BGA welding spot shearing performance prediction method based on machine learning and finite element analysis
By combining machine learning and finite element analysis, the shear performance of BGA solder joints is predicted, and the problem of difficulty in modeling the solder joint structure and insufficient research on the impact of IMC thickness changes is solved, and efficient solder joint reliability evaluation and R&D efficiency improvement is achieved.
Patent Information
- Application Number
- CN202410690271.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Priority Date
- 2024-05-28
- Filing Date
- 2024-05-30
- Publication Date
- 2024-11-08
AI Technical Summary
The prior art is difficult to effectively predict the shear performance of BGA solder joints, especially because the solder joint structure is difficult to substitute for a simple mechanical model, and the impact of IMC thickness changes on the shear performance is not studied carefully enough, which makes it difficult to guarantee the reliability of solder joints.
Using machine learning and finite element analysis methods, we collect experimental data of lead-free solder alloys, establish machine learning models, combine finite element analysis to simulate the shear performance of solder joints, and evaluate the shearing of IMC layer thickness after solder joint aging. performance.
The test cost of obtaining tensile properties of solder alloys is reduced, the R&D efficiency is improved, the impact of tensile properties of solder alloys and IMC thickness on the shear crack initiation strength of solder joints is determined, and the reliability prediction accuracy of solder joints is improved.
Smart Images

Figure CN118918988A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of electronic packaging material reliability, and in particular to a BGA solder joint shear performance prediction method based on machine learning and finite element analysis. Background Art
[0002] Nowadays, the three-level packaging widely used in electronic products and equipment adopts micro-connection. The number of solder joints is increasing, the size is getting smaller, and the packaging density is getting higher. The solder joints have become the weakest link in the packaging process. The reliability of the solder joints will directly determine the reliability of electronic products and equipment. According to statistics, the failure of the interconnection solder joints inside the device is the main cause of the failure of electronic components. And 55% of the solder joint failures are caused by the thermal stress caused by warping under temperature changes due to the mismatch of thermal expansion coefficients between various materials in the packaging structure.
[0003] The bottom surface of the electronic device of the ball grid array package (BGA) can be used as pins, and has a shorter average wire length, which greatly improves the integration and has better high-speed performance. The thermal stress on the solder joints of this packaging mode can be decomposed into shear stress parallel to the solder joint connection interface and tensile stress perpendicular to the interface, mainly shear stress. At present, the reliability of BGA solder joints is mainly measured by the shear performance of solder joints. In the field of electronic packaging, in order to ensure the realization of high-reliability solder joints, we should test and analyze the shear strength of solder joints.
[0004] The process of solder joints in use is equivalent to the process of thermal aging. The growth of IMC layer thickness during thermal aging is considered to be the main factor affecting the shear performance of solder joints. However, the research on the effect of IMC thickness change on the shear performance of solder joints in the past was not detailed and comprehensive enough, especially because the structure of solder joints is difficult to substitute into simple mechanical models. Therefore, further research is needed to link IMC thickness change with shear performance. Summary of the invention
[0005] In view of the above situation, in order to overcome the defects of the prior art, the present invention provides a BGA solder joint shear performance prediction method based on machine learning and finite element analysis, which at least solves some of the problems in the prior art.
[0006] To achieve the above purpose, the technical solution adopted by the present invention is as follows:
[0007] A method for predicting shear performance of BGA solder joints based on machine learning and finite element analysis comprises the following steps:
[0008] S1. Collect experimental data on alloy composition and tensile properties of lead-free solder to establish an initial data set.
[0009] S2, using alloy composition as input and tensile properties as output, using machine learning model for modeling and training; the model is validated using leave-one-out cross validation method;
[0010] S3, designing alloy composition to construct virtual samples, and inputting the virtual samples into the machine learning model to obtain prediction results;
[0011] S4. Establish a finite element model for BGA solder ball shear test and use the displacement method of the pusher to perform finite element analysis;
[0012] S5. The thickness of the IMC layer of the solder joint is used as a control factor to evaluate the shear performance of the solder joint after aging and to determine the resistance of solder joints with different compositions to aging.
