Вы находитесь на странице: 1из 17

Journal of Membrane Science 147 (1998) 217233

Applications of catalytic inorganic membrane reactors to renery products


J.N. Armor1
Air Products and Chemicals, Inc., 7201 Hamilton Blvd, Allentown, PA 18195, USA Received 21 January 1998; received in revised form 1 April 1998; accepted 2 April 1998

Abstract Catalytic membrane reactors are reviewed as applied to opportunities and applications within petroleum reneries. Since so many inorganic membranes take advantage of H2 permselectivity and H2 demands are increasing in a renery, there are a number of interesting process applications being considered. H2 production can be enhanced by using Pd based membranes for dehydrogenation, oxydehydrogenation, and decomposition reactions. Permselective H2 membranes could be used for carrying out selective hydrogenations of organic substrates and coupled reactions. These membranes have been also considered for enhancing steam reforming reactions for the production of bulk H2, the water gas shift reaction, and the conversion of natural gas to syngas and liquid fuels. Dense oxide membranes are also being developed for the selective oxidation of CH4 to syngas. For many of these processes, the formation of carbon during steam reforming or dehydrogenation reactions will always be a huge hurdle towards any successful commercial application of Pd membranes to such processes. In any of these applications one has to understand production problems associated with the metal membranes, the renery demands for high purity H2, and the reactor fabrication hurdles; these will be evaluated with recent examples. For all these applications, the critical issues that need to be resolved for the commercial use of catalytic membrane reactors will be discussed. # 1998 Elsevier Science B.V. All rights reserved. Keywords: Membrane catalysis; Membrane reactors; Inorganic membranes; Renery products

1. Introduction There has been an intense, worldwide effort on membrane catalysis since the 1980s and these efforts have been summarized in a number of recent review articles (for example, [18]). Incorporating inorganic membranes into reactors has also been surveyed for a number of reactions [9,10]. A number of groups are

1 Tel.: +1 610 481 5792; fax: +1 610 481 2989; e-mail: armorjn@apci.com

working to bring this technology to a demonstrated commercial application, particularly for catalytic membrane reactors. The success of these groups will depend on a number of key process steps that make up this problem, which include:  the membrane;  a device to contain the membrane;  an attractive chemical process to couple with the use of the membrane. It is important to appreciate that without an acceptable, targetted utility for catalytic membrane reactors, they will not ever move into real applications.

0376-7388/98/$19.00 # 1998 Elsevier Science B.V. All rights reserved. PII: S0376-7388(98)00124-0

218

J.N. Armor / Journal of Membrane Science 147 (1998) 217233

One should keep in mind that membranes are costly materials and their use is favorable when one has a high pressure feed, high concentrations of permeable materials in the feed, minimal amount of contaminants present to impact the membrane performance, need for high purity product, or the permeate (the portion of the feed that passes through the membrane) product is acceptable at lower pressures [11]. Both inorganic and polymer membranes are used for separation operations; however, membranes have not been used with catalytic reactors to enhance the production of some chemical(s). There are a limited number of reactor applications for polymer membranes because of their low upper use temperature; hence, this review will focus on a variety of inorganic membranes which may offer more opportunities over a wider range of operating temperatures. It will become apparent that the use of membranes with catalytic reactors is dependent on understanding the limitations of membranes and their strengths. Therefore, the approach taken by this review will be to briey summarize what we know about likely inorganic membrane compositions, rather than provide an in depth review of these materials. With this as a stepping stone, we will then consider a number of process operations and summarize how such membranes can be used. The size and importance of the worldwide renery industry provides a number of potential applications for membrane catalysis and that is the focal point of this paper. Finally, a summary of critical issues and hurdles around each process will be provided which is intended to help focus the reader at areas of opportunity needed to make this technology succeed commercially. 1.1. What is driving the interest in this topic? While membranes have been used commercially for many years for the separation of gases, their application to catalysis is driven by a number of features. By using a membrane to separate a particular reactant or product from a stream, one hopes to drive a reaction further in shifting the concentration on a reactant or product as it moves through the membrane. A membrane also offers a barrier to prevent two incompatible reactants, such as H2 and O2 from being on the same side of the reactor. In addition, the concept of producing a puried product (by using the membrane to

perform a selective separation) provides the added bonus of enhancing the value of a product. Since separation and purication are key steps in the production of chemicals, there is keen interest in incorporating a membrane into a reactor in order to reduce the number of operating units (and cost) in a production facility. 1.2. Traditional approaches The current approach taken by most investigators in this eld is to use coated mesoporous supports as membranes. Mesoporous substrates allow one to use a rigid support which is commercially available, such as the 40 A alpha alumina tubular membranes which are widely used [12]. A working layer is then deposited on this support with the intent of keeping the layer as thin as possible, but continuous. Where possible, other groups are using permselective, dense oxide material (such as ion transport materials) or metal alloy membranes. Use of a permselective membrane layer (a membrane having a high selectivity for one component of a multicomponent feed) assures a pure permeate product which does not require further purication. 1.3. Membrane axioms One often starts working in this topic and then nds that certain desired factors might have to be compromised to produce a working device. A membrane is going to add cost to any process, so for it to be cost effective, it must give one other advantages that outweigh the added cost. The original driving force behind catalytic membrane reactors is often to reduce unit operations. To do so, one cannot substitute it with more expensive or exotic methods of separation. There are critical operational features of a membrane that one needs to be successful. These include:  A material with a very good separation factor (small values such as 1.1, 2.2 are not going to drive one to spend the extra cost). Separation factors of >5 or approaching infinity are what one will ultimately need.  High flux (rate of flow of permeating species per unit area of membrane surface).

J.N. Armor / Journal of Membrane Science 147 (1998) 217233

219

High quality membrane materials: the membrane material must be stable over many months of operation and the working layer must be flawless. Many membrane reactors are demonstrated by using a sweep gas or a vacuum to assist in the permeation of a specie, but in a commercial operation those will not be acceptable; ultimately, one has to be able to drive the separation with a method which does not dilute the permeating product (thus losing one of the key attributes of a membrane) or demand high energy input to perform additional separation. 2. Types of membrane reactors There are two types of catalytic membrane reactors that one can envision:  A microporous (<20 A) membrane layer wherein the catalyst is contained within the wall of the membrane, or the catalyst is deposited on the outer surface of the membrane.  A permselective membrane which allows only one component through. This provides the advantage of not requiring any further separation of the permeated gas(es). Here the permselective membrane can be the catalyst. Alternatively, the permselective membrane may contain the catalyst within the membrane layer. 2.1. Summary of available inorganic membranes There are at least ve types of inorganic membranes: precipitated oxides, zeolites, carbon, dense oxides, and dense metal. Advances in catalytic membrane reactor technology are limited by the absence of commercially acceptable inorganic membranes capable of molecular size or permselective separations. Since the membrane material is a critical part of the membrane reactor, I will offer a summary of their limitations and their weaknesses, in order to focus on the areas for improvement. Many reviews (for example, [1315]) have been written about these materials. Some of the early work in this topic was done with ne oxide particles which were precipitated onto a mesoporous support. By producing small particles and packing them tightly together on top of a mesoporous

