High Energy Pulse Amplification in Mid-IR Range Using Er:ZBLAN Fibers

Abstract

Laser sources operating in Mid-IR (2 – 5 µm) spectral range have many important applications in spectroscopy (e.g. environmental monitoring), medicine (e.g. laser surgery), and fundamental science (e.g. attosecond pulse generation and metrology). Fiber lasers offer technological advantages of high efficiency, compactness, stability, and high mode quality, but typically operate in Near-IR spectral range, because they are based on technologically-mature fused-silica glass which is ill-suited for signal transmission beyond 2µm.

This thesis explores high energy pulse generation in Mid-IR (~3µm) using ZBLAN glass based optical fiber lasers. ZBLAN glasses are becoming the material of choice for fiber lasers accessing mid-IR, since their transmission-window long-wavelength edge extends to ~4 µm, much longer than that of fused-silica glass. Our work focused on Er-doped ZBLAN fibers, which offer significant practical advantages, such as compatibility with standard telecom-grade pump diodes operating at 980nm, where pump-to-signal conversion efficiency can exceed quantum defect limit due to beneficial energy transfer up-conversion processes, and thus leads to record high average powers in Mid-IR.

The main achievement of this work is that we extended ns-scale pulsed energies achievable with Er:ZBLAN fiber by an order of magnitude (from ~10^2 µJ to ~1mJ) while preserving diffraction-limited (i.e. single transverse mode) output beam quality. This was enabled by using Er:ZBLAN fibers with core sizes of 30µm and 50µm which significantly exceed single-mode limit, and developing techniques of preserving single-mode propagation of high energy pulses in these large mode area fibers. This work serves as the basis for the ongoing work on developing the first femtosecond-pulse fiber CPA system operating in Mid-IR spectral region, as well as for future work on spatial and temporal coherent combining of multiple Mid-IR fiber lasers.

As a related effort, we proposed and simulated several novel machine learning algorithms to optimize coherent pulse stacking (CPS) performance. CPS is a coherent time domain pulse combining technique. Among all methods, we find that the MSPGD achieves best performance and can be potentially advantageous for CPS to increase the system operation efficiency.