Separation Technologies for the Industries of the Future (1998)

The panel identified five issues that were common to the chemical and petroleum refining industries: the availability of lower cost oxygen; methods for removing acid gases and hydrogen separations; improved distillation technologies; the development of isomer separations; and methods for recovering components from dilute gaseous and aqueous streams. The panel also identified two technology areas with the potential to meet these needs and three enabling technologies.

Both industries need lower cost oxygen, which is an important product for one segment and an important feedstock for other segments of the chemical industry and an important feedstock for methane conversion in the petroleum refining industry. Technologies of interest include reactive metal complex sorbents, polymer membranes, and dense, perovskite-type oxide membranes. Because oxygen is already quite cheap, however, economic analyses of research projects in this area are essential. The potential cost reduction should be determined, as well as the extent to which industries would be impacted.

The removal of acid gases and hydrogen separation processes are important to both industries. The chemical industry requires methods for removing acid gases from process streams. The petroleum refining industry requires improved methods for removing acid gases from natural gas and process streams, for recovering hydrogen from gaseous waste streams that are currently burned for fuel, and

ultimately for the recovery of hydrogen from hydrogen sulfide. Potentially applicable technologies include very thin, high-flux membranes, such as polyimides, polyaramides, and polypyrrolones, as well as various polymer-inorganic hybrid materials and inorganic membranes.

Improved distillation technologies would benefit both industries by reducing the amount of energy consumed in separation processes where distillation is still the best method. Technologies of interest include heat cascading, vapor compression, combined chemical synthesis and distillation, and combined membrane processes or other processes, such as adsorption, and distillation.

Improved isomer separations would benefit both industries. The chemical industry needs new separation technologies for the production of single-isomer chiral compounds. The petroleum refining industry needs improved isomer separations, such as the separation of p-xylene from other xylenes. Technologies of interest include high-performance liquid chromatography, crystallization, and zeolite and carbon molecular sieve membranes.

The separation and recovery of valuable components, such as metal ions, and the removal of contaminants, such as VOCs or particulates, from dilute gaseous and aqueous waste streams presents important problems for both industries. Technologies of interest include reactive metal complex sorbents/chemically facilitated transport membranes, reducing agents, electrically aided membrane separation, continuous adsorption processes, air oxidation combined with absorption, highselectivity gas separation membranes, and pervaporation.

The panel identified two technology areas with the potential for addressing some of the needs described above: separation processes associated with chemical reactions and separation methods with multiple driving forces.

Traditional chemical engineering separation methods rely mostly on differences in physical properties (e.g., boiling point, size, solubility) between the components of a mixture. If physical properties are similar or if a high specificity is required, separation methods that rely on chemical differences, rather than physical differences, may be useful.

Reactive metal complex sorbents have been used extensively for the separation and recovery of metal ions. Activated carbons, widely used for purifying water, are

mainly physical adsorbents. Greater selectivity is achievable by using sorbents with strong chemical interactions, such as metal coordination complexes. Metal complexes in solution or immobilized on solid surfaces for the separation of small molecules, such as Cu(I) complexes for the recovery of carbon monoxide from process gases and Ag(I)+ complexes for the separation of olefins from hydrocarbons, are increasingly being used (Doyle, 1981; Ho et al., 1988). In order for specialty membranes using complexation to be economical, however, either the market must be large enough to justify development costs or the base technology must apply to a broad range of problems with only minor modifications.

The high specificity of metal coordination complexes is exploited in chemically facilitated transport membranes, which allow for continuous, noncyclical, separation processes (O'Hara and Bohrer, 1989). In a chemically facilitated transport membrane, a reversible coordination complex or other reversibly binding species essentially acts as a highly specific mobile ''carrier'' for the desired component of the mixture. In ideal situations, highly selective separations are possible. Existing chemically facilitated transport membranes consist largely of supported liquids or emulsions containing carrier species (Way et al., 1982). Recently, solid-state chemically facilitated transport membranes using immobilized carriers have been suggested (Nishide et al., 1994). Although their transport rates are generally lower than for liquid systems, solid-state membranes should be easier to incorporate into commercial modules. Because of the potential variety and architecture of coordinating ligands, the scope is considerable for the design and synthesis of metal complexes for use in highly specific chemistry-based separation processes.

