Gas Separation Membrane Materials: A Perspective

Perspective pubs.acs.org/Macromolecules

Gas Separation Membrane Materials: A Perspective Richard W. Baker and Bee Ting Low* Membrane Technology and Research, Inc., 39630 Eureka Drive, Newark, California 94560, United States ABSTRACT: The membrane gas separation industry is 35 years old and growing at a significant rate. Development of higher selectivity and higher permeance membranes would result in faster growth. This paper will analyze the barriers that have inhibited the development of these membranes. We start by reviewing the lessons that can be drawn from the past 35 years of experience. We then review the needs and most promising research directions for new materials in current and future membrane applications.

■

INTRODUCTION The first successful industrial membrane gas separation systems were built by Monsanto in 1979−1980. The systems separated hydrogen from the nitrogen, argon, and methane in ammonia synthesis plant purge gas. This was an easy separation for membranes: the economics of hydrogen recovery were good, and within a few years, Monsanto had installed several dozen systems worldwide. The membrane gas separation industry was off the ground. Since the 1980s, gas separation membranes have found applications in a variety of processes, most importantly nitrogen separation from air, CO2 from natural gas, and hydrogen from various refinery and petrochemical process streams. Membranes are also used on a smaller scale in many other applications.1,2 If improved membrane materials could be developed, the technology would be used more widely in the existing applications and in many others as well. The search for these better membrane materials has spurred a huge effort by many research groups, but for the most part, the results have been disappointing. Despite the synthesis and evaluation of hundreds, perhaps even thousands, of new materials, more than 90% of current commercial membranes are made from fewer than 10 membrane materials, most of which have been in use for decades. All the membranes in current commercial use are polymeric and separate gas mixtures by the solution diffusion mechanism. Our review will therefore concentrate on this type of membrane. We will only briefly describe the chemistry of membrane materials research. A number of recent reviews have covered this work in detail.3−6 Instead, this paper will analyze the barriers that have inhibited the development of better membrane materials and describe some potential solutions. We will also describe some of the most promising future applications and research approaches that may lead to useful new membranes. Lessons Learned. On the basis of the research results of the past 35 years, we have made a list of lessons learned. Other research groups have noted some of these issues before, and some may have additional lessons or may object to ours. © 2014 American Chemical Society

Whatever the case, those who do not learn from the mistakes of the past are doomed to repeat them. Pure-Gas Measurements Are Poor Predictors of the Performance of Industrial Membranes. Pure-gas permeance measurements are much easier to perform than gas mixture measurements. For this reason, essentially all new materials are initially characterized by measuring low-pressure, ambienttemperature, pure-gas permeabilities and selectivities, usually with a vacuum on the permeate side of the membrane. Measurement of the same permeants with gas mixtures is much less common, particularly measurements with gas mixtures at industrially relevant operating pressures and temperatures. This is a problem because the results of pure-gas measurements are often far from what is found with mixed-gas streams treated in industrial processes, particularly for separations involving condensable gases. The separation of CO2 from natural gas is a well-known example of this problem. New materials are routinely reported with CO2/CH4 pure-gas selectivities of 50 or more, combined with high CO2 permeabilities. Yet the membrane material installed in essentially all membrane natural gas processing plants remains cellulose acetate, with a selectivity, in use, of 10−15. This is because most materials, when tested with CO2− methane mixtures at the high pressures encountered in natural gas processing, lose much of their selectivity. Replacing pure-gas measurements with CO2−methane mixture measurements, although producing substantially more relevant results, does not completely solve the problem. Natural gas is not a simple mixture of CO2 and methane; it also contains 5−15% of a range of C2−C6 light hydrocarbons. More importantly, the gas also contains about 1000 ppm of water and up to 500 ppm of BTEX aromatics (benzene, toluene, ethylbenzene, and xylene).7 This may not sound like much, but at normal natural gas pressures of 20−60 bar, these components are close to their dew points (unit activity). Received: July 18, 2014 Revised: September 3, 2014 Published: September 17, 2014 6999

