On Atomic Line Opacities for Modeling Astrophysical Radiative Transfer

In astrophysics, atomic transition line opacity is a primary source of uncertainty in theoretical calculations of radiative transfer. Much of this uncertainty is dominated by the inability to resolve the lines in frequency, leading to the use of approximate frequency-averaged treatments, often employing the `line-expansion formalism’. In this short paper we assess the usage of this formalism in simulations, specifically the prominent Eastman & Pinto 1993 formula (hereafter EP93). As a case study, we reproduce EP93 opacities from the commonly-used STELLA simulations. The latter previously yielded orders of magnitude discrepancy in observed emission relative to similar simulations from our group. The discrepancy is due to differences in line opacity treatment. We show that the widely used EP93 expansion opacity substantially underestimates photon emissivity and reprocessing rates, even when it correctly captures photon mean-free-paths. We also highlight the importance of introducing micro-plasma electron excitation level cutoffs in the equation of state (EOS) for calculating opacity. We propose a new method for calculating emissivity, based on a modification of the simple frequency-bin averaged opacity method, in a way that incorporates the effect of expansion on effective line strength. This formulation should reduce the overestimation of the opacity that may occur with the simple averaging method. To our knowledge, no fully-consistent coarse-frequency solution currently exists for line modeling in these systems. Finally, we describe new features in our updated publicly available high-resolution frequency-dependent opacity table.

💡 Research Summary

This paper revisits the treatment of atomic line (bound‑bound) opacity in astrophysical radiative‑transfer calculations, focusing on the widely used Eastman & Pinto (1993) expansion‑opacity formalism (hereafter EP93). The authors begin by reproducing the line opacities employed in the STELLA code, which are based on the Blinnikov et al. (1998) data set (B98). By comparing these with their own high‑resolution opacity tables (M23, 2023), they identify two principal sources of discrepancy.

First, in the bound‑free (photo‑ionization) regime, the implementation of the Hummer & Mihalas (1988) occupation‑probability cutoff for highly excited electron states dramatically changes the hydrogen partition function. When the cutoff is omitted and the maximum principal quantum number is limited to nmax ≈ 400, the authors can reproduce the B98 bound‑free opacity to within a factor of a few. With the full Hummer‑Mihalas treatment, the bound‑free opacity can differ by orders of magnitude, reflecting the sensitivity of the equation of state (EOS) to the treatment of ionization and level populations.

Second, for bound‑bound lines, the EP93 prescription computes an “expansion opacity” χexp,i = ρ κexp,i that depends on the ratio ν/Δν and the expansion time texp, together with the Sobolev optical depth τl of each line. While this formalism captures the reduction of photon mean free paths as lines are Doppler‑shifted in an expanding flow, it does not provide a consistent emissivity term. The authors show that the frequency‑averaged opacity h κν,i (used for emission/absorption in their own code) can be tens to hundreds of times larger than the EP93 opacity for the same physical conditions. In the limit texp → 0 the two converge, but as texp grows the EP93 opacity quickly saturates because each strong line contributes at most a factor