support, one can now produce a pore structure with <20 A micropores. Such densely packed oxide particles are subject to surface chemistry of the oxide, delamination, and occasional aws in the surface, which can only be avoided by making the layer thicker at the expense of permeability (rate of ow of gas through a membrane of 1 cm thickness with a crosssectional area of 1 cm2 and a transmembrane differential pressure of 1 cm Hg). These oxide compositions are grown at intermediate temperatures where the oxides are still in a metastable form and subject to additional phase changes or reactivity of outer surface hydroxyl groups with water or other feed components. Heating such materials above and below their preparation temperature can often cause collapse or pore size changes. These hurdles have limited the approach of precipitation (as a means of controlling pore size) to laboratory operations. In addition, such membranes are not easily scaled to large membrane devices. There has been a recent tidal wave of work with zeolite membranes which are usually prepared by precipitating or growing a zeolite working layer on top of a mesoporous support (for example, see [16 23]). The zeolite layer is still relatively thick [$10 microns] and permselectivity is difcult to achieve. Zeolites do not tend to grow in a sheet like topology, but rather as prisms. The growth of a zeolite on a surface often proceeds with stacks of zeolite crystals in soldier-like array. Thus the zeolite layer is often grown quite thick in order to reduce the chance for interparticle voids between adjoining crystallites. As Jansen and Coker [23] recently pointed out, a major challenge is the crystallization of a zeolitic lm without pinholes while maintaining a very thin microporous phase for optimal ux. Beyond the preparation challenges, there are operational ones. With zeolite membranes the separation of bulky organics is complicated by the fact that these materials also adsorb on the surface of the pore and can actually block smaller molecules from permeating while being able to slip through themselves. Adsorption of components on the surface can be very temperature dependent, thus changing the characteristics of the membrane as one changes the temperature. The pore sizes available with zeolites are really very limited, so the pore size control that one can achieve will also be limited. In most cases permselective separation will not be

220

J.N. Armor / Journal of Membrane Science 147 (1998) 217233

possible, so post separation may still be required. Some very preliminary examples of zeolite membrane catalysis are described in [20], especially with regard to using a zeolite membrane to contain a catalyst or where the zeolite is embedded in a polymer membrane. As Delft workers [20] concluded in 1996, ``each zeolite will require its own and detailed study in combination with the supports selected. Altogether it is expected that the learning curve will be long before arriving at industrially operating zeolitic membranes for separation and catalysis''. Carbon based membranes are beginning to emerge (for example, [2426]). These are usually grown by decomposition of an organic material, which produces a dense layer having a microporous sieving network. Granular carbon materials have been used commercially for years for molecular size separation by adsorption [28]. Using vapor deposition techniques it is possible to grow extended layers of carbon sieving surface. Rao and Sircar [29,30] have demonstrated the use of carbon coated supports for recovery of H2 from targetted renery streams. In these selective surface ow membranes, H2 is rejected from the membrane and it allows one to recover high pressure H2 in an enriched retentate stream (the portion of the feed that is rejected by the membrane and not allowed to permeate). Tubular carbon molecular sieve membranes have recently been reported for the separation of gaseous mixtures [24,27]. It is still difcult to achieve these as continuous, strong tubes free of microcracks. These carbon membranes are in very limited supply and often very small, which limits their evaluation and application. Dense oxide membranes are being widely considered for O2 transport. These non-porous oxide compositions permselectively allow only O2 to permeate as oxygen atoms through the membrane layer via oxygen atom vacancies [31,32]. The driving force for O2 transport is a differential O2 partial pressure across the membrane. For these materials the kinetics of O2 incorporation at the oxide surface will limit transport of O2 through the membrane when the membrane layer becomes small (<$25 m). This O2 transport can also be enhanced by the application of a catalyst to the surface of the membrane [33,34]. Most of the current interest evolves around mixed cation conductors where divalent cations are substituted for trivalent cations. Charge compensation takes place by

the creation of oxygen vacancies and electronic holes. They too are in limited supply. While some have been made into large tubular devices, their long term durability is a concern [35]. Another disadvantage of these materials lies with the need to operate them at very high temperatures (>7008C) in order to get sufcient O2 permeability; however, for reactions such as partial oxidation of methane or power generation cycles, these temperatures could be achieved from reaction or heat recovery. Dense metal membranes have enjoyed a great deal of attention largely because they are commercially available in a variety of compositions and can be made into large devices. Unfortunately, there are only a limited number of permselective type of separations that one can achieve: Pd based alloys for H2 permselectivity or Ag based alloys for O2 permselectivity. Development of these membranes has been limited by fabrication capabilities and availability of a wide composition of alloys. Pd membranes can undergo phase changes which can cause catastrophic failure [36] of the membrane due to expansion of the lattice resulting in microcracks in the bulk metal. These phase changes are very pressure and temperature dependent. In the 1960s commercially manufactured Pd diffusers were used to extract H2 from waste process gas streams. These were 50 ft long, 100 m thick Pd/Ag alloy tubes which operated at 4006008C with a P20 atm. It appears that this business was terminated in less than one year when multiple cases of pinholes repeatedly developed in these membranes. One group in Russia [37] has had unique success in developing a multimetallic Pd alloy which they claim resists cracking. Further, they claim that these have been used in a pilot unit employing a unique reactor design and seals. Four membrane columns, each 10 m high were used for two years for the recovery of H2 from an NH3 purge gas at 200 atm feed to produce pure H2 at 30 atm with 96% H2 recovery. (This work has not been veried by others in the open literature.) In order to minimize operational problems, much of the focus in the current literature has shifted to deposition of Pd alloys onto the surface of mesoporous supports. For complete and awless surface coverage, this still requires a relatively thick Pd layer which limits ux. Use of Pd membranes must be balanced against their demonstrated limitations given in Table 1.

J.N. Armor / Journal of Membrane Science 147 (1998) 217233 Table 1 Current limitations for any commercial use of Pd membranes

221

 Best membranes have limited life (months) which is often due to cracking or pinhole formation. Since pure H2 is desired, this is unacceptable  Alloys of Pd can undergo surface inhomogeneity during long term operation  Sensitivity of Pd to traces of iron, which causes pinholes (this can be minimized by using aluminized steel for piping ahead of the membrane)  Need for ultra thin, continuous layers of Pd in order to maximize H2 flux  Low surface area of metals requires complex membrane reactor designs to maximize surface to volume ratio  Pd membrane alone is not a sufficient catalyst for many reactions; one may still need to have a catalyst present thus giving rise to mass transfer concerns (see Section 2.2 below)  Pd is a precious, commodity metal whose price is very subject to sometimes unpredictable market forces

Fig. 1. Schematic of laboratory membrane reactor system [9]. The catalyst would be contained within the tubular membrane inside the furnace.

