Experiment-Driven Atomistic Materials Modeling: A Case Study Combining X-Ray Photoelectron Spectroscopy and Machine...

Area of Science:

• Computational Materials Science
• Atomistic Modeling
• Machine Learning in Materials

Background:

• Reconciling atomistic simulations with experimental data is a significant challenge in materials science.
• Traditional methods rely on extensive structure optimization, which is inefficient and not always successful.
• Understanding the limits of oxygen incorporation in amorphous carbon (a-CO) is crucial for its applications.

Purpose of the Study:

• To introduce a general method combining atomistic machine learning (ML) with experimental observables for 'design-by-experiment' materials modeling.
• To determine the maximum oxygen content in amorphous carbon (a-CO) before decomposition into CO and CO₂.
• To elucidate the atomistic structure of a-CO using a combination of computational and experimental data.

Main Methods:

• Developed a novel approach integrating atomistic ML with experimental observables.
• Employed grand-canonical Monte Carlo with a modified Hamiltonian formalism for generating chemically sound configurations.
• Utilized an ML-based X-ray photoelectron spectroscopy (XPS) model trained on GW and DFT data, coupled with an ML interatomic potential.

Main Results:

• Identified a-CO structures consistent with experimental XPS predictions and energetically favorable according to DFT.
• Accurately deconvolved XPS spectra into motif contributions using network analysis, revealing experimental interpretation inaccuracies.
• Provided atomistic insights into the structure of a-CO, demonstrating the method's capability for experiment-driven modeling.

Conclusions:

• The proposed ML-driven method enables the generation of atomistic structures compatible with experimental data by design.
• This approach allows for the direct elucidation of material structures from experimental observables with high realism.
• The study provides critical insights into the structural limits and stability of oxygenated amorphous carbon (a-CO).