Rapid defect detection method for chip packaging process
Through multi-angle image acquisition and deep learning algorithms, we can identify chip packaging defects and generate detailed reports, solving the problems of low detection efficiency and poor accuracy in the existing technology, achieving efficient defect detection and process optimization, and improving chip packaging quality and production efficiency.
Patent Information
- Application Number
- CN202411085719.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-08
- Publication Date
- 2024-11-26
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The prior art has low defect detection efficiency and poor accuracy in chip packaging processes, and cannot correct defects in time, affecting production efficiency and product quality.
Using industrial cameras, optical microscopes and X-ray imaging systems for multi-angle all-round shooting, combined with deep learning algorithms for image processing and defect feature extraction, identify and classify defects, generate detailed defect reports, and adjust and optimize packaging processes based on report results. .
It significantly improves the speed and accuracy of defect detection, reduces the dependence of manual inspection, shortens the production cycle, reduces the defect rate and rework rate, and improves the packaging quality and production efficiency.
Smart Images

Figure CN119027383A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of chip packaging, and in particular to a method for quickly detecting defects in chip packaging technology. Background Art
[0002] The chip packaging process plays a vital role in the modern electronic manufacturing process, and the packaging quality directly affects the performance and life of the chip. With the continuous increase in chip functions and the improvement of integration, the requirements for packaging technology are becoming increasingly stringent. In the traditional packaging process, the connection quality of the solder joints, the reliability of the packaging materials, and the accurate positioning of the chip are all important factors affecting the packaging quality. High-quality packaging can not only protect the chip from the external environment, but also ensure the stability of the electrical connection and reliable thermal management. However, due to the complexity of the packaging process, defects in the packaging process such as poor solder joints, cracks in the packaging materials, and chip displacement are still common. These defects may cause chip failure and affect the normal operation of the entire electronic equipment.
[0003] Existing technologies mainly rely on manual inspection or simple machine vision systems to identify and classify packaging defects, but these methods have many limitations. First, manual inspection is inefficient, easy to fatigue, and greatly affected by subjective factors, making it difficult to ensure the consistency and accuracy of inspection. Second, traditional machine vision systems have limited capabilities in processing complex images and identifying subtle defects, making it difficult to adapt to the needs of modern high-density integrated circuit packaging. In addition, existing defect detection methods usually cannot provide feedback functions, resulting in defects in the packaging process cannot be corrected in a timely manner, affecting production efficiency and product quality. Summary of the invention
[0004] Based on the above purpose, the present invention provides a method for quickly detecting chip packaging process defects.
[0005] A method for rapidly detecting defects in chip packaging process comprises the following steps:
[0006] S1, Image acquisition: Use industrial cameras, optical microscopes and X-ray imaging systems to take multi-angle and all-round photos of the packaged chips to obtain high-definition images of the chip surface and internal structure;
[0007] S2, image processing: processing the acquired images, including denoising, enhancement and correction;
[0008] S3, defect feature extraction and classification: The processed images are analyzed through deep learning algorithms, defect feature parameters are extracted and classified, and different types of defects are identified, including:
[0009] S31, data preparation: collect and annotate defect image data, and perform preprocessing;
[0010] S32, Model Design: Design a model architecture including convolutional layers, pooling layers, fully connected layers, and Softmax classification layers through deep learning algorithms;
[0011] S33, model training: dividing the image data into a training set and a validation set, and using the training set to perform model training;
[0012] S34, Model Validation: Use the validation set to evaluate model performance and adjust model architecture and parameters;
[0013] S35, defect detection and classification: input the processed image in S2 into the trained model to perform defect detection and classification;
[0014] S36, result output: output classification results and defect characteristic parameters;
[0015] S4, positioning and marking: using image processing technology to locate the classified defects and mark them on the image;
[0016] S5, defect assessment: Combine the defect type and location to assess the impact of defects on chip performance, generate defect reports, and provide a basis for optimizing chip packaging processes;
[0017] S6, Feedback and Optimization: Based on the defect assessment results, feedback is provided to the production line to adjust and optimize the packaging process to reduce the occurrence of defects.
