Process flow database construction method based on super-junction mosfet

Abstract

The application relates to the technical field of process flow database construction, and relates to a process flow database construction method based on super junction MOSFET preparation, which comprises the following steps: confirming super junction MOSFET process flow knowledge, obtaining database design specifications according to the super junction MOSFET process flow knowledge, establishing an automatic data acquisition interface, constructing a database framework according to the automatic data acquisition interface and the database design specifications, confirming a set of to-be-simulated process parameters, obtaining a set of associated data based on the set of to-be-simulated process parameters and the database framework, constructing an optimized forward prediction machine learning model and an optimized reverse design machine learning model according to the set of associated data, and completing the process flow database construction based on the super junction MOSFET preparation based on the optimized forward prediction machine learning model and the optimized reverse design machine learning model. The application can improve the optimization efficiency of the preparation process and the consistency of the data.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of process flow database construction, and particularly relates to a process flow database construction method based on super-junction MOSFET preparation. BACKGROUND

[0002] MOSFET is a common semiconductor device, widely used in electronic circuits for switching and amplification functions. Super-junction is a special MOSFET structure, which realizes higher voltage withstand capability and lower on-resistance by optimizing the structure design of the device. Engineering process refers to the sequence and combination of preparation steps. Database is a systematic data storage structure for storing, retrieving and managing data.

[0003] Traditional preparation process data is scattered in different experimental records and engineers' notes, which is difficult to manage and analyze systematically. Secondly, the traditional optimization method relying on human experience is low in efficiency, and it is difficult to quickly find the optimal combination of process parameters. Therefore, how to improve the optimization efficiency of the preparation process and the consistency of the data is a technical problem to be solved. SUMMARY

[0004] The present application provides a process flow database construction method based on super-junction MOSFET preparation, and a computer readable storage medium, which mainly aims to improve the optimization efficiency of the preparation process and the consistency of the data.

[0005] To achieve the above purpose, the present application provides a process flow database construction method based on super-junction MOSFET preparation, which comprises:

[0006] Receiving process flow database construction instructions, confirming super-junction MOSFET process flow knowledge according to the process flow database construction instructions, and obtaining database design specifications according to the super-junction MOSFET process flow knowledge;

[0007] Establishing an automatic data acquisition interface, wherein the automatic data acquisition interface comprises a device layer interface, a measurement layer interface and a file layer interface;

[0008] According to the automatic data acquisition interface and the database design specification, a database framework is constructed, and a set of process parameters to be simulated is confirmed;

[0009] Based on the set of process parameters to be simulated and the database framework, a set of associated data is obtained, wherein the set of process parameters to be simulated comprises a plurality of process parameters to be simulated, the set of associated data comprises a plurality of associated data, and the process parameters to be simulated and the associated data are one-to-one corresponding;

[0010] According to the correlation data set, an optimized forward prediction machine learning model and an optimized reverse design machine learning model are constructed, and based on the optimized forward prediction machine learning model and the optimized reverse design machine learning model, a process flow database construction based on the super junction MOSFET preparation is completed.

[0011] Optionally, the design specification according to the super junction MOSFET process flow knowledge acquisition database comprises:

[0012] According to the super junction MOSFET process flow knowledge acquisition data entity set and the data entity relationship set, data entities are extracted from the data entity set in sequence, and the extracted data entities are all subjected to the following operations:

[0013] According to the data entity, a node type is confirmed, and according to the data entity, a target data entity relationship is confirmed from the data entity relationship set;

[0014] According to the target data entity relationship, a relationship type is confirmed, according to the node type and the relationship type, a database mode is constructed, and whether the database mode conforms to a preset process flow chart structure is judged;

[0015] If the database mode conforms to the preset process flow chart structure, the process flow chart structure is taken as a final database mode;

[0016] If the database mode does not conform to the preset process flow chart structure, whether the database mode conforms to a preset graph structure requirement is judged;

[0017] If the database mode conforms to the preset graph structure requirement, a pre-constructed graph database is taken as the final database mode;

[0018] If the database mode does not conform to the preset graph structure requirement, whether the database mode conforms to a preset relationship schema is judged;

[0019] If the database mode conforms to the preset relationship schema, a pre-constructed relational database is taken as the final database mode;

[0020] The final database mode is summarized to obtain a database design specification.

[0021] Optionally, the correlation data set is obtained based on the to-be-simulated process parameter set and the database framework, comprising:

[0022] To-be-simulated process parameters are extracted from the to-be-simulated process parameter set in sequence, and a simulation input file is obtained by using a pre-constructed workflow engine to perform file conversion on the to-be-simulated process parameters, wherein the workflow engine comprises a result analysis module;

[0023] A simulation output text file and a simulation output binary file are obtained by using a pre-constructed TCAD software to perform a simulation operation on the simulation input file;

[0024] The simulation output text file is read by the result analysis module to obtain the electrical index parameter, and the simulation output binary file is read by the result analysis module to obtain the internal device distribution parameter, wherein the internal device distribution parameter includes: the doping concentration distribution data and the electric field intensity distribution data;

[0025] The internal device distribution parameter is extracted according to the database framework to obtain the feature parameter, and the structured simulation result data is generated according to the electrical index parameter and the feature parameter;

[0026] The structured simulation result data is associated and bound with the to-be-simulated process parameter to obtain the associated data, and the associated data is summarized to obtain the associated data set.

[0027] Optionally, the internal device distribution parameter is extracted according to the database framework to obtain the feature parameter, including:

[0028] The electric field intensity distribution data in the internal device distribution parameter is subjected to numerical analysis to obtain the electric field peak value and the electric field peak coordinate;

[0029] The blank dictionary is created according to the database framework, the electric field peak value and the electric field peak coordinate are stored by using the blank dictionary, and the stored electric field parameter is obtained;

[0030] The path point set is extracted from the doping concentration distribution data according to the preset doping concentration integral path, the path points are extracted from the path point set in sequence, and the following operations are performed on the extracted path points:

[0031] It is judged whether the path point is located on the preset grid point or not;

[0032] If the path point is located on the preset grid point, the doping concentration value of the path point is obtained, and the doping concentration value is taken as the doping concentration integral value;

[0033] If the path point is not located on the preset grid point, the path point coordinate corresponding to the path point is confirmed from the doping concentration distribution data, and the adjacent grid point coordinate set is obtained according to the path point coordinate;

[0034] The adjacent grid point coordinates are extracted from the adjacent grid point coordinate set in sequence, the path point coordinate and the adjacent grid point coordinate are combined to obtain the path coordinate group, and the grid weight and the path distance of the path coordinate group are calculated;

[0035] The grid weight and the path distance are summarized respectively to obtain the grid weight set and the path distance set, wherein the path coordinate group corresponds to the grid weight and the path distance one by one;

[0036] The doping concentration integral value is obtained based on the coordinate set of adjacent grid points, the grid weight set, and the path distance set. The existing electric field parameters and the doping concentration integral value are used as characteristic parameters.

[0037] Optionally, the calculation of the grid weights and path distances of the path coordinate set includes:

[0038] Based on the coordinates of adjacent grid points in the path coordinate group, determine the x-coordinate and y-coordinate of adjacent grid points; based on the coordinates of path points in the path coordinate group, determine the x-coordinate and y-coordinate of path points.

[0039] The difference in x-coordinates is calculated based on the x-coordinates of adjacent grid points and the x-coordinates of path points; the difference in y-coordinates is calculated based on the y-coordinates of adjacent grid points and the y-coordinates of path points.

[0040] The grid weight is calculated based on the difference between the horizontal and vertical coordinates, where the grid weight is the product of the difference between the horizontal and vertical coordinates.

[0041] Calculate the path distance based on the x-coordinates of adjacent grid points, the y-coordinates of adjacent grid points, the x-coordinates of path points, and the y-coordinates of path points.

