12]. For instance,Jiang and coworkers elegantly fabricated smooth sub-8 nm ultrathinpolyamide nanofilms at a free aqueous-organic interface [6], achievinga significantly enhanced permeance by providing a shorter transportingdistance. However, the decreased thickness may in return lead to re-duced mechanical strength, which significantly affects the practicalapplication of membranes. As for the surface modification, it has beendemonstrated that the co-deposition of polydopamine, poly-etheylenimine, and silica nanoparticles in the fabrication of nanofil-tration (NF) membranes can produce tripled water flux compared with

that of the pristine membranes [10]. Nevertheless, the surface mod-ification strategy is usually conducted with several tedious modificationsteps, and the surface activation of the substrate is necessary for thenext deposition. Alternatively, researchers developed a facile method tobuild additional pores in the membranes by incorporating porous ma-terials into the separation layer, thus producing additional channels formolecular transferring. Here, the incorporated materials exhibit twodifferent mechanisms to modify the pores. One is to narrow the poresizes as a result of the growth of materials along the pore walls, and theother is that the inherent pores in the grown materials allow water topermeate through. Therefore, the separation selectivity can be en-hanced with less expense on permeability.

With the favourable development of porous materials, especiallymicroporous materials, numerous of such materials have been in-corporated into the membrane matrix to enhance the membrane per-meance, examples are metal-organic frameworks (MOFs) [13,14],zeolites [15], and covalent organic frameworks (COFs) [11,12]. Toachieve a more pronounced performance, these porous materials aretypically incorporated into polymeric matrixes as additives, thus pro-viding additional channels for molecular transport with an enhancedpermeance. Nevertheless, this incorporation approach is failed to take

https://doi.org/10.1016/j.memsci.2019.01.034Received 11 December 2018; Received in revised form 18 January 2019; Accepted 18 January 2019

⁎ Corresponding author.E-mail address: yongwang@njtech.edu.cn (Y. Wang).

Journal of Membrane Science 576 (2019) 116–122

Available online 25 January 20190376-7388/ © 2019 Elsevier B.V. All rights reserved.

T

full advantages of these microporous materials and the amount of ad-dition is limited (usually< 50%) to avoid the formation of cracks. Evenworse, the channels of these microporous fillers maybe blocked bypolymer chains. Alternatively, researchers also developed other variousmethods, such as interfacial synthesis [16], solvothermal synthesis[17], and assembly of exfoliated nanosheets [18], to prepare con-tinuous selective layer constructed by these porous materials, leading toan excellent nanofiltration performance for dye separation. For in-stance, Hong and coworkers constructed a selective layer using two-dimensional MOF nanosheets, producing ultrafast and energy-savingNF membranes [18]. In this case, the separation performance can beeasily controlled by the thickness of stacked nanosheets. However,these efforts mostly focus on the preparation of NF membranes bene-fitting from the inherent advantages of these novel materials or themodification of ultrafiltration (UF) membranes using their microporouschannels. Thus, the membrane preparation with a wider separationcategory bridging a gap between UF and NF still remains unexplored.

Among above-mentioned novel crystalline materials, COFs are nowdeveloped as a new family of crystalline polymeric networks with well-defined and inherent porosities [19–21]. Their architectures can beprecisely customized by molecular precursors based on reticularchemistry and they exhibit low densities and highly ordered channels of~1–5 nm, making them excellent candidates for numerous applicationssuch as sensors [22,23], catalysis [24,25], gas storages [26,27], andmembrane separations [28,29]. Furthermore, benefitting from theirfavourable hydrophilicity and good water stability, imine-based COFshave been proposed as prominent building blocks for the preparation ofmembranes used for water treatment. A pioneering work was reportedby Banerjee and coworkers, they established a facile and up-scalablemethodology for the fabrication self-standing COF membranes bybaking precursor pairs [30]. Alternatively, other methods includingsolvothermal synthesis [31] and interfacial crystallization [32] havebeen actively developed for the preparation of COF membranes used fordyeing water treatment. Likewise, these methods only focus on theproduction of continuous COF layers, in which the separation me-chanism is mainly determined by their inherent channels and the per-formance can only be adjusted by their inherent pore sizes originatedfrom the structure of building blocks. Thus, preparation of COFs intothe membrane form with tunable pore sizes ranging from ~2–20 nmremains a great challenge, but if achieved, could present promises fortargeted molecular separation applications ranging from UF to NFfeatured with enhanced permeances.