dx.doi.org/10.1021/ma501488s | Macromolecules 2014, 47, 6999−7013

Macromolecules

Perspective

or a high product recovery rate, but not both at the same time. This means operators of one-stage membrane processes must be ready to accept some product loss in the nonproduct gas stream leaving the process. Many applications cannot tolerate this loss of product. Process engineers use multistep or multistage membrane recycle processes to simultaneously achieve high purity and high recovery.15−17 Recycle operation requires compression equipment, and in most industrial membrane processes, the cost of the compression equipment needed is significantly larger than the membrane skid cost, and the cost of the power used by this equipment is the largest operating expense. Minimizing these costs means optimizing the process design to balance compression equipment costs against membrane selectivity and permeance. High membrane permeance is always good, but the optimum selectivity will depend on the process design and operating pressures. The optimum selectivity may not be the highest value possible, and different membrane selectivities may be optimum for different portions of an industrial separation plant.16 An iterative discussion between membrane material developers and process engineers is helpful in setting achievable targets for membrane performance. Process engineers are often ready to trade off selectivity for permeance and stability. Useful Membrane Materials Must Be Capable of Being Formed into Thin Membranes and Packaged into Large Area Membrane Modules. No matter how permeable a membrane material is used, most industrial processes require large membrane areas. Today’s industrial gas separation plants contain 1000−500 000 m2 of membrane. To minimize the membrane area required, membrane materials must be formed into thin membranes with a selective layer 0.1−1.0 μm thick. If a membrane material cannot be formed easily and reproducibly into a membrane this thin, it is unlikely to find a commercial use. For example, the difficulty of fabricating large areas of thin defect-free zeolite or carbon membranes has been the principal barrier preventing deployment of these and a number of similar, otherwise promising, membrane materials. Before Beginning a Membrane Optimization Program, Some Long-Term Stability Data Are Required. A question any producer of membrane gas separation equipment will have heard is “how long do these filters last?” This reasonable question is often left unaddressed in reports of new gas separation materials. Industrially, useful membranes need a lifetime of three to five years. Membranes that cannot achieve this lifetime are unlikely to be used. Having said this, even the best of today’s membranes show a decline in permeance of 30% or more over a three or four year lifetime, with most of the decline being in the first six months of operation. Chemical stability is sometimes an issue (polyimide membranes, for example, are degraded by trace amounts of amines), but physical stability is the more common problem. Loss of free volume in glassy polymers was mentioned above as a known limitation of some materials. Zeolite and carbonized membranes are less susceptible to free volume loss but can be slowly poisoned by minor components in the feed that lodge in the interstices of the membrane and stay there. Other membranes slowly compact under pressure. Long-term stability tests can be time-consuming. However, even simple mixed-gas test results lasting a few weeks can reveal obvious performance problems and should be conducted before embarking on any serious membrane optimization program. Finally, on a more positive note:

Significant amounts of these components dissolve in the membrane material, and the resulting plasticization and softening of the material significantly degrade the permeation properties of the membrane. Pure-gas measurements are a necessary first step in the new material screening process but must be followed by gas mixture measurements at realistic conditions. Permeation Properties of Thin Membranes Are Very Different from Thick Films. Most new polymer materials are evaluated as dense films 50−150 μm thick, but the selective layer of membranes in industrial use is usually 0.1−1.0 μm thick, supported on some type of high-permeability porous support that provides mechanical strength. Makers of thin membranes have known for many years that it is difficult to obtain the high selectivity and permeance that film data suggest should be possible. The difficulty of making completely defectfree thin membranes and the negative contribution of the support layer on membrane performance account for some of this problem, but this is only part of the story. Freshly made thin composite membranes will often lose 25% of their permeance within a few days and another 25% over the next week or two. Recent work by Paul and others from the University of Texas, Austin, has shown that this degradation in performance is due to reordering of the polymer chains making up the membrane, reducing the number and size of the free volume elements that contribute to gas permeation.8,9 The free volume elements move to the surface like tiny bubbles, leaving a viscous liquid. In a 50−150 μm thick film, the volume elements must move a considerable distance to escape, so reordering and elimination of free volume can take months or even years to become important. However, the rate of loss of the volume elements is proportional to the square of the film thickness. This means thin membranes 0.1−1.0 μm thick show signs of reordering and loss of permeance within a few days. In recent years, development of polymer membranes with high free volumes has become a hot research topic.10−13 These membranes are made from polymers with exceptionally rigid backbonessubstituted polyacetylenes and fused ring polybenzodioxanes, for example. Because of their exceptionally high free volume, these polymers are extremely permeable when measured as thick films and can have interesting selectivities for important gas mixtures. Although widely studied in the laboratory, these materials have found no industrial application because of the drastic reduction in permeance they experience over time when formed into thin membranes. The first two lessons learned and described above highlight the need to evaluate candidate membrane materials as thin membranes under realistic test conditions. It is possible that some of the hundreds of new membrane materials synthesized over the past 35 years, if evaluated this way, would be winners and significantly better than the existing standard materials. But absent realistic test data to separate the wheat from the chaff, these materials are doomed to stay in the laboratory. To Make Useful Membranes, You Have To Understand the Process. A significant disconnect exists between membrane materials scientists and industrial process engineers trying to develop new separation technology. Most membrane researchers are aware of the trade-off between permeability and selectivity. You can have high permeability or high selectivity, but it is hard to get both at the same time.14 Process engineers have other trade-offs to consider. One of the most important is the trade-off between product purity and product recovery. In a one-stage membrane process, you can have high product purity 7000

dx.doi.org/10.1021/ma501488s | Macromolecules 2014, 47, 6999−7013

Macromolecules

Perspective

Membrane Processes Can Tolerate High-Cost Membrane Materials. Many current membrane gas separation applications can tolerate a cost of $50/m2 for the membrane material. Some high value separations could probably support a higher cost. The amount of material used to produce a square meter of useful membrane varies over a wide range, depending on the way the membrane is made. The two most commonly used commercial membrane configurations are illustrated in Figure 1. Loeb−Sourirajan membranes are made from a single material

Figure 2. Comparison of permeation through a pore-flow membrane and a solution-diffusion membrane.

solution-diffusion model, it is assumed that no permanent pores exist and the gas dissolves in the membrane material as in a liquid. The dissolved gas then diffuses through the membrane by random diffusion down a concentration gradient.18 Pore-flow and solution-diffusion membranes have been studied since the work of Barrer and Van Amerongen in the 1950s and 1960s, but only dense polymeric solution-diffusion membranes have been used in commercial operations. In this review, we will focus on permeation through dense polymeric solution-diffusion membranes. Solution-Diffusion Membranes. Two basic assumptions underlie the solution-diffusion model. The first is that fluids on either side of the membranes are in equilibrium with the membrane material at the interface. The second is that the concentration and pressure differences across the membranes are represented only as the concentration gradient of dissolved permeate within the membrane. It can then be shown that Ji, the volume (molar) flux of component i through the membrane [cm3 (STP)/(cm2 s)], is given by the equation

