• Aqueous Phase Reforming
• Bimetallic catalyst
• Catalysis
• Computational Study
• Steam reforming

Hydrogen can play a role in the development of renewable energies, which will certainly play a crucial role in the energy transition of the future. Catalytic reforming processes are well-established techniques to obtain hydrogen with high efficiency of conversion. Indeed, steam reforming of biomass or biomass-derived alcohols such as methanol accounts for a great amount of hydrogen production (> 95 %) in refinery and chemical industries. However, a very high temperature under the harsh reaction condition of steam reforming causes catalytic deactivation by coking or catalytic sintering. So, having an optimum yield of hydrogen production at comparatively lower temperatures becomes critical. Experimental observations reveal that introducing an effective promotor into monometallic catalyst may represent one efficient way to stabilize and improve activities of steam reforming metal catalysts. An example of adding an effective promotor is indeed reported in our results described in chapter 3. Within a strict collaboration with experimental colleagues, it has been shown that Cu-In catalysts on silica supports shows an increase of H2 product selectivity when compared with Cu/SiO2 experimentally. The reason for the enhanced catalytic activities of Cu/SiO2 by adding indium were not clear. We applied a powerful technique as the Global Optimization search (GO) to propose finite cluster models of (partially oxidized) monometallic Cu and bimetallic Cu-In catalysts on a model silica support. The GO predictions invariably produced a “necklace” or “belt” of partially oxidized In atoms (InOx phase) at the interface with the silica support in all lowest-energy structures. In is found to act as an electron donor and transfers electrons to the Cu/SiO2 phase, thus modulating the electronic properties of Cu in Cu-In systems. The average CLS predictions are in line with the experimental CLS data, indicating the reliability of our Cu and Cu-In models. We propose that H2O dissociation is a key step to determine the catalytic activity in CH3OH steam reforming on Cu and Cu-In catalyst. We examined the energetics for H2O dissociation on Cu and Cu-In clusters and the overall result shows that Cu-In can catalyze the direct dissociation of H2O to H* and O* products with smaller energy barriers, in good agreement with and rationalizing experimental data. Thus, In acts as a promoter in the Cu/SiO2 system to enhance catalytic activity for CH3OH steam reforming. Focusing on the stability of bimetallic catalysts under e.g. aqueous solutions, there are various systems of bimetallic alloys in which the leaching of one of the metals into the solution occurs due to the difference in the electropositive properties of the metals. The catalytic stability of metal alloys in the reactor indeed strongly depends on the reaction conditions. The leaching process is a dynamic process, and complicated due to being composed of several different reactions going through unknown configurations. Bimetallic Pt-Mn catalysts have been reported experimentally as excellent catalysts for steam reforming and aqueous phase reforming. However, experiment indicate that Mn is leached out from the catalyst at some point of the catalytic runs, thus resulting in a lack of catalytic stability. Due to the complexity of this leaching process, theoretical investigations of this process in the literature are scarce and the mechanism of the leaching process of bimetallic alloys is still unclear. In chapter 4, we addressed this challenging problem and tried to produce some knowledge in this field by studying the oxidation and dealloying of PtMn under oxygen coverage condition. We aimed at understanding the reaction mechanism in the initial steps of oxidation of Pt-Mn models. Since this is a complex transformation with many possible phenomena that can potentially happen, we used an equally complex approach. We combined Density-Functional Theory (DFT) with neural network potentials and meta-dynamics simulations to accelerate the mechanistic search. As an important outcome, we proposed an effective scheme to incrementally sample the Potential Energy Surface (PES) of the reaction and finally extract important intermediates to augment the structural database to produce putative reaction paths that are finally fed into NEB simulations to predict accurate energetics at the DFT level. The obtained results from this mechanistic study provided (to the best of our knowledge, for the first time) a detailed mechanism of oxidation and leaching of Pt-Mn systems with the reasonable energy barriers in the range of 0.9 – 1.0 eV. Aqueous phase reforming (APR) is a promising reaction for the valorization of biomass-derived compounds for hydrogen production. APR occurs at a milder reaction conditions and lower processing costs compared to traditional steam