[0018] Optionally, the S1 specifically includes:
[0019] S11, prepare equipment: use a rotating platform or multi-axis robotic arm to install a high-resolution industrial camera, optical microscope, and X-ray imaging system, and calibrate the camera position and angle;
[0020] S12, multi-angle shooting: fix the chip on a rotating platform or robotic arm, set a predetermined shooting angle and position, use an industrial camera to shoot chip surface images from multiple angles, use an optical microscope to obtain chip surface image details such as solder joints, cracks, etc., and use an X-ray imaging system to obtain chip internal structure images.
[0021] Optionally, the S2 specifically includes:
[0022] S21, image denoising: use median filtering to remove noise in the image and improve image clarity;
[0023] S22, image enhancement: histogram equalization is used to improve the contrast and details of the image, making defects more obvious;
[0024] S23, image correction: using perspective transformation to correct geometric distortion in the image, and using color correction to correct color distortion in the image;
[0025] S24, Image stitching and reconstruction: Image stitching technology is used to stitch images taken from different angles into a panoramic image to ensure the complete display of the chip surface and internal structure. A three-dimensional reconstruction algorithm is used to synthesize two-dimensional image data from multiple angles into a three-dimensional image to provide a comprehensive view of the chip structure.
[0026] Optionally, the S31 specifically includes:
[0027] S311, Dataset Construction: Collect a large amount of chip image data containing various defects, including poor solder joints, packaging material cracks and chip displacement, and annotate the chip images to mark the defect types and defect features;
[0028] S312, data preprocessing: normalize the image data, scale the pixel values to the range of [0, 1] or [-1, 1], and perform data enhancement (such as rotation, translation, flipping, etc.) to increase data diversity and prevent overfitting.
[0029] Optionally, the S32 specifically includes:
[0030] Convolutional layer: extract local features of the image through multiple convolutional layers;
[0031] Pooling layer: Reduce the dimension of the feature map through maximum pooling;
[0032] Fully connected layer: The extracted features are mapped to the classification label space through the fully connected layer;
[0033] Softmax classification layer: Use the Softmax function to map the output to the [0,1] interval, indicating the probability of each category.
[0034] Optionally, the S4 uses image processing technology to locate the classified defects, specifically including:
[0035] S41, edge detection: using the Canny edge detector to detect the edges of defective areas in the image;
[0036] S42, contour detection: forming a contour by connecting adjacent edge pixels to determine the shape and boundary of the defect;
[0037] S43, defect area calculation: determine the location and size of the defect by calculating the bounding box and the minimum enclosing rectangle.
[0038] Optionally, marking on the image in S4 specifically includes:
[0039] S44, color marking: Use different colors to mark different types of defects for easy distinction, for example: red for poor solder joints, green for packaging material cracks, and blue for chip displacement;
[0040] S45, Label marking: Use text labeling formula to mark text labels of defect types on images.
[0041] Optionally, the S5 specifically includes:
[0042] S51, define indicators for evaluating the impact of defects on chip performance, including defect area, defect density and defect location. The larger the defect area, the greater the impact on chip performance. The larger the defect density, the greater the impact on chip performance. The more critical the defect location (such as near the solder joint, chip edge), the greater the impact on chip performance.
[0043] S52. Different types of defects have different effects on chip performance. Bad solder joints may lead to unstable electrical connections, resulting in poor or disconnected signal transmission. Evaluate the defect area and number of solder joint locations, especially the locations of key solder joints. Cracks in packaging materials may cause uneven stress on the chip, increase the risk of chip damage, and affect heat dissipation performance. Evaluate the length, width, and location of the cracks, especially whether the cracks extend into the chip. Chip displacement may cause misalignment between the chip and the packaging substrate, affecting electrical connections and mechanical stability. Evaluate the distance and direction of the displacement, especially whether it exceeds the allowable tolerance range.
