Membrane separation science has enjoyed tremendous progress since the first synthesis of membranes almost 40 years ago, which was driven by strong technological needs and commercial expectations. As a result, the range of successful applications of membranes and membrane processes is continuously broadening. An additional change lies in the nature of membranes, which is now extended to include liquid and gaseous materials, biological or synthetic. Membranes are understood to be thin barriers between two phases through which transport can take place under the action of a driving force, typically a pressure difference and generally a chemical or electrical potential difference.

An attempt at functional classification of membranes would have to include diverse categories such as gas separation, pervaporation, reverse osmosis, micro- and ultrafiltration, and biomedical separations. The dominant application of membranes is certainly the separation of mixed phases or fluids, homogeneous or heterogeneous. Separation of a mixture can be achieved if the difference in the transport coefficients of the components of interest is sufficiently large. Membranes can also be used in applications other than separation targeting: They can be employed in catalytic reactors, energy storage and conversion systems, as key components of artificial organs, as supports for electrodes, or even to control the rate of release of both useful and dangerous species.

In order to meet the requirements posed by the aforementioned applications, membranes must combine several structural and functional properties.

Microporous inorganic membranes are potentially useful in gas separation in emerging areas such as catalytic reactors, gasification of coal, molten-carbonate and solid-electrolyte fuel cells, and water decomposition by thermochemical reactions. If the feed or product gases can be separated at elevated temperatures specific to each process, the energy required for purification could be greatly reduced. Advances in the development of inorganic membranes have been quite rapid in recent years. For example, in 1991 the reported CO2/N2 selectivity at ambient temperature was less than 10, but by 1997 it had improved to approximately 100.

The permeation rate and permselectivity of porous inorganic membranes are dependent on the microstructures of membrane/support composites such as pore size and distribution, porosity, tortuosity, and the affinity between permeating species and pore walls. Figure 1 shows the relationship between molecular weight and kinetic diameter (calculated from minimum equilibrium cross-sectional diameter) for a selected series of molecules. Hydrogen and helium are smaller and lighter than the others. Structural isomers such as n-C4H10 and i-C4H10 have the same mass but quite different sizes. Therefore, the control of micropores is of critical importance in these cases. However, the molecular masses and sizes of CO2 and N2 are not greatly different; thus the difference in affinity is important for separation of these molecules.

In order to achieve effective separation of small-molecule gases, the membrane pores should be smaller than 2 nm. In the case of mesopores or macropores, gases permeate with low selectivities through these pores. In this article, preparation processes and permeation properties of porous inorganic membranes are reviewed, and permeation mechanisms are discussed.

Here we will attempt a brief overview of recent synthetic efforts for micropore and lower-end mesopore membranes. We will not address the very important classes of nonporous membranes, such as dense metals and solid electrolytes with applications in H2 and O2 separations, or meso- and macroporous membranes, which find applications in food processing and water treatment. Microporous materials provide high permselectivities for molecules encountered in the chemical-processing industry but suffer from low intrinsic permeabilities. Therefore, in order to bring microporous membrane materials to commercial applications, functional composites with small effective thicknesses (in the micron or submicron range) must be developed. For example, to achieve economical membrane-reactor sizes, fluxes as high as 0.1 mol/(m2 s) are desirable. Approaches to microporous membranes include modification of mesoporous membranes by sol-gel and chemical-vapor-deposition (CVD) techniques, carbonization of polymers to form molecular-sieve carbon, and polycrystalline-film growth of zeolites and other molecular sieves.

Microporous carbon is widely used for liquid or gas purification because of its strong adsorptive properties and high surface area. It is also used for air separation by pressure swing adsorption (PSA), relying on its adsorptive and molecular-sieving properties. From the standpoint of applications, microporous carbons are classified into activated carbons with pore size 0.8–2 nm, and ultramicroporous carbons or carbon molecular sieves with pores 0.3–0.6 nm. Activated carbons are used because of their strong adsorption properties, while carbon molecular sieves are useful on account of their molecular-sieving as well as adsorption properties.

A better understanding of membrane microstructure and its role in the mechanisms of different separation processes is important in the development of new and more complex types of membranes. This understanding is particularly significant in the case of inorganic membranes because the permeation and separation properties are determined by the porous properties of this microstructure, which in turn is controlled by the synthesis route. This review will cover three topics: first, the origin and development of porosity in inorganic materials; second, the general definitions and terminology used to describe porous media; and third, a brief description of several techniques that can provide pore-structure characteristics of membranes.

