A semiconductor workpiece manufacturing detection system and method

Abstract

The present invention relates to the field of semiconductors, and discloses a detection system and method for semiconductor workpiece production, comprising: an image acquisition module, which is used to be deployed on a production line to obtain image data before, during and after raw material preparation, establish corresponding standard image reference systems according to different products, and compare the currently acquired image data with the data in the standard image reference system in real time to determine whether there is an abnormality; a working condition acquisition module, which is used to obtain the operating parameters of each workpiece production equipment, and regularly receive and convert the equipment working condition parameters according to a preset acquisition cycle to determine whether there is an abnormality; through real-time image detection and equipment detection, after one party has detected a problem, the other party is quickly called for verification to improve the accuracy of the detection and prevent misjudgment of abnormalities. When the same problem occurs, it can be indexed in the database and quickly associated with a suitable solution, thereby alleviating computing pressure.

Description

Technical Field

[0001] The present invention relates to the field of semiconductor technology, and in particular to a detection system and method for manufacturing semiconductor workpieces. Background Art

[0002] In the production of semiconductor devices, from semiconductor single wafers to the final product, in order to ensure that the product performance is qualified, stable and reliable, and has a high yield rate, all process steps must have strict and specific requirements according to the production conditions of various products. Therefore, corresponding systems and precise monitoring measures must be established in the production process, starting with semiconductor process detection;

[0003] However, the existing inspection systems and methods for semiconductor workpiece manufacturing still have some shortcomings, such as:

[0004] 1. During the detection process, the means are often relatively simple, and judgment is made only through simple image detection, which can easily cause large errors, making the accuracy of the detection results low, difficult to verify by multiple parties, and difficult to respond in time when problems arise. Frequent occurrence of similar problems often wastes a lot of computing power;

[0005] 2. The detection method is limited to the production line, and it is difficult to conduct secondary tracking of the product quality outside the production line. When problematic products appear, it is difficult to trace the fault and it is difficult to help the system complete self-inspection of the problem, so the functionality is low. Summary of the invention

[0006] 1. Technical issues to be resolved

[0007] In view of the above-mentioned shortcomings of the prior art, the present invention provides a semiconductor workpiece manufacturing detection system and method, which can effectively solve the problems of the prior art.

[0008] (II) Technical solution

[0009] To achieve the above objectives, the present invention is implemented through the following technical solutions:

[0010] The present invention discloses a semiconductor workpiece manufacturing detection system, comprising:

[0011] The image acquisition module is used to be deployed on the production line to obtain image data before, during and after raw material preparation. A corresponding standard image reference system is established according to different products. The current image data is compared with the data in the standard image reference system in real time to determine whether there is any abnormality.

[0012] The working condition acquisition module is used to obtain the operating parameters of each workpiece manufacturing equipment. According to the preset acquisition cycle, it regularly receives and converts the equipment working condition parameters to determine whether there are any abnormalities;

[0013] The attribute determination module is used to obtain the abnormal parameters pointed to by the abnormal judgment, classify them accordingly, issue classification labels, and upload them to the database;

[0014] The problem analysis module is used to analyze based on abnormal parameters, output the detection report to the management end, automatically analyze the solution, receive and upload the manual intervention solution, and record and confirm the solution data that effectively solves the abnormal parameters, extract key features, and upload them to the database;

[0015] The same type index module is used to obtain the classification label pointed to by the current abnormal parameter when receiving the abnormal parameter, extract the key features of the data, and index through its key features within the classification label to match the valid solution data pointed to by the same type of abnormal parameter;

[0016] The summary evaluation module is used to obtain the effective solution data captured by all similar index modules, evaluate based on the original data of the current abnormal parameters, extract and display the solution data with the highest effectiveness;

[0017] The sharing module is used to provide a shared solution data platform. After passing the identity authentication, the abnormal parameters are input to provide query interaction services;

[0018] The network distribution module is used to provide IoT support and provide network permissions for each network-using module.

[0019] Furthermore, the image acquisition module is interactively connected to the association module via a wireless network, and the association module is interactively connected to the working condition acquisition module via a wireless network. The association module serves as a jump end. When the image acquisition module and the working condition acquisition module each obtain abnormal data, the data of the other party's associated abnormal data area is called and re-acquired as a reference for auxiliary judgment.