[0044] S53, generate a defect report, which should include the type, location and area of the defect.
[0045] Optionally, the S6 specifically includes:
[0046] S61, based on the defect assessment results, determine adjustment and optimization strategies, including;
[0047] S62, forming a specific implementation plan for the determined adjustment and optimization strategy, including process parameter adjustment, operation step optimization and equipment improvement;
[0048] S63, feed back the adjustment and optimization plan to the production line, and the implementation plan includes the following contents.
[0049] Optionally, the defect features in S311 include cold welds, welds that are too large or too small, weld voids, crack length, crack width, crack location, horizontal offset distance, and vertical offset distance.
[0050] Beneficial effects of the present invention:
[0051] The present invention, through detailed analysis of defect types and locations, formulates targeted process adjustment and optimization strategies, such as adjusting welding temperature and time, improving packaging materials, and optimizing chip positioning. These measures can effectively reduce the occurrence of defects such as poor solder joints, packaging material cracks, and chip displacement, and significantly improve the overall quality and reliability of chip packaging.
[0052] The present invention builds a model through a deep learning algorithm to identify different types of defects, greatly improving the detection speed and accuracy, reducing the reliance on manual detection, and thus shortening the production cycle. Using the defect assessment results, a detailed process optimization plan is implemented, such as improving welding and packaging equipment, optimizing operating steps, and real-time monitoring of key process parameters. These optimization measures not only reduce the defect rate and rework rate, reduce the scrap rate and production costs, but also improve the operating efficiency of the production line.
[0053] The present invention enhances the quality control capability of the packaging process by establishing a systematic defect classification, evaluation and feedback mechanism. Detailed defect reports and evaluation results provide a scientific basis for process optimization, so that every adjustment and optimization has a basis to follow. It not only improves the efficiency and quality of the current production line, but also lays a solid foundation for future technology upgrades and process innovations. BRIEF DESCRIPTION OF THE DRAWINGS
[0054] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0055] Figure 1 A schematic diagram of a method flow chart of an embodiment of the present invention;
[0056] Figure 2 Schematic diagram of S4 process according to an embodiment of the present invention. DETAILED DESCRIPTION
[0057] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. At the same time, it is explained here that in order to make the embodiments more detailed, the following embodiments are the best and preferred embodiments, and those skilled in the art may also adopt other alternatives to implement some known technologies; and the accompanying drawings are only for more specific description of the embodiments, and are not intended to specifically limit the present invention.
[0058] It should be noted that the references to "one embodiment", "an embodiment", "an exemplary embodiment", "some embodiments" and the like in the specification indicate that the embodiments described may include specific features, structures or characteristics, but not every embodiment may include the specific features, structures or characteristics. In addition, when a specific feature, structure or characteristic is described in conjunction with an embodiment, it should be within the knowledge of a person skilled in the art to implement such feature, structure or characteristic in conjunction with other embodiments (whether or not explicitly described).
[0059] In general, a term can be understood, at least in part, from its use in context. For example, depending, at least in part, on the context, the term "one or more" as used herein can be used to describe any feature, structure, or characteristic in the singular sense, or can be used to describe a combination of features, structures, or characteristics in the plural sense. Additionally, the term "based on" can be understood as not necessarily intended to convey an exclusive set of factors, but can instead, depending, at least in part, on the context, allow for the presence of other factors that are not necessarily explicitly described.
