CN113239602A

CN113239602A - SRAM single event upset cross section prediction method based on Monte Carlo method

Info

Publication number: CN113239602A
Application number: CN202110336380.6A
Authority: CN
Inventors: 刘红侠; 鲍一豪; 王树龙; 陈树鹏
Original assignee: Xidian University
Current assignee: Xidian University
Priority date: 2021-03-29
Filing date: 2021-03-29
Publication date: 2021-08-10

Abstract

The invention discloses a SRAM single event upset section prediction method based on a Monte Carlo method, which comprises the following steps: establishing a device model; establishing an SRAM circuit according to the device model; uniformly dividing a device model of the SRAM circuit along the length direction to establish a composite sensitive body; acquiring LET thresholds of the SRAM circuit at different incidence positions and different incidence depths of the composite sensitive body to enable the SRAM circuit to turn over; based on an LET threshold value, according to a preset number of incident particles, the incident particles are incident to the composite sensitive body to obtain a particle number ratio for enabling the SRAM circuit to turn over; and obtaining a turnover cross section distribution transfer curve of the device model under different incident energies according to the population ratio of the SRAM circuit to be turned. The invention carries out the single-particle Monte Carlo simulation of the device and the method for measuring the turning section, simulates the interaction among particles, calculates the nuclear physical processes such as energy absorption, collision and the like, completes the simulation of the random process of the particles, and completes the turning section statistics of the random process by carrying out multi-particle bombardment.

Description

SRAM single event upset cross section prediction method based on Monte Carlo method

Technical Field

The invention belongs to the technical field of radiation resistance of semiconductor devices, and particularly relates to a SRAM single event upset cross section prediction method based on a Monte Carlo method.

Background

After the 21 st century, more and more electronic systems have been used in space, including military, civil, commercial, and other fields. Therefore, the research and development of electronic devices and overall systems with high stability become a large research target in the space irradiation direction in China. In the space environment, however, the operation of electronic devices can be affected by the exposure of space. In 1962, Wallmark and Marcus both pointed out that devices failed during their manufacture and operation, guessing that device operation failure could be due to both temporary and permanent spatial radiation-induced failures, and they thought that electronic device sensitivity to spatial radiation would create a number of problems in the future. In fact, in fifteen years from 1971 to 1986, 1589 faults of total thirty more artificial geosynchronous orbit satellites launched by the countries of cosu et al occurred, and in such a huge number of faults, the faults caused by space irradiation account for as much as 71% of the total number, so that researchers in various countries know that new requirements should be made on space electronic systems from different angles.

Space irradiation environments are complex and contain significant quantities of different particles including protons, neutrons and cosmic rays, which, upon bombardment into the device, undergo physical processes such as nuclear reactions with the target material. The most important Effects are Single Event Effects (SEE) and total Dose Effects (TID). The single-particle effect refers to a radiation effect which causes abnormal changes of the state of a microelectronic device when a single high-energy particle passes through a sensitive area of the device, and includes single-particle overturn, single-particle locking, single-particle burning and the like.

However, the existing software for researching the single event effect mainly comprises Sentaurus TCAD simulation software, but the software is mainly used for simulating the contents of the electrical characteristics, the device process and the like of the device, and the simulation of the single event bombardment effect on the device in the real space is not accurate and intuitive enough, and the measurement effect on the turnover section is not good.

Disclosure of Invention

In order to solve the problems in the prior art, the invention provides a SRAM single event upset cross section prediction method based on a Monte Carlo method. The technical problem to be solved by the invention is realized by the following technical scheme:

a SRAM single event upset cross section prediction method based on a Monte Carlo method comprises the following steps:

establishing a device model;

establishing an SRAM circuit according to the device model;

uniformly dividing a device model of the SRAM circuit along the length direction to establish a composite sensitive body comprising a plurality of sensitive areas;

acquiring LET thresholds of the SRAM circuit, which enable the SRAM circuit to turn over at different incidence positions and different incidence depths of the composite sensitive body;

based on the LET threshold, according to a preset number of incident particles, the composite sensitive body is incident to obtain a particle number ratio for enabling the SRAM circuit to turn over;

and obtaining a turnover cross section distribution transfer curve of the device model under different incident energies according to the population ratio of the SRAM circuit to be turned.

In one embodiment of the present invention, building a device model includes:

and establishing a device model by using an SDE module of a Sentaurus TCAD simulation tool.

In one embodiment of the present invention, building an SRAM circuit from the device model comprises:

defining a device model used by each transistor of the SRAM circuit using an SDEVICE module of a Sentaurus TCAD simulation tool;

performing input and output setting on the defined device model by using the SDEVICE module;

connecting all the device models with input and output settings to build the SRAM circuit.

