A method for detecting the thickness of a custom area silver paste

Abstract

The application provides a self-defined area silver paste thickness detection method, and relates to the technical field of semiconductor packaging manufacturing and detection. The method comprises the following steps: performing global scanning on a substrate by three-dimensional laser scanning to obtain three-dimensional point cloud information of the substrate, generating a corresponding substrate image based on the three-dimensional point cloud information, and setting a silver paste area to be detected on the substrate image. The silver paste area contains a plurality of detection single points. The self-defined selection can eliminate areas that do not need to be detected, thereby shortening the detection time and improving the detection efficiency. The self-defined area silver paste thickness detection method performs global scanning on a substrate by three-dimensional laser scanning to obtain three-dimensional point cloud information, and generates a corresponding substrate image based on the information. A self-defined silver paste area is set on the image, and areas that do not need to be detected are eliminated, thereby optimizing the detection process, achieving efficient and accurate silver paste thickness detection, and significantly improving the detection efficiency and shortening the detection time.

Description

Technical Field

[0001] This invention relates to the field of semiconductor packaging manufacturing and testing technology, specifically a method for detecting the thickness of silver paste in a custom area. Background Technology

[0002] In the semiconductor packaging field, sintering silver paste technology is a key process for achieving efficient electrical interconnection and thermal conductivity between chips and substrates. This technology uses high-temperature sintering to form a dense interconnected structure in the silver paste layer, the quality of which directly determines the performance and reliability of the final packaged device. The printing thickness of the silver paste is a core process parameter, and its uniformity has a decisive impact on the volatilization of the organic carrier and the densification behavior of silver particles during the sintering process. In current mass production environments, monitoring the thickness of silver paste mostly relies on manual sampling inspection. Operators use handheld contact thickness gauges or fixed laser thickness gauges to perform single-point measurements at a few pre-set fixed locations, and compare the average of the discrete measurement results with the process specification threshold to determine whether the printing quality of the entire sheet of silver paste is up to standard. Some production lines have introduced automated equipment, but they still follow the basic paradigm of pre-selecting a small number of points for measurement. This method is superior to purely manual operation in terms of efficiency and data consistency.

[0003] The most critical flaw in existing technologies lies in their "point-to-surface" measurement principle, which cannot accurately reproduce the overall and local three-dimensional morphology of the silver paste printing layer. Especially when the shape of the silver paste area is complex or has minute features, sparse discrete point measurements will miss thickness anomalies in key areas. It cannot effectively capture fatal defects such as uneven local thickness, steep edge drops, or internal depressions caused by fluctuations in the printing process. Undetected morphological defects will directly lead to insufficient organic volatilization or incomplete connection during subsequent sintering, ultimately resulting in deterioration or even failure of device interconnect reliability. Existing methods cannot meet the urgent need for comprehensive and refined control of silver paste printing quality. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a method for detecting the thickness of silver paste in a custom area. The technical problem this invention aims to solve is how to address the issues of low efficiency, poor accuracy, and inconsistent judgments in silver paste thickness detection through three-dimensional laser scanning and a custom detection area method.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for detecting the thickness of silver paste in a custom area, comprising: S1. The substrate is scanned in its entirety by three-dimensional laser scanning to obtain the three-dimensional point cloud information of the substrate. A corresponding substrate image is generated based on the three-dimensional point cloud information. A silver paste area to be detected is set on the substrate image. The silver paste area contains several detection points. The custom selection can remove areas that do not need to be detected to shorten the detection time and improve the detection efficiency. S2. Set the number of rows and columns of the detection points within the silver paste area. The number of detection points can be dynamically adjusted according to the detection accuracy requirements to more realistically reflect the actual shape of the silver paste area. S3. Based on the number of detection points and the area of ​​the silver paste region, calculate the spatial coordinate position of the detection point within the silver paste region, map the spatial coordinate position to the corresponding point in the three-dimensional point cloud information, and establish the correspondence between the detection point and the three-dimensional point cloud information; S4. Extract the height value of the detected single point based on the correspondence; S5. Perform statistical processing on the height value, compare the result of the statistical processing with the preset thickness judgment standard to generate a comparison dataset, and determine whether the thickness of the silver paste in the silver paste area is qualified based on the comparison dataset and output the thickness detection result.

