ADAPTIVE AND AUTOMATIC DETERMINATION OF SYSTEM PARAMETERS

• US 20120130525A1
• Filed: 01/14/2011
• Published: 05/24/2012
• Est. Priority Date: 11/18/2010
• Status: Active Grant

• Current Assignee: Taiwan Semiconductor Manufacturing Company Limited
• Original Assignee: Taiwan Semiconductor Manufacturing Company Limited
• Inventors: TSAI, Po-Feng, HO, Chia-Tong, WU, Sunny, WANG, Jo Fei, MOU, Jong-I
• Granted Patent: US 8,606,387 B2
• Time in Patent Office: Days
• US Class Current: 700/108
• CPC Class Codes: G05B 13/0265 the criterion being a learning criterion
• CPC Class Codes: G05B 19/41875 characterised by quality surveillance of production
• CPC Class Codes: G05B 2219/32188 Teaching relation between controlling parameters and quality parameters
• CPC Class Codes: G05B 2219/32189 Compare between original solid model and measured manufactured object
• CPC Class Codes: G05B 2219/45031 Manufacturing semiconductor wafers
• CPC Class Codes: Y02P 90/02 Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]
• CPC Class Codes: Y02P 90/80 Management or planning

Abstract

A MIMO optimizer is used to identify tunable process parameters for processing equipment.

Claims

1. A method of automatically determining process parameters for processing equipment, said method comprising:

processing at least one substrate in the processing equipment;

collecting data on process monitors for the at least one substrate after processing is completed;

receiving the collected data by a multiple-input-multiple-output (MIMO) optimization system;

revising a sensitivity matrix used by a MIMO optimizer of the MIMO optimization system using the collected data and an adaptive-learning algorithm; and

determining a set of process parameters for the processing equipment by the MIMO optimizer, wherein the MIMO optimizer uses the revised sensitivity matrix to determine the process parameters.

2. The method of claim 1, wherein the sensitivity matrix includes linear relationships between multiple said process monitors and multiple said process parameters of the processing equipment;

and wherein the process monitors are outputs of the MIMO optimizer and the process parameters are inputs of the MIMO optimizer.

3. The method of claim 1, further comprising:

defining an optimization object function of the MIMO optimizer to minimize the sum of absolute values of differences between the collected data on the process monitors and their corresponding specifications, wherein the absolute values are weighted.

4. The method of claim 3, wherein the optimization object function includes a penalty term that gives different weights to changes in different process parameters.

5. The method of claim 1, wherein the MIMO optimizer includes constraints for the process parameters and/or the process monitors.

6. The method of claim 1, wherein the MIMO optimizer uses Newton'"'"'s method, genetic method or simulated annealing method.

7. The method of claim 1, further comprising:

receiving a request for a number of sets of process parameters.

8. The method of claim 1, wherein the adaptive-learning algorithm minimizes a filtered energy function, wherein a energy function of the filtered energy function is defined to be the square of a difference between a simulated value of a process parameter and a predicted value of the process parameter, and wherein the adaptive-learning algorithm revises the sensitivity matrix by determining a new set of sensitivity coefficients for the revised sensitivity matrix.

9. The method of claim 8, wherein the adaptive-learning algorithm uses a weight to reflect the reliability of the collected data on the process monitors and a learning rate coefficient to determine how fast the sensitivity matrix is revised.

10. The method of claim 8, wherein the filtered energy function is an exponentially weighted moving average (EWMA) of the energy function.

11. A method of qualifying processing equipment for production, said method comprising:

(1) processing at least one substrate in the processing equipment;

(2) collecting data on process monitors for the at least one substrate after processing is completed;

(3) determining if the collected data are within specifications of the process monitors,

if the answer is yes, determining that the processing equipment is qualified for production, and

if the answer is no, determining a set of process parameters for the processing equipment by a multiple-input-multiple-output (MIMO) optimizer;

and repeating (1)-(3) until the processing equipment is qualified.

12. The method of claim 11, further comprising:

revising a sensitivity matrix used by the MIMO optimizer using the collected data and an adaptive-learning algorithm, wherein the revised sensitivity matrix is used in said determining the set of process parameters.

13. The method of claim 11, wherein the method is practiced after the processing equipment undergoes a maintenance process.

14. The method of claim 11, wherein the processing equipment is selected from the group consisting of deposition, etching, planarization, lithography, dopant implant, annealing, or cleaning equipment for semiconductor fabrication.

15. A computer-readable storage medium embodying instructions for causing, when executed by a processor, the processor to perform a method of automatically determining process parameters for processing equipment, said medium comprising:

instructions for causing the processing equipment to process at least one substrate;

instructions for collecting data on process monitors for the at least one substrate after the substrate has been processed by the processing equipment;

instructions for revising a sensitivity matrix used by a MIMO optimizer using the collected data and an adaptive-learning algorithm; and

instructions for determining a set of process parameters for the processing equipment by the MIMO optimizer, wherein the MIMO optimizer uses the revised sensitivity matrix to determine the process parameters.

16. The computer-readable storage medium of claim 15, further comprising:

instructions for receiving a request for a number of sets of process parameters.

17. The computer-readable storage medium of claim 15, wherein the sensitivity matrix includes linear relationships between multiple said process monitors and multiple said process parameters of the processing equipment;

and wherein the process monitors are outputs of the MIMO optimizer and the process parameters are inputs of the MIMO optimizer.

18. The computer-readable storage medium of claim 15, further comprising instructions for defining an optimization object function of the MIMO optimizer to minimize the sum of absolute values of differences between the collected data on the process monitors and their corresponding specifications.

19. A computer-readable storage medium embodying instructions for causing, when executed by a processor, the processor to perform a method of qualifying processing equipment for production, said medium comprising:

first instructions for causing the processing equipment to process at least one substrate;

second instructions for collecting data on process monitors for the at least one substrate after the substrate has been processed by the processing equipment;

third instructions for determining if the collected data are within specifications of the process monitors, and for,

if the answer is yes, determining that the process equipment is qualified for production, and

if the answer is no, determining a set of process parameters for the processing equipment by a multiple-input-multiple-output (MIMO) optimizer, and repetitively performing the first through third instructions until the processing equipment is qualified.

20. The computer-readable storage medium of claim 19, further comprising:

instructions for revising a sensitivity matrix used by the MIMO optimizer using the collected data and an adaptive-learning algorithm, wherein the revised sensitivity matrix is used in said determining the set of process parameters.