Automatic semiconductor scratch analysis system

Abstract

The present invention provides an automatic semiconductor scratch analysis system, characterized by comprising: a wafer defect map acquisition module, configured to acquire a defect map detected by a detection machine; a new coordinate system definition module, configured to redefine the coordinates of the defect map; a cluster analysis module, configured to analyze, on the basis of a DBSCAN clustering algorithm, whether a cluster is present in a wafer defect map, wherein if the cluster is present, it is determined that there is a scratch, and if the cluster is not present, it is determined that there is no scratch; a scratch defect analysis module, configured to analyze the wafer defect map in which the scratch is present; and a scratch result display module, configured to display a scratch defect analysis result in the wafer defect map. The automatic semiconductor scratch analysis system of the present invention can automatically analyze scratch defects present in a semiconductor defect map, thereby reducing the requirements on manual measurement, thus reducing errors and subjectivity caused by human factors.

Description

An automated semiconductor scratch analysis system Technical Field

[0001] This invention relates to the field of semiconductor quality inspection technology, and in particular to an automatic semiconductor scratch analysis system. Background Technology

[0002] In semiconductor manufacturing, defect analysis of wafer surfaces is a key step in ensuring product quality. The current practice is to manually measure the surface after scratches are found. The current practice of using manual measurement to analyze wafer surface scratches has the following disadvantages: (1) Subjectivity: Manual measurement is easily affected by personal experience and judgment, leading to inconsistencies and errors in measurement results. (2) Inefficiency: The manual measurement process is time-consuming, especially when a large number of wafers need to be inspected, which greatly reduces production efficiency. (3) Poor repeatability: Measurements by different engineers may differ, resulting in poor repeatability and making it difficult to ensure data consistency. (4) Difficulty in data integration: Since the defect coordinate systems generated under different equipment and process conditions may be different, it is very difficult to manually integrate these data into a unified database. (5) Poor automation: Due to reliance on manual operation, it is difficult to achieve automation and intelligence of the entire production process. (6) Imperfect feedback mechanism: The feedback of the results of manual measurement and analysis to the production process may be delayed, affecting the timeliness of problem solving. Summary of the Invention

[0003] In order to solve the above problems, the present invention aims to provide an automatic semiconductor scratch analysis system.

[0004] To achieve the above objectives, the present invention adopts the following technical solution:

[0005] This invention provides an automatic semiconductor scratch analysis system, characterized by comprising: a wafer defect map acquisition module for acquiring a defect map obtained by an inspection machine; a new coordinate system definition module for redefining the coordinates of the defect map; a cluster analysis module for analyzing whether clusters exist in the wafer defect map based on the DBSCAN clustering algorithm, identifying scratches if clusters exist and no scratches if no clusters exist; and a scratch defect analysis module for analyzing wafer defect maps with scratches.

[0006] The scratch result display module shows the scratch defect analysis results on the wafer defect map.

[0007] Furthermore, the automatic semiconductor scratch analysis system provided by the present invention may also have the following feature: in the coordinate system redefined by the new coordinate system definition module, the origin is determined by the transverse tangent and the longitudinal tangent of the wafer, and the origin is located on the longitudinal tangent below the intersection of the transverse tangent and the longitudinal tangent.

[0008] Furthermore, the automatic semiconductor scratch analysis system provided by this invention may also have the following feature: the analysis process of the cluster analysis module is as follows:

[0009] Step 2-1, define the two parameters eps and min_samples in the DBSCAN clustering algorithm:

[0010] eps is the neighborhood radius, used here to define the distance threshold between data points;

[0011] min_samples is used to define the minimum number of data points in the neighborhood of a core point;

[0012] Step 2-2: Traverse all data points. If a data point's eps neighborhood contains at least min_samples data points, then mark that point as a core point.

[0013] Steps 2-3: Starting from a core point, assign the core point and all points within its eps neighborhood to the same cluster; recursively, if a point in a cluster is a core point, add all points within its eps neighborhood to that cluster; repeat the above steps until all core points have been assigned to a cluster.

