Sub-100-nW-threshold Raman silicon laser designed by a machine-learning method that optimizes the product of the cavity Q-factors.

Abstract

Raman silicon lasers based on photonic crystal nanocavities with a threshold of several hundred microwatts for continuous-wave lasing have been realized. In particular, the threshold depends on the degree of confinement of the excitation light and the Raman scattering light in the two nanocavity modes. Here, we report lower threshold values for Raman silicon nanocavity lasers achieved by increasing the quality (Q) factors of the two cavity modes. By using an optimization method based on machine learning, we first increase the product of the two theoretical Q values by a factor of 17.0 compared to the conventional cavity. The experimental evaluation demonstrates that, on average, the actually achieved product is more than 2.5 times larger than that of the conventional cavity. The input-output characteristic of a Raman laser with a threshold of 90 nW is presented and the lowest threshold obtained in our experiments is 40 nW.