A method for verifying accuracy of CZT irradiation damage simulation

Abstract

The application discloses a kind of verification methods of tellurium zinc cadmium (CZT) irradiation damage simulation accuracy, the method based on two software packages of SRIM software simulates the irradiation damage of CZT material, uses machine learning modeling to obtain regression curve to verify the accuracy of simulation result, improves simulation calculation speed.In the ion projection range distribution data set, the fitting effect of the linear regression model of machine learning is better than that of logarithmic regression model, and the determination coefficient of linear regression in the projection range fitting of H ion and He ion is greater than 0.98, while the determination coefficient of logarithmic regression model is only greater than 0.60.The method of the application has important significance and application prospect in the fields of energy, public safety, military, nuclear industry, nuclear medicine, scientific research and aerospace, energy acquisition, safety monitoring and radiation protection.

Description

Technical Field

[0001] This invention relates to a method for verifying the accuracy of irradiation damage simulation of zinc cadmium telluride, belonging to the field of optoelectronic materials and devices. Background Technology

[0002] Cadmium zinc telluride (CdZnTe), abbreviated as CZT, is a high-performance group II-VI compound semiconductor with a band gap of 1.63 eV at room temperature. It belongs to a direct-transition band structure, which is close to the ideal band gap value for solar cell materials. Furthermore, CZT exhibits an absorption coefficient as high as 10⁻⁶ in the visible light band. 5 cm -1 A thickness of only 1 μm is sufficient to absorb over 90% of photons with energies exceeding its bandgap. Therefore, CZT is a relatively ideal material for the absorber layer of solar cells. In recent years, it has been widely used in the field of solar cells. Furthermore, detectors made from CZT possess characteristics such as ultra-high resolution, no need for liquid nitrogen cooling, small size, and portability, making them a new generation of nuclear radiation detectors and a hot research topic in the international high-tech field in recent years.

[0003] When used as a material in radiation detectors, CZT crystals inevitably interact with high-energy particles such as electrons, protons, X-rays, and gamma rays. This interaction significantly impacts the defects and energy levels within the crystal. Detectors primarily rely on the interaction between the probe rays and the detector material to generate new electron-hole pairs, i.e., non-equilibrium charge carriers. These non-equilibrium charge carriers are collected by external equipment and output as electrical signals. The generation and transport of these new electron-hole pairs are closely related to defects within the crystal. Furthermore, the radiation damage effects generated during nuclear radiation, medical applications, and astronomical exploration can greatly influence the inherent defects within CZT crystals. Therefore, studying the radiation damage effects of high-energy particles on CZT crystals and the shielding against radiation damage is highly significant for the application of CZT crystals in high-energy radiation detectors.

[0004] The SRIM software uses the Monte Carlo method for simulation. Therefore, when the number of ions is set to 99,999, it must calculate the interaction between the ion and the target and record its trajectory one ion at a time. This whole process is very time-consuming, and the simulation time increases with the number of target layers, thickness, types of elements contained, and atomic number of the incident ions. Summary of the Invention

[0005] In order to solve the prior art problems, the purpose of the present application is to overcome the shortcomings of the prior art, provide a verification method for simulation accuracy of tellurium-zinc-cadmium irradiation damage, introduce linear regression and logarithmic linear regression models of machine learning to fit data, establish a model, so that the distribution law in other energy ranges can be predicted and its correctness can be verified. The present application simulates various conditions systematically and perfectly by using SR, TRIM software and machine learning algorithm, establishes a regression model by using the method of machine learning, so as to realize a method which can effectively verify the correctness of the simulation of CZT damage. The method of the present application can simulate the irradiation damage of charged ions in the cosmic space to CZT material by using a computer program, and can play a greater role in the field of aerospace.

[0006] In order to achieve the above-mentioned purposes, the present application adopts the following technical solutions:

[0007] A verification method for simulation accuracy of CZT irradiation damage, using open source SRIM software, simulating the motion process of H and He ions by using SR mode, the irradiation ions are H and He ions, and the target material is CZT; H and He ions are bombarded on the CZT target material, the energy loss and penetration ability data of H and He ions in the target material are analyzed by using the SR software package in the SRIM software, and the results are analyzed and compared, and a linear regression and logarithmic regression model is established by using a machine learning algorithm to fit the result analysis; wherein, the irradiation damage effect of H and He ions is simulated by using TRIM mode, and the vacancy.txt file of the simulated data is substituted into the formula:

[0008]

[0009] The vacancy production rate of the CZT crystal can be obtained from formula (1-1); wherein represents the injection dose, represents the atomic density, # of vacancies represents the vacancy number, atoms represents the atomic weight, and Dpa represents displacement damage, that is, how many atomic displacements per unit distance.

