The adsorption and mass transport of CO2 and CH4 in CHA zeolite were studied experimentally. First, large and well-defined CHA crystals with varying Si/Al ratios and morphologies ideal for adsorption studies were prepared. Then, adsorption isotherms were measured, and adsorption parameters were estimated from the data.
Trang 1Available online 29 January 2022
1387-1811/© 2022 The Authors Published by Elsevier Inc This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
in ultra-thin CHA membranes
Mojtaba Sinaei Nobandegani*, Liang Yu , Jonas Hedlund
Chemical Technology, Luleå University of Technology, SE-971 87, Luleå, Sweden
A R T I C L E I N F O
Keywords:
Adsorption
Mass transport
Surface barrier
Surface diffusion
Activation energy
A B S T R A C T The adsorption and mass transport of CO2 and CH4 in CHA zeolite were studied experimentally First, large and well-defined CHA crystals with varying Si/Al ratios and morphologies ideal for adsorption studies were prepared Then, adsorption isotherms were measured, and adsorption parameters were estimated from the data In the next step, permeation experiments for pure components and mixtures were conducted for a defect-free CHA mem-brane with a Si/Al ratio of 80 and a thickness of 600 nm over a wide temperature range A maximum selectivity
of 243 in combination with a CO2 permeance of 70 × 10− 7 mol/(m2 s Pa) was observed for a feed of an equimolar
CO2/CH4 mixture at 273 K and 5.5 bar Finally, a simple model accounting for adsorption and diffusion through the surface barriers and the interior of the pores of the membrane was fitted to the permeation data The fitted model indicated that the surface barrier was a surface diffusion process at the pore mouth with higher activation energy than the diffusion process within the pores The model also showed that the highly selective mass transport in the membrane was mostly a result of a selective surface barrier and, to a lesser extent, a result of adsorption selectivity
1 Introduction
Natural gas and biogas are mainly composed of a mixture of methane
and carbon dioxide [1–3], and the removal of CO2 is usually required to
satisfy grid and fuel specifications Water scrubbing, pressure swing
adsorption (PSA), amine sorption, cryogenic separation, and membrane
techniques [4–13] have been employed to remove CO2 from CH4
However, the existing technologies have some drawbacks such as low
selectivity, complexity, high energy consumption, and high cost [14]
Due to their high efficiency, low energy demand, compact equipment,
and straightforward operation, membrane-based techniques have been
studied intensively [11,15] and polymeric membranes have been used
for gas separation on a large scale However, polymeric membranes
display relatively poor selectivity, permeability, and stability, which
makes them disadvantageous and less applicable For instance, for
cel-lulose acetate membranes that are used on a large scale for CO2/CH4
separation, a CO2 permeance of approximately 0.6 × 10− 7 mol/(m2 s Pa)
in combination with a CO2/CH4 ideal selectivity of 35 has been observed
in the laboratory [16] For commercial polymeric membranes, the CO2
permeance is even lower; for example, polyetherimide (Ultem® 1000)
has an indicated CO2 permeance of 0.09 × 10− 7 mol/(m2 s Pa) coupled
with a CO2/CH4 selectivity of 40 [17] Zeolites are ceramic materials with well-defined pores and much higher chemical and thermal stabil-ities than polymeric materials and have been used as adsorbents for industrial gas upgrading [18–20] Ceramic zeolite membranes have the potential to display a higher selectivity, permeability, and stability than polymeric membranes for gas separations; however, zeolite membranes have not yet been commercialized for gas separations Consequently, much research and development work has been devoted to zeolite membranes during the past decades [21]
The pore system of CHA zeolite has a window size of 3.7 × 3.7 Å Because this window size is in between the kinetic diameters of CO2 (3.3 Å) and CH4 (3.8 Å), CHA zeolite can separate CO2/CH4 mixtures by molecular sieving [22–29] The CHA membranes with different chemi-cal compositions and Si/Al ratios but the same CHA pore system have been reported for CO2/CH4 separation “Pure silica” (implying an infinite Si/Al ratio) CHA membranes [30] and “high silica” (implying a finite Si/Al ratio) membranes [25,31,32] have also been reported These
“pure silica” and “high silica” CHA membranes are alternatively denoted
as SSZ-13 membranes There are also reports on SAPO-34 membranes [33,34], in which the CHA framework comprises phosphate in addition
to silica and alumina
* Corresponding author
E-mail address: mojtaba.nobandegani@ltu.se (M.S Nobandegani)
Contents lists available at ScienceDirect Microporous and Mesoporous Materials
journal homepage: www.elsevier.com/locate/micromeso
