Received: 10 January 2019
Revised: 21 March 2019
Accepted: 8 April 2019
DOI: 10.1002/aic.16611
EDITOR’S CHOICE: AICHE LETTER: SEPARATIONS: MATERIALS, DEVICES AND PROCESSES
Ultra-thin skin carbon hollow fiber membranes for sustainable molecular separations Chen Zhang
| Rachana Kumar | William J. Koros
School of Chemical & Biomolecular Engineering, Georgia Institute of Technology, Atlanta, Georgia
Significance A transformative platform is reported to derive ultra-thin carbon molecular sieve
Correspondence William J. Koros, School of Chemical & Biomolecular Engineering, Georgia Institute of Technology, 311 Ferst Drive, Atlanta, GA 30332. Email:
[email protected] Chen Zhang, School of Chemical & Biomolecular Engineering, Georgia Institute of Technology, 311 Ferst Drive, Atlanta, GA 30332. Email:
[email protected]
(CMS) hollow fiber membranes from dual-layer precursor hollow fibers with indepen-
Present address Chen Zhang, Department of Chemical and Biomolecular Engineering, University of Maryland 4418 Stadium Dr. College Park, MD 20742.
KEYWORDS
dently tuned skin layer and substrate properties. These ultra-thin CMS hollow fiber membranes show attractive CO2/CH4 separation factors and excellent CO2 permeances up to ~1,400% higher than state-of-the-art asymmetric CMS hollow fiber membranes. They provide a unique combination of permeance and selectivity competitive with zeolite membranes, but with much higher membrane packing density and potentially much lower costs.
carbon molecular sieve membranes, hollow fiber membranes, molecular separations, natural gas purification, ultra-thin membranes
Funding information Basic Energy Sciences, Grant/Award Number: DE-FG02-04ER15510; Shell International Exploration and Production Inc.
that are more rigid than flexible polymers.6,7 Rigid membrane mate-
1 | I N T RO D UC T I O N
rials can provide exceptional entropic diffusion selectivities for closely Molecularly selective membranes can enable sustainable large-scale
sized and/or shaped molecular pairs.8 Carbon molecular sieves (CMS)
molecular separations by reducing the required thermal energy inputs
represent a class of rigid membrane materials having excellent bal-
and CO2 footprints.1 One of the most commercially successful exam-
ances between transport properties, tunability, and scalability.9 CMS
ples is membrane desalination accounting for over 50% of worldwide
materials comprise micropores made by packing imperfections of
2
desalination capacity. The breakthrough allowing its large-scale prac-
graphene-like sheets. These graphene-like sheets are populated by
tice was the invention of thin-film composite membranes comprising
molecular-size features that allow precise differentiation between
selective crosslinked aromatic polyamides. The intrinsic water perme-
molecules with only 0.1–0.2 Å size difference.10 CMS materials are
ability of the material is moderate; however, defect-free ultra-thin
formed by controlled pyrolysis of polymer precursors. By tuning pre-
(~100–500 nm) separation (skin) layers can be formed via in situ inter-
cursor chemistry and pyrolysis conditions, the pore structure of CMS
facial polymerization, thereby enabling attractive productivity in prac-
materials can be controlled to provide desirable transport properties
tical spiral-wound modules.4,5 Indeed, economically translating high-
for target molecular separations (e.g., O2/N2, CO2/CH4, C2H4/C2H6,
performance materials into scalable devices with ultra-thin skin layers
C3H6/C3H8, xylene isomers).8,11-14 Notably, the family of 6FDA poly-
is key to enable large-scale application of any molecularly selective
imides
membrane.
DETDA/DABA) are outstanding precursor materials.11,15,16 Pyrolysis
3
(e.g.,
6FDA/BPDA-DAM,
6FDA-mPDA/DABA,
6FDA-
Expanding the boundaries of large-scale membrane separations
of these advanced polyimides has been shown to create CMS mate-
beyond desalination, however, requires advanced membrane materials
rials with remarkably high CO2 permeability (~7,200–22,000 Barrer)
1 of 7
© 2019 American Institute of Chemical Engineers
wileyonlinelibrary.com/journal/aic
AIChE Journal. 2019;65:e16611. https://doi.org/10.1002/aic.16611
2 of 7
ZHANG ET AL.
and simultaneously attractive CO2/CH4 ideal selectivity (~30–50) suit-
the core polymer dope. Compositions of the polymer dopes (Table S1)
able for membrane-based natural gas purification.
and spinning parameters (Table S2) can be found in the Supporting
Two challenges of CMS membranes must be addressed before a
Information.
