Presentation Information

[9p-B31-2]Understanding Carbon Material Formation Mechanisms through Reactive Species Diagnostics and Machine Learning

〇Hiroki Kondo1, Yusuke Ando2, Naoki Ueoka2, Kenji Ishikawa3, Takayoshi Tsutsumi3, Masaru Hori3, Mineo Hiramatsu4, Yutaka Matsuo2 (1.Kyushu Univ., 2.Nagoya Univ., 3.Nagoya Univ., Plasma, 4.Meijo Univ.)

Keywords:

Carbon nanomaterials,Plasma-enhanced chemical vapor deposition,Machine learning

Plasma-assisted processes are widely used for the synthesis and modification of carbon materials. Reactive species generated in plasmas play essential roles in surface reactions, structure formation, and material functionalities. However, the coexistence of numerous reactive species makes it difficult to quantitatively understand their individual contributions and synergistic effects. In this study, reactive-species diagnostics using optical emission spectroscopy (OES) and quadrupole mass spectrometry (QMS) were combined with explainable machine learning to investigate carbon material formation mechanisms. Hydrogenated amorphous carbon (a-C:H) films deposited by plasma-enhanced chemical vapor deposition were used as a model system. SHAP-based contribution analysis quantitatively evaluated the effects of dissociated and polymerized hydrocarbon species on film properties. The results revealed that carbon-rich unsaturated species contribute to improved etch resistance through increased carbon density, whereas hydrogen-rich species promote surface hydrogen termination and reduce etch resistance. The proposed approach provides mechanistic insights into the relationships among reactive species, surface reactions, structure formation, and material properties, and is applicable to a wide range of carbon materials and nanostructures.