Inverse design of a binary waveguide crossing by the particle swarm optimization algorithm

Abstract

This paper describes the inverse design of the particle swarm optimization algorithm combined with the three-dimensional finite-difference time-domain simulation to design a waveguide crossing that resembles a binary code. The device consists of 15 × 15 air holes in a 220-nm silicon slab on a 3-μm-thick SiO2 substrate with one input port and three output waveguide ports. The designed device has a small footprint of 4 μ m 2 and a short simulation time of 1.7 h. In the wavelength range of 1.5– 1 . 6 μ m , the device has insertion loss I L < 0 . 85 dB and crosstalk X T < − 14 . 5 dB. This device tolerates air hole-position disordering of η p = 23%, and hole-radius disordering of η r = 35%, so it is appropriate to use in the complementary metal–oxide–semiconductor fabrication process. The paper also proposes integrated interconnections that contain 2 × 2 waveguide crossings and have I L < 2 . 9 dB and X T < − 13 dB, and 3 × 3 waveguide crossings that have I L < 5 and X T < − 12 . 5 dB.

Keywords

• Particle swarm optimization algorithm
• Waveguide crossing
• Disordering