The "separation" process can be a synergistic combination of usually separate processes, such as chemical synthesis and physical separation. Important combinations include catalytic distillation and reactive distillation. In addition, the extent of conversion in equilibrium-controlled reactions can be increased by the reversible absorption of one of the products. This type of "sorption enhanced process" (Carvill et al., 1996) is illustrated in the water-gas shift reaction:

By using a solid zeolite adsorbent for water, at 250°C, the equilibrium shifts to yield almost completely carbon monoxide until the adsorbent is depleted. Continuous production of carbon monoxide is possible by a combination of reaction, sorption, and sorbent regeneration steps.

Chemical synthesis could also be combined with a closely coupled, but separate, membrane separation device. This would be most useful for equilibrium processes and would require a membrane selective for the particular product. An example is the selective production of para-xylene by an equilibrium redistribution of mixed isomeric xylenes coupled with selective transport of the product through a membrane. Synergistic processes that combine chemical synthesis with distillation, sorption, and membranes can, in principle, lead to more energy-efficient and materialsefficient chemical processing, especially for equilibrium-controlled reactions. For each of these hybrid systems, however, significant challenges will have to be overcome to achieve selectivity with the individual catalytic synthesis, distillation, sorption, or membrane separation steps.

The most intimate combination of a separation process with chemical synthesis occurs in a membrane reactor, in which the membrane and catalyst are one and the same. Membrane reactors can potentially increase the efficiency of chemical synthesis because the reaction and separation steps are combined into a single process. These devices can improve efficiency by biasing equilibria to produce the desired product.

The potential benefits of membrane reactors include, for example, the reduction of the formation of coke in dehydrogenation reactions and improved separation of homogeneous catalysts from reaction product mixtures. Membrane reactors that use immobilized enzyme catalysts have been used for synthesis, as in the conversion of sucrose to glucose. Gryaznov (1986) pioneered the concept of membrane reactors with inorganic systems utilizing H₂-permeable palladium spiral membranes for hydrocarbon hydrogenation/dehydrogenation reactions. This method is particularly advantageous for the equilibrium-limited dehydrogenation reaction of alkanes, where the removal of H₂ by permeation can increase product yield. Recent research has been focused on porous inorganic membranes, which offer better transport rates than palladium or other dense systems but have lower selectivities. The inorganic membranes usually consist of porous alumina, glass, or other oxides impregnated with the appropriate catalysts, such as platinum for the dehydrogenation of hydrocarbons to olefin and molybdenum sulfides for the dissociation of H₂S to sulfur and hydrogen.

One of the most enticing developments is the use of dense oxide, high-temperature (800°C to 900°C) oxide ion transport ceramic membranes in a membrane reactor for the selective oxidation of methane to synthesis gas with air (Balachandran et al., 1997). These O₂-permselective membrane systems will most likely be used to conduct other in situ selective oxidation reactions, such as the synthesis of olefin oxides. These applications will, however, require the development of O₂-transport

membranes that can operate at much lower temperatures (< 500°C) at which selective oxidation catalysts can effectively be used.

Oxygen could potentially be produced from air at lower cost by the use of dense, perovskite-type oxide membranes. The membranes could be used in the membrane reactor configuration for these reactions (see Figure 4-1). The state-of-the-art membrane materials are focused on the perovskite structure (ABO₃) with alkaline earth cations as the A site and transition metal cations as the B site (Lin and Zeng, 1996). The major outstanding issues include membrane stability and thickness and the high-temperature requirement (> 800°C). Changes in thermal expansion and defect structures have been observed even at low oxygen pressures (below 10^-6 atm).

The implementation of membrane reactor technology has been limited by the availability of membranes with the required properties. For a membrane to work in the reactor configuration shown in Figure 4-1, it must be stable in the reaction environment, selectively transport desired species at high rates, and be easily incorporated into modules.

