lable at ScienceDirect

Microporous and Mesoporous Materials 228 (2016) 45e53

Contents lists avai

Microporous and Mesoporous Materials

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

Ionothermal synthesis of LTA-type aluminophosphate molecular sievemembranes with gas separation performance

Xiaolei Li a, b, Keda Li a, Shuo Tao a, b, Huaijun Ma a, Renshun Xu a, Bingchun Wang a,Ping Wang a, b, Zhijian Tian a, c, *

a Dalian National Laboratory for Clean Energy, Dalian Institute of Chemical Physics, Chinese Academy of Sciences, Dalian 116023, Chinab University of Chinese Academy of Sciences, Beijing 100049, Chinac State Key Laboratory of Catalysis, Dalian Institute of Chemical Physics, Chinese Academy of Sciences, Dalian 116023, China

a r t i c l e i n f o

Article history:Received 18 January 2016Received in revised form10 March 2016Accepted 20 March 2016Available online 22 March 2016

Keywords:AluminophosphateIonothermal synthesisLTA membraneGas separation

* Corresponding author. Dalian National LaboratoInstitute of Chemical Physics, Chinese Academy of Sc

E-mail address: tianz@dicp.ac.cn (Z. Tian).

http://dx.doi.org/10.1016/j.micromeso.2016.03.0261387-1811/© 2016 Elsevier Inc. All rights reserved.

a b s t r a c t

LTA-type aluminophosphate molecular sieve membranes with gas separation performance were ion-othermally synthesized by using d-alumina substrates as both the supports and the aluminum sources.The effects of the synthesis parameters, such as the concentrations of H3PO4, TAMOH, and HF, and thedetailed formation process of the membrane were thoroughly investigated using X-ray diffraction (XRD),scanning electron microscopy (SEM), and energy dispersive X-ray spectroscopy (EDX). Furthermore, thesubsequent membrane syntheses by using recycled mother liquids were also studied. The resultsdemonstrated that continuous and compact LTA molecular sieve membranes can be prepared over arelatively wide range of the synthesis composition. The membranes prepared with fresh and recycledmother liquids exhibit the same crystallinity, morphology and the gas separation performance. Typically,for single-component gases at 293 K, the ideal separation factors of H2/CO2, H2/O2, H2/N2, and H2/CH4, are10.9, 8.1, 6.8, and 4.8, respectively, which suggests the good gas separation performance of themembranes.

© 2016 Elsevier Inc. All rights reserved.

1. Introduction

Molecular sieve membranes are highly attractive for applica-tions in the separation of gases and liquids due to their uniqueproperties such as uniform micropores, superior hydrothermal andchemical stabilities [1,2]. Generally, there are twomajor approachesto fabricate molecular sieve membranes: an in situ method and asecondary (seeded) growth method [2]. The in situ method is themost widely used, inwhich themembrane is prepared by the directcrystallization of molecular sieves on a support. As for the in situmethod, the membrane quality depends largely on the physico-chemical properties of the support surface [3,4], as well as thesynthetic procedures and conditions [5,6]. The secondary growthmethod, which involves coating seeds on the substrate followed bysecondary growth, is more practical in preparing a high qualitymembrane [7,8]. However, the seeding step complicates the

ry for Clean Energy, Dalianiences, Dalian 116023, China

synthesis process, especially on a tubular support. Both of thesemethods are typically performed under hydrothermal conditions athigh autogenous pressure, which introduces safety concerns andhinders scale-up preparation [9e11]. Further, the utilization ratio ofreactant materials is relative low since the crystallization takesplace mainly in the solution phase, and only a fraction of nutrientsare consumed for the growth of themembrane, which increases thesynthesis cost. In addition, the waste mother liquids containingtoxic organic amines also bring environmental pollution.

To date, much attention is paid to aluminosilicate-basedmolecular sieve membranes [12e14], whereas investigations to-ward the synthesis and characterization of aluminophosphate-based membranes are somewhat limited [15e19]. Ionothermalsynthesis has been proved to be an effective approach to syn-thesize microporous aluminphosphate molecular sieves [20e27].Compared with traditional hydrothermal and solvothermal syn-theses, ionothermal synthesis can be performed at ambientpressure due to the negligible vapor pressure of ionic liquids.Moreover, ionothermal synthesis provides an alternative methodto fabricate aluminophosphate molecular sieve membranes. Forinstance, AlPO4-11 and SAPO-11 films with anti-corrosion

Table 1Details of initial solution compositions, crystallization conditions and productstructures.