Figure 1. Two main types of gas separation membranes. Loeb− Sourirajan membranes are made from a single polymer, so about 50 g of polymer is used per m2 of membrane. Multilayer composites are made from several different polymers. As little as 1−2 g of the expensive selective polymer per m2 of membrane is used in this type of membrane.

by a precipitation process. Because the same polymer is used in the microporous support and the selective layer, about 50 g of polymer is used per m2 of membrane. Multilayer composite membranes contain several layers made from different polymers. This is useful because the support membrane represents the bulk of the mass of the membranes and can be made from a relatively low cost material. As little as 1 or 2 g of a polymer with a high cost per m2 of membrane is sufficient to make the selective layer. It follows from the above that 1 kg of the selective material is sufficient to make about 20 m2 of a Loeb−Sourirajan membrane or as much as 500−1000 m2 of a multilayer composite membrane. Since gas separation membrane applications can support a material cost of $50/m2, expensive membrane materials costing as much as $25 to $50 thousand/ kg can be used in industrial plants if the appropriate membrane fabrication technique is used.

Ji =

DiK i(pi(o) − pi(l) ) S

(1)

where pi(o) is the partial pressure of component i at the feed side of the membrane; pi(l) is the partial pressure of i at the permeate side of the membrane; Di is the permeate diffusion coefficient [cm2/s], which is a measure of the mobility of individual molecules in the membrane; and Ki is the sorption coefficient with units [cm3 (STP) of component i/cm3 of polymer pressure]. The product DiKi can be written as Pi, called the membrane permeability, and measures the ability of a membrane material to permeate gas.a Pi = DiK i (2) The best measure of the ability of a membrane to separate two gases is the ratio of the permeabilities, αij, called the membrane selectivity

■

MEMBRANE STRUCTURE AND MECHANISMS OF GAS PERMEATION The way gases and vapors permeate membranes depends on the membrane material and gas composition. From the very early days of membrane permeation, two models were used to describe the mechanism of permeation. These models are illustrated in Figure 2. In the pore-flow model, gases permeate the membrane through tiny pores and gas mixtures are separated by some form of molecular filtration. In the

αij =

⎛ D ⎞⎛ K ⎞ Pi = ⎜⎜ i ⎟⎟⎜⎜ i ⎟⎟ Pj ⎝ Dj ⎠⎝ Kj ⎠

(3)

Di/Dj, the ratio of the diffusion coefficients of the two gases, can be viewed as the mobility selectivity, reflecting the different size of the two molecules. Ki/Kj, the ratio of the sorption coefficients, can be viewed as the sorption selectivity, reflecting the relative solubility of the two gases in the membrane material. 7001

dx.doi.org/10.1021/ma501488s | Macromolecules 2014, 47, 6999−7013

Macromolecules

Perspective

Sorption Coefficients. The sorption coefficient Ki for a particular gas i is surprisingly constant in a wide range of chemically different polymers. This is because gas sorption in most polymers behaves as though the polymers were ideal liquids. It can be shown that all ideal liquids should have the same sorption for the same gas and that the sorption of different gases is inversely proportional to their saturation vapor pressure.19 A convenient measure of the saturation vapor pressure is the gas boiling point or critical temperature, and so the sorption of gases can be conveniently correlated with the gas critical temperature as shown in Figure 3. Sorption data for

Figure 4. Diffusion coefficient as a function of molar volume for a variety of gas permeants in water, natural rubber, and poly(vinyl chloride), a glassy polymer. This type of plot was first drawn by Gruen23 and has been revised by many others since. Figure 3. Solubilities as a function of critical temperature (Tc) for a typical glassy polymer (polysulfone) and a typical rubbery polymer (silicone rubber) compared with values for an ideal liquid.20

do not change by more than a factor of 2−3 in all of these materials. The ratios of the sorption coefficients, that is, the sorption selectivity terms KCO2/KCH4, are even more tightly grouped. This is because if a polymer has a higher or lower sorption coefficient than the ideal value for one gas, the coefficients for other gases are usually affected in the same way. In contrast to the sorption data, the diffusion coefficients shown in Table 1 are very different for different materials. For example, the diffusion coefficient for CO2 varies by 3 orders of magnitude from a high of 2200 × 10−8 cm2/s in silicone rubber to a low of 0.56 × 10−8 cm2/s in a Kapton polyimide. The ratio of the CO2 and CH4 diffusion coefficients is closer, but still differ by 10-fold between the most permeable and the least permeable polymer. Much of the difference in membrane selectivity for different materials is due to differences in this mobility selectivity term. The data in Table 1 also highlight a trade-off between membrane permeability (P) and membrane selectivity (α). The polymers in Table 1 are listed in order of decreasing CO2 permeability, from silicone rubber (PCO2 = 3800 barrer) to the Kapton polyimide (PCO2 = 2.7 barrer). This is almost the reverse of the order of overall membrane selectivity: from silicone rubber (αCO2/CH4 = 3.2) to Kapton (αCO2/CH4 = 46). This trade-off was noticed and organized into a very useful plot by Robeson.14,27 An example of a Robeson plot for the separation of O2/N2 mixtures is shown in Figure 5. Each point on the figure represents the pure-gas selectivity and permeability of a particular polymer. Many materials are shown, but for gas separation applications, the most permeable material at a particular selectivity is of the most interest. The line that links these materials is called the upper bound, beyond which no better materials are known. The inverse relationship

