High-performance anti-reflective coatings for TARSIS on the near UV range

In this work, we present a procedure to obtain optimized designs of the antireflection coatings of TARSIS optics. TARSIS, acronym for Tetra-Arm Super-Ifu Spectrograph, is a new Integral Field Spectrograph that incorporate image slicers. It will explore the sky in the near ultraviolet with an unprecedented field of view, and will be installed on the 3.5m telescope of the Calar Alto Observatory. In astronomical observations, we aim to maximize the amount of light of a telescope that reach the final stage of the instrument. However, the blue-UV range is particularly challenging because the optical materials available, for both the lenses and the coatings, are limited due to light absorption. Times increase dramatically, and the design and manufacturing processes are more complex. Particularly for coatings, we find the best design that maximize optical throughput combined with high robustness to manufacturing tolerances. TARSIS requires 40 different coatings designs. We achieve multi-layer coatings with: 1) very high transmittance, 2) at the mean angle on convex and concave surfaces, 3) minimum sensitivity to thin film deposition tolerances, and 4) minimum number of layers. To reflect the performance of the multi-layer coatings, we defined a merit function. We conducted a global search to find the best high-transmittance solutions with the fewest layers. We introduced random thickness variations into each solution in order to evaluate the effect on the merit function. We have developed a pseudo-color evaluation tool that shows design reliability over wavelength. This visualization allowed us to directly compare the performance of equivalent solutions. Finally, we evaluated that the optimized solutions did in fact provide a high level of transmission over a range of incidence angles. In summary, our designs have a high average transmittance > 99%, just four layers and are highly insensitive to thickness variations which speed-up the manufacturing process.