Single-walled carbon nanotubes (SWCNTs) are nanomaterials widely used in industrial and biomedical applications. Computational studies have been actively applied for the last couple of decades to investigate physical, chemical, and mechanical properties of carbon nanotubes. Here we report for the first time a data-driven nano-QSAR study for a series of 85 SWCNTs with sp³-hybridized defects, to investigate various contributing factors and predict photoluminiscence properties of SWCNTs. Two machine learning (ML) models, namely Quantitative Structure-Property Relationships (QSPR) models were developed, to predict photoluminescence wavelength (nm) and oscillator strength of investigated SWCNTs. Molecular structures of SWCNTs were used as input information, on several levels, where encoded information was convalent functional groups, as well as the part of the SWCNT of a certain size. The obtained ML models showed good performances for the Emission Wavelength (nm) in the training set, R²=0.98 and validation set, R²=0.96. The model for Oscillator Strength values was also developed, where the determination coefficients were R²=0.92 and R²=0.87 for training and validation set, respectively. The QSPR models were internally validated by 10-fold cross-validation and y-scrambling experiments, confirming that the models have high levels of robustness and predictivity. The proposed approach provides an important tool to predict the photoluminescence properties of SWCNTs, considering the part of the structure, related to the defect location, ligand and type of the nanotube, avoiding lengthy and complex calculations, assisting in design of SWCNTs with improved photoluminescence properties.

Amirreza Daghighi

Appearances