In one embodiment of the present invention, the SRAM circuit includes transistor N1, transistor N2, transistor P3, transistor P4, transistor Q5, and transistor Q6, wherein,

the drain of the transistor N1 is connected to the drain of the transistor P3, the source of the transistor N1 is grounded, the gate of the transistor N1 is connected to the gate of the transistor P3, the drain of the transistor N2 is connected to the drain of the transistor P4, the source of the transistor N2 is grounded, the gate of the transistor N2 is connected to the gate of the transistor P4, the source of the transistor P3 and the source of the transistor P4 are both connected to a power supply terminal Vcc, the drain of the transistor Q5 is connected to the bit line BL NOT, the source of the transistor Q5 is connected to the gate of the transistor N2 and the gate of the transistor P4, the gate of the transistor Q5 is connected to the word line WL, the drain of the transistor Q6 is connected to the bit line BL, the source of the transistor Q6 is connected to the gate of the transistor N1 and the gate of the transistor P3, and the gate of the transistor Q6 is connected to the word line WL.

In one embodiment of the present invention, the transistor N1 and the transistor N2 are N-type fets, the transistor P3 and the transistor P4 are P-type fets, and the transistor Q5 and the transistor Q6 are switches.

In one embodiment of the present invention, obtaining the LET threshold value of the SRAM circuit for flipping the SRAM circuit at different incident positions and different incident depths of the composite sensor comprises:

the SRAM circuit is turned by utilizing different incidence positions and different incidence depths of LETs incident to the composite sensitive body;

and calculating the LET threshold value for enabling the SRAM circuit to be overturned according to a dichotomy.

In an embodiment of the present invention, based on the LET threshold, obtaining a population ratio for inverting the SRAM circuit according to a preset number of incident particles incident on the composite sensor includes:

obtaining a charge threshold value according to the LET threshold value and the incident depth based on a charge threshold value calculation formula;

based on a weight calculation formula, obtaining weights of different incident depths according to the LET threshold and the incident depth;

based on a weighted total charge calculation formula, carrying out weighted summation according to weights of different incident depths and the number of charges collected by the ith sensitive area of the composite sensitive body to obtain weighted total charges;

judging the relationship between the weighted total charge and the charge threshold, if the weighted total charge is larger than the charge threshold, turning over the SRAM circuit, and if the weighted total charge is smaller than the charge threshold, not turning over the SRAM circuit;

counting the number of all incident particles which enable the SRAM circuit to turn over in a preset number of incident particles;

and obtaining the particle number ratio for inverting the SRAM circuit according to the number of incident particles for inverting the SRAM circuit and the preset number.

In one embodiment of the present invention, the charge threshold calculation formula is:

Q_crit，_L＝LET_crit，_L*L_LET

wherein Q is_crit,LIndicating threshold of charge, LET_crit，_LDenotes the LET threshold, L_LETIndicating the depth of incidence.

In one embodiment of the present invention, the weighted total charge calculation formula is:

wherein Q is_collRepresenting the weighted total charge, a, collected after incidence of a particle on said composite susceptor_iRepresents a weight, Q_gen,iRepresents a particleThe number of charges collected by the ith sensitive area incident to the composite sensitive body.

In an embodiment of the invention, the SRAM single event upset cross section prediction method further includes adding a shielding layer on the device model.

The invention has the beneficial effects that:

the invention carries out the single-particle Monte Carlo simulation of the device and the method for measuring the turning section, accurately simulates the interaction among particles, calculates the nuclear physical processes such as energy absorption, collision and the like, completes the simulation of the random process of the particles, and completes the turning section statistics of the random process by carrying out multi-particle bombardment.

The present invention will be described in further detail with reference to the accompanying drawings and examples.

Drawings

FIG. 1 is a schematic flow chart of a SRAM single event upset cross section prediction method based on the Monte Carlo method according to an embodiment of the present invention;

FIG. 2 is a circuit diagram of an SRAM provided in an embodiment of the present invention;

FIG. 3 is a process diagram of a circuit device hybrid simulation using Sentaurus simulation software according to an embodiment of the present invention;

FIG. 4 is a schematic diagram of LET threshold curves generated by a device SRAM in which heavy ions are incident on different positions of an X axis according to an embodiment of the present invention;

fig. 5 is a schematic diagram of SRAMLET threshold versus incident depth for a bulk silicon device according to an embodiment of the present invention;

fig. 6 is a schematic diagram illustrating the relationship between LET threshold and incident depth of an FDSOI device according to an embodiment of the present invention;

FIG. 7 is a schematic diagram of an embodiment of the present invention, illustrating the inversion of Beta particles and Gamma particles as an incident particle beam;

FIG. 8 is a schematic diagram of an alternative embodiment of the present invention for inverting Beta particles and Gamma particles as the incident particle beam;

fig. 9 is a schematic diagram of a bulk silicon and FDSOI device flipped over according to an embodiment of the present invention;

fig. 10 is a schematic diagram of another bulk silicon and FDSOI device flip scenario provided by embodiments of the present invention;

FIG. 11 is a graph showing a cross-sectional turn-over of incident particles such as protons, Alpha particles, Beta particles, and Gamma particles according to an embodiment of the present invention;

FIG. 12 is a graph of a reversed cross-section of a heavy ion incident on a particle according to an embodiment of the present invention;

fig. 13 is a graph of an inverted cross-section of a shield layer provided by an embodiment of the present invention with the thickness increased to 10mm and 100 mm.