Sample H3PO4:TMAOH:HF:ILs (molar ratio) T (�C) t (h) Product

1 1.0:1.0:0.7:100 180 24 LTA2 2.0:1.0:0.7:100 180 24 LTA3 4.0:1.0:0.7:100 180 24 LTA4 6.0:1.0:0.7:100 180 24 LTA5 8.0:1.0:0.7:100 180 24 LTAþAEL6 4.0:0:0.7:100 180 24 AEL7 4.0:0.5:0.7:100 180 24 LTAþAEL8 4.0:0.7:0.7:100 180 24 LTA9 4.0:2.0:0.7:100 180 24 LTA10 4.0:4.0:0.7:100 180 24 LTA11 4.0:1.0:0:100 180 24 Amorphous12 4.0:1.0:0.2:100 180 24 LTA13 4.0:1.0:1.0:100 180 24 LTA14 4.0:1.0:0.7:100 180 2 LTA15 4.0:1.0:0.7:100 180 6 LTA16 4.0:1.0:0.7:100 180 12 LTA

Table 2The amounts of unreacted H3PO4 and added reagents.

Membranes Unreacted H3PO4 (g)a Added reagents

H3PO4 (g) TMAOH (g) HF (g)

M1 0 0.39 0.091 0.014M2 0.253 0.14 0.046 0.007M3 0.298 0.10 0.046 0.007

a Measured by ICP-OES. M1, M2 andM3 represent the membranes prepared fromthe first, second and third syntheses cycles, respectively.

X. Li et al. / Microporous and Mesoporous Materials 228 (2016) 45e5346

performance have been fabricated by in situ and electrochemicalionothermal synthesis, respectively [28,29]. Recently, we havedeveloped a substrate surface conversion method to ion-othermally prepare aluminophosphate molecular sieve mem-branes [30,31]. In the synthesis, a moderately active d-alumina (ametastable phase between g and a-alumina) substrate acts asboth the support and the sole aluminum source. The detailedformation mechanism of a CHA-type membrane shows that thecrystallization takes place merely on the surface of the substratevia a solid-state transformation mechanism [32]. Due to theexcellent thermal stability of ionic liquids, we expect that themother liquids could be reused, which will reduce the synthesiscost and environmental pollution.

The LTA-type aluminophosphate molecular sieve (AlPO4-42) isa representative member of the AlPO4-n family, which consists ofthree-dimensional 8-ring channels with a pore diameter of0.4 nm [33]. The small pore size, together with its neutralframework, makes the aluminophosphate LTA membrane apromising material for the separation of light gases by the mo-lecular sieving effect. Huang et al. reported the hydrothermalpreparation of an AlPO4-LTA membrane with high hydrogenselectivity by the secondary seeded method, in which theexpensive crown ether compound Kryptofix 222 was used as thestructure-directing agent [18,34].

In the present work, we report a facile and green ionothermalroute for the synthesis of LTA-type AlPO4 membranes with gasseparation performance. The effects of the synthesis parametersand the formation process of the membrane were investigated indetail using XRD, SEM, and EDX techniques. The subsequentmembrane syntheses by using recycled mother liquids were alsoinvestigated. The gas permeation properties of the membranesprepared with freshly made and recycled mother liquids weredetermined using the WickeeKallenbach technique.

2. Experimental

2.1. Materials

Phosphoric acid (H3PO4, 85 wt% in water, AR), hydrofluoric acid(HF, 40 wt% in water, AR) and tetramethylammonium hydroxide(TMAOH, 25 wt% inwater, AR) were purchased from Tianjin KermelChemical Reagent Co. and used as received without further purifi-cation. 1-butyl-3-methylimidazolium bromide ([BMIm]Br), wasprepared by neutralization of redistilled N-methylimidazole (KaileChemical Factory, Zhejiang, China, 99.9%) and 1-bromobutane(Sinopharm Chemical Reagent Co., Ltd.) according to the procedurereported elsewhere [21]. d-alumina disks with 2.0 mm thicknessand 20 mm diameter were prepared according to the method re-ported elsewhere [30].

2.2. Molecular sieve membrane syntheses using fresh mother liquid

Table 1 summarizes the detailed synthesis conditions andproduct structures. The typical synthesis procedure can bedescribed as follows: the synthesis solution was prepared by mix-ing 21.9 g (0.1 mol) of [BMIm]Br and the required quantities ofH3PO4, TMAOH and HF with vigorous stirring. Notably, no addi-tional aluminum source was added into the synthesis solution.Afterward, the solution was transferred into a Teflon-lined auto-clave in which a d-alumina substrate prepared in-house was placedvertically with a Teflon holder. The crystallizationwas performed inan air oven at 180 �C for 0.5e24 h. The as-synthesized samples waswashed with distilled water by ultrasonication and dried overnightat 110 �C. To remove the template, the samples were calcined at500 �C for 8 h with heating and cooling rates of 0.2 �C/min.

