Method and system of semiconductor fabrication fault analysis

• US 7,031,860 B2
• Filed: 09/22/2004
• Issued: 04/18/2006
• Est. Priority Date: 09/22/2004
• Status: Expired due to Fees

• Current Assignee: Taiwan Semiconductor Manufacturing Company Limited
• Original Assignee: Taiwan Semiconductor Manufacturing Company Limited
• Inventors: Liang, Shih-Tsung, Tai, Hsin-Chih
• Primary Examiner(s): Barlow, John
• Assistant Examiner(s): PRETLOW, DEMETRIUS R
• Application Number:
  US10/947,090
• Publication Number:
  US 20060064271A1
• Time in Patent Office: 573 Days
• Field of Search: 702/121
• US Class Current: 702/84
• CPC Class Codes: G05B 19/41875 characterised by quality surveillance of production
• CPC Class Codes: G05B 2219/45031 Manufacturing semiconductor wafers
• CPC Class Codes: Y02P 90/02 Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]

Abstract

A system and method for semiconductor fabrication fault analysis. The storage device stores test records. The program module receives a study lot identity, acquires suspect fabrication issues corresponding to the study lot identity, acquires a number of comparative wafer lot identities processed by the same fabrication tool and fabrication recipe for each fabrication issue, defines the comparative wafer lot identities having the same failed cluster groups as similar failed lot identities, calculates a similarity score for each similar failed lot identity, calculates a causal score according to the similarity scores for each suspect fabrication issue, and arranges the suspect fabrication issues according to causal scores thereof.

Claims

1. A method for semiconductor fabrication fault analysis, the method comprising the steps of:

receiving a wafer lot identity as a study lot identity, the wafer lot identify having at least one corresponding test record;

acquiring at least one suspect fabrication issue corresponding to the study lot identity;

acquiring a number of comparative wafer lot identities which have been processed by the same fabrication tool and fabrication recipe as the study lot identity, each of the comparative wafer lot identities having at least one corresponding test record;

comparing the test records of the study lot identity with the test records of the comparative wafer lot identities to identify comparative wafer lot identities having a failure relating to the suspect fabrication issue which is similar to a failure of the study lot identity;

calculating a similarity score for each comparative wafer lot identity having the similar failure; and

calculating a causal score according to the similarity scores of the comparative wafer lot identities having the similar failure for the suspect fabrication issue.

2. The method of claim 1, wherein the study lot identity is received from a user, a software application resident in the system, or a remote computer system via a computer network.

3. The method of claim 1, wherein the suspect fabrication issue comprises at least one selected from the group consisting of fabrication tools, fabrication recipes, and fabrication operations.

4. The method of claim 1, wherein the number of comparative lot identities prior to the study lot identity is not less than three.

5. The method of claim 1, wherein the number of comparative lot identities next to the study lot identity is not less than three.

6. The method of claim 1, further comprising a step of arranging the suspect fabrication issues according to causal scores thereof.

7. The method of claim 6, further comprising the step of displaying the ordered suspect fabrication issues on a display device.

8. The method of claim 6, further comprising the step of storing the ordered suspect fabrication issues in the storage device.

9. The method of claim 1, wherein the test record comprises a Wafer Acceptance Test (WAT) record.

10. The method of claim 1, wherein the suspect issue is acquired by querying a plurality of Quality Function Deployment (QFD) records and the test records, each QFD record storing information regarding a test parameter belonging to a cluster group and a plurality of correlation values between the test parameter and a plurality of fabrication issues, each test record storing test information regarding whether the wafer lot has failed the test parameters.

11. The method of claim 1, wherein the comparative lot identity is acquired by querying a plurality of lot processing records for each fabrication issue, each lot processing record storing fabrication information regarding a wafer lot which has been processed by a fabrication tool according to a fabrication recipe.

12. The method of claim 1, wherein the similarity score (SS) for the comparative wafer lot is calculated by a first equation if the comparative wafer lot identities corresponding to different failed test parameters from the study lot identity, the first equation being:

$\mathrm{SS}=\frac{\underset{i=1}{\overset{n}{\sum <br><br>}}<br><br><br><br>\mathrm{PSVi}}{n},$

PSVi representing the i(th) correlation value between the test parameter and the fabrication issue, and n representing a total number of PSVs.

13. The method of claim 12, wherein the similarity score (SS) for the comparative wafer lot is calculated by a second equation if the comparative wafer lot identities corresponding to any of the same test parameters as those found in the study lot, the second equation being:

$\mathrm{SS}=\frac{\underset{i=1}{\overset{n}{\sum <br><br>}}<br><br><br><br>\mathrm{PSVi}}{n}*\mathrm{\alpha <br><br>},$

PSVi representing the i(th) correlation value between the test parameter and the fabrication issue, n representing a total number of PSVs, and α

being a predetermined constant larger than 1.

