Efficiency-speed tradeoff in slow-light silicon photonic modulators

he purpose of this paper is twofold. First, we discuss the efficiency-speed tradeoff in slow-light (SL) silicon photonic (SiP) modulators. For this, a comprehensive model for the electro-optic (EO) response of lumped-electrode SL Mach-Zehnder modulators (SL-MZMs) is presented. The model accuracy is verified by comparing it to experiments. Our analysis shows that slowing down the optical wave helps to enhance efficiency by increasing the interaction time between the optical wave and the uniform voltage across lumped electrodes, but at the cost of limiting the EO bandwidth. Then, we investigate SL-MZMs with traveling-wave (TW) electrodes whose dynamic interaction is predicted using a distributed circuit model. Having been solved by the finite-difference time-domain (FDTD) method, the model shows that TW SL-MZMs are capable of improving both efficiency and speed under an optimized SL effect. We also compare SL-MZMs with conventional MZMs (C-MZM) considering a figure of merit (FOM) that combines key parameters such as efficiency, loss, and EO bandwidth. We show that the additional loss of SL waveguides significantly impacts the preferred modulator choice at different baudrates. The second aim of this paper is to examine different design strategies to reduce Vπ of C-MZMs in order to meet the requirement of COMS driver using 1) a longer phase shifter, 2) higher doping densities, and 3) the SL effect. It is shown that the SL effect provides the best overall performance among the three. Indeed, only the SL effect offers simultaneous improvement in Vπ, footprint, and EO bandwidth; the other approaches provide Vπ reduction but at the cost of reduced speed or enlarged footprint (or even both).