Elemental analysis of neutron-induced gamma-ray spectra is a significant technology in the detection of chemical agents, explosives, etc. The hard part of this problem is the very complicated and uncertain background signals of the gamma-ray spectra. Also, the background signals are always changing as the searched objects change, thus further complicating the gamma-ray spectra analysis process. We can define a typical or average background spectrum if the variation of background spectrum is not too large, then we use this background spectrum to identify a gamma-ray signal.

We tested both the direct summation approach and the Gaussian fitting approach in our computer algorithms. We found these two different approaches have individual advantages and disadvantages when they are applied in calculations of the signal significance level. In the end, we combined the direct summation approach with the Gaussian fitting approach in our computer algorithms in the actual calculations of the signal significance level.

Based on our previous preliminary MCNP simulations results, we used phosphorus powder to simulate the chemical agent sarin and used our automated computer algorithms to calculate the single-line and multiple-line signal significance levels. Presented in this paper are some results in which we used our experimental data to test our average background spectrum and our computer algorithms. Our calculated results show the average background spectra that we defined are appropriate for the elemental analysis in searching for the chemical agent, and our computer algorithms, in which we combined the direct summation approach with the Gaussian fitting approach, function well. From a multiple-line analysis, our calculated results show that in the real application of this neutron-induced gamma-ray detection technique the detection time can be reduced to 15 s or less for detecting small quantities of chemical agents.