Pinhole palladium

A classic example of a composite membrane is that patented in 1916 by Snelling who used porous ceramics to support dense layers of palladium, 25 p,m thick [1]. Variations of his theme remain at the forefront of research [2, 3]. Examples include use of porous alumina, silica, perovskites, and stainless steel to support thin layers of Pd and its alloys [2, 3]. Snelling addressed issues of perforations and pinholes... 

For ill-designed composite membranes, for example, formed by depositing palladium onto substrates which it does not wet, surface tension will force the thin film to contract and ball up if the palladium atoms acquire sufficient surface mobility. Pinholes may form as a prelude to complete de-wetting, or pinholes may remain from the initial fabrication if the palladium did not fully wet its substrate. Kinetics of de-wetting is accelerated at elevated temperature and in the presence of adsorbates such as CO, which increase surface mobility of Pd. If molten metals do not wet ceramics, they will be expelled from ceramic pores. During sintering of cermets, Pd and other metals will not adhere to the ceramic phase, if the metal and ceramic do not wet. 

Palladium cermets have a number of advantages over thin films of Pd supported by porous ceramics. For systems which wet, sintering cermets at very high temperatures (well above membrane operating temperatures) produces dense, pinhole-free composites (see Fig. 8.5). Because Pd is closely confined within a matrix of ceramic and because small, individual, micron-size Pd crystallites already possess a small surface-to-volume ratio of low surface energy, the Pd has relatively low driving... 

Complex substrate modifications involving intermediate layers and palladium alloy deposition methods are often required for superior membrane performance. Modification of a membrane support surface before palladium deposition by sintering on smaller particles can create a smoother surface with smaller pores, facilitating the deposition of a defect-free palladium layer. Nickel microparticles have been sintered together to form a porous support that was sputter-coated with palladium and then copper [118]. Thermal treatment at 700 °C for 1 h promoted reflow to create a durable, pinhole-free membrane with a Pd-Cu-Ni alloy film. In another case, starting with commercially available PSS with a 0.5 pm particle filtration cut-ofF, submicron nickel particles were dispersed on the surface, vacnium sintered for 5 h at 800 °C, and then sputtered with UN [159]. The nickel particles created a smoother surface with smaller pores, so a thinner palladium alloy layer... 

For a composite membrane that is relatively free of pinholes, there is apparently a limiting palladium thickness that depends primarily on support quaUty and somewhat on deposition method. For example, unmodified PSS requires at least 10 pm of palladium to bridge a majority of the pores [149, 227]. Invariably, a few macrodefects remain that preclude the attainment of perfect hydrogen permselectivity. On porous asymmetric a-alumina supports with 100-200 nm pore size on the surface, approximately 7 pm appears to be the limiting thickness that is proba-... 

So, if thin membranes are a good idea, what about ultra-thin membranes that are only 1 or 2 (tm thick. Unfortunately, fabrication and handling costs do not decrease linearly with reduction in thickness of the permselective membrane. Experience has shown that qualitatively the relationship between fabricated cost and membrane thickness is approximately as shown in Fig. 5.7. Initially, as the thickness of the membrane is reduced, there is a corresponding reduction in the cost, largely driven by a reduction in the mass of expensive palladium used in the membrane. However, as the membrane thickness is reduced to approximately <5 i-im, the fabrication cost increases faster than can be offset by the reduction in the amount of palladium. A major reason that fabrication costs increase so dramatically is the reduction in the yield of satisfactory membrane product (i.e., membrane product without pinholes, tears, and similar defects). The membrane must be free of defects which would compromise the permselectivity of the membrane. This requirement becomes increasingly difficult to satisfy as membranes become exceptionally thin. 

Barbieri et al. [523] prepared palladium membranes by solvated metal atom deposition, a deposition method that created a 0.1-pm thick film of palladium on an alumina tube. Another membrane was prepared from a palladium/silver alloy (21 wt.% silver), which had a thickness of 10 pm. Both membranes suffered from pinholes and cracks and thus did not show infinite hydrogen selectivity. However, conversion exceeding the thermodynamic equilibrium could still be achieved for methane steam reforming at temperatures exceeding 400 °C. 

Deposition should be performed in controlled conditions to obtain the formation of pinhole-free, adherent palladium film a typical laboratory apparatus is shown in Fig. 3.3. The membrane support is inserted in a reactor and maintained in rotation at constant velocity by a variable-speed motor to allow the removal from the reaction zone of nitrogen produced by hydrazine reduction (Equation [3.5]).The reactor is immersed in a thermostatic bath to keep the temperature at a constant value. Hydrazine can be periodically added from the tube inserted in the middle of the reactor, while nitrogen can be evacuated from the reactor through the tube on the right side. 

---