Palladium Under Pressure: Multiscale Modeling of Nanoalloys for Hydrogen Sensing

Doctoral thesis, 2025

Pd nanoalloys offer a spark-free, plasmonic, and highly tunable platform for H₂ sensing, which is a crucial aspect of a safe hydrogen economy. Owing to a fortunate combination of thermodynamic and optical properties, Pd nanoparticles rapidly absorb H₂, which induces a measurable optical shift. Pure Pd sensors suffer, however, from issues such as hysteresis during hydrogenation and CO poisoning. Alloying with Au and Cu have been found to mitigate these issues, but introduces challenges related to long-term stability and performance.

The vast configurational space accessible through alloying and nanostructuring makes computational modeling an efficient route to understand and optimize nanoalloys for H₂ sensing. This thesis develops multiscale models to better understand the optical and thermodynamic properties of Pd nanoalloys for H₂ sensing, anchored at the atomic scale via first-principles calculations.

The effect of alloying Pd with Au and Cu on the surface composition and adsorbate coverages under different environments is studied via cluster expansion models. It is found that Pd segregates to the surface in H₂ and CO environments, due to strong adsorption, while Au segregates to the surface under vacuum conditions. Cu shows a more complex behavior, with a non-trivial preference for the subsurface layer under most conditions, and only modest presence in the top surface layer. The H–CO coadsorption behavior is primarily governed by the fabrication conditions, dictating whether Pd or Au segregates to the surface, while tuning the exact bulk alloy composition has only a minor effect. The experimentally observed role of Cu in mitigating CO poisoning must therefore go beyond adsorption thermodynamics, potentially by providing energetically feasible H absorption paths through the surface when the energetically most favorable paths are blocked by CO.

The H sensitivity of nanodisk devices is optimized by combining atomic-scale dielectric functions with continuum electrodynamic simulations of nanoalloy structures. Single disk simulations suggest that the H-induced plasmon shift is limited by an interplay between localized surface plasmonic resonances and interband transitions. In addition, a computational platform for designing optimal nanoarray-based sensors for specific targets is presented, paving the way for future efforts in multiplexed sensor design.