| Hydrogen purification membranes are traditionally made from pure Pd and its alloys because these materials offer high throughputs while remaining 100% selective and catalyzing the dissociation/recombination reactions at the surface. However, the high cost of Pd has necessitated the development of newer, cheaper purification systems. The Group V refractory metals Ta, Nb, and V promise even higher permeability at a lower cost, but they are susceptible to forming oxide layers that are impervious to hydrogen. One proposed solution has been to remove these surface layers and apply thin, protective coatings of Pd to form a metallic triple layer.;The electrochemical permeation technique was used to study hydrogen transport through pure Pd, Pd-Ag, and Nb substrates coated with Pd or Pd-Ag. A multilayer diffusion model was developed, in accordance with conditions of the experiments, which can extend to any n-layered system and account for general non-equilibrium interface conditions. Exact solutions were obtained in the Laplace domain and inverted numerically using an appropriate algorithm.;Graphical representations of concentration or flux were produced for various example systems to demonstrate multilayer phenomena associated with different surface conditions or with changes in relative layer thickness, diffusivity, and solubility. A discussion on the effects of non-equilibrium includes further graphical examples and focuses on the deviations from equilibrium behavior.;The time lag tL for a general triple layer with non-equilibrium interfaces is derived completely. The significance of each term is discussed, and special limiting forms of tL are obtained and their physical implications are explained. It is also shown how tL can be attained simply from knowledge of F(s). Parametric analyses are performed with respect to both the time lag tL and the breakthrough time tb, the latter having no analytical form for multilayered systems to this point. Expressions for tb in a symmetric triple layer are derived that are accurate in certain parametric regions. These regions are evaluated through an error analysis comparing the analytical tb with the single layer approximation and the exact value, which shows that Pd/Nb is a system for which the analytical formula applies. Formulas for non-equilibrium cases are developed and discussed as well.;Experimental data from Pd/Nb/Pd specimens is described well by the equilibrium model. Calculated values of k for Pd/Nb are in agreement with theory and past studies. Since k = KPd/KNb << 1 and DPd < DNb, the downstream Pd layer has a barrier effect that causes accumulation in the Nb layer. The overall process is not diffusion-limited with respect to Nb, so D Nb is not evaluable from tL or the full transient. DNb was instead determined from the breakthrough time tb and was consistent with previously reported values. Results suggest that the equilibrium assumption was valid at the interfaces and that bulk behavior was observed. The results are discussed in the context of some common concerns regarding metallic composites, including: anomalous behavior in thin films, presence of discrete intermixing layers at the interfaces, and interface disturbances resulting from non-equilibrium conditions. |
