Over-gauge product automatic sorting method based on EAP

Abstract

The invention discloses an over-gauge product automatic sorting method based on EAP, and relates to the technical field of sorting, a manufacturing execution system determines whether a wafer is qualified according to a detection value reported by a measuring machine, and generates a corresponding mark; calculating a necessary sorting index in combination with the batch confusion index and the rework complexity index, comparing the necessary sorting index with a preset threshold value, and judging whether the wafers must be sorted by the sorting machine or not; and when the index is lower than a threshold value, the system intelligently controls the wafer to complete sorting operation on the processing machine or the sorting machine according to the sorting capacity of the processing machine. In this way, the high-load state of the sorting machine can be effectively relieved; meanwhile, whether the wafer must pass through the sorting machine or not can be judged according to specific conditions, and the stability of the operation process is ensured; therefore, whether the sorting machine is bypassed or not can be intelligently and dynamically determined on the premise of keeping the wafer classification accuracy, so that the utilization rate of the sorting machine is further reduced, and the overall production efficiency is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of sorting, in particular to an automatic sorting method for out-of-specification products based on EAP. BACKGROUND

[0002] In the semiconductor manufacturing process, each wafer needs to go through multiple complex process procedures and be detected between procedures to ensure that the process parameters meet the preset standards. After detection, the qualified wafers and unqualified wafers usually need to be separated by a sorter so that the qualified wafers continue to the next process and the unqualified wafers enter the rework or abnormal processing path. However, with the continuous improvement of production capacity, the use frequency of the sorter has increased dramatically, becoming a key bottleneck in the production link. In order to alleviate this contradiction, in recent years, under the coordinated control of the manufacturing execution system (MES) and the equipment automation program (EAP), attempts have been made to use the wafer extraction and sorting functions of the measurement machine and some process machines to complete the screening and circulation of some wafers at the machine end, thereby reducing the dependence on the sorter, improving production efficiency, and reducing equipment investment.

[0003] However, the existing method still has some shortcomings: on the one hand, when the process machine itself does not have wafer extraction or sorting capability, the wafer still needs to rely on the sorter to complete the sorting, and it is difficult to effectively alleviate the high load state of the sorter; on the other hand, even if the process machine has wafer extraction and sorting functions, if there are batch mixed loading, complex rework paths, etc., the judgment of whether the wafer needs to pass through the sorter often depends on human experience, lacks a unified and quantitative judgment mechanism, and is prone to unstable operation processes. Therefore, a new automatic sorting method for out-of-specification products based on EAP is urgently needed, which can intelligently and dynamically decide whether to bypass the sorter while maintaining the accuracy of wafer classification, thereby further reducing the usage rate of the sorter and improving overall production efficiency. SUMMARY

[0004] The purpose of the present application is to solve the above-mentioned problems and provide an automatic sorting method for out-of-specification products based on EAP.

[0005] The present application provides an automatic sorting method for out-of-specification products based on EAP, which comprises: In the manufacturing execution system, it is determined whether the wafer is qualified according to the detection value reported by the measurement machine, and a corresponding qualified mark or unqualified mark is generated; The manufacturing execution system receives the batch confusion index and the rework complexity index wafer mark, combines the batch information and the rework path information, and calculates the must-sort index; The must-sort index is compared with a preset must-sort index threshold to determine whether the wafer must be sorted by the sorter; When the must-sorting index is less than the preset must-sorting index threshold, the manufacturing execution system controls the wafer to complete sorting on the processing machine or the sorting machine according to the sorting capability information of the processing machine.