breakthrough similar to the one made by thin-film composite mem-
The dual-layer precursor hollow fibers were soaked in a 10 wt%
branes can occur. First, 6FDA-polyimides are more expensive than
VTMS/hexane solution for 24 hr followed by hydrolysis/condensation
commercially available polymers. Producing CMS membranes can be
in saturated water vapor.18 The hollow fiber skin layer was then in situ
costly if the entire precursor hollow fiber consists of these advanced
hybridized by sequential soaking in two monomer solutions. The first
polymers. Second, although translating CMS materials to scalable hol-
solution comprises 0.001–0.1 wt% diethyltoluenediamine (DETDA)
low fiber formats was demonstrated, a platform has not been devel-
dissolved in hexane, and the second solution comprises 0.001–0.1 wt
oped to create CMS hollow fiber membranes with ultra-thin (<1 μm)
% trimesoyl chloride (TMC) dissolved in hexane. Following hybridiza-
skin layers. The porous substrate of precursor hollow fibers can col-
tion, the precursor hollow fibers were dried in a vacuum oven at
lapse during pyrolysis, providing CMS hollow fiber membranes with
150 C for 12 hr prior to controlled pyrolysis under continuous purge
thick skin layers (15–50 μm).
Treating precursor hollow fibers
(200 cc/min) of ultra-high-purity Argon at 550 C. Details regarding
with silane (e.g., vinyltrimethoxysilane, VTMS) prior to pyrolysis sup-
heating protocols and construction of CMS hollow fiber modules can
presses substrate collapse.18 Skin layer thickness of asymmetric CMS
be found in the Supporting Information.
17,18
hollow fiber membranes prepared by this method can be reduced to
Separation performance of the ultra-thin CMS hollow fiber mem-
~3–6 μm, which is still much thicker than the precursor hollow fibers.
branes was studied using the constant-pressure permeation method
Further reducing CMS hollow fiber skin layer thickness (ideally to that
with an equimolar CO2/CH4 mixture at 100 psia and 35 C. Membrane
of the precursor hollow fibers, i.e., <1 μm) will provide significantly
permeate pressure was kept at 1 atm. Permeate flow rate was mea-
higher permeances to reduce required membrane area and system
sured using a bubble flow meter (10 mL) and compositions were ana-
footprint. Clearly, addressing the two challenges will drastically
lyzed using a Varian-430 gas chromatograph (GC). The stage-cut,
enhance the economic viability of CMS membranes.
which is the percentage of feed mixture that permeates through the
In this letter, we report a transformative ultra-thin CMS hollow
membrane, was kept less than 1%. Permeation data were collected
fiber platform that addresses both challenges using Matrimid®
after bubble flow meter and GC readings became stabilized. A mini-
(MA) and 6FDA/BPDA-DAM (6F) as precursors. The terminology used
mum of three GC injections were made for each individual membrane
to refer to these ultra-thin CMS hollow fibers is ULT CMS-X#, where
module. Permeation data of individual membrane modules can be
X refers to the sheath precursor material and # refers to the CMS skin
found in Table S3. Scanning electron microscope (SEM) images were
layer thickness in μm. The intrinsic structural and transport properties
obtained in a LEO 1530 field emission SEM. Fourier transform infra-
of CMS materials derived from these two precursors have been
red spectroscopy (FT-IR) was done using a Bruker Tensor
19-21
reported in the literature.
By using dual-layer precursor hollow
27 spectrometer.
fibers, this novel platform enables creation of advanced CMS hollow fiber membranes with ultra-thin skin layers. These advanced structures provide highly attractive permeance more than one order of magnitude higher than state-of-the-art asymmetric CMS hollow fiber membranes. Additionally, the approach allows derivation of CMS hollow fiber membranes from 6FDA-polyimides with minimal consumption (up to 98% less) of the expensive polymer. Last, we report an innovative in situ hybridization method to effectively repair skin layer defects of CMS hollow fiber membranes.
3 | RESULTS AND DISCUSSION CMS hollow fiber membranes are traditionally derived from monolithic precursor hollow fibers (Figure 1a). Monolithic Matrimid® precursor hollow fibers (Figure 1b) have skin layers ~1 μm. Following silane treatment and pyrolysis, the porous substrate underneath the skin layer partially collapsed, and the skin layer thickness of the asymmetric CMS hollow fiber membrane (CMS-MA) increased to ~6 μm (Figure 1c). To create CMS hollow fiber membranes with ultra-thin
2 | EXPERIMENTAL METHODS
skin layers, substrate collapse of precursor hollow fibers must be totally avoided. Investigating substrate collapse of asymmetric CMS
The dual-layer precursor hollow fibers were formed using dry-jet wet-
hollow fiber membranes suggests that a highly porous substrate can
quench spinning by coextruding the sheath and core polymer dopes
provide improved collapse resistance during pyrolysis (Figure S4).
