| Abstract: | This study aims to utilize the manufacturing impurities found in illicit heroin to predict whether the samples had originated from a common production batch. From highly cut heroin samples, twelve frequently extractable neutral/acidic impurities were selected for investigation. A gas chromatographic method was optimized using three locally seized samples containing the target impurities and a control sample consisting of n-alkanes. The method optimization procedures are reported in detail. The capability of the optimized method was further assessed using five simulated links prepared from unrelated heroin seizures. With the aid of principal component analysis (PCA), it was found that the statistical data pretreatment (normalization followed by standardization, N + S) showed promising results for sample classification using profiles obtained with the optimized method. Similarly, 252 street doses of highly cut samples were analyzed and the profiles were statistically decomposed by PCA after the N + S pretreatment. The statistics suggested that the locally seized samples could have originated from more than one production batch. |
