Machine learning-assisted Ru-N bond regulation for ammonia synthesis.

Area of Science:

• Materials Science
• Catalysis
• Computational Chemistry

Background:

• Ruthenium intermetallic compounds (Ru-IMCs) are promising for ammonia synthesis catalysis.
• Challenges exist in designing Ru-IMC catalysts due to complex phase behavior and numerous parameters.

Purpose of the Study:

• To leverage machine learning (ML) and model mining for efficient Ru-IMC catalyst design.
• To identify key descriptors for optimizing ammonia synthesis activity.

Main Methods:

• Combined machine learning and model mining techniques.
• Generated a 2D activity volcano plot based on N2 and N adsorption energies.
• Employed Density Functional Theory (DFT) calculations, N2-Temperature Programmed Desorption (TPD), and Fourier-Transform Infrared Spectroscopy (FT-IR).

Main Results:

• Identified the coordinating atom's radius to Ruthenium as a critical parameter for catalyst performance.
• Developed a highly active catalyst, Sc1/8Nd7/8Ru2, achieving 8.18 mmol m-2 h-1 at 0.1 MPa and 400 °C.
• Revealed that Ru-N interaction is governed by d-p orbital hybridization.

Conclusions:

• Machine learning significantly aids in the exploration and design of intermetallic compounds for ammonia synthesis.
• The findings highlight the potential of ML-guided materials discovery for catalytic applications.