[0006] Preferably, the shape of the silver paste area is a rectangle, a polygon, or any closed area defined by the user.

[0007] Preferably, the detection points adopt a uniform grid distribution method, and the spacing between adjacent detection points is equal in the row direction and column direction, so as to achieve uniform distribution of the detection points in the region.

[0008] Preferably, the spatial coordinate position is calculated using bilinear interpolation, and the formula for bilinear interpolation is as follows: in, For the first Line number The spatial coordinates of the detected single point in the 3D point cloud information are given in units of... , Indicates the row number index. Indicates column index, The spatial coordinates of the lower left corner of the silver paste region in the three-dimensional point cloud information are given in units of 1. , The spatial coordinates of the lower right corner of the silver paste region in the 3D point cloud information are given in units of 1. , The spatial coordinates of the upper left corner of the silver paste region in the three-dimensional point cloud information are given in units of 1. , The number of rows is dimensionless. It is a sequence number, dimensionless.

[0009] Preferably, the height value is calculated using a point-to-plane distance formula, which includes the plane equation of the spatial coordinate position and a preset reference plane. The point-to-plane distance formula is as follows: in, Let be the spatial coordinates of the i-th detected point in the 3D point cloud, in units of . , The direction cosine coefficient of the preset reference plane is dimensionless. The direction cosine coefficient of the preset reference plane is dimensionless. The direction cosine coefficient of the preset reference plane is dimensionless. The planar offset term of the preset reference plane, in units of , Let be the magnitude of the normal vector, which is dimensionless. The height value of the detected single point, in units of .

[0010] Preferably, the statistical processing includes calculating the maximum, minimum, and average values ​​of the height values.

[0011] Preferably, the preset thickness determination criteria include upper limit type, lower limit type, and interval type. The upper limit type stipulates that the thickness of the silver paste area cannot exceed the maximum thickness limit. When the maximum height value of the detection point is less than or equal to the maximum thickness limit, it is deemed qualified. The lower limit type stipulates that the thickness of the silver paste area cannot be lower than the minimum thickness limit. When the minimum height value of the detection point is greater than or equal to the minimum thickness limit, it is deemed qualified. The interval type stipulates that the thickness of the silver paste must be within the allowable thickness range. When the minimum height value of the detection point is not lower than the lower limit of the range and the maximum height value does not exceed the upper limit of the range, it is deemed qualified, or when the average height value of the detection point is within the thickness range, it is deemed qualified.

[0012] Preferably, the thickness detection results include thickness distribution statistics, pass / fail judgment conclusions, and a visual chart of the detection area, enabling rapid, accurate, and automated detection of the thickness of silver paste in a custom area.

[0013] This invention provides a method for detecting the thickness of silver paste in a custom area. It has the following beneficial effects: This custom-region silver paste thickness detection method uses 3D laser scanning to perform a full-area scan of the substrate, acquiring 3D point cloud information and generating a corresponding substrate image based on this information. A custom silver paste region is defined on this image, and the detection process is optimized by removing areas that do not need to be detected. This achieves efficient and accurate silver paste thickness detection, significantly improving detection efficiency and shortening detection time.

