Louise Fortunato, Sulaf Assi, Paul Kneller and David Osselton Identifying Counterfeit Tobacco Using Portable Fourier Transform Infrared Spectroscopy Louise Fortunato, Sulaf Assi, Paul Kneller and David Osselton Department of Archaeology, Anthropology and Forensic Science, Bournemouth University, UK. BH12 5BB Objective To identify counterfeit tobacco using portable Fourier transform infrared spectroscopy and chemometrics. Results and Discussion Hit quality index (HQI) The threshold used for HQI was 95%. Figure 1 shows the MSC treated FTIR spectra of authentic and counterfeit tobacco products. Both the aforementioned spectra had major peaks at 3296, 2914, 2847, 1582, 1380, 1315, 1093, 1025 and 1000, which respectively correspond to following groups: OH, CH3, CH3CO, C=C, CH2CH2CH3, NO2, CH2, OC(CH3)3 and C(CH3)3 (Socrates 2001). The difference in the spectra was mainly attributed to the intensity of absorption; which was lower for counterfeit tobacco (0.103-0.295 au) than the authentic tobacco (0.186-0.463 au). Consequently, HQI method was not able to identify counterfeit tobacco products which gave mismatches for authentic tobacco product. Principle Component Analysis (PCA) Likewise, PCA showed to be an effective method in identifying counterfeit tobacco. PCA extracts the important information and expresses it in orthogonal values called principal components (PC), it also explains the variation in a set of variables (Dunteman 1989). In this respect, the first three PC scores accounted for 93.44% of the variance among the data which corresponded mainly to (3357; 2926; 2890; 1621; 1395; 1252; 1025 cm-1). When the scores were plotted distinct clusters were observed for the authentic and counterfeit tobacco products (Figure 4). Introduction Counterfeit tobacco is a growing problem in the UK, thus new and rapid techniques need to be developed for their quick identification (Assi 2014). Fourier Transform Infrared (FTIR) is one of the most used techniques for the identification of chemical constituents and the understanding of chemical structures. Moreover the fingerprint characters provide a unique characterisation to chemical compounds (Griffiths 2007). Experimental Materials: A total of 34 tobacco products (27 authentic and seven counterfeit) (Table 1) were measured using the Bruker ALPHA FTIR spectrometer equipped with Platinum attenuated total reflectance (ATR) diamond. The nicotine content of the authentic and counterfeit tobacco products were in the range of 0.3-1.27 and 0.9-1.27 mg respectively. Method: Cigarette content for each batch was and stored in 4 mm glass vials. Then, three subsamples (1 – 2 mg each) were taken from each vial and placed in direct contact with the ATR accessory for spectral collection. Each spectrum was the average of 16 scans (resolution of 4 cm-1) over the wavenumber range 400 to 4000 cm-1. Figure 2: MSC treated FTIR spectra of authentic (blue) and counterfeit (red) tobacco products. K-mean clustering However, K-mean clustering was more successful in identifying counterfeit tobacco products. Hence, the coefficient observed for all authentic products was 1.2; whereas, the counterfeit had coefficient above 1.4 (Figure 2). Spectral treatment: Raw spectra of authentic and counterfeit tobacco were exported into Unscrambler 9.2 where multiple signal correction (MSC), hit quality index (HQI), K-mean clustering and principal component analysis (PCA) were applied (Esbensen 2002). Figure 4: PCA of authentic (blue) and counterfeit (red) batches using the FTIR spectra with MSC pre-treatment. Conclusion FTIR offered a rapid and accurate method for identifying counterfeit tobacco products. K-mean clustering and PCA showed to be more accurate in identification than HQI. References Assi, S., Kneller, P., Osselton, D., and Moorey, P., 2014. Identifying Organic Impurities in Counterfeit and Illicit Tobacco Using Portable ATR-FT-IR Spectroscopy. Raman Technology for Today's Spectroscopists [online]. 29 (8), 40-50. Dunteman, G, 1989. Principal Component Analysis [online]. London: SAGE. Esbensen, K.H., et al, 2002. Multivariate Data Analysis: In Practice: an Introduction to Multivariate Data Analysis and Experimental Design [online]. 5th edition. Camo Process AS. Griffiths, P., 2007. Fourier Transform Infrared Spectrometry. 2nd edition. New Jersey: Wiley. Socrates, G., 2001. Infrared and Raman Characteristic Group Frequencies. 3rd edition. England: Wiley. Figure 1: The Bruker Alpha FTIR instrument equipped with ATR accessory (left), and the authentic and counterfeit tobacco products (right). Acknowledgements The Trading Standards for supplying the counterfeit tobacco products. Figure 3: K-mean clustering of authentic (A1-A13) and counterfeit (C1-C7) tobacco products.