[0013] In a further embodiment, in step S1, the alloy components of the lead-free solder include Sn, Ag, Cu, Bi, In and Ti.
[0014] In a further embodiment, in step S1, the tensile properties include yield strength, tensile strength, plastic strain and elongation in the strengthening stage.
[0015] In a further embodiment, in steps S2 and S3, the machine learning model is a random forest model.
[0016] In a further embodiment, in step S3, the lead-free solder alloy is predicted to include, by weight percentage, 1.0-3.8% Ag, 0.5-0.7% Cu, 0.5-5.5% Bi, 2.5-5.5% In, 0.1-0.8% Ti, and the remainder is Sn.
[0017] In a further solution, in step S4, the finite element model is a semi-symmetrical model, and the model structure parameters are: solder ball diameter 760 μm, shear height 60 μm, IMC layer thickness 3 μm and 7 μm.
[0018] In a further solution, in step S4, the step of establishing a finite element model for a BGA solder ball shear test includes:
[0019] Abaqus software was used to establish a semi-symmetrical model for BGA solder ball shear test. The area between the pusher and the solder ball contact point and the IMC layer interface was divided into a minimum block of 5×20×20μm. 3 Divide the area close to the 1 / 4 pushed to one side into a minimum block of 5×7×7μm 3 The other parts are divided by free mesh;
[0020] The solder alloy is a viscoplastic material, and the IMC layer and the pusher are linear elastic materials;
[0021] Boundary conditions: The solder alloy is connected to the IMC layer, restricting the movement of the bottom of the IMC layer, applying constraints perpendicular to the symmetry plane to the entire symmetry plane, and selecting the x-axis as the moving direction of the pusher.
[0022] In order to fully simulate the deformation of the solder ball caused by the contact between the push knife and the solder ball, the interaction between the push knife and the solder ball surface adopts the standard surface-to-surface contact between the push knife and the solder ball contact surface and the solder ball surface. The interaction properties are hard contact and no friction. In the initial state, a certain distance is kept between the push knife and the solder ball.
[0023] In a further solution, in step S4, the step of performing finite element analysis using the displacement method of the push tool includes:
[0024] The reaction force received by the pusher when it is displaced is the corresponding shear force received by the solder joint. The judgment condition for solder joint cracking is that the stress value received by a certain node exceeds the tensile strength. The equivalent stress of the node where the tensile force in the solder joint first reaches the tensile strength is analyzed, so as to obtain the finite element analysis results of the BGA ball shear test under the shear force of the pusher, and obtain the relationship between the shear force and the maximum equivalent stress value.
[0025] In a further embodiment, the BGA solder joint shear performance prediction method based on machine learning and finite element analysis further includes the following steps:
[0026] BGA solder joints with predicted performance were prepared and compared with SAC387 alloy through shear performance test to evaluate the prediction accuracy of this method for the shear performance of BGA solder joints.
[0027] In a further embodiment, the BGA solder joint shear performance prediction method based on machine learning and finite element analysis further includes the following steps:
[0028] The results predicted by the finite element analysis in step S4 are compared with the finite element simulation results of the SAC387 alloy to evaluate the shear strength; a shear test is performed on the BGA solder joints of the solder alloy with the predicted composition and the SAC387 to evaluate the accuracy of the prediction.
[0029] In a further solution, in step S5, the IMC layer thickness and the shear performance of the solder alloy are both measured in accordance with actual tests. The IMC layer thickness of 3 μm and 7 μm are approximate values of the SAC387 alloy before aging and after aging at 150°C for 25 days, respectively.
[0030] Compared with the prior art, the present invention has the following beneficial effects:
[0031] 1. The present invention innovatively uses machine learning strategies to predict the tensile properties of alloyed solder alloys, which can reduce the economic and time costs of experiments, reduce the experimental costs of obtaining the tensile properties of solder alloys, and improve research and development efficiency.
