Study On Common Modeling Analytic Technology Of Alpha Particle Spectra For Semiconductor Si Detector

The measurement technology of alpha-particle spectra based on semiconductor Si detectors has become an important radioactive analysis method for measuring alpha-nuclides in environmental,biological,and nuclear technology-related samples.The analysis of alpha-particle spectra is the basis of qualitative and quantitative analysis for alpha-nuclei.Almost every alpha-nucleus has multiple energy alpha-rays,and alpha-particles with similar energy are common between the same or different nucleus.The alpha-particle spectra have the unique low-energy tailing due to the source self-absorption,absorption of the medium,absorption of the detector entrance window?dead layer?,the incomplete collection of electric charge,electronic noise and other factors.Current detectors,even those with the highest energy resolution,cannot completely separate these peaks.Therefore,under the limited energy resolving power of the detector,the accurate analysis of the alpha-particle spectra is particularly important for analyzing the species and radioactivity of alpha-nuclides.Researchers at home and abroad have carried out many years of research on the analysis of alpha-particle spectra,and proposed many alpha-particle detector response function models.Due to the particularity of low-energy tailing of alpha-particle spectra,the research focuses on the description of low-energy tailings.At present,researchings have proved that the alpha-particle detector response function based on the convolution of a Gaussian function with one or more exponential functions is the best.However,since most alpha-nuclides are also accompanied with low-energy converted electrons and X-ray emission during decay,the coincidence of alpha-particles with low-energy electrons and photons is inevitable in the limited response time of the detectors.The energy deposited in the detector is higher than that of the alpha-particle itself.Therefore,high energy tailing is also present in the alpha-particle spectra.This is particularly evident in alpha-nuclides such as ²⁴¹Am and ²⁴³Am,and therefore it is necessary to improve the alpha-particle spectra response function.On the other hand,based on the analysis of the alpha-particle spectra detector response function,a weighted nonlinear least squares fitting method is usually used.The estimation of the initial values of the parameters is the key to successful or accurate fittin.The method is mainly based on empirical values.For a function model with multiple parameters,the manual adjustment is not stable and time-consuming,and the result is often poor.Therefore,the study of initial value acquisition method and optimization is contributed to improve the accuracy and automation of the spectra analysis.In this work,aiming at the above common problems,based on the in-depth study of predecessors'work,a new model and a new method for analyzing alpha-particle spectra measured by semiconductor Si detectors are studied.The main research contents and conclusions are as follows:?1?The formation mechanism of alpha particle spectra and the physical process of semiconductor Si detector response are systematically analyzed.The coincidence of alpha particles and converted electrons or photons is contributed to high energy tailing.A new EMG-Landau model with a high energy tailing alpha-particle detector response function is established based on the multi-exponential function convolution with Gaussian model.The Landau distribution term is increased as a high-energy tailing.The model was tested by IAEA reference alpha-particle spectra,EUROMET-5535reference alpha spectra and transuranic nuclide alpha-particle spectra experiment.The results show that the EMG-Landau model can more fully and accurately describe the alpha-particle spectra,and can effectively and accurately fit alpha-particle spectra with high energy tailing.For the alpha-nuclides with obvious coincidence effects,such as²⁴³Am and ²⁴¹Am,the EMG-Landau model has a better fitting than the EMG model;for the alpha spectra with low resolution and low statistics,the EMG-Landau model is consistent with the EMG model.?2?Aiming at estimation of the initial value for parameters in the weighted non-linear least square fitting of alpha-particle spectra,a method is proposed to obtain the initial value of parameters based on statistical moments,but this method is suitable for the detector response function of Gaussian convoluted with an exponential.This paper also proposes a method for obtaining initial values of parameters based on genetic algorithm with Re?² function as fitness function.The parameter initial value estimation of the multi-parameter alpha-particle spectra detector response function was accurately performed.?3?Applying the iterative deconvolution method to the analysis of alpha particle spectra,and comparing the three most ideal particle spectra iterative deconvolution algorithms:Boosted-Gold,Richardson–Lucy,and MAP;The indicators of accuracy and speed of operation indicate that the Boosted-Gold iterative algorithm is better than the other two algorithms.In the iterative deconvolution process,this paper proposes an alpha-particle Monte Carlo simulation spectrum broadening algorithm based on the alpha-particle detector response function,and then accurately broaden of alpha-particle simulation spectrum.Then the Monte Carlo simulation alpha-particle spectra and spectra interpolation algorithm is gathered to establish the alpha-particle spectra detector response matrix,which lays the foundation for accurate iterative deconvolution analysis of alpha particle spectra.The main innovations of the thesis are the following:?1?A new alpha-particle spectra detector response function model with high energy tailings is proposed that is EMG-Landau model.The model can more accurately describe the alpha-particles spectra.?2?A method for obtaining the initial values of alpha-particle spectra parameters based on statistical moments is proposed.?3?Aiming at the limitation of the application of initial values acquisition method based on statistical moments,a method of obtaining initial values of alpha-particle spectra parameters based on genetic algorithm is proposed.?4?The iterative deconvolution method is applied to the analysis of alpha-particle spectra.An alpha-particle Monte Carlo simulation spectrum broadening algorithm was established,and a Monte Carlo simulation alpha-particle spectra and spectra interpolation algorithm was gathered to establish the alpha-particle spectra detector response matrix.In summary,on the basis of previous studies,this paper presents a new alpha-particle spectra detector response function model for the common problem in?-particle spectra analysis of semiconductor Si detectors,and proposes initial value acquisition method of parameters for alpha-particle spectrum fitting.Alpha particle Monte Carlo spectrum broadening algorithm was established,and the iterative deconvolution method was applied to the alpha-particle spectra analysis,which enriched the analysis method of alpha-particle spectra.This study has reference value to improve the analysis accuracy of alpha-particle spectra and to accumulate research experience for automated analysis.