Related Experiment Video
Updated: Aug 6, 2026
03:54
Aqueous Synthesis of Plasmonic Gold-Tin Alloy Nanoparticles
Published on: March 15, 2024
Surface-structure search with variable composition and periodicity via machine learning and evolutionary algorithms:
1Materials DX Research Center, National Institute of Advanced Industrial Science and Technology (AIST), Tsukuba, Japan.
Science and Technology of Advanced Materials
|July 23, 2026
Summary
This study introduces a new framework for predicting stable surface structures by treating periodicity and composition as dynamic. This approach accurately identifies complex surface configurations, accelerating materials discovery for catalysis and electronics.
Area of Science:
- Materials Science
- Computational Chemistry
- Surface Science
Background:
- First-principles structure prediction is crucial for functional materials discovery.
- Surface structure searches are hindered by fixed in-plane periodicity and composition assumptions.
Purpose of the Study:
- To develop a global search framework for identifying stable surface structures.
- To enable dynamic exploration of superlattice periodicity and stoichiometry.
- To overcome limitations of existing surface structure search algorithms.
Main Methods:
- Integration of an evolutionary algorithm with surface-specific variation operators.
- Symmetry-enriched initialization for enhanced search efficiency.
- Bayesian optimization utilizing atomic cluster expansion descriptors for accelerated screening.
Main Results:
- Successfully identified oxygen-induced surface structures on FCC Pt(111) and diamond Ge(100) surfaces, matching experimental data.
- Discovered Sn alloying motifs on FCC Au(111) consistent with known surface-structure trends.
- Achieved accurate structure prediction with significantly reduced first-principles evaluations.
Conclusions:
- The developed framework provides a general route for exploring complex materials landscapes.
- It enables the direct identification of stable surface structures across varying supercell shapes and compositions.
- This approach is applicable to fields like heterogeneous catalysis, electronics, and spintronics where coupled structural and compositional factors are key.