[0010] Preferably, the verification method for simulation accuracy of CZT irradiation damage comprises the following steps:

[0011] (1) Data acquisition process:

[0012] Two software packages based on SRIM software are used to simulate the irradiation damage of CZT material:

[0013] The SR software package is used to simulate the electron stopping power and nuclear stopping power of CZT for different charged ions, and the penetration ability of different charged ions is simulated;

[0014] Using TRIM software package, the damage of materials and the production of vacancies in displacement damage under different incident energies and angles are simulated.

[0015] (2) Machine learning modeling process:

[0016] Based on machine learning algorithm, using python programming language, importing sklearnpandas library, reading data set, preprocessing data set, dividing data set into training set and test set, establishing linear regression model, using linear regression model and logistic regression model respectively, taking incident energy of ion as characteristic value and projected range as target value.

[0017] Preferably, in the step (2), the data set is 70-100 groups, wherein the test set accounts for 20-30%, and the data set is divided into 50-70 groups of training set and 20-30 groups of test set.

[0018] Preferably, in the step (2), for H ion, the final trained linear regression model is as formula (1-2):

[0019] ProjectedRange=339.30692898×E-122147.77882447519 (1-2)

[0020] Preferably, in the step (2), for H ion, the final trained logistic regression model is as formula (1-3):

[0021] ProjectedRange=exp(0.00103467×E+9.10695465877085) (1-3)

[0022] Preferably, in the step (2), for He ion, the final trained linear regression model is as formula (1-4):

[0023] ProjectedRange=41.02740264×E-6684.187246266345 (1-4)

[0024] Preferably, in the step (2), for He ion, the final trained logistic regression model is as formula (1-5):

[0025] ProjectedRange=exp(0.0007475×E+8.463578713631032) (1-5)

[0026] Preferably, in the step (2), the energy range is 10keV-10MeV, and one hundred points are set.

[0027] Compared with the prior art, the present application has the following obvious substantial features and advantages:

[0028] 1. The verification method of the simulation accuracy of CZT irradiation damage can simulate the irradiation damage effects of CZT under various irradiation conditions without damaging the material itself, simulate the vacancy generation rate and distribution of CZT irradiation damage, and solve the problem of the position of the irradiation damage vacancy which is difficult to display in actual experiments.

[0029] 2. The verification method of the simulation accuracy of CZT irradiation damage uses a machine learning method to establish a regression model, which can fit the characteristic curve of irradiation damage according to the irradiation parameters, ensure the correctness, and solve the problem of the simulation software needing to run 99999 points in the simulation procedure, thereby saving a lot of time.

[0030] 3. The present application has important significance and application prospect in the fields of energy acquisition, safety monitoring, radiation protection, and the like. BRIEF DESCRIPTION OF DRAWINGS

[0031] Figure 1 The comparison chart of the logarithmic regression and linear regression models of H ions of the preferred embodiment of the present application and the true value.

[0032] Figure 2 The comparison chart of the logarithmic regression and linear regression models of He ions of the preferred embodiment of the present application and the true value. DETAILED DESCRIPTION

[0033] The following embodiment of the verification method of the simulation accuracy of CZT irradiation damage uses the open source SRIM software to simulate the motion process of H and He ions by using the SR mode, the irradiation ions are H and He ions, and the target material is CZT; the H and He ions are used to bombard the CZT target material, the energy loss and penetration ability data of the H and He ions in the target material are analyzed by using the SR software package in the SRIM software, and the results are analyzed and compared, and a machine learning algorithm is used to establish a linear regression and logarithmic regression model to fit the result analysis; wherein the TRIM mode is used to simulate the irradiation damage effects of H and He ions, and the vacancy.txt file of the simulation data is substituted into the formula:

[0034]

[0035] The vacancy generation rate of the CZT crystal can be obtained from formula (1-1); wherein represents the implantation dose, represents the atomic density, # of vacancies represents the number of vacancies, atoms represents the atomic weight, and Dpa represents the displacement damage, i.e., how many atomic displacements per unit distance.

[0036] The following embodiment verifies the accuracy of the simulation of the irradiation damage of cadmium zinc telluride, introduces a linear regression and a logarithmic linear regression model of machine learning to fit the data, establishes a model, and thus can predict the distribution law in other energy ranges and verify its correctness.

