Contents lists available at ScienceDirect

Journal of Membrane Science

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

Organic template-free synthesis of high-quality CHA type zeolite membranesfor carbon dioxide separation

Eunhee Janga, Sungwon Honga, Eunjoo Kima, Nakwon Choib, Sung June Choc, Jungkyu Choia,⁎

a Department of Chemical & Biological Engineering, College of Engineering, Korea University, 145 Anam-Ro, Seongbuk-gu, Seoul 02841, Republic of Koreab Center for BioMicrosystems, Brain Science Institute, Korea Institute of Science and Technology (KIST), Hwarang-ro 14-gil, Seongbuk-gu, Seoul 02792, Republic of Koreac Department of Chemical Engineering, Chonnam National University, 77 Yongbong-ro, Buk-gu Gwangju 61186, Republic of Korea

A R T I C L E I N F O

Keywords:Chabazite zeolitesSeeded or secondary growthTemplate-free synthesisCracksCarbon dioxide separation

A B S T R A C T

Microporous chabazite (CHA) zeolite is very promising for CO2 capture because of its appropriate pores withmolecular dimensions for the preferential adsorption of CO2 molecules. Herein, CHA type zeolite particles andmembranes were prepared by using a seeded growth method in the absence of an organic structure directingagent (OSDA) or template. After substantial effort to find appropriate and reliable conditions for obtainingcontinuous CHA type zeolite membranes, it was recognized that the formation of these membranes is a highlysensitive function of the Si/Al ratio in the synthetic precursor. Using the appropriate Si/Al ratio of ~50, OSDA-free CHA type zeolite membranes were manufactured with high reproducibility. The resulting OSDA-free CHAtype zeolite membranes showed maximum CO2/N2 and CO2/CH4 separation factors of ~12.5± 3.8 and~28.8±6.9, respectively, with a moderate CO2 permeance of ~1 × 10−7 mol m−2 s−1 Pa−1. Notably, undermore realistic wet conditions (i.e., in the presence of H2O vapor), the separation performance at temperaturesabove 75 °C was comparable to that obtained under dry conditions, although permeation was hindered below50 °C, apparently due to the strong adsorption of H2O vapor.

1. Introduction

Zeolite membranes have been widely studied owing to their highperformance for separating molecular mixtures based on their intrinsicmolecular sieving ability and/or their capability for preferential ad-sorption [1,2]. Furthermore, the high thermal and chemical stabilitiesalso render zeolite membranes good candidates for separating in-dustrially important mixtures [2–4]. Desirably, zeolite membranesshould be able to separate mixtures of components with close boilingpoints (e.g., xylene isomers [2,5], butene isomers [3,6–8], and aceticacid/water mixtures [9,10]), for which conventional thermodynamics-based separation processes are not viable options.

The secondary growth method, nowadays regarded as a reliablezeolite membrane synthesis methodology [11], requires the batch-wisehydrothermal growth of a seed layer. Very often, organic structure di-recting agents (OSDAs; e.g., 1-adamantylamine for DDR [12], N,N,N-trimethyl-1-adamantammonium cation (TMAda+) for CHA [13,14],and tetrapropylammonium cation for MFI zeolites [15]) are addedduring secondary growth to achieve the reproducible production ofhigh-quality zeolite membranes [16–19]. In the secondary growthmethodology, the successful membrane formation is contingent on

minimizing defect formation after the inevitable time- and energy-consuming thermal activation process (i.e., calcination) [20,21]. Anappropriate calcination step is a requisite even for well-intergrown as-synthesized zeolite membranes to ensure their intrinsic separationability. Indeed, the calcination step often results in the uncontrollableformation of unwanted defects that provide non-selective pathways topermeates and thus nullifies the intrinsic separation ability of thezeolite membrane [22,23]. Although alternative activation approacheshave been introduced [24–26], a membrane fabrication method thatdoes not require the calcination step is highly attractive for realizing theintrinsic separation ability of the zeolite membrane [27]. Thus, theorganic template-free synthesis of high-performance zeolite membranesis highly desirable.

Hydrothermal growth without OSDAs has enabled the synthesis ofseveral types of zeolites such as MFI [28–30], CHA [31–35], and BEA[36–38]. As the main advantages, OSDA-free syntheses have cost-ef-fectiveness due to the non-use of the OSDAs and energy-efficiency dueto the lack of requirement for the calcination step [39,40]. An extensiveliterature survey showed that OSDA-free synthesis of MFI type zeolitemembranes has been intensively studied [5,20,28,29,41,42]. Some ofthese OSDA-free MFI zeolite membranes showed good performance for

https://doi.org/10.1016/j.memsci.2017.11.068Received 19 September 2017; Received in revised form 23 November 2017; Accepted 26 November 2017

⁎ Corresponding author.E-mail address: jungkyu_choi@korea.ac.kr (J. Choi).

Journal of Membrane Science 549 (2018) 46–59

Available online 02 December 20170376-7388/ © 2017 Elsevier B.V. All rights reserved.

T

the separation of mixtures (hydrogen-hydrocarbon (C1-C4) [42] andn-/i-C4H10 [28]). It was noted that the separation performances ofOSDA-free MFI zeolite membranes were comparable to and sometimeshigher than that of the MFI membranes prepared with the OSDA [42].However, it appears that the successful formation of MFI membranes inthe absence of OSDAs is largely limited to a very narrow region ofcompositional and hydrothermal conditions [43]. The high sensitivityand/or low reproducibility in the OSDA-free synthesis of MFI, CHA, andBEA zeolites have been addressed [30,35,38]. The data indicate theimportance of a rigorous approach toward OSDA-free membrane fab-rication; to this end, the main factors, such as the Si/Al ratio and thecation species in the synthetic precursor, the amount of seed employed,and hydrothermal conditions, should be comprehensively elucidated.Based on such a rigorous approach, a reliable OSDA-free synthesisprotocol can be secured and adopted for membrane manufacturing.Although parameter studies to determine the optimal conditions towardhigh-performance OSDA-free zeolite membranes have been conducted,the final successful synthetic route is presented in most studies[5,20,28,33,41,42].

Compared with MFI and BEA zeolites, the successful OSDA-freesynthesis of 8-membered ring (MR) small pore zeolites such as CHA andDDR has not been widely reported, as it is plausibly difficult to form thesmall pore channels and large cavities without appropriate OSDAs [44].Among the 8-MR zeolites, chabazite (CHA) zeolites with a pore size of0.37 × 0.42 nm2 are promising for separating CO2 (0.33 nm) fromlarger molecules such as N2 (0.364 nm) or CH4 (0.38 nm); CO2/N2 andCO2/CH4 separations are respectively relevant to post-combustioncarbon capture and bio-gas/natural gas upgrading. To date, most CHAmembranes have been synthesized in the presence of the OSDATMAda+ and have been shown to be effective for CO2 separation[19,45,46]. However, the use of expensive OSDAs such as TMAdaOHand/or the co-SDA tetraethylammonium hydroxide (TEAOH) for syn-thesizing the high-quality CHA membranes hampers their practical use[47]. Thus, alternative synthetic approaches with inexpensive OSDAs(e.g., N,N,N-dimethylethylcyclohexylammonium bromide [47]) andeven without OSDAs have been introduced [31–35]. In particular, twotypes of the OSDA-free CHA and SAPO-34 membranes showed goodperformance for the separation of H2O/ethanol mixtures [31] and CO2/CH4 [33], respectively. Despite their promise, to the best of ourknowledge, the effects of the above-mentioned influential factors on thesynthesis of OSDA-free CHA membranes have not yet been addressedand discussed in a comprehensive way. Because the OSDA-free synth-esis of small pore zeolites is highly challenging [44], a rigorous analysisof the synthesis of OSDA-free CHA zeolites under various syntheticconditions is desirable. The delineation of any correlations between theseeded growth of OSDA-free CHA particle and membrane formationshould be remarkably instructive for obtaining CHA membranes as wellas membranes of other zeolites.

In this study, we first elucidated the effect of two critical factors (theSi/Al ratio in the synthetic precursor and the hydrothermal reactiontime for seeded growth) on the crystallization of OSDA-free CHA typezeolite particles in the seeded growth method. The obtained under-standing of the various synthetic routes for OSDA-free CHA type zeoliteparticles was directly extended to the fabrication of OSDA-free CHAtype zeolite membranes. In particular, we focused on achieving highreproducibility of the OSDA-free CHA type zeolite membranes. Theresulting OSDA-free CHA type zeolite membranes that do not requirethe aforementioned energy-intensive, time-consuming calcination pro-cess were subjected to carbon capture tests employing CO2/N2 andCO2/CH4 mixtures. Furthermore, we investigated the effect of H2Ovapor in the feed on the CO2 perm-selectivity of the OSDA-free CHAmembrane, as the H2O vapor, the 3rd largest component in both cases,plays a critical role in determining the final separation performance ofzeolite membranes. Finally, we conducted a long-term test on theOSDA-free CHA type zeolite membranes under wet conditions for up to3 d in an effort to confirm their feasibility for membrane-based

separation processes.

2. Experimental

2.1. Synthesis of conventional CHA type zeolite particles

First, SSZ-13 seed particles were synthesized according to a con-ventional literature method by employing TMAdaOH as an OSDA [19].Specifically, certain amounts of TMAdaOH (SACHEM Inc.), NaOH(Sigma-Aldrich), Al(OH)3 (Sigma-Aldrich), and fumed silica (Cab-O-SilM5, Cabot) were sequentially added to deionized (DI) water. The finalmolar composition was 20 NaOH: 5 Al(OH)3: 100 SiO2: 4400 H2O: 20TMAdaOH. This precursor was thoroughly mixed in a shaking machineovernight and then poured into a Teflon liner. The Teflon liner wasplaced in a stainless steel autoclave. The hydrothermal reaction wascarried out at 160 °C for 4 d under rotation in a forced convection oven.After completing the hydrothermal reaction, the resulting solid parti-cles were recovered by a combination of centrifuging, decanting, andwashing with fresh DI water. Calcination was performed at 550 °C at aramp rate of 1 °C/min under air flow at 200 cc/min. These SSZ-13particles were furthered used as seeds in the synthesis of OSDA-freeCHA type zeolite particles and membranes. For convenience, conven-tional SSZ-13 particles, obtained using TMAdaOH as an OSDA, wereused as a reference and are hereinafter denoted as C-SSZ-13 particles.