an ideal liquid and two very different polymerspolysulfone, a tough amorphous glassy polymer, and silicone rubber, a soft elastomerare shown.20 The sorption coefficients of gases in both materials are close and also close to the calculated sorption of an ideal liquid. Most other conventional polymers have gas sorption in the range shown. The only outliers are cases where there are specific interactions between polymers and gases, such as perfluorinated polymers that have anomalously low hydrocarbon solubilities21 and some ultrahigh free volume polymers, which have anomalously high gas sorptions.22 Diffusion Coefficients. In contrast to gas sorption, gas diffusion coefficients vary widely depending on the polymer material. Some data that illustrate this result are shown in Figure 4. The diffusion coefficients of gases in natural rubber are high and within an order of magnitude of the diffusion coefficient of the same gas in a liquid such as water. On the other hand, the diffusion coefficients of the same gases in unplasticized poly(vinyl chloride)a tough, high glass transition amorphous glassy polymerare 5−8 orders of magnitude smaller. As the size of the gas permeants increases, the permeant diffusion coefficient also falls very sharply. Most polymers fall between these two curves, which represent the extremes of what is normally seen. Figures 3 and 4 provide a basis to rationalize the mobility and sorption selectivity contributions to the overall membrane selectivity given in eq 3. Table 1 shows pure-gas sorption and diffusion coefficients for CO2 and methane in a variety of polymer materials. The sorption coefficients for CO2 or CH4 7002

dx.doi.org/10.1021/ma501488s | Macromolecules 2014, 47, 6999−7013

Macromolecules

Perspective

Table 1. Diffusion and Sorption Selectivities for Carbon Dioxide/Methane for a Variety of Polymers24−26

polymer rubbery polymers silicone rubber polyisoprene glassy polymers poly(ethylene terephthalate) polystyrene polycarbonate polysulfone Kapton polyimide

permeability PCO2 (barrer) 3800 153

diffusion coefficient DCO2 (10−8 cm2/s) 2200 125

sorption coefficient KCO2 (cm3 (STP)/(cm3 atm))

sorption coefficient KCH4 (cm3 (STP)/(cm3 atm))

total selectivity αCO2/CH4

diffusion selectivity DCO2/DCH4

sorption selectivity KCO2/KCH4

1.29 0.94

0.42 0.26

3.2 5.1

1.1 1.4

3.1 3.6

17.2

4.46

2.9

0.83

27.3

7.8

3.5

12.4 6.8 5.6 2.7

8.50 3.20 2.00 0.56

1.1 1.6 2.1 3.6

0.38 0.40 0.59 0.95

15.8 19 22 46

5.5 4.7 5.9 11.9

2.9 4.0 3.7 3.8

When originally developed, the position of the upper bound lines in the Robeson plots was a completely empirical result. But Freeman has since shown that the upper bound curves shown in Figure 6 can be represented by the expression ln ai / j = ln βi / j − λi / j ln Pi

(4)

or λi / j

ai / j = βi / j /Pi

(5)

where i and j represent the two gases, λi/j is the slope of the lines in Figure 6 for the two gases, and ln βi/j is the intercept at ln Pi = 0.28 These parameters have physical significance and can be calculated with reasonable accuracy from first principles. The slope λi/j depends only on the size difference of the gas pair, and βi/j depends on the gas condensability. An implication of Freeman’s derivation, consistent with what Robeson found, is that the position of the upper bound is unlikely to vary as more polymeric solution-diffusion membranes are developed. Most of the materials that have significantly crossed the upper bound in recent yearsfor example, polymers of intrinsic microporosity and thermally rearranged polymershave a pore diffusion−molecular sieving component in their transport mechanism. In these materials, a portion of the permeation process is rapid diffusion in micropores, followed by a slower rate-limiting diffusion−molecular sieving step at pore constrictions. This leads to anomalously high permeabilities. In polymeric solution-diffusion materials, a high selectivity between two gases is best achieved when the permeable component is both smaller and more condensable than the nonpermeating component; for example, the separation of CO2 from methane. CO2 is slightly smaller than methane, giving mobility selectivities of 2−10. CO2 is also more condensable than methane, giving sorption selectivities of 3−4. The result is that all membranes preferentially permeate CO 2 with selectivities that range from 6 to 40. In some applications, the sorption selectivity and the mobility selectivity in eq 3 are opposed. For example, the separation of hydrogen from CO2 is an important separation in hydrogen production plants. CO2 (bp = −56 °C) is more condensable than hydrogen (bp = −253 °C), so the sorption selectivity term KCO2/KH2 always favors permeation of CO2. On the other hand, CO2 (kinetic diameter = 3.3 Å) is larger than hydrogen (kinetic diameter = 2.9 Å), so the mobility selectivity DCO2/DH2 always favors permeation of hydrogen. Depending on the nature of the polymer, it is possible to produce membranes that selectively permeate CO2 using the sorption selectivity or

Figure 5. Oxygen/nitrogen selectivity as a function of oxygen permeability. This plot by Robeson shows the wide range of selectivity and permeability combinations achieved by conventional materials.14

between permeability and selectivity is clear. Robeson plots have been created for other gas pairs of interest. The upper bound lines taken from some of these plots are shown in Figure 6. These upper bound lines are useful in showing the separation performances that can be expected for currently best membrane materials in ideal conditions.

