Estimation device, estimation method, and program
The estimation device and method improve SEU cross section estimation accuracy in semiconductor devices by adjusting model parameters to align measured and estimated SEU probabilities using neutron and proton irradiation, addressing the accuracy issues in existing methods.
Patent Information
- Application Number
- JP2024114503
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Current Assignee / Owner
- Filing Date
- 2024-07-18
- Publication Date
- 2026-01-29
AI Technical Summary
Existing methods for estimating the Single Event Upset (SEU) cross section in semiconductor devices suffer from reduced accuracy due to the degeneracy of model parameters, as they rely on convolution integral values for determining the SEU probability, making it difficult to derive the model function with high precision.
An estimation device and method that involves irradiating semiconductor devices with neutrons and protons, counting SEUs, measuring SEU probabilities, and adjusting model parameters to match measured and estimated values, using a combination of neutron and proton irradiation to correct the model function and enhance accuracy.
Enables precise estimation of the SEU cross section by correcting the model function based on measured SEU probabilities, improving accuracy through a hybrid approach that combines neutron and proton irradiation experiments.
Smart Images
Figure 2026013836000001_ABST
Abstract
Description
[Technical Field]
[0001] The present disclosure relates to an estimation device, an estimation method, and a program for estimating an SEU cross section. [Background technology]
[0002] When neutrons are incident on a semiconductor device, an error occurs that rewrites the bit information in the device, causing the device to malfunction. This error is called an SEU (Single Event Upset). The probability of a neutron SEU occurring per unit time is calculated using the following equation (1).
[0003]
number
[0004] In equation (1), RSEU is the SEU occurrence probability, En is the neutron energy, Φn(En) is the normalized neutron energy distribution (hereinafter referred to as the "neutron spectrum Φn(En)"), and σSEU(En) is the SEU cross section.
[0005] Therefore, in order to estimate the SEU probability RSEU according to the neutron spectrum in various environments such as above ground, underground, and at aviation altitude, it is necessary to obtain the SEU cross section σSEU(En) specific to each semiconductor device. For this reason, it is desirable to establish a method for estimating the SEU cross section σSEU(En) that can be applied to various semiconductor devices.
[0006] Non-Patent Document 1 discloses a method for estimating the SEU probability RSEU by replacing the SEU cross section σSEU(En) of a semiconductor device shown in equation (1) with a model function fSEU(En,p) using a model parameter p. The model parameter p is an N-dimensional vector, and is represented by p={p0, p1, . . . , pN}. [Prior art documents] [Non-patent literature]
[0007] [Non-Patent Document 1] Estimation of neutronenergy-dependent SEU crosssections of semiconductor devices without fast TOF-analysis function(UCANS10-Abstract-book- FINAL.pdf,p.102) Summary of the Invention [Problem to be solved by the invention]
[0008] In the above-mentioned Non-Patent Document 1, the SEU occurrence probability RSEU shown in equation (1) is replaced with fSEU(En,p) as shown in the following equation (2). The SEU occurrence probability RSEU can be expressed by a function Rcalc(p) that depends on the model parameter p.
[0009]
number
[0010] The SEU probability RSEU is experimentally measured for various neutron sources with known neutron spectra Φn(En), and the model parameter p for which Rcalc(p) = RSEU is determined. By determining the model parameter p (p = {p0, p1, , pN}), the model function fSEU(En, p) shown in equation (2) can be derived. This in turn allows the SEU cross section σSEU(En) to be estimated.
[0011] However, in the technology disclosed in Non-Patent Document 1, the model parameter p is determined based on the SEU probability RSEU, which is a convolution integral value. This can lead to degeneration of each component of the model parameter p, or a decrease in the accuracy of determining the model parameter p. That is, in the above-mentioned equation (2), "RSEU" on the left side and "Rcalc(p)" on the right side are both integral values, and since these integral values are compared, it is difficult to derive the model function fSEU(En,p) with high accuracy.