2.2. Synthesis of OSDA-free CHA type zeolite particles

Along with the C-SSZ-13 particles, we synthesized OSDA-free CHAtype zeolite particles via the seeded growth method. Here, the C-SSZ-13particles played the role of nuclei, while the alkali metal cations (Na+

and K+) were used as inorganic SDAs to grow the CHA type zeoliteparticles from the seed particles. Specifically, the C-SSZ-13 particleswere added to a synthetic precursor with a molar composition of 100silica: x NaAlO2 (Sigma-Aldrich, Al (50–56 wt%): Na (40–45 wt%)): 70NaOH (Sigma-Aldrich): 18 KOH (Sigma-Aldrich): 10000 H2O at variousvalues of x (x = 0, 1, 2, and 5, corresponding to nominal Si/Al ratios of∞, 100, 50, and 20, respectively). For preparation of the syntheticprecursor, certain amounts of NaOH (pellet form), KOH (pellet form),NaAlO2, and silica (Cab-O-Sil M5, Cabot) were added to DI water. Toform a homogeneous precursor, the mixture was further blended on ashaking machine for 2 d. After the mixture became homogeneous andalmost translucent, ~0.1 g of the C-SSZ-13 particles was added to 30 gof the synthetic precursor, followed by additional mixing with theshaking machine for 1 d. The final mixture was poured into a Teflonliner and the Teflon liner was moved to a stainless steel autoclave forreaction. The hydrothermal reaction was carried out at 175 °C for dif-ferent times (1, 2, and 3 d) under rotation in a forced convection oven.After completing the hydrothermal reaction by quenching with tapwater, the solid particles, synthesized in the absence of OSDAs, wererecovered by repeated centrifugation, decanting, and washing with DIwater. For convenience, the resulting particles are referred to as P_x_yd,where P represents the OSDA-free particles and x and y indicate thenominal Si/Al ratio and hydrothermal reaction time (in d), respectively.

2.3. Synthesis of OSDA-free CHA type zeolite membranes

Porous α-alumina discs with a thickness of ~2 mm and diameter of~22 mm were prepared by following a method reported elsewhere [48]and were used as supports for the OSDA-free CHA type zeolite mem-branes. The C-SSZ-13 (for synthesis, see Section 2.1) were deposited onα-alumina discs via dip-coating. Prior to dip coating, a seed suspensionwas prepared by adding ~0.05 g of the C-SSZ-13 particles to ~40 mL ofethanol followed by sonication for ~20 min. One side of the α-aluminadisc, which was previously polished with a sand paper, was broughtinto contact with the seed suspension for 30 s and the disc was with-drawn from the seed suspension and dried for 30 s under ambient

E. Jang et al. Journal of Membrane Science 549 (2018) 46–59

47

conditions. This dip-coating procedure was repeated four times in anattempt to cover the disc surface. The C-SSZ-13 particles deposited onthe α-alumina disc, now forming a SSZ-13 seed layer, were calcined at450 °C for 4 h at a ramp rate of 1 °C/min under air flow at 100 cc/min.For secondary growth, a synthetic precursor was prepared by using thesame procedure used for synthesis of the OSDA-free particles, as de-scribed in Section 2.2. Accordingly, synthetic precursor comprised 100silica: 70 NaOH: 18 KOH: x NaAlO2 (x = 0, 1, 2, and 5): 10000 H2O bymole. The α-alumina disc with the seeded side facing down was placedin a tilted position in a Teflon liner and the prepared synthetic pre-cursor was then added. The Teflon liner was mounted in an autoclave.The hydrothermal reaction for secondary growth was carried out at175 °C for a certain period under static conditions; the oven tempera-ture was increased from room temperature to the target temperature(175 °C) at a rate of ~5 °C/min. The final reaction duration includes thetime for heat ramping. After a fixed hydrothermal reaction time, thereaction was quenched by immersing the autoclave under tap water.The recovered membrane samples were washed with DI water andfurther soaked in DI water overnight to remove undesired impurities.Subsequently, the membrane samples were slowly dried at room tem-perature over 3 d and further dried at 100 °C in an oven before per-forming the gas permeation experiment. For convenience, the resultingmembrane samples are referred to as M_x_yd, similar to the nomen-clature adopted for the particle samples in Section 2.2.

2.4. Characterization

Scanning electron microscope (SEM) images were obtained by usinga Hitachi S-4300 instrument; the surfaces of all the particle and mem-brane samples were Pt-coated at 15 mA for 100 s. X-ray diffraction(XRD) patterns were obtained by using a Rigaku Model D/Max-2500V/PC diffractometer (Japan) with Cu Kα radiation (λ = 0.154 nm). Thesimulated XRD patterns of CHA and MOR zeolites were obtained byusing the Mercury software (available from the CambridgeCrystallographic Data Centre; CCDC) with a crystallographic informa-tion file (CIF) that was downloaded from the International ZeoliteAssociation (IZA). N2 adsorption isotherms of some particle sampleswere obtained at 77 K by using an ASAP 2020 (Micromeritics Inc.) in-strument. Fluorescent confocal optical microscopy (FCOM) analysis wasperformed according to the method described in the literature [17],except for the use of a solid state laser with the wavelength of 488 nmas the source and oil between a membrane sample and a glass in thesample holder.

For structural analysis of the CHA zeolites, X-ray diffraction datawere collected in reflection mode using a Rigaku Model D/MAX UltimaIII (Japan) instrument with Cu Kα radiation (λ = 0.154 nm). The ac-celerating voltage and current were 40 kV and 40 mA, respectively.Data for the sample were obtained at room temperature in flat-platemode with a step size of 0.02° for a scan time of 10 s per step over the2θ range of 2–100°. The diffraction patterns were indexed using theDICVOL06 program implemented in the FullProf program suite [49,50].The initial structure of the framework of the CHA zeolite, acquired fromthe Database of Zeolite Structures (http://www.iza-structure.org/databases/), was utilized for Le Bail refinement to determine latticeparameters [51]. Le Bail refinement was performed by using the Riet-veld method in the JANA2006 package [52]. The low angle XRD dif-fraction data below 5° was excluded for the Le Bail refinement owing tothe large background component.

Separation of CO2/N2 and CO2/CH4 mixtures by using the OSDA-free CHA type zeolite membranes was conducted using a home-madepermeation system in the Wicke-Kallenbach mode; the total pressure ofboth the feed and permeate sides was held at ~1 atm. The partialpressures of CO2 and N2 (or CH4) in the CO2/N2 and CO2/CH4 mixturesused for the permeation tests under dry feed conditions were 50.5 kPaand 50.5 kPa, respectively (referred to as DRY CO2/N2 or DRY CO2/CH4, respectively), while the partial pressures of CO2, N2 (or CH4), and

H2O used for the CO2/N2 and CO2/CH4 separation tests under wet feedconditions were 49 kPa, 49 kPa, and 3 kPa, respectively (referred to asWET CO2/N2 or WET CO2/CH4, respectively). For the wet feed condi-tion, an equimolar CO2 and N2 (or CH4) mixture gas stream was allowedto go through a water-containing gas bubbler at ~25 °C in order toinclude water vapor in the dry binary mixture. The total flow rate of thefeed mixture and the He sweep was maintained at ~100 mL min−1. Asan internal standard for reliable gas chromatographic analysis,~5 mL min−1 of CH4 for the CO2/N2 mixtures and of H2 for the CO2/CH4 mixtures were added to the permeate stream carried to a gaschromatograph (GC) column by the He sweep gas. A GC (YL 6100 GCsystem, YOUNG LIN, South Korea) equipped with a packed column (6 ft× 1/8” Porapak T) and thermal conductivity detector (TCD) was usedfor on-line detection of the CO2/N2 permeates, whereas a GC (YLInstrument, 6500 GC System) equipped with a capillary column (30 m× 0.320 mm GS-GasPro) and a pulsed discharge ionization detector(PDD) was used for on-line detection of the CO2/CH4 permeates.

3. Results and discussion

3.1. OSDA-free CHA type zeolite particles

3.1.1. Morphological investigation of OSDA-free particlesFig. 1 shows the SEM images of the particles obtained by using the

seeded growth method in the absence of TMAdaOH as an OSDA; the Si/Al ratios and reaction times were varied. In all cases, the morphologiesand sizes of the OSDA-free particles were different from those of the C-SSZ-13 particles (Fig. S1a), which were added to serve as seeds for theseeded growth. This difference suggests that the C-SSZ-13 seed particlesunderwent decomposition during the seeded growth. For the 1 d seededgrowth, while P_20_1d was mainly composed of aggregated plate-likeparticles (Fig. 1a1), P_x_1d (x = 50, 100, and ∞) comprised small, ir-regular-shaped grains (Fig. 1b1-d1). For a longer duration of 2 d,P_20_2d and P_50_2d (Fig. 1a2-b2) had small, irregular-shaped grains,similar to P_x_1d (x = 50, 100, and ∞). On the contrary, P_100_2dcomprised sharp, plate-like particles with minor needle-like particles(Fig. 1c2), similar to the morphology of MOR zeolites reported in [53],whereas P_∞_2d was still comprised of the small, irregular-shapedgrains (similar to P_∞_1d), though some undefined, large particles wereobserved, as indicated by the yellow arrow in Fig. 1d2. A longerduration of 3 d resulted in a pronounced change in the morphology ofthe resulting particles (Fig. 1b3-d3). Specifically, P_x_3d (x = 100 and∞) was likely to grow into larger particles with a more defined mor-phology, while the particles of P_20_3d still had the small, irregularshape, and P_50_3d contained some larger particles (indicated byyellow arrows in Fig. 1b3). The estimated yields of all the syntheses aresummarized in Table 1. The yields for the syntheses with nominal Si/Alratios of 100 and ∞ suggest little or no seeded growth after 1 d,whereas a longer duration led to larger particles with sharp edges (seeFig. 1c1-c3 and 1d1-d3). This change in the particle morphology mightsuggest a gradual change toward another zeolite phase. In contrast, inthe syntheses with lower nominal Si/Al ratios (i.e., 20 and 50), theirregular-shaped morphology was preserved for up to 3 d and the cor-responding yields increased monotonically (see Fig. 1a1-a3 or 1b1-b3).This suggests preservation of the original zeolite phase and enhancedsynthesis with time, though some impurities were co-generated, as in-dicated by the yellow arrows in Fig. 1b3.