The driving force in a separation process can be enhanced by coupling different physical and chemical phenomena. Separation processes involving a phase change or molecular diffusion rely on a thermodynamic driving force defined by the difference between the chemical potential of the transferring species in the source and the chemical potential in the recipient. In most instances, the driving force is approximated by the difference in concentration of the species. Multiple driving forces exist either when a naturally occurring driving force for a specific operation is enhanced by an intervention that changes the system thermodynamics or when two or more separation techniques are combined. An example of the first category is the addition of an electrical potential across a membrane to facilitate the transport of a species through the membrane. An example of the second is using an adsorption step to break an azeotrope formed by one of the products of distillation.

Figure 4-1

Membrane reactor configuration used in dehydrogenation technology.

Affinity-based separation has become more important in recent years because of its application to biological processes and biologically produced species. The technique most often uses a chromatographic operation and makes possible the removal of a specific species from solutions containing a number of other slightly different compounds. This method uses alternations in the thermodynamics of the separation process to remove the affinity materials (i.e., reverse the sorption).

The principles of affinity chromatography are illustrated in Figure 4-2, where a specific substance has been attached to a solid support, and the attached compound (receptor) is a companion to the one to be separated from a mixture of species that are similar in all respects but the one used as the basis of separation, which might be chemical functionality, steric hindrance, or a combination of the two. The companion compounds might be an enzyme and an enzyme inhibitor, an antibody and an antigen, a nucleic acid and a nucleic base, a hormone and a receptor, or a vitamin and an appropriate carrier protein. In the operation, mixture conditions are adjusted, and the solution is fed to the chromatographic column. The desired species attach to the receptor while the undesired species pass through the column. After the undesired species have been purged, conditions in the column are altered, for example by adjusting pH, to release the desired species from the receptor. This highly selective process should be investigated for use with separation techniques other than chromatography.

Figure 4-2

Schematic illustration of affinity chromatography.

Electrochemistry has the potential to combine reaction and separation into a single process step. Traditional chemical engineering unit operations are often described as a reaction, a separation of products from unreacted material, and a recycling of unreacted materials back to the reaction step. With electrochemical methods, the splitting of a salt into an acid and a base, for example, can be done in a single bipolar electrodialysis unit that combines the separation of the anions and cations with the production of the needed protons and hydroxyl ions (Mani, 1991). Other examples include the production of hydrogen iodide without by-products (Weinberg, 1997); electrochemical separation in gas separation technology (Winnick, 1990); and electrodialysis for the separation of ions from liquids (Strathmann, 1979).

Pressure and electric fields could be additive to selectively enhance or retard the transport of a species across a membrane. A combined-effect membrane could be constructed with both sufficient strength to permit the application of a significant pressure field and ionic transport capability. Electrode material could be applied to either side of the membrane. The promise of electrochemical synthesis/separation methods is that they will be able to synthesize pure products under moderate conditions and with high energy efficiency, while simultaneously performing separation processes. Electrochemical synthesis/separation methods will require advances in materials and the basic understanding of physico-chemical phenomena in electrochemical processing.

Reversible complexation of organics with organic agents is a promising new separation technology that is currently being researched. Acetylenic inclusion compounds (organic molecules that can trap a specific component of a mixture in their crystal lattice) can be used for the recovery of NaOH from aqueous solution (Toda, 1987). In addition, some classes of organic compounds (e.g., cyclodextrins and calixaranes that are porous at the molecular level) can function as size-selective hosts in separation processes.

In a number of cases, chemical reactions can be used to enhance the driving force for separation. One example is the synthesis of methyl acetate from acetic acid and methanol. Reactive distillation is used to separate the methyl acetate from the reactants and products resulting from the synthesis. Successful application of these technologies could make separation easier or even unnecessary.

