Inverse-designed grating couplers for hybrid silicon nitride photonics using metaheuristic methods

Silicon nitride (Si 3 N 4 ) has emerged as an attractive photonic platform for implementing passive optical components due to its ultra-low propagation losses and broad transparency window. However, its low refractive index contrast presents persistent design challenges for low-loss surface grating couplers, resulting in poor radiation and sub-optimal grating-to-fiber mode matching. High-index overlays situated atop the Si 3 N 4 structure are a promising strategy to tackle these limitations. Although this hybrid configuration offers enhanced design flexibility, it also introduces additional complexity into the design space, rendering conventional optimization methods inefficient. To address this, in this work, we investigate inverse-designed grating couplers employing metaheuristic optimization techniques tailored for the hybrid α −Si/Si 3 N 4 platform. In particular, we apply three distinct metaheuristic algorithms—particle swarm optimization (PSO), grey wolf optimization (GWO), and genetic algorithm (GA)—in conjunction with finite-difference time-domain (FDTD) simulations. Our results demonstrate that all three methods can effectively explore the multi-dimensional design space and achieve low fiber-chip coupling losses at the telecom waveband. This work establishes a scalable inverse design framework for the development of high-performance grating couplers on hybrid Si 3 N 4 platforms, demonstrating the practical viability of metaheuristic optimization for low-loss input/output optical interfaces. The proposed approach not only advances current coupling performance but also paves the way for future innovations in the automated design of photonic chip interfaces for next-generation on-chip systems.