3.1.2. Phases evaluation of OSDA-free particlesThe XRD patterns of all the OSDA-free particles shown in Fig. 1 are

presented in Fig. 2. The 1 d seeded growth gave rise to pure CHAzeolites (Fig. 2a) irrespective of the Si/Al ratio (i.e., P_x_1d; x = 20, 50,100, and ∞). The XRD pattern of P_100_1d showed a low signal-to-noise ratio, as compared with the patterns of P_20_1d and P_50_1d,indicating unfavorable growth of the zeolite in the former. Further-more, the XRD pattern of P_∞_1d indicated a lower degree of

E. Jang et al. Journal of Membrane Science 549 (2018) 46–59

48

Fig. 1. SEM images of P_20 (1st row), P_50 (2nd row), P_100 (3rd row), and P_∞ (4th row) for various synthesis times of 1 (1st column), 2 (2nd column), and 3 d (3rd column). Yellow arrows areused to indicate particles of other phases. Blue dashed lines are used to designate the different morphologies of the obtained particles shown in (c2), (c3), and (d3). All scale bars represent5 µm.

Table 1Yields and zeolite structure types of P_20_xd, P_50_xd, P_100_xd, and P_∞_xd (x = 1, 2, and 3).a

Hydrothermal reaction time (d)

1 2 3

Yield (%)b Phase Yield (%)b Phase Yield (%)b Phase

P_20a 15 CHA 19.1 CHA 33 CHAP_50a 2.1 CHA 7.4 CHA+MORe 37 CHA+MORe

P_100a N/Ac CHAd 4.9 MOR 30 MORP_∞a N/Ac CHAd N/Ac CHAd+MORe N/Ac MOR

a In the sample label of P_x_yd, P represents the OSDA-free particles and x and y indicate the nominal Si/Al ratio and hydrothermal reaction time (in d), respectively.b Yield: (increased weight after drying - seed amount)/silica amount in the precursor.c The amount of particles recovered was less than the given seed amount (here, 0.1 g) in the precursor.d A CHA phase seemingly resulted from the dissolution of the C-SSZ-13 seed particles during seeded growth.e A minor portion of MOR zeolite was present in the mixture of MOR and CHA zeolite particles.

E. Jang et al. Journal of Membrane Science 549 (2018) 46–59

49

crystallization of the CHA zeolite. For the cases of P_100_1d andP_∞_1d, the XRD patterns, as well as the very low yields, indicate thatthe C-SSZ-13 particles, which are supposed to serve as the seeds, werepartially dissolved and/or collapsed instead of proceeding to crystalgrowth. Thus, the corresponding SEM image in Fig. 1d1 reveals amorphology comprising dissolved and/or collapsed C-SSZ-13 seedparticles. For the 2 d seeded growth, we found that P_20_2d had thepure CHA zeolite, whereas P_100_2d and P_x_2d (x = 50 and ∞) con-tained MOR zeolite as the major phase and a very small quantity of theMOR phase, respectively. After the longer reaction time of 3 d, the MORzeolite phases were pronounced in P_x_3d (x = 100 and ∞), whereasthe phases of P_x_3d (x = 20 and 50) were still comparable to those ofP_x_2d (x = 20 and 50). From the SEM images in Fig. 1, one canconclude that some particles with a different morphology (indicated byyellow arrows in Fig. 1b3 and 1d2) were associated with the MORzeolite particles and the particles with a more defined morphologyshown in Fig. 1c2, 1c3, and 1d3 were MOR zeolite particles. From thevarious synthesis results, it appears that given the duration of seededgrowth, a lower but finite Al content in the synthetic precursor favoredtransformation of the CHA structure into the MOR structure (i.e., P_100series), indicating the important role of Al atoms in the synthesis ofCHA zeolite during OSDA-free synthesis. In addition, a longer reactiontime resulted in phase transformation from CHA to MOR zeolites in thecases of P_50, P_100, and P_∞, though among them P_50 exhibited thelowest degree for phase transformation, also supporting the importanceof Al content. This phase transformation may be correlated with theaforementioned pronounced morphological change observed in theSEM images (Fig. 1c1-c3 and 1d1-d3). This phase transformation is ingood agreement with the previous report that a prolonged reaction timeresulted in the synthesis of the undesired MOR type zeolite [32]. Fromthe SEM and XRD characterizations, the OSDA-free synthesis withnominal Si/Al ratios of less than and equal to ~50 was appropriate forobtaining CHA zeolite particles as the major phase using a reaction timeof up to 3 d.

3.1.3. Structural properties of OSDA-free CHA particlesAmong the synthesized particles, we chose three representative

OSDA-free CHA particles (P_20_1d, P_20_2d, and P_50_1d) and furthermeasured their N2 adsorption isotherms at 77 K, along with that of theC-SSZ-13 seed particles as a reference (Fig. 3). The BET surface area ofthe SSZ-13 particles (740±3.8 m2 g−1) was comparable to the re-ported values (611–775 m2 g−1) [54–57]. However, the BET surface

areas of P_20_1d, P_20_2d, and P_50_1d were found to be lower at557± 2.0, 491±1.7, and 397±2.2 m2 g−1, respectively. It appearsthat the OSDA-free synthesis led to a reduction of the effective pore sizeof the resulting particles, which could in turn be attributed to the ca-tions present in the CHA zeolite framework [58] because of the low Si/Al ratio. Indeed, it was reported that the OSDA-free particles of smallpore zeolites such as CHA, RHO, and KFI tend to have a low Si/Al ratio(generally, Si/Al ≤ ~10) [44,59,60], and accordingly, contain a largeamount of cations. Moreover, the N2 adsorption amounts of the OSDA-free CHA particles are lower than those of the CHA type zeolite andzeotype (SAPO-34) particles synthesized with organic templates[32,33]. Similarly, the three OSDA-free CHA type zeolite particlessynthesized in this study also had low Si/Al ratios of ~4.1–4.2 and ~5.5(Table 2) compared with their nominal Si/Al ratios of 20 and 50, re-spectively; thus, it is reasonable to consider that cations compensatingthe charge balance of Al3+ in the framework were present inside theframework.

The changes in the cell parameters of the three OSDA-free CHAparticles were also elucidated in an effort to comprehend their lowerBET surface areas. The XRD peaks of the OSDA-free particles weregenerally shifted to lower 2θ values (Fig. 2a-b and S2), indicating anexpansion of the unit cell parameters. A previous study [32] also re-ported a left shift of the XRD peaks of template-free CHA particles,though such phenomenon was not addressed or discussed. In this study,we further attempted to estimate the cell parameters of P_20_1d, P_20_2d, and P_50_1d by using the Le bail refinement. This refinement re-vealed that all three particles had slightly longer lattice parameters interms of the a (or b) axis and c axis than those of the C-SSZ-13 particles(Table 2). This increase was presumably due to framework expansion,which in turn originated from electrostatic repulsion between the alkalications in the pore structure; the Le bail refinement data for the C-SSZ-13 particles and the template-free CHA particles (P_20_1d, P_20_2d, andP_50_1d) are compared with their respective XRD patterns in Figs. S3-S6. The lattice expansion depends on the amount and/or size of cationspresent inside the zeolite pore structure [61,62]. Given that the OSDA-free particles have lower BET surface areas, the alkali cations (here,Na+ and K+), which were presumably present in excess in the CHAzeolite (Table 2), blocked the micropores and thus inhibited the diffu-sion of N2 into the pores [58].

In general, we found that the actual Si/Al ratio of the OSDA-freeparticles determined by the energy dispersive X-ray (EDX) analysis wasnot comparable to the nominal Si/Al ratio of the corresponding

Fig. 2. XRD patterns of the particles shown in Fig. 1; P_20, P_50, P_100, and P_∞ obtained at various synthesis times of (a) 1 d (b) 2 d, and (c) 3 d. XRD pattern of C-SSZ-13 particles usedas seeds is included in (a). The simulated XRD pattern of all-silica CHA zeolite is included in (a)-(c) and the XRD pattern simulated for MOR zeolite is included in (b)-(c) for comparison.The arrows indicate the peak of MOR zeolite as a minor phase in P_50_2d, P_50_3d, and P_∞_2d. For easy comprehension, the samples that contained the MOR zeolite as a major phase areindicated by MOR in parentheses next to the sample name. For easy comparison, gray dashed lines that indicate the XRD peaks of the (101) and (3 1 1) planes in the simulated XRDpattern of CHA zeolites are included.

E. Jang et al. Journal of Membrane Science 549 (2018) 46–59

50

synthetic precursors. Instead, the OSDA-free particles were formed withhigher Al content (the Si/Al ratios were approximately five times lowerthan those used for synthesis of the C-SSZ-13 particles). Accordingly,the concentration of cations in the three OSDA-free particles was esti-mated to be ten times higher than those in the C-SSZ-13 particles(Table 2), and thus the effective pore size became smaller, resulting inthe lower BET surface areas (Table 2). The insertion of the excess ca-tions into the porous structure of the CHA zeolite was also reflected bythe increased lattice parameters (Table 2). This implies that thesynthesis of OSDA-free particles that maintain the intrinsic properties ofthe original SSZ-13 zeolite is quite challenging.