2.2. Reactor issues Once one has an acceptable membrane layer to use with a process, then one realizes that there are device related problems. A very simplistic schematic of a catalytic membrane reactor is illustrated in Fig. 1. Perhaps, the most difcult and common problem is that of the seal between the membrane and the rest of the reactor unit (for example, the large quartz extensions of Fig. 1). Any membrane layer, as part of a catalytic reactor, will have to be encapsulated into a device to capture the permeating gas and handle the retentate stream. The seal between the membrane layer and the device often becomes a process limitation driven by materials incompatibility. For example, it is difcult to x Pd alloys to steel or ceramic walls and take the entire device through temperature changes from ambient temperature to <9008C. In any production unit as one starts and restarts the

reactor unit, the seals between the membrane and the rest of the reactor unit can fail. With Pd, Pd coated, or dense oxide membranes, coefcients of thermal expansion are different enough that on repeated temperature cycling the seals can weaken and fail, thus destroying the purpose of a membrane layer. Another limitation of the reactor system is the inherent low surface area of the most popular membrane materials. Metals or dense oxides do not have a great deal of available physical surface area. Reaction productivity can become limited by the amount of surface area available that the permeating species can see. Unlike a catalyst particle, the current popular membrane materials have no real microporosity, and hence offer very limited surface area for gas separation. Fortunately, this physical limitation can be circumvented by building elaborate cross-ow channel like designs [32] which maximize the gas to surface contact areas. However, this requires organizations which have the novel membrane compositions to also having the ability to fabricate these materials into complex shapes. This can be achieved, but not on a wide, commercially available level at this time. A more severe process limitation that has yet to be resolved with these membrane reactors lies with the issue of mass transport. One is trying to link a catalysis reaction with a separation. As we have seen, the ux through these materials can be a limiting feature of their operation. Thick membranes slow the transport of permeating species. One can pack a tubular membrane with a catalyst, heat the catalyst up, and easily get a reaction to occur. However, the products of that reaction must now nd their way to the walls of the membrane and permeate the membrane without rst being swept away in the retenate stream; this is controlled by the process feed rates. Looking at this

222

J.N. Armor / Journal of Membrane Science 147 (1998) 217233

Fig. 2. H2 transport hurdles from catalyst to permeate side. Note, for ease of display, the diagram shows the catalyst particles separated from the membrane surface; this is not necessarily the case.

problem (see Fig. 2) on a molecular level: the reactants must rst get to the surface of the catalyst particle, then some adsorption and dissociation must occur on the surface of the catalyst particle. Reaction on the catalyst can proceed producing a permeating product, such as H2. Now the H2 must desorb from the surface of the catalyst, ow past other catalyst particles (or other adsorbing surfaces) and arrive at the membrane surface. In the case of H2 on Pd metal, the molecular H2 must dissociate on the surface and permeate the Pd layer, then re-associate on the permeate side as H2 where it is then collected as a pure product. Dissociation on the membrane surface and permeation of the H2 must all occur at a feed rate which permits sufcient time for this permeation to occur. Thus, one sees that the operation of a membrane reactor can become mass transfer limited if one does not have a membrane with sufcient ux to satisfy the productivity of the catalyst. Much of the early work in this eld achieved some interesting catalysis which was driven by a membrane, but at the cost of reactor productivity. Recent studies have considered various reactor ow models for optimal dehydrogenation [38]. Most of the work with catalytic membrane reactors has been done with the catalyst contained within a tubular membrane. Alternatively, the membrane separation unit can exist as a separate unit from the catalytic reactor [39]. A few examples of these arrangements will be given below where specic reactions are described. One of the clear hurdles in the development of inorganic membranes for catalytic membrane reactors

is the life of the membrane. This topic has received limited attention. In addition, the corollary to this limitation is the ability to repair the membrane while in operation. In the commercial polymer based membranes, a key to their successful development was a means to repair the membrane [40] when they failed during operation. To the author's knowledge, the current solution for lab based membrane catalytic reactors which fail (such as due to pinhole formation) is to replace the membrane; on a commercial scale this will likely be unacceptable. Thus, one sees that current development of the eld of catalytic membrane reactors is limited by not having sufcient membrane materials which is further aggravated by not being able to fabricate those materials within a device having productivities acceptable to the chemical industry. The balance of this review will summarize what progress has been reported in the literature as applied to renery operations, and then reach some conclusions about where there are opportunities for this technology. 3. Refinery opportunities If one looks at the complex operation of a typical world scale petroleum renery [41,42], there are a multitude of reactions occurring. Many of these are already efciently operated by way of heat and mass balance with regard to optimal productivities. However, there is one apparent opportunity that would

J.N. Armor / Journal of Membrane Science 147 (1998) 217233

223

seem to match nicely with the current features of membranes. Reneries are now net consumers of H2, whereas a decade ago, they were net producers of H2 [43]. All over the world, large H2 plants are being built adjoining a renery in order to meet the renery's immense thirst for H2. H2 is the method of choice for desulfurizing fuels. As those environmental demands get tougher, more and more H2 is needed. Years ago the H2 needs were easily met by the coproduction of H2 which occurred during the formation of aromatics by dehydrogenation reactions. Now emission controls on the level of benzene in gasoline has reduced the demand for benzene as an octane enhancer in gasoline and produced less H2 within the renery. Before expanding upon applications for H2, it should be pointed out that in anticipating any opportunities for membrane catalysis within reneries, one must appreciate the issue of scale. Renery operations are massive in size and scope; many unit operations involving multiple processing units demand scale and operational integration into existing operations. The scale of many of these potential processes may demand that some of these membrane operations considered below will have to involve sizable reactor volumes and surfaces, thus necessitating potentially sizable capital costs. Early applications could involve smaller pilot units, but eventually some sizable and engineering demanding scaleup may be required. Some related issues of energy management, economy of scale and capital costs are described in a recent article by Lange et al. [44]. Membrane science has often focused around the selective separation of H2 either by a metal membrane or a microporous layer which would easily permeate the smaller H2 molecule. A good deal of the reactions in membrane catalysis deal with H2 producing reactions, such as dehydrogenation of alkanes and oxydehydrogenation (in the latter, the H2 is converted to water). Within a renery there are a number of other processes (see Table 2) which use or produce H2 which one might envision satisfying with a membrane reactor. These include selective hydrogenation, steam reforming of hydrocarbons, water gas shift, and the conversion of remote natural gas to syngas and liquid fuels. These will now be considered in greater detail below. The author views it unlikely that in the near term (<10 years) that membrane reactors in reneries

Table 2 Opportunities within refineries  Dehydrogenation reactions - Alkanes to simple olefins - Reforming of alkanes by dehydrocyclization  Oxydehydrogenation  Catalytic decomposition of H2S; HI, H2O  Hydrogenation reactions  Steam reforming with membrane reactors  Water gas shift reaction  Conversion of remote natural gas to syngas and liquid fuels

would assist with methane coupling to C2s or for selective oxidations over Ag membranes. There is very little data in the open literature on the use of Ag membranes or their fabrication. What data are available, often deals with very low conversions. 3.1. Dehydrogenation Classical works of Itoh [45] and Gryaznov [46] in the 1980s led the way for others to build small membrane reactors for the dehydrogenation of alkanes. Itoh's early work with a 0.05 in thick Pd/ Ag membrane tube containing a 0.5% Pt/Al2O3 catalyst was able to achieve 99% conversion for the dehydrogenation of cyclohexane to benzene. His reactor design is illustrated in Fig. 3, and used an argon sweep to carry away the H2 permeating the Pd membrane. The reaction over the catalyst was run at 2008C and 1 atm pressure using an argon stream saturated with cyclohexane vapor. If one reads the article carefully, the actual feed rate over the catalyst had to be maintained very low in order to achieve substantial conversions, thus residence times ranged from 2.5 to 14.5 min. Space time yields of benzene were 0.0126 mol of benzene/kg-hr catalyst, which corresponds to about 1 g benzene/kg catalyst/hr (about 1000 times lower than commercial catalyst units!). Kikuchi and coworkers [4750] have extended this work into a number of other reactions. More recently, the membranes have been made of a Pd alloy coated onto a mesoporous membrane support. In one study, a 20 micron Pd lm was deposited onto a mesoporous Al2O3 tubular membrane [50]. Isobutane was passed over a Pt/Al2O3 catalyst contained within the membrane. The yield of isobutylene rose from the equili-

