Surface reconstruction from microscopic images in optical lithography.

Area of Science:

• Materials Science
• Computer Vision
• Metrology

Background:

• Accurate 3D surface reconstruction of silicon wafers is crucial for semiconductor manufacturing quality control.
• Existing methods may lack the precision needed to capture subtle variations in wafer surfaces.
• Scanning Electron Microscopy (SEM) provides high-resolution 2D images, but 3D reconstruction requires advanced algorithms.

Purpose of the Study:

• To develop an advanced method for reconstructing the 3D surface topography of silicon wafers from SEM images.
• To integrate physical models of the SEM acquisition system with prior knowledge of circuit patterns.
• To quantify deviations between designed circuit patterns and actual manufactured surfaces.

Main Methods:

• A shape-from-shading technique incorporating a shape prior was employed.
• The reconstruction was framed as an optimization problem with data-fidelity and prior surface terms.
• An irradiance equation modeled the SEM acquisition, while smoothness and pattern shape priors were applied.
• A deformation field was introduced to account for manufacturing process variability and nonlinear elastic changes.

Main Results:

• The method successfully reconstructs 3D silicon wafer surfaces from 2D SEM images.
• Two key outputs were generated: the reconstructed 3D surface and a deformation field.
• The deformation field quantifies the deviation between expected circuit designs and the actual manufactured surfaces.

Conclusions:

• The developed method offers a robust approach to 3D surface reconstruction for silicon wafers.
• The technique provides valuable insights into manufacturing process deviations by mapping design to reality.
• This approach enhances metrology capabilities in semiconductor fabrication by combining imaging, physical models, and shape priors.