Methods and apparatuses for utilizing adaptive predictive algorithms and determining when to use the adaptive predictive algorithms for virtual metrology

Patent metadata

Field	Value
Publication number	US-10409231-B2
Application number	US-201815889906-A
Country	US
Kind code	B2
Filing date	Feb 6, 2018
Priority date	Feb 16, 2010
Publication date	Sep 10, 2019
Grant date	Sep 10, 2019

Abstract

Described herein are methods, apparatuses, and systems for determining adaptive predictive algorithms for virtual metrology. In some embodiments, a computer implemented method identifies a plurality of predictive algorithms. The method determines when to use one or more of the plurality of predictive algorithms to predict one or more virtual metrology variables in a manufacturing facility.

First claim

What is claimed is:

1. A computer implemented method for predicting product yield, the method comprising:
receiving metrology data in a measurement system configured to measure and control production activities within a deposition process or an etch process;
receiving the metrology data in a multi-algorithm predictive subsystem that is designed to facilitate switching between predictive algorithms;
identifying, with the multi-algorithm predictive subsystem, the plurality of predictive algorithms including first and second adaptive predictive algorithms to predict a value of product yield in a manufacturing facility;
identifying a yield threshold that is based on predictions of yield data and actual measurement of yield data to determine when to switch between the first and second adaptive predictive algorithms based on a level of error between predicted yield data and actual measured yield data;
comparing the yield data predicted with the yield threshold when actual measured yield data for a current prediction cycle is not available;
invoking, when the actual measured yield data for a current prediction cycle is not available, the first adaptive predictive algorithm if the yield for prediction is less than or approximately equal to the yield threshold to update a prediction equation to adapt to dynamics of the manufacturing facility;
determining a predicted measurement data based upon the first or second adaptive predictive algorithm, and
reporting to the measurement system, a difference between a predicted measurement data and actual measurement data, wherein the actual measurement data and predicted measurement data at least correspond to a thickness, film uniformity, or critical dimension average of a substrate, at an end of a process run or after the deposition process or the etch process.

2. The computer implemented method of claim 1 , further comprising:
analyzing end-of-line data to determine measured yield data.

3. The computer implemented method of claim 1 , further comprising:
identifying a quality prediction value for each predictive algorithm;
identifying a quality prediction value for each prediction of data; and
performing a normalized weighted calculation of the predictions where weights are a function of the quality prediction value for each predictive algorithm if the actual measurement data is not available.

4. The computer implemented method of claim 1 , wherein the actual yield data is one or more of a product yield, yield factor, yield excursion variables or metrics such as product yield trends, lot level analysis of product yield, yield correlation to manufacturing processes, or statistical analysis of the product yield.

5. The computer implemented method of claim 4 , further comprising:
predicting a probability of failure or time to failure based on the actual yield data.

6. The computer implemented method of claim 4 , further comprising:
assessing a quality of the yield prediction;
switching between the predictive algorithms based on an error threshold determined between the predictions of yield data and actual measured yield data; and
invoking a next algorithm in the plurality of predictive algorithms.

7. The computer implemented method of claim 4 , further comprising:
utilizing feedback of the actual measured yield data to continually tune the prediction algorithms.

8. A computer-readable non-transitory storage medium comprising executable instructions to cause a processor to perform operations, the instructions comprising:
receiving metrology data in a measurement system configured to measure and control production activities within a deposition process or an etch process;
receiving the metrology data in a multi-algorithm predictive subsystem that is designed to facilitate switching between predictive algorithms;
identifying, with the multi-algorithm predictive subsystem, the plurality of predictive algorithms including first and second adaptive predictive algorithms to predict a value of product yield in a manufacturing facility;
identifying a yield threshold that is based on predictions of yield data and actual measurement of yield data to determine when to switch between the first and second adaptive predictive algorithms based on a level of error between predicted yield data and actual measured yield data;
comparing the yield data predicted with the yield threshold when actual measured yield data for a current prediction cycle is not available; and
invoking, when the actual measured yield data for a current prediction cycle is not available, the first adaptive predictive algorithm if the yield for prediction is less than or approximately equal to the yield threshold to update a prediction equation to adapt to dynamics of the manufacturing facility;

determining a predicted measurement data based upon the first or second adaptive predictive algorithm, and
reporting to the measurement system, a difference between a predicted measurement data and actual measurement data, wherein the actual measurement data and predicted measurement data at least correspond to a thickness, film uniformity, or critical dimension average of a substrate, at an end of a process run or after the deposition process or the etch process.

9. The computer-readable non-transitory storage medium of claim 8 , further comprising:
analyzing end-of-line data to determine measured yield data.

10. The computer-readable non-transitory storage medium of claim 8 , further comprising:
identifying a quality prediction value for each predictive algorithm;
identifying a quality prediction value for each prediction of data; and
performing a normalized weighted calculation of the predictions where weights are a function of the quality prediction value for each predictive algorithm if the actual measurement data is not available.

11. The computer-readable non-transitory storage medium of claim 8 , wherein the actual yield data is one or more of a product yield, yield factor, yield excursion variables or metrics such as product yield trends, lot level analysis of product yield, yield correlation to manufacturing processes, or statistical analysis of the product yield.

12. The computer-readable non-transitory storage medium of claim 11 , further comprising:
predicting a probability of failure or time to failure based on the actual yield data.

13. The computer-readable non-transitory storage medium of claim 11 , further comprising:
assessing a quality of the yield prediction;
switching between the predictive algorithms based on an error threshold determined between the predictions of yield data and actual measured yield data; and
invoking a next algorithm in the plurality of predictive algorithms.

14. The computer-readable non-transitory storage medium of claim 11 , further comprising:
utilizing feedback of the actual measured yield data to continually tune the prediction algorithms.

15. A computer system comprising:
a memory to store a plurality of predictive algorithms; and
at least one processing device, coupled to the memory, that is configured to execute processing logic to:
receive metrology data in a measurement system configured to measure and control production activities within a deposition process or an etch process;
receive the metrology data in a multi-algorithm predictive subsystem that is designed to facilitate switching between predictive algorithms;
identify, with the multi-algorithm predictive subsystem, the plurality of predictive algorithms including first and second adaptive predictive algorithms to predict a value of product yield in a manufacturing facility;

Assignees

• Applied Materials Inc

Inventors

• Moyne James

Classifications

• G05B2219/31465

  Determine which variables of the system to be monitored · CPC title

• G05B13/026 Primary

  using a predictor · CPC title

• G05B23/0297 Primary

  Reconfiguration of monitoring system, e.g. use of virtual sensors; change monitoring method as a response to monitoring results · CPC title

• G06N5/04 Primary

  Inference or reasoning models · CPC title

• G06Q50/04

  Manufacturing · CPC title