Detailed Description

The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

Example one

Referring to fig. 1, fig. 1 is a schematic flow chart of a method for predicting a single event upset cross section of an SRAM based on a monte carlo method according to an embodiment of the present invention. The SRAM single event upset cross section prediction method based on the monte carlo method of the present embodiment includes steps 1 to 6, where:

step 1, establishing a device model.

Specifically, a device model is established by using an SDE module of a Sentaurus TCAD simulation tool, the device model is a 3D model, and the device model may be, for example, FDSOI or other types of semiconductor devices.

And 2, establishing an SRAM circuit according to the device model.

In the present embodiment, referring to fig. 2, the SRAM circuit includes a transistor N1, a transistor N2, a transistor P3, a transistor P4, a transistor Q5, and a transistor Q6, wherein a drain of the transistor N1 is connected to a drain of the transistor P3, a source of the transistor N1 is grounded, a gate of the transistor N1 is connected to a gate of the transistor P3, a drain of the transistor N2 is connected to a drain of the transistor P4, a source of the transistor N2 is grounded, a gate of the transistor N2 is connected to a gate of the transistor P4, a source of the transistor P3 and a source of the transistor P4 are both connected to a power supply terminal Vcc, a drain of the transistor Q5 is connected to a bit line BL, a source of the transistor Q5 is connected to a gate of the transistor N2 and a gate of the transistor P4, a gate of the transistor Q5 is connected to a word line WL, a drain of the transistor Q6 is connected to a bit line BL, a source of the transistor Q6 is connected to a gate of the transistor N1 and a gate of the transistor P3, the gate of transistor Q6 is connected to the word line WL.

The SRAM circuit contains six transistors, namely transistor N1, transistor N2, transistor P3, transistor P4, transistor Q5, and transistor Q6. Data (Bit) in the SRAM circuit is stored in four structures of a transistor N1, a transistor N2, a transistor P3 and a transistor P4, wherein the four structures are all field effect transistors and are in cross coupling. The transistors N1 and N2 in the four transistor structures belong to N-FETs, and the transistors N1 and N2 form a pull-down network structure; the transistor P3 and the transistor P4 belong to a P-FET, and the transistor P3 and the transistor P4 form a pull-up network structure; the transistors Q5 and Q6 are switch transistors, the transistors Q5 and Q6 are controlled to be turned on and off by the word line WL, and the bit lines BL and BL are not configured to be set to high or low for writing high or low or for reading.

The basic operating cells of an SRAM circuit have three states: standby state, reading state, writing state. When the word line WL is low, the transistor Q5 and the transistor Q6 are turned off, and the voltages at the nodes V2 and V1 in fig. 2 are latched in the latch formed by the pull-up and pull-down network, so that the state can be stored as long as Vcc is kept stable, which is called idle state. The SRAM circuit in an idle state stores charges, a word line WL is written to a high level, the transistor Q5 and the transistor Q6 are turned on, the nodes V1 and V2 precharge and share the charges and transfer to the BL and BL negations, and the amplifier circuit can read a latched charge value based on a change in a potential difference between the BL and BL negations, which is referred to as a read state. The bit lines BL and BL are not precharged to the potential to be written, then the word line WL is set to the high potential, the original states of the nodes V2 and V1 are released and synchronized to the potential not corresponding to BL and BL, and then the word line WL is set to the low potential, so that the state can be written into SRAM, which is called the writing state.

In the present embodiment, the SRAM circuit in the idle state is selected as the test circuit, and in this case, the transistor P3, the transistor P4, the transistor N1, and the transistor N2 all contain electric charges while the transistor Q5 and the transistor Q6 are in the off state. At this time, the initial node V1 is at a high potential, the node V2 is at a low potential, and the space is irradiated at the position of the transistor N1.

In a specific embodiment, step 2 may specifically comprise step 2.1 to step 2.3, wherein:

step 2.1, using the SDEVICE module of the Sentaurus TCAD simulation tool to define the device model used by each transistor of the SRAM circuit.

Specifically, in the present embodiment, each transistor of the SRAM circuit uses the device model established in step 1, so the device model used by each transistor can be defined according to the specific form of each transistor in the SDEVICE module.

And 2.2, carrying out input and output setting on the defined device model by using the SDEVICE module.

Specifically, each transistor of the SRAM circuit is set at the SDEVICE module according to its input and output.

And 2.3, connecting all device models with input and output settings to establish the SRAM circuit.

Specifically, transistors provided with input and output are connected in an SRAM circuit.

In the present embodiment, as shown in fig. 3, a process diagram of the present embodiment when circuit device hybrid simulation is performed using Sentaurus simulation software is shown. As can be seen from fig. 3, when performing hybrid simulation, first, a required device model is established by using SDE, then devices used by various transistors are defined by using SDIVICE module, and input and output settings are performed on the device model already defined by SDE. The inside of the SDEVICE module is divided into several different parts, namely, Electrode, File, Physci, Plot, Math and Solve, and a Device part, a System part and the like are also arranged on a circuit with various devices.

