Visible and Mid-Infrared Supercontinuum Generation and Their Respective Application to 3D imaging and Stand-off Reflection Spectroscopy.

Abstract

The thesis describes broadband supercontinuum (SC) generation in optical fibers for both the visible and mid-infrared regions of the spectrum, and their respective application to 3D imaging and stand-off reflection spectroscopy. Both SC sources leverage mature telecom technology, and are based on a common all-fiber integrated platform comprising a ~1.55 micron distributed feedback seed laser diode amplified to high peak powers in two stages of cladding pumped Erbium or Erbium-Ytterbium fiber amplifiers. A visible SC extending from 0.45-1.20 microns with 0.74 W of time-averaged power is demonstrated using a two step process. The output of the Er-Yb power amplifier is frequency doubled to ~0.78 micron using a periodically poled lithium niobate crystal, followed by non-linear spectral broadening in 2m of high nonlinearity photonic crystal fiber. Numerical simulations based on solving the generalized non-linear Schrödinger equation are also presented to verify the underlying SC generation mechanisms and predict further improvements. The above SC source is used in a Fourier domain line scan interferometer to measure the height and identify shape defects of ~300 micron high solder balls in a ball grid array. The 3D imaging system has an axial resolution of ~125 nm, transverse resolution of ~15 microns, and an angular measurement range between 20 to 60 degrees depending on the sample surface roughness. The mid-infrared SC source is generated by pumping a 9m long ZrF4-BaF2-LaF3-AlF3-NaF (ZBLAN) fiber to obtain a spectrum spanning 0.8-4.3 microns with 3.9 W time-averaged power. The output power is linearly scalable with pump power, but requires optimization of the critical splices and thermal management of the gain fiber and pump diodes to ensure stable high power operation. Finally, an application of the mid-IR SC is demonstrated by measuring the diffuse reflection spectra of solid samples at a stand-off distance of 5 m and 100 ms integration time. The samples can be distinguished using a correlation algorithm based on distinct spectral features in the reflection spectrum. Signal to noise ratio calculations show that the distance is limited by space constraints in our lab and can be extended to ~150 m.