[0032] 2. The BGA ball shear test finite element model used in the present invention determines the influence of the tensile properties of the solder alloy and the IMC thickness on the shear crack initiation strength of the solder joint, as well as the decreasing law of the shear crack initiation strength of the BGA solder joint after the solder alloy has aged to a certain extent. BRIEF DESCRIPTION OF THE DRAWINGS
[0033] Figure 1 is a simplified flow chart of the innovative method used by the present invention.
[0034] Figure 2 is a flow chart of the method for predicting shear performance of solder alloy solder joints of the present invention;
[0035] Figure 3 Schematic diagram of the finite element model of the present invention; (a) represents the solder joint model, (b) represents the model load, 1 represents the solder joint, 2 represents the IMC layer, and 3 represents the shear pusher;
[0036] Figure 4 1 is the leave-one-out cross validation result of the prediction of the four characteristics of tensile properties in Example 3 of the present invention; (a) represents the tensile strength, (b) represents the elongation, (c) represents the strain corresponding to the tensile strength, and (d) represents the yield strength;
[0037] Figure 5 The figures are the comparison results of the predicted tensile properties and the actual tensile curves of Examples 1-4 of the present invention; (a) represents Examples 1-2, and (b) represents Examples 3-4;
[0038] Figure 6 1 is a stress distribution diagram of the finite element analysis results of Examples 1-4 of the present invention; the arrows in the figure indicate the crack initiation points; (a) in the figure represents Example 1, (b) represents Example 2, (c) represents Example 3, and (d) represents Example 4;
[0039] Figure 7 The present invention compares stress shear force curves based on finite element results of predicted tensile properties and actual tensile properties. DETAILED DESCRIPTION
[0040] The following non-limiting examples can enable those skilled in the art to more fully understand the present invention, but do not limit the present invention in any way. The following content is only an exemplary description of the scope of the present invention, and those skilled in the art can make various changes and modifications to the invention of the present invention based on the disclosed content, and they should also fall within the scope of the present invention.
[0041] The thermal stress on the solder joints of the ball grid array package can be decomposed into shear stress parallel to the solder joint connection interface and tensile stress perpendicular to the interface, mainly shear stress. At present, the reliability of BGA solder joints is mainly measured by the shear performance of the solder joints. The process of solder joints in use is equivalent to the process of thermal aging. The growth of the IMC layer thickness during the thermal aging process is considered to be the main factor affecting the shear performance of the solder joints. However, the research on the effect of IMC thickness change on the shear performance of solder joints in the past was not detailed and comprehensive enough, especially because the structure of the solder joint is difficult to substitute into a simple mechanical model, so the connection between IMC thickness change and shear performance still needs further research.
[0042] By combining finite element analysis with machine learning, the present invention can reduce experimental costs through machine learning strategies and predict the shear performance of BGA solder joints in advance through finite element models, eliminating the experimental costs of obtaining the tensile properties of solder alloys, improving research and development efficiency, and determining the influence of the tensile properties of solder alloys and IMC thickness on the shear crack initiation strength of solder joints.
[0043] To solve the above problems, the embodiment of the present invention provides a BGA solder joint shear performance prediction method based on machine learning and finite element analysis, and the simplified flow chart is as follows: Figure 1 As shown, the machine learning model is first trained by the random forest algorithm to predict the tensile properties of solder alloys with different compositions; then the predicted tensile properties and the thickness of the IMC layer are used as the control factors of the finite element analysis model, and the shear force magnitude when the crack of the solder joints of different embodiments is obtained through finite element simulation analysis, and the shear performance of the corresponding solder joints of each embodiment is evaluated. This method can quickly and effectively predict the shear strength of solder joints with different compositions without experiments through the method of calculating two materials, which plays a guiding role in the design of lead-free solder compositions corresponding to high-reliability solder joints.