[0042] Optionally, obtaining the doping concentration integral value based on the coordinate set of adjacent grid points, the grid weight set, and the path distance set includes:

[0043] Obtain the concentration value set of adjacent grid points based on the coordinate set of adjacent grid points, and calculate the doping concentration value of the path point based on the grid weight set and the concentration value set of adjacent grid points.

[0044] Extract path distances sequentially from the path distance set, determine the target path coordinate set based on the path distances, and determine the concentration values ​​of adjacent grid points of the target based on the target path coordinate set;

[0045] The trapezoidal area is calculated based on the doping concentration value of the path point, the path distance, and the concentration values ​​of the adjacent grid points of the target. The trapezoidal areas are then summarized to obtain the trapezoidal area set.

[0046] Calculate the sum of the trapezoidal areas of the trapezoidal area set, and use the sum of the trapezoidal areas as the integral value of the doping concentration.

[0047] Optionally, the formula for calculating the doping concentration value at the path point is as follows:

[0048] in, This indicates the doping concentration value at the path point. This indicates the number of grid weights in the grid weight set. Indicates the first in the grid weight set Each grid weight, Indicates the first The concentration values ​​of adjacent grid points corresponding to each grid weight.

[0049] Optionally, the constructing the optimized forward prediction machine learning model and the optimized reverse design machine learning model according to the correlation data set comprises:

[0050] The correlation data set is stored in a database framework to obtain a virtual experiment data set, and a process flow tracing function, an SPC control chart analysis function and an interactive correlation analysis tool are developed based on the virtual experiment data set;

[0051] The data analysis visualization unit is obtained based on the process flow tracing function, the SPC control chart analysis function and the interactive correlation analysis tool;

[0052] The forward prediction machine learning model is constructed based on the virtual experiment data set, and the reverse design machine learning model is constructed according to the forward prediction machine learning model;

[0053] The data analysis visualization unit, the forward prediction machine learning model and the reverse design machine learning model are integrated in the system user interface to obtain an integrated system user interface;

[0054] The role access permission is set for the integrated system user interface to obtain an updated system user interface;

[0055] The optimized forward prediction machine learning model and the optimized reverse design machine learning model are obtained based on the updated system user interface.

[0056] Optionally, the obtaining the optimized forward prediction machine learning model and the optimized reverse design machine learning model based on the updated system user interface comprises:

[0057] The incremental data set is obtained, wherein the incremental data set comprises actual production process parameters and measured performance data, the incremental data set is merged with the virtual experiment data set to obtain an updated experiment data set;

[0058] The forward prediction machine learning model and the reverse design machine learning model in the updated system user interface are trained and optimized by using the updated experiment data set to obtain the optimized forward prediction machine learning model and the optimized reverse design machine learning model.

[0059] To achieve the above-mentioned purpose, the application further provides a process flow database construction system based on super junction MOSFET, comprising:

[0060] A database design specification confirmation module is configured to receive a process flow database construction instruction, confirm super junction MOSFET process flow knowledge according to the process flow database construction instruction, and obtain a database design specification according to the super junction MOSFET process flow knowledge;

[0061] The data acquisition interface establishing module is configured to establish an automated data acquisition interface, wherein the automated data acquisition interface comprises a device layer interface, a measurement layer interface and a file layer interface.

[0062] The database framework constructing module is configured to construct a database framework according to the automated data acquisition interface and a database design specification, and to determine a set of process parameters to be simulated.

[0063] The database constructing completion module is configured to obtain a set of associated data based on the set of process parameters to be simulated and the database framework, wherein the set of process parameters to be simulated comprises a plurality of process parameters to be simulated, the set of associated data comprises a plurality of associated data, the process parameters to be simulated and the associated data are in one-to-one correspondence, an optimized forward prediction machine learning model and an optimized reverse design machine learning model are constructed according to the set of associated data, and a process flow database construction based on super junction MOSFET preparation is completed based on the optimized forward prediction machine learning model and the optimized reverse design machine learning model.

[0064] To solve the above problems, the present application further provides an electronic device, which comprises:

[0065] A memory configured to store at least one instruction;

[0066] A processor configured to execute the instruction stored in the memory to implement the process flow database construction method based on super junction MOSFET preparation as described above.

[0067] To solve the above problems, the present application further provides a computer readable storage medium, which stores at least one instruction, and the at least one instruction is executed by a processor in an electronic device to implement the process flow database construction method based on super junction MOSFET preparation as described above.

[0068] The present application is to solve the problems described in the background art, the present application receives process flow database construction instructions, confirms the super junction MOSFET process flow knowledge according to the process flow database construction instructions, and obtains the database design specification according to the super junction MOSFET process flow knowledge, the present application receives the construction instruction, so that the system is clear about the task requirement, and can carry out the subsequent work targetedly, avoids blind operation, improves the efficiency and accuracy of building the database, confirms the super junction MOSFET process flow knowledge, systematically organizes the dispersed professional knowledge, provides a solid theoretical basis and data source for the design of subsequent database, ensures the professionalism and integrity of the database content, at the same time, according to the process flow knowledge acquisition database design specification, it is ensured that the structure and content of the database meet the industry standards and actual application requirements, improve the usability and maintainability of the database, establish an automatic data acquisition interface, wherein the automatic data acquisition interface includes: device layer interface, measurement layer interface and file layer interface, the automatic data acquisition interface of the present application can automatically acquire data from equipment, measuring instruments and files, reduces manual intervention, greatly improves the speed and efficiency of data acquisition, saves time, avoids errors and deviations that may occur in the process of manual data collection, ensures the accuracy and reliability of the data, provides a high-quality data basis for subsequent data analysis and model construction, according to the automatic data acquisition interface and the database design specification, a database framework is constructed, and a set of to-be-simulated process parameters is confirmed, the present application combines the automatic data acquisition interface and the database design specification to construct the database framework, so that the database has a clear structure and good organization, facilitating data storage, management and query, improving the performance and scalability of the database, based on the set of to-be-simulated process parameters and the database framework, a set of associated data is obtained, wherein the set of to-be-simulated process parameters includes a plurality of to-be-simulated process parameters, the set of associated data includes a plurality of associated data, and the to-be-simulated process parameters correspond to the associated data one by one, the present application obtains the set of associated data corresponding to the to-be-simulated process parameters one by one, establishes the connection between the process parameters and the related data, which helps to deeply analyze the influence of process parameters on the preparation process, and provides strong data support for optimizing the process, according to the set of associated data, an optimization forward prediction machine learning model and an optimization reverse design machine learning model are constructed, based on the optimization forward prediction machine learning model and the optimization reverse design machine learning model, the process flow database construction based on the super junction MOSFET preparation is completed, the optimization forward prediction machine learning model can predict the preparation process result according to the input process parameters, help engineers understand the process effect in advance, provide reference for process adjustment and optimization, reduce the number of trial and error, improve the production efficiency, the optimization reverse design machine learning model can deduce the appropriate process parameters according to the expected preparation result, provide a new idea and method for process design, and help to quickly design a preparation process that meets the requirements. Therefore, the present application can improve the optimization efficiency of the preparation process and the consistency of the data. BRIEF DESCRIPTION OF DRAWINGS

[0069] Figure 1 A flowchart of a process flow database construction method based on super junction MOSFET preparation provided by an embodiment of the present application is shown in the figure.

[0070] Figure 2 A function module diagram of a process flow database construction system based on super junction MOSFET preparation provided by an embodiment of the present application is shown in the figure.

[0071] Figure 3 A structural diagram of an electronic device implementing the process flow database construction method based on super junction MOSFET preparation provided by an embodiment of the present application is shown in the figure.

[0072] Reference signs are explained as follows:

[0073] 1. Electronic device; 10. Processor; 11. Memory; 12. Bus.

[0074] The implementation, functional features and advantages of the present application will be further explained with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION

[0075] It should be understood that the specific embodiments described herein are merely intended to explain the present application and not to limit the present application.

