Chabazite-Type Zeolite Membranes for Effective CO2 Separation: The

Subscriber access provided by University of Florida | Smathers Libraries

Energy, Environmental, and Catalysis Applications

Chabazite Type Zeolite Membranes for Effective CO2 Separation: The Roles of Hydrophobicity and Defect Structure Minseong Lee, Sungwon Hong, Dongjae Kim, Eunjoo Kim, Kyunghwan Lim, Jae Chil Jung, Dr. Hannes Richter, Jong-Ho Moon, Nakwon Choi, Jaewook Nam, and Jungkyu Choi ACS Appl. Mater. Interfaces, Just Accepted Manuscript • DOI: 10.1021/acsami.8b18854 • Publication Date (Web): 07 Jan 2019 Downloaded from http://pubs.acs.org on January 10, 2019

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 40 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

ACS Applied Materials & Interfaces

Chabazite Type Zeolite Membranes for Effective CO2 Separation: The Roles of Hydrophobicity and Defect Structure Minseong Lee,1,† Sungwon Hong,1,† Dongjae Kim,2 Eunjoo Kim,1 Kyunghwan Lim,1 Jae Chil Jung,3 Hannes Richter,4 Jong-Ho Moon,5 Nakwon Choi,6 Jaewook Nam,7 and Jungkyu Choi1,* 1

Department of Chemical and Biological Engineering, College of Engineering, Korea University, 145

Anam-ro, Seongbuk-gu, Seoul 02841, Republic of Korea 2

School of Chemical and Biological Engineering, Seoul National University, Seoul 08826, Republic of

Korea 3

FINETECH. Co., Ltd., 53-60 Jinwisandan-ro, Jinwi-myeon, Pyeongtaek-si, Gyeonggi-do 17709,

Republic of Korea 4

Fraunhofer Institute for Ceramic Technologies and Systems IKTS, Michael-Faraday-Str. 1, 07629

Hermsdorf, Germany 5

Greenhouse Gas Research Laboratory, Korea Institute of Energy Research, 152 Gajeong-ro, Yuseong-

gu, Daejeon 34129, Republic of Korea 6

Center for BioMicrosystems, Brain Science Institute, Korea Institute of Science and Technology

(KIST), 5 Hwarang-ro 14-gil, Seongbuk-gu, Seoul 02792, Republic of Korea 7

Institute of Chemical Process, School of Chemical and Biological Engineering, Seoul National

University, Seoul 08826, Republic of Korea †

These two authors equally contributed to this work.

* Corresponding Author E-mail address: [email protected], Phone: +82-2-3290-4854, and Fax: +82-2-926-6102 Keywords: Siliceous zeolite film; chabazite; secondary growth; CO2 permselectivity; hydrophobicity; defects. 1 ACS Paragon Plus Environment

ACS Applied Materials & Interfaces 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 2 of 40

Abstract

Chabazite (CHA) type zeolites are promising for the separation of CO2 from larger molecules, such as N2 (relevant to post-combustion carbon capture) and CH4 (relevant to natural gas/biogas upgrading). In particular, the pore size of CHA zeolites (0.37 × 0.42 nm2) can recognize slight molecular size differences between CO2 (0.33 nm) and the larger N2 (0.364 nm) or CH4 (0.38 nm) molecules, thus allowing separation in favor of CO2 through CHA membranes. Furthermore, the siliceous constituents in the CHA zeolite can reduce the adsorption capacity toward the smaller H2O molecule (0.265 nm) and, thus, the H2O permeation rate. This is highly desirable for securing good molecular sieving ability with CO2 permselectivity in the presence of H2O vapor. Indeed, a siliceous CHA film obtained with a nominal Si/Al ratio of 100 (CHA_100) showed high CO2/N2 and CO2/CH4 separation performance, especially in the presence of H2O vapor; ~13.4 CO2/N2 and ~37 CO2/CH4 separation factors (SFs) at 30 °C. These SFs were higher than the corresponding values (~5.2 CO2/CH4 SFs and ~31 CO2/CH4 SFs) under dry conditions; such improvement could be ascribed to defect blocking by physisorbed water molecules. Finally, the contribution of molecular transport through zeolitic and non-zeolitic parts was quantitatively analyzed by combining information extracted from image processing of fluorescence confocal optical microscopy images with a one dimensional permeation model. It appears that ~19% and ~20% of the total CO2 permeance for CHA_100 were reduced due to transport inhibition by the physisorbed water molecules on the membrane surface and defect, respectively.

2 ACS Paragon Plus Environment

Page 3 of 40 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


1. Introduction

Zeolites are aluminosilicate crystalline materials that have regular, rigid microporous structures. Because the size of the pore structures is close to those of permanent gases, a zeolite, if appropriately selected, can act as a molecular sieve, functioning analogously to a conventional sand/stone/gravel sieve used on construction sites. Indeed, zeolite membranes have high potential for distinguishing the minute differences between the shape and size of permanent gases, thus achieving energy-efficient gas separation.12

If a zeolite is chosen such that its pore size lies between the size of CO2 and other larger molecules, the

molecular sieve can act as a good CO2 separator. In this respect, eight membered-ring (8 MR) zeolites, whose maximum pore size is ~0.43 nm,3 are promising molecular sieves for the separation of CO2 (0.33 nm) from other larger gas molecules such as N2 (0.364 nm) and CH4 (0.38 nm). Effective CO2/N2 and CO2/CH4 separation is attractive to many researchers because of the importance of regulating flue gas emissions from fossil fuel combustion and in the upgrading of natural gas and biogas.

Thus, two 8 MR zeolites with the structure codes CHA (chabazite) and DDR (decadodecasil 3R) have been intensively investigated for use as membranes, and these zeolites show high CO2 separation performance.4-9 In particular, a CHA zeotype (SAPO-34) has been the main focus of investigation so far.10-13 However, the incorporation of heterogeneous atoms, such as Al and P, into the CHA type framework, which favors the adsorption of CO2, is not favorable for CO2/N2 separation in the presence of H2O vapor (from now on, denoted H2O) in the feed. The representative molar compositions of N2, CO2, and H2O in the flue gas stream emitted from a coal-fired power plant are ~70-77%, ~10-16%, and ~6-12%, respectively.14-17 For dry CO2/N2 separation, increasing the hydrophilicity (mainly via the control of Si/Al ratios)18 of a zeolite membrane with large pore sizes (for example, NaY) is a sound approach.19-20 Although the bulk concentration of H2O vapor can be reduced to some extent by lowering the temperature before membranebased separation processes are carried out, the permeation rate of CO2, which is determined by a convoluted coupling of its adsorption and diffusion properties, will be significantly decreased in the presence of H2O 3 ACS Paragon Plus Environment


Page 4 of 40

in the feed.21 The degree of this permeation rate reduction is pronounced if the preferred CO2 sorptionbased approaches are adopted as means to achieve high CO2 separation performance. This can be ascribed to the stronger adsorption of H2O than CO2 in zeolites.22-23 In addition, because of the smaller size of H2O (0.265 nm), the size-based exclusion of H2O while favoring CO2 transport will be extremely challenging. Therefore, hydrophobic siliceous zeolites whose interactions with H2O are minimized24 are suitable for the preparation of zeolite membranes with high CO2 separation performance even in the presence of H2O.8, 25

In particular, the temperature of the flue gas passing a desulfurization unit lies in the range of ~5070 °C and the total pressure is close to ~1 atm.15, 26 Moreover, a low fraction of CO2 (~10-15%) in the flue gas requires a chemisorption-based approach to secure high CO2/N2 separation performance. In fact, the facilitated transport of CO2 through polymeric and mixed matrix membranes is promising for CO2/N2 separations, even under wet conditions,27-29 though the long-term stabilities of these membranes should be verified before practical application.30-31 In addition, biogas streams are mainly equi-molar CO2/CH4 mixtures containing a minor portion of H2O vapor (~3-6 vol%) at ~40-70 °C.32-33 We found that most studies in the literature were done to investigate CO2/N2 and CO2/CH4 separation performance at ~25-30 °C under dry conditions, instead of the more realistic wet conditions at temperatures of ~50-70 °C.

Indeed, a siliceous, hydrophobic CHA membrane fabricated by using the secondary growth method has been reported to separate CO2 from both N2 and CH4.4 The siliceous, hydrophobic property of the CHA membranes has been shown to be effective in obtaining high separation factors (SFs) under wet conditions. Despite the high promise of the CHA membrane, its complicated manufacturing procedure requires simplification to allow facile and reproducible membrane fabrication. Hydrophobic DDR zeolite membranes also showed good CO2 permselectivities under wet conditions.7-8 Both cases indicate the desirability of hydrophobic membrane constituents for securing high CO2 permselectivities in H2Ocontaining feeds.


Page 5 of 40 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


In this study, we adopted the secondary growth method for the fabrication of siliceous CHA membranes. First, a uniform CHA seed layer, which was mainly comprised of plate-like CHA particles, was formed on a porous α-Al2O3 disc via a selective deposition method.5 The seed layer was further hydrothermally inter-grown to form a continuous, siliceous CHA membrane. For secondary growth, three different nominal Si/Al ratios of 20, 100, and infinity in the synthetic precursor were used. This approach was used to study the effect of the Si/Al ratio on the membrane properties systematically and, thus, the separation performances. Among the three, the most siliceous and, thus, most hydrophobic CHA membrane showed much improved separation performance for both CO2/N2 and CO2/CH4 mixtures under wet conditions compared to that under dry conditions. In addition, fluorescence confocal optical microscopy (FCOM) revealed that the CHA membranes showed distinctive features in the defect (crack) structure depending on the Si/Al ratios in the synthetic precursors. Image processing of the FCOM images allowed us to correlate the defect structure with the separation performance quantitatively. Furthermore, this relationship was used to elucidate the effect of the hydrophobicity and defect structure of the CHA membrane on the CO2 permselectivities in wet feeds. Finally, we compared the CO2 permselectivity of the CHA membranes in the presence of H2O vapor in the feed with those of CHA and other type zeolite membranes in the literature and considered the performance reported under both dry and wet conditions.

2. Experimental 2.1. Seed crystal synthesis and seed deposition

All-silica CHA (Si-CHA) zeolite particles were synthesized following a literature procedure.34 N,N,N-trimethyl-1-adamantammonium hydroxide (TMAdaOH, 25 wt% in water, Sachem Inc.) was poured into distilled (DI) water in a Teflon beaker. Ethanol (anhydrous 200 proof, ≥ 99.5%, Sigma-Aldrich) was subsequently added to the mixture, followed by the addition of tetraethyl orthosilicate (TEOS 98%, SigmaAldrich), which served as the silica source. The synthesis mixture was hydrolyzed overnight at room temperature and was further dried until the ratio of SiO2 to water reached ~2.4. After that, the synthesis 5 ACS Paragon Plus Environment


Page 6 of 40

mixture became almost solid. Finally, an HF solution (48%, Sigma-Aldrich) was added drop-wise to the solid-like synthesis mixture. The final molar composition was 10 SiO2: 5 TMAdaOH: 5 HF: 30 H2O. The resultant synthesis mixture was transferred to a Teflon liner in an autoclave for reaction. The autoclave containing the synthesis mixture was placed on a rotating rack in a preheated oven at 155 °C, and after completing the reaction for 42 h with rotation, the reaction was quenched with tap water. The resulting particles were recovered by repeating centrifugation and washing five times. In addition, the recovered particles were calcined at ~600 °C for 12 h at a ramp rate of 1 °C·min-1 under air flow (200 mL·min-1).

