Method for predicting atmospheric neutron see error rate based on heavy ion and proton data

Abstract

The present application relates to a method for predicting atmospheric neutron SEE error rate based on heavy ion and proton data, comprising the following steps: obtaining heavy ion SEE cross section of electronic devices by carrying out heavy ion SEE test; obtaining proton SEE cross section by carrying out proton SEE test; obtaining LET spectrum of proton and silicon nuclear reaction product by simulation; determining equivalent thickness t' by combining heavy ion and proton SEE cross section of electronic devices; obtaining LET spectrum of neutron and silicon nuclear reaction product by simulating neutron and silicon nuclear reaction; predicting neutron SEE cross section by combining heavy ion SEE cross section, LET spectrum of neutron and silicon nuclear reaction product and equivalent thickness t'; and predicting atmospheric neutron SEE error rate by combining atmospheric neutron energy spectrum and neutron SEE cross section. The method disclosed in the present application can predict atmospheric neutron SEE error rate only by obtaining SEE cross section of heavy ion and proton by experiment, and the obtained error rate prediction result is simpler and more reliable than that of traditional method.

Description

Technical Field

[0001] This invention belongs to the field of radiation effects of electronic devices, specifically relating to a method for predicting the SEE error rate of atmospheric neutrons based on heavy ion and proton data. Background Technology

[0002] Atmospheric neutrons are produced by high-energy particles from outer space entering the atmosphere and reacting with their nuclei. Their energy ranges from thermal neutrons to GeV, and their energy spectrum and flux are influenced by altitude, latitude, solar activity, and geomagnetic variations. Neutrons can induce single-event effects (SEE) by ionizing secondary ions produced from nuclear reactions with materials inside electronic devices. The hazards of atmospheric neutrons to aircraft have received widespread attention and importance from the international aviation community, leading to the development of standards such as JESD 89A and IEC 62396 for testing methods of neutron SEE in microelectronic devices and for assessing their resistance to atmospheric radiation environments. The potential harm of atmospheric neutron SEE to high-reliability industries such as finance, big data, and intelligent vehicles is also increasingly attracting attention.

[0003] Predicting the single-event effect (SEE) error rate induced by atmospheric neutrons is a core aspect of evaluating the resistance of electronic devices to atmospheric neutron radiation. Currently, there are two main methods for predicting the SEE error rate. One method involves conducting white-light neutron irradiation experiments on the device. White-light neutrons are produced by bombarding heavy elements with high-energy protons, and their forward energy spectrum shape is close to that of atmospheric neutrons, but their fluence rate is millions of times higher. This can be used to accelerate the single-event effect of atmospheric neutrons on the device. The acceleration factor is generally considered to be the ratio of the neutron fluence rate above 1 or 10 MeV. It is worth noting that although the energy spectrum shape of white-light neutrons is similar to that of atmospheric neutrons at a specific location, there are still differences. These differences introduce significant uncertainty into the prediction of the atmospheric neutron SEE error rate.

[0004] Another approach involves conducting quasi-monoenergetic neutron irradiation experiments on the device to obtain the relationship between the neutron SEE cross section and energy, and then combining this with the atmospheric neutron energy spectrum to obtain the atmospheric neutron SEE error rate. Quasi-monoenergetic neutrons generally consist of protons and... 7 In addition to single-energy neutrons, Li nuclear reactions also produce low-energy tail neutrons, which makes it difficult to obtain the neutron SEE cross section corresponding to the peak energy of the single-energy neutron. Commonly used iterative deconvolution tail correction methods require pre-assuming the curve shape of the neutron SEE cross section as a function of energy, which leads to a large error in the obtained neutron SEE cross section. Ultimately, this results in a large error in the atmospheric neutron SEE error rate obtained based on it. Therefore, it is necessary to seek new methods for predicting the atmospheric neutron SEE error rate. Summary of the Invention

[0005] To address the shortcomings of existing technologies, the present invention aims to provide a method for predicting the atmospheric neutron SEE error rate based on heavy ion and proton data. This method can, to a certain extent, avoid the significant uncertainties present in traditional prediction methods based on white light neutron and quasi-monoenergetic neutron SEE experiments. Starting from the single-event effect mechanism of heavy ions, protons, and neutrons, and based on the experimentally obtained heavy ion and proton SEE cross sections, it achieves the prediction of the neutron SEE cross section. Furthermore, it combines this prediction with the atmospheric neutron energy spectrum to calculate the atmospheric neutron SEE error rate, thus expanding a new approach to obtaining the atmospheric neutron SEE error rate.