[0006] Optionally, in the manufacturing execution system, whether the wafer is qualified is determined according to the detection value reported by the measurement machine, and a corresponding qualified mark or unqualified mark is generated. Specifically, the key parameter qualified range of the measurement machine is set in the manufacturing execution system, and after receiving the detection data of the measurement machine, the batch confusion index and the rework complexity index detection data are compared with the key parameter qualified range set in advance in the manufacturing execution system. When the batch confusion index and the rework complexity index detection data are within the batch confusion index and the rework complexity index qualified range, a qualified mark is generated for the corresponding wafer. When the batch confusion index and the rework complexity index detection data exceed the batch confusion index and the rework complexity index qualified range, an unqualified mark is generated for the corresponding wafer.

[0007] Optionally, the step of calculating the must-sorting index in combination with the batch information and the rework path information is as follows: The batch confusion index is calculated according to the batch information, the rework complexity index is calculated according to the rework path information, the batch confusion index and the rework complexity index are normalized, the normalized batch confusion index and the rework complexity index are weighted and summed to obtain the must-sorting index.

[0008] Optionally, the step of calculating the batch confusion index is as follows: The total number of wafers in the target wafer carrier is obtained, wherein there are several different batches, and the number of wafers in each corresponding batch is recorded; The total number of wafers corresponding to the batch with the largest number is subtracted from the total number of wafers corresponding to the batch with the smallest number, and the result of the subtraction is divided by the total number of wafers to obtain the maximum contrast intensity coefficient of the batch distribution; The overall proportion of the number of wafers in each batch is calculated to obtain the number proportion of each batch, and the number proportion of all batches is taken as a proportion sequence; and the difference between adjacent proportions in the proportion sequence is multiplied to obtain a multiplication factor; The proportion of each batch is inversely proportional to the transformation, and the result of the inverse proportional transformation is accumulated to obtain the proportion reciprocal sum; The proportion reciprocal sum is added to the maximum contrast intensity coefficient and then added to the multiplication factor to obtain the batch confusion index.

[0009] Optionally, the step of calculating the rework complexity index is as follows: ​Suppose the whole production process has M processes, and the number of processes that the unqualified wafer needs to pass through again during the rework process is L. Divide the number of processes that the unqualified wafer needs to pass through again during the rework process L by the total number of processes M of the production process to obtain the rework process ratio; Suppose the serial number of the earliest process involved in the rework path is A, and the serial number of the latest process is B. The span length of the rework path is B-A. Combine the span length with the number of rework processes to obtain the rework span factor. The formula for calculation is: In the formula, S represents the rework span factor. Multiply the rework process ratio and the rework span factor, and take the square root of the multiplication result as the rework complexity index.

[0010] Optionally, the step of comparing the must sort index with the preset must sort index threshold to determine whether the wafer must be sorted by the sorting machine is: Compare the must sort index and the preset must sort index threshold. If the must sort index is not less than the preset must sort index threshold, the wafer must be sorted by the sorting machine. If the must sort index is less than the preset must sort index threshold, the wafer does not necessarily have to be sorted by the sorting machine.

[0011] Optionally, when the must sort index is less than the preset must sort index threshold, the manufacturing execution system controls the wafer to complete sorting on the process machine or the sorting machine according to the sorting capability information of the process machine. The step is: When the process machine does not support wafer extraction, the wafer is sorted by the sorting machine. The qualified wafer enters the next process, and the unqualified wafer enters the rework or abnormal processing. When the process machine supports wafer extraction but does not support sorting, the wafer first completes the wafer extraction operation on the process machine, and then completes the sorting by the sorting machine. When the process machine supports wafer extraction and sorting, and it is determined that the sorting machine is not needed, the wafer directly completes the sorting on the process machine. The qualified wafer enters the discharge port and goes to the next process. The unqualified wafer is retained in the loading port for rework or abnormal processing.