2.3. Molecular sieve membrane syntheses using recycled motherliquids

In the recyclable syntheses, about 21 g of mother liquids werecollected by simply removing the membrane. Prior to the secondand third syntheses cycles, the chemical composition of the motherliquids was adjusted by adding proper amount of H3PO4, HF andTMAOH. The added amount of H3PO4was determined by ICP-OES tokeep the H3PO4/[Bmim]Br ratio constant. The added amount ofTMAOH and HF was half of the initial amount. The detailed syn-thesis information is presented in Table 2. The other procedure wasidentical to that prepared with fresh reactants. The amount of theunreacted H3PO4 and compensated reagents is listed in Table 2. Theother synthetic procedures are identical to that prepared usingfresh reactant materials.

2.4. Characterization of the membranes

XRD analysis was carried out on a PANalytical X'Pert PROdiffractometer fitted with a CuKa radiation source (l ¼ 1.5418 Å)operating at 40 mA and 40 kV. The relative crystallinity wascalculated based on the ratio of the intensities of the Bragg reflec-tion (200) for LTA for each sample compared to sample 3. SEM wasperformed on a Phenom scanning electron microscope (FEI Elec-tron Optics). EDX was carried out on a Hitachi S4800 field emissionscanning electron microscope. Inductively coupled plasma opticalemission spectroscopy (ICP-OES) analysis was carried out on aPerkineElmer Optima 7300 DV spectrometer.

Single and mixture gas permeation tests were performed ac-cording to procedure outlined in reference [35] at 293 K followedby amodifiedWickeeKallenbach technique. The permeate sidewasswept with N2 (except for the N2 permeation measurement whereCH4 was employed as the sweep gas). The fluxes of the feed and

Fig. 1. XRD patterns of the LTA-type AlPO4 membranes prepared with molar compo-sition of xH3PO4:1.6TMAOH:1.0HF:100[BMIm]Br. (a) x ¼ 1.0, (b) x ¼ 2.0, (c) x ¼ 4.0, (d)x ¼ 6.0, (e) x ¼ 8.0.

X. Li et al. / Microporous and Mesoporous Materials 228 (2016) 45e53 47

sweep gases were determined by mass flow controllers. The pres-sures at the feed and permeate side were both kept at 0.1 MPa. Thegas concentrations of the permeate side were measured using anon-line gas chromatography (Agilent 7890A).

Fig. 2. Top view SEM images of LTA membranes prepared with molar composition of xH

3. Results and discussion

3.1. Effects of the synthesis parameters

To prepare a uniform and compact LTA molecular sieve mem-brane, the effects of the synthesis parameters, such as the con-centrations of H3PO4, TMAOH, and HF, on the formation of LTAmembranes are investigated in detail.

3.1.1. Effect of the concentration of H3PO4

As the phosphorus source, H3PO4 plays a vital role in the for-mation of the molecular sieve membranes. To investigate the effectof the concentration of H3PO4 on the synthesis of LTAmembranes, aseries of samples were prepared with a molar composition ofxH3PO4:1.0TMAOH:0.7HF:100[BMIm]Br at 180 �C for 24 h Figs. 1and 2 show the XRD patterns and corresponding SEM images ofthe samples prepared with different concentrations of H3PO4 (x),respectively. As shown in Fig. 1, pure phase of LTA can be obtainedwith x ranging from 1 to 6. As shown in Fig. 2a only a portion of thesupport was covered by LTA molecular sieves when xwas 1.0. Withthe increase of H3PO4 concentration, the coverage of the LTAmembrane increased. When x was increased to 2.0, the surface ofthe support was covered with block-shaped crystals and a contin-uous LTA membrane was obtained (Fig. 2b). When x was furtherincreased to 4.0 and 6.0, the LTA membranes were continuous andpossessed similar surface morphology (Fig. 2c and d). As the con-centration of H3PO4 continued to increase, a mixture of LTA and AELwas obtained (Fig. 1e). This can be explained that excessive amountof H3PO4 (x ¼ 8.0) destroyed the cooperative structure-directingeffect of ionic liquid cations and the TMAþ [36]. In conclusion, a

3PO4:1.6TMAOH:1.0HF:100[BMIm]Br. (a) x ¼ 1.0, (b) x ¼ 2.0, (c) x ¼ 4.0, (d) x ¼ 6.0.

Fig. 3. (aed) Top view SEM images of LTA membranes prepared with molar composition of 4.0H3PO4:yTMAOH:0.7HF:100[BMIm]Br. (a) y ¼ 0.5, (b) y ¼ 0.7, (c) y ¼ 2.0, (d) y ¼ 4.0.(eef) Top view SEM images of LTA membranes prepared with molar composition of 4.0H3PO4:1.0TMAOH:zHF:100[BMIm]Br. (e) z ¼ 0.2, (f) z ¼ 1.0.

X. Li et al. / Microporous and Mesoporous Materials 228 (2016) 45e5348

continuous LTA membrane could be obtained when the H3PO4concentration was within the range 2.0e6.0.