Figure 6. Upper bound Robeson selectivity/permeability lines for a number of commercially important gas separations. This figure allows the trade-off between selectivity and permeability to be estimated for the best available membrane materials. 7003

dx.doi.org/10.1021/ma501488s | Macromolecules 2014, 47, 6999−7013

Macromolecules

Perspective

where better membranes would make a significant difference.7 About 15% of natural gas must be treated to remove CO2, so the gas can meet the pipeline specification of cyclic chain > branched chain. The general approach used to make the membranes is to prepare segmented block copolymers consisting of hard segments not swollen by the hydrocarbon oil (to provide mechanical strength) and soft segments through which the permeation takes place. Crosslinking is used to control swelling. Typical membrane materials are polyester−polyimide and polyurea−polyurethane block copolymers.

■

■

CONCLUSIONS The membrane gas separation industry is now 35 years old. In the existing applicationsnitrogen from air, hydrogen from light hydrocarbons, and hydrocarbons from nitrogen and methanethe membrane materials used have not changed significantly in a decade. And there has been little commercial driving force for the development of better materials for established membrane separations. The glaring exception is the separation of CO2 from natural gas. The existing industry standard is cellulose acetate. More selective, higher permeance membranes would greatly enhance the competitiveness of membrane technology in this large application. There are many reports of new materialscharacterized to date only as films and with separation performance reported only as low-pressure pure-gas datathat appear to have better properties than cellulose acetate. But so far, when these materials are characterized as thin membranes with gas mixtures at industry relevant operating conditions, the separation performances of all of them has come up short. Some recent studies with crosslinked polyimides look promising. Some PIM and TR polymers have also been reported with exceptional properties, but their membrane permeation properties decline over time. There is a need for better membranes in several emerging applications areas. In most of these applications, the ability of conventional materials to provide the required separations has been examined, and it will be difficult, but not impossible, to find conventional polymers that are significantly better than what has been developed to date. The best hopes for significant improvements are in the novel materials category; key examples include MOF mixed-matrix membranes for olefin/paraffin separations and high free volume thermally rearranged (TR) and PIM membranes for separations of oxygen from air, olefins from paraffins, and perhaps even for CO2 from natural gas. But these materials will go nowhere unless their developers face up to some of the real-world deficiencies of their membranes: poisoning by contaminants such as water or toluene and slow (and sometimes not-so-slow) changes in membrane selectivity and permeance over time. Also, there is a need to address how to transform new membrane materials from 100 μm thick films made as small stamps to membranes with submicron thick selective layers made as large membrane area modules. Finally, the separation of vapor/vapor mixtures is an area where development of new selective materials and predictive structure−property studies would be useful. The size of this potential application is huge, and there is a real need for basic permeation data. Here, too, the value of the data would be greatly enhanced if it were measured at the conditions likely to be found in the final application: vapor mixtures at temperatures above 100 °C. Use of membrane technology for vapor/ vapor separations has reached the pilot-plant scale for a number of applications. Examples include a report of methanol

AUTHOR INFORMATION

Corresponding Author

*Tel +1 650-543-3392; e-mail [email protected] (B.T.L.). Notes

The authors declare no competing financial interest. Biographies

Dr. Baker received his doctorate in physical chemistry in 1966 at Imperial College, London, where he studied under Professor R. M. Barrer, one of the pioneers of membrane science. Subsequently, he joined Amicon Corporation, Lexington, MA, and developed a series of ultrafiltration membranes now sold under the name Diaflow. While at Alza Corporation (Palo Alto, CA) from 1971 to 1974, he collaborated in the development of the Ocusert ocular delivery system. In 1974, he cofounded Bend Research, Inc. (Bend, OR), where he was the Director of Research until 1981. Dr. Baker founded Membrane Technology and Research, Inc. (MTR), in 1982 as a research and development company specializing in membrane technology. He served as MTR’s President for 25 years. In that time, MTR became a leading membrane research, development, engineering, and production company, concentrating on the development of membranes and membrane systems for industrially and environmentally significant separations. Dr. Baker currently serves as MTR’s Principal Scientist. Dr. Baker is the author of more than 100 papers and over 100 patents, all in the membrane area. Three editions of his book, Membrane Technology and Applications, were published in 2000, 2004, and 2012. He serves on the editorial board of the Journal of Membrane Science. He was a cofounder of the North American Membrane Society (NAMS). In 2002, he was recipient of the first NAMS Alan S. Michaels Award for Innovation in Membrane Science and Technology. 7011

