Plasma Oxidation of Copper: Molecular Dynamics Study with Neural Network Potentials

Area of Science:

• Materials Science
• Plasma Physics
• Surface Chemistry

Background:

• Thin oxide film formation is critical for semiconductor manufacturing.
• Current plasma etching processes for copper patterning lack precise thickness control.
• Developing self-limiting oxidation is key for advanced semiconductor fabrication.

Purpose of the Study:

• To develop a validated neural network potential for copper-oxygen interactions.
• To investigate plasma oxidation of copper surfaces using large-scale molecular dynamics.
• To understand the mechanisms of oxide growth and identify strategies for self-limiting oxidation.

Main Methods:

• Developed and validated a neural network potential for Cu-O interactions against first-principles calculations.
• Employed large-scale molecular dynamics simulations at atomic resolution to model plasma oxidation.
• Analyzed ion impact effects, local heating, and oxide layer formation dynamics.

Main Results:

• An amorphous copper oxide (CuO) layer up to 2.5 nm thick was formed.
• Plasma induces local heating, with ion energy influencing the heating range and oxide growth rate.
• A quasicrystalline oxide layer forms beneath the amorphous layer, slowing growth; an activation energy of 1.0 eV was determined.

Conclusions:

• Plasma oxidation of copper is influenced by ion energy and local heating effects.
• A strategy involving lower substrate temperature and increased plasma power is proposed for self-limited oxidation.
• The findings offer insights for optimizing plasma etching processes in the semiconductor industry.