3.1.2. Effect of the concentration of TMAOHIn ionothermal synthesis of aluminophosphate molecular sieves,

the amount of introduced TMAOH not only affects the dynamics ofthe crystallization process but also alters the phase selectivity of thecrystallization reaction [37,38]. Our previous study has shown thatTMAþ could play a cooperative structure-directing role togetherwith [BMIm]þ to direct the formation of the LTA-type molecularsieves [39]. In the present work, to obtain a continuous membrane,the effect of TMAOH was also investigated. The TMAOH concen-tration in the ionic liquid was variedwhile the molar composition ofother reactants was kept constant at 4.0H3PO4:yTMAOH:0.7HF:100[BMIm]Br. Figs. S1 and 3 show the XRD patterns and corresponding

SEM images of the samples prepared with different concentrationsof TMAOH (y), respectively. As shown in Fig. S1a, in the absence ofTMAOH, only AEL was obtained. When the concentration of TMAOH(y) was 0.5, a mixture of LTA and AEL was obtained, suggesting thatthe amount of TMAOHwas not sufficient to direct the formation of apure LTA phase (Figs. S1b and 3a). As shown in Fig. S1, pure phase ofLTA can be obtained when y was above 0.7. A continuous LTAmembrane was obtained when y was above 0.7 (Fig. 3b). When ywas increased to 1.0 and 2.0, the morphology of the LTAmembranesdid not change obviously (Figs. 2b and 3c). As shown in Fig. 3d, withfurther increasing y to 4.0, a large number of small crystals appearedand no continuous LTA membrane was obtained on the supportsurface. This is probably due to excessive amounts of the amineneutralized theH3PO4 in the solution, which slows down the growthrate of LTA molecular sieves. Therefore, the optimal TMAOH

Fig. 4. Crystallinity curve of LTA membranes prepared with a molar ratio of4.0H3PO4:1.0TMAOH:0.7HF:100[BMIm]Br at 180 �C.

X. Li et al. / Microporous and Mesoporous Materials 228 (2016) 45e53 49

concentration for the formation of a continuous LTA membrane wasin the range 0.7e2.0.

3.1.3. Effect of the concentration of HFAs reported previously, HF playsmineralizing and co-templating

roles in the ionothermal synthesis of aluminumphosphate molec-ular sieves [39e41]. In the LTA-type AlPO4 molecular sieve, F� ionsreside in the D4R units to stabilize the structure [39,40]. To studythe effect of HF on the membrane formation, its concentration wasvaried while the molar composition of other reactants was keptconstant at 4.0H3PO4:1.0TMAOH:zHF:100[BMIm]Br. In the absenceof HF, no LTA molecular sieve crystal was formed on the support,indicating that the LTA membrane cannot be prepared in theabsence of HF (not shown). At the same time, excessive HF is notbeneficial for LTA membrane preparation. When the HF concen-tration (z) was increased to 1.0, a large quantity of small crystalsappeared (Fig. 3f). Thus, the optimal HF concentration for LTAgrowth on the substrate was in the range 0.2e0.7 (Figs. 2c and 3e).

Based on the above results, it can be concluded that continuousLTA molecular sieve membranes can be prepared over a relativelywide range of the synthesis composition. For the solution molarcomposition xH3PO4:yTMAOH:zHF:100[BMIm]Br, a continuous LTAmembrane can be obtained when x, y and z are within the ranges2.0e6.0, 0.7e2.0 and 0.2e0.7, respectively. All these membraneshave similar surface morphology and exhibit high compactness.Previous reports have shown that the amount of water plays acrucial role in ionothermal synthesis [41,42]. It is worth to mentionthat, in the present work, the amount of water brought by the re-actants was high enough to ensure complete crystallization. Inaddition, the water content has no remarkable effect on the phaseselectivity since LTA membranes can be synthesized over a widesynthetic window. Therefore, the water content has no obviouseffect on formation of the membranes.

3.2. The formation process of the LTA membrane

To demonstrate the formation process of the LTA membrane,the samples prepared with a molar composition of 4.0H3PO4:1.0T-MAOH:0.7HF:100[BMIm]Br were characterized at different reactiontime intervals. According to the XRD characterization results shownin Fig. 4, the LTA phase first appeared after 2 h of crystallization.

During the period of 2e12 h, the relative crystallinity increasedsharply, suggesting the continued growth of the LTA molecularsieves. When the synthesis time was prolonged to 12 h, the relativecrystallinity was close to 100%. The relative crystallinity increasedslightly when the synthesis time was extended to 24 h, which sug-gested that the crystallization was completed. The correspondingSEM images shown in Fig. 5 were consistent with the XRD results. Asshown in Fig. 5a, some small LTA crystals appeared on the substrateafter 2 h of crystallization. The cross-sectional view shows that theLTA layer was composed of small LTA grains (Fig. 5b). When theduration was increased from 2 h to 12 h, the LTA grains grew largerand fused with each other. The coverage of the LTA molecular sieveson the substrate increased, and the LTA layer grew compactly. LTAcrystals coveredmost of the substrate after reaction for 12 h (Fig. 5e).However, several voids remained between crystals. As the synthesistime prolonged to 24 h, the LTA crystals were crowd and graduallyfused together, followed by bridging the gaps between them. Finally,a continuous membrane was obtained (Fig. 5g). It is worthmentioning that although some boundary defects can be seen fromthe top view image of themembrane, the cross-sectional view imageshows that these defects are shallow and the inner part of themembrane is compact and continuous.

