Data processing model training and FCBGA packaging structure material matching method with cover

Abstract

The invention discloses a data processing model training and FCBGA packaging structure material matching method with a cover. According to the specific implementation scheme, a three-dimensional finite element model of a target packaging structure is constructed based on geometric parameters and material parameters of the target packaging structure; with the material parameters as design variables, inputting M groups of material parameter combinations generated through target test design based on the material parameters into the three-dimensional finite element model to obtain a data set; and extracting part of data from the data set based on the simulation test task as a training data set, and training the to-be-trained model based on the training data set to obtain a data processing model. According to the technical scheme, mechanical property analysis can be carried out on different materials according to the simulation result, so that the efficiency of evaluating the reliability of the packaging structure is improved, and the trial and error cost is reduced.

Description

Technical Field

[0001] This application relates to the field of semiconductor manufacturing technology, specifically to a data processing model training and material matching method for a covered FCBGA package structure. Background Technology

[0002] With the rapid updates and iterations of electronic devices, chip sizes are increasing, and packaging structure designs are becoming increasingly complex. Higher demands are emerging in chip design, packaging processes, and material properties. Simultaneously, with the continuous increase in verification costs, finite element analysis (FEA) has gradually become an important means of evaluating the reliability of packaging structures. However, traditional trial-and-error methods cannot directly analyze the mechanical properties of different materials based on simulation results, thus failing to efficiently evaluate the reliability of packaging structures. Therefore, a data processing scheme is urgently needed to achieve efficient evaluation of packaging structure reliability. Summary of the Invention

[0003] This paper presents a data processing model training and material matching method for FCBGA package structure with cap, aiming to solve the problem that traditional trial and error methods cannot directly analyze the mechanical properties of different materials based on simulation results, thus making it impossible to efficiently evaluate the reliability of the package structure.

[0004] Firstly, a data processing model training method is provided, including:

[0005] Based on the geometric and material parameters of the target packaging structure, a three-dimensional finite element model of the target packaging structure is constructed.

[0006] Using material parameters as design variables, M sets of material parameter combinations generated through target experimental design based on the material parameters are input into the three-dimensional finite element model to obtain a dataset; wherein, the dataset includes M sets of material parameter combinations and corresponding M sets of target simulation results; M is a positive integer;

[0007] Based on the simulation test task, a portion of the dataset is extracted as a training dataset, and the model to be trained is trained based on the training dataset to obtain the data processing model; wherein, the data processing model includes a data processing sub-model corresponding to the simulation test task.

[0008] Secondly, a method for matching materials in a capped FCBGA package structure is also provided, including:

[0009] Obtain the target material parameters for the capped FCBGA package structure;

[0010] The target material parameters are used as design variables, and N sets of target material parameter combinations are generated based on the target material parameters through target experimental design; where N is a positive integer;

[0011] Based on the simulation test task, select the data processing sub-model to be called from the data processing model; the data processing sub-model is obtained based on the data processing model training method in the first aspect.

[0012] Input N sets of target material parameter combinations into the data processing sub-model to obtain the optimal material parameter combination of the capped FCBGA package structure output by the data processing sub-model.

[0013] Configure the capped FCBGA package structure according to the optimal combination of material parameters.

[0014] Thirdly, this application provides a data processing model training apparatus, comprising:

[0015] The building module is used to construct a three-dimensional finite element model of the target packaging structure based on its geometric and material parameters.

[0016] The first generation module is used to input M sets of material parameter combinations generated by target experimental design based on material parameters into the three-dimensional finite element model, using material parameters as design variables, to obtain a dataset; wherein, the dataset includes M sets of material parameter combinations and corresponding M sets of target simulation results; M is a positive integer;

[0017] The training module is used to extract a portion of the dataset as a training dataset based on the simulation test task, and to train the model to be trained based on the training dataset to obtain the data processing model; wherein, the data processing model includes a data processing sub-model corresponding to the simulation test task.

[0018] Fourthly, this application provides a material matching device for a capped FCBGA package structure, comprising:

[0019] The first acquisition module is used to acquire the target material parameters of the covered FCBGA package structure;

[0020] The second generation module is used to take the target material parameters as design variables and generate N sets of target material parameter combinations based on the target material parameters through target experimental design; where N is a positive integer;

[0021] The selection module is used to select the data processing sub-model to be called from the data processing model according to the simulation test task; the data processing sub-model is obtained based on the data processing model training method in the first aspect.

[0022] The second acquisition module is used to input N sets of target material parameter combinations into the data processing sub-model to obtain the optimal material parameter combination of the capped FCBGA package structure output by the data processing sub-model.

[0023] The configuration module is used to configure the covered FCBGA package structure according to the optimal combination of material parameters.

[0024] Fifthly, this application also provides an electronic device, which includes a processor and a memory, wherein the memory stores a computer program, and when the processor invokes the computer program in the memory, it executes the data processing model training method provided in the first aspect and / or the material matching method for the covered FCBGA package structure provided in the second aspect.

