Ranged fault signatures for fault diagnosis

• US 7,934,125 B2
• Filed: 12/19/2007
• Issued: 04/26/2011
• Est. Priority Date: 05/07/2006
• Status: Active Grant

• Current Assignee: Applied Materials Incorporated
• Original Assignee: Applied Materials Incorporated
• Inventors: Harvey, Jerry Lynn Jr., Schwarm, Alexander T.
• Primary Examiner(s): Beausoliel; Robert
• Assistant Examiner(s): Riad; Amine
• Application Number:
  US12/004,667
• Publication Number:
  US 20080125898A1
• Time in Patent Office: 1,224 Days
• Field of Search: 714/25
• US Class Current: 714/25
• CPC Class Codes: G05B 23/0251 Abstraction hierarchy, e.g. "complex systems", i.e. system is divided in subsystems, subsystems are monitored and results are combined to decide on status of whole system
• CPC Class Codes: G05B 23/0281 Quantitative, e.g. mathematical distance; Clustering; Neural networks; Statistical analysis

Abstract

Process data is analyzed, the process data having been generated during a manufacturing process to detect a fault. One or more process variables of the manufacturing process that contributed to the fault are determined. A relative contribution of each of the one or more process variables to the fault is determined. A fault signature having relative contribution ranges for at least one of the one or more process variables is generated, the relative contribution ranges based on the determined relative contributions.

Claims

1. A method of generating fault signatures, implemented by a computing system programmed to perform the following, comprising:

analyzing process data generated during a manufacturing process to detect a fault;

determining two or more process variables of the manufacturing process that contributed to the fault;

determining relative contributions of the two or more process variables to the fault and confidence ranges of the determined relative contributions;

assigning contribution ranks to the two or more process variables based on the relative contributions and the confidence ranges, wherein process variables having overlapping confidence ranges are assigned overlapping ranking ranges; and

generating a fault signature that includes ranked process variables, at least two of the ranked process variables having overlapping ranking ranges, the ranking ranges being ranges of consecutively numbered rankings based on the determined relative contributions.

2. The method of claim 1, further comprising:

associating the fault signature with an existing fault class.

3. The method of claim 1, further comprising:

generating a new fault class; and

associating the fault signature with the new fault class.

4. The method of claim 3, wherein generating the new fault class includes storing one or more parameters that collectively define the fault class in a fault diagnosis database.

5. The method of claim 1, wherein only process variables having a contribution that exceeds a contribution threshold are included in the fault signature.

6. The method of claim 1, wherein the confidence ranges are statistical confidence ranges.

7. The method of claim 1, further comprising:

storing the fault signature in a fault diagnosis database.

8. The method of claim 1, wherein detecting the fault comprises:

calculating statistics for the process data, wherein the process data is raw process data;

comparing the calculated statistics to corresponding statistics of a multivariate statistical model; and

detecting that a model metric for the multivariate statistical model exceeds a control limit based on the comparison.

9. A computer readable storage medium including instructions that, when executed by a processing system, cause the processing system to implement a method of generating fault signatures comprising:

analyzing process data generated during a manufacturing process to detect a fault;

determining two or more process variables of the manufacturing process that contributed to the fault;

determining relative contributions of the two or more process variables to the fault and confidence ranges of the determined relative contributions;

assigning contribution ranks to the two or more process variables based on the relative contributions and the confidence ranges, wherein process variables having overlapping confidence ranges are assigned overlapping ranking ranges; and

generating a fault signature that includes ranked process variables, at least two of the ranked process variables having overlapping ranking ranges, the ranking ranges being ranges of consecutively numbered rankings based on the determined relative contributions.

10. The computer readable storage medium of claim 9, the method further comprising:

associating the fault signature with an existing fault class.

11. The computer readable storage medium of claim 9, the method further comprising:

generating a new fault class; and

associating the fault signature with the new fault class.

12. The computer readable storage medium of claim 11, wherein generating the new fault class includes storing one or more parameters that collectively define the fault class in a fault diagnosis database.

13. The computer readable storage medium of claim 9, wherein only process variables having a contribution that exceeds a contribution threshold are included in the fault signature.

14. The computer readable storage medium of claim 9, wherein the confidence ranges are statistical confidence ranges.

15. The computer readable storage medium of claim 9, the method further comprising:

storing the fault signature in a fault diagnosis database.

16. The computer readable storage medium of claim 9, wherein detecting the fault comprises:

calculating statistics for the process data, wherein the process data is raw process data;

comparing the calculated statistics to corresponding statistics of a multivariate statistical model; and

detecting that a model metric for the multivariate statistical model exceeds a control limit based on the comparison.

17. A system for monitoring processes of a manufacturing apparatus, comprising:

a processor to execute a fault detector to receive process data, and to detect a fault by calculating statistics for the process data, comparing the calculated statistics to corresponding statistics of a multivariate statistical model, and detecting that a model metric of the multivariate statistical model exceeds a control limit based on the comparison; and

the processor to execute a fault diagnoser coupled with the fault detector, the fault diagnoser to determine two or more process variables of the manufacturing process that contributed to the fault, determine relative contributions of the two or more process variables to the fault and confidence ranges of the determined relative contributions, assign contribution ranks to the two or more process variables based on the relative contributions and the confidence ranges, wherein process variables having overlapping confidence ranges are assigned overlapping ranking ranges, and generate a fault signature that includes ranked process variables, at least two of the ranked process variables having overlapping ranking ranges, the ranking ranges being ranges of consecutively numbered rankings based on the determined relative contributions.

18. The process monitoring system of claim 17, wherein the fault diagnoser is further configured to associate the fault signature with an existing fault class.

19. The process monitoring system of claim 17, wherein the fault diagnoser is further configured to generate a new fault class, associate the fault signature with the new fault class.

20. The process monitoring system of claim 19, further comprising:

a database to store one or more of the fault signature and the new fault class.

21. The process monitoring system of claim 17, wherein the confidence ranges are statistical confidence ranges.

22. A method of dividing fault signatures, implemented by a computing system programmed to perform the following, comprising:

analyzing process data generated during a manufacturing process to detect a fault;

determining two or more process variables that contributed to the fault;

determining a relative contribution of each of the two or more process variables;

determining a fault signature that matches the fault by comparing relative contributions of each of the two or more process variables to relative contribution ranges of a plurality of fault signatures, a fault signature matching the fault if the relative contributions of the two or more process variables to the fault match relative contribution ranges of the two or more process variables in the fault signature;

dividing the determined fault signature into two distinct fault signatures, each of the two distinct fault signatures having a different subset of the fault contribution ranking ranges of the process variables in the determined fault signature;

adding a new fault class; and

associating the new fault class with one of the distinct fault signatures.