dx.doi.org/10.1021/ma501488s | Macromolecules 2014, 47, 6999−7013

Macromolecules

Perspective

(5) Stern, S. A. J. Membr. Sci. 1994, 94, 1−65. (6) Xiao, Y.; Low, B. T.; Hosseini, S. S.; Chung, T. S.; Paul, D. R. Prog. Polym. Sci. 2009, 34, 561−580. (7) Baker, R. W.; Lokhandwala, K. Ind. Eng. Chem. Res. 2008, 47, 2109−2121. (8) Huang, Y.; Paul, D. R. Ind. Eng. Chem. Res. 2007, 46, 2342−2347. (9) Rowe, B. W.; Freeman, B. D.; Paul, D. R. In Membrane Engineering for the Treatment of Gases; Drioli, E., Barbier, G., Eds.; Royal Society of Chemistry: Cambridge, UK, 2011; Vol. 1, Chapter 3, pp 58−83. (10) Park, H. B.; Jung, C. H.; Lee, Y. M.; Hill, A. J.; Pas, S. J.; Mudie, S. T.; Van Wagner, E.; Freeman, B. D.; Cookson, D. J. Science 2007, 318, 254−258. (11) Pinnau, I.; Cassillas, C. G.; Morisato, A.; Freeman, B. D. J. Polym. Sci., Part B: Polym. Phys. 1996, 43, 2613−2621. (12) McKeown, N. B.; Budd, P. M. Macromolecules 2010, 43, 5163− 5176. (13) Du, N.; Dal-Cin, M. M.; Pinnau, I.; Nicalek, A.; Robertson, G. P.; Guiver, M. D. Macromol. Rapid Commun. 2011, 32, 631−636. (14) Robeson, L. M. J. Membr. Sci. 1991, 62, 165−185. (15) Baker, R. W.; Wijmans, J. G.; Kaschemekat, J. H. J. Membr. Sci. 1998, 151, 55−62. (16) Huang, Y.; Merkel, T. C.; Baker, R. W. J. Membr. Sci. 2014, 463, 33−40. (17) Zolandz, R. R.; Fleming, G. K. In Membrane Handbook; Ho, W. S., Sirkar, K. E., Eds.; Van Nostrand-Reinhold: New York, 1992; p 54. (18) Wijmans, J. G.; Baker, R. W. J. Membr. Sci. 1995, 107, 1−21. (19) Denbeigh, K. In The Principles of Chemical Equilibrium; Cambridge University Press: Cambridge, 1960. (20) Freeman, B. D.; Pinnau, I. ACS Symp. Ser. 1999, 733, 1−27. (21) Merkel, T. C.; Pinnau, I.; Prabhakar, R.; Freeman, B. D. Gas and Vapor Transport Properties of Perfluoropolymers. In Materials Science of Membranes for Gas and Vapor Separations; Freeman, B. D., Pinnau, I., Yampolskii, Y., Eds.; John Wiley and Sons: New York, 2006. (22) Merkel, T. C.; Bondar, V.; Nagai, K.; Freeman, B. D. Macromolecules 1999, 32, 370−374. (23) Gruen, F. Experimentia 1947, 3, 490−492. (24) Tanaka, K.; Kita, H.; Okano, M.; Okamoto, K.-I. Polymer 1992, 33, 585−592. (25) Koros, W. J.; Coleman, M. R.; Walker, D. R. B. Annu. Rev. Mater. Sci. 1992, 22, 47−89. (26) Merkel, T. C.; Bondar, V. I.; Nagai, K.; Freeman, B. D.; Pinnau, I. J. Polym. Sci., Part B: Polym. Phys. 2000, 38, 415−434. (27) Robeson, L. M. J. Membr. Sci. 2008, 320, 390−400. (28) Freeman, B. D. Macromolecules 1999, 32, 375−380. (29) Merkel, T. C.; Zhou, M.; Baker, R. W. J. Membr. Sci. 2012, 389, 441−450. (30) Puleo, A. C.; Paul, D. R.; Kelley, S. S. J. Membr. Sci. 1989, 47, 301−332. (31) White, L. S.; Blinka, T. A.; Kloczewski, H. A.; Wang, I.-F. J. Membr. Sci. 1995, 103, 73−82. (32) Vanherck, K.; Koeckelberghs, G.; Vankelecom, I. F. J. Prog. Polym. Sci. 2013, 38, 874−896. (33) Wind, J. D.; Staudt-Bickel, C.; Paul, D. R.; Koros, W. J. Macromolecules 2003, 36, 1882−1888. (34) Ma, C.; Koros, W. J. J. Membr. Sci. 2013, 428, 251−259. (35) Masuda, T.; Isobe, E.; Higashimura, T.; Takada, K. J. Am. Chem. Soc. 1983, 105, 7473−7474. (36) Budd, P. M.; McKeown, N. B. Polym. Chem. 2010, 1, 63−68. (37) Du, N.; Park, H. B.; Dal-Cin, M. M.; Guiver, M. D. Energy Environ. Sci. 2012, 5, 7306−7322. (38) Du, N.; Dal-Cin, M. M.; Robertson, G. P.; Guiver, M. D. Macromolecules 2012, 45, 5134−5139. (39) Sanders, D. F.; Smith, Z. P.; Ribeiro, C. P.; Guo, R.; McGrath, J. E.; Paul, D. R.; Freeman, B. D. J. Membr. Sci. 2012, 409, 232−241. (40) Han, S. H.; Lee, Y. M. In Membrane Engineering for the Treatment of Gases; Drioli, E., Barbieri, G., Eds.; The Royal Society of Chemistry: Cambridge, 2011; Vol. 1, Chapter 4, p 116.