[0012] The beneficial effects of the present application are: The application provides an automatic sorting method for out-of-specification products based on EAP. A manufacturing execution system determines whether a wafer is qualified according to a detection value reported by a measurement machine and generates a corresponding label. A sorting index is calculated by combining a batch confusion index and a rework complexity index, and compared with a preset threshold to determine whether the wafer must be sorted by a sorting machine. When the index is lower than the threshold, the system intelligently controls the wafer to complete the sorting operation on the processing machine or the sorting machine according to the sorting capacity of the processing machine. In this way, the high load state of the sorting machine can be effectively relieved, and it can be determined whether the wafer must pass through the sorting machine according to specific conditions to ensure the stability of the operation process. In this way, it can be intelligently and dynamically determined whether to bypass the sorting machine while maintaining the accuracy of wafer classification, thereby further reducing the utilization rate of the sorting machine and improving the overall production efficiency. BRIEF DESCRIPTION OF DRAWINGS

[0013] The application will be further described below with reference to the drawings.

[0014] Figure 1 A flowchart of an automatic sorting method for out-of-specification products based on EAP. DETAILED DESCRIPTION

[0015] The technical solutions in the embodiments of the application will be clearly and completely described below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, rather than all the embodiments of the application. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the application.

[0016] Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the application.

[0017] The application provides an automatic sorting method for out-of-specification products based on EAP. Referring to Figure 1 , Figure 1 A flowchart of an automatic sorting method for out-of-specification products based on EAP provided by the application. The method comprises the following steps: In the manufacturing execution system, whether a wafer is qualified is determined according to a detection value reported by a measurement machine, and a corresponding qualified label or unqualified label is generated; After receiving the batch confusion index and the rework complexity index wafer label, the manufacturing execution system combines the batch information and the rework path information to calculate a sorting index; The sorting index is compared with a preset sorting index threshold to determine whether the wafer must be sorted by a sorting machine; When the required sorting index is less than the preset required sorting index threshold, the manufacturing execution system controls the wafers to complete sorting on the process equipment or sorter based on the sorting capacity information of the process equipment.

[0018] Based on the present invention, an EAP-based automatic sorting method for oversized products can effectively alleviate the high load on the sorting machine. At the same time, it can determine whether the wafer must pass through the sorting machine according to the specific situation, ensuring the stability of the operation process. In this way, while maintaining the accuracy of wafer classification, it can intelligently and dynamically decide whether to bypass the sorting machine, thereby further reducing the utilization rate of the sorting machine and improving the overall production efficiency.

[0019] In one embodiment, in the manufacturing execution system, determining whether a wafer is qualified based on the inspection values ​​reported by the measurement machine and generating a corresponding qualified or unqualified mark specifically involves: Set the acceptance range of key parameters for the measuring machine in the manufacturing execution system, and after receiving the test data from the measuring machine, compare the batch disorder index and rework complexity index test data with the acceptance range of key parameters set in the manufacturing execution system in advance. When the batch disorder index and rework complexity index test data are within the acceptable range of the batch disorder index and rework complexity index, a qualified mark is generated for the corresponding wafer; When the batch disorder index and rework complexity index test data exceed the acceptable range of the batch disorder index and rework complexity index, a non-conforming mark is generated for the corresponding wafer.

[0020] It should be noted that the acceptable range of key parameters for metrology equipment refers to the upper and lower limits of detection parameters set by the Manufacturing Execution System (MES) based on process design specifications, historical measurement data, and process control requirements to ensure the stability of wafer quality before it enters the next process step in semiconductor manufacturing. This range typically includes key indicators directly related to the quality of the current process, such as film thickness, linewidth, electrical characteristics, and defect density, and is stored in the MES as numerical ranges. The setting method is as follows: before process development or mass production, process engineers determine the allowable fluctuation range of each parameter through experimental verification and statistical analysis of the target process, and enter this range into the MES. When the system receives the detection data reported by the metrology equipment, it can automatically compare and determine whether the data falls within the batch disorder index and rework complexity index range. For example, in the thin film deposition process, if the process requires a film thickness within the allowable range of 100nm to 105nm, then when the measurement instrument detects a thickness of 102nm, the manufacturing execution system determines that it falls within the acceptable range and generates a pass mark; when the detection result is 108nm, it is determined to be outside the range and a fail mark is generated. This method ensures that subsequent sorting processes can be based on objective standards of measurement results, avoiding the uncertainty caused by differences in manual judgment.