No single membrane material will have all of the desired characteristics for membrane separation technologies. Important properties for emerging membrane materials include impact strength, flexibility, and thermal stability, as well as the properties required for a particular separation (i.e., flux and selectivity). The development of materials for specific applications will only be economical if the application is widely used, so research should be focused on membranes that will be useful for a variety of applications.

Currently available commercial polymer membranes include polysulfone, polyimide, ethylcellulose, and polycarbonate. These materials are easy to form, have high flux rates and efficiencies, and have been used successfully. However, they also have several limitations. First, they degrade in harsh chemical or high-temperature environments. Polymers that are stable in these environments are brittle and difficult to process. Second, applications, such as the production of oxygen-rich air and the removal of acid gas from natural gas, will require membranes with higher selectivity and higher flux. Polyimides, polyaramides, and polypyrrolones are among the polymers that are emerging or already in use, but membranes made of these polymers tend to have high selectivity but a low flux due to their thick active layers. Thinner membranes have been produced to try to increase flux, but these membranes could lose their selectivity due to imperfections. Finally, swelling is a problem for polymer membranes in some applications, such as organic separations. Swelling problems can potentially be overcome by combining hard segments for stability with soft segments for selectivity.

Inorganic materials are also being used to construct membranes. Membranes made out of microporous inorganics and nonporous metals also have significant limitations, however. Emerging applications include the use of thin, nanostructured, dense metal membranes for the separation of gases, such as hydrogen. Other applications include the use of molecular sieve membranes, such as zeolite and carbon molecular sieve materials, for the separation of isomers and dense, perovskite-type oxide membranes to produce high-purity oxygen.

Sorbent materials separate components by three methods: the formation of stronger bonds with certain components than with others; the exclusion of some components and inclusion of others based on their relative geometries (molecular sieving); and separation based on differences in intraparticle diffusion rates. Sorbent materials in the first category represent by far the largest tonnage of

sorbents sold; sorbents in the third category are by far the smallest. Most sorbents are inorganic (adsorbents), although a few are polymeric with selectivity based on the relative solubilities of components in the polymer.

In order for a sorbent to be regenerated for further use, the bonds formed between the sorbent and the sorbate (the material being sorbed) must be reversible. The degree of reversibility is a critical factor for the economic viability of sorption processes. The few adsorbents that bind components irreversibly can only be used to remove relatively small amounts of components (about 100 kg/day) or less because of the high cost of removing and replacing the spent adsorbent.

So-called hydrophobic sorbents, such as activated carbon, silicalite, and various resins, tend to be selective for less polar components and tend to select among hydrophobic species based roughly on boiling points. Hydrophilic adsorbents (e.g., activated alumina, silica gel, and zeolite molecular sieves) are selective for polar compounds and form relatively strong bonds with water as compared to a wide range of organic species.

The number of adsorbent types, not including a variety of irreversible adsorbents, is relatively small (less than 10), but most have many variations. Thus, activated carbons can be produced from various carbon-containing sources and can be heat-treated in various ways to produce differences in inherent bond strengths with sorbates, different pore sizes, and different particle shapes. As a result, hundreds of separate activated-carbon products are being produced commercially.

The newest class of sorbents consists of special shapes that facilitate movement between adsorption and desorption zones. One example consists of small, spherical, uniformly-sized particles of polymer sorbents that are used in processes that combine fluidized and moving beds and that conceptually resemble a conventional absorption/desorption process. A second example consists of large, monolithic shapes of metal ioxides or activated carbon usually in the form of a large wheel. The wheel rotates several times per hour through adsorbing and desorbing zones.

On-line detection of the composition of material streams in separation processes would benefit the chemical and petroleum industries. Carefully controlling reaction conditions is an effective strategy for minimizing undesired side reactions, and hence reducing the need for subsequent separation or purification steps. Much of the technology for the fine control of reactions is commercially available, including excellent means of controlling key parameters, such as flow, temperature, pressure, and mass. Feedback algorithms are often well known, and there is ample computing power and memory to implement these on a real-time basis.