Dr. Low received her bachelor and doctorate degrees in chemical and biomolecular engineering in 2006 and 2010 at the National University of Singapore (NUS). Her thesis work with Professor Tai-Shung Chung at NUS was centered on the design of polymer membranes for hydrogen enrichment and natural gas sweetening. She also worked one year as a NUS Research Fellow, leading a project on polymer membranes for postcombustion CO2 capture from flue gas. Dr. Low joined Membrane Technology and Research, Inc. (MTR), in December 2010 as a Senior Research Scientist. Her first project at MTR was to develop energy-efficient membrane systems for building applications. She also worked with MTR teams on optimizing the performance of oxygen enrichment and natural gas dehydration membranes. Her work in the past two to three years has been focused on new approaches to develop membranes for the separation of carbon dioxide from natural gas and the separation of propylene/ propane mixtures. Dr. Low earned several honors and awards for her university work. In 2010, Dr. Low received an award in the Green Talents Competition sponsored by the German Federal Ministry of Education and Research and given to young researchers with potential in the field of sustainable development. In 2011, with the support of BMBF, she participated in one month of research study at the Institute of Energy Research, Forschungszentrum Jülich in Germany. Dr. Low has coauthored 14 publications and book chapters.

■

ACKNOWLEDGMENTS The authors thank Dr. Timothy C. Merkel and Dr. Meijuan Zhou for their valuable suggestions and perspective in writing this paper.

■

ADDITIONAL NOTE The permeability of gases through membranes is most commonly measured in barrer, defined as 1 × 10−10 cm3 (STP) cm/(cm2 s cmHg) and named after R. M. Barrer, a pioneer in gas permeability measurements. The term Ji/(pi(o) − pi(l)), best called the pressure-normalized flux or permeance, is often measured in terms of gas permeation units (gpu), where 1 gpu is defined as 1 × 10−6 cm3 (STP)/(cm2 s cmHg). Occasional IUPAC purists insist on writing permeability in terms of mol m/(m2 s Pa) (1 barrer = 0.33 × 10−15 mol m/(m2 s Pa)), but fortunately this has not caught on. a

■

REFERENCES

(1) Bernardo, P.; Drioli, E.; Golemme, G. Ind. Eng. Chem. Res. 2009, 48, 4638−4663. (2) Baker, R. W. Ind. Eng. Chem. Res. 2002, 41, 1393−1411. (3) Sanders, D. F.; Smith, Z. P.; Guo, R.; Robeson, L. M.; McGrath, J. E.; Paul, D. R.; Freeman, B. D. Polymers 2013, 54, 4729−4761. (4) Yampolskii, Y. Macromolecules 2012, 45, 3298−3311. 7012

dx.doi.org/10.1021/ma501488s | Macromolecules 2014, 47, 6999−7013

Macromolecules

Perspective

(76) Berchtold, K. A.; Singh, R. P.; Young, J. S.; Dudeck, K. W. J. Membr. Sci. 2012, 415−416, 265−270. (77) Sandberg, J.; Wang, C.; Larsson, M. Int. J. Energy Eng. 2013, 3, 28−33. (78) Bhide, B. O.; Stern, S. A. J. Membr. Sci. 1991, 62, 13−35. (79) Bezzu, C. G.; Carta, M.; Tonkins, A.; Jansen, J. C.; Bernardo, P.; Bazzarelli, F.; McKeown, N. B. Adv. Mater. 2012, 24, 5930−5933. (80) Carta, M.; Malpass-Evans, R.; Croad, M.; Rogan, Y.; Jansen, J. C.; Bernardo, P.; Bazzarelli, F.; McKeown, N. B. Science 2013, 339, 303−307. (81) Ghanem, B. S.; Swaidan, R.; Litwiller, E.; Pinnau, I. Adv. Mater. 2014, 26, 3688−3692. (82) Lau, C. H.; Nguyen, P. T.; Hill, M. R.; Thornton, A. W.; Konstas, K.; Doherty, C. M.; Mulder, R. J.; Bourgeois, L.; Liu, A. C. Y.; Sprouster, D. J.; Sullivan, J. P.; Bastow, T. J.; Hill, A. J.; Gin, D. L.; Noble, R. D. Angew. Chem., Int. Ed. 2014, 53, 5322−5326. (83) Minhas, B. S.; Chuba, M. R.; Saxton, R. J. U.S. Patents 6,649,061, 2003; 6,702,945, 2004. (84) White, L. S.; Wormsbecher, R. F.; Lesemann, M. U.S. Patent 7,041,212, 2006. (85) Schucker, R. C. U.S. Patents 4,983,338, 1991; 5,063,186, 1991; 5,055,632, 1991. (86) Matsui, S.; Paul, D. R. J. Membr. Sci. 2003, 213, 67−83. (87) Huang, Y.; Baker, R. W.; Vane, L. M. Ind. Eng. Chem. Res. 2010, 49, 3760−3768. (88) Wijmans, J. G.; Baker, R. W.; Mairal, A. P. U.S. Patent 6,899,743, 2005. (89) Morigami, Y.; Konedo, M.; Abe, J.; Kita, H.; Okamoto, K. Sep. Purif. Sci. 2001, 25, 251−260.