[0021] In one embodiment, after receiving the wafer markers for the batch disorder index and rework complexity index, the manufacturing execution system calculates the required sorting index by combining the batch information and rework path information. In one implementation method, the step of calculating the required sorting index by combining batch information and rework path information is as follows: The batch disorder index is calculated based on batch information, and the rework complexity index is calculated based on rework path information. Both the batch disorder index and the rework complexity index are normalized. Finally, the normalized batch disorder index and the rework complexity index are weighted and summed to obtain the mandatory sorting index. The formula for calculation is: In the formula, TG is the mandatory sorting index, and XC and XZ are the normalized batch disorder index and rework complexity index, respectively. These represent the preset weighting coefficients for the batch disorder index and rework complexity index after multiple normalization processing, respectively. All are greater than 0; It should be noted that the above-mentioned normalization methods for removing dimensions include Min-Max normalization, Z-Score standardization, etc., which will not be elaborated here. Settings should be set according to the actual situation, generally They are equal and their sum is 1, for example, It can be 0.5 or 0.5.

[0022] In one embodiment, the batch disorder index is calculated as follows: Suppose the total number of wafers in the target wafer carrier is T, and there are N different batches, with batch numbers as follows: The corresponding number of wafers in each batch is ; The maximum contrast ratio of the batch distribution is obtained by subtracting the total number of wafers from the total number of wafers from the batch with the largest number of wafers, and then dividing the result by the total number of wafers; the closer it is to 0, the more uniform the batches are. Calculate the overall proportion of wafers in each batch to obtain the quantity ratio of each batch, and take the quantity ratio of all batches as a ratio sequence; calculate the product of the differences between adjacent ratios in the ratio sequence as a multiplication factor to measure the degree of unevenness in the proportion between batches; the larger the value of this factor, the more severe the difference in the proportion between adjacent batches and the more unbalanced the distribution. The proportion of each batch is inversely transformed, and the results of the inverse transformation are summed to obtain the sum of the reciprocals of the proportions. This operation can amplify the influence of the number of batches and their distribution sparseness. When the number of batches is large and the proportion of each batch is small, this summation result will increase rapidly. The batch disorder index is obtained by adding the proportional derivative, the maximum contrast intensity coefficient, and the multiplication factor.

[0023] It should be noted that during the calculation of the batch disorder index, all raw data are automatically collected and structured by the Manufacturing Execution System (MES). Specifically, the total number of wafers in the target wafer carrier can be read in real time by the MES system. This information includes the current carrier number and a complete list of its built-in wafers. Different batches of wafers are distinguished based on the batch identifiers (such as batch number, process code, timestamp, etc.) automatically entered when the wafers are put into storage or onto the machine. The MES system will perform batch classification statistics on the wafers in the carrier to obtain the quantity of all different batches and the number of wafers corresponding to each batch. The ratio sequence is the ratio of the quantity of each batch to the total number of wafers, which is automatically calculated by the system based on the statistical batch quantity. The multiplication factor of the difference between adjacent ratios is obtained by the system calculating the difference item by item in the sorted ratio sequence and multiplying them sequentially. The inverse ratio transformation and the reciprocal sum are also obtained by the MES system performing mathematical transformations item by item based on the existing ratio values. As for the maximum contrast intensity coefficient, the MES system can directly obtain this parameter by comparing the difference between the batch with the largest number and the batch with the smallest number in the carrier and dividing it by the total number of wafers. All of the above data can be automatically statistically calculated before measurement or manufacturing process without manual intervention, ensuring the objectivity, stability and efficiency of the batch disorder index generation process.