More often than not, the limiting components for controlling reactions are the sensors. The chemical and petroleum refining industries need accurate, selective

chemical sensors that can maintain performance in harsh or extreme manufacturing environments. In distillation processes, sensors that could determine compositions on-line at various points within a distillation column would be an enormous help in controlling and optimizing performance by allowing distillation columns to operate as close as possible to minimum reflux. These rapidly responding sensors would have to be capable of detecting minor differences in the structure of chemical compounds or fractions in a potentially harsh environment.

The growing need for sensors is apparent when product specifications are strict. Uncertainties in the distillation column are usually dealt with by operating the column very conservatively, which often leads to product purities that exceed specifications. The consequences of this practice are shown in Figure 4-3, which plots energy costs as a function of product purity. Note that the slope of the plot increases as purity increases, meaning that the incremental energy cost associated with increasing purity by a fixed amount increases as the base value increases. The figure also shows the effect on energy costs of high and low process variability for the same mean operating value of product purity. The mean operating value has

Figure 4-3

Schematic illustration of the relationship between energy costs, product purity, and process variability. Energy costs increase with increased product purity. Increased process variability results in increased energy costs with no increase in product purity.

been selected so that no portion of the product is below purity specifications. Processes that can be tightly controlled to lower the uncertainty in terms of purity have significantly lower energy costs than processes that cannot be tightly controlled. Columns operated in the mode just described often become bottlenecks in the production process.

Improvements in control technology are likely to occur as the result of improved sensors, control algorithms that improve feed-forward, capabilities, and artificial intelligence and fuzzy logic methodologies to update process models and controller operations. However, industrial experience has shown that frequent maintenance and calibration of sensors is important. Fouling is likely to occur, as well as interference from minor contaminants. Early research and development should include the use of realistic industrial process streams and contaminants. Although the cost of a sensor is clearly important, a robust solution to a challenging detection problem may be worth significant capital expenditures. Finally, the interest in using biological sensors to detect chemical composition has increased significantly.

The ability to separate a mixture as efficiently as possible is closely related to making the best possible use of the driving forces. In most separation processes, the thermodynamic description of equilibrium between two phase is an indispensable tool in understanding driving forces. Research in phase equilibrium is hardly new, and extensive information is available. Nevertheless, some aspects of phase equilibrium have been so little researched that some separation processes cannot be optimally designed. The negative consequences of suboptimal designs include excessive energy use, less-than-maximum separation or recovery, and high investment costs.

Several aspects of phase equilibrium are lacking data and correlations. One of these areas is mixtures in which one or more components of importance are at very low concentrations (e.g., parts per billion). These are the same conditions encountered in the removal of already small concentrations of pollutants from waste streams, as well as in the purification of final products. This near-infinite dilution is difficult to quantify because, not only do we not have analytical-chemistry techniques for accurately determining extremely low concentrations, but also the maximum deviation from ideal-phase behavior almost always occurs at these concentrations. As a result, phase-equilibrium correlations derived from data at less extreme compositions often yield vapor-liquid and liquid-liquid equilibrium estimates that can be wrong by a factor of at least two or more in the near-infinite dilution range. This degree of error can cause substantial departures from the optimal design. Systems that involve, for example, liquid-solid equilibria are also

subject to major problems in the very-low-concentration range. Accurate activity coefficient data for molecules in near-infinite dilution, and use of this data in modeling and design, are critical.

A second area for research is systems in which reaction accompanies the attainment of equilibrium. A typical example is the absorption of H₂S into amine and other basic solutions. Ionization of the H₂S can occur in the liquid phase, and, in some systems, reaction kinetics can be slow enough that a combined reaction rate/phase equilibrium design model must be derived. High-quality models are rarely available in these cases. A third research area might be called extreme condition equilibria. Extreme conditions include temperatures well above 100°C or well below 0°C and pressures in the supercritical-fluid range. Supplying the critical thermodynamic information for separation processes in the areas described above will require combining advanced analytical-chemistry technology with phase-equilibrium equipment and deriving correlations that accurately describe and generalize the experimental data.