1 Introduction
The field detection of explosives has received considerable attention
globally for many years due to its importance in forensic applications,
defense and security, and environmental control. High explosives (HEs)
have been used for decades in military practice and mining excavation.
These are considered as a contamination source which can generate
poisoning of population humans and animals, and producing genetic
diseases that threaten human health. [1,2] In
order to monitor the quality of HE-contaminated soils, this study
focuses on the detection of HE in soil using mid-infrared (MIR) laser
spectroscopy with a Quantum Cascade Laser (QCL) source as a remote
method of analysis. [3-13] QCL spectroscopy first
demonstrated in 1994 by Faist et al. [14]offers several benefits over conventional or thermal source MIR
spectroscopy, such as room temperature operation, small beam sizes, long
lifetimes, low energy consumption, long-term power stability, and
fine-tuning of the output frequency. [15] Several
groups have demonstrated the capability of remote sensing of HE and
others analytes using MIR laser spectroscopy with QCLs.[14, 16-27] The typical methods used to detect HE
is destructive, require sampling, transferring samples to the lab, and
performing a proper treatment of the sample for later detection. These
methods include protocols based on gas chromatography-mass spectroscopy
(GC-MS), gas chromatography-chemiluminescence (GC-CL), ion mobility
spectrometry (IMS), [28] immunosensors,[29] electrophoresis, [30]fluorescence, [31] electrochemical methods,[30,32] high-pressure liquid chromatography
(HPLC), [33,34] and HPLC-MS.[33] However, in situ detection of HE in
the soil is not easy due to the presence of solid interfering materials
such as organic and inorganic compounds, which vary for each type of
soil [35-37] which makes this detection a
challenge for the analyst. Other studies conducted by this research
group involved the characterization [38-40]interactions, [41-43] and detection[44-45] of HE in soil using Raman and FT-IR
spectroscopy. [46,47] In all of these cases, the
detection was marginally possible because the crystalline particles of
HE had to be found in the solid matrix by microscopy to achieve the
detection.
Today, AI methods are becoming more popular because they have
demonstrated to be a rapidly evolving research area that offers
sophisticated and advanced approaches capable of addressing complicated
and challenging problems. Besides, AI-based systems have a variety of
applications in different sectors, such as engineering, economics,
medicine, military, marine sciences, and others.[48] Therefore, AI allows the transfer of human
knowledge to machines through analytical models and learning from the
data. This is a task that can be accomplished by soft-computing
methodologies. [49]
AI uses minimum information (spectra of neat explosives and clean soil)
for the development of the Machine Learning (ML) models without the
necessity of experimental data of the mixes. AI: Self-Simulated Learning
Artificial Intelligence (SSLAI) models were tested with real spectra of
experimental mixes of HE/soils. SSLAI models do not need to be trained
with real contaminated soil samples or real mixes of HE/soils. The model
would only have to be fed with spectra of neat HE and soils for the
model to train itself. This provides the possibility of HE detection in
field applications with the advantages that the natural solid matrices
could be unknown.