[0014] By employing a uniform grid distribution for single-point detection and bilinear interpolation to calculate spatial coordinates, the height values ​​within the silver paste region can be accurately extracted and compared with a preset thickness standard. This method not only improves detection accuracy but also comprehensively reflects the thickness distribution of the silver paste region through statistical analysis and result visualization, ensuring the accuracy and reliability of the detection results. Attached Figure Description

[0015] Figure 1 It is a flowchart of the overall method for realizing an invention; Figure 2 This is a schematic diagram illustrating the generation of the detection area and points for realizing the invention; Figure 3 It is a logical flowchart for data processing and result determination in implementing an invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] Example 1 like Figure 1-3 As shown, this embodiment of the invention provides a method for detecting the thickness of silver paste in a custom area, including: S1. Performing a full-area scan of the substrate using a three-dimensional laser scanner to obtain three-dimensional point cloud information of the substrate; generating a corresponding substrate image based on the three-dimensional point cloud information; setting a silver paste area to be detected on the substrate image; the silver paste area containing several detection points; and custom selection to remove areas that do not need to be detected to shorten the detection time and improve detection efficiency. The shape of the silver paste area can be rectangular, polygonal, or any closed region defined by the user.

[0018] Improved detection accuracy and efficiency: Comprehensive information about the substrate is obtained through 3D laser scanning, and single detection points with custom area selection and uniform grid distribution can more accurately reflect the actual morphology of the silver paste area, while reducing interference from unnecessary areas and shortening the detection time.

[0019] S2. Set the number of rows and columns of detection points within the silver paste area. The number of detection points can be dynamically adjusted according to the detection accuracy requirements to more accurately reflect the actual shape of the silver paste area. Detection points are distributed in a uniform grid pattern, with adjacent detection points having equal spacing in both row and column directions to achieve uniform distribution of detection points within the area.

[0020] Flexible region definition and adaptive adjustment: It supports defining detection regions of arbitrary shapes according to different needs and can dynamically adjust the number of detection points, which enhances the flexibility and adaptability of the method and is suitable for detection tasks of different sizes and complexities.

[0021] S3. Based on the number of detected points and the area of ​​the silver paste region, calculate the spatial coordinates of each detected point within the silver paste region. Map these spatial coordinates to the corresponding points in the 3D point cloud information, establishing a correspondence between the detected points and the 3D point cloud information. The spatial coordinates are calculated using bilinear interpolation. The formula for bilinear interpolation is: in, For the first Line number The spatial coordinates of a single detected point in the 3D point cloud information, in units of , Indicates the row number index. Indicates column index, The coordinates of the lower left corner of the silver paste region in the 3D point cloud information are given in units of 1. , The coordinates of the lower right corner of the silver paste region in the 3D point cloud information are given in units of 1. , The coordinates of the upper left corner of the silver paste region in the 3D point cloud information are given in units of 1. , The number of rows is dimensionless. It is a sequence number, dimensionless.

[0022] Precise position mapping based on 3D point clouds: By mapping the detected single point to 3D point cloud information through bilinear interpolation, the spatial coordinates of each detected single point can be obtained more accurately, ensuring the high accuracy of the detection results.

[0023] S4. Extract the height value of the detected single point based on the correspondence. The height value is calculated using the point-to-plane distance formula, which includes the spatial coordinate position and the plane equation of the preset reference plane. The point-to-plane distance formula is: in, Let be the spatial coordinates of the i-th detected point in the 3D point cloud, in units of . , The direction cosine coefficient of the preset reference plane is dimensionless. The direction cosine coefficient of the preset reference plane is dimensionless. The direction cosine coefficient of the preset reference plane is dimensionless. For the plane offset of the preset reference plane, the unit is , Let be the magnitude of the normal vector, which is dimensionless. To detect the height value of a single point, the unit is... .

[0024] Height value calculation and standard comparison: By calculating the distance from a point to a plane, the height value of each test point is accurately obtained. Combined with the statistical processing of the maximum, minimum and average values, the thickness variation of the silver paste layer can be fully reflected. By comparing with the preset standard, the pass rate can be judged, which improves the comprehensiveness and reliability of the test.

