A method for estimating the reference range of single event upset rate of devices on orbit

Abstract

The present invention discloses a method for estimating the reference range of the single event upset rate of a device on orbit. The method obtains the experimental data of the single event upset cross section (σ) and the incident ion parameter (LET) of the device by conducting a ground heavy ion irradiation test; the single event upset saturation cross section (σ sat ), single event upset LET threshold (LET th ), the depth of the device sensitive area (d), the length of the device funnel (F) and other parameters are estimated; each parameter is adjusted within the estimated range, fitted using the Weibull curve, and the Monte Carlo simulation tool is used to carry out simulation to obtain the on-orbit single-particle flip rate; finally, the relationship between each parameter and the average figure of merit of the on-orbit flip rate is obtained, thereby determining the reference range of the on-orbit single-particle flip rate of the device; this method can obtain the reference range of the on-orbit single-particle flip rate of devices, especially commercial devices, and effectively guide the selection of aerospace devices.

Description

Technical Field

[0001] The present invention relates to a method for estimating a reference interval of an on-orbit single-particle upset rate of a device, and belongs to the technical field of verification of the ability of aerospace integrated circuits to resist single-particle effects in space. Background Art

[0002] The space radiation environment contains a large number of particles (electrons, neutrons, protons and heavy ions) and rays. These radiation particles interact with semiconductor devices and produce single-particle effects, causing the semiconductor device state to be disturbed or permanently fail, thereby inducing spacecraft failure. Before aerospace devices are used in aerospace equipment, their single-particle resistance performance needs to be evaluated on the ground. On the ground, heavy ion accelerators are usually used to simulate single-particle effects and obtain single-particle upset error rates under specific space radiation conditions. As the device process size decreases, factors such as charge sharing and incident angle make it very difficult to estimate the on-orbit single-particle upset rate. Currently, the following problems exist in the calculation of the on-orbit single-particle upset rate of aerospace components:

[0003] (1) Single event upset saturation cross section (σ sat ), single event upset LET threshold (LET th ) and other parameters become more difficult to obtain, and the accuracy of the obtained data is not high.

[0004] (2) It is generally believed that the σ-LET curve follows the Weibull distribution. Currently, the fitting of single-particle Weibull curves mostly adopts the visual fitting method, and the curve fitting accuracy is not high.

[0005] (3) To obtain the depth of the sensitive area and the length of the funnel, it is necessary to dissect the device specifically, obtain the device cross-section, and then combine the process reverse modeling calibration method to obtain the preliminary range.

[0006] Therefore, it is necessary to develop a method to estimate the reference range of the on-orbit single-particle upset rate of a device. This method can reduce the difficulty of obtaining key parameters, improve the fitting accuracy of the single-particle Weibull curve, and ultimately obtain the accurate on-orbit single-particle upset rate reference range of aerospace components, effectively guiding the selection of aerospace components. Summary of the Invention

[0007] The technical problem solved by the present invention is: to overcome the shortcomings of the existing technology, to provide a method for estimating the reference range of the on-orbit single-particle upset rate of a device, thereby reducing the difficulty of obtaining key parameters and eliminating the need to pay excessive attention to the internal parameter information of the device. By using the Monte Carlo simulation tool, the sensitivity of different parameter changes to the single-particle upset rate of the device can be verified, thereby making it more convenient to obtain the on-orbit single-particle upset rate reference range of the device, especially commercial devices, and effectively guiding the selection of aerospace devices.

[0008] The technical solution of the present invention is:

[0009] A method for estimating a reference range of an on-orbit single event upset rate of a device comprises the following steps:

[0010] (1) Conduct ground-based heavy ion irradiation tests to obtain experimental data on the device's single-event upset cross section σ and incident ion parameter LET;

[0011] (2) Estimation of the single event upset saturation cross section σ sat , single event upset threshold LET th , the range of the device sensitive area depth d, and the device funnel length F;

[0012] (3) Each parameter is adjusted within the estimated range, and the single-particle Weibull curve is fitted using an appropriate estimation method. Finally, a Monte Carlo simulation tool is used to perform simulations to obtain the single-particle on-orbit flip rate and the average figure of merit of the on-orbit flip rate of the device under specific spatial conditions.

