3:15 PM - 3:30 PM
[13p-N321-7] Designing L3 nanocavity with machine learning by asymmetrically shifting the air holes (II) ~Demonstration of experimental Q value of 160,000~
Keywords:photonic crystal, machine Learning, L3 nanocavity
In recent years, a method of optimizing the air hole position of a two-dimensional photonic crystal nanoresonator by utilizing machine learning has attracted attention. In order to achieve a high Q value, it is effective to emphasize the suppression of radiation loss based on rotational symmetry. On the other hand, by introducing structural asymmetry into photonic crystals, it is possible to control optical characteristics other than the Q value, such as polarization and radiation patterns, so it is important to develop a method for simultaneously optimizing various optical characteristics. A structure in which the air hole positions are asymmetrically shifted has a higher degree of freedom in design than a symmetric structure. This high degree of freedom may be advantageous when optimizing multiple figure of merit including Q value in a well-balanced manner or when searching for a structure that is robust against air hole fluctuations. Last time, we reported a design Q value of over 200,000 in an L3 resonator with asymmetrically shifted air holes by design using machine learning. This time, we report the designed L3 resonator on the SOI substrate and evaluated its optical characteristics.