Monitor of plasma processes with multivariate statistical analysis of plasma emission spectra

• US 6,153,115 A
• Filed: 10/23/1997
• Issued: 11/28/2000
• Est. Priority Date: 10/23/1997
• Status: Expired due to Fees

• Current Assignee: Massachusetts Institute of Technology
• Original Assignee: Massachusetts Institute of Technology
• Inventors: Le, Minh, Smith, Taber H., Chen, Kuang Han, Boning, Duane S., Sawin, Herbert H.
• Primary Examiner(s): Gulakowski, Randy
• Assistant Examiner(s): OLSEN, ALLAN W
• Application Number:
  US08/956,575
• Time in Patent Office: 1,132 Days
• Field of Search: 156/345, 438/9, 438/16, 216/60, 216/67, 216/72, 118/663, 118/695, 118/696, 118/698, 118/708, 118/712
• US Class Current: 216/60
• CPC Class Codes: G01N 21/73 using plasma burners or torches
• CPC Class Codes: H01J 37/32935 Monitoring and controlling tubes by information coming from the object and/or discharge

Abstract

Plasma process analysis techniques are provided. The intensity of each of a number, P, of a plurality of radiation wavelengths that are emitted from a plasma process are monitored as the process proceeds. Indications of P-dimensional correlations between the intensities of the P monitored wavelengths are produced as the process proceeds. Then the produced correlation indications are compared with a prespecified correlation indication generated based on historical conditions for the plasma process, to determine the status condition of the process as the process proceeds. With this technique, the use of a priori, expected, specific templates is not required for evaluating radiation emission data during a plasma process. Instead the techniques investigate and discover the multiple complex correlations that form between various radiation emission wavelengths during a plasma process, and do not impose an expectation for a specific correlation or trend between the various wavelengths. The discovered correlations found to exist between the radiation wavelengths are then employed for monitoring a plasma process based on the discovered correlations. The analysis techniques enables evaluation of interactions occurring across the entire spectrum of detected radiation emission wavelengths, and thus can accomplish detection and analysis of changes in a given plasma process due to shifts in the electrical and physical process environment as well as changes in a given process due to procession through stages of the process.

Claims

1. A method for monitoring the status condition of a plasma process as the plasma process proceeds, the method comprising the steps of:

monitoring the intensity of each of a number, P, of a plurality of radiation wavelengths that are emitted from the plasma process as the process proceeds;

producing indications of P-dimensional correlations between the intensities of the P monitored wavelengths as the process proceeds; and

comparing the produced correlation indications with a prespecified correlation indication generated based on historical conditions for the plasma process, to determine the status condition of the process as the process proceeds.

2. The method of claim 1 wherein the emitted radiation wavelengths that are monitored comprise radiation wavelengths between about 200 nm and about 800 nm.

3. The method of claim 1 wherein the number, P, of monitored radiation wavelengths is at least 2.

4. The method of claim 3 wherein the number, P, of monitored radiation wavelengths is at least about 10.

5. The method of claim 4 wherein the number, P, of monitored radiation wavelengths is at least about 100.

6. The method of claim 5 wherein the number, P, of monitored radiation wavelengths is at least about 500.

7. The method of claim 1 wherein the plasma process for which the status condition is determined is a plasma etch process.

8. The method of claim 7 wherein the plasma process for which the status condition is determined is a main etch stage of a plasma etch process.

9. The method of claim 8 wherein the step of comparing produced correlation indications with a prespecified correlation indication comprises a comparison to determine if a condition of etch endpoint has been reached.

10. The method of claim 9 wherein the plasma process for which the status condition is determined is a plasma etch process employing reactive plasma species for plasma etching oxide.

11. The method of claim 1 further comprising a last step of controlling process parameters of the plasma process in response to the process status condition determination.

12. The method of claim 1 further comprising:

a first step of carrying out a number, K, of historical plasma processes during each of which the intensity of each of the P radiation wavelengths are monitored; and

a second step of producing the prespecified correlation indication based on the intensities of the P monitored wavelengths for the process conditions of the K historical processes carried out.

