R ES E A RC H

◥ molecular sieves, ceramics, zeolites, various nano-

REVIEW SUMMARY materials (e.g., graphene, graphene oxide, and
metal organic frameworks), and their mixtures
with polymers have been explored. Simultaneously,
MEMBRANES
global challenges such as climate change and
◥
rapid population growth
Maximizing the right stuff: The ON OUR WEBSITE

Read the full article

stimulate the search for
efficient water purification

trade-off between membrane at http://dx.doi.

org/10.1126/
science.aab0530
and energy-generation tech-
nologies, many of which
are membrane-based. Ad-
permeability and selectivity ..................................................

clude wastewater reuse from shale gas extraction

ditional driving forces in-

Ho Bum Park, Jovan Kamcev, Lloyd M. Robeson, and improvement of chemical and petrochemical
Menachem Elimelech, Benny D. Freeman* separation processes by increasing the use of
light hydrocarbons for chemicals manufacturing.

BACKGROUND: Synthetic membranes are through a membrane. Based on increasing OUTLOOK: Opportunities for advancing mem-
used for desalination, dialysis, sterile filtra- molecular understanding of both biological branes include (i) more mechanically, chem-
tion, food processing, dehydration of air and and synthetic membranes, key design criteria ically, and thermally robust materials; (ii)
other industrial, medical, and environmental for new membranes have emerged: (i) prop- judiciously higher permeability and selec-
applications due to low energy requirements, erly sized free-volume elements (or pores), (ii) tivity for applications where such improve-

Downloaded from https://www.science.org at Open University of Israel on November 22, 2021

compact design, and mechanical simplicity. narrow free-volume element (or pore size) dis- ments matter; and (iii) more emphasis on
New applications are emerging from the water- tribution, (iii) a thin active layer, and (iv) highly fundamental structure/property/processing
energy nexus, shale gas extraction, and en- tuned interactions between permeants of in- relations. There is a pressing need for mem-
vironmental needs such as carbon capture. terest and the membrane. Here, we discuss the branes with improved selectivity, rather than
All membranes exhibit a trade-off between permeability/selectivity trade-off, highlight sim- membranes with improved permeability, es-
permeability—i.e., how fast molecules pass through ilarities and differences between synthetic and pecially for water purification. Modeling at all
a membrane material—and selectivity—i.e., to biological membranes, describe challenges for length scales is needed to develop a coherent
what extent the desired molecules are sepa- existing membranes, and identify fruitful areas molecular understanding of membrane prop-
rated from the rest. However, biological mem- of future research. erties, provide insight for future materials
branes such as aquaporins and ion channels design, and clarify the fundamental basis
are both highly permeable and highly selective. ADVANCES: Many organic, inorganic, and hy- for trade-off behavior. Basic molecular-level
Separation based on size difference is common, brid materials have emerged as potential mem- understanding of thermodynamic and diffu-
but there are other ways to either block one branes. In addition to polymers, used for most sion properties of water and ions in charged
component or enhance transport of another membranes today, materials such as carbon membranes for desalination and energy ap-
plications such as fuel cells is largely incom-
plete. Fundamental understanding of membrane
structure optimization to control transport
of minor species (e.g., trace-organic contam-
inants in desalination membranes, neutral
compounds in charged membranes, and heavy
hydrocarbons in membranes for natural gas
separation) is needed. Laboratory evaluation
of membranes is often conducted with highly
idealized mixtures, so separation performance
in real applications with complex mixtures is
poorly understood. Lack of systematic under-
standing of methodologies to scale promising
membranes from the few square centimeters
needed for laboratory studies to the thousands
of square meters needed for large applications
stymies membrane deployment. Nevertheless,
opportunities for membranes in both existing
From intrinsic permeability/selectivity trade-off to practical performance in membranes. and emerging applications, together with an
Polymer membranes for liquid and gas separation applications obey a permeability/selectivity expanding set of membrane materials, hold
trade-off—highly permeable membranes have low selectivity and vice versa—largely due to great promise for membranes to effectively
broad distributions of free-volume elements (or pores in porous membranes) and nonspecific
interactions between small solutes and polymers. We highlight materials approaches to overcome
address separations needs.
▪
this trade-off, including the development of inorganic, isoporous, mixed matrix, and aquaporin
membranes. Further, materials must be processed into thin, typically supported membranes,
Author affiliations are available in the full article online.
fashioned into high surface/volume ratio modules, and used in optimized processes. Thus, factors
*Corresponding author. Email: [email protected]
that govern the practical feasibility of membranes, such as mechanical strength, module design, Cite this article as H. B. Park et al., Science 356, eaab0530
and operating conditions, are also discussed. (2017). DOI: 10.1126/science.aab0530

Park et al., Science 356, 1137 (2017) 16 June 2017 1 of 1

R ES E A RC H

◥ li, j, depends on the ratio of gas molecule di-

REVIEW ameters, li, j = (dj /di)2 − 1, where dj and di are
the kinetic diameters of the larger and smaller
gases, respectively, and upper-bound behavior
MEMBRANES was found for all gas pairs reported. The front
factor, bi, j, depends on gas solubility, li, j, and an

Maximizing the right stuff: The adjustable constant, f, related to the average dis-
tance between polymer chains and chain stiffness.
This model was based on five hypotheses: (i) gas
trade-off between membrane permeation occurs via the solution-diffusion mech-
anism (see Box 1); (ii) gas diffusion is an activated

permeability and selectivity process, described by an Arrhenius equation;

(iii) the activation energy of diffusion, ED, of
a gas molecule depends on the square of its
Ho Bum Park,1 Jovan Kamcev,2 Lloyd M. Robeson,3 effective size; (iv) a universal linear free-energy
Menachem Elimelech,4 Benny D. Freeman2* relation exists between ED and the front factor
in the Arrhenius equation; and (v) gas solubility
Increasing demands for energy-efficient separations in applications ranging from water depends on gas molecule condensability, as ex-
purification to petroleum refining, chemicals production, and carbon capture have pressed, for example, by the Lennard-Jones well
stimulated a vigorous search for novel, high-performance separation membranes. Synthetic depth of the gas molecule.
membranes suffer a ubiquitous, pernicious trade-off: highly permeable membranes lack The 1991 upper-bound results (21) were re-
selectivity and vice versa. However, materials with both high permeability and high visited in 2008 with a much larger database and
selectivity are beginning to emerge. For example, design features from biological membranes notable advances in the search for more per-

Downloaded from https://www.science.org at Open University of Israel on November 22, 2021

have been applied to break the permeability-selectivity trade-off. We review the basis for the meable and selective polymers (23). In most cases,
permeability-selectivity trade-off, state-of-the-art approaches to membrane materials only modest shifts in the upper bound were
design to overcome the trade-off, and factors other than permeability and selectivity that observed despite many studies between 1991
govern membrane performance and, in turn, influence membrane design. and 2008 aimed at preparing materials to exceed
the upper bound. This point is illustrated in Fig.

S
1A where the 1991 and 2008 upper bounds are
ynthetic membranes, based largely on ing scalability, and small footprint (1–3). However, superimposed. Notably, and in agreement with
polymers, are widely used in gas separa- all synthetic membranes are subject to a trade-off the upper-bound model (22), the slopes, li, j, of
tions (e.g., air dehydration; O2/N2 separa- between permeability and selectivity, as well as the upper bound did not change, but the posi-
tion; hydrogen purification; and CO2, H2S, practical challenges such as fouling, degrada- tion of the upper bound, bi, j, moved as shown
and higher hydrocarbon removal from natu- tion, and material failure that limit their use. in Fig. 1A for all gas pairs considered.
ral gas), water purification (e.g., desalination, ultra- The most significant changes in bi, j values
pure water production, drinking water treatment, Origin of the permeability-selectivity were with He-based gas pairs (specifically He/H2),
and municipal and industrial wastewater treat- trade-off where glassy perfluorinated polymers came to
ment and reuse) (1–3), bioprocessing (e.g., sterile The commercialization of polymeric membranes dominate the upper bound (23). Perfluoropolymers
filtration, protein concentration, and buffer ex- for gas separations in the late 1970s catalyzed a exhibit unique gas solubility characteristics rela-
change) (4, 5), medical applications (e.g., dialy- sustained search for materials with better sepa- tive to aromatic and other hydrocarbon polymers,
sis, blood oxygenation, and drug delivery) (6), ration properties. As the database of gas per- for reasons still unresolved at a fundamental level
food processing (e.g., beer and wine clarification meation properties on various materials expanded, (24, 25). These solubility characteristics account
and demineralization of whey, juices, and sugar) researchers discovered a trade-off between gas for the presence of perfluoropolymers at the up-
(7, 8), chemicals production (e.g., chlor-alkali pro- permeability, Pi (see Box 1 for definition), and per bound for some gas pairs.
cess to produce chlorine and sodium hydroxide) selectivity, ai,j = Pi /Pj, where i represents the For other gas pairs, bi, j values changed due to
(8), batteries (9), and fuel cells (10, 11). Potential more permeable gas of the i, j gas pair (e.g., i = introduction of new materials. Two classes of poly-
applications include energy generation (12, 13); O2 and j = N2 in air separation) (21). During the mers—PIMs (polymers of intrinsic microporosity,
energy storage (14); environmental applications 1980s, permeability data on six common gases e.g., polybenzodioxanes) (26) and TRs (thermally
such as carbon capture (15) and selective removal (He, H2, O2, N2, CO2, and CH4) were compiled, rearranged polymers, e.g., polybenzoxazoles)
of ions (e.g., nitrates and phosphates) contributing and the trade-off relationship between perme- (27)—gave exceptional performance. For CO2/CH4,
to surface water eutrophication (16); organic ability and selectivity was analyzed. Polymers several TR polymers significantly exceeded the up-
solvent recovery (17); pharmaceutical purification having the highest selectivity at a given perme- per bound. The bases for their unusually high
(18); catalyst recovery (18); and membrane crys- ability lay near or on a line, called the upper permeability/selectivity combinations are (i) high gas
tallization (19), distillation (3, 19), and emulsifi- bound, obeying the following relation (21) solubility, an inherent characteristic of high free-
cation (20). In many applications, membranes li;j volume glassy polymers such as PIMs and TR
are favored over other processes due to advan- ai;j ¼ bi;j Pi ð1Þ polymers; (ii) high gas diffusion coefficients,
tages in energy efficiency, simplicity, manufactur- which are also a consequence of high free vol-
where li,j and bi,j are parameters depending on ume; and (iii) for several gas pairs (e.g., O2/N2
1
Department of Energy Engineering, Hanyang University, the gas pair. An example for air separation (i.e., and CO2/CH4), unusually high gas diffusion se-
Seoul 04763, Republic of Korea. 2Department of Chemical O2/N2 separation) is shown in Fig. 1A. More per- lectivity, suggesting size and size distribution
Engineering, Texas Materials Institute, Center for Research in
Water Resources, and Center for Energy and Environmental
meable polymers tend to have lower selectivity, of free-volume elements in a range particularly
Research, The University of Texas at Austin, 10100 Burnet and vice versa. This study (21) became the stan- suitable for separating these gas molecules based
Road, Building 133 (CEER), Austin, TX 78758, USA. dard against which permeability and selectivity on sub-Angstrom differences in effective gas mol-
3
Department of Materials Science and Engineering, Lehigh values for new and improved membrane mate- ecule size (28).
University, 1801 Mill Creek Road, Macungie, PA 18062, USA.
4
Department of Chemical and Environmental Engineering,
rials were measured. A comparison of rubbery versus glassy (i.e.,
Yale University, New Haven, CT 06520-8286, USA. A theoretical model of the permeability/ flexible versus rigid) polymers (29) revealed that
*Corresponding author. Email: [email protected] selectivity trade-off (22) revealed that the slope, gas solubilities are uniformly higher in glassy