https://doi.org/10.1016/j.micromeso.2022.111716
Received 30 October 2021; Received in revised form 30 December 2021; Accepted 20 January 2022
Trang 2In previous studies [35,36], we have prepared and evaluated CHA
membranes for the separation of CO2/CH4 mixtures A high CO2/CH4
separation factor of 99 in combination with a high CO2 permeance of 60
×10− 7 mol/(m2 s Pa) was observed for a feed of an equimolar CO2/CH4
mixture at room temperature A high separation factor in combination
with high permeance is a desirable membrane property The observed
CO2 permeance was approximately 2–20 times higher than that reported
for CHA membranes in the literature [25,37,38] This high permeance
was attributed to the very thin CHA film (approximately 450 nm)
sup-ported on a highly permeable support Furthermore, the CO2 permeance
was about 100 times higher than that typically observed for polymeric
membranes, e.g., cellulose acetate membranes in the laboratory [16],
and more than 600 times larger than the indicated permeance for
commercial polyetherimide (Ultem® 1000) membranes [17]
Conse-quently, these highly permeable CHA membranes are promising for
in-dustrial applications, but the fundamental mass transfer process in thin
membranes has hitherto been poorly understood A fundamental
un-derstanding of the mass transfer process is essential for the development
of tools for engineering and, in the next step, to enable the design of
industrial CO2/CH4 separation processes
Adsorption, surface diffusion, and desorption are the main mass
transfer steps in nanoporous materials In sufficiently large crystals, the
surface diffusion step must be rate-limiting Krishna et al modeled the
mass transport of molecules through zeolite membranes with a thickness
of 50 μm using the Maxwell–Stefan equations to describe the surface
diffusion process in the pores [39,40] Similar work has also been
re-ported by Kapteijn et al for silicalite-1 membranes having a thickness of
20–60 μm [41–44] Surface barriers may influence the mass transfer as
first described by Bülow et al [45], and in small crystals and thin
membranes, the mass transfer may even be limited by the surface barrier
[46–48] The effect of surface barriers on the molecular mass transport
in zeolites has been studied by K¨arger and Bülow’s groups using various
experimental methods such as micro-imaging, NMR tracer desorption,
frequency response (FR), and barometric (or piezometric) techniques
[49–54] However, the origin of the surface barrier has been unknown
[51], although pore narrowing and pore blockage have been suggested
as possible reasons for the barrier [55] We have studied the surface
barrier in ultra-thin MFI and CHA membranes by careful permeation
experiments over a wide temperature range [56] The results indicated
that the surface barrier was the rate-limiting mass transfer step and that
it was a surface diffusion process with higher activation energy than that
for the surface diffusion process within the pores It appeared that the
activation energy was higher because there were fewer molecular
in-teractions at the pore mouth than within the pores themselves The pore
mouth was in direct contact with the gas phase where the concentration
of molecules was very low compared to the concentration in the pores
Consequently, the origin of the surface barrier may be due to the
dif-ference in geometries between the pore mouth and the pore interior
In our previous work [56], the adsorption parameters were taken
from the literature However, the chemical properties of the zeolite, such
as the Si/Al ratio and the concentration of silanol groups, affect the
adsorption parameters In addition, most of the reported adsorption data
has been determined in a narrow temperature range [27,57–59]
Determination of the parameters in a temperature range similar to the
membrane experiments may serve to avoid systematic errors caused by
taking the two measurements at different temperatures Thus, the
determination of adsorption parameters for a zeolite with the same
chemical properties as the zeolite in the membrane as well as the use of
similar temperature ranges for the adsorption and membrane
experi-ments are essential to accurately determine the surface permeability and
the corresponding activation energy Experimental studies of both
adsorption and permeation over CHA zeolites are rare, however, due to
the more extensive experimental work that they require
In the present work, the adsorption isotherms of CO2 and CH4 were
measured for CHA crystals with Si/Al ratios of 45, 77, and ∞ over a wide
temperature range of 150–350 K; the adsorption parameters were
