Preferential Oxidation of Carbon Monoxide on Structured Supports

Abstract

Hydrocarbon fuels must be reformed in a series of steps to provide hydrogen for use in proton exchange membrane fuel cells (PEMFCs). Preferential oxidation (PROX) is one method to reduce the CO concentration to < 10 ppm in the presence of ~40% H&#8322;, CO&#8322;, and steam. This will prevent CO poisoning of the PEMFC anode. The PROX reaction requires an active, selective, and stable catalyst. Structured supports, such as ceramic monoliths, can be used for the PROX reaction. Metal foams offer advantages over the traditional ceramic monolith: higher thermal conductivity, radial mixing and heat transport, and a durable, low density, high strength structure.

Reaction studies were conducted on catalyzed structured supports using a fixed bed adiabatic reactor with an online non-dispersive IR gas analyzer. A study on ceramic monoliths showed that higher Fe loadings promoted on 5 wt% Pt / g-Al&#8322;O&#8323; increased CO and O&#8322; conversions and decreased CO selectivity. A study on metal foams showed that lower cell density and higher pores per inch foams exhibited higher CO conversions and selectivity. Under most operating conditions, the CO conversion and selectivity of the best 5 wt% Pt / 0.5 wt% Fe metal foam were comparable to the ceramic monolith. Comparison tests showed lower CO conversion and selectivity for the ceramic foam and the metal monolith compared to the metal foam and ceramic monolith.

Two important effects limit the PROX reaction: the reverse water-gas-shift (r-WGS) reaction, and transport resistances. Under adiabatic conditions, the r-WGS reaction made it impossible to achieve low outlet CO concentrations. The metal foam showed less r-WGS activity compared to the ceramic monolith. The effects of space velocity and linear velocity were studied independently using various catalyst lengths and flow rates. The CO conversion increased at higher linear velocities, which suggested mass transfer resistance between the bulk gas and the catalyst surface.

Carbon monoxide pulse chemisorption and temperature programmed desorption (TPD) were used to determine the number of active metal sites. Pulse chemisorption and TPD experiments on the blank reactor and the blank metal foam wrapped with glass wool insulation showed no CO adsorption. No CO adsorption was observed from pulse chemisorption tests on the ceramic fiber insulation, however, considerable CO desorption was observed from TPD tests. On catalyzed supports, an elongated tail in the pulse chemisorption tests was attributed to CO adsorption on either the g-Al&#8322;O&#8323; washcoat or on surface impurities. The pulse chemisorption and TPD results did not describe the catalyst performance results. Higher values of the number of CO adsorbed on the catalyst surface did not correspond to higher CO conversions from the PROX reaction.