A mathematical study on membranes and catalysis for hydrogen production from ammonia

Ammonia is a promising carbon-free H2 carrier that has a superior H2 content and energy density compared to alternative carriers. To use ammonia as a hydrogen carrier, the hydrogen stored in its chemical bond has to be recovered through ammonia decomposition (NH3 ⇌ 0.5 N2 + 1.5 H2). The endothermic reaction and subsequent separation of H2 are energy intensive steps. A catalytic membrane reactor can be used to circumvent thermodynamic limitations, combining catalytic reaction and separation in a single unit, to separate the H2 at high purity for applications such as fuel cells or combustion. Carbon membranes could become an interesting separation method for this application, due to their potential for a high permselectivity, good permeability, thermal- and chemical stability, high tuneability, and relatively low-cost of production materials. However, the mechanisms of transport through carbon membranes as well was the kinetics of ammonia decomposition in such systems are not yet fully understood. Understanding these phenomena is crucial for improved membrane development, catalyst development, and their integration within a single reactor unit. This dissertation therefore focuses on the development of mathematical models to describe and investigate ammonia decomposition in carbon membrane-assisted packed bed membrane reactors. After the general introduction reported in the first Chapter, a general modelling approach is developed in Chapter 2 to describe gas permeation through carbon membranes. Currently, not many carbon membrane models exist due to the complex nature and combination of transport phenomena. The developed approach assumes a bundle of pores with parallel resistances, using the pore size distribution as a weight factor to sum the different transport phenomena. Imposing realistic boundary conditions, the model is fit to pure- and mixed gas experimental data using global optimization algorithms. In Chapter 3, several LHHW-type kinetic models are developed to describe ammonia decomposition on a commercial Ru/Al2O3 catalyst. An experimental dataset obtained from microreactor measurements are used to fit the models. To assess their applicability at larger scale, a validation is performed with measurements in a larger packed bed reactor. In Chapter 4, a microkinetic model is developed to study the ammonia decomposition mechanism in greater detail. A hybrid strategy is developed that integrates DFT-based kinetic parameters with a neural network-enhanced global fitting routine to fit the microkinetic model. This model identifies NH-H bond scission as the rate-determining step and highlights NH2* and H* as the most abundant reaction intermediates. Results are validated with a thermodynamic partition function analysis, yielding excellent agreement with experimental enthalpy and entropy values. The proposed framework enables accurate model refinement beyond the limitations of first-principles methods and offers a scalable approach for guiding catalyst development in ammonia decomposition and related systems. Chapter 5 integrates the fitted carbon membrane permeation model and a new LHHW-type kinetic model derived based on the microkinetic model results from Chapter 4 in a 1D heterogeneous reactor model. A preliminary reactor study is performed to investigate ammonia decomposition in a packed bed membrane reactor. Transport phenomena are investigated and an initial outlook is given on the potential of this technology. Finally, the overall conclusions and recommendations for future research are discussed in Chapter 6.