WO2022267750A1

WO2022267750A1 - Modeling method and modeling apparatus, and electronic device and storage medium

Info

Publication number: WO2022267750A1
Application number: PCT/CN2022/093257
Authority: WO
Inventors: 薛小帝
Original assignee: 海光信息技术股份有限公司
Priority date: 2021-06-25
Filing date: 2022-05-17
Publication date: 2022-12-29
Also published as: CN113343630A; CN113343630B

Abstract

A modeling method and a modeling apparatus, and an electronic device and a storage medium. The modeling method comprises: acquiring a model base, wherein the model base comprises a plurality of initial models, and each initial model comprises a group of model parameters and a corresponding group of simulated physical parameters, which are generated through simulation; acquiring a plurality of groups of test data; and obtaining a target model on the basis of a statistical distribution calculation and according to the plurality of groups of test data, the plurality of initial models in the model base, and the model parameters and simulated physical parameters of the plurality of initial models. By means of the modeling method, the problems of a manual iteration process being complex and tedious and having relatively low efficiency can be solved, an automatic process for secondary development of a model can be realized, a large amount of test data can be processed, with the processing efficiency, the processing speed and the accuracy being high, and a customized secondary development of the model can be completed on the basis of specific range requirements and the test data.

Description

Modeling method and device, electronic device and storage medium

This application claims the priority of Chinese patent application No. 202110713138.6 submitted on June 25, 2021, and the content disclosed in the above Chinese patent application is quoted in full for all purposes as a part of this application.

technical field

Embodiments of the present disclosure relate to a modeling method, a modeling device, electronic equipment, and a storage medium.

Background technique

Integrated circuit performance analysis circuit simulation program (Simulation Program With Integrated Circuits Emphasis, SPICE) is a language simulator software for circuit description and simulation, which can be used to detect the integrity of the connection and function of the circuit, and to predict the performance of the circuit. Behavior. SPICE is currently the most widely used circuit-level simulation program in the device design industry, mainly used for simulation of analog circuits and mixed-signal circuits.

Contents of the invention

At least one embodiment of the present disclosure provides a modeling method, including: obtaining a model library, wherein the model library includes a plurality of initial models, each initial model includes a set of model parameters and a corresponding set of simulations generated through simulation Physical parameters; obtaining multiple sets of test data; according to the multiple sets of test data, multiple initial models in the model library, and model parameters and simulated physical parameters of the multiple initial models, based on statistical distribution calculations, the target model is obtained .

For example, in the modeling method provided by an embodiment of the present disclosure, a set of model parameters included in each initial model includes fitting parameters and candidate parameters, and according to the multiple sets of test data, multiple The initial model and the model parameters and simulation physical parameters of the multiple initial models are calculated based on statistical distribution to obtain the target model, including: based on each set of test data in the multiple sets of test data, from the multiple sets of test data in the model library Multiple fitting models are selected from the initial model, wherein each set of test data corresponds to multiple fitting models, and the simulation physical parameters included in the fitting model corresponding to each set of test data satisfy the first condition; for each set of test data , performing statistical distribution calculation on the fitting parameters in the model parameters of the fitting model corresponding to the set of test data, and obtaining the fitting parameters corresponding to the maximum probability value as an alternative fitting parameter; corresponding to the multiple sets of test data respectively According to the target physical parameters and the target fitting parameters, from multiple initial Select the candidate model in the model, and use the candidate parameters in the model parameters of the candidate model as the target candidate parameters, so as to obtain the target model; wherein, the simulation physical parameters of the candidate model and the model parameters The fitting parameters satisfy the second condition, and the model parameters of the target model include the target fitting parameters and the target candidate parameters.

For example, in the modeling method provided by an embodiment of the present disclosure, each set of test data includes a plurality of test physical parameters, and the first condition includes: each simulated physical parameter in the simulated physical parameters included in the fitting model Each corresponding test physical parameter in the test data corresponding to the fitting model is respectively equal, or each of the simulated physical parameters included in the fitting model is the same as the test corresponding to the fitting model The sum of the differences of each corresponding test physical parameter in the data is less than the first threshold.

For example, in the modeling method provided by an embodiment of the present disclosure, the second condition includes: the simulated physical parameters of the candidate model are equal to the target physical parameters, and the model parameters of the candidate model are The fitting parameters are equal to the target fitting parameters, or each of the fitting parameters in the simulation physical parameters and model parameters of the alternative model is equal to the target physical parameters and the target fitting parameters The sum of the differences of each corresponding parameter is smaller than the second threshold.

For example, in the modeling method provided by an embodiment of the present disclosure, the statistical distribution calculation includes normal distribution calculation.

For example, in the modeling method provided by an embodiment of the present disclosure, according to the multiple sets of test data, the multiple initial models in the model library, and the model parameters and simulation physical parameters of the multiple initial models, based on statistics The distribution calculation to obtain the target model further includes: optimizing the target fitting parameters to update the target fitting parameters.

For example, in the modeling method provided by an embodiment of the present disclosure, optimizing the target fitting parameters to update the target fitting parameters includes: performing simulation based on the target fitting parameters to obtain a comparative physical parameter; judging whether the similarity between the comparison physical parameter and the multiple sets of test data is greater than the similarity threshold; if the similarity is greater than the similarity threshold, supplement the test data, and recalculate according to the supplemented test data the target fitting parameters to update the target fitting parameters.

For example, in the modeling method provided by an embodiment of the present disclosure, obtaining the model library includes: defining the values of the model parameters of the original model, and obtaining multiple value combinations based on the values; based on the multiple The combination of values simulates the original model to obtain multiple sets of simulated physical parameters to obtain a model library including the multiple initial models; wherein the multiple value combinations are respectively used as models of the multiple initial models parameter.

For example, in the modeling method provided by an embodiment of the present disclosure, the original model is simulated based on the multiple value combinations to obtain the multiple sets of simulated physical parameters, so as to obtain the The model library includes: based on the multiple value combinations, using a script file to simulate the original model to obtain the multiple sets of simulated physical parameters, so as to obtain the model library including the multiple initial models.

For example, in the modeling method provided by an embodiment of the present disclosure, the original model includes a Berkeley Short Channel Insulated Gate Field Effect Transistor SPICE model.

For example, in the modeling method provided by an embodiment of the present disclosure, the set of model parameters includes at least mobility correction parameters, source-drain channel current correction parameters, and threshold voltage drift parameters, and the set of simulation physical parameters includes at least threshold voltage, effective drive current, and leakage current.

For example, in the modeling method provided by an embodiment of the present disclosure, two of the mobility correction parameter, the source-drain channel current correction parameter, and the threshold voltage drift parameter are used as the fitting parameters, and the The other one of the mobility correction parameter, the source-drain channel current correction parameter, and the threshold voltage drift parameter is used as the candidate parameter.

For example, in the modeling method provided by an embodiment of the present disclosure, the multiple sets of test data are obtained based on a wafer acceptability test or a wafer screening test.

For example, in the modeling method provided by an embodiment of the present disclosure, the modeling method is used for secondary development of the SPICE model based on the measured characteristics of the product.

At least one embodiment of the present disclosure also provides a modeling device, including: a first acquisition unit configured to acquire a model library, wherein the model library includes a plurality of initial models, and each initial model includes a set of model parameters and corresponding A set of simulated physical parameters generated through simulation; the second acquisition unit is configured to acquire multiple sets of test data; the calculation unit is configured to obtain multiple sets of test data, multiple initial models in the model library, and the The model parameters and simulation physical parameters of multiple initial models are calculated based on statistical distribution to obtain the target model.