[0024] It should be noted that the batch disorder index measures the degree of batch disorder and distribution complexity of wafers in a target carrier. It comprehensively considers the maximum batch quantity difference, the cumulative effect of differences between batch proportions, and the dispersion of batch quantities. A higher batch disorder index indicates a larger number of wafer batches in the current carrier, closer proportions between batches, and a more uniform distribution, or extremely irregular batch differences, resulting in blurred batch boundaries and a complex distribution. Without forced sorting by a sorting machine, this can easily lead to subsequent process steps mixing wafers from different batches, causing disruptions in the production chain and difficulties in quality traceability. Therefore, it is essential to... Thorough sorting by a sorting machine is necessary to ensure that different batches are strictly separated. For example, if a carrier contains 20 wafers, 15 of which belong to the same batch and 5 to another batch, the batch confusion index will be relatively low because the differences between batches are obvious and easy to distinguish manually or by machine. However, if the 20 wafers come from 5 different batches, with 4 wafers from each batch, the batch confusion index will be close to 1, indicating that the batch mixing situation is extremely serious. If the wafers are directly handed over to the process equipment, different batches of wafers may be incorrectly mixed, leading to the failure of production management and quality control. Therefore, in this case, the system will determine that all wafers must be sorted by a sorting machine.

[0025] It's important to note that the reason for calculating the batch disorder index using the above method, rather than other common methods (such as directly using proportion variance or entropy), is primarily because this method can simultaneously characterize batch mixing from multiple perspectives, thus avoiding the potential bias caused by a single indicator. First, introducing the "maximum contrast intensity coefficient" captures the extreme differences between the most and least numerous batches, ensuring that even situations with significant quantity imbalances between only two batches can be identified. Second, by calculating the multiplication factor of the differences between adjacent proportions in the proportion sequence, the impact of local proportion differences in batch distribution can be amplified, accurately reflecting the uneven distribution among batches. Third, by inversely transforming and summing the proportions, the sparsity can be amplified when the number of batches increases and the proportion of each batch is generally small, revealing hidden disorder risks even when the quantities appear balanced. Finally, the results from these three dimensions are combined into a unified batch disorder index, considering both the extreme differences between overall batches and local differences and overall sparsity, avoiding the problem of traditional single-indicator calculation methods over-reliance on a single dimension. The index calculated in this way ensures stable and interpretable results, and does not rely on manually set weights or unknown factors, making it suitable for automated judgment directly within the MES system. Its advantages lie in its ability to maintain sensitivity and robustness across different batch distribution scenarios. It avoids missing severely mixed batches and effectively prevents misjudgments caused by occasional fluctuations in batch quantity distribution, thereby improving the accuracy and stability of wafer sorting decisions.

[0026] In one embodiment, the rework complexity index is calculated as follows: Let M be the number of processes in the entire production process, and L be the number of processes that a defective wafer needs to go through again during rework. Divide L by the total number of processes M in the entire production process to get the rework process ratio. This ratio reflects the scale of the rework path in the overall process flow. The larger the ratio, the larger the scale of rework. Let A be the sequence number of the earliest process involved in the rework path, B be the sequence number of the latest process, and BA be the span length of the rework path. Combining the span length with the number of rework processes, we obtain the rework span factor, calculated using the following formula: , where S represents the rework span factor; Multiply the rework process ratio by the rework span factor, and take the square root of the result as the rework complexity index.

[0027] It should be noted that in the calculation of the rework complexity index, all the required data can be automatically obtained and retrieved through the process management module and the exception handling record module in the Manufacturing Execution System (MES), ensuring that the entire calculation process is online, structured, and without human intervention. Specifically, the total number of processes in the entire production process comes from the standard process flow template set in the MES. This template predefines the complete processing path for each product, and the process flow number and sequence information can be directly called. The number of rework processes comes from the rework plan automatically generated by the MES system after the defective wafer is judged as defective. This plan clearly lists each process that needs to be re-executed, and the system can count the number of rework processes accordingly. The sequence numbers of the earliest and latest processes in the rework path are automatically identified by sorting the rework process numbers. The system can automatically obtain the corresponding process position by comparing the minimum and maximum values ​​of the process numbers in the rework process list, thereby calculating the span. The entire data acquisition process is based entirely on the existing production configuration information and running process data in the MES system. All data are structured fields that support high-frequency calls and real-time updates, making them suitable for continuous rework complexity identification and dynamic decision support in high-cycle production environments.

