Method for acquiring single event upset cross section and sensitive region of device based on machine learning

Abstract

The invention discloses a method for acquiring a single event upset cross section and a sensitive area of a device based on machine learning, and relates to the technical field of semiconductor analysis, and the system discloses the following steps: S1, simulating a heavy ion single event effect device of a target chip, and acquiring a machine learning sample with a single event upset label; s2, taking the machine learning sample with the single event upset label as the input of a machine learning classification algorithm, and determining a decision boundary where single event upset occurs; and S3, mapping the single event upset decision boundary extracted by the machine learning algorithm to the target chip to obtain the single event effect sensitive region of the chip. The method for acquiring the single event upset cross section and the sensitive area of the device is provided by combining simulation data of the single event effect device, rapid evaluation of the sensitivity of the heavy ion single event effect of the chip can be achieved through the method, and the test cost is greatly saved.

Description

Technical Field

[0001] The present invention relates to the field of semiconductor analysis technology, and more specifically, to a method for acquiring a single particle flip cross section and a sensitive area of ​​a device based on machine learning. Background Art

[0002] Semiconductor devices used in space applications are affected by high-energy particles released from the space radiation environment. When a single high-energy particle hits a semiconductor device, it causes the device's logic state to flip, current and voltage to perturb, and even damage the device. This radiation effect is called a single-particle effect. Common single-particle effects include single-particle flips, single-particle latches, single-particle transients, and single-particle burnouts. According to statistics, on-orbit failures of spacecraft caused by radiation effects account for about 45% of total failures, of which single-particle effects account for nearly 86%. Therefore, single-particle effects are an issue that requires special attention in the space application of semiconductor devices.

[0003] In order to evaluate the space adaptability of semiconductor devices, it is usually necessary to conduct heavy ion accelerated irradiation tests to obtain the single particle effect cross section of semiconductor devices and then calculate their on-orbit error rate. However, the cost of heavy ion accelerated irradiation tests is very high (20,000 yuan / hour for the test machine) and difficult to operate (it requires the design of a dedicated test system and special sample processing). Summary of the invention

[0004] In view of the deficiencies in the prior art, the object of the present invention is to provide a method for acquiring the single particle upset cross section and sensitive area of ​​a device based on machine learning.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] The method for obtaining the single event upset cross section and sensitive area of ​​the device based on machine learning includes the following steps:

[0007] Step S1: heavy ion single event effect device simulation of the target chip to obtain machine learning samples with single event upset labels;

[0008] Step S2: using the machine learning sample with the SEP label as the input of the machine learning classification algorithm to determine the decision boundary of the SEP;

[0009] Step S3: The single event upset decision boundary extracted by the machine learning algorithm is mapped to the target chip to obtain the single event effect sensitive area of ​​the chip.

[0010] Furthermore, the heavy ion single particle effect device of the target chip is simulated to obtain machine learning samples with single particle flip labels. Specifically, a three-dimensional TCAD simulation structure of the target chip is constructed. The ions are incident from the upper surface of the device in a direction perpendicular to the YZ plane. The range of the incident ions is constant. The characteristic quantities of the incident ions are set according to the characteristics of the single particle effect induced by heavy ions in space, including the incident position (y, z axis coordinates) and the secondary ion LET value. Based on the simulation results, it is evaluated whether the corresponding incident ions have caused a single particle flip. By conducting single particle effect device simulations with different combinations of characteristic quantities, machine learning samples with single particle flip labels are obtained.

[0011] Furthermore, the upper surface of the device corresponds to a plane with coordinate X=0.

[0012] Furthermore, the range of the incident ions is set to cover the thickness of the active region of the device.

[0013] Furthermore, the machine learning samples with single particle flip labels are used as input of the machine learning classification algorithm to determine the decision boundary of the single particle flip. Specifically, corresponding to each LET value, all points with corresponding (yi, zi) coordinates are marked in the yz plane, and the points where the single particle flip occurs are represented by “·”, and the points where the single particle flip does not occur are represented by “X”. The machine learning classification algorithm is used to divide the data points in the yz plane into two sets according to whether the single particle flip occurs, and the decision boundary is determined.

[0014] Furthermore, the area inside the decision boundary is the region where the single particle upset occurs, and the area outside the decision boundary is the region where the single particle upset does not occur.

[0015] Furthermore, the area within the decision boundary at a certain LET value is the single particle upset cross section corresponding to the LET value.