The seed deposition procedure has been reported elsewhere.5,

35

Briefly, ~0.05 g of Si-CHA

particles were added to ~50 mL of anhydrous toluene in a specially designed glass reactor with flowing argon (~20 mL·min-1). After that, a home-made α-Al2O3 disc sandwiched between two cover glasses was fixed on a comb-shaped Teflon holder and placed in the suspension inside the glass reactor. Detailed information about the manufacturing of the home-made α-Al2O3 discs (~21 mm in diameter and ~2 mm in thickness) and the placement of the disc on the Teflon holder can be found elsewhere.36 About 50 mL of anhydrous toluene was poured into the glass reactor under an argon environment. Subsequently, the glass reactor was sealed with Parafilm and ultra-sonicated (UC-10P, JEIO Tech, South Korea) for ~20 min. The seeded α-Al2O3 disc was calcined at 450°C for 4 h at a ramp rate of 1 °C·min-1.

Fig. S1a in the Supporting Information shows the scanning electron microscopy (SEM) images of the Si-CHA particles obtained following the protocol described above. As reported in our previous studies,5, 35

a majority of cubic-like Si-CHA particles having a wide size distribution were observed, as well as a

minor portion of plate-like Si-CHA particles (indicated by red arrows in Fig. S1a). Unfortunately, it was impossible to estimate the size of the plate-like Si-CHA particles, as they were present in an aggregated form (Fig. S1a). Instead, the measurement of ~200 cubic-like Si-CHA particles led to the size estimation of 8.6 ± 6.2 µm, supporting the aforementioned wide size distribution. Despite being the minor portion, the plate-like Si-CHA particles were selectively deposited onto α-Al2O3 discs to form a uniform seed layer by 6 ACS Paragon Plus Environment

Page 7 of 40 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


controlling the access to the disc surface (Fig. S1b) following above-mentioned approach reported in our previous study.5 Because of the broad size distribution, the large cubic-like Si-CHA particles were not appropriate for forming a uniform seed layer, which is critical in forming a continuous membrane via the secondary growth methodology.37-38 Fig. S1c shows that the deposited plate-like CHA particles were preferentially aligned. In particular, the X-ray diffraction (XRD) peak corresponding to the h0h-plane was pronounced, while the others observed in the powder XRD patterns almost vanished. The resulting seed layer was comparable to that reported in the literature.5

2.2. CHA film growth

Three types of CHA zeolite membranes were synthesized by varying the Si/Al ratio in the synthetic precursor. First, a certain amount of TMAdaOH, Al(OH)3 (for a finite Si/Al ratio), and CAB-O-SIL (CABO-Sil® M-5, Cabot Corporation) as a silica source were added to deionized water, and the mixture was further shaken by using a shaking machine (Lab Companion, SI-300R, South Korea) overnight at room temperature. After that, a certain amount of NaOH was added to the well-mixed synthetic precursor, followed by shaking overnight at room temperature. The final molar compositions of the three synthetic precursors for secondary growth were 20 TMAdaOH: 100 SiO2: 4,400 H2O: 20 NaOH: x Al(OH)3, where x = 0, 1, and 5 correspond to the nominal Si/Al ratios of infinity, 100, and 20, respectively. The prepared synthetic precursor was poured into a Teflon liner where a seeded α-Al2O3 disc had been placed vertically with the help of a Teflon holder. The Teflon liner was placed in an autoclave and the autoclave was moved to an oven preheated to 160 °C. The hydrothermal reaction for secondary growth was carried out at 160 °C for 6 d. After the reaction had finished, the autoclave was quenched with tap water. The as-synthesized membranes were recovered from the autoclave, washed with copious water, and dried at room temperature. The dried membranes were further calcined at 550 °C for 12 h at a heating ramp rate of 0.5 °C·min-1 under air flow (200 mL·min-1). For convenience, the calcined membrane samples are referred to as CHA_x, where x indicates the nominal Si/Al ratio in the synthetic precursor. 7 ACS Paragon Plus Environment


Page 8 of 40

2.3. Characterizations SEM images of CHA particles, layer, and membranes were obtained by using a Hitachi S-4300. Before analysis, all samples were coated by Pt-sputtering (Hitachi S-4300). The crystallinity of the Si-CHA zeolites and any orientation of the Si-CHA layer and membranes were confirmed by XRD (Rigaku Model D/Max-2500V/PC diffractometer, Japan) in the θ/2θ configuration using Cu Kα radiation (40 kV, 100 mA, λ = 0.154 nm). The simulated powder XRD pattern of CHA zeolite was generated using the Mercury program (from www.ccdc.cam.ac.uk). The crystallographic information file (CIF) of CHA zeolite was downloaded from www.iza-online.org. From the membrane surface to the part of the α-Al2O3 support that is adjacent to the CHA film, the chemical contents were analyzed by energy dispersive X-ray analysis (EDX, Hitachi S-4800); this was used to estimate the hydrophobicity of the membranes. In addition, the contact angles of water droplets on the membrane samples were measured at room temperature using an optical microscope (General Type Phoenix 300, Surface Electro Optics) and recorded with time. To investigate the inner structures of the CHA membranes, FCOM (Zeiss LSM-700) was used after the membranes had been impregnated with a fluorescent dye (fluorescein sodium salt, empirical formula: C20H10Na2O5, Sigma-Aldrich (Product No.: F6377)) having an approximate diameter of ~1 nm. Detailed information about the FCOM measurements can be found in other report.39 In this study, membrane dyeing was conducted for 12, 24, and 96 h. In addition, a simple screening test suitable for excluding highly defective membranes was used. This test involves placing several droplets of 1 mM fluorescent dye solution on the membrane surfaces and observing the dye distribution with time. Detailed information about the image processing and its combination with the 1-D permeation model can be found elsewhere.40

The Wicke-Kallenbach mode was adopted for measuring the permeation rates of CO2 and N2 (or CH4). The total pressures of both feed and permeate sides were maintained at 1 atm. To examine the effect of water on the membrane, the feed gas was humidified. Specifically, the partial pressures of CO2/N2/H2O (or CO2/CH4/H2O) were 48.5 kPa/48.5 kPa/3 kPa, respectively. A total flow rate of 100 mL⸱min-1 of CO2 and N2 (or CH4) on a dry basis was fed to the membrane, and the CO2/N2 or CO2/CH4 mixture permeated 8 ACS Paragon Plus Environment

Page 9 of 40 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


through the membrane. The permeate was carried by He sweep gas (100 mL⸱min-1) and was analyzed online by a gas chromatograph (GC) system (YL 6100 GC for CO2/N2 and YL 6500 GC for CO2/CH4, Young Lin Instruments, South Korea) equipped with a packed column (6 ft × 1/8” Porapak T for CO2/N2 and 30 m × 0.320 mm GS-GasPro for CO2/CH4) and a thermal conductivity detector (TCD) for CO2/N2 analysis and a pulsed discharge ionization detector (PDD) for CO2/CH4 analysis. For reliable analysis, ~5 mL⸱min-1 of CH4 for CO2/N2 and H2 for CO2/CH4 measurements were added to the swept permeate gas stream before reaching the GC system. In addition, permeation tests with respect to the different feed humidities and longer durations at 50 °C were conducted with CHA_100 and _∞. To evaluate the effect of the humidity on the separation performance of the membrane samples at 50 °C, the permeation tests were conducted at relative humidities of 0%, 26%, 60%, and 100% at 50 °C for 12 h each. The relative humidities of 0%, 26%, 60%, and 100% at 50 °C correspond to water vapor partial pressures of approximately 0, 3, 7, and 12 kPa, respectively. The long-term stability tests were conducted at 50 °C at the saturated water vapor of ~12 kPa. To expedite any deactivation, the membrane samples were heated to 200 °C between the long-term measurements at 50 °C. Specifically, the permeation tests were performed at 50 °C for 3 d, then at 200 °C for 2 d, and then again at 50 °C for 3 d.

3. Results and discussion 3.1. Formation of CHA membranes

Fig. 1a1-c1 and 1a2-c2 reveals that the secondary growth of the seed layer (shown in Fig. S1b) led to well-intergrown, continuous films at SEM resolution for all three Si/Al ratio cases (20, 100, and ∞) considered in this study. CHA_20 and _∞ were comprised of grains similar to the cubic-like CHA particles (Fig. S1a). It appears that these grains were aligned in the diagonal direction along the membrane thickness. As a result, the protruding edges of the cubic-like particles led to rough surfaces in CHA_20 and _∞. On the other hand, the grains on the top surface of CHA_100 were smooth and larger. The thicknesses of CHA_20, _100, and _∞ were estimated to be 2.2, 9.2, and 2 μm, respectively (Fig. 1a2-c2). Further 9 ACS Paragon Plus Environment


measurements provided averaged and standard deviation values for the membrane thickness (Table S1). Given that all-silica CHA seed layer in this study, an optimal Si/Al ratio was required for facile membrane growth, apparently arising from the compatible growth of the all-silica CHA particles with a synthetic precursor with a high Si/Al ratio. Nevertheless, the little amount of Al sources, seemingly leached from the α-Al2O3 disc during secondary growth for CHA_∞, was likely inappropriate for facile growth. 5 μm

5 μm (a2)

CHA film

Normalized intensity

(a3) Si

2.2 μm

Al

α-Al2O3 disc

*

Calcined Si-CHA zeolite (3 11) (104) (003) (102)(20 1) (202) (3 12) (420)

(2 10)

10

15

20

2 θ (˚)

5 μm (b2)

25

30

(b3)

9.2 μm

Si

CHA film

α-Al2O3 disc

Al

Membrane CHA_100

(2 10)

(c2)

(20 1)

*

Calcined Si-CHA zeolite powder

5

10

5 μm

5 μm (c1)

(101)

5

5 μm (b1)

Membrane CHA_20


(a1)

15

20

2 θ (˚)

25

30

(c3)

CHA film

Si

2 μm

α-Al2O3 disc

Al

Membrane CHA_∞


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 10 of 40

*

Calcined Si-CHA zeolite powder

5

10

15

20

2 θ (˚)

25

30

Fig. 1. SEM images in top view (left) and cross-sectional view (middle) along with the corresponding XRD patterns (right) of (a1)-(a3) CHA_20, (b1)-(b3) CHA_100, and (c1)-(c3) CHA_∞. In (a2)-(c2), the EDX results obtained along the yellow lines and the estimated membrane thicknesses are shown. In (a3)-(c3), the XRD pattern of calcined Si-CHA zeolite powder is included as a reference, and asterisks (*) indicate the XRD reflections arising from the α-Al2O3 disc.


