Dynamics of growing carbon nanotube interfaces probed by machine learning-enabled molecular simulations

Area of Science:

• Materials Science
• Nanotechnology
• Computational Chemistry

Background:

• Carbon nanotubes (CNTs) are promising materials for advanced technologies.
• Uniform CNT structure is crucial for performance, but defects form during high-temperature growth.
• Atomic-level understanding of defect formation and healing at the tube-catalyst interface is lacking.

Purpose of the Study:

• To unveil the atomic-level mechanisms of CNT formation, including nucleation, growth, and defect dynamics.
• To investigate the role of the tube-catalyst interface in CNT structural integrity.
• To explore conditions enabling defect-free CNT growth.

Main Methods:

• Development and application of DeepCNT-22, a machine learning force field (MLFF).
• Driving molecular dynamics simulations to observe CNT growth processes.
• Analyzing the dynamic behavior and configurational entropy of the CNT-edge at the tube-catalyst interface.

Main Results:

• The tube-catalyst interface is highly dynamic, with significant fluctuations in CNT-edge chirality.
• Continuous spiral growth is not the general mechanism; significant configurational entropy exists at the growing edge.
• Defects form stochastically but heal effectively at low growth rates and high temperatures, enabling defect-free growth.

Conclusions:

• MLFF-driven simulations provide unprecedented insights into CNT growth mechanisms.
• Defect healing at the tube-catalyst interface is key to producing long, defect-free CNTs.
• The study challenges existing models and fills critical knowledge gaps in CNT synthesis.