Density Functional Tight-Binding Models for Band Structures of Transition-Metal Alloys and Surfaces across the...

Area of Science:

• Computational chemistry
• Materials science
• Condensed matter physics

Background:

• First-principles simulations are vital for understanding chemical reactions and bonding.
• Machine learning accelerates simulations but often loses electronic structure data.
• Electronic band structures are key to understanding reactivity in heterogeneous catalysis.

Purpose of the Study:

• To develop efficient computational methods for predicting electronic descriptors in large catalytic systems.
• To preserve ab initio electronic structure information lost in typical machine learning models.
• To enable high-throughput studies of complex catalytic systems beyond DFT computational limits.

Main Methods:

• Development of Shell-Optimized Atomic Confinement (SOAC) DFTB parametrizations.
• Focus on electronic-part-only parameters for d-block transition metals.
• Application to both bulk and surface band structures.

Main Results:

• SOAC DFTB parameters enable efficient prediction of electronic band structures for all d-block transition metals.
• These parameters allow for the study of large structures and high-throughput screening.
• The method preserves fundamental ab initio electronic structure insights.

Conclusions:

• SOAC DFTB parametrizations offer a computationally efficient approach to study electronic properties of transition metal catalysts.
• This method bridges the gap between high-accuracy first-principles methods and computationally feasible large-scale simulations.
• Enables deeper understanding of catalytic reactivity through preserved electronic structure information.