Deep learning neural network designed large dimensional neutron focusing metalens

Neutron focusing lenses are very important for various applications in fundamental science, industry, or even medicine. Conventional Fresnel zone plates (FZP) were successfully used as neutron focusing lenses. However, several difficulties are arisen when larger dimension (at the order of 10 cm) focal lenses are demanded for some applications. Here, we proposed a neuron network metalens design method based on D²NN algorithm. In this work, several neutron power focusing metalenses were designed for neutron energies from cold to epithermal, with size lengths from a few mm to 140 mm, and a focal length range from 60 mm to 13.5 m. Based on our simulations, clear focusing patterns were observed and were consistent with previous experiments. The estimation value of power focusing efficiency could be as high as 20%.