[0014] Steps 2-4: Mark boundary points and noise points: Points that are assigned to a cluster but are not core points are marked as boundary points, and points that do not belong to any cluster are marked as noise points.

[0015] Furthermore, the automatic semiconductor scratch analysis system provided by the present invention may also have the following feature: wherein the scratch defect analysis module includes calculating the distance from the point to the straight line and the angle of the scratch.

[0016] Furthermore, the automatic semiconductor scratch analysis system provided by this invention may also have the following feature: the calculation process for the distance from a point to a line is as follows:

[0017] Step 3-1, construct the formula for calculating the distance from a point to a line:

[0018] Assuming the line passes through points (x1, y1) and (x2, y2), the distance d from point (x0, y0) to the line is expressed by the following formula:

[0019] Step 3-2, construct the vector:

[0020] Constructing direction vectors It refers to the vector from point (x1, y1) to point (x0, y0):

[0021] Constructing point-line vectors It refers to the vector from point (x1, y1) to point (x2, y2):

[0022] Step 3-3, calculate the projection vector:

[0023] Point-line vector In direction vector Projection vector on The formula is expressed as follows:

[0024] in, It is the dot product of a point vector, a line vector, and a direction vector, calculated using the following formula:

[0025] It is the squared magnitude of the direction vector, calculated using the following formula:

[0026] (x2-x1) 2 +(y2-y1) 2

[0027] Then, the projection vector The calculation formula is as follows:

[0028] Steps 3-4: Calculate the distance from the point to the line.

[0029] The distance d from a point to a line is also known as the point-line vector. and projection vector The difference in modulus:

[0030] After calculation, the distance d from the point to the line is expressed as:

[0031] Where P x and P y These are the projection vectors. The x and y components.

[0032] Furthermore, the automatic semiconductor scratch analysis system provided by this invention may also have the following feature: the calculation process for the scratch angle is as follows:

[0033] Let all the points of the scratch be (x i ,y i ), where i = 1, 2, ..., n, and n is the total number of scratch points. The direction angle θ of the scratch is expressed by the following formula:

[0034] Where, Δy=max(y i )-min(y i) represents the span of the scratch in the y-direction.

[0035] Δx=max(x i )-min(x i () represents the span of the scratch in the x-direction.

[0036] The above directional angle θ is expressed in radians. To convert it to degrees, the formula is as follows:

[0037] The scratch angle, θ, is output by the scratch defect analysis module. degree .

[0038] Furthermore, the automatic semiconductor scratch analysis system provided by the present invention is characterized by further comprising: a machine information database module for storing machine information, including the machine Arm width; a machine information matching module for searching and matching corresponding machine information from the machine information database module after the scratch result display module displays the analysis results; and a matching result output module for outputting the matching results of the machine information matching module.

[0039] Furthermore, in the semiconductor scratch automatic analysis system provided by the present invention, it is characterized by further including: a machine information supplementation module, used by the user to input and supplement the machine information in the machine information database module.

[0040] Compared with the prior art, the present invention has the following beneficial effects:

[0041] 1) Unified coordinate system: The semiconductor scratch automatic analysis system of the present invention solves the problem of computational complexity and inconsistency caused by different coordinate systems in the prior art by introducing a new coordinate system with the lower left corner of the wafer tangent as the origin.

[0042] 2) Automated measurement: The method of automatically calculating the scratch width in the semiconductor scratch automatic analysis system of this invention reduces the need for manual measurement, thereby reducing errors and subjectivity caused by human factors.

[0043] 3) Improved analysis speed: The semiconductor scratch automatic analysis system of this invention improves the speed of defect analysis through automated analysis, meeting the needs of large-scale production in the semiconductor industry.

[0044] 4) Improved accuracy: The semiconductor scratch automatic analysis system of this invention uses mathematical models and automation methods to reduce measurement errors and improve the accuracy of analysis results.

[0045] 5) Reduce the influence of human factors: The automatic semiconductor scratch analysis system of this invention reduces the interference of subjective human factors during the analysis process and improves the reliability of the results.

