High-Q Silicon Photonic Crystal Ring Resonator Based on Machine Learning

We propose and demonstrate a silicon-based photonic crystal ring resonator (PCRR) with a high quality (Q) factor based on machine learning. The elliptical optimization of the key holes is exploited to effectively reduce the tangential k-vector component inside the leaky region, contributing to a significant improvement in the Q factor. To further enhance the optimization efficiency, we propose a novel approach that combines the optimization of the elliptical holes with machine learning techniques (including the backpropagation neural network, grey wolf optimizer algorithm and genetic algorithm). Consequently, the high Q factors of the PCRRs are efficiently explored. To the best of our knowledge, it is the first time to realize the record theoretical Q factors beyond one million for the silicon PCRRs with a compact radius of 2.1 mu m, and the experimental Q factor of 7.67 x 10(5) is three times larger than the previously reported highest values. The proposed PCRR exhibits various merits such as a high Q factor, excellent mode flexibility, strong structural scalability and good tolerance, making it widely applicable in the important fields of filtering, laser sources and sensing. More importantly, the proposed optimization model can be extended to the efficient optimization designs of other microcavities.