3.2. OSDA-free CHA type zeolite membranes

3.2.1. Effect of the Si/Al ratio on formation of the membranesThe selective deposition of SSZ-13 particles with a size of ~700 nm

(Fig. S1a) on α-alumina supports resulted in the formation of a uniformSSZ-13 seed layer (Fig. S1b), which is a pre-requisite for intergrowth toachieve continuous membranes. Among the four nominal Si/Al ratios(20, 50, 100, and ∞) used in the synthesis of the OSDA-free particles(Figs. 1 and 2), it was recognized that secondary growth for the for-mation of a continuous membrane occurred only with the Si/Al ratio of50, strongly supporting the critical role of the Si/Al ratio during seeded(or secondary) growth (Fig. 4). Throughout the manuscript, the termsseeded and secondary growth are used interchangeably. In the case ofM_20_1d, aggregated OSDA-free particles were formed throughout the

surface (Fig. 4a1) but in a non-continuous way (Fig. 4a2). This couldpossibly be attributed to an overproduction of nuclei, seemingly gen-erated from dissolution of the C-SSZ-13 seed particles under the low Si/Al ratio environment. In the case of M_50_1d, although some crackswere observed in the top-view SEM image (indicated by red arrows inFig. 4b1), at the given SEM resolution, these cracks did not appear to bepropagated into the interface between the film and support, as evi-denced by the cross-sectional-view SEM image (Fig. 4b2). Similarcracks were also observed in other organic template-free CHA mem-brane in the literature [31], indicating a difficulty in avoiding crackseven in the organic template-free synthetic approach. For M_100_1d,long, oval-shaped grains were formed in a non-continuous manner(Fig. 4c1-c2). Accordingly, the bare α-alumina support was observed, asindicated by yellow arrows. Likewise, isolated thick, disc-like particleswere observed in M_∞_1d where the bare α-alumina support was sig-nificantly exposed, as indicated by yellow arrows (Fig. 4d1). Under thelatter two conditions (i.e., the Si/Al ratios of 100 and ∞), where the Alcontent in the synthetic precursors was apparently insufficient, con-tinuous OSDA-free membranes could not be fabricated after secondarygrowth. In fact, this trend is consistent with the finding that the use of alower Al ratio during synthesis of the OSDA-free particles hindered theformation of CHA zeolite particles (Figs. 1 and 2). The XRD patterns ofthe OSDA-free membranes, except for M_∞_1d, confirmed the CHAzeolite structure (Fig. 4a3-c3). In contrast, M_∞_1d comprised a dif-ferent phase, namely, the MOR zeolite structure (Fig. 4d3), and thus,the thick, disc-like particles can be regarded as MOR zeolite grains. This

Fig. 3. N2 adsorption isotherms of (a) C-SSZ-13 seed particles,(b) P_20_1d, (c) P_20_2d, and (d) P_50_1d at 77 K. The filledand vacant symbols represent adsorption and desorptionpoints, respectively.

Table 2Structural parameters of C-SSZ-13 particles, P_20_1d, P_20_2d, and P_50_1d estimated via the Le Bail Refinement and EDX data.

BET surface area (m2/g) Structural parametersb EDX analysis (Atomic %)

a /Ǻ c /Ǻ V /Ǻ3 Na/Al K/Al Si/Al (Na+K)/(Si+Al)

C-SSZ-13 particles 740± 3.8 13.5920(10) 14.7532(15) 2360.4(3) 0.45 – 23 0.02P_20_1da 557±2.0 13.776(2) 14.894(4) 2447.9(9) 0.35 0.81 4.1 0.23P_20_2da 491±1.7 13.7580(15) 14.859(2) 2435.8(5) 0.27 0.81 4.2 0.21P_50_1da 397±2.2 13.7670(9) 14.8597(12) 2439.1(3) 0.49 0.90 5.5 0.21

a In the sample label of P_x_yd, P represents the OSDA-free particles and x and y indicate the nominal Si/Al ratio and hydrothermal reaction time (in d), respectively.b The values of RP (profile factor) in the Le Bail Refinement for C-SSZ-13 seed particles, P_20_1d, P_20_2d, and P_50_1d were 4.31, 4.84, 5.45, and 4.83, respectively, while the GOF

(Goodness of Fit) values in the same order were 7.18, 6.44, 7.33, and 5.91, respectively.

E. Jang et al. Journal of Membrane Science 549 (2018) 46–59

51

result also indicates that the lower amount of Al facilitated transfor-mation and/or growth of undesired MOR zeolite structures from the C-SSZ-13 seed particles. Therefore, the amount of Al cations, i.e., an ap-propriate Si/Al ratio, is also a key factor for achieving continuousOSDA-free CHA zeolite membranes via the secondary growth metho-dology. From the syntheses using various Si/Al ratios (in Fig. 4), it canbe concluded that optimal, but unfortunately narrow-range, conditionsare required for the synthesis of well-intergrown OSDA-free CHA typezeolite films; in particular, the conditions used for synthesis forM_50_1d were found to be optimal for the formation of a continuousfilm.

3.2.2. CO2/N2 separation performance of OSDA-free CHA membranesThe CO2/N2 separation performance of the membrane samples

shown in Fig. 4 was investigated as a function of the temperature up to~200 °C under dry conditions (Fig. 5). M_20_1d (Fig. 5a) exhibited amaximum (max) CO2/N2 separation factor (SF) of ~1; considering the

CO2/N2 SF of ~0.8 for the bare α-alumina support based on Knudsendiffusion, this performance can be regarded as extremely poor. Incontrast, the max CO2/N2 SF of M_50_1d was as high as 12.5±3.8 at75 °C (Fig. 5b). Because the max CO2/N2 SF was estimated to be ~20 bymolecular simulation [63], the separation performance of M_50_1d in-dicates the feasibility of the template-free CHA membranes. It is notedthat the permeance behavior of the CO2 and N2 molecules throughM_50_1d was quite unique compared to that through the other CHAtype zeolite membranes. In general, CHA zeolite membranes exhibit amonotonic decrease of both the CO2 permeance and the correspondingCO2/N2 SF with increasing temperature under dry conditions, as ad-sorption-based separation is likely to be dominant [46,64,65]. In con-trast, the max CO2/N2 SF for M_50_1d was observed at 75 °C, while theCO2 permeance increased monotonically from 30 to 100 °C. This maybe because the effective pore size of the OSDA-free CHA zeolite wassmaller than that of the C-SSZ-13 particles (Fig. 3). The additional in-crease in the nominal Si/Al ratio did not allow for the formation of a

Fig. 4. Top-view (left) and cross-sectional-view (middle) SEM images of (a1)-(a2) M_20_1d, (b1)-(b2) M_50_1d, (c1)-(c2) M_100_1d, and (d1)-(d2) M_∞_1d along with (a3)-(d3) thecorresponding XRD patterns (right). Yellow arrows indicate the bare alumina region, where the seeded particles were not intergrown during secondary growth. The black scale barsrepresent 20 µm and the asterisks (*) indicate the XRD peaks from the α-Al2O3 disc. (For interpretation of the references to color in this figure legend, the reader is referred to the webversion of this article)

E. Jang et al. Journal of Membrane Science 549 (2018) 46–59

52

continuous membrane in the case of M_100_1d and M_∞_1d (Fig. 4c1-d1), and thus, neither membranes exhibited any separation ability forthe CO2/N2 mixtures (Fig. 5c-d). The N2 molecules, which are lighterthan CO2 (N2 molecular weight (28) vs. CO2 molecular weight (44)),could permeate both membranes faster with a corresponding CO2/N2

SF of ~0.8, indicative of dominant Knudsen diffusion.

3.2.3. Effect of hydrothermal reaction times on the formation of continuousmembranes

Using the nominal Si/Al ratio of 50, we further varied the durationof the hydrothermal reaction (0.5 d, 1.5 d, and 2 d), since M_50_1dshowed marked CO2/N2 separation performance (Fig. 5b). The longerdurations of 1.5 d and 2 d also resulted in the synthesis of continuousmembranes with surface morphologies that were almost identical tothat of M_50_1d (Fig. 6b-d). However, the shorter duration of 0.5 d wasnot sufficient to complete the intergrowth of the seed particles, leading

to the co-existence of string- and plate-like particles on the top surface(Fig. 6a). It appears that the plate-like particles were initially formed,along with a majority of string-like particles (Fig. 6a); further inter-growth formed continuous membranes (Fig. 6b-d). Intriguingly, thestring-like particles were observed irrespective of the secondary growthtime (Fig. 6a-d). In addition to M_50_1d (Fig. 6b), some cracks, in-dicated by red arrows in Fig. 6c-d, were also found for M_50_xd (x =1.5 and 2). The corresponding XRD patterns shown in Fig. 6e confirmthat all membranes obtained after secondary growth for 1–2 d had thepure CHA zeolite structure with a minor left-shift of the XRD peaks, aspreviously observed in the XRD patterns of OSDA-free particles (Fig. 2and Table 2).

Furthermore, the CO2/N2 separation performance of the series ofsamples obtained at different secondary growth times (M_50_xd; x =0.5, 1.5, and 2) was evaluated under dry conditions (Fig. 7). As ex-pected from the similarity of the morphologies in the top-view SEM

Fig. 5. CO2/N2 separation performance of M_x_1d (x= (a) 20, (b) 50, (c) 100, and (d) ∞) as a function oftemperature up to 200 °C under dry conditions. In allgraphs, gray dashed lines represent the CO2/N2 SF of~0.8, determined by the Knudsen diffusion. In ad-dition, the red dashed lines, which represent theCO2/N2 SF of 10 (ideal selectivity determined frommultiplication of permeation selectivity and diffu-sion selectivity [66]), are included for eye guidance.Finally, the separation performances of M_50_1ddisplayed with the error bars in (b) were obtainedfrom the permeation measurement of three differentsamples. (For interpretation of the references to colorin this figure legend, the reader is referred to the webversion of this article)

Fig. 6. SEM images of M_50_xd (x = (a) 0.5, (b) 1, (c) 1.5, and (d) 2) along with (e) their XRD patterns as well as the simulated XRD pattern of all-silica CHA zeolite. Red arrows indicatesome observed cracks. The black scale bars represent 10 µm and the asterisks (*) indicate the XRD peak from the α-Al2O3 disc. For fair comparison, the intensity of the XRD peakcorresponding to the α-Al2O3 disc was used to normalize the XRD patterns of the membrane samples. (For interpretation of the references to color in this figure legend, the reader isreferred to the web version of this article)

E. Jang et al. Journal of Membrane Science 549 (2018) 46–59

53

images (Fig. 6b-d), the CO2/N2 separation performance of M_50_1.5dand M_50_2d were comparable to that of M_50_1d with the minor dif-ference being the decreased permeance of CO2 and N2 moleculesthrough the membranes. The reduction in the permeance observed inFig. 7b-d can be associated with the thicker membrane obtained withthe increased synthesis time (Fig. S7). Along with the SEM character-ization (Fig. 6a), the constant CO2/N2 SF of ~0.8 for M_50_0.5d overthe evaluated temperature range up to 200 °C also supports the in-complete intergrowth of the seed layer (Fig. 7a). From these multipleattempts, we could conclude that the secondary growth of the SSZ-13seed layer with a synthetic precursor (with a nominal Si/Al ratio of~50) for the duration of ~1 d is optimal for the fabrication of high-quality OSDA-free CHA membranes in a reproducible manner. Notably,synthesis of the OSDA-free CHA membranes requires a duration herein,~1–1.5 d that is comparable to ~2 d needed to acquire conventionalSSZ-13 membranes using TMAdaOH as an OSDA [67]. Despite thedrawback that the smaller effective pore size of the template- or OSDA-free zeolite membranes results in lowered permeance [42], this simplerprocess based on template-free secondary growth is considerably ben-eficial for the realization of large-scale zeolite membrane manu-facturing.