[0044] A method for predicting the shear performance of BGA solder joints based on machine learning and finite element analysis, the specific flow chart is as follows: Figure 2 As shown, the specific steps include:
[0045] S1. Collect experimental data on alloy composition and tensile properties of lead-free solder to establish an initial data set; alloy composition includes Sn, Ag, Cu, Bi, In and Ti; tensile properties include yield strength, tensile strength, plastic strain and elongation in the strengthening stage.
[0046] S2. With alloy composition as input and tensile properties as output, the random forest model in the machine learning model is used for modeling and training. The model is verified by the leave-one-out cross-validation method.
[0047] S3. Design alloy composition to construct a virtual sample, and then input the virtual sample into the machine learning model to obtain a prediction result; obtain a lead-free solder alloy according to the prediction result; obtain a lead-free solder alloy according to the prediction result, which includes, by weight percentage: 1.0-3.8% Ag, 0.5-0.7% Cu, 0.5-5.5% Bi, 2.5-5.5% In, 0.1-0.8% Ti, and the remainder is Sn.
[0048] S4. Establish a semi-symmetric finite element model for BGA solder ball shear test, such as Figure 3 As shown, the solder joint shape is spherical. Considering the growth trend of IMC, it is close to flat after long-term aging, so the interface between the IMC layer and the solder can be simplified to a plane. A finite element semi-symmetric model of BGA ball shear test is established using Abaqus software. The steps are as follows: A semi-symmetric ball shear test model of BGA ball shear test is established using Abaqus software. The area between the push knife and the solder ball contact point and the IMC interface is divided into a minimum block of 5×20×20μm. 3 Divide the area close to the 1 / 4 pushed to one side into a minimum block of 5×7×7μm 3 The other parts are divided by free mesh. The model structure parameters are shown in Table 1:
[0049] Table 1
[0050] Solder joint structure parameters Numeric Welding spot diameter 760μm Cutting height 60μm
[0051] In a further scheme, the solder alloy is a viscoplastic material, the tensile properties of the solder alloy are predicted by the machine learning model, and the elastic properties of all materials are measured by nanoindentation;
[0052] The IMC layer and the pusher are linear elastic materials;
[0053] The parameters of each material in the BGA model are shown in Table 2 below:
[0054] Table 2
[0055] Solder joint structure Material Young's modulus(GPa) Poisson's ratio Push Knife Tungsten steel 450 0.25 IMC Cu6Sn5 100 0.35 Solder Alloy SnAgCuBiInTi lead-free solder alloy 55 0.35
[0056] The thickness of the IMC layer is approximately 3 μm and 7 μm for the SAC387 alloy before aging and after aging at 150 °C for 25 days, respectively;
[0057] Boundary conditions: The solder alloy is connected to the IMC layer, restricting the movement of the bottom of the IMC layer, applying a constraint perpendicular to the symmetry plane to the entire symmetry plane, and selecting the x-axis as the direction of movement of the pusher;
[0058] In order to fully simulate the deformation of the solder ball caused by the contact between the push knife and the solder ball, the interaction between the push knife and the solder ball surface adopts the standard surface-to-surface contact between the push knife and the solder ball contact surface and the solder ball surface. The interaction properties are hard contact and no friction. In the initial state, a certain distance is kept between the push knife and the solder ball.
[0059] The finite element analysis is performed using the displacement method of the pusher. The reaction force received by the pusher during displacement is the corresponding shear force received by the weld. The thrust direction of the pusher is as follows: Figure 3 (b) The judgment condition for solder joint cracking is that the stress value of a certain unit exceeds the tensile strength. The equivalent stress of the node where the tensile force in the solder joint first reaches the tensile strength is analyzed, so as to obtain the finite element analysis results of the BGA ball shear test under the push knife shear force, and the relationship between the shear force and the maximum equivalent stress value is obtained.
[0060] S5. The thickness of the IMC layer of the solder joint is used as a control factor to evaluate the shear performance of the solder joint after aging and to determine the resistance of solder joints with different compositions to aging.