®
from a multichannel spinneret. To form dual-layer Matrimid /
With monolithic precursor hollow fibers, however, it's challenging to
Matrimid® precursor hollow fibers, both the sheath and core polymer
independently control the substrate porosity without introducing skin
dopes comprise Matrimid® 5218 polyimide (Huntsman Advanced
layer defects. To overcome this challenge, we derived CMS hollow
Materials). To form dual-layer 6F/Matrimid® precursor hollow fibers,
fiber membranes from dual-layer precursor hollow fibers with tailored
the sheath polymer dope comprises 6FDA/BPDA-DAM polyimide
chemistry and substrate morphology (Figure 1d). A dual-layer precur-
and the core polymer dope comprises Matrimid® 5218 polyimide. To
sor hollow fiber consists of an outside sheath layer and an inside core
increase core layer porosity of the dual-layer precursor hollow fibers,
layer formed by co-extruding sheath and core polymer dopes. By dis-
polyvinylpyrrolidone (Mw~1,300,000, Sigma-Aldrich) was dissolved in
solving pore formers22 (e.g., polyvinylpyrrolidone, PVP) in the core
3 of 7
ZHANG ET AL.
F I G U R E 1 Formation of ultra-thin CMS hollow fiber membranes. (a) Schematic showing formation of asymmetric CMS hollow fiber membranes using monolithic precursor hollow fibers; (b, c) scanning electron microscope (SEM) images of monolithic Matrimid® precursor hollow fibers and asymmetric CMS hollow fiber membrane (skin layer 6 μm) derived thereof (CMS-MA); (d) schematic showing formation of ultra-thin CMS hollow fiber membranes using dual-layer precursor hollow fibers comprising the same polymer in the sheath and core layer; (e, f) SEM images of dual-layer Matrimid®/Matrimid® precursor hollow fibers and ultra-thin (skin layer 1 μm) CMS hollow fiber membranes (ULT CMSMA1) derived thereof; (g, h) SEM images of dual-layer Matrimid®/Matrimid® precursor hollow fibers and ultra-thin (skin layer 0.5 μm) CMS hollow fiber membranes (ULT CMS-MA0.5) derived thereof; (i) schematic showing formation of ultra-thin CMS hollow fiber membranes using dual-layer precursor hollow fibers comprising different polymers in the sheath and core layer; (j, k) SEM images of dual-layer 6F/Matrimid® precursor hollow fibers and ultra-thin (skin layer 0.5 μm) CMS hollow fiber membranes (ULT CMS-6F0.5) derived thereof. 6F, 6FDA/BPDADAM; CMS, carbon molecular sieve; HF, hollow fiber; MA, Matrimid®; ULT, ultra-thin [Color figure can be viewed at wileyonlinelibrary.com] polymer dope, we found that the substrate porosity of dual-layer ®
®
Without substrate collapse, skin layer thickness of ultra-thin
precursor hollow fiber can be dramatically
CMS hollow fiber membranes is determined by sheath layer thick-
increased (Figure 1e). Fortunately, a dense sheath layer can still be
ness of the dual-layer precursor hollow fibers. The sheath layer
formed as it relies upon independently extruded sheath polymer dope
thickness can be controlled by spinning parameters of the dual-layer
without dissolved pore formers. With the aid of silane treatment, sub-
precursor hollow fibers. A smaller ratio of sheath versus core dope
strate collapse was totally avoided during pyrolysis at 550 C
flow rate will provide dual-layer precursor hollow fibers with thinner
(Figure 1f). This gives CMS hollow fiber membranes (ULT CMS-MA1)
sheath layer. For example, by reducing the flow rate ratio from 1:30
with much more open substrate and ultra-thin skin layer (~1 μm).
to 1:50 (Table S2), we reduced sheath layer thickness of the dual-
Interestingly, a clear boundary is seen between the skin layer and the
layer Matrimid®/Matrimid® precursor hollow fibers from ~1 to
Matrimid /Matrimid
highly porous substrate, which does not exist in asymmetric CMS hol-
~0.5 μm (Figure 1g). The reduction contributes to thinner skin layer
low fiber membranes derived from monolithic precursor hollow fibers
(~0.5 μm, Figure 1h) of ultra-thin CMS hollow fiber membranes (ULT
(Figure 1c).