[0013] (4) Obtain the relationship between each parameter change and the average figure of merit of the on-orbit flip rate, so as to determine the reference range of the device's on-orbit single-particle flip rate.

[0014] The specific implementation of step (1) is as follows:

[0015] (2.1) Determine the ion type to be tested by referring to the single-particle test data of the device that is closest to the estimated device structure and process;

[0016] (2.2) Develop a single-event experimental system for the device to be evaluated and conduct single-event upset experiments on a heavy ion accelerator;

[0017] (2.3) Count the number of single-element upsets and the total incident ion dose, and obtain the data of the single-element upset cross section and the incident ion parameter LET.

[0018] In step (2.3), the calculation formula for the single-event upset cross section σ(i) under the incident condition of the i-th ion is:

[0019] σ(i)=N(i) / φ(i)

[0020] Where φ(i) is the total injection of the i-th ion incident vertically on the device surface;

[0021] N(i) is the number of single event upsets in the case of the i-th ion incident.

[0022] The implementation of step (2) is as follows:

[0023] (3.1) In the ground-based heavy ion irradiation test, the incident ions should cover the range from the ions with the smallest LET value that triggers the single event upset of the device to the ions with the largest LET value that reaches saturation of the single event upset cross section. The minimum LET value should be reasonably set, and the single event upset LET threshold should be 0. <LET th <The minimum LET value set;

[0024] (3.2) Single event upset saturation cross section -5σ max <σ sat <5σ max , σ max is the single-event upset cross section when the LET of the incident ion is maximum;

[0025] (3.3) Based on the structural parameters and process parameters of the device to be estimated, the device simulation tool TCAD is used to determine the range of the device sensitive area depth and device funnel length.

[0026] The specific implementation steps of step (3) are:

[0027] (4.1) The single-event upset saturation cross section and the single-event upset LET threshold are adjusted within the estimated range, and the single-event Weibull curve parameters are obtained by performing Weibull fitting using the least squares method;

[0028] The specific form of the single-particle Weibull curve is:

[0029] σ(LET)=σ sat (1-exp{-[(LET-LET th ) / W] S})

[0030] σ(LET) represents the single-element upset cross section at the LET value, W represents the width parameter of the Weibull curve, and S represents the shape parameter of the Weibull curve;

[0031] After taking the natural logarithm twice on both sides of the above formula, we can get:

[0032] ln(LET-LET th )=(1 / S)×ln{ln[σ sat / (σ sat -σ(LET))]}+ln(W)

[0033] So it becomes a linear equation y=mx+b, where

[0034] y=ln(LET-LET th )

[0035] m=1 / S

[0036] x=ln{ln[σ sat / (σsat -σ(LET))]}

[0037] b=ln(W)

[0038] By performing linear regression analysis based on the least squares method and finding the slope m and intercept b in the linear equation, the width parameter W and shape parameter S of the Weibull curve can be found.

[0039] (4.2) Adjust the depth of the sensitive area and the length of the device funnel within the estimated range, and combine the parameters of the obtained single-particle Weibull curve to obtain all the parameters required for Monte Carlo simulation;

[0040] (4.3) Simulate in Monte Carlo simulation tool to obtain the single particle on-orbit flip rate under specific spatial conditions.

[0041] In step (3), the calculation method of the average merit of the on-track flip rate MR is:

[0042]

[0043] N is the total number of single-particle on-track flip rates obtained in step (3), Rate n is the on-orbit flipping rate of the nth single particle.