[0060] like Figure 1-Figure 2 As shown, the chip packaging process defect rapid detection method comprises the following steps:
[0061] S1, Image acquisition: Use industrial cameras, optical microscopes and X-ray imaging systems to take multi-angle and all-round photos of the packaged chips to obtain high-definition images of the chip surface and internal structure;
[0062] S2, image processing: processing the acquired images, including denoising, enhancement and correction;
[0063] S3, defect feature extraction and classification: The processed images are analyzed through deep learning algorithms, defect feature parameters are extracted and classified, and different types of defects are identified, including:
[0064] S31, data preparation: collect and annotate defect image data, and perform preprocessing;
[0065] S32, Model Design: Design a model architecture including convolutional layers, pooling layers, fully connected layers, and Softmax classification layers through deep learning algorithms;
[0066] S33, model training: dividing the image data into a training set and a validation set, and using the training set to perform model training;
[0067] The cross entropy loss function is used as the loss function for the classification task during model training;
[0068]
[0069] Among them, y i is the true label, is the predicted probability;
[0070] S34, Model Validation: Use the validation set to evaluate model performance and adjust model architecture and parameters;
[0071] S35, defect detection and classification: input the processed image in S2 into the trained model to perform defect detection and classification;
[0072] S36, result output: output classification results and defect characteristic parameters;
[0073] S4, positioning and marking: using image processing technology to locate the classified defects and mark them on the image;
[0074] S5, defect assessment: Combine the defect type and location to assess the impact of defects on chip performance, generate defect reports, and provide a basis for optimizing chip packaging processes;
[0075] S6, Feedback and Optimization: Based on the defect assessment results, feedback is provided to the production line to adjust and optimize the packaging process to reduce the occurrence of defects.
[0076] S1 specifically includes:
[0077] S11, prepare equipment: use a rotating platform or multi-axis robotic arm to install a high-resolution industrial camera, optical microscope, and X-ray imaging system, and calibrate the camera position and angle;
[0078] S12, multi-angle shooting: fix the chip on a rotating platform or robotic arm, set a predetermined shooting angle and position, use an industrial camera to shoot chip surface images from multiple angles, use an optical microscope to obtain chip surface image details such as solder joints, cracks, etc., and use an X-ray imaging system to obtain chip internal structure images.
[0079] S2 specifically includes:
[0080] S21, image denoising: use median filtering to remove noise in the image and improve image clarity;
[0081] Median filtering replaces the central pixel value by calculating the median of the pixel values in the pixel neighborhood, expressed as:
[0082] Let I(x,y) be the pixel value of the image at position (x,y), I med (x, y) is the pixel value after filtering, then the calculation is:
[0083] I med (x,y)=median{I(x+i,y+j)},(i,j)∈W;
[0084] Where W is the neighborhood window, which represents a group of pixels around the central pixel;
[0085] S22, image enhancement: histogram equalization is used to improve the contrast and details of the image, making defects more obvious;
[0086] Histogram equalization is expressed as: the gray value of the input image is r k , the gray value after equalization is s k , then the calculation is:
[0087]
[0088] Where L is the number of gray levels, usually 256, p r (r j ) is the gray value r j The probability density function of is the normalized value of the grayscale histogram;
[0089] S23, image correction: using perspective transformation to correct geometric distortion in the image, and using color correction to correct color distortion in the image;
[0090] Perspective transformation: Let the perspective transformation matrix be H, the input image coordinates be (x, y), and the transformed coordinates be (x′, y′), expressed as:
[0091]
[0092] Among them, H is a 3×3 matrix, which is calculated by the coordinates of four sets of corresponding points;
[0093] Color correction: Assume that the image is in LAB color space, L is brightness, A and B are color components, and the A and B components after correction are expressed as:
[0094]
[0095] in, and are the average values of the A and B components in the image, respectively, and 128 is the center value of the LAB color space;
[0096] S24, Image stitching and reconstruction: Image stitching technology is used to stitch images taken from different angles into a panoramic image to ensure the complete display of the chip surface and internal structure. A three-dimensional reconstruction algorithm is used to synthesize two-dimensional image data from multiple angles into a three-dimensional image to provide a comprehensive view of the chip structure.