CMS-MA0.5). Clearly, ultra-thin CMS hollow fiber membranes with
4 of 7
ZHANG ET AL.
even thinner skin layer (<0.5 μm) can be made by further tuning spin-
to derive defect-free CMS hollow fiber membranes, which has limited
ning parameters of dual-layer precursor hollow fibers.
the scope of suitable precursor chemistry and structures for CMS
This highly versatile platform was successfully adapted to dual-layer
membrane formation. To address this challenge, we developed a
precursor hollow fibers comprising different sheath and core layer poly-
method to enhance separation factors of defective CMS membranes
mers (Figure 1i). We derived ultra-thin CMS hollow fiber membranes
by hybridizing precursor hollow fiber skin layers with in situ synthe-
using a 6FDA-polyimide with minimal need for the advanced polymer
sized polyamide. The hybridization occurs by sequentially soaking the
(2 wt% of the entire hollow fiber, Table S2). Sheath layer (~0.5 μm) of
precursor hollow fibers in diamine DETDA/hexane and acid chloride
®
the dual-layer 6F/Matrimid
precursor hollow fibers comprises
TMC/hexane solutions. Condensation polymerization of the highly
6FDA/BPDA-DAM polyimide (Figure 1j). The core layer does not con-
reactive monomers forms branched or crosslinked polyamide, filling
tribute to formation of the CMS membrane skin layer and thus only
the skin layer defects (Figure 2a).26 As the hybridized precursor hol-
comprises the less expensive Matrimid® polyimide. As shown in
low fibers undergo pyrolysis, the crosslinked polyamides inside the
Figure 1k, pyrolysis of the dual-layer 6F/Matrimid® precursor hollow
defects were presumably transformed into CMS, thereby eliminating
fibers gives CMS hollow fiber membranes (ULT CMS-6F0.5) with highly
skin layer defects on CMS hollow fiber membranes.
open substrate and crack-free ultra-thin skin layer (~0.5 μm). This result
Ultra-thin CMS hollow fiber membranes (ULT CMS-MA1) were
may seem surprising considering the different thermal expansion coeffi-
derived from dual-layer Matrimid®/Matrimid® precursor hollow fibers
23
It's hypothesized that as
hybridized with 0.1 wt% DETDA/0.1 wt% TMC prior to pyrolysis. The
the sheath and core layers undergo thermal decomposition during
hybridization dramatically increased CO2/CH4 separation factor of
pyrolysis, they were both transformed into disordered defective
the ultra-thin CMS hollow fiber membrane from ~7.9 to ~36.7,
graphene-like sheets with similar chemistry and thermal expansion
suggesting that hybridization was highly effective to repair skin layer
coefficients. Hence, an integral skin layer can be formed without crack-
defects of CMS hollow fiber membranes. To our best knowledge, this
ing. The ultra-thin CMS hollow fiber membranes were evaluated for
is the first time that a method to repair CMS hollow fiber membranes
CO2/CH4 separation using an equimolar CO2/CH4 feed mixture at
skin defect is reported. Notably, the ultra-thin CMS hollow fiber mem-
100 psia and 35 C. Compared with asymmetric CMS hollow fibers with
branes provide excellent CO2 permeance ~1,177 GPU. This may seem
cients of the sheath and core layer polymers.
24
these specific ultra-thin CMS hollow fibers show
surprising because the crosslinked polyamide may introduce signifi-
unattractive CO2/CH4 separation factor (~7.9). This result indicates that
cant mass transfer resistance if a continuous film is formed on precur-
defects exist in the skin layer of the ultra-thin CMS hollow fiber mem-
sor hollow fiber skin layer due to hybridization. Unlike interfacial
branes, which was likely because the dual-layer precursor hollow fibers
polymerization of monomers dissolved in two immiscible solvents
were defective (Figure 2a). This was evidenced by low O2/N2 ideal
(e.g., water and hexane),27 the hybridization method relies upon
selectivities of the dual-layer precursor hollow fibers (Table S4).
monomers dissolved in the same solvent (hexane). The diamine mono-
thicker skin layers,
Skin layer defects of precursor hollow fibers can be repaired by 25
mer solution was sorbed into the defects at the membrane surface,
caulking ; however, to our best knowledge, methods that can repair
thereby providing reactants for film growth inside the defects, and
skin layer defects of CMS hollow fiber membranes have not been
possibly at the membrane surface surrounding the defects. As the
reported. Defect-free precursor hollow fibers are generally preferred
defect density was low, we anticipate that a continuous polyamide film
F I G U R E 2 Repairing skin layer defects of ultra-thin CMS hollow fiber membranes. (a) Schematic showing repairing CMS hollow fiber membrane skin defects by in situ hybridization. The inset represents chemical structure of crosslinked polyamide formed by highly reactive monomers (DETDA and TMC); (b) effects of hybridization monomer concentrations on CO2/CH4 separation performance of ultra-thin CMS hollow fiber membranes (ULT CMS-MA1); (c) comparing FT-IR spectra of dual-layer Matrimid®/Matrimid® precursor hollow fiber prior to and after hybridization with 0.1 wt% DETDA/0.1 wt% TMC. The spectra of bulk polyamide formed by solution polymerization of DETDA and TMC are shown for reference. CMS, carbon molecular sieve; DETDA, diethyltoluenediamine; FT-IR, Fourier transform infrared spectroscopy; TMC, trimesoyl chloride; ULT, ultra-thin [Color figure can be viewed at wileyonlinelibrary.com]
5 of 7
ZHANG ET AL.