3.2.4. Elucidation of non-zeolitic, defective structuresFurthermore, we, for the first time, attempted to investigate the

structure of defects such as cracks and grain-boundary defects in thehigh-performance OSDA- or template-free membranes (here, M_50_1d)by using FCOM analysis. Despite the lack of a calcination step, M_50_1dobviously had defects, mainly cracks, throughout the membrane surfaceas shown in Fig. 8, implying that the formation of defects could not beavoided. Indeed, a previous study showed that crack formation in anorganic template-free MFI membrane was inevitable after drying abovea certain temperature [68]. Although the MFI membrane showed thedefect formation after drying at ~250 °C, M_50_1d, which had beendried at ~100 °C for activation, showed some cracks as shown in theSEM image (Fig. 4b1). The Si/Al ratios of the MFI [68] and CHA(P_50_1d, Table 2) particles obtained by seeded growth were ~10 and~5.5, respectively. Considering this, it could be inferred that more Alatoms were incorporated into the framework in the corresponding CHAmembrane sample (here, M_50_1d), which had the Si/Al ratio of ~2.1(Fig. S7b2). As mentioned above, similar crack formation was also

observed on the organic template-free CHA membrane with the Si/Alratio of 3.2 [31]. This suggests that the formation of defects was pre-sumably due to insufficient intergrowth among the polycrystallinegrains during the hydrothermal secondary growth [69]. In particular, asit is extremely challenging to synthesize the Al-rich LTA and FAU typemembranes for gas separations [70], it appears that the CHA mem-branes (M_50_xd; x = 1, 1.5, and 2) with the low Si/Al ratios of~2.1–3.8 (Fig. S7b2-d2) still suffered from the undesired crack for-mation. Despite the presence of some cracks, all membranes M_50_xd (x= 1, 1.5, and 2) showed good CO2/N2 separation performances(Fig. 7b-d), supporting the minor presence of the cracks.

The cross-sectional-view FCOM images reveal that two types ofdefects were present; (1) one type defect that propagated fully down tothe interface between M_50_1d and the α-alumina support and (2) theother type defect that existed near the surface, as respectively indicatedby yellow and red arrows in Fig. 8. Although the defects that propa-gated fully down to the interface would deteriorate the membrane se-paration performance, the density of these defects was significantlylower than that of the defects present near the surface (indicated by thesmaller number of yellow arrows in Fig. 8) and seemingly lower thanthat in the high CO2 perm-selective DDR membrane [17]. Referring to aprevious study on the investigation of drying temperature on crackformation in a MFI membrane [68], further drying of M_50_1d at thehigher temperature would result in the eventual propagation of cracksdown to the interface. In addition, we demonstrated the FCOM resultsof the DDR membrane [17] in Fig. S8, according to the display formatused for Fig. 8. It appears that the fully propagated cracks (indicated byyellow arrows) along the membrane thickness existed more in the CO2

perm-selective DDR membrane (with the CO2/N2 SF being as high as~13.2 at 30 °C), showing the beneficial organic-template free sec-ondary growth methodology. Further analysis showed that most of thedefect-free portions seemingly consisted of ~10–40 grains, as estimatedfrom Fig. S9a-b (yellow dots). Desirably, almost no grain-boundarydefects were observed around the individual grains in M_50_1d, ascompared to the ~10–30 µm thick MFI type zeolite membrane, whichshowed poor molecular sieving ability [71,72]. Moreover, the OSDA-free synthetic protocol allowed for the formation of well-intergrownmembrane constituents (Fig. S9c-d). In combination, these featuresindicate that the approach employing OSDA-free secondary growth iseffective for avoiding the considerable generation of defects that serve

Fig. 7. CO2/N2 separation performance for M_50_xd(x = (a) 0.5 (b) 1, (c) 1.5, and (d) 2) under dryconditions. In all graphs, the gray dashed lines re-present the CO2/N2 SF of ~0.8, determined fromKnudsen diffusion and red dashed lines, which re-present the CO2/N2 SF of 10 (the ideal permeationselectivity determined from multiplication of thesorption selectivity and the diffusion selectivity[66]), are included for eye guidance. (For inter-pretation of the references to color in this figure le-gend, the reader is referred to the web version of thisarticle)

E. Jang et al. Journal of Membrane Science 549 (2018) 46–59

54

as non-selective pathways along the membrane thickness. Thus, acontinuous CHA membrane (here, M_50_1d) with good performancewas obtained via the OSDA-free secondary growth method (Fig. 4b1-b3), which should afford good-quality molecular sieving (Fig. 5b) pri-marily through the dominant zeolitic part (Fig. 8).

3.2.5. CO2/N2 and CO2/CH4 separation performancesEncouraged by the high CO2/N2 separation performance, we further

examined the CO2/N2 separation performance of M_50_1d under wetconditions (Fig. 9b), as H2O vapor is the 3rd largest component in theflue gas stream generated from fossil fuel-fired power plants [73,74].Considering that the removal of H2O vapor prior to membrane-basedseparation is energy-intensive [75], robust CO2-selective separationcapacity of zeolite membranes, especially at ~50–75 °C, is highly de-sirable. At the moderate temperatures of 30–50 °C, no permeationthrough M_50_1d was detectable at the limit of the TCD detector (es-timated to be as low as ~1 × 10−9 mol m−2 s−1 Pa−1), mainly due tostrong inhibition by the H2O vapor. Because of the prominent hydro-philic portion of M_50_1d, as supported by the EDX data in Fig. S7b2,H2O vapor was preferentially adsorbed inside the CHA zeolite and re-duced the effective pore size, similar to the permeation behavior of NaYmembranes under wet conditions [64]. It was noted that the Si/Al ratio(~2.1) of M_50_1d shown in Fig. S7b2 was lower than that (~5.5) of thepowder counterpart (P_50_1d) (Table 2), apparently due to the addi-tional framework incorporation of Al atoms leached from α-Al2O3 disc[42]. Despite the negligible permeances up to ~50 °C, above 75 °C, thecapacity to separate CO2/N2 mixtures for WET CO2/N2 was recoveredas the affinity for H2O adsorption was weakened. Notably, the maxCO2/N2 SF of M_50_1d was as high as 10.0± 1.0 at 125 °C with a

corresponding CO2 permeance of 7.5 × 10−8 mol m−2 s−1 Pa−1. Al-though the OSDA-free membrane would not be useful at low tem-peratures such as flue gas temperatures, this membrane can still beapplied to CO2 separation under high temperature and wet feed con-ditions [76,77]. In fact, the representative temperature range of~50–75 °C for the flue gas stream originates from the desulfurizationprocess, which usually requires copious amounts of water for cooling[78]. Therefore, the CO2 perm-selective membrane (M_50_1d) can po-tentially be applied to CO2/N2 separation at higher temperatures (up to150 °C as shown in Fig. 9b) prior to the desulfurization process.

In addition to the good CO2/N2 separation performance, M_50_1dshowed a max CO2/CH4 SF as high as 28.8± 6.9 at 30 °C under DRYCO2/CH4 (Fig. 9c). This value is larger than the max CO2/N2 SF (12.5),apparently due to the larger molecular size of CH4 (0.38 nm) relative tothat of N2 (0.364 nm). Although the permeance of CO2 would be morestrongly inhibited by the larger CH4, we found that for the CO2/N2

(Fig. 9a) and CO2/CH4 mixtures (Fig. 9c), the permeance of CO2

through the membrane was comparable, indicating the weak interac-tion of CO2 with CH4. However, it was noted that the CO2 permeanceunder WET CO2/CH4 was lower than that under WET CO2/N2, in-dicating that CO2 molecules are obviously more strongly inhibited bythe larger CH4 molecules in the presence of H2O, as evident fromFig. 9c-d. As observed from the wet CO2/N2 permeation test, no per-meance of CO2 and CH4 molecules was not detected below 50 °C underwet conditions (the corresponding detection limit was approximated as2 × 10−10 mol m−2 s−1 Pa−1). Nevertheless, both molecules per-meated the membrane above 75 °C (Fig. 9d). For WET CO2/CH4, themax CO2/CH4 SF and CO2 permeance were 11.3±2.4 and 6.2 ×10−9 mol m−2 s−1 Pa−1 at 75 °C, respectively. Considering the

Fig. 8. (a)-(e) Cross-sectional-view and (f)-(g) top-view FCOM images of M_50_1d. The top-view FCOMimages in (f) and (g) were obtained from the posi-tions indicated by the yellow lines in (a)-(e). Thecross-sectional-view FCOM images in (a)-(e) wereobtained from the position indicated by the fiveyellow green lines in (f)-(g). In (a)-(g), the two typesof defects that (1) propagated into the interface be-tween M_50_1d and the α-Al2O3 disc and (2) existednear the surface are indicated by yellow and redarrows, respectively. The surface and interface ofM_50_1d are designated by white dashed lines in (a)-(e). (For interpretation of the references to color inthis figure legend, the reader is referred to the webversion of this article)

E. Jang et al. Journal of Membrane Science 549 (2018) 46–59

55

representative temperature of ~40–70 °C and water content of 3–6 vol% in the biogas stream [80,81], a slight increase of the feed tempera-ture for example, to 100–125 °C, is needed to secure modest CO2 per-meance (6.2 × 10−8 mol m−2 s−1 Pa−1 at 125 °C) with a marked CO2/CH4 SF (8.8) under wet conditions.