Fig. 6a illustrates a cross-sectional SEM image of an LTA mem-brane. As can be seen, the synthesized membrane is compact andcontinuous. The EDX line scanning reveals that there is no sharptransition between the AlPO4 LTA membrane and the support. TheP/Al ratio of the surface portion of the substrate, with thickness of10 mm, is close to 1, which is consistent with the stoichiometry ofLTA molecular sieves. The P/Al ratio decreases gradually along thearrow in Fig. 6a, and eventually reaches to zerowhen the d-aluminasubstrate is contacted. The intermediate layer is approximate30 mm thick. The nature of the intermediate layer was amorphousaluminophosphate gel since its P/Al ratio was below 1. From theSEM image, we can see that there is no obvious boundary betweenthe LTA membrane and the intermediate layer.

In our previous work, the detailed formation process of thesubstrate surface conversion method has been demonstrated[30,32]. In this work, the surface area of the d-alumina substratereacts with the solution to form an aluminophosphate precursorgel layer at initial stage. LTA crystals first nucleate at the layer/so-lution interface, where all the nutrients required for molecularsieve nucleation and crystallization are abundant. As time in-creases, a large number of small LTA molecular sieve grains appearon the substrate surface. Subsequently, the LTA grains grow largerand fuse together. After the formation of a continuous LTA mem-brane, the reaction between the substrate and the reaction solutionis hindered. Eventually, an AlPO4 LTAmembrane of 10 mm thicknessis formed on the substrate surface.

3.3. Recycling of mother liquids

The initial LTA membrane (M1) was prepared with molarcomposition of 4.0H3PO4:1.0TMAOH:0.7HF:100[BMIm]Br at 180 �Cfor 24 h. In the present synthesis, since the crystallization takesplace merely on the surface of the substrate and no obvious crystalis formed in the reaction solution. Therefore, mother liquids can beeasily recycled by removing the substrate without additionaltreatment. The recycled mother liquids contain mainly unreactedionic liquids, H3PO4, HF, organic amine, a small amount of waterfrom the reagents, and negligible amount of aluminum dissolvedfrom the d-alumina substrate.

Prior to the second and third syntheses cycles, the chemicalcomposition of mother liquids was adjusted by adding properamount of reactants. The added amount of H3PO4 was determinedby ICP-OES to keep the H3PO4/[Bmim]Br ratio constant. The

Fig. 5. SEM top views of LTA-type membranes synthesized with a molar ratio of 4.0H3PO4:1.0TMAOH:0.7HF:100[BMIm]Br at 180 �C for (a) 2 h, (c) 6 h, (e) 12 h and (g) 24 h; SEMcross-sectional views of LTA-type membranes synthesized at 180 �C for (b) 2 h, (d) 6 h, (f) 12 h and (h) 24 h.

X. Li et al. / Microporous and Mesoporous Materials 228 (2016) 45e5350

Fig. 6. (a) Cross-sectional SEM image of the LTA membrane prepared with a molarratio of 4.0H3PO4:1.0TMAOH:0.7HF:100[BMIm]Br at 180 �C for 24 h. The arrow rep-resents the path of the EDX line scan. (b) P/Al ratio along the path of the EDX linescanning.

Fig. 8. Single gas permeances of different gases as function of their kinetic diametersat 293 K. The inset shows the ideal separation factor for H2 relative to other gases.

X. Li et al. / Microporous and Mesoporous Materials 228 (2016) 45e53 51

remaining amount of HF and TMAOH inmother liquids was difficultto determine, however since the membranes can be synthesizedover a wide synthetic window and a little change of the chemicalcomposition has negligible effect on the membrane synthesis.Therefore, the amount of TMAOH and HF can be safely adjusted byadding half of the initial amount. LTA membranes prepared by the

Fig. 7. XRD patterns of the LTA-type AlPO4 membranes that were prepared withrecycled mother liquids at 180 �C for 24 h (a) M2, (b) M3.

second and third uses of mother liquids are denoted as M2 and M3,respectively. As shown in Fig. 7, M2 and M3 exhibit the typicaldiffraction peaks of the LTA structure with good crystallinity. Fig. S2shows the SEM images of the membranes M2 and M3. As can beseen, these membranes are compact and have the samemorphology with the membrane M1, which indicates the suc-cessful synthesis of high quality of LTA membranes by recyclingmother liquids. From the cost and environmental perspectives, thisprocess is extremely beneficial since the utilization ratio of thestarting materials is improved.

