The combined LIBS and ICP HRMS analysis of 13 tea samples are studied in view of identification of tea geographical origin. The elemental signature provided by LIBS spectra is treated by principal component analysis followed by partial least square discriminant analysis and factorial discriminant analysis. Selected element lines are found efficient to discriminate most sample groups. Data analysis model is improved by variable selection and the isotopic ratio 11B/10B was employed to improve the prediction capacity of the model. The alkaline earth: Ba, Ca, Mg, Sr and alkaline Rb, Na are easily detected by the LIBS system and these elements are important to classify sample according to their geographical origin. Minor elements like P, S, Fe, B … also bring discriminant information. A five clusters model gave best correct identification in a cross validation test (94.2%). This method also allowed to identify the origin of four unknown teas. In this study the use of FDA or PLS DA after the PCA examination of the LIBS/ICP MS data led to similar conclusions for fast classification of the tea samples and identification of the geographical origin of the four unknown teas.
Keywords: ICP HRMS; LIBS; Multi-spectral analysis; PCA; Tea samples.
Copyright © 2020 Elsevier B.V. All rights reserved.