How Flat is a Plateau? Evolution of Late-Time TDE Disks

Late-time light curve plateaus in tidal disruption events (TDEs) are often approximated as flat and time-independent. This simplification is motivated by theoretical modeling of spreading late time TDE disks, which often predicts slow light curve evolution. However, if time evolution can be detected, late-time light curves will yield more information than has been previously accessible. In this work, we re-examine late-time TDE data to test how well the flat plateau assumption holds. We use Markov Chain Monte Carlo to estimate the maximum likelihood for a family of theory-agnostic models and apply the Akaike information criterion to find that that roughly one third of our sample favors evolving plateaus, one third favors truly flat plateaus, and one third shows no statistically significant evidence for any plateau. Next, we refit the TDEs that exhibit statistically significant plateaus using a magnetically elevated $\alpha$-disk model, motivated by the lack of clear thermal instability in late time TDE light curves. From these model-dependent fits, we obtain estimates for the supermassive black hole (SMBH) mass, the mass of the disrupted star, and the $\alpha$ parameter itself. Fitted $\alpha$ values range from $10^{-3}$ to 0.4 (the mean fitted $\alpha=10^{-1.8}$, with scatter of 0.6 dex), broadly consistent with results from magnetohydrodynamic simulations. Finally, we estimate the timescales of disk precession in magnetically elevated TDE models. Theoretically, we find that disk precession times may be orders of magnitude shorter than in unmagnetized Shakura-Sunyaev disks, and grow in time as $T_{\rm prec}\propto t^{35/36}$; empirically, by using fitted $\alpha$ parameters, we estimate that late time disks may experience $\sim$few-10 precession cycles.