Designing the Next Generation of Chemical Separation Membranes

See allHide authors and affiliations

Science  06 May 2011:
Vol. 332, Issue 6030, pp. 674-676
DOI: 10.1126/science.1203771

Synthetic membranes are used in many separation processes, from industrial-scale ones—such as separating atmospheric gases for medical and industrial use, and removing salt from seawater—to smaller-scale processes in chemical synthesis and purification. Membranes are commonly solid materials, such as polymers, that have good mechanical stability and can be readily processed into high–surface area, defect-free, thin films. These features are critical for obtaining not only good chemical separation but also high throughput. Membrane-based chemical separations can have advantages over other methods—they can take less energy than distillation or liquefaction, use less space than absorbent materials, and operate in a continuous mode. In some cases, such as CO2 separations for CO2 capture, their performance must be improved. We discuss how membranes work, and some notable new approaches for improving their performance.

Membranes work by forming a barrier between two phases (e.g., salt water and fresh water) that restricts the movement of some molecules while letting others through (1). Separation is driven by a difference in concentration or pressure (or both) across the membrane—pressurization is the main energy input. Membranes are either dense or porous, depending on how the molecules move across the barrier. In dense membranes, molecules dissolve into the material and diffuse through it (1). The product of a molecule's solubility and diffusivity is its permeability P, and membranes are designed so that two (or more) molecules have different P values (see the figure, panel A).

Chemical separation with dense and porous membranes.

(A) For dense membranes, orange and green molecules move through the membranes at different rates because they have different permeabilities P. The Robeson plot shows that conventional dense membranes separate mainly via differences in diffusivity, and performance is limited by an “upper bound.” (B) Nanoporous membranes separate via molecular size differences. The examples shown are for pores that interconnect into a 3D network. With uniform pore sizes, it is possible to get complete separation (smaller molecules pass through—they have a higher molecular flux; larger ones are completely blocked). With nonuniform pores, the largest pore sizes (i.e., a distribution) dictate the selectivity, and both molecules can pass through.

In contrast, porous membranes for chemical separations should have pores that are ideally the size of single small molecules (≤1 nm), and molecules pass through as a gas or a liquid solution. Molecules can be separated by size because the nanopores act as a screen or sieve (1). Uniform nanopores with the correct size must be continuous across the membrane. If there is a distribution of pore sizes, most of the molecules will pass through the largest pores first (path of least resistance), thereby compromising selectivity (see the figure, panel B).

The limitation of conventional dense polymer membranes is best illustrated by gas separations. Differences in P mainly arise from diffusion differences of each gas component in the polymer film—solubilities tend to be similar. Materials that are more permeable (i.e., that have higher throughput and would process gas faster) also have a more open structure and thus have lower selectivity. This compromise leads to an upper bound for separation performance that is shown graphically by plots generated by Robeson (2).

A new approach in the design of dense membranes is to use room-temperature ionic liquids (RTILs) in various morphologies. RTILs are liquid-phase organic salts (i.e., ionic compounds) with negligible vapor pressure (avoiding evaporation losses), high thermal stability, and intrinsic solubility for certain gases. Unlike conventional polymers, RTILs perform gas separations via solubility differences. They have been applied to CO2 separations, but the wide variety of RTILs allows for many candidate separations. Also, RTIL derivatives can be prepared that allow the molecules to be polymerized to form solids so that membranes can be prepared in various morphologies, such as films (3), solid composite structures with RTIL within the structure (3), and gels (4). For RTIL morphologies, the limit on a Robeson plot is determined by the performance of the liquid itself as a membrane because the other morphologies have the same solubility selectivity but different diffusion rates.

Poly(RTIL) films have recently been tested for gas separations up to 40 atm pressure for mixtures of CO2 and CH4. The conditioning or morphology change caused by the incorporation of gases such as CO2 at these high pressures is reversible compared to conventional polymers that exhibit irreversible conditioning at these pressures (5). Examples of improved separations include CO2 from N2 for postcombustion cleanup of flue gas from power plants.

For nanoporous membranes, several methods have recently been developed that afford materials with uniform molecular-size pores. For example, deposition techniques have been successfully used to reduce the pore size of commercial nanoporous polymer and ceramic membranes down to molecular dimensions (6). Recent advances in blending organic polymers with inorganic zeolites have afforded viable composite membranes with uniform pore sizes in the 0.3- to 0.7-nm range for light gas separations, such as CO2, N2, and CH4 (7).

Similar approaches for making polymer-based composite membranes containing metal-organic framework compounds (8), carbon nanotubes (9), and peptide nanotubes (10) as the porous component have also been found to be promising for separating different-size ions in water and mixtures of light gases. “Molecular square” coordination compounds (11) and macrocyclic surfactants (12) also form membranes with molecular-size pores when applied as ordered thin layers on membrane supports. Ordered surfactant liquid crystal assemblies formed in water have successfully been polymerized into membranes with three-dimensional (3D) interconnected pores, smaller than 1 nm in diameter, that act as molecular sieves for water desalination (13). Selectively etched phase-segregated block copolymers (14) and colloidal crystal assemblies (15) are two promising platforms with uniform pores in the 10- to 100-nm range. Such materials are useful for macromolecular or protein separations, but methods are needed to bring the pore sizes down to molecular dimensions if they are to be useful for small-molecule separations.

Other factors are also important for making practical membranes with high throughput, including high pore density, pore continuity, and the ability to form defect-free thin films. Several of the approaches listed above form 1D columnar pores (6, 912) that need to be aligned in the flow direction and packed closely together for high membrane flux. Materials with 3D-interconnected pores (7, 8, 1315) have the advantage that the pores need not be aligned to be continuous across the membrane, and are not easily blocked. These materials also often have better overall pore densities. Although many of the above materials can be processed into films, only a handful have been formed into films thin enough (6, 8, 11, 12, 14) to achieve high fluxes (i.e., thinner membranes have less flow resistance). For researchers working on new dense or porous materials for membrane applications, it is important to consider not only the design factors that afford better separation selectivity but also the factors that afford good productivity.

The future directions for these new membrane materials are very promising, primarily because of the enormous chemical flexibility of their base structures. The separation properties for the application of interest can be tuned, as can operational parameters such as stability and longevity. In addition, functional additives such as selective complexing agents can be incorporated into these new classes of membrane materials, providing exciting new opportunities for enhancing separation performance.


View Abstract

Stay Connected to Science

Navigate This Article