Multivariate calibration for carbon nanotubes in the environment using the microwave induced heating method

Abstract

With continuous increase in production and application of various types of carbon nanotubes (CNTs), the presence of a mixture of different types of CNTs in the environment cannot be neglected. Since distinct types of CNTs have different influences on environmental and human health, the environmental quantification of each type of CNTs has become a critical step in all CNTs-related studies. However, most of existing quantification methods have not been implemented to determine mass/concentration of individual type of CNTs in multi-component samples. Therefore, the goal of the present paper is to develop Chemometrics-based multivariate calibration approaches for simultaneously quantifying individual type of CNTs in the environment with the microwave induced temperature rise spectra. Motivated by successful applications of partial least square regression (PLS), least square-support vector machine (LS-SVM) and artificial neural networks (ANN) in measuring specific contaminants in mixtures, the potential of applying these techniques in predicting quantities of Single-walled CNTs, Multi-walled CNTs and carboxylate Multi-walled CNTs in environmental matrices (agricultural soil and anaerobic sludge) was investigated in this study. Our results revealed that the developed LS-SVM model presented high R2 and low root mean square error of prediction (RMSEP) in both 2-component and 3-component matrices, while the resulted ANN model was only accurate in the 2-component matrix. The PLS model was found to be ineffective in interpreting relationship between the microwave induced temperature rises and mass of CNTs as indicated by small R2 and large RMSEP.