Femtosecond supercontinuum source in the C-band with equalized spectra through evolutionary strategy based adaptive optimization of nonlinear spectral broadening

Abstract

Broadband femtosecond supercontinuum sources find applications in fields such as Optical Coherence Tomography, fluorescence lifetime imaging, and frequency metrology. A mechanism to achieve the required spectral bandwidth is to broaden the output of a femtosecond laser source in nonlinear media such as highly nonlinear fibers (HNLF) utilizing a combination of nonlinear effects such as self-phase modulation (SPM) and four-wave mixing (FWM). However, conventional spectral broadening often suffers from supercontinua with degraded spectral flatness. The profile of the broadened spectrum depends on the properties of the medium, as well as the power and the temporal profile of the input pulse. The pulse can be shaped before broadening to improve the supercontinuum spectrum. However, the envelope is highly sensitive to the pulse spectral phase, potentially time-varying, resulting in a sub-optimal performance with any single pass optimization approach. Here, we overcome this by adaptively optimizing the input pulse by perturbing the spectral phase in an automated closed-control loop. A Fourier pulse shaper modifies the C-band sub-picosecond pulses from a mode-locked fiber laser before spectral broadening in HNLF. An evolutionary strategy algorithm is used to process the measured spectrum and adaptively optimize the spectral phase to realize a smooth supercontinuum with a broad Gaussian spectrum iteratively. We allowed the spectral phase to evolve with multiple variables across the pulse. We achieved a 4X bandwidth enhancement of the input pulse with high fidelity between the supercontinuum spectra and the target Gaussian shape. Spectral fluctuations were <3dB across the bandwidth of the generated supercontinuum.