determined from the isotherms In the next step, permeation experi-ments for the same components in their pure forms and as mixtures were conducted for an ultra-thin CHA membrane with a Si/Al ratio of 80 The crystals and membrane were synthesized in fluoride media, which has been shown to eliminate silanol groups (i.e., the concentration of silanol groups should be very low or even zero in both the crystals and the membrane) [60,61] Furthermore, the synthesis of the crystals was optimized to produce large and well-defined crystals with morphologies that were optimal for adsorption studies (e.g., a low external area/-internal area ratio) Finally, a simple mathematical model [56] ac-counting for adsorption and surface diffusion through the two surface barriers and the pores of the membrane was fitted to more extensive permeation data for the CHA membrane than in previous work [56] The permeation experiments were conducted over a wider temperature range than the CO2 adsorption experiments, i.e from 210 to 450 K This also allowed a more precise estimation of the surface permeability and the corresponding activation energy In addition, the permeance selec-tivity (usually denoted as permselecselec-tivity), adsorption selecselec-tivity, and surface permeability selectivity were evaluated, which led to a deeper understanding of the selective mass transfer processes
2 Material and methods
2.1 Synthesis of CHA crystals
To synthesize relatively large pure silica CHA crystals in fluoride media [35], distilled water, colloidal silica (40%, Ludox AS-40), N,N, N-trimethyl-1-adamantyl ammonium hydroxide (TMAdaOH 25%, SACHEM, Inc.), and hydrofluoric acid (48%) were mixed in a plastic bottle and stirred overnight at room temperature The mixture was then freeze-dried, and a small amount of water was added to obtain a gel with
a molar composition of 1.0 SiO2:1.4 TMAdaF:9.4H2O The gel was placed in an autoclave that was kept in an oven at 175 ◦C for 1 day The crystals were purified by repeated centrifugation and re-dispersion in a 0.1 M NH3 solution a total of 6 times This sample will be furthermore denoted as Si-CHA Two additional CHA samples with Si/Al ratios of 45 and 77 were synthesized using a similar procedure; these samples will be furthermore denoted as CHA45 and CHA77, respectively These samples were prepared by adding aluminum isopropoxide (99.99%, Sigma-Aldrich) to the synthesis gel, followed by stirring for 15 min before freeze-drying The compositions of the synthesis mixtures used to prepare CHA45 and CHA77 were 1.0 SiO2:0.01 Al2O3:1.4 TMA-daF:9.4H2O and 1.0 SiO2:0.005 Al2O3:1.4 TMAdaF:9.4H2O, respec-tively Finally, the crystals were calcined at 480 ◦C in ambient air for 16
h to remove the template molecules from the pores
2.2 CHA membranes
CHA membranes supported on graded α-alumina discs with a diameter of 25 mm were provided by ZeoMem Sweden AB The thick-ness of the top layer of the support was about 35 μm with a pore size of approximately 100 nm, and the thickness of the base layer was 3 mm with a pore size of about 3 μm
2.3 Characterization
Scanning electron microscope (SEM) images of the samples were recorded by using an extreme-high-resolution SEM (XHR-SEM) (Magellan 400, FEI Company, Eindhoven, The Netherlands) The in-strument was operated using an accelerating voltage of 3 kV and a probe current of 6.3 pA No conductive coating was applied to the samples prior to imaging A PANalytical Empyrean X-ray diffractometer equip-ped with a Cu LFF HR X-ray tube and a PIXcel3D detector was employed
to record XRD patterns of the zeolite crystals and the membrane in the 2θ range of 5◦–35◦ The accelerating voltage and current were 45 kV and
40 mA, respectively The Si/Al ratios of the CHA crystals were measured
Trang 3by inductively coupled plasma-sector field mass spectroscopy (ICP-
SFMS, ALS Analytica) The samples were prepared by digesting 0.1 g
zeolite powder in LiBO2 followed by dissolving in HNO3 Loss on ignition
was estimated by heating the sample to 1000 ◦C
2.4 Adsorption and permeation experiments
A Micromeritics ASAP 2020 Plus instrument equipped with a
Micromeritics Cryostat I was used to measure the adsorption isotherms
of CO2 (99.995%) and CH4 (99.9995%) at pressures up to 125 kPa The
CO2 and CH4 isotherms were measured over the temperature range of
230–350 K and 150–300 K, respectively A lower temperature range was
selected for CH4 to arrive at sufficient adsorption The samples were
degassed under vacuum conditions at 350 ◦C for 12 h before
measure-ment The equilibrium time for CO2 and CH4 was 40 and 630 s,
respectively To evaluate membrane quality, the permeance of H2 and
SF6 was measured at a feed pressure of 2 bar(a) and a permeate pressure
of 1 bar(a) at room temperature Since H2 molecules are small enough to
permeate the CHA pores while SF6 molecules can only permeate defects,
a high H2/SF6 permeance ratio indicates a high membrane quality
The membrane was mounted in a stainless steel cell and sealed with
graphite gaskets for permeation measurements over a wide temperature