[0028] It's important to note that the rework complexity index measures the complexity of the process paths involved in reworking defective wafers. Specifically, it reflects the proportion of rework operations in the entire production process and the dispersion of rework steps in time or sequence. Its core function is to determine whether the current rework operation is highly customized, has strong process jumps, or is highly dispersed, thus assisting the system in deciding whether pre-sorting by a sorting machine is necessary. A higher index value indicates a higher proportion of rework steps in the entire process and a larger process span, representing a wider distribution of rework paths, more dispersed jump locations, and longer time spans. In such cases, if wafers are directly sent to the processing equipment without sorting, mismatched machine process parameters, uncoordinated equipment operation rhythms, or non-standardized work paths are highly likely to cause processing failures, chaotic wafer flow, or process failures, thereby affecting the stability of the entire production line and the consistency of product quality. For example, if a defective wafer only needs to be reworked to an adjacent process, the rework complexity index is relatively low. The system can allow it to be sent to the machine along with other qualified wafers in its original carrier and screened through the wafer extraction logic. However, if the rework path of another wafer spans more than a dozen processes and is distributed across multiple discontinuous segments before and after the process, its rework complexity index will increase significantly. The system will automatically determine that the wafer must be pre-separated by a sorting machine and an independent rework path must be established, ensuring that the rework plan is completely decoupled from the process path. Through this index, the system can achieve a quantitative assessment of the complexity of the rework path, avoiding potential errors caused by differences in human judgment, thereby improving rework management efficiency and production stability at the system level.

[0029] The reason for using the above method to calculate the rework complexity index, rather than traditional methods such as counting the number of processes, counting process jumps, or setting levels based on expert experience, is primarily because this method can more accurately, structurally, and without human intervention, comprehensively reflect the dual characteristics of structural complexity and process span of the rework path in the production process, possessing stronger engineering feasibility and system decision support value. Specifically, this method quantifies the "coverage" of rework operations in the overall process by using the ratio of the number of rework processes to the length of the entire process flow, which can truly reflect whether the rework plan deviates from the standard path. On the other hand, it reflects the distribution density of the rework process in the process by using the ratio of the span between the earliest and latest processes in the rework path to the number of rework processes, avoiding the error of ignoring process jumps by judging solely by quantity. The results of these two dimensions are multiplied and fused, and the square root is taken, which effectively suppresses the excessive amplification of the overall judgment by outliers in one dimension, making the calculation results both sensitive and stable. Meanwhile, this calculation method does not involve weight adjustment, empirical parameters, or machine learning modeling, and does not introduce uncontrollable factors. It has the advantages of being reproducible, comparable, and embeddable into MES or EAP systems for automatic execution, making it suitable for real-time decision-making and batch processing scenarios. After normalizing the final result through an exponential function, the threshold for judging rework complexity becomes clearer and more adjustable. This allows the system to maintain consistent sorting standards and execution logic in production scenarios with diverse rework paths, multiple batches, and frequent task switching, fundamentally improving the controllability of process scheduling and the traceability efficiency of wafer quality.

[0030] In one embodiment, the step of comparing the mandatory sorting index with a preset mandatory sorting index threshold to determine whether a wafer must be sorted by a sorter is as follows: The required sorting index is compared with the preset required sorting index threshold. If the required sorting index is not less than the preset required sorting index threshold, the wafer must be sorted by the sorter. If the required sorting index is less than the preset required sorting index threshold, the wafer does not necessarily have to be sorted by the sorter.