Herein, we aim to fabricate COF-based membranes with adjustableperformances by controlling the growth of COF crystallites on poroussubstrates. The imine-based COF crystallites were synthesized onanodic aluminum oxide (AAO) substrates with 20 nm pore sizes via asimple solvothermal reaction between amines and aldehydes. As a re-sult, pore sizes of AAO substrates can be easily tuned through adjustingthe synthesis durations, temperatures, and monomer concentrations.With the growth of COF crystallites, a gradually narrowed pore chan-nels and eventually a dense COF selective layer can be obtained,eventuating in an adjustment separation performance ranging from UFto NF (Scheme 1). To the best of our knowledge, this is the first report ofconstruction of COF crystallites on porous substrates aimed for variousseparation processes, bridging a gap between UF and NF.

2. Experimental section

2.1. Materials

Synthesis monomers including 1,3,5triformylphloroglucinol (Tp,95%) and p-phenylenediamine (Pa, 97%) were purchased fromTongChuangYuan Pharmaceutical (Sichuan, China) and Aladdin, re-spectively. Sodium hydroxide (NaOH, 96%), acetic acid (AA, 99.5%), 1,4-dioxane (99.5%), mesitylene (99.5%) were obtained from local sup-pliers. Porous AAO substrates (2.5 cm, Whatman) with an average pore

size of ~20 nm and porosity of ~30% were used as supports for COFgrowth. Proteins including bovine serum albumin (BSA, 67 kDa) andovalbumin (OVA, 45 kDa) were obtained from Sigma-Aldrich andAladdin, respectively. Phosphate buffer solution (PBS) tablets werepurchased from MP Biomedicals and used for preparation of aqueousprotein solutions. Monodispersed gold nanoparticles with a particle sizeof 10 nm, dispersed in water, were obtained from British BiocellInternational (BBI) and diluted before filtration tests. Dyes includingacid orange 7 (AO7), eriochrome black T (EB-T), acid fuchsin (AF),reactive blue 19 (RB19), and Congo red (CR) were purchased from localsuppliers and used to evaluate the nanofiltration performance of theproduced membranes. Deionized water (DI water, conductivity: 8–20 µscm-1, pH=6.8, Wahaha Co.) was used in all the experiments. All che-micals were used as received without further purification.

2.2. Synthesis of TpPa-AAO membrane

The TpPa-AAO membrane was prepared by in situ fabricationthrough a simple solvothermal synthesis between Tp (6.3mg,0.03mmol) and Pa (4.9 mg, 0.045mmol) under the presence of AA(12M, 80 μL) (Scheme S1). First, the monomer pairs were dissolved inthe mixed solvent (5 mL 1, 4-dioxane and 5mL mesitylene) to form atransparent homogeneous solution in a Teflon-lined stainless steel au-toclave, then the catalyst, AA, was added and the solution was aged forabout 5min to generate a certain amount of crystallites. The AAOsubstrate was then horizontally placed into the aged solution afterbubbling with nitrogen for 5min. The autoclave was then heated at80 °C or 100 °C for various durations, thus-produced TpPa-AAO mem-branes were thoroughly washed with 1,4-dioxane and ethanol severaltimes, and then naturally dried overnight.

2.3. Synthesis of TpPa powders

Similar to the synthesis of TpPa-AAO membranes, The TpPa pow-ders were produced using the traditional solvothermal synthesis.Specifically, 0.3 mmol Tp (63mg) and 0.45mmol Pa (49mg) weredissolved in the mixture containing 7.5 mL 1, 4-dioxane and 7.5mLmesitylene followed by the addition of 1mL AA (12M) and nitrogenbubbling for 5min, then the autoclave containing COF monomers washeated at 120 °C for 72 h. Thus-obtained powders were thoroughlywashed with 1, 4-dioxane and ethanol and drying at 100 °C overnightbefore further characterizations.

2.4. Characterizations

Morphologies and elemental analysis of the membranes were per-formed on a field-emission scanning electron microscope (SEM, HitachiS-4800, Japan) attached with an energy-dispersive spectrometry (EDX)

Scheme 1. Schematic illustration of synthesis of TpPa-AAO membranes for UFand NF.

X. Shi et al. Journal of Membrane Science 576 (2019) 116–122

117

detector. The X-ray diffraction (XRD) patterns of the membranes wereobtained at room temperature using a Rigaku SmartLab X-ray dif-fractometer operating in the 2θ range of 0.6–30° and at a scanning rateof 0.02° s-1. The fourier transform infrared spectra (FTIR) were de-termined using a spectrometer (Nicolet 8700) in the scanning range of4000–600 cm-1. The measurement on water contact angles of preparedmembranes were performed on a contact angle goniometer (DropMeterA100, Maist).