And 3, uniformly dividing the device model of the SRAM circuit along the length direction to establish a composite sensitive body comprising a plurality of sensitive areas.

In this embodiment, the composite sensitive body is a device simulation model established by Geant4, the device model for dividing the sensitive region is a device to be irradiated in the SRAM circuit, in this embodiment, the device to be irradiated is the transistor N1, that is, only the transistor N1 is irradiated, the transistor N1 is uniformly divided into x portions along the length direction, and meanwhile, y incident depths are provided, so that the sensitive region includes x y portions in total. For example, if the transistor N1 is uniformly divided into 11 parts along the length direction, and the incidence depth is 3, the composite sensor has 3 × 11 different sensitive regions, which refers to the positions where the devices are incident.

And 4, acquiring LET thresholds of the SRAM circuit, which enable the SRAM circuit to turn over at different incidence positions and different incidence depths of the composite sensitive body.

In a specific embodiment, step 4 may specifically include steps 4.1 to 4.2, wherein:

and 4.1, turning the SRAM circuit by utilizing different incident positions and different incident depths of the LET incident composite sensitive body.

Specifically, in this embodiment, the SRAM circuit is turned at different incident positions and different incident depths by adjusting the incidence position and the incident depth of the LET (linear energy transfer), which is equivalent to the energy of the particles incident into the device.

And 4.2, calculating an LET threshold value for enabling the SRAM circuit to be overturned according to the dichotomy.

Specifically, in this embodiment, a Sentaurus TCAD simulation tool performs irradiation processing on a transistor N1 of an SRAM circuit, then different LET values for inverting the SRAM circuit are taken, and a dichotomy is used for these LET values in the Sentaurus TCAD simulation tool to obtain LET thresholds for inverting at different incident depths.

And 5, based on the LET threshold value, obtaining the population ratio for turning the SRAM circuit according to a preset number of incident particles incident into the composite sensitive body.

Specifically, in this embodiment, a preset number of incident particles are incident on the sensitive region of the composite sensitive body, which is not tolerant to the incident depth, so as to obtain how many incident particles of the preset number of incident particles can turn over the SRAM circuit, and a ratio of the number of the incident particles turned over by the SRAM circuit to the preset number is a particle count ratio.

In a particular embodiment, step 5 may particularly comprise steps 5.1 to 5.6, wherein:

step 5.1, obtaining a charge threshold value according to the LET threshold value and the incident depth based on a charge threshold value calculation formula, wherein the charge threshold value calculation formula is as follows:

Q_crit，L＝LET_crit，L*L_LET

wherein Q is_crit,LIndicating threshold of charge, LET_crit，LDenotes the LET threshold, L_LETDenotes the depth of incidence, Q_crit,LA charge threshold, LET, representing the ability of a particle to flip an SRAM circuit at an incident depth of L_crit，LIndicating the LET value at which a particle can flip the SRAM circuit at an incident depth of L.

Step 5.2, obtaining weights of different incident depths according to the LET threshold and the incident depth based on a weight calculation formula, wherein the weight calculation formula is as follows:

L₁α₁LET_crit,k+(L₂-L₁)α₂LET_crit,k+(L₃-L₂)α₃LET_crit,k… (L_k-L_k-1)α_kLET_crit,k＝L₁LET_crit,k

wherein L is_iI is more than or equal to 1 and less than or equal to k, k represents the number of incident depths in the current formula, LET_crit,kDenotes an incident depth of L_kLET threshold of_iRepresenting the weight of the ith incident depth.

For example, there are three incident depths, 56nm, 256nm, and 326nm, respectively, then:

56α₁LET_crit,56＝56LET_crit,56

56α₁LET_crit,256+200α₂LET_crit,256＝56LET_crit,56

56α₁LET_crit,326+200α₂LET_crit,326+70α₃LET_crit,326＝56LET_crit,56

therefore, α can be obtained by the above three equations₁,α₂,α₃。

And 5.3, based on a weighted total charge calculation formula, carrying out weighted summation according to weights of different incident depths and the number of charges collected by the ith sensitive area of the composite sensitive body to obtain weighted total charges, wherein the weighted total charge calculation formula is as follows:

wherein Q is_collRepresenting the weighted total charge, a, collected after incidence of a particle on said composite susceptor_iRepresents a weight, Q_gen,iRepresenting the number of charges collected by a particle incident on the ith sensitive area of the composite sensitive body.

And 5.4, judging the relation between the weighted total charge and the charge threshold, turning the SRAM circuit if the weighted total charge is larger than the charge threshold, and not turning the SRAM circuit if the weighted total charge is smaller than the charge threshold.

Specifically, after the weighted total charge and the charge threshold corresponding to one particle are obtained, the relationship between the weighted total charge and the charge threshold can be compared, if the weighted total charge is greater than the charge threshold, it indicates that the particle can flip the SRAM circuit, and if the weighted total charge is less than the charge threshold, the SRAM circuit cannot be flipped.

And 5.5, counting the number of all incident particles which turn over the SRAM circuit in the preset number of incident particles.