was not able to cover the entire membrane surface. This hypothesis
~1,400% higher CO2 permeance, respectively. Similarly, ULT CMS-
was supported by comparing morphology of the precursor hollow fiber
6F0.5 with ultra-thin skin layer (~0.5 μm) was ~830% more permeable
skin layer before and after hybridization (Figures S6 and S7). It is worth-
than asymmetric CMS hollow fiber membranes (CMS-6F) derived
while to note that this is quite different from formation of polyamide
from monolithic 6FDA/BPDA-DAM precursor hollow fibers.31 While
thin-film composite membranes often studied for desalination. With
CO2/CH4 separation factors of ultra-thin CMS hollow fiber mem-
highly porous substrates (e.g., ultrafiltration membranes), the surface
branes were ~15–25% lower than asymmetric CMS hollow fiber
pore density is sufficiently high that the polyamide film formed at indi-
membranes with thicker skin layers, they remain highly attractive for
vidual pore surface can connect to provide a continuous film.28
energy-efficient natural gas purification.32
Our results further suggest that separation performance of the
One would expect ULT CMS-MA0.5 (skin layer ~0.5 μm) with 50%
ultra-thin CMS hollow fiber membranes can be tuned by optimization
thinner skin layer to provide 100% higher CO2 permeance than ULT
of monomer concentrations (Figure 2b). By reducing monomer con-
CMS-MA1 (skin layer ~1 μm). However, ULT CMS-MA0.5 (CO2
centration from 0.1 to 0.001 wt%, CO2 permeance of the ultra-thin
permeance ~1,310 GPU) was only slightly more permeable than ULT
CMS hollow fiber membrane (ULT CMS-MA1) increased from 1,177
CMS-MA1 (CO2 permeance ~1,177 GPU). This was likely due to exis-
to 1,452 GPU. CO2/CH4 separation factor dropped to ~18, which was
tence of the so-called “hyperskin” at the outmost region of CMS hollow
still quite attractive for bulk acid gas removal.29 It is hypothesized that
fiber membrane skin layer.33 The hyperskin has extremely low perme-
at higher monomer concentrations, a larger percentage of skin layer
ability, and limits membrane permeance as the membrane skin layer
defects can be filled by polyamides providing higher membrane sepa-
thickness approaches that of the hyperskin. Eliminating the hyperskin
ration factors. We studied surface chemistry of the precursor hollow
will be useful to further increase CO2 permeance; however, is not within
fibers before and after hybridization using FT-IR. The characteristic
the scope of the current work. Permeance change due to physical aging
band (1,637 cm−1) of amide linkages30 was not observed in dual-layer
is often discussed for CMS membranes.31,34 Separation performance of
precursor hollow fiber hybridized with 0.1 wt% DETDA/0.1 wt% TMC
a ultra-thin CMS hollow fiber membrane module (M7, Table S3) was
(Figure 2c). This supports our hypothesis that the crosslinked polyam-
periodically measured as it was stored under 100 psia pure CO2 for
ide only exists inside skin layer defects and its concentration in the
275 days (Figure 3b). The separation performance stabilized ~100 days
polyimide skin layer was possibly too low for FT-IR detection.
following membrane formation, before which CO2 permeances dropped
We compared CO2/CH4 separation performance (Figure 3a) of
by ~35% and CO2/CH4 separation factor increased by ~23%. The stabi-
ultra-thin CMS hollow fiber membranes (ULT CMS-MA1, ULT CMS-
lized CO2 permeances ~850 GPU remains highly attractive (840% higher
MA0.5, and ULT CMS-6F0.5) derived from dual-layer precursor hol-
than the asymmetric CMS hollow fiber membranes).