3.4. Gas permeation

The aluminophosphate LTA membrane is expected to separategas molecules by differences in their size because of its small poresize of approximately 0.4 nm. The gas permeation properties ofmembranes M1, M2 and M3 were tested by single gas permeationand mixture gas separation. All the flux of the single gases H2, CO2,N2, O2, and CH4, as well as the binary mixture gases of H2 with CO2,N2, O2, and CH4, were determined using the WickeeKallenbachtechnique.

Fig. 8 shows the single gas permeances for membranes M1, M2andM3. As can be seen, themembranesM2 andM3 have nearly thesame gas permeances and ideal separation factors as M1. The per-meances decreasewith the increase of the kinetic diameters. H2 hasthe highest permeance because its kinetic diameter is smaller thanthose of the other gases. For the membrane M1, the H2 permeanceis 8.1 � 10�7 mol m�2 s�1 Pa�1. The inset of Fig. 8 shows the idealseparation factors of H2 over CO2, O2, N2, and CH4, which aredefined as the ratio of the single gas permeances [43]. The idealseparation factors of H2/CO2, H2/O2, H2/N2, and H2/CH4 for themembrane M1 are 10.9, 8.1, 6.8, and 4.8, respectively. The idealseparation factors are all higher than those of the correspondingKnudsen selectivities of 4.7, 4.0, 3.7, and 2.8, respectively, whichindicate that the LTA-type AlPO4 membrane displays H2 selectivity.For aluminosilicate and silicoaluminophosphate zeolite mem-branes applied in gas separation, such as SAPO-34 [44,45], zeolitesilicalite-1 [46], zeolite Y [47], and zeolite T [48], high CO2 per-meance can be expected because their polar structures lead tostrong adsorption affinity toward CO2. Owing to the neutralframework of LTA structure which is comprised of alternating AlO4and PO4 tetrahedra, the present LTA membrane shows no CO2 af-finity. The CO2 permeance is lower than that of O2, N2 and CH4

Table 3Mixture gas separation performances of LTA-type aluminophosphate molecular sieve membranes for equimolar binary gas mixtures.

Membrane M1 M2 M3

H2/CO2 H2 permeances (mol m�2 s�1 Pa�1) 4.9 � 10�7 5.8 � 10�7 5.6 � 10�7

CO2 permeances (mol m�2 s�1 Pa�1) 7.2 � 10�8 8.7 � 10�8 7.9 � 10�8

Mixture separation factor 6.8 6.7 7.1H2/O2 H2 permeances (mol m�2 s�1 Pa�1) 5.2 � 10�7 5.3 � 10�7 5.1 � 10�7

O2 permeances (mol m�2 s�1 Pa�1) 9.1 � 10�8 1.0 � 10�7 8.4 � 10�8

Mixture separation factor 5.7 5.3 6.1H2/N2 H2 permeances (mol m�2 s�1 Pa�1) 4.9 � 10�7 5.4 � 10�7 5.4 � 10�7

N2 permeances (mol m�2 s�1 Pa�1) 7.9 � 10�8 1.1 � 10�7 9.9 � 10�8

Mixture separation factor 6.2 4.9 5.5H2/CH4 H2 permeances (mol m�2 s�1 Pa�1) 5.7 � 10�7 6.1 � 10�7 5.9 � 10�7

CH4 permeances (mol m�2 s�1 Pa�1) 1.3 � 10�7 1.5 � 10�7 1.3 � 10�7

Mixture separation factor 4.4 4.1 4.5

X. Li et al. / Microporous and Mesoporous Materials 228 (2016) 45e5352

because its molecular weight is the largest, which is consistent withprevious reports on AlPO4 molecular sieve membranes [15,17,34].

Table 3 summarizes the mixture gas separation performance ofthe membranes M1, M2 and M3. As demonstrated in Table 3, themembranes M2 andM3 have similar mixture gas separation factorsas M1, which suggests the good performance of the membranesprepared by recycling mother liquids. As for M1, the mixture sep-aration factors of H2/CO2, H2/O2, H2/N2, and H2/CH4, are 6.8, 5.7, 6.2,and 4.4, respectively, with H2 permeances over4.9 � 10�7 mol m�2 s�1 Pa�1. All of the mixture separation factorsare lower than the corresponding ideal separation factors. The ex-istence of a second component results in a decrease in the per-meance of H2, for the second gas molecules block the path of H2molecules.

All these gas permeation properties indicate that the gaspermeation behavior is predominantly controlled by the molecularsieving effect of the LTA membrane and the gas molecules mainlypermeate through the pores of LTA aluminophosphate molecularsieves. The membranes prepared by recycling mother liquids havenearly the same gas permeance properties as the membrane pre-pared using fresh mother liquid, suggesting that the approach toprepare LTA membranes by using recycled mother liquids is valid.