[0044] The specific implementation steps of step (4) are:

[0045] (1) Plot the change of each parameter and the average figure of merit of the on-orbit flip rate into a line graph;

[0046] (2) Analyze the line graph data to obtain the degree of influence of each parameter on the single-particle flip rate, and then determine the best and worst cases of the single-particle flip rate based on the sensitive parameters, and finally obtain the reference range of the on-orbit single-particle flip rate.

[0047] Compared with the prior art, the present invention has the following beneficial effects:

[0048] (1) The present invention uses experimental data to calculate the single-event upset saturation cross section (σ sat ), single event upset LET threshold (LET th ) was used to estimate the reference interval, and then the single-particle test data were fitted with a Weibull curve using the least squares method. The single-particle Weibull curve finally obtained had a good fitting degree and high confidence.

[0049] (2) The present invention uses Monte Carlo simulation tools to obtain the effects of different parameters on the sensitivity of single-particle flip rate, rather than focusing on the physical process of energy transmission after single-particle incidence.

[0050] (3) The present invention does not require high precision of the internal structure and process parameters of the device, and can obtain the reference range of the on-orbit single-particle upset rate of the device, especially the commercial device, to effectively guide the selection of aerospace devices. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 It is a schematic flow chart of the present invention;

[0052] Figure 2 is a single particle Weibull curve according to an embodiment of the present invention;

[0053] Figure 3 is a curve showing the variation of variable key parameters with the average figure of merit of error rate according to an embodiment of the present invention; DETAILED DESCRIPTION

[0054] Example:

[0055] The implementation steps of the method for determining the reference interval of the on-track single event upset rate of a 28nm SRAM memory circuit of the present invention are as follows:

[0056] 1. Conduct ground-based heavy ion irradiation tests to obtain experimental data on the device's single-event upset cross section (σ) and incident ion parameter (LET);

[0057] (1) Refer to the single-particle test data of the device that is closest to the target device structure and process to determine the type of ions to be selected. The selected ions should not only include the ions at which the target device just begins to experience a single-particle upset, but also include the ions at which the single-particle upset reaches the single-particle upset saturation cross section. Generally, no less than five data points need to be obtained. For example, Xilinx's Artix 7FPGA also uses a 28nm CMOS process. After reviewing the published literature, it was found that its internal CRAM structure still experiences a single-particle upset when C ions (LET value is 1.73) are incident. Therefore, it can be inferred that the single-particle upset LET threshold of the device under test is also less than 3MeV·cm2 / mg. Combined with the actual situation of domestic accelerators, the selected ions are shown in Table 1.

[0058] Table 1 Related parameters of selected ions

[0059]

[0060] (2) Develop a single-particle test system and conduct single-particle upset tests on a heavy ion accelerator.

[0061] For example, in a memory circuit single-event test system, we use a dynamic read / write mode to acquire single-event upset data. During device irradiation, we periodically perform a test pattern writing, waiting, reading, and comparison process. The test pattern can be all "0s," all "1s," 55, AA, a slanted triangle, and other data. However, to mask the impact of different patterns on single-event upsets, the same pattern should be used for each test within the same set of tests.

[0062] (3) Count the number of single-event upsets and the total incident ion flux to obtain the data of the single-event upset cross section and the incident ion LET. The data statistics are shown in Table 2.

[0063] Table 2 Single event upset data statistics

[0064]

[0065] Among them, the calculation formula of the single-particle upset cross section is

[0066] σ(i)=N(i) / φ(i)

[0067] Where φ(i) is the total injection of the i-th ion incident vertically on the device surface,

[0068] N(i) is the number of single event upsets in the case of ion incidence in the i-th case,

[0069] σ(i) is the single-event upset cross section for the i-th ion incident case.

[0070] 2. Single event upset saturation cross section (σ sat ), single event upset LET threshold (LET th ), device sensitive area depth (d), device funnel length (F) and other parameters are estimated;

[0071] The SEE saturation cross section and SEE LET threshold are key parameters for Weibull fitting and must be obtained experimentally. However, due to inherent errors in SEE irradiation test results and the limited number of data points obtained due to machine time constraints, it is impossible to obtain precise SEE saturation cross section and SEE LET threshold values. Therefore, we estimate the range of parameter variation and divide the variable parameters into different levels within the possible range.