[0097] S31 specifically includes:
[0098] S311, Dataset Construction: Collect a large amount of chip image data containing various defects, including poor solder joints, packaging material cracks and chip displacement, and annotate the chip images to mark the defect types and defect features;
[0099] S312, data preprocessing: normalize the image data, scale the pixel values to the range of [0, 1] or [-1, 1], and perform data enhancement (such as rotation, translation, flipping, etc.) to increase data diversity and prevent overfitting.
[0100] S32 specifically includes:
[0101] Convolutional layer: extract local features of the image through multiple convolutional layers. The convolution operation can detect edges, corners and other detailed features in the image;
[0102] Convolution operation formula: F i,j,k =∑ m,n I i+m,j+n ·K m,n,k ;
[0103] Among them, F i,j,k is the kth channel of the convolutional layer output feature map, I is the input image, and K is the convolution kernel;
[0104] Pooling layer: Reduce the dimension of the feature map through maximum pooling, reduce the amount of calculation and enhance the robustness of the features;
[0105] Max pooling formula: P i,j,k =max m,n F i+m,j+n,k ;
[0106] Among them, P i,j,k Output the kth channel of the feature map of the pooling layer;
[0107] Fully connected layer: The extracted features are mapped to the classification label space through the fully connected layer, and the fully connected layer converts the high-dimensional feature vector into classification probability;
[0108] Fully connected layer formula: y = Softmax(W x + b);
[0109] Among them, W is the weight matrix, x is the input feature vector, b is the bias, and y is the classification probability;
[0110] Softmax classification layer: Use the Softmax function to map the output to the [0,1] interval, indicating the probability of each category;
[0111] Softmax function formula:
[0112] Among them, z i is the unnormalized probability value of the i-th category.
[0113] S4 uses image processing technology to locate the classified defects, including:
[0114] After S3 defect classification, each pixel or area of the image has been labeled as a specific type of defect (such as poor solder joints, packaging material cracks, and chip displacement);
[0115] S41, edge detection: using the Canny edge detector to detect the edges of defective areas in the image;
[0116] The Canny edge detector can effectively detect edges in images, expressed as:
[0117] edges(x,y)=Canny(I(x,y),threshold1,threshold2);
[0118] Among them, I(x,y) is the input image, threshold1 and threshold2 are high and low thresholds;
[0119] S42, contour detection: forming a contour by connecting adjacent edge pixels to determine the shape and boundary of the defect, the steps are as follows:
[0120] (1) The edges in the image are detected by the Canny edge detector. The binary image after edge detection is represented as E(x, y), where E(x, y) is the pixel value of the binary edge image at position (x, y);
[0121] (2) Contour detection is to determine the connected area of the edge based on the value of the binary edge image E(x,y), including:
[0122] Contour definition: A contour is a series of edge pixels connected together, represented by C i , where i is the index of the contour, and each contour C i A set of pixels (x j ,y j );
[0123] Composition, expressed as: C i ={(x j ,y j )|E(x j ,y j )=1}, where j is the index of the pixel in the contour;
[0124] Contour set: All contours in the entire image form a set, expressed as: Where n is the number of wheels detected;
[0125] S43, defect area calculation: determine the position and size of the defect by calculating the bounding box and the minimum enclosing rectangle, the specific steps are as follows;
[0126] Contour feature calculation: For each detected contour C i , calculate its bounding box and minimum enclosing rectangle features, where;
[0127] Bounding box: The bounding box is the smallest rectangle that contains the outline and is defined as:
[0128] B i =(x min ,ymin ,x max ,y max ),in:
[0129]
[0130] Minimum Enclosing Rectangle: The minimum enclosing rectangle is the rectangle with the smallest area that contains the outline, and is defined as:
[0131] R i ={(x,y)|(x,y)∈R 2 And (x,y) is a pair containing C i The smallest rectangle of}.