[0076] Embodiments of the present application provide a process flow database construction method based on super junction MOSFET preparation. The execution subject of the process flow database construction method based on super junction MOSFET preparation includes but is not limited to at least one of electronic devices such as a server and a terminal which can be configured to execute the method provided by the embodiments of the present application. In other words, the process flow database construction method based on super junction MOSFET preparation can be executed by software or hardware installed in a terminal device or a server device, and the software can be a blockchain platform. The server includes but is not limited to a single server, a server cluster, a cloud server or a cloud server cluster, etc.

[0077] Referring to Figure 1 A flowchart of a process flow database construction method based on super junction MOSFET preparation provided by an embodiment of the present application is shown in the figure. In this embodiment, the process flow database construction method based on super junction MOSFET preparation includes:

[0078] S1, receiving a process flow database construction instruction, confirming super junction MOSFET process flow knowledge according to the process flow database construction instruction, and obtaining database design specifications according to the super junction MOSFET process flow knowledge.

[0079] It should be explained that the process flow database construction instruction refers to an instruction initiated by an operator for starting the construction of the process flow database. The process flow database construction instruction makes the system clear about the task requirements, enabling targeted follow-up work, avoiding blind operation, and improving the efficiency and accuracy of database construction. MOSFET is a common semiconductor device widely used in electronic circuits for switching and amplification functions. It is composed of three layers of metal, oxide and semiconductor, and controls the flow of current by controlling the gate voltage. Super junction is a special MOSFET structure that achieves higher voltage withstand capability and lower on-resistance by optimizing the device structure design.

[0080] In detail, the design specification of the super junction MOSFET process flow knowledge acquisition database includes:

[0081] According to the super junction MOSFET process flow knowledge acquisition data entity set and the data entity relationship set, the data entities are extracted from the data entity set in sequence, and the extracted data entities are all executed as follows:

[0082] According to the data entity, the node type is confirmed, and the target data entity relationship is confirmed from the data entity relationship set according to the data entity;

[0083] According to the target data entity relationship, the relationship type is confirmed, the database mode is constructed according to the node type and the relationship type, and it is judged whether the database mode conforms to the preset process flow chart structure;

[0084] If the database mode conforms to the preset process flow chart structure, the process flow chart structure is taken as the final database mode;

[0085] If the database mode does not conform to the preset process flow chart structure, it is judged whether the database mode conforms to the preset graph structure requirement;

[0086] If the database mode conforms to the preset graph structure requirement, the pre-constructed graph database is taken as the final database mode;

[0087] If the database mode does not conform to the preset graph structure requirement, it is judged whether the database mode conforms to the preset relationship schema;

[0088] If the database mode conforms to the preset relationship schema, the pre-constructed relational database is taken as the final database mode;

[0089] The final database mode is summarized to obtain the database design specification.

[0090] It should be explained that the super-junction MOSFET process flow knowledge refers to the process and technical knowledge related to the manufacture of super-junction MOSFET, which covers the entire process from raw material preparation to finished product testing, including information on various process steps, process parameters, material usage, equipment operation, and quality control. The data entity set refers to the collection of all data entities that have independent existence in the super-junction MOSFET process flow. For example, data entities are process steps, wafers, measurement data, simulation tasks, and equipment. The data entity relationship is a collection of various relationships that exist between data entities. For example, the relationship between equipment and process steps: a certain photolithography machine is used to perform the photolithography step, the relationship between materials and process steps: a certain photoresist is used for the photolithography step, the sequence relationship between process steps: epitaxial growth must be completed before ion implantation, etc. Node type refers to the type used to represent data entities in the database. Relationship type refers to the type used to represent the relationship between data entities in the database. For example, relationship types are usage relationship, sequence relationship, and association relationship. Target data entity relationship refers to the data entity relationship corresponding to the data entity in the data entity relationship set. The relationship paradigm refers to a series of rules and standards in the design of a relational database, used to ensure the normalization of the database table structure and the consistency of the data. The main purpose of the relationship paradigm is to reduce data redundancy and avoid data update anomalies (such as insertion anomalies, deletion anomalies, and update anomalies), thereby improving the efficiency and maintainability of the database. A relational database is a database management system based on the relational model. It uses tables to store data, and each table is composed of rows and columns. The relationship paradigm and the relational database are both prior art and will not be described here.

[0091] Importantly, the constructing the database schema according to the node type and the relationship type means that the node type and the relationship type are combined to form the structure of the database, which can reflect the entities in the super junction MOSFET process flow and the relationships between the entities. The process flow graph structure means the graphical representation of the data entities and the entity relationships in the super junction MOSFET process flow, which can intuitively show various complex relationships and dynamic properties in the process flow. The graph structure requirement means the requirement for the process flow graph structure when the database schema is constructed. The database design specification means a series of rules and standards that need to be followed when the database is constructed, for example, the correctness of the relationship is ensured through the foreign key constraint mechanism or the data leakage and illegal access are prevented through the user permission control mechanism. The database schema conforms to the preset process flow graph structure means that each node (data entity) in the database schema is consistent with the node in the process flow graph and the relationship between the nodes in the database schema is consistent with the relationship in the process flow graph. The database schema does not conform to the preset process flow graph structure means that a certain node (data entity) in the database schema is inconsistent with the node in the process flow graph and the relationship between a certain node in the database schema is inconsistent with the relationship in the process flow graph. The explanation of the database schema conforming to the preset graph structure requirement and the database schema not conforming to the preset graph structure requirement is similar to the meaning of the database schema conforming to the preset process flow graph structure and the database schema not conforming to the preset process flow graph structure, which will not be repeated here. The database schema conforms to the preset relationship normal form: if the database schema conforms to the first normal form in the relationship normal form, it means that each field in the database schema is a single value, there is no repeated group or multi-value field; if the database schema conforms to the second normal form in the relationship normal form, it means that each field in the database schema is a single value, there is no repeated group or multi-value field and all non-primary attributes completely depend on the primary key; if the database schema conforms to the third normal form in the relationship normal form, it means that the database schema conforms to the first normal form, the second normal form and all non-primary attributes directly depend on the primary key, but cannot depend on other non-primary attributes. The non-primary attribute means the field in the table that is not the primary key. It needs to be explained that the process flow graph structure of the application is constructed based on the existing technology of UML modeling specification and SQL DDL database definition language.

[0092] S2, establishing an automated data acquisition interface, wherein the automated data acquisition interface comprises a device layer interface, a measurement layer interface and a file layer interface.

[0093] It needs to be explained that the step of establishing an automated data acquisition interface is to develop or configure a gateway, real-time capture actual operating parameters (temperature, power, etc.) from process equipment through standard protocols (such as SECS / GEM), integrate with measurement instruments, automatically upload measurement result files, and automatically associate with wafer records in the database through wafer ID, develop analysis programs, automatically read simulation result files, extract key data (such as breakdown voltage) and store into the database. The purpose of establishing an automated data acquisition interface in the present application is to automatically obtain data from equipment, measurement instruments and files, reducing manual intervention, greatly improving the speed and efficiency of data acquisition, saving time. At the same time, it also avoids errors and deviations that may occur in the process of manual data collection, ensuring the accuracy and reliability of the data, and providing a high-quality data basis for subsequent data analysis and model construction. The device layer interface is the interface between the automated data acquisition interface and the physical equipment, and the main function is to directly collect data from various production equipment (such as lithography machines, oxidation furnaces). The measurement layer interface is the interface between the automated data acquisition interface and the measurement equipment (such as electrical parameter testers, optical measurement equipment), which is used to ensure the time synchronization of measurement data and equipment operation data for accurate analysis. The file layer interface is the interface between the automated data acquisition interface and the file storage system (such as log files), which is used to parse the data in the file into a structured format for storage in the database.

[0094] S3, constructing a database framework according to the automated data acquisition interface and database design specifications, and confirming a set of to-be-simulated process parameters.