[0072] The single event upset LET threshold is generally greater than 0. By properly setting the minimum LET value (B ion), the estimation accuracy can be improved. For example, for 28nm memory, our threshold estimation range is 0 to 1MeV·cm 2 / mg, the estimated accuracy is already very high;

[0073] The maximum LET of space ions is 120 MeV·cm 2 / mg, so we typically select ions with the highest LET values ​​available from domestic accelerators, such as Ta ions, or increase the LET value by varying the incident angle. For example, for 28nm memory, based on the overall trend of the experimental curve, we found that the SEE cross section approaches saturation under Ta ion incidence. Therefore, we estimate the SEE saturation cross section to be 0.3, 0.5, 1, 1.5, 2, and 5 times the SEE cross section of Ta ions, respectively.

[0074] The device sensitive area depth and device funnel length are key parameters for Monte Carlo simulation based on IRPP. There are several problems in determining these two parameters:

[0075] (1) Usually, we know very little about the internal parameters of the device, especially for commercial devices. We need to dissect the device, obtain the device cross-section, and then combine it with the process reverse modeling calibration method to obtain it;

[0076] (2) Some literature suggests that the funnel effect can accelerate charge collection and increase the probability of single-particle upsets. However, some researchers do not take the funnel effect into account when making predictions. This means that even if accurate parameters are obtained, the flip rate estimation may not be accurate.

[0077] (3) Usually we regard the depletion region as the sensitive region. The Space Radiation software recommends setting the thickness of the sensitive region to 1μm and checking the differences caused by different thicknesses of the sensitive region.

[0078] Based on the above conclusions and relevant data, we estimate that the depth of the device's sensitive area is between 0.1μm and 2μm. The device funnel length is set to vary from 0.01μm to 1μm.

[0079] The estimated parameter range is shown in Table 3. The variable parameters are divided into 6 levels to cover all possible ranges.

[0080] Table 3 Estimated parameter ranges

[0081] Simulation estimated parameters Level 1 Level 2 Level 3 Level 4 Level 5 Level 6 Depth of device sensitive area (μm) 0.1 0.2 0.5 0.8 1 2 Device funnel length (μm) 0 0.1 0.2 0.5 0.8 1 <![CDATA[Single event upset saturation cross section (cm 2 )]]> 0.3X 0.5X 1X 1.5X 2X 5X <![CDATA[Single Event Upset LET Threshold (MeV.cm 2 / mg)]]> 0.01 0.05 0.1 0.2 0.5 1

[0082] Where X represents the multiple of the Ta ion incident cross section, and 0.3X represents 0.3 times the Ta ion incident cross section.

[0083] 3. Adjust each parameter within the estimated range, use the Weibull curve for fitting, and finally use the Monte Carlo simulation tool to carry out simulation.

[0084] (1) First, the single-particle upset saturation cross section and the single-particle upset LET threshold are adjusted within the estimated range, and the Weibull fitting is performed using the least squares method to obtain the single-particle Weibull curve parameters.

[0085] The specific form of the single-particle Weibull curve is:

[0086] σ(LET)=σ sat (1-exp{-[(LET-LET th ) / W] S})

[0087] After taking the natural logarithm twice on both sides of the equation, we get:

[0088] ln(LET-LET th )=(1 / S)×ln{ln[σ sat / (σ sat -σ(LET))]}+ln(W)

[0089] So it becomes a linear equation y=mX+b, where

[0090] y=ln(LET-LET th )

[0091] m=1 / S

[0092] x=ln{ln[σ sat / (σ sat -σ(LET))]}

[0093] b=ln(W)