Page 11 of 40 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


First, the SEM images of CHA_20 and _100 in Fig. 1a1 and 1c1 suggest a preferred out-of-plane orientation, seemingly because of the protruding, regular grains. The corresponding XRD patterns, however, confirmed a random out-of-plane orientation (Fig. 1a3 and 1c3), indicating the failure of the preferred alignment of the grains. In addition, the XRD pattern of CHA_100 showed that it did not have a pronounced degree of preferred out-of-plane orientation, though the intensities of some XRD peaks, such as those corresponding to the (2 10) and (20 1) planes, were slightly increased relative to those in the calcined SiCHA powder (Fig. 1b3). In particular, the increased intensity of the XRD peak corresponding to the (2 10) plane indicates that the 8 MR channels (as the orthogonality between the 8 MR channels along the [100] axis and the (2 10) plane was confirmed by the formula given in the literature41) were positioned in the out-of-plane direction (that is, perpendicular to the zeolite surface) and the corresponding CO2 permeance would be increased by a shorter passage through the membrane. To the best of our knowledge, achievement of such out-of-plane orientation perpendicular to the (2 10) plane is reported here for the first time, though its degree was not pronounced. Referring to comparable secondary growth conditions but distinct out-ofplane orientation in the literature,42 it is reasonable to conjecture that the h0h-oriented seed layer adopted in this study along with an optimal Si/Al ratio (here, ~100) contributed to achieving the preferred out-ofplane orientation.

3.2. CO2 permselectivities of the CHA membranes 3.2.1. CO2/N2 separation performances under dry conditions

Despite the continuity observed at SEM resolution, the CHA film synthesized with a nominal Si/Al ratio of 20 (CHA_20) showed very poor CO2/N2 separation performance under dry conditions. The maximum (max) CO2/N2 SF was estimated to be 2.2 ± 0.2, which is relatively close to that (~0.8) determined based on Knudsen diffusion (Fig. 2a1). In addition, the poor CO2/N2 separation performance of CHA_20 was similar to that of the CHA membranes reported in our previous study.25 The low max CO2/N2 SF of 11 ACS Paragon Plus Environment


Page 12 of 40

2.2 was much smaller than the expected value of ~10-20 through the CHA membranes,25, 43 indicating the presence of pronounced, unwanted non-zeolitic parts in CHA_20. Compared to CHA_20, CHA_100 and _∞ showed higher max CO2/N2 SFs of ~5.2 ± 0.1 and ~7.3 ± 2.0, respectively (Fig. 2b1-c1). It was noted that the CO2/N2 SFs monotonically decreased with temperature, similar to the trend of the previously reported SSZ-13 membranes with the max CO2/N2 SF being ~10-11 at ~20-25 °C.4, 6 Among the three membranes in this study, CHA_∞ showed the highest max CO2/N2 SF at 30 °C under dry conditions. This suggests that the small amount of Al in the secondary growth solution is beneficial for obtaining a lessdefective membrane structure. The monotonic increase in the separation performance with increase in Si/Al ratio in the synthetic precursor is similar to that observed in a previous study.42 However, in that paper, it was reported that the use of a small amount of Al in the synthetic precursor (for example, Si/Al ratio of 125) led to the formation of undesired AFI zeolite membranes. However, despite the absence of Al, the synthesis of high-performance CHA membranes was still possible (CHA_∞ in Fig. 2c1). To the best of our knowledge, the secondary growth of the CHA seed layer with a nominal Si/Al ratio of infinity in the synthetic precursor is the first example of the fabrication of a high-performance CHA membrane (here, CHA_∞).


Page 13 of 40

Dry condition 10

10

Wet condition

-6

(a1) Membrane CHA_20

100 10

-6

10

-7

CO2

CO2

-7

N2

10

-8

10

-9

10

-6

10

-7

-8

0.5 10

-9

0

50

100

150

CO2/N2 SF

1

CO2/N2 SF through a bare α-Al2O3 disc

Permeance (mol·m-2·s-1·Pa-1)

10

CO2/N2 SF

10

N2

CO2/N2 SF of 10

10

200

1 0

100 10-6

(b1) Membrane CHA_100

50

100

150

200

CO2

10

N2

-7

N2

10 10

-8

0.5 10

-9

-8

10

CO2/N2 SF

CO2/N2 SF

1

1 10

-9

10

-6

0

50

100

150

200

0

100 10-6

(c1) Membrane CHA_∞

50

-7

10

-8

10

-9

100

150

200

CO2

10

N2

-7

N2

10 10

-8

0.5 10

-9

CO2/N2 SF

10

CO2/N2 SF

1 0

50

100

0.5 100

(c2) Membrane CHA_∞

CO2

10

0.5 100

(b2) Membrane CHA_100

CO2

10

100

(a2) Membrane CHA_20

150

200

Temperature (ºC)

CO2/N2 Separation Factor (SF)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


1 0

50

100

150

200

0.5

Temperature (ºC)

Fig. 2. CO2/N2 permeances and the CO2/N2 SFs vs. temperature for (a1)-(a2) CHA_20, (b1)-(b2) CHA_100, and (c1)-(c2) CHA_∞ under dry (left) and wet (right) conditions, respectively. For comparison, a CO2/N2 SF of 10, theoretically estimated as an ideal permeation selectivity through CHA membranes,25 is indicated by the red dashed line in (b2) and (c2). In addition, the CO2/N2 SFs (~0.8), determined under the Knudsen diffusion regime, are indicated by blue dashed lines.

In addition, we noticed that the CO2 permeance through CHA_100 was higher than that through CHA_∞ (Fig. 2b1-c1), even though the corresponding membrane (~9.2 μm) was much thicker than CHA_∞ (~2 μm) (Fig. 1b2-c2). Because the permeance across a membrane is generally reciprocal to the membrane



Page 14 of 40

thickness, such behavior indicates a beneficial contribution of other membrane properties. Among the possible factors, we ascribed the good CO2 permeance to the preferred alignment of the 8 MR channels in CHA_100 (Fig. 1b3). As mentioned above, the XRD peak arising from the (2 10) plane, which is perpendicular to the 8 MR channel in the [100] axis, was pronounced. Furthermore, the intensity of the XRD peak corresponding to the (003) plane, which is perpendicular to the 6 MR pore in the [001] axis, was reduced. This indicates that the 8 MR channel was more aligned along the membrane thickness, while the mass transfer inhibition by the 6 MR pores was significantly weakened. It seems that the preferred alignment of the 8 MR channels helped to increase the CO2 permeance, despite the greater thickness of CHA_100. Nevertheless, at this point, we cannot exclude a possible contribution of molecular transport through non-zeolitic defects to the higher CO2 permeance, and this will be discussed in more detail in Section 3.4.

Previously, Si-CHA zeolite seed particles have been synthesized in the presence of hydrofluoric acid (HF), where F- served as an anionic counterpart with respect to the cationic structure directing agent (SDA, here TMAda+).34 However, the direct adoption of the synthetic protocol for the secondary growth of the CHA seed layer would not work, apparently due to the dissolution of the pre-deposited CHA seed particles by HF. Recently, the use of TMAdaF as a SDA (which can be considered to have the combined role of TMAdaOH and HF) instead of the conventional TMAdaOH SDA has been shown to be effective for the fabrication of CHA membranes via secondary growth.44 Nevertheless, the use of TMAdaOH is highly desirable for manufacturing high-performance CHA membranes, because it does not require an additional, sophisticated procedure to obtain TMAdaF.


Page 15 of 40 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


3.2.2. CO2/N2 separation performances under wet conditions

Considering the third main component, H2O vapor, in the CO2-containing streams, we investigated the CO2 permselectivities of the three CHA membranes under wet conditions. Specifically, CHA_20, which showed poor CO2/N2 separation performance (max SF of ~2.2 at 30 °C) under dry conditions also exhibited a low CO2/N2 separation performance (max SF of ~1.6 at 100 °C) under wet conditions (Fig. 2a2), confirming the pronounced presence of undesired non-zeolitic defects. The increase in the nominal Si/Al ratio of the secondary growth precursor from 20 to 100 led to an improvement in the max CO2/N2 SF from ~5.2 ± 0.1 under dry conditions to ~13.4 ± 1.8 under wet conditions at the same temperature, 30 °C (CHA_100 in Fig. 2b1-b2). This trend was also observed for CHA_∞ (Fig. 2c1-c2). However, a difference was found; although the CO2/N2 SF of CHA_100 monotonically decreased with temperature (Fig. 2b2), that of CHA_∞ showed a max CO2/N2 SFs of ~15 in the temperature range of 50-75 °C, not 30 °C, under wet conditions (Fig. 2c2). Considering the representative temperature (~50-75 °C) of flue gas, the comparable CO2/N2 SFs of ~15 at ~50-75 °C through CHA_∞ is highly desirable for effective membranebased carbon captures. Notably, despite the use of the same seed layer, the appropriate choice of Si/Al ratio in the secondary growth precursor was critical for obtaining high-performance CO2 permselective CHA membranes. Moreover, the feed composition of the simulated post-combustion flue gas (here, at a CO2 partial pressure of ~15 kPa) was considered to investigate the separation performances of CHA_100 and _∞ in the presence of H2O in the feed (Fig. S2). The results are comparable with those observed for equimolar CO2/N2 binary feeds (Fig. 2b2-c2). This result could be ascribed to almost the linear adsorption behavior of both CO2 and N2 molecules inside the CHA zeolites.25

Despite previous results concerning water-assisted CO2 permselectivity, the monotonic decrease in the CO2/N2 SF of the SSZ-13 membrane with increase in temperature is undesirable for practical use.4 In particular, considering the plausible temperature fluctuations of flue gas streams around the representative temperature (~50 °C), the establishment of the max separation performance in the temperature range of 15 ACS Paragon Plus Environment


Page 16 of 40

~50-75 °C is highly desirable. In this respect, CHA_∞ prepared in this study, which had a CO2/N2 SF greater than 10 in the presence of H2O vapor in a temperature range of ~50-75 °C, meets the above-mentioned requirement for practical use.

The improved CO2/N2 separation performance under wet conditions can be ascribed to the hydrophobic surface of the CHA membranes. Because of the smaller size of the H2O molecule (0.265 nm), preventing the entry of H2O into the CHA zeolite pores is not a sound approach for minimizing the H2O permeation rate. Elucidation of coupled features due to the water adsorption/condensation on the defects and membrane surface will be required. Instead, the control of the hydrophobicity of the surface of the CHA membranes is effective for reducing the adsorption of H2O and, thus, its permeation rate. The EDX analysis shown in Fig. 1a2-c2 reveals that there is a siliceous component near the surface in all three CHA membranes. Considering the trend in the distribution of Al atoms along the membrane thickness, it appears that CHA_20 had a higher portion of Al atoms compared to CHA_∞. It was noted that the proportion of Si atoms approached zero after passing the interface in CHA_∞, whereas that in CHA_20 did not reach zero, suggesting the possible existence of CHA zeolites deep inside the α-Al2O3 disc support. In contrast, a higher siliceous portion was observed throughout the membrane thickness for CHA_100. Accordingly, CHA_100 could be considered as the most hydrophobic membrane of the three prepared membranes.

To clarify the hydrophobicities of the CHA membranes, we measured the contact angles of water droplets on the membrane surfaces (Fig. S3). The water droplet on CHA_20 was immediately infiltrated into the membrane sample, apparently into the non-zeolitic defects, thus failing a reliable test. CHA_∞, which clearly contained a lower Al portion than CHA_20 (Fig. 1a2 and 1c2), had a water droplet contact angle of ~25°. This contact angle indicates the lower hydrophobicity of CHA_∞, especially compared to CHA_100 (contact angle of ~80°). Unlike highly defective CHA_20, CHA_100 and _∞ both held the water droplet for up to duration of 5 min (Fig. S3), suggesting narrower or fewer defects in these two membranes. Both the EDX and contact angle measurements led us to the conclusion that the surface of CHA_100 was 16 ACS Paragon Plus Environment

Page 17 of 40 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


most hydrophobic, followed by CHA_∞, while the surface of CHA_20 was not continuous, possibly because of the insufficient intergrowth during secondary growth.