13. The method of claim 1 further comprising a step of producing the prespecified correlation indication during the plasma process prior to the step of comparing the produced correlation indications with the prespecified correlation indication.

14. The method of claim 1 further comprising a step of producing, for each of the monitored intensities of the P radiation wavelengths, a digital intensity value signal to be used for producing the P-dimensional correlation indications as the plasma process proceeds.

15. The method of claim 1 further comprising a step of updating the prespecified correlation indication based on the produced correlation indication.

16. The method of claim 15 wherein the updating step comprises applying an exponentially-weighted moving average, based on historical process condition drift, to the prespecified correlation indication.

17. The method of claim 1 wherein the step of monitoring the intensities of the P radiation wavelengths comprises the steps of:

monitoring the intensity of each of a number, J, of radiation wavelengths, where J>

P; and

filtering the J monitored wavelength intensities to select those P wavelengths that each meet a prespecified intensity criterion.

18. The method of claim 1 wherein the step of producing indications of the P-dimensional intensity correlations comprises producing P-dimensional intensity correlation values, and wherein the step of comparing the produced correlation indications with a prespecified correlation indication comprises determining if each produced P-dimensional intensity correlation value is within a prespecified variation of intensity correlation values that is characteristic of historical process conditions for the plasma process, a produced intensity correlation value determined to not be within the prespecified variation of correlation values indicating a change in the status condition of the process.

19. The method of claim 18 wherein the step of determining if each produced P-dimensional intensity correlation value is within a prespecified variation of correlation values comprises, for a given intensity correlation value:

producing a measure of the difference between the given intensity correlation value and a mean of intensity correlation values characteristic of historical process conditions, and

comparing the difference measure with a control limit that is prespecified for a maximum allowable variation of correlation values and generated based on historical process conditions.

20. The method of claim 19 wherein the step of determining if each produced P-dimensional intensity correlation value is within a prespecified variation of correlation values further comprises indicating a change in the status condition of the plasma process after a minimum number, c, of difference measures are compared with the prespecified control limit and found to indicate that the corresponding produced intensity correlation values are not within the prespecified correlation value variation.

21. The method of claim 19 further comprising:

a first step of carrying out a number, K, of historical plasma processes during each of which the intensity of a number, P, of a plurality of radiation wavelengths is monitored over a duration of time, T; and

a second step of computing the mean, M(P), of the radiation wavelength intensity values for the P wavelengths over the time duration, T, for the K historical processes, and computing an inverse covariance, S, of a matrix of the radiation wavelength intensity values for the P wavelengths over the time duration, T, for the K historical processes;

and wherein the step of producing a measure of the difference between a given intensity correlation value and a mean of intensity correlation values comprises producing a difference measure given as;

space="preserve" listing-type="equation">T.sup.2 (t)=(D(P,t)-M(P))*S*(D(P,t)-M(P)).sup.T,

where t is the time during the plasma process to which the given intensity correlation value corresponds;

D(P, t) is a matrix of the P radiation intensities monitored at time t;

M(P) is the mean of the radiation wavelength intensity values for the P wavelengths over the time duration, T, for the K historical processes;

S is the inverse covariance of the matrix of the radiation wavelength intensity values for the P wavelengths over the time duration, T, for the K historical processes; and

the superscript T denotes transposition.

22. The method of claim 21 wherein the control limit is based on a confidence value and a distribution of the P radiation intensities monitored over the time duration, T, for the K historical processes.

23. The method of claim 22 wherein the control limit is based on a chi-squared distribution of the P radiation intensities monitored over the time duration, T, for the K historical processes.

24. The method of claim 1 wherein the step of producing indications of the P-dimensional intensity correlations comprises producing indications of maximum variations in correlations between the intensities of the P monitored radiation wavelengths during the plasma process as the plasma process proceeds, and wherein the step of comparing the produced correlation indications with a prespecified correlation indication comprises comparing the produced indications of maximum variations in correlations with prespecified maximum variation indications that are characteristic of historical process conditions, a produced indication of maximum correlation variation determined not to match a prespecified maximum variation indication representing a change in the status condition of the plasma process.