reforming. Since APR is an innovative process, many scientists have devoted their efforts to find the optimal catalytic system that maximizes activity, selectivity and stability. In addition, under the reaction condition of APR, products and intermediates can be produced in both gas and liquid phases, thus increasing the complexity of the reaction. Further studies and investment of resources in the APR process are needed in order to promote its use for hydrogen production at the industrial scale. In chapter 5, our research has been focused on explaining the catalytic performance of Pt-Mn catalyst for APR of methanol. Within a strict collaboration with experimental groups, the synthesis and catalytic testing of bimetallic Pt-Mn was achieved experimentally and investigated theoretically. In the presence of Mn, the Pt-Mn system shows a better performance for catalyzing the APR reaction of CH3OH. Investigating the APR energetics on monometallic Pt and bimetallic Pt-Mn can help interpret and understand the catalytic performance of Pt and Pt-Mn systems. The reaction mechanism of CH3OH-APR mainly involves two main chemical reaction steps which are methanol decomposition and successive water-gas shift reaction. An important outcome from our studies is that hydrogen abstraction from CHxOH species results in high coverage of hydrogen on the Pt catalyst due to the strong interaction between Pt surface atoms and hydrogen under APR reaction conditions. To the best of our knowledge, this is the first report of CH3OH-APR reaction under hydrogen coverage. We examined the detailed reaction mechanism of CH3OH-APR at high hydrogen coverage on Pt and Pt-Mn catalyst. H2O dissociation to OH + H is the rate determining step for CH3OH-APR on Pt catalyst. The Mn surface atom activates H2O to produce OH* and H* with a smaller barrier. Remarkably, both Mn on the surface and Mn underneath can weaken the interaction between hydrogen and Pt surface atoms, thus alleviating the hydrogen poisoning issue and helping to promote hydrogen evolution. In chapter 7, we deal with a different system in which experimental partners introduced In as a promotor into Ni-In catalyst for steam reforming of oxygenates, finding that Ni-In catalyst shows a better performance for hydrogen production. Remarkably, in the presence of In, coke formation is suppressed, thus enhancing the catalytic stability of bimetallic Ni-In. Theoretically, we evaluate the stability of In-doped Ni systems by calculating the mixing energy of the alloys. In atoms mix well with Ni atoms in the surface layer, but are not likely to penetrate to the sublayers of Ni in the first stage of the doping process (low In content). In surface atoms inhibit carbon adsorption on In sites and weaken the adsorption strength of carbon on the surface of Ni-In, thus reducing carbon-adsorption-induced phenomena such as e.g. restructuring to (311) facets, which are responsible for the formation of carbon nanotubes leading to coke formation and catalyst deactivation. In chapter 6, we explored a completely novel sampling approach, that we name “conformal sampling”. The idea here is to exploit the knowledge and structural databases obtained from methanol oxidation on Pt to accelerate constructing the potential energy surface of methanol APR on other metal systems, assuming it occurs via a similar reaction mechanism, with the help of Machine Learning techniques. To achieve this, we combined several computational techniques including rescaling (affine transformations) to duplicate the database from a worked-out case (Pt metal under high coverage conditions) to 8 cases of different metals (under low-coverage conditions), applying NEB simulations using a higher order equivariant message passing neural network potential (MACE-NNP) to search for transition state structures and activation energies of reaction. We demonstrate that our approach can speed up building potential energy surface (PES) of methanol APR on diverse metal cases with accurate energy landscape and energy barrier, actually with an accuracy below 0.1 eV in energy barriers (to the best of our knowledge, for the first time in this type of studies) which seems sufficient to quickly screen candidate catalysts. It will be very interesting to apply our approach to explore other chemical reactions to further validate our novel approach. In conclusion, we can summarize a general perspective. We think that further studies in catalyst design for blue hydrogen production should be oriented toward finding a balance between catalytic stability and activity. Finding the optimum conditions to equilibrate between these two factors will improve the efficiency of catalysts in their long-term use. We hope that the obtained results from this thesis will provide useful guidelines in the design and development of efficient metal catalysts for blue hydrogen production via steam and aqueous-phase reforming processes.