Stress test method and test system for integrated circuit packaging material
By applying static loads to integrated circuit packaging materials and using neural network models to determine the critical point of plastic deformation, the test control mode is adjusted, thus solving the problem of insufficient test accuracy in existing technologies and achieving higher test accuracy and reliability.
Patent Information
- Application Number
- CN202511350691.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-09-22
- Publication Date
- 2026-01-06
AI Technical Summary
Existing testing methods cannot accurately reflect the performance of integrated circuit packaging materials in practical applications, affecting their quality and reliability.
By applying a static load to the packaging material under test, acquiring deformation images, and calling a neural network model to determine the critical point at which plastic deformation occurs, the test control mode, including stress control and strain control, is adjusted according to the critical point to improve the accuracy and reliability of stress testing.
This improves the accuracy and reliability of stress testing for integrated circuit packaging materials, ensuring the application of appropriate test control modes at different deformation stages.
Smart Images
Figure CN121275475A_ABST
Abstract
Description
Technical Field
[0001] This disclosure relates to the field of materials mechanics testing technology, and in particular to a stress testing method and system for integrated circuit packaging materials. Background Technology
[0002] Integrated circuit packaging materials are subjected to various mechanical stresses during production and use. These stresses can easily lead to material deformation, cracking, or other damage, thereby affecting the performance and reliability of integrated circuits.
[0003] Therefore, testing the packaging materials under different mechanical stress conditions is crucial for improving the quality and reliability of integrated circuit packaging.
[0004] However, existing testing methods cannot accurately reflect the performance of encapsulation materials in practical applications. Summary of the Invention
[0005] To address the aforementioned technical problems, this disclosure provides a stress testing method and system for integrated circuit packaging materials. This system can determine the test control mode based on the characteristics of the packaging material under test at different deformation stages, thereby improving the accuracy and reliability of stress testing.
[0006] In a first aspect, embodiments of this disclosure provide a stress testing method for integrated circuit packaging materials, including:
[0007] A static load is applied to a first sample of the packaging material to be tested, and deformation images of the sample are acquired according to a set sampling rate; wherein, the sampling rate is greater than a first threshold, and the deformation images include elastic deformation images and plastic deformation images of the sample;
[0008] The first neural network model is invoked to output the critical point where plastic deformation occurs based on the deformation image;
[0009] Based on the critical point of plastic deformation, a test control mode is determined, and stress testing is performed on the second sample of the packaging material to be tested.
[0010] The test control mode differs before and after the critical point.
[0011] Optionally, determining the test control mode based on the critical point of plastic deformation and performing stress testing on the second sample of the packaging material to be tested includes:
[0012] Before the critical point of plastic deformation occurs, a stress control mode is used to perform stress testing on the second sample of the packaging material under test, wherein the stress control mode performs stress testing on the second sample of the packaging material under test by controlling the load;
[0013] After the critical point of plastic deformation, a strain control mode is used to perform stress testing on the second sample of the packaging material under test, wherein the strain control mode performs stress testing on the second sample of the packaging material under test by controlling the strain rate.
[0014] Optionally, determining the test control mode based on the critical point of plastic deformation and performing stress testing on the second sample of the packaging material to be tested further includes:
[0015] Hardware response delay of stress testing equipment;
[0016] Based on the hardware response delay, the switching between the stress control mode and the strain control mode is controlled.
[0017] Optionally, the plastic deformation image includes a necking deformation image.
[0018] Optionally, the step of calling the first neural network model to output the critical point of plastic deformation based on the deformation image includes:
[0019] The first neural network model is invoked, and based on the deformation image and the timestamps corresponding to each deformation image, the critical point at which plastic deformation occurs is output.
[0020] Optionally, the timestamp includes any of the following:
[0021] Relative timestamp;
[0022] Physical timestamp.
[0023] Optionally, before invoking the first neural network model and outputting the critical point of plastic deformation based on the deformation image, the method further includes:
[0024] Calculate the strain field distribution in the deformation image to obtain the strain contour map of the deformation image.
[0025] In a second aspect, embodiments of this disclosure provide a stress testing system for integrated circuit packaging materials, comprising:
[0026] An image sensor is configured to acquire deformation images of a first sample of packaging material under test when a static load is applied to the sample at a set sampling rate; wherein the sampling rate is greater than a first threshold, and the deformation images include elastic deformation images and plastic deformation images of the sample;
[0027] A first processor, coupled to the image sensor, is configured to invoke a first neural network model and output the critical point of plastic deformation based on the deformation image;
[0028] A stress testing device, coupled to the first processor, is configured to determine a test control mode based on the critical point where the plastic deformation occurs, and to perform stress testing on a second sample of the packaging material to be tested, wherein the test control mode is different before and after the critical point.