[0012] The present disclosure has been made in view of the above circumstances, and an object thereof is to provide an estimation device, an estimation method, and a program that are capable of estimating an SEU cross section with high accuracy. [Means for solving the problem]
[0013] An estimation device according to one aspect of the present disclosure is an estimation device for estimating an SEU cross section of a semiconductor device, the estimation device comprising: a measurement unit that counts the number of SEUs that occur when the semiconductor device is irradiated with neutrons or protons from a radiation source and measures an SEU probability; a setting unit that sets a model function that simulates the SEU cross section and sets an equation for calculating the SEU probability based on the model function; a calculation unit that calculates an estimate of the SEU probability using the equation based on spectrum information of the neutrons irradiated from the radiation source; and a comparison unit that compares the measured and estimated SEU probability values and changes model parameters included in the model function so that the measured and estimated values match, wherein the setting unit changes the model function using the model parameters changed by the comparison unit, and corrects the model function based on the SEU probability measured by the measurement unit when the semiconductor device is irradiated with protons from the radiation source, and estimates the corrected model function as the SEU cross section.
[0014] An estimation method according to one aspect of the present disclosure is an estimation method for estimating an SEU cross section of a semiconductor device, the method comprising: a measurement unit counting the number of SEUs that occur when the semiconductor device is irradiated with neutrons or protons emitted from a radiation source to measure an SEU probability; a setting unit setting a model function that simulates the SEU cross section and setting an arithmetic expression for calculating the SEU probability based on the model function; a calculation unit calculating an estimate of the SEU probability using the arithmetic expression based on spectrum information of the neutrons irradiated from the radiation source; a comparison unit comparing the measured and estimated SEU probability values and changing model parameters included in the model function so that the measured and estimated values match; the setting unit modifying the model function using the changed model parameters; and correcting the model function based on the SEU probability measured by the measurement unit when the semiconductor device is irradiated with protons from the radiation source, and estimating the corrected model function as the SEU cross section.
[0015] One aspect of the present disclosure is a program for causing a computer to function as the estimation device. [Effects of the Invention]
[0016] According to the present disclosure, it is possible to estimate the SEU cross section with high accuracy. [Brief explanation of the drawings]
[0017] [Figure 1] FIG. 1 is a block diagram showing the configuration of an estimation device according to an embodiment. [Figure 2] FIG. 2 is an explanatory diagram showing a specific configuration of a radiation source, where (a) shows a case where a neutron source is not used, and (b) shows a case where a neutron source is used. [Figure 3] FIG. 3 is an explanatory diagram showing each data when a semiconductor device is irradiated with neutron spectra from a plurality of radiation sources and the SEU occurrence probability is measured. [Figure 4] FIG. 4 is an explanatory diagram showing each data when calculating the SEU occurrence probability using a model function based on the neutron spectra of multiple radiation sources. [Figure 5] FIG. 5 is an explanatory diagram showing an example of correcting the model function based on the SEU occurrence probability when irradiated with monoenergetic protons. [Figure 6] FIG. 6 is a block diagram showing the hardware configuration of this embodiment. DETAILED DESCRIPTION OF THE INVENTION
[0018] Hereinafter, an embodiment will be described with reference to the drawings. Fig. 1 is a block diagram showing the configuration of an estimation apparatus 100 according to the embodiment. As shown in Fig. 1, the estimation apparatus 100 according to the embodiment includes an experimental apparatus 1 and a calculation apparatus 2. The estimation apparatus 100 estimates the SEU cross-sectional area of a semiconductor device 13.
[0019] The experimental device 1 includes a radiation source 11 and a measurement unit 12. The calculation device 2 includes a setting unit 21, a calculation unit 22, and a comparison unit 23.
[0020] The radiation source 11 irradiates radiation such as protons and neutrons onto a semiconductor device 13 to be measured. The radiation source 11 outputs spectrum information of the radiation to the measurement unit 12 and a calculation unit 22 (described later).