It was noted that the dry CO2/N2 SF of CHA_∞ was higher that of CHA_100 (Fig. 2b1-c1), possibly indicating a lower degree of defects in CHA_∞. Despite the expected lower degree of defects in CHA_∞, its wet CO2/N2 separation performance was inferior to that of CHA_100, which showed higher CO2 permeance and a marked CO2/N2 SF at ~30 °C (Fig. 2b2-c2). This suggests the convoluted roles of the zeolitic and defect parts as well as the membrane surface property. Because CHA_100 was the most hydrophobic membrane (Figs. 1a2 and S3), the degree of CO2 permeance reduction arising from the presence of H2O vapor was much lower than that for CHA_∞ (Fig. 2b1-b2 and 2c1-c2). Despite the lower CO2 permeance, CHA_∞ exhibited significant CO2/N2 SFs (~9.7-15.2) throughout ~30-75 °C (Fig. 2c2). This steady CO2/N2 SF at ~30-75 °C indicates robustness with respect to the plausible temperaturefluctuation of the flue gas or biogas stream of ~50 °C in the presence of H2O vapor. Considering that the final CO2/N2 separation performance of CHA membranes with some defects is a coupled function of (1) the non-zeolitic defects associated with the physisorption of H2O vapor and (2) intrinsic zeolitic parts,4 the CO2/N2 separation performance shown in Fig. 2b2-c2 needs to be related to the defect structure, and this will be addressed and discussed below (Sections 3.3-3.4).

The good separation performance of CHA membranes under wet conditions (Fig. 2c2) is not expected to deteriorate with time at temperatures below 100 °C, mainly because of the robust CHA zeolite structure. To confirm this expectation, the less hydrophobic CHA_∞ was tested first, because this is more likely to show deactivation if any occurs. The response of CHA_∞ with respect to the simulated CO2/N2 feed (a molar composition of 15:85 on a dry basis) at ~50 °C with relative humidities of 0%, 26%, 60%, and 100% was investigated (Fig. 3a). An enhancement in the CO2 permselectivity (equivalent to that observed at 50 °C in Fig. S2b) caused by water vapor was observed immediately after the addition of water vapor (from 0% to 26%) and was maintained. Furthermore, the addition of more water vapor to relative 17 ACS Paragon Plus Environment


humidities of 60% and 100% reduced the permeances of both CO2 and N2 molecules to the same extent so that the corresponding CO2/N2 SF remained constant at about 27. In addition, the rigorous long-term stability test of CHA_∞ in the simulated CO2/N2 feed at 50 °C with a relative humidity of 100% confirmed its robustness (Fig. 3b). Even after its exposure to harsher feed conditions (at 200 °C for 2 d), the CO2/N2 separation performance on reaching the target temperature of 50 °C was fully recovered and maintained for


an additional 3 d (Fig. 3b), revealing the robustness of the CHA zeolite inorganic membranes. 10

-6

10

-7

50 10-6

at 50 °C

(b) CHA_ ∞

at RH ~100%

50

200 °C

RH 0%

CO2/N2 SF RH ~60%

RH ~26%

RH ~100%

10

10

50 °C

-7

50 °C

CO2

10 CO2/N2 SF

10

CO2

-8

10

-8

N2

10

N2

-9

10

10

0

6

12

18

24

30

36

-6

(c) CHA_100

at 50 °C

RH 0%

RH ~60%

RH ~26% CO2

10

-9

0

50 10-6

24

48

72

at RH ~100%

-7

10

-8

10

-7

10

-8

50 °C

CO2

10 N2

CO2/N2 SF

-9

0

6

12

50

200 °C 50 °C

N2

10

1 96 120 144 168 192

(d) CHA_100

RH ~100%

CO2/N2 SF

10

1 42

CO2/N2 SF


(a) CHA_∞

CO2/N2 SF

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 40

18

24

30

36

1 42

10

-9

0

24

48

Measurement time (h)

72

1 96 120 144 168 192

Measurement time (h)

Fig. 3. Permeances of CO2 and N2 through (a) CHA_∞ and (c) CHA_100 for binary mixtures with a respective molar composition of 15:85 at relative humidities of 0%, 26%, 60%, and 100% at 50 °C. Longterm hydrothermal stabilities of (b) CHA_∞ and (d) CHA_100 at 50 °C; (1) initially, a membrane sample was exposed to a relative humidity of 100% at 50 °C for 3 d, then (2) the membrane sample was treated for 3 d at a higher temperature (200 °C) in an effort to expedite any deactivation, and (3) the sample was exposed again to a relative humidity of 100% at 50 °C for 3 d.


Page 19 of 40 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


In addition, the more hydrophobic CHA_100 showed an improved and steady CO2/N2 SF immediately after adding water vapor to the dry feed (Fig. 3c). Desirably, such CO2/N2 SF was well preserved. Further addition of water vapor (relative humidities of 60% and 100%) selectively reduced the permeance of N2 molecules and, accordingly, increased the CO2/N2 SF monotonically. This confirms the desirable hydrophobicity of the zeolite membranes for separation under wet conditions. The long-term stability test of CHA_100 in the simulated CO2/N2 feed at 50 °C with a relative humidity of 100% (Fig. 3d) also confirmed its high suitability for practical use (that is, low CO2 partial pressures and under wet conditions). Apparently, the harsh treatment at 200 °C for 2 d did not damage the membrane performance; the CO2/N2 separation performance before the treatment was recovered well and maintained for an additional 3 d (Fig. 3d). At this point, we would like to mention that the simple tuning of the Si/Al ratio in the synthetic precursor allowed for achieving the appropriate combination of a desirable out-of-plane orientation, high hydrophobicity, and some degree of defects and thus, securing the high-flux and highselectivity in favor of CO2 molecules even under wet conditions.

3.2.3. CO2/CH4 separation performances

In addition to the CO2/N2 separation performance, we also measured the CO2/CH4 separation performances of CHA_100 and _∞ (Fig. 4). As expected from their high CO2/N2 separation performance (Fig. 2), both CHA_100 and _∞ also showed good CO2/CH4 separation performances under dry conditions, having corresponding max CO2/CH4 SFs as high as ~31 and ~61, respectively, at 30 °C (Fig. 4a and 4c). The observation of higher CO2/CH4 SFs than CO2/N2 SFs was expected, because the diffusional behavior inside the microporous channels is a function of molecular size (CH4; 0.38 nm vs. N2; 0.364 nm).45-46 Furthermore, the CO2/CH4 permeation behaviors of both CHA_100 and _∞ under wet conditions (Fig. 4b and 4d) exhibited a similar trend to that of the CO2/N2 separation performance under wet conditions (Fig. 2b2-c2); that is, the CO2 permeance decreased, but the permeance of the larger CH4 molecule was more significantly decreased at 30 °C. It was noted that the permeance of CH4 was much lower than that of N2, 19 ACS Paragon Plus Environment


Page 20 of 40

apparently due to the significant inhibition of the larger CH4 molecule by the physisorbed water molecules. Nevertheless, the wet CO2/CH4 SFs at 30 °C did not increase as much for CHA_100 and decreased for CHA_∞; (1) dry SF of 30.7 vs. wet SF of 37.2 for CHA_100 and (2) dry SF of 60.8 vs. wet SF of 16.5 for CHA_∞. Compared to that in the CO2/N2 binary mixture, the CO2 permeance reduction was more significant in the CO2/CH4 binary mixture, as the presence of the larger CH4 molecule was likely to disfavor the transport of CO2. Nevertheless, it was found that for CHA_100, the reduction in CO2 permeance was not considerable, whereas that for CHA_∞ was significantly decreased. This can be ascribed to the surface hydrophobicity/hydrophilicity, as reflected by the higher and lower contact angles of water droplets of 80° for CHA_100 and 25° for CHA_∞, respectively (Fig. S3). Considering the fact that the temperature of biogas streams is ~40-70 °C and they contain ~3-6% H2O vapor, CHA_100 and _∞, which showed wet CO2/CH4 SFs of 35.6 and 27.4, respectively, at 50 °C, are suitable for the separation of CO2 molecules from biogas streams.


Page 21 of 40 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Fig. 4. Permeance of equi-molar CO2 and CH4 mixtures and their corresponding CO2/CH4 SFs through (a)(b) CHA_100 and (c)-(d) CHA_∞. In addition, the CO2/CH4 SFs (~0.6) determined under the Knudsen diffusion regime are indicated by the blue dashed lines.

At this point, we would like to mention that the monotonic decreasing trend of CO2/N2 and CO2/CH4 SFs with an increase in temperature is not desirable for reliable use. The maximum performance should be achieved at ~50 °C, a representative flue gas and biogas temperature, and be insensitive to small changes in temperature. The low Si/Al ratio in the synthetic precursor could contribute to the formation of more defects in the zeolite membranes (as observed form CHA_20 in this study), possibly because of the relatively high polarity and repulsion of the growing zeolite grains.47 Concerning CO2/N2 and CO2/CH4 separations through SSZ-13 membranes in wet conditions, the permeances of CO2, N2, and CH4 all decreased at temperatures below 100 °C but the corresponding SFs were increased because of the greater inhibition of the slowly permeating, larger N2 and CH4 molecules by H2O.42 The Si/Al ratio in the synthetic precursor significantly affects the hydrophobicity of the final membrane surface and constituents, whereas



Page 22 of 40

defect formation is a sensitive function of thermal processing. As both factors influence the final performance, the identification of defects and the elucidation of their effects on the performance, which will be discussed in Sections 3.3-3.4, are also important for understanding and achieving high CO2 permselectivities in realistic feeds (here, water-vapor-containing CO2 streams).

3.2.4. Evaluation of the separation performance of CHA_100 and _∞

As shown in Fig. 2, there was a trade-off relationship between the separation performances (in terms of CO2 permeance and the CO2/N2 SF) for CHA_100 and _∞. In particular, this trend was pronounced in the CO2 separation performance under wet conditions (Fig. 2b2-c2). Specifically, the CO2 permeance and CO2/N2 SF of CHA_100 around 50 °C (a representative temperature of flue gas and biogas) were higher and lower than those of CHA_∞, respectively (Fig. 2b2-c2). Under this circumstance, the product of CO2 permeance (relevant to the recovery of a target molecule on the permeate side) and CO2/N2 SF (relevant to the purity of the target molecule on the permeate side)48 is a good indicator of the final separation performance. Thus, we plotted the product as a function of temperature for CO2/N2 separation (Fig. 5a). For clarity, a reasonable target temperature range of ~30-75 °C is shaded in Fig. 5a. The product values for both membranes in the dry feed were comparable to each other. However, under wet conditions, the product values for CHA_100 were higher than (at 30-50 °C) or comparable to (at 75 °C) those for CHA_∞ in the shaded area. This strongly indicates that CO2 separation by CHA_100 was more feasible. This trend was similar for CO2/CH4 separations (Fig. 5b), where CHA_100 also showed better (at 30-50 °C) or comparable (at 75 °C) overall separation performance in the wet CO2/CH4 feed between 30 and 75 °C (also shaded in Fig. 5b). In addition, the noticeable increasing and decreasing trend of the product value for hydrophilic CHA_∞ can be explained by the significant reduction of CO2 permeance at a lower temperature by adsorbed water molecules (~3 kPa vapor pressure in the feed) and the gradual recovery of CO2 permeance back to the value observed under dry conditions with increasing temperature.