224

J.N. Armor / Journal of Membrane Science 147 (1998) 217233

Fig. 3. Dehydrogenation of cyclohexane [45].

brium value of 6% (demonstrated in the absence of a membrane) to 23% at 4008C with the membrane present. They did report some deactivation of the membrane over a period of 350 h which was due to carbon buildup on the catalyst. The rate of H2 production was limited by catalyst productivity. Brinker and coworkers [51] studied the dehydrogenation of propane using a Pd lm coated on a mesoporous Al2O3 support in a device containing a commercial Amoco dehydrogenation catalyst. Propylene yields rose from the equilibrium value of 30% (in the absence of a membrane) to 40% at 5508C. Once again, they too reported catalyst deactivation due to carbon deposition on the catalyst. Another type of dehydrogenation reaction is reforming via dehydrocyclization of heptane. Ali and Baiker [52] recently described a system of two plug-ow reactors with an interstage Pd/Ag membrane for ex situ H2 separation from the product stream of the rst reactor using a commercial Pt/ Al2O3 catalyst for toluene production. At 4008C at 17 atm they obtained 65% higher toluene yield than a system without the membrane unit. The authors pointed out that a more selective catalyst might perform even better. Once again deactivation of the membrane was observed due to carbon deposition. The latter is very dependent upon the H2/hydrocarbon ratio in the feed to the second reactor. Another dehydrogenation reaction of value to the petrochemical industry is ethylbenzene dehydrogenation to styrene. This has also attracted a lot of interest in the membrane community. Liu and coworkers [53] studied a two stage packed bed reactor followed by a membrane reactor vs a reactor without a membrane unit. The tubular 40 A alumina mesoporous membrane was packed with a commercial K promoted

iron oxide catalyst used for styrene production. They did observe a 4% yield enhancement to styrene with the membrane in the system. Unfortunately, carbon deposition rapidly reduced H2 permeability even with a cofeed of steam added to the reaction feed. These studies on dehydrogenation bring out some important factors which were not highlighted in earlier remarks about membrane and device limitations. With dehydrogenation one has two serious process issues that emerge: carbon deposition and catalyst activity. Raich and Foley [54] studied dehydrogenation of alkanes in a membrane reactor and concluded that these devices were very limited by the availability of dehydrogenation catalysts with high reactivity. Most commercially available dehydrogenation catalysts do not have high turnover numbers; further, they are limited by excessive carbon formation which accumulates on the catalyst. One must remember that these dehydrogenation reactions are endothermic reactions. Their high demand for heat means that the products of the reaction can get trapped within the pores of the catalyst. The reaction rates are slow which means that the olenic products will tend to polymerize before permeating away from the catalyst particles. Further the operation of a membrane complicates this process even further. In commercial dehydrogenation operations, the generation of H2 actually serves to reduce carbon formation tendency. In a membrane reactor, one is trying to remove the H2 as soon as it forms which only aggravates the formation of carbon. Carbon which forms during dehydrogenation can hurt in at least three ways: it can block the surface of the catalyst; it can foul the reactor and plug the unit; or it can coat the membrane layer and thus block further H2 permeation through the membrane layer. None of these are acceptable without some type of repetitive

J.N. Armor / Journal of Membrane Science 147 (1998) 217233

225

Fig. 4. Simulation study on propane dehydrogenation [56].

regeneration. In addition, Ziemecki [55] has reported that CO/H2 reactions over a Pd membrane resulted in the formation of interstitial carbides of Pd in the membrane itself. Thus one has to also worry about carbon altering the composition and performance of the Pd membrane. All these performance limitations seem to suggest that membrane reactors employing dehydrogenation reactions may be impractical. A recent economic study by a group in Norway [56] examined the use of ceramic membranes for propane dehydrogenation. In the absence of sufcient process data, they assumed that one was working with an ideal membrane employing a conventional commercial propane dehydrogenation catalyst (which operates with continual regeneration). A simulation study was carried out on the block scheme illustrated in Fig. 4 which employed catalyst regeneration. They assumed H2 permeabilities of 1.3105 mol/m2 s Pa at 6008C with a H2/C3H8 separation factor of 3.99 as measured by others [57] with a ceramic membrane reactor. Using process data from an Oleex dehydrogenation process, they established boundary conditions. They concluded that a membrane operating with Knudsen (gases which can be separated due to differences in molecular masses arising from their frequent collision with the walls) [13] diffusion selectivity is not going to be commercially attractive because of the large amount of propane lost in the permeate. They concluded that the only possibility to use inorganic membranes in an adiabatic reactor concept for propane dehydrogenation was to use membranes with much higher ideal separation factor to avoid reverse ow of steam (the sweep gas) across the membrane into the retenate stream. An isothermal membrane reactor was evaluated and compared to the commer-

cial Oleex process. With a price differential of $300/ kton for propylene vs propane, they concluded that a membrane process cannot be economically feasible using even the most optimistic devices. In the long run, one will also have to face the high probability of coke formation in the membrane reactor which will force the incorporation of a continual regeneration of the membrane with steam. Thus, the author suspects that in the near term (<10 years) the use of inorganic membranes for alkane dehydrogenation is an incompatible application of membrane reactors. Even if one had a working membrane, one would need a much more active catalyst than exists today, and then one cannot accept continual formation of carbon which is an unacceptable result of dehydrogenation reactions. There are at least three huge hurdles (membrane, more active catalyst, and carbon deposition) which limit the application of membranes to assist dehydrogenation reactions. As Raich and Foley [54] concluded, one needs to invest more effort in developing much improved dehydrogenation catalysts. After that, one must tackle thermodynamics and the tendency for carbon formation in these reactions. The successful application of membranes to assist catalytic dehydrogenation cannot occur without these catalyst and process improvements. 3.2. Steam reforming of methane There have been some recent studies on the use of palladium based membranes for application to steam methane reforming (SMR). SMR technology is the major route to industry's production of merchant H2 on a worldwide scale. This is a very endothermic

