PREDICTION METHOD AND APPARATUS FOR SUBSTRATE PROCESSING APPARATUS

Abstract

A prediction method for a substrate processing apparatus is to predict processing results from operation data on the substrate processing apparatus during a procedure for processing a target processing substrate in a processing chamber of the substrate processing apparatus. The method includes the steps of: collecting operation data obtained; and obtaining a moving average of a preset number of sets of data using the processing result data collected at the data collection step. The method further includes the steps of: performing multivariate analysis using the operation data collected at the data collection step and the moving average processing result data obtained at the moving average processing step; and predicting processing results using operation data obtained when a target processing substrate, other than the target processing substrate used to obtain the correlation at the analysis step, is processed on a basis of the correlation.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

The above and other objects and features of the present invention will become apparent from the following description of embodiments given in conjunction with the accompanying drawings, in which:

FIG. 1 is a sectional view of a plasma processing apparatus according to an embodiment of the present invention;

FIG. 2 is a block diagram showing a detailed example of a prediction apparatus in the plasma processing apparatus;

FIG. 3 is a diagram showing a detailed example of a film structure on which etching processing is performed in the plasma processing apparatus;

FIG. 4 is a diagram showing a detailed example of data on a critical dimension (CD) shift amount, obtained before moving average processing is performed;

FIG. 5 is a diagram showing CD shift amount values obtained before moving average processing is performed;

FIG. 6 is a diagram showing data about CD shift amount values when the number of pieces of data required to obtain each moving average is set to 3, and moving average processing is performed;

FIG. 7 is a diagram showing data about CD shift amount values when the number of pieces of data required to obtain each moving average is set to 5, and moving average processing is performed;

FIG. 8 is a diagram showing data about CD shift amount values when the number of pieces of data required to obtain each moving average is set to 10, and moving average processing is performed;

FIG. 9A is a diagram showing the comparison of a predicted value, obtained from a model created from CD shift amount values, on which moving average processing was not performed, with a measured value;

FIG. 9B is a diagram showing a correlation between the predicted value and the measured value of FIG. 9A;

FIG. 10A is a diagram showing the comparison of a predicted value, obtained from a model created from CD shift amount values, on which moving average processing has been performed based on the basic number of pieces of data “3” with a measured value;

FIG. 10B is a diagram showing a correlation between the predicted value and the measured value of FIG. 10A;

FIG. 11A is a diagram showing the comparison of a predicted value, obtained from a model created from CD shift amount values, on which moving average processing is performed based on the basic number of pieces of data “5”, with a measured value;

FIG. 11B is a diagram showing a correlation between the predicted value and the measured value of FIG. 11A;

FIG. 12A is a diagram showing the comparison of a predicted value, obtained from a model created from CD shift amount values, on which moving average processing is performed based on the basic number of pieces of data “10”, with a measured value;

FIG. 12B is a diagram showing a correlation between the predicted value and the measured value of FIG. 12A; and

FIG. 13 is a diagram showing an example of the management of a CD shift amount using the predicted value of FIG. 11A.

Claims

1. A prediction method for a substrate processing apparatus, the prediction method predicting processing results from operation data on the substrate processing apparatus during a procedure for processing a target processing substrate in a processing chamber of the substrate processing apparatus, the prediction method comprising: a data collection step of collecting operation data obtained whenever the target processing substrate is processed, and processing result data obtained by measuring status of the target processing substrate;a moving average processing step of obtaining a moving average of a preset number of sets of data using the processing result data collected at the data collection step, thus obtaining moving average processing result data;an analysis step of performing multivariate analysis using the operation data collected at the data collection step and the moving average processing result data obtained at the moving average processing step, thus obtaining a correlation between the operation data and the moving average processing result data; anda prediction step of predicting processing results using operation data obtained when a target processing substrate, other than the target processing substrate used to obtain the correlation at the analysis step, is processed on a basis of the correlation.

2. The prediction method of claim 1, wherein the moving average processing step is performed such that: when processing result data, existing before and after maintenance of the substrate processing apparatus, is included in the processing result data, the processing result data is divided into groups for respective sections defined by the maintenance; andfor each group, moving averages of a preset number of pieces of data are obtained using only processing result data belonging to the group, thus obtaining moving average processing result data.

3. The prediction method of claim 2, wherein the moving average processing step is performed such that, for each group, moving average processing result data corresponding to considered processing result data is calculated while the considered processing result data is shifted by one piece.

4. The prediction method of claim 3, wherein the moving average processing step is performed such that: until a number of pieces of data preceding the considered processing result data reaches the preset number of pieces of data, an average value is obtained using all pieces of processing result data preceding the considered processing result data, and the average value is taken as moving average processing result data corresponding to the considered processing result data; andwhen the number of pieces of data preceding the considered processing result data reaches the preset number of pieces of data or more, an average value is obtained using only processing result data that immediately precedes the considered processing result data and corresponds to the preset number of pieces of data, and the obtained average value is taken as moving average processing result data corresponding to the considered processing result data.

5. The prediction method of claim 4, wherein the moving average processing step is performed so that the number of pieces of data required to obtain each moving average is preset for each group.

6. The prediction method of claim 5, wherein the number of pieces of data required to obtain each moving average is preset according to the number of pieces of processing result data belonging to each group.

7. The prediction method of claim 5, wherein the number of pieces of data required to obtain each moving average is one falling within a range between 2 and 10.