Park et al., Science 356, eaab0530 (2017) 16 June 2017 1 of 10

R ES E A RC H | R E V IE W

than rubbery polymers when compared at equiv-

alent permeabilities, due to additional sorption
of gas into the nonequilibrium excess volume
inherently present in glassy polymers (i.e., dual-
mode sorption) (30). Thus, glassy polymers dom-
inate the upper bound mainly due to higher gas
solubility than rubbery polymers (29). When
compared at equivalent free-volume levels, rub-
bery and glassy polymers have very similar dif-
fusion selectivities (29). That is, for most polymers,
diffusion selectivity differences between glassy
and rubbery polymers only modestly favor glassy
polymers, in contrast to generalizations previously
reported and accepted in the literature.
Various approaches to exceed the upper bound
have been explored, including surface modifica-
tion (31, 32), facilitated transport (33), phase-
separated polymer blends (34), and mixed-matrix
membranes (MMMs) (35). Although the upper-
bound concept as originally formulated only applies
to homogeneous polymer membranes, compar-
ing permeability and selectivity data on upper-
bound plots remains a popular way to gauge

Downloaded from https://www.science.org at Open University of Israel on November 22, 2021

membrane material performance. In addition
to the upper-bound pairs possible from the six
gases noted above, upper-bound relationships
have been reported for other important gas pairs,
such as propylene/propane (36), ethylene/ethane
(37), and N2/NF3 (38).
Other examples of such trade-off behavior
have emerged in virtually all synthetic polymer
membranes, including desalination membranes
(39) (Fig. 1B), forward osmosis membranes (40–42),
porous ultrafiltration (UF) membranes (43) (Fig.
1C), polymer electrolyte membranes for fuel cells Fig. 1. Upper-bound relations in polymer membranes. (A) O2/N2 separation (23); (B) water/salt
(44) (Fig. 1D), pervaporation membranes (45), separation (39); (C) protein/water separation in porous ultrafiltration membranes [1/Sa is the
and ion-exchange membranes (46). The funda- separation factor of bovine serum albumin (BSA) from water and other small solutes (e.g., salts and
mental physics of small-molecule transport through sugars)], and hydraulic permeability is the rate of transport through the membrane of water and
dense, nonporous polymers (e.g., for gas separation any solutes not retained by the membrane (43); and (D) polymer electrolyte membranes, where ionic
and water/salt separation (47)) are different from conductivity for membranes at a given level of water uptake (i.e., water sorption) is apparently
those of proteins through porous UF membranes limited by an upper-bound type relationship (44).
and ion transport through charged polymers
in electrically driven separations (3, 5, 8, 43).
Nevertheless, upper-bound behavior is observed
in all cases, suggesting the generality of this Box 1. Solution-diffusion transport mechanism in nonporous polymers. Gas permeability
phenomenon in both dense and porous mem- is defined as Pi = NiL/Dp, where Ni is the steady-state flux of gas i through the membrane,
branes. Additionally, materials used to form L is membrane thickness, and Dp is the transmembrane pressure difference (or partial pressure
either porous or dense membranes do not nec- difference, in the case of gas mixtures). The three-step solution-diffusion mechanism is
essarily have to be polymeric for the resulting the accepted framework for small-molecule transport in nonporous polymers (120). A penetrant
membranes to exhibit permeability-selectivity molecule (i) dissolves into the upstream (i.e., high concentration) side of a membrane, (ii) diffuses
trade-offs (48). through the membrane, and (iii) desorbs from the downstream (i.e., low concentration) side
Common features of the dense polymer mem- of a membrane. The second step is the rate-limiting step in all current membranes, and the
branes mentioned above are a broad distribution rate-limiting step in gas diffusion through polymers is the local segmental dynamics of the
of free-volume element sizes, as shown in Fig. 2, polymer chains, which open transient gaps (i.e., free-volume elements) through which gas
A and B, and relatively nonspecific interactions diffusion occurs. In this framework, permeability is written as Pi = SiDi, where Si and Di are
governing solubility of small molecules in poly- the penetrant solubility and diffusivity in the polymer, respectively, and ai,j = (Si/Sj) × (Di/Dj),
mers (3, 27, 30). Changing a polymer structure where Si/Sj and Di/Dj are the solubility and diffusivity selectivities, respectively. The same
to give more rapid permeation often enhances mechanism applies to water and salt transport in RO membranes and other membrane
permeability of larger species more than smaller processes involving nonporous materials (120). The gas permeability unit is the barrer, where
species, resulting in increased permeability but 1 barrer ¼ 1010 cm
3
ðSTPÞ cm
. Gas flux is often normalized by pressure to give permeance, Pi /L =
cm2 s cmHg
decreased selectivity. This pattern has been as- cm ðSTPÞ
Ni/Dp, which is reported in gas permeation units (GPU), where 1 GPU ¼ 106 cm
3
2 s cmHg. In
cribed to a lack of ability to increase free-volume
element size while simultaneously narrowing RO desalination membranes, the equivalent expression for water flux, Nw, is Nw = A(Dp – Dp),
free-volume element size distribution (27). Mem- where A (i.e., permeance) contains the membrane permeability to thickness ratio, Dp is the
branes with very narrow pore size distributions applied hydrostatic pressure, and Dp is the osmotic pressure difference across the membrane,
are “isoporous” membranes (i.e., those with uni- generated by the difference in salt concentration on either side of the membrane (121).
form pore sizes), as shown in Fig. 2, C and D. Such

Park et al., Science 356, eaab0530 (2017) 16 June 2017 2 of 10

R ES E A RC H | R E V IE W

membranes can have both high permeability

and high selectivity because the isoporous geome-
try contributes to membranes with no tortuosity
(high permeability) and precise, uniform pore
size (high selectivity). However, practical devel-
opment of these membranes is in its infancy.

Design approaches for improved

permeability and selectivity
Biological membranes, such as potassium ion
channels and aquaporins (Fig. 2, E and F), have
extremely high selectivity-permeability combina-
tions, which has stimulated recent efforts aimed
at (i) direct incorporation of such structures into
membranes (49), (ii) theoretical studies aimed at
understanding optimal structures (Fig. 2G) that
might yield high permeability and selectivity, or
(iii) synthetic membrane structures that mimic
or are inspired by one or more elements of bi-
ological membranes. So far, incorporation of, for
example, aquaporins into membranes has been
done via assimilation of aquaporins into vesicles
and integration of the resulting vesicles into

Downloaded from https://www.science.org at Open University of Israel on November 22, 2021

membranes, but there are no successful, repro-
ducible studies demonstrating that this strategy
can produce highly selective membranes (48).
Thus, much remains uncertain about their ability
to be processed into the large-scale, defect-free
structures required for practical applications or
whether they can maintain adequate transport
and selectivity properties upon exposure to com-
plex, real-world feed mixtures for extended periods
of time. We return to these points below.
Certain structural changes in polymers used
for gas separation, such as thermal rearrange-
ment, narrow an otherwise broad free-volume
element size distribution. This contributes to
higher permeability-selectivity combinations, but
such materials still have a wide distribution of
free-volume elements relative to biological mem-
branes (27). Block copolymers innately self-assemble
into well-defined structures with regular perio-
dicity. This phenomenon has been harnessed to
prepare ~15-nm-diameter isoporous membranes
with complete rejection of virus particles and
high water flux (50). This concept was extended
to prepare isoporous UF membranes via non-
solvent-induced phase inversion (NIPS), an in-
dustrial process used to produce many current
membranes, providing a potential route to large-
scale production of such structures (51). Optimi-
zation of this process led to water permeances
of 3200 L m−2h−1bar−1, an order of magnitude
higher than conventional NIPS membranes of
the same average pore size, coupled with a se-
lectivity of 87 for separating bovine serum albumin
(MW = 67 kDa, diameter ~6.8 nm) from globulin-g
(MW = 150 kDa, diameter ~14 nm), proteins too
close in size to be separated by conventional UF
Fig. 2. Evolution of cavity (or pore) size distributions in membranes. Cavity (or pore) size membranes (52). The high water permeance in
distributions in separation membranes, ranging from (A) broad cavity-size distributions in dense such isoporous membranes was due to higher
polymers, such as those used in gas separations (113) and (B) water purification (e.g., porous porosity and lower tortuosity at similar pore size
ultrafiltration membranes) (114) to narrow pore size distributions in (C) isoporous UF membranes relative to conventional UF membranes. Typically,
formed via self-assembly of block copolymers (~34-nm pore diameter) (52), (D) graphene such self-assembled structures can be used to make
nanomesh having ~30-nm pores prepared via block copolymer lithography (115), (E) potassium ion isoporous membranes with pore sizes ranging
channel (116), (F) aquaporin (93), and (G) the effect of hourglass pore design parameters (pore from 3 to 20 nm, in the range of UF membranes.
opening angle, a, and aspect ratio, L/a) on water permeability (117). Using mixtures of different block copolymers,