226

J.N. Armor / Journal of Membrane Science 147 (1998) 217233

reaction (reaction (1)) that operates at $8008C and at $20 atm pressure in order to achieve near CH4 H2 O 3 CO 3H2 (1) equilibrium conversions and to meet the customers need for high pressure H2. By using a Pd alloy membrane which is packed with traditional SMR catalyst, the intent is to shift the reaction to produce more H2 at lower operating temperatures. The Pd based membrane also would produce pure H2 thus simplifying the current operation which includes extensive H2 purication steps [58]. In 1991 Kikuchi and coworkers [59] showed that SMR also can be enhanced using an alumina supported Ni catalyst contained within an 80 m Pd on Pd/23% Ag alloy membrane coated onto a porous glass tube. At a steam/CH43, they achieved enhanced conversion of CH4 approaching 80% at 1 atm and 5008C vs an equilibrium value of $42%. They used 13 g of catalyst within 25 cm2 of membrane at a feed rate of 25 cm3 per minute with a sweep rate of 400 cm3 per min. Thus, productivity rates were low and diluted H2 is produced. Increasing the sweep rate enhanced the H2 productivity but the rate was limited by the permeability of the membrane. Interestingly, increasing CH4 pressure to 8 atm, increased CH4 conversion. Jorgensen et al. from Haldor Topse also showed [60] that this concept could operate at a lab scale using a commercial Ni/MgO SMR catalyst contained within a 100 m Pd/Ag tubular alloy membrane. At 6 atm and 5008C, they obtained CH4 conversions of 51% compared to equilibrium values of 21%. They found that they had to operate at higher steam/methane ratios (which is undesirable from a process point of view) in order to avoid carbon formation. Members of the same group at Haldor Topse carried out [61] a process economic analysis of an SMR membrane reactor assuming ideal conditions with a 2 m Pd alloy coated ceramic tube. In their model, they incorporated tubular Pd membranes directly into a traditional SMR catalyst bed. They found that they had to operate at >6508C to maintain sufcient CH4 conversions and that membrane assisted SMR was only attractive if the cost of electricity was very low and the membranes were 100% selective to H2. This is because, they found it necessary to run at a lower permeate product pressure, which then required costly compression to typical

H2 pressure levels in commercial SMR units. Their conclusion, based on current information, was that a membrane integrated process is currently not competitive with traditional SMR. Another process economic study on a membrane assisted SMR reactor concept was reported by Sogge and Strom [62]. They concluded the use of size selective mesoporous silica membranes was not cost attractive. However, making some ideal assumptions, they concluded that Pd/Al2O3 membranes offer $18% lower investment costs vs conventional SMR. Thus, they believe that there is incentive for further development due to the economic potential of using a membrane reactor approach. They also extended their analysis to methanol synthesis from H2/CO, but here they concluded that a membrane reactor approach was not economically attractive due to the high compression costs associated with methanol synthesis. In summary, with regard to SMR, there does appear to be greater opportunity for application of membrane reactors, but signicant challenges do remain. One must develop membranes with high H2 ux in order to be competitive with current SMR technology, the membranes must be highly selective (preferentially permselective to H2), and the membrane device must be fabricated with durable, high temperature seals to the rest of the process unit. Attention must be paid to the possibility of carbon formation on the membrane, since commercial SMR reactors are noted for their long term reliability with 5 year catalyst life [63,64]. The high reaction temperature for SMR places a big hurdle on near term implementation of catalytic membrane reactors to SMR technology. 3.3. Water gas shift The water gas shift (WGS) reaction is another valuable process operation within a renery. It is used to convert CO to H2 or to remove CO2. The reaction, described by (2), is an exothermic reaction and typically gives high conversion; hence, there is less CO H2 O 3 CO2 H2 (2) incentive in applying membrane reactor technology to this reaction. There have been a few membrane reactor studies to enhance water gas shift. Drioli and coworkers [65] recently reported some studies using some very thin 0.2 m Pd coated, 25 cm long tubular

J.N. Armor / Journal of Membrane Science 147 (1998) 217233

227

Al2O3 membranes. The membrane was lled with a commercial low temperature WGS catalyst and the reactor operated at 3228C. Because the Pd coating was so thin, some N2 permeated the membrane with H2. With a steam/CO0.96, conversions were always below the equilibrium level of 78% H2 expected. Kikuchi and coworkers [66] used a 20 m layer of Pd coated onto a porous glass tube to enhance CO conversion and H2 production using a commercial FeCr high temperature water gas shift catalyst at 4008C and 1 atm with steam/CO2/1. Conversion of CO was $88% with a feed rate of 100 cm3/min and an argon sweep of 400 cm3/min vs an equilibrium value of $78%. They were able to demonstrate that the use of a membrane permitted one to reduce the amount of steam needed to achieve reasonable levels of CO conversion. Ross and Xue have provided [37] a preliminary report of the use of a porous membrane unit which is separate from the catalyst bed in modeling the water gas shift reaction. They envisioned using recirculation of reactants and products in order to continually drive reaction (2) to the right. Separating the membrane from the catalyst unit removes some of the problems associated with membrane stability and the seals to the device, especially since the membrane no longer has to operate at the same temperature as the catalyst. Of course this adds more complexity and potential cost to the overall process operation. In a later report, Ross and coworkers [67] evaluated a WGS membrane reactor for CO2 removal in IGCC systems. Using a team approach, they incorporated system integration studies, catalyst research, modeling, and bench scale membrane reactor studies. With an integrated membrane reactor for CO2 removal, the net efciency of the process is 42.8% with CO2 recovery vs 40.5% for IGCC with conventional CO2 removal. While the development of the process is considered to be technologically feasible, it became clear that the technology of high selectively, inorganic membrane manufacturing and high temperature ceramic materials engineering is not sufciently mature, thus requiring further development. 3.4. Methane to syngas With the availability of natural gas, there has been intense interest lately to convert CH4 by selective

oxidation to mixtures of CO/H2 (reaction (3)). This reaction is difcult to control CH4 1O2 3 CO 2H2 2 (3) because of over oxidation of the CH4 to CO2 at the high temperatures needed to initiate the reaction. Thus, a number of groups are studying the use of O2 selective, dense ceramic membranes for assisting this reaction. The membrane allows one to separate the methane from the O2, which provides an extra margin of safety in avoiding explosive mixtures of these two reactants. Since the membrane material is permselective to O2, one can use air as a feed. The diffusion of O2 through the membrane wall allows one to add small amounts of O2 to the CH4 on the permeate side. A catalyst layer is often incorporated where the permeating O2 and CH4 react. The group at Amoco and Argonne have worked [35,68,69] with non-perovskite type oxides for permselective O2 transport. Hollow tube membranes of SrFeCo oxides have been fabricated up to 30 cm in length with 6.5 mm diameter. A Rh based reforming catalyst was contained inside the tubular membrane, and a mixture of 80% CH4 and 20% argon was passed through the inside of the tube while air was passed over the outside of the tube. Optimal operating temperature was 8508C, during which >98% CH4 conversion with 90% CO selectivity was achieved for $70 h. They found they had to employ a gold mesh to physically separate the Rh reforming catalyst from direct contact with the membrane wall in order to avoid loss of lattice oxygen. Unfortunately, long term life tests indicated that there was a 43% decrease in O2 ux over a period of 1000 h. This steady deactivation of the membrane material is unacceptable and indicates that more work is needed to dene an acceptable membrane material. In a similar effort, another group at BP Chemicals have fabricated [70] a tubular membrane reactor from the mixed conductor, La0.2Sr0.8Fe0.8Cr0.2Ox and have operated this reactor for the partial oxidation of methane for >1000 h at 11008C. Eltron Research has studied substituted brownmillerite compounds as alternative materials for mixed conducting ceramic membranes. Selected substitution of metal ions into the brownmillerite lattice having a large population of oxide ion vacancies has led to new materials with high ionic conductivities [71]. They have recently demonstrated the combined partial oxi-

228

J.N. Armor / Journal of Membrane Science 147 (1998) 217233

Fig. 5. Ion transport membrane mediated partial oxidation of methane [71].