Page 23 of 40

2.0

2.0

10

10

1.5

(a) CO2/N2 separation performance

10

-4

1.5

(b) CO2/CH4 separation performance B

2.0

-5

0.5

1.5

Flue gas temperature range

Filled: DRY Open: WET

0.0

1.0

0.0

0.5

1.0

1.5

B C

10

1.5

0.0

0.5

50

1.0

A

100

1.5

150

2.0

200

10

0.5

1.0

1.5

B C

2.0

A 0.5

0.0

0.0

0

1.0

-6

0.0

-7

Filled: DRY Open: WET

0.5

0.0

10

0.5

C Membrane CHA_100 Membrane CHA_∞

1.0 2.0

-5

2.0

A

-6

B

B

B

C Membrane CHA_100 Membrane CHA_∞

1.0

B

10

-4

B

10

CO2 permeance × SF (mol·m-2·s-1·Pa-1)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


0.0

-7

Biogas temperature range 0.5

1.0

1.5

2.0

A

0

Temperature (ºC)

50

100

150

200

Temperature (ºC)

Fig. 5. (a) CO2 permeance × CO2/N2 SF and (b) CO2 permeance × CO2/CH4 SF as a function of temperature for CHA_100 and _∞. For comparison, the performances measured under dry (filled symbol) and wet (open symbol) conditions are included. For clarity, the representative temperature range of flue gas and biogas is shaded.

3.3. Investigation of the defects in the CHA membranes through FCOM

In addition to the measurements of CO2 permselectivities of CHA_20, _100, and _∞, the defect structures were imaged via FCOM39,

49-50

to understand the relationship between this structure and the

separation performance. FCOM analysis is an attractive technique for studying the structure of defects embedded in a membrane without damaging the membrane structure. Fig. 6 reveals that despite the significant discrepancy in the separation performance of CHA_20 and _∞ (Fig. 2a1-a2 and 2c1-c2), the cracks observed in the FCOM images were similar. Specifically, the cross-sectional view FCOM images show that the cracks, which constituted a defect network, as shown in the top-view FCOM images, propagated all the way down to the interface between the CHA membrane and the α-Al2O3 disc. Although it is challenging to determine the contribution of the cracks to the permeation rates in both CHA_20 and _∞ because of the limited optical resolution, the permeation results in both dry and wet conditions shown in Fig. 2a1-a2 and 2c1-c2 confirm that CHA_20 was more defective than CHA_∞. Thus, the sizes or widths



Page 24 of 40

of the cracks in CHA_20 were apparently larger than those in CHA_∞. Despite the above-mentioned limited resolution, the FCOM images of the thicker CHA_100 membrane indicate that although the cracks also penetrated to the interface, the number of cracks was much larger than those in the other two membranes.

Fig. 6. FCOM cross-sectional and top view images obtained for (a1)-(a3) CHA_20 (24 h), (b1)-(b3) CHA_100 (96 h), and (c1)-(c3) CHA_∞ (24 h); the numbers inside the parenthesis indicate the dyeing duration. The top view images in (a2)-(c2) and (a3)-(c3) were taken near the CHA membrane surface and the interface between the CHA membrane and α-Al2O3 disc, respectively, as indicated by white horizontal white lines in the corresponding cross-sectional images in (a1)-(c1).


Page 25 of 40 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Considering that the dyeing process is kinetically relevant, we systematically varied the dyeing duration from 12 h through 24 h to 96 h (Figs. S4-S5). This approach is beneficial for obtaining a fundamental understanding of the defect structure. Figs. S4-S5 reveal that the degree of crack dyeing in CHA_20 and _∞ with respect to the dyeing duration was similar; the cracks, which propagated from the membrane surface to the interface, were accessed by the dye molecules after 12 h (Figs. S4a2-S5a2 and S4c2-S5c2) and saturated after 24 h (Figs. S4a3-S5a3 and S4c3-S5c3). However, dyeing for 96 h resulted in a marked difference; while it was almost impossible to distinguish the cracks in CHA_20 because of the extremely high concentration of dye molecules (Figs. S4a4 and S5a4), additional cracks were observed for CHA_∞, but these ended in the middle section (indicated by the yellow arrows in Figs. S4c4 and S5c4). For better comparison, additional FCOM images are shown in Figs. S6-S7. Compared to those of CHA_20 and _∞, the cracks in CHA_100 were slowly dyed with an increase in dyeing duration (Fig. S4b2-b4 and S5b2-b4). Unless the thickness of CHA_100 is similar to those of CHA_20 and _∞, a fair evaluation of the size and number of defects is challenging. Nevertheless, it is reasonable to conjecture that the size of the defects in CHA_100 was comparable to or smaller than that of CHA_∞, whereas the number of defects or defect density of CHA_100 was higher than that of CHA_∞.

Although the FCOM images provide detailed information about the defect structure, the dyeing process required for FCOM characterizations also can reveal the degree of undesired defects. The inset in Fig. S4 shows that the initially transparent water on the permeate side of CHA_20 became colored after 12 h and the concentration of dye on the permeate side gradually increased with time up to 96 h. This was interpreted as a result of the steady passage of dye molecules along the defective cracks. Thus, the size and/or number of defects in CHA_20 was significant compared to those of the other membranes (CHA_100 and _∞).

Combining the FCOM characterization (defect structure inside the CHA membranes) with the water contact angle results (hydrophilicity of the surface on the CHA membranes), a simple, rapid, but 25 ACS Paragon Plus Environment


Page 26 of 40

reliable evaluation tool that can determine the success of a membrane fabrication was developed. Specifically, a dye solution, which was originally allowed to diffuse into the defect structure in the CHA membranes for FCOM characterizations, was dropped onto the membrane surface and its diffusion into the membrane was tracked with time (Fig. S8). Technically, this tool can be considered as a time-resolved visualization of the water contact angle measurement. Considering that optical characterization is not suitable for the selective characterization of a zeolite membrane with thicknesses on the order of 100 nm, the simple observation of the diffusional behavior of the dropped dye molecules can provide a useful insight into the formation of undesired structures. First, this simple test (Fig. S8a) reveals that all the dye molecules on CHA_20 penetrated the defects, immediately after dropping. This result was consistent with the FCOM images (Figs. S4-S5) and water contact angle results (Fig. S3a). These characterizations indicate that CHA_20 was the most defective membrane. Secondly, the surface of CHA_100 was more hydrophobic than that of CHA_∞ (Fig. S8b-c), as already indicated by their respective water contact angles (CHA_100: 80° vs. CHA_∞: 25° as shown in Fig. S3). However, this test does not allow for the differentiation of differences in the defect structures, in contrast to FCOM results. In short, this simple test was, to the best of our knowledge, proposed for the first time in this study and can be used as a quick screening process to aid in the exclusion of highly defective membranes.

3.4. Elucidation of the CO2 separation performance based on quantitative membrane properties

To date, the qualitative analysis of defects in zeolite membranes based on the FCOM technique has mainly focused on identifying the defect density and distribution and correlating these factors with the final separation performance.2, 49, 51 However, quantitative information about the size and number of defects and an association with the final membrane performance would be useful.50 Thus, the quantitative analysis of defects in CHA_20 and _∞ was conducted based on an approach reported in our previous study;40 the resulting processed defects are schematically illustrated in Fig. S9. However, in the case of CHA_100, the differentiation of bright and dark spots in the FCOM image was not possible because of the presence of 26 ACS Paragon Plus Environment

Page 27 of 40

broad, extremely bright spots (Fig. 6b1-b3). The quantitative properties extracted from the FCOM images were complemented with a 1-D permeation model, which was used to estimate the representative properties (tortuosity and porosity) of the defects. The relevant properties are summarized in Table S2.

3.5 nm

2.1 nm

16

~0 nm

14

Porosity (%)

100

Defect healing Defect

0.13% 0.24%

Zeolite 46%

100%

≈

CHA_20

≈

8 4

99.76%

99 0

10 6

19% 99.87%

54%

12

81%

CHA_∞

≈

100%

Post-treated SSZ-13 Membrane

2 0

CO2 molar flux (mmol∙m-2∙s-1)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Fig. 7. Porosity of non-zeolitic parts (that is, defects) and zeolitic parts, and their corresponding contribution to the total CO2 molar fluxes at 30 °C for CHA_20 and _∞, as well as those of the post-treated and thus, plausibly defect-free SSZ-13 membrane.40 Left and right columns for each membrane refer to the porosity and CO2 molar flux, respectively. Columns with and without patterns indicate the portions of each value originating from the defect and zeolite, respectively. Above each column, the estimated defect sizes are shown.

Fig. 7 graphically summarizes the defect size and porosity and the contribution of the defects to the final molar flux of the faster permeating component (i.e., CO2). Specifically, the defect sizes for CHA_20 and _∞ were estimated to be 3.5 and 2.1 nm, respectively, and the corresponding porosities were 0.24% and 0.13%, respectively. As mentioned above, it appears that the distributions of defects in CHA_20 and _∞ were comparable to each other (Fig. 6a1-a3 and 6c1-c3) and, accordingly, their pixel-based area



Page 28 of 40

fractions were similar (Table S2). However, a rigorous analysis based on a combination of image processing and permeation modeling reveals that the defect size (~3.5 nm) in CHA_20 was ~1.7 times of that (~2.1 nm) in CHA_∞. In turn, this fact confirms the lower CO2 permselectivity of CHA_20 under dry conditions (Fig. 2a1) than that of CHA_∞ (Fig. 2c1). Furthermore, the CO2 molar fluxes passing through the defects in CHA_20 and _∞ at 30 °C under dry conditions accounted for 46% and 19% of the total CO2 molar flux through the membranes, respectively (Fig. 7). Surprisingly, it was noted that despite the porosity of 0.24%, the molar flux of CO2 through the defects in CHA_20 was pronounced, indicating the importance of synthetic (here, nominal Si/Al ratio) or post-treatment methods for the effective reduction of the defects in CHA membranes. Indeed, the reduction in porosity from 0.24% to 0.13% could decrease the molar flux through the defects from 46% to 19% (Fig. 7). This is consistent with the previous conclusion that a very low number of defects (less than 1%) significantly degrade the zeolite membrane performance.52-53 In fact, the difference in the defect sizes (3.5 nm for CHA_20 vs. 2.1 nm for CHA_∞) was key to the substantial change in their CO2 permselectivities (Fig. 2a1 and 2c1).

As already inferred from the CO2 separation performance in Fig. 2b1-c1, the more defective CHA_100 had a lower max CO2/N2 SF (5.2) under dry conditions (for convenience, referred to as dry max CO2/N2 SF) than that (7.3) of CHA_∞. Thus, the visualized defect structure along with the hydrophobicity (as indicated by water droplet contact angle (Fig. S3b-c) and EDX results (Fig. 1b2-c2)) could account for the permeation behavior (Fig. 2b2-c2) of both membranes under wet conditions. The reduction in CO2 permeance under wet conditions through CHA_∞ (less hydrophobic) seemingly resulted from the combined adsorption of water vapor in the defects and on the membrane surface. This preserved the high molecular sieving ability for CO2 (Fig. 2c2). In contrast, CHA_100, which was more hydrophobic and had a larger number of defects, provided high CO2 permeance because of the high hydrophobicity. Although some water vapor was adsorbed on the defects, the CO2 molecules could pass through the preferentially out-of-plane oriented CHA zeolite channels.


