Method and apparatus for real time monitoring of wafer attributes in a plasma etch process

• US 5,654,903 A
• Filed: 11/07/1995
• Issued: 08/05/1997
• Est. Priority Date: 11/07/1995
• Status: Expired due to Term

• Current Assignee: Bell Semiconductor, LLC (Hilco Inc. (d/b/a Hilco Global))
• Original Assignee: Lucent Technologies, Inc. (Nokia Corporation)
• Inventors: Reitman, Edward A., Ibbotson, Dale E., Lee, Tseng-Chung
• Primary Examiner(s): Trammell, James P.
• Assistant Examiner(s): Bui, Bryan
• Application Number:
  US08/553,118
• Time in Patent Office: 637 Days
• Field of Search: 364/551.01, 364/569, 364/580, 364/578, 156/625.1, 156/626.1, 156/627.1, 156/643.1, 395/21, 395/22, 395/68, 216/63, 216/67, 216/59, 216/60
• US Class Current: 702/81
• CPC Class Codes: G05B 13/027 using neural networks only
• CPC Class Codes: H01J 37/32935 Monitoring and controlling tubes by information coming from the object and/or discharge

Abstract

The present invention provides a method and apparatus for monitoring the state of an attribute of a product during the manufacturing process. The invention employs an intelligent system trained in the relationship between the signatures of the manufacturing process and the product attribute as a function of time. In one embodiment the intelligent system comprises two intelligent systems: the first, trained in the relationship between the signatures of the manufacturing process and one or more signatures of the state of the attribute as a function of time; and the second, trained in the relationship between the product-state signatures as a function of time and the product attribute as a function of time. The disclosed apparatus can be an integral part of the manufacturing machinery allowing the process to continue until the exact moment when the desired state is achieved.

Claims

1. A method of monitoring in real time, the state of an attribute of a product during the manufacturing process of said product, comprising the steps of:

selecting one or more process signatures which characterize said manufacturing process and which have an indirect relationship to the state of said attribute;

selecting one or more product-state signatures which characterize said attribute;

collecting said process signatures in situ, over a period of time;

inputting said previously collected process signatures to an intelligent system which has been trained in the relationship between said process signatures over a period of time and said product attribute as a function of time, said intelligent system having an output representing said attribute as a function of time; and

using said output of said intelligent system to determine when a desired state of said attribute is found and to alter said manufacturing process.

2. A method in accordance with claim 1 wherein said intelligent system comprises two intelligent systems and wherein said inputting step further comprises the steps of:

inputting said previously collected process signatures to a first intelligent system which has been trained in the relationship between said process signatures over a period of time and one or more of said product state signatures as a function of time, said first intelligent system having an output representing one or more of said product state signature; and

inputting said output of said first intelligent system into a second intelligent system which has been trained in the relationship between said product state signatures as a function of time and said attribute as a function of time, said second intelligent system having an output representing said attribute as a function of time.

3. A method in accordance with claim 1 wherein said selecting one or more process signatures step includes choosing one or more metrics representative of the material structures of the product.

4. A method in accordance with claim 1 wherein said intelligent system comprises one or more neural networks.

5. A method in accordance with claim 1 wherein said intelligent system comprises one or more recurrent networks.

6. A method in accordance with claim 1 wherein said intelligent system comprises one or more time delay networks.

7. A method in accordance with claim 1 wherein said product is a semiconductor wafer.

8. A method in accordance with claim 7 wherein said semiconductor wafer attribute is the post processing thickness of a film on said semiconductor wafer.

9. A method of monitoring in real time, the state of an attribute of a semiconductor wafer during plasma etch processing, comprising the steps of:

selecting one or more process signatures which characterize said plasma etch process and which have a cause and effect relationship to the state of said attribute;

selecting one or more wafer-state signatures which characterize said attribute;

collecting said process signatures in situ, over a period of time;

inputting said process signatures to an intelligent system which has been trained in the relationship between said process signatures and said attribute as a function of time, said intelligent system having an output representing said attribute as a function of time; and

using the output of said intelligent system to determine when a desired state of said attribute is found and to alter said plasma etch process.

10. A method in accordance with claim 9 wherein said intelligent system comprises two intelligent systems and wherein said inputting step further comprises the steps of:

inputting said previously collected process signatures to a first intelligent system which has been trained in the relationship between said process signatures over a period of time and one or more of said product state signatures as a function of time, said first intelligent system having an output representing one or more of said product state signature; and

inputting said output of said first intelligent system into a second intelligent system which has been trained in the relationship between said product state signatures as a function of time and said attribute as a function of time, said second intelligent system having an output representing said attribute as a function of time.

11. A method in accordance with claim 9 wherein said wafer-state signature includes the ellipsometric trace of the semiconductor wafer.

12. A method in accordance with claim 9 wherein said wafer-state signature includes the interferometric trace of the semiconductor wafer.

13. A method in accordance with claim 9 wherein said selecting step includes choosing one or more metrics representative of the material structures of the semiconductor wafer.

14. A method in accordance with claim 9 wherein said selecting step includes choosing one or more metrics representative of the pattern density of the semiconductor wafer.

15. A method in accordance with claim 9 wherein said first and second intelligent system comprises one or more neural networks.

16. A method in accordance with claim 9 wherein said intelligent system comprises one or more recurrent networks.

17. A method in accordance with claim 9 wherein said intelligent system comprises one or more time delay networks.

18. A method in accordance with claim 9 wherein said attribute is the post etch thickness of the semiconductor wafer.

19. Apparatus for monitoring in real time, the state of an attribute of a product during the manufacturing process, comprising:

means for collecting one or more process signatures which characterize said manufacturing process and which have an indirect relationship to the state of said attribute;

an intelligent system which has been trained in the relationship between said process signatures and said attribute as a function of time;

input means for inputting said process signatures to said intelligent system over a period of time; and

means for creating an output signal from said intelligent system, said output signal containing information for determining when a desired state of the attribute is obtained and to stop said manufacturing process.

20. An apparatus in accordance with claim 19 wherein said intelligence system comprises two intelligent systems and further comprising:

a first intelligent system which has been trained in the relationship between said process signatures and one or more product state signatures which characterize said attribute and whose output represents said one or more product state signatures of said attribute;

input means for inputting said process signatures to said intelligent first system over a period of time;

a second intelligent system which has been trained in the relationship between said product state signatures as a function of time and said attribute as a function of time;

input means for inputting said output of said first intelligent system into said second intelligent system over a period of time; and

means for creating an output signal from said second intelligent system, said output signal containing information for determining one of a desired state of the attribute is obtained and to stop said manufacturing process.

21. An apparatus in accordance with claim 19 wherein said intelligent system comprises one or more neural networks.

22. An apparatus in accordance with claim 19 wherein said intelligent system comprises one or more recurrent networks.

23. An apparatus in accordance with claim 19 wherein said intelligent system comprises one or more time delay networks.

24. An apparatus in accordance with claim 19 wherein said product is a semiconductor wafer.

25. An apparatus in accordance with claim 24 wherein said manufacturing process is plasma etch processing.