[0095] Importantly, the step of constructing a database framework according to the automated data acquisition interface and database design specifications is to create an empty database structure (create tables, define indexes, etc.) in a database management system (such as Neo4j, PostgreSQL), then deploy and configure each automated data acquisition interface developed previously, and finally connect the output of the automated data acquisition interface to the corresponding tables and fields of the already created database framework. The set of to-be-simulated process parameters refers to the collection of all to-be-simulated process parameters that need to be simulated in the super-junction MOSFET process simulation. The present application combines the automated data acquisition interface and the database design specification to construct the database framework, so that the database has a clear structure and good organization, facilitating data storage, management and query, and improving the performance and scalability of the database.

[0096] S4, obtaining an associated data set based on the set of to-be-simulated process parameters and the database framework, wherein the set of to-be-simulated process parameters includes a plurality of to-be-simulated process parameters, the associated data set includes a plurality of associated data, and the to-be-simulated process parameters and the associated data correspond one-to-one.

[0097] In detail, the obtaining of the associated data set based on the to-be-simulated process parameter set and the database framework comprises:

[0098] The to-be-simulated process parameters are sequentially extracted from the to-be-simulated process parameter set, and a pre-constructed workflow engine is used for file conversion of the to-be-simulated process parameters to obtain a simulation input file, wherein the workflow engine comprises a result analysis module.

[0099] The simulation input file is executed by using a pre-constructed TCAD software to obtain a simulation output text file and a simulation output binary file.

[0100] The simulation output text file is read by using the result analysis module to obtain electrical index parameters, and the simulation output binary file is read by using the result analysis module to obtain device internal distribution parameters, wherein the device internal distribution parameters comprise doping concentration distribution data and electric field intensity distribution data.

[0101] The device internal distribution parameters are subjected to feature extraction according to the database framework to obtain feature parameters, and the structured simulation result data is generated according to the electrical index parameters and the feature parameters.

[0102] The structured simulation result data is associated and bound with the to-be-simulated process parameters to obtain associated data, and the associated data is summarized to obtain an associated data set.

[0103] It should be explained that the to-be-simulated process parameter set of the present application is set by a process engineer or a researcher according to actual process requirements. For example, the to-be-simulated process parameters are epitaxial growth temperature, ion implantation dose, photolithography alignment accuracy, etc. The workflow engine is an automatic tool responsible for triggering and executing each step in order, including file conversion, submission of simulation tasks, monitoring of states, and calling of the result analysis module. It is used to ensure that the entire process runs automatically without human intervention. File conversion refers to the process of converting to-be-simulated process parameters from one format to another format. For example, the to-be-simulated process parameters are converted from the JSON format to the input deck file format of the TCAD software. The simulation input file is a file generated after file conversion, which is used for input into the simulation software for simulation. The result analysis module is used to analyze the file output by the simulation software, which can extract key electrical index parameters and device internal distribution parameters from the simulation output file and convert them into a structured data format. The TCAD software is a professional software for semiconductor device and process simulation, which can simulate the physical behavior of the device according to the input process parameters and device structure, and output the electrical performance and internal physical field distribution of the device. The simulation output text file is a text file containing electrical index parameters output by the TCAD software after simulation is completed, which records the simulation results such as breakdown voltage, on-resistance, threshold voltage, etc.

[0104] It can be understood that the simulation output binary file refers to a binary file output by the TCAD software after simulation is completed, containing distribution parameters in the device, and storing physical field distribution data in the device in a binary format. The electrical index parameter refers to a parameter extracted from a simulation output text file and used to describe the electrical performance of the device. The internal distribution parameter of the device refers to a parameter extracted from the simulation output binary file and used to describe the internal physical field distribution of the device. The doping concentration distribution data refers to the distribution of the doping concentration in the device at different positions. The electric field intensity distribution data refers to the distribution of the electric field intensity in the device at different positions. The doping concentration distribution data and the electric field intensity distribution data of the present application are stored in the form of two-dimensional arrays, and each array element represents a doping concentration value and an electric field intensity value at a position. The structured simulation result data refers to structured data format composed of electrical index parameters and feature parameters. The association data refers to data obtained by associating and binding the structured simulation result data with the to-be-simulated process parameters. The association data set refers to a collection of association data obtained after all to-be-simulated process parameters are simulated.

[0105] In detail, the feature extraction of the internal distribution parameter of the device according to the database framework to obtain the feature parameter comprises:

[0106] Numerical analysis is performed on the electric field intensity distribution data in the internal distribution parameter of the device to obtain an electric field peak value and an electric field peak coordinate;

[0107] A blank dictionary is created according to the database framework, and the electric field peak value and the electric field peak coordinate are stored by using the blank dictionary to obtain stored electric field parameters;

[0108] According to a preset doping concentration integral path, a path point set is extracted from the doping concentration distribution data, path points are sequentially extracted from the path point set, and the following operations are performed on the extracted path points:

[0109] It is judged whether the path point is located on a preset grid point;

[0110] If the path point is located on the preset grid point, the doping concentration value of the path point is obtained, and the doping concentration value is taken as a doping concentration integral value;

[0111] If the path point is not located on the preset grid point, the path point coordinates corresponding to the path point are confirmed from the doping concentration distribution data, and a set of adjacent grid point coordinates is obtained according to the path point coordinates;

[0112] Adjacent grid point coordinates are sequentially extracted from the set of adjacent grid point coordinates, the path point coordinates and the adjacent grid point coordinates are combined to obtain a path coordinate group, and the grid weight and the path distance of the path coordinate group are calculated;

[0113] The grid weight and the path distance are respectively aggregated to obtain a grid weight set and a path distance set, wherein the path coordinate group is one-to-one corresponding to the grid weight and the path distance;

[0114] The doping concentration integral value is obtained based on the adjacent grid point coordinate set, the grid weight set and the path distance set, and the stored electric field parameter and the doping concentration integral value are used as the characteristic parameters.

[0115] It should be explained that the numerical analysis of the electric field intensity distribution data in the device internal distribution parameter refers to the numerical analysis of the electric field intensity distribution data by using the numerical analysis method. The electric field peak value refers to the maximum value of the electric field intensity in the device. The electric field peak coordinate refers to the coordinate where the maximum value of the electric field intensity is located. The blank dictionary is an initially empty dictionary used for storing the extracted characteristic parameters. The doping concentration integral path refers to a path in the device which is set in advance and used for calculating the integral value of the doping concentration. The path point set refers to the set of path points obtained after discretization of the doping concentration integral path. The grid point refers to a discrete point defined in the doping concentration distribution data, and the coordinates of the points correspond one-to-one to the doping concentration values. The doping concentration value refers to the numerical value of the doping concentration at a certain position. The path point coordinate refers to the coordinate of the path point in the device coordinate system. The stored electric field parameter refers to the electric field peak value and the electric field peak coordinate which have been stored in the blank dictionary. The adjacent grid point coordinate set obtained according to the path point coordinate refers to keeping the horizontal coordinate of the path point coordinate unchanged, performing plus one and minus one operations on the vertical coordinate to obtain two adjacent grid point coordinates, keeping the vertical coordinate of the path point coordinate unchanged, performing plus one and minus one operations on the horizontal coordinate to obtain two adjacent grid point coordinates, and aggregating all the adjacent grid point coordinates to obtain the adjacent grid point coordinate set. The path coordinate group refers to a group of coordinates composed of the path point coordinate and the adjacent grid point coordinate.

[0116] In detail, the calculation of the grid weight and the path distance of the path coordinate group comprises:

[0117] The adjacent grid point horizontal coordinate and the adjacent grid point vertical coordinate are confirmed according to the adjacent grid point coordinates in the path coordinate group, and the path point horizontal coordinate and the path point vertical coordinate are confirmed according to the path point coordinates in the path coordinate group;

[0118] The horizontal coordinate difference value is calculated based on the adjacent grid point horizontal coordinate and the path point horizontal coordinate, and the vertical coordinate difference value is calculated based on the adjacent grid point vertical coordinate and the path point vertical coordinate;

[0119] The grid weight is calculated according to the horizontal coordinate difference value and the vertical coordinate difference value, wherein the grid weight is the product of the horizontal coordinate difference value and the vertical coordinate difference value;

[0120] The path distance is calculated according to the adjacent grid point horizontal coordinate, the adjacent grid point vertical coordinate, the path point horizontal coordinate and the path point vertical coordinate.