[0025] Sixthly, this application also provides a computer-readable storage medium having a computer program stored thereon, the computer program being loaded by a processor to execute the data processing model training method provided in the first aspect and / or the material matching method for the covered FCBGA package structure provided in the second aspect.

[0026] According to the technical solution disclosed herein, by constructing a dedicated training dataset to train the data processing model for differentiated simulation test task scenarios, it is possible to achieve deep coupling between the simulation test task and the data processing model capabilities; furthermore, it is possible to perform mechanical property analysis on different materials based on simulation results, thereby improving the efficiency of evaluating the reliability of the packaging structure and reducing trial and error costs. Attached Figure Description

[0027] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0028] Figure 1 This is a schematic diagram of the data processing model training process provided in some embodiments of this application;

[0029] Figure 2 This is a schematic diagram of some material parameters of the dataset provided in some embodiments of this application;

[0030] Figure 3 This is a schematic diagram of the finite element model of a covered FCBGA package provided in some embodiments of this application;

[0031] Figure 4 This is a schematic diagram of the material matching process for a covered FCBGA package structure provided in some embodiments of this application;

[0032] Figure 5 This is a schematic diagram of the warp value variation provided in some embodiments of this application;

[0033] Figure 6This is a schematic diagram of one embodiment of the data processing model training apparatus provided in some embodiments of this application;

[0034] Figure 7 This is a schematic diagram of an embodiment of the material matching device for a covered FCBGA package structure provided in this application.

[0035] Figure 8 This is a schematic diagram of an embodiment of the electronic device provided in this application. Detailed Implementation

[0036] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0037] In the description of the embodiments of this application, it should be understood that the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Therefore, features defined with "first" and "second" may explicitly or implicitly include one or more of the stated features. In the description of the embodiments of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0038] The use of "applies to" or "configured to" in this application implies open and inclusive language, which does not exclude the applicability to or configuration to devices performing additional tasks or steps. Additionally, the use of "based on" implies openness and inclusivity, because processes, steps, calculations, or other actions "based on" one or more of the stated conditions or values ​​may in practice be based on additional conditions or values ​​beyond those stated.

[0039] In existing technologies, with the rapid updates and iterations of electronic devices, chip sizes are increasing, and packaging structure designs are becoming increasingly complex. Higher demands are also emerging in chip design, packaging processes, and material properties. Simultaneously, with rising verification costs, finite element simulation has gradually become an important tool for evaluating the reliability of packaging structures. However, due to the increasing complexity of packaging structure designs, the number of components and materials within the packaging structure is growing, and different materials are coupled together. This makes it impossible for traditional trial-and-error methods to directly analyze the mechanical properties of different materials based on simulation test results, thus hindering efficient evaluation of packaging structure reliability. Therefore, a data processing solution is urgently needed to achieve efficient evaluation of packaging structure reliability.

[0040] To at least partially address one or more of the aforementioned problems and other potential issues, this disclosure proposes a data processing model training scheme. The scheme includes: constructing a three-dimensional finite element model of the target packaging structure based on its geometric and material parameters; inputting M sets of material parameter combinations generated through target experimental design based on the material parameters into the three-dimensional finite element model, using the material parameters as design variables, to obtain a dataset; extracting a portion of the dataset as a training dataset based on simulation test tasks, and training the model to be trained based on the training dataset to obtain a data processing model; wherein the data processing model includes a data processing sub-model corresponding to the simulation test task. Thus, by constructing a dedicated training dataset to train the data processing model for differentiated simulation test task scenarios, deep coupling between the simulation test task and the data processing model's capabilities can be achieved; furthermore, mechanical property analysis of different materials can be performed based on simulation results, thereby improving the efficiency of evaluating the reliability of the packaging structure and reducing trial-and-error costs.

[0041] Figure 1 This is a schematic diagram of the data processing model training process according to an embodiment of the present disclosure, such as... Figure 1 As shown, the method includes at least the following steps:

[0042] S101: Construct a three-dimensional finite element model of the target packaging structure based on its geometric and material parameters;

[0043] S102: Using material parameters as design variables, input the M sets of material parameter combinations generated by the target experimental design based on the material parameters into the three-dimensional finite element model to obtain a dataset; wherein, the dataset includes the M sets of material parameter combinations and the corresponding M sets of target simulation results; M is a positive integer;

[0044] S103: Extract a portion of the data from the dataset based on the simulation test task as a training dataset, and train the model to be trained based on the training dataset to obtain a data processing model; wherein, the data processing model includes a data processing sub-model corresponding to the simulation test task.

[0045] In some embodiments, the target package structure refers to a specific semiconductor package form to be simulated using finite element methods. For example, the target package structure may be a lidd flip-chip ball grid array (FCBGA) structure; alternatively, it may be a 2.5D integrated circuit package structure. It should be noted that the above are merely illustrative examples and are not intended to limit all possible structures of the target package structure; an exhaustive list is not provided here.