[0031] It should be noted that in the actual wafer processing, by comparing the mandatory sorting index with a preset mandatory sorting index threshold, the manufacturing execution system can be provided with a clear and quantifiable judgment standard to dynamically decide whether the current wafer must undergo sorting by the sorting machine. If the mandatory sorting index is not less than the threshold, it indicates that the wafer's batch origin is complex and the rework path is extensive, posing a high risk of contamination and uncertainty in the rework process. The system will directly determine that the wafer must be sorted by the sorting machine to ensure the clarity and controllability of the subsequent processing path.

[0032] In one embodiment, when the required sorting index is less than a preset required sorting index threshold, the manufacturing execution system controls the wafer to complete the sorting process on the process equipment or sorter based on the sorting capacity information of the process equipment. When the process equipment does not support wafer extraction, the wafers are sorted by a sorting machine. Qualified wafers enter the next process, while unqualified wafers are reworked or handled abnormally. When the process equipment supports wafer extraction but not sorting, the wafers are first extracted on the process equipment and then sorted by the sorting machine. When the process equipment supports both wafer extraction and sorting, and it is determined that no sorting machine is needed, the wafers are sorted directly on the process equipment. Qualified wafers enter the unloading port and are sent to the next process, while unqualified wafers are kept at the loading port for rework or abnormal handling.

[0033] It should be noted that when the required sorting index is below a preset threshold, the Manufacturing Execution System (MES) will flexibly arrange the subsequent operation path of the wafers based on the actual sorting capacity information of the process equipment. This achieves intelligent optimization and resource coordination of the sorting process. When the process equipment does not have a wafer extraction function, the wafers must undergo standardized screening by a sorter. The sorter physically separates qualified and unqualified wafers. Qualified wafers are sent to the next process according to the flow, while unqualified wafers are marked and guided to the rework process or trigger an exception handling mechanism. If the process equipment has a wafer extraction function but does not support sorting, the wafers are first pre-processed by selecting target wafers through a wafer extraction operation in the process equipment, and then sent to the sorter for final screening. This mode has strong scheduling flexibility when the sorter is running and there is a waiting time. This reduces idle waiting time. When the process equipment supports both wafer extraction and sorting, and the sorting index must be below a threshold, the system can skip the sorting machine and complete the sorting operation directly inside the process equipment. The EAP system marks qualified wafers and automatically puts them into the carrier corresponding to the unloading port and sends them to the next process. Unqualified wafers are kept at the loading port and automatically transferred to the rework or abnormal handling process. This method not only saves sorting machine resources, reduces the number of handling operations and cycle time, but also reduces the error risk caused by equipment handover, and improves the stability and intelligence level of the entire operation process. For example, if only two wafers in a batch are unqualified and the process equipment has the ability to extract and sort, the process equipment can directly complete the sorting without the need for sorting by the Sorter machine, thus improving the utilization efficiency of the equipment.

[0034] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the patent coverage of this invention.

Claims

1. An automated sorting method for oversized products based on EAP, characterized in that, Includes the following steps: In the manufacturing execution system, the wafer is judged to be qualified based on the test values ​​reported by the measurement machine, and a corresponding qualified or unqualified mark is generated. After receiving the wafer markers for the batch disorder index and rework complexity index, the manufacturing execution system calculates the required sorting index by combining the batch information and rework path information. The mandatory sorting index is compared with a preset mandatory sorting index threshold to determine whether the wafer must be sorted by the sorter. When the required sorting index is less than the preset required sorting index threshold, the manufacturing execution system controls the wafers to complete sorting on the process equipment or sorter based on the sorting capacity information of the process equipment.