[0061] In a further embodiment, the BGA solder joint shear performance prediction method based on machine learning and finite element analysis further includes:
[0062] (1) Prepare BGA solder joints with predicted performance and compare them with SAC387 alloy through shear performance test to evaluate the prediction accuracy of this method for the shear performance of BGA solder joints;
[0063] (2) The finite element model in step S4 is used to predict the tensile properties based on machine learning and the crack initiation force in the shear test obtained through simulation analysis, and the shear strength is evaluated by comparing with the finite element simulation results of the SAC387 alloy; a shear test is performed on the solder alloy with the predicted composition and the BGA solder joint of SAC387 to evaluate the accuracy of the prediction.
[0064] The following specific embodiments and comparative examples take lead-free BGA solder joint materials with specific components as an example to illustrate the BGA solder joint shear performance prediction method based on machine learning and finite element analysis of the present invention.
[0065] Example 1
[0066] Lead-free BGA solder joint materials, by weight percentage, include: 89.4% Sn, 3.8% Ag, 0.7% Cu, 1.5% Bi, 4.4% In, 0.128% Ti. The structural parameters of the finite element analysis model are: solder ball diameter 760μm, shear height 60μm, IMC thickness 3μm.
[0067] Example 2
[0068] Lead-free BGA solder joint materials, by weight percentage, include: 89.4% Sn, 3.8% Ag, 0.7% Cu, 1.5% Bi, 4.4% In, 0.128% Ti. The structural parameters of the finite element analysis model are: solder ball diameter 760μm, shear height 60μm, IMC thickness 7μm.
[0069] Example 3
[0070] Lead-free BGA solder joint materials, by weight percentage, include: 94% Sn, 3.8% Ag, 0.7% Cu, 1.5% Bi. The structural parameters of the finite element analysis model are: solder ball diameter 760μm, shear height 60μm, IMC thickness 3μm.
[0071] Example 4
[0072] Lead-free BGA solder joint materials, by weight percentage, include: 94% Sn, 3.8% Ag, 0.7% Cu, 1.5% Bi. The structural parameters of the finite element analysis model are: solder ball diameter 760μm, shear height 60μm, IMC thickness 7μm.
[0073] Comparative Example 1
[0074] Lead-free BGA solder joint material, by weight percentage, includes: 95.5% Sn, 3.8% Ag, 0.7% Cu. The structural parameters of the finite element analysis model are: solder ball diameter 760μm, shear height 60μm, IMC thickness 3μm.
[0075] Comparative Example 2
[0076] Lead-free BGA solder joint materials, by weight percentage, include: 95.5% Sn, 3.8% Ag, 0.7% Cu. The structural parameters of the finite element analysis model are: solder ball diameter 760μm, shear height 60μm, IMC thickness 7μm.
[0077] Test Example 1
[0078] The shear force, pusher displacement and crack height when the finite element analysis model in the above embodiment and comparative example generates cracks are shown in Table 3, and the shear test results are shown in Table 4. Among them, the predicted value is the simulation result of the tensile curve predicted by the machine learning method; the true value is the simulation result of the tensile curve measured by the actual experiment.
[0079] Table 3
[0080]
[0081] Table 4
[0082] Group Shear strength(MPa) Maximum shear force(N) Example 1 105.12 33.44 Example 2 103.74 32.34 Example 3 95.41 30.35 Example 4 89.75 28.55 Comparative Example 1 79.69 24.84 Comparative Example 2 61.09 19.04
[0083] As can be seen from Table 3, the crack initiation shear force estimated by the method of predicting the tensile curve by machine learning in the four embodiments differs from the shear force estimated by the actual tensile curve by 0.29N, 0.26N, 0.06N and 0.14N, respectively, with an error of less than 5%. The predicted pusher displacement and crack height are also basically the same as the simulation results of the real tensile curve. This shows that the use of machine learning methods to predict tensile performance and apply it to finite element models has certain guiding significance for the evaluation of shear performance.