2.5. Filtration and separation tests

The water permeance and rejection performance of thus-producedTpPa-AAO membranes were measured using a dead-end filtrationsystem (Amicon 8003, Millipore) under a trans-membrane pressure of0.5 bar. Water permeance (A, L m-2 h-1 bar-1) of the membrane wasdetermined by normalizing the volume (V) of permeate collected duringthe time of Δt per unit area (S) per unit pressure (P), and calculated bythe following equation:

=A V SΔt P/( ) (1)

Protein solutions with a concentration of 0.5 g L-1 and 10-nm goldnanoparticles were used to evaluate the ultrafiltration performance.The nanofiltration performance of the membrane with a dense COFlayer was characterized by filtrating 50 ppm dye solutions. To avoid theconcentration polarization, the separation measurements were con-ducted with stirring at a speed of 600 rpm. The concentrations of feed(Cf), permeate (Cp), and retention (Cr) were analyzed by a UV–vis de-tector (NanoDrop 2000c, Thermo) at the maximum absorption wave-length of each dye. The rejection (R, %) of the various dyes was cal-culated by the following equation:

= − ×R C C(1 / ) 100%p f (2)

3. Results and discussion

The first stage of preparation of a composite membrane involvesageing of the synthetic solution for about 5min at room temperature.During this stage, the initial translucent solution changed to be cloudywith the generation of seed crystals (Fig. S1). The second stage involvesthe deposition and growth of seeds on the AAO surface, eventuating inthe formation of the COF layer. Here we should note that the existenceof seeds is crucial for the production of water-permeable TpPa-AAOmembranes. Because the generated seeds will deposit onto the supportsurface by gravity to form a growth layer, and this layer can prevent thegeneration of a coverage on the bottom of the AAO substrates. (Fig. S2).Subsequently, a gradually narrowed pores and a continuous COF se-lective layer can be achieved with the condensation of monomer pairsalong the seeds.

After synthesis at 100 °C for 40min, the physical appearance of theAAO support is changed from white to deep red, implying the growth ofCOFs onto the substrate (Fig. 1a). The successful fabrication of theTpPa-AAO membrane has been confirmed by FTIR analysis. As illu-strated in Fig. 1b, the produced powders and composite membranesshow characteristic stretching bands at 1575 cm-1 (-C˭C), and 1253 cm-1

(-C-N), matching well with the reported references [33]. Moreover, thedisappearance of -N-H and aldehyde –C˭O stretching bands at3100–3300 and 1639 cm-1, respectively, demonstrates a completeconsumption of the starting materials. The XRD characterization wasthen conducted to ascertain the ordered structure of the producedpowders and TpPa layers (Fig. 1c). The powders display an intense peakat lower 2θ value of ~4.7°, which corresponds to the reflection from the100 plane [33]. The broad peak at higher 2θ value of ~27° arises due tothe 001 plane reflection, this peak indicates the π-π stacking betweensuccessive layers of the TpPa. The TpPa-AAO membranes also showmoderate peaks at the corresponding 2θ values, confirming the for-mation of stacked TpPa layer on the AAO substrate. The weak peak

intensity for FTIR and XRD analysis on the AAO substrate can be as-cribed to the small amount of TpPa deposited on the substrate.