3.2.6. Comparison of the separation performance with literature dataFig. 10a presents a comparison of the CO2/N2 separation perfor-

mance of M_50_1d with those of documented zeolite membranes underboth dry and wet conditions. In coal-fired power plants, the operationspans a wide temperature range from the coal-fired boiler temperature(~1100–1600 °C) to the stack gas temperature (~50–75 °C) [82].Herein, the membrane performance was evaluated in the range of100–110 °C, which corresponds to that expected prior to the flue gasdesulfurization process. As mentioned above, at the selected tempera-tures, M_50_1d showed marked CO2 perm-selectivity under WET CO2/N2. In this temperature range, FAU [64] and CVD-treated CHA [66]zeolite membranes had a trade-off between the separation performanceunder dry and wet conditions, exhibiting high CO2 permeance and lowCO2/N2 SF under dry conditions vs. low CO2 permeance and high CO2/N2 SF under wet conditions. This suggests that at the high temperatureof ~100 °C, H2O molecules could still be adsorbed in the membranesand hamper the permeation of CO2. More importantly, the permeance

of the larger N2 molecule was further decreased, mainly due to in-hibition by the adsorbed H2O molecules. This difference in the degreeof inhibition by the adsorbed H2O molecules led to an increase in theCO2/N2 SF under wet conditions. In contrast, M_50_1d, as well as SAPO-34 [65] and SSZ-13 [46] membranes, exhibited different permeationbehavior, in which the CO2/N2 SFs were comparable under both dryand wet conditions, whereas under wet conditions, the CO2 and N2

permeance both decreased to a similar extent relative to the corre-sponding values under dry conditions. The fact that the CO2/N2 SF(~8.8) of M_50_1d at ~100 °C under dry conditions was almost twice aslarge as those of the FAU and CVD-treated CHA zeolite membranessuggests a lower degree of non-zeolitic defects. Notably, the CO2/N2

separation performance of M_50_1d was comparable to that of theSAPO-34 membrane obtained by using TEAOH and dipropylamine asOSDAs [65], making it promising for practical use under harsh condi-tions.

We further compared the CO2/CH4 separation performance ofM_50_1d with those of other membranes; SSZ-13 [19,46], DDR [83],and SAPO-34 membranes [23] (Fig. 10b). For Fig. 10b, we consideredthe CO2/CH4 separation performance at 20–30 and 100 °C. It was notedthat both M_50_1d and other CHA and DDR membranes showed goodCO2/CH4 separation performance under dry conditions. When the feedtemperature was increased from 20 to 30 °C to 100 °C, the CO2/CH4 SF

Fig. 9. (a)-(b) CO2/N2 (upper) and (c)-(d) CO2/CH4

(lower) separation performance of M_50_1d underdry (left) and wet (right) conditions as a function oftemperature up to 200 °C. Red dashed lines, whichrepresent the CO2/N2 SF of 10 in (a)-(b) and CO2/CH4 SF of 17 in (c)-(d) (the ideal permeation se-lectivities determined from multiplication of thesorption selectivity and the diffusion selectivity[46,66,79]), are included for eye guidance. Underwet conditions, no permeation of CO2, N2, and CH4

molecules could be detected (below the dark pinkdashed line; detection limit) at the temperaturelower than and equal to 50 °C (left side of the bluedashed line) in (b) and (d). (For interpretation of thereferences to color in this figure legend, the reader isreferred to the web version of this article)

Fig. 10. (a) CO2/N2 separation factor or selectivityvs. CO2 permeance for M_50_1d along with that ofother zeolite membranes (SAPO-34 [65], NaY [64],CVD-treated CHA [66], and SSZ-13 [46]) under bothdry and wet conditions at 100–110 °C. (b) CO2/CH4

SF vs. CO2 permeance of M_50_1d along with those ofother zeolite membranes (SAPO-34 [23], SSZ-13[19,46], DDR [83]); permeation tests were con-ducted at 20–30 and 100 °C under dry and wetconditions. In (a), the red dashed line, which re-presents the CO2/N2 SF of 10, is included for eyeguidance, while in (b), the red dashed lines representthe CO2/CH4 SFs of 10 and 20. In (a), the solid ar-rows indicate the change in the separation perfor-mance from those under dry conditions to thoseunder wet conditions. In (b), the dashed arrows in-dicate the temperatures during separation perfor-

mance measurement from low (20–30 °C) to high (100 °C).

E. Jang et al. Journal of Membrane Science 549 (2018) 46–59

56

and the CO2 permeance of the SSZ-13, SAPO-34, and DDR membranesconcomitantly decreased; specifically, the CO2/CH4 SFs were decreasedby almost half. On the contrary, M_50_1d showed a different behavior,whereby the CO2 permeance increased without any noticeable de-gradation of the CO2/CH4 SF.

The CO2/CH4 separation performance of M_50_1d under the wetcondition was inferior to that of the other SSZ-13 membrane at~100 °C, mainly because of the above-mentioned significant hindranceby the adsorbed H2O molecules. The degree of degradation of themembrane performance under WET CO2/CH4 was much higher forM_50_1d, for which the Si/Al ratio was presumably lower than that ofothers. Nevertheless, at the higher temperature of 100 °C, where theadsorption of H2O molecules is weakened, the membrane exhibitedgood separation performance under WET CO2/CH4 with a CO2/CH4 SFas high as ~10 at 100 °C. To be attractive for large-scale use, the re-duced CO2 permeance of M_50_1 under WET CO2/CH4 could be in-creased by adopting high-flux, asymmetric supports [84,85].

3.2.7. Long-term stability of OSDA-free CHA membranesFinally, the long-term stability of M_50_1d was evaluated at 100 °C

under the wet condition for a sufficiently long duration (3 d) underlaboratory settings (Fig. 11a) to ensure its robust use of CO2/N2 se-paration. During the continuous measurements, the CO2/N2 separationperformance was well maintained without any pronounced degrada-tion, suggesting preservation of the structural integrity of the CHAzeolite in the presence of H2O. Specifically, the average CO2/N2 SF at100 °C was 10.9± 0.3 and the average CO2 permeance was 6.3 ×10−8 mol m−2 s−1 Pa−1. Furthermore, a long-term stability test ofM_50_1d up to 3 d, carried out in an effort to evaluate the reliability ofthe CO2/CH4 separation ability at 100 °C under the wet condition, didnot indicate any significant degradation, also supporting the highstructural robustness of M_50_1d (Fig. 11b). Here, we emphasize thatthe high long-term stability of M_50_1d supports the effectiveness of theapproach based on the template- or OSDA-free secondary growth asobserved for other membranes that showed stabilities for water/ethanol[31,58] and CO2/CH4 separations [33].

4. Conclusions

CHA type zeolite particles and membranes were successfully syn-thesized in the absence of an OSDA via hydrothermal growth of seedparticles and seed layers, respectively. Specifically, very cheap in-organic reagents (KOH, NaOH, and NaAlO2) were used as SDAs insteadof the conventional OSDA, TMAdaOH, and thus, the calcination stepcould be omitted. In the organic template-free synthesis of the particles,a nominal Si/Al ratio lower than or equal to ~50 and a reaction time of~1 d were found to be optimal for acquiring high purity CHA zeolites.In contrast, a lower Al content and prolonged hydrothermal reactiontime led to formation of the undesired MOR type zeolite. In addition,synthetic conditions similar to those employed for the particle synthesiswere directly extended to the intergrowth of a SSZ-13 seed layer forobtaining OSDA-free CHA type zeolite membranes. Similarly, a nominal

Si/Al ratio of 50 and a hydrothermal reaction time of ~1–1.5 d wererequired for achieving CO2-selective CHA zeolite films with high re-producibility. In fact, the CO2 separation performance of M_50_1d wascomparable to those of conventional CHA membranes prepared withOSDAs. Although no permeate was detected below ~50 °C, apparentlydue to inhibition by H2O in the wet feed, high CO2/N2 and CO2/CH4

separation performance was achieved at the higher temperature of~75 °C, where the strength of adsorption of H2O vapor was less pro-nounced. The CO2/N2 and CO2/CH4 separation performance under wetconditions was well maintained up to ~125–150 °C. Long-term stabilitytests for the separation of CO2/N2 and CO2/CH4 mixtures at 100 °Cunder the wet condition for up to ~3 d showed no noticeable de-gradation, supporting the high structural robustness of the OSDA-freeCHA type zeolite membranes. In an effort to scale up these OSDA-freeCHA membranes, we are currently attempting to extend the manu-facturing protocol, proven to be effective on porous discs, to mem-branes on tubular supports.

Acknowledgements

This work was supported by the Korea CCS R&D Center (KCRC)(2014M1A8A1049309), by the International Research & DevelopmentProgram (2016K1A3A1A48954031), and by the Super Ultra LowEnergy and Emission Vehicle (SULEEV) Engineering Research Center(2016R1A5A1009592) through National Research Foundation (NRF) ofKorea. These grants were funded by the Korea government (Ministry ofScience and ICT). In addition, this research was supported by KoreaUniversity Future Research Grant. A part of SEM characterizations wascarried out at the Korea Basic Science Institute (KBSI).

Appendix A. Supplementary material

Supplementary data associated with this article can be found in theonline version at http://dx.doi.org/10.1016/j.memsci.2017.11.068.

References

[1] J. Caro, M. Noack, P. Kölsch, R. Schäfer, Zeolite membranes - state of their devel-opment and perspective, Microporous Mesoporous Mater. 38 (2000) 3–24.

[2] Z.P. 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, Microstructural optimi-zation of a zeolite membrane for organic vapor separation, Science 300 (2003)456–460.

[3] H. Voß, A. Diefenbacher, G. Schuch, H. Richter, I. Voigt, M. Noack, J. Caro, Buteneisomers separation on titania supported MFI membranes at conditions relevant forpractice, J. Membr. Sci. 329 (2009) 11–17.