low fibers with silane treatment and hybridization (0.1 wt%
Competing with zeolite membranes has been challenging for poly-
DETDA/0.1 wt% TMC). Clearly, CO2 permeances of the ultra-thin
mer membranes and polymer-derived membranes (e.g., CMS mem-
CMS hollow fiber membranes were remarkably higher than conven-
branes, mixed-matrix membranes).35-41 We compared the ultra-thin
tional asymmetric CMS hollow fiber membranes. Compared with
CMS hollow fiber membranes with tubular zeolite membranes and
CMS-MA with thicker skin layer (~6 μm), ULT CMS-MA1 (skin layer
other polymer or polymer-derived hollow fiber membranes for CO2/
~1 μm), and ULT CMS-MA0.5 (skin layer ~0.5 μm) provide ~1,200 and
CH4 separation (Figure 3c). It should be noted that permeation data
F I G U R E 3 Evaluation of membrane separation performance. (a) CO2/CH4 separation performance of ultra-thin CMS hollow fiber membranes (ULT CMS-MA1, ULT CMS-MA0.5, ULT CMS-6F0.5). The ultra-thin CMS hollow fiber membranes were all pyrolyzed (550 C) from precursors hybridized with 0.1 wt% DETDA/0.1 wt% TMC. Separation performance of asymmetric CMS hollow fiber membranes (CMS-MA and CMS-6F) pyrolyzed (550 C) from monolithic precursors are also shown for comparison. (b) Periodically measured CO2/CH4 separation performance of a ultra-thin CMS hollow fiber membrane module within 275 days. The module was stored under 100 psia pure CO2 between permeation measurements. (c) Comparing CO2/CH4 separation performance of hollow fiber membranes and tubular membranes (Table S5). The solid gray line represents the converted (assuming 1 μm skin layer) Robeson upper bond for the CO2/CH4 pair. Solid squares represent ultra-thin CMS hollow fiber membranes reported in this work. (Hollow triangles: tubular zeolite membranes; hollow squares: asymmetric CMS hollow fiber membranes derived from monolithic precursor hollow fibers; hollow circles: polymeric hollow fiber membranes). CMS, carbon molecular sieve; DETDA, diethyltoluenediamine; MA, Matrimid®; TMC, trimesoyl chloride; ULT, ultra-thin [Color figure can be viewed at wileyonlinelibrary.com]
6 of 7
ZHANG ET AL.
measured using films were not used for comparison as film-based membrane formats (e.g., plate-and-frame, spiral-wound) have much lower packing density.42 Remarkably, the ultra-thin CMS hollow fiber membranes offer competitive CO2 permeance and CO2/CH4 separation factors with selected tubular zeolite membranes (e.g., SiCHA, SSZ-13, NaX, KY) under similar testing conditions.43-45 SAPO-34 is possibly the most extensively studied and arguably the highestperforming zeolite membrane material for the CO2/CH4 pair.35,46,47 Compared with tubular SAPO-34 membranes, the ultra-thin CMS hollow fiber membranes (ULT CMS-6F0.5) show competitive CO2 permeance (2,546 GPU) with lower (24.1), yet attractive CO2/CH4 separation factors under similar measurement conditions. By further tuning pyrolysis and hybridization conditions, a closing-up of the gap can be expected. Hollow fiber membranes can provide up to ~10× higher packing density with much lower manufacturing cost than tubular zeolite membranes.48 Notably, ultra-thin CMS hollow fiber membranes are the only hollow fiber membranes that exceed the converted (assuming 1 μm skin layer) Robeson upper bound (Figure 3c).49 By translating high-performance CMS materials to scalable hollow fibers with ultra-thin skin layers, this work can potentially transform membrane-based gas and vapor separations as thin-film composite membranes accomplished for desalination.
4 | CO NC LUSIO NS We report novel ultra-thin CMS hollow fiber membranes derived from tunable dual-layer precursor hollow fibers for sustainable CO2/CH4 separation. These advanced CMS membranes provide outstanding CO2 permeances (1,177–2,546 GPU) one order of magnitude higher than state-of-the-art structures while maintaining highly attractive CO2/CH4 separation factors (~24–37). Additionally, using dual-layer precursor hollow fibers, we have shown that ultra-thin CMS hollow fiber membranes can be economically derived from advanced 6FDA-polyimide precursors. Given the exceptional tunability of CMS materials, it is clear that the ultra-thin CMS hollow fiber platform can be extended to other economically important molecular separations (e.g., CO2/N2, olefins/ paraffins, and hydrocarbon isomers) with a broad impact.
ACKNOWLEDGMENTS The authors gratefully acknowledge financial support from Shell International Exploration and Production, Inc. We thank Dr. Joseph M. Mayne and Dr. P. Jason Williams for useful discussions. W.J.K. acknowledges additional financial support from Office of Basic Energy Science of the U.S. Department of Energy (Grant DEFG02-04ER15510) and equipment support through the Specialty Separations Center at Georgia Institute of Technology.