For example, in the modeling device provided by an embodiment of the present disclosure, a set of model parameters included in each initial model includes fitting parameters and candidate parameters; the calculation unit includes a fitting model determination unit, a first statistical distribution calculation unit, a second statistical distribution calculation unit, and a target model determination unit; the fitting model determination unit is configured to select from multiple initial models in the model library to obtain multiple A fitting model, wherein each set of test data corresponds to a plurality of fitting models, and the simulation physical parameters included in the fitting model corresponding to each set of test data satisfy the first condition; the first statistical distribution calculation unit is configured to, for For each group of test data, perform statistical distribution calculation on the fitting parameters in the model parameters of the fitting model corresponding to the group of test data, and obtain the fitting parameters corresponding to the maximum probability value as an alternative fitting parameter; the second statistics The distribution calculation unit is configured to perform statistical distribution calculation on the alternative fitting parameters corresponding to the plurality of sets of test data, and obtain the alternative fitting parameters corresponding to the maximum probability value as the target fitting parameter; the target model determination unit configuration In order to select a candidate model from a plurality of initial models in the model library according to the target physical parameter and the target fitting parameter, and use the candidate parameters in the model parameters of the candidate model as target candidate parameters, thereby The target model is obtained; wherein, the simulation physical parameters of the alternative model and the fitting parameters in the model parameters meet the second condition, and the model parameters of the target model include the target fitting parameters and the target candidate parameters .

For example, in the modeling device provided by an embodiment of the present disclosure, each set of test data includes a plurality of test physical parameters, and the first condition includes: each simulated physical parameter in the simulated physical parameters included in the fitting model Each corresponding test physical parameter in the test data corresponding to the fitting model is respectively equal, or each of the simulated physical parameters included in the fitting model is the same as the test corresponding to the fitting model The sum of the differences of each corresponding test physical parameter in the data is less than the first threshold.

For example, in the modeling device provided in an embodiment of the present disclosure, the second condition includes: the simulated physical parameters of the candidate model are equal to the target physical parameters, and the model parameters of the candidate model are The fitting parameters are equal to the target fitting parameters, or each of the fitting parameters in the simulation physical parameters and model parameters of the alternative model is equal to the target physical parameters and the target fitting parameters The sum of the differences of each corresponding parameter is smaller than the second threshold.

For example, in the modeling device provided in an embodiment of the present disclosure, the calculation unit further includes an optimization unit; the optimization unit is configured to optimize the target fitting parameters, so as to update the target fitting parameters.

For example, in the modeling device provided in an embodiment of the present disclosure, the optimization unit includes a first subunit, a second subunit, and a third subunit; the first subunit is configured to be based on the target fitting parameter , performing simulation to obtain a comparison physical parameter; the second subunit is configured to determine whether the similarity between the comparison physical parameter and the multiple sets of test data is greater than a similarity threshold; the third subunit is configured to, if If the similarity is greater than the similarity threshold, the test data is supplemented, and the target fitting parameters are recalculated according to the supplemented test data, so as to update the target fitting parameters.

At least one embodiment of the present disclosure further provides an electronic device, including the modeling apparatus provided by any embodiment of the present disclosure.

At least one embodiment of the present disclosure also provides an electronic device, including: a processor; a memory including one or more computer program modules; wherein the one or more computer program modules are stored in the memory and configured To be executed by the processor, the one or more computer program modules include instructions for implementing the modeling method provided by any embodiment of the present disclosure.

At least one embodiment of the present disclosure further provides a storage medium for storing non-transitory computer-readable instructions, and when the non-transitory computer-readable instructions are executed by a computer, the modeling provided by any embodiment of the present disclosure can be realized method.

Description of drawings

In order to illustrate the technical solutions of the embodiments of the present disclosure more clearly, the accompanying drawings of the embodiments will be briefly introduced below. Obviously, the accompanying drawings in the following description only relate to some embodiments of the present disclosure, rather than limiting the present disclosure .

Figure 1 is a flowchart of secondary development of a SPICE model;

Fig. 2 is a schematic flowchart of a modeling method provided by some embodiments of the present disclosure;

Fig. 3 is a schematic flow chart of step S10 in Fig. 2;

FIG. 4 is a schematic flow chart of step S30 in FIG. 2;

Fig. 5A is one of the schematic diagrams of statistical distribution calculation in the modeling method provided by some embodiments of the present disclosure;

FIG. 5B is the second schematic diagram of statistical distribution calculation in the modeling method provided by some embodiments of the present disclosure;

Fig. 6 is a logical schematic diagram of a modeling method provided by some embodiments of the present disclosure;

FIG. 7 is a schematic flowchart of another modeling method provided by some embodiments of the present disclosure;

FIG. 8 is a schematic flow chart of step S35 in FIG. 7;

FIG. 9 is a schematic flowchart of a modeling method provided by some embodiments of the present disclosure;

Fig. 10 is a schematic diagram of comparing the SPICE model obtained by the modeling method provided by some embodiments of the present disclosure with the original SPICE model and test data;

Fig. 11 is a schematic block diagram of a modeling device provided by some embodiments of the present disclosure;

Fig. 12 is a schematic block diagram of an electronic device provided by some embodiments of the present disclosure;

Fig. 13 is a schematic block diagram of another electronic device provided by some embodiments of the present disclosure;

Fig. 14 is a schematic block diagram of another electronic device provided by some embodiments of the present disclosure; and

Fig. 15 is a schematic diagram of a storage medium provided by some embodiments of the present disclosure.

detailed description

In order to make the purpose, technical solutions and advantages of the embodiments of the present disclosure clearer, the technical solutions of the embodiments of the present disclosure will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present disclosure. Apparently, the described embodiments are some, not all, embodiments of the present disclosure. Based on the described embodiments of the present disclosure, all other embodiments obtained by persons of ordinary skill in the art without creative effort fall within the protection scope of the present disclosure.

Unless otherwise defined, the technical terms or scientific terms used in the present disclosure shall have the usual meanings understood by those skilled in the art to which the present disclosure belongs. "First", "second" and similar words used in the present disclosure do not indicate any order, quantity or importance, but are only used to distinguish different components. Likewise, words like "a", "an" or "the" do not denote a limitation of quantity, but mean that there is at least one. "Comprising" or "comprising" and similar words mean that the elements or items appearing before the word include the elements or items listed after the word and their equivalents, without excluding other elements or items. Words such as "connected" or "connected" are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "Down", "Left", "Right" and so on are only used to indicate the relative positional relationship. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

Usually, in the design stage, it is necessary to establish a SPICE model for devices such as transistors, and then realize the design through simulation. In the actual tape-out process, due to the iteration of the fab process, there may be deviations between the actual product characteristics and the SPICE model provided by it. Operations such as updating the SPICE model through the fab have problems such as long cycle and many difficulties, which affect the project progress and the new product launch cycle. Therefore, it is necessary to carry out the secondary development of the SPICE model based on the actual characteristics of the mass-produced products, so that the SPICE model obtained after the secondary development matches the actual product characteristics, which has great practical significance and effect on failure analysis, new product design, etc. .

Figure 1 is a flow chart of the secondary development of a SPICE model. As shown in Figure 1, the current secondary development of the SPICE model generally adopts the forward fitting method. That is, the parameters of the SPICE model are manually modified first, and then the physical parameters are generated by the simulator. Compare simulation-generated physical parameters with measured data. If the difference between the physical parameters generated by the simulation and the measured data is large and does not meet the requirements, the SPICE model parameters are manually modified again, and then the physical parameters are obtained through simulation, and the simulated physical parameters are compared with the measured data. At this time, the model parameters are usually modified based on the designer's experience and theoretical judgment. If the physical parameters generated by the simulation are not significantly different from the measured data and meet the requirements, the current SPICE model is a model that matches the actual product characteristics, thus completing the secondary development of the SPICE model. Using the above method, the model parameters are continuously revised based on experience and theoretical judgment, so as to iteratively generate better model parameters to complete the secondary development of the SPICE model.

However, when carrying out the secondary development of the SPICE model, it takes a lot of time to manually modify the model parameters and perform iterative forward fitting, and the requirements for designers are high, so the secondary development of the model is more difficult. Moreover, when fitting a large amount of measured data, there will be problems such as insufficient precision, which will affect the accuracy of the obtained SPICE model.

At least one embodiment of the present disclosure provides a modeling method, a modeling device, electronic equipment, and a storage medium. This modeling method can solve the complex and cumbersome and low-efficiency problems of the manual iterative process, can realize the automatic process for the secondary development of the model, can handle a large amount of test data, has high processing efficiency, fast processing speed, high accuracy, and The customized secondary development of the model can be completed based on specific scope requirements and test data.

Hereinafter, embodiments of the present disclosure will be described in detail with reference to the accompanying drawings. It should be noted that the same reference numerals will be used in different drawings to refer to the same elements already described.

