Method for measuring semiconductor device

1. A method for measuring a semiconductor device, the method comprising:

defining an interest area and an acceptable area in a chip area on a wafer;

performing a first measurement of the chip area with a spectral imaging device to acquire spectrum data of the chip area;

assuming a distribution of spectrum data of a first pixel in the acceptable area is a normal distribution;

calculating a distance from a central point on the normal distribution to second pixels in the interest area;

selecting a position of a second pixel having a distance from the central point on the normal distribution greater than a predetermined range, among the second pixels, as a candidate position;

performing a second measurement of the candidate position; and

predicting a measurement of pixels other than the interest area, using the second measurement of the candidate position, predicting the measurement including:

learning a prediction model based on a supervised learning algorithm, using the second measurement; and

predicting a measurement of pixels other than the interest area, using the learned prediction model.

2. The method as claimed in claim 1 , wherein assuming the distribution of spectrum data of the first pixel is a normal distribution includes:

assuming the distribution of spectrum data of the first pixel is a multivariate normal distribution.

3. The method as claimed in claim 1 , wherein calculating a distance from the central point on the normal distribution to the second pixel includes:

calculating a Mahalanobis distance to the second pixel from a central pixel of the distribution of spectrum data of the first pixel; and

calculating a distance from the central point on the normal distribution for the second pixel, using the Mahalanobis distance.

4. The method as claimed in claim 3 , further comprising:

generating a Mahalanobis distance map in which the Mahalanobis distance of the second pixel is expressed by a numerical value of intensity,

wherein selecting a position of a pixel as the candidate position includes:

selecting a position of a pixel having an intensity that exceeds a predetermined value, among second pixels, as the candidate position.

5. The method as claimed in claim 1 , wherein the acceptable area is defined inside the interest area.

6. The method as claimed in claim 1 , wherein the second pixel is in the interest area but is not in the acceptable area.

7. The method as claimed in claim 1 , wherein the supervised learning algorithm includes a regression analysis algorithm or a classification algorithm.

8. A method for measuring a semiconductor device, the method comprising:

defining an interest area and an acceptable area in a chip area on a wafer;

performing a first measurement of the chip area with a spectral imaging device to acquire spectrum data of the chip area;

calculating a Mahalanobis distance on a second pixel in the interest area from a distribution of spectrum data of a first pixel in the acceptable area;

generating a Mahalanobis distance map in which the Mahalanobis distance of the second pixel indicates an intensity value;

selecting a position of a pixel having an intensity value that exceeds a predetermined value as a candidate position; and

performing a second measurement of the candidate position.

9. The method as claimed in claim 8 , wherein the acceptable area is defined inside the interest area.

10. The method as claimed in claim 9 , wherein the second pixel is in the interest area but is not in the acceptable area.

11. The method as claimed in claim 8 , wherein the spectrum data includes multi-dimensional spectrum data.

12. The method as claimed in claim 8 , further comprising:

predicting a measurement of pixels other than the interest area, using the second measurement of the candidate position.

13. The method as claimed in claim 12 , wherein the chip area includes a first chip area and a second chip area, and

wherein predicting the measurement includes:

predicting a measurement of pixels in the second chip area using the second measurement performed on the first chip area.

14. The method as claimed in claim 12 , wherein predicting a measurement includes:

learning a prediction model based on a supervised learning algorithm, using the second measurement; and

predicting a measurement of pixels other than the interest area, using the learned prediction model.

15. The method as claimed in claim 14 , wherein the supervised learning algorithm includes a regression analysis algorithm or a classification algorithm.

16. A method for measuring a semiconductor device, the method comprising:

defining an interest area and an acceptable area in a chip area on a wafer;

performing a first measurement of the chip area with a spectral imaging device to acquire spectrum data of the chip area;

assuming a distribution of spectrum data of a first pixel in the acceptable area is a normal distribution;

calculating a distance from a central point on the normal distribution to second pixels in the interest area, wherein calculating a distance from the central point on the normal distribution to the second pixel includes:

calculating a Mahalanobis distance to the second pixel from a central pixel of the distribution of spectrum data of the first pixel; and

calculating a distance from the central point on the normal distribution for the second pixel, using the Mahalanobis distance;

selecting a position of a second pixel having a distance from the central point on the normal distribution greater than a predetermined range, among the second pixels, as a candidate position; and

performing a second measurement of the candidate position.