And 5.6, obtaining the particle number ratio for inverting the SRAM circuit according to the number of the incident particles for inverting the SRAM circuit and the preset number.

Specifically, the ratio of the number of incident particles that flip the SRAM circuit to the preset number is the ratio of the number of particles that flip the SRAM circuit.

And 6, obtaining a turnover cross section distribution transfer curve of the device model under different incident energies according to the population ratio of the SRAM circuit to be turned over.

Specifically, in nuclear physics and particle physics, cross section is a term used to express the possibility of occurrence of interactions between particles. The reaction cross section is a description of the probability of a reaction occurring when a nuclear reaction occurs, and is used to describe the probability of a specific reaction occurring between the incident particle and the nucleus of the target material. The present embodiment describes the probability of a single event upset occurring in a sensitive body by using the concept of a cross section, and the calculation is shown as follows:

where Δ n represents the number of particles that cause a single event upset of the sensor, even if the number of incident particles that flip the SRAM circuit, n represents the total number of incident particles obtained by the sensor, i.e., the preset number, S represents the area of the cross section of the target nucleic acid, i.e., the area of the flip section, and σ represents the flip section.

In this embodiment, the horizontal axis of the distribution transfer curve of the flip section is the energy of the incident particle, and the horizontal axis is the flip section, where the energy is the energy of the incident particle for flipping the SRAM circuit, and the energy is the threshold flipping energy, and the conversion relationship between the energy and the charge can be calculated according to the energy of 3.6eV for generating one electron-hole pair, so that the conversion relationship between the charge threshold for flipping the SRAM circuit and the threshold flipping energy is shown as the following formula:

wherein E is_crit,LRepresenting threshold flip energy, Q_crit,LRepresenting a charge threshold.

In addition, in this embodiment, a shielding method may be used for a device that may flip an SRAM circuit, so that the energy of the irradiated particles is absorbed and does not enter a sensitive region of the device, so as to reinforce the device, in this embodiment, a shielding layer is added to the device that may absorb the energy of the irradiated particles, so that the energy of the irradiated particles is absorbed, and the shielding layer may be a silicon shielding layer, and the thickness of the shielding layer may be 10 to 100 mm.

The specific process of the incident particle incident composite sensor in this embodiment includes: (1) constructing a system target body geometric structure; (2) constructing a system target material; (3) setting a particle type; (4) setting particle occurrence events and particle gun energy; (5) setting a particle gun angle; (6) setting electromagnetic field and other physical processes for controlling particle interaction; (7) setting a sensitive detector; (8) configuring an event generating function; (9) storing events and track results; (10) visualizing particle trajectories in the detector; (11) the simulation data is captured and analyzed.

The embodiment uses the multithreading capability to accelerate the simulation, greatly shortens the simulation speed and improves the fine granularity of the simulation acquisition. The method comprises the steps of starting a multithreading option in a cmake installation compiling stage of Geant4, namely adding a configuration item-DGEANT 4_ BUILD _ MULTHREADEDED ═ ON; in a vis.mac configuration file, the number of threads can be set based on the number of CPU cores and whether hyper-threads are supported; introducing a following header file before LET.cc, namely main function definition, and writing a code which is compatible with a hyper-thread which is not opened downwards, so that the header file can be automatically selected by a Geant4 according to whether the hyper-thread is opened or not in a cmake compiling stage by the Geant 4; the method comprises the steps of using G4MTRunManager in LET.cc, namely main function configuration, and using a static method G4Threading, wherein G4GetNumberOfCores () is used for acquiring the core number; creating an actionlnitiation function at actionlnitiation.cc; creating a main thread execution function BuildForMaster, wherein the main thread execution function needs to call a SetMaster method of a runAction instance to be referred to as true; creating a secondary thread execution function Build, wherein the secondary thread execution function needs to call a SetMaster method of a runAction instance to refer to false, meanwhile, using the runAction instantiation EventAction class to control the generation step of Event, using the eventAction instantiation SteppingAction class to control the generation step of Steppng, and finally instantiating a PrimaryGeneratoriAction class to control the related steps of particle generation.

The invention introduces Geant4 simulation software to carry out single-particle Monte Carlo simulation of the device and determine the method of the turnover section, accurately simulates the interaction between particles, calculates nuclear physical processes such as energy absorption and collision and the like, completes the simulation of the random process of the particles, completes the turnover section statistics of the random process by carrying out multi-particle bombardment, and finally completes the device-level reinforcement.

In order to illustrate the effectiveness of the SRAM single-event upset section prediction and reinforcement method based on the monte carlo method proposed in this embodiment, a total dose simulation of FDSOI at 28nm is performed, specifically:

simulation software

Sentaurus，Geant4

Second, simulation device

28nm FDSOI, N-type FDSOI and P-type FDSOI with width-to-length ratio of 0.08um/0.02 um.

