Exploration of Weak Peak Detection Method Based on Gamma Spectra Deconvolution

To improve the analytical performance of low resolution gamma spectrometers, a new method based on spectrum deconvolution to detect weak peaks was put forward. First of all, with SuperMC Monte Carlo simulation code, a geometric model matching the measured conditions was established, and then the system response matrices of a NaI(Tl) spectrometer was constructed. Secondly, in order to ensure the accuracy of the gamma spectrum deconvolution, the nonlinear least square spectrum stabilization technique was applied to stabilize the measured spectra. Thirdly, the long time measured background spectrum was used for background deduction. And then the L-R deconvolution algorithm based on the system response matrices was used to improve the saliency of weak peaks. Finally, the symmetric zero-area transformation method was used to detect the weak peaks. To verify the feasibility and applicability of the method, the gamma spectra of weak source ¹³⁷Cs disturbed by the Compton scattering continuum of high activity ⁶⁰Co were simulated, and the gamma spectra of weak ¹³⁷Cs and ¹⁵²Eu sources disturbed by the Compton scattering continuum of high activity ⁶⁰Co were measured, as well as the gamma spectra of weak source ¹³⁷Cs. The simulated and measured results show that the significance of the weak peaks of ¹³⁷Cs and ¹⁵²Eu is markedly improved after deconvolution, and the peak-finding success rate increases greatly. Specifically, the peak-finding success rate of the weak peak ¹³⁷Cs increases from 25.9% to 42.0%, 73.4% to 90.8%, and 61.4 % to 75.2% when it is overwhelmed by the environmental background, the Compton scattering of ⁶⁰Co and the statistical fluctuation, respectively. The peak-finding success rate of all weak peaks of ¹⁵²Eu disturbed by the Compton scattering continuum of different degrees, are improved to varying degrees, with an increase of 6.0% to 97.0%. Especially, the peak-finding success rate of 295.9 keV weak peak of ¹⁵²Eu is significantly elevated from 3.0% to 100%, as well as 1 528.1 keV weak peak of ¹⁵²Eu from 6.1% to 97.0%. Therefore, it is proved that the method based on spectrum deconvolution to detect weak peaks can effectively improve the weak peaks detection capability of low resolution spectrometers. The performance of the method is restricted by the error of the system response matrices, which is attributed to the mismatch between the Monte Carlo geometric model and the measured conditions, so the method is mainly suitable for the fixed measurement in the laboratory. Furthermore, this method is not only applicable to NaI(Tl) measurement system, but also to other spectrometers such as CsI(Tl), LaBr₃(Ce), CdZnTe and HPGe, and is also suitable for application scenarios with shielding measures and anti-coincidence techniques. But, the system response matrices matching the measurement system needs to be established and applied.