Park et al., Science 356, eaab0530 (2017) 16 June 2017 3 of 10

R ES E A RC H | R E V IE W

the pore size was reduced to ~1.5 nm, while ity (130 at 120°C), and a strong temperature de- 16,000 to meet water purity requirements, and
still maintaining high water flux (53), yielding pendence of H2 permeance, implying activated aquaporins have essentially infinite selectivity
nanofiltration (NF) membranes, thereby poten- molecular transport through the ZIF-8 pores. When and a permeance of 1.6 × 106 L m−2h−1bar−1 per
tially opening a large set of practically important defects were sealed with poly(dimethylsiloxane), aquaporin (68).
separations (e.g., dye removal from textile waste- the H2/C3H8 selectivity increased to 370, but H2 Graphene oxide (GO) has a high density of
water). The detailed formation mechanism of permeance decreased to 750 GPU. ZIF-8 has an oxygen-containing functional groups (e.g., hydroxyl,
such membranes is still under debate (54). Block effective aperture (i.e., pore) size of ~4.0 Å, so it epoxy, and carboxyl) and some vacancy defects
copolymers, such as those used to prepare the may be of interest for propylene/propane separa- in the 2D carbon lattice. GO can be prepared
original isoporous membranes, are expensive tion, a large-scale, highly energy-intensive sepa- inexpensively on a large scale by oxidation and
relative to traditional polymers used in water- ration (36). More recently, Kwon et al. reported exfoliation of graphite. GO sheets can be readily
purification membranes. If isoporous membrane heteroepitaxially grown ZIF membranes consisting dispersed in water or organic solvents, providing
were made entirely of such block copolymers, of two different ZIF layers (60). Submicrometer- facile means for membrane fabrication via well-
cost would likely be a considerable roadblock for thick ZIF-67 membranes were heteroepitaxially developed membrane-coating technology using
such materials, restricting their use to high-value, grown from ZIF-8 seed layers on alumina sup- GO dispersions (69, 70). Differing from nano-
low-volume separations (e.g., biomedical applica- ports. Addition of a ZIF-8 overlayer (~300 nm) porous graphene, adjustable 2D nanochannels
tions) rather than large-scale drinking or waste- on a ZIF-67 membrane led to very high propylene/ between adjacent GO sheets can be used for
water purification applications. propane separation factor (~200) with high pro- selective molecular transport. GO membranes
Inorganic materials offer efficient ways to pylene permeability, well above the upper bound can have high water-transport rates (71), because
tailor pore sizes and shapes and achieve narrow for both polymer and carbon membranes (36). the oxidized domains act as spacers to separate
pore size distributions, potentially leading to high However, since alumina is a brittle ceramic, this adjacent GO sheets and facilitate water molecule
permeability and/or high selectivity. For example, approach does not solve the basic mechanical intercalation, and the pristine graphitic domains
metal-organic frameworks (MOFs) have been property challenge facing such materials, ren- in GO sheets create a network of capillaries allow-
studied for gas separation and other related dering scale-up of these membranes unlikely. ing almost frictionless flow of water, similar to

Downloaded from https://www.science.org at Open University of Israel on November 22, 2021

applications (55). MOFs are synthesized by con- Graphene and other two-dimensional (2D) atom- water transport through carbon nanotubes (72).
necting inorganic and organic units via covalent ically thin materials (e.g., graphdiyne, graphyne, In addition, GO membranes can display sieving
bonds to form permanently porous crystalline hBN, and MoS2, 2D polymers based on polypheny- properties in aqueous solutions (69), blocking
structures. Numerous MOFs with various well- lene, porphyrin, and cyclohexa-m-phenylene, etc.) all solutes with hydrated radii larger than 0.45 nm,
defined pore sizes and channels have been reported have attracted attention because of the minimum which is inadequate for desalination applications,
(56). Due to their high porosity (~50% of the possible thickness, high mechanical strength, given that, for example, a hydrated sodium ion
crystal volume); high surface area (~10,000 m2g−1), chemical stability, and ability to create selec- is smaller than this.
which exceeds traditional inorganic materials tive nanoscale pores in their rigid 2D lattices GO membranes have also been explored for
(e.g., zeolites and carbon molecular sieves); and (61). Two distinct strategies to use such materials gas separations. Li et al. (73) fabricated ultrathin
extensive ability to chemically tailor MOFs, they (e.g., graphene) for membranes have been reported: GO membranes (e.g., ~1.8 nm thick) that exhibited
have been widely studied as membrane mate- (i) creating nanopores (or defects) in the basal plane high H2/CO2 and H2/N2 selectivity. Kim et al. (63)
rials. Crystalline MOF membranes on porous (62) or (ii) tailoring the 2D nanochannels inherently demonstrated that hydrated thin GO membranes
substrates have been explored, particularly for present between stacked 2D sheets (e.g., graphene can be used as CO2-selective membranes because
gas separation applications. Like zeolite mem- oxide or reduced graphene oxide) (63). Simulations permeance of all gases, except CO2, decreased
branes, continuous MOF membranes are prepared suggest that monolayer porous graphene could be as feed humidity increased. Condensed water
primarily by in situ and seeded-secondary growth highly permeable and selective, with higher separa- molecules in the pores, or between GO sheets,
methods. More recently, zeolitic imidazole frame- tion performance than polymer membranes. Gas hindered transport of relatively noncondensable
works (ZIFs), a subclass of MOFs, have been of separation and water/ion separation properties small molecules such as N2. As with graphene
interest for gas separations, mainly because their have been measured using such porous graphene and other 2D atomically thin materials discussed
pore size is similar to the size of small gas mol- membranes (62, 64). Precisely tuning the nano- before, fabrication of defect-free, large-scale GO
ecules and they can exhibit high chemical and channels between stacked 2D graphene sheets, membranes for liquid or gas separation is ex-
thermal stability (57, 58). or controlling the pore size and achieving high tremely challenging, due to their poor mechanical
Although continuous, self-supporting thin films porosity with large-area graphene membranes, properties, which markedly limits the practicality
of MOFs would be preferred for membrane ap- pose considerable technological challenges due to of such membranes.
plications, they are not mechanically robust enough defect control in raw graphene monolayers. There- On the laboratory scale, a 10 cm2 (10−3 m2)
to form large surface area membranes. Conse- fore, such materials cannot be used for industrial, membrane sample is often sufficient to generate
quently, they are fabricated on mechanically stable large-area membranes without major breakthroughs initial characterization data. However, industrial
porous supports (58). Additionally, such crystalline in large-scale, reproducible manufacturing. production of membranes for large-scale appli-
materials in the thin-film state contain intrinsic or As an example of the properties of such ma- cations can require 1000 to 100,000 m2 of mem-
extrinsic defects (e.g., grain boundaries) between terials, Surwade et al. created small nanopores branes, all of which must be prepared with defect
single crystalline domains, leading to nonselective (~1 nm) via oxygen plasma irradiation at a density densities, for gas separations, below about 1 cm2
flow (59). Many studies focus on minimizing such of ~1012 cm−2 in monolayer graphene membranes of defects for every 105 cm2 of membrane surface
defects to achieve high selectivity and reducing the placed on a microscale aperture. The resulting area (74, 75). Such a 106 to 108 scale-up in the size
thickness of the active layer to achieve high per- membranes exhibited high water permeability of defect-free membranes is often a major stumbl-
meance. For example, Nair and colleagues. coated (64). In one sample, the driving force normalized ing block for introduction of new materials as
ZIF-8 onto polymeric hollow fibers using two- water flux (i.e., permeance) was 250 L m−2h−1bar−1, membranes and is exacerbated if the new materials
solvent interfacial synthesis (58). They used highly whereas a state-of-the-art reverse osmosis (RO) cannot be readily transformed into thin, defect-
gas-permeable, porous poly(amide-imide) hollow membrane would have a permeance of about free membranes using conventional membrane-
fibers as a support and prepared a continuous ZIF-8 2.3 L m−2h−1bar−1 (65–68). However, this per- formation methods (3).
membrane on the lumen surface of the hollow forated graphene membrane had poor salt/water One popular approach to address difficulties
fibers via interfacial microfluidic membrane pro- selectivity, ranging from about 5 to 200 (the wide in preparing large-surface-area, defect-free, ultra-
cessing. These membranes showed clear molecular range presumably due to sample-to-sample vari- thin membranes of promising nanomaterials (e.g.,
sieving properties—high H2 permeance [3000 gas ability). For comparison, a state-of-the-art RO carbon nanotubes, graphene, zeolites, and MOFs)
permeation units (GPU)], high H2/C3H8 selectiv- membrane requires a salt/water selectivity of is the use of mixed-matrix membranes (MMMs).

Park et al., Science 356, eaab0530 (2017) 16 June 2017 4 of 10

R ES E A RC H | R E V IE W

MMMs commonly consist of a dispersed nano- compared the CO2/CH4 selectivity with MMMs,
material phase and a continuous polymer matrix, including a bulky-type MOF, nanosized MOF,
hoping to marry the high intrinsic permeability and MOF nanosheets, based on the same MOF
and separation properties of advanced nano- material (85). Only MOF nanosheets in the MMM
materials with the robust processing and mechan- showed improved selectivity and no significant
ical properties of polymers (Fig. 3). Most studies permeability loss relative to the neat polymer
involving MMMs have focused on gas separation membrane. Similar effects can be observed in
membranes (76), in which molecular sieving par- other nanosheet-incorporated MMMs. When GO
ticles such as zeolites, carbons, and MOFs have nanosheets are incorporated into polyether-based
been incorporated into a polymer matrix to alter copolymer membranes for CO2 separation, CO2/N2
their performance. More recently, nanotubes and selectivity increased significantly compared with
nanosheets have been considered as the dispersed the pristine polymer membrane with no perme-
phase (77, 78). Koros and colleagues theoretically ability penalty, even at low loadings (e.g., below
analyzed the potential of MMMs, predicting gas 0.1 weight %) (86).
separation performance beyond Robeson’s upper
bound (particularly for O2/N2 separation) (35). Comparison of transport characteristics
However, most experimental results fall below of synthetic and biological membranes
the theoretical maximum because of unfavorably Fig. 3. In designing mixed-matrix membranes Unlike synthetic membranes, biological mem-
large interfacial gaps between the dispersed to overcome the upper bound, compatibility branes exhibit both high permeability and high
nanomaterials and polymer matrix and relatively between filler and polymer, filler particle selectivity. For example, the potassium ion chan-
limited dispersed phase loading due to particle size and shape, and homogeneous filler dis- nel in cell membranes is thousands of times more
agglomeration (79). As such, much of the effort tribution are important factors. (Case 1) permeable to potassium than sodium ions, despite
to enhance separation performance in MMMs Molecular sieving fillers (e.g., CMS and zeolites) the smaller ionic (i.e., crystallographic) size of

Downloaded from https://www.science.org at Open University of Israel on November 22, 2021