[0046] In some embodiments, the core inputs for constructing a three-dimensional finite element model of the target packaging structure based on the target packaging structure include geometric parameters and material parameters. Specifically, the geometric parameters are one of the core inputs for constructing the three-dimensional finite element model of the target packaging structure, ensuring that the three-dimensional finite element model is consistent with the actual packaging in spatial morphology, providing an accurate geometric basis for simulation. Exemplarily, these geometric parameters may include: core functional unit parameters (such as length, width, thickness, and edge chamfer radius), interconnect structure parameters (such as the original diameter of solder balls and the height after welding), and packaging structure parameters (such as length, width, height, and edge fillet radius). The material parameters are also one of the core inputs for constructing the three-dimensional finite element model of the target packaging structure, ensuring that the three-dimensional finite element model matches the actual packaging in physical properties, providing an accurate material performance basis for simulation. Exemplarily, these material parameters may include: substrate material parameters, interconnect structure material parameters (such as solder balls and bonding wires), packaging material parameters (such as packaging colloid), and cover plate material parameters. It should be noted that the above are merely illustrative examples and do not constitute a limitation on all possible contents of the geometric and material parameters; they are simply not exhaustive.

[0047] In some embodiments, the target experimental design is used to generate M sets of material parameter combinations using the material parameters of the target packaging structure as design variables. For example, the target experimental design can employ an orthogonal experimental design, using the material parameters of the target packaging structure as design variables, and obtaining M sets of material parameter combinations by constructing an orthogonal table. In this way, each level (e.g., value range) of each design variable (e.g., material parameter) encounters each other's level (e.g., value range) the same number of times, and the level combinations of any two design variables are evenly distributed in the selected experimental scheme (material parameter combinations), ensuring both comprehensive coverage of the design variable level space and ensuring that the influence of each design variable on the experimental results can be analyzed independently, without any bias of duplication or omission. As another example, the target experimental design can employ a uniform experimental design, using the material parameters of the target packaging structure as design variables, and obtaining M sets of material parameter combinations by constructing a uniform design table. In this way, generating combinations through a uniform design table ensures that the projections of any two variables are uniformly distributed on the plane. As another example, the target experimental design can employ a screening experimental design, using the material parameters of the target packaging structure as design variables, and obtaining M combinations of material parameters through a screening method based on partial factorial design. Thus, if there are many candidate material parameters, but only a few key material parameters that significantly affect performance (such as warpage) need to be identified, without the need for a comprehensive analysis of interactions, a screening experimental design can be used to eliminate ineffective material parameters with a minimal sample size.

[0048] In some embodiments, the target simulation results are key indicators generated by simulating the target packaging structure using a three-dimensional finite element model to evaluate the reliability of the target packaging structure. Specifically, the target simulation results may include: stress simulation results, strain simulation results, warpage simulation results, and fracture simulation results.

[0049] In some embodiments, the dataset includes stress simulation results, strain simulation results, warpage simulation results, and fracture simulation results corresponding to M sets of material parameter combinations. Figure 2 A schematic diagram of some material parameters of the dataset is shown.

[0050] For example, taking a covered FCBGA structure as an example, the steps for training the data processing model may include:

[0051] Step 1: Construct a three-dimensional finite element model based on the geometric and material parameters of the covered FCBGA structure. The model specifically includes five parts: Lid, adhesive, thermal interface material, bottom filler, and substrate. Then, perform finite element simulation on the above three-dimensional finite element model after performing component assembly, material setting, material assignment, analysis step setting, load setting, boundary condition setting, mesh generation, mesh setting, and simulation job setting. Figure 3 A schematic diagram of the finite element model of the covered FCBGA package is shown.

[0052] Step two involves using the material parameters of five materials—Lid, adhesive, thermal interface material, underfill adhesive, and substrate—as design variables. For each material parameter, a suitable parameter range is selected, and within each range, three or four values ​​are taken at equal intervals as pre-selected values. Multiple combinations of material parameters are generated through target experimental design on these pre-selected values. This ensures that the selected combinations of material parameters cover the entire feature space. Simultaneously, multiple pre-selected values ​​are also taken at equal intervals within the parameter range, and their endpoints are removed before linear combination to obtain multiple combinations of material parameters.

[0053] Step 3: Input the selected multiple sets of material parameter combinations into the 3D finite element model for finite element simulation to obtain the target simulation results corresponding to each set of material parameter combinations. Here, the target simulation results can include: stress simulation results, strain simulation results, warpage simulation results, and fracture simulation results. Obtain the target simulation results corresponding to M sets of material parameter combinations to generate a dataset. Based on the simulation test task, extract a portion of the data from the dataset as a training dataset, and train the model to be trained based on the training dataset to obtain a data processing model. The data processing model includes a data processing sub-model corresponding to the simulation test task. Here, the model to be trained is a deep learning-based neural network model. Through multiple rounds of iterative training, the model gradually learns the nonlinear mapping relationship between the material parameter combinations and the simulation results. The neural network model adopts a fully connected neural network structure, containing three hidden layers, which, together with the input layer and the output layer, form a five-layer fully connected architecture. Its hyperparameters are selected through mesh optimization. The input layer has ten variables, each hidden layer contains 50 neurons, and the output layer contains the optimal material parameter combination of the target variable and the optimal simulation result (such as the warpage value) corresponding to the optimal material parameter combination.