A distinction will be made between two techniques: static and dynamic methods. Dynamic methods are based on membrane permeation characteristics and are used routinely to assess membrane performance. This review, however, will be restricted to static methods, which can provide details of porous microstructure directly. These include the following: stereology (microscopic techniques such as scanning electron microscopy [SEM]); intrusive methods (e.g., mercury porosimetry, physisorption of gases, calorimetrie methods, and nuclear magnetic resonance [NMR] analysis); and nonintrusive methods (e.g., radiation scattering with neutrons and x-rays, wave propagation, ion-beam analysis, and positron lifetime spectroscopy). Finally, the suitability of these techniques in the characterization of membrane pore structure will be discussed.

Porous materials can be formed in several ways, although the following three are most important in the synthesis of membrane structures.

The capability of membranes to affect differently, both qualitatively and quantitatively, the transport rates of chemical species of dissimilar chemical structure through their interior space renders them attractive for use in many separation problems. Extensive research efforts have thus been undertaken on the preparation and characterization of membrane materials and the study of the transport processes involved in their use in separation applications. The study of the transport of gaseous species through the pore space of porous membranes and the analysis and understanding of the mechanisms that are involved in this process are a very important, if not the most important, element in the development of membranebased separation processes.

The resistance that a gaseous species encounters as it is transported through the pore space of a porous membrane is a function of its molecular properties, of its interaction with the material that makes up the walls of the pores, and of the membrane pore structure. Gaseous transport in pores can take place through various mechanisms, whose contribution to the overall transport rate of a particular species is, in general, determined by the strength of the interactions of the molecules of that species with the pore walls and by the relative magnitudes of three length scales that characterize the molecular size, the distance between pore walls, and the density of the fluid in the pore space.

The development of high-temperature processes and tighter environmental regulations requires utilization of efficient gas-separation processes that will provide high fluxes, high selectivity of separation, and the ability to operate at elevated temperatures. Dense inorganic membranes and membrane reactors are especially well suited for high-temperature reactions and separations, due in part to their thermal stability and high separation selectivity (in theory, infinite). Furthermore, membrane reactors offer an inherent advantage of combining reaction, product concentration, and separation in a single-unit operation for the improvement of process economics and waste minimization.

The classification of membrane reactors can either be by membrane material and geometry or by the configuration of the reactor. Porous and dense membranes in both tubular and disk forms have been used for membrane reactors. The membrane can either be catalytically active (catalytic membrane reactor [CMR]) or simply act as a separation medium. In the latter case, the catalyst is packed in the reactor, whose walls are formed by the membrane (packed-bed membrane reactor [PBMR]). In addition, if the membrane is also catalytically active, the reactor is called a packed-bed catalytic membrane reactor (PBCMR).

The principal materials from which porous inorganic (ceramic) membranes are made are alumina, zirconia, and glass. Alumina and zirconia membranes are usually asymmetric and composite, with a porous support (0.5–2.0 mm thick) for mechanical strength and one or more thin layers for carrying out separations.

On the other hand, glass membranes, such as Vycor and microporous glass, have symmetric pores. Materials commonly used as the porous support are alumina, granular carbon, sintered metal, and silicon carbide.

Pervaporation is a membrane process used to separate liquid mixtures. Separation is achieved by a combination of evaporation and membrane permeation. As a result, the process offers the possibility of removing dissolved volatile organic compounds (VOCs) from water, dehydrating organic solvents, and separating mixtures of components with close boiling points or azeotropes that are difficult to separate by distillation or other means.

A schematic diagram of the pervaporation process is shown in Figure 1. In the example shown, the feed liquid is a solution of toluene in water which contacts one side of a membrane that is selectively permeable to toluene. The permeate, enriched in toluene, is removed as a vapor from the other side of the membrane. The driving force for the process is the difference in the partial vapor pressures of each component in the feed liquid and the permeate gas. This driving force can be increased by raising the temperature of the feed liquid to increase its vapor pressure or by decreasing the permeate gas pressure. The permeate gas pressure can be adjusted by using a vacuum pump, but industrially the most economical method is to cool and condense the vapor. Condensation spontaneously generates a vacuum. The permeate vapor pressure is then determined by the temperature of the permeate condenser and the composition of the permeate liquid generated by cooling and condensing the permeate vapor.

Pervaporation membranes are made by coating a thin layer of selective polymer material onto a microporous support.