range using equipment that has been detailed in previous work [62] The
membrane was dried at 573 K for 6 h in a flow of dry He, and then the
permeation experiments with pure CO2 and CH4 were conducted in the
temperature ranges of 220–450 K and 210–450 K, respectively The
membrane experiments were carried out in a slightly wider temperature
range than the temperature range for the CO2 adsorption measurements
in order to more accurately determine the activation energy for surface
permeability The pure components were fed to the membrane through a
mass flow controller, and the pressure on the feed side of the membrane
was controlled by a backpressure regulator set to 1.5 or 2 bar(a) The
pressure on the permeate side was maintained at 1 bar(a) The permeate
flow was measured by a bubble flowmeter Finally, permeation
experi-ments for the feeds comprised of the CO2/CH4 mixtures with molar
ra-tios of 50/50 and 80/20 were carried out at feed pressures of 3 and 6 bar
(a) and a permeate pressure of 1 bar(a) A drum-type flowmeter was
used to measure the permeate flow rate, and the composition of the
permeate was analyzed using an online GC (Micro GC 490, Agilent)
2.5 Modeling 2.5.1 Gas adsorption
To consider the heterogeneity of the adsorbate, Toth adsorption isotherms were fitted to the measured adsorption data [63]:
C = C sat
bP
[
1 + (bP) t]1/
t
(1)
In this equation, C represents the adsorbed concentration and C sat
represents the adsorbed concentration at saturation The parameter b is the affinity constant, t is the Toth heterogeneity parameter, and P is the pressure The parameter C sat was estimated by fitting the isotherm to the adsorption data recorded at the lowest temperature, while the
param-eters b and t were fitted at all temperatures The heat of adsorption (ΔH ads.) for the three samples with varying Si/Al ratios was estimated by fitting the van’t Hoff equation to the experimental data:
ln b =− ΔH ads.
ΔS ads.
where ΔS ads is the adsorption entropy, which was assumed to be con-stant for all three samples
2.5.2 Gas permeation
Fig 1 illustrates the mass transfer process in the zeolite membrane at
steady conditions, and for component i, the flux through the zeolite film
J i
f can be described as [56]:
p D i
α i D i+α i α i
p L + α i
p D i
(
C f eq,i− C p eq,i
)
where α i f and α i p are the surface permeabilities of i at the surface barrier
at the feed side f and permeate side p of the zeolite film, respectively D i
is the diffusion coefficient of i in the zeolite pores, ε is the fractional pore volume (0.382 for CHA [40]), and L is the zeolite film thickness C f
eq,i and
permeate sides, respectively As shown in previous work [56], the mass transfer process in these thin membranes is controlled by the surface barrier and not by the diffusion process inside the pores Adsorption equilibrium was assumed at the feed and permeate sides of the zeolite film, and consequently, the concentrations in the zeolite pores at the
feed (C f eq,i ) and permeate (C p eq,i) sides were estimated from the
Fig 1 Schematic of the mass transfer process through a zeolite membrane C f
eq denotes the concentration within the pores in equilibrium with the feed gas with
(partial) pressure p f g , C f b is the concentration at the other side of the barrier, C p
b is the concentration within the pores before the barrier at the permeate side, and C p
eq is
the concentration after the barrier within the pores in equilibrium with the gas with (partial) pressure p p
g at the permeate side of the membrane
Trang 4corresponding pressures (p f
g and p p) using Equations (1) and (2) with the fitted adsorption parameters
The surface permeability is a function of the concentration and
temperature as follows [56]:
1 − C
sat
)nexp
(
Eα R
( 1
300−
1
T
))
(4)
In Equation (4), E α is the activation energy for surface permeability,
parameter n is equal to 1.2 [56] The diffusion coefficient D is considered
Fig 2 SEM images of CHA crystals and a membrane: a) Si-CHA, b) CHA77, c) CHA45, d) Cross-sectional view of a CHA membrane, e) High resolution image of the
cross-section rotatated 90◦anti-clockwise as compared to the image in d), and f) Top-view of a CHA membrane
Trang 5to be a function of temperature and is independent of loading [64]:
The flux through the support J s was considered to be a combination
of Poiseuille flow and Knudsen diffusion [56]:
J s=
(
B0P
194K0
RT
̅̅̅̅̅
T M
√ )
dP
where B 0 (1.90 × 10− 16 m2) is the Poiseuille structural parameter, K 0
(2.40 × 10− 9 m) is the Knudsen structural parameter, M is the molar
mass (g/mol), μ is the viscosity (N⋅s/m2), and x is the thickness (35 μm)
of the top layer of the support, which is where the main mass transfer
resistance in the support is generated
The Sutherland model was used to estimate the viscosity [65]:
μ=μ ref
(
T
T ref
