Radiatively-Cooled Mass Transfer: Disk Properties and L2 outflows across Mass Transfer Rates

TL;DR

This work investigates radiatively cooled mass transfer in a binary consisting of a $10\,M_\odot$ donor and a $5\,M_\odot$ accretor by performing 3D hydrodynamic simulations with cooling across MT rates $\dot{M}_{\rm donor} \sim 10^{-5}-10^{-1}\,M_\odot\,\mathrm{yr}^{-1}$. By varying orbital separation, the study reveals a transition from conservative MT at low rates to strongly non-conservative MT with equatorial L2 outflows at higher rates, with the lost mass carrying specific angular momentum near $h_{\rm L2}$. Cooling elevates the disk and circumbinary outflow luminosities to $L \sim 10^{38}-10^{39}$ erg s$^{-1}$ and lowers/cools gas to $T_{\rm eff} \sim 10^{2}-10^{4}$ K depending on the MT rate, implying UV/optical/IR observational signatures. These results inform binary orbital evolution, circumbinary material formation, and the progenitor physics of luminous transients and compact-object mergers. The findings provide quantitative benchmarks for population syntheses and guide expectations for detecting and interpreting L2-driven circumbinary outflows in massive binaries.

Abstract

High rates of stable mass transfer (MT) occur for some binary star systems, resulting in luminous transients and circumbinary outflows. We perform hydrodynamical simulations of a $10 \ M_\odot$ donor star and a $5\ M_\odot$ point mass secondary, incorporating approximate effects of radiative cooling. By varying the orbital separation of the system, we probe MT rates between $10^{-5}$ and $10^{-1} M_\odot$/yr. Mass flows from the donor into an accretion disk, with significant equatorially-concentrated outflows through the outer Lagrange point L2 occurring for MT rates $\gtrsim 10^{-3} M_\odot$/yr, while the MT remaining mostly conservative for lower MT rates. In all cases, any outflowing gas approximately carries the specific angular momentum of L2. The gas cooling luminosity $L$ and temperature increases with MT rate, with $L \sim 10^{5} L_\odot$ and $T \sim 10^4 \, {\rm K}$ for simulations featuring the strongest outflows, with contributions from both the accretion disk and circumbinary outflow. The most luminous transients associated with mass outflows will be rare due to the high MT rate requirement, but generate significant optical emission from both the accretor's disk and the circumbinary outflow.

Figures (10)

• Figure 1: The density $\rho$ in the equatorial plane, for simulations of varying MT rate. The orbital separation $a$ and Lagrange points are labeled. The snapshots, representative of the simulations once they have reached a quasi-steady state, are labeled with $t$ in both code units and physical units (days). The full domain extends to $r=5a$, but these plots extend to $r=3a$ to zoom in near the accretion disk. Note the different density scale in each case.
• Figure 2: Top panel: The fraction of material retained in the accretion disk, $\beta$, versus the MT rate $\dot{M}_{\rm donor}$ for our simulations. Bottom panel: the same quantity, plotted versus $\dot{M}_{\rm donor}$ in units of the Eddington MT rate (Eq. \ref{['eddington']}). The vertical dashed line shows where $\dot{M}_{\rm donor} = \dot{M}_{\rm edd}$. The values shown are time-averaged once the simulations reach a quasi-steady state, at $t$ greater than $\sim300$ (code units).
• Figure 3: The predicted fraction of material escaping through L2, $f_{\rm L2}$, for various $a$ and MT rates. This follows the figures of lu_rapid_2023, but adapted for a 10 $M_\odot$ primary and 5 $M_\odot$ primary. Solid lines show contours of constant $f_{\rm L2}$, and the cyan dotted line shows where $\dot{M}_{\rm donor} = \dot{M}_{\rm edd}$. Stars show where our simulations lie in this parameter space, with the internal color corresponding to their $f_{\rm L2}$ value. The highest_mdot simulation is not shown for convenience, but would lie outside the right border of this plot.
• Figure 4: The cooling luminosity calculated within the accretion disk (dashed lines, Eq. \ref{['Ldisk']}, and the entire computational domain (solid lines, Eq. \ref{['Ltot']}). Curves are smoothed using a rolling average, and values are not shown at $t<200$ because the MT rate has not ramped up to its steady-state value.
• Figure 5: The gas temperature $T$ in the equatorial plane for all simulations. The snapshots, representative of the simulations once they have reached a quasi-steady state, are labeled with $t$ in both code units and physical units (days). The $T$ floor is reached in the outer regions of the mid_mdot simulation and almost everywhere in the low_mdot simulation (see text for details).