dation of methane with steam and CO2 reforming reactions using a membrane reactor. One side of the reactor in Fig. 5 is exposed to an air ow, while the permeate side is exposed to 80% CH4 in helium. Their 8.6 mm tubular membrane was coated with a Niperovskite catalyst for the reforming reaction on the outside of the tube. Initial operation for >3500 h has been demonstrated with a H2/CO ratio of 1.82 at 9008C. A large multiyear effort in this area has recently been funded by United States Department of Energy with an award of $84 million over a period of eight years [32,72,73]. This total award supports a cooperative agreement to be administered by Air Products & Chemicals of a consortium including several other major laboratories and companies, such as Chevron and ARCO. The consortium is assembled from companies having special skills in the many different aspects of technology needed to succeed in membrane reactors. The program comprises three phases, with the rst phase funded at $14 million. The nal phase is the construction of a pre-commercial plant to handle 15 million cu ft/day of natural gas. The goal of the plant is to produce rened products equivalent to those made from crude oil at $20/barrel. The staged program intends to develop a viable ceramic membrane which permeates O2 selectively and to react the permeated O2 with natural gas to produce syngas in one step, thus eliminating the need for cryogenic air separation to

produce the required volume of O2. Additional unit operations will convert syngas to higher value petroleum-like products. Successful development of this technology would allow access to remote sources of natural gas and production of liquid hydrocarbons would provide higher value added products than syngas alone. Later in 1997 a global consortium was announced between BP, Praxair, Amoco, Statoil, and Sasol which would also focus on the use of ceramic membranes for the conversion of natural gas to syngas and further to liquid fuels [31]. Recently, Galuszka et al. [74] studied CH4 to syngas and included some preliminary process engineering calculations. They used a 10 m Pd coating on an Al2O3 tubular membrane lled with 5% Pd/Al2O3 catalyst. A feed of 3:1:4 CH4/O2/N2 was passed through the catalyst at 5008C with an argon sweep on the permeate side. The argon assisted in the transport of H2 away from the catalyst with a resulting increase in H2 yield from 18% to 36% (vs the use of catalyst without a membrane). Methane conversion jumped from 27% to 40%. However, about 30% of the H2 produced remained in the retenate stream, and after 20 h they saw extensive amounts of lamentous carbon on the membrane. Obviously, the extensive carbon formation presents a serious process hurdle. They also calculated what amount of membrane surface they would need to convert the H2 permeate (at 1 cm3/cm2] to supply a 1000 ton/day methanol synthesis plant. The Pd membrane surface area would be 97 000 m2 which would require about 1200 kg of Pd [42 000 oz] for one plant. The Pd quantities calculated above bring out another often ignored issue. If these estimates are within reason, one has to wonder whether there is sufcient Pd available at the right price to build multiple world class plants using Pd based membrane reactors. Palladium costs are subject to supply and demand of this precious metal. Availability of Pd is dependent on the different ore compositions around the world. In 1989, there were 3.2 million troy ounces of Pd and Pt produced from mining operations, and about 70 000 ou was recycled from automobile exhaust catalyst recovery [75]. The bulk of the yearly demand for the world supply of Pd goes to electrical components ($1.3 million ounces) with another sizable portion (1.1 million ounces) going to dental supplies. The demand in the auto catalyst and jewelry markets was about 400 000 ou, leaving about

J.N. Armor / Journal of Membrane Science 147 (1998) 217233

229

250 000 ou for existing miscellaneous uses. These all suggest that if a sudden, new demand for sizable levels of Pd were created, the market would have to respond by raising prices, given that large surplus supplies do not exist. Such cost and supply issues could make such membrane reactors less attractive. 3.5. Applications for zeolite membranes There is a great deal of enthusiasm in the literature for the potential of zeolite membranes based upon the established utility of zeolites in extrudate form for the separation of hydrocarbons. If one could fabricate them into membranes, it might be possible to carry out catalytic and separation schemes in one unit, such as the isomerization of n-butane to isobutane; the recovery of H2 from off gases; membrane distillation; and monomer recovery in vent gases [76]. As pointed out in the earlier section on inorganic membranes one has to temper these applications with the fact that performance of such microporous sieves with large organic molecules will be very temperature dependent, since at lower temperatures adsorption can block the zeolite pores. Morooka et al. [77] have seen that for Y zeolite membranes deposited on Al2O3, CO2 in a feed of CO2 and N2, can block N2 permeation. Many early measurements of permeabilities through these membranes have been reported on single component feeds. Nobel and coworkers [16] have shown that this

is not satisfactory; instead one has to measure selectivities of the individual components in the actual mixture one intends to separate. The recent review by the Deft University group [76] provided an interesting assessment of the current technology in this eld relative to the current level of materials available. They concluded that unlike distillation, membrane separations do not have the economies of scale. Therefore, opportunities using zeolite membranes in catalytic processes will be limited to small or medium scale operations. Costs to compete against existing alternative technologies dictate that any early commercial examples will come when there is a need for a new unit or a replacement. 3.6. Selective surface flow carbon membranes By decomposing carbon precursors onto mesoporous supports, workers at Air Products [29] have pioneered the use of sub 10 A porous membranes for size selective separations (see Fig. 6). They have been able to prepare 2.5 m carbon layers with 68 A pores bundled together into large pre-commercial process units. To date these have been demonstrated for H2 enrichment in renery off gas applications. For example, in a feed of 41% H2, 20.2% CH4, 9.5% C2H6, 9.4% C3H8, and 19% C4H10 at 228C and 4.4 atm, the higher hydrocarbons (such as C4H10, etc.) are strongly adsorbed at the carbon pore mouth and block the

Fig. 6. Mechanism of separation with SSFTM membranes [30].

230

J.N. Armor / Journal of Membrane Science 147 (1998) 217233

Table 3 Permeabilities of components of a mixed gas feed [30] Gas H2 CH4 C2H6 C3H8 C4H10 P (barrer) 1.2 1.3 7.7 25.4 112.3 S over H2 from mixture 1.1 6.5 21.3 94.4 Pure gas PHC/PH2 5.1 6.6 2.3 1.2

permeation of smaller molecules like H2 (see Table 3). The H2 is thus recovered at high pressure from the retenate side producing an enriched H2 stream of 56% H2, which is compressed and sent to a pressure swing adsorption system to produce 99.99% H2 while the purge gas from the adsorption unit can be used as sweep gas on the membrane unit. Such a system gives 100% C4 rejection, 91.1% C3H8, 67.5% C2H6, and 36% CH4 (see Fig. 7) rejection. 3.7. Miscellaneous applications in refineries Kikuchi and Chen [78] also studied another H2 producing reaction, CO2 reforming of methane (reaction (4)). Like SMR, this is another reaction which operates at high temperature CO2 CH4 3 2CO 2H2 (4) and pressure; however, it is complicated by an even greater tendency to form carbon from CO decomposi-

tion. With a traditional SMR catalyst of Ni/Al2O3, they observed a signicant enhancement in CH4 conversion from 26% to 99% at 5008C but with excessive coke formation. Using a Pt/Al2O3 catalyst reduced the level of coke make, but it was still unacceptable. The bulk of the examples of membrane catalysis in the literature have dealt with dehydrogenation reactions. However, these same membranes could be used for selective hydrogenation reactions as well. The Pd membranes used for permselective hydrogenation are not very good for different types of hydrogenations, just somespecicsubstrates[79].Inmost casesa catalyst must be incorporated into the membrane reactor to get sufcient hydrogenation. Since the catalyst itself is a verygoodagentforcarryingoutthehydrogenation,there are limited advantages to incorporating the catalyst within a Pd membrane. One minor advantage is the modest productivity that one seems to get for using permeated hydrogen as a reducing agent [80]. A more important, but limiting application, is to use a Pd membrane to extract H2 from an impure process stream (especially if it contains impurities that might poison a catalyst) as a way to carry out selective hydrogenations of renery products. 4. Summary We have seen that the development of catalytic membrane reactors is actually a multistep task. Currently, adequate membrane materials do not yet exist

Fig. 7. SSFTM membrane for recovery of H2 from refinery off gas [30].