)32/
T ref+S
In this model, S is the Sutherland constant equal to 275 and 179 K for
CO2 and CH4, respectively The parameter μ ref is the viscosity of the gas
at the reference temperature T ref of 273 K, which is 1.37 × 10− 5 and
1.03 × 10− 5 (N s)/m2 for CO2 and CH4, respectively
The adsorption selectivity, permeance selectivity, surface
perme-ability selectivity, and driving force were estimated according to
Equations (8)–(11):
f
CO2
θ f
CH4
(8)
CO2
π f
CH4
(9)
CO2
α f
CH4
(10)
Here, θ is the loading (θ = C eq
C sat) and π represents the permeance (π i =
J i
p f−p p
i ), where J i and p i are the flux and partial pressure, respectively, of
component i These selectivities are denoted as “ideal” when estimated
for pure components and as “mixture” when estimated for mixtures
3 Results and discussion
3.1 General characterization of CHA crystals and membrane
The Si/Al ratios of the CHA45 and CHA77 crystals were determined
by ICP-SFMS to be 45 and 77, respectively The Si/Al ratio of a CHA membrane was estimated to be 80 by first performing an ion exchange of the membrane to the Cs+form and then measuring the Cs signal by EDS analysis, as described in previous work [66,67]
Fig 2 shows SEM images of the crystals and CHA membrane The CHA crystals displayed the typical pseudo-cubic habit [28,35] No crystals with other morphologies or any amorphous materials were observed by the SEM The width of the crystals was approximately 10
μm with a narrow size distribution for the three samples (Fig 2a–c) Consequently, the ratio of the external-to-internal surface areas of these large crystals was as small as 1/1000 (i.e., the adsorption data reflected only the internal surface of the crystals) Fig 3a shows the XRD patterns recorded for the CHA crystals (black traces) All observed reflections are typical for the CHA phase, as indicated by the reference pattern (ICDD-00-052-0784) for CHA crystals (blue bars) No signal from amorphous material was observed
Fig 2d and e shows SEM images of the cross-section of a CHA membrane A continuous film with a thickness of around 600 nm (Fig 2d) was observed In addition, the pores of the support were completely open, which indicated that the support was highly perme-able (Fig 2d) Furthermore, this demonstrated that the mass transfer in the support could be described using Equation (6) with parameters fitted
to the permeation data for the support (without zeolite) Fig 2e shows a high-resolution image of the cross-section of the film, which is rotated
90◦anticlockwise compared to Fig 2d The high-resolution image of the cross-section (Fig 2e) shows that the grain boundaries within the film were closed The SEM image of the membrane surface (Fig 2f) dem-onstrates that the film was comprised of well inter-grown zeolite crys-tals, and no cracks or pinholes were present The XRD pattern of the CHA membrane (black trace in Fig 3b) only displayed reflections from the CHA and alumina phases, which confirmed the high purity of the CHA membrane
3.2 Gas adsorption
The points in Fig 4 represent the measured adsorption isotherms of
Fig 3 XRD patterns of the as-synthesized: a) CHA crystals and b) CHA membrane (black trace) Blue bars represent reflections from the reference database and the
red bar represents the reflection from the α-alumina support (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)
Trang 6Fig 4 Adsorption isotherms of single components over CHA crystals Measured data are shown by points and curves represent the fitted model
Trang 7CO2 and CH4 for the CHA crystals The adsorption measurements were
repeated and almost the same results were obtained, see Fig S1a All the
isotherms appeared to be of type I [68], which is typical for microporous
materials, although saturation was not reached even at the lowest
investigated temperatures The observed CO2/CH4 ideal adsorption
selectivity was around 4.3 at 298 K and 100 kPa for Si-CHA, which is
comparable with the reported adsorption selectivity of 4.06 at 298 K and
100 kPa [29]
Fig 4 also shows that the Toth adsorption isotherms (curves) are
fitted well to the adsorption data; the R-squared values (>0.99) are
summarized in Table S1 Single site Langmuir isotherms were also fitted
to the adsorption data, but the fit was not as good, particularly for CH4 at
low temperatures, as illustrated by much lower R-squared values, see
Table S1 Fig S2 shows Toth and Langmuir isotherms fitted to CO2
adsorption data over Si-CHA crystals It shows that a Langmuir isotherm
cannot be fitted well to the data recorded at the lowest temperature, and
that the Toth isotherm can be fitted well to data recorded at all
tem-peratures To determine the adsorption capacities at saturation, the Toth
adsorption isotherms were fitted to the adsorption data recorded at the
lowest temperatures, e.g., 230 and 150 K for CO2 and CH4, respectively