[0046] 6) Improved data comparison: The automatic semiconductor scratch analysis system of this invention is still equipped with an organic platform information matching module to compare with the data in the database, quickly match possible problematic equipment, and improve the efficiency of problem diagnosis. Attached Figure Description

[0047] Figure 1 is a block diagram of the automatic semiconductor scratch analysis system in an embodiment of the present invention;

[0048] Figure 2 is a flowchart of the automatic semiconductor scratch analysis system in an embodiment of the present invention;

[0049] Figure 3 is a schematic diagram of the definition of the new coordinate system in an embodiment of the present invention;

[0050] Figure 4 is a schematic diagram showing the scratch defect analysis results in an embodiment of the present invention. Detailed Implementation

[0051] To make the technical means, creative features, objectives and effects of this invention easier to understand, the following embodiments, in conjunction with the accompanying drawings, will specifically illustrate the technical solution of this invention.

[0052] <Example>

[0053] Referring to Figure 1, this embodiment provides an automatic semiconductor scratch analysis system, which includes the following logical function modules for computer program execution: wafer defect map acquisition module 101, new coordinate system definition module 102, cluster analysis module 103, scratch defect analysis module 104, scratch result display module 105, machine information database module 106, machine information matching module 107, machine information output module 108, and machine information supplementation module 109.

[0054] The wafer defect map acquisition module 101 is used to acquire the defect map obtained by the inspection equipment. The defect map is a wafer defect map, which is output by the monitoring equipment on the production line through self-inspection, and the file format is klarf. The monitoring equipment is communicatively connected to the semiconductor scratch automatic analysis system of the embodiment, and the wafer defect map acquisition module 101 automatically acquires the defect map uploaded by the monitoring equipment.

[0055] The new coordinate system definition module 102 redefines the coordinates of the defect map.

[0056] The cluster analysis module 103 analyzes whether there are clusters in the wafer defect map based on the DBSCAN clustering algorithm. If clusters exist, they are identified as having scratches; if clusters do not exist, they are identified as not having scratches.

[0057] The scratch defect analysis module 104 analyzes the defect map of the wafer with scratches.

[0058] The scratch result display module 105 displays the scratch defect analysis results on the wafer defect map.

[0059] The machine information database module 106 stores machine information, including the machine Arm width. This information forms the Process Arm database.

[0060] The machine information matching module 107 is used to search for and match the corresponding machine information from the database.

[0061] The machine information output module 108 is used to output the matching results of the machine information matching module 107.

[0062] The machine information supplementation module 109 is used by users to input and supplement the machine information in the machine information database module 106.

[0063] Referring to Figure 2, the workflow of the semiconductor scratch automatic analysis system in this embodiment is as follows:

[0064] Step 1: Redefine the new coordinates of the defect map using the new coordinate system definition module 102.

[0065] Referring to Figure 3, in the redefined coordinate system, the origin is determined by the transverse and longitudinal tangents of the wafer. The origin (Die(0,0) in Figure 3) is located on the longitudinal tangent below the intersection of the transverse and longitudinal tangents. The position of the origin is preset by the user.

[0066] Step 2: The cluster analysis module 103 analyzes the wafer defect map for the presence of clusters based on the DBSCAN clustering algorithm. If a cluster is found, it is considered that there is a scratch; if no cluster is found, it is considered that no scratch is found.

[0067] The analysis process of the cluster analysis module is as follows:

[0068] Step 2-1, define the two parameters eps and min_samples in the DBSCAN clustering algorithm:

[0069] eps is the neighborhood radius, used here to define the distance threshold between data points;

[0070] min_samples is used to define the minimum number of data points in the neighborhood of a core point;

[0071] Step 2-2: Traverse all data points. If a data point's eps neighborhood contains at least min_samples data points, then mark that point as a core point.

[0072] Steps 2-3: Starting from a core point, assign that core point and all points within its eps neighborhood to the same cluster;

[0073] Recursively, if a point in a cluster is a core point, then all points within its eps neighborhood are also added to that cluster;

[0074] Repeat the above steps until all core points have been assigned to a cluster;

[0075] Steps 2-4: Mark boundary points and noise points: Points that are assigned to a cluster but are not core points are marked as boundary points, and points that do not belong to any cluster are marked as noise points.