One of the advantages of this in situ solvothermal synthesis is theflexibility of controlling the growth amount of TpPa simply by varyingthe synthesis conditions, leading to a customizable separation perfor-mance of the resulting membranes. Thus, we have prepared UF TpPa-AAO membranes under short synthesis durations and NF membranesfor dye removal under longer synthesis durations. We performed SEMcharacterization to gain insight into the morphology of the producedmembranes. As shown in Fig. 2a, there are some discrete crystalliteswith a size of ~25 nm deposited on the AAO substrate after synthe-sizing at 80 °C for 20min. The condensation between starting pre-cursors facilitates the development of TpPa on the AAO surface, and theTpPa is preferentially produced along the previously formed crystallitesto knit a continuous layer. As expected, increasingly more TpPa crys-tallites are produced and intergrown with their neighbors after syn-thesizing for 30 and 40min (Fig. 2b, c), producing gradually narrowedpores with enhanced selectivity. Interestingly, a continuous and defect-free TpPa selective layer can be obtained after only 50min of synthesis(Fig. 2d), which is also validated by the formation of dense film afterdegrading the AAO substrate (Fig. S3). Intriguingly, the time for theformation of a continuous COF layer is much faster than that of otherreported solvothermal synthesis on varied substrates (typically for 3days) [31,34], endowing this approach with a relatively high efficiencytrait for large-scale production. The quicker formation of the COF layersis mainly because that we used AAO substrates with the average poresizes of ~20 nm while other works used substrates having bigger poreswith sizes of several hundreds of nanometers. Smaller pores requiremuch less time to be narrowed to generate a covering layer. To furtherreduce the fabrication time, the synthesis temperature was increased to100 °C with results given in Fig. 2e-h. Similar to the results under lowertemperature of 80 °C, the AAO surface was gradually covered by TpPawith prolonging synthesis durations, which also leads to a narrowedpore size and finally a dense TpPa layer. In this case, the time for theformation of a continuous COF layer is reduced to 40min, and the COFgrowth is faster than the synthesis at 80 °C as the higher temperaturefacilitates the Schiff-based reaction between amines and aldehydes.Here we should note that the membrane performance can hardly con-trolled with synthesis temperature up to 120 °C due to the excessivegrowth of TpPa crystallites. We then checked the cross-sectional mor-phology of the TpPa-AAO membrane, and the result shows that theproduced membrane has a composite structure with a COF layer on thetop surface (Fig. 3a). Moreover, there are numerous COF crystallitesgenerated on the AAO pore walls (Fig. 3b top), showing a distinct dif-ference compared with the pristine AAO substrate which has a smoothpore wall (Fig. S4). Furthermore, the amount of these crystallite par-ticles is gradually decreased from top to bottom and the pore wall of thebottom still presents the original smooth morphology (Fig. 3b). Theelement analysis of the membrane was then conducted on the surfaceand cross section. A uniform distribution of C and N elements (char-acteristic elements from TpPa) can be observed in the surface EDXmapping (Fig. S5), demonstrating the uniform growth of TpPa on themembrane surface [31]. Besides, the atomic ratio of nitrogen to carbon(N/C) is in line with that of the TpPa molecular structure (Fig. S6). Toshed more light on the formation of TpPa-AAO membranes, we con-ducted a line scanning on the cross-section of the membrane. The Celement (TpPa characteristic element) presents a gradient distributionon the basis of the location from the top surface to the membranebottom (Fig. 3c). Interestingly, the C content is steeply decreased whenthe location varies from 0 to 10 µm, and then achieves a plateau,matching well with the change in the crystallite amount of the crosssection at different locations (Fig. 3b). This C element distributionclearly indicates the TpPa is mainly generated on top of the AAO sub-strate [35]. When the substrate was immersed into the ripened pre-cursor solution containing crystallites, these crystallites will deposit onthe AAO surface by gravity and small-sized crystallites will infiltrate

X. Shi et al. Journal of Membrane Science 576 (2019) 116–122

118

into the inner pores. With further solvothermal synthesis, the formedcrystallites can hardly permeate into the narrowed AAO pores and growon the surface, thus producing a TpPa layer with a significant gradient.

Considering the pore sizes of AAO supports can be preciselymodulated through developed TpPa crystallites, we expect that theseparation performance of thus-produced TpPa-AAO membranes can beeffectively governed. As illustrated in Fig. 4a, The pristine AAO supportshows a water permeance about ~2231 Lm-2 h-1 bar-1 with a negligibleBSA rejection (< 10%). After synthesizing for only 20min at 80 °C, theBSA rejection can be remarkably improved to ~51.4% with a slightly

decreased permeance to ~1779 Lm-2 h-1 bar-1. An extension of thesynthesis duration from 30 to 40min elevates BSA rejection from~88.6% to ~98.6% while reduces the permeance from ~1263 to~289 Lm-2 h-1 bar-1. Given an effective balance between permeanceand rejection performance, a membrane with 30min TpPa growth isappropriate as a ~43% sacrifice in the water permeance will in returnattain a decuple improvement in BSA rejection. Notably, the selectivitycan be significantly improved with the formation of a continuous TpPaselective layer, because the inherent micropores of about 1.83 nm ori-ginated from TpPa would be the only channels for molecules

Fig. 1. Characterization of the TpPa-AAO membrane after synthesizing under 100 °C for 40min (a) Physical appearance of the pristine AAO substrate (top) and TpPa-AAO membrane (bottom). (b) FTIR spectra of monomer pairs, TpPa powders, and TpPa-AAO membranes. (c) XRD patterns of simulated TpPa, TpPa powders, andTpPa-AAO membranes.

Fig. 2. Surface SEM images of the TpPa-AAO membranes. (a–d) Synthesized at 80 °C for 20, 30, 40, and 50min (e–h) Synthesized at 100 °C for 10, 20, 30, and 40min.

Fig. 3. Investigation on the COF formation of the TpPa-AAO membrane after synthesizing at 100 °C for 40min (a) The cross-sectional morphology of entire thickness.(b) The cross-sectional morphologies at top, 6 µm, 12 µm, and bottom. (c) C elemental distribution in the entire cross section of the membrane.

