Taşınabilir Raman Spektrometresi ile Patlayıcı Öncülü Olarak Toprakta Amonyum Nitrat Tayini; Kantitatif Analiz için Zemin Düzeltme Yöntemi Uygulaması

Abstract

In simple terms, explosives can be defined as substances that are capable of producing an explosion and produce a large volume of expanding gas in a very short period, creating a strong destructive impact on the target. Explosives have become a main instrument for irregular armed groups due to reasons such as being able to act in small groups, creating fear and panic in the society, and obtaining easy results with minimum risk. Military and commercial explosives are rarely preferred by terrorist organizations due to their high prices and the difficulties in their procurement and also to avoid law enforcement attention and detection methods. Instead of using such explosives, they prefer to manufacture their own explosives (Homemade explosives, HMEs) by combining commercially available chemical materials that have low cost and can be easily obtained from the market. Significant amount of Ammonium nitrate (AN) is frequently used by terrorists as an explosive precursor in HMEs preparation. During the manufacturing of AN-based HME, small amounts of AN is expected to spill on the ground, therefore soil-chemical mixture samples taken from the scene are sent to the laboratories, in order to identify possible AN presence as a result of the analyzes made there. This is a time- consuming process for the security forces, delays early detection of suspects and may prevent timely measures against possible AN-based IED attacks.

In this study, it was aimed to achieve real-time and on site determination of AN in soil mixtures with portable Raman spectrometer. For this purpose, portable Raman spectroscopy, samples of Van top soil, pure AN and mixtures of AN and top soil in various concentrations were used in the experimental phase of the thesis. 19 AN-soil mixtures (5% to 95% AN w/w) were prepared and deposited in vials, then 10 Raman measurements were performed for each sample. The obtained data were transferred to the computer. In order to eliminate noises in measured Raman spectrum, firstly, adaptive-degree polynomial filter (ADPF) was applied to the Raman spectral data for noise smoothing. As ADPF does not provide a significant smoothing even at high AN concentrations, Asymmetric Least Squares (ALS) method has been applied to eliminate background noise in the spectra. At this stage, an estimated baseline was generated for each spectrum by applying the ALS method to the raw Raman data, and the obtained average baseline was subtracted from the original spectrum to gain (AN) signals. Finally, an appropriate regression model was constructed to determine the concentration of AN in AN-soil mixtures using the noise-removed data. In order to verify the compatibility of the proposed model for real field conditions, the effectiveness of algorithm was tested with several concentrations of AN-top soil (taken from Van, Hakkari and Adana provinces) mixtures. As a result of this process, prediction success between 79% to 117% was achieved.