[0054] In some embodiments, after training the model to be trained based on the training dataset to obtain the data processing model, a portion of the data can be extracted from the dataset as a test dataset. The neural network is then tested using the test dataset to evaluate its generalization ability and computational accuracy, ensuring the model's stability on unseen data. Test results show that the machine learning model's accuracy remains stable under unknown parameter combinations. Its coefficient of determination is 0.9998, and its mean squared error is 0.005. Finally, the reliability (warpage) of the target packaging structure was selected for verification. The Broyden-Fletcher-Goldfarb-Shanno Algorithm (BFGS) was used to solve for the optimal warpage value and the corresponding material parameter combination. The obtained optimal warpage value and the corresponding material parameter combination were input into the finite element simulation model for simulation verification. The results show that the optimal material parameter combination obtained by the machine learning model does indeed significantly reduce the warpage of the target packaging structure.

[0055] It should be noted that the modeling methods for the aforementioned machine learning models are not limited to artificial neural networks; they can also be implemented using methods such as multinomial regression, support vector machines, and decision trees. Furthermore, the optimization solution can be achieved using methods such as genetic algorithms, simulated annealing, and traversal.

[0056] The solution of this disclosure involves constructing a three-dimensional finite element model of the target packaging structure based on its geometric and material parameters. Using material parameters as design variables, M sets of material parameter combinations generated through target experimental design are input into the three-dimensional finite element model to obtain M sets of target simulation results. The target simulation results corresponding to each of the M sets of material parameter combinations are then obtained to generate a dataset. Based on the simulation test task, a portion of the data is extracted from the dataset as a training dataset, and the model to be trained is trained using this training dataset to obtain a data processing model. Thus, by constructing a dedicated training dataset to train the data processing model for differentiated simulation test task scenarios, deep coupling between the simulation test task and the data processing model's capabilities can be achieved. Furthermore, the mechanical properties of different materials can be analyzed based on the simulation results, thereby improving the efficiency of evaluating the reliability of the packaging structure and reducing trial-and-error costs.

[0057] In this embodiment of the disclosure, the simulation test task includes stress simulation test, and the data processing model includes a first data processing sub-model. Based on the simulation test task, a portion of the data is extracted from the dataset as a training dataset, and the model to be trained is trained based on the training dataset to obtain the data processing model. The process includes: based on the stress simulation test, extracting stress simulation results corresponding to M sets of material parameter combinations from the dataset as a first training data subset, and training the model to be trained based on the first training data subset to obtain the first data processing sub-model.

[0058] In some embodiments, this stress simulation test is a test task used to evaluate the reliability (stress value) of the target package structure. For example, in a flip-chip ball grid array package with a cap, if the stress at critical locations (such as solder balls and lid bonding points) exceeds the stress the material can withstand, it may cause cracking or debonding, leading to electrical connection failure or mechanical damage to the package. In stress simulation testing, the smaller the stress value and the more uniform the distribution, the higher the reliability of the target package structure.

[0059] In some embodiments, the first data processing sub-model is used to predict the stress values ​​corresponding to different combinations of material parameters of the target packaging structure, and output the optimal combination of material parameters corresponding to the optimal stress value, thereby guiding the material selection and optimization design of the packaging structure.

[0060] In this way, by using multiple combinations of material parameters and the corresponding stress simulation results to train the model to be trained, the stress value corresponding to each combination of material parameters can be quickly predicted and the material parameter combination corresponding to the optimal stress value can be directly output, which helps to improve the efficiency and accuracy of optimizing the material parameters of the target packaging structure.

[0061] In this embodiment of the disclosure, the simulation test task includes strain simulation test, and the data processing model includes a second data processing sub-model. Based on the simulation test task, a portion of the data is extracted from the dataset as a training dataset, and the model to be trained is trained based on the training dataset to obtain the data processing model. This includes: based on the strain simulation test, extracting the strain simulation results corresponding to M sets of material parameter combinations from the dataset as a second training data subset, and training the model to be trained based on the second training data subset to obtain the second data processing sub-model.

[0062] In some embodiments, this strain simulation test is a test task used to evaluate the reliability (strain value) of the target package structure. For example, excessive strain between the chip and the substrate can compromise the integrity of the underfill adhesive; if the substrate strain exceeds its tolerance limit, it can lead to wiring breakage. In strain simulation testing, the higher the strain value is within the material's allowable deformation range, the higher the reliability of the target package structure.