[0121] It should be explained that the adjacent grid point abscissa and the adjacent grid point ordinate respectively refer to the abscissa value of the adjacent grid point and the ordinate value of the adjacent grid point. The path point abscissa and the path point ordinate respectively refer to the abscissa value of the path point and the ordinate value of the path point. The abscissa difference value refers to the absolute difference value between the abscissa of the adjacent grid point and the abscissa of the path point. The ordinate difference value refers to the absolute difference value between the ordinate of the adjacent grid point and the ordinate of the path point. The abscissa difference value and the ordinate difference value are used to quantify the relative position relationship of the path point and each adjacent grid point around the path point in the X and Y directions. The closer the path distance, the smaller the coordinate difference value, and the greater the grid weight. Through the calculation of the grid weight, when the path point continuously moves, the estimated value will also smoothly transition without jumping, which is crucial for the subsequent calculation of the doping concentration integral value. The calculation of the path distance according to the abscissa of the adjacent grid point, the ordinate of the adjacent grid point, the abscissa of the path point and the ordinate of the path point refers to the calculation of the abscissa of the adjacent grid point, the ordinate of the adjacent grid point, the abscissa of the path point and the ordinate of the path point using the Euclidean distance to obtain the path distance. In the calculation of the doping concentration integral value, the length of the line segment between each path point and the adjacent grid point needs to be known, therefore, the Euclidean distance between the adjacent grid point coordinates and the path point is calculated.

[0122] In detail, the doping concentration integral value is obtained based on the adjacent grid point coordinate set, the grid weight set and the path distance set, which includes:

[0123] The adjacent grid point concentration value set is obtained according to the adjacent grid point coordinate set, and the path point doping concentration value is calculated based on the grid weight set and the adjacent grid point concentration value set.

[0124] The path distance is sequentially extracted from the path distance set, the target path coordinate group is confirmed according to the path distance, and the target adjacent grid point concentration value is confirmed according to the target path coordinate group.

[0125] The trapezoidal area is calculated according to the path point doping concentration value, the path distance and the target adjacent grid point concentration value, the trapezoidal areas are summarized to obtain a trapezoidal area set.

[0126] The trapezoidal area sum of the trapezoidal area set is calculated, and the trapezoidal area sum is taken as the doping concentration integral value.

[0127] It should be explained that the adjacent grid point concentration value refers to the doping concentration value of the adjacent grid point in the doping concentration distribution data. The adjacent grid point concentration value set refers to the set of doping concentration values of all adjacent grid points around the path point. The target path coordinate group refers to the path coordinate group corresponding to the current extracted path distance. The target adjacent grid point concentration value refers to the doping concentration value corresponding to the adjacent grid point in the target path coordinate group. In the step of calculating the trapezoidal area according to the path point doping concentration value, the path distance and the target adjacent grid point concentration value, the trapezoidal area is { (path point doping concentration value + target adjacent grid point concentration value) path distance) / 2. The trapezoidal area set refers to the set of all trapezoidal areas. The trapezoidal area sum refers to the sum obtained by accumulating all trapezoidal areas. The present application applies the trapezoidal rule for numerical integration to calculate the contribution of each small path distance to the doping concentration integral value. At the same time, the present application combines bilinear interpolation and trapezoidal integration method to ensure that the doping concentration integral value extracted from discrete grid data has high reliability.

[0128] In detail, the calculation formula of the path point doping concentration value is as follows:

[0129] Among them, represents the path point doping concentration value, represents the number of grid weights in the grid weight set, represents the first grid weight in the grid weight set, represents the adjacent grid point concentration value corresponding to the first grid weight.

[0130] Importantly, the present application obtains the accurate doping concentration value of each path point on the doping concentration integral path, which is the data basis for integral calculation. TCAD simulation only provides the concentration value on the regular grid point, while the doping concentration integral path of the present application is arbitrarily set, and the path point is rarely exactly on these grid points. Therefore, the concentration value of those path points not on the grid points must be obtained by calculation, and the formula provides a scientific and reliable estimation method.

[0131] S5, constructing an optimized forward prediction machine learning model and an optimized reverse design machine learning model according to the associated data set, and completing the construction of a process flow database based on the preparation of super-junction MOSFET based on the optimized forward prediction machine learning model and the optimized reverse design machine learning model.

[0132] In detail, the step of constructing an optimized forward prediction machine learning model and an optimized reverse design machine learning model according to the associated data set comprises:

[0133] The correlation data set is stored in a database framework to obtain a virtual experimental data set, and process flow tracing function, SPC control chart analysis function and interactive correlation analysis tools are developed based on the virtual experimental data set;

[0134] A data analysis visualization unit is obtained based on the process flow tracing function, the SPC control chart analysis function and the interactive correlation analysis tools;

[0135] A forward prediction machine learning model is constructed based on the virtual experimental data set, and a reverse design machine learning model is constructed according to the forward prediction machine learning model;

[0136] A system user interface is created, the data analysis visualization unit, the forward prediction machine learning model and the reverse design machine learning model are integrated in the system user interface, and an integrated system user interface is obtained;

[0137] Role access permissions are set for the integrated system user interface, and an updated system user interface is obtained;

[0138] Optimized forward prediction machine learning models and optimized reverse design machine learning models are obtained based on the updated system user interface.

[0139] It should be explained that the virtual experiment data set refers to the associated data set stored in the database framework. The present application stores the associated data set in the database framework to ensure the integrity and accessibility of the data. The process flow traceability function can realize the traceability of all process steps and the process parameters to be simulated experienced by the wafer ID. Statistical analysis is performed on a certain key performance parameter (such as Rds(on)) in the virtual experiment data set, the mean and standard deviation are calculated, and a control chart with upper and lower control limits is drawn. The control chart is used for the stability of the process parameters to be simulated. If the data point is not located in the control chart, it means that the process parameters to be simulated have large fluctuation and poor robustness, and are not suitable for production. The SPC control chart analysis function is a statistical process control monitoring of the process parameters and the electrical performance indicators. The interactive correlation analysis tool is a tool that can generate scatter plots and trend charts for any process parameters and performance parameters. The data analysis visualization unit refers to a module for displaying the process flow traceability function, the SPC control chart analysis function and the interactive correlation analysis tool to the user. The steps of constructing the forward prediction machine learning model based on the virtual experiment data set are: dividing the virtual experiment data set into a training set and a test set, training the forward prediction machine learning model on the training set using the gradient boosting decision tree algorithm, and verifying the prediction accuracy through the test set. The construction of the reverse design machine learning model according to the forward prediction machine learning model refers to the construction of the reverse design machine learning model around the trained forward prediction machine learning model using the Bayesian optimization framework. The present application constructs the forward prediction machine learning model and the reverse design machine learning model, which are prior art and will not be described here. The process flow traceability function backend uses the Cypher language of the graph database (such as Neo4j) for efficient relationship traversal, and the recursive query of the relational database can also be used; the front end directly displays the whole process through timeline, table or flowchart components. The SPC control chart analysis function backend relies on Pandas, NumPy and other libraries to calculate the mean, standard deviation and control limit, and the front end uses ECharts or Highcharts chart library to render the data into monitoring charts to evaluate the process stability. The interactive correlation analysis tool backend receives any parameter combination passed by the front end through flexible RESTful API and executes dynamic queries, and the front end provides drop-down selectors and chart linkage to enable users to generate scatter plots or trend charts in real time.

