Differentiable wave propagation method for shape optimization of freeform optics beyond the paraxial approximation

Area of Science:

• Optics and Photonics
• Computational Science
• Materials Science

Background:

• Freeform optics enable precise light control for beam shaping and mode conversion.
• Designing micro-optical elements demands efficient, low-runtime simulation methods.
• Traditional wave propagation methods (WPM) are fast but lack differentiability for inverse design.

Purpose of the Study:

• To develop a differentiable formulation of WPM for gradient-based inverse design.
• To enable efficient simulation of volumetric micro-optical elements.
• To overcome limitations of paraxial methods in simulating high diffraction angles.

Main Methods:

• Formulating a differentiable wave propagation method (WPM).
• Utilizing smoothstep functions for soft spatial domain partitioning.
• Implementing gradient-based optimization for inverse design.

Main Results:

• Achieved a differentiable WPM framework for micro-optics inverse design.
• Successfully simulated a freeform hologram with high diffraction angles.
• Demonstrated computational efficiency and general applicability.

Conclusions:

• The differentiable WPM provides an efficient and general framework for micro-optics inverse design.
• This method overcomes the limitations of non-differentiable WPM and traditional paraxial approaches.
• Enables the design of complex micro-optical elements previously unattainable.