focuses on reducing the incompatibility between often lead to selectivity increases and perme- sodium, and exhibits permeation rates (~108 ions/s)
inorganic materials and organic polymers, while ability decreases (35). (Case 2) Molecular approaching the diffusion limit (87). Polymer
achieving homogeneous dispersions at high load- sieving fillers with nano size or nanosheet membranes often exhibit little selectivity for
ing levels (80, 81). However, despite well over a shapes (e.g., MOFs nanocrystals or 2D nano- ions of like valence and transport such ions or-
decade of sustained efforts to develop large-scale, sheets) can improve both permeability and ders of magnitude more slowly. For example,
practical MMMs, there are no commercial ex- selectivity (55). (Case 3) Fillers with interfacial McGrath et al. reported NaCl and KCl perme-
amples of such membranes. Uniformly dispers- voids, even when homogeneously dispersed, ability coefficients of 3.8 × 10−8 cm2 s−1 and 4.4 ×
ing nanomaterials in thin polymer selective layers can result in increased permeability and 10−8 cm2 s−1, respectively, in a di-sulfonated
(e.g., <100 nm thick) via continuous processes decreased selectivity (55). poly(arylene ether sulfone) membrane (88). KCl
(e.g., hollow fiber spinning) to make reproducible, permeation was more rapid than NaCl permea-
large-surface-area membranes, without introduc- tion by a factor of ~1.2, whereas in potassium ion
ing selectivity-destroying defects while maintain- number of nonselective pathways for gas trans- channels, K+ is thousands of times more perme-
ing sufficient mechanical properties to permit port, resulting in high ethylene/ethane separa- able than Na+ (87, 89).
manufacturing and use of the resulting mem- tion performance with increasing MOF content Rates of K+ transport in ion channels may be
branes, has proven an arduous undertaking due (55). Incorporation of larger MOFs (i.e., 100 to up to 108 ions/s (87). The diameter of the selec-
to the large number of processing variables affect- 200 nm) did not show performance improve- tivity filter is approximately 0.3 nm, so the area
ing the final material structure. ment relative to that of membranes containing available for ion transport is 0.071 nm2. Thus, the
In principle, in the absence of significant inter- smaller MOF nanocrystals. K+ flux through the selectivity filter is 0.235 mol/
facial gaps and with homogeneous dispersion, In addition to the factors discussed above, filler (cm2 s). A typical extracellular K+ concentra-
inclusion of inorganic materials can improve shape and orientation are important. Anisotrop- tion is 4 mM, and a typical intracellular K+
membrane performance by increasing selectivity, ically shaped (e.g., sheetlike) inorganic materials concentration is 155 mM (90). The length of the
permeability, or both. Interfacial compatibility, may be better than isotropically shaped materials selectivity filter is 1.2 nm (87). On this basis, the
interaction energy between nanomaterials and for mixed-matrix membranes, because MMMs permeability coefficient of K+, PK þ , is estimated
polymers, type of nanomaterial (e.g., zeolites, containing nanoscopically small, thin nanosheets as follows (39, 91): PK þ ¼ JK þ L=DCK þ , where
silica, carbon molecular sieves, activated carbon, can allow high separation performance at much JK þ is the flux of K+, L is the length of the
MOFs, carbon nanotubes, metal oxides, meso- lower loading than that required for isotropic selectivity filter, and DCK þ is the K+ concentra-
porous materials, nonporous materials, and nano- fillers. For example, 2D nanomaterials [i.e., those tion difference between the extracellular and
sheets), nanomaterial concentration, orientation with one dimension on the nanoscale (e.g., MOF intracellular compartments. Thus, the K+ perme-
of anisotropic nanomaterials, and permeation nanosheets and graphene oxide)], rather than ability of an ion channel, PK þ , is 1.9 × 10−4 cm2 s−1,
properties of the polymer continuous phase all 0D [all three dimensions on the nanoscale (e.g., ~4 orders of magnitude faster than that reported
influence membrane performance (77, 82, 83). carbon molecular sieves, silica, and MOF crystals)] by McGrath et al.
Scalable processing of MMMs is less difficult or 1D [two dimensions on nanoscale (e.g., carbon As a further comparison, the Na+ permeability
than processing membranes made solely from nanotubes and aluminosilicate nanotubes)] nano- of a Nafion synthetic ion exchange membrane
novel nanomaterials (e.g., MOFs and carbon materials, may be preferred because they can be was calculated using data reported by Pintauro
nanotubes) but can be much more difficult than incorporated into ultrathin membranes such as and Bennlon (92). The Na+ concentration in
processing polymers alone. Successfully includ- the selective skin layers of thin-film composite Nafion equilibrated with a 1 M NaCl solution
ing such nanomaterials into thin, selective poly- or hollow fiber membranes. MOF nanosheets was 1.21 mol L−1 (swollen membrane). The diffusion
mer layers less than 100 nm thick requires small exhibiting high gas permeance and high gas coefficient of Na+ in the membrane at the same
filler size. Bachman et al. reported MOF nano- selectivity (e.g., H2/CO2 separation) have been conditions was 2.62 × 10−6 cm2 s−1. The Na+ per-
crystals dispersed in a polyimide membrane for developed (84). When such MOF nanosheets are meability, PNaþ , was calculated using the solution-
olefin/paraffin separation (55). Small MOF nano- incorporated into a polymer matrix, the resulting diffusion model (91): PNaþ ¼ CNa m
þ DNaþ =CNaþ ,
s
m +
crystals (17 to 18 nm), having high external surface MMMs can improve gas separation performance where CNa þ is the Na concentration in the mem-
s +
area, led to a greater fraction of polymer at the by eliminating nonselective permeation pathways brane, CNa þ is the Na concentration in the solu-

polymer/MOF interface, thereby reducing the as compared with isotropic MOFs. Rodenas et al. tion contiguous to the membrane (1 M), and DNaþ

Park et al., Science 356, eaab0530 (2017) 16 June 2017 5 of 10

R ES E A RC H | R E V IE W

some applications cannot use or benefit from

higher flux. To date, only a handful of families
of polymers have been commercialized as sepa-
ration membranes (3). High permeability is one
component of high flux but not the only one.
To obtain high flux, one may design a material
with a higher permeability coefficient, make an
existing membrane thinner, or increase the driving
force for transport. However, there are limitations
and drawbacks to each of these approaches.
Figure 4 shows cross sections of typical gas
separation and RO membranes. All membranes
for these applications are formed in solvent-based
Fig. 4. Morphology of state-of-the-art membranes. Examples of (A) hollow fiber gas separation processes that lead to a thin (~100 nm) selective
membrane (118) and (B) flat sheet RO membrane (91). membrane layer situated on a porous support,
which provides mechanical strength to the thin
selective membrane. This is about the thinnest
is the sodium diffusion coefficient in the mem- porin channels themselves. Although aquapor- that can be achieved in large-scale membrane
brane. Thus, the Na+ permeability in Nafion was ins have the potential to create highly permeable, production processes with current procedures
3.1 × 10−6 cm2 s−1, nearly two orders of magni- salt-rejecting membranes, no studies to date without introducing intolerable amounts of
tude lower than that of the potassium ion chan- have demonstrated higher performance data pinhole defects (i.e., through pores) that allow
nel, demonstrating the high permeability of ion for aquaporin-based membranes than for state- selectivity-destroying bulk convective flow (74, 75).
channels relative to synthetic membranes. of-the-art thin-film composite desalination mem- The porous support may be 50 to 200 (or more)

Downloaded from https://www.science.org at Open University of Israel on November 22, 2021

In aquaporins, water can transport as fast as branes (48). mm thick and typically has surface porosity values
~3 × 109 molecules/s through the water channel, Although some initial fouling and cleaning in the range of 1 to 10%, with small surface pores
whereas H3O+ and small neutral solutes are data have been reported for aquaporin-based (<100 nm) to provide a relatively smooth surface
almost completely excluded (48, 93–95). Such membranes (49), further exploration of the long- for the nonporous selective layer (99, 100).
transport properties are simply unattainable term robustness of such membranes in actual Figure 5A shows an example of the effect of
with current synthetic membranes (68). These desalination processes is needed. Examples of support mass transfer resistance on upper-bound
combinations of high selectivity and permeation such studies include fouling properties, tolerance performance of a thin-film composite (TFC) gas
are ascribed to several common structural fea- to common membrane-cleaning chemistries, and separation membrane consisting of a thin selec-
tures in both aquaporins and potassium ion chan- stability of AQP-based membranes. Moreover, tive separating layer on top of a porous support.
nels: (i) uniform pore sizes of just the right size to higher selectivity is needed to be competitive Such membrane structures are widely used com-
accommodate the desired ion or water molecule; with existing RO membranes. Depending upon mercially because the separation layer thickness
(ii) hourglass-shaped pores, with a very short its source, water slated for desalination can con- can be quite thin (~100 nm) and, in many cases,
selectivity filter [i.e., 12 Å long in the potassium tain complex mixtures of ions and other com- the support can be made from a low-cost, me-
channel (87) and ~20 Å in AQP1 aquaporin (93)], ponents, some of which (e.g., mercury ions) are chanically robust engineering thermoplastic (e.g.,
flanked by wider regions (i.e., vestibules) to accom- known to disable water transport in aquaporins polysulfone). This approach allows one to use
modate rapid ion or water transport to and from (93). Studies to date have mainly focused on potentially exotic, expensive materials for the
the selectivity filter; and (iii) exquisitely tuned simple salts (e.g., NaCl), so the performance of separating layer, because so little of this material
molecular interactions to energetically and en- such membranes for separating other common is used per square meter of membrane area that
tropically facilitate partitioning of desired spe- ions, or mixtures of ions, is largely unknown. the overall cost of the membrane can be quite
cies into the channel (87, 93). Ability to remove solutes such as boron and low low. However, if the thin selective layer becomes
Direct incorporation of bacterial water chan- molar mass neutral molecules, which are prob- extremely thin (i.e., much less than 100 nm), the
nel proteins [aquaporin Z (AQP)] into synthetic lematic for existing RO membranes, has not distance a solute molecule must travel through
membranes can lead to improved combinations been reported. Similar considerations will apply the selective layer of the membrane to reach a
of water transport and salt rejection in de- to other biomimetic membrane structures under pore in the porous support can become greater
salination membranes. In 2007, Kumar et al. development. than the selective layer thickness, and flux does
reported successful insertion of AQP proteins not increase as expected with decreasing thick-
into synthetic triblock copolymer vesicles to form Design constraints beyond permeability ness. In gas separation membranes, addition of
spherical cavities (~117-nm radius) with AQP units and selectivity a high-permeability, thin “gutter” layer between
spanning the vesicle wall (96). The rate of water Many studies focus exclusively on making mem- the thin selective separating layer and the porous
transport through the vesicle walls increased 5 brane materials with better permeability and/or support helps mitigate this phenomenon, but
to 10 times upon introduction of AQP. To create better selectivity. However, the ultimate perform- simply making thinner selective layers does not
a functional membrane, AQP-containing vesicles ance of a membrane is gauged by flux (or per- guarantee flux increases (99). Although the porous
were ruptured onto a porous support membrane meance) and ability to separate mixtures of support is often assumed to contribute negligibly
to form a planar bilayer, which acted as the practical interest. High flux depends on using to mass transfer resistance across a membrane (i.e.,
selective barrier of the membrane (97). Such high-permeability materials and making thin all of the mass transfer resistance is due to the
membranes generally exhibited low salt rejec- membranes from those materials, which is why thin selective membrane itself), this approxima-
tion due to unavoidable defects. Another approach the separating layer of current membranes is tion becomes less and less valid as membrane
is to incorporate AQP-containing vesicles into often less than 100 nm thick. Making such thin, materials become thinner and more permeable,
the aqueous diamine solution commonly used defect-free membranes on a large scale in a leading to fluxes that do not increase as the thick-
to prepare NF and RO membranes by inter- reproducible fashion is a critical barrier to in- ness of the thin selective membrane decreases and
facial polymerization (98). However, separation troduction of new membrane materials. Moreover, selectivities that are far below intrinsic values of
performance of membranes prepared by this as discussed further below, not every separation the separation material of interest (101). Thus,
simple approach is mostly dependent on the demands or even benefits from membranes with design of new membrane materials without con-
surrounding polymer matrix, not on the aqua- ultrahigh permeability or selectivity. Additionally, sideration of membrane support properties can