[0094] We can perform linear regression analysis based on the least squares method to find the slope m and intercept B in the linear equation, and then find the width parameter W and shape parameter S of the Weibull curve. At the same time, we can calculate the linear correlation coefficient r value that describes the quality of the linear fit. The closer r is to 1, the better the curve fit is. It is generally considered that the curve is strongly correlated when it is greater than 0.8. The single-particle upset threshold we selected is 0.1MeV.cm 2 / mg, the single particle upset saturation cross section is 1X, i.e. 1.43E-03cm 2 / device, the calculated S is 1.687 and W is 24.783MeV.cm 2 / mg. The fitting parameter R is 0.9357, indicating that the curve fit is very good. Figure 2 shown.

[0095] (2) Adjust the sensitive area and the length of the device funnel within the estimated range. For example, we set the sensitive area to 0.2 μm and the length of the device funnel to 0.5 μm. Combined with the parameters of the single-particle Weibull curve obtained in the previous step, all the parameters required for Monte Carlo simulation are obtained, as shown in Table 3.

[0096] Table 2 All parameters required for Monte Carlo simulation

[0097]

[0098] (3) Perform simulation in the Monte Carlo simulation tool to obtain the single-particle on-orbit flip rate under specific spatial conditions.

[0099] For example, Space Radiation software, a comprehensive tool for calculating the space and atmospheric radiation environment and radiation effects of spacecraft and aircraft, is currently used by most US aerospace companies for radiation environment and effects analysis. Using a GEO orbit, 90% worst case, and equivalent 3mm aluminum shielding, and substituting the parameters shown in Table 2 in the Weibull model, the single-event upset rate (SER) was calculated to be 5.15E-3 times / device-day.

[0100] 4. Obtain the relationship between each parameter change and the average merit of the on-orbit flip rate, so as to determine the reference range of the device's on-orbit single-event flip rate

[0101] (1) The change of each parameter and the average merit of the flip rate are plotted into a broken line graph. According to our classification of parameters, there are 1296 possible combinations of variable parameters according to the level classification. Then, the error rate of each point in the broken line graph is calculated by 216 single particle flip cross sections. The calculation method of the average merit of the flip rate is N is the total number of single-particle on-orbit flip rates obtained, Rate n is the on-orbit flipping rate of the nth single particle.

[0102] The variation of variable parameters with the average merit of on-orbit flip rate is as follows: Figure 3 shown.

[0103] (2) By analyzing the line graph data, the reference range of the on-orbit single-particle flip rate can be obtained.

[0104] For example, as the single-particle upset saturation cross section increases, the on-orbit single-particle upset rate increases. Therefore, based on the impact of changes in different parameters on the sensitivity of the on-orbit single-particle upset rate, a single-particle upset rate reference range can be obtained. The parameter that is most sensitive to the single-particle upset rate can also be obtained based on the changes in the parameters. Subsequently, a more detailed analysis of this parameter can be performed to obtain more accurate conclusions.

[0105] Figure 1It is a schematic flow chart of the present invention.

[0106] Parts of the present invention that are not described in detail belong to common knowledge among those skilled in the art.

