Recognition of diffraction-grating profile using a neural network classifier in optical scatterometry

Area of Science:

• Optics and Materials Science
• Artificial Intelligence in Manufacturing

Background:

• Optical scatterometry is crucial for semiconductor metrology, but inverse problems require predefined models.
• Accurate geometrical identification of diffraction gratings is essential for process control.

Purpose of the Study:

• To develop an artificial neural network (ANN) classifier for identifying diffraction grating geometry.
• To enable pre-characterization identification of grating structures using ellipsometric signatures.

Main Methods:

• Development of an ANN classifier trained on ellipsometric signatures.
• Testing the ANN on manufactured sinusoidal photoresist and trapezoidal SiO2 gratings.
• Utilizing optical scatterometry principles for structural analysis.

Main Results:

• The ANN successfully identified the geometry of both sinusoidal and trapezoidal diffraction gratings.
• Ellipsometric signatures were effectively correlated with specific grating structures.
• The method provides a rapid pre-characterization step.

Conclusions:

• ANN-based classification of ellipsometric data is a viable method for identifying diffraction grating geometry.
• This approach can enhance the efficiency of metrology in semiconductor fabrication.
• The developed classifier shows promise for automated process control.