J.N. Armor / Journal of Membrane Science 147 (1998) 217233

231

with a large surface area on a commercial scale to meet the rigors of process operations. Given an acceptable membrane material, then one has to pay attention to sealing the membrane within some type of reactor, and then devising a process where one can balance the material demands of the membrane and the reactor with some chemistry of value. 4.1. Membrane challenges which remain There are key features that are needed for the preparation of membranes for membrane reactors [2]. Membranes which operate with low selectivities are not going to be attractive, because they will demand further separation to isolate a pure gaseous permeate product of value. These include:  need for crack free, uniform, sub 8 A microporous inorganic membranes which are stable to extended use above 2008C and also are chemically inert under operating conditions;  fabrication of very thin, flawless membranes over large, complex supports;  membrane materials which are not susceptible to poisoning or fouling. Membrane material needs can also benet from dense metal oxides which can rapidly transport O2 at <5008C (in order to save energy and to use them for synthesis of many specialty chemical products). 4.2. Membrane reactor challenges Membranes operate best when there are high feed pressures (to provide the driving force for separation); thus, the reactor needs to be able to operate at high feed pressures. Key reactor issues that remain include:  developing membranemesoporous support combinations that are compatible with regard to no delamination of the membrane layer upon repetitive cycling of temperature;  durable, high temperature seals to attach the membrane to the reactor;  proper balance of the reactor and catalyst with regard to heat and mass transport issues. 4.3. Reaction challenges It is clear that the membrane and the device must also be compatible with the chemistry. Membranes are

attractive where high purities are required; hence, one has to look for product areas where a high purity permeate stream has value. Contaminants can destroy the performance of many membrane materials; therefore, one is looking for process feed streams which have no objectionable components for the membrane separation being considered. Products that are desired at high pressure have less attractiveness for a membrane operation, since they would demand a greater pressure driving force to maintain membrane ux.  There are very few chemical synthesis reactions which one can run at >8008C, so that limits the utility of current high temperature O2 transport membranes.  Reactions which readily generate copious amounts of byproduct which can foul a membrane surface are not acceptable for long term reactor operation.  More economic studies are needed which look at the entire process, evaluate feedstock costs and conditions, and identify key hurdles as they impact the separation features of membrane reactors.  The cost of a membrane will probably have a significant impact on the final cost of the products; hence, one needs to pursue reactions of real value which offer distinctive product opportunities to offset the cost and hesitation in using a membrane reactor system on a commercial scale. 4.4. Production of H2 in refineries There are increasing demands for H2 within reneries, and it would seem that one might be able to use the H2 permselectivity of Pd for this purpose. However, one must remember that the desirable product is high pressure H2, and compression costs are often not acceptable. In addition, the attractive feature about permselective membranes is that one produces a pure gas product. However, if one has to use a sweep gas to enhance the transport of H2, this means that the H2 now has to be puried, meaning additional and often unacceptable processing costs to upgrade the H2. Finally, very high mass transport is needed in order to move the product through the membrane before it escapes with the retenate stream. Since dehydrogenation reactions always produce some carbon byproduct, it would appear that they are a very long way, if ever, from near term commercialization. The high temperatures of steam reforming

232

J.N. Armor / Journal of Membrane Science 147 (1998) 217233 [10] G. Sarracco, G. Versteeg, W. van Swaaij, J. Membr. Sci. 95 (1994) 105. [11] R. McBride, D. McKinley, Chem. Eng. Prog. 61 (1965) 81. [12] R. Soria, Catal. Today 25 (1995) 285. [13] R.R. Bhave, Inorganic Membranes Synthesis, Characteristics and Applications, Van Nostrand Reinhold, New York, 1991. [14] L. Chu, M. Anderson, J. Membr. Sci. 110 (1996) 141. [15] R. deLange, J. Hekkink, K. Keizer, A. Burggraaf, J. Membr. Sci. 99 (1995) 57. [16] J. Coronas, J. Falconer, R. Noble, AIChE J. 43 (1997) 1797. [17] Y. Yan, M. Davis, G. Gavalas, J. Membr. Sci. 126 (1997) 53. [18] Y. Yan, M. Davis, G. Gavalas, J. Membr. Sci. 126 (1997) 95. [19] J. Fehlner, Z. Zhang, US Patent 5 618 435 (1997). [20] M. den Exter, J. Jansen, J. van de Graaf, F. Kapteijn, J. Moulijn, H. van Bekkum, in: H. Chon, S. Woo, S.-E. Park (Eds.), Recent Advances and New Horizons in Zeolite Science and Technology, Stud. Surf. Sci. 102 (1996) 413. [21] S. Barri, G. Bratton, T. Naylor, US Patent 5 567 664 (1996). [22] T. Bein, Chem. Mater. 8 (1996) 1636. [23] K. Jansen, E. Coker, Current Opinions in Solid State & Materials Sci. 1 (1996) 65. [24] H. Kita, H. Maeda, K. Tanaka, K. Okamoto, Chem Lett. (1997) 179. [25] F. Katsaros, T. Steriotis, A. Stubos, A. Mitropoulos, N. Kanellopoulos, S. Tennison, Microporous Matls 8 (1997) 171. [26] V. Linkov, R. Sanderson, E. Jacobs, J. Membr. Sci. 95 (1994) 93. [27] A. Damle, S. Gangwal, V. Venkataraman, Gas Sep. Purif. 8 (1994) 137. [28] J.N. Armor, Carbon molecular sieves for air separation, in: E.F. Vansant (Ed.), Separation Technology, Elsevier, Amsterdam, Netherlands, 1994, pp. 163200. [29] M. Rao, S. Sircar, J. Abrardo, W. Baade, US Patent 5 332 424 (1994). [30] M. Rao, S. Sircar, J. Membr. Sci. 85 (1993) 253. [31] A. Sammells, M. Schwartz, J. White, World Organization Patent 97-41060-A1 (1997). [32] J. Kilner, S. Bensen, J. Lane, D. Waller, Chem. Ind. 22 (1997) 907. [33] M. Carolan, P. Dyer, US Patent 5 534 471 (1996). [34] M. Carolan, P. Dyer, US Patent 5 569 633 (1996). [35] U. Balachandran, J. Dusek, P. Maiya, B. Ma, R. Mieville, M. Kleefisch, C. Udovich, Catal. Today 36 (1997) 265. [36] A. Darling, Plat. Met. Rev. 2 (1958) 16. [37] V. Mordkovich, Y. Baichtock, M. Sosna, Plat. Met. Rev. 36 (1992) 90. [38] N. Itoh, Catal. Today 25 (1995) 351. [39] J. Ross, E. Xue, Catal. Today 25 (1995) 291. [40] J.S. Henis, M.K. Tripodi, Sep. Sci. Tech. 15 (1980) 1059. [41] R. Shreve, J. Brink, Jr., Chemical Process Industries, 4th ed., McGraw-Hill, New York, 1977. [42] W. Leffler, Petroleum Refining for the Non-Technical Person, 2nd ed., PennWell Books, Tulsa, OK, 1975. [43] P. Courty, A. Chauvel, Catal. Today 29 (1996) 3. [44] J.-P. Lange, K. deJong, J. Ansorge, P. Tijm, Stud. Surf. Sci. Catal. 107 (1997) 81. [45] N. Itoh, AIChE J. 33 (1987) 1576.