Page 29 of 40 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


Table 1. Summary of the physical properties and CO2/N2 permeation results at 30 °C of CHA_100 and _∞ along with those of another SSZ-13 membrane reported in the literature.40 Crack density

Dry CO2 permean ce × 107 (mol∙m2∙s-1∙Pa-1)

Dry CO2/ N2 SF

Wet CO2 permeanc e × 107 (mol∙m2∙s-1∙Pa-1)

Wet CO2/ N2 SF

Wet/Dr y CO2 permea nce (%)

100Wet/Dr y CO2 permea nce (%)

Wet/D ry CO2/N 2 SF

Ref.

3.1

Middle

1.4

4.6

0.37

10

26

74

2.2

40

80

N/A

Relative ly High

2.3

5.2

1.4

13.4

61

39

2.6

25

2.1

Middle

1.5

7.3

0.37

9.7

25

75

1.3

Membran e sample

Thick ness (μm)

Water contac t angle (°)

Defe ct size (nm)

SSZ-13

~3.5

78

CHA_100

~9.3a

CHA_∞

~2.3a

a

This study This study

Thicknesses of CHA_100 and _∞ were adopted from the averaged values given in Table S1.

Compared to that under dry conditions, the membrane separation performance under wet conditions is a strongly couple function of the zeolite structure, surface hydrophobicity/hydrophilicity, and size/number of defects. Thus, we have summarized all possible membrane properties relevant to the permeation behavior of CHA_100 and _∞ and their CO2/N2 permeation results at 30 °C in Table 1, where those of another SSZ-13 membrane (for convenience, referred to as SSZ-13 in this section)40 are also given for comparison. First, we considered the effects of defects (defect (crack) porosity and defect size) on the CO2/N2 SFs of CHA_∞ and SSZ-13. The porosity difference between the two membranes (0.13% for CHA_∞ and 0.27% for SSZ-13 in Table S2) has a significant effect on their respective dry max CO2/N2 SFs of 7.3 and 4.6. Such an observation is consistent with the relationship between the porosity and dry max CO2/N2 SF for CHA_20 and _∞. More specifically, given the similar area fraction of CHA_20 and _∞ (Table S2), the crack size was critical in determining the final separation performance. Thus, the crack size of CHA_100, which could not be determined from its FCOM images, will be less than ~3.1 nm to achieve the dry max CO2/N2 SF of 5.2, because the corresponding crack density was higher than that of CHA_∞ and SSZ-13.

For a fair comparison, we focused on the permeation behavior caused by the change from dry to wet conditions. Specifically, the CO2 permeance (wet CO2 permeance) and CO2/N2 SF 30 °C (wet CO2/N2 SF) under wet conditions were compared with those (dry CO2 permeance and dry CO2/N2 SF) under dry 29 ACS Paragon Plus Environment


Page 30 of 40

conditions (Table 1); the respective ratios are referred to as wet/dry CO2 permeance and wet/dry CO2/N2 SF, respectively. For CHA_∞ and SSZ-13, we can reasonably speculate that water vapor was adsorbed in the defects and further adsorbed on the membrane surface of the less hydrophobic CHA_∞. Assuming that all the defects were occupied and blocked by water vapor, for CHA_∞, the reduction in CO2 permeation (i.e., wet/dry CO2 permeance) was caused by defect blocking (~19%; inferred from the result in Fig. 7) and surface inhibition (~56%). In a similar manner, for SSZ-13, the contributions to the reduction in CO2 permeance arising from defect blocking and surface inhibition were ~51%40 and ~23%, respectively. This degree of surface inhibition is consistent with the membrane surface hydrophobicity/hydrophilicity (contact angle of water droplets of ~25° for CHA_∞ and ~78° for SSZ-13). Reasonably, the reduction due to surface inhibition in CHA_100 (contact angle of water droplets of ~80°) will be comparable to or slightly less than ~23%, whereas that due to defect blocking will be greater than ~19%. Considering the total reduction of ~39% for CHA_100 (Table 1), the degrees of reduction due to defect blocking and surface inhibition were estimated to be 20% and 19%, respectively. It was noted that the less degree of surface inhibition due to the physisorbed water molecules was observed for hydrophobic SSZ-13 (~23%) and CHA_100 (~19%) against less hydrophobic CHA_∞ (~56%). The relative portions of CO2 permeances due to defect blocking and surface inhibition along with those of wet CO2 permeances are displayed in Fig. 8.


3.0x10-7 Defect blocking Surface inhibition Wet permeance

2.5x10-7 2.0x10-7 1.5x10-7

20%

Dry permeance

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


CO2 permeance (mol·m-2·s-1·Pa-1)

Page 31 of 40

19%

1.0x10-7 5.0x10-8

61%

19% 51% 56% 23% 25%

0.0 CHA_100

CHA_∞

26%

SSZ-13 Membrane

Fig. 8. Estimated CO2 permeances as a result of defect blocking (vertical pattern) and surface inhibition (upward diagonal pattern), when dry feeds were changed from wet feeds at 30 °C. The analysis results for three type membranes (CHA_100 and _∞ in this study and a SSZ membrane in the literature40) were displayed. For analysis, dry CO2 permeance was assumed to be equal to the sum of (1) wet CO2 permeance and (2) reduced CO2 permeances due to defect blocking and surface inhibition, which were both affected by physisorbed water molecules. In addition, the permeation data of CO2 molecules were obtained through the respective membranes at 30 °C with respect to the dry feed (50.5 kPa CO2/50.5 kPa N2) and wet feed (49.5 kPa CO2/49.5 kPa N2/3 kPa H2O vapor). Specifically, in the cases of CHA_100 and _∞, the permeation data of CO2 molecules are identical to those at 30 °C in Fig. 2. Subsequently, we turned to the values of the wet/dry CO2/N2 SF; 2.2, 2.6, and 1.3 for SSZ-13, CHA_100, and CHA_∞, respectively. It appears that the water-blocking defects in SSZ-13 and CHA_100 contributed to increasing their molecular sieving abilities, as indicated by the higher wet/dry CO2/N2 SFs of 2.2 and 2.6, respectively (Table 1). However, CHA_∞, which had the lowest degree of defects as indicated by the highest dry max CO2/N2 SF of 7.3, was unlikely to gain marked benefit from the blocking of defects with water molecules, because the defect effects were much less significant than those of the other two membranes. Instead, a significant loss in CO2 permeance without any improvement of CO2/N2



Page 32 of 40

SFs occurred for less hydrophobic CHA_∞ by surface inhibition (up to ~56%) because of the physisorbed water molecules.

We also investigated the absolute values of the CO2 permeances obtained through the zeolitic and non-zeolitic parts (Fig. 8). First, the relative portion of molecular transport through the cracks in CHA_100 (~20%) was comparable to that in CHA_∞ (~19%). However, the corresponding CO2 permeance (~4.6 × 10-8 mol·m-2·s-1·Pa-1) was almost twice of that (~2.8 × 10-8 mol·m-2·s-1·Pa-1) in CHA_∞ (Fig. 8). Then, the water-blocking of higher density defects in CHA_100 (Fig. 6b) could effectively help to achieve a higher molecular sieving in favor of CO2 under wet conditions (Fig. 2b2-c2). Second, in less hydrophobic CHA_∞, the CO2 permeance (~8.2 × 10-8 mol·m-2·s-1·Pa-1) reduced by the surface inhibition was much higher those (~3.3-4.4× 10-8 mol·m-2·s-1·Pa-1) for CHA_100 and SSZ-13 (Fig. 8). Finally, we would like to estimate the CO2 permeance solely due to the zeolite parts for both CHA_100 and _∞. For this, the CO2 permeance due to the defect-blocking was subtracted from the dry CO2 permeance. Specifically, it was estimated to be as high as 1.8 × 10-7 mol·m-2·s-1·Pa-1 for CHA_100 and this value was higher than the expected value of 1.2 × 10-7 mol·m-2·s-1·Pa-1 in CHA_∞. This strongly indicates a beneficial role of the aforementioned preferred out-of-plane orientation in CHA_100. As the out-of-plane orientation, though not pronounced, in CHA_100 clearly reveals the benefit of pore/channel alignments for securing high molar flux, the control of out-ofplane orientation in CHA membranes is an important task in the future.

3.5. Comparison of the separation performance of CHA_100 and _∞ with the literature data

Fig. 9a-b shows a summary of the CO2/N2 and CO2/CH4 separation performance of CHA_100 and _∞, as well as those of other zeolite membranes. Only the CO2 separation performances measured at ~4050 °C under dry and wet conditions are collected in Fig. 9. As already shown in Fig. 2c1-c2, the CO2/N2 SFs of CHA_∞ for the wet equi-molar CO2/N2 mixture were increased compared to those under dry conditions. This trend had been also found for other hydrophobic zeolite membranes (DDR8 and SSZ-134, 32 ACS Paragon Plus Environment

Page 33 of 40 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


40).

However, hydrophilic NaY membranes (Si/Al ratio of ~1.5-3) showed the opposite trend; although

these membranes showed a marked CO2/N2 separation performance under dry conditions, they lost this ability completely under wet conditions (water vapor pressure of ~2.6 kPa).21 This indicates the importance of hydrophobicity in membranes for securing high performance under wet conditions, as addressed and emphasized before.7 In addition, CHA_∞ was not significantly hydrophobic (as reflected by the lower water droplet contact angle; 25° vs. 80° for CHA_100, as shown in Fig. S3). Accordingly, the CO2 permeance for CHA_∞ decreased more under wet conditions. Despite the much lowered CO2 permeance in CHA_∞, some defects, where water vapor was apparently condensed, were beneficial for securing a high CO2/N2 SF under wet conditions.4 Nevertheless, the optimal defect size should be satisfied, as the wider defects (here, ~3.5 nm for CHA_20, as shown in Fig. 7) negated the separation performance, regardless of the presence of water vapor in the feed (Fig. 2a1-a2).

Fig. 9. (a) CO2/N2 SF vs. CO2 permeance and (b) CO2/CH4 SF vs. CO2 permeance through CHA_100 and _∞, as well as those of other zeolite membranes; in (a), DDR,8 CVD-treated CHA,25 NaY,21 SSZ-134, 40 and in (b), DDR8 and SSZ-134, 40 zeolites. Black and red symbols represent the CO2 separation performance under dry and wet conditions, respectively. For eye guidance, the CO2 permeance through the bare α-Al2O3 disc (with a pore size of 150 nm) is indicated by blue dashed lines, while the corresponding CO2 permselectivity is pointed by black arrows. In addition, the CO2/N2 SF of 10 and CO2/CH4 SF of 50 are included in the form of blue dashed lines.



