Tidal disruption event rates across cosmic time: forecasts for LSST, Roman, and JWST and their constraints on the supermassive black hole mass function

TL;DR

This work constructs a semi-empirical, redshift-dependent model for TDE rates by tying the local TDE rate to the evolving SMBH mass function and galaxy-scale properties, including dust obscuration, nuclear density, mergers, and IMF variations. The model shows a rise in the volumetric TDE rate toward cosmic noon ($z\sim2$) followed by a decline at higher redshift, with the turnover strongly dependent on the BHMF; galaxy-density effects can dominate at intermediate redshifts, while BHMF evolution governs the high-redshift behavior. Foreseeing upcoming surveys, the authors forecast TDE yields for LSST, Roman, and JWST COSMOS-Web, highlighting how LSST constrains the rate normalization, Roman provides high-purity $z>1$ samples, and JWST probes the high-redshift tail and SMBH seeding. They also outline a method to use flux-limited LSST TDE samples to directly constrain the redshift evolution of the BHMF, enabling population-level tests of SMBH growth that complement AGN-based approaches.

Abstract

Measuring the mass distribution of supermassive black holes (SMBHs) over cosmic time remains particularly challenging for the low mass ($M_{\bullet}\lesssim10^8~M_\odot$) population at $z>1$. This population is also the most sensitive to SMBH seeding and early growth models. In this work we construct a semi-empirical model for the redshift evolution of the TDE rate under multiple SMBH mass function prescriptions, and show that the observed redshift-dependent rate of TDEs is very sensitive to the SMBH mass function and its evolution with redshift. We further incorporate galaxy-scale processes that evolve with redshift -- namely, increasing galaxy nuclear stellar densities, enhanced galaxy-galaxy merger rates, dust obscuration, and a possible top-heavy IMF at early cosmic times -- and quantify their combined impact on the TDE rate. We find that including these effects generally results in a volumetric TDE rate that increases with redshift until a maximum near cosmic noon, before declining at higher redshift where SMBHs that can disrupt stars become increasingly scarce. We forecast TDE rates in the Rubin LSST and the Roman High Latitude Time Domain Survey, alongside expectations for serendipitous TDE rates in the JWST COSMOS-Web survey. Finally, we provide a methodology for using a flux-limited survey of TDEs in LSST to directly constrain the redshift evolution of the SMBH mass function.

Figures (11)

• Figure 1: The two SMBH mass functions used in this work at $z\sim0$ through $z=6$. The solid line is the semi-empirical model from Shankar2009 and the dashed line is the ILLUSTRIS simulation Genel2014.
• Figure 2: Evolution of the assumed TDE obscuration fraction as a function of redshift is within the blue region. The redshift evolution of the obscuration fraction for typical AGN, as modeled in Gilli2022 is compared in red. The TDE obscuration fraction is lower at low redshifts because of the lack of a dusty torus, but grows with ISM column density.
• Figure 3: Evolution of galaxy central surface density as a function of redshift. Lines are solid where measured, and dashed where Barro2017 is extrapolated to higher redshifts. It can be seen that the extrapolation of the star-forming sample is a good prediction of what is later observed in the CEERS Ormerod2024Finkelstein2023 data taken with JWST. The quiescent extrapolation can predict the two highest-redshift quiescent galaxies for which this density is measured, deGraaff2025Carnall2023. We use the $1$ kpc extrapolation to predict the evolution of central galaxy density.
• Figure 4: Scalings of all rate modifications as a function of redshift. On the left y-axis, a value of $1$ is equivalent to the local value. All values at $1$ is the local TDE rate. On the right y-axis, we show the volumetric TDE rate.
• Figure 5: Volumetric rate of TDEs as a function of redshift after applying all redshift-dependent rate modifications. The red line uses the Shankar2009 BHMF model and the blue uses the ILLUSTRIS simulation. The shaded regions are the $1\sigma$ confidence intervals using Monte Carlo sampling of all uncertain parameters. Previous calculations also using the Shankar2009 BHMF were done in Kochanek2016 and are plotted in the grey dashed line. They were calibrated to an older TDE rate measurement from vanVelzen2014 so we re-calibrate it to the Yao2023 local rate, and compare it to our BHMF-only rate (the dashed red line).