[0021] 2(a) shows a case where a neutron source 15 is not used, and FIG. 2(b) shows a case where a neutron source 15 is used. As shown in FIG. 2, the radiation source 11 includes a proton accelerator 14 and a neutron source 15.
[0022] The proton accelerator 14 irradiates protons. The neutron source 15 emits neutrons when irradiated with protons from the proton accelerator 14. The radiation source 11 can select either a first mode in which the protons irradiated from the proton accelerator 14 irradiate the neutron source 15 to generate neutrons, and the neutrons are irradiated onto the semiconductor device 13, as shown in FIG. 2(b), or a second mode in which the protons irradiated from the proton accelerator 14 are irradiated directly onto the semiconductor device 13 without passing through the neutron source 15, as shown in FIG. 2(a).
[0023] That is, the radiation source 11 includes a proton accelerator 14 that irradiates monoenergetic protons and a neutron source 15 that generates neutrons when irradiated with the monoenergetic protons. When irradiating the semiconductor device 13 with neutrons, the protons irradiated from the proton accelerator 14 are irradiated onto the neutron source to generate neutrons.
[0024] The radiation source 11 can be switched between a first mode in which neutrons are irradiated onto the semiconductor device 13, and a second mode in which protons are irradiated onto the semiconductor device 13. In the second mode, the energy of the protons irradiated from the proton accelerator 14 may be set to, for example, 20 MeV or more.
[0025] 1 counts the number of SEUs that occur when radiation is irradiated onto the semiconductor device 13. The measurement unit 12 measures the probability of SEU occurrence in the semiconductor device 13 based on the spectrum information of the radiation irradiated from the radiation source 11 and the number of SEUs that occur. More specifically, the measurement unit 12 measures the probability of SEU occurrence based on the number of SEUs that occur when neutrons are irradiated from the radiation source 11 and the number of SEUs that occur when protons of more than 20 MeV are irradiated from the radiation source 11.
[0026] The measuring unit 12 outputs the measured SEU probability to the comparing unit 23 (described later). The measured SEU probability includes the measured SEU probability when the semiconductor device 13 is irradiated with neutrons and the measured SEU probability when the semiconductor device 13 is irradiated with protons. That is, the measuring unit 12 counts the number of SEUs that occur when the semiconductor device 13 is irradiated with neutrons or protons emitted from the radiation source 11, and measures the SEU probability.
[0027] The setting unit 21 sets a model function fSEU(En,p) that indicates the relationship between neutron energy En and the SEU occurrence probability RSEU, and a model parameter p that is used in the model function. As described above, the number of SEU occurrences RSEU can be expressed by the following equation (2).
[0028]
number
[0029] In equation (2), RSEU is the SEU probability, En is the neutron energy, Φn(En) is the neutron spectrum, σSEU(En) is the SEU cross section, and fSEU(En,p) is the model function. “Rcalc(p)” is a function that expresses the SEU probability RSEU using the model parameter p.
[0030] The setting unit 21 acquires the SEU probability measurement value measured by the measurement unit 12 when protons are irradiated from the radiation source 11 (the second mode described above), and corrects the model function based on this measurement value. The correction of the model function will be described later. The setting unit 21 optimizes the model parameter p so that the model function fSEU(En,p) matches the SEU cross section σSEU(En). The SEU cross section σSEU(En) is derived based on the optimized model function fSEU(En,p). That is, the setting unit 21 sets a model function that simulates the SEU cross section, and sets an arithmetic expression for calculating the SEU probability based on this model function.
[0031] The calculation unit 22 calculates the SEU probability Rcalc(p) based on the spectrum information of the radiation (neutrons, protons) output from the radiation source 11, the model function fSEU(En,p) set by the setting unit 21, and the model parameter p. The calculation unit 22 outputs the calculated estimated value of the SEU probability to the comparison unit 23. The estimated value of the SEU probability includes an estimated value of the SEU probability when the semiconductor device 13 is irradiated with neutrons (first mode). That is, the calculation unit 22 obtains the estimated value of the SEU probability using an arithmetic formula based on the spectrum information of the neutrons irradiated from the radiation source 11.