[0063] In some embodiments, the second data processing sub-model is used to predict the strain values ​​corresponding to different combinations of material parameters of the target packaging structure, and output the optimal combination of material parameters corresponding to the optimal strain value, thereby guiding the material selection and optimization design of the packaging structure.

[0064] In this way, by using multiple combinations of material parameters and the corresponding strain simulation results to train the model to be trained, the strain value corresponding to each combination of material parameters can be quickly predicted and the material parameter combination corresponding to the optimal strain value can be directly output, which helps to improve the efficiency and accuracy of optimizing the material parameters of the target packaging structure.

[0065] In this embodiment of the disclosure, the simulation test task includes a warpage simulation test, and the data processing model includes a third data processing sub-model. Based on the simulation test task, a portion of the data is extracted from the dataset as a training dataset, and the model to be trained is trained based on the training dataset to obtain the data processing model. This includes: based on the warpage simulation test, extracting the warpage simulation results corresponding to M sets of material parameter combinations from the dataset as a third training data subset, and training the model to be trained based on the third training data subset to obtain the third data processing sub-model.

[0066] In some embodiments, this warpage simulation test is a test task used to evaluate the reliability (warpage value) of the target package structure. Specifically, the warpage value is the bending deformation caused by the difference in the coefficient of thermal expansion of the package materials, and it is a key indicator affecting the reliability of the target package structure during assembly and use. In the warpage simulation test, a larger warpage value is more likely to cause problems such as soldering misalignment and uneven stress on solder balls between the target package structure and the printed circuit board (PCB), while a smaller warpage value indicates higher reliability of the target package structure.

[0067] In some embodiments, the third data processing sub-model is used to predict the warpage value corresponding to different combinations of material parameters of the target packaging structure, and output the optimal combination of material parameters corresponding to the optimal warpage value, thereby guiding the material selection and optimization design of the packaging structure.

[0068] In this way, by using multiple combinations of material parameters and the corresponding warpage simulation results to train the model to be trained, the warpage value corresponding to each combination of material parameters can be quickly predicted and the material parameter combination corresponding to the optimal warpage value can be directly output, which helps to improve the efficiency and accuracy of optimizing the material parameters of the target packaging structure.

[0069] In this embodiment of the disclosure, the simulation test task includes fracture simulation test, and the data processing model includes a fourth data processing sub-model. Based on the simulation test task, a portion of the data is extracted from the dataset as a training dataset, and the model to be trained is trained based on the training dataset to obtain the data processing model. This includes: based on the fracture simulation test, extracting the fracture simulation results corresponding to M sets of material parameter combinations from the dataset as a fourth training data subset, and training the model to be trained based on the fourth training data subset to obtain the fourth data processing sub-model.

[0070] In some embodiments, this fracture simulation test is a test task used to evaluate the reliability (fracture value) of the target package structure. The fracture simulation test directly simulates the potential fracture failure process of the target package structure (such as solder ball cracking or colloid delamination). In the fracture simulation test, the lower the fracture value, the higher the reliability of the target package structure.

[0071] In some embodiments, the fourth data processing sub-model is used to predict the fracture value corresponding to different combinations of material parameters of the target packaging structure, and output the optimal combination of material parameters corresponding to the optimal fracture value, thereby guiding the material selection and optimization design of the packaging structure.

[0072] In this way, by using multiple combinations of material parameters and the corresponding fracture simulation results to train the model to be trained, the fracture value corresponding to each combination of material parameters can be quickly predicted and the material parameter combination corresponding to the optimal fracture value can be directly output, which helps to improve the efficiency and accuracy of optimizing the material parameters of the target packaging structure.

[0073] Figure 4 This is a schematic flowchart illustrating the material matching process for a covered FCBGA package structure according to an embodiment of the present disclosure, as shown below. Figure 4 As shown, the method includes at least the following steps:

[0074] S401: Obtain the target material parameters for the capped FCBGA package structure;

[0075] S402: Use the target material parameters as design variables, and generate N sets of target material parameter combinations based on the target material parameters through target experimental design; where N is a positive integer;

[0076] S403: Based on the simulation test task, select the data processing sub-model to be called from the data processing model; the data processing sub-model is obtained based on the above data processing model training method.

[0077] S404: Input N sets of target material parameter combinations into the data processing sub-model to obtain the optimal material parameter combination of the capped FCBGA package structure output by the data processing sub-model;

[0078] S405: Configure the covered FCBGA package structure according to the optimal combination of material parameters.

[0079] In some embodiments, the target material parameters refer to the parameters of five materials in the covered FCBGA package structure: Lid, adhesive, thermal interface material, underfill adhesive, and substrate.

[0080] In some embodiments, the target material parameter combination refers to using the parameters of five materials—Lid, adhesive, thermal interface material, underfill adhesive, and substrate—of the covered FCBGA package structure as design variables, and generating N sets of target material parameter combinations through target experimental design based on the parameters of these five materials.