At least one embodiment of the present disclosure provides a modeling method. The modeling method includes: obtaining a model library, the model library includes a plurality of initial models, each initial model includes a set of model parameters and a corresponding set of simulated physical parameters generated through simulation; obtaining multiple sets of test data; The data, multiple initial models in the model library, and the model parameters and simulation physical parameters of the multiple initial models are calculated based on the statistical distribution to obtain the target model.

Fig. 2 is a schematic flowchart of a modeling method provided by some embodiments of the present disclosure. As shown in Fig. 2, in some embodiments, the modeling method includes the following operations.

Step S10: Obtain a model library, the model library includes a plurality of initial models, each initial model includes a set of model parameters and a corresponding set of simulated physical parameters generated through simulation;

Step S20: Acquiring multiple sets of test data;

Step S30: According to multiple sets of test data, multiple initial models in the model library, and model parameters and simulated physical parameters of the multiple initial models, the target model is obtained based on statistical distribution calculation.

For example, in step S10, the established and stored model library can be obtained, or the model library can be directly established in this step. For example, taking the transistor as an example, in order to accurately describe the parameters of the transistor, the Berkeley Short-channel IGFET Model (BSIM) can be used to form a model library. The BSIM model is a commonly used SPICE model, which can more accurately simulate transistor performance and calculate various parameters.

For example, the model library includes multiple initial models, and the initial model is a BSIM model. Each initial model includes a set of model parameters and a corresponding set of simulated physical parameters generated through simulation. For example, in some examples, a set of model parameters may include at least three model parameters, which are respectively a mobility correction parameter, a source-drain channel current correction parameter, and a threshold voltage drift parameter. Based on these model parameters, a set of corresponding simulation physical parameters can be obtained by simulating the model. The set of simulated physical parameters may include at least three simulated physical parameters, which are threshold voltage, effective driving current and leakage current, respectively. For example, in some examples, A represents a mobility correction parameter, B represents a source-drain channel current correction parameter, C represents a threshold voltage drift parameter, x represents a threshold voltage, y represents an effective driving current, and z represents a leakage current. The values of A, B, and C determine the characteristics of the model, and the parameters x, y, and z that reflect the characteristics of the model can be obtained by simulating the model. For example, different values of A, B, and C correspond to different initial models. In the model library, the model parameters A, B, and C of different initial models will not be exactly the same, so as to distinguish different initial models. The simulation physical parameters x, y, and z of different initial models may be completely different, partly the same, or completely the same.

It should be noted that in the embodiments of the present disclosure, the modeling object of the SPICE model is not limited to transistors, but can be any device in integrated circuits, chips, etc. Although this disclosure uses transistors as an example for illustration, this does not constitute Limitations on Embodiments of the Disclosure. Correspondingly, the model parameters are not limited to mobility correction parameters, source-drain channel current correction parameters, and threshold voltage drift parameters, and the simulation physical parameters are not limited to threshold voltage, effective drive current, and leakage current. The model parameters and simulation physical parameters are specific The category may be determined according to actual requirements and the type of model used, which is not limited in the embodiments of the present disclosure. The number of a set of model parameters is not limited to 3, but can also be any number such as 2, 4, 5, etc. Similarly, the number of a set of simulation physical parameters is not limited to 3, and can also be 2, 4, 5, etc. Any number, such as 4, 5, etc., is not limited in the embodiments of the present disclosure. The initial model in the model library is not limited to the BSIM model, and may also be other types of models, which may be determined according to actual requirements, which is not limited in the embodiments of the present disclosure.

For example, in some examples, as shown in FIG. 3 , the above step S10 may further include the following operations.

Step S11: Define the values of the model parameters of the original model, and obtain multiple value combinations based on the values;

Step S12: Simulating the original model based on multiple value combinations to obtain multiple sets of simulated physical parameters to obtain a model library including multiple initial models.

For example, in step S11, the original model can adopt the BSIM model, that is, the Berkeley short-channel insulated gate field-effect transistor SPICE model, and the model parameters can be mobility correction parameters, source-drain channel current correction parameters, and threshold voltage drift parameters, That is, A, B, and C described above (the mobility correction parameter is represented by A, the source-drain channel current correction parameter is represented by B, and the threshold voltage drift parameter is represented by C). Define the respective values of A, B, and C to obtain multiple value combinations, that is, to obtain multiple sets of A, B, and C. For example, the values of A, B, and C determine the model characteristics, so multiple sets of A, B, and C correspond to multiple different models. For example, the values of different groups A, B, and C are not exactly the same, so that the multiple models are different from each other.

For example, within the value range of A, values can be uniformly selected with a certain step size, thereby obtaining multiple values of A. Similarly, B and C can also take values evenly with a certain step size within their respective value ranges, so as to obtain multiple B values and multiple C values. Multiple sets of A, B, and C can be obtained by permuting and combining multiple values of A, multiple values of B, and multiple values of C. For example, in some examples, the value range of A may be 0~1 or 0.7~1.2, the value range of B may be 0~1 or 0.7~1.2, the value range of C may be -1~+1, The step size of each value of A, B, and C can be 0.01 or 0.001. It should be noted that the above value range and step size are exemplary rather than restrictive, and the value range and step size can be set according to actual requirements.

For example, in some examples, assuming that N1 values of A are obtained, N2 values of B are obtained, and N3 values of C are obtained, the number of combinations of values of A, B, and C that can be obtained is: N1 *N2*N3, where N1, N2, and N3 are all positive integers.

For example, in step S12, the original model is simulated based on multiple value combinations, that is, the above multiple sets of A, B, and C are respectively written into the original model for simulation, thereby obtaining multiple sets of simulated physical parameters. For example, the simulated physical parameters include threshold voltage, effective driving current, and leakage current, namely x, y, and z described above (the threshold voltage is represented by x, the effective driving current is represented by y, and the leakage current is represented by z). Multiple sets of model parameters correspond one-to-one to multiple sets of simulation physical parameters, that is, multiple sets of A, B, and C correspond to multiple sets of x, y, and z. A set of A, B, and C corresponds to a model, and a corresponding set of x, y, and z is obtained through simulation.

For example, in some examples, the script file may be used to simulate the original model based on multiple value combinations, so as to obtain multiple sets of simulated physical parameters, so as to obtain a model library including multiple initial models. For example, the script file may be written in any applicable language, and may also be executed in any applicable order and manner, which is not limited by the embodiments of the present disclosure. By using the script file to simulate the original model, all value combinations can be efficiently and quickly traversed without manual operation to perform multiple simulations, thereby improving the simulation efficiency.

Multiple sets of models corresponding to A, B, and C are used as initial models, thereby obtaining a model library including multiple initial models. Multiple value combinations (that is, multiple sets of A, B, and C) are used as model parameters of multiple initial models, each initial model corresponds to a set of model parameters A, B, and C, and each initial model also corresponds to a A set of simulation physical parameters x, y, z generated by simulation.

For example, as shown in FIG. 2, in step S20, multiple sets of test data are acquired. For example, the test data may be obtained by testing an actual product, that is, the test data is actually measured data. In the case where a set of simulated physical parameters included in the initial model in the model library are threshold voltage, effective drive current, and leakage current, a set of test data is the threshold voltage, effective drive current, and leakage current obtained from testing the actual product. current. For example, the parameter category of the test data is the same as that of the simulated physical parameter, thereby facilitating subsequent analysis and calculation. It should be noted that when the simulated physical parameter is a parameter of another category, the type of the test data is also adjusted accordingly, as long as the parameter category of the test data is the same as that of the simulated physical parameter.

For example, multiple sets of test data can be obtained based on wafer acceptance test (Wafer Acceptance Test, WAT) or wafer screening test (Wafer Sort, WS). Of course, the embodiments of the present disclosure are not limited thereto, and other methods may also be used to obtain test data, which may be determined according to actual requirements.

For example, in step S30, according to multiple sets of test data, multiple initial models in the model library, model parameters and simulated physical parameters of the multiple initial models, the target model is obtained based on statistical distribution calculation. For example, the target model is a SPICE model obtained through secondary development based on the measured characteristics of the product, and the above steps S10 to S30 can realize the secondary development.

For example, in some examples, as shown in FIG. 4 , the above step S30 may further include the following operations.