25. The method of claim 1 wherein the step of producing indications of the P-dimensional intensity correlations comprises producing indications of maximum variations in correlations between the intensities of the P monitored radiation wavelengths during the plasma process as the plasma process proceeds, and wherein the step of comparing the produced correlation indications wit h a prespecified correlation indication comprises comparing th e produced indications of maximum variations in correlations with prespecified maximum variation indications that are characteristic of historical process conditions, a produced indication of maximum ncorrelatio n variation determined to match a prespecified maximum variation indication representing a change in the status condition of the plasma process.

26. The method of claim 24 wherein the step of producing indications of maximum variations in correlations between the intensities of the P monitored radiation wavelengths comprises computing a number, S, of principal components of the variation in wavelength intensity correlation for the P radiation wavelengths monitored, each computed principal component corresponding to one of the P radiation wavelengths and each computed principal component being characterized by a computed eigenvalue magnitude, and wherein the step of comparing the produced indications of maximum variations in correlations with prespecified maximum variation indications that are characteristic of historical process conditions comprises comparing each of the number, S, of computed eigenvalue magnitudes with a corresponding prespecified eigenvalue magnitude that refers to a common one of the P radiation wavelengths.

27. The method of claim 24 wherein the step of producing indications of maximum variations in correlations between the intensities of the P monitored radiation wavelengths comprises computing a number, S, of principal components of the variation in wavelength intensity correlation for the P radiation wavelengths monitored, each computed principal component corresponding to one of the P radiation wavelengths and each computed principal component being characterized by a computed eigenvector orientation, and wherein the step of comparing the produced indications of maximum variations in correlations with prespecified maximum variation indications that are characteristic of historical process conditions comprises comparing each of the number, S, of computed eigenvector orientations with a corresponding prespecified eigenvector orientation that refers to a common one of the P radiation wavelengths.

28. The method of claim 27 further comprising the first steps of:

carrying out a number, K, of historical plasma processes during each of which the intensity of a number, P, of a plurality of radiation wavelengths is monitored over a duration of time, T;

computing a number, P, of historical principal components of the variation in wavelength intensity correlation for the P radiation wavelengths monitored over the time duration, T, for each of the K historical processes, each historical principal component being characterized by a historical orientation and a historical eigenvalue magnitude;

ranking the historical principal components for each of the K historical processes, in descending order based on historical eigenvalue magnitude;

retaining the first S historical principal components for each of the K historical processes based on a prespecified percentage in variation of intensity correlations that can be defined by the first S historical principal components for each of the K historical processes;

computing an average historical principal component for each of the S historical principal components across the K historical processes; and

computing an angle between each historical principal component eigenvector orientation and the eigenvector orientation of the corresponding computed average historical principal component; and

and wherein the step of comparing each of the number, S, of computed eigenvector orientations with a corresponding prespecified eigenvector orientation that refers to a common one of the P radiation wavelengths comprises comparing each of the number, S, of the computed eigenvector orientations with the historical eigenvector orientation of a corresponding one of the S computed average historical principal components.

29. The method of claim 28 further comprising a step of setting a prespecified threshold angle based on a prespecified percentage in variation of the computed historical eigenvector orientation angles for the K historical processes;

and further comprising the steps of;

computing an angle between each of the computed eigenvector orientations of the number, S, of the computed principal components and the historical eigenvector orientation of a corresponding one of the S computed average historical principal components; and

comparing each of the computed angles with the prespecified threshold angle, a computed angle determined to be less than the prespecified threshold angle indicating a change in the status condition of the plasma process.