Third, simulation process

The implementation idea of the embodiment can be divided into two steps, firstly, a 6T-SRAM circuit is constructed through Sentaurus, the position and the incident depth of a sensitive area are adjusted, LET thresholds for turning the SRAM at different positions and different incident depths of the 6T-SRAM are obtained, and after the turned LET thresholds are obtained, the following formula 3-1 can be extracted by combining the charges collected based on the weight:

wherein Q_collRepresenting the total charge collected, alpha_iRepresenting weights, which can be derived from Sentaurus simulationsIs calculated to obtain the LET threshold value of, Q_gen,iWhich represents the number of charges collected in different regions, and which is scaled to the deposition energy, can be calculated as the 3.6eV energy at which an electron-hole pair is generated.

Q_crit＝LET_crit*L_LET (3-2)

Equation 3-2 is used to represent the charge threshold, where Q_critRepresenting the electric charge, LET_critDenotes the LET threshold, L_LETIndicating the depth of incidence.

Geant4 supports the definition of device structures of different materials and different dopings, in the embodiment, three-dimensional structure parameters based on the device and the X-axis position of bulk silicon in the Sentaurus simulation process are divided into 11 incidence areas, the Z-axis position is 0 nm-56 nm Fin, the Z-axis position is 56 nm-256 nm Fin, the Z-axis position is 256 nm-326 nm substrate, and the divided three areas are totally 33 to define a composite sensitive body model.

For bulk silicon devices, the entire channel-to-substrate sensitive region is divided into 33 sensitive regions, and based on equations 3-1 and 3-2, the weight of the incident depth can be obtained by solving equations 3-3, 3-4, and 3-5. Three incident depths are respectively 56nm, 256nm and 326nm, and the formula is shown as follows:

56α₁LET_crit,56＝56LET_crit,56 (3-3)

56α₁LET_crit,256+200α₂LET_crit,256＝56LET_crit,56 (3-4)

56α₁LET_crit,326+200α₂LET_crit,326+70α₃LET_crit,326＝56LET_crit,56 (3-5)

in this embodiment, the three incident depths are 56nm, 256nm, and 326nm, respectively, and the three incident depths are substituted into the solution formula to obtain the depth direction weight α₁、α₂、α₃Are 1, 0.546 and 0.78667, and then assuming that-14 nm to 14nm of the incident depth of 56nm, that is, the weight of the LET threshold at the center position is 1, the weights of the other parts are multiplied by the weight α in the depth direction according to the ratio of the 56nmLET threshold to the LET threshold at the position_nThat is to sayThe region weight is obtained. Similarly, for the FDSOI device, since the α coefficient of the incident depth does not need to be considered, the weight of the center position may be set to 1, and the position weight may be obtained after calculating the weight of the bulk silicon device as shown in table 3.1.

TABLE 3.1 bulk silicon/SOI FinFET Turn-over weights

For the judgment of the overturning condition, the calculation can be carried out based on the LET product of the incidence depth 56nm and the central position of bulk silicon and FDSOI devices respectively, and since 3.6eV is required for depositing 1 electron-hole pair, the conversion relation of the overturning charge threshold and the threshold overturning energy is shown as the formula 3-6, wherein Q is_crit,56Can be solved by equations 3-7.

Q_crit,56＝LET_crit,56×0.056um (3-7)

Based on the fact that a bulk silicon device is at the center position, the threshold flipping LET of the 56nm incident depth is 0.0059pC/um and FDSOI is at the center position, the threshold flipping LET of the 56nm incident depth is 0.0052pC/um, the threshold flipping energy of the bulk silicon FinFET is 7.434keV according to the formula 3-6, the threshold flipping energy of the FDSOI is 6.552keV, 1000 events (namely particles) generated by a single Run in the Geant4, and a plurality of steps (line segments) generated by each Event are added in a weighting mode.

Geant4 can obtain the energy deposition area name, so the deposition energy can be calculated by weight according to the area position of Step deposition. And through the weighting and the results of the electric charge quantities collected by the collected particles in different areas and through the previously obtained overturning LET threshold value as an overturning condition, 1000 beams of particles are incident, and the number ratio of finally overturned particles is analyzed, so that the distribution and transfer curves of the overturning cross sections of the two devices under different incident energies can be obtained finally.

Fourth, simulation result and analysis

As shown in fig. 4, since the FDSOI device possesses a buried oxide layer, the LET value required for switching to occur at an incident depth of 56nm is the same as that at 326 nm. Meanwhile, the graph shows that the device is most sensitive to overturn at the position of 0-20 nm, the generated LET threshold is 0.0049pC/um, the overturn at the position of-100 nm is least sensitive, and the overturn LET threshold is 0.11 pC/um.

As shown in fig. 5 and 6, the SRAMLET threshold is a function of the incident depth for bulk silicon devices, and the LET threshold is a function of the incident depth for FDSOI devices. The LET threshold values of the SOI device are the same at the position values above 50nm, because the buried oxide layer of the SOI device can prevent the device from collecting leakage current, and therefore the single-particle upset effect can be prevented. The LET threshold has been obtained so far that the SRAM flips. And after the overturning LET threshold is obtained, establishing a Geant4 bulk silicon model and an FDSOI model for next simulation.