[0132] The marking on the image in S4 specifically includes:
[0133] S44, color marking: Use different colors to mark different types of defects for easy distinction, for example: red for poor solder joints, green for packaging material cracks, and blue for chip displacement;
[0134] S45, label marking: Use the text labeling formula to mark the text label of the defect type on the image. The text labeling formula is expressed as:
[0135] putText(I(x,y),label,(x,y),font,scale,color,thickness);
[0136] Where label is the label of the defect type, (x, y) is the position of the text, font is the font type, scale is the font size, color is the text color, and thickness is the font thickness;
[0137] Through the above steps, image processing technology can be effectively used to locate the classified defects and mark them on the image to facilitate subsequent defect evaluation.
[0138] S5 specifically includes:
[0139] S51, define indicators for evaluating the impact of defects on chip performance, including defect area, defect density and defect location. The larger the defect area, the greater the impact on chip performance. The larger the defect density, the greater the impact on chip performance. The more critical the defect location (such as near the solder joint, chip edge), the greater the impact on chip performance.
[0140] The defect area is expressed as: A = (x max -x min )×(y max -y min );
[0141] Among them, (xmin ,y min ) and (x max ,y max ) are the bounding box coordinates of the defect area;
[0142] S52. Different types of defects have different effects on chip performance. Bad solder joints may lead to unstable electrical connections, resulting in poor or disconnected signal transmission. Evaluate the defect area and number of solder joint locations, especially the locations of key solder joints. Cracks in packaging materials may cause uneven stress on the chip, increase the risk of chip damage, and affect heat dissipation performance. Evaluate the length, width, and location of the cracks, especially whether the cracks extend into the chip. Chip displacement may cause misalignment between the chip and the packaging substrate, affecting electrical connections and mechanical stability. Evaluate the distance and direction of the displacement, especially whether it exceeds the allowable tolerance range.
[0143] S53, generate a defect report, which shall include the type, location and area of the defect;
[0144] Through the above content, a detailed defect report is generated to evaluate the impact of defects on chip performance and provide a scientific basis for the optimization of chip packaging process.
[0145] S6 specifically includes:
[0146] S61, based on the defect assessment results, determine the adjustment and optimization strategies, including:
[0147] (1) Optimization of poor solder joints:
[0148] Adjust welding temperature: Make sure the welding temperature is within the appropriate range, avoid being too high or too low;
[0149] Optimize welding time: adjust welding time to ensure that the solder joint is fully melted and cooled;
[0150] Improve welding equipment: Use higher precision welding equipment to reduce the variation of weld spot size;
[0151] (2) Optimization of packaging material cracks:
[0152] Improve material formulation: use more flexible packaging materials to reduce material stress;
[0153] Optimize packaging process: adjust packaging pressure and time to avoid excessive stress on the material;
[0154] Enhanced quality inspection: Add crack detection steps during the packaging process to detect and repair cracks in a timely manner;
[0155] (3) Optimization of chip displacement:
[0156] Improve positioning accuracy: Use high-precision positioning system to ensure accurate chip placement;
[0157] Optimize the installation process: improve the chip installation sequence and method to reduce the risk of displacement;
[0158] Strengthen process control: monitor chip position in real time and make timely adjustments if deviation is found;
[0159] S62, forming a specific implementation plan for the determined adjustment and optimization strategy, including process parameter adjustment, operation step optimization and equipment improvement;
[0160] S63, feedback the adjustment and optimization plan to the production line. The implementation plan includes the following:
[0161] (1) Process parameter adjustment:
[0162] Welding temperature: According to the defect analysis results, adjust the welding temperature control system and set the appropriate temperature range;
[0163] Welding time: Optimize the welding time control program to ensure the accurate welding time of each welding point;
[0164] Packaging pressure and time: adjust the pressure and time parameters of the packaging machine to ensure uniform force during the packaging process;
[0165] (2) Operation steps optimization:
[0166] Welding steps: Improve the welding operation process to ensure that each welding point is evenly heated;
[0167] Packaging steps: Optimize packaging operation steps to reduce material stress concentration;
[0168] Chip installation steps: Improve the chip installation process to ensure accurate chip positioning;
[0169] (3) Equipment improvement:
[0170] Welding equipment: Use higher precision welding equipment to improve the consistency of solder joints;
[0171] Packaging equipment: Improve the pressure control system of packaging equipment to reduce material cracks;
[0172] Positioning system: Use high-precision positioning system to ensure accurate placement of the chip;
[0173] By analyzing the defect assessment results and formulating targeted adjustment and optimization plans, the occurrence of defects can be effectively reduced and the quality and production efficiency of the chip packaging process can be improved.