[0029] Optionally, the stress testing system further includes:
[0030] The second processor, coupled to the image sensor, is configured to calculate the strain field distribution in the deformation image and obtain the strain cloud map of the deformation image.
[0031] Thirdly, embodiments of this disclosure provide a storage medium having a computer program stored thereon, the computer program being executed by a processor to perform the stress testing method steps for integrated circuit packaging materials described in any of the above embodiments. Attached Figure Description
[0032] To more clearly illustrate the technical solutions in the embodiments of this disclosure or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of this application. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.
[0033] Figure 1 A flowchart illustrating a stress testing method for integrated circuit packaging materials, consistent with some embodiments of this disclosure, is shown.
[0034] Figure 2 A structural example diagram of a stress testing system for integrated circuit packaging materials, consistent with some embodiments of this disclosure, is shown. Detailed Implementation
[0035] Testing packaging materials under different mechanical stress conditions is crucial for improving the quality and reliability of integrated circuit packaging. However, existing testing methods cannot accurately reflect the performance of packaging materials in practical applications.
[0036] To address the aforementioned problems, this disclosure provides a stress testing method for integrated circuit packaging materials. By applying a static load to a first sample of the packaging material under test and acquiring deformation images of the sample at a set sampling rate, the probability of acquiring deformation images during the transition from elastic to plastic deformation can be increased when the sampling rate exceeds a first threshold. By invoking a first neural network model, the critical point of plastic deformation can be output based on the deformation images, where the deformation images include both elastic and plastic deformation images of the sample. By determining a test control mode based on the critical point of plastic deformation, stress testing is performed on a second sample of the packaging material under test, thereby improving the accuracy and reliability of the stress test.
[0037] Figure 1 A flowchart illustrating a stress testing method for integrated circuit packaging materials, consistent with some embodiments of this disclosure, is shown. (Refer to...) Figure 1 In some embodiments, the stress testing method for integrated circuit packaging materials may include S101, S102, and S103.
[0038] S101, apply a static load to the first sample of the packaging material to be tested, and acquire the deformation image of the sample according to the set sampling rate.
[0039] Wherein, the sampling rate is greater than a first threshold, and the deformation image includes the elastic deformation image and the plastic deformation image of the sample.
[0040] A static load refers to a load whose magnitude, direction, and point of application do not change significantly over a certain period of time, or change so slowly that its dynamic effects can be ignored. When such a load acts on an object, it causes static deformation and static stress, meaning that the object's response (such as displacement and stress distribution) is in a stable state and does not fluctuate significantly over time. In other words, a static load has a low strain rate.
[0041] In some embodiments, the sample of the packaging material to be tested may include an epoxy molding compound sample.
[0042] In some embodiments, the sample of the packaging material to be tested may include a polyimide film sample.
[0043] It is understood that the embodiments disclosed herein do not impose specific limitations on the packaging material samples to be tested, and the above embodiments are merely illustrative examples.
[0044] In some embodiments, the test packaging material sample can be prepared by compression molding of the test packaging material.
[0045] In some embodiments, the test packaging material sample can be manufactured by injection molding the test packaging material.
[0046] In some embodiments, a displacement control mode can be used to apply a static load to the test packaging material sample. The displacement control mode controls the movement speed of the actuator. For example, in a universal testing machine, the actuator's movement speed is set to 1 mm / min.
[0047] In some embodiments, a strain control mode can be used to apply a static load to the test encapsulation material sample. The strain control mode controls the actual strain rate of the sample. For example, in a universal testing machine, a constant actual strain rate of 0.001 s⁻¹ can be maintained.
[0048] It is understood that the embodiments of this disclosure do not impose specific limitations on the way static loads are applied. In some embodiments, the moving speed of the actuator can be set to 5 mm / min.
[0049] Since the transition from elastic deformation to plastic deformation of a sample often occurs within a very short strain range, if the sampling rate of the deformation image of the sample is too low, the transition process from elastic deformation to plastic deformation may be missed. Therefore, it is necessary to set the sampling rate to be greater than the first threshold.
