Abstract
Optical nanocavities formed by defects in a two-dimensional photonic crystal (PC) slab can simultaneously realize a very small modal volume and an ultrahigh quality factor (Q). Therefore, such nanocavities are expected to be useful for the enhancement of light–matter interaction and slowdown of light in devices. In the past, it was difficult to design a PC hole pattern that makes sufficient use of the high degree of structural freedom of this type of optical nanocavity, but very recently, an iterative optimization method based on machine learning was proposed that efficiently explores a wide parameter space. Here, we fabricate and characterize an L3 nanocavity that was designed by using this method and has a theoretical Q value of 29 × 106 and a modal volume of 0.7 cubic wavelength in the material. The highest unloaded Q value of the fabricated cavities is 4.3 × 106; this value significantly exceeds those reported previously for an L3 cavity, i.e., ≈2.1 × 106. The experimental result shows that the iterative optimization method based on machine learning is effective in improving cavity Q values.
Highlights
Optical nanocavities that are based on artificial defects in a twodimensional (2D) photonic crystal (PC) slab have attracted much attention as device elements for light control
We have reported on the fabrication and performance characterization of an L3 cavity with a design Q value of 29 × 106
The design was obtained in a previous study by optimizing 25 degrees of freedom of the L3 cavity using an iterative optimization method based on machine learning
Summary
Optical nanocavities that are based on artificial defects in a twodimensional (2D) photonic crystal (PC) slab have attracted much attention as device elements for light control. Because the prediction in the above-mentioned machine-learning method becomes more incorrect as the new pattern departs from the vicinity of the parameter space region of the initial training dataset, it has not been possible to apply this method to optimizations where large shifts in the hole positions are required to find a significantly improved cavity (In the case of the above-mentioned results, the largest amount of shift was about 2%–3% of the lattice constant.) To solve this issue, a method has been proposed that determines the Qdesign values of the final candidate structures by a first-principles method, adds them to the training dataset, and iterates these steps (i.e., generates a regression function, improves the candidates, evaluates their Qdesign values by first-principles, and adds these values to the training dataset).. These values significantly exceed the corresponding values of the previously reported L3 cavities, i.e., about 2.1 × 106 and 1.1 × 106, respectively (see Table I)
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