[0020] Furthermore, the image acquisition module merges the areas where pixels are connected into a whole through connected domain extraction, obtains the key areas in the image, extracts the connected domain of the workpiece image after threshold processing, marks the areas where pixels are not connected into different connected domains, extracts and connects the edges of the workpiece, sets the width threshold range, and extracts the contours that meet the range.

[0021] Furthermore, the image data obtained by the attribute determination module is processed by distortion correction to express the imaging model, and the calculation formula is:

[0022]

[0023] Where (u, v) is the ideal imaging two-dimensional pixel coordinate value, (u d ,v d) is the actual imaging two-dimensional pixel coordinate value, K is the radial distortion coefficient, when K>0, it is positive distortion, the object image is concave inward; when K<0, it is negative distortion, the object image is convex outward.

[0024] Furthermore, after the attribute determination module classifies the data, it needs to confirm whether there is feedback data from the same type index module. If there is no feedback data, the classified data is packaged and submitted to the problem analysis module. Otherwise, the data transmission to the problem analysis module is temporarily suspended.

[0025] Furthermore, the evaluation result of the summary evaluation module is calculated by a loss function to measure the positioning accuracy of the target detection model with the calculation result, and the calculation formula of the evaluation result is:

[0026]

[0027] In the formula, R represents the evaluation result, n represents the number of samples, and f(X i ) is the predicted value of the i-th sample, y i is the true value of the i-th sample.

[0028] Furthermore, the network distribution module is interactively connected to a node acquisition module via a wireless network, and the node acquisition module is used to issue control authority to each production node, support traceability of product information, and provide an intervention interaction interface after login.

[0029] Furthermore, the node acquisition module is interactively connected to a feedback module via a wireless network, and the feedback module is used to obtain anomalies submitted by uninspected finished products. After being triggered, the node acquisition module is connected to the product information to carry out traceability, and obtain the detection and operation data of the image acquisition module and the working condition acquisition module of the relevant batches, and perform reference analysis to determine whether there are anomalies in the production cycle.

[0030] Furthermore, the image acquisition module, the working condition acquisition module and the attribute determination module are connected through electrical signal communication, the attribute determination module and the problem analysis module and the same type index module are interactively connected through a wireless network, the same type index module and the summary evaluation module are interactively connected through a wireless network, the summary evaluation module and the sharing module are interactively connected through a wireless network, and the sharing module and the network distribution module are interactively connected through a wireless network.

[0031] A method for detecting semiconductor workpiece manufacturing, comprising the following steps:

[0032] Step 1: Obtain image data and equipment operation data on the production line to determine whether there is any abnormality;

[0033] Step 2: Classify the abnormal data and index it in the database according to the classification characteristics. If there is related data, retrieve the processing solution data of the same type of data, evaluate it and then reference it;

[0034] Step 3: If the relevant data does not exist in the database, then calculate, analyze the solution and apply for human intervention;

[0035] Step 4: After obtaining an effective solution, record the solution and the problem it points to and classify it into the database;

[0036] Step 5: After obtaining product abnormality data externally, trace the detection equipment of related batches of related products to analyze whether there are any abnormalities in the production cycle of the product.

[0037] (III) Beneficial effects

[0038] Compared with the known public technology, the technical solution provided by the present invention has the following beneficial effects:

[0039] 1. The present invention uses real-time image detection and equipment detection. After one party detects a problem, it quickly calls the other party for verification to improve the accuracy of detection and prevent misjudgment of abnormalities. When the same problem occurs, it can be indexed in the database and quickly associated with a suitable solution, thereby alleviating computing pressure. As the production process continues, it continues to learn and solve problems in a timely manner when they occur.

[0040] 2. The present invention adds measures to trace the source of product quality problems, obtains product quality problems outside the production line, finds related production lines, and calls their test data under the production batch of the problematic product, thereby helping users analyze the source of the problem and conducting a secondary evaluation of the system's detection capabilities to promote its continuous improvement. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention, and for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0042] Figure 1 A schematic diagram of a semiconductor workpiece manufacturing inspection system;

[0043] The numbers in the figure represent: 1. Image acquisition module; 2. Working condition acquisition module; 3. Attribute determination module; 4. Problem analysis module; 5. Similar indexing module; 6. Summary evaluation module; 7. Sharing module; 8. Distribution network module; 9. Association module; 10. Node acquisition module; 11. Feedback module. DETAILED DESCRIPTION

[0044] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0045] The present invention will be further described below in conjunction with the embodiments.