Step S31: Based on each set of test data in multiple sets of test data, select multiple fitting models from multiple initial models in the model library, each set of test data corresponds to multiple fitting models, and the fitting model corresponding to each set of test data The simulated physical parameters included in the model satisfy the first condition;

Step S32: For each group of test data, perform statistical distribution calculation on the fitting parameters in the model parameters of the fitting model corresponding to the group of test data, and obtain the fitting parameters corresponding to the maximum probability value as an alternative fitting parameter;

Step S33: Perform statistical distribution calculation on the candidate fitting parameters respectively corresponding to multiple sets of test data, and obtain the candidate fitting parameters corresponding to the maximum probability value as the target fitting parameters;

Step S34: According to the target physical parameters and the target fitting parameters, select a candidate model from a plurality of initial models in the model library, and use the candidate parameters in the model parameters of the candidate model as the target candidate parameters, thereby obtaining the target model.

For example, in step S31, multiple fitting models are selected from multiple initial models in the model library based on each set of test data, and each set of test data corresponds to multiple fitting models. Here, the selected initial model is referred to as a fitted model. For example, the initial models in the model library can be traversed, and the selection can be made according to whether the simulation physical parameters included in the initial models satisfy the first condition. If the first condition is satisfied, the initial model is selected as the fitting model; if the first condition is not satisfied, the initial model is not selected. By traversing all the initial models in the model library, multiple fitting models corresponding to each set of test data can be selected and obtained.

For example, each set of test data includes a plurality of test physical parameters, and the plurality of test physical parameters are threshold voltage, effective driving current and leakage current obtained from the test respectively. For example, x' represents the threshold voltage obtained by the test, y' represents the effective driving current obtained by the test, and z' represents the leakage current obtained by the test.

For example, in some examples, the above-mentioned first condition may be: each simulated physical parameter included in the fitted model is equal to each corresponding tested physical parameter in the test data corresponding to the fitted model. That is, for a set of test data x', y', z', if x, y, z of an initial model are equal to x', y', z' respectively, ie x=x', y=y ', z=z', then select the initial model as the fitting model corresponding to the set of test data x', y', z'. Since the test data and the simulated physical parameters are completely equal, the fitting model obtained in this way has a high degree of matching, which is beneficial to improve the subsequent calculation precision and accuracy.

For example, in some other examples, the above-mentioned first condition may also be: the difference between each simulated physical parameter in the simulated physical parameters included in the fitted model and each corresponding tested physical parameter in the test data corresponding to the fitted model The sum of the values is less than the first threshold. That is, for a set of test data x', y', z', if the sum of the differences between x, y, z and x', y', z' of an initial model is less than the first threshold, ie ( x-x')+(y-y')+(z-z')<K1, then the initial model is selected as the fitting model corresponding to the set of test data x', y', z'. Here, K1 represents the first threshold, and the first threshold can be set according to actual needs, and can be set to any value. Since the test data is close to the simulated physical parameters, a large number of fitting models can be obtained in this way, which is conducive to increasing the number of samples for subsequent calculations, and it is easier to obtain results through statistical distribution calculations.

For example, a set of model parameters for each initial model (and fitted model) includes fitted parameters and candidate parameters. For example, when a set of model parameters includes mobility correction parameters, source-drain channel current correction parameters, and threshold voltage drift parameters, two of the mobility correction parameters, source-drain channel current correction parameters, and threshold voltage drift parameters are used as pseudo The combination parameter, the mobility correction parameter, the source-drain channel current correction parameter, and the threshold voltage drift parameter are used as a candidate parameter. For example, in some examples, the fitting parameters in a set of model parameters are A and B (ie, the mobility correction parameter and the source-drain channel current correction parameter), and the candidate parameter is C (ie, the threshold voltage drift parameter).

For example, in step S32, for each set of test data, statistical distribution calculation is performed on the fitting parameters in the model parameters of the fitting model corresponding to the set of test data, and the fitting parameter corresponding to the maximum probability value is obtained as an alternative fitting parameter. combined parameters. For example, this step executes the first statistical distribution calculation in the modeling method, and the fitting parameter corresponding to the maximum probability value may refer to the fitting parameter with the highest distribution proportion. For example, a set of test data corresponds to multiple fitting models, and the fitting parameters A and B among the model parameters of these fitting models are calculated for statistical distribution, so as to obtain the fitting parameters A and B corresponding to the maximum probability value, and the are called alternative fitting parameters. Here, the fitting parameters A and B are calculated as a two-dimensional array. For example, the normal distribution calculation can be performed on multiple sets of fitting parameters A and B, so as to obtain a set of fitting parameters A and B corresponding to the maximum probability value as an alternative fitting parameter. For example, by performing calculations on each set of test data, alternative fitting parameters corresponding to each set of test data can be obtained, thereby obtaining multiple sets of alternative fitting parameters.

Taking a set of test data x', y', z' as an example, first select the fitting model corresponding to the set of test data x', y', z' from the model library. For example, taking the first condition that each simulation physical parameter included in the fitting model is equal to each corresponding test physical parameter in the test data corresponding to the fitting model as an example, you can directly choose to satisfy the condition x The initial model of =x', y=y', z=z' is used as the fitting model, and the model parameters A, B, and C of the fitting model are obtained.

Of course, the way to select and obtain the fitted model is not limited to the one described above, and the following ways can also be used. As shown in Figure 5A, when x=x', select the initial model, and traverse all models in the model library, thereby obtaining n1 initial models to obtain n1 groups A, B, C; when y=y' , select the initial model, traverse all the models in the model library, and thus obtain n2 initial models to obtain n2 groups A, B, C; when z=z', select the initial model, and traverse all the models in the model library model, thus obtaining n3 initial models to obtain n3 groups A, B, and C. Then, select A, B, and C included in groups n1, n2, and n3, and the initial model corresponding to A, B, and C is the fitting model. For example, Figure 5A shows two fitted models, Fitted Model-1 and Fitted Model-2.

As shown in FIG. 5A , after the fitting model and the model parameters A, B, and C of the fitting model are obtained, statistical distribution calculation is performed on the fitting parameters A, B among the model parameters A, B, and C of the fitting model. For example, one set of fitting parameters is A2, B2, another set of fitting parameters is A3, B7, of course, there are other fitting parameters, which are not shown in Fig. 5A. After statistical distribution calculation, the fitting parameters corresponding to the maximum probability value can be obtained, which are called alternative fitting parameters. In the example shown in Figure 5A, the alternative fitting parameters are A2, B2.

For example, as shown in FIG. 4 , in step S33 , the statistical distribution calculation is performed on the candidate fitting parameters respectively corresponding to multiple sets of test data, so as to obtain the candidate fitting parameter corresponding to the maximum probability value as the target fitting parameter. For example, this step executes the second statistical distribution calculation in the modeling method, and the candidate fitting parameter corresponding to the maximum probability value may refer to the candidate fitting parameter with the highest distribution proportion. Since the corresponding candidate fitting parameters are calculated for each set of test data in step S32, multiple sets of candidate fitting parameters can be obtained for multiple sets of test data. Statistical distribution calculations are performed on multiple sets of candidate fitting parameters, so that the candidate fitting parameters corresponding to the maximum probability value can be obtained, which are called target fitting parameters. Here, the alternative fitting parameters A, B are calculated as a two-dimensional array. For example, the normal distribution calculation can be performed on multiple sets of alternative fitting parameters A and B, so as to obtain a set of alternative fitting parameters A' and B' corresponding to the maximum probability value as the target fitting parameters. Through calculation, a set of A' and B' can be finally obtained, which are the target fitting parameters.

For example, as shown in FIG. 5B , in some examples, after calculating corresponding candidate fitting parameters for N groups of test data (such as test data_1 to test data_N), these candidate fitting parameters Perform the statistical distribution calculation again. For example, N is a positive integer. For example, the alternative fitting parameters corresponding to test data_1 are A2, B2, the alternative fitting parameters corresponding to test data_2 are A3, B2, and the alternative fitting parameters corresponding to test data_N-1 are A2, B2, the alternative fitting parameters corresponding to the test data_N are A6 and B6, and the alternative fitting parameters corresponding to other test data are not shown in FIG. 5B. Through statistical distribution calculation, the candidate fitting parameters corresponding to the maximum probability value can be obtained, which are called target fitting parameters. Here, the alternative fitting parameters A, B are calculated as a two-dimensional array. For example, the normal distribution calculation may be performed on multiple sets of candidate fitting parameters, so as to obtain a set of candidate fitting parameters corresponding to the maximum probability value as the target fitting parameters.

