Wafer-scale waveguide sidewall roughness scattering loss characterization by image processing.

Area of Science:

• Photonics and Materials Science
• Optoelectronics and Integrated Circuits
• Nanotechnology and Surface Science

Background:

• Photonic integrated circuits (PICs) are crucial for high-performance optoelectronic devices, but optical losses from waveguide roughness limit their efficiency.
• Existing methods like Atomic Force Microscopy (AFM) for roughness measurement are costly and not scalable for industrial applications.
• Sidewall roughness, a major contributor to scattering losses, is typically larger than surface roughness in UV-lithography-based fabrication.

Purpose of the Study:

• To develop and validate a scalable method for characterizing optical loss in thin-film waveguides using 2D SEM imaging.
• To correlate scattering loss estimations from 2D SEM image analysis with theoretical models.
• To establish a new benchmark for optical loss in waveguides at visible wavelengths.

Main Methods:

• Utilized 2D high-resolution scanning electron microscope (SEM) imaging to evaluate sidewall roughness profiles of waveguides.
• Applied theoretical Payne model to estimate scattering loss from 2D image-based roughness profiles.
• Characterized optical loss in homemade nitride and oxide thin-film waveguides on silicon wafers.

Main Results:

• Achieved a record low optical loss of 0.075 dB/cm for the fundamental transverse electric (TE0) mode at 633 nm across 24 waveguide devices.
• Demonstrated 100% success in edge detection for image processing, enabling accurate estimation of autocorrelation functions and optical mode loss.
• Showcased strong correlation between experimental loss measurements and estimations derived from 2D SEM image analysis.

Conclusions:

• 2D SEM imaging provides a scalable and effective approach for characterizing waveguide sidewall roughness and estimating optical losses in PICs.
• The developed method offers valuable insights for optimizing fabrication processes to minimize optical losses.
• This work paves the way for efficient wafer-scale loss characterization in the manufacturing of PICs.