reactions appear to limit near term commercial application using membrane catalytic reactors until the many signicant technical and engineering hurdles described earlier are resolved. Dense oxide membranes for the selective oxidation of CH4 to syngas appear attractive but considerable hurdles remain. Beyond all these major applications, there are probably niche opportunities that might emerge sooner because of less severe temperature and pressure constraints. 4.5. What does the future look like? It would appear that in the next 10 years there are lots of hurdles to surmount, but there are also lots of opportunities. One has to dene and focus on the critical technical and process issues. As these are solved the science can move ahead to clear the hurdles. It is clear that we are many years away from successful commercialization of large scale catalytic membrane reactor technology on a wide basis. Acknowledgements I wish to thank Air Products and Chemicals for the permission to publish this work, and the interest in this topic by the Japan High Polymer work group which rst encouraged me to try to summarize opportunities on this topic. I also wish to thank Paul Dyer and Rao Madhuker of Air Products and Chemicals for carefully reviewing this paper. References
[1] J.N. Armor, Appl. Catal. 49 (1989) 125. [2] J.N. Armor, Chemtech (1992) 557563. [3] G. Saracco, V. Specchia, Catal. Rev.-Sci. Eng. 36 (1994) 305 384. [4] J. Shu, B.P.A. Grandjean, A. Van Neste, S. Kaliaguine, Can. J. Chem. Eng. 69 (1991) 10361060. [5] M.P. Harold, C. Lee, A.J. Burggraaf, K. Keizer, V.T. Zaspalis, R.S.A. de Lange, MRS Bulletin, April 1994, pp. 3439. [6] N. Itoh, Sekiyu Gakkaishi 33 (1990) 136146. [7] H.P. Hsieh, Catal. Rev.-Sci. Eng. 33 (1991) 170. [8] J.-A. Dalmon, in: G. Ertl, H. Knozinger, J. Weitkamp (Eds.), Handbook of Heterogeneous Catalysis, VCH, Weinheim, Germany, 1997, pp. 13871398. [9] J. Zaman, A. Chakma, J. Membr. Sci. 92 (1994) 128.

J.N. Armor / Journal of Membrane Science 147 (1998) 217233 [46] [47] [48] [49] [50] [51] [52] [53] [54] [55] [56] [57] [58] [59] [60] [61] [62] [63] [64] V.M. Gryaznov, Plat. Met. Rev. 30 (1986) 6872. E. Kikuchi, Hyomen 31 (1993) 18. E. Kikuchi, Catal. Today 25 (1995) 333. E. Kikuchi, Cattech 1 (1997) 67. T. Maksuda, I. Koike, N. Kubo, E. Kikuchi, Appl. Catal. A 96 (1993) 3. J. Collins, R. Schwartz, R. Sehgal, T. Ward, C.J. Brinker, G. Hagen, C. Udovich, Ind. Eng. Chem. Res. 35 (1996) 4398. J. Ali, A. Baiker, Appl. Catal. A 140 (1996) 99. G. Gallaher Jr., T. Gerdes, P. Liu, Sep. Sci. Tech. 28 (1993) 309. B.A. Raich, H. Foley, Appl. Catal. A 129 (1995) 167. S. Ziemecki, Catalysis (1987) 625. H. van Veen, P. Alderliesten, Proceedings of the Fourth Workshop on Optimization of Catalytic Membrane Reactor Systems, May 1997. H. Van Veen, J. Tol, C. Engelen, H. Veringa, Key Eng. Mater. 61 (1991) 593. H. Gunardson, Industrial Gases in Petrochemical Processing, Marcel Dekker, New York. S. Uemiya, N. Sato, H. Ando, T. Matsuda, E. Kikuchi, Appl. Catal. 67 (1991) 223. S. Laegsgaard, P. Hjlund Nielsen, P. Lehrmann, Catal. Today 25 (1995) 303. K. Aasberg-Petersen, C. Nielsen, S. Laegsgaard Jorgensen, Proceedings of the Fourth European Workshop on Methane Activation, June 1997, Limerick, Ireland. J. Sogge, T. Strom, Stud. Surf. Sci. 101 (1997). D. Ridler, M.V. Twigg, in: V. Twigg (Ed.), Catalyst Handbook, 2nd ed., Wolfe, Frome, UK, 1989, p. 225. J.N. Armor, Appl. Catal., under review.

233

[65] A. Basile, A. Criscuoli, F. Santella, E. Drioli, Gas Sep. Purif. 10 (1996) 243. [66] S. Uemiya, N. Sato, H. Ando, E. Kikuchi, Ind. Eng. Chem. Res. 30 (1991) 585. [67] M. Bracht, P. Alderliesten, R. Kloster, R. Pruschek, G. Haupt, E. Xue, J. Ross, M. Koukou, N. Papayannoakos, Energy Convers. Manage. 38 (1997) S159. [68] U. Balachandran, R. Poeppel, R. Mieville, T. Kobylinski, Bull. Amer. Cer. Soc. 74 (1995) 71. [69] U. Balachandran et al., Appl. Catal. A 133 (1995) 19. [70] T. Mazenec, T. Cable, J. Frye, Jr., R. Kliewer, US Patent 5 306 411 (1994). [71] M. Schwartz, J. White, M. Myers, S. Deych, A. Sammells, Proceedings of the AIChE 1997 Spring National Meeting, Houston, TX, March 1997. [72] New Technology Week, 9 June 1997, p. 10. [73] Chemical Week, 28 May 1997, p. 13. [74] J. Galuszka, R. Pandey, S. Ahmed, Proceedings of Fifth European Workshop on Methane Activation, June 1997, Limerick, Ireland. [75] US Bureau of Mines, Report No. IC9338 (1993). [76] F. Kapteijn, J. van de Graaf, J. Moulijn, Proceedings of Fourth Workshop on Catalytic Membrane Reactors, Oslo, Norway, May 1997. [77] S. Morooka et al., Ind. Eng. Chem. Res. 36 (1997) 649. [78] E. Kikuchi, Y. Chen, Natural Gas Conversion IV, Stud. Surf. Sci. Catal. 107 (1997) 547. [79] T.S. Farris, J.N. Armor, Appl. Catal. A 96 (1993) 2532. [80] J.N. Armor, T.S. Farris, in: L. Guzzi, F. Solymosi, P. Tetenyi (Eds.), New Frontiers in Catalysis, Proceedings of the 10th International Congress on Catalysis, July 1992, Budapest, Hungary, Elsevier, Amsterdam, 1993, p. 1363.

Вам также может понравиться