X. Shi et al. Journal of Membrane Science 576 (2019) 116–122

119

transferring in this condition. Herein, the TpPa-AAO membrane with50min growth was used to separate water from AF dye molecules. Witha continuous TpPa selective layer, the membrane can reject ~93.5% AFmolecules with a water permeance of ~14 Lm-2 h-1 bar-1. Thus, wesuccessfully build a bridge between UF for BSA separation and NF fordye removal by simply tuning the synthesis time. Moreover, the bridgecan be faster established under a higher synthesis temperature of 100 °C(Fig. 4b). For example, the BSA rejection is improved to ~92.3% with awater permeance about 1173 Lm-2 h-1 bar-1 after 20 min, and themembrane can separate ~97.2% AF molecules from water after 40min.Typically, this decreased synthesis time enhances the efficiency ofmembrane fabrication and is beneficial to a practical production.Moreover, a decreased monomer concentration will in return lead to aslowed synthesis rate (Fig. S7), which is similar to the result synthesizedunder a lower temperature.

To further validate the separation performance, the membranesprepared at 80 °C for various durations was evaluated in a number ofseparation tests. Different kinds of molecules, such as AF, OVA, BSA,and 10-nm gold nanoparticles were filtered through the TpPa-AAOmembranes. As shown in Fig. 4c, the membranes present rejection ratesof ~93.5%, ~88.2%, ~88.6%, and ~95.2% for AF, OVA, BSA, and Auparticles, respectively, which keeps a high separation efficiency about90% for different molecules. Hence, the membranes synthesized for anappropriate time are expected to separate targeted molecules withspecific sizes from water. Notably, the resultant membranes exhibitcomparable or even better UF performance than other reported mem-branes prepared by polymers [36], MOF [37], and graphene oxide (GO)[38] (Fig. 4d, Table S1), and show excellent separation performancebeyond the upper bound of UF membranes (Fig. S8) [39]. For instance,the TpPa-AAO membrane shows a BSA rejection comparable to poly-imide-modified AAO membrane by atomic layer deposition (ALD)

technique [40], but with a much higher water permeance. We believethat this superior water permeance is related to the synergetic effect ofmultiple features of developed imine-based TpPa. An outstanding hy-drophilicity (Fig. S9) of the imine-based COFs benefits to the rapiddiffusion of water molecules through membranes [31]. More im-portantly, intrinsic micropores of TpPa will also provide additionalnanochannels for water transport compared with deposited materialshaving no pores. Based on the above mentioned benefits, we can realizethe production of TpPa-AAO membranes with significantly enhancedselectivities and acceptable decline in water permeance.

As discussed above, the resulting membranes with continuous COFlayers display a great potential in dye separation from water. Hence, tofurther evaluate the dye separation performance of the preparedmembranes, five dye molecules, AO7, EB-T, AF, RB19, and CR, dis-solved in water with a concentration of 50 ppm were chosen as thefeeds. As shown in Fig. 5, the solution color change after filtrationclearly demonstrates a highly promising rejection rate for the producedTpPa-AAO membrane. For instance, the membrane exhibits a rejectionas high as ~99.4% for CR (Mw: 696.68 gmol-1, dimension:~2.63 nm×0.73 nm). Interestingly, dyes with a relatively small di-mension ranging from ~1.3 to 1.6 nm also can be effectively excludedthough the intrinsic channel size of TpPa is calculated to be ~1.83 nm.Specifically, the membrane presents rejection rates of ~97.2%,~98.7%, and ~99.1% for AF (Mw: 585.54 gmol-1, dimension:~1.26 nm×1.04 nm), RB19 (Mw: 626.54 gmol-1, dimension:~1.48 nm×1.10 nm), and EB-T (Mw: 461.38 gmol-1, dimension:~1.60 nm×1.02 nm), respectively. This experimental result can beexplained by the inter-growth of TpPa crystallites, which further nar-rows down the produced pore sizes due to staggered channels [31].Another reason for this finding might be the electrostatic repulsionbetween the negatively charged TpPa layer and dyes [29]. Besides,

Fig. 4. Separation performances of the synthesized TpPa-AAO membranes. Membranes synthesized at 80 °C (a) and 100 °C (b) for various durations. (c) Separationperformance of the TpPa-AAO membranes prepared at 80 °C for 20, 30, 40, and 50min as a function of equivalent diameter of each rejected substance. (d)Ultrafiltration performance of the membrane in comparison with other membranes, details are given in Table S1.