[0037] The above scheme is further described in combination with specific embodiment examples, and preferred embodiments of the application are described in detail as follows:

[0038] In this embodiment, H ions and He ions with an energy range of 10 keV to 10 MeV are used to bombard the CZT target, the energy loss and penetration ability of the ions in the target are analyzed by the SR software package in the SRIM software, the results are comprehensively analyzed and compared, and a linear regression and a logarithmic regression model are established by using a machine learning algorithm to fit the results, and the results are shown in Figure 1 and Figure 2 .

[0039] A method for verifying the correctness of the simulation of CZT damage by computer software, comprising the following steps:

[0040] (1) Simulate the energy loss and penetration ability of different charged ions in the CZT target

[0041] First, the ion is selected as H ion, and the relative atomic mass of the SR software is used by default; the minimum energy of the H ion is selected as 10 keV, and the maximum energy is selected as 10 MeV; the target is set as CZT, the stoichiometric ratio of the elements is 1:1, and the density is set as 6.20 g / cm 3 ; the output file is converted to obtain the energy loss and penetration ability of the H ion in the target; then the ion is selected as He ion, and the relative atomic mass of the He element of the SR software package is used by default; similarly, the minimum energy of the He ion is selected as 10 keV, and the maximum energy is selected as 10 MeV; the target is set as CZT, the stoichiometric ratio of the elements is 1:1, and the density is set as 6.20 g / cm 3 ; the output file is converted to obtain the energy loss and penetration ability of the He ion in the target; by comparing and analyzing the energy loss and penetration ability of the H ion and the He ion, the ratio of the electron stopping power of the He ion to the electron stopping power of the H ion and the ratio of the nuclear stopping power can be obtained;

[0042] (2) Fitting and prediction of ion projection range distribution

[0043] By introducing the linear regression and logistic regression model of machine learning to fit the data, the model is established, so that the distribution rule in other energy range can be predicted; the projected range of 10keV-10MeV H ion and He ion is taken as the data set for training; the incident energy of the ion is taken as the characteristic value, and the projected range is taken as the target value, a total of 79 data sets; according to the principle of 35% test set, the data set is divided into 52 training sets and 27 test sets; linear regression model and logistic regression model are used respectively, and then the linear regression and logistic regression modeling results are analyzed and fitted.

[0044] In the step (2), for H ion, the finally trained linear regression model is as formula (1-2):

[0045] ProjectedRange=339.30692898×E-122147.77882447519 (1-2)

[0046] In the step (2), for H ion, the finally trained logistic regression model is as formula (1-3):

[0047] ProjectedRange=exp(0.00103467×E+9.10695465877085) (1-3)

[0048] In the step (2), for He ion, the finally trained linear regression model is as formula (1-4):

[0049] ProjectedRange=41.02740264×E-6684.187246266345 (1-4)

[0050] In the step (2), for He ion, the finally trained logistic regression model is as formula (1-5):

[0051] ProjectedRange=exp(0.0007475×E+8.463578713631032) (1-5)

[0052] Simulation result analysis:

[0053] In this embodiment, the irradiation damage and implanted ion concentration distribution of H ion and He ion on CZT material are simulated based on SRIM software, the data therein are trained and fitted by using machine learning algorithm, the actual damage can be approximately simulated, the protection effect of shielding layer on target material is simulated and analyzed, the influence mechanism of charged ion on CZT irradiation damage and how to select shielding layer to avoid these damages are deeply studied, and the ion projected range distribution is fitted and predicted. The simulation result analysis of the embodiment is as follows:

[0054] (1) When H ions and He ions are incident on the CZT target, the electron stopping power of the target for the charged ions is much greater than the nuclear stopping power, so the incident energy of the charged ions is mainly lost through interaction with the outer electrons, and the energy lost through interaction with the target atoms accounts for only a small part. The irradiation damage produced therein includes ionization damage and displacement damage, and most of the energy of the ions is used to cause ionization damage. The greater the atomic number of the incident ions, the greater the energy damage per unit distance, and the more serious the irradiation damage per unit distance, so the penetration ability of the target is weaker. And with the increase of the incident energy, the difference in irradiation damage per unit distance between ions with large atomic numbers and ions with small atomic numbers will become larger and larger.

[0055] (2) In the ion projection range distribution data set, the fitting effect of the linear regression model of machine learning is better than that of the logarithmic regression model, and the determination coefficients of the linear regression in the projection range fitting of H ions and He ions are all greater than 0.98, while the determination coefficients of the logarithmic regression model are only greater than 0.60.