The parameters b and t were then estimated by fitting the Toth
adsorption isotherms to the data recorded at all temperatures Finally,
the parameters ΔH ads and ΔS ads were estimated from the fitted b-values
by fitting the van’t Hoff equation (Equation (2)) to the data As
illus-trated by the van’t Hoff plots in Fig S3, the fit was excellent (R2 >0.99)
The fitted parameters are presented in Tables 1 and 2 and discussed
below
The fitted adsorbed concentration at saturation was 35.0 and 30.0
kmol/m3 for CO2 and CH4, respectively These values are quite similar to
those estimated by configurational-bias Monte Carlo (CBMC) simulation
and reported by Krishna et al (34.98 and 30.61 kmol/m3 for CO2 and
CH4, respectively) [40] A higher CO2 adsorption capacity is mainly be
an effect of the smaller size of the CO2 molecule compared to the CH4
molecule As shown in Tables 1 and 2, the same adsorbed concentration
at saturation was observed for all samples independent of the Si/Al ratio
This can be rationalized by the facts that only a small amount of
aluminum was introduced in the samples CHA77 and CHA45, and that
protons are the counterions, which should result in a minor influence of
the pore volume accessible for pore filling by the adsorbates
The estimated b-values at 300 K for Si-CHA of 2.9 × 10− 6 and 5.25 ×
10− 7/Pa for CO2 and CH4, respectively, are also quite similar to those
reported by Krishna et al (1.7 × 10− 6 and 6.1 × 10− 7/Pa for CO2 and
CH4, respectively) [40] The higher b-value for CO2 should be an effect
of the larger polarizability of CO2 compared to that of CH4 [69] Higher
should be an effect of the increased basicity and polarity of the
frame-work [70,71], and for CH4, this should be an effect of the increased
polarity [72,73] caused by the introduction of Al in the zeolite
The fitted Toth heterogeneity parameter (t) deviated further from
unity by decreasing the Si/Al ratio This indicated that the adsorption
sites became more heterogeneous when more Al was introduced in the
zeolite, as observed for other zeolites [74–77] The fitted heterogeneity
parameter was also lower for CH4 than for CO2, which indicated that the
adsorption sites for CH4 are more heterogeneous than those for CO2
The heats of adsorption ΔH ads for CO2 and CH4 were estimated
within the ranges of − 26.75 to − 25.82 kJ/mol and − 17.76 to − 17.23 kJ/mol, respectively, which are close to the values reported by other groups [27,40,78] A more negative heat of adsorption is expected for adsorption systems with larger polarities [47,79], which is in concert with the observed heat of adsorption for CO2
Fig 5 shows plots of -ΔH ads (Fig 5a), b-values (Fig 5b), and t
(Fig 5c) as a function of the Al/Si ratio, i.e., the inverse of the more common Si/Al ratio More details can be found in Fig S4 It is evident
that the parameters -ΔH ads and b are increasing nearly linearly with the Al/Si ratio, while the parameter t is decreasing nearly linearly with the Al/Si ratio, as shown by the fitted lines The values of -ΔH ads. , b, and t for
the membrane with an Al/Si ratio of 1/80 were estimated from these linear dependencies, and the estimated values are given in Tables 1 and
2 These estimated values differ only slightly from the literature data that
we used in previous work [56]
3.3 Single-component permeation experiments
The permeances of pure H2 and SF6 over the membrane were measured to be 52 × 10− 7 and 7 × 10− 11 mol/(m2 s Pa), respectively The H2/SF6 permeance ratio was as high as 75,000, which was indica-tive of a high membrane quality [25,30] and shows permeation data should reflect only mass transfer in the pores of the zeolite, and not in the defects This ratio is much higher than the ratios previously reported for CHA membranes [25,30], which were in the range of 200–600
In the next step, single-component permeation experiments with CO2
and CH4 at feed pressures of 1.5 and 2 bar(a) were carried out at various temperatures The maximum CO2 fluxes were observed at 280 K and were 0.42 and 0.88 mol/(m2⋅s) for feed pressures of 1.5 and 2 bar(a), respectively (Fig 6a) Comparable results were obtained from repeating the permeation measurement, e.g Fig S1b shows the results for CH4
permeation at 2 bar(a) feed pressure The corresponding CO2 per-meances were as high as 82 × 10− 7 and 86 × 10− 7 mol/(m2 s Pa), which
is similar to the permeance of 78 × 10− 7 mol/(m2 ⋅ s ⋅ Pa) that we have reported for an ultra-thin MFI membrane [80] and significantly higher than the permeance reported by Falconer et al for SSZ-13 membranes [23] High CO2 permeance must have been a result of the low membrane thickness of 600 nm in combination with a highly permeable and open support as observed by the SEM At the same temperature, considerably lower CH4 fluxes of 2.6 × 10− 3 and 5.8 × 10− 3 mol/(m2⋅s) were observed, which corresponded to low CH4 permeances of 0.51 × 10− 7
