|
As calibrations of AXAF mirrors in the 2-12 keV range are near completion, we can report that the archive is complete for the scheduled number of mirrors over the 5-12 keV range, and also over the 2010-6200 eV range. Analysis of the results for optical constants and coating layer parameters is now proceeding rapidly. To date, we have derived optical constants from 39 mirrors over 5-12 keV using our existing Fresnel Equation model, with uniform layers of Ir, Cr, and Zerodur, and the roughness algorithm of Nevot and Croce. The analysis method has been presented in our earlier papers, but its application has been expanded to all three varieties of witness mirrors use in AXAF's qualification and production coating runs. For the first and most consistent variety, which have 1 angstrom roughness, reflectances are indistinguishable from mirror to mirror save for thickness variations between coating runs. Residuals of the fits for optical constants become large when reflectance values below 18 percent are included in the fits. If such data points are ignored, values of (delta) (E) and (beta) (E) very much like those of Henke et al. over 5-11 keV are obtained. Residuals are at the 0.6-0.8 percent level, which meets calibration requirements but exceeds experimental noise. For the second variety of witness mirror, polished to obtain roughness specification similar to the flight mirrors, fits approximately within the noise level of the measurements may be obtained over 5-11 keV, for reflectance values down to 5 percent. The (delta) (E) is essentially the same as that obtained from the 1 angstrom flats; however (beta) (E) is higher by approximately 3-4 percent systematically, and the mirror-to-mirror variation is larger. The third variety of mirror was obtained with 5-7 angstrom nominal roughness. Residuals to the fits are large for the entire range of angles, in some case exceeding 2 percent with a sinusoidal character through the critical angle. Coating layer depths are similar to those found for the other mirror types, with good precision. We discuss results along with possible improvements to the model and experiments to verify it.
|
