Growth Mechanisms of Carbon Nanotubes reveled by Machine Learning Force Field-Driven Simulations

Carbon nanotubes (CNTs) hold the potential to revolutionize nanoelectronics by outperforming silicon-based devices.
To realize this technological promise, controlled growth of defect-free CNTs must be achieved through a deep understanding of atomic-scale growth mechanisms.
Traditional molecular dynamics (MD) simulations have been limited by short timescales, hindering progress in this area.
Here we pioneer the use of machine learning force fields to enable efficient and accurate MD simulations of single-walled carbon nanotube (SWCNT) growth on iron catalysts.
Our method facilitates near-microsecond timescale simulations, yielding the growth of long, defect-free nanotubes and delivering unprecedented atomic-level insights into the entire growth process, including the evolution of the tube-catalyst interface and the formation and healing of defects.
Our findings emphasize the maximization of the configurational entropy at the tube-catalyst interface and reveal that ultralong, defect-free CNTs can be synthesized through careful control of carbon supply and temperature.