[0032] The comparison unit 23 compares the measured value of the SEU probability due to neutrons and protons measured by the measurement unit 12 with the estimated value of the SEU probability based on the neutron spectrum information calculated by the calculation unit 22, and calculates the deviation between them. Based on the calculated deviation, the comparison unit 23 adjusts the model parameter p so that the measured value and the estimated value match, and feeds this back to the setting unit 21. In other words, the comparison unit 23 compares the measured value and the estimated value of the SEU probability, and changes the model parameter p included in the model function so that the measured value and the estimated value match.
[0033] The setting unit 21 changes the model function fSEU(En,p) based on the model parameter p fed back from the comparison unit 23. That is, the setting unit 21 changes the model function fSEU(En,p) so that the measured value and estimated value of the SEU occurrence probability Rcalc(p) match. Furthermore, the setting unit 21 corrects the model function using the measured value of the SEU occurrence probability when the semiconductor device 13 is irradiated with protons.
[0034] That is, the setting unit 21 changes the model function using the model parameter p changed by the comparison unit 23, corrects the model function based on the SEU occurrence probability measured by the measurement unit 12 when the radiation source 11 irradiates the semiconductor device 13 with protons, and estimates the corrected model function as the SEU cross-sectional area.
[0035] Specific processing procedures for setting the model function fSEU(En,p) will be described below with reference to Figures 3 to 5. Figure 3 is an explanatory diagram showing each piece of data when measuring the SEU probability RSEU by irradiating the semiconductor device 13 with neutron spectra Φn(En) from multiple radiation sources. Figure 4 is an explanatory diagram showing each piece of data when calculating the SEU probability Rcalc(p) using a model function based on the neutron spectra Φn(En) from multiple radiation sources.
[0036] FIG. 3(a) is a graph showing the neutron spectrum Φn(En) irradiated from multiple radiation sources in the radiation source 11. For example, four types of neutrons from radiation sources A1, A2, A3, and A4 are irradiated toward a semiconductor device 13. Irradiating the semiconductor device 13 with neutrons from each of the radiation sources A1 to A4 causes an SEU to occur in the semiconductor device 13. The number of SEU occurrences is measured by the measurement unit 12. As a result, a measured value of the SEU occurrence probability RSEU from each of the radiation sources A1 to A4 is obtained, as shown in FIG. 3(c), for example. Note that the SEU cross section σSEU(En) shown in FIG. 3(b) is unknown.
[0037] Fig. 4(a) is a graph showing the neutron spectrum Φn(En) output from the radiation source 11 to the calculation unit 22. Fig. 4(b) is a graph showing the model function fSEU(En,p) set by the setting unit 21. As shown in Fig. 4(a), the calculation unit 22 receives input of neutron spectrum information similar to that shown in Fig. 3(a) above.
[0038] The calculation unit 22 calculates the SEU probability Rcalc(p) using the above-mentioned equation (2) based on the neutron spectrum Φn(En) shown in FIG. 4(a) and the model function fSEU(En,p) shown in FIG. 4(b). The model function shown in FIG. 4(b) is the model function set by the setting unit 21. As a result, for example, as shown in FIG. 4(c), the SEU probability Rcalc(p) corresponding to each of the radiation sources A1 to A4 is obtained. "Rcalc(p)" is the SEU probability that depends on the model parameter p.
[0039] The comparison unit 23 compares the measured values of the SEU occurrence probability due to each of the radiation sources A1 to A4 measured by the measurement unit 12 (graph in FIG. 3(c)) with the estimated values of the SEU occurrence probability due to each of the radiation sources A1 to A4 calculated by the calculation unit 22 (graph in FIG. 4(c)). The comparison unit 23 calculates the deviation between the measured value RSEU of the SEU occurrence probability and the estimated value Rcalc(p) of the SEU occurrence probability, and calculates a model parameter p so that the measured value and the estimated value match.