[0174] The defect features in S311 include cold solder joints, solder joints that are too large or too small, solder joint voids, crack length, crack width, crack location, horizontal offset distance, and vertical offset distance.
[0175] The present invention covers any substitution, modification, equivalent method and scheme made on the essence and scope of the present invention. In order to make the public have a thorough understanding of the present invention, specific details are described in detail in the following preferred embodiments of the present invention, but those skilled in the art can fully understand the present invention without the description of these details. In addition, in order to avoid unnecessary confusion about the essence of the present invention, well-known methods, processes, procedures, components and circuits are not described in detail.
[0176] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.
Claims
1. A rapid detection method for chip packaging process defects, characterized in that: The following steps are involved: S1, image acquisition: Use industrial cameras, optical microscopes and X-ray imaging systems to take multi-angle and all-round photos of the packaged chips to obtain images of the chip surface and internal structure; S2, image processing: processing the acquired images, including denoising, enhancement and correction; S3, defect feature extraction and classification; The processed images are analyzed through deep learning algorithms to extract defect feature parameters and classify them to identify different types of defects, including: S31, data preparation: collect and annotate defect image data, and perform preprocessing; S32, Model Design: Design a model architecture including convolutional layers, pooling layers, fully connected layers, and Softmax classification layers through deep learning algorithms; S33, model training: dividing the image data into a training set and a validation set, and using the training set to perform model training; S34, Model Validation: Use the validation set to evaluate model performance and adjust model architecture and parameters; S35, defect detection and classification: input the processed image in S2 into the trained model to perform defect detection and classification; S36, result output: output classification results and defect characteristic parameters; S4, positioning and marking: using image processing technology to locate the classified defects and mark them on the image; S5, defect assessment: Combine defect type and location to assess the impact of defects on chip performance and generate a defect report; S6, Feedback and Optimization: Based on the defect assessment results, feedback is provided to the production line to adjust and optimize the packaging process to reduce the occurrence of defects.
2. The chip packaging process defect rapid detection method according to claim 1, characterized in that: The S1 specifically includes: S11, prepare equipment: use a rotating platform or multi-axis robotic arm to install industrial cameras, optical microscopes, and X-ray imaging systems, and calibrate the camera positions and angles; S12, multi-angle shooting: fix the chip on a rotating platform or robotic arm, set a predetermined shooting angle and position, use an industrial camera to shoot chip surface images from multiple angles, use an optical microscope to obtain the details of the chip surface image, and use an X-ray imaging system to obtain the chip internal structure image.
3. The method for rapid detection of chip packaging process defects according to claim 1, characterized in that: The S2 specifically includes: S21, image denoising: median filtering is used to remove noise in the image and improve the clarity of the image; S22, image enhancement: histogram equalization is used to improve the contrast and details of the image, making defects more obvious; S23, image correction: using perspective transformation to correct geometric distortion in the image, and using color correction to correct color distortion in the image; S24, Image stitching and reconstruction: Image stitching technology is used to stitch images taken from different angles into a panoramic image, and a 3D reconstruction algorithm is used to synthesize 2D image data from multiple angles into a 3D image, providing a comprehensive view of the chip structure.
