Improved differentiation of NPS analogs through the application of chemometric methods to GC-solid deposition-FTIR spectra

Novel psychoactive substances pose a significant analytical challenge for forensic laboratories. Gas Chromatography with Infrared Spectroscopy (GC-IR) is typically presented as producing visually distinct spectra. However, this is not the case for all compounds. In this study, we showcase three synthetic cathinone analogs (methylone, N-ethylpentylone, and pentylone) which produce extremely visually similar solid state IR spectra.
A primary dataset of these analogs was generated at the Amsterdam Police Laboratory. Libraries were created using two different sample preparation methods. The same compounds were also analyzed at the Virginia Department of Forensic Science (DFS). It was observed that changes in either instrument or sample preparation were enough to pose challenges to both the visual assessment and a library matching algorithm. Additionally, week-to-week variation was observed within the primary dataset. Principal Component Analysis (PCA) in combination with mahalanobis distances for objective comparison was assessed. A leave-one-sample-out cross validation accurately identified 100% of the samples in the primary dataset.
This study shows how the application of chemometrics to spectral GC-IR data can substantiate the instrument’s differentiation capabilities and provide valuable objective support for a compound identification. In addition, limitations in the consistency of GC-IR spectra over time and across instruments and sample preparation methods were observed which could affect how the forensic community utilizes these techniques. Specifically, this study shows that shared GC-IR libraries might pose selectivity limitations. Therefore, laboratories must exercise caution if shared/generic GC-IR libraries are utilized for casework, and it is recommended that instrument and solvent specific libraries are generated.