[0040] The comparison unit 23 feeds back the calculated model parameter p to the setting unit 21. The setting unit 21 changes the model function fSEU(En,p) based on the fed back model parameter p. As a result, it is possible to derive a model function fSEU(En,p) that approximates the unknown SEU cross section σSEU(En) shown in FIG. 3(b).
[0041] Furthermore, the setting unit 21 corrects the graph of the model function fSEU(En,p) based on the measured value of the SEU occurrence probability when the semiconductor device 13 is irradiated with protons using the radiation source 11 in the second mode described above (see FIG. 2(a)).
[0042] Specifically, the radiation source 11 irradiates the semiconductor device 13 with monoenergetic protons of spectrum Φp exceeding 20 MeV. It is known that in the energy band exceeding 20 MeV, the SEU cross section caused by neutrons is equivalent to the SEU cross section caused by protons. In this embodiment, the radiation source 11 is set to the second mode, the SEU probability RSEU using monoenergetic protons is measured, and a part of the model function fSEU(En,p) of the SEU cross section σSEU(En) is corrected.
[0043] FIG. 5 is an explanatory diagram showing an example of correcting the model function fSEU(En,p) based on the SEU probability when irradiated with monoenergetic protons. As shown in FIG. 5(a), a measured SEU probability is obtained when a semiconductor device 13 is irradiated with protons from radiation sources A1 and A2 with proton energies exceeding 20 MeV. For example, the data shown in FIG. 5(b) is obtained. The setting unit 21 corrects the model function fSEU(En,p) using the data for the SEU probability RSEU shown in FIG. 5(b). Specifically, the graph showing the relationship between neutron energy and the model function fSEU(En,p) shown in FIG. 5(c) is corrected using the data shown in FIG. 5(b).
[0044] That is, when neutrons are irradiated onto the semiconductor device 13 in the first mode described above, it is not possible to irradiate with a single energy, but rather neutrons containing a mixture of various energies are irradiated, which makes it difficult to estimate the degree of error caused by neutrons of which energy.
[0045] On the other hand, in the second mode described above, the semiconductor device 13 can be irradiated with monoenergetic protons of more than 20 [MeV], making it possible to measure the degree of error that occurs with protons of a specific energy. In this embodiment, the model function fSEU(En,p) is corrected using a measurement value of the SEU occurrence probability obtained when the semiconductor device 13 is irradiated with protons of a high level of energy (for example, 20 [MeV] or more), making it possible to derive a more reliable model function fSEU(En,p).
[0046] As described above, the estimation apparatus 100 according to this embodiment is an estimation apparatus 100 for estimating the SEU cross section of the semiconductor device 13, and includes a measurement unit 12 that counts the number of SEUs that occur when the semiconductor device 13 is irradiated with neutrons or protons emitted from the radiation source 11 and measures the SEU probability, a setting unit 21 that sets a model function simulating the SEU cross section and sets an arithmetic expression for calculating the SEU probability based on the model function, a calculation unit 22 that calculates an estimate of the SEU probability using the arithmetic expression based on spectrum information of the neutrons irradiated from the radiation source 11, and a comparison unit 23 that compares the measured and estimated SEU probability values and changes model parameters included in the model function so that the measured and estimated values match. The setting unit 21 changes the model function using the model parameters changed by the comparison unit 23, and corrects the model function based on the SEU probability measured by the measurement unit 12 when the semiconductor device 13 is irradiated with protons from the radiation source 11, and estimates the corrected model function as the SEU cross section.
[0047] In this embodiment, a model parameter p is set by comparing the measured value of the SEU occurrence probability with the estimated value, and a model function fSEU(En,p) is derived using this model parameter p. Furthermore, the model function fSEU(En,p) is corrected using the measured value of the SEU occurrence probability obtained when the semiconductor device 13 is irradiated with protons of high energy (for example, 20 MeV or more).