(41) Baker, R. W.; Lokhandwala, K. A.; Jacobs, M. L.; Gottschlich, D. E. Chem. Eng. Prog. 2000, 96, 51−57. (42) Ohlrogge, K.; Wind, J.; Behling, R. D. Sep. Sci. Technol. 1995, 30, 1625−1638. (43) Pinnau, I.; He, Z. J. Membr. Sci. 2004, 244, 227−233. (44) Stern, S. A.; Shah, V. M.; Hordey, B. J. J. Polym. Sci., Phys. Ed. 1987, 25, 1263−1298. (45) Pinnau, I.; Toy, L. G. J. Membr. Sci. 1996, 116, 199−209. (46) Thomas, S.; Pinnau, I.; Du, N.; Guiver, M. D. J. Membr. Sci. 2009, 338, 1−4. (47) Eldrige, R. B. Ind. Eng. Chem. Res. 1993, 32, 2208−2212. (48) Baker, R. W., DaCosta, A. R.; Daniels, R.; Pinnau, I.; He, Z. U.S. Patent 6,271,319, 2001. (49) Gottschlich, D. E.; Roberts, D. L. U.S. Department of Energy Report DOE/ID-10301, 1990. (50) Xu, L.; Rungta, M.; Brayden, M. K.; Martinez, M. V.; Stears, B. A.; Barbay, G. A.; Koros, W. J. J. Membr. Sci. 2012, 423, 314−323. (51) Hughes, R. D.; Mahoney, J. A.; Steigelmann, E. F. In Recent Developments in Separation Science; CRC Press: Boca Raton, FL, 1986; p 173. (52) Merkel, T. C.; Blanc, R.; Ciobanu, I.; Firat, B.; Swuarlim, A.; Zeid, J. J. Membr. Sci. 2013, 447, 177−189. (53) Burns, R. L.; Koros, W. J. J. Membr. Sci. 2003, 211, 299−309. (54) Shimazu, A.; Miyazaki, T.; Maeda, M.; Ikeda, K. J. Polym. Sci., Part B: Polym. Phys. 2000, 38, 2525−2536. (55) Roman I. C.; Simmons, J. W.; Ekiner, O. M. U.S. Patent 7,399,897, 2008. (56) Liu, C.; Kulprathipanja, S.; Hilock, A. M. W.; Husain, S.; Koros, W. J. In Advanced Membrane Technology and Applications; John Wiley: West Sussex, 2005; p 787. (57) Mahajan, R.; Koros, W. J. Polym. Eng. Sci. 2002, 42, 1420−1431. (58) Jeazet, H. B. T.; Staudt, C.; Janiak, C. Dalton Trans. 2012, 41, 14003−14027. (59) Thornton, A. W.; Dubbeldam, D.; Liu, M. S.; Ladewig, B. P.; Hill, A. J.; Hill, M. R. Energy Environ. Sci. 2012, 5, 7637−7646. (60) Kwon, H. T.; Jeong, E.-K. J. Am. Chem. Soc. 2013, 135, 10763− 10768. (61) Zhang, C.; Dai, Y.; Johnson, J. R.; Karvan, O.; Koros, W. J. J. Membr. Sci. 2012, 389, 34−42. (62) Merkel, T. C.; Lin, H.; Wei, X.; Baker, R. W. J. Membr. Sci. 2010, 359, 126−139. (63) Favre, E. J. Membr. Sci. 2007, 294, 50−59. (64) Ramasubramanian, K.; Ho, W. S. W. Curr. Opin. Chem. Eng. 2011, 1, 47−54. (65) Blume, I.; Pinnau, I. Composite membrane, method of preparation and use. U.S. Patent 4,963,165, 1990. (66) Car, A.; Stropnik, C.; Yave, W.; Peinemann, K. V. Adv. Funct. Mater. 2008, 18, 2815−2823. (67) Yave, W.; Car, A.; Funari, S. S.; Nunes, S. P.; Peinemann, K.-V. Macromolecules 2010, 43, 326−333. (68) Yave, W.; Car, A.; Wind, J.; Peinemann, K. V. Nanotechnology 2010, 21, 395303−395310. (69) Hirayama, Y.; Kase, Y.; Tanihara, N.; Sumiyama, Y.; Kusuki, Y.; Haraya, K. J. Membr. Sci. 1999, 160, 87−99. (70) Sforca, M. L.; Yoshida, I. V. P.; Nunes, S. P. J. Membr. Sci. 1999, 159, 197−207. (71) Brinkmann, T.; Wolff, T.; Pauls, J.-R. “Post-Combustion Processes Employing Polymeric Membranes”, 2nd International Conference on Energy Process Engineering Efficient Carbon Capture for Coal Power Plants, June 20−22, 2011, in Frankfurt/Main, Germany. (72) Kim, T.-J.; Vralstad, H.; Sandru, M.; Hagg, M.-B. J. Membr. Sci. 2013, 428, 218−224. (73) Lin, H.; Van Wager, E.; Freeman, B. D.; Toy, L. G.; Gupta, R. P. Science 2006, 311, 639−642. (74) Zhao, Y.; Ho, W. S. W. J. Membr. Sci. 2012, 415−416, 132−138. (75) Pesiri, D. R.; Jorgensen, B.; Dye, R. C. J. Membr. Sci. 2003, 218, 11−18. 7013

dx.doi.org/10.1021/ma501488s | Macromolecules 2014, 47, 6999−7013