[0006] To achieve the above objectives, the technical solution adopted by this invention is: a method for predicting the SEE error rate of atmospheric neutrons based on heavy ion and proton data, the method comprising the following steps:

[0007] S1. Conduct heavy-ion SEE experiments on electronic devices, obtain their heavy-ion SEE cross sections, and perform Weibull fitting on the relationship between the heavy-ion SEE cross sections and the LET values ​​of heavy ions.

[0008] S2. Conduct proton SEE experiments on electronic devices, select protons of different energies to irradiate electronic devices, and obtain the proton SEE cross section of electronic devices.

[0009] S3. Monte Carlo simulations were performed on the reaction between protons and silicon nuclei to obtain the LET spectra of the reaction products of protons and silicon nuclei with different energies;

[0010] S4. Based on the heavy ion SEE cross section of electronic devices and the LET spectrum of the reaction products of protons with silicon nuclei at different energies calculated by simulation, the proton SEE cross section is predicted, and the equivalent thickness t′ that makes the prediction result most consistent with the experiment is determined.

[0011] S5. Monte Carlo simulation of the reaction between neutrons and silicon nuclei was performed to obtain the LET spectra of the reaction products of neutrons and silicon nuclei at different neutron energies;

[0012] S6. The neutron SEE cross section based on electronic devices, the LET spectrum of neutron-silicon nucleus reaction products at different neutron energies calculated by simulation, and the equivalent thickness t′ are used to predict the neutron SEE cross section.

[0013] S7. Predict the error rate of atmospheric neutron SEE by combining the neutron SEE cross section and the atmospheric neutron energy spectrum.

[0014] Furthermore, in step S1, the electronic device is irradiated with heavy ions of four or more different LET values ​​to obtain the heavy ion SEE cross section of the electronic device.

[0015] Furthermore, in step S3, Monte Carlo software is used to simulate N. inA proton with energy E is incident on a thin silicon layer of thickness H. The LET values ​​generated by the reaction between the proton and the silicon nucleus are statistically analyzed. The number of products within the interval is used to calculate the LET spectrum of nuclear reaction products. Where n Si The number of Si atoms per unit volume is denoted by L, and L represents the LET value of the nuclear reaction product.

[0016] Furthermore, in step S3, the proton energy used in the Monte Carlo simulation should be the same as the energy used in the experiment, or cover the energy range of the proton energy used in the experiment.

[0017] Furthermore, in step S4, according to the formula Predict the proton SEE cross section, where E is the proton energy, t is the equivalent thickness, and σ is the cross section. HI For heavy ion SEE cross section, n Si The number of Si atoms per unit volume. The LET spectra are of the products of the reaction between protons and silicon nuclei at different energies.

[0018] Furthermore, in step S4, when predicting the proton SEE cross section, the predicted proton SEE cross sections at multiple proton energies are compared with the experimentally obtained proton SEE cross sections to find the equivalent thickness value t′ that makes the two closest overall.

[0019] Furthermore, in step S5, Monte Carlo software is used to simulate N. in A thin silicon layer of thickness H is incident with neutrons of energy E. The LET values ​​generated by the reaction of neutrons with silicon nuclei are statistically analyzed. The number of products ΔN within the interval is used to calculate the LET spectrum of nuclear reaction products.

[0020]

[0021] Furthermore, in step S6, according to the formula Predict the neutron SEE cross section.

[0022] Furthermore, in step S7, according to the formula The error rate of atmospheric neutron SEE is predicted, where C is the number of units contained in the electronic device. For the atmospheric neutron energy spectrum, σ n (E) represents the neutron SEE cross section.

[0023] Furthermore, the energy of the neutrons used in the Monte Carlo simulation in step S5 should cover the atmospheric neutron energy range.

[0024] Furthermore, the SEE test in step S1 includes the SEU test, and the SEE cross section includes the SEU cross section.