For example, by performing steps S32 and S33, after two statistical distribution calculations, a set of target fitting parameters can be obtained, that is, a unique set of A' and B' can be obtained.

For example, as shown in Figure 4, in step S34, according to the target physical parameters and the target fitting parameters, a candidate model is selected from a plurality of initial models in the model library, and the candidate parameters in the model parameters of the candidate model are As the target candidate parameters, the target model is obtained. For example, the target physical parameters also include threshold voltage, effective driving current, and leakage current. The values of each parameter in the target physical parameters are preset values, which can be determined according to actual needs, such as the product requirements that need to be met or satisfied , which is not limited by the embodiments of the present disclosure. For example, x" represents the threshold voltage in the target physical parameter, y" represents the effective driving current in the target physical parameter, and z" represents the leakage current in the target physical parameter. For example, the target fitting parameter is a value obtained in the above step S33 Group A', B'.

Here, the selected initial model is referred to as a candidate model. For example, the initial model in the model library can be traversed, and the selection can be made according to whether the simulation physical parameters included in the initial model and the fitting parameters in the model parameters meet the second condition, that is, according to x, y, z, A, The selection is made based on whether the value of B satisfies the second condition. If the second condition is met, the initial model is selected as an alternative model; if the second condition is not met, the initial model is not selected.

For example, in some examples, the above second condition may be: the simulation physical parameters of the candidate model are equal to the target physical parameters, and the fitting parameters in the model parameters of the candidate model are equal to the target fitting parameters. That is, if x, y, z of an initial model are equal to x", y", and z" respectively, and A, B of the initial model are equal to A', B' respectively, that is, x=x", y=y", z=z", A=A', B=B', then select the initial model as the candidate model. Since the above-mentioned corresponding parameters are equal to each other, the matching degree of the candidate model obtained in this way is high, which is conducive to obtaining more accurate results.

For example, in some other examples, the above-mentioned second condition may also be: each of the simulation physical parameters and the fitting parameters of the model parameters of the alternative model corresponds to each of the target physical parameters and the target fitting parameters The sum of the differences of the parameters is smaller than the second threshold. That is, if the sum of the differences between x, y, z, A, B and x", y", z", A', B' of an initial model is less than the second threshold, that is, (x-x")+ (y-y")+(z-z")+(A-A')+(B-B')<K2, the initial model is selected as an alternative model. Here, K2 represents the second threshold, and the second threshold can be set according to actual needs, and can be set to any value. Since the above-mentioned corresponding parameters are relatively close to each other, it is easier to obtain an alternative model in this way, which is conducive to obtaining the final result.

For example, the candidate model can be obtained directly through the above method. Alternatively, in some examples, the target fitting parameters obtained in step S33 can also be optimized, and an alternative model can be obtained according to the optimized target fitting parameters. The optimization of the target fitting parameters will be described later, here I won't repeat them here.

For example, after the candidate model is obtained, a candidate parameter among the model parameters of the candidate model can be obtained, such as a threshold voltage drift parameter among the model parameters of the candidate model, and used as a target candidate parameter, denoted by C'. Thus, the aforementioned target fitting parameters A', B' and the target candidate parameter C' obtained here are taken as a set of model parameters, and the model corresponding to the set of model parameters A', B', and C' is the target model. The model parameters of the target model include target fitting parameters A', B' and target candidate parameters C'.

Through the above method, the target model with model parameters A', B', and C' can be obtained. The target model is a model obtained by secondary development of the SPICE model based on the measured characteristics of the product, thus completing the secondary SPICE model. develop. The target model matches the actual product characteristics, which is helpful for failure analysis and new product design.

The modeling method provided by the embodiments of the present disclosure is used for secondary development of the SPICE model based on the measured characteristics of the product, which can solve the problem of complex and cumbersome manual iteration process and low efficiency, and can realize the automated process for the secondary development of the model. The process can be calculated automatically without human intervention or based on the experience and judgment of the designer. The modeling method can handle a large amount of test data, and has high processing efficiency, fast processing speed and high accuracy. Moreover, by limiting the scope of test data, the customized secondary development of the model can be completed based on specific range requirements and test data, so that the obtained model can well reflect the product characteristics within the limited test data range.

This modeling method incorporates the idea of neural network. Different from the traditional manual iterative method, this modeling method uses the model library as the sample set of the neural network, adopts the method of simulation traversal and combined with the neural network to strengthen learning, forms an automatic process, and converts the SPICE model problem into a neural network problem, by This improves modeling efficiency and accuracy. As shown in Figure 6, by establishing a multi-layer structure such as the model parameter layer (that is, the aforementioned A, B, C), the simulation layer (that is, the simulator), and the physical layer (that is, the aforementioned x, y, z), Using multiple sets of test data x', y', z' for training, the target model can be obtained, and the secondary development of the SPICE model can be completed.

Fig. 7 is a schematic flowchart of another modeling method provided by some embodiments of the present disclosure. For example, in some embodiments, as shown in FIG. 7, the modeling method may further include step S35. Steps S31-S34 in the modeling method are basically the same as steps S31-S34 shown in FIG. 4, where No longer.

Step S35: Optimizing the target fitting parameters to update the target fitting parameters.

For example, in step S35, the target fitting parameters are optimized, and the updated target fitting parameters are used in the operation of selecting a candidate model in the subsequent step S34.

For example, as shown in FIG. 8 , in some examples, the above step S35 may further include the following operations.

Step S351: Based on the target fitting parameters, perform simulation to obtain comparison physical parameters;

Step S352: judging whether the similarity between the compared physical parameters and multiple sets of test data is greater than the similarity threshold;

Step S353: If the similarity is greater than the similarity threshold, supplement the test data, and recalculate the target fitting parameters according to the supplemented test data, so as to update the target fitting parameters.

For example, in step S351, according to the target fitting parameters (the aforementioned A', B'), the emulator is used to perform simulation to obtain the comparison physical parameters, and the comparison physical parameters include, for example, threshold voltage (indicated by x"'), effective Drive current (denoted by y"') and leakage current (denoted by z"'). In the simulation, C in the model parameters is uncertain, so multiple values of C can be used to match the target fitting parameter A ', B', so as to simulate the values of each C separately, thus obtaining multiple sets of physical parameters for comparison.

For example, in step S352, it is judged whether the similarity between the compared physical parameter and multiple sets of test data is greater than a similarity threshold. For example, there are multiple groups of physical parameters compared, corresponding to different C values. For different C values, judge whether the group of comparison physical parameters is the same or similar to the test data obtained under the same C value, so as to judge whether the comparison physical parameters are within the value range of C and the test data of multiple groups as a whole Whether the similarity on is greater than the similarity threshold. For example, the similarity threshold may be set according to actual requirements, for example, according to the similarity that needs to be achieved, which is not limited in the embodiments of the present disclosure.

For example, comparing the similarity between a physical parameter and multiple sets of test data can be defined in various ways. For example, in some examples, physical parameters can be fitted to obtain the first fitting curve, multiple sets of test data can be fitted to obtain the second fitting curve, and then the first fitting curve and the second fitting curve can be judged. The sum of the difference values or the variance of the difference values of the fitting curve within the preset range is used as the similarity. For example, in some other examples, the sum of differences between multiple sets of comparison physical parameters and multiple sets of test data may be calculated respectively, and then the sum of differences may be used as the similarity. In the above two similarity definition methods, the larger the value of the similarity, the greater the difference between the comparison physical parameters and the test data, and the smaller the similarity value, the smaller the difference between the comparison physical parameters and the test data .

It should be noted that the definition of the similarity between the comparison physical parameters and the test data can be determined according to the data type and actual needs, as long as it can reflect the similarity between the comparison physical parameters and the test data, the implementation of the present disclosure Examples are not limited to this. According to different definition methods, the numerical value of the similarity degree and the difference between the comparison physical parameters and the test data can be positively correlated or negatively correlated, which is determined by the definition method of the similarity degree.