[0140] It can be understood that the creation of the system user interface refers to developing a unified Web application using front-end technologies such as React, Vue.js. The integrated system user interface refers to the interface obtained after integrating the data analysis visualization unit, the forward prediction machine learning model, and the reverse design machine learning model in the system user interface. Users can access the data analysis visualization unit, the forward prediction machine learning model, and the reverse design machine learning model through the interface to achieve one-stop data analysis and process design. The setting of role access permissions for the integrated system user interface refers to defining different user roles (such as administrators, engineers, analysts, etc.), setting access permissions for each role, and limiting the function modules and data that users can access. For example, administrators can access all function modules and data, engineers can only access the forward prediction machine learning model and the reverse design machine learning model, and analysts can only access the data analysis visualization unit. The updated system user interface refers to the interface obtained after setting role access permissions for the integrated system user interface. The interface displays different function modules and data according to the user's role to ensure that users can only access the content within their permission scope.

[0141] In detail, the obtaining of the optimized forward prediction machine learning model and the optimized reverse design machine learning model based on the updated system user interface comprises:

[0142] An incremental data set is obtained, wherein the incremental data set includes actual production process parameters and measured performance data. The incremental data set is merged with the virtual experimental data set to obtain an updated experimental data set.

[0143] The updated experimental data set is used to train and optimize the forward prediction machine learning model and the reverse design machine learning model in the updated system user interface, and the optimized forward prediction machine learning model and the optimized reverse design machine learning model are obtained.

[0144] It should be explained that the actual production process parameters refer to the process parameters used in the actual production process. The measured performance data refer to the device performance data obtained by measurement in the actual production process. Said merging the incremental data set with the virtual experiment data set refers to converting the actual production process parameters and the measured performance data in the incremental data set into the same format as the virtual experiment data set, and adding the data in the incremental data set as new records to the virtual experiment data set. Said method for training and optimizing the forward prediction machine learning model and the reverse design machine learning model in the system user interface by using the updated experiment data set is the same as the method for constructing the forward prediction machine learning model based on the virtual experiment data set and constructing the reverse design machine learning model according to the forward prediction machine learning model, which will not be repeated here. The optimized forward prediction machine learning model refers to the model obtained by training and optimizing the forward prediction machine learning model in the system user interface by using the updated experiment data set. The optimized reverse design machine learning model refers to the model constructed around the trained optimized forward prediction machine learning model by using the Bayesian optimization framework. The optimized forward prediction machine learning model can predict the results of the preparation process according to the input process parameters, help engineers understand the process effect in advance, provide reference for process adjustment and optimization, reduce the number of trial and error, and improve the production efficiency. The optimized reverse design machine learning model can deduce suitable process parameters according to the expected preparation results, provide a new idea and method for process design, and help to quickly design a preparation process that meets the requirements. The present application integrates the two models into the database framework, enriches the functions of the database, so that the database is not only a data storage platform, but also has the ability of intelligent analysis and decision support, and provides more comprehensive services for the research and production of super junction MOSFET preparation process.

[0145] The present application is to solve the problems described in the background art, the present application receives process flow database construction instructions, confirms the super junction MOSFET process flow knowledge according to the process flow database construction instructions, and obtains the database design specification according to the super junction MOSFET process flow knowledge, the present application receives the construction instruction, so that the system is clear about the task requirement, and can carry out the subsequent work targetedly, avoids blind operation, improves the efficiency and accuracy of building the database, confirms the super junction MOSFET process flow knowledge, systematically organizes the dispersed professional knowledge, provides a solid theoretical basis and data source for the design of subsequent database, ensures the professionalism and integrity of the database content, at the same time, according to the process flow knowledge acquisition database design specification, it is ensured that the structure and content of the database meet the industry standards and actual application requirements, improve the usability and maintainability of the database, establish an automatic data acquisition interface, wherein the automatic data acquisition interface includes: device layer interface, measurement layer interface and file layer interface, the automatic data acquisition interface of the present application can automatically acquire data from equipment, measuring instruments and files, reduces manual intervention, greatly improves the speed and efficiency of data acquisition, saves time, avoids errors and deviations that may occur in the process of manual data collection, ensures the accuracy and reliability of the data, provides a high-quality data basis for subsequent data analysis and model construction, according to the automatic data acquisition interface and the database design specification, a database framework is constructed, and a set of to-be-simulated process parameters is confirmed, the present application combines the automatic data acquisition interface and the database design specification to construct the database framework, so that the database has a clear structure and good organization, facilitating data storage, management and query, improving the performance and scalability of the database, based on the set of to-be-simulated process parameters and the database framework, a set of associated data is obtained, wherein the set of to-be-simulated process parameters includes a plurality of to-be-simulated process parameters, the set of associated data includes a plurality of associated data, and the to-be-simulated process parameters correspond to the associated data one by one, the present application obtains the set of associated data corresponding to the to-be-simulated process parameters one by one, establishes the connection between the process parameters and the related data, which helps to deeply analyze the influence of process parameters on the preparation process, and provides strong data support for optimizing the process, according to the set of associated data, an optimization forward prediction machine learning model and an optimization reverse design machine learning model are constructed, based on the optimization forward prediction machine learning model and the optimization reverse design machine learning model, the process flow database construction based on the super junction MOSFET preparation is completed, the optimization forward prediction machine learning model can predict the preparation process result according to the input process parameters, help engineers understand the process effect in advance, provide reference for process adjustment and optimization, reduce the number of trial and error, improve the production efficiency, the optimization reverse design machine learning model can deduce the appropriate process parameters according to the expected preparation result, provide a new idea and method for process design, and help to quickly design a preparation process that meets the requirements. Therefore, the present application can improve the optimization efficiency of the preparation process and the consistency of the data.

[0146] As Figure 2 shown in FIG. 1, it is a functional module diagram of the process flow database construction system based on super junction MOSFET preparation provided by an embodiment of the present application.

[0147] The process flow database construction system based on super junction MOSFET preparation 100 can be installed in an electronic device. According to the functions to be realized, the process flow database construction system based on super junction MOSFET preparation 100 can include a database design specification confirmation module 101, a data acquisition interface establishment module 102, a database framework construction module 103, and a database construction completion module 104. The modules of the present application can also be referred to as units, which refer to a series of computer program segments that can be executed by an electronic device processor and can complete a fixed function, which are stored in the memory of the electronic device.

[0148] The database design specification confirmation module 101 is configured to receive a process flow database construction instruction, confirm super junction MOSFET process flow knowledge according to the process flow database construction instruction, and obtain a database design specification according to the super junction MOSFET process flow knowledge.

[0149] The data acquisition interface establishment module 102 is configured to establish an automated data acquisition interface, wherein the automated data acquisition interface includes a device layer interface, a measurement layer interface, and a file layer interface.

[0150] The database framework construction module 103 is configured to construct a database framework according to the automated data acquisition interface and the database design specification, and confirm a set of process parameters to be simulated.

[0151] The database construction completion module 104 is configured to obtain a set of associated data based on the set of process parameters to be simulated and the database framework, wherein the set of process parameters to be simulated includes a plurality of process parameters to be simulated, the set of associated data includes a plurality of associated data, the process parameters to be simulated and the associated data correspond to each other, an optimized forward prediction machine learning model and an optimized reverse design machine learning model are constructed according to the set of associated data, and the process flow database based on super junction MOSFET preparation is constructed based on the optimized forward prediction machine learning model and the optimized reverse design machine learning model.

[0152] In detail, the modules in the process flow database construction system based on super junction MOSFET preparation 100 in the embodiment of the present application adopt the same technical means as the process flow database construction method based on super junction MOSFET preparation described in the above Figure 1 , and can produce the same technical effects, which will not be described here again.

[0153] AsFigure 3 Fig. 1 is a structural schematic diagram of an electronic device for implementing the process flow database construction method based on super-junction MOSFET preparation according to an embodiment of the present application.

[0154] The electronic device 1 can include a processor 10, a memory 11 and a bus 12, and can further include a computer program stored in the memory 11 and executable on the processor 10, such as the process flow database construction method program based on super-junction MOSFET preparation.

