Inverse design of 3D nanophotonic devices with structural integrity using auxiliary thermal solvers.

3D additive manufacturing enables the fabrication of nanophotonic structures with subwavelength features that control light across macroscopic scales. Gradient-based optimization offers an efficient approach to design these complex and non-intuitive structures. However, expanding this methodology from 2D to 3D introduces complexities, such as the need for structural integrity and connectivity. This work introduces a multi-objective optimization method to address these challenges in 3D nanophotonic designs. Our method combines electromagnetic simulations with an auxiliary heat-diffusion solver to ensure continuous material and void connectivity. By modeling material regions as heat sources and boundaries as heat sinks, we optimize the structure to minimize the total temperature, thereby penalizing disconnected regions that cannot dissipate thermal loads. Alongside the optical response, this heat metric becomes part of our objective function. We demonstrate the utility of our algorithm by designing two 3D nanophotonic devices. The first is a focusing element. The second is a waveguide junction, which connects two incoming waveguides for two different wavelengths into two outgoing waveguides, which are rotated by 90° to the incoming waveguides. Our approach offers a design pipeline that generates digital blueprints for fabricable nanophotonic materials, paving the way for practical 3D nanoprinting applications.