[0025] S5. Perform statistical processing on the height values, compare the results with preset thickness judgment standards to generate a comparison dataset, and determine whether the thickness of the silver paste in the area is qualified based on the comparison dataset, and output the thickness detection results. Statistical processing includes calculating the maximum, minimum, and average height values. Preset thickness judgment standards include upper limit type, lower limit type, and interval type. The upper limit type stipulates that the thickness of the silver paste area cannot exceed the maximum thickness limit. When the maximum height value of a single detection point is less than or equal to the maximum thickness limit, it is considered qualified. The lower limit type stipulates that the thickness of the silver paste area cannot be lower than the minimum thickness limit. When the minimum height value of a single detection point is greater than or equal to the minimum thickness limit, it is considered qualified. The interval type stipulates that the silver paste thickness must be within the allowable thickness range. When the minimum height value of a single detection point is not lower than the lower limit of the range and the maximum height value does not exceed the upper limit of the range, it is considered qualified, or when the average height value of a single detection point is within the thickness range, it is considered qualified. The thickness detection results include thickness distribution statistics, qualification judgment conclusions, and a visualization chart of the detection area, realizing fast, accurate, and automated detection of the silver paste thickness in a custom area.

[0026] Visualization and automated processing of results: The test results not only include thickness distribution statistics, but also provide qualification judgment and regional visualization charts, which facilitate intuitive analysis and quick decision-making, realizing automated, rapid detection and quality control of silver paste thickness.

[0027] This method accurately detects the thickness of silver paste through 3D laser scanning and custom area settings. By mapping uniformly distributed detection points to 3D point cloud data, the height value is calculated and statistically analyzed, and the thickness is judged to be acceptable based on preset standards. Visualization and automated processing of the results improve detection efficiency and the accuracy of quality control.

[0028] Example 2 This embodiment is based on a custom-defined region silver paste thickness detection method. It uses bilinear interpolation to accurately map the spatial coordinates of detection points within the silver paste region to 3D point cloud data, providing accurate spatial data for subsequent thickness detection. The specific implementation is as follows: 1. Data Preparation Given the 3D point cloud coordinates of the silver paste region, the spatial coordinates of the lower left, lower right, and upper left corners of this region are obtained through 3D scanning, as follows: The coordinates are the lower left corner of the silver paste area.

[0029] The coordinates are the lower right corner of the silver paste area.

[0030] The coordinates are the top left corner of the silver paste area.

[0031] The detection grid set within the silver paste area is as follows: and The location of each detection point needs to be calculated based on the grid.

[0032] 2. Bilinear interpolation method For each detection point The formula for calculation using bilinear interpolation is: in, For the first Line number The spatial coordinates of a single detected point in the 3D point cloud information, in units of , Indicates the row number index. Indicates column index, The coordinates of the lower left corner of the silver paste region in the 3D point cloud information are given in units of 1. , The coordinates of the lower right corner of the silver paste region in the 3D point cloud information are given in units of 1. , The coordinates of the upper left corner of the silver paste region in the 3D point cloud information are given in units of 1. , The number of rows is dimensionless. It is a sequence number, dimensionless.

[0033] 3. Example Calculation Suppose we calculate the detection point in the 3rd row and 4th column. Spatial coordinates, known: lower left corner of the silver paste area .

[0034] bottom right corner of the silver paste area .

[0035] upper left corner of the silver paste area The grid is 5 rows and 5 columns, that is... and .

[0036] Substitute the above data into the formula: .

[0037] .

[0038] Therefore, the detection point in the 3rd row and 4th column Spatial coordinates are .

[0039] 4. Coordinate Mapping and Result Verification After calculating the spatial coordinates of all detection points using bilinear interpolation, the detection points will be accurately mapped in the 3D point cloud.

[0040] The coordinates of each detection point will be used for subsequent height calculation and thickness determination.

[0041] This method accurately maps the spatial coordinates of each detection point within the silver paste area to 3D point cloud data, ensuring that the position of the detection point in the 3D point cloud matches the actual silver paste area. This process provides reliable spatial data support for subsequent silver paste thickness calculation and quality assessment, improving detection accuracy and automation, thereby achieving efficient and accurate silver paste thickness detection and ensuring that product quality meets set standards.