For example, in step S353, if the similarity is greater than the similarity threshold, the test data is supplemented, and the target fitting parameters are recalculated according to the supplemented test data, so as to update the target fitting parameters. For example, when the similarity is greater than the similarity threshold, it means that the comparison physical parameters are quite different from the test data, so it is necessary to optimize the target fitting parameters A' and B'. For example, the test data can be supplemented, that is, the sample size of the test data is increased, and then the target fitting parameters are recalculated through two statistical distribution calculations in steps S32 and S33, thereby updating the target fitting parameters. For example, after the updated target fitting parameters are obtained, step S34 may be directly executed, that is, the candidate model is selected directly using the updated target fitting parameters. For example, steps S351 and S352 can also be executed again based on the updated target fitting parameters. If the similarity does not meet the requirements, then continue to supplement the test data, update the target fitting parameters again, and stop updating the target fitting parameters until the similarity meets the requirements. combined parameters.

Through the above method, the feedback method can be used to obtain more accurate target fitting parameters, thereby improving the accuracy of calculation and facilitating the subsequent acquisition of a more accurate target model.

Fig. 9 is a schematic flowchart of a modeling method provided by some embodiments of the present disclosure. In some examples, as shown in FIG. 9 , the specific execution process of the modeling method is as follows. First, define the parameters of the mobility correction parameters, source-drain channel current correction parameters, and threshold voltage drift parameters (such as the aforementioned A, B, and C) in the model parameters, and prepare the original model (such as the BSIM model). Then, the simulation traverses (that is, the simulation is performed sequentially for multiple value combinations of A, B, and C) to generate a model library. Next, obtain WAT or WS test data (such as the aforementioned x', y', z') as training sample values. Then, the first statistical distribution calculation is performed, that is, the aforementioned step S32 is executed to obtain multiple sets of candidate fitting parameters respectively corresponding to each set of test data. Next, perform the second statistical distribution calculation, that is, execute the aforementioned step S33 to obtain a set of target fitting parameters (such as the aforementioned A', B').

Then, write parameter verification is performed based on the target fitting parameters, that is, it is judged whether the similarity between the comparison physical parameters obtained according to the target fitting parameters and multiple sets of test data meets the requirements. If the requirements are not met, continue to supplement the WAT or WS test data, and then calculate again to obtain the updated target fitting parameters. If the requirements are met, based on the target fitting parameters and in conjunction with the target physical parameters, the candidate model is selected in the model library, thereby obtaining the candidate parameters in the model parameters of the candidate model as the target candidate parameters (such as the aforementioned C '). Finally, the target fitting parameters and target candidate parameters together form a new set of model parameters A', B', and C', and the model corresponding to this set of model parameters is the target model, that is, the SPICE model obtained from the secondary development , just export the SPICE model.

Fig. 10 is a schematic diagram of comparing the SPICE model obtained by the modeling method provided by some embodiments of the present disclosure with the original SPICE model and test data. As shown in FIG. 10 , the original SPICE model is obtained based on theoretical design, for example, and its performance deviates greatly from the actual test data of the product. The SPICE model obtained by using the modeling method provided by the embodiments of the present disclosure is a SPICE model obtained by secondary development. It can be seen that the model has a high degree of agreement with the test data and can more accurately reflect the characteristics of the actual product.

It should be noted that the modeling method provided by the embodiment of the present disclosure is not limited to the steps described above, and may further include more steps. The execution sequence of the various steps is not limited, although the above described various steps in a specific order, this does not constitute a limitation to the embodiments of the present disclosure.

At least one embodiment of the present disclosure further provides a modeling device. The modeling device can solve the complex and cumbersome and low-efficiency problems of the manual iteration process, can realize the automatic process for the secondary development of the model, can process a large amount of test data, has high processing efficiency, fast processing speed, high accuracy, and The customized secondary development of the model can be completed based on specific scope requirements and test data.

Fig. 11 is a schematic block diagram of a modeling device provided by some embodiments of the present disclosure. As shown in FIG. 11 , the modeling device 100 includes a first acquisition unit 110 , a second acquisition unit 120 , and a calculation unit 130 . The modeling device 100 can be used for secondary development of the SPICE model based on the measured characteristics of the product.

The first obtaining unit 110 is configured to obtain a model library. The model library includes multiple initial models, and each initial model includes a set of model parameters and a corresponding set of simulated physical parameters generated through simulation. For example, the first obtaining unit 110 may execute step S10 of the modeling method as shown in FIG. 2 . The second acquiring unit 120 is configured to acquire multiple sets of test data. For example, the second acquiring unit 120 may execute step S20 of the modeling method as shown in FIG. 2 . The calculation unit 130 is configured to obtain the target model based on statistical distribution calculation according to multiple sets of test data, multiple initial models in the model library, model parameters and simulated physical parameters of the multiple initial models. For example, the computing unit 130 may execute step S30 of the modeling method as shown in FIG. 2 .

For example, the first acquisition unit 110, the second acquisition unit 120, and the calculation unit 130 may be hardware, software, firmware, or any feasible combination thereof. For example, the first acquisition unit 110, the second acquisition unit 120, and the calculation unit 130 may be dedicated or general-purpose circuits, chips or devices, or may be a combination of processors and memories. Regarding specific implementation forms of the first acquiring unit 110 , the second acquiring unit 120 , and the calculating unit 130 , the embodiment of the present disclosure does not limit it.

It should be noted that, in the embodiment of the present disclosure, each unit of the modeling device 100 corresponds to each step of the aforementioned modeling method. For the specific functions of the modeling device 100, please refer to the relevant description of the modeling method above. I won't repeat them here. The components and structures of the modeling device 100 shown in FIG. 11 are exemplary rather than limiting, and the modeling device 100 may also include other components and structures as required.

For example, each initial model includes a set of model parameters including fitted parameters and candidate parameters.

For example, the calculation unit 130 includes a fitting model determination unit, a first statistical distribution calculation unit, a second statistical distribution calculation unit, and a target model determination unit.

The fitting model determination unit is configured to select and obtain a plurality of fitting models from a plurality of initial models in the model library based on each set of test data in the plurality of sets of test data. Each set of test data corresponds to multiple fitting models, and the simulated physical parameters included in the fitting model corresponding to each set of test data satisfy the first condition. For example, each set of test data includes multiple test physical parameters.

The first condition includes: each simulated physical parameter in the simulated physical parameters included in the fitted model is equal to each corresponding tested physical parameter in the test data corresponding to the fitted model, or, the simulated physical parameters included in the fitted model The sum of the differences between each simulation physical parameter among the parameters and each corresponding test physical parameter in the test data corresponding to the fitting model is less than the first threshold.

The first statistical distribution calculation unit is configured to, for each set of test data, perform statistical distribution calculation on the fitting parameters in the model parameters of the fitting model corresponding to the set of test data, and obtain the fitting parameter corresponding to the maximum probability value as a backup Choose fitting parameters.

The second statistical distribution calculation unit is configured to perform statistical distribution calculation on the candidate fitting parameters respectively corresponding to multiple sets of test data, and obtain the candidate fitting parameter corresponding to the maximum probability value as the target fitting parameter.

The target model determination unit is configured to select a candidate model from multiple initial models in the model library according to the target physical parameters and target fitting parameters, and use the candidate parameters in the model parameters of the candidate model as target candidate parameters, thereby obtaining target model. For example, the simulation physical parameters of the candidate model and fitting parameters in the model parameters satisfy the second condition, and the model parameters of the target model include target fitting parameters and target candidate parameters.

The second condition includes: the simulation physical parameters of the alternative model are equal to the target physical parameters, and the fitting parameters in the model parameters of the alternative model are equal to the target fitting parameters, or, the simulation physical parameters of the alternative model and the model parameters The sum of the differences between each of the fitting parameters and the target physical parameter and each corresponding parameter of the target fitting parameters is less than a second threshold.

For example, the above statistical distribution calculation includes normal distribution calculation.

For example, computing unit 130 also includes an optimization unit. The optimization unit is configured to optimize the target fitting parameters to update the target fitting parameters. For example, the optimization unit further includes a first subunit, a second subunit and a third subunit. The first subunit is configured to perform simulation based on the target fitting parameters to obtain the comparison physical parameters. The second subunit is configured to judge whether the similarity between the compared physical parameter and multiple sets of test data is greater than a similarity threshold. The third subunit is configured to supplement the test data in response to the similarity being greater than the similarity threshold, and recalculate the target fitting parameters according to the supplemented testing data, so as to update the target fitting parameters.

