Accretion disks in (repeating) partial tidal disruption events: rapid state transitions, UV plateaus and flares from disk-remnant collisions

Abstract

Tidal disruption events which repeat on timescales of months-to-years represent an unambiguous signature of a partial disruption, with the surviving stellar remnant returning to pericentre to be repeatedly stripped by tidal forces. These systems therefore offer the best laboratories to study the differences between partial and full disruptions. One noteworthy observational difference between the two systems is that all known X-ray bright repeating TDEs show rapid transitions between thermal, non-thermal and completely dim states on timescales much shorter than full (non-repeating) TDEs. We argue this can be simply understood as being due to the reduction in fuel supply available to the disk, and that these systems provide evidence that all tidal disruption events undergo a state transition at Eddington ratios $f_{\rm edd} = L_{\rm bol}/L_{\rm edd} \sim 0.01$, similar to X-ray binaries. {As part of this calculation we derive a general expression for the time taken for a TDE disk to fall to a given Eddington fraction, which will be of use to both full and partial TDE science.} Perhaps surprisingly, the late-time optical/UV plateau luminosity observed from these systems is largely unaffected by this reduction in fuel supply, provided the outer disk remains in a thermal state for long enough for this emission to be detected. We then show that collisions between the returning stellar remnant and the disk formed from the last passage will produce potentially observable X-ray flares ($L_{\rm flare} \simeq 10^{42}$ erg/s), but that they are likely to be very difficult to detect as they are generally short-lived ($t_{\rm flare} \simeq 0.1-1$ hr).

Figures (7)

• Figure 1: Upper: the X-ray light curves (defined here as the integrated luminosity between photon energies of 0.3 and 10 keV) for fiducial stellar and black hole parameters (see text), and different values of the impact parameter $\beta$ (displayed on plot). Even weak disruptions $\beta/\beta_c \sim 1/2$, where only $\sim 10\%$ of the mass is stripped off the star can produce observable X-ray flares. Lower: the Eddington luminosity ratio of the disk systems, as a function of impact parameter $\beta/\beta_c$. By a horizontal dashed line we display an Eddington ratio of $1\%$, highlighting the different timescales at which the partial and full disruptions cross this transitional scale.
• Figure 2: The time taken (in days) for the accretion flow to reach an Eddington ratio of $1\%$, from the time of first disk formation, as a function of normalised impact parameter $\beta/\beta_c$. The dashed lines show the analytical scaling argument put forward in this work. For sufficiently small impact parameters $\beta/\beta_c \sim 1/2$ state transitions could occur on timescales of hundreds of days. Deviations from the predicted scaling relationship at very low $\beta/\beta_c$ occur as the disk has not reached the $t \gg t_{\rm visc}$ asymptotic regime when $1\%$ Eddington is reached. The deviation at high $\beta/\beta_c$ results from the relativistic formation radius of the disk, which leads to modifications from the strictly Newtonian analytical scaling arguments used in the text.
• Figure 3: The light curves of the accretion flows shown in Figure \ref{['fig:partial+full']}, as observed at UV frequencies ($\nu = 10^{15}$ Hz). We see that the UV plateau luminosity roughly $\sim 1000$ days post disruption shows only a weak dependence on $\beta/\beta_c$, with $\beta/\beta_c = 1$ disruptions brightest (as can be understood analytically). At significantly later times ($\sim$ decades) partial disruptions show a more pronounced $\beta$ dependence, although the disks are likely to have undergone a state transitions by this point, which may well modify their behavior.
• Figure 4: The multi wavelength light curves of AT2018fyk, data from Pasham24fyk. The evolving UV luminosity is shown by blue points (left vertical axis), the evolving X-ray luminosity by purple diamonds (right vertical axis), while X-ray upper limits are shown by orange inverted triangles. The rough timescales at which different state transitions occur are shown by vertical dashed lines.
• Figure 5: Optical emission decay indices inferred from a large sample of ZTF TDEs (see text for more information). As the distribution is not cleanly bimodal around $-9/4$ and $-5/3$ it is clear that the observed decay index inferred from optical light curves cannot be used to distinguish between different types of stellar disruptions.