Claims

1. A method for estimating the reference interval of the single event upset rate of a device on orbit, characterized in that The following steps are involved: Step 1: Conduct ground-based heavy ion irradiation tests to obtain experimental data on the device's single-event upset cross section σ and incident ion parameter LET; Step 2: Estimate the single event upset saturation cross section σ sat , single event upset threshold LET th , the range of the device sensitive area depth d, and the device funnel length F; Step 3: Adjust each parameter within the estimated range, use appropriate estimation methods to fit the single-particle Weibull curve, and finally use Monte Carlo simulation tools to perform simulations to obtain the single-particle on-orbit flip rate and the average figure of merit of the on-orbit flip rate of the device under space conditions. Step 4: Obtain the relationship between each parameter change and the average figure of merit of the on-orbit upset rate, thereby determining the reference range of the device's on-orbit single event upset rate; The specific implementation steps of step 3 are: Step 3.1: Adjust the SEE saturation cross section and SEE LET threshold within the estimated range, and use the least squares method to perform Weibull fitting to obtain the SEE Weibull curve parameters. The specific form of the single-particle Weibull curve is: σ(LET)=σ sat (1-exp{-[(LET-LET th ) / W] S }) σ(LET) represents the single-element upset cross section at the LET value, W represents the width parameter of the Weibull curve, and S represents the shape parameter of the Weibull curve; After taking the natural logarithm twice on both sides of the above formula, we can get: ln(LET-LET th )=(1 / S)×ln{ln[σ sat / (σ sat -σ(LET))]}+ln(W) So it becomes a linear equation y=mx+b, where y=ln(LET-LET th ) m=1 / S x=ln{ln[σ sat / (s sat -σ(LET))]} b=ln(W) By performing linear regression analysis based on the least squares method and finding the slope m and intercept b in the linear equation, the width parameter W and shape parameter S of the Weibull curve can be found. In step 3.2, the depth of the sensitive area and the length of the device funnel are adjusted within the estimated range. Combined with the parameters of the obtained single-particle Weibull curve, all the parameters required for Monte Carlo simulation are obtained. Step 3.3: Perform simulation in the Monte Carlo simulation tool to obtain the single-particle on-orbit flip rate under space conditions.

2. The method for estimating a reference range of an on-orbit single event upset rate of a device according to claim 1, characterized in that: The specific implementation of step 1 is as follows: Step 1.1 Determine the ion type to be tested by referring to the single-particle test data of devices with similar structure and process to the estimated device; Step 1.2: Develop a single-event experimental system for the device to be evaluated and conduct single-event upset tests on a heavy ion accelerator. Step 1.3 counts the number of single-event upsets and the total incident ion flux to obtain data on the single-event upset cross section and the incident ion parameter LET.

3. The method for estimating a reference range of an on-orbit single event upset rate of a device according to claim 2, characterized in that: In step 1.3, the calculation formula for the single event upset cross section σ(i) under the case of the i-th ion incident is σ(i) = N(i) / φ(i) Where φ(i) is the total injection of the i-th ion incident vertically on the device surface; N(i) is the number of single event upsets in the case of the i-th ion incident.

4. The method for estimating a reference range of an on-orbit single event upset rate of a device according to claim 1, wherein: The implementation of step 2 is as follows: Step 2.1 In the ground heavy ion irradiation test, the incident ions should cover the range from the ions with the smallest LET value that trigger the single event upset of the device to the ions with the largest LET value that reaches saturation of the single event upset cross section. The minimum LET value should be reasonably set, and the single event upset LET threshold should be 0. <LET th <The minimum LET value set; Step 2.2 Single event upset saturation cross section -5σ max <σ sat <5σ max , σ max is the single-event upset cross section when the LET of the incident ion is maximum; In step 2.3, based on the structural parameters and process parameters of the device to be estimated, the device simulation tool TCAD is used to determine the range of the device sensitive area depth and the device funnel length.

5. The method for estimating a reference range of an on-orbit single event upset rate of a device according to claim 1, characterized in that: In step 3, the calculation method of the average merit of the on-orbit flip rate MR is: N is the total number of single-particle on-orbit flip rates obtained in step 3, Rate n is the on-orbit flipping rate of the nth single particle.

6. The method for estimating a reference range of an on-orbit single event upset rate of a device according to claim 5, characterized in that: The specific implementation steps of step 4 are: Step 4.1: Plot the change of each parameter and the average figure of merit of the on-orbit flip rate into a line graph; Step 4.2 analyzes the line graph data to obtain the degree of influence of each parameter on the single-event upset rate. Then, based on the sensitive parameters, the optimal and worst-case scenarios of the single-event upset rate are determined, and finally the reference range of the on-orbit single-event upset rate is obtained.

Images