Method and apparatus for self-learning and self-improving a semiconductor manufacturing tool

• US 9,424,528 B2
• Filed: 04/23/2014
• Issued: 08/23/2016
• Est. Priority Date: 03/08/2008
• Status: Active Grant

• Current Assignee: Tokyo Electron Limited
• Original Assignee: Tokyo Electron Limited
• Inventors: Kaushal, Sanjeev, Patel, Sukesh Janubhai, Sugishima, Kenji
• Primary Examiner(s): KASENGE, CHARLES R
• Application Number:
  US14/259,696
• Publication Number:
  US 20140229409A1
• Time in Patent Office: 853 Days
• Field of Search: 700/28, 700/47, 700/52, 700/100, 700/103, 700/105, 700/108, 700/109, 700/110, 700/121, 700/104, 702/81, 702/82, 702/83, 702/84, 702/182, 702/183, 702/184, 438/5, 438/14, 438/16, 438/17, 706/12, 706/13, 706/16, 706/23, 706/25
• US Class Current: 1/1
• CPC Class Codes: G05B 13/0265 the criterion being a learning criterion
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 5/04 Inference or reasoning models

Abstract

Performance of a manufacturing tool is optimized. Optimization relies on recipe drifting and generation of knowledge that capture relationships among product output metrics and input material measurement(s) and recipe parameters. Optimized recipe parameters are extracted from a basis of learned functions that predict output metrics for a current state of the manufacturing tool and measurements of input material(s). Drifting and learning are related and lead to dynamic optimization of tool performance, which enables optimized output from the manufacturing tool as the operation conditions of the tool changes. Features of recipe drifting and associated learning can be autonomously or externally configured through suitable user interfaces, which also can be drifted to optimize end-user interaction.

Claims

1. A system, comprising:

a memory storing computer executable components; and

a processor configured to execute the following computer executable components stored in the memory;

a drift component that adjusts a manufacturing recipe for an asset produced by a manufacturing tool based at least on a set of variables and a probability distribution function to generate an adjusted manufacturing recipe for the asset, wherein a magnitude of change between a recipe parameter included in the manufacturing recipe and an adjusted recipe parameter included in the adjusted manufacturing recipe is determined based at least on the probability distribution function; and

an objective autonomous learning engine that determines a function that predicts output metrics for the asset based on the adjusted manufacturing recipe, and relaxes a constraint for the output metrics to infer the function for the manufacturing tool.

2. The system of claim 1, wherein the drift component adjusts the manufacturing recipe for the asset based on a sequence in which to modify a set of recipe parameters associated with the manufacturing recipe.

3. The system of claim 1, further comprising an interaction component that receives data associated with the probability distribution function to adjust the recipe parameter included in the manufacturing recipe.

4. The system of claim 1, further comprising an autonomous optimization engine that extracts a set of updated recipe parameters from a set of input measurements and the function to generate another manufacturing recipe within a predefined tolerance of a target value for the output metrics.

5. The system of claim 1, wherein the drift component adjusts the recipe parameter included in the manufacturing recipe based on the probability distribution function to generate the adjusted recipe parameter.

6. The system of claim 1, wherein the objective autonomous learning engine relaxes the constraint based on a ranking of weights for the output metrics.

7. A method, comprising:

receiving, by a device comprising a processor, a manufacturing recipe for an asset produced by a manufacturing tool;

generating, by the device, a modified manufacturing recipe by modifying the manufacturing recipe based on a set of variables and a probability distribution function that determines a magnitude of change between a parameter value associated with the manufacturing recipe and a modified parameter value associated with the modified manufacturing recipe; and

determining, by the device, one or more functions that predict asset output metrics for the asset based on the modified manufacturing recipe, comprising relaxing one or more constraints for the asset output metrics to infer the one or more functions.

8. The method of claim 7, further comprising learning, by the device, a set of relationships associated with product output based on the modifying of the manufacturing recipe.

9. The method of claim 8, further comprising generating, by the device, another recipe to accomplish a target product output based at least in part on the set of relationships.

10. The method of claim 9, further comprising determining, by the device, whether the target product output is fulfilled in response to determining that generation of the modified manufacturing recipe is successful.

11. The method of claim 7, wherein the modifying further comprises modifying the manufacturing recipe by randomizing one or more parameters associated with the manufacturing recipe.

12. The method of claim 7, wherein the modifying further comprises modifying the manufacturing recipe based on a sequence in which to modify a set of parameter values associated with the manufacturing recipe.

13. The method of claim 7, further comprising receiving, by the device, data associated with the probability distribution function from a user interface.

14. The method of claim 13, further comprising extracting, by the device, a set of parameter values from a set of input measurements and the one or more functions to generate another manufacturing recipe within a predefined tolerance of a target value for the asset output metrics.

15. The method of claim 13, wherein the generating the modified manufacturing recipe comprises generating the modified parameter value by adjusting the parameter value associated with the manufacturing recipe based on the probability distribution function.

16. The system of claim 13, wherein the relaxing comprises relaxing the one or more constraints for the asset output metrics based on a ranking of weights for the asset output metrics.

17. A non-transitory computer-readable medium comprising a set of code instruction retained therein that, in response to execution, cause a computing system including at least one processor to perform operations, comprising:

receiving a manufacturing recipe for an asset produced by a manufacturing tool;

generating another manufacturing recipe by adjusting the manufacturing recipe based on a set of variables and a probability distribution function that determines a degree of change between a set of recipe parameters included in the manufacturing recipe and a set of other recipe parameters included in the other manufacturing recipe; and

determining a function that predicts a set of output metrics for the asset based on the other manufacturing recipe, comprising relaxing a constraint for the set of output metrics to infer the function for the manufacturing tool.

18. The non-transitory computer-readable medium of claim 17, wherein the adjusting further comprises adjusting the manufacturing recipe by randomizing one or more parameters associated with the manufacturing recipe.

19. The non-transitory computer-readable medium of claim 17, wherein the adjusting further comprises adjusting the manufacturing recipe based on a sequence in which to adjust the set of recipe parameters included in the manufacturing recipe.

20. The non-transitory computer-readable medium of claim 17, further comprising learning a set of relationships associated with product output based on the adjusting of the manufacturing recipe.