Park et al., Science 356, eaab0530 (2017) 16 June 2017 6 of 10

R ES E A RC H | R E V IE W

lead to performance much worse than that ex- to 100 L m−2 h−1 bar−1 produces a meager 1.0% gain in product purity (3). In gas separation mem-
pected based on results from free-standing films. reduction in energy consumption. Similar con- branes, this concept is quantified by the pressure
Such support mass transfer resistance has been clusions were obtained when modeling RO brackish ratio (i.e., the ratio of feed to permeate pressure)
recognized, for example, to have pronounced water desalination (102). Although high-permeance divided by membrane selectivity (3). Pressure
performance-limiting effects in the current gen- RO membranes may have the potential to reduce ratios may be set by economic considerations
eration of forward osmosis membranes (40) required membrane area, this solution will not largely dependent on process conditions (i.e.,
and has long been important in gas separation be practical because severe concentration polar- independent of membrane properties). In gas
membranes (99). ization (CP) induced by the high water flux will separation applications, typical pressure ratios
In water purification membranes, RO de- significantly hinder process performance (1). In are 5 to 15 (106). Designing materials with se-
salination provides an example of the insignif- addition, membrane fouling, which is already a lectivity values much greater than the pressure
icance of high-permeability membranes on process major problem with current-generation mem- ratio yields little or no improvement in product
efficiency. Numerous studies have suggested that branes, is exacerbated at higher water fluxes purity, as shown in Fig. 5B. Membranes used in
developing high-permeance RO membranes could (103, 104). Rather than increasing membrane industrial hydrogen separations (3) and those
reduce energy consumption in desalination, permeance, increasing the water-solute selectivity considered for postcombustion carbon capture,
which is proportional to the applied hydraulic (or minimizing solute passage) of RO membranes for example, have low pressure ratios, limiting
pressure to drive water permeation across the should be a very important goal for improving the need for highly selective membranes (106).
membrane (1, 72, 96). However, energy consump- process performance and yielding higher-quality Wessling and colleagues recently demonstrated
tion is constrained by the conventional single- product water (48, 102). Current challenges with the use of upper-bound properties of membranes
stage operation of RO systems, in which the existing membranes include the removal of boron coupled with process modeling to identify eco-
minimum hydraulic pressure, and hence the in seawater desalination, minimization of the pas- nomically optimal combinations of permeability
minimum specific energy (i.e., energy per vol- sage of low–molecular weight toxic contaminants and selectivity for nitrogen removal from natural
ume of product water), is equal to the osmotic in water and wastewater reuse [e.g., arsenic, gas (107). Such studies for other gas and liquid
pressure of the exiting brine (1, 102). Recent fluoride, and many uncharged solutes, such as separations of interest would be desirable to

Downloaded from https://www.science.org at Open University of Israel on November 22, 2021

process modeling for RO seawater desalination endocrine disruptors increasingly found in water provide appropriate targets for materials design.
indicates that increasing membrane water per- (105)], and production of high-purity water for In addition to having excellent transport prop-
meance from 2 L m−2 h−1 bar−1, the current level various industries with minimal deionization steps erties, membrane materials must be mechanically
of commercial TFC desalination membranes, to after the RO stage (102). robust to survive manufacturing, installation, and
10 L m−2 h−1 bar−1, results in only a 3.7% re- Membrane separations are often limited by long-term use. For example, seawater RO mem-
duction in energy consumption (102). Moreover, available driving force, so increases in mem- branes are routinely used at pressures of 55 bar,
a further 10-fold increase in permeance from 10 brane material selectivity result in little or no and gas separation membranes may operate at

Fig. 5. Effect of membrane support and operating conditions on a hypothetical material with separation properties above the upper bound.
separation characteristics of gas separation membranes. (A) Mem- The stars [right side of (A)] denote permeance and selectivity of thin-film
brane support. (B) operating conditions. (A) shows the 2008 upper-bound composite membranes of this material using different supports, showing
data for CO2/N2 separation translated into the selectivity and permeance that a high flux support is needed to reach desirable performance (blue
(i.e., pressure-normalized flux) that could be achieved by placing a thin shaded region) for postcombustion CO2 capture (15). The procedure for
(100 nm) membrane onto a slow (103 GPU), medium (104 GPU), or fast generating the upper-bound lines in the selectivity-permeance plot is
(105 GPU) porous support membrane. For brevity, only 20% of the 2008 described here (119). (B) shows the effect of varying membrane selectivity
upper-bound data have been shown on the selectivity-permeance plot. The on permeate vapor concentration for a vapor separation membrane
dashed lines indicate the expected performance of materials lying on the operating at a feed/permeate pressure ratio of 20 and 1% vapor in
upper bound. The star [left side of (A)] denotes permeability/selectivity of the feed.

Park et al., Science 356, eaab0530 (2017) 16 June 2017 7 of 10

R ES E A RC H | R E V IE W

pressures exceeding 100 bar (3). High-salinity bution, which can break the conventional upper 2. M. A. Shannon et al., Science and technology for water
brines (e.g., much greater than seawater, such as bound. purification in the coming decades. Nature 452, 301–310
(2008). doi: 10.1038/nature06599; pmid: 18354474
water produced from hydraulic fracking) would However, many questions remain about both 3. R. W. Baker, Membrane Technology and Applications (John
require mechanically more robust RO membranes new and existing membranes. For example, our Wiley & Sons, Inc., ed. 3, Hoboken, NJ, 2012).
to enable operation at ultrahigh pressures. For fundamental understanding of the basic thermo- 4. R. van Reis, A. Zydney, Bioprocess membrane technology.
example, using the extremely saline water (34% dynamics and diffusion properties of water and J. Membr. Sci. 297, 16–50 (2007). doi: 10.1016/
j.memsci.2007.02.045
salinity) from the Dead Sea would require mem- ions in polymers used in RO and electrodialysis 5. H. Lutz, Ultrafiltration for Bioprocessing: Development
branes operating at pressures of more than 250 bar membranes and many energy applications (e.g., and Implementation of Robust Processes (Elsevier,
(108). Details of the mechanical property require- fuel cells, batteries, and reverse electrodialysis) is Amsterdam, 2015).
ments and design criteria, both for membranes at an extremely rudimentary level (111), although 6. D. F. Stamatialis et al., Medical applications of membranes:
Drug delivery, artificial organs and tissue engineering.
and for their porous supports, are currently poorly recent advances in this area are allowing for J. Membr. Sci. 308, 1–34 (2008). doi: 10.1016/
understood, in part due to the lack of mechanical prediction of ion sorption and transport in highly j.memsci.2007.09.059
property data accompanying reports of new mem- charged ion exchange membranes (111, 112). While 7. K.-V. Peinemann, S. P. Nunes, L. Giorno, Membrane
brane materials. theoretical models have been proposed for the Technology: Volume 3: Membranes for Food Applications
(Wiley-VCH, Weinheim, Germany, 2010).
Membrane materials must also be chemically upper bound for gas separation membranes (22) 8. T. Sata, Ion Exchange Membranes: Preparation,
and thermally tolerant to conditions encountered and desalination membranes (47), no such analog Characterization, Modification and Application (Royal Society
during operation. For example, oxidatively stable exists today for other separations, such as electro- of Chemistry, London, 2004).
(i.e., chlorine tolerant) RO membranes remain an dialysis, although the empirical evidence is over- 9. P. Arora, Z. J. Zhang, Battery separators. Chem. Rev. 104,
4419–4462 (2004). doi: 10.1021/cr020738u;
elusive goal. Such materials are needed because whelming that such a trade-off exists (46). There pmid: 15669158
aqueous chlorine (e.g., hypochlorite) is widely is a real need for fundamental modeling, at length 10. S. J. Peighambardoust, S. Rowshanzamir, M. Amjadi,
used to control biological activity in water streams scales ranging from atomistic to continuum, to Review of the proton exchange membranes for fuel cell
being fed to desalination processes, and current provide rational guidance for designing future applications. Int. J. Hydrogen Energy 35, 9349–9384
(2010). doi: 10.1016/j.ijhydene.2010.05.017
aromatic polyamide–based polymers, the state of membranes. More recently developed nanomate-

Downloaded from https://www.science.org at Open University of Israel on November 22, 2021

