GAGA for nonreciprocal emitters: genetic algorithm gradient ascent optimization of compact magnetophotonic crystals.

Fundamental limits of thermal radiation are imposed by Kirchhoff's law, which assumes the electromagnetic reciprocity of a material or material system. Thus, breaking reciprocity can enable breaking barriers in thermal efficiency engineering. In this work, we present a subwavelength, 1D photonic crystal composed of Weyl semimetal and dielectric layers, whose structure was optimized to maximize the nonreciprocity of infrared radiation absorptance in a planar and compact design. To engineer an ultra-compact absorber structure that does not require gratings or prisms to couple light, we used a genetic algorithm (GA) to maximize nonreciprocity in the design globally, followed by the application of the numerical gradient ascent (GAGA) algorithm as a local optimization to further enhance the design. We chose Weyl semimetals as active layers in our design as they possess strong, intrinsic nonreciprocity, and do not require an external magnetic field. The resulting GAGA-generated 1D magnetophotonic crystal offers high nonreciprocity (quantified by absorptance contrast) while maintaining an ultra-compact design with much fewer layers than prior work. We account for both s- and p-polarized absorptance spectra to create a final, eight-layer design suitable for thermal applications, which simultaneously minimizes the parasitic, reciprocal absorptance of s-polarized light.