8. The prediction method of claim 1, wherein the prediction method is operated such that: a management value range is set to have a certain width based on a target value of processing results so as to manage the processing results, an upper prediction error range is set to have a certain width based on an upper limit of the management value range, and a lower prediction error range is set to have a certain width based on a lower limit of the management value range;if a predicted value for status of the target processing substrate obtained at the prediction step is within an allowable prediction range when a range from a lower limit of the upper prediction error range to an upper limit of the lower prediction error range is set to the allowable prediction range, status of the target processing substrate is determined to be normal;if the predicted value is included in the upper prediction error range or the lower prediction error range even though the predicted value departs from the allowable prediction range, the status of the target processing substrate is determined based on a measured value obtained by measuring the target processing substrate; andif the predicted value departs from the allowable prediction range, and also departs from the upper prediction error range and the lower prediction error range, the status of the target processing substrate is determined to be abnormal.

9. The prediction method of claim 8, wherein each prediction error range is set according to a standard error between the predicted value and the measured value.

10. The prediction method of claim 8, wherein the predicted value is a processing dimension of the target processing substrate.

11. The prediction method of claim 1, wherein the processing result data is a processing dimension of the target processing substrate.

12. The prediction method of claim 1, wherein the operation data is electronic data obtained from a plurality of detectors provided in the substrate processing apparatus.

13. The prediction method of claim 1, wherein the analysis step uses Partial Least Squares (PLS) as the multivariate analysis.

14. A prediction apparatus for a substrate processing apparatus, the prediction apparatus predicting processing results from operation data on the substrate processing apparatus during a procedure for processing a target processing substrate in a processing chamber of the substrate processing apparatus, the prediction apparatus comprising: data collection unit for collecting operation data obtained whenever the target processing substrate is processed, and processing result data obtained by measuring status of the target processing substrate;moving average processing unit for obtaining a moving average of a preset number of pieces of data using the processing result data collected by the data collection unit, thus obtaining moving average processing result data;analysis unit for performing multivariate analysis using the operation data collected by the data collection unit and the moving average processing result data obtained by the moving average processing unit, thus obtaining a correlation between the operation data and the moving average processing result data; andprediction unit for predicting processing results using operation data obtained when a target processing substrate, other than the target processing substrate used to obtain the correlation by the analysis unit, is processed on a basis of the correlation.

15. The prediction apparatus of claim 14, wherein the moving average processing unit is operated such that: when processing result data, existing before and after maintenance of the substrate processing apparatus, is included in the processing result data, the processing result data is divided into groups for respective sections defined by the maintenance; andfor each group, moving averages of a preset number of pieces of data are obtained using only processing result data belonging to the group, thus obtaining moving average processing result data.

16. The prediction apparatus of claim 15, wherein the moving average processing unit calculates moving average processing result data corresponding to considered processing result data for each group while shifting the considered processing result data by one piece.

17. The prediction apparatus of claim 16, wherein the moving average processing unit is operated such that: until a number of pieces of data preceding the considered processing result data reaches the preset number of pieces of data, an average value is obtained using all pieces of processing result data preceding the considered processing result data, and the average value is taken as moving average processing result data corresponding to the considered processing result data; andwhen the number of pieces of data preceding the considered processing result data reaches the preset number of pieces of data or more, an average value is obtained using only processing result data that immediately precedes the considered processing result data and corresponds to the preset number of pieces of data, and the obtained average value is taken as moving average processing result data corresponding to the considered processing result data.

18. The prediction apparatus of claim 17, wherein the number of pieces of data required to obtain each moving average is set in advance for each group.

19. The prediction apparatus of claim 18, wherein the number of pieces of data required to obtain each moving average is set in advance according to the number of pieces of processing result data belonging to each group.

20. The prediction apparatus of claim 18, wherein the number of pieces of data required to obtain each moving average is one falling within a range between 2 and 10.

21. The prediction apparatus of claim 14, wherein the prediction apparatus is operated such that: a management value range is set to have a certain width based on a target value of processing results so as to manage the processing results, an upper prediction error range is set to have a certain width based on an upper limit of the management value range, and a lower prediction error range is set to have a certain width based on a lower limit of the management value range;if a predicted value for status of the target processing substrate obtained at the prediction step is within an allowable prediction range when a range from a lower limit of the upper prediction error range to an upper limit of the lower prediction error range is set to the allowable prediction range, status of the target processing substrate is determined to be normal;if the predicted value is included in the upper prediction error range or the lower prediction error range even though the predicted value departs from the allowable prediction range, the status of the target processing substrate is determined based on a measured value obtained by measuring the target processing substrate; andif the predicted value departs from the allowable prediction range, and also departs from the upper prediction error range and the lower prediction error range, the status of the target processing substrate is determined to be abnormal.

22. The prediction apparatus of claim 21, wherein each prediction error range is set according to a standard error between the predicted value and the measured value.

23. The prediction apparatus of claim 21, wherein the predicted value is a processing dimension of the target processing substrate.

24. The prediction apparatus of claim 14, wherein the processing result data is a processing dimension of the target processing substrate.

25. The prediction apparatus of claim 14, wherein the operation data is electronic data obtained from a plurality of detectors provided in the substrate processing apparatus.

26. The prediction apparatus of claim 14, wherein the analysis unit uses Partial Least Squares (PLS) as the multivariate analysis.