Delayed Radio Flares in Tidal Disruption Events from Star-Disk Collision Outflows

Abstract

A growing fraction of tidal disruption events (TDEs) exhibit radio emission that rises only years after the optical or infrared flare, indicating delayed outflow activity. In some events the outflow is inferred to be slow ($\sim 0.02 \, c$) and massive ($\gtrsim 0.01-0.1 M_{\odot}$), challenging models such as delayed jets and disk state transitions. We propose a new mechanism for such delayed outflows: repeated collisions between a TDE accretion disk and a pre-existing stellar extreme-mass-ratio-inspiral (EMRI) orbiting the black hole. In this scenario, the delay reflects the viscous time required for the initially compact TDE disk to expand and intercept the EMRI orbit, rather than delayed jet launching or off-axis viewing effects. Once star-disk collisions commence, repeated impacts eject outflows with velocities comparable to the orbital speed, $v_{\rm w} \sim 0.02-0.1c$. We develop a time-dependent model for the coupled evolution of the spreading disk and EMRI-induced mass-loss, identifying regimes where the outflow is dominated by disk material or ablated stellar debris. Depending on disk viscosity, orbital period, and collision efficiency, masses $\sim (10^{-3}-1) \, \rm M_\odot$ can be launched with energies up to $10^{51} \rm \, erg$, years after the TDE. These outflows produce radio emission through interaction with circumnuclear material or earlier TDE ejecta, consistent with observed late-time radio re-brightening. This model predicts a connection between delayed radio flares and EMRI-hosting systems, potentially including those exhibiting quasi-periodic eruptions (QPEs) powered by star-disk collisions, though the conditions for bright radio flares may not always match those necessary for detectable QPEs.