ABSTRACT

Membrane reactors are a combination of a chemical reactor and a membrane separation device. The membrane should be permselec-tive for one of the products of the chemical reaction and the mem-brane must be in the vicinity of the chemical reaction. The goal is to drive an equilibrium-limited reaction toward complete conversion by selectively removing a product from the reaction zone. For high-temperature reforming reactions, polymeric membranes are not suitable, but inorganic membranes are chemically and thermally sta-ble under the reaction conditions. Many reports have been published describing the effective use of membrane reactors for a variety of high-temperature reactions, including methanol-steam reforming.1-3Palladium-alloy membranes are a good choice for use in a mem-brane reactor if hydrogen is a principal product. Alternatively, a thermally and chemically stable membrane that is permselective for carbon dioxide would also be a good choice, although at the present no such membranes are commercial available.Favorable attributes of membrane reactors include the potential for a smaller and less costly overall system by virtue of combining reaction and separation into one unit operation. One may also expect the forward reaction rates will be enhanced (since reverse reaction rates are necessarily decreased), thereby further reducing

the size of the reactor. Also, conversion beyond normal equilibrium limitations can be achieved in a single pass, reducing the requirement for recycling unconverted feedstock.Drawbacks include a very challenging design that may drive cost up and typically poor mass transfer when catalyst and membrane are placed within the same physical space. Since methanol-steam reforming is not an equilibrium-limited reaction (the reverse reaction kinetics are slow), application of membrane reactors to methanol-steam reforming is not a recommended approach. This chapter will examine the concept of membrane reactors as applied to methanol-steam reforming and explain why this technical approach is disfavored. Although many technical papers have been published with titles suggesting that they address the application of membrane reactors to methanol reforming, the majority of these papers really discuss the reaction of carbon monoxide with water to yield hydrogen within a membrane reactor. This is an important distinction, and the remainder of this chapter will deal first with methanol-steam reforming and then the water-gas shift reaction, and the suitability of membrane reactors for each reaction. 5.1 Reactor Performance

The goal of a membrane reactor is to draw off a product species from a chemical reactor during the course of reaction. In this case, the reaction of interest is methanol-steam reforming. The two principal reaction products are hydrogen and carbon dioxide. Carbon monox-ide is not considered a principal reaction product; it is considered an intermediate product and is better converted to carbon dioxide and hydrogen via the water-gas shift reaction.The chemical reactor must cause the generation of hydrogen and carbon dioxide. Since there are no commercial, high-temperature, carbon-dioxide-selective membranes, we will focus the discussion on a hydrogen-selective membrane. As the reaction proceeds, hydrogen must come in contact with, and diffuse through, the permselective membrane.To understand how the membrane will perform in the context of a membrane reactor, first the performance of the reactor must

be established. As an example, consider a simple slab reactor: i.e., a rectangular reactor that is long relative to its width and height (see Fig. 5.1). It is assumed that the reactor is just large enough so that complete conversion of methanol (and water) occurs immediately prior to the exit from the reactor. It is also assumed that the con-version is directly proportional to the distance the feed stream has traveled along the length of the reactor. In other words, 50% con-version occurs when the feed stream has traveled halfway along the length of the reactor. The example shown in Fig. 5.1 indicates 10 segments, all of equiv-alent length, so we may consider that 10% conversion of methanol (and water) occurs within each of the 10 segments. For this discus-sion it does not matter what is the operating temperature or pres-sure, but only that the conditions and catalyst are such that 10% conversion of the reactants occurs within each reactor segment. Also, the shape and size of the segments is not relevant; only that the assumed conversion of reactants occurs within each segment.Making a further assumption that the feed stream is a 1:1 molar ratio of methanol and water and that the reaction proceeds accord-ing to the ideal stoichiometry (CH3OH + H2O = CO2 + 3H2) a plot of hydrogen partial pressure and hydrogen mole-fraction as a function

of distance traveled along the length of the slab reactor is presented in Fig. 5.2. Initially, there is no hydrogen present and it builds up as the distance along the reactor progresses toward the end of the reactor. At complete conversion the reformate stream (product gas) is 75% hydrogen. In a membrane reactor this reaction profile largely governs the performance of the hydrogen-selective membrane since the reactor and membrane module are intimately associated within the same vessel. This means that the hydrogen partial pressure is increas-ing-from initially zero-as the feed stream enters and passes through the reactor/membrane module. Although this example used a simple geometry (slab reactor), any geometry will yield conceptu-ally similar results.