Use of a portable near- infrared transmittance spectrometer (NIT-38, NIR Technology, Australia) was evaluated for the authentication of tablets. Operating in the third overtone region (720–1090 nm, 10 nm wavelength increment), the device was used to analyse tablets in transmission mode using a custom-made cell. Authentic tablets of two proprietary products were studied: cialis (n = 14 tablets, from four batches); levitra (n = nine tablets, from three batches). Counterfeit versions of these two products were also analysed: counterfeit cialis (n = 22 tablets, from eight batches); counterfeit levitra (n = 19 tablets, from several batches). Data were converted from transmittance to apparent absorbance and scatter corrected using standard normal variate (SNV) transform. Classification models were constructed for each product from principal component scores and used UNEQ for classification. The classification ability of each UNEQ model was verified using the scores of the other authentic and counterfeit tablets. An unsupervised learning method (Kohonen self-organising map (SOM) ) was also studied to further assess the utility of transmission measurements in the third overtone region for identification. Compressed wafers of the active pharmaceutical ingredients (API) and major excipients were prepared and analysed in transmission. These data and those of the analysed tablets were used to train the SOM. UNEQ classification models were able to correctly identify authentic tablets and differentiate them from counterfeits (p < 0.05). The SOM showed close association of authentic tablets with their respective APIs and revealed the presence of several clusters in the counterfeit samples suggesting several sources of origin. Sufficient chemical information was therefore found to exist in the spectra to enable tablet authentication and cluster counterfeits to reveal their likely number of sources of origin.

Keywords: portable, NIR, spectrometer, tablet, authentication, counterfeit, detection, SNV, UNEQ, principal components analysis, self-organising map