As shown in fig. 7, when the incident particle beam is Beta particle and also Gamma particle, the cross section of the single event upset effect of the device is 0, and it is known that the LET values of the two particles are lower relative to those of proton and Alpha particle, so the reduction of the energy deposition effect leads to the reduction of the number of electron-hole pairs in the device, and therefore the SRAM circuit is not flipped. As can be seen, the flip-sectional curve of the Alpha particles was observed, and when the flip-sectional value reached 0.64 × 10^-15cm²The probability of flip occurring is 50%, when the incident energy of the SRAM beam for FDSOI is in the range of 6 x 10^-2MeV and 6.5MeV, and for bulk silicon devices the incident energy is 4.7 x 10^-2MeV and 32 MeV. When the incident particle is proton, the flip section curve of proton is observed, the flip section value of SRAM of FDSOI device is not as high as that of Alpha particle, and the maximum value of the cross section is 0.555 x 10 when the incident particle energy is 0.1MeV^-15cm²Almost only Alpha particles produce about half of the extremum, which accounts for the difference in energy deposition to the device between protons and Alpha particles; meanwhile, the corresponding energy of the body silicon device for generating the turnover section reaching the half value is 6 x 10^-2MeV and 1.4 MeV. The results demonstrate that first Beta particles and Gamma particlesBecause the energy deposition efficiency of the device is very low, the two particles can not cause the device to generate the SRAM circuit single-particle upset effect of 28 nm; and secondly, compared with the common bulk silicon device, the FDSOI device has excellent radiation resistance.

When the incident position is changed to make the particle beam incident at the position where the device is least sensitive (-102nm), the generated turning cross section curve is as shown in fig. 8, and it can be known that four kinds of particles are equally not turned, which also proves the effect of the incident position on the single particle irradiation effect.

As shown in fig. 9, both bulk and FDSOI devices SRAM curves at lower energies from almost 0 to a relatively high level (1.28 x 10)^-15cm) and different incident particles produce different curves of reversed cross section, and heavy ions (B11) with low atomic number reach the reversed cross section (6.4 x 10)^-13cm) threshold incident energy is only 0.26MeV, while heavy ions with high atomic number (Bi209) reach the flip section (6.4 x 10)^-13cm) threshold incident energy can reach 0.78 MeV.

FIG. 10 shows the generated flip-flop cross-sectional curve when the incident particle is at-102 nm. Comparing the curve of the turning cross section incident at the central position of the 0nm device in the previous experiment, it can be known that when the initial incident energy of heavy ions with higher atomic number in the bulk silicon device is in a range of more than 101MeV, the turning cross section of most incident particles (Xe 132 and Bi209 particles) is rapidly reduced to 0 value under different energies, and the trend of the curve illustrates the fact that the incident sensitivity degree is lower than the position sensitivity degree of the previous incident, and the SRAM circuit cannot turn; when the irradiated device is FDSOI, the graph shows that a great deal of incident particle species are affected, all particles reach the minimum value at 105MeV, the turnover section values of the whole particle energy incidence range of a plurality of particles (12, N14 and the like) are connected to 0 value, and the incidence energy reduced by the turnover section values of other heavy ions (r40, Cr52, Ni58, Kr84, Xe132 and the like) is greatly reduced relative to the bulk silicon, which shows that the irradiation resistance of the FDSOI device is far higher than that of the bulk silicon device, and the FDSOI device is consistent with the fact.

In summary, in the embodiment, energy deposition effects of bulk silicon and FDSOI devices in different regions are obtained by exploring different particle incidence positions and LET thresholds where the particles are incident deep to generate flipping, and thus a device model is established by using Geant4, and the devices are bombarded by different types of particles to obtain flipping cross sections and weighted deposition dose energy spectrums of the devices, which are in line with expected conclusions, and prove the effectiveness of the invention.

In addition, there are two general solutions to the single event effect, one is protection, and the other is redundancy and recovery.

For the protection mode, firstly, Alpha particles released by radioactive decay of the packaging material need to be considered, and the generation reason is related to the purity of the packaging material, so that the introduction of uranium and thorium impurities needs to be removed as much as possible, and the high-purity packaging material can enable the content of 5-10 alphas/cm²-hr decreased to 0.001 alphas/cm²-hr, while the boron-containing species may release neutrons, replacing the BPSG medium with another medium may reduce neutron production. The device structure also has an influence on the occurrence of the single event effect, for example, the protective capability of the SOI structure verified by the embodiment for high-energy heavy ions can be improved by more than ten times due to the fact that the effective area of the SOI structure is smaller than that of a bulk silicon structure, and in addition, any other SOI structure can reduce the Q_collAnd increase Q_critAll the above methods can suppress the generation of the single event effect to some extent, and therefore, the detailed description thereof is omitted. Another physical method is to add a shielding layer, and the method of using a thick shielding layer to absorb the energy of the irradiated particles so that the irradiated particles do not enter the sensitive region of the device any more is verified to be an effective method.