[0025] The beneficial technical effects of this invention are as follows: The method for predicting the error rate of atmospheric neutron SEE based on heavy ion and proton data disclosed in this invention is based on the fact that both protons and neutrons are single-event effects induced by the ionization of secondary particles produced by nuclear reactions. The correlation between these two methods and the single-event effect cross-section induced by heavy ions is the same, and the equivalent thickness should be the same value when predicting the proton SEE cross-section and the neutron SEE cross-section. Based on this, a new approach to predicting the error rate of atmospheric neutron SEE is opened up, enabling the acquisition of the atmospheric neutron SEE error rate without white light or quasi-monoenergetic neutron SEE experiments, while avoiding, to a certain extent, the large uncertainties present in these two prediction methods. Since it is based on the correlation between the single-event effect mechanisms of protons, neutrons, and heavy ions, it should theoretically have higher accuracy, laying a reliable foundation for evaluating the atmospheric neutron radiation resistance of electronic devices. Attached Figure Description

[0026] Figure 1 This is a flowchart of the method for predicting the SEE error rate of atmospheric neutrons based on heavy ion and proton data according to Embodiment 1 of the present invention.

[0027] Figure 2 The heavy-ion SEU cross section and its Weibull fit of the SRAM device of model CY7C1318CV18 obtained by the experiment shown in Embodiment 1 of the present invention;

[0028] Figure 3 The LET spectrum of the reaction products of protons and silicon nuclei in the 1-100 MeV energy range, as shown in Embodiment 1 of the present invention, is obtained from simulation calculations.

[0029] Figure 4 This is a comparison diagram of the predicted proton SEU cross section and the experimental value when the value of t is 1.55 μm, as shown in Embodiment 1 of the present invention.

[0030] Figure 5 The JEDEC standard atmospheric neutron energy spectrum is shown in Embodiment 1 of the present invention;

[0031] Figure 6 The LET spectrum of the reaction products of neutrons and silicon nuclei in the 1-10000 MeV energy range is shown in Embodiment 1 of the present invention.

[0032] Figure 7 This is the predicted neutron SEU cross section when the sample size is 1.55 μm, as shown in Embodiment 1 of the present invention. Detailed Implementation

[0033] The present invention will now be further described with reference to the accompanying drawings and specific embodiments.

[0034] Example 1

[0035] Taking the SEE error rate prediction of the SRAM with model number CY7C1318CV18 in the atmospheric neutron radiation environment in the JEDEC standard as an example, the method for predicting the atmospheric neutron SEE error rate based on heavy ion and proton data disclosed in this invention is illustrated.

[0036] like Figure 1 As shown, this embodiment of the invention provides a method for predicting the SEE error rate of atmospheric neutrons based on heavy ion and proton data. The method includes the following steps:

[0037] S1. Conduct heavy-ion SEE experiments on electronic devices to obtain the heavy-ion SEE cross section of the electronic devices, and perform Weibull fitting on the relationship between the heavy-ion SEE cross section of the electronic devices and the LET value of heavy ions.

[0038] Heavy-ion SEE experiments were conducted on electronic devices. The devices were irradiated with heavy ions of four or more different LET (Linear Energy Transfer) values. The heavy-ion SEE cross sections of the electronic devices were obtained, and the relationship between the cross sections and LET was fitted using Weibull fitting. The fitted heavy-ion SEE cross section curve is denoted as σ. HI (L), where L represents the LET value of the heavy ion.

[0039] The SRAM, model CY7C1318CV18, has a feature size of 90 nm and a capacity of 18 Mbit. Heavy-ion single-event upset (SEU) experiments were performed on the SRAM. SEU is a type of single-event effect (SEE), such as... Figure 2 As shown, the SEU cross sections and their Weibull fitting parameters under heavy ion irradiation with four different LET values ​​were obtained.

[0040] S2. Conduct proton SEE experiments on electronic devices by irradiating the electronic devices with protons of various energies and obtaining the proton SEE cross sections of the electronic devices.

[0041] Following the previous example, proton SEU experiments were conducted on SRAM at various energies up to 100 MeV to obtain its proton SEU cross-section.