[0084] The magnitude of the crack initiation strength in the finite element simulation results can be used as a basis for evaluating the shear strength of the solder joint. For Example 1 and Example 3, it can be seen from Tables 3 and 4 that the shear force of crack initiation is positively correlated with the shear strength. It is shown that the shear strength of the solder joint can also be evaluated according to the shear stress magnitude at the time of crack initiation in the finite element simulation results. The shear strength of Example 2 and Example 4 is the test result when the thickness of the IMC layer at the solder joint interface is approximately 7μm after aging treatment. During the aging process of the actual solder alloy, the tensile properties of the solder alloy itself will also decrease. By comparing Example 1, Example 2, Example 3 and Example 4, it can be seen from Tables 3 and 4 that the results of the finite element analysis can predict the change in the shear strength after changing the thickness of the interface IMC layer, which has a certain guiding significance and can reflect the influence of the interface IMC thickness on the performance of the solder joint. In addition, the finite element analysis method can also predict the location where the solder joint crack may occur, and judge the degree of deformation that the solder joint can resist under a sufficiently large external force based on the displacement of the push knife. It provides help for further research on the reliability of the solder joint.
[0085] Through the application of machine learning technology and finite element analysis methods, the shear strength of solder joints with different components and the impact of interface structure changes on shear strength can be effectively predicted. The weak position of the solder joint and the degree of resistance to deformation can also be determined to achieve the expected technical effect.
[0086] Test Example 2
[0087] The leave-one-out cross-validation results of the random forest algorithm constructed based on the alloy composition characteristics for the tensile properties of the above Examples 1-4 are as follows: Figure 4 As shown in the figure, for a small amount of data, the Pearson correlation coefficient R of the four tensile properties using the leave-one-out cross-validation method is greater than 0.8, which has a good prediction accuracy.
[0088] The comparison results of the predicted tensile properties and the actual tensile curves of Examples 1-4 are as follows Figure 5 As shown in the figure, the tensile curve drawn based on the predicted tensile properties is consistent with the actual tensile curve to a certain extent, indicating that the tensile properties of solder alloys with different compositions predicted by the machine learning model trained with the random forest algorithm can guide the simulation of finite element analysis.
[0089] Figure 6 The stress distribution diagram of the finite element analysis results of the BGA solder joint model when cracks occur in the solder joints of Examples 1-4. The position indicated by the arrow is the crack initiation position. The crack initiation position of all solder joints is between the contact surface between the shear pusher and the solder joint and the IMC layer. The crack heights of different embodiments are shown in Table 3. The displacement of the shear pusher in Table 3 can reflect the deformation of the solder joint. Figure 6 It can also be reflected in.
[0090] Figure 7 To predict the relationship between the shear force and the stress at the crack initiation point of the solder joint in the finite element analysis results under the tensile properties and the actual tensile curve. Since the actual tensile curve does not distinguish between elastic and plastic deformation, there are some differences in the shear force-stress curves of the actual tensile curve and the predicted tensile curve in the elastic deformation stage, but this difference gradually decreases in the plastic deformation stage. The two curves intersect when approaching crack initiation, and the shear force at the crack initiation obtained by simulation analysis is finally close. This shows that the tensile properties of solder alloys with different compositions predicted by machine learning methods are close to the results obtained by finite element analysis using the actual tensile curve.
[0091] The above description of the embodiments is to facilitate the understanding and use of the invention by those skilled in the art. It is obvious that those skilled in the art can easily make various modifications to these embodiments and apply the general principles described herein to other embodiments without creative work. Therefore, the present invention is not limited to the above embodiments, and improvements and modifications made by those skilled in the art based on the disclosure of the present invention without departing from the scope of the present invention should be within the scope of protection of the present invention.
