Improving the estimation quality of parameters in kinetic models for hydriding/dehydriding reactions: An OED study

• We proved improper data processing and experimental design can cause large disparity in parameter estimate.
• We found the estimate of pre-exponential factor C is subject to prohibitively large error (>10%).
• We suggested minimizing the uncertainty of k estimate to gain more accurate overall estimate on other parameters.
• We proposed a practical OED-based sequential procedure able to reduce uncertainty in parameter estimate by half.

Among the hydrogen storage properties of solid materials, e.g., metal hydrides, the reaction kinetics is of great importance and is often investigated by fitting reaction rate curves to given models. Thanks to such models, key parameters such as activation energies can be found. However, it is noteworthy that the parameters thus obtained often show great disparities even for the same material. In addition to experimental inaccuracies, these disparities may also be attributed to the improper processing of the data and to suboptimal experimental design. In this paper, we present some theoretical tools based on optimal experimental design (OED) to analyze and optimize the uncertainty on the estimated parameters. The uncertainty is derived from asymptotic statistics and it is shown to be closely linked to the sensitivity of the experiment to the parameters. We found that by applying a new OED-based procedure, the estimate of the kinetic parameters can be effectively improved.