ABSTRACT
"Physics-Based Design Methods for Adiabatic Mode Transformation in Integrated Photonics"
Integrated photonics is a growing field of Applied Physics, and photonic integrated circuits (PICs) have reached very-large-scale integration (VLSI), with more than 105 components demonstrated today. This growth is also reflected by the expanding PIC market, which has been reported to have a Compound Annual Growth Rate (CAGR) of approximately 10.8%. However, as the scaling and applications of PICs continue to grow, so will the limitations of current design methodologies. The Integrated Photonic Systems Roadmap–International (IPSR-I) outlines several challenges and scaling gaps between laboratory prototypes and ready-to-use commercial systems. Several of these challenges are linked to mode transformation theory, including the efficient transfer of optical power between fibers, waveguides, layers, materials, and guided modes.
Linear inverse tapers are widely used as a baseline because they are simple and robust. More advanced adiabatic profiles can provide enhanced performance, but their usefulness depends on how well the design method captures the local physics of modal evolution. Many current approaches to this challenge avoids physics-driven design by using simple geometric profiles, empirical optimization, or full-wave simulation sweeps. Simulation-heavy approaches such as FDTD are powerful but can be expensive for routine design generation, while some semi-analytic or local design methods can fail when their local assumptions break down. Therefore, there is a need for design workflows that connect taper geometry directly to the physics of mode coupling and radiation loss, rather than treating the taper shape only as a geometric or numerical optimization variable.
This dissertation develops and compares four design methods for adiabatic mode trans-formation, using silicon-nitride Si3N4 edge couplers as the testing platform and a standard linear edge coupler as the reference design. Two of these methods are numerical and provide direct optimization or validation references. These are the Numerical Adiabatic Mode Evolution Structure (NAMES) and the finite-difference time-domain (FDTD)-derived optimization. NAMES uses sectional EME loss minimization and is historically important as a proof-of-concept design method. The FDTD-derived method is developed in this work as a propagation-based optimization and refinement approach for the taper. The other two methods are physics-based design methods: the monotonic mode-tracking, or Mono, approach and the coupled local-mode theory (CMT) method. Mono uses local overlap sensitivity to develop a local adiabaticity, or loss, term for its design rule, while CMT models coupling from the target TE00 mode to the radiation continuum. The CMT method is the main physics-driven contribution of this dissertation. Together, these workflows are evaluated using theory, Lumerical simulations, fabricated devices, and experimental measurements.
Across the studied platforms, the adiabatic taper designs are evaluated against linear references using both simulation and experimental measurement. The AIM platforms provide the strongest validation cases for the physics-based workflows, showing that adiabatic tapering can improve simulated performance and provide useful design guidance when the fabricated device and launch conditions are close to the simulation assumptions. The Applied Nanotools comparison shows a more complicated case, where simulations predicted improvement from the adiabatic profiles, but the measured results showed substantial disagreement with the models. This result is not interpreted as a failure of CMT, but as evidence that lossless models can be incomplete when taper-region loss, launch distortion, or polished-facet quality differ from the assumed model. Among the methods studied, CMT is identified as the most promising future workflow because it directly connects the taper shape to coupling between the target mode and the radiation continuum. More broadly, this work provides a physics-based design and comparison framework for adiabatic mode transformers beyond edge couplers, including transitions between waveguide dimensions, material platforms, layers, and guided-mode families.
Advisor: Doug Petkie
Committee members: James Eakin, Raisa Trubko, Lyubov Titova and Sathwik Bharadwaj
Zoom link: https://wpi.zoom.us/j/91527422389