[0056] The above-mentioned embodiment of the verification method for the accuracy of the irradiation damage simulation of cadmium zinc telluride (CZT) of the present application simulates the irradiation damage of the CZT material based on two software packages of the SRIM software, uses machine learning modeling to obtain a regression curve to verify the accuracy of the simulation results, and improves the simulation calculation speed. In the ion projection range distribution data set, the fitting effect of the linear regression model of machine learning is better than that of the logarithmic regression model, and the determination coefficients of the linear regression in the projection range fitting of H ions and He ions are all greater than 0.98, while the determination coefficients of the logarithmic regression model are only greater than 0.60. The above-mentioned embodiment of the present application has important significance and application prospect in the fields of energy, public safety, military, nuclear industry, nuclear medicine, scientific research, aerospace and the like, energy acquisition, safety monitoring and radiation protection.

[0057] The above describes the embodiments of the present application in combination with the drawings, but the present application is not limited to the above-mentioned embodiments, and can be changed in various ways according to the purpose of the invention of the present application. Any change, modification, replacement, combination or simplification made according to the spirit and principles of the technical solution of the present application shall be an equivalent replacement mode, as long as it meets the purpose of the present application, as long as it does not deviate from the technical principles and inventive concept of the present application, and belongs to the protection scope of the present application.

Claims

1. A method for verifying the accuracy of CZT irradiation damage simulation, characterized in that: Using the open-source SRIM software, the movement process of H and He ions was simulated in SR mode. The irradiated ions were H and He ions, and the target material was CZT. The CZT target was bombarded with H and He ions. The energy loss and penetration ability data of H and He ions in the target were analyzed using the SR package in the SRIM software. The results were analyzed and compared, and machine learning algorithms were used to establish linear regression and logarithmic regression models for fitting analysis. The TRIM mode was used to simulate the irradiation damage effect of H and He ions. The simulation data from the vacancy.txt file were substituted into the formula. The vacancy generation rate of CZT crystal can be obtained from formula (1-1); where Indicates the injection dose. # of vacancies represents the number of vacancies, atoms represents the atomic weight, and Dpa represents the displacement damage, i.e., how many atoms are displaced per unit distance. The method for verifying the accuracy of CZT irradiation damage simulation includes the following steps: (1) Data acquisition process: Two software packages based on SRIM were used to simulate and study the radiation damage of CZT materials: The SR software package was used to simulate the electron blocking and nuclear blocking abilities of CZT against different charged ions, and to simulate the penetration ability of different charged ions. The TRIM software package was used to simulate the irradiation damage of materials by ions with different incident energies and incident angles, as well as the generation of vacancies in displacement damage. (2) Machine learning modeling process: Based on machine learning algorithms, using the Python programming language, importing the sklearn pandas library, reading in the dataset, preprocessing the dataset, dividing the dataset into training and testing sets, establishing a linear regression model, and using both linear regression and logarithmic regression models, with the incident energy of the ion as the feature value and the projected range as the target value.

2. The method for verifying the accuracy of CZT irradiation damage simulation according to claim 1, in step (2), the dataset consists of 70 to 100 groups, of which the test set accounts for 20 to 30%, and the dataset is divided into 50 to 70 training groups and 20 to 30 test groups.

3. The method for verifying the accuracy of CZT irradiation damage simulation according to claim 1, in step (2), for H ions, the final trained linear regression model is as shown in equation (1-2): ProjectedRange=339.30692898×E-122147.77882447519 (1-2).

4. The method for verifying the accuracy of CZT irradiation damage simulation according to claim 1, in step (2), for H ions, the final trained logarithmic regression model is as shown in equation (1-3): ProjectedRange=exp(0.00103467×E+9.10695465877085) (1-3).

5. The method for verifying the accuracy of CZT irradiation damage simulation according to claim 1, in step (2), for He ions, the final trained linear regression model is as shown in equation (1-4): ProjectedRange=41.02740264×E-6684.187246266345 (1-4).

6. The method for verifying the accuracy of CZT irradiation damage simulation according to claim 1, in step (2), for He ions, the final trained logarithmic regression model is as shown in equation (1-5): ProjectedRange=exp(0.0007475×E+8.463578713631032) (1-5).

7. In the method for verifying the accuracy of CZT irradiation damage simulation according to claim 1, in step (2), the energy range is 10keV to 10MeV, and one hundred points are set.

Images