30. The method of claim 1 wherein the step of producing indications of the P-dimensional intensity correlations comprises producing P-dimensional intensity correlation values, and wherein the step of comparing the produced correlation indications with a prespecified correlation indication comprises the steps of:

determining if each produced P-dimensional intensity correlation value is within a prespecified variation of intensity correlation values that is characteristic of historical process conditions for the plasma process;

for a produced intensity correlation value determined to not be within the prespecified variation of correlation values, producing indications of maximum variations in correlations between the intensities of the P monitored radiation wavelengths during the plasma process as the plasma process proceeds;

comparing the produced indications of maximum variations in correlations with prespecified maximum variation indications that are characteristic of historical process conditions, a produced indication of maximum correlation variation determined not to match a prespecified maximum variation indication representing a change in the status condition of the plasma process.

31. The method of claim 1 wherein the step of producing indications of the P-dimensional intensity correlations comprises producing P-dimensional intensity correlation values, and wherein the step of comparing the produced correlation indications with a prespecified correlation indication comprises the steps of:

determining if each produced P-dimensional intensity correlation value is within a prespecified variation of intensity correlation values that is characteristic of historical process conditions for the plasma process;

for a produced intensity correlation value determined to be within the prespecified variation of correlation values, producing indications of maximum variations in correlations between the intensities of the P monitored radiation wavelengths during the plasma process as the plasma process proceeds;

comparing the produced indications of maximum variations in correlations with prespecified maximum variation indications that are characteristic of historical process conditions, a produced indication of maximum correlation variation determined not to match a prespecified maximum variation indication representing a change in the status condition of the plasma process.

32. Apparatus for monitoring the status condition of a plasma process as the plasma process proceeds, comprising:

means for monitoring the intensity of each of a number, P, of a plurality of radiation wavelengths that are emitted from the plasma process as the process proceeds;

means for producing indications of P-dimensional correlations between the intensities of the P monitored wavelengths as the process proceeds; and

means for comparing the produced correlation indications with a prespecified correlation indication generated based on historical conditions for the plasma process, to determine the status condition of the process as the process proceeds.

33. The apparatus of claim 32 wherein the means for comparing comprises means for comparing the produced correlation indications with a prespecified correlation indication generated based on historical conditions for the plasma process, to determine if a condition of etch endpoint has been reached during a plasma etch process as the process proceeds.

34. The apparatus of claim 32 further comprising means for controlling parameters of the plasma process in response to process status condition determination.

35. The apparatus of claim 32 further comprising:

means for carrying out a number, K, of historical plasma processes during each of which the intensity of each of the P radiation wavelengths are monitored; and

means for generating the prespecified correlation indication based on the intensities of the P monitored wavelengths for the process conditions of the K historical processes carried out.

36. The apparatus of claim 32 further comprising means for updating the prespecified correlation indication based on the produced correlation indication.

37. The apparatus of claim 32 further comprising:

means for monitoring the intensity of each of a number, J, of radiation wavelengths emitted from the plasma process as the process proceeds, where J>

P; and

means for filtering the J monitored wavelength intensities to select those P wavelengths that each meet a prespecified intensity criterion.

38. The apparatus of claim 32 wherein the means for monitoring the radiation wavelength intensities comprises means for monitoring the radiation wavelength intensities between about 200 nm and about 800 nm.

39. The apparatus of claim 32 wherein the means for producing indications of the P-dimensional intensity correlations comprises means for producing P-dimensional intensity correlation values, and wherein the means for comparing the produced correlation indications with a prespecified correlation indication comprises means for determining if each produced P-dimensional intensity correlation value is within a prespecified variation of intensity correlation values that is characteristic of historical process conditions for the plasma process, a produced intensity correlation value determined to not be within the prespecified variation of correlation values indicating a change in the status condition of the process.

40. The apparatus of claim 32 wherein the means for producing indications of the P-dimensional intensity correlations comprises means for producing indications of maximum variations in correlations between the intensities of the P monitored radiation wavelengths during the plasma process as the plasma process proceeds, and wherein the means for comparing the produced correlation indications with a prespecified correlation indication comprises means for comparing the produced indications of maximum variations in correlations with prespecified maximum variation indications that are characteristic of historical process conditions, a produced indication of maximum correlation variation determined not to match a prespecified maximum variation indication representing a change in the status condition of the plasma process.