[4] W.G. Sun, X.W. Wang, J.H. Yang, J.M. Lu, H.L. Han, Y. Zhang, J.Q. Wang,Pervaporation separation of acetic acid-water mixtures through Sn-substitutedZSM-5 zeolite membranes, J. Membr. Sci. 335 (2009) 83–88.

[5] W. Yuan, Y.S. Lin, W. Yang, Molecular sieving MFI-type zeolite membranes forpervaporation separation of xylene isomers, J. Am. Chem. Soc. 126 (2004)4776–4777.

[6] J. Priegnitz, The Selective Separation of Butene-1 from a C4 Hydrocarbon MixtureEmploying Zeolites X and Y, US 3,723,561, 1973.

[7] B. Tijsebaert, C. Varszegi, H. Gies, F.-S. Xiao, X. Bao, T. Tatsumi, U. Müller, D. DeVos, Liquid phase separation of 1-butene from 2-butenes on all-silica zeolite RUB-41, Chem. Commun. (2008) 2480–2482.

Fig. 11. Evaluation of long-term stability of M_50_1dfor the separation of (a) CO2/N2 and (b) CO2/CH4

binary mixtures under the wet condition at 100 °Cfor 72 h.

E. Jang et al. Journal of Membrane Science 549 (2018) 46–59

57

[8] H. Richter, H. Voß, I. Voigt, A. Diefenbacher, G. Schuch, F. Steinbach, J. Caro, High-flux ZSM-5 membranes with an additional non-zeolite pore system by alcohol ad-dition to the synthesis batch and their evaluation in the 1-butene/i-butene se-paration, Sep. Purif. Technol. 72 (2010) 388–394.

[9] T. Sano, S. Ejiri, K. Yamada, Y. Kawakami, H. Yanagishita, Separation of acetic acid-water mixtures by pervaporation through silicalite membrane, J. Membr. Sci. 123(1997) 225–233.

[10] G. Li, E. Kikuchi, M. Matsukata, Separation of water-acetic acid mixtures by per-vaporation using a thin mordenite membrane, Sep. Purif. Technol. 32 (2003)199–206.

[11] M.A. Snyder, M. Tsapatsis, Hierarchical nanomanufacturing: from shaped zeolitenanoparticles to high-performance separation membranes, Angew. Chem. -Int. Ed.46 (2007) 7560–7573.

[12] J. Van den Bergh, W. Zhu, J. Gascon, J.A. Moulijn, F. Kapteijn, Separation andpermeation characteristics of a DD3R zeolite membrane, J. Membr. Sci. 316 (2008)35–45.

[13] S.I. Zones, Zeolite SSZ-13 and its Method of Preparation US 4,544,538, 1985.[14] K. Kida, Y. Maeta, K. Yogo, Preparation and gas permeation properties on pure silica

CHA-type zeolite membranes, J. Membr. Sci. 522 (2017) 363–370.[15] G.T. Kokotailo, S.L. Lawton, D.H. Olson, W.M. Meier, Structure of synthetic zeolite

ZSM-5, Nature 272 (1978) 437–438.[16] C.T.P. Tung, H.S. Kim, K.B. Yoon, Growth of uniformly oriented silica MFI and BEA

zeolite films on substrates, Science 334 (2011) 1533–1538.[17] E. Kim, S. Hong, E. Jang, J.H. Lee, J.C. Kim, N. Choi, C.H. Cho, J. Nam, S.K. Kwak,

A.C.K. Yip, J. Choi, An oriented, siliceous deca-dodecasil 3R (DDR) zeolite film foreffective carbon capture: insight into its hydrophobic effect, J. Mater. Chem. A 5(2017) 11246–11254.

[18] T. Tomita, K. Nakayama, H. Sakai, Gas separation characteristics of DDR typezeolite membrane, Microporous Mesoporous Mater. 68 (2004) 71–75.

[19] H. Kalipcilar, T.C. Bowen, R.D. Noble, J.L. Falconer, Synthesis and separationperformance of SSZ-13 zeolite membranes on tubular supports, Chem. Mater. 14(2002) 3458–3464.

[20] J. Hedlund, M. Noack, P. Kölsch, D. Creaser, J. Caro, J. Sterte, ZSM-5 membranessynthesized without organic templates using a seeding technique, J. Membr. Sci.159 (1999) 263–273.

[21] S. Mintova, J. Hedlund, V. Valtchev, B.J. Schoeman, J. Sterte, Preparation of ZSM-5films from template free precursors, J. Mater. Chem. 7 (1997) 2341–2342.

[22] Y. Zhang, B. Tokay, H.H. Funke, J.L. Falconer, R.D. Noble, Template removal fromSAPO-34 crystals and membranes, J. Membr. Sci. 363 (2010) 29–35.

[23] S. Li, J.L. Falconer, R.D. Noble, SAPO-34 membranes for CO2/CH4 separation, J.Membr. Sci. 241 (2004) 121–135.

[24] J. Choi, H.K. Jeong, M.A. Snyder, J.A. Stoeger, R.I. Masel, M. Tsapatsis, Grainboundary defect elimination in a zeolite membrane by rapid thermal processing,Science 325 (2009) 590–593.

[25] S. Heng, P.P.S. Lau, K.L. Yeung, M. Djafer, J.C. Schrotter, Low-temperature ozonetreatment for organic template removal from zeolite membrane, J. Membr. Sci. 243(2004) 69–78.

[26] C.W. Jones, K. Tsuji, T. Takewaki, L.W. Beck, M.E. Davis, Tailoring molecular sieveproperties during SDA removal via solvent extraction, Microporous MesoporousMater. 48 (2001) 57–64.

[27] E. Kim, J. Choi, M. Tsapatsis, On defects in highly a-oriented MFI membranes,Microporous Mesoporous Mater. 170 (2013) 1–8.

[28] S. Gopalakrishnan, T. Yamaguchi, S.I. Nakao, Permeation properties of templatedand template-free ZSM-5 membranes, J. Membr. Sci. 274 (2006) 102–107.

[29] G. Li, E. Kikuchi, M. Matsukata, ZSM-5 zeolite membranes prepared from a cleartemplate-free solution, Microporous Mesoporous Mater. 60 (2003) 225–235.

[30] R. Lai, G.R. Gavalas, ZSM-5 membrane synthesis with organic-free mixtures,Microporous Mesoporous Mater. 38 (2000) 239–245.

[31] Y. Hasegawa, H. Hotta, K. Sato, T. Nagase, F. Mizukami, Preparation of novelchabazite (CHA)-type zeolite layer on porous α-Al2O3 tube using template-freesolution, J. Membr. Sci. 347 (2010) 193–196.

[32] H. Imai, N. Hayashida, T. Yokoi, T. Tatsumi, Direct crystallization of CHA-typezeolite from amorphous aluminosilicate gel by seed-assisted method in the absenceof organic-structure-directing agents, Microporous Mesoporous Mater. 196 (2014)341–348.

[33] H. Shi, Organic template-free synthesis of SAPO-34 molecular sieve membranes forCO2-CH4 separation, RSC Adv. 5 (2015) 38330–38333.

[34] B. Liu, Y. Zheng, N. Hu, T. Gui, Y. Li, F. Zhang, R. Zhou, X. Chen, H. Kita, Synthesisof low-silica CHA zeolite chabazite in fluoride media without organic structuraldirecting agents and zeolites, Microporous Mesoporous Mater. 196 (2014) 270–276.

[35] X. Li, H. Kita, H. Zhu, Z. Zhang, K. Tanaka, K.-i. Okamoto, Influence of the hy-drothermal synthetic parameters on the pervaporative separation performances ofCHA-type zeolite membranes, Microporous Mesoporous Mater. 143 (2011)270–276.

[36] Y. Kamimura, S. Tanahashi, K. Itabashi, A. Sugawara, T. Wakihara, A. Shimojima,T. Okubo, Crystallization behavior of zeolite beta in OSDA-free, seed-assistedsynthesis, J. Phys. Chem. C 115 (2011) 744–750.

[37] G. Majano, L. Delmotte, V. Valtchev, S. Mintova, Al-rich zeolite beta by seeding inthe absence of organic template, Chem. Mater. 21 (2009) 4184–4191.

[38] Y. Tang, X. Liu, S. Nai, B. Zhang, Template-free synthesis of beta zeolite membraneson porous α-Al2O3 supports, Chem. Commun. 50 (2014) 8834–8837.

[39] N. Rangnekar, N. Mittal, B. Elyassi, J. Caro, M. Tsapatsis, Zeolite membranes–areview and comparison with MOFs, Chem. Soc. Rev. 44 (2015) 7128–7154.

[40] K. Iyoki, K. Itabashi, T. Okubo, Progress in seed-assisted synthesis of zeoliteswithout using organic structure-directing agents, Microporous Mesoporous Mater.189 (2014) 22–30.

[41] M. Kanezashi, J. O’Brien, Y.S. Lin, Template-free synthesis of MFI-type zeolitemembranes: permeation characteristics and thermal stability improvement ofmembrane structure, J. Membr. Sci. 286 (2006) 213–222.

[42] M. Pan, Y.S. Lin, Template-free secondary growth synthesis of MFI type zeolitemembranes, Microporous Mesoporous Mater. 43 (2001) 319–327.

[43] N.Y. Kang, B.S. Song, C.W. Lee, W.C. Choi, K.B. Yoon, Y.-K. Park, The effect ofNa2SO4 salt on the synthesis of ZSM-5 by template free crystallization method,Microporous Mesoporous Mater. 118 (2009) 361–372.

[44] M. Moliner, C. Martinez, A. Corma, Synthesis strategies for preparing useful smallpore zeolites and zeotypes for gas separations and catalysis, Chem. Mater. 26 (2014)246–258.

[45] N. Kosinov, C. Auffret, G.J. Borghuis, V.G.P. Sripathi, E.J.M. Hensen, Influence ofthe Si/Al ratio on the separation properties of SSZ-13 zeolite membranes, J. Membr.Sci. 484 (2015) 140–145.

[46] N. Kosinov, C. Auffret, C. Gücüyener, B.M. Szyja, J. Gascon, F. Kapteijn,E.J.M. Hensen, High flux high-silica SSZ-13 membrane for CO2 separation, J. Mater.Chem. A 2 (2014) 13083–13092.