X. Shi et al. Journal of Membrane Science 576 (2019) 116–122

120

because AO7 molecules (Mw: 350.32 gmol-1, dimension: ~1.37 nm×0.72 nm) have a small dimension together with a short axial length of0.72 nm, the membrane displays a slightly reduced rejection of~80.2%. In each case, the concentration of dye molecules in the re-tention is notably increased (Fig. S10), implying the separation me-chanism is not caused by the adsorption of dyes. The static adsorptionresults also show that the concentration remains nearly unchanged afterthe adsorption tests, confirming that adsorption plays a negligible rolein removing molecules from the solution (Fig. S11). It is noteworthythat dyes are often discharged along with organic solvents. Hence, tofurther extend the application of the produced composite membranewith a dense TpPa selective layer, we have performed the organic sol-vent nanofiltration tests by using ethanol as solvent. The membranedisplays an ethanol permeance of ~31 Lm-2 h-1 bar-1 after synthesizedat 100 °C for 40min. CR molecules dissolved in ethanol with a con-centration of 50 ppm was then filtrated through the membrane. Simi-larly, a colorless filtrate is obtained with a highly desired rejection of~90.5% for thus-prepared membrane (Fig. S12). Here, a slightly re-duced rejection than that of water system can be caused by the di-minished electrostatic repulsion effect in ethanol solution. Bringingtogether the advantages of imine-based COF materials, the membranedelivers a comparable rejection performance together with a higherethanol permeance than other membranes [41,42].

4. Conclusion

In summary, performance-adjustable imine-linked COF membraneshave been successfully prepared via a facile solvothermal synthesis onmesoporous AAO substrates. With rising synthesis durations, the pro-duced membranes can effectively reject small molecules from 10-nmgold particles through BSA proteins to various dyes, thus tuningmembrane performance from UF to NF category. Due to sufficient in-herent micropores combined with desired hydrophilicity of synthesizedTpPa layer, the fabricated membrane within UF category showcases adesired BSA rejection about 92.3% together with a relatively high waterpermeance of up to ~1173 Lm-2 h-1 bar-1, which is ~2–9 time higherthan that of other membranes with similar BSA rejections. For NF ca-tegory, dye molecules with a size above 1.3 nm can be effectively re-moved from water by the membrane with a continuous TpPa selectivelayer. Benefitted from traits of solvent resistance and well-definedchannels, the TpPa-AAO membrane also presents an excellent separa-tion performance for dye molecules in organic solvents. We anticipatethat the fabrication of COF membranes used for precisely targeted se-paration process can be realized by choosing appropriate synthesismonomers and conditions.

Acknowledgement

Financial support from the National Basic Research Program ofChina (2015CB655301), Natural Science Foundation of China(21825803), and the Jiangsu Natural Science Foundation(BK20150063) is gratefully acknowledged. We also thank the Programof Excellent Innovation Teams of Jiangsu Higher Education Institutionsand the Project of Priority Academic Program Development of JiangsuHigher Education Institutions (PAPD).

Appendix A. Supplementary material

Supplementary data associated with this article can be found in theonline version at doi:10.1016/j.memsci.2019.01.034.

References

[1] X. Yao, L. Guo, X. Chen, J. Huang, M. Steinhart, Y. Wang, Filtration-based synthesisof micelle-derived composite membranes for high-flux ultrafiltration, ACS Appl.Mater. Interfaces 7 (2015) 6974–6981.

[2] X. Shi, Z. Xu, C. Huang, Y. Wang, Z. Cui, Selective swelling of electrospun blockcopolymers: from perforated nanofibers to high flux and responsive ultrafiltrationmembranes, Macromolecules 51 (2018) 2283–2292.

[3] L. Ding, Y. Wei, L. Li, T. Zhang, H. Wang, J. Xue, L.-X. Ding, S. Wang, J. Caro,Y. Gogotsi, MXene molecular sieving membranes for highly efficient gas separation,Nat. Commun. 9 (2018) 155.

[4] W. Li, P. Su, Z. Li, Z. Xu, F. Wang, H. Ou, J. Zhang, G. Zhang, E. Zeng, Ultrathinmetal-organic framework membrane production by gel-vapour deposition, Nat.Commun. 8 (2017) 406.

[5] V.S. Bobrov, N.G. Digurov, V.V. Skudin, Propane dehydrogenation using catalyticmembrane, J. Membr. Sci. 253 (2005) 233–242.

[6] Z. Jiang, S. Karan, A.G. Livingston, Water transport through ultrathin polyamidenanofilms used for reverse osmosis, Adv. Mater. 30 (2018) 1705973.

[7] H.B. Park, J. Kamcev, L.M. Robeson, M. Elimelech, B.D. Freeman, Maximizing theright stuff: the trade-off between membrane permeability and selectivity, Science356 (2017).

[8] S. Karan, Z. Jiang, A.G. Livingston, Sub-10 nm polyamide nanofilms with ultrafastsolvent transport for molecular separation, Science 348 (2015) 1347–1351.