Claims
1. A method for predicting shear performance of BGA solder joints based on machine learning and finite element analysis, It is characterized in that The following steps are involved: S1. Collect experimental data on alloy composition and tensile properties of lead-free solder to establish an initial data set. S2, using alloy composition as input and tensile properties as output, using machine learning models for modeling and training; The model was validated using the leave-one-out cross-validation method; S3, designing alloy composition to construct virtual samples, and inputting the virtual samples into the machine learning model to obtain prediction results; S4. Establish a finite element model for BGA solder ball shear test and use the displacement method of the pusher to perform finite element analysis; S5. The thickness of the IMC layer of the solder joint is used as a control factor to evaluate the shear performance of the solder joint after aging and to determine the resistance of solder joints with different compositions to aging.
2. According to claim 1, a BGA solder joint shear performance prediction method based on machine learning and finite element analysis, It is characterized in that In step S1, the alloy components of the lead-free solder include Sn, Ag, Cu, Bi, In and Ti.
3. According to the method for predicting the shear performance of BGA solder joints based on machine learning and finite element analysis according to claim 1, It is characterized in that In step S1, the tensile properties include yield strength, tensile strength, plastic strain and elongation in the strengthening stage.
4. According to claim 1, a BGA solder joint shear performance prediction method based on machine learning and finite element analysis, It is characterized in that In step S2 and step S3, the machine learning model is a random forest model.
5. According to the method for predicting the shear performance of BGA solder joints based on machine learning and finite element analysis according to claim 1, It is characterized in that In step S3, the lead-free solder alloy is predicted to include, by weight percentage, 1.0-3.8% Ag, 0.5-0.7% Cu, 0.5-5.5% Bi, 2.5-5.5% In, 0.1-0.8% Ti, and the remainder is Sn.
6. According to claim 1, a BGA solder joint shear performance prediction method based on machine learning and finite element analysis, It is characterized in that In step S4, the finite element model is a semi-symmetrical model, and the model structure parameters are: solder ball diameter 760 μm, shear height 60 μm, IMC layer thickness 3 μm and 7 μm.
7. According to claim 1, a method for predicting BGA solder joint shear performance based on machine learning and finite element analysis, It is characterized in that In step S4, the step of establishing a finite element model for BGA solder ball shear test includes: Abaqus software was used to establish a semi-symmetrical model for BGA solder ball shear test. The area between the pusher and the solder ball contact point and the IMC layer interface was divided into a minimum block of 5×20×20μm. 3 Divide the area close to the 1 / 4 pushed to one side into a minimum block of 5×7×7μm 3 The other parts are divided by free mesh; The solder alloy is a viscoplastic material, and the IMC layer and the pusher are linear elastic materials; Boundary conditions: The solder alloy is connected to the IMC layer, limiting the movement of the bottom of the IMC layer, applying a constraint perpendicular to the symmetry plane to the entire symmetry plane, and selecting the x-axis as the push tool movement direction.
8. According to claim 1, a method for predicting BGA solder joint shear performance based on machine learning and finite element analysis, It is characterized in that In step S4, the step of performing finite element analysis using the displacement method of the push tool includes: The equivalent stress of the node where the tensile force in the solder joint first reaches the tensile strength is analyzed, so as to obtain the finite element analysis results of the BGA ball shear test under the push knife shear force, and the relationship between the shear force and the maximum equivalent stress value is obtained.
9. A method for predicting shear performance of BGA solder joints based on machine learning and finite element analysis according to any one of claims 1 to 8, It is characterized in that The following steps are also included: BGA solder joints with predicted performance were prepared and compared with SAC387 alloy through shear performance test to evaluate the prediction accuracy of this method for the shear performance of BGA solder joints.
10. A method for predicting BGA solder joint shear performance based on machine learning and finite element analysis according to claim 9, It is characterized in that The following steps are also included: The predicted results were compared with the finite element simulation results of SAC387 alloy to evaluate the shear strength; shear tests were performed on BGA solder joints of the predicted solder alloy and SAC387 to evaluate the accuracy of the prediction.
