OUP accepted manuscript

Abstract

Recent monitoring campaigns designed to map the accretion regime in AGN show major discrepancies with models where the optical/ultraviolet (UV) is produced by X-ray-illuminated, optically thick disc material within a few hundred gravitational radii. However, these campaigns only monitored X-rays below $10$ keV, whereas the bolometric luminosity for most of these AGN peaks above $50$ keV. We use data from the recent multiwavelength campaign by \cite{E17} on NGC 4151 - the only AGN bright enough to be monitored at higher energies with \textit{Swift} BAT. We develop a spectral-timing model with a hot corona, warm Comptonisation, and outer standard disc. This fits the time-averaged spectrum well, but completely fails to match the UV variability predicted from the X-ray lightcurve. However, it reveals that NGC 4151 had a bolometric luminosity around $1.4\%$ of the Eddington luminosity during this campaign, close to the luminosity at which AGN show a `changing-state' transition, where the broad optical lines disappear. Stellar mass black holes show a state transition at a similarly low Eddington fraction, which is broadly interpreted as the inner disc being replaced by an optically thin flow. We find that the UV lightcurve can instead be matched by reprocessing of the X-ray flux on size scales of the broad line region (BLR; $1.5-20$ light-days) and rule out there being optically thick material inwards of this, as expected if the thin disc is replaced by the flow below the inner radius of the BLR. These results emphasise the need for even longer-timescale, multiwavelength monitoring campaigns on variable AGN.