and 0.57 × 10− 7 mol/(m2 s Pa) at feed pressures of 1.5 and 2 bar(a), respectively Consequently, a high maximum ideal CO2/CH4 permeance selectivity of 160 was observed (see Fig 6b) This selectivity is much higher than the reported selectivities of 54 [30] and 76 [35] at com-parable test conditions and membrane types
Based on the parameters estimated from the adsorption data (see
Tables 1 and 2), Equation (3) was fitted to the data Since the mass transfer process is controlled by the surface barrier in thin membranes [56], the experimental data could not be used to determine the diffusion
coefficient; thus, diffusion coefficients (D i) were taken from the litera-ture (2.5 × 10− 9 and 5 × 10− 11 m2/s at 300 K for CO2 and CH4, respectively [40]) These diffusion coefficients were assumed to be in-dependent of loading, while the surface permeability at zero concen-tration (α *), the activation energy of the surface permeability (E α), and
Table 1
Fitted parameters for CO2 adsorption in CHA
3 ) Ref ΔH ads (kJ/mol)
Trang 8diffusion in the pores (E Diffusion) were fitted to the experimental data As
shown by Teixeira et al [81], the activation energy for diffusion E Diffusion
can be correctly estimated from experimental data independent of
crystal size Consequently, the activation energy for diffusion was
esti-mated from the experimental data in the present work, which has also
been demonstrated in previous work [56] The fitted parameters are
summarized in Table 3, and the agreement between the fitted model and
the experimental data is illustrated in Fig 6a
A surface permeability at zero concentration (α *) of 2.0 × 10− 5 m/s
was observed for CH4, which is 40 times lower than that of CO2, which
was 8.0 × 10− 4 m/s This is presumed to be the main reason for the
highly selective mass transfer of CO2 across the membrane For a deeper
analysis of this selective mass transfer, the ideal surface permeability
selectivity is plotted in Fig 6b This selectivity was 25 at 450 K and
increased to 1500 at 220 K As shown in Fig 6b, the ideal surface permeability selectivity was always much higher than the ideal adsorption selectivity, and the difference was particularly large at low temperatures The highly selective mass transfer through the membrane was mostly an effect of the selective surface barrier, i.e., a high ideal surface permeability selectivity The high surface permeability selec-tivity at low temperatures was largely due to the high adsorbed con-centration of CO2 at low temperatures, which resulted in a small denominator in Equation (4) and thereby a large surface permeability for CO2 For instance, at 230 K and a feed pressure of 2 bar(a), C f CO2 was 33.1 kmol/m3, which is close to the C sat of CO2 (35 kmol/m3) This produced a denominator value in Equation (4) of 0.03 and an α CO2 of 5.8
×10− 3 m/s Under the same conditions, C f
CH4 was 14.8 kmol/m3, which
is less than half of the C sat of CH4 (30 kmol/m3) This resulted in a
Table 2
Fitted parameters for CH4 adsorption in CHA
3 ) Ref ΔH ads (kJ/mol)
[ 40 ] [ 27 ]
Fig 5 a) -ΔH ads , b) b-values, and c) t-values as a function of the Al/Si ratio Circular red-filled and empty blue symbols indicate the measured values for CO2 and CH4
adsorption for the crystals, respectively Lines are fitted to the data by linear regression and the stars indicate the estimated values for the membrane (For inter-pretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)
Trang 9denominator value in Equation (4) of approximately 0.44, a small α CH4
value of 6.1 × 10− 6 m/s, and an ideal surface permeability selectivity of
943 In addition, the activation energy for surface permeability E α of
CH4 (16.5 kJ/mol) was larger than that of CO2 (13.0 kJ/mol) (see
Table 3) Thus, there was a higher temperature sensitivity and a greater
reduction of the surface permeability for CH4 at lower temperatures,
which supports the observation that surface permeability selectivity
increases with decreasing temperature
Fig 6b also shows the driving force for the mass transfer across the
membrane expressed as the loading difference θ f− θ p, which is a func-tion of the temperature and corresponds to the concentrafunc-tion difference
C f
eq in Equation (3) In the case of CO2 and a feed pressure of 2 bar (a), a maximum driving force was observed at about 285 K This maximum driving force was a result of increasing CO2 adsorption
to-wards saturation at the feed side (C f
eq) when the temperature was decreasing to 285 K At temperatures lower than 285 K, the driving force
of CO2 was reduced due to increased adsorption towards saturation at
the permeate side (C p
eq) The presence of the maximum driving force for
CO2 mass transfer was the main reason for the maximum CO2 flux at 280
K, as shown in Fig 6a However, the maximum CO2 flux was observed at