[0081] In some embodiments, the simulation test task refers to a test task used to verify the reliability of the covered FCBGA package structure. Verifying the reliability of the covered FCBGA package structure may include: stress simulation testing, strain simulation testing, warpage simulation testing, and fracture simulation testing.

[0082] For example, if the simulation test task is a reliability (warpage) simulation test of a covered FCBGA package structure, the material matching step of the covered FCBGA package structure may include:

[0083] Step 1: Obtain the target material parameters for the flip-chip ball grid array package with a lid; specifically, this includes parameters for five parts: lid, adhesive, thermal interface material, underfill adhesive, and substrate.

[0084] Step two, which involves using the parameters of five materials—Lid, adhesive, thermal interface material, underfill adhesive, and substrate—as design variables, and generating N sets of target material parameter combinations based on these parameters through target experimental design, may include:

[0085] Step a: Select the coefficient of thermal expansion and modulus of the five materials (Lid, adhesive, thermal interface material, underfill adhesive, and substrate) as design variables (ten in total).

[0086] Step b: Select a suitable parameter range from the coefficients of thermal expansion and moduli of five materials: Lid, adhesive, thermal interface material, underfill adhesive, and substrate; within the parameter range, take three or four values ​​at equal intervals as pre-selected values.

[0087] Step c: The above pre-selected values ​​are used to generate 64 material parameter combinations through target experimental design; and 12 points are taken at equal intervals within the parameter range and the two endpoints are removed for linear combination to generate 10 material parameter combinations.

[0088] Step 3: Based on the reliability (warpage) simulation test of the covered FCBGA package structure, call the third data processing sub-model from the data processing model; the third data processing sub-model is obtained based on the above data processing model training method.

[0089] Step four: Input the above 74 sets of target material parameter combinations into the third data processing sub-model to obtain the optimal material parameter combination for the capped FCBGA package structure output by the third data processing sub-model. Among the warpage values ​​corresponding to the 72 sets of target material parameter combinations, this material parameter combination has the smallest warpage value. Figure 5 A schematic diagram of the curve showing the change in warpage value is shown, such as... Figure 5 As shown, the optimal combination of material parameters does significantly reduce the warpage of the target package structure. Configuring a covered FCBGA package structure based on this optimal combination of material parameters can ensure the reliability (e.g., warpage) of the covered FCBGA package structure.

[0090] In this way, by optimizing multiple sets of target material parameters for the capped FCBGA package structure based on simulation test tasks, the optimal combination of material parameters can be accurately located, which helps to improve the reliability of the capped FCBGA package structure, thereby reducing trial and error costs and R&D cycle.

[0091] This disclosure provides a data processing model training apparatus, such as... Figure 6 As shown, the data processing model training device includes:

[0092] Module 601 is used to construct a three-dimensional finite element model of the target packaging structure based on the geometric and material parameters of the target packaging structure.

[0093] The first generation module 602 is used to input M sets of material parameter combinations generated by target experimental design based on material parameters into a three-dimensional finite element model, using material parameters as design variables, to obtain a dataset; wherein, the dataset includes M sets of material parameter combinations and corresponding M sets of target simulation results; M is a positive integer;

[0094] The training module 603 is used to extract a portion of the data from the dataset as a training dataset based on the simulation test task, and to train the model to be trained based on the training dataset to obtain a data processing model; wherein, the data processing model includes a data processing sub-model corresponding to the simulation test task.

[0095] In some embodiments, the simulation test task includes stress simulation test, the data processing model includes a first data processing sub-model, and the training module 603 includes: extracting stress simulation results corresponding to M sets of material parameter combinations from the dataset based on the stress simulation test as a first training data subset, and training the model to be trained based on the first training data subset to obtain the first data processing sub-model.

[0096] In some embodiments, the simulation test task includes strain simulation test, the data processing model includes a second data processing sub-model, and the training module 603 includes: extracting strain simulation results corresponding to M sets of material parameter combinations from the dataset based on the strain simulation test as a second training data subset, and training the model to be trained based on the second training data subset to obtain the second data processing sub-model.

[0097] In some embodiments, the simulation test task includes a warpage simulation test, and the data processing model includes a three-data-processing sub-model. The training module 603 includes: extracting the warpage simulation results corresponding to M sets of material parameter combinations from the dataset based on the warpage simulation test as a third training data subset, and training the model to be trained based on the third training data subset to obtain the third data-processing sub-model.

[0098] In some embodiments, the simulation test task includes fracture simulation test, and the data processing model includes a fourth data processing sub-model. The training module 603 includes: extracting fracture simulation results corresponding to M sets of material parameter combinations from the dataset based on the fracture simulation test as a fourth training data subset, and training the model to be trained based on the fourth training data subset to obtain the fourth data processing sub-model.

