Data specialist Edoardo Spadetto developed accurate computational methods for calculating correlation energies in molecules and periodic materials. The greatest challenge was the exponential scaling of quantum calculations with system size. He implemented Møller–Plesset perturbation theory (MP2) and Random Phase Approximation (RPA) using numerical atomic orbitals in the AMS-BAND software. The motivation was to enable accurate predictions of material properties and molecular interactions, particularly van der Waals forces and surface chemistry, which are crucial for heterogeneous catalysis, drug design, and materials engineering. He developed stabilization techniques using projector methods, extended pair atomic density fitting to periodic systems, and implemented efficient periodic Hartree–Fock calculations to make these methods computationally feasible for large systems.
Spadetto demonstrates that accurate quantum-chemical calculations can be performed on large molecules and materials using efficient approximation methods. Pair atomic density fitting with projector methods prevents numerical instabilities while maintaining chemical accuracy. For periodic materials, he implemented RPA correlation-energy calculations using new damping techniques that stabilize convergence, enabling accurate predictions of lattice constants and surface interactions. He validated the methods using benchmarks ranging from small molecules to crystalline materials, showing strong agreement with the literature. Importantly, these methods capture van der Waals interactions and correlation effects that are missed by simpler density functional theory, enabling studies of realistic catalytic surfaces and material interfaces.
Spadetto’s research enables more accurate simulations for designing materials and understanding surface reactions, which is crucial for developing better industrial catalysts. For example, accurately modeling the interactions between molecules and metal oxide substrates helps design efficient catalytic converters or improved chemical production processes. The pharmaceutical industry benefits from better predictions of drug–molecule interactions. Materials scientists can design improved batteries, solar cells, and electronic devices. Implementation in the widely used AMS software makes these methods accessible to researchers worldwide. The methods are particularly valuable for organic electronics and materials where van der Waals forces dominate.
More information on the thesis