4. Conclusions

In summary, we presented a facile ionothermal approach for thesynthesis of LTA-type aluminophosphate molecular sieve mem-branes with gas separation performance. Compact LTA molecularsieve membranes could be prepared over wide ranges of the con-centrations of H3PO4, TMAOH, and HF. Furthermore, the motherliquids could be recycled and reused for subsequent cycles ofmembrane preparations, which improves the utilization ratio of thestartingmaterials. Themembranes preparedwith freshlymade andrecycled mother liquids exhibit the same crystallinity, morphologyand gas permeation performance. Typically, for single-componentgases at 293 K, the ideal separation factors of H2/CO2, H2/O2, H2/N2, and H2/CH4 are found to be 10.9, 8.1, 6.8, and 4.8, respectively,indicating that the LTA membrane displays molecular sieving per-formance in gas permeation tests. The present approach based onthe substrate surface conversion method and recycling motherliquids can effectively decrease the waste of raw materials andenvironmental pollution. We believe this cost-effective and envi-ronmentally benignmethod could inspire researchers to synthesizeother types of molecular sieve membranes.

Acknowledgments

We thank the financial supports from Strategic Priority ResearchProgram of the Chinese Academy of Sciences (Grant No.

XDA07020300) and the National Natural Science Foundation ofChina (Grant No. 21373214).

Appendix A. Supplementary data

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.micromeso.2016.03.026.

References

[1] Z. Lai, G. Bonilla, I. Diaz, J.G. Nery, K. Sujaoti, M.A. Amat, E. Kokkoli, O. Terasaki,R.W. Thompson, M. Tsapatsis, D.G. Vlachos, Science 300 (2003) 456e460.

[2] J. Caro, M. Noack, P. Kolsch, R. Schafer, Microporous Mesoporous Mater. 38(2000) 3e24.

[3] J.L.H. Chau, C. Tellez, K.L. Yeung, K.C. Ho, J. Membr. Sci. 164 (2000) 257e275.[4] A.W.C. van den Berg, L. Gora, J.C. Jansen, M. Makkee, T. Maschmeyer, J. Membr.

Sci. 224 (2003) 29e37.[5] X. Li, Y. Yan, Z. Wang, Ind. Eng. Chem. Res. 49 (2010) 5933e5938.[6] A. Huang, W. Yang, Sep. Purif. Technol. 61 (2008) 175e181.[7] J. Caro, M. Noack, Microporous Mesoporous Mater. 115 (2008) 215e233.[8] Y. Peng, X. Lu, Z. Wang, Y. Yan, Angew. Chem. Int. Ed. 54 (2015) 5709e5712.[9] C.S. Cundy, P.A. Cox, Chem. Rev. 103 (2003) 663e701.

[10] L. Tosheva, V.P. Valtchev, Chem. Mater. 17 (2005) 2494e2513.[11] X. Meng, F.-S. Xiao, Green routes for synthesis of zeolites, Chem. Rev. 114

(2014) 1521e1543.[12] A. Huang, F. Liang, F. Steinbach, J. Caro, J. Membr. Sci. 350 (2010) 5e9.[13] Z. Wang, Q. Ge, J. Shao, Y. Yan, J. Am. Chem. Soc. 131 (2009) 6910e6911.[14] Y. Peng, H. Lu, Z. Wang, Y. Yan, J. Mater. Chem. A 2 (2014) 16093e16100.[15] G.Q. Guan, T. Tanaka, K. Kusakabe, K.I. Sotowa, S. Morooka, J. Membr. Sci. 214

(2003) 191e198.[16] L. Tosheva, E.P. Ng, S. Mintova, M. Holzl, T.H. Metzger, A.M. Doyle, Chem.

Mater. 20 (2008) 5721e5726.[17] A. Huang, F. Liang, F. Steinbach, T.M. Gesing, J. Caro, J. Am. Chem. Soc. 132

(2010) 2140e2141.[18] M.L. Carreon, S.G. Li, M.A. Carreon, Chem. Commun. 48 (2012) 2310e2312.[19] C. Ji, Y. Tian, Y. Li, Y.S. Lin, Microporous Mesoporous Mater. 186 (2014) 80e83.[20] E.R. Cooper, C.D. Andrews, P.S. Wheatley, P.B. Webb, P. Wormald, R.E. Morris,

Nature 430 (2004) 1012e1016.[21] E.R. Parnham, R.E. Morris, Chem. Mater. 18 (2006) 4882e4887.[22] R.E. Morris, Angew. Chem. Int. Ed. 47 (2008) 442e444.[23] Y.P. Xu, Z.J. Tian, S.J. Wang, Y. Hu, L. Wang, B.C. Wang, Y.C. Ma, L. Hou, J.Y. Yu,

L.W. Lin, Angew. Chem. Int. Ed. 45 (2006) 3965e3970.[24] Y. Wei, Z.J. Tian, H. Gies, R.S. Xu, H.J. Ma, R.Y. Pei, W.P. Zhang, Y.P. Xu, L. Wang,

K.D. Li, B.C. Wang, G.D. Wen, L.W. Lin, Angew. Chem. Int. Ed. 49 (2010)5367e5370.