As shown in fig. 11, is a graph of the reversed cross-section generated when the incident particles are protons, Alpha particles, Beta particles, and Gamma particles. From the left graph in fig. 11, it can be seen that in the different low, medium and high initial incident energy ranges, the four particles have no flip effect, and only the Alpha particles have a minute peak at 10-2MeV, which is not more than 1 × 10-16cm, while the Alpha, Beta and Gamma particles have no peak in the full energy range. The deposition energy curves for different incident energies of protons and Alpha particles are selected as the right-hand graph in fig. 11.

As shown in fig. 12, which is a graph of an inverted cross section when the incident particles are heavy ions, the maximum value of the curve is far beyond the value of the light particles as can be seen from the left side of fig. 12, and therefore the radiation-resistant effect of the shielding layer made of silicon material with 1mm on the heavy ions is very general. According to the deposition energy diagram, the numerical value of the overturning section is very large in the incident energy range of more than 500MeV, and the effect of the shielding layer on resisting single particles is verified; meanwhile, different curves are observed, so that the deposition energy peak value is increased along with the increase of the atomic number of the incident particle, and a larger turnover section value is obtained.

As shown in fig. 13, which is a graph of the reversed cross-section when the shield layer is increased to 10mm and 100 mm. It can be seen from the figure that the shielding layer effect of 10mm is much better than that of 1mm, but the particles with smaller atomic number such as B11, C12, N14 and Ne20 still have small peak value at 103Me, and the peak value exceeds 3 x 10 when the incident energy is more than 106MeV^-16MeV. When the thickness of the shield layer continues to increase to 100mm, it can be seen that the curve is very gentle and that no single event upset effect occurs for all particles.

In conclusion, the reinforcing method of the invention shows that the 10mm silicon shielding layer can prevent most particles from turning in the earth, and the 100mm silicon shielding layer can completely prevent the particles from turning in the earth.

In the description of the present invention, the terms "first" and "second" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance or implying any number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of the present invention, "a plurality" means two or more unless specifically defined otherwise.

In the description herein, references to the description of the term "one embodiment," "some embodiments," "an example," "a specific example," or "some examples," etc., mean that a particular feature, structure, material, or characteristic data point described in connection with the embodiment or example is included in at least one embodiment or example of the invention. In this specification, the schematic representations of the terms used above are not necessarily intended to refer to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples. Furthermore, various embodiments or examples described in this specification can be combined and combined by those skilled in the art.

The foregoing is a more detailed description of the invention in connection with specific preferred embodiments and it is not intended that the invention be limited to these specific details. For those skilled in the art to which the invention pertains, several simple deductions or substitutions may be made without departing from the spirit of the invention, which should be construed as belonging to the scope of the invention.

Claims

1. A SRAM single event upset cross section prediction method based on a Monte Carlo method is characterized by comprising the following steps:

establishing a device model;

establishing an SRAM circuit according to the device model;

and obtaining a turnover cross section distribution transfer curve of the device model under different incident energies according to the population ratio enabling the SRAM circuit to turn over.

2. The SRAM single-event upset cross-section prediction method based on the Monte Carlo method as claimed in claim 1, wherein establishing a device model comprises:

3. The SRAM single-event upset cross-section prediction method based on the Monte Carlo method as claimed in claim 1, wherein building the SRAM circuit according to the device model comprises:

4. The SRAM single event upset cross-section prediction method based on the Monte Carlo method of claim 1 or 3, wherein the SRAM circuit comprises a transistor N1, a transistor N2, a transistor P3, a transistor P4, a transistor Q5 and a transistor Q6, wherein,

5. The SRAM single event upset cross-section prediction method based on the Monte Carlo method as claimed in claim 4, wherein the transistors N1 and N2 are NFETs, the transistors P3 and P4 are PFETs, and the transistors Q5 and Q6 are switches.

6. The SRAM single-event upset cross-section prediction method based on the Monte Carlo method as claimed in claim 1, wherein obtaining LET thresholds for the SRAM circuit to flip at different incident positions and different incident depths of the composite sensor comprises:

7. The SRAM single-particle upset cross section prediction method based on the Monte Carlo method as claimed in claim 1, wherein based on the LET threshold, obtaining the particle number ratio for the SRAM circuit to upset according to the composite sensor incident by a preset number of incident particles, comprises:

obtaining weights of different incidence depths according to the LET threshold and the incidence depth based on a weight calculation formula;

and obtaining the number ratio of the particles which enable the SRAM circuit to be overturned according to the number of the incident particles which enable the SRAM circuit to be overturned and the preset number.

8. The SRAM single-event upset cross-section prediction method based on the Monte Carlo method as claimed in claim 7, wherein the charge threshold calculation formula is:

Q_crit，L＝LET_crit，L*L_LET

wherein Q is_crit,LIndicating threshold of charge, LET_crit，LDenotes the LET threshold, L_LETIndicating the depth of incidence.

9. The SRAM single-event upset cross-section prediction method based on the Monte Carlo method as claimed in claim 7, wherein the weighted total charge calculation formula is:

10. The SRAM single-event upset cross-section prediction method based on the Monte Carlo method as claimed in claim 1, further comprising adding a shielding layer on the device model.