Method and apparatus for monitoring a process by employing principal component analysis

• US 6,896,763 B2
• Filed: 01/14/2003
• Issued: 05/24/2005
• Est. Priority Date: 07/07/1999
• Status: Expired due to Term

• Current Assignee: Dimitris Lymberopoulos, Jed Davidow, Lalitha Balasubramhanya, Moshe Sarfaty
• Original Assignee: Dimitris Lymberopoulos, Jed Davidow, Lalitha Balasubramhanya, Moshe Sarfaty
• Inventors: Sarfaty, Moshe, Lymberopoulos, Dimitris, Balasubramhanya, Lalitha, Davidow, Jed
• Primary Examiner(s): Goudreau, George A.
• Application Number:
  US10/341,696
• Publication Number:
  US 20030136511A1
• Time in Patent Office: 861 Days
• Field of Search: 156/345.24, 156/345.25, 118/723.E
• US Class Current: 156/345.24
• CPC Class Codes: G01N 21/274 Calibration, base line adjustment, drift correction
• CPC Class Codes: G01N 21/3563 for analysing solids; Preparation of samples therefor
• CPC Class Codes: G01N 21/359 using near infrared light
• CPC Class Codes: G01N 2201/1293 resolving multicomponent spectra
• 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

A method and apparatus for monitoring a process by employing principal component analysis are provided. Correlated attributes are measured for the process to be monitored (the production process). Principal component analysis then is performed on the measured correlated attributes so as to generate at least one production principal component; and the at least one production principal component is compared to a principal component associated with a calibration process (a calibration principal component). The calibration principal component is obtained by measuring correlated attributes of a calibration process, and by performing principal component analysis on the measured correlated attributes so as to generate at least one principal component. A principal component having a feature indicative of at least one of a desired process state, process event and chamber state then is identified and is designated as the calibration principal component.

Claims

1. An apparatus adapted to monitor a chamber comprising:

a measurement apparatus adapted to collect optical emission spectroscopy (OES) data for electromagnetic radiation emitted by a plasma of the chamber; and

a processing mechanism coupled to the measurement apparatus, the processing mechanism adapted to;

receive OES data for electromagnetic radiation emitted by the plasma of the chamber;

for a series of windows of the received OES data;

perform principal component analysis to compute a respective principal component for each window of the received OES data; and

calculate an inner product of the principal component computed for each window of the received OES data and a calibration principal component computed for a window of OES data that corresponds to a previously performed calibration process; and

detect at least one of health of the chamber and a fault of the chamber based on one or more of the calculated inner products.

2. The apparatus of claim 1 wherein the calibration principal component comprises a principal component of a window of OES data of a calibration process performed in a non-faulted chamber.

3. The apparatus of claim 1 wherein the calibration principal component comprises a principal component of a window of OES data of a calibration process performed in the chamber.

4. The apparatus of claim 1 wherein the measurement apparatus comprises a detector adapted to detect a plurality of electromagnetic emissions from a plasma.

5. The apparatus of claim 4 wherein the detector comprises a detector selected from the group consisting of a CCD array, a photodiode array, a spectrometer and a photomultiplier.

6. The apparatus of claim 4 wherein the detector is adapted to detect electromagnetic emissions having wavelengths from about 180 to 850 nanometers.

7. The apparatus of claim 4 further comprising an optical fiber coupled to the detector, the optical fiber adapted to collect electromagnetic emissions from the plasma and to transmit the electromagnetic emissions to the detector.

8. The apparatus of claim 1 wherein the processing mechanism is further adapted to detect a transition in the calculated inner products.

9. The apparatus of claim 1 wherein the processing mechanism is further adapted to mean-center the received OES data before performing principal component analysis.

10. The apparatus of claim 1 wherein the processing mechanism is adapted to detect at least one of health of the chamber and a fault of the chamber while a process is being performed within the chamber.

11. An apparatus adapted to monitor a chamber comprising:

a measurement apparatus adapted to measure correlated attribute data of a plasma of the chamber; and

a processing mechanism coupled to the measurement apparatus, the processing mechanism adapted to;

receive correlated attribute data for the plasma of the chamber;

for a series of windows of the received correlated attribute data;

perform principal component analysis to compute a respective principal component for each window of the received correlated attribute data; and

calculate an inner product of the principal component computed for each window of the received correlated attribute data and a calibration principal component computed for a window of correlated attribute data that corresponds to a previously performed calibration process; and

detect at least one of health of the chamber and a fault of the chamber based on one or more of the calculated inner products.

12. The apparatus of claim 11 wherein the correlated attribute data comprises at least one of electromagnetic radiation emitted by the plasma and RF power associated with the plasma of the chamber.