Efficient Verilog-A based time-domain variability analysis method for passive photonic integrated circuits

The performance of silicon photonic integrated circuits (PICs), especially wavelength filters, can be highly sensitive to variations in the fabrication process due to the large refractive index contrast of the silicon on insulator platform. This paper proposes an easy-to-implement and efficient time-domain variability analysis method for passive PICs. The method utilizes the polynomial chaos expansion technique to construct Verilog-A based models for estimating the statistical information of stochastic passive PICs. In comparison to existing methods, this approach is considerably easy to implement, efficient, and exhibits superior scalability, particularly as the numbers of ports and random parameters in the studied PICs increase. The technique is demonstrated via the time-domain variability analysis of a ring-resonator-based wavelength filter and a Mach-Zehnder interferometer-based demultiplexer filter.