Calibrating and stabilizing spectropolarimeters with charge shuffling and daytime sky measurements

Well-calibrated spectropolarimetry studies at resolutions of R > 10 000 with signal-to-noise ratios (S/Ns) better than 0.01% across individual line profiles, are becoming common with larger aperture telescopes. Spectropolarimetric studies require high S/N observations and are often limited by instrument systematic errors. As an example, fiber-fed spectropolarimeters combined with advanced line-combination algorithms can reach statistical error limits of 0.001% in measurements of spectral line profiles referenced to the continuum. Calibration of such observations is often required both for cross-talk and for continuum polarization. This is not straightforward since telescope cross-talk errors are rarely less than ~1%. In solar instruments like the Daniel K. Inouye Solar Telescope (DKIST), much more stringent calibration is required and the telescope optical design contains substantial intrinsic polarization artifacts. This paper describes some generally useful techniques we have applied to the HiVIS spectropolarimeter at the 3.7 m AEOS Telescope on Haleakala. HiVIS now yields accurate polarized spectral line profiles that are shot-noise limited to 0.01% S/N levels at our full spectral resolution of 10 000 at spectral sampling of ~100 000. We show line profiles with absolute spectropolarimetric calibration for cross-talk and continuum polarization in a system with polarization cross-talk levels of essentially 100%. In these data the continuum polarization can be recovered to one percent accuracy because of synchronized charge-shuffling model now working with our CCD detector. These techniques can be applied to other spectropolarimeters on other telescopes for both night and daytime applications such as DKIST, TMT, and ELT which have folded non-axially symmetric foci