Absolute calibration method for nanosecond-resolved, time-streaked, fiber optic light collection, spectroscopy systems

Abstract

Summary form only given. Presented is a convenient method to calibrate fast (<;1ns resolution) streaked, fiber optic light collection, spectroscopy systems. Such a system is used to collect spectral data on plasmas generated in the vacuum gap of electron beam diodes fielded on the RITS-6 accelerator (8-12MV, 140-200kA). On RITS, plasma light is collected through a small diameter optical fiber and recorded on a fast streak camera at the output of 1m Czerny-Turner monochromator. To calibrate such a system, it is necessary to efficiently couple light from a spectral lamp into the small diameter fiber, split it into its spectral components (10 Angstroms or less resolution), and record it on a streak camera with 1ns or less temporal resolution. One method of doing this is with a DC short arc lamp. For this paper, a 300W xenon arc lamp was used. Since the radiance of the xenon arc varies from the cathode to the anode, just the area around the tip of the cathode (“hotspot”) was imaged onto the fiber, producing the highest intensity output. To compensate for chromatic aberrations, the signal was optimized at each wavelength measured. Output power at each wavelength was measured using 10nm bandpass interference filters and a calibrated photodetector. These measurements give power at discrete wavelengths across the spectrum, and when linearly interpolated, provide a calibration curve for the lamp. The fiber is then attached to the entrance of the monochromator and a spectrum is taken. The shape of the spectrum is determined by the collective responsivity of the optics, monochromator, and streak tube across the spectral region of interest. The ratio of this curve to the measured bandpass filter curve at each wavelength produces a correction factor (Q) curve. This curve is then applied to the experimental data and the resultant spectra are given in absolute intensity units (photons/sec/cm <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> /steradian/nm). Error analysis shows this method to be accurate to within +/-20%.