Deep learning enabled robust wavefront sensing for active beam smoothing with a continuous phase modulator

Abstract

In laser systems requiring a flat-top distribution of beam intensity, beam smoothing is a critical technology for enhancing laser energy deposition onto the focal spot. The continuous phase modulator (CPM) is a key component in beam smoothing, as it introduces high-frequency continuous phase modulation across the laser beam profile. However, the presence of the CPM makes it challenging to measure and correct the wavefront aberration of the input laser beam effectively, leading to unwanted beam intensity distribution and bringing difficulty to the design of the CPM. To address this issue, we propose a deep learning enabled robust wavefront sensing (DLWS) method to achieve effective wavefront measurement and active aberration correction, thereby facilitating active beam smoothing using the CPM. The experimental results show that the average wavefront reconstruction error of the DLWS method is 0.04 μm in the root mean square, while the Shack–Hartmann wavefront sensor reconstruction error is 0.17 μm.

Keywords

• beam smoothing
• continuous phase modulator
• deep learning
• wavefront sensing