Dense Ceramic Membranes for Hydrogen Separation

Novel cermet (i.e., ceramic-metal composite) membranes have been developed for separating hydrogen from product streams that are generated during coal gasification, methane partial oxidation, and watergas shift reactions. The hydrogen permeation rate in the temperature range of 600-900°C has been measured for three classes (ANL-1, -2, and -3) of cermet membranes. ANL-3 membranes provided the highest hydrogen flux: ≈20 cm(STP)/min-cm for a 40-μm-thick sample at 900C using 100% H2 as the feed gas. The effects of membrane thickness and hydrogen partial pressure on hydrogen flux indicate that the bulk diffusion of hydrogen is rate-limiting for ANL-3 membranes with thickness >40 μm. ANL-3 membranes were tested in simulated syngas (66% H2, 33% CO, 1% CO2) at several temperatures for times approaching ≈200 h, and no degradation in performance was observed. The lack of degradation in simulated syngas suggests that the membrane is chemically stable and may be suitable for long-term operation. INTRODUCTION The U.S. Department of Energy's Office of Fossil Energy sponsors a wide variety of research, development, and demonstration programs that aim to maximize the use of vast domestic fossil resources and ensure a fuel-diverse energy sector while responding to global environmental concerns. Development of cost-effective, membrane-based reactor and separation technologies is of significant interest for applications in advanced fossil-based power and fuel technologies. Because concerns over global climate change are driving nations to reduce CO2 emissions, hydrogen is considered the fuel of choice for the electric power and transportation industries. Although it is likely that renewable energy sources will ultimately be used to generate hydrogen, technologies based on fossil fuels will supply hydrogen in the interim. The purpose of this work is to develop dense hydrogen-permeable membranes for separating hydrogen from mixed gases at commercially significant fluxes under industrially relevant operating conditions. Of particular interest is the separation of hydrogen from product streams that are generated during coal gasification (IGCC), methane partial oxidation, and water-gas shift reactions. Because the membrane will separate hydrogen without using electrodes or an external power supply (i.e., its operation will be nongalvanic), it requires materials that exhibit suitable electronic and protonic conductivities as well as high hydrogen diffusivity and solubility. Good mechanical properties will also be necessary to withstand operating stresses. In addition, the fabricated materials must be thin and dense in order to maximize the hydrogen flux and maintain high hydrogen selectivity. Membrane development at Argonne National Laboratory (ANL) and the National Energy Technology Laboratory (NETL) focused initially on BaCe0.8Y0.2O3-δ (BCY), because it is a mixed proton/electron conductor with a high total electrical conductivity [1, 2] and may therefore yield a high hydrogen flux without using electrodes or electrical circuitry. Despite having a high total electrical conductivity, however, its electronic component of conductivity is insufficient to support a high nongalvanic hydrogen flux [3, 4]. In order to increase the electronic conductivity, and thereby increase the hydrogen flux, we have developed various cermet (i.e., ceramic-metal composite) membranes, in which a metal powder is dispersed in a ceramic matrix [5, 6]. In these cermets, the metal enhances the hydrogen permeability of the ceramic phase by increasing the electronic conductivity of the composite. If a metal with high hydrogen permeability is used, it may also provide an additional transport path for the hydrogen. The cermet membranes in this paper are classified as ANL-1, -2, or -3, based on the hydrogen transport properties of the metal and matrix phases. ANL-1 membranes contain a metal with low hydrogen permeability that is distributed in a matrix of hydrogen-permeable BCY. ANL-2 membranes also have a BCY matrix, but contain a metal with high hydrogen permeability (i.e., a hydrogen transport metal). In ANL-3 membranes, a hydrogen transport metal is dispersed in a ceramic matrix with low hydrogen permeability, e.g., Al2O3 or BaTiO3. Specific membranes are identified by a number and a letter, where the number represents the type of membrane, e.g. ANL-3 membranes, and the letter indicates a particular combination of metal and matrix phases, e.g., ANL-3a is an ANL-3 membrane that contains "metal-a" in a matrix of Al2O3, whereas ANL-3b is another ANL-3 membrane that contains a different metal and/or a different ceramic matrix. In making general comments regarding an entire class of membranes, a number is used alone without any letter. The first class of membranes, ANL-1, contains a metal with low hydrogen permeability in a hydrogenpermeable matrix, BCY. The hydrogen flux through ANL-1a is higher than that through monolithic BCY because the metal increases the overall electronic conductivity of the membrane. The hydrogen flux was increased with ANL-2a membranes by replacing the metal of ANL-1a with a hydrogen transport metal, i.e., a metal that has a high hydrogen permeability. Although BCY and the metal phase both contribute to the hydrogen flux through ANL-2 membranes, most of the hydrogen diffuses through the metal [7]. Because BCY contributes relatively little to the overall permeation rate of ANL-2 membranes, exhibits poor mechanical properties, and is chemically unstable under some conditions of interest, ANL-3 membranes were developed. In ANL-3 membranes, the BCY matrix of ANL-1 and -2 membranes is replaced by a ceramic with superior mechanical properties and thermodynamic stability, e.g., Al2O3 or ZrO2, to form a membrane that gives a higher hydrogen flux and has higher strength and greater chemical stability. In this paper, we discuss the state of development of the various cermet membranes (ANL-1, -2, and -3) and compare their hydrogen permeation rates. EXPERIMENTAL BCY powder was prepared at ANL as previously described [5]. All membranes were prepared to contain 40 vol.% metal phase, except where otherwise noted. BCY and metal powders were mixed together to prepare powders for ANL-1a and -2a membranes. Powders for ANL-3 membranes were prepared by mixing one of two hydrogen transport metals with ceramic powders that are reported to be poor proton conductors [8]. Powders were pressed uniaxially to prepare disks (≈22 mm in diameter x 2 mm thick) for sintering. Sintering conditions were selected on the basis of the membrane composition. ANL-1a and -2a membranes were sintered for 5 h in 4% H2/balance Ar at 1420°C. ANL-3a membranes were sintered for 5 h at 1400°C in 4% H2/balance He, ANL-3b membranes were sintered at 1350 C for 12 h in air, and ANL-3d membranes were sintered in air for 5 h at 1390°C. To test permeation, a sintered disk was polished with 600-grit SiC polishing paper and then affixed to an Al2O3 tube that was part of the assembly shown in Fig. 1. A seal formed when the assembly was heated to 950°C and spring-loaded rods squeezed a gold ring between the membrane and the Al2O3 tube. One side of the sample was purged with 4% H2/balance He during sealing, while the other side was purged with 100 ppm H2/balance N2. The leakage rate following this procedure was typically <10% of the total permeation flux.