Criteria for drug identification by thin layer chromatography and near infrared reflectance spectroscopy

A study was conducted of drug identification by thin layer chromatography using computer search techniques. This investigated the criteria for drug identification as used by pharmacopoeias, and various organisational bodies. The effect of varying standard control conditions, the use of standardising reference standards and running conditions were investigated. Computer based search methods incorporating fixed and moving windows, mean list length and discrepancy index were then used to identify drugs based upon Rf data. A probability of identification and the ability to rank drugs in order of likeliness of identification was produced from the Rf data. Achieving the current pharmacopoeial criteria and the feasibility of setting rigid criteria is discussed. Near Infrared (NIR) spectroscopy is a rapid non-destructive method of analysis which requires little or no sample preparation. The rapid, non-invasive identification of pure drug substances and tablets from a database of over 300 drugs by NIR is described, making use of simple chemometrics. An investigation of sample presentation and physical effects upon NIR spectra is presented. Novel drug profiling studies are described that differentiate between different tablet manufacturers demonstrate identification based upon NIR data. Methods include the analysis of pure drugs and tablets by NIR, the setting up and validation of the database and identification of unknowns against the database. The data were analysed and the methods optimised using a number of chemometric procedures such as second derivatives, correlation spectral matching, wavelength distances, polar co-ordinates, centres of gravity and scanning probability windows. Identification by NIR spectroscopy is demonstrated for unknown pure drugs and tablets with varied success for the type of substance, tablet and chemometric method used. A low number of mis-identifications were reported. A ranking method of identification based upon spectral match is incorporated with optimisation and criteria to be used in the general method