For example, the first acquisition unit 110 includes a definition unit and a simulation unit. The definition unit is configured to define values of model parameters of the original model, and obtain multiple value combinations based on the values. The simulation unit is configured to simulate the original model based on multiple value combinations to obtain multiple sets of simulated physical parameters to obtain a model library including multiple initial models. For example, multiple value combinations are respectively used as model parameters of multiple initial models. In some examples, the simulation unit is further configured to use a script file to simulate the original model based on multiple value combinations to obtain multiple sets of simulated physical parameters to obtain a model library including multiple initial models.

For example, the original model includes the Berkeley Short Channel Insulated Gate Field Effect Transistor SPICE model, also known as the BSIM model. A set of model parameters at least includes mobility correction parameters, source-drain channel current correction parameters and threshold voltage drift parameters, and a set of simulation physical parameters at least includes threshold voltage, effective driving current and leakage current. Two of the mobility correction parameters, source-drain channel current correction parameters, and threshold voltage drift parameters are used as fitting parameters, and the other one of the mobility correction parameters, source-drain channel current correction parameters, and threshold voltage drift parameters is used as a candidate parameter . For example, multiple sets of test data are obtained based on wafer acceptability testing or wafer screening testing.

At least one embodiment of the present disclosure further provides an electronic device. The electronic device can solve the complex and cumbersome and low-efficiency problems of the manual iteration process, can realize the automatic process for the secondary development of the model, can process a large amount of test data, has high processing efficiency, fast processing speed, high accuracy, and can Complete the customized secondary development of the model based on specific scope requirements and test data.

Fig. 12 is a schematic block diagram of an electronic device provided by some embodiments of the present disclosure. As shown in FIG. 12 , the electronic device 200 includes a modeling device 210 . For example, the modeling device 210 may be the modeling device 100 shown in FIG. 11 . For related descriptions about the electronic device 200 , reference may be made to the above description about the modeling apparatus 100 , which will not be repeated here.

At least one embodiment of the present disclosure also provides an electronic device, the electronic device includes a processor and a memory, one or more computer program modules are stored in the memory and configured to be executed by the processor, one or more computer programs The program modules are used to realize the modeling method provided by any embodiment of the present disclosure. The electronic device can solve the complex and cumbersome and low-efficiency problems of the manual iteration process, can realize the automatic process for the secondary development of the model, can process a large amount of test data, has high processing efficiency, fast processing speed, high accuracy, and can Complete the customized secondary development of the model based on specific scope requirements and test data.

Fig. 13 is a schematic block diagram of another electronic device provided by some embodiments of the present disclosure. As shown in FIG. 13 , the electronic device 300 includes a processor 310 and a memory 320 . Memory 320 is used to store non-transitory computer readable instructions (eg, one or more computer program modules). The processor 310 is configured to execute non-transitory computer-readable instructions, and when the non-transitory computer-readable instructions are executed by the processor 310, one or more steps in the modeling method described above may be performed. The memory 320 and the processor 310 may be interconnected by a bus system and/or other forms of connection mechanisms (not shown).

For example, the processor 310 may be a central processing unit (CPU), a graphics processing unit (GPU), a digital signal processor (DSP), or other forms of processing units having data processing capabilities and/or program execution capabilities, such as field programmable Gate array (FPGA), etc.; for example, the central processing unit (CPU) can be X86 or ARM architecture, etc. The processor 310 can be a general-purpose processor or a special-purpose processor, and can control other components in the electronic device 300 to perform desired functions.

For example, memory 320 may include any combination of one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and/or nonvolatile memory. The volatile memory may include random access memory (RAM) and/or cache memory (cache), etc., for example. Non-volatile memory may include, for example, read only memory (ROM), hard disks, erasable programmable read only memory (EPROM), compact disc read only memory (CD-ROM), USB memory, flash memory, and the like. One or more computer program modules can be stored on the computer-readable storage medium, and the processor 310 can run one or more computer program modules to realize various functions of the electronic device 300 . Various application programs, various data, and various data used and/or generated by the application programs can also be stored in the computer-readable storage medium.

It should be noted that, in the embodiments of the present disclosure, for the specific functions and technical effects of the electronic device 300 , reference may be made to the description about the modeling method above, which will not be repeated here.

Fig. 14 is a schematic block diagram of another electronic device provided by some embodiments of the present disclosure. As shown in FIG. 14 , the electronic device 400 is, for example, suitable for implementing the modeling method provided by the embodiment of the present disclosure. The electronic device 400 may be a terminal device or a server. It should be noted that the electronic device 400 shown in FIG. 14 is only an example, which does not impose any limitation on the functions and application scope of the embodiments of the present disclosure.

As shown in FIG. 14 , the electronic device 400 may include a processing device (such as a central processing unit, a graphics processing unit, etc.) Various appropriate actions and processes are executed by programs in the memory (RAM) 43 . In the RAM 43, various programs and data necessary for the operation of the electronic device 400 are also stored. The processing device 41, the ROM 42, and the RAM 43 are connected to each other by a bus 44. An input/output (I/O) interface 45 is also connected to the bus 44 .

Typically, the following devices can be connected to the I/O interface 45: input devices 46 including, for example, a touch screen, touchpad, keyboard, mouse, camera, microphone, accelerometer, gyroscope, etc.; including, for example, a liquid crystal display (LCD), speaker, vibration an output device 47 such as a computer; a storage device 48 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 49. The communication means 49 may allow the electronic device 400 to communicate with other electronic devices wirelessly or by wire to exchange data. Although FIG. 14 shows electronic device 400 having various means, it should be understood that it is not required to implement or have all of the means shown, and electronic device 400 may alternatively implement or have more or fewer means.

For example, according to an embodiment of the present disclosure, the modeling method shown in FIG. 2 can be implemented as a computer software program. For example, embodiments of the present disclosure include a computer program product including a computer program carried on a non-transitory computer readable medium, the computer program including program code for executing the above-mentioned modeling method. In such an embodiment, the computer program may be downloaded and installed from the network via the communication means 49, or installed from the storage means 48, or installed from the ROM 42. When the computer program is executed by the processing device 41, the functions defined in the modeling method provided by the embodiments of the present disclosure can be realized.

At least one embodiment of the present disclosure further provides a storage medium for storing non-transitory computer-readable instructions. When the non-transitory computer-readable instructions are executed by a computer, the modeling method provided by any embodiment of the present disclosure can be implemented. . Using this storage medium can solve the complex and cumbersome and low-efficiency problems of the manual iteration process, realize the automated process for the secondary development of the model, and be able to process a large amount of test data with high processing efficiency, fast processing speed, and high accuracy. And the customized secondary development of the model can be completed based on specific scope requirements and test data.

Fig. 15 is a schematic diagram of a storage medium provided by some embodiments of the present disclosure. As shown in FIG. 15 , the storage medium 500 is used to store non-transitory computer readable instructions 510 . For example, one or more steps in the modeling method described above may be performed when the non-transitory computer readable instructions 510 are executed by a computer.

For example, the storage medium 500 can be applied to the above-mentioned electronic devices. For example, the storage medium 500 may be the memory 320 in the electronic device 300 shown in FIG. 13 . For example, for related descriptions about the storage medium 500, reference may be made to the corresponding description of the memory 320 in the electronic device 300 shown in FIG. 13 , which will not be repeated here.

The following points need to be explained:

(1) Embodiments of the present disclosure The drawings only relate to the structures involved in the embodiments of the present disclosure, and other structures may refer to common designs.

(2) In the case of no conflict, the embodiments of the present disclosure and the features in the embodiments can be combined with each other to obtain new embodiments.

The above description is only a specific implementation manner of the present disclosure, but the protection scope of the present disclosure is not limited thereto, and the protection scope of the present disclosure should be based on the protection scope of the claims.

