CN-121997008-A - Semiconductor yield root cause determination method, device, equipment and storage medium

Abstract

The application provides a method, a device, equipment and a storage medium for determining the root cause of a semiconductor yield, wherein the method comprises the steps of obtaining original data corresponding to a target manufacturing process, preprocessing the original data to obtain target data, determining doubtful degree data corresponding to the original characteristic values according to a preset index calculation mode aiming at all the original characteristic values in the target data, deleting the original characteristic values, meeting preset exclusion conditions, of the doubtful degree data in the target data to obtain alternative characteristic values, and determining the target characteristic values corresponding to the target manufacturing process from the alternative characteristic values according to the doubtful degree data of all the alternative characteristic values. Therefore, the final determined target characteristic value can be more in line with the business logic of the actual semiconductor scene by calculating the suspicion degree data based on a preset index calculation mode and performing characteristic elimination according to a preset elimination condition, and the accuracy of determining the root cause of the yield can be improved.

Inventors

• DUAN LILI
• ZHOU TAO

Assignees

• 上海集成电路研发中心有限公司
• 上海微迈睿科技有限公司

Dates

Publication Date

20260508

Application Date

20241101

Claims (10)

1. 1. A method for determining root cause of semiconductor yield, comprising: Acquiring original data corresponding to a target manufacturing process, and preprocessing the original data to obtain target data; for each initial characteristic value in the target data, determining suspicion degree data corresponding to the initial characteristic value according to a preset index calculation mode; Deleting initial characteristic values of which the suspicion degree data meet preset exclusion conditions from the target data to obtain alternative characteristic values; And determining a target characteristic value corresponding to the target manufacturing process from the candidate characteristic values according to the suspicion degree data of each candidate characteristic value.
2. 2. The method of claim 1, wherein the raw data includes yield index data and manufacturing process data; The yield index data comprises at least one of wafer test data corresponding to the target manufacturing process and defect number data in the target manufacturing process, and the data type of the manufacturing process data comprises at least one of station identification, machine station identification, chamber identification, process formula identification and material identification.
3. 3. The method according to claim 2, wherein preprocessing the raw data to obtain target data comprises: performing format conversion on the original data to obtain alternative data; determining a target state identifier corresponding to each wafer in the target manufacturing process according to the yield index data; And adding the target state identifier into the alternative data to obtain the target data.
4. 4. The method according to claim 1, wherein the determining the suspicion data corresponding to the initial feature value according to the preset index calculation method includes: analyzing and processing each feature type in the target data based on a preset variance analysis mode to obtain a significant difference parameter corresponding to each initial feature value in the feature type, and/or, Determining the ratio of the number of abnormal wafers corresponding to the initial characteristic value to the number of wafers corresponding to the initial characteristic value to obtain a first abnormal proportion corresponding to the initial characteristic value, and/or, Determining the ratio of the number of abnormal wafers corresponding to the initial characteristic value to the total number of abnormal wafers corresponding to the characteristic type to which the initial characteristic value belongs to obtain a second abnormal proportion corresponding to the initial characteristic value, and/or, And determining the ratio of a target sum value corresponding to the initial characteristic value to the total number of wafers corresponding to the characteristic type to which the initial characteristic value belongs to obtain the common proportion of the abnormal wafers corresponding to the initial characteristic value, wherein the target sum value is the sum value of the number of abnormal wafers corresponding to the initial characteristic value and the number of normal wafers corresponding to other initial characteristic values except the initial characteristic value in the characteristic type to which the initial characteristic value belongs.
5. 5. The method according to claim 4, wherein the method further comprises: If the significant difference parameter of the initial characteristic value is larger than a first preset threshold value, determining that the initial characteristic value meets the preset exclusion condition, or Determining a first initial feature value with the largest number of abnormal wafers in the feature type according to the feature type to which the initial feature value belongs, and determining that each initial feature value corresponding to the feature type meets the preset exclusion condition if a first abnormal proportion corresponding to the first initial feature value is smaller than a second preset threshold value, or If the second abnormal proportion corresponding to the first initial characteristic value is larger than a third preset threshold value and the ratio of the number of wafers corresponding to the first initial characteristic value to the total number of wafers corresponding to the characteristic type is larger than a fourth preset threshold value, determining that each initial characteristic value in the characteristic type meets the preset exclusion condition, or If the common proportion of the abnormal wafers corresponding to the first initial characteristic value is smaller than a fifth preset threshold value, determining that all initial characteristic values corresponding to the characteristic types meet the preset exclusion condition.
6. 6. The method according to any one of claims 1 to 5, wherein determining a target feature value corresponding to the target manufacturing process from among the candidate feature values based on suspicion data of the respective candidate feature values, comprises: Weighting and summing the suspicion degree data of the alternative characteristic values aiming at each alternative characteristic value to obtain target suspicion degree data corresponding to the alternative characteristic value; and determining the target characteristic value from the alternative characteristic value according to the target suspicion degree data.
7. 7. A semiconductor yield cause determination apparatus, comprising: the acquisition module is used for acquiring original data corresponding to a target manufacturing process and preprocessing the original data to obtain target data; The first determining module is used for determining suspicion degree data corresponding to the initial characteristic values according to a preset index calculation mode aiming at each initial characteristic value in the target data; The deleting module is used for deleting the initial characteristic value of which the suspicion degree data meets the preset exclusion condition from the target data to obtain an alternative characteristic value; And the second determining module is used for determining a target characteristic value corresponding to the target manufacturing process from the candidate characteristic values according to the suspicion degree data of each candidate characteristic value.
8. 8. The semiconductor yield root cause determining device is characterized by comprising a processor and a memory; The memory stores computer-executable instructions; The processor executes computer-executable instructions stored in the memory to implement the semiconductor yield root cause determination method of any one of claims 1 to 6.
9. 9. A computer-readable storage medium having stored therein computer-executable instructions for implementing the semiconductor yield root cause determination method of any one of claims 1 to 6 when the computer-executable instructions are executed.
10. 10. A computer program product comprising a computer program which, when executed, implements the semiconductor yield root cause determination method of any one of claims 1 to 6.