11. Y. Wang, K. S. Chen, J. Mishler, S. C. Cho, X. C. Adroher, A
the art for desalination membranes, have poor rials and MMMs would also benefit from an review of polymer electrolyte membrane fuel cells:
chlorine tolerance (109). Polymeric gas separa- increased focus on fundamental modeling. In Technology, applications, and needs on fundamental
tion membranes are rarely deployed in applica- all cases, better understanding of structure- research. Appl. Energy 88, 981–1007 (2011). doi: 10.1016/
j.apenergy.2010.09.030
tions exceeding 100°C due to a lack of stability property-performance relationships in new, as 12. J. W. Post, H. V. M. Hamelers, C. J. N. Buisman, Energy recovery
of such membranes at high temperatures. Sep- well as existing, membrane materials is urgently from controlled mixing salt and fresh water with a reverse
arations such as those found in precombustion needed. electrodialysis system. Environ. Sci. Technol. 42, 5785–5790
carbon capture could benefit from membranes For all membrane materials, it is critical to (2008). doi: 10.1021/es8004317; pmid: 18754509
13. B. E. Logan, M. Elimelech, Membrane-based processes for
with much better thermal stability (110). Design- consider factors other than permeability and sustainable power generation using water. Nature 488,
ing membranes that are resistant to plasticization selectivity in materials design. Today, we typically 313–319 (2012). doi: 10.1038/nature11477; pmid: 22895336
(i.e., penetrant-induced swelling and subsequent use rather elaborate, nonequilibrium, continuous 14. R. S. Kingsbury, K. Chu, O. Coronell, Energy storage by
loss of separation properties) will be required for processes involving large amounts of volatile, reversible electrodialysis: The concentration battery.
J. Membr. Sci. 495, 502–516 (2015). doi: 10.1016/
membranes to be used in applications such as potentially toxic solvents to manufacture mem- j.memsci.2015.06.050
olefin/paraffin separation (36, 37). Most impor- branes. An advantage to this approach is the 15. T. C. Merkel, H. Lin, X. Wei, R. Baker, Power plant post-
tant, membrane materials must be scalable to be widespread use of composite membranes, where combustion carbon dioxide capture: An opportunity for
manufactured in defect-free, large surface areas. a thin (~100 nm) separating membrane is ap- membranes. J. Membr. Sci. 359, 126–139 (2010).
doi: 10.1016/j.memsci.2009.10.041
plied to a porous support membrane, which can 16. L. M. Blaney, S. Cinar, A. K. SenGupta, Hybrid anion
Outlook be made of a relatively inexpensive engineering exchanger for trace phosphate removal from water and
Increasing demand for energy-efficient gas and thermoplastic. Such composite membranes allow wastewater. Water Res. 41, 1603–1613 (2007). doi: 10.1016/
water separations, coupled with growing avail- the use of novel, expensive materials as the sep- j.watres.2007.01.008; pmid: 17306856
17. L. S. White, Development of large-scale applications in
ability of nanomaterials and a deeper under- arating membrane, because so little of the sep- organic solvent nanofiltration and pervaporation for chemical
standing of the structural features of biological arating membrane is required for a given area of and refining processes. J. Membr. Sci. 286, 26–35 (2006).
membranes that give them excellent permeabil- membrane. Nevertheless, novel processing strat- doi: 10.1016/j.memsci.2006.09.006
ity and selectivity, has stimulated substantial egies to simplify membrane manufacturing and 18. P. Marchetti, M. F. Jimenez Solomon, G. Szekely,
A. G. Livingston, Molecular separation with organic solvent
research aimed at overcoming the permeability/ scale-up would contribute remarkably to accel-
nanofiltration: A critical review. Chem. Rev. 114, 10735–10806
selectivity trade-off. Membranes in use today erating deployment of new membrane materials. (2014). doi: 10.1021/cr500006j; pmid: 25333504
were discovered, in many cases, by serendipity. Development of solvent-free membrane manufac- 19. E. Curcio, E. Drioli, Membrane distillation and related
Molecular-level design and insight, including turing processes or processes using environmen- operations: A review. Separ. Purif. Rev. 34, 35–86 (2005).
advanced simulation and modeling, will be crit- tally benign, inexpensive solvents (e.g., water) doi: 10.1081/SPM-200054951
20. S. M. Joscelyne, G. Trägårdh, Membrane emulsification: A
ical for breakthroughs going forward. For exam- would bring about disruptive, sustainable changes literature review. J. Membr. Sci. 169, 107–117 (2000).
ple, the approaches used to prepare membranes in membrane fabrication. Issues such as process doi: 10.1016/S0376-7388(99)00334-8
today do not allow for independent control of constraints, material processability, and long-term 21. L. M. Robeson, Correlation of separation factor versus
permeation properties of interest [e.g., water stability in the process environment where the permeability for polymeric membranes. J. Membr. Sci. 62,
165–185 (1991). doi: 10.1016/0376-7388(91)80060-J
transport cannot be controlled independent of membranes will operate should be built into re-
22. B. D. Freeman, Basis of permeability/selectivity tradeoff
salt (or other solute) transport, and the transport search efforts to identify, as early as possible, relations in polymeric gas separation membranes.
of one gas, e.g., CO2, cannot be systematically materials that can be deployed for large-scale, Macromolecules 32, 375–380 (1999). doi: 10.1021/
manipulated without changing the transport practical challenges, such as the growing need ma9814548
of other gases, e.g., CH4], even though such pro- for clean water, treatment of waste products from 23. L. M. Robeson, The upper bound revisited. J. Membr. Sci.
320, 390–400 (2008). doi: 10.1016/j.memsci.2008.04.030
perties would be highly desirable. Whereas tra- hydraulic fracturing, and climate change. 24. W. Song, P. J. Rossky, M. Maroncelli, Modeling
ditional membranes for water and gas separations alkane+perfluoroalkane interactions using all-atom potentials:
are largely based on polymers and are subject Failure of the usual combining rules. J. Chem. Phys. 119,
RE FERENCES AND NOTES 9145–9162 (2003). doi: 10.1063/1.1610435
to the permeability-selectivity trade-off, many
1. M. Elimelech, W. A. Phillip, The future of seawater 25. T. C. Merkel, I. Pinnau, R. Prabhakar, B. D. Freeman, in
new materials and design approaches (e.g., bio- desalination: Energy, technology, and the environment. Materials Science of Membranes, Y. Yampolskii, I. Pinnau,
inspired, biomimetic, or MMMs) offer the hope Science 333, 712–717 (2011). doi: 10.1126/science.1200488; B. D. Freeman, Eds. (John Wiley & Sons Ltd., Chichester, UK,
of better control over pore size and size distri- pmid: 21817042 2006), pp. 251–270.

Park et al., Science 356, eaab0530 (2017) 16 June 2017 8 of 10

R ES E A RC H | R E V IE W

26. P. M. Budd, N. B. McKeown, D. Fritsch, Free volume and 47. H. Zhang, G. M. Geise, Modeling the water permeability and 68. A state-of-the-art commercial RO membrane (Dow SEAMAXX
intrinsic microporosity in polymers. J. Mater. Chem. 15, 1977 water/salt selectivity tradeoff in polymer membranes. 440i) has a reported water volumetric flow rate of 17,000
(2005). doi: 10.1039/b417402j J. Membr. Sci. 520, 790–800 (2016). doi: 10.1016/ gallons per day through an active membrane area of 41 m2
27. H. B. Park et al., Polymers with cavities tuned for fast j.memsci.2016.08.035 when operated as designed (i.e., feed salt concentration =
selective transport of small molecules and ions. Science 318, 48. J. R. Werber, C. O. Osuji, M. Elimelech, Materials for next- 32,000 mg/L NaCl and feed pressure = 55.2 bar absolute)
254–258 (2007). doi: 10.1126/science.1146744; generation desalination and water purification membranes. (65). The water flux through this membrane is approximately
pmid: 17932294 Nat. Rev. Mater. 1, 16018 (2016). doi: 10.1038/ 65 L m−2 hour−1. Normalizing for the driving force for water
28. L. M. Robeson, M. E. Dose, B. D. Freeman, D. R. Paul, Analysis natrevmats.2016.18 transport (i.e., Dp  Dp ¼ 55:2 bar  27:1 bar ¼ 28:1 bar),
of the transport properties of thermally rearranged (TR) 49. X. Li et al., Nature gives the best solution for desalination: the permeance is 2.3 L m−2 hour−1 bar−1. The osmotic pressure
polymers and polymers of intrinsic microporosity (PIM) Aquaporin-based hollow fiber composite membrane with difference, Dp, was calculated using the van’t Hoff relation (i.e.,
relative to upper bound performance. J. Membr. Sci. 525, superior performance. J. Membr. Sci. 494, 68–77 (2015). p = CRT). The salt concentration in the permeate (96 mg/L)
18–24 (2017). doi: 10.1016/j.memsci.2016.11.085 doi: 10.1016/j.memsci.2015.07.040 was calculated from the reported salt rejection (99.70%) and a
29. L. M. Robeson, Q. Liu, B. D. Freeman, D. R. Paul, 50. S. Y. Yang et al., Nanoporous membranes with ultrahigh feed salt concentration (32,000 mg/L) (65). Aquaporins are
Comparison of transport properties of rubbery and glassy selectivity and flux for the filtration of viruses. Adv. Mater. 18, reported to transport water at rates of 3 × 109 molecules per
polymers and the relevance to the upper bound 709–712 (2006). doi: 10.1002/adma.200501500 second through pores that are 0.28 nm in diameter (93–95).
relationship. J. Membr. Sci. 476, 421–431 (2015). 51. K. V. Peinemann, V. Abetz, P. F. Simon, Asymmetric Thus, the water flux through an aquaporin is 52 × 105 L m−2
doi: 10.1016/j.memsci.2014.11.058 superstructure formed in a block copolymer via phase hour−1, nearly 5 orders of magnitude greater than that in the
30. D. F. Sanders et al., Energy-efficient polymeric gas separation separation. Nat. Mater. 6, 992–996 (2007). doi: 10.1038/ commercial RO membrane. This estimate accounts for water
membranes for a sustainable future: A review. Polymer (Guildf.) nmat2038; pmid: 17982467 transport through a single aquaporin channel. Such channels,
54, 4729–4761 (2013). doi: 10.1016/j.polymer.2013.05.075 52. X. Qiu et al., Selective separation of similarly sized proteins configured in a close-packed array, would have a packing
31. I. K. Meier, M. Langsam, H. C. Klotz, Selectivity enhancement with tunable nanoporous block copolymer membranes. density of 0.25% (66), so the maximum hypothetical flux
via photooxidative surface modification of polyimide air ACS Nano 7, 768–776 (2013). doi: 10.1021/nn305073e; through a flat sheet membrane comprising aquaporins at
separation membranes. J. Membr. Sci. 94, 195–212 (1994). pmid: 23252799 this packing density would be 1.3 × 104 L m−2 hour−1, which
doi: 10.1016/0376-7388(93)E0174-I 53. H. Yu et al., Self-assembled asymmetric block copolymer is ~200 times as high as that of the commercial RO
32. C. H. Park et al., Nanocrack-regulated self-humidifying membranes: Bridging the gap from ultra- to nanofiltration. membrane. This calculation does not include effects of
membranes. Nature 532, 480–483 (2016). doi: 10.1038/ Angew. Chem. Int. Ed. 54, 13937–13941 (2015). doi: 10.1002/ concentration polarization, support layer resistance,
nature17634; pmid: 27121841 anie.201505663; pmid: 26388216 fouling, or other effects that would likely be observed in
33. H. Nishide, T. Suzuki, H. Kawakami, E. Tsuchida, Cobalt 54. V. Abetz, Isoporous block copolymer membranes. Macromol. practice, so the value reported here should be used only to

Downloaded from https://www.science.org at Open University of Israel on November 22, 2021

