Describing the Nonuniversal Galaxy Merger Timescales in IllustrisTNG: Effects of Host Halo Mass, Baryons, and Sample Selection

Galaxy merger timescales are crucial for understanding and modeling galaxy formation in our hierarchically structured Universe. However, previous studies have reported widely varying dependencies of merger timescales on initial orbital parameters and mass ratio at the first crossing of $r_{\rm vir}$. Using IllustrisTNG simulations, we find that these dependencies vary with host halo mass, suggesting that discrepancies in prior studies may arise from differences in the systems analyzed. Specifically, in low-mass halos, merger timescales show a stronger dependence on initial orbital parameters, while in high-mass halos, this dependence weakens. To account for these variations, we present a fitting formula that incorporates host mass dependence, achieving a logarithmic scatter smaller than 0.15 dex. Comparing dark matter-only and baryonic simulations, we observe similar merger timescales for circular orbits but notable differences for radial orbits. In halos with $M_{\rm{host}} < 10^{12.5} h^{-1} M_{\odot}$, mergers in dark matter-only runs take longer than in baryonic runs, whereas the trend reverses in more massive halos. We attribute these differences to the competing effects of tidal disruption by central galaxy disks and the resistance of baryonic satellites to tidal stripping. Finally, we extend our model to predict merger timescales from any starting radius within the halo. By fitting the extended model to the entire infall sample, we find that using only the merger sample can underestimate merger timescales, particularly for low mass ratios. Our model provides a valuable tool for improving semi-analytical and empirical models of galaxy formation.