[0050] For example, the first threshold can be 20Hz, 50Hz, 100Hz, etc.
[0051] It is understood that the embodiments of this disclosure do not impose specific limitations on the first threshold, as long as an image of the transformation process of the sample from elastic deformation to plastic deformation can be acquired.
[0052] Elastic deformation images typically show uniform deformation and regular strain distribution. For example, when a static load is applied to a sample of packaging material under test, the sample elongates uniformly throughout the elastic deformation image, without any obvious local bulges or depressions.
[0053] In plastic deformation images, the packaging material under test will exhibit phenomena such as localized deformation concentration, abrupt changes in strain gradient, and intensified surface texture distortion.
[0054] S102, invoke the first neural network model, and output the critical point where plastic deformation occurs based on the deformation image.
[0055] In some embodiments, the first neural network model may include a Convolutional Neural Network (CNN) model. A CNN typically includes an input layer, convolutional layers, pooling layers, fully connected layers, and an output layer. The size of the convolutional kernels in each convolutional layer can be set according to the size of the image data input to the input layer and the feature perception requirements. In some embodiments, the image data input to the input layer may, for example, be a deformed image with a size of 1024*1024.
[0056] In some embodiments, before calling the first neural network model to output the critical point of plastic deformation based on the deformation image, the strain field distribution in the deformation image can also be calculated to obtain the strain cloud map of the deformation image.
[0057] For example, a high-resolution sample deformation image can be acquired using an optical strain measurement system at a sampling rate greater than a first threshold. The optical strain measurement system can calculate the displacement and strain field distribution of the first sample of the packaging material under test during static loading and output a strain contour map of the deformation image.
[0058] In some embodiments, the image data input to the input layer of the first neural network model can be, for example, a strain cloud map of a deformation image with a size of 1024*1024, which is more conducive to the first neural network model performing feature extraction of the deformation image.
[0059] In some embodiments, the features of the deformation image may include strain distribution features and strain value features.
[0060] In some embodiments, a first neural network model may be invoked to output the critical point at which plastic deformation occurs, based on the deformation image and the timestamps corresponding to each deformation image.
[0061] By employing the above embodiments, and by precisely synchronizing the timestamps of the deformation image and the applied static load, a one-to-one correspondence between the deformation image and the static load is established, which can further improve the accuracy of the critical point of plastic deformation output by the first neural network model.
[0062] In some embodiments, the critical point at which plastic deformation occurs can be represented by the strain value corresponding to the permanent plastic deformation of the first sample of the packaging material under test.
[0063] In some embodiments, the critical point at which plastic deformation occurs can be represented by the stress value corresponding to the permanent plastic deformation of the first sample of the packaging material under test.
[0064] In some embodiments, a new channel with the same size as the deformed image can be created, where each pixel of the channel is no longer image information but time information.
[0065] In some embodiments, relative timestamps can be used. A relative timestamp refers to a timestamp that indicates the chronological order and relative position of deformed images within a sequence of acquired deformed images, and may be independent of absolute time. For example, the frame number of each deformed image can be used as the time information. Specifically, deformed images can be numbered frame by frame, for example, numbered 1, 2, 3, ..., t. For the t-th frame of the deformed image, all pixel values in its time channel are filled with 't'.
[0066] Using the above embodiments, a one-to-one correspondence between deformation images and static loads is established using relative timestamps. Since relative timestamps are an absolutely stable and noise-free time measurement with no measurement error and independent of the acquisition system, the computational load of the first neural network model can be reduced and the risk of overfitting can be reduced.
[0067] In some embodiments, physical timestamps can be used. A physical timestamp is a continuous quantity with clear physical meaning, acquired synchronously with the deformed image sequence, which can mark the state of the external real world corresponding to each frame of the deformed image. For example, real time can be recorded synchronously. The real time is then used as the pixel value of the time channel.
[0068] Using the above embodiments, since physical timestamps can provide physical laws, establishing a one-to-one correspondence between deformation images and static loads through physical timestamps is beneficial for feature extraction of the first neural network model, thereby further improving the accuracy of the critical point of plastic deformation output by the first neural network model.
[0069] In some embodiments, physical timestamps may be used. For example, static load values may be recorded synchronously and used as pixel values for time channels.
