Abstract
AbstractHydrogen is one of the most potential clean energies in the 21st century. Although it can be produced by solar light, electricity, etc., the purification process is still very costly step. To lower the cost and get higher purity and large amount of hydrogen, palladium-silver membrane supported on the porous substrate is used in hydrogen purification. The palladium-silver alloy membrane has been extensively studied in recent years due to the absence of α/β phase transition and embrittlement associated with hydrogen permeation. In addition to the application for ultra high purity hydrogen separation, it is also widely utilized in recovery or purification of hydrogen in petroleum industry, hydrogen reactor in fuel cells, and methane-steam reforming.In this study, silver powder agglomerate was used to prepare porous substrate. The porous silver substrate was obtained by uniaxial mechanical pressing and sintering at 350℃ in air for 1 hr. Prior to sputter deposition, the surface was modified by filling with Pd70Ag30 nanoparticles to reduce the pore size and cleaning with acetone to remove the residual particles. After the surface modification, a 5μm-thick Pd-Ag membrane was deposited on the surface by dc magnetron sputtering. For the tube substrate, the membrane was deposited with a rotating design. The surface and cross-section morphologies of the substrate and membrane were examined by scanning electron microscopy. The compositions of the membrane were calibrated by X-ray diffraction. Besides, the permeation rate of hydrogen at 250℃ was measured by a gas permeation apparatus.It was observed that cleaning surface with acetone was very important for the modification process. A dense and continuous Pd-Ag membrane could be uniformly prepared on the modified tube surface. A 5μm-thick membrane was successfully deposited on the 2cm-long silver tube surface. The maximum hydrogen pressure the tube membrane could support was 2.5 kg/cm2. The thermal stability and durability at different pressures was very good even after 48 cycles of test. The Sievert’s law was not obeyed in this study.