14. The method of claim 13, wherein the causal score is calculated by a third equation:

$\mathrm{CS}=\left(1-\mathrm{\beta <br><br>}\right)*\left(\underset{i=1}{\overset{\mathrm{n1}}{\sum <br><br>}}<br><br><br><br>\mathrm{LENi}-1\right)/\left(\mathrm{Wt}/2\right)\bigwedge <br><br>2+\mathrm{\beta <br><br>}*\underset{j=1}{\overset{\mathrm{n2}}{\sum <br><br>}}<br><br><br><br>\left(\mathrm{SSj}*\left(\mathrm{WSSt}/\mathrm{Wt}\right)*\mathrm{WSSj}\right),$

β

being a predetermined constant between 0 and 1, n1 representing a total number of continuity failed patterns for the fabrication issue, n2 representing a total number of similar failed lots for the fabrication issue, LENi representing a length of the (i)th continuity failed pattern, Wt representing the sum of all weighted values, SSj representing the similar score of the (j)th similar failed lot identity, WSSt representing the sum of the weighted values of the similar failed lot identities, and WSSj representing the weighted value of the (j)th similar failed lot identity.

15. The method of claim 1, wherein the causal score is calculated by equation:

$\mathrm{CS}=\left(1-\mathrm{\beta <br><br>}\right)*\left(\underset{i=1}{\overset{\mathrm{n1}}{\sum <br><br>}}<br><br><br><br>\mathrm{LENi}-1\right)/\left(\mathrm{Wt}/2\right)\bigwedge <br><br>2+\mathrm{\beta <br><br>}*\underset{j=1}{\overset{\mathrm{n2}}{\sum <br><br>}}<br><br><br><br>\left(\mathrm{SSj}*\left(\mathrm{WSSt}/\mathrm{Wt}\right)*\mathrm{WSSj}\right),$

β

being a predetermined constant between 0 and 1, n1 representing a total number of continuity failed patterns for the fabrication issue, n2 representing a total number of similar failed lots for the fabrication issue, LENi representing a length of the (i)th continuity failed pattern, Wt representing the sum of all weighted values, SSj representing the similar score of the (j)th similar failed lot identity, WSSt representing the sum of the weighted values of the similar failed lot identities, and WSSj representing the weighted value of the (j)th similar failed lot identity.

16. A system for semiconductor fabrication fault analysis, comprising:

a storage device, capable of storing a plurality of test records;

a first module for receiving a wafer lot identity as a study lot identity having at least one corresponding test record;

a second module for acquiring a plurality of suspect fabrication issues corresponding to the study lot identity; and

a third module for acquiring a number of comparative wafer lot identities which have been processed by the same fabrication tool and fabrication recipe as the study lot identity, each of the comparative wafer lot identities having at least one corresponding test record, the third module further being for comparing the test records of the study lot identity with the test records of the comparative wafer lot identities to identify comparative wafer lot identities having a failure relating to the suspect fabrication issue which is similar to a failure of the study lot identity, the third module further being for calculating a similarity score for each comparative wafer lot identity having the similar failure, and calculating a causal score according to the similarity scores of the comparative wafer lot identities having the similar failure for the suspect fabrication issue.

17. The system of claim 16, wherein the study lot identity is received from one of:

a user;

a software application resident in the system;

or

a remote computer system via a computer network;

or

a combination thereof.

18. The system of claim 16, wherein the suspect fabrication issues comprise one of:

fabrication tools;

fabrication recipes;

fabrication operations;

or

a combination thereof.

19. The system of claim 16, wherein the number of comparative lot identities prior to the study lot identity is not less than three.

20. The system of claim 16, wherein the number of comparative lot identities next to the study lot identity is not less than three.

21. The system of claim 16, further comprising a fourth module for arranging the suspect fabrication issues according to causal scores thereof.

22. The system of claim 21, wherein the ordered suspect fabrication issues are:

displayed on a display device;

stored in the storage device;

or

both.

23. The system of claim 21, further comprising means for displaying the ordered suspect fabrication issues.

24. The system of claim 16, wherein the test record comprises a Wafer Acceptance Test (WAT) record.

25. The system of claim 16, wherein the suspect issue is acquired by querying a plurality of Quality Function Deployment (QFD) records and the test records, each QFD record storing information regarding a test parameter belonging to a cluster group and a plurality of correlation values between the test parameter and a plurality of fabrication issues, each test record storing test information regarding whether the wafer lot has failed the test parameters.

26. The system of claim 16, wherein the comparative lot identity by querying a plurality of lot processing records for each fabrication issue, each lot processing record storing fabrication information regarding a wafer lot which has been processed by a fabrication tool according to a fabrication recipe.

