Investigation of a time-effective approach to the discrimination and interpretation of inkjet printer inks using micro Raman spectroscopy

As submitted by the proposer: The Department of Forensic Science at Sam Houston State University proposes a collaborative research endeavor with the Criminal Investigative Division, Treasury Obligations Section of the United States Secret Service. Our goal is to investigate if Raman data gathered from the three microscopic colored spots of inkjet printed documents constitute, all together, a chemical signature sufficiently discriminating to provide reliable investigative leads in a time-effective and non-destructive manner. In a forensic setting, spectroscopic data are typically obtained to evaluate similarities between questioned specimens and reference samples to address questions about source attributions. Data produced in this project are considered instead to evaluate their suitability to produce investigative leads in cases where a suspected printer needs to be developed (i.e., sourcing). This work focuses on inkjet printer inks because more than 60% of all counterfeit banknotes classified by the USSS are produced using this technology. Given their relatively low cost and their wide occurrence, however, their use is not exclusively encountered in problems of counterfeit, but they can be submitted to questioned document examiners in a variety of cases, such as extortions, questioned contracts, ID documents, or anonymous letters. Although Raman spectroscopy is already a well-established method for the characterization of colorants (both dyes and pigments), its potential to address meaningful investigative questions has not been fully explored yet. Typically, inkjet printer produce microscopic spots that allow obtaining a chemical signature from the three main colored components (cyan,magenta, and yellow) and a considerable number of sample. A sensible, objective, and fast classifer is then required for conducting spectral comparisons. Various multivariate statistical methods will be tested. These include the traditional methods such as principal components analysis or linear discriminant analysis, but also statistical techniques that are less explored in forensic disciplines known as ensemble methods (i.e., bagging, boosting and random forests) and functional data analysis. During this research it is expected to develop a statistical approach that can reliably and objectively differentiate between samples from different sources based on minor spectral differences and to evaluate if different color spots from the same source can be treated as independent pieces of information. This study will also compare to the traditional method of thin layer chromatography to inform questioned document examiners about the complementarities between the two approaches. Note: This project contains a research and/or development component, as defined in applicable law. ca/ncf