Page 34 of 40

The same trend as that observed for permeation for CHA_100 and _∞ in dry and wet CO2/N2 feeds (Fig. 9a) was also observed for CO2/CH4 separations (Fig. 9b), with a minor difference being the comparable CO2/CH4 SFs under dry and wet conditions. This trend can be explained by the good dry CO2/CH4 performance through the 8 MR zeolite membranes. Although a negative effect of non-zeolitic defects on the final performance could not be excluded completely, the defects present in CHA_100 and _∞ (as shown in Fig. 6) did not provide a non-selective pathway for the larger CH4 molecules under dry conditions. Under wet conditions, water molecules physisorbed in the defects would not contribute to a significant reduction in the CH4 permeance; instead, the transport of the faster permeating species, CO2, was inhibited. This reduction in the CO2 permeance was still much higher for the less hydrophobic CHA_∞. Considering the fact that a more hydrophobic DDR membrane (water contact angle of 95°)8 showed a similar pattern to that of CHA_∞ (Fig. 9b), it appears that an appropriate combination of hydrophobicity and defect density/size is required to achieve high CO2/CH4 separation performance under wet conditions. Nevertheless, the minimization of defects is always desirable for securing high performance under dry conditions.40 In particular, CHA_100 showed the highest molar flux of CO2 in both CO2/N2 and CO2/CH4 binary mixtures, having the comparable CO2 permselectivities under wet conditions (Fig. 9). Specifically, the CO2 permeances at 50 °C under wet conditions were as high as ~20-25% of the CO2 permeance through bare supports with ~150 nm pore size. For better comparison, we also added the CO2 permselectivities of MFI type zeolite54-56 and metal organic framework57-61 membranes to Fig. 9 and the resulting graphs are displayed in Figs. S10 and S11, respectively, indicating the superior performances of the CHA membranes reported here. Finally, we would like to mention that although the difference in the kinetic diameters of CH4 (0.38 nm) and N2 (0.364 nm) was extremely small, the resulting performance in favor of CO2 molecules was quite distinct under both dry and wet conditions.


Page 35 of 40 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


4. Conclusions

First, a uniform CHA seed layer was prepared via the physical attachment of plate-like Si-CHA seed particles to an α-Al2O3 disc. The limited accessibility for seed particles to reach the surface of the αAl2O3 disc was key to obtaining a uniform seed layer. Despite being the minor portion, the plate-like particles dominated the seed layer. The subsequent secondary growth of the seed layer with a synthetic precursor that had nominal Si/Al ratios of 100 and infinity allowed for the formation of continuous CHA membranes (denoted CHA_100 and _∞, respectively). It appears that a higher, but optimal Si/Al ratio was required for facile growth, as CHA_100 had the largest membrane thickness (~9.3 µm vs. ~2 µm in both CHA_20 and _∞).

The resulting CHA membranes showed modest CO2 SFs (5.2 for CHA_100 and 7.3 for CHA_∞ at 30 °C) when using dry CO2/N2 binary mixtures. However, they showed improved CO2/N2 SFs as high as 13.4 for CHA_100 and 9.7 for CHA_∞ at 30 °C, when using a wet feed. Such improvement was ascribed to the defect blocking by physisorbed water molecules and pronounced for more defective CHA_100. The trend observed for CO2/N2 separation was similar to that for CO2/CH4 separation, but the dry and wet CO2/CH4 SFs were comparable to each other. Considering the CO2 molar flux (relevant to CO2 recovery) and CO2/N2 SF (relevant to CO2 purity) in a coupled way, CHA_100 showed a marked activity for CO2 capture in the temperature range of 30-75 °C (representative of flue gas temperatures). In addition, the degree of CO2 permeance reduction (75%) due to the presence of water vapor in the feed was much larger for less hydrophobic CHA_∞ (water contact angle: 25°) than that (39%) observed for CHA_100 (water contact angle: 80°), supporting the desirable role of hydrophobic membrane constituents. In particular, the CO2 molar fluxes through zeolitic parts and non-zeolitic parts could be effectively deciphered by complementing the information extracted from the image processing of FCOM images with a one dimensional permeation model. From this, we could conclude that increasing hydrophobicity was desirable



Page 36 of 40

for securing high CO2 permselectivities under a water vapor-containing realistic feed condition, where defects with proper size/density could be well blocked by physisorbed water molecules.

Acknowledgements

This work was financially supported by the Korea CCS R&D Center (KCRC) (2014M1A8A1049309) and by the Super Ultra Low Energy and Emission Vehicle (SULEEV) Center (2016R1A5A1009592) through National Research Foundation (NRF) funded by the Korea government (Ministry of Science and ICT). In addition, this work was supported by the Korea-Germany joint program through NRF in the MSIT of Korea (2016K1A3A1A48954031) and in the German Federal Ministry of Education and Research (01DR17015A). This research was also supported by Korea University Future Research Grant. The FCOM characterizations were carried out at KIST and some SEM measurements were conducted at KBSI (Seoul Center).

Supporting Information

Table of estimated thickness of membrane layers, table of tortuosity, area fraction, and porosity of defects, SEM images of Si-CHA particles and a Si-CHA seed layer and their XRD patterns, CO2/N2 permeances and CO2/N2 SFs versus temperature with respect to 15:85 CO2/N2 mixtures in wet and dry conditions, images of water droplets applied to membrane surfaces, cross-sectional view and top view FCOM images of the membranes, images showing the spread of dye droplets on membranes, and schematic tilted and top view images of the defect structure in the membranes.


Page 37 of 40 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


References 1. Jeon, M. Y.; Kim, D.; Kumar, P.; Lee, P. S.; Rangnekar, N.; Bai, P.; Shete, M.; Elyassi, B.; Lee, H. S.; Narasimharao, K.; Basahel, S. N.; Al-Thabaiti, S.; Xu, W. Q.; Cho, H. J.; Fetisov, E. O.; Thyagarajan, R.; DeJaco, R. F.; Fan, W.; Mkhoyan, K. A.; Siepmann, J. I.; Tsapatsis, M., Ultra-Selective High-Flux Membranes from Directly Synthesized Zeolite Nanosheets. Nature 2017, 543, 690-691. 2. Tung, C. T. P.; Kim, H. S.; Yoon, K. B., Growth of Uniformly Oriented Silica MFI and BEA Zeolite Films on Substrates. Science 2011, 334, 1533-1538. 3. Breck, D. W., Zeolite Molecular Sieves: Structure, Chemistry, and Use. John Wiley & Sons, Inc.: 1974. 4. Kosinov, N.; Auffret, C.; Guecueyener, C.; Szyja, B. M.; Gascon, J.; Kapteijn, F.; Hensen, E. J. M., High Flux High-Silica SSZ-13 Membrane for CO2 Separation. Journal of Materials Chemistry A: Materials for Energy and Sustainability 2014, 2, 13083-13092. 5. Kim, E.; Cai, W.; Baik, H.; Choi, J., Uniform Si-CHA Zeolite Layers Formed by a Selective Sonication-Assisted Deposition Method. Angew. Chem., Int. Ed. 2013, 52, 5280-5284. 6. Kalipcilar, H.; Bowen, T. C.; Noble, R. D.; Falconer, J. L., Synthesis and Separation Performance of SSZ-13 Zeolite Membranes on Tubular Supports. Chem. Mater. 2002, 14, 3458-3464. 7. Himeno, S.; Tomita, T.; Suzuki, K.; Nakayama, K.; Yajima, K.; Yoshida, S., Synthesis and Permeation Properties of a DDR-Type Zeolite Membrane for Separation of CO2/CH4 Gaseous Mixtures. Ind. Eng. Chem. Res. 2007, 46, 6989-6997. 8. Kim, E.; Hong, S.; Jang, E.; Lee, J. H.; Kim, J. C.; Choi, N.; Cho, C. H.; Nam, J.; Kwak, S. K.; Yip, A. C. K.; Choi, J., An Oriented, Siliceous Deca-Dodecasil 3R (DDR) Zeolite Film for Effective Carbon Capture: Insight into Its Hydrophobic Effect. Journal of Materials Chemistry A: Materials for Energy and Sustainability 2017, 5, 11246-11254. 9. Yu, L.; Holmgren, A.; Zhou, M.; Hedlund, J., Highly Permeable CHA Membranes Prepared by Fluoride Synthesis for Efficient CO2/CH4 Separation. Journal of Materials Chemistry A: Materials for Energy and Sustainability 2018, 6, 6847-6853. 10. Venna, S. R.; Carreon, M. A., Amino-Functionalized SAPO-34 Membranes for CO2/CH4 and CO2/N2 Separation. Langmuir 2011, 27, 2888-2894. 11. Tian, Y. Y.; Fan, L. L.; Wang, Z. Y.; Qiu, S. L.; Zhu, G. S., Synthesis of a SAPO-34 Membrane on Macroporous Supports for High Permeance Separation of a CO2/CH4 Mixture. J. Mater. Chem. 2009, 19, 7698-7703. 12. Li, S. G.; Fan, C. Q., High-Flux SAPO-34 Membrane for CO2/N2 Separation. Ind. Eng. Chem. Res. 2010, 49, 4399-4404. 13. Li, S. G.; Falconer, J. L.; Noble, R. D., Improved SAPO-34 Membranes for CO2/CH4 Separations. Adv. Mater. 2006, 18, 2601-2603. 14. D'Alessandro, D. M.; Smit, B.; Long, J. R., Carbon Dioxide Capture: Prospects for New Materials. Angew. Chem., Int. Ed. 2010, 49, 6058-6082. 15. Merkel, T. C.; Lin, H. Q.; Wei, X. T.; Baker, R., Power Plant Post-Combustion Carbon Dioxide Capture: An Opportunity for Membranes. J. Membr. Sci. 2010, 359, 126-139. 16. Singh, D.; Croiset, E.; Douglas, P. L.; Douglas, M. A., Techno-Economic Study of CO2 Capture from an Existing Coal-Fired Power Plant: MEA Scrubbing vs. O2/CO2 Recycle Combustion. Energy Conv. Manag. 2003, 44, 3073-3091. 17. Senior, C. L.; Sarofim, A. F.; Zeng, T. F.; Helble, J. J.; Mamani-Paco, R., Gas-Phase Transformations of Mercury in Coal-Fired Power Plants. Fuel Process. Technol. 2000, 63, 197-213. 18. Sawamura, K.; Furuhata, T.; Sekine, Y.; Kikuchi, E.; Subramanian, B.; Matsukata, M., Zeolite Membrane for Dehydration of Isopropylalcohol-Water Mixture by Vapor Permeation. ACS Appl. Mater. Interfaces 2015, 7, 13728-13730. 19. Kusakabe, K.; Kuroda, T.; Morooka, S., Separation of Carbon Dioxide from Nitrogen using IonExchanged Faujasite-Type Zeolite Membranes Formed on Porous Support Tubes. J. Membr. Sci. 1998, 148, 13-23. 37 ACS Paragon Plus Environment