[0042] Example 3 This embodiment is based on a custom area silver paste thickness detection method. It utilizes three-dimensional laser scanning technology to automatically detect the silver paste thickness in specific areas of the substrate, ensuring that it meets preset quality standards. The specific implementation method is as follows: 1. Data Collection and Preparation A 3D laser scanner was used to scan the substrate to obtain complete point cloud data. The unit of the scanned data is micrometers, and the scanning resolution is 1 millimeter with a sampling interval of 1 millimeter.

[0043] A specific area on the substrate surface is selected as the silver paste detection area. Height measurements will be performed at 100 individual detection points within this area.

[0044] Assuming the silver paste detection area is 200mm×200mm, the coordinates of the lower left corner of the area are (100000, 100000) and the coordinates of the upper right corner are (120000, 120000), that is, the coordinate range is 200mm×200mm.

[0045] The area contains 100 detection points evenly distributed in 10 rows and 10 columns, with a spacing of 20 mm between each detection point.

[0046] Detect single-point coordinates: Spatial coordinates of each detection point The data originates from 3D point cloud data, and the specific coordinates of a single point are obtained through laser scanning.

[0047] The following is a partial list of data: Table 1: Test Data Table.

[0048] 2. Set the reference surface parameters Select the reference plane: The reference plane is set to be parallel to The reference plane is a plane that passes through the center point of the substrate. The normal vector of the reference plane is... Bias term .

[0049] Reference surface equation: The equation of the datum surface can be expressed as: Substitute the known parameters The equation of the datum surface is obtained as follows: This indicates that the reference plane is parallel to the substrate surface.

[0050] 3. Calculate the height value of a single detection point. The height value of each detection point is calculated using the following formula, and the formula for the distance from a point to the plane is: in, Let be the spatial coordinates of the i-th detected point in the 3D point cloud, in units of . , The direction cosine coefficient of the preset reference plane is dimensionless. The direction cosine coefficient of the preset reference plane is dimensionless. The direction cosine coefficient of the preset reference plane is dimensionless. For the plane offset of the preset reference plane, the unit is , Let be the magnitude of the normal vector, which is dimensionless. To detect the height value of a single point, the unit is... .

[0051] 4. Calculate the height value of a single detection point. Suppose we extract the coordinates of three single detection points from 3D point cloud data: For the first detection point The height value is: .

[0052] For the second detection point The height value is: .

[0053] For the third detection point The height value is: .

[0054] This process calculates the same height value for all 100 detection points.

[0055] 5. Height value statistics and qualification judgment Statistical processing: The height values ​​of all detected points were calculated, and the following statistics were obtained: Maximum value: The maximum height is 6000μm.

[0056] Minimum value: The minimum height is 5000μm.

[0057] Average value: The average height is 5500 μm.

[0058] Preset thickness judgment criteria: Thickness compliance is determined according to preset standards: Upper limit type: The maximum thickness cannot exceed 6000μm.

[0059] Lower limit type: The minimum thickness cannot be less than 4000μm.

[0060] For interval type: the thickness should be between 5000μm and 6000μm.

[0061] Pass / Fail Judgment: The maximum value of 6000μm meets the upper limit type standard.

[0062] The minimum value of 5000μm meets the lower limit standard.

[0063] The interval-type standard is met because the height values ​​of all detection points are within the range of 5000μm to 6000μm.

[0064] Therefore, the thickness of the silver paste in the test area meets the preset standard and is judged to be qualified.

[0065] Through the above steps, automated detection of the silver paste thickness is effectively achieved. Utilizing 3D point cloud data and pre-set thickness criteria, the system accurately calculates the height of each detection point and performs acceptance testing through statistical analysis. The detection results show that the silver paste thickness meets the set upper limit, lower limit, and interval standards, ensuring that the silver paste thickness in the detected area is qualified. This method improves detection efficiency and accuracy while providing a reliable quality control tool, suitable for thickness detection and quality assurance in the production process.