[0042] S3. Simulate N using Monte Carlo software. in A proton with energy E is incident on a thin silicon layer of thickness H. The LET value of the p+Si nuclear reaction is statistically analyzed. If the number of products ΔN within the interval is given, then the LET spectrum of the nuclear reaction products is:

[0043]

[0044] Where nSi The number of Si atoms per unit volume. The LET spectra of p+Si nuclear reaction products corresponding to protons with different energies can be simulated and calculated using formula (1). The proton energy used in the simulation should be the same as the energy used in the experiment, or cover the energy range of the protons used in the experiment.

[0045] Continuing from the previous example, we used Geant4 software to perform Monte Carlo simulations of the p+Si nucleus reaction, such as... Figure 3 As shown, LET spectra of the reaction products of protons with silicon nuclei at different energies were obtained.

[0046] S4, based on the heavy ion SEE cross section σ HI And the LET spectrum of p+Si nuclear reaction products calculated by simulation For the proton SEE section σ p To make predictions, that is

[0047]

[0048] Where E is the proton energy and t is the equivalent thickness. In the prediction, only t is unknown. Different values ​​of t result in different predicted proton SEE cross-section curves. Therefore, proton SEE cross-sections at multiple proton energies used in the experiment can be obtained through interpolation and other methods. These cross-sections are then compared with the proton SEE cross-sections obtained in the experiment to find the value of t that makes the two most similar overall, denoted as t′.

[0049] Continuing from the previous example, such as Figure 4 As shown, calculations show that when t is 1.55 μm, the proton SEU cross section predicted by formula (2) best matches the experimental value. Therefore, t′ is set to 1.55 μm.

[0050] S5. Monte Carlo simulations were performed on the reaction between neutrons and silicon nuclei to obtain LET spectra of n+Si nuclei reaction products at various neutron energies. The energies of the neutrons used in the simulation should cover the atmospheric neutron energy range.

[0051] A Monte Carlo simulation of the neutron-silicon nucleus reaction was performed using a method similar to that in step S3. Specifically, Monte Carlo software was used to simulate the N... in A thin silicon layer of thickness H is incident with neutrons of energy E. The LET values ​​generated by the n+Si nuclear reaction are statistically analyzed. The number of products ΔN within the range was used to obtain the LET spectra of n+Si nuclear reaction products at various neutron energies.

[0052] Following the previous example, Figure 5 The data represents the JEDEC standard atmospheric neutron energy spectrum, covering the energy range of 1–10000 MeV. Monte Carlo simulations of n+Si nuclear reactions were performed using Geant4 software, such as… Figure 6As shown, the LET spectra of the reaction products of neutrons and silicon nuclei in the energy range of 1–10000 MeV were obtained.

[0053] S6, Based on the heavy ion SEE cross section σ HI And the LET spectrum of n+Si nuclear reaction products calculated by simulation Predicting the neutron SEE cross section, i.e.

[0054]

[0055] This allows for the prediction of the neutron SEE cross section in the atmospheric neutron energy region.

[0056] Continuing from the previous example, such as Figure 7 As shown, when t′ is 1.55 μm, the predicted neutron SEU cross section is obtained by calculation according to formula (3).

[0057] S7. Predict the error rate of atmospheric neutron SEE by combining the neutron SEE cross section and the atmospheric neutron energy spectrum, i.e.

[0058]

[0059] Where C represents the number of units contained in the electronic device. This is the atmospheric neutron energy spectrum, measured in cm⁻¹. -2 s -1 MeV -1 .

[0060] Following the previous example, the single-event flip rate of the SRAM under JEDEC standard atmospheric neutron radiation environment was finally predicted to be 7.62 × 10⁻⁶ according to formula (4). -6 upsets / hour.

[0061] As can be seen from the above embodiments, the method for predicting the atmospheric neutron SEE error rate based on heavy ion and proton data disclosed in this invention does not require the use of white light or quasi-monoenergetic neutron SEE experiments. It can predict the atmospheric neutron SEE error rate simply by conducting experiments to obtain the SEE cross sections of heavy ions and protons. Moreover, the error rate prediction results obtained are simpler and more reliable than traditional methods, laying a reliable foundation for evaluating the atmospheric neutron radiation resistance of electronic devices.

[0062] The method described in this invention is not limited to the embodiments described in the specific implementation. Other implementation methods derived by those skilled in the art based on the technical solution of this invention also fall within the scope of technical innovation of this invention.