Page 38 of 40

20. Hasegawa, Y.; Tanaka, T.; Watanabe, K.; Jeong, B. H.; Kusakabe, K.; Morooka, S., Separation of CO2-CH4 and CO2-N2 Systems Using Ion-Exchanged FAU-Type Zeolite Membranes with Different Si/Al Ratios. Korean J. Chem. Eng. 2002, 19, 309-313. 21. Gu, X. H.; Dong, J. H.; Nenoff, T. M., Synthesis of Defect-Free FAU-Type Zeolite Bembranes and Separation for Dry and Moist CO2/N2 Mixtures. Ind. Eng. Chem. Res. 2005, 44, 937-944. 22. Wang, Y.; Levan, M. D., Adsorption Equilibrium of Binary Mixtures of Carbon Dioxide and Water Vapor on Zeolites 5A and 13X. J. Chem. Eng. Data 2010, 55, 3189-3195. 23. Wang, Y.; Levan, M. D., Adsorption Equilibrium of Carbon Dioxide and Water Vapor on Zeolites 5A and 13X and Silica Gel: Pure Components. J. Chem. Eng. Data 2009, 54, 2839-2844. 24. Zhu, Z. G.; Xu, H.; Jiang, J. G.; Wu, H. H.; Wu, P., Hydrophobic Nanosized All-Silica Beta Zeolite: Efficient Synthesis and Adsorption Application. ACS Appl. Mater. Interfaces 2017, 9, 27273-27283. 25. Kim, E.; Lee, T.; Kim, H.; Jung, W. J.; Han, D. Y.; Baik, H.; Choi, N.; Choi, J., Chemical Vapor Deposition on Chabazite (CHA) Zeolite Membranes for Effective Post-Combustion CO2 Capture. Environ. Sci. Technol. 2014, 48, 14828-14836. 26. Zhao, L.; Riensche, E.; Menzer, R.; Blum, L.; Stolten, D., A Parametric Study of CO2/N2 Gas Separation Membrane Processes for Post-Combustion Capture. J. Membr. Sci. 2008, 325, 284-294. 27. Deng, L. Y.; Kim, T. J.; Hagg, M. B., Facilitated Transport of CO2 in Novel PVAm/PVA Blend Membrane. J. Membr. Sci. 2009, 340, 154-163. 28. Low, B. T.; Zhao, L.; Merkel, T. C.; Weber, M.; Stolten, D., A Parametric Study of the Impact of Membrane Materials and Process Operating Conditions on Carbon Capture from Humidified Flue Gas. J. Membr. Sci. 2013, 431, 139-155. 29. Qiao, Z. H.; Wang, Z.; Zhang, C. X.; Yuan, S. J.; Zhu, Y. Q.; Wang, J. X.; Wang, S. C., PVAmPIP/PS Composite Membrane with High Performance for CO2/N2 Separation. Aiche J. 2013, 59, 215-228. 30. Li, Y. F.; Wang, S. F.; Wu, H.; Guo, R. L.; Liu, Y.; Jiang, Z. Y.; Tian, Z. Z.; Zhang, P.; Cao, X. Z.; Wang, B. Y., High-Performance Composite Membrane with Enriched CO2-Philic Groups and Improved Adhesion at the Interface. ACS Appl. Mater. Interfaces 2014, 6, 6654-6663. 31. Li, Y. F.; Xin, Q. P.; Wu, H.; Guo, R. L.; Tian, Z. Z.; Liu, Y.; Wang, S. F.; He, G. W.; Pan, F. S.; Jiang, Z. Y., Efficient CO2 Capture by Humidified Polymer Electrolyte Membranes with Tunable Water State. Energy Environ. Sci. 2014, 7, 1489-1499. 32. Favre, E.; Bounaceur, R.; Roizard, D., Biogas, Membranes and Carbon Dioxide Capture. J. Membr. Sci. 2009, 328, 11-14. 33. Turco, M.; Ausiello, A.; Micoli, L., Treatment of Biogas for Feeding High Temperature Fuel Cells: Removal of Harmful Compounds by Adsorption Processes. Springer: 2016. 34. Diaz-Cabanas, M. J.; Barrett, P. A.; Camblor, M. A., Synthesis and Structure of Pure SiO2 Chabazite: The SiO2 Polymorph with the Lowest Framework Density. Chem. Commun. 1998, 1881-1882. 35. Kim, E.; Lee, M.; Baik, H.; Han, D. Y.; Ha, J. M.; Choi, J., On the Synthesis and Characterization of All-silica CHA Zeolite Particles. Microporous Mesoporous Mat. 2014, 184, 47-54. 36. Choi, J.; Ghosh, S.; King, L.; Tsapatsis, M., MFI Zeolite Membranes from a- and Randomly Oriented Monolayers. Adsorption 2006, 12, 339-360. 37. Choi, J.; Ghosh, S.; Lai, Z. P.; Tsapatsis, M., Uniformly a-Oriented MFI Zeolite Films by Secondary Growth. Angew. Chem., Int. Ed. 2006, 45, 1154-1158. 38. Lai, Z.; Bonilla, G.; Diaz, I.; Nery, J. G.; Sujaoti, K.; Amat, M. A.; Kokkoli, E.; Terasaki, O.; Thompson, R. W.; Tsapatsis, M.; Vlachos, D. G., Microstructural Optimization of a Zeolite Membrane for Organic Vapor Separation. Science 2003, 300, 456-460. 39. Bonilla, G.; Tsapatsis, M.; Vlachos, D. G.; Xomeritakis, G., Fluorescence Confocal Optical Microscopy Imaging of the Grain Boundary Structure of Zeolite MFI Membranes Made by Secondary (Seeded) Growth. J. Membr. Sci. 2001, 182, 103-109. 40. Hong, S.; Kim, D.; Jeong, Y.; Kim, E.; Jung, J. C.; Choi, N.; Nam, J.; Yip, A. C. K.; Choi, J., Healing of Microdefects in SSZ-13 Membranes via Filling with Dye Molecules and Its Effect on Dry and Wet CO2 Separations. Chem. Mater. 2018, 30, 3346-3358. 38 ACS Paragon Plus Environment

Page 39 of 40 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


41. Otte, H. M.; Crocker, A. G., Crystallographic Formulae for Hexagonal Lattices. Phys. Stat. Sol. 1965, 9, 441-450. 42. Kosinov, N.; Auffret, C.; Borghuis, G. J.; Sripathi, V. G. P.; Hensen, E. J. M., Influence of the Si/Al Ratio on the Separation Properties of SSZ-13 Zeolite Membranes. J. Membr. Sci. 2015, 484, 140-145. 43. Krishna, R.; van Baten, J. M., Segregation Effects in Adsorption of CO2-Containing Mixtures and Their Consequences for Separation Selectivities in Cage-Type Zeolites. Sep. Purif. Technol. 2008, 61, 414-423. 44. Kida, K.; Maeta, Y.; Yogo, K., Pure Silica CHA-Type Zeolite Membranes for Dry and Humidified CO2/CH4 Mixtures Separation. Sep. Purif. Technol. 2018, 197, 116-121. 45. Krishna, R., Diffusion in Porous Crystalline Materials. Chem. Soc. Rev. 2012, 41, 3099-3118. 46. Krishna, R.; van Baten, J. M., A Molecular Dynamics Investigation of the Diffusion Characteristics of Cavity-Type Zeolites with 8-Ring Windows. Microporous Mesoporous Mat. 2011, 137, 83-91. 47. Caro, J.; Albrecht, D.; Noack, M., Why Is It So Extremely Difficult to Prepare Shape-Selective AlRich Zeolite Membranes like LTA and FAU for Gas Separation? Sep. Purif. Technol. 2009, 66, 143-147. 48. Kim, E.; Choi, J.; Tsapatsis, M., On Defects in Highly a-Oriented MFI Membranes. Microporous Mesoporous Mat. 2013, 170, 1-8. 49. Choi, J.; Jeong, H. K.; Snyder, M. A.; Stoeger, J. A.; Masel, R. I.; Tsapatsis, M., Grain Boundary Defect Elimination in a Zeolite Membrane by Rapid Thermal Processing. Science 2009, 325, 590-593. 50. Snyder, M. A.; Vlachos, D. G.; Nikolakis, V., Quantitative Analysis of Membrane Morphology, Microstructure, and Polycrystallinity via Laser Scanning Confocal Microscopy: Application to NaX Zeolite Membranes. J. Membr. Sci. 2007, 290, 1-18. 51. Lai, Z. P.; Tsapatsis, M.; Nicolich, J. R., Siliceous ZSM-5 Membranes by Secondary Growth of bOriented Seed Layers. Adv. Funct. Mater. 2004, 14, 716-729. 52. Korelskiy, D.; Ye, P.; Nabavi, M. S.; Hedlund, J., Selective Blocking of Grain Boundary Defects in High-Flux Zeolite Membranes by Coking. Journal of Materials Chemistry A: Materials for Energy and Sustainability 2017, 5, 7295-7299. 53. Yu, M.; Falconer, J. L.; Noble, R. D., Characterizing Nonzeolitic Pores in MFI Membranes. Ind. Eng. Chem. Res. 2008, 47, 3943-3948. 54. Bernal, M. P.; Coronas, J.; Menendez, M.; Santamaria, J., Separation of CO2/N2 mixtures using MFItype zeolite membranes. Aiche J. 2004, 50, 127-135. 55. Lindmark, J.; Hedlund, J., Modification of MFI membranes with amine groups for enhanced CO2 selectivity. J. Mater. Chem. 2010, 20, 2219-2225. 56. Nicolas, C. H.; Pera-Titus, M., Synthesis and characterisation of nanocomposite Ge-MFI-alumina membranes for CO2/N2 separation from wet flue gases. Microporous Mesoporous Mat. 2012, 153, 254262. 57. Betard, A.; Bux, H.; Henke, S.; Zacher, D.; Caro, J.; Fischer, R. A., Fabrication of a CO2-selective membrane by stepwise liquid-phase deposition of an alkylether functionalized pillared-layered metalorganic framework [Cu2L2P]n on a macroporous support. Microporous Mesoporous Mat. 2012, 150, 7682. 58. Xie, Z. Z.; Li, T.; Rosi, N. L.; Carreon, M. A., Alumina-supported cobalt-adeninate MOF membranes for CO2/CH4 separation. Journal of Materials Chemistry A: Materials for Energy and Sustainability 2014, 2, 1239-1241. 59. Venna, S. R.; Carreon, M. A., Highly Permeable Zeolite Imidazolate Framework-8 Membranes for CO2/CH4 Separation. J. Am. Chem. Soc. 2010, 132, 76-+. 60. Zou, X. Q.; Zhang, F.; Thomas, S.; Zhu, G. S.; Valtchev, V.; Mintova, S., Co3(HCOO)6 Microporous Metal-Organic Framework Membrane for Separation of CO2/CH4 Mixtures. Chem.-Eur. J. 2011, 17, 12076-12083. 61. Liu, Y. Y.; Zeng, G. F.; Pan, Y. C.; Lai, Z. P., Synthesis of highly c-oriented ZIF-69 membranes by secondary growth and their gas permeation properties. J. Membr. Sci. 2011, 379, 46-51.



Table of Contents

Performance with respect to the wet CO2/N2 feed

10 CO2 permeance × CO2/N2 separation factor (mol·m-2·s-1·Pa-1)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 40 of 40

10

-4

□: CHA_100 ○: CHA_∞

Si/Al ratio = 100 (CHA_100)

Si/Al ratio = ∞ (CHA_∞)

-5

HH22OO CO2 CO 2

Flue gas temperature range

10

-6

10

-7

0

50

N2 or CH4 N 2 Fluorescein Sodium salt

100

150

200

Temperature (ºC)


Chabazite-Type Zeolite Membranes for Effective CO2 Separation: The

Recommend Documents