[0070] Using the above embodiments, since the application process of static load may be uneven depending on the method of static load loading, and static load is the root cause of deformation of the first sample of the packaging material under test, it can provide richer information than simply the passage of time. Therefore, when the application process of static load is uneven, the load value can be used as the pixel value of the time channel, thereby further improving the accuracy and robustness of the critical point of plastic deformation output by the first neural network model.
[0071] In some embodiments, to further improve the accuracy and robustness of the critical point of plastic deformation output by the first neural network model, the deformation image of the sample may also include a necking deformation image.
[0072] In some embodiments, the first neural network model may include VGG16.
[0073] In some embodiments, the first neural network model may include ResNet50.
[0074] It is understood that the embodiments disclosed herein do not impose specific limitations on the first neural network model.
[0075] S103, based on the critical point of plastic deformation, determine the test control mode and perform stress test on the second sample of the packaging material to be tested.
[0076] The test control mode differs before and after the critical point.
[0077] In some embodiments, before the critical point of plastic deformation, a stress control mode can be used to perform stress testing on the second sample of the packaging material under test, wherein the stress control mode performs stress testing on the second sample of the packaging material under test by controlling the load. For example, in a universal testing machine, the applied static load can be set to increase at a constant rate, for example, at a constant rate of 0.00025 MPa / s. After the critical point of plastic deformation, a strain control mode can be used to perform stress testing on the second sample of the packaging material under test, wherein the strain control mode performs stress testing on the second sample of the packaging material under test by controlling the strain rate. For example, in a universal testing machine, a constant true strain rate of 0.001 s⁻¹ can be maintained.
[0078] Using the above embodiments, before the critical point of plastic deformation, the second sample of the packaging material under test undergoes elastic deformation and can recover its deformation; therefore, it is necessary to accurately measure the modulus through stress control mode. After the critical point of plastic deformation, the second sample of the packaging material under test undergoes plastic deformation and permanent deformation; therefore, it is necessary to maintain stable loading through strain control. As can be seen, the stress testing method can determine the test control mode according to the characteristics of the packaging material under test at different deformation stages, thereby improving the accuracy and reliability of stress testing.
[0079] In some embodiments, determining the test control mode based on the critical point of plastic deformation and performing stress testing on the second sample of the packaging material to be tested further includes: measuring the hardware response delay of the stress testing equipment; and controlling the switching between the stress control mode and the strain control mode based on the hardware response delay.
[0080] For example, in a universal testing machine, there may be an actuator response delay. Specifically, during stress testing, when a "stop loading" command is issued to the actuator, there is a time required for the command to be issued and for the actuator to complete the action. Therefore, the time difference between the time the command is issued and the time the actuator completes the command can be considered as the response delay of the mechanical transmission between the actuator and the universal testing machine. Furthermore, when controlling the switching between the stress control mode and the strain control mode, the switching command can be issued in advance, and the advance time can be set to the hardware response delay obtained from the test.
[0081] By adopting the above embodiments, it is possible to further ensure that appropriate test control modes are applied at different deformation stages of the packaging material under test, thereby further improving the accuracy and reliability of stress testing.
[0082] This disclosure also provides a stress testing system for integrated circuit packaging materials. (See also...) Figure 2 The diagram shown is a structural example of a stress testing system for integrated circuit packaging materials, consistent with some embodiments of this disclosure. In some embodiments, the stress testing system T may include:
[0083] Image sensor T1 is configured to acquire deformation images of a first sample of packaging material under test when a static load is applied to the sample at a set sampling rate; wherein the sampling rate is greater than a first threshold, and the deformation images include elastic deformation images and plastic deformation images of the sample.
[0084] The first processor T2 is coupled to the image sensor T1 and is configured to call a first neural network model to output the critical point of plastic deformation based on the deformation image.
[0085] In some embodiments of this disclosure, the first processor may be implemented using an integrated circuit. In some embodiments, the first processor may include a central processing unit (CPU), a microprocessor, or a field-programmable gate array (FPGA), etc.
[0086] The stress testing device T3, coupled to the first processor T2, is configured to determine a test control mode based on the critical point of the occurrence of plastic deformation, and to perform stress testing on the second sample of the packaging material to be tested, wherein the test control mode is different before and after the critical point.
[0087] Using the above embodiments, an image sensor acquires deformation images of a first sample of the packaging material under test when a static load is applied, at a set sampling rate. When the sampling rate is greater than a first threshold, the probability of acquiring deformation images during the transition from elastic deformation to plastic deformation of the sample can be increased. A first processor calls a first neural network model to output the critical point of plastic deformation based on the deformation images, where the deformation images include both elastic and plastic deformation images of the sample. Based on the critical point of plastic deformation, a stress testing device determines a test control mode and performs stress testing on a second sample of the packaging material under test, thereby improving the accuracy and reliability of the stress test.