porphyrin-mediated oxygen transport in a polymer Rapid Commun. 36, 10–22 (2015). doi: 10.1002/ compare intrinsic permeation properties of the biological and
membrane: effect of the cobalt porphyrin structure on the marc.201400556; pmid: 25451792 synthetic membrane, not practically achievable fluxes of an
oxygen-binding reaction, oxygen-diffusion constants, and 55. J. E. Bachman, Z. P. Smith, T. Li, T. Xu, J. R. Long, Enhanced aquaporin-based membrane, because effects ignored in this
oxygen-transport efficiency. J. Phys. Chem. 98, 5084–5088 ethylene separation and plasticization resistance in polymer analysis would be limited in a real membrane based
(1994). doi: 10.1021/j100070a023 membranes incorporating metal-organic framework on such ultrapermeable materials (67).
34. L. M. Robeson, Polymer blends in membrane transport nanocrystals. Nat. Mater. 15, 845–849 (2016). doi: 10.1038/ 69. R. K. Joshi et al., Precise and ultrafast molecular sieving
processes. Ind. Eng. Chem. Res. 49, 11859–11865 (2010). nmat4621; pmid: 27064528 through graphene oxide membranes. Science 343, 752–754
doi: 10.1021/ie100153q 56. H. Furukawa, K. E. Cordova, M. O’Keeffe, O. M. Yaghi, The (2014). doi: 10.1126/science.1245711; pmid: 24531966
35. C. M. Zimmerman, A. Singh, W. J. Koros, Tailoring mixed chemistry and applications of metal-organic frameworks. 70. D. R. Dreyer, S. Park, C. W. Bielawski, R. S. Ruoff, The
matrix composite membranes for gas separations. Science 341, 1230444 (2013). doi: 10.1126/science.1230444; chemistry of graphene oxide. Chem. Soc. Rev. 39, 228–240
J. Membr. Sci. 137, 145–154 (1997). doi: 10.1016/S0376- pmid: 23990564 (2010). doi: 10.1039/B917103G; pmid: 20023850
7388(97)00194-4 57. B. R. Pimentel, A. Parulkar, E. K. Zhou, N. A. Brunelli, 71. R. R. Nair, H. A. Wu, P. N. Jayaram, I. V. Grigorieva, A. K. Geim,
36. R. L. Burns, W. J. Koros, Defining the challenges for C3H6/C3H8 R. P. Lively, Zeolitic imidazolate frameworks: Next-generation Unimpeded permeation of water through helium-leak-tight
separation using polymeric membranes. J. Membr. Sci. 211, materials for energy-efficient gas separations. ChemSusChem graphene-based membranes. Science 335, 442–444 (2012).
299–309 (2003). doi: 10.1016/S0376-7388(02)00430-1 7, 3202–3240 (2014). doi: 10.1002/cssc.201402647; doi: 10.1126/science.1211694; pmid: 22282806
37. M. Rungta, C. Zhang, W. J. Koros, L. Xu, Membrane-based pmid: 25363474 72. J. K. Holt et al., Fast mass transport through sub-2-nanometer
ethylene/ethane separation: The upper bound and 58. A. J. Brown et al., Separation membranes. Interfacial carbon nanotubes. Science 312, 1034–1037 (2006).
beyond. AIChE J. 59, 3475–3489 (2013). doi: 10.1002/ microfluidic processing of metal-organic framework hollow doi: 10.1126/science.1126298; pmid: 16709781
aic.14105 fiber membranes. Science 345, 72–75 (2014). doi: 10.1126/ 73. H. Li et al., Ultrathin, molecular-sieving graphene oxide
38. S. Park et al., The polymeric upper bound for N2/NF3 science.1251181; pmid: 24994649 membranes for selective hydrogen separation. Science 342,
separation and beyond; ZIF-8 containing mixed matrix 59. D. S. Sholl, R. P. Lively, Defects in metal-organic frameworks: 95–98 (2013). doi: 10.1126/science.1236686; pmid: 24092739
membranes. J. Membr. Sci. 486, 29–39 (2015). doi: 10.1016/ Challenge or opportunity? J. Phys. Chem. Lett. 6, 3437–3444 74. J. M. S. Henis, M. K. Tripodi, The developing technology of
j.memsci.2015.03.030 (2015). doi: 10.1021/acs.jpclett.5b01135; pmid: 26268796 gas separating membranes. Science 220, 11–17 (1983).
39. G. M. Geise, H. B. Park, A. C. Sagle, B. D. Freeman, 60. H. T. Kwon, H. K. Jeong, A. S. Lee, H. S. An, J. S. Lee, doi: 10.1126/science.220.4592.11; pmid: 17736148
J. E. McGrath, Water permeability and water/salt selectivity Heteroepitaxially grown zeolitic imidazolate framework 75. J. M. S. Henis, M. K. Tripodi, Composite hollow fiber
tradeoff in polymers for desalination. J. Membr. Sci. 369, membranes with unprecedented propylene/propane membranes for gas separation: The resistance model
130–138 (2011). doi: 10.1016/j.memsci.2010.11.054 separation performances. J. Am. Chem. Soc. 137, approach. J. Membr. Sci. 8, 233–246 (1981). doi: 10.1016/
40. D. L. Shaffer, J. R. Werber, H. Jaramillo, S. Lin, M. Elimelech, 12304–12311 (2015). doi: 10.1021/jacs.5b06730; S0376-7388(00)82312-1
Forward osmosis: Where are we now? Desalination 356, pmid: 26364888 76. S. Husain, W. J. Koros, Mixed matrix hollow fiber membranes
271–284 (2015). doi: 10.1016/j.desal.2014.10.031 61. V. Singh et al., Graphene based materials: Past, present made with modified HSSZ-13 zeolite in polyetherimide
41. N. Y. Yip et al., Thin-film composite pressure retarded and future. Prog. Mater. Sci. 56, 1178–1271 (2011). polymer matrix for gas separation. J. Membr. Sci. 288,
osmosis membranes for sustainable power generation from doi: 10.1016/j.pmatsci.2011.03.003 195–207 (2007). doi: 10.1016/j.memsci.2006.11.016
salinity gradients. Environ. Sci. Technol. 45, 4360–4369 62. S. P. Koenig, L. Wang, J. Pellegrino, J. S. Bunch, Selective 77. R. Adams et al., in Encyclopedia of Membrane Science
(2011). doi: 10.1021/es104325z; pmid: 21491936 molecular sieving through porous graphene. Nat. Nanotechnol. and Technology, E. M. V. Hoek, V. V. Tarabara, Eds.
42. J. Wei, X. Liu, C. Qiu, R. Wang, C. Y. Tang, Influence of 7, 728–732 (2012). doi: 10.1038/nnano.2012.162; (John Wiley & Sons, Inc., Hoboken, NJ, 2013).
monomer concentrations on the performance of polyamide- pmid: 23042491 78. B. Seoane et al., Metal-organic framework based mixed
based thin film composite forward osmosis membranes. 63. H. W. Kim et al., Selective gas transport through few- matrix membranes: A solution for highly efficient CO2
J. Membr. Sci. 381, 110–117 (2011). doi: 10.1016/ layered graphene and graphene oxide membranes. Science capture? Chem. Soc. Rev. 44, 2421–2454 (2015).
j.memsci.2011.07.034 342, 91–95 (2013). doi: 10.1126/science.1236098; doi: 10.1039/C4CS00437J; pmid: 25692487
43. A. Mehta, A. L. Zydney, Permeability and selectivity analysis pmid: 24092738 79. T. T. Moore, W. J. Koros, Non-ideal effects in organic–inorganic
for ultrafiltration membranes. J. Membr. Sci. 249, 245–249 64. S. P. Surwade et al., Water desalination using nanoporous materials for gas separation membranes. J. Mol. Struct. 739,
(2005). doi: 10.1016/j.memsci.2004.09.040 single-layer graphene. Nat. Nanotechnol. 10, 459–464 (2015). 87–98 (2005). doi: 10.1016/j.molstruc.2004.05.043
44. L. Robeson, H. Hwu, J. McGrath, Upper bound relationship for doi: 10.1038/nnano.2015.37; pmid: 25799521 80. H. Ha, J. Park, K. Ha, B. D. Freeman, C. J. Ellison, Synthesis
proton exchange membranes: Empirical relationship and 65. Dow FilmTec Seamaxx Element Product Data Sheet, http:// and gas permeability of highly elastic poly(dimethylsiloxane)/
relevance of phase separated blends. J. Membr. Sci. 302, msdssearch.dow.com/PublishedLiteratureDOWCOM/ graphene oxide composite elastomers using telechelic
70–77 (2007). doi: 10.1016/j.memsci.2007.06.029 dh_0948/0901b80380948b83.pdf?filepath=liquidseps/pdfs/ polymers. Polymer (Guildf.) 93, 53–60 (2016). doi: 10.1016/
45. C. P. Ribeiro, B. D. Freeman, D. S. Kalika, S. Kalakkunnath, noreg/609-50126.pdf&fromPage=GetDoc (May 8, 2016). j.polymer.2016.04.016
Aromatic polyimide and polybenzoxazole membranes for the 66. M. Grzelakowski, M. F. Cherenet, Y. Shen, M. Kumar, A 81. H. Ha et al., Gas permeation and selectivity of poly
fractionation of aromatic/aliphatic hydrocarbons by framework for accurate evaluation of the promise of (dimethylsiloxane)/graphene oxide composite elastomer
pervaporation. J. Membr. Sci. 390-391, 182–193 (2012). aquaporin based biomimetic membranes. J. Membr. Sci. 479, membranes. J. Membr. Sci. 518, 131–140 (2016).
doi: 10.1016/j.memsci.2011.11.042 223–231 (2015). doi: 10.1016/j.memsci.2015.01.023 doi: 10.1016/j.memsci.2016.06.028
46. G. M. Geise, M. A. Hickner, B. E. Logan, Ionic resistance and 67. D. Cohen-Tanugi, R. K. McGovern, S. H. Dave, J. H. Lienhard, 82. G. Dong, H. Li, V. Chen, Challenges and opportunities for
permselectivity tradeoffs in anion exchange membranes. J. C. Grossman, Quantifying the potential of ultra-permeable mixed-matrix membranes for gas separation. J. Mater. Chem.
ACS Appl. Mater. Interfaces 5, 10294–10301 (2013). membranes for water desalination. Energy Environ. Sci. 7, A Mater. Energy Sustain. 1, 4610 (2013). doi: 10.1039/
doi: 10.1021/am403207w; pmid: 24040962 1134–1141 (2014). doi: 10.1039/C3EE43221A c3ta00927k

Park et al., Science 356, eaab0530 (2017) 16 June 2017 9 of 10

R ES E A RC H | R E V IE W

83. A. A. Gusev, O. Guseva, Rapid mass transport in mixed matrix Technol. 49, 1436–1444 (2015). doi: 10.1021/es5044062; 118. L. Liu, E. S. Sanders, S. S. Kulkarni, D. J. Hasse, W. J. Koros,
nanotube/polymer membranes. Adv. Mater. 19, 2672–2676 pmid: 25564877 Sub-ambient temperature flue gas carbon dioxide capture via
(2007). doi: 10.1002/adma.200602018 101. I. Pinnau, J. G. Wijmans, I. Blume, T. Kuroda, K.-V. Peinemann, Matrimid® hollow fiber membranes. J. Membr. Sci. 465,
84. Y. Peng et al., Metal-organic framework nanosheets as Gas permeation through composite membranes. J. Membr. Sci. 49–55 (2014). doi: 10.1016/j.memsci.2014.03.060
building blocks for molecular sieving membranes. Science 37, 81–88 (1988). doi: 10.1016/S0376-7388(00)85070-X 119. The gas permeance of a thin film composite membrane, JC ,
346, 1356–1359 (2014). doi: 10.1126/science.1254227; 102. J. R. Werber, A. Deshmukh, M. Elimelech, The critical need for can be calculated using the series resistance model (10)
pmid: 25504718 increased selectivity, not increased water permeability, for 1=Jc ¼ ‘A =PA þ 1=JB , where PA is the gas permeability (in
85. T. Rodenas et al., Metal-organic framework nanosheets in desalination membranes. Environ. Sci. Technol. Lett. 3, units of barrer) of the active layer, JB is the gas permeance (in
polymer composite materials for gas separation. Nat. Mater. 14, 112–120 (2016). doi: 10.1021/acs.estlett.6b00050 units of GPU) of the porous support, and ‘A is the thickness of
48–55 (2015). doi: 10.1038/nmat4113; pmid: 25362353 103. D. J. Miller, S. Kasemset, D. R. Paul, B. D. Freeman, the separation membrane (i.e., active layer). The permeance-
86. J. Shen et al., Membranes with fast and selective gas- Comparison of membrane fouling at constant flux and selectivity data in Fig. 4A were calculated using this model.
transport channels of laminar graphene oxide for efficient constant transmembrane pressure conditions. J. Membr. Sci. The CO2 permeance of the composite membrane was
2 1
CO2 capture. Angew. Chem. Int. Ed. Engl. 54, 578–582 454, 505–515 (2014). doi: 10.1016/j.memsci.2013.12.027 calculated from JCO c
2
¼ ð‘A =PCO A
2
þ 1=JCO
B Þ and the N2
(2015). pmid: 25378197 104. R. Zhang et al., Antifouling membranes for sustainable permeance of the composite membrane was calculated from
87. D. A. Doyle et al., The structure of the potassium channel: water purification: Strategies and mechanisms. Chem. Soc. JNc 2 ¼ ð‘A =PNA 2 þ 1=JBN2 Þ1. Finally, the selectivity, aCCO2 =N2 , was
Molecular basis of K+ conduction and selectivity. Science Rev. 45, 5888–5924 (2016). doi: 10.1039/C5CS00579E; calculated from aCCO2 =N2 ¼ JcCO2 =JcN2 . To demonstrate the
280, 69–77 (1998). doi: 10.1126/science.280.5360.69; pmid: 27494001 importance of the porous support layer, the N2 permeance of
pmid: 9525859 105. S. A. Snyder et al., Role of membranes and activated carbon the porous support layer in Fig. 4A is varied over a range of
88. J. E. McGrath, H. B. Park, B. D. Freeman, Chlorine resistant in the removal of endocrine disruptors and pharmaceuticals. values (1000, 10,000, and 100,000 GPU) representing slow,
desalination membranes based on directly sulfonated poly Desalination 202, 156–181 (2007). doi: 10.1016/ medium, and fast support membranes, respectively.
(arylene ether sulfone) copolymers. U.S. Patent 8,028,842, j.desal.2005.12.052 Assuming Knudsen selectivity in the porous support, the CO2
(4 October 2011). 106. Y. Huang, T. C. Merkel, R. W. Baker, Pressure ratio and its permeance of the pporous support
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi layer is then calculated
89. G. M. Geise, D. R. Paul, B. D. Freeman, Fundamental water impact on membrane gas separation processes. J. Membr. Sci. from JCOB
2
¼ JBN2 MN2 =MCO2 ¼ 0:798JBN2 , where MN2 is the
and salt transport properties of polymeric materials. 463, 33–40 (2014). doi: 10.1016/j.memsci.2014.03.016 molecular mass of N2, and MCO2 is the molecular
Prog. Polym. Sci. 39, 1–42 (2014). doi: 10.1016/ 107. B. Ohs, J. Lohaus, M. Wessling, Optimization of membrane mass of CO2. The dashed lines in the selectivity-permeance
j.progpolymsci.2013.07.001 based nitrogen removal from natural gas. J. Membr. Sci. 498, plot in Fig. 4A were generated as described below.
90. B. Hille, Ion Channels of Excitable Membranes, Third Edition 291–301 (2016). doi: 10.1016/j.memsci.2015.10.007 The upper-bound relationship for CO2/N2 is (23)
(Sinauer Associates, Inc., Sunderland, MA, 2001). 108. A. P. Straub, A. Deshmukh, M. Elimelech, Pressure-retarded PCO2 ¼ 30; 967; 000ðaCO2 =N2 Þ2:888 . This equation was used