Description

Semiconductor yield root cause determination method, device, equipment and storage medium Technical Field The present application relates to the field of semiconductor manufacturing technologies, and in particular, to a method, an apparatus, a device, and a storage medium for determining a root cause of a semiconductor yield. Background Semiconductor manufacturing is a highly sophisticated engineering technique, and the production process is relatively complex, typically involving hundreds or thousands of process steps, each involving a different tool (Equipment), chamber (Recipe), recipe (Recipe), etc. On this basis, when a problem occurs in the yield of semiconductors, it is often difficult for a worker to quickly locate the Root Cause (Root Cause) in a large amount of data. The root cause of the yield is a root cause of the yield decrease in the semiconductor manufacturing process, and specifically may be an abnormal processing process in the semiconductor manufacturing process. The abnormality processing process is generally represented as an abnormal change index of the entire semiconductor manufacturing process, or an abnormal change index in the semiconductor intermediate manufacturing process, or the like. In order to quickly determine the yield root cause, a data processing mode based on statistics is generally adopted in the related art, and the yield root cause is determined based on a statistical index, wherein the yield root cause finally determined by the yield root cause determining method does not accord with business logic of a semiconductor actual scene, so that the accuracy of determining the yield root cause is not high. Disclosure of Invention The application provides a method, a device, equipment and a storage medium for determining the root cause of the semiconductor yield, which can enable the root cause of the yield finally determined to accord with the actual business logic of a semiconductor manufacturing scene, and improve the accuracy of the root cause determination of the yield. In a first aspect, an embodiment of the present application provides a method for determining a root cause of a semiconductor yield, including: Acquiring original data corresponding to a target manufacturing process, and preprocessing the original data to obtain target data; for each initial characteristic value in the target data, determining suspicion degree data corresponding to the initial characteristic value according to a preset index calculation mode; Deleting initial characteristic values of which the suspicion degree data meet preset exclusion conditions from the target data to obtain alternative characteristic values; And determining a target characteristic value corresponding to the target manufacturing process from the candidate characteristic values according to the suspicion degree data of each candidate characteristic value. In one possible implementation, the raw data includes yield index data and manufacturing process data; The yield index data comprises at least one of wafer test data corresponding to the target manufacturing process and defect number data in the target manufacturing process, and the data type of the manufacturing process data comprises at least one of station identification, machine station identification, chamber identification, process formula identification and material identification. In a possible implementation manner, the preprocessing the raw data to obtain target data includes: performing format conversion on the original data to obtain alternative data; determining a target state identifier corresponding to each wafer in the target manufacturing process according to the yield index data; And adding the target state identifier into the alternative data to obtain the target data. In a possible implementation manner, the determining the suspicion degree data corresponding to the initial feature value according to a preset index calculation manner includes: analyzing and processing each feature type in the target data based on a preset variance analysis mode to obtain a significant difference parameter corresponding to each initial feature value in the feature type, and/or, Determining the ratio of the number of abnormal wafers corresponding to the initial characteristic value to the number of wafers corresponding to the initial characteristic value to obtain a first abnormal proportion corresponding to the initial characteristic value, and/or, Determining the ratio of the number of abnormal wafers corresponding to the initial characteristic value to the total number of abnormal wafers corresponding to the characteristic type to which the initial characteristic value belongs to obtain a second abnormal proportion corresponding to the initial characteristic value, and/or, And determining the ratio of a target sum value corresponding to the initial characteristic value to the total number of wafers corresponding to the characteristic type to which the initial cha