METHOD AND APPARATUS FOR DETECTING FAULT IN THE SEMICONDUCTOR MENUFACTURING PROCESS AND RECORDING MEDIUM THEREOF

• US 20150318221A1
• Filed: 11/26/2014
• Published: 11/05/2015
• Est. Priority Date: 05/02/2014
• Status: Abandoned Application

• Current Assignee: Bistel, Inc., Industry Academic Cooperation Foundation
• Original Assignee: Bistel, Inc.
• Inventors: PARK, Chan Young, KIM, Chang Ouk
• Application Number:
  US14/555,451
• Publication Number:
  US 20150318221A1
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G01B 5/043 for measuring length
• CPC Class Codes: G05B 2219/45031 Manufacturing semiconductor wafers
• 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
• CPC Class Codes: G06F 18/2135 based on approximation criteria, e.g. principal component analysis
• CPC Class Codes: H01L 21/67253 Process monitoring, e.g. flow or thickness monitoring
• CPC Class Codes: H01L 22/12 for structural parameters, e.g. thickness, line width, refractive index, temperature, warp, bond strength, defects, optical inspection, electrical measurement of structural dimensions, metallurgic measurement of diffusions electrical measurement of diffusions H01L22/14
• CPC Class Codes: H01L 22/20 Sequence of activities consisting of a plurality of measurements, corrections, marking or sorting steps

Abstract

A method of detecting a fault in a semiconductor manufacturing process. The method includes obtaining measured data and reference data regarding at least one parameter related to semiconductor manufacturing conditions in a process included in a semiconductor manufacturing process during a pre-set period of time; converting the measured data and the reference data by using at least one principal component parameter obtained via principal component analysis with respect to the measured data and the reference data; calculating a similarity between the converted measured data and the converted reference data; and detecting a fault in the process based on the calculated similarity. As a result, production efficiency of a semiconductor manufacturing process may be improved.

Claims

1. A method of detecting a fault in a semiconductor manufacturing process, the method comprising:

obtaining measured data and reference data regarding at least one parameter related to semiconductor manufacturing conditions in a process included in a semiconductor manufacturing process during a pre-set period of time;

converting the measured data and the reference data by using at least one principal component parameter obtained via principal component analysis with respect to the measured data and the reference data;

calculating a similarity between the converted measured data and the converted reference data; and

detecting a fault in the process based on the calculated similarity.

2. The method of claim 1, wherein the calculating of the similarity comprises obtaining at least one from among first distance yardstick information, second distance yardstick information, and third distance yardstick information between the converted measured data and the converted reference data,

the first distance yardstick information is information based on an angle formed between a first principal component parameter of the converted measured data and a first principal component parameter of the converted reference data,

the second distance yardstick information is information based on a ratio between distribution of the converted measured data and distribution of the converted reference data, and

the third distance yardstick information is information based on a difference between the average of the converted measured data and the average of the converted reference data.

3. The method of claim 2, wherein the second distance yardstick information is determined based on a ratio between an n^th eigenvalue of the converted measured data and an n^th eigenvalue of the converted reference data.

4. The method of claim 2, wherein the detecting of the fault comprises:

determining whether the first distance yardstick information, the second distance yardstick information, and the third distance yardstick information exceed first through third critical values, respectively; and

,

if at least one from among the first through third distance yardstick information exceeds the corresponding critical value, determining that there is a fault in the designated process.

5. The method of claim 4, wherein the first through third critical values are determined based on distances between components included in the converted reference data and the median of the distances between the components.

6. The method of claim 5, wherein the first through third critical values are determined by applying the Hampel'"'"'s outlier detection model to distances between components included in the converted reference data and the median of the distances between the components.

7. The method of claim 1, further comprising, if a fault is detected in a process, notifying information regarding parameters included in the fault-detected process and information regarding a semiconductor manufacturing process included in the fault-detected process to a user.

15. A computer readable recording medium having recorded thereon a computer program for implementing the method of claim 1.

8. A fault detecting device for detecting a fault in a semiconductor manufacturing process, the fault detecting device comprising:

an input unit, which obtains measured data and reference data regarding at least one parameter related to semiconductor manufacturing conditions in a process included in a semiconductor manufacturing process during a pre-set period of time;

a control unit, which converts the measured data and the reference data by using at least one principal component parameter obtained via principal component analysis with respect to the measured data and the reference data and calculates a similarity between the converted measured data and the converted reference data; and

a detection unit, which detects a fault in the process based on the calculated similarity.

9. The fault detecting device of claim 8, wherein the control unit obtains at least one from among first distance yardstick information, second distance yardstick information, and third distance yardstick information between the converted measured data and the converted reference data,

the first distance yardstick information is information based on an angle formed between a first principal component parameter of the converted measured data and a first principal component parameter of the converted reference data,

the second distance yardstick information is information based on a ratio between distribution of the converted measured data and distribution of the converted reference data, and

the third distance yardstick information is information based on a difference between the average of the converted measured data and the average of the converted reference data.

10. The fault detecting device of claim 9, wherein the second distance yardstick information is determined based on a ratio between an n^th eigenvalue of the converted measured data and an n^th eigenvalue of the converted reference data.

11. The fault detecting device of claim 9, wherein the detection unit determines whether the first distance yardstick information, the second distance yardstick information, and the third distance yardstick information exceed first through third critical values, respectively;

and,

if at least one from among the first through third distance yardstick information exceeds the corresponding critical value, the detection unit determines that there is a fault in the designated process.

12. The fault detecting device of claim 11, wherein the first through third critical values are determined based on distances between components included in the converted reference data and the median of the distances between the components.

13. The fault detecting device of claim 12, wherein the first through third critical values are determined by applying the Hampel'"'"'s outlier detection model to distances between components included in the converted reference data and the median of the distances between the components.

14. The fault detecting device of claim 8, further comprising an output unit, which, if a fault is detected in a process, notifies information regarding parameters included in the fault-detected process and information regarding a semiconductor manufacturing process included in the fault-detected process to a user.