[9] Q. Wang, X. Wang, Z. Wang, J. Huang, Y. Wang, PVDF membranes with simulta-neously enhanced permeability and selectivity by breaking the tradeoff effect viaatomic layer deposition of TiO2, J. Membr. Sci. 442 (2013) 57–64.

[10] Y. Lv, Y. Du, W.Z. Qiu, Z.K. Xu, Nanocomposite membranes via the codeposition ofpolydopamine/polyethylenimine with silica nanoparticles for enhanced mechanicalstrength and high water permeability, ACS Appl. Mater. Interfaces 9 (2017)2966–2972.

[11] B.P. Biswal, H.D. Chaudhari, R. Banerjee, U.K. Kharul, Chemically stable covalentorganic framework (COF)-polybenzimidazole hybrid membranes: enhanced gasseparation through pore modulation, Chemistry 22 (2016) 4695–4699.

[12] C.C. Zou, Q.Q. Li, Y.Y. Hua, B.H. Zhou, J.G. Duan, W.Q. Jin, Mechanical synthesis ofCOF nanosheet cluster and its mixed matrix membrane for efficient CO2 removal,ACS Appl. Mater. Interfaces 9 (2017) 29093–29100.

[13] 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.

[14] T.Y. Liu, H.G. Yuan, Y.Y. Liu, D. Ren, Y.C. Su, X. Wang, Metal-organic frameworknanocomposite thin films with interfacial bindings and self-standing robustness forhigh water flux and enhanced ion selectivity, ACS Nano 12 (2018) 9253–9265.

[15] N.X. Wang, X.T. Li, L. Wang, L.L. Zhang, G.J. Zhang, S.L. Ji, Nanoconfined zeoliticimidazolate framework membranes with composite layers of nearly zero thickness,ACS Appl. Mater. Interfaces 8 (2016) 21979–21983.

[16] Y. Hu, J. Wei, Y. Liang, H. Zhang, X. Zhang, W. Shen, H. Wang, Zeolitic imidazolateframework/graphene oxide hybrid nanosheets as seeds for the growth of ultrathinmolecular sieving membranes, Angew. Chem. Int. Ed. Engl. 55 (2016) 2048–2052.

[17] A. Huang, Q. Liu, N. Wang, Y. Zhu, J. Caro, Bicontinuous zeolitic imidazolate fra-mework ZIF-8@GO membrane with enhanced hydrogen selectivity, J. Am. Chem.Soc. 136 (2014) 14686–14689.

[18] H. Ang, L. Hong, Polycationic polymer-regulated assembling of 2D MOF nanosheetsfor high-performance nanofiltration, ACS Appl. Mater. Interfaces 9 (2017)28079–28088.

[19] S.Y. Ding, W. Wang, Covalent organic frameworks (COFs): from design to appli-cations, Chem. Soc. Rev. 42 (2013) 548–568.

[20] A.P. Cote, A.I. Benin, N.W. Ockwig, M. O'Keeffe, A.J. Matzger, O.M. Yaghi, Porous,crystalline, covalent organic frameworks, Science 310 (2005) 1166–1170.

[21] S. Kandambeth, V. Venkatesh, D.B. Shinde, S. Kumari, A. Halder, S. Verma,R. Banerjee, Self-templated chemically stable hollow spherical covalent organicframework, Nat. Commun. 6 (2015) 6786.

[22] W. Li, C.X. Yang, X.P. Yan, A versatile covalent organic framework-based platformfor sensing biomolecules, Chem. Commun. 53 (2017) 11469–11471.

[23] H. Singh, V.K. Tomer, N. Jena, I. Bala, N. Sharma, D. Nepak, A. De Sarkar,K. Kailasam, S.K. Pal, A porous, crystalline truxene-based covalent organic

Fig. 5. Dye separation performance of the TpPa-AAO membrane prepared at100 °C for 40min. Inserted photographs show the color change before (left) andafter (right) filtration.

X. Shi et al. Journal of Membrane Science 576 (2019) 116–122

121

framework and its application in humidity sensing, J. Mater. Chem. A 5 (2017)21820–21827.

[24] Q. Fang, S. Gu, J. Zheng, Z. Zhuang, S. Qiu, Y. Yan, 3D microporous base-func-tionalized covalent organic frameworks for size-selective catalysis, Angew. Chem.Int. Ed. Engl. 53 (2014) 2878–2882.

[25] S. Lu, Y. Hu, S. Wan, R. McCaffrey, Y. Jin, H. Gu, W. Zhang, Synthesis of ultrafineand highly dispersed metal nanoparticles confined in a thioether-containing cova-lent organic framework and their catalytic applications, J. Am. Chem. Soc. 139(2017) 17082–17088.