[0076] Step 3: Analyze the scratched wafer defect map using the scratch defect analysis module 104.

[0077] The scratch defect analysis module calculates the distance from a point to a line and the angle of the scratch. The distance from a point to a line refers to the origin of the new coordinate system, i.e., Die(0,0) in Figure 3, and the line refers to the scratch, i.e., the straight line formed by cluster fitting.

[0078] (1) The calculation process for the distance from a point to a line is as follows:

[0079] Step 3-1, construct the formula for calculating the distance from a point to a line:

[0080] Suppose the line passes through points (x1, y1) and (x2, y2), and its slope m is:

[0081] The equation of the line is:

[0082] Transform the equation y-y1=m(x-x1) into its standard form: Ax+By+C=0.

[0083] (y2-y1)x-(x2-x1)y+[(x2-x1)y1-(y2-y1)x1]=0

[0084] Then we have:

[0085] A=y2-y1,B=-(x2-x1),C=(x2-x1)y1-(y2-y1)x1

[0086] Therefore, the distance from the origin (x0, y0) to the line Ax + By + C = 0 can be expressed by the formula:

[0087] Substituting A, B, and C into formula (5), we obtain the formula for the final distance d:

[0088] Step 3-2, construct the vector:

[0089] Constructing direction vectors It refers to the vector from point (x1, y1) to point (x0, y0):

[0090] Constructing point-line vectors It refers to the vector from point (x1, y1) to point (x2, y2):

[0091] Step 3-3, calculate the projection vector:

[0092] Projection vector This represents the point-to-line vector from the point (x0, y0) to the line. The projection along the straight line can be understood as its projection onto the direction vector. The component on the line, that is, the projection of the point (x0, y0) onto the line, represents the "component" of the point onto the line in that direction.

[0093] Point-line vector In direction vector Projection vector on The formula is expressed as follows:

[0094] in, It is the dot product of a point vector, a line vector, and a direction vector, calculated using the following formula:

[0095] It is the squared magnitude of the direction vector, since the magnitude of the direction vector is:

[0096] so, The calculation formula is as follows:

[0097] (x2-x1) 2 +(y2-y1) 2

[0098] Then, the projection vector The calculation formula is as follows:

[0099] Steps 3-4: Calculate the distance from the point to the line.

[0100] The distance d from a point to a line is also known as the point-line vector. and projection vector The difference in modulus:

[0101] Substituting the previously known parameters into the formula, and after calculation,

[0102] Therefore, the distance d from the point to the line can be obtained:

[0103] Where P x and P y These are the projection vectors. The x and y components.

[0104] (2) The calculation process for the angle of the scratch is as follows:

[0105] Steps 3-5: Let all points of the scratch be (x... i ,y i ), where i = 1, 2, ..., n, and n is the total number of scratch points. The direction angle θ of the scratch is expressed by the following formula:

[0106] Where, Δy=max(y i )-min(y i ) represents the span of the scratch in the y-direction.

[0107] Δx=max(x i )-min(x i () represents the span of the scratch in the x-direction.

[0108] The above directional angle θ is expressed in radians. To convert it to degrees, the formula is as follows:

[0109] The scratch angle, θ, is output by the scratch defect analysis module. degree .

[0110] The scratch result display module 105 displays the scratch defect analysis results on the wafer defect map. Referring to Figure 4, the figure shows the scratch result of a case study. The scratch is the distance from the point to the line, and the angle is the angle of the scratch.

[0111] Step 4: After the system provides the scratch results, the machine information matching module 107 searches the database for corresponding machine information, and the machine information output module 108 outputs the matching results. This step quickly matches potentially problematic equipment, improving the efficiency of problem diagnosis.

[0112] Step 5: When no matching result is found, the user can supplement the machine information in the machine information database module 106 by inputting the machine information supplementation module.

[0113] The above embodiments are merely preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. For those skilled in the art, the present invention can have various modifications and variations. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention are included within the scope of protection of the present invention.