Claims

1. A method for predicting the SEE error rate of atmospheric neutrons based on heavy ion and proton data, the method comprising the following steps: S1. Conduct heavy-ion SEE experiments on electronic devices, obtain their heavy-ion SEE cross sections, and perform Weibull fitting on the relationship between the heavy-ion SEE cross sections and the LET values ​​of heavy ions. S2. Conduct proton SEE experiments on electronic devices, select protons of different energies to irradiate electronic devices, and obtain the proton SEE cross section of electronic devices. S3. Monte Carlo simulations were performed on the reaction between protons and silicon nuclei to obtain the LET spectra of the reaction products of protons and silicon nuclei with different energies; S4. Based on the heavy ion SEE cross section of electronic devices and the LET spectrum of the reaction products of protons with silicon nuclei at different energies calculated by simulation, the proton SEE cross section is predicted, and the equivalent thickness t′ that makes the prediction result most consistent with the experiment is determined. S5. Monte Carlo simulations were performed on the reaction between neutrons and silicon nuclei to obtain the LET spectra of the reaction products of neutrons and silicon nuclei at different neutron energies. S6. The neutron SEE cross section based on electronic devices, the LET spectrum of neutron-silicon nucleus reaction products at different neutron energies calculated by simulation, and the equivalent thickness t′ are used to predict the neutron SEE cross section. S7. Predict the error rate of atmospheric neutron SEE by combining the neutron SEE cross section and the atmospheric neutron energy spectrum.

2. The method for predicting the SEE error rate of atmospheric neutrons based on heavy ion and proton data as described in claim 1, characterized in that: In step S1, heavy ion irradiation of electronic devices with four or more different LET values ​​is used to obtain the heavy ion SEE cross section of the electronic devices.

3. The method for predicting the SEE error rate of atmospheric neutrons based on heavy ion and proton data as described in claim 2, characterized in that: In step S3, Monte Carlo software is used to simulate N. in A proton with energy E is incident on a thin silicon layer of thickness H. The LET values ​​generated by the reaction between the proton and the silicon nucleus are statistically analyzed. The number of products ΔN within the interval is used to calculate the LET spectrum of nuclear reaction products. Where n Si The number of Si atoms per unit volume is denoted by L, and L represents the LET value of the nuclear reaction product.

4. The method for predicting the SEE error rate of atmospheric neutrons based on heavy ion and proton data as described in claim 3, characterized in that: In step S3, the proton energy used in the Monte Carlo simulation should be the same as the energy used in the experiment, or cover the energy range of the proton energy used in the experiment.

5. The method for predicting the SEE error rate of atmospheric neutrons based on heavy ion and proton data as described in claim 4, characterized in that: In step S4, according to the formula Predict the proton SEE cross section, where E is the proton energy, t is the equivalent thickness, and σ is the cross section. HI For heavy ion SEE cross section, n Si The number of Si atoms per unit volume. The LET spectrum is for the products of the reaction between protons and silicon nuclei.

6. The method for predicting the SEE error rate of atmospheric neutrons based on heavy ion and proton data as described in claim 5, characterized in that: In step S4, when predicting the proton SEE cross section, the predicted proton SEE cross sections at multiple proton energies are compared with the experimentally obtained proton SEE cross sections to find the equivalent thickness value t′ that makes the two closest overall.

7. The method for predicting the SEE error rate of atmospheric neutrons based on heavy ion and proton data as described in claim 6, characterized in that: In step S6, according to the formula Predicting the neutron SEE cross section, where The LET spectrum is for the products of the reaction between neutrons and silicon nuclei.

8. The method for predicting the SEE error rate of atmospheric neutrons based on heavy ion and proton data as described in claim 7, characterized in that: In step S7, according to the formula The error rate of atmospheric neutron SEE is predicted, where C is the number of units contained in the electronic device. For the atmospheric neutron energy spectrum, σ n (E) represents the neutron SEE cross section.

9. The method for predicting the SEE error rate of atmospheric neutrons based on heavy ion and proton data as described in claim 1, characterized in that: The neutron energies used in the Monte Carlo simulation in step S5 should cover the atmospheric neutron energy range.

10. The method for predicting the SEE error rate of atmospheric neutrons based on heavy ion and proton data as described in claim 1, characterized in that: The SEE test described in step S1 includes a single-event flip test.

Images