[0046] Example 1

[0047] A semiconductor workpiece manufacturing detection system of this embodiment, such as Figure 1 As shown, including:

[0048] The image acquisition module 1 is used to be deployed on the production line to obtain image data before, during and after raw material preparation, establish a corresponding standard image reference system according to different products, and compare the currently acquired image data with the data in the standard image reference system in real time to determine whether there is an abnormality. The image acquisition module 1 merges the pixel-connected areas into a whole through connected domain extraction to obtain the key area in the image, extracts the connected domain of the workpiece image after threshold processing, marks the pixel-unconnected areas into different connected domains, extracts and connects the workpiece edge, sets the width threshold range, and extracts the contour that meets the range;

[0049] The working condition acquisition module 2 is used to obtain the operating parameters of each workpiece manufacturing equipment, and regularly receives and converts the equipment working condition parameters according to the preset acquisition cycle to determine whether there is an abnormality. The image acquisition module 1 is interactively connected to the association module 9 through a wireless network. The association module 9 is interactively connected to the working condition acquisition module 2 through a wireless network. The association module 9 serves as a jump end. When the image acquisition module 1 and the working condition acquisition module 2 each obtain abnormal data, the data of the other party's associated abnormal data area is called and re-collected as a reference for auxiliary judgment;

[0050] The attribute determination module 3 is used to obtain the abnormal parameters pointed to by the abnormal judgment, and to classify them accordingly, and to issue classification labels and upload them to the database. After the attribute determination module 3 classifies the data, it is necessary to confirm whether there is feedback data from the same type index module 5. If there is no feedback data, the classified data is packaged and submitted to the problem analysis module 4. Otherwise, the data transmission to the problem analysis module 4 is temporarily suspended. The image data obtained by the attribute determination module 3 is expressed as an imaging model through distortion correction processing. The calculation formula is:

[0051]

[0052] Where (u, v) is the ideal imaging two-dimensional pixel coordinate value, (u d ,v d ) is the actual imaging two-dimensional pixel coordinate value, K is the radial distortion coefficient, when K>0 is positive distortion, the object image is concave inward; when K<0 is negative distortion, the object image is convex outward;

[0053] The problem analysis module 4 is used to analyze based on abnormal parameters, output the detection report to the management end, automatically analyze the solution, receive and upload the manual intervention solution, and record and confirm the solution data that effectively solves the abnormal parameters, extract key features, and upload them to the database;

[0054] The same type indexing module 5 is used to obtain the classification label pointed to by the current abnormal parameter upon receiving the abnormal parameter, extract the key features of the data, and index the effective solution data pointed to by the same type abnormal parameter within the classification label by its key features;

[0055] The summary evaluation module 6 is used to obtain the effective solution data captured by all similar index modules 5, evaluate according to the original data of the current abnormal parameters, extract and display the solution data with the highest effectiveness, and calculate the evaluation result of the summary evaluation module 6 through the loss function to measure the positioning accuracy of the target detection model with the calculation result. The calculation formula of the evaluation result is:

[0056]

[0057] In the formula, R represents the evaluation result, n represents the number of samples, and f(X i ) is the predicted value of the i-th sample, y i is the true value of the i-th sample;

[0058] The sharing module 7 is used to provide a sharing solution data platform, input abnormal parameters after passing identity authentication, and provide query interaction services;

[0059] The network distribution module 8 is used to provide Internet of Things support and provide network permissions for each network-using module.

[0060] like Figure 1 As shown, the image acquisition module 1, the working condition acquisition module 2 and the attribute determination module 3 are connected through electrical signal communication, the attribute determination module 3 and the problem analysis module 4 and the same type index module 5 are interactively connected through a wireless network, the same type index module 5 and the summary evaluation module 6 are interactively connected through a wireless network, the summary evaluation module 6 and the sharing module 7 are interactively connected through a wireless network, and the sharing module 7 and the distribution network module 8 are interactively connected through a wireless network.