[47] X. Wang, Q.M. Wu, C.Y. Chen, S.X. Pan, W.P. Zhang, X.J. Meng, S. Maurer,M. Feyen, U. Müller, F.S. Xiao, Atom-economical synthesis of a high silica CHAzeolite using a solvent-free route, Chem. Commun. 51 (2015) 16920–16923.

[48] J. Choi, S. Ghosh, L. King, M. Tsapatsis, MFI zeolite membranes from a- and ran-domly oriented monolayers, Adsorption 12 (2006) 339–360.

[49] A. Boultif, D. Louër, Powder pattern indexing with the dichotomy method, J. Appl.Cryst. 37 (2004) 724–731.

[50] J. Rodríguez-Carvajal, Recent advances in magnetic structure determination byneutron powder diffraction, Physica B 192 (1993) 55–69.

[51] A. Le Bail, Whole powder pattern decomposition methods and applications: a ret-rospection, Powder Diffr. 20 (2005) 316–326.

[52] V. Petříček, M. Dušek, L. Palatinus, Crystallographic computing system JANA2006:general features, Z. Krist. 229 (2014) 345–352.

[53] P. Sharma, P. Rajaram, R. Tomar, Synthesis and morphological studies of nano-crystalline MOR type zeolite material, J. Colloid Interf. Sci. 325 (2008) 547–557.

[54] L. Sommer, A. Krivokapić, S. Svelle, K.P. Lillerud, M. Stöcker, U. Olsbye, Enhancedcatalyst performance of zeolite SSZ-13 in the methanol to olefin reaction afterneutron irradiation, J. Phys. Chem. C 115 (2011) 6521–6530.

[55] A. Shishkin, H. Kannisto, P.A. Carlsson, H. Härelind, M. Skoglundh, Synthesis andfunctionalization of SSZ-13 as an NH3-SCR catalyst, Catal. Sci. Technol. 4 (2014)3917–3926.

[56] L. Sommer, D. Mores, S. Svelle, M. Stöcker, B.M. Weckhuysen, U. Olsbye, Mesoporeformation in zeolite H-SSZ-13 by desilication with NaOH, Microporous MesoporousMater. 132 (2010) 384–394.

[57] I. Lezcano-Gonzalez, U. Deka, B. Arstad, A. Van Yperen-De Deyne, K. Hemelsoet,M. Waroquier, V. Van Speybroeck, B.M. Weckhuysen, A.M. Beale, Determining thestorage, availability and reactivity of NH3 within Cu-chabazite-based ammoniaselective catalytic reduction systems, Phys. Chem. Chem. Phys. 16 (2014)1639–1650.

[58] R. Zhou, Y. Li, B. Liu, N. Hu, X. Chen, H. Kita, Preparation of chabazite membranesby secondary growth using zeolite-T-directed chabazite seeds, MicroporousMesoporous Mater. 179 (2013) 128–135.

[59] Y. Ji, J. Birmingham, M.A. Deimund, S.K. Brand, M.E. Davis, Steam-dealuminated,OSDA-free RHO and KFI-type zeolites as catalysts for the methanol-to-olefins re-action, Microporous Mesoporous Mater. 232 (2016) 126–137.

[60] Y. Ji, M.A. Deimund, Y. Bhawe, M.E. Davis, Organic-free synthesis of CHA-typezeolite catalysts for the methanol-to-olefins reaction, ACS Catal. 5 (2015)4456–4465.

[61] J. Scherzer, J.L. Bass, F.D. Hunter, Structural characterization of hydrothermallytreated lanthanum Y zeolites. I. Framework vibrational spectra and crystal struc-ture, J. Phys. Chem. 79 (1975) 1194–1199.

[62] M. Martis, A.J. Smith, C. Tang, J.E. Parker, T.I. Hyde, M.J. Watson, X. Baucherel,S. Kohara, T. Wakihara, G. Sankar, Tracking the structural changes in pure andheteroatom substituted aluminophosphate, AIPO-18, using synchrotron based X-raydiffraction techniques, Phys. Chem. Chem. Phys. 15 (2013) 11766–11774.

[63] S. Li, J.L. Falconer, R.D. Noble, R. Krishna, Modeling permeation of CO2/CH4, CO2/N2, and N2/CH4 mixtures across SAPO-34 membrane with the Maxwell−Stefanequations, Ind. Eng. Chem. Res. 46 (2007) 3904–3911.

[64] X. Gu, J. Dong, T.M. Nenoff, Synthesis of defect-free FAU-type zeolite membranesand separation for dry and moist CO2/N2 mixtures, Ind. Eng. Chem. Res. 44 (2005)937–944.

[65] S. Li, C.Q. Fan, High-flux SAPO-34 membrane for CO2/N2 separation, Ind. Eng.Chem. Res. 49 (2010) 4399–4404.

[66] E. Kim, T. Lee, H. Kim, W.-J. Jung, D.-Y. Han, H. Baik, N. Choi, J. Choi, Chemicalvapor deposition on chabazite (CHA) zeolite membranes for effective post-com-bustion CO2 capture, Environ. Sci. Technol. 48 (2014) 14828–14836.

[67] Y. Zheng, N. Hu, H. Wang, N. Bu, F. Zhang, R. Zhou, Preparation of steam-stablehigh-silica CHA (SSZ-13) membranes for CO2/CH4 and C2H4/C2H6 separation, J.Membr. Sci. 475 (2015) 303–310.

[68] M. Lassinantti, F. Jareman, J. Hedlund, D. Creaser, J. Sterte, Preparation and eva-luation of thin ZSM-5 membranes synthesized in the absence of organic templatemolecules, Catal. Today 67 (2001) 109–119.

[69] Z. Chen, D. Yin, Y. Li, J. Yang, J. Lu, Y. Zhang, J. Wang, Functional defect-patchingof a zeolite membrane for the dehydration of acetic acid by pervaporation, J.Membr. Sci. 369 (2011) 506–513.

[70] J. Caro, D. Albrecht, M. Noack, Why is it so extremely difficult to prepare shape-selective Al-rich zeolite membranes like LTA and FAU for gas separation? Sep. Purif.Technol. 66 (2009) 143–147.

[71] G. Bonilla, M. Tsapatsis, D.G. Vlachos, G. Xomeritakis, Fluorescence confocal

E. Jang et al. Journal of Membrane Science 549 (2018) 46–59

58

optical microscopy imaging of the grain boundary structure of zeolite MFI mem-branes made by secondary (seeded) growth, J. Membr. Sci. 182 (2001) 103–109.

[72] M.A. Snyder, Z. Lai, M. Tsapatsis, D.G. Vlachos, Combining simultaneous re-flectance and fluorescence imaging with SEM for conclusive identification ofpolycrystalline features of MFI membranes, Microporous Mesoporous Mater. 76(2004) 29–33.

[73] T.C. Merkel, H. Lin, X. Wei, R. Baker, Power plant post-combustion carbon dioxidecapture: an opportunity for membranes, J. Membr. Sci. 359 (2010) 126–139.

[74] D.M. D'Alessandro, B. Smit, J.R. Long, Carbon dioxide capture: prospects for newmaterials, Angew. Chem. -Int. Ed. 49 (2010) 6058–6082.

[75] M.T. Snider, H. Verweij, Gas sorption studies on zeolite Y membrane materials forpost-combustion CO2 capture in coal-fired plants, Microporous Mesoporous Mater.192 (2014) 3–7.

[76] R. Bredesen, K. Jordal, O. Bolland, High-temperature membranes in power gen-eration with CO2 capture, Chem. Eng. Process. 43 (2004) 1129–1158.

[77] M. Ostwal, R.P. Singh, S.F. Dec, M.T. Lusk, J.D. Way, 3-Aminopropyltriethoxysilanefunctionalized inorganic membranes for high temperature CO2/N2 separation, J.Membr. Sci. 369 (2011) 139–147.

[78] H. Zhai, E.S. Rubin, P.L. Versteeg, Water use at pulverized coal power plants withpostcombustion carbon capture and storage, Environ. Sci. Technol. 45 (2011)2479–2485.

[79] R. Krishna, J.M. van Baten, Segregation effects in adsorption of CO2-containing

mixtures and their consequences for separation selectivities in cage-type zeolites,Sep. Purif. Technol. 61 (2008) 414–423.

[80] M. Turco, A. Ausiello, L. Micoli, Treatment of Biogas for Feeding High TemperatureFuel Cells: Removal of Harmful Compounds by Adsorption Processes, Springer,2016.

[81] E. Favre, R. Bounaceur, D. Roizard, Biogas, membranes and carbon dioxide capture,J. Membr. Sci. 328 (2009) 11–14.

[82] R. Quinn, R.J. Kitzhoffer, J.R. Hufton, T.C. Golden, A high temperature lithiumorthosilicate-based solid absorbent for post combustion CO2 capture, Ind. Eng.Chem. Res. 51 (2012) 9320–9327.

[83] S. Himeno, T. Tomita, K. Suzuki, K. Nakayama, K. Yajima, S. Yoshida, Synthesis andpermeation properties of a DDR-type zeolite membrane for separation of CO2/CH4

gaseous mixtures, Ind. Eng. Chem. Res. 46 (2007) 6989–6997.[84] J. Hedlund, J. Sterte, M. Anthonis, A.J. Bons, B. Carstensen, N. Corcoran, D. Cox,

H. Deckman, W. De Gijnst, P.P. de Moor, F. Lai, J. McHenry, W. Mortier, J. Reinoso,High-flux, MFI membranes, Microporous Mesoporous Mater. 52 (2002) 179–189.

[85] K.V. Agrawal, B. Topuz, T.C.T. Pham, T.H. Nguyen, N. Sauer, N. Rangnekar,H. Zhang, K. Narasimharao, S.N. Basahel, L.F. Francis, C.W. Macosko, S. Al-Thabaiti, M. Tsapatsis, K.B. Yoon, Oriented MFI membranes by gel-less secondarygrowth of sub-100 nm MFI-nanosheet seed layers, Adv. Mater. 27 (2015)3243–3249.

E. Jang et al. Journal of Membrane Science 549 (2018) 46–59

59