290749 Pattern Recognition and Quantification of Chemical Agent Simulant in Bottles Using 488 Nm Probe Optic Fiber Raman Spectroscopy

Monday, October 29, 2012
Hall B (Convention Center )
Amanda Figueroa1, Nataly Galan1, Yanh Pacheco1, William Ortiz2, Leonardo Pacheco3 and Samuel Hernandez4, (1)Chemistry, Center for Chemical Sensors and Development-UPRM, Mayaguez, PR, (2)Chemistry, Center for Chemical Sensors and Development, Mayaguez, PR, (3)Center for Chemical Sensors and Development-UPRM, Mayaguez, PR, (4)Chemistry, Center for Chemical Censors and Development-UPRM, Mayaguez, PR

Chemical Warfare agents have been used since World War I and employed by terrorists today to disable their targets. Multivariable analysis techniques such as PLS-DA and Neural Networks were compared to discriminate commercial products with and without the chemical warfare simulant. Neural Networks predominate over PLSDA when using large amounts of variant data.  The potential use of chemometrics techniques applied to experimental data can improve predictions and discrimination of chemical warfare agents concealed in commercial beverage products. Raman Spectroscopy experiments were employed at a 488 nm excitation wavelength for detection of a Chemical Warfare Agent Simulant (CWAS): triethyl phosphate (TEP) inside various commercial bottles: green-plastic, green-glass, clear-plastic, clear-glass, amber-glass and white plastic. Aqueous solutions were also used to perform experiments of discrimination on various bottle materials in commercial beverage products. A custom built experimental setup with optical fibers (OF) was used in this study. Mixtures of TEP simulant with: water, soda, milk, beer and juice were analyzed. Depending on the material, values for Limits of Detection (LOD) were reported for bottle material at less than 1% and Limits of Quantification (LOQ) were as low as 3%. True Positive (sensitivity) values of 1 and False Positive values of 0 for some commercial beverage mixtures were also reported in Partial Least Squares Discriminant Analysis (PLS-DA).  Calibration models included PLS regression to determine Limits of Detection and Limits of Quantification (LOQ). In fact, Neural Networks accurately predicted over 90% of all testing data for various commercial bottle materials. Normalization of integration times for different bottle materials helped lower Limits of Detection and Quantification for aqueous and commercial beverage solutions. Applications of these models could be utilized in airport, military and first responder equipment to enhance the critical data collected.

Extended Abstract: File Not Uploaded
See more of this Session: Student Poster Session: Food, Pharmaceutical, and Biotechnology
See more of this Group/Topical: Student Poster Sessions