Method of data compression using principal components analysis

• US 6,675,137 B1
• Filed: 01/26/2000
• Issued: 01/06/2004
• Est. Priority Date: 09/08/1999
• Status: Expired due to Fees

• Current Assignee: Advanced Micro Devices, Inc.
• Original Assignee: Advanced Micro Devices, Inc.
• Inventors: Yue, Hongyu, Toprac, Anthony John
• Primary Examiner(s): Frejd, Russell
• Assistant Examiner(s): THOMSON, WILLIAM D
• Application Number:
  US09/491,882
• Time in Patent Office: 1,441 Days
• Field of Search: 703/2, 703/6, 703/13, 703/11, 703/12, 703/9, 702/179, 702/180, 702/181, 702/194, 700/110, 700/121, 356/300
• US Class Current: 703/2
• CPC Class Codes: A61K 39/00 Medicinal preparations containing antigens or antibodies materials for immunoassay G01N33/53
• CPC Class Codes: C07K 14/195 from bacteria
• CPC Class Codes: G01N 21/71 thermally excited
• CPC Class Codes: G01N 21/73 using plasma burners or torches
• CPC Class Codes: G05B 2219/32187 Correlation between controlling parameters for influence on quality parameters
• CPC Class Codes: G05B 2219/45031 Manufacturing semiconductor wafers
• CPC Class Codes: H01J 2237/334 Etching
• CPC Class Codes: H01J 2237/3341 Reactive etching
• CPC Class Codes: H01J 37/32082 Radio frequency generated discharge H01J37/32357, H01J37/32366, H01J37/32394 and H01J37/32403 take precedence
• CPC Class Codes: H01J 37/32935 Monitoring and controlling tubes by information coming from the object and/or discharge
• CPC Class Codes: H01J 37/32963 End-point detection
• CPC Class Codes: H01J 37/32972 Spectral analysis
• CPC Class Codes: H01J 37/3299 Feedback systems
• CPC Class Codes: H01L 21/67069 for drying etching
• CPC Class Codes: H01L 21/67253 Process monitoring, e.g. flow or thickness monitoring
• CPC Class Codes: H01L 22/20 Sequence of activities consisting of a plurality of measurements, corrections, marking or sorting steps
• CPC Class Codes: H01L 22/26 Acting in response to an ongoing measurement without interruption of processing, e.g. endpoint detection, in-situ thickness measurement endpoint detection arrangements in CMP apparatus B24B37/013, in discharge apparatus H01J37/32

Abstract

A method is provided for compressing data. The method includes collecting data representative of a process. The method further includes scaling at least a portion of the collected data to generate mean values and mean-scaled values for the collected data. The method also includes calculating Scores from the mean-scaled values for the collected data using at most first, second, third and fourth Principal Components derived from a model using archived data sets and saving the Scores and the mean values.

Claims

1. A method for compressing data, the method comprising:

collecting data, representative of a process;

scaling at least a portion of the collected data to generate mean values and mean-scaled values for the collected data;

calculating Scores from the mean-scaled values for the collected data using at most first, second, third and fourth Principal Components derived from a model using archived data sets; and

saving the Scores and the mean values.

2. The method of claim 1, wherein calculating the Scores from the collected data using at most the first, second, third and fourth Principal Components derived from the model includes building the model.

3. The method of claim 2, wherein building:

the model includes saving Loadings corresponding to, at most first, second, third and fourth Principal Components derived from a plurality of data representative of previous processes.

4. The method of claim 1 further comprising:

reconstructing the data representative of the process using the Scores and the mean values.

5. The method of claim 2 further comprising:

reconstructing the data representative of the process using the Scores and the mean values.

6. The method of claim 3 further comprising:

reconstructing the data representative of the process using the Scores and the mean values.

7. The method of claim 1, wherein scaling the at least the portion of the collected data to generate the mean values and the mean-scaled values for the collected data includes generating the mean values and the mean-scaled values for the collected data from the model built using a plurality of data representative of previous processes.

8. The method of claim 3, wherein saving Loadings corresponding to at most the first, second, third and fourth Principal Components includes using an eigenanalysis method.

9. The method of claim 3, wherein saving Loadings corresponding to at most the first, second, third and fourth Principal Components includes using a singular value decomposition method.

10. The method of claim 3, wherein saving Loadings corresponding to at most the first, second, third and fourth Principal Components includes using a power method.

11. A computer-readable, program storage device encoded with instructions that, when executed by a computer, perform a method, the method comprising:

collecting data representative of a process;

scaling at least a portion of the collected data to generate mean values and mean-scaled values for the collected data;

calculating Scores from the mean-scaled values for the collected data using at most first, second, third and fourth Principal Components derived from a model using archived data sets; and

saving the Scores and the mean values.

12. The computer-readable, program storage device of claim 11, wherein calculating the Scores from the collected data using at most the first, second, third and fourth Principal Components derived from the model includes building the model.

13. The computer-readable, program storage device of claim 12, wherein building the model includes saving Loadings corresponding to at most first, second, third and fourth Principal Components derived from a plurality of data representative of previous processes.