ORCID Chen Zhang
https://orcid.org/0000-0002-0071-2898
William J. Koros
https://orcid.org/0000-0001-5873-0899
RE FE RE NCE S 1. Sholl DS, Lively RP. Seven chemical separations to change the world. Nature. 2016;532:435-437. 2. Imbrogno J, Belfort G. Membrane desalination: where are we, and what can we learn from fundamentals? Annu Rev Chem Biomol Eng. 2016;7:29-64. 3. Cadotte JE. Interfacially synthesized reverse osmosis membrane. US patent 4277344A1981. 4. Petersen RJ. Composite reverse osmosis and nanofiltration membranes. J Membr Sci. 1993;83:81-150. 5. Geise GM, Paul DR, Freeman BD. Fundamental water and salt transport properties of polymeric materials. Prog Polym Sci. 2014;39:1-42. 6. Koros WJ, Zhang C. Materials for next-generation molecularly selective synthetic membranes. Nat Mater. 2017;16:289-297. 7. Lin JYS. Molecular sieves for gas separation. Science. 2016;353: 121-122. 8. Singh A, Koros WJ. Significance of entropic selectivity for advanced gas separation membranes. Ind Eng Chem Res. 1996;35:1231-1234. 9. Sanyal O, Zhang C, Wenz GB, et al. Next generation membranes— using tailored carbon. Carbon. 2018;127:688-698. 10. Zhang C, Koros WJ. Ultraselective carbon molecular sieve membranes with tailored synergistic sorption selective properties. Adv Mater. 2017;29:1701631. 11. Steel KM, Koros WJ. An investigation of the effects of pyrolysis parameters on gas separation properties of carbon materials. Carbon. 2005;43:1843-1856. 12. Koh D-Y, McCool BA, Deckman HW, Lively RP. Reverse osmosis molecular differentiation of organic liquids using carbon molecular sieve membranes. Science. 2016;353:804-807. 13. Ma XL, Lin YS, Wei XT, Kniep J. Ultrathin carbon molecular sieve membrane for propylene/propane separation. AIChE J. 2016;62: 491-499. 14. Salinas O, Ma XH, Litwiller E, Pinnau I. Ethylene/ethane permeation, diffusion and gas sorption properties of carbon molecular sieve membranes derived from the prototype ladder polymer of intrinsic microporosity (PIM-1). J Membr Sci. 2016;504:133-140. 15. Fu S, Sanders ES, Kulkarni SS, Koros WJ. Carbon molecular sieve membrane structure–property relationships for four novel 6FDA based polyimide precursors. J Membr Sci. 2015;487:60-73. 16. Qiu W, Zhang K, Li FS, Zhang K, Koros WJ. Gas separation performance of carbon molecular sieve membranes based on 6FDAmPDA/DABA (3:2) polyimide. ChemSusChem. 2014;7:1186-1194. 17. Xu L, Rungta M, Koros WJ. Matrimid® derived carbon molecular sieve hollow fiber membranes for ethylene/ethane separation. J Membr Sci. 2011;380:138-147. 18. Bhuwania N, Labreche Y, Achoundong CSK, et al. Engineering substructure morphology of asymmetric carbon molecular sieve hollow fiber membranes. Carbon. 2014;76:417-434. 19. Adams JS, Itta AK, Zhang C, Wenz GB, Sanyal O, Koros WJ. New insights into structural evolution in carbon molecular sieve membranes during pyrolysis. Carbon. 2019;141:238-246. 20. Kiyono M, Williams PJ, Koros WJ. Effect of polymer precursors on carbon molecular sieve structure and separation performance properties. Carbon. 2010;48:4432-4441. 21. Rungta M, Wenz GB, Zhang C, et al. Carbon molecular sieve structure development and membrane performance relationships. Carbon. 2017;115:237-248. 22. Marino T, Blasi E, Tornaghi S, Di Nicolo E, Figoli A. Polyethersulfone membranes prepared with Rhodiasolv® Polarclean as water soluble green solvent. J Membr Sci. 2018;549:192-204. 23. Gnesin I, Gnesin B, Nekrasov A. Ways to reduce the negative consequences of the thermal expansion coefficient mismatch of a carboncarbon composite and a protective silicide coating. J Alloys Compd. 2018;767:803-810.