[0066] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for detecting the thickness of silver paste in a custom area, characterized in that, include: S1. The substrate is scanned in its entirety by three-dimensional laser scanning to obtain the three-dimensional point cloud information of the substrate. A corresponding substrate image is generated based on the three-dimensional point cloud information. A silver paste area to be detected is set on the substrate image. The silver paste area contains several detection points. S2. Set the number of rows and columns of the detection point within the silver paste area; S3. Based on the number of detection points and the area of ​​the silver paste region, calculate the spatial coordinate position of the detection point within the silver paste region, map the spatial coordinate position to the corresponding point in the three-dimensional point cloud information, and establish the correspondence between the detection point and the three-dimensional point cloud information; S4. Extract the height value of the detected single point based on the correspondence; S5. Perform statistical processing on the height value, compare the result of the statistical processing with the preset thickness judgment standard to generate a comparison dataset, and determine whether the thickness of the silver paste in the silver paste area is qualified based on the comparison dataset and output the thickness detection result.

2. The method for detecting the thickness of silver paste in a custom area according to claim 1, characterized in that: The shape of the silver paste area can be rectangular, polygonal, or any closed area defined by the user.

3. The method for detecting the thickness of silver paste in a custom area according to claim 1, characterized in that: The detection points are distributed in a uniform grid, and the spacing between adjacent detection points is equal in both the row and column directions.

4. The method for detecting the thickness of silver paste in a custom area according to claim 1, characterized in that: The spatial coordinates were calculated using bilinear interpolation, and the formula for bilinear interpolation is as follows: ， in, For the first Line number The spatial coordinates of the detected single point in the three-dimensional point cloud information are listed below. Indicates the row number index. Indicates column index, The spatial coordinates of the lower left corner of the silver paste region in the three-dimensional point cloud information are: The spatial coordinates of the lower right corner of the silver paste region in the three-dimensional point cloud information are: The spatial coordinates of the upper left corner of the silver paste region in the three-dimensional point cloud information are: For the number of rows, For column numbers.

5. The method for detecting the thickness of silver paste in a custom area according to claim 1, characterized in that: The height value is calculated using the point-to-plane distance formula, which includes the plane equation of the spatial coordinate position and a preset reference plane. The point-to-plane distance formula is as follows: ， in, Let be the spatial coordinates of the i-th detected point in the 3D point cloud, in units of . , The direction cosine coefficient of the preset reference plane, The direction cosine coefficient of the preset reference plane, The direction cosine coefficient of the preset reference plane, For the plane offset term of the preset reference plane, Let the normal vector be the magnitude. The height value of the detected single point.

6. The method for detecting the thickness of silver paste in a custom area according to claim 1, characterized in that: The statistical processing includes calculating the maximum, minimum, and average values ​​of the height.

7. The method for detecting the thickness of silver paste in a custom area according to claim 1, characterized in that: The preset thickness judgment criteria include upper limit type, lower limit type, and interval type. The upper limit type stipulates that the thickness of the silver paste area cannot exceed the maximum thickness limit. When the maximum height value of the detection point is less than or equal to the maximum thickness limit, it is judged as qualified. The lower limit type stipulates that the thickness of the silver paste area cannot be lower than the minimum thickness limit. When the minimum height value of the detection point is greater than or equal to the minimum thickness limit, it is judged as qualified. The interval type stipulates that the thickness of the silver paste must be within the allowable thickness range. When the minimum height value of the detection point is not lower than the lower limit of the range and the maximum height value does not exceed the upper limit of the range, it is judged as qualified, or when the average height value of the detection point is within the thickness range, it is judged as qualified.

8. The method for detecting the thickness of silver paste in a custom area according to claim 1, characterized in that: The thickness detection results include thickness distribution statistics, pass / fail judgment conclusions, and a visual chart of the detection area.

Images