A Palladium Membrane Reactor (PMR) system is under consideration for the tritium plant of the International Thermonuclear Experimental Reactor (ITER). The ITER reactor exhaust will contain tritiated impurities such as water and methane. Tritium will need to be recovered from these impurities for environmental and economic reasons. The PMR is a combined permeator and catalytic reactor. Shift catalysts are used to foster reactions such as water-gas shift, H2O + CO -> H2 + CO2, and methane steam reforming CH4 + H2O -> 3H2 + CO. Due to thermodynamic limitations these reactions only proceed to partial completion. Thus, a Pd/Ag membrane, which is exclusively permeable to hydrogen isotopes, is incorporated into the reactor. By maintaining a vacuum on the permeate side of the membrane, product hydrogen isotopes are removed, enabling the reactions to proceed toward completion.

Previously reported results show that the PMR is capable of extremely high decontamination factors at a feed composition of nominally 1/3 methane, 1/3 water, and 1/3 argon. This paper gives results for varied inlet compositions of relevance to ITER. The inlet impurities composition ranged from pure CH4 to pure H2O. For the pure H2O case, CO addition was required so that the water-gas-shift reaction can occur. For the pure CH4 case, oxygen addition was required so that reactions such as methane-steam reforming could take place and so that carbon deposition could be avoided.

Data show that PMR performance does not depend on whether hydrogen originates from methane or water. However, performance is dependent on the total quantity of hydrogen isotopes in the feed. This dependence indicates that permeation rate (not reaction rate) is the rate controlling step for the geometry and operating conditions in these experiments. Performance is also dependent on the residence time in the reactor. Performance was found to be only weakly dependent on the rate of CO or O2 injection. The decontamination factor (DF) was determined as a function of carbon-to-oxygen ratio in the feed. It was found that good DFs were observed over a wide range of C:O feed ratios. From this it was concluded that precise feed composition control is not required for effective PMR performance, making PMR operation relatively simple.