27. The system of claim 16, wherein the similarity score (SS) for the comparative wafer lot is calculated by a first equation if the comparative wafer lot identities corresponding to different failed test parameters from the study lot identity, the first equation being:

$\mathrm{SS}=\frac{\underset{i=1}{\overset{n}{\sum <br><br>}}<br><br><br><br>\mathrm{PSVi}}{n},$

PSVi representing the i(th) correlation value between the test parameter and the fabrication issue, and n representing a total number of PSVs.

28. The system of claim 27, wherein the similarity score (SS) for the comparative wafer lot is calculated by a second equation if the comparative wafer lot identities corresponding to any of the same test parameters as those found in the study lot, the second equation being:

$\mathrm{SS}=\frac{\underset{i=1}{\overset{n}{\sum <br><br>}}<br><br><br><br>\mathrm{PSVi}}{n}*\mathrm{\alpha <br><br>},$

PSVi representing the i(th) correlation value between the test parameter and the fabrication issue, n representing a total number of PSVs, and α

being a predetermined constant larger than 1.

29. The system of claim 28, wherein the causal score is calculated by a third equation:

$\mathrm{CS}=\left(1-\mathrm{\beta <br><br>}\right)*\left(\underset{i=1}{\overset{\mathrm{n1}}{\sum <br><br>}}<br><br><br><br>\mathrm{LENi}-1\right)/\left(\mathrm{Wt}/2\right)\bigwedge <br><br>2+\mathrm{\beta <br><br>}*\underset{j=1}{\overset{\mathrm{n2}}{\sum <br><br>}}<br><br><br><br>\left(\mathrm{SSj}*\left(\mathrm{WSSt}/\mathrm{Wt}\right)*\mathrm{WSSj}\right),$

β

being a predetermined constant between 0 and 1, n1 representing a total number of continuity failed patterns for the fabrication issue, n2 representing a total number of similar failed lots for the fabrication issue, LENi representing a length of the (i)th continuity failed pattern, Wt representing the sum of all weighted values, SSj representing the similar score of the (j)th similar failed lot identity, WSSt representing the sum of the weighted values of the similar failed lot identities, and WSSj representing the weighted value of the (j)th similar failed lot identity.

30. The system of claim 16, wherein the causal score is calculated by a first equation:

$\mathrm{CS}=\left(1-\mathrm{\beta <br><br>}\right)*\left(\underset{i=1}{\overset{\mathrm{n1}}{\sum <br><br>}}<br><br><br><br>\mathrm{LENi}-1\right)/\left(\mathrm{Wt}/2\right)\bigwedge <br><br>2+\mathrm{\beta <br><br>}*\underset{j=1}{\overset{\mathrm{n2}}{\sum <br><br>}}<br><br><br><br>\left(\mathrm{SSj}*\left(\mathrm{WSSt}/\mathrm{Wt}\right)*\mathrm{WSSj}\right),$

β

being a predetermined constant between 0 and 1, n1 representing a total number of continuity failed patterns for the fabrication issue, n2 representing a total number of similar failed lots for the fabrication issue, LENi representing a length of the (i)th continuity failed pattern, Wt representing the sum of all weighted values, SSj representing the similar score of the (j)th similar failed lot identity, WSSt representing the sum of the weighted values of the similar failed lot identities, and WSSj representing the weighted value of the (j)th similar failed lot identity.

31. A machine-readable storage medium for storing a computer program which when executed performs a method of semiconductor fabrication fault analysis, the method comprising the steps of:

receiving a wafer lot identity as a study lot identity, the wafer lot identify having at least one corresponding test record;

acquiring at least one suspect fabrication issue corresponding to the study lot identity;

acquiring a number of comparative wafer lot identities which have been processed by the same fabrication tool and fabrication recipe as the study lot identity, each of the comparative wafer lot identities having at least one corresponding test record;

comparing the test records of the study lot identity with the test records of the comparative wafer lot identities to identify comparative wafer lot identities having a failure relating to the suspect fabrication issue which is similar to a failure of the study lot identity;

calculating a similarity score for each comparative wafer lot identity having the similar failure; and

calculating a causal score according to the similarity scores of the comparative wafer lot identities having the similar failure for the suspect fabrication issue.

32. The machine-readable storage medium of claim 31, wherein the study lot identity is received from one of:

a user;

a software application resident in the system;

a remote computer system via a computer network;

or

a combination thereof.

33. The machine-readable storage medium of claim 31, wherein the suspect fabrication issues comprise:

fabrication tools;

fabrication recipes;

fabrication operations;

or

a combination thereof.

34. The machine-readable storage medium of claim 31, wherein the number of comparative lot identities prior to the study lot identity is not less than three.

35. The machine-readable storage medium of claim 31, wherein the number of comparative lot identities next to the study lot identity is not less than three.