Waste tobacco leaves derived carbon dots for tetracycline detection: Improving quantitative accuracy with the aid of chemometric model

The recycling and reutilization of biomass wastes are significant for environmental protection and sustainable development. Recently, there have many studies on utilizing biomass wastes to produce carbon dots . Whereas, the spectrum shift effect that occurs in the quantitative application of carbon dots as fluorescent probes limits the accuracy of the quantitative analysis . In this work, waste tobacco leaves were used as the carbon source for synthesizing a novel carbon dots (CDs(WTL)) through a facile hydrothermal method . The CDs(WTL) possess a series of excellent properties, including good water solubility, well stability, and high fluorescence quantum yield . The fluorescent intensity of the CDs(WTL) can be quenched by tetracycline (TC) obviously, but there is a spectrum shift. In order to use the CDs(WTL) as fluorescent probes to quantify TC with higher accuracy, a quantification fluorescence model (QFM) was introduced to overcome this spectrum shift effect that often occurs. The coefficient of determination ( R 2 ) of traditional quantification model (TQ), partial least squares (PLS), and QFM are 0.9672, 0.9834, and 0.9991, respectively; the average relative predictive error (ARPE) of TQ, PLS, and QFM are 8.8%, 4.5%, and 3.9% for the spiked water samples, and 21.9%, 22.0%, and 2.9% for spiked tablet samples, respectively. The obtained results suggest that QFM is more accurate than PLS and TQ for the TC detection. By utilizing QFM, the spike recoveries (mean ± standard deviation) in three kinds of real tablet samples produced by different manufacturers are 98.9 ± 3.6%, 102.5 ± 6.2%, and 98.5 ± 2.7%, respectively; the spike recovery in river water samples is 99.4 ± 5.0%. In addition, high performance liquid chromatography (HPLC) was used as a reference method, the F and t tests suggest that there are no significant differences on the precision and accuracy between QFM and HPLC methods.