An integrated Raman spectroscopy and self-organizing feature map chemometrics analysis for the discrimination of gel inks

• Main Author: Asri, Muhammad Naeim Mohamad
• Format: Thesis
• Language: English
• Published: 2022
• Subjects: QD71-142 Analytical chemistry
• Online Access: http://eprints.usm.my/57339/1/MUHAMMAD%20NAEIM%20BIN%20MOHAMAD%20ASRI-FINAL%20THESIS%20P-SKD000618%28R%29%20-24%20pages.pdf; http://eprints.usm.my/57339/

One of the techniques often used by forensic document examiner in the analysis of questioned documents is Raman spectroscopy. This technique has proven effective in rapidly and accurately characterising ink with the advantage of not causing permanent damage to the analysed document. Gel ink pen was first introduced to the market around 1990s by the Sakura Colour Product Corporation, Japan as an alternative daily writing instrument other than ballpoint-pen. Being marketed in varieties of attractive colours and also environmentally friendly, gel ink pen has become very popular among consumers therefore it appearances in cases involving document fraud is anticipated. Unfortunately, studies involving gel ink pen in the perspective of questioned document are limited compared to ball-point pen. In forensic investigation, it was reported that 80% handwritten documents were written using ballpoint pens. Therefore, due to the lack of objectivity in interpreting gel inks, this study proposed using novel chemometric techniques for discriminating gel-pen inks in forensic investigation. This study show the comprehensive conceptual work flow with conventional chemometrics techniques of Principal Component Analysis (PCA), Hierarchical Cluster Analysis (HCA) and novel Self-Organising Feature Maps (SOFM) in tandem with Raman spectroscopy to discriminate gel inks of three different colours i.e. blue, red and black before and after undergoing ageing process for nearly three years. Preliminary dissolution studies performed by dissolving gel inks in a varieties of organic solvents show that gel inks of pigment based colourants do not normally dissolve in organic solvents while Thin Layer Chromatography (TLC) is only successful to dye-based colourant gel inks. Queries made to the acquired blue, red and black gel inks Raman spectra against a database containing Raman spectra of 200 pigments, suggested that CI Pigment Blue 15: 1 and 15: 3, could be the main pigment in blue gel inks conversely Red DPP BO (23180 – CI Pigment Red) and XSL Poppy Red (26308 – CI Pigment 112) could be the main pigment in red gel inks however main pigments in black gel inks remain unidentified. In most cases, Hierarchical Cluster Analyses (HCA) have recorded lesser number of clusterings compared to Principal Component Analyses (PCA). In general, both conventional chemometrics techniques are unable to discriminate the gel inks according to their brands however this is not the case for the novel SOFM. All SOFM maps are in line with the classification made by the visual Raman comparison of inks. The misclassified sample (HCA and PCA) was successfully resolved using the SOFM model signifying its capability for both discrimination and classification purposes. Cross validations employing the k-validation strategy have recorded 100% correct classification rates for all the SOFM models, thus signifies the robustness and potential of SOFM for discrimination of gel inks. As for the aged gel inks, SOFM has also successfully sourced the aged inks to their fresh counterparts. This study reported the first use of SOFM for discrimination and classification of gel inks. The findings of this study did not only communicate the comprehensive work flow for gel ink analysis but most importantly signify the potential of SOFM to be employed in tandem with Raman spectroscopy as an alternative pattern recognition technique to the conventional chemometrics techniques that can be highly useful in document fraud investigation involving gel inks.