[0088] In some embodiments, the stress testing system T may further include: a second processor (not shown), coupled to the image sensor T1, configured to calculate the strain field distribution in the deformation image and obtain a strain contour map of the deformation image.
[0089] This disclosure also provides a storage medium storing a computer program, which, when executed by a processor, performs the stress testing method for integrated circuit packaging materials described in any of the above embodiments. Specific steps can be found in the foregoing embodiments and will not be repeated here.
[0090] In some embodiments of this disclosure, the storage medium may be various suitable readable storage media such as optical discs, hard disk drives, and solid-state drives.
[0091] In this disclosure, unless otherwise expressly specified and limited, ordinal numbers, such as “first”, “second”, etc., are used only to distinguish and describe related objects, and should not be construed as indicating or implying the relative importance or order between related objects. Furthermore, ordinal numbers do not represent the quantity of related objects.
[0092] In the above embodiments, the descriptions of each embodiment have their own emphasis. Parts not described in detail or recorded in a certain embodiment can be referred to in the relevant descriptions of other embodiments. Furthermore, the above embodiments can be freely combined as needed. Although the embodiments of this disclosure have been disclosed above, this disclosure is not limited thereto. Any person skilled in the art can make various modifications and alterations without departing from the spirit and scope of this disclosure.
Claims
1. A method for stress testing of integrated circuit packaging materials, comprising: The method comprises: applying a static load to a first sample of a to-be-tested packaging material, and collecting deformation images of the sample at a set sampling rate, wherein the sampling rate is greater than a first threshold, and the deformation images comprise elastic deformation images and plastic deformation images of the sample; calling a first neural network model to output a critical point of plastic deformation occurrence based on the deformation images; determining a test control mode based on the critical point of plastic deformation occurrence, and performing a stress test on a second sample of the to-be-tested packaging material; wherein the test control mode is different before and after the critical point.
2. The method of claim 1, wherein, The method further comprises: before the critical point of plastic deformation occurrence, performing the stress test on the second sample of the to-be-tested packaging material using a stress control mode, wherein the stress control mode controls the load to perform the stress test on the second sample of the to-be-tested packaging material; after the critical point of plastic deformation occurrence, performing the stress test on the second sample of the to-be-tested packaging material using a strain control mode, wherein the strain control mode controls a strain rate to perform the stress test on the second sample of the to-be-tested packaging material.
3. The method of claim 2, wherein, The method further comprises: measuring a hardware response delay of a stress test device; controlling switching of the stress control mode and the strain control mode based on the hardware response delay.
4. The method of claim 1, wherein, The plastic deformation images comprise necking deformation images.
5. The method of claim 2, wherein, The method further comprises: calling the first neural network model to output the critical point of plastic deformation occurrence based on the deformation images and time stamps corresponding to the deformation images.
6. The method of claim 5, wherein, The time stamps comprise any one of: a relative time stamp; a physical time stamp.
7. The method of claim 1, wherein, Before calling the first neural network model to output the critical point of plastic deformation occurrence based on the deformation images, the method further comprises: calculating a strain field distribution in the deformation images to obtain a strain cloud map of the deformation images.
8. A stress testing system for integrated circuit package materials, comprising: The method comprises: an image sensor configured to collect deformation images of a first sample of a to-be-tested packaging material when a static load is applied to the sample at a set sampling rate, wherein the sampling rate is greater than a first threshold, and the deformation images comprise elastic deformation images and plastic deformation images of the sample; a first processor coupled to the image sensor and configured to call a first neural network model to output a critical point of plastic deformation occurrence based on the deformation images; a stress test device coupled to the first processor and configured to determine a test control mode based on the critical point of plastic deformation occurrence, and perform a stress test on a second sample of the to-be-tested packaging material, wherein the test control mode is different before and after the critical point.
9. The system of claim 8, wherein, The method further comprises: a second processor coupled to the image sensor and configured to calculate a strain field distribution in the deformation images to obtain a strain cloud map of the deformation images.
10. A storage medium having stored thereon a computer program, characterized in that The computer program, when run by a processor, performs the method steps of any of claims 1-7 for stress testing of an integrated circuit package material.