Vibrational characterization and machine learning-based classification of defects in TWCNTs

Carbon nanotubes (CNTs) have remarkable electrical, mechanical, and thermal capabilities, defects like pinholes and point vacancies are common in their lattice structures. Although there is no standardized technique for detecting these flaws, they can have a substantial impact on their characteristics. It is less to known how these issues impact the vibration characteristics of three-walled carbon nanotubes (TWCNTs). The study presents a new way to analyze vibrations for finding defects in TWCNTs. Researchers used molecular structural mechanics model to simulate 648 TWCNT samples and identified their vibration patterns through finite element (FE) analysis. The basic natural frequency drops as the diameter and point vacancies increase, according to the results. In chiral TWCNTs, this trend starts after 10–19 gm mass, but in armchair and zigzag TWCNTs, the frequency increases rapidly after 10–20 gm mass. Armchair and zigzag configurations had similar frequencies to chiral TWCNTs, whether there were 0% or 0.5% vacancy defects. The following techniques were used to classify defects: Logistic regression (LR), Decision Tree Classifier (DT), K-Nearest Neighbors (K-NN), Support Vector Machine (SVM), and Naïve Bayes Classifier. SVM successfully managed complicated boundaries, Naïve Bayes operated on probability, KNN identified based on closeness, DTs caught non-linear patterns, and LR provided interpretability. The possibility of integrating vibrational analysis and machine learning (ML) for accurate TWCNT fault diagnosis was demonstrated by a DT classifier that obtained 100% test accuracy.