[0048] In this embodiment, by combining a neutron irradiation experiment (first mode) that can indirectly estimate a wide neutron energy range with a proton irradiation experiment (second mode) that can directly estimate a model function in a narrow energy range, it becomes possible to improve the estimation accuracy of the SEU cross section.
[0049] The estimating device 100 of the present embodiment described above can be, for example, a general-purpose computer system including a CPU (Central Processing Unit, processor) 901, a memory 902, a storage 903 (HDD: Hard Disk Drive, SSD: Solid State Drive), a communication device 904, an input device 905, and an output device 906, as shown in Fig. 6. The memory 902 and the storage 903 are storage devices. In this computer system, the CPU 901 executes a predetermined program loaded on the memory 902, thereby realizing each function of the estimating device 100.
[0050] The estimation device 100 may be implemented by one computer or by multiple computers, or may be a virtual machine implemented on a computer.
[0051] The program for the estimating device 100 can be stored in a computer-readable recording medium such as an HDD, an SSD, a Universal Serial Bus (USB) memory, a Compact Disc (CD), or a Digital Versatile Disc (DVD), or can be distributed via a network. The computer-readable recording medium is, for example, a non-transitory recording medium.
[0052] The present disclosure is not limited to the above-described embodiments, and various modifications are possible within the scope of the present disclosure. [Explanation of symbols]
[0053] 1. Experimental equipment 2 Computing device 11 Radiation source 12 Measuring part 13 Semiconductor Devices 14 Proton Accelerator 15 Neutron Source 21 Setting section 22 Calculation section 23 Comparison section 100 Estimator
Claims
1. An estimation apparatus for estimating an SEU cross section of a semiconductor device, comprising: a measurement unit that counts the number of SEUs generated when the semiconductor device is irradiated with neutrons or protons emitted from a radiation source, and measures the SEU generation probability; a setting unit that sets a model function that simulates the SEU cross section and sets an arithmetic expression that calculates the SEU occurrence probability based on the model function; a calculation unit that calculates an estimated value of the SEU occurrence probability using the formula based on spectrum information of neutrons irradiated from the radiation source; a comparison unit that compares the measured value and estimated value of the SEU occurrence probability and changes a model parameter included in the model function so that the measured value and estimated value match; Equipped with The setting unit changes the model function based on the model parameters changed by the comparison unit, corrects the model function based on the SEU occurrence probability measured by the measurement unit when the radiation source irradiates the semiconductor device with protons, and estimates the corrected model function as the SEU cross section. Estimation device.
2. A radiation source for irradiating neutrons and protons is provided, The radiation source irradiates the semiconductor device with protons having an energy of 20 [MeV] or more. The estimation device according to claim 1 .
3. A radiation source for irradiating neutrons and protons is provided, the radiation source includes a proton accelerator that irradiates monoenergetic protons, and a neutron source that generates neutrons when irradiated with the monoenergetic protons; When irradiating the semiconductor device with neutrons, protons irradiated from the proton accelerator are irradiated onto the neutron source to generate neutrons. The estimation device according to claim 1 .
4. 1. A method for estimating an SEU cross section of a semiconductor device, comprising: a measurement unit counting the number of SEUs generated when the semiconductor device is irradiated with neutrons or protons emitted from a radiation source, and measuring the SEU generation probability; a setting unit setting a model function that simulates the SEU cross section and setting an arithmetic expression that calculates the SEU occurrence probability based on the model function; a calculation unit that calculates an estimated value of the SEU occurrence probability using the formula based on spectrum information of neutrons irradiated from the radiation source; a comparison unit that compares the measured value and the estimated value of the SEU occurrence probability and changes a model parameter included in the model function so that the measured value and the estimated value match; The setting unit changes the model function based on the changed model parameters, and corrects the model function based on the SEU occurrence probability measured by the measurement unit when the radiation source irradiates the semiconductor device with protons, and estimates the corrected model function as the SEU cross section. Estimation method.
5. A program that causes a computer to function as the estimation device according to claim 1.