[0016] Compared with the prior art, the present invention has the following beneficial effects:

[0017] The present invention combines the simulation data of single particle effect devices and proposes a method for obtaining the single particle flip cross section and sensitive area of ​​the device. The method of the present invention can realize the rapid evaluation of the sensitivity of the chip to the heavy ion single particle effect, which greatly saves the test cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a flowchart of the present invention;

[0019] Figure 2 A display diagram of the three-dimensional simulation structure of the target chip and the coordinates of the incident position;

[0020] Figure 3 A collection display diagram is provided to show whether a single-particle upset has occurred;

[0021] Figure 4 This is a diagram showing the single-particle effect sensitive area of ​​the chip. DETAILED DESCRIPTION

[0022] Reference Figure 1-Figure 4 , a method for obtaining a device single-event upset cross section and a sensitive area based on machine learning includes the following steps:

[0023] Step S1: heavy ion single event effect device simulation of the target chip to obtain machine learning samples with single event upset labels;

[0024] Step S11: Conduct single-particle effect device simulation of the target chip. Construct a three-dimensional TCAD simulation structure of the target chip; ions are incident from the upper surface of the device (corresponding to the plane with coordinate X=0) in a direction perpendicular to the YZ plane, and the range of the incident ions is constant (the range is set to cover the thickness of the active area of ​​the device). According to the characteristics of the single-particle effect induced by heavy ions in space, the characteristic quantities of the incident ions are set, including the incident position (y, z axis coordinates) and the secondary ion LET value, and based on the simulation results, it is evaluated whether the corresponding incident ions cause a single-particle upset. By conducting single-particle effect device simulations with different combinations of characteristic quantities, machine learning samples with single-particle upset labels are obtained.

[0025]

[0026] Table 1 Labeled machine learning samples

[0027] Step S2: using the machine learning sample with the SEP label as the input of the machine learning classification algorithm to determine the decision boundary of the SEP;

[0028] Step S21: For each LET value, mark all points with corresponding (yi, zi) coordinates in the yz plane. Points with single particle upsets are represented by "·", and points without single particle upsets are represented by "X". Use a machine learning classification algorithm to divide the data points in the yz plane into two sets according to whether a single particle upset occurs, and determine the decision boundary. The area within the decision boundary is the area where a single particle upset occurs, and the area outside the decision boundary is the area where a single particle upset does not occur;

[0029] Step S3: Map the SEP decision boundary extracted by the machine learning algorithm onto the target chip to obtain the SEP sensitive area of ​​the chip. The area within the decision boundary (SEP sensitive area) at a certain LET value is the SEP cross section corresponding to that LET value.

[0030] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0031] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented by software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website site, computer, server or data center to another website site, computer, server or data center by wired or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state hard disk.

[0032] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0033] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0034] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0035] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0036] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage media include: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories (ROM), random access memories (RAM), magnetic disks or optical disks.

[0037] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A method for obtaining the single event upset cross section and sensitive area of ​​a device based on machine learning, characterized in that: The steps include: Step S1: heavy ion single event effect device simulation of the target chip to obtain machine learning samples with single event upset labels; Step S2: using the machine learning sample with the SEP label as the input of the machine learning classification algorithm to determine the decision boundary of the SEP; Step S3: The single event upset decision boundary extracted by the machine learning algorithm is mapped to the target chip to obtain the single event effect sensitive area of ​​the chip.

2. The method for obtaining the single event upset cross section and sensitive area of ​​a device based on machine learning according to claim 1, characterized in that: The heavy ion single particle effect device of the target chip is simulated to obtain machine learning samples with single particle flip labels. Specifically, a three-dimensional TCAD simulation structure of the target chip is constructed. The ions are incident from the upper surface of the device in a direction perpendicular to the YZ plane. The range of the incident ions is constant. The characteristic quantities of the incident ions are set according to the characteristics of the single particle effect induced by heavy ions in space, including the incident position (y, z axis coordinates) and the secondary ion LET value. Based on the simulation results, it is evaluated whether the corresponding incident ions have caused a single particle flip. By conducting single particle effect device simulations with different combinations of characteristic quantities, machine learning samples with single particle flip labels are obtained.

3. The method for obtaining the single event upset cross section and sensitive area of ​​a device based on machine learning according to claim 2, characterized in that: The upper surface of the device corresponds to the plane with coordinate X=0.

4. The method for obtaining the single event upset cross section and sensitive area of ​​a device based on machine learning according to claim 2, characterized in that: The range of the incident ions is set to cover the thickness of the active region of the device.

5. The method for obtaining the single event upset cross section and sensitive area of ​​a device based on machine learning according to claim 1, characterized in that: The machine learning samples with single particle upset labels are used as the input of the machine learning classification algorithm to determine the decision boundary of the single particle upset. Specifically, for each LET value, all points with corresponding (yi, zi) coordinates are marked in the yz plane. Points where single particle upsets occur are represented by "·", and points where single particle upsets do not occur are represented by "X". The machine learning classification algorithm is used to divide the data points in the yz plane into two sets according to whether single particle upsets occur, and the decision boundary is determined.

6. The method for obtaining the single event upset cross section and sensitive area of ​​a device based on machine learning according to claim 5, characterized in that: The area inside the decision boundary is the region where single-particle upset occurs, and the area outside the decision boundary is the region where single-particle upset does not occur.

7. The method for obtaining the single event upset cross section and sensitive area of ​​a device based on machine learning according to claim 1, characterized in that: The area within the decision boundary at a certain LET value is the single particle upset cross section corresponding to that LET value.

Images