[14a-B415-1~9] 3.11 Photonic structures and phenomena

[14a-B415-1] Experimental evaluation of Raman silicon nanocavity laser designed by machine learning

〇(M1)Taro Kawakatsu¹, Takashi Asano², Susumu Noda², Yasushi Takahashi¹ (1.Osaka Pref. Univ., 2.Kyoto Univ.)

Keywords:photonic crystal, Raman scattering, machine learning

We have developed a silicon Raman laser using a two-dimensional photonic crystal nanocavity. This device achieves a low threshold of 1 micro Wat or less by confining pump light and Stokes Raman scattered light in two high Q-value resonance modes (referred to as pump mode and Stokes mode) formed in a heterostructure nanocavity I do. The minimum lasing threshold reported so far was 120 nW, the experimental Qp value in the pump mode was 400,000, and the experimental QS value in the Stokes mode was 2.35 million. It is important to raise the Q value of the two resonance modes and further lower the oscillation threshold in order to expand the possibilities of basic and applied research of this device. In recent years, it has become possible to increase the design Q value of two modes simultaneously by using machine learning. In this study, we experimentally evaluated the Q factor and threshold of a nanocavity silicon Raman laser designed by machine learning.