[0155] The memory 11 includes at least one type of readable storage medium, such as flash memory, mobile hard disk, multimedia card, card-type memory (e.g. SD or DX memory, etc.), magnetic memory, disk, optical disk, etc. In some embodiments, the memory 11 can be an internal storage unit of the electronic device 1, such as a mobile hard disk of the electronic device 1. In other embodiments, the memory 11 can also be an external storage device of the electronic device 1, such as a plug-in mobile hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc. equipped on the electronic device 1. Further, the memory 11 includes both the internal storage unit and the external storage device of the electronic device 1. The memory 11 can be used not only to store application software and various data installed on the electronic device 1, such as the code of the process flow database construction method program based on super-junction MOSFET preparation, but also to temporarily store data that has been output or will be output.

[0156] The processor 10 can be composed of integrated circuits in some embodiments, such as a single packaged integrated circuit or a plurality of packaged integrated circuits with the same or different functions, including one or more combinations of central processing unit (CPU), microprocessor, digital processing chip, graphics processor and various control chips, etc. The processor 10 is the control unit of the electronic device, which connects various components of the entire electronic device through various interfaces and lines, executes or runs programs or modules stored in the memory 11 (such as the process flow database construction method program based on super-junction MOSFET preparation, etc.), and calls data stored in the memory 11, to perform various functions and process data of the electronic device 1.

[0157] The bus 12 can be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus, etc. The bus 12 can be divided into an address bus, a data bus, a control bus, etc. The bus 12 is configured to enable connection and communication between the memory 11, the at least one processor 10, etc.

[0158] Figure 3 Only the electronic device with components is shown, and those skilled in the art can understand that, Figure 3 The structure shown does not constitute a limitation on the electronic device 1, and can include fewer or more components than shown, or combine certain components, or different component arrangements.

[0159] For example, although not shown, the electronic device 1 can also include a power supply (such as a battery) to power each component. Preferably, the power supply can be logically connected to the at least one processor 10 through a power management device, so that the power management device can implement functions such as charge management, discharge management, and power consumption management. The power supply can also include one or more direct current or alternating current power supplies, recharging devices, power supply fault detection circuits, power supply converters or inverters, power supply status indicators, etc. The electronic device 1 can also include various sensors, Bluetooth modules, Wi-Fi modules, etc., which are not described here.

[0160] Further, the electronic device 1 can also include a network interface, which can optionally include a wired interface and / or a wireless interface (such as a WI-FI interface, a Bluetooth interface, etc.), and is typically used to establish a communication connection between the electronic device 1 and other electronic devices.

[0161] Optionally, the electronic device 1 can also include a user interface, which can be a display (Display), an input unit (such as a keyboard (Keyboard)), and optionally a standard wired interface, a wireless interface. Optionally, in some embodiments, the display can be an LED display, a liquid crystal display, a touch liquid crystal display, an OLED (Organic Light-Emitting Diode) touch, etc. The display can also be appropriately referred to as a display screen or a display unit, and is used to display information processed in the electronic device 1 and to display a visualized user interface.

[0162] The process flow database construction method program stored in the memory 11 in the electronic device 1 is a combination of multiple instructions, which can realize the following when running in the processor 10:

[0163] receiving a process flow database construction instruction, confirming super junction MOSFET process flow knowledge according to the process flow database construction instruction, and obtaining database design specifications according to the super junction MOSFET process flow knowledge;

[0164] establishing an automatic data acquisition interface, wherein the automatic data acquisition interface includes a device layer interface, a measurement layer interface, and a file layer interface;

[0165] constructing a database framework according to the automatic data acquisition interface and the database design specifications, and confirming a set of process parameters to be simulated;

[0166] obtaining a set of associated data based on the set of process parameters to be simulated and the database framework, wherein the set of process parameters to be simulated includes multiple process parameters to be simulated, the set of associated data includes multiple associated data, and the process parameters to be simulated and the associated data correspond one-to-one;

[0167] constructing an optimized forward prediction machine learning model and an optimized reverse design machine learning model according to the set of associated data, and completing the process flow database construction based on the super junction MOSFET preparation based on the optimized forward prediction machine learning model and the optimized reverse design machine learning model.

[0168] Specifically, the specific implementation method of the processor 10 to the above instructions can refer to Figures 1 to 3 The descriptions of related steps in the corresponding embodiments are not repeated here.

[0169] Further, the modules / units integrated in the electronic device 1 can be stored in a computer readable storage medium if they are realized in the form of software function units and sold or used as independent products. The computer readable storage medium can be volatile or non-volatile. For example, the computer readable medium can include any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory).

[0170] The application also provides a computer readable storage medium, which stores a computer program, and the computer program can realize the following when executed by a processor of an electronic device:

[0171] Receive a process flow database construction instruction, confirm super junction MOSFET process flow knowledge according to the process flow database construction instruction, and acquire database design specifications according to the super junction MOSFET process flow knowledge;

[0172] An automatic data acquisition interface is established, wherein the automatic data acquisition interface comprises a device layer interface, a measurement layer interface and a file layer interface;

[0173] A database framework is constructed according to the automatic data acquisition interface and the database design specifications, and a set of to-be-simulated process parameters is confirmed;

[0174] A set of associated data is acquired based on the set of to-be-simulated process parameters and the database framework, wherein the set of to-be-simulated process parameters comprises a plurality of to-be-simulated process parameters, the set of associated data comprises a plurality of associated data, and the to-be-simulated process parameters and the associated data are in one-to-one correspondence;

[0175] An optimized forward prediction machine learning model and an optimized reverse design machine learning model are constructed according to the set of associated data, and a process flow database construction based on super junction MOSFET preparation is completed based on the optimized forward prediction machine learning model and the optimized reverse design machine learning model.

[0176] In several embodiments provided in the present application, it should be understood that the disclosed devices, systems and methods can be implemented in other ways. For example, the above-described system embodiments are only illustrative, and actual implementation can have another division way.

[0177] The modules described as separate components can or can not be physically separated, and the components shown as modules can or can not be physical units, i.e., they can be located in one place or distributed on multiple network units. Part or all of the modules can be selected to achieve the purpose of the embodiment according to actual needs.

[0178] In addition, each functional module in each embodiment of the present application can be integrated in one processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of hardware plus software functional modules.

[0179] It is obvious for those skilled in the art that the present application is not limited to the details of the above exemplary embodiments, and the present application can be implemented in other specific forms without departing from the spirit or essential characteristics of the present application.

[0180] Finally, it should be noted that the above examples are merely intended to illustrate the technical solutions of the present application and not to limit the present application. Although the present application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present application.

Claims

1. A process flow database construction method based on super-junction MOSFET preparation, characterized in that, The method comprises: receiving process flow database construction instructions, confirming super junction MOSFET process flow knowledge according to the process flow database construction instructions, and acquiring database design specifications according to the super junction MOSFET process flow knowledge; establishing an automatic data acquisition interface, wherein the automatic data acquisition interface comprises a device layer interface, a measurement layer interface, and a file layer interface; constructing a database framework according to the automatic data acquisition interface and the database design specifications, and confirming a set of process parameters to be simulated; acquiring a set of associated data based on the set of process parameters to be simulated and the database framework, wherein the set of process parameters to be simulated comprises a plurality of process parameters to be simulated, the set of associated data comprises a plurality of associated data, and the process parameters to be simulated and the associated data correspond one-to-one; constructing an optimized forward prediction machine learning model and an optimized reverse design machine learning model according to the set of associated data, comprising: storing the set of associated data in the database framework to obtain a set of virtual experimental data, and developing a process flow tracing function, an SPC control chart analysis function, and an interactive correlation analysis tool based on the set of virtual experimental data; acquiring a data analysis visualization unit based on the process flow tracing function, the SPC control chart analysis function, and the interactive correlation analysis tool; constructing a forward prediction machine learning model based on the set of virtual experimental data, and constructing a reverse design machine learning model according to the forward prediction machine learning model; creating a system user interface, integrating the data analysis visualization unit, the forward prediction machine learning model, and the reverse design machine learning model in the system user interface to obtain an integrated system user interface; setting role access permissions for the integrated system user interface to obtain an updated system user interface; acquiring the optimized forward prediction machine learning model and the optimized reverse design machine learning model based on the updated system user interface, comprising: acquiring an incremental data set, wherein the incremental data set comprises actual production process parameters and measured performance data, merging the incremental data set with the virtual experimental data set to obtain an updated experimental data set; training and optimizing the forward prediction machine learning model and the reverse design machine learning model in the updated system user interface using the updated experimental data set to obtain the optimized forward prediction machine learning model and the optimized reverse design machine learning model; completing the process flow database construction based on the super junction MOSFET preparation based on the optimized forward prediction machine learning model and the optimized reverse design machine learning model.