2. The automatic sorting method for oversized products based on EAP according to claim 1, characterized in that, In the manufacturing execution system, the wafer's pass / fail status is determined based on the inspection values ​​reported by the measurement machine, and corresponding pass or fail marks are generated. Specifically: Set the acceptance range of key parameters for the measuring machine in the manufacturing execution system, and after receiving the test data from the measuring machine, compare the batch disorder index and rework complexity index test data with the acceptance range of key parameters set in the manufacturing execution system in advance. When the batch disorder index and rework complexity index test data are within the acceptable range of the batch disorder index and rework complexity index, a qualified mark is generated for the corresponding wafer; When the batch disorder index and rework complexity index test data exceed the acceptable range of the batch disorder index and rework complexity index, a non-conforming mark is generated for the corresponding wafer.

3. The automatic sorting method for oversized products based on EAP according to claim 1, characterized in that, The steps to calculate the required sorting index by combining batch information and rework path information are as follows: The batch disorder index is calculated based on the batch information, and the rework complexity index is calculated based on the rework path information. The batch disorder index and the rework complexity index are normalized. The normalized batch disorder index and the rework complexity index are then weighted and summed to obtain the mandatory sorting index.

4. The automatic sorting method for oversized products based on EAP according to claim 3, characterized in that, The calculation steps for the batch disorder index are as follows: Obtain the total number of wafers in the target wafer carrier, which contains several different batches, and record the number of wafers in each batch. The maximum contrast intensity coefficient of the batch distribution is obtained by subtracting the total number of wafers from the total number of wafers in the batch with the largest number of wafers, and then dividing the result by the total number of wafers. Calculate the overall proportion of wafer quantity in each batch to obtain the quantity ratio of each batch, and take the quantity ratio of all batches as a ratio sequence; and calculate the product of the differences between adjacent ratios in the ratio sequence as a multiplication factor. The proportion of each batch is inversely transformed, and the results of the inverse transformation are summed to obtain the sum of the reciprocals of the proportions; Adding the reciprocal of the proportion to the maximum contrast intensity coefficient, and then adding the multiplication factor, yields the batch disorder index.

5. The automatic sorting method for oversized products based on EAP according to claim 3, characterized in that, The calculation steps for the rework complexity index are as follows: Let M be the total number of processes in the entire production process, and L be the number of processes that a defective wafer needs to go through again during rework. Divide L by the total number of processes M in the entire production process to get the rework process ratio. Let A be the sequence number of the earliest process involved in the rework path, B be the sequence number of the latest process, and BA be the span length of the rework path. Combining the span length with the number of rework processes, we obtain the rework span factor, calculated using the following formula: , where S represents the rework span factor; Multiply the rework process ratio by the rework span factor, and take the square root of the result as the rework complexity index.

6. The automatic sorting method for oversized products based on EAP according to claim 1, characterized in that, The steps for determining whether wafers must be sorted by a sorting machine are as follows: (The required sorting index is compared with a preset required sorting index threshold.) The required sorting index is compared with the preset required sorting index threshold. If the required sorting index is not less than the preset required sorting index threshold, the wafer must be sorted by the sorter. If the required sorting index is less than the preset required sorting index threshold, the wafer does not need to be sorted by the sorter.

7. The automatic sorting method for oversized products based on EAP according to claim 1, characterized in that, When the required sorting index is less than the preset required sorting index threshold, the manufacturing execution system controls the wafers to complete the sorting process on the process equipment or sorter based on the sorting capacity information of the process equipment. When the process equipment does not support wafer extraction, the wafers are sorted by a sorting machine. Qualified wafers enter the next process, while unqualified wafers are reworked or handled abnormally. When the process equipment supports wafer extraction but not sorting, the wafers are first extracted on the process equipment and then sorted by the sorting machine. When the process equipment supports both wafer extraction and sorting, and it is determined that no sorting machine is needed, the wafers are sorted directly on the process equipment. Qualified wafers enter the unloading port and are sent to the next process, while unqualified wafers are kept at the loading port for rework or abnormal handling.

Images