A fast and reliable 2d-ir spectroscopic technique for herbal leaves classification

Abstract

The safety and quality of herbal products are of great importance. People take herbal products with the intention for health improvement and overall well-being. Both chromatographic and spectroscopic techniques have been proven for their effectiveness in phytochemical fingerprinting and profiling. The present study describes the application of advanced two dimensional correlation infrared spectroscopy (2D-IR), and followed by chemometrics for herbal leaves classification. The 2D-IR could ascertain small difference and reveal for any overlapping peaks compared to conventional FTIR technique. There were 13 herbs selected and their dried leaves in fine powder form were used for spectral scanning from 400 to 4000 cm−1. The absorption signals were analysed by principal component analysis (PCA) and hierarchical clustering analysis (HCA). It was found that the wavenumber of 1800–900 cm−1 was the dominant region of identification and discrimination. The region was found to have signals attributed to the functional groups of C[sbnd]O, C[dbnd]O and C[dbnd]C stretching, and C[sbnd]H and O[sbnd]H bending. Both PCA and HCA could cluster the IR spectral data of the herbs into 4 distinct groups. There was no significant relationship of leaf morphology and phytoconstituents similarity. Somehow, hairy (Strobilanthes crispa and Morus alba) and waxy (Eurycoma longifolia and Citrus hystrix) leaf surfaces of the herbs are still grouped together after data extraction and reduction using the pattern recognition tools. The classification was also verified by introducing another 3 new spectral data (S. crispa, M. alba and Ficus deltoidea) into the statistical analysis. The new dataset was found to be located at the position near to their individual plant species in both score plot and dendrogram. Although the absorption signals of the new dataset were obtained from their aqueous crude extract, 2D-IR spectroscopic analysis was still capable to capture the key features of discrimination for classification in a precise and rapid manner.