14. The computer-readable, program storage device of claim 11 further comprising:

reconstructing the data representative of the process using the Scores and the mean values.

15. The computer-readable, program storage device of claim 12 further comprising:

reconstructing the data representative of the process using the Scores and the mean values.

16. The computer-readable, program storage device of claim 13 further comprising:

reconstructing the data representative of the process using the Scores and the mean values.

17. The computer-readable, program storage device of claim 11, wherein scaling the at least the portion of the collected data to generate the mean values and the mean-scaled values for the collected data includes generating the mean values and the mean-scaled values for the collected data from the model built using a plurality of data representative of previous processes.

18. The computer-readable, program storage device of claim 13, wherein saving Loadings corresponding to at most the first, second, third and fourth Principal Components includes using an eigenanalysis method.

19. The computer-readable, program storage device of claim 13, wherein saving Loadings corresponding to at most the first, second, third and fourth Principal Components includes using a singular value decomposition method.

20. The computer-readable, program storage device of claim 13, wherein saving Loadings corresponding to at most the first, second, third and fourth Principal Components includes using a power method.

21. A computer programmed to perform a method, the method comprising:

collecting data representative of a process;

scaling at least a portion of the collected data to generate mean values and mean-scaled values for the collected data;

calculating Scores from the mean-scaled values for the collected data using at most first, second, third and fourth Principal Components derived from a model using archived data sets; and

saving the Scores and the mean values.

22. The computer of claim 21, wherein calculating the Scores from the collected data using at most the first, second, third and fourth Principal Components derived from the model includes building the model.

23. The computer of claim 22, wherein building the model includes saving Loadings corresponding to at most first, second, third and fourth Principal Components derived from a plurality of data representative of previous processes.

24. The computer of claim 21 further comprising:

reconstructing the data representative of the process using the Scores and the mean values.

25. The computer of claim 22 further comprising:

reconstructing the data representative of the process using the Scores and the mean values.

26. The computer of claim 23 further comprising:

reconstructing the data representative of the process using the Scores and the mean values.

27. The computer of claim 21, wherein scaling the at least the portion of the collected data to generate the mean values and the mean-scaled values for the collected data includes generating the mean values and the mean-scaled values for the collected data from the model built using a plurality of data representative of previous processes.

28. The computer of claim 23, wherein saving Loadings corresponding to at most the first, second, third and fourth Principal Components includes using an eigenanalysis method.

29. The computer of claim 23, wherein saving Loadings corresponding to at most the first, second, third and fourth Principal Components includes using a singular value decomposition method.

30. The computer of claim 23, wherein saving Loadings corresponding to at most the first, second, third and fourth Principal Components includes using a power method.

31. A method of compressing optical emission spectroscopy data collected during an etching process, comprising:

scaling at least a portion of the collected data to generate mean values and mean-scaled values for the collected data;

calculating approximate Scores from the mean-scaled values for the collected data using a plurality of principal components derived from archived optical emission spectroscopy data sets; and

storing the approximate Scores and mean values as compressed optical emission spectroscopy data.

32. The method of claim 31, wherein calculating the approximate Scores from the mean-scaled values for the collected data using the plurality of principal components comprises calculating the approximate Scores using at most first, second, third and fourth principal components derived from a model.

33. The method of claim 32, wherein calculating the approximate Scores using at most first, second, third and fourth principal components derived from the model comprises determining Loadings corresponding to at most first, second, third and fourth Principal Components derived from the archived optical emission spectroscopy data sets.

34. The method of claim 33, wherein determining the Loadings comprises determining the Loadings using at least one of an eigenanalysis method, a singular value decomposition, and a power method.

35. The method of claim 31, further comprising reconstructing the data representative of the semiconductor etching process using the compressed optical emission spectroscopy data.

36. The method of claim 31, wherein storing the approximate Scores and mean values as the compressed optical emission spectroscopy data comprises storing the compressed optical emission spectroscopy data having a compression ratio of at least approximately 100:

1, relative to the collected data representative of the etching process.

37. A computer-readable, program storage device encoded with instructions that, when executed by a computer, perform a method, the method comprising:

scaling at least a portion of data collected during an etching process to generate mean values and mean-scaled values for the collected data;

calculating approximate Scores from the mean-scaled values for the collected data using a plurality of principal components derived from archived optical emission spectroscopy data sets; and

storing the approximate Scores and mean values as compressed optical emission spectroscopy data.

38. The device of claim 37, wherein calculating the approximate Scores from the mean-scaled values for the collected data using the plurality of principal components comprises calculating the approximate Scores using at most first, second, third and fourth principal components derived from a model based on the archived optical emission spectroscopy data sets.

39. The device of claim 38, wherein the method further comprises reconstructing the data representative of the semiconductor etching process using the compressed optical emission spectroscopy data.

40. The device of claim 38, wherein storing the approximate Scores and mean values as the compressed optical emission spectroscopy data comprises storing the compressed optical emission spectroscopy data having a compression ratio of at least approximately 100:

1, relative to the collected data representative of the etching process.