[0099] The data processing model training device of this disclosure can train the data processing model by constructing a dedicated training dataset for different simulation test task scenarios, thereby achieving deep coupling between simulation test tasks and data processing model capabilities; and can perform mechanical property analysis on different materials based on simulation results, thereby improving the efficiency of evaluating the reliability of packaging structures and reducing trial and error costs.

[0100] This disclosure provides a material matching device for a capped FCBGA package structure, such as... Figure 7 As shown, the material matching device for the covered FCBGA package structure includes:

[0101] The first acquisition module 701 is used to acquire the target material parameters of the covered FCBGA package structure;

[0102] The second generation module 702 is used to take the target material parameters as design variables and generate N sets of target material parameter combinations based on the target material parameters through target experimental design; where N is a positive integer;

[0103] Module 703 is used to select the data processing sub-model to be called from the data processing model according to the simulation test task; the data processing sub-model is obtained based on the above-mentioned data processing model training method.

[0104] The second acquisition module 704 is used to input N sets of target material parameter combinations into the data processing sub-model to obtain the optimal material parameter combination of the covered FCBGA package structure output by the data processing sub-model.

[0105] Configuration module 705 is used to configure the covered FCBGA package structure according to the optimal combination of material parameters.

[0106] The material matching device for the covered FCBGA package structure in this embodiment optimizes multiple sets of target material parameters of the covered FCBGA package structure according to the simulation test task, which can accurately locate the optimal combination of material parameters, help improve the reliability of the covered FCBGA package structure, and thus reduce trial and error costs and development cycle.

[0107] The acquisition, storage, and application of user personal information involved in the technical solution disclosed herein comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0108] Figure 8 This is a structural block diagram of a computer device according to an embodiment of the present disclosure. Figure 8 As shown, the computer device includes a memory 810 and a processor 820. The memory 810 stores a computer program that can run on the processor 820. The number of memories 810 and processors 820 can be one or more. The memory 810 can store one or more computer programs, which, when executed by the computer device, cause the computer device to perform the methods provided in the above-described method embodiments. The computer device may also include a communication interface 830 for communicating with external devices and performing data exchange and transmission.

[0109] If the memory 810, processor 820, and communication interface 830 are implemented independently, they can be interconnected via a bus to communicate with each other. This bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. This bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 8 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0110] Optionally, in a specific implementation, if the memory 810, processor 820, and communication interface 830 are integrated on a single chip, then the memory 810, processor 820, and communication interface 830 can communicate with each other through an internal interface.

[0111] It should be understood that the aforementioned processor can be a Central Processing Unit (CPU), or other general-purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field-Programmable Gate Arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. General-purpose processors can be microprocessors or any conventional processor. It is worth noting that the processor can be a processor supporting Advanced Reduced Instruction Set Machines (ARM) architecture.

[0112] Furthermore, the aforementioned memory may include read-only memory and random access memory, and may also include non-volatile random access memory. The memory may be volatile or non-volatile, or may include both. Non-volatile memory may include read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. Volatile memory may include random access memory (RAM), which serves as an external cache. Many forms of RAM are available by way of example, but not limitation. Examples include Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), Synchronous DRAM (SDRAM), Double Data Rate Synchronous DRAM (DDR SDRAM), Enhanced Synchronous DRAM (ESDRAM), Synchlink DRAM (SLDRAM), and Direct RAMBUS RAM (DR RAM).

[0113] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. A computer program product includes one or more computer instructions. When the computer instructions are loaded and executed on a computer, all or part of the flow or function according to the embodiments of this disclosure is generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer instructions can be transmitted from one website, computer, server, or data center to another via wired (e.g., coaxial cable, fiber optic, Digital Subscriber Line, DSL) or wireless (e.g., infrared, Bluetooth, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access, or a data storage device such as a server or data center that integrates one or more available media. The available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., Digital Versatile Discs (DVDs)), or semiconductor media (e.g., Solid State Disks (SSDs)). It is worth noting that the computer-readable storage media mentioned in this disclosure can be non-volatile storage media; in other words, they can be non-transient storage media.

[0114] Those skilled in the art will understand that all or part of the steps of the above embodiments can be implemented by hardware, or by a program instructing related hardware. The program can be stored in a computer-readable storage medium, such as a read-only memory, a disk, or an optical disk.

[0115] In the description of the embodiments of this disclosure, the reference to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refers to a specific feature, structure, material, or characteristic described in connection with that embodiment or example, which is included in at least one embodiment or example of this disclosure. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of those different embodiments or examples.

[0116] The above are merely exemplary embodiments of this disclosure and are not intended to limit this disclosure. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this disclosure should be included within the protection scope of this disclosure.

