What powers the 3000-day light curve of SN 2006gy?

SN 2006gy was the most luminous supernova (SN) ever observed at the time of its discovery and the first of the newly defined class of superluminous supernovae (SLSNe). The extraordinary energetics of SN 2006gy and all SLSNe (\u3e1051 erg) require either atypically large explosion energies (e.g. pair-instability explosion) or the efficient conversion of kinetic into radiative energy (e.g. shock interaction). The mass-loss characteristics can therefore offer important clues regarding the progenitor system. For the case of SN 2006gy, both a scattered and thermal light echo from circumstellar material (CSM) have been reported at later epochs (day ~800), ruling out the likelihood of a pair-instability event and leading to constraints on the characteristics of the CSM. Owing to the proximity of the SN to the bright host-galaxy nucleus, continued monitoring of the light echo has not been trivial, requiring the high resolution offered by the Hubble Space Telescope (HST) or ground-based adaptive optics (AO). Here, we report detections of SN 2006gy using HST and Keck AO at ~3000 d post-explosion and consider the emission mechanism for the very late-time light curve.While the optical light curve and optical spectral energy distribution are consistent with a continued scattered-light echo, a thermal echo is insufficient to power the K\u27-band emission by day 3000. Instead, we present evidence for late-time infrared emission from dust that is radiatively heated by CSM interaction within an extremely dense dust shell, and we consider the implications on the CSM characteristics and progenitor system