2. The process flow database construction method based on super junction MOSFET fabrication according to claim 1, wherein, The database design specifications are acquired according to the super junction MOSFET process flow knowledge, comprising: acquiring a set of data entities and a set of data entity relationships according to the super junction MOSFET process flow knowledge, sequentially extracting data entities from the set of data entities, and performing the following operations on the extracted data entities: confirming a node type according to the data entity, and confirming a target data entity relationship from the set of data entity relationships according to the data entity; confirming a relationship type according to the target data entity relationship, constructing a database schema according to the node type and the relationship type, and judging whether the database schema conforms to a preset process flow chart structure; If the database mode conforms to the preset process flowchart structure, the process flowchart structure is taken as the final database mode; If the database mode does not conform to the preset process flowchart structure, it is judged whether the database mode conforms to the preset graph structure requirement; If the database mode conforms to the preset graph structure requirement, the pre-constructed graph database is taken as the final database mode; If the database mode does not conform to the preset graph structure requirement, it is judged whether the database mode conforms to the preset relation schema; If the database mode conforms to the preset relation schema, the pre-constructed relation database is taken as the final database mode; The final database mode is summarized to obtain the database design specification.

3. The process flow database construction method based on super junction MOSFET fabrication according to claim 2, wherein, The obtaining of the associated data set based on the to-be-simulated process parameter set and the database framework comprises: The to-be-simulated process parameters are extracted from the to-be-simulated process parameter set in sequence, and the to-be-simulated process parameters are file-converted by using the pre-constructed workflow engine to obtain simulation input files, wherein the workflow engine comprises a result analysis module; The simulation input files are executed by simulation operation by using the pre-constructed TCAD software to obtain simulation output text files and simulation output binary files; The simulation output text files are read by using the result analysis module to obtain electrical index parameters, and the simulation output binary files are read by using the result analysis module to obtain device internal distribution parameters, wherein the device internal distribution parameters comprise doped concentration distribution data and electric field intensity distribution data; The device internal distribution parameters are feature-extracted according to the database framework to obtain feature parameters, and the structured simulation result data is generated according to the electrical index parameters and the feature parameters; The structured simulation result data is associated and bound with the to-be-simulated process parameters to obtain associated data, and the associated data is summarized to obtain an associated data set.

4. The process flow database construction method based on super junction MOSFET fabrication according to claim 3, wherein, The feature-extracting of the device internal distribution parameters according to the database framework to obtain feature parameters comprises: The electric field intensity distribution data in the device internal distribution parameters are subjected to numerical analysis to obtain an electric field peak value and an electric field peak coordinate; A blank dictionary is created according to the database framework, the electric field peak value and the electric field peak coordinate are stored by using the blank dictionary to obtain stored electric field parameters; Path point sets are extracted from the doped concentration distribution data according to a preset doped concentration integral path, path points are extracted from the path point sets in sequence, and the following operations are performed on the extracted path points: It is judged whether the path point is located on a preset grid point; If the path point is located on the preset grid point, a doped concentration value of the path point is obtained, and the doped concentration value is taken as a doped concentration integral value; If the path point is not located on the preset grid point, path point coordinates corresponding to the path point are confirmed from the doped concentration distribution data, and adjacent grid point coordinate sets are obtained according to the path point coordinates; Adjacent grid point coordinates are extracted from the adjacent grid point coordinate sets in sequence, the path point coordinates and the adjacent grid point coordinates are combined to obtain path coordinate groups, and grid weights and path distances of the path coordinate groups are calculated; The grid weights and the path distances are summarized respectively to obtain a grid weight set and a path distance set, wherein the path coordinate groups correspond to the grid weights and the path distances one by one; The doping concentration integral value is obtained based on the adjacent grid point coordinate set, the grid weight set and the path distance set, and the stored electric field parameter and the doping concentration integral value are taken as characteristic parameters.

5. The process flow database construction method based on super junction MOSFET fabrication according to claim 4, wherein, The grid weight and the path distance of the path coordinate group are calculated, including: The adjacent grid point horizontal coordinate and the adjacent grid point vertical coordinate are confirmed according to the adjacent grid point coordinates in the path coordinate group, and the path point horizontal coordinate and the path point vertical coordinate are confirmed according to the path point coordinates in the path coordinate group; The horizontal coordinate difference value is calculated based on the adjacent grid point horizontal coordinate and the path point horizontal coordinate, and the vertical coordinate difference value is calculated based on the adjacent grid point vertical coordinate and the path point vertical coordinate; The grid weight is calculated according to the horizontal coordinate difference value and the vertical coordinate difference value, wherein the grid weight is the product of the horizontal coordinate difference value and the vertical coordinate difference value; The path distance is calculated according to the adjacent grid point horizontal coordinate, the adjacent grid point vertical coordinate, the path point horizontal coordinate and the path point vertical coordinate.

6. The process flow database construction method based on super junction MOSFET fabrication according to claim 5, wherein, The doping concentration integral value is obtained based on the adjacent grid point coordinate set, the grid weight set and the path distance set, including: The adjacent grid point concentration value set is obtained according to the adjacent grid point coordinate set, and the path point doping concentration value is calculated based on the grid weight set and the adjacent grid point concentration value set; The path distance is sequentially extracted from the path distance set, the target path coordinate group is confirmed according to the path distance, and the target adjacent grid point concentration value is confirmed according to the target path coordinate group; The trapezoidal area is calculated according to the path point doping concentration value, the path distance and the target adjacent grid point concentration value, the trapezoidal areas are summarized to obtain a trapezoidal area set; The trapezoidal area sum of the trapezoidal area set is calculated, and the trapezoidal area sum is taken as the doping concentration integral value.

7. The process flow database construction method based on super junction MOSFET fabrication according to claim 6, wherein, The calculation formula of the path point doping concentration value is as follows: wherein, represents a path point doping concentration value, represents a number of grid weights in a grid weight set, represents a thgrid weight in a grid weight set, represents a thadjacent grid point concentration value corresponding to the thgrid weight.

8. A system for constructing a process flow database using the super junction MOSFET-based fabrication process of claim 1, wherein, The system includes: A database design specification confirmation module is configured to receive a process flow database construction instruction, confirm super-junction MOSFET process flow knowledge according to the process flow database construction instruction, and obtain a database design specification according to the super-junction MOSFET process flow knowledge. A data acquisition interface establishment module is configured to establish an automatic data acquisition interface, wherein the automatic data acquisition interface includes a device layer interface, a measurement layer interface and a file layer interface. A database framework construction module is configured to construct a database framework according to the automatic data acquisition interface and the database design specification, and confirm a set of to-be-simulated process parameters. A database construction completion module is configured to obtain a set of associated data based on the set of to-be-simulated process parameters and the database framework, wherein the set of to-be-simulated process parameters includes a plurality of to-be-simulated process parameters, the set of associated data includes a plurality of associated data, the to-be-simulated process parameters and the associated data correspond to each other, an optimized forward prediction machine learning model and an optimized reverse design machine learning model are constructed according to the set of associated data, and the process flow database construction based on the super-junction MOSFET preparation is completed based on the optimized forward prediction machine learning model and the optimized reverse design machine learning model.

Images