Downloaded from https://www.science.org at Open University of Israel on November 22, 2021

91. G. M. Geise et al., Water purification by membranes: The role osmosis for power generation from salinity gradients: Is it to obtain CO2/N2 selectivities for CO2 permeability values
of polymer science. J. Polym. Sci. Polym. Phys. 48, viable? Energy Environ. Sci. 9, 31–48 (2016). doi: 10.1039/ ranging from 10−4 to 105 barrer. Subsequently, N2 perme-
1685–1718 (2010). doi: 10.1002/polb.22037 C5EE02985F ability values were calculated from PN2 ¼ PCO2 =aCO2 =N2. The
92. P. N. Pintauro, D. N. Bennlon, Mass transport of electrolytes 109. H. B. Park, B. D. Freeman, Z. B. Zhang, M. Sankir, selectivity and permeance values for the composite mem-
in membranes. 2. Determination of NaCl equilibrium and J. E. McGrath, Highly chlorine-tolerant polymers for branes were then calculated according to procedures
transport parameters for nafion. Ind. Eng. Chem. Res. 23, desalination. Angew. Chem. Int. Ed. Engl. 47, 6019–6024 described above.
234–243 (1984). (2008). doi: 10.1002/anie.200800454; pmid: 18604786 120. J. G. Wijmans, R. W. Baker, The solution-diffusion model:
93. P. Agre, Aquaporin water channels (Nobel Lecture). Angew. 110. T. C. Merkel, M. Zhou, R. W. Baker, Carbon dioxide capture A review. J. Membr. Sci. 107, 1–21 (1995). doi: 10.1016/0376-
Chem. Int. Ed. Engl. 43, 4278–4290 (2004). doi: 10.1002/ with membranes at an IGCC power plant. J. Membr. Sci. 389, 7388(95)00102-I
anie.200460804; pmid: 15368374 441–450 (2012). doi: 10.1016/j.memsci.2011.11.012 121. D. R. Paul, Reformulation of the solution-diffusion theory of
94. J. S. Jung, G. M. Preston, B. L. Smith, W. B. Guggino, P. Agre, 111. J. Kamcev et al., Partitioning of mobile ions between ion reverse osmosis. J. Membr. Sci. 241, 371–386 (2004).
Molecular structure of the water channel through aquaporin exchange polymers and aqueous salt solutions: Importance doi: 10.1016/j.memsci.2004.05.026
CHIP. The hourglass model. J. Biol. Chem. 269, 14648–14654 of counter-ion condensation. Phys. Chem. Chem. Phys. 18,
(1994). pmid: 7514176 6021–6031 (2016). doi: 10.1039/C5CP06747B; AC KNOWLED GME NTS
95. G. M. Preston, T. P. Carroll, W. B. Guggino, P. Agre, pmid: 26840776 We are grateful to many members of the membrane community
Appearance of water channels in Xenopus oocytes 112. J. Kamcev, D. R. Paul, G. S. Manning, B. D. Freeman, worldwide who contributed articles, advice, and perspectives
expressing red cell CHIP28 protein. Science 256, 385–387 Predicting salt permeability coefficients in highly swollen, during the preparation of this review. H.B.P and B.D.F.
(1992). doi: 10.1126/science.256.5055.385; pmid: 1373524 highly charged ion exchange membranes. ACS Appl. Mater. acknowledge financial support by the Korea CCS R&D Center
96. M. Kumar, M. Grzelakowski, J. Zilles, M. Clark, W. Meier, Interfaces 9, 4044–4056 (2017). doi: 10.1021/ (KCRC) grant funded by Ministry of Science, ICT, and Future
Highly permeable polymeric membranes based on the acsami.6b14902; pmid: 28072514 Planning from the Korean government (grant 2016910057). The
incorporation of the functional water channel protein 113. M. L. Greenfield, D. N. Theodorou, Geometric analysis of work of J.K. and B.D.F. was supported by the Australian-
Aquaporin Z. Proc. Natl. Acad. Sci. U.S.A. 104, 20719–20724 diffusion pathways in glassy and melt atactic polypropylene. American Fulbright Commission for the award to B.D.F. of the
(2007). doi: 10.1073/pnas.0708762104; pmid: 18077364 Macromolecules 26, 5461–5472 (1993). doi: 10.1021/ U.S. Fulbright Distinguished Chair in Science, Technology,
97. H. Wang et al., Highly permeable and selective pore- ma00072a026 and Innovation sponsored by the Commonwealth Scientific and
spanning biomimetic membrane embedded with aquaporin 114. S. R. Wickramasinghe, S. E. Bower, Z. Chen, A. Mukherjee, Industrial Research Organization (CSIRO); the Division of
Z. Small 8, 1185–1190, 1125 (2012). doi: 10.1002/ S. M. Husson, Relating the pore size distribution of Chemical Sciences, Geosciences, and Biosciences, Office of
smll.201102120; pmid: 22378644 ultrafiltration membranes to dextran rejection. J. Membr. Sci. Basic Energy Sciences of the U.S. Department of Energy (grant
98. Y. Zhao et al., Synthesis of robust and high-performance 340, 1–8 (2009). doi: 10.1016/j.memsci.2009.04.056 DE-FG02-02ER15362); and the International Institute for Carbon
aquaporin-based biomimetic membranes by interfacial 115. J. Bai, X. Zhong, S. Jiang, Y. Huang, X. Duan, Graphene Neutral Energy Research (WPI-I2CNER), sponsored by the World
polymerization-membrane preparation and RO performance nanomesh. Nat. Nanotechnol. 5, 190–194 (2010). Premier International Research Center Initiative (WPI), MEXT,
characterization. J. Membr. Sci. 423-424, 422–428 (2012). doi: 10.1038/nnano.2010.8; pmid: 20154685 Japan. The work of J.K. was also sponsored by the National
doi: 10.1016/j.memsci.2012.08.039 116. J. H. Morais-Cabral, Y. Zhou, R. MacKinnon, Energetic Science Foundation (NSF) Graduate Research Fellowship under grant
99. M. Kattula et al., Designing ultrathin film composite optimization of ion conduction rate by the K+ selectivity filter. DGE-1110007. The work of M.E. was supported by the National
membranes: The impact of a gutter layer. Sci. Rep. 5, 15016 Nature 414, 37–42 (2001). doi: 10.1038/35102000; Science Foundation through the Engineering Research Center for
(2015). doi: 10.1038/srep15016; pmid: 26456377 pmid: 11689935 Nanotechnology-Enabled Water Treatment (ERC-1449500) and by
100. X. Lu, L. H. Arias Chavez, S. Romero-Vargas Castrillón, J. Ma, 117. S. Gravelle et al., Optimizing water permeability through the grant CBET 1437630. We also thank E. Zumalt and P. Wiseman for
M. Elimelech, Influence of active layer and support layer hourglass shape of aquaporins. Proc. Natl. Acad. Sci. U.S.A. preparing the graphic for the print summary.
surface structures on organic fouling propensity of thin-film 110, 16367–16372 (2013). doi: 10.1073/pnas.1306447110;
composite forward osmosis membranes. Environ. Sci. pmid: 24067650 10.1126/science.aab0530

Park et al., Science 356, eaab0530 (2017) 16 June 2017 10 of 10

 Maximizing the right stuff: The trade-off between membrane permeability and
selectivity
Ho Bum ParkJovan KamcevLloyd M. RobesonMenachem ElimelechBenny D. Freeman

Science, 356 (6343), eaab0530. • DOI: 10.1126/science.aab0530

Filtering through to what's important

Membranes are widely used for gas and liquid separations. Historical analysis of a range of gas pair separations
indicated that there was an upper bound on the trade-off between membrane permeability, which limits flow rates, and
the selectivity, which limits the quality of the separation process. Park et al. review the advances that have been made
in attempts to break past this upper bound. Some inspiration has come from biological membranes. The authors also

Downloaded from https://www.science.org at Open University of Israel on November 22, 2021

highlight cases where the challenges lie in areas other than improved separation performance.
Science, this issue p. eaab0530

View the article online

https://www.science.org/doi/10.1126/science.aab0530
Permissions
https://www.science.org/help/reprints-and-permissions

Use of think article is subject to the Terms of service

Science (ISSN 1095-9203) is published by the American Association for the Advancement of Science. 1200 New York Avenue NW,
Washington, DC 20005. The title Science is a registered trademark of AAAS.
Copyright © 2017 The Authors, some rights reserved; exclusive licensee American Association for the Advancement of Science. No claim
to original U.S. Government Works