Claims

1. A data processing model training method, characterized in that, The method includes: Based on the geometric and material parameters of the target packaging structure, a three-dimensional finite element model of the target packaging structure is constructed. Using the material parameters as design variables, M sets of material parameter combinations generated through target experimental design based on the material parameters are input into the three-dimensional finite element model to obtain a dataset; wherein, the dataset includes M sets of the material parameter combinations and corresponding M sets of target simulation results; where M is a positive integer; Based on the simulation test task, a portion of the data is extracted from the dataset as a training dataset, and the model to be trained is trained based on the training dataset to obtain a data processing model; wherein, the data processing model includes a data processing sub-model corresponding to the simulation test task.

2. The method according to claim 1, characterized in that, The simulation test task includes stress simulation testing, and the data processing model includes a first data processing sub-model. The step of extracting a portion of data from the dataset based on the simulation test task as a training dataset, and training the model to be trained based on the training dataset to obtain the data processing model, includes: Based on the stress simulation test, the stress simulation results corresponding to the M sets of material parameter combinations are extracted from the dataset as the first training data subset, and the training model is trained based on the first training data subset to obtain the first data processing sub-model.

3. The method according to claim 1, characterized in that, The simulation test task includes strain simulation testing, and the data processing model includes a second data processing sub-model. The step of extracting a portion of data from the dataset based on the simulation test task as a training dataset, and training the model to be trained based on the training dataset to obtain the data processing model, includes: Based on the strain simulation test, the strain simulation results corresponding to the M sets of material parameter combinations are extracted from the dataset as the second training data subset, and the training model is trained based on the second training data subset to obtain the second data processing sub-model.

4. The method according to claim 1, characterized in that, The simulation test task includes a warpage simulation test, and the data processing model includes a third data processing sub-model. The step of extracting a portion of data from the dataset based on the simulation test task as a training dataset, and training the model to be trained based on the training dataset to obtain the data processing model, includes: Based on the warping simulation test, the warping simulation results corresponding to the M sets of material parameter combinations are extracted from the dataset as the third training data subset, and the training model is trained based on the third training data subset to obtain the third data processing sub-model.

5. The method according to claim 1, characterized in that, The simulation test task includes fracture simulation testing, and the data processing model includes a fourth data processing sub-model. The process involves extracting a portion of the dataset from the simulation test task as a training dataset, and training the model to be trained based on the training dataset to obtain the data processing model, including: Based on the fracture simulation test, the fracture simulation results corresponding to the M sets of material parameter combinations are extracted from the dataset as the fourth training data subset, and the training model is trained based on the fourth training data subset to obtain the fourth data processing sub-model.

6. A method for matching materials in a capped FCBGA package structure, characterized in that, The method includes: Obtain the target material parameters for the capped FCBGA package structure; The target material parameters are used as design variables, and N sets of target material parameter combinations are generated based on the target material parameters through target experimental design; where N is a positive integer; According to the simulation test task, a data processing sub-model to be called is selected from the data processing model; the data processing sub-model is obtained based on the data processing model training method according to any one of claims 1 to 5. The N sets of target material parameter combinations are input into the data processing sub-model to obtain the optimal material parameter combination of the capped FCBGA package structure output by the data processing sub-model. The capped FCBGA package structure is configured according to the optimal combination of material parameters.

7. A data processing model training device, characterized in that, The device includes: A construction module is used to construct a three-dimensional finite element model of the target packaging structure based on the geometric and material parameters of the target packaging structure. The first generation module is used to input M sets of material parameter combinations generated by target experimental design based on the material parameters into the three-dimensional finite element model, using the material parameters as design variables, to obtain a dataset; wherein, the dataset includes M sets of the material parameter combinations and corresponding M sets of target simulation results; M is a positive integer; The training module is used to extract a portion of the data from the dataset as a training dataset based on the simulation test task, and to train the model to be trained based on the training dataset to obtain a data processing model; wherein, the data processing model includes a data processing sub-model corresponding to the simulation test task.

8. A material matching device for an FCBGA package structure with a cap, characterized in that, The device includes: The first acquisition module is used to acquire the target material parameters of the covered FCBGA package structure; The second generation module is used to take the target material parameters as design variables and generate N sets of target material parameter combinations based on the target material parameters through target experimental design; wherein, N is a positive integer; The selection module is used to select the data processing sub-model to be invoked from the data processing model according to the simulation test task; the data processing sub-model is obtained based on the data processing model training method according to any one of claims 1 to 5. The second acquisition module is used to input N sets of target material parameter combinations into the data processing sub-model to obtain the optimal material parameter combination of the covered FCBGA package structure output by the data processing sub-model. A configuration module is used to configure the covered FCBGA package structure according to the optimal combination of material parameters.

9. An electronic device, characterized in that, It includes a processor and a memory, wherein the memory stores a computer program, and when the processor calls the computer program in the memory, it executes the data processing model training method as described in any one of claims 1 to 5 or the material matching method for the covered FCBGA package structure as described in claim 6.

10. A computer-readable storage medium, characterized in that, It stores a computer program, which is loaded by a processor to execute the data processing model training method according to any one of claims 1 to 5 or the material matching method for the covered FCBGA package structure according to claim 6.

Images