[26] C.J. Doonan, D.J. Tranchemontagne, T.G. Glover, J.R. Hunt, O.M. Yaghi,Exceptional ammonia uptake by a covalent organic framework, Nat. Chem. 2(2010) 235–238.

[27] N. Huang, X. Chen, R. Krishna, D.L. Jiang, Two-dimensional covalent organic fra-meworks for carbon dioxide capture through channel-wall functionalization,Angew. Chem. Int. Ed. Engl. 54 (2015) 2986–2990.

[28] L. Valentino, M. Matsumoto, W.R. Dichtel, B.J. Marinas, Development and perfor-mance characterization of a polyimine covalent organic framework thin-filmcomposite nanofiltration membrane, Environ. Sci. Technol. 51 (2017)14352–14359.

[29] R. Wang, X. Shi, A. Xiao, W. Zhou, Y. Wang, Interfacial polymerization of covalentorganic frameworks (COFs) on polymeric substrates for molecular separations, J.Membr. Sci. 566 (2018) 197–204.

[30] S. Kandambeth, B.P. Biswal, H.D. Chaudhari, K.C. Rout, H.S. Kunjattu, S. Mitra,S. Karak, A. Das, R. Mukherjee, U.K. Kharul, R. Banerjee, Selective molecularsieving in self-standing porous covalent-organic-framework membranes, Adv.Mater. 29 (2017).

[31] H.W. Fan, J.H. Gu, H. Meng, A. Knebel, J. Caro, High-flux membranes based on thecovalent organic framework COF-LZU1 for selective dye separation by nanofiltra-tion, Angew. Chem. Int. Ed. Engl. 57 (2018) 4083–4087.

[32] K. Dey, M. Pal, K.C. Rout, H.S. Kunjattu, A. Das, R. Mukherjee, U.K. Kharul,R. Banerjee, Selective molecular separation by interfacially crystallized covalentorganic framework thin films, J. Am. Chem. Soc. 139 (2017) 13083–13091.

[33] S. Kandambeth, A. Mallick, B. Lukose, M.V. Mane, T. Heine, R. Banerjee,Construction of crystalline 2D covalent organic frameworks with remarkable che-mical (acid/base) stability via a combined reversible and irreversible route, J. Am.Chem. Soc. 134 (2012) 19524–19527.

[34] H. Lu, C. Wang, J.J. Chen, R. Ge, W.G. Leng, B. Dong, J. Huang, Y.A. Gao, A novel3D covalent organic framework membrane grown on a porous alpha-Al2O3 sub-strate under solvothermal conditions, Chem. Commun. 51 (2015) 15562–15565.

[35] H. Yang, H. Wu, Z. Yao, B. Shi, Z. Xu, X. Cheng, F. Pan, G. Liu, Z. Jiang, X. Cao,Functionally graded membranes from nanoporous covalent organic frameworks forhighly selective water permeation, J. Mater. Chem. A 6 (2018) 583–591.

[36] H. Yang, Z. Wang, Q. Lan, Y. Wang, Antifouling ultrafiltration membranes by se-lective swelling of polystyrene/poly(ethylene oxide) block copolymers, J. Membr.Sci. 542 (2017) 226–232.

[37] H. Sun, B. Tang, P. Wu, Development of hybrid ultrafiltration membranes withimproved water separation properties using modified superhydrophilic metal-or-ganic framework nanoparticles, ACS Appl. Mater. Interfaces 9 (2017)21473–21484.

[38] J. Xue, S. Wang, X. Han, Y. Wang, X. Hua, J. Li, Chitosan-functionalized grapheneoxide for enhanced permeability and antifouling of ultrafiltration membranes,Chem. Eng. Technol. 41 (2018) 270–277.

[39] A. Mehta, A.L. Zydney, Permeability and selectivity analysis for ultrafiltrationmembranes, J. Membr. Sci. 249 (2005) 245–249.

[40] H. Wang, M. Wei, Z. Zhong, Y. Wang, Atomic-layer-deposition-enabled thin-filmcomposite membranes of polyimide supported on nanoporous anodized alumina, J.Membr. Sci. 535 (2017) 56–62.

[41] D.Y. Xing, S.Y. Chan, T.-S. Chung, The ionic liquid [EMIM]OAc as a solvent tofabricate stable polybenzimidazole membranes for organic solvent nanofiltration,Green Chem. 16 (2014) 1383–1392.

[42] D. Fritsch, P. Merten, K. Heinrich, M. Lazar, M. Priske, High performance organicsolvent nanofiltration membranes: development and thorough testing of thin filmcomposite membranes made of polymers of intrinsic microporosity (PIMs), J.Membr. Sci. 401–402 (2012) 222–231.

X. Shi et al. Journal of Membrane Science 576 (2019) 116–122

122