Modeling bias in supermassive black hole spin measurements

X-ray reflection spectroscopy (or iron line method) is a powerful tool to probe the strong gravity region of black holes, and currently is the only technique for measuring the spin of the supermassive ones. While all the available relativistic reflection models assume thin accretion disks, we know that several sources accrete near or above the Eddington limit and therefore must have thick accretion disks. In this work, we employ the Polish donut model for the description of thick disks. We thus estimate the systematic error on the spin measurement when a source with a thick accretion disk is fitted with a thin disk model. Our results clearly show that spin measurements can be significantly affected by the morphology of the accretion disk. Current spin measurements of sources with high mass accretion rate are therefore not reliable.