Claims

A modeling method comprising:

Obtaining a model library, wherein the model library includes a plurality of initial models, each initial model includes a set of model parameters and a corresponding set of simulated physical parameters generated through simulation;

Obtain multiple sets of test data;

According to the multiple sets of test data, the multiple initial models in the model library, and the model parameters and simulated physical parameters of the multiple initial models, the target model is obtained based on statistical distribution calculation.
The modeling method according to claim 1, wherein a set of model parameters included in each initial model includes fitting parameters and candidate parameters,

According to the multiple sets of test data, multiple initial models in the model library, and model parameters and simulated physical parameters of the multiple initial models, based on statistical distribution calculations, the target model is obtained, including:

Based on each set of test data in the multiple sets of test data, a plurality of fitting models are selected from a plurality of initial models in the model library, wherein each set of test data corresponds to a plurality of fitting models, and each set of test data corresponds to The simulated physical parameters included in the fitted model satisfy the first condition;

For each group of test data, the fitting parameters in the model parameters of the fitting model corresponding to the group of test data are calculated for statistical distribution, and the fitting parameters corresponding to the maximum probability value are obtained as alternative fitting parameters;

Perform statistical distribution calculation on the alternative fitting parameters corresponding to the plurality of sets of test data respectively, and obtain the alternative fitting parameters corresponding to the maximum probability value as the target fitting parameters;

According to the target physical parameters and the target fitting parameters, a candidate model is selected from a plurality of initial models in the model library, and the candidate parameters in the model parameters of the candidate model are used as target candidate parameters, thereby obtaining said target model;

Wherein, the simulation physical parameters of the candidate model and the fitting parameters of the model parameters satisfy the second condition, and the model parameters of the target model include the target fitting parameters and the target candidate parameters.
The modeling method according to claim 2, wherein each group of test data comprises a plurality of test physical parameters,

The first condition includes:

Each of the simulated physical parameters included in the fitting model is equal to each corresponding test physical parameter in the test data corresponding to the fitting model, or

A sum of differences between each simulated physical parameter included in the fitting model and each corresponding test physical parameter in the test data corresponding to the fitting model is less than a first threshold.
The modeling method according to claim 2 or 3, wherein the second condition comprises:

The simulated physical parameters of the alternative model are equal to the target physical parameters, and the fitting parameters in the model parameters of the alternative model are equal to the target fitting parameters, or

The sum of the difference between each of the simulation physical parameters and the fitting parameters in the model parameters of the candidate model and each corresponding parameter of the target physical parameter and the target fitting parameters is less than the second threshold.
The modeling method according to any one of claims 2-4, wherein, based on the multiple sets of test data, multiple initial models in the model library, and model parameters and simulation physical parameters of the multiple initial models , based on the statistical distribution calculation, the target model is obtained, which also includes:

Optimizing the target fitting parameters to update the target fitting parameters.
The modeling method according to claim 5, wherein optimizing the target fitting parameters to update the target fitting parameters comprises:

Based on the target fitting parameters, performing simulation to obtain comparison physical parameters;

Judging whether the similarity between the compared physical parameters and the multiple sets of test data is greater than a similarity threshold;

If the similarity is greater than the similarity threshold, supplement test data, and recalculate the target fitting parameters according to the supplemented test data, so as to update the target fitting parameters.
The modeling method according to any one of claims 1-6, wherein obtaining the model library comprises:

Define the values of the model parameters of the original model, and obtain multiple value combinations based on the values;

Simulating the original model based on the multiple value combinations to obtain multiple sets of simulated physical parameters to obtain a model library including the multiple initial models;

Wherein, the multiple value combinations are respectively used as model parameters of the multiple initial models.
The modeling method according to claim 7, wherein the original model is simulated based on the multiple value combinations to obtain the multiple sets of simulated physical parameters, so as to obtain a model library including the multiple initial models ,include:

Based on the multiple value combinations, the script file is used to simulate the original model to obtain the multiple sets of simulated physical parameters, so as to obtain a model library including the multiple initial models.
The modeling method according to any one of claims 1-8, wherein the set of model parameters at least includes mobility correction parameters, source-drain channel current correction parameters, and threshold voltage drift parameters, and the set of simulation physical parameters Including at least threshold voltage, effective drive current and leakage current.
The modeling method according to claim 9, wherein two of the mobility correction parameter, the source-drain channel current correction parameter, and the threshold voltage drift parameter are used as the fitting parameters,

The other one of the mobility correction parameter, the source-drain channel current correction parameter, and the threshold voltage drift parameter is used as the candidate parameter.
The modeling method according to any one of claims 1-10, wherein the multiple sets of test data are obtained based on wafer acceptability test or wafer screening test.
The modeling method according to any one of claims 1-11, wherein the modeling method is used for secondary development of the SPICE model based on the measured characteristics of the product.
A modeling device comprising:

The first acquisition unit is configured to acquire a model library, wherein the model library includes a plurality of initial models, and each initial model includes a set of model parameters and a corresponding set of simulated physical parameters generated through simulation;

The second acquisition unit is configured to acquire multiple sets of test data;

The calculation unit is configured to obtain a target model based on statistical distribution calculation according to the multiple sets of test data, the multiple initial models in the model library, and the model parameters and simulated physical parameters of the multiple initial models.
The modeling apparatus according to claim 13, wherein the set of model parameters included in each initial model includes fitting parameters and candidate parameters;

The calculation unit includes a fitting model determination unit, a first statistical distribution calculation unit, a second statistical distribution calculation unit and a target model determination unit;

The fitting model determining unit is configured to select a plurality of fitting models from a plurality of initial models in the model library based on each set of test data in the plurality of sets of test data, wherein each set of test data corresponds to a plurality of Fitting the model, the simulation physical parameters included in the fitting model corresponding to each set of test data satisfy the first condition;

The first statistical distribution calculation unit is configured to, for each set of test data, perform statistical distribution calculation on the fitting parameters in the model parameters of the fitting model corresponding to the set of test data, and obtain the fitting parameters corresponding to the maximum probability value as follows: as an alternative fitting parameter;

The second statistical distribution calculation unit is configured to perform statistical distribution calculation on the candidate fitting parameters respectively corresponding to the plurality of sets of test data, and obtain the candidate fitting parameter corresponding to the maximum probability value as the target fitting parameter;

The target model determining unit is configured to select a candidate model from a plurality of initial models in the model library according to the target physical parameter and the target fitting parameter, and select the candidate model among the model parameters of the candidate model Parameters are used as target candidate parameters, thereby obtaining the target model;

Wherein, the simulation physical parameters of the candidate model and the fitting parameters of the model parameters satisfy the second condition, and the model parameters of the target model include the target fitting parameters and the target candidate parameters.
The modeling device according to claim 14, wherein each set of test data includes a plurality of test physical parameters,

The first condition includes:

Each of the simulated physical parameters included in the fitting model is equal to each corresponding test physical parameter in the test data corresponding to the fitting model, or

A sum of differences between each simulated physical parameter included in the fitting model and each corresponding test physical parameter in the test data corresponding to the fitting model is less than a first threshold.
The modeling device according to claim 14 or 15, wherein the second condition comprises:

The simulated physical parameters of the alternative model are equal to the target physical parameters, and the fitting parameters in the model parameters of the alternative model are equal to the target fitting parameters, or

The sum of the difference between each of the simulation physical parameters and the fitting parameters in the model parameters of the candidate model and each corresponding parameter of the target physical parameter and the target fitting parameters is less than the second threshold.
The modeling device according to any one of claims 13-16, wherein the calculation unit further includes an optimization unit;

The optimization unit is configured to optimize the target fitting parameters to update the target fitting parameters.
The modeling device according to claim 17, wherein said optimization unit comprises a first subunit, a second subunit and a third subunit;

The first subunit is configured to perform simulation based on the target fitting parameter to obtain a comparison physical parameter;

The second subunit is configured to determine whether the similarity between the compared physical parameter and the multiple sets of test data is greater than a similarity threshold;

The third subunit is configured to supplement test data if the similarity is greater than the similarity threshold, and recalculate the target fitting parameters according to the supplemented test data, so as to update the target fitting parameters.
An electronic device, comprising the modeling device according to any one of claims 13-18.
An electronic device comprising:

processor;

memory, including one or more computer program modules;

Wherein, the one or more computer program modules are stored in the memory and are configured to be executed by the processor, and the one or more computer program modules include components for implementing any of claims 1-12. Instructions for the modeling method described above.
A storage medium for storing non-transitory computer-readable instructions, when the non-transitory computer-readable instructions are executed by a computer, the modeling method described in any one of claims 1-12 can be realized.