Computational polarized Raman microscopy on sub-surface nanostructures with sub-diffraction-limit resolution.

Area of Science:

• Materials Science
• Nanotechnology
• Optical Physics

Background:

• Raman microscopy is a powerful tool for chemical analysis but typically limited by diffraction.
• Imaging buried nanostructures with high resolution remains a significant challenge in materials characterization.

Purpose of the Study:

• To demonstrate Raman microscopy with sub-diffraction resolution for imaging sub-surface nanostructures.
• To extract chemical, strain, and temperature information from buried nanostructures.
• To improve spatial resolution while maintaining the flexibility of Raman microscopy.

Main Methods:

• Utilizing polarized Raman microscopy optimized for edge enhancement and nanostructure contrast.
• Implementing fast computational deconvolution methods, specifically the cosine transform method (O(Nlog N) complexity).
• Testing the technique on CMOS poly-Si nanostructures buried beneath complex dielectrics (0.3–6 µm depth).

Main Results:

• Achieved sub-diffraction resolution imaging of nanostructures several microns below the sample surface.
• Demonstrated accurate extraction of chemical, strain, and temperature details.
• Computational deconvolution significantly reduced processing time (below image acquisition time).
• Relative errors for feature sizes, chemical concentrations, and fill factors were approximately 10% compared to ground truth.
• Absolute error for the smallest 230 nm poly-Si feature was approximately 25 nm.

Conclusions:

• The combined approach of polarized Raman microscopy and fast computational deconvolution enables high-resolution imaging of buried nanostructures.
• This technique overcomes diffraction limits and offers a flexible, efficient method for nanoscale characterization of subsurface materials.
• The method shows high accuracy, making it suitable for analyzing complex semiconductor devices and other buried nanostructures.