Claims

1. An automatic semiconductor scratch analysis system, characterized in that, include: The wafer defect map acquisition module is used to acquire the defect map obtained by the inspection equipment; The new coordinate system definition module redefines the coordinates of the defect map; The cluster analysis module uses the DBSCAN clustering algorithm to analyze whether clusters exist in the wafer defect map. If clusters exist, it is considered that there are scratches; if no clusters exist, it is considered that no scratches have been found. The scratch defect analysis module analyzes the defect images of wafers with scratches. The scratch result display module shows the scratch defect analysis results on the wafer defect map.

2. The semiconductor scratch automatic analysis system as described in claim 1, characterized in that: in, In the coordinate system redefined by the new coordinate system definition module, the origin is determined by the transverse and longitudinal tangents of the wafer, and the origin is located on the longitudinal tangent below the intersection of the transverse and longitudinal tangents.

3. The automatic semiconductor scratch analysis system as described in claim 1, characterized in that: in, The analysis process of the cluster analysis module is as follows: Step 2-1, define the two parameters eps and min_samples in the DBSCAN clustering algorithm: eps is the neighborhood radius, used here to define the distance threshold between data points; min_samples is used to define the minimum number of data points in the neighborhood of a core point; Step 2-2: Traverse all data points. If a data point's eps neighborhood contains at least min_samples data points, then mark that point as a core point. Steps 2-3: Starting from a core point, assign that core point and all points within its eps neighborhood to the same cluster; Recursively, if a point in a cluster is a core point, then all points within its eps neighborhood are also added to that cluster; Repeat the above steps until all core points have been assigned to a cluster; Steps 2-4: Mark boundary points and noise points: Points that are assigned to a cluster but are not core points are marked as boundary points, and points that do not belong to any cluster are marked as noise points.

4. The semiconductor scratch automatic analysis system as described in claim 1, characterized in that: in, The scratch defect analysis module includes calculating the distance from the point to the line and the angle of the scratch.

5. The semiconductor scratch automatic analysis system as described in claim 4, characterized in that: in, The calculation process for the distance from a point to a line is as follows: Step 3-1, construct the formula for calculating the distance from a point to a line: Assuming the line passes through points (x1, y1) and (x2, y2), the distance d from point (x0, y0) to the line is expressed by the following formula: Step 3-2, construct the vector: Constructing direction vectors It refers to the vector from point (x1, y1) to point (x0, y0): Constructing point-line vectors It refers to the vector from point (x1, y1) to point (x2, y2): Step 3-3, calculate the projection vector: Point-line vector In direction vector Projection vector on The formula is expressed as follows: in, It is the dot product of a point vector, a line vector, and a direction vector, calculated using the following formula: It is the squared magnitude of the direction vector, calculated using the following formula: (x2-x1) 2 +(y2-y1) 2 Then, the projection vector The calculation formula is as follows: Steps 3-4: Calculate the distance from the point to the line. The distance d from a point to a line is also known as the point-line vector. and projection vector The difference in modulus: After calculation, the distance d from the point to the line is expressed as: Where P x and P y These are the projection vectors. The x and y components.

6. The semiconductor scratch automatic analysis system as described in claim 4, characterized in that: in, The calculation process for the angle of the scratch is as follows: Let all the points of the scratch be (x i ,y i ), where i = 1, 2, ..., n, and n is the total number of scratch points. The direction angle θ of the scratch is expressed by the following formula: Where, Δy=max(y i )-min(y i ) represents the span of the scratch in the y-direction. Δx=max(x i )-min(x i () represents the span of the scratch in the x-direction. The above directional angle θ is expressed in radians. To convert it to degrees, the formula is as follows: The scratch angle output by the scratch defect analysis module is θd. egree .

7. The semiconductor scratch automatic analysis system as described in claim 1, characterized in that, Also includes: The machine information database module stores machine information, including the machine Arm width. The machine information matching module searches for and matches the corresponding machine information from the machine information database module after the scratch result display module displays the analysis results. The matching result output module is used to output the matching results of the machine information matching module.

8. The automatic semiconductor scratch analysis system as described in claim 7, characterized in that, Also includes: The machine information supplementation module is used by users to input and supplement the machine information in the machine information database module.

Images