4. The method for rapid detection of chip packaging process defects according to claim 1, characterized in that: The S31 specifically includes: S311, data set construction: collect chip image data containing various defects, including poor solder joints, packaging material cracks and chip displacement, and annotate the chip images to mark the defect types and defect features; S312, data preprocessing: normalize the image data, scale the pixel values to the range of [0, 1] or [-1, 1], and perform data enhancement to increase the diversity of the data through rotation, translation, and flipping.
5. The chip packaging process defect rapid detection method according to claim 1, characterized in that: The S32 specifically includes: Convolutional layer: extract local features of the image through multiple convolutional layers; Pooling layer: Reduce the dimension of the feature map through maximum pooling; Fully connected layer: The extracted features are mapped to the classification label space through the fully connected layer; Softmax classification layer: Use the Softmax function to map the output to the [0,1] interval, indicating the probability of each category.
6. The method for rapid detection of chip packaging process defects according to claim 1, characterized in that: In S4, the classified defects are located by using image processing technology, which specifically includes: S41, edge detection: using the Canny edge detector to detect the edges of defective areas in the image; S42, contour detection: forming a contour by connecting adjacent edge pixels to determine the shape and boundary of the defect; S43, defect area calculation: determine the location and size of the defect by calculating the bounding box and the minimum enclosing rectangle.
7. The method for rapid detection of chip packaging process defects according to claim 6, characterized in that: The marking on the image in S4 specifically includes: S44, color marking: Use different colors to mark different types of defects for easy distinction; S45, Label marking: Use text labeling formula to mark text labels of defect types on images.
8. The method for rapid detection of chip packaging process defects according to claim 7, characterized in that: The S5 specifically includes: S51, define indicators for evaluating the impact of defects on chip performance, including defect area, defect density and defect location. The larger the defect area, the greater the impact on chip performance. The larger the defect density, the greater the impact on chip performance. The more critical the defect location, the greater the impact on chip performance. S52, different types of defects have different effects on chip performance. Bad solder joints lead to unstable electrical connections, resulting in poor or disconnected signal transmission. The defect area and number of solder joints should be evaluated. Cracks in packaging materials lead to uneven stress on the chip, increasing the risk of chip damage and affecting heat dissipation performance. The length, width and position of the cracks should be evaluated. Chip displacement leads to misalignment between the chip and the packaging substrate, affecting electrical connections and mechanical stability. The distance and direction of displacement should be evaluated. S53, generate a defect report, which should include the type, location and area of the defect.
9. The method for rapid detection of chip packaging process defects according to claim 1, characterized in that: The S6 specifically includes: S61, determine adjustment and optimization strategies based on defect assessment results; S62, forming a specific implementation plan for the determined adjustment and optimization strategy, including process parameter adjustment, operation step optimization and equipment improvement; S63, feeds back the adjustment and optimization plan to the production line.
10. The method for rapid detection of chip packaging process defects according to claim 4, characterized in that: The defect features in S311 include cold solder joints, solder joints that are too large or too small, solder joint voids, crack length, crack width, crack location, horizontal offset distance, and vertical offset distance.
Citation Information
Patent Citations
Semiconductor product defect detection method and device, electronic equipment and storage medium
CN112017986A
Silicon wafer surface defect detection method based on deep convolutional neural network
CN115409824A
Chip appearance defect automatic detection method, electronic equipment and storage medium
CN117011260A
Multi-chip packaging defect detection method and system
CN118195422A
Chip packaging defect detection method and system based on machine vision
CN118392891A
Cited By
Chip defect detection and optimization system based on knowledge graph reasoning
CN119671776A
Chip defect detection and optimization system based on knowledge graph reasoning
CN119671776B
Medicine logistics management device and method based on machine vision
CN119671981A
Ultrathin multi-layer memory chip stacking and packaging method and system
CN119742237A
Automatic packaging test method and system for micro-fluidic chip
CN119803301A
