Highly efficient variable step-size binary search algorithm for ultra-compact cyclic three-mode converters

The recent advancements in nano-optics have created a strong demand for ultra-compact, miniaturized photonic devices. However, the mainstream inverse design algorithms face challenges in achieving compact size, efficient performance, and superior process manufacturability simultaneously. The present study proposes an exceptionally efficient segmented hierarchical variable step-size binary search-based digital inverse design method, aiming to transcend these limitations. The concept of “pixel splitting” is introduced for the first time in this paradigm. By layering the optimization process in several segments, the size of each pixel also changes dynamically, achieving both high efficiency and superior performance. The chosen examples of cyclic mode converters serve as typical complex proof-of-concept structures. The simulation results demonstrate that this method not only maintains excellent manufacturability but also enhances design efficiency over three times and exhibits superior performance. The experimental demonstration of two types of cyclic mode converters (CMCs), with dimensions of 3.84 × 9.6 μm2 and 3.84 × 12 μm2, respectively, based on this model, showcases their significantly smaller footprints compared to previously reported devices. The characterizations of these two CMCs demonstrate minimal insertion losses (<2.86 and <3.26 dB) and minimal crosstalk (<−11.89 and <−9.24 dB) across bandwidths of 50 nm. The data transmission of 3 × 60 Gbps on-off-keying signals and 3 × 80 Gbps four-level pulse amplitude signals are also demonstrated successfully. The proposed variable step-size binary search offers a novel approach for efficiently designing compact digital nano-photonic devices, addressing limitations such as slow convergence, sensitivity to initial patterns, and susceptibility to local optima. In addition, ultra-compact CMCs provide an effective solution for reducing link damage in mode division multiplexing transmission systems.