a slightly lower temperature than the temperature at which the maximum driving force occurred, due to the increasing surface perme-ability of CO2 with decreasing temperature As mentioned above, a maximum ideal permeance selectivity of 160 was observed at 280 K, which was because the maximum driving force for CH4 occurred at a lower temperature (220 K) than the temperature at which the maximum driving force for CO2 occurred (285 K)
Table 3 demonstrates that the activation energy for surface
Fig 6 a) Single-component CO2 and CH4 fluxes and b) Ideal CO2/CH4 permeance selectivity, ideal surface permeability selectivity, ideal adsorption selectivity, and driving force Filled symbols and lines show the experimental data and model, respectively, for a feed pressure of 2 bar(a), while empty symbols and dashed lines show the experimental data and model, respectively, for a feed pressure of 1.5 bar(a) The permeate pressure is 1 bar(a) in all cases
Table 3
Fitted parameters in Equation (3)
Single Component
E 300K
E 300K
Mixture
α * at 300 K (m/s) 3.0 × 10 − 5 8.0 × 10 − 4
Trang 10permeability was higher than the activation energy for diffusion in the
pores This indicates that surface permeation was the limiting mass
transfer step across the membrane and, consequently, this higher
acti-vation energy caused the surface barrier [56] Furthermore, the ratio
Lα0α L
D(α0 +α L)can be used to determine if either the surface permeability or the
diffusivity was limiting the mass transfer [56] For single-component
permeation of CO2 and CH4, this ratio was 0.18 and 0.13,
respec-tively These low ratios indicate that the surface barrier controls the
mass transport in the thin membranes K¨arger et al [50,51] observed
increasing surface permeability with increasing concentration of
adsorbed molecules However, no mathematical description of this
de-pendency was suggested
In our previous work [56], we showed that when Equation (4)
accurately describes the surface permeability, it can be fitted well to our
experimental data as well as to the experimental data reported by K¨arger
et al [50,51] Equation (4) is similar to the HIO model derived for
surface diffusion [82] under the assumption that molecules jump from
site to site, and if a site is occupied, the molecule is scattered to another
site In the HIO model, it is further assumed that molecular interactions
are negligible, which results in n = 1.0 The successful fitting of Equation
(4) to the experimental data suggests that the surface permeation
pro-cess is a surface diffusion propro-cess [56] We observed that the fitted
activation energy for surface permeability E α was higher than the
acti-vation energy for surface diffusion E Diffusion within the pores and that the
molecular interactions increased n to 1.2, which reduced these
activa-tion energies We suggested that the surface barrier is a result of the
geometrical differences between the pore mouth and the interior of the
pores, with more significant interactions within the pores than at the pore mouth Consequently, it appeared that the surface barrier is a geometrical effect The excellent fit between the model and the more extensive experimental data for a CHA membrane at two different pressures and over a wide temperature range in the present work in-dicates that Equation (3) provides an adequate description of the mass transfer and that Equation (4) provides an adequate description of the temperature and concentration dependencies of the surface permeability
3.4 Mixture permeation experiments
The points in Fig 7a represent, as a function of temperature, the experimental permeation data for a feed of CO2/CH4 mixtures with compositions of 50/50 and 80/20 (molar ratios) at a feed pressure of 5.5 bar(a), and a permeate pressure of 1 bar(a) The observed CO2 flux was consistently about two orders of magnitude higher than the observed
CH4 flux The membrane was highly CO2 selective across the entire studied temperature range and a maximum separation factor of 156 was observed at 273 K (Fig S5), which corresponded to a mixture permeance selectivity of 243 (Fig 7b) Table 4 summarizes CO2/CH4 separation data reported for zeolite and MOF membranes in the literature and in the present work
As shown in Fig 7b, the selective mass transfer across the membrane was a result of the high mixture surface permeability selectivity (approximately 54 at 300 K) and high mixture adsorption selectivity (6.7 at 300 K) The curves in Fig 7a illustrate that the same model and
Fig 7 a) Fluxes observed for CO2/CH4 mixtures with compositions of 50/50 and 80/20 (molar ratios), b) Mixture selectivities and driving forces for a 50/50 CO2/
CH4 mixture, and c) Mixture selectivities and driving forces for an 80/20 CO2/CH4 mixture Points indicate experimental data and curves indicate the fitted model