[25] E.-P. Ng, L. Itani, S.S. Sekhon, S. Mintova, Chem. Eur. J. 16 (2010)12890e12897.

[26] E.-P. Ng, J.-P. Ghoy, H. Awala, A. Vicente, R. Adnan, T.C. Ling, S. Mintova,CrystEngComm 18 (2016) 257e265.

[27] S. Tao, X. Li, R. Xu, D. Li, G. Zhang, H. Ma, Y. Xu, Z. Tian, Chem. Commun. 52(2016) 2253e2256.

[28] R. Cai, M. Sun, Z. Chen, R. Munoz, C. O'Neill, D.E. Beving, Y. Yan, Angew. Chem.Int. Ed. 47 (2008) 525e528.

[29] T. Yu, W. Chu, R. Cai, Y. Liu, W. Yang, Angew. Chem. Int. Ed. 54 (2015)13032e13035.

[30] K. Li, Z. Tian, X. Li, R. Xu, Y. Xu, L. Wang, H. Ma, B. Wang, L. Lin, Angew. Chem.Int. Ed. 51 (2012) 4397e4400.

[31] K. Li, X. Li, Y. Wang, D. Li, H. Liu, R. Xu, Z. Tian, Acta Chim. Sin. 71 (2013)573e578.

[32] X. Li, K. Li, H. Ma, R. Xu, S. Tao, Z. Tian, Microporous Mesoporous Mater. 217(2015) 54e62.

X. Li et al. / Microporous and Mesoporous Materials 228 (2016) 45e53 53

[33] Ch. Baerlocher, L.B. McCusker, Database of Zeolite Structures. http://www.iza-structure.org/databases/.

[34] A. Huang, J. Caro, Chem. Commun. 47 (2011) 4201e4203.[35] H. Bux, F. Liang, Y. Li, J. Cravillon, M. Wiebcke, J. Caro, J. Am. Chem. Soc. 131

(2009) 16000e16001.[36] S. Tao, R. Xu, X. Li, D. Li, H. Ma, D. Wang, Y. Xu, Z. Tian, J. Colloid Interface Sci.

451 (2015) 117e124.[37] L. Wang, Y.P. Xu, Y. Wei, J.C. Duan, A.B. Chen, B.C. Wang, H.J. Ma, Z.J. Tian,

L.W. Lin, J. Am. Chem. Soc. 128 (2006) 7432e7433.[38] R. Pei, Y. Wei, K. Li, G. Wen, R. Xu, Y. Xu, L. Wang, H. Ma, B. Wang, Z. Tian,

W. Zhang, L. Lin, Dalton Trans. 39 (2010) 1441e1443.[39] R.Y. Pei, Z.J. Tian, Y. Wei, K.D. Li, Y.P. Xu, L. Wang, H.J. Ma, Mater. Lett. 64

(2010) 2118e2121.[40] E.J. Fayad, N. Bats, C.E.A. Kirschhock, B. Rebours, A.A. Quoineaud, J.A. Martens,

Angew. Chem. Int. Ed. 49 (2010) 4585e4588.

[41] H.J. Ma, Z.J. Tian, R.S. Xu, B.C. Wang, Y. Wei, L. Wang, Y.P. Xu, W.P. Zhang,L.W. Lin, J. Am. Chem. Soc. 130 (2008) 8120e8121.

[42] D.Y. Khoo, W.-M. Kok, R.R. Mukti, S. Mintova, E.-P. Ng, Solid State Sci. 25(2013) 63e69.

[43] W.J. Koros, Y.H. Ma, T. Shimidzu, Pure Appl. Chem. 68 (1996) 1479e1489.[44] S.G. Li, J.L. Falconer, R.D. Noble, Adv. Mater. 18 (2006) 2601e2603.[45] Y.Y. Tian, L.L. Fan, Z.Y. Wang, S.L. Qiu, G.S. Zhu, J. Mater. Chem. 19 (2009)

7698e7703.[46] H. Guo, G. Zhu, H. Li, X. Zou, X. Yin, W. Yang, S. Qiu, R. Xu, Angew. Chem. Int.

Ed. 45 (2006) 7053e7056.[47] G. Clet, L. Gora, N. Nishiyama, J.C. Jansen, H. van Bekkum, T. Maschmeyer,

Chem. Commun. (2001) 41e42.[48] Y. Cui, H. Kita, K. Okamoto, J. Mater. Chem. 14 (2004) 924e932.