[0061] In the specific implementation of this embodiment, through real-time image detection and equipment detection, after one party detects a problem, the other party is quickly called for verification to improve the accuracy of detection and prevent misjudgment of abnormalities. When the same problem occurs, it can be indexed in the database and quickly associated with a suitable solution, thereby alleviating computing power pressure. As the production process continues, continuous learning is carried out and problems are solved in a timely manner when they arise.

[0062] Example 2

[0063] This embodiment also provides a semiconductor workpiece manufacturing detection method, comprising the following steps:

[0064] Step 1: Obtain image data and equipment operation data on the production line to determine whether there is any abnormality;

[0065] Step 2: Classify the abnormal data and index it in the database according to the classification characteristics. If there is related data, retrieve the processing solution data of the same type of data, evaluate it and then reference it;

[0066] Step 3: If the relevant data does not exist in the database, then calculate, analyze the solution and apply for human intervention;

[0067] Step 4: After obtaining an effective solution, record the solution and the problem it points to and classify it into the database;

[0068] Step 5: After obtaining product abnormality data externally, trace the detection equipment of related batches of related products to analyze whether there are any abnormalities in the production cycle of the product.

[0069] Example 3

[0070] In this embodiment, Figure 1As shown, the network distribution module 8 is interactively connected to a node acquisition module 10 via a wireless network. The node acquisition module 10 is used to issue control authority to each production node, support the traceability of product information, and provide an intervention interaction interface after logging in. The node acquisition module 10 is interactively connected to a feedback module 11 via a wireless network. The feedback module 11 is used to obtain anomalies submitted by uninspected finished products. After being triggered, the node acquisition module 10 is connected to the product information to carry out traceability, and the detection and operation data of the image acquisition module 1 and the working condition acquisition module 2 of the relevant batches are obtained for reference analysis to determine whether there are anomalies in the production cycle.

[0071] Through this setting, product quality problems can be obtained outside the production line, related production lines can be found, and their inspection data can be called under the production batch of the problematic product, thereby helping users analyze the source of the problem and conducting a secondary evaluation of the system's detection capabilities to promote its continuous improvement.

[0072] In summary, when the present invention is used, after one party has detected a problem, it quickly calls the other party for verification through real-time image detection and equipment detection, so as to improve the accuracy of detection and prevent misjudgment of abnormalities. When the same problem occurs, it can be indexed in the database and quickly associated with a suitable solution, thereby alleviating the pressure of computing power. As the production process continues, it continues to learn and solve problems in a timely manner when they occur.

[0073] By adding measures to trace product quality issues, obtaining product quality issues outside the production line, finding related production lines, and calling their test data under the production batch of the problematic product, users can analyze the source of the problem, conduct a secondary assessment of the system's detection capabilities, and promote its continuous improvement.

[0074] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements will not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A semiconductor workpiece manufacturing detection system, characterized in that: include: An image acquisition module (1) is used to be deployed on a production line to acquire image data before, during and after raw material preparation, establish a corresponding standard image reference system based on different products, and compare the currently acquired image data with the data in the standard image reference system in real time to determine whether there is any abnormality; The working condition acquisition module (2) is used to obtain the operating parameters of each workpiece manufacturing device, and regularly receive and convert the equipment working condition parameters according to a preset acquisition cycle to determine whether there is an abnormality; The attribute determination module (3) is used to obtain the abnormal parameters pointed to by the abnormal judgment, classify them accordingly, issue classification labels, and upload them to the database; The problem analysis module (4) is used to analyze based on abnormal parameters, output a detection report to the management end, automatically analyze solutions, receive and upload manual intervention solutions, and record and confirm the solution data that effectively solves the abnormal parameters, extract key features, and upload them to the database; A similar indexing module (5) is used to obtain the classification label pointed to by the current abnormal parameter upon receiving the abnormal parameter, extract the key features of the data, and perform indexing by the key features within the classification label to match the valid solution data pointed to by the similar abnormal parameter; The summary evaluation module (6) is used to obtain the effective solution data captured by all similar index modules (5), evaluate according to the original data of the current abnormal parameters, extract and display the solution data with the highest effectiveness; The sharing module (7) is used to provide a sharing solution data platform, input abnormal parameters after passing identity authentication, and provide query interaction services; The network distribution module (8) is used to provide Internet of Things support and provide network permissions for each network-using module; The image data obtained by the attribute determination module (3) is processed by distortion correction to express the imaging model, and the calculation formula is: ； In the formula, is the ideal imaging two-dimensional pixel coordinate value, is the actual imaging two-dimensional pixel coordinate value, K is the radial distortion coefficient, when K>0 is positive distortion, the object image is concave inward; when K<0 is negative distortion, the object image is convex outward; After the attribute determination module (3) classifies the data, it is necessary to confirm whether there is feedback data from the same type index module (5). If there is no feedback data, the classified data is packaged and submitted to the problem analysis module (4). Otherwise, the data transmission to the problem analysis module (4) is temporarily suspended. The evaluation result of the summary evaluation module (6) is calculated by a loss function, and the calculation result is used to measure the positioning accuracy of the target detection model. The calculation formula of the evaluation result is: ； In the formula, R represents the evaluation result, n represents the number of samples, is the predicted value of the i-th sample, is the true value of the i-th sample.