7 of 7
24. Zhang C, Zhang K, Cao Y, Koros WJ. Composite carbon molecular sieve hollow fiber membranes: resisting support densification via silica particle stabilization. Ind Eng Chem Res. 2018;57:16051-16058. 25. Ekiner OM, Hayes RA, Manos P. Reactive post treatment for gas separation membranes. US patent 5091216A1990. 26. Morgan PW. Interfacial polymerization. In Mark HF, ed. Encyclopedia of Polymer Science and Technology. Hoboken, NJ: Wiley; 2011. 27. Ali Z, Pacheco F, Litwiller E, Wang Y, Han Y, Pinnau I. Ultra-selective defect-free interfacially polymerized molecular sieve thin-film composite membranes for H2 purification. J Mater Chem A. 2018;6:30-35. 28. Ghosh AK, Hoek EMV. Impacts of support membrane structure and chemistry on polyamide–polysulfone interfacial composite membranes. J Membr Sci. 2009;336:140-148. 29. Havas D, Lin H. Optimal membranes for biogas upgrade by removing CO2: high permeance or high selectivity? Sep Sci Technol. 2017;52: 186-196. 30. Jimenez Solomon MF, Bhole Y, Livingston AG. High flux membranes for organic solvent nanofiltration (OSN)—interfacial polymerization with solvent activation. J Membr Sci. 2012;423-424:371-382. 31. Wenz GB, Koros WJ. Tuning carbon molecular sieves for natural gas separations: a diamine molecular approach. AIChE J. 2017;63: 751-760. 32. Baker RW. Future directions of membrane gas separation technology. Ind Eng Chem Res. 2002;41:1393-1411. 33. Sanyal O, Hicks ST, Bhuwania N, et al. Cause and effects of hyperskin features on carbon molecular sieve (CMS) membranes. J Membr Sci. 2018;551:113-122. 34. Xu L, Rungta M, Hessler J, et al. Physical aging in carbon molecular sieve membranes. Carbon. 2014;80:155-166. 35. Zhou R, Ping EW, Funke HH, Falconer JL, Noble RD. Improving SAPO-34 membrane synthesis. J Membr Sci. 2013;444:384-393. 36. Jeon MY, Kim D, Kumar P, et al. Ultra-selective high-flux membranes from directly synthesized zeolite nanosheets. Nature. 2017;543:690. 37. Rangnekar N, Mittal N, Elyassi B, Caro J, Tsapatsis M. Zeolite membranes - a review and comparison with MOFs. Chem Soc Rev. 2015; 44:7128-7154. 38. Sanders DE, Smith ZP, Guo RL, et al. Energy-efficient polymeric gas separation membranes for a sustainable future: a review. Polymer. 2013;54:4729-4761. 39. Baker RW, Low BT. Gas separation membrane materials: a perspective. Macromolecules. 2014;47:6999-7013.
ZHANG ET AL.
40. Carreon ML, Li S, Carreon MA. AlPO-18 membranes for CO2/CH4 separation. Chem Commun. 2012;48:2310-2312. 41. Kanezashi M, O'Brien-Abraham J, Lin YS, Suzuki K. Gas permeation through DDR-type zeolite membranes at high temperatures. AIChE J. 2008;54:1478-1486. 42. Baker RW. Membrane Technology and Applications. 2nd ed. West Sussex, England, UK: Wiley; 2004. 43. Kida K, Maeta Y, Yogo K. Preparation and gas permeation properties on pure silica CHA-type zeolite membranes. J Membr Sci. 2017;522: 363-370. 44. Kosinov N, Auffret C, Gücüyener C, et al. High flux high-silica SSZ-13 membrane for CO2 separation. J Mater Chem A. 2014;2:13083-13092. 45. Hasegawa Y, Tanaka T, Watanabe K, Jeong B-H, Kusakabe K, Morooka S. Separation of CO2-CH4 and CO2-N2 systems using ionexchanged FAU-type zeolite membranes with different Si/Al ratios. Korean J Chem Eng. 2002;19:309-313. 46. Li SG, Falconer JL, Noble RD. Improved SAPO-34 membranes for CO2/CH4 separations. Adv Mater. 2006;18:2601-2603. 47. Li S, Martinek JG, Falconer JL, Noble RD, Gardner TQ. High-pressure CO2/CH4 separation using SAPO-34 membranes. Ind Eng Chem Res. 2005;44:3220-3228. 48. Koros WJ, Lively RP. Water and beyond: expanding the spectrum of large-scale energy efficient separation processes. AIChE J. 2012;58: 2624-2633. 49. Robeson LM. The upper bound revisited. J Membr Sci. 2008;320: 390-400.
SUPPORTING INF ORMATION Additional supporting information may be found online in the Supporting Information section at the end of this article.
How to cite this article: Zhang C, Kumar R, Koros WJ. Ultrathin skin carbon hollow fiber membranes for sustainable molecular separations. AIChE J. 2019;65:e16611. https://doi. org/10.1002/aic.16611