Relative and absolute intensity calibrations of a modern broadband echelle spectrometer

We report on relative and absolute intensity calibrations of a modern broadband echelle spectrometer (type ESA 3000(R) trademark of LLA Instruments GmbH, Berlin) for use in the diagnostics of low-temperature plasma. This type of device measures simultaneously complete emission spectra in the spectral range from 200 to 800 nm with a spectral resolution of several picometres by using more than 90 spectral orders, causing a strongly structured efficiency function. The assumptions and approximations entering the calibration procedure under these conditions are discussed in section 3. For coping with the strongly structured efficiency function a continuum light source is needed, which covers the entire spectral range. Furthermore, the variation of its intensity must be low enough to ensure that neither statistical errors perturb the calibration in regions with low photon flux and/or low efficiency, nor local memory overflow in regions with high photon flux or high efficiency. In our case this requires that during calibration over the whole spectral range of the spectrometer the counts per pixel in one measurement vary at highest by a factor 10 to 12. Usual broadband light sources do not meet this latter requirement. We, therefore, use an uncalibrated 'composite'source, an adjustable combination of a standard tungsten strip lamp and a deuterium lamp, and calibrate the spectrometer in a two-step process against the tungsten strip lamp and well-known rovibrational intensity distributions in the emission spectra of NO and N2. We adjust the composite source in a way to produce a perturbation-free first approximation of an (uncalibrated) efficiency function, which is then corrected and thus calibrated by comparison with the (secondary) standards mentioned above. For absolute calibration we use the tungsten strip lamp. The uncertainty attained in this way for the relative calibration depends on the wavelength and varies between 5% and 10%. For the absolute calibration we obtained an uncertainty of 12%. We further discuss problems caused by the non-uniform spectral efficiency and dispersion of the spectrometer, which complicate the calibration procedure.