2. A semiconductor workpiece manufacturing detection system according to claim 1, characterized in that: The image acquisition module (1) is interactively connected to the association module (9) via a wireless network, and the association module (9) is interactively connected to the working condition acquisition module (2) via a wireless network. The association module (9) serves as a jump end. When the image acquisition module (1) and the working condition acquisition module (2) respectively acquire abnormal data, the association module (9) calls the data of the other party's associated abnormal data area and re-acquires the data as a reference for auxiliary judgment.

3. A semiconductor workpiece manufacturing detection system according to claim 1, characterized in that: The image acquisition module (1) combines pixel-connected areas into a whole through connected domain extraction, obtains key areas in the image, extracts connected domains from the workpiece image after threshold processing, marks pixel-unconnected areas into different connected domains, extracts and connects workpiece edges, sets a width threshold range, and extracts contours that meet the range.

4. A semiconductor workpiece manufacturing detection system according to claim 1, characterized in that: The network distribution module (8) is interactively connected to a node acquisition module (10) via a wireless network. The node acquisition module (10) is used to issue control permissions to each production node, support product information traceability, and provide an intervention interaction interface after login.

5. A semiconductor workpiece manufacturing detection system according to claim 4, characterized in that: The node acquisition module (10) is interactively connected to a feedback module (11) via a wireless network. The feedback module (11) is used to obtain anomalies submitted by uninspected finished products. After being triggered, the node acquisition module (10) is connected to carry out source tracing based on product information, and the detection and operation data of the image acquisition module (1) and the working condition acquisition module (2) of the relevant batch are obtained for reference analysis to determine whether there are anomalies in the production cycle.

6. A semiconductor workpiece manufacturing detection system according to claim 1, characterized in that: The image acquisition module (1), the working condition acquisition module (2) and the attribute determination module (3) are connected via electrical signal communication; the attribute determination module (3) is interactively connected with the problem analysis module (4) and the same type index module (5) via a wireless network; the same type index module (5) is interactively connected with the summary evaluation module (6) via a wireless network; the summary evaluation module (6) is interactively connected with the sharing module (7) via a wireless network; and the sharing module (7) is interactively connected with the network distribution module (8) via a wireless network.

7. A method for detecting semiconductor workpiece manufacturing, the method being an implementation method of a semiconductor workpiece manufacturing detection system as claimed in any one of claims 1 to 6, characterized in that: The following steps are involved: Step 1: Obtain image data and equipment operation data on the production line to determine whether there is any abnormality; Step 2: Classify the abnormal data and index it in the database according to the classification characteristics. If there is related data, retrieve the processing solution data of the same type of data, evaluate it and then reference it; Step 3: If the relevant data does not exist in the database, then calculate, analyze the solution and apply for human intervention; Step 4: After obtaining an effective solution, record the solution and the problem it points to and classify it into the database; Step 5: After obtaining product abnormality data externally, trace the detection equipment of related batches of related products to analyze whether there are any abnormalities in the production cycle of the product.

Images