Method and system for semiconductor process control and monitoring by using a data quality metric

• US 8,849,615 B2
• Filed: 01/11/2010
• Issued: 09/30/2014
• Est. Priority Date: 01/30/2009
• Status: Active Grant

• Current Assignee: Globalfoundries US Incorporated (GlobalFoundries, Inc.)
• Original Assignee: GlobalFoundries, Inc.
• Inventors: Kost, Daniel
• Primary Examiner(s): Breene, John
• Assistant Examiner(s): BLOSS, STEPHANIE E
• Application Number:
  US12/685,022
• Publication Number:
  US 20100198556A1
• Time in Patent Office: 1,723 Days
• Field of Search: 702/183, 702/185
• US Class Current: 702/185
• CPC Class Codes: G05B 23/0221 Preprocessing measurements, e.g. data collection rate adjustment; Standardization of measurements; Time series or signal analysis, e.g. frequency analysis or wavelets; Trustworthiness of measurements; Indexes therefor; Measurements using easily measured parameters to estimate parameters difficult to measure; Virtual sensor creation; De-noising; Sensor fusion; Unconventional preprocessing inherently present in specific fault detection methods like PCA-based methods
• CPC Class Codes: G05B 23/024 Quantitative history assessment, e.g. mathematical relationships between available data; Functions therefor; Principal component analysis [PCA]; Partial least square [PLS]; Statistical classifiers, e.g. Bayesian networks, linear regression or correlation analysis; Neural networks

Abstract

During fault detection of a production process, the influence of the data communication process for communicating measurement readings to the fault detection system may be taken into consideration. In one illustrative embodiment, a data rate related parameter may be used as an input variable for a data reduction procedure, thereby enabling an efficient assessment of the quality of the remaining “real” process and tool parameters.

Claims

1. A method of monitoring a production process for processing microstructure devices in at least one process tool, the method comprising:

performing a data communication process so as to communicate a plurality of measurement data at least several times from said at least one process tool to a fault detection system during processing of a microstructure device;

determining a data communication quality metric in said fault detection system, said data communication quality metric comprising a momentary data rate between successive measurements generated for each of the plurality of measurement data and indicating at least a degree of completeness of said measurement data; and

performing a fault detection process in said fault detection system using said plurality of measurement data and said data communication quality metric as inputs to the fault detection process.

2. The method of claim 1, wherein said measurement data represents values of a plurality of process parameters and wherein performing said fault detection process comprises performing a multivariate data reduction using each of said plurality of process parameters as an input variable.

3. The method of claim 2, wherein performing said multivariate data reduction comprises using said data communication quality metric as a further input variable.

4. The method of claim 2, wherein performing said multivariate data reduction comprises performing at least one of a principal component analysis (PCA) or a k nearest neighbor (KNN) analysis.

5. The method of claim 1, wherein determining said momentary data rate further comprises subtracting respective time stamps of two subsequent measurement data items that are associated with the same process parameter.

6. The method of claim 1, wherein performing said fault detection process comprises indicating a fault condition of said production process when a first process quality metric is outside a valid process quality range and said data communication quality metric is within a predefined valid communication quality range.

7. The method of claim 6, wherein performing said fault detection process further comprises disregarding a second process quality metric that is outside said predefined valid process quality range when said data communication quality metric is outside said predefined valid communication quality range.

8. The method of claim 7, wherein performing said fault detection process further comprises determining, on the basis of said data communication quality metric, a contribution of said measurement data to a third process quality metric that is within said predefined valid process quality range.

9. The method of claim 8, wherein determining said contribution comprises identifying a part of said measurement data as non-influential data when said data communication quality metric exceeds a predefined threshold.

10. A method of fault detection in a semiconductor manufacturing process, the method comprising:

communicating a plurality of measurement data for each of a plurality of process parameters of said manufacturing process to a fault detection system;

performing a data reduction process in said fault detection system by using each of said plurality of process parameters as an input variable;

determining a momentary data rate metric between successive measurements for each of the measurement data provided by a data communication process used for communicating said measurement data;

using said momentary data rate metric as a further input variable of said data reduction process; and

assessing said semiconductor manufacturing process by using a quality metric obtained by said data reduction process and generated using said input variables and a data rate contribution value indicating a contribution of said momentary data rate metric to said quality metric.

11. The method of claim 10, wherein assessing said semiconductor manufacturing process comprises indicating a fault condition of said semiconductor manufacturing process when said quality metric is outside a predefined valid process quality range and said data rate metric is within a predefined valid value range.

12. The method of claim 11, wherein assessing said semiconductor manufacturing process comprises indicating said quality metric as non-conclusive when being outside of said predefined valid process quality range and when said data rate metric is outside said predefined valid value range.

13. The method of claim 10, wherein determining said momentary data rate metric comprises using respective time stamps associated with each of said measurement data.

14. The method of claim 10, wherein performing said data reduction process comprises establishing a model of said semiconductor manufacturing process on the basis of reference measurement data of said manufacturing process.

15. The method of claim 14, wherein said model is established using one of a principal component analysis (PCA) or a k nearest neighbor (KNN) analysis.

16. The method of claim 10, wherein said measurement data of said plurality of process parameters is obtained for each of a plurality of substrates that are processed in said semiconductor manufacturing process.

17. The method of claim 10, wherein determining said momentary data rate metric comprises determining a time interval between two subsequent measurement data items that are associated with the same process parameter.

18. A fault detection system, comprising:

an interface configured to receive a plurality of measurement data from at least one process tool used for processing a substrate, said measurement data being indicative of a plurality of process parameters of said at least one process tool; and

a fault detection module connected to said interface and configured to determine a data communication quality metric indicating at least a momentary data rate between successive measurements of a data communication of said interface when receiving said measurement data for each of the plurality of measurement data, said fault detection module further being configured to determine a fault condition of said at least one process tool using said data communication quality metric and said measurement data as inputs to said fault detection module.

19. The fault detection system of claim 18, wherein said fault detection module has implemented therein at least one of an principal component analysis (PCA) algorithm or a k nearest neighbor analysis (KNN) algorithm.

20. The fault detection system of claim 18, wherein the fault detection module determines said momentary data rate by determining a time interval between two subsequent measurement data items that are associated with the same process parameter.