Self-consistent 3D radiative transfer for kilonovae: directional spectra from merger simulations

Abstract

We present three-dimensional radiative transfer calculations for the ejecta from a neutron star merger that include line-by-line opacities for tens of millions of bound-bound transitions, composition from an r-process nuclear network, and time-dependent thermalization of decay products from individual $α$ and $β^-$ decay reactions. In contrast to expansion opacities and other wavelength-binned treatments, a line-by-line treatment enables us include fluorescence effects and associate spectral features with the emitting and absorbing lines of individual elements. We find variations in the synthetic observables with both the polar and azimuthal viewing angles. The spectra exhibit blended features with strong interactions by Ce III, Sr II, Y II, and Zr II that vary with time and viewing direction. We demonstrate the importance of wavelength-calibration of atomic data using a model with calibrated Sr, Y, and Zr data, and find major differences in the resulting spectra, including a better agreement with AT2017gfo. The synthetic spectra for near-polar inclination show a feature at around 8000 A, similar to AT2017gfo. However, they evolve on a more rapid timescale, likely due to the low ejecta mass (0.005 M$_\odot$) as we take into account only the early ejecta. The comparatively featureless spectra for equatorial observers gives a tentative prediction that future observations of edge-on kilonovae will appear substantially different from AT2017gfo. We also show that 1D models obtained by spherically averaging the 3D ejecta lead to dramatically different direction-integrated luminosities and spectra compared to full 3D calculations.

Figures (8)

• Figure 1: Direction-integrated luminosity versus time for the models 3D AD1, 3D AD2, 1D AD1, 1D AD2, the 3D gray opacity model of Collins:2023ha, and inferred bolometric luminosity of AT2017gfo Smartt:2017kw.
• Figure 2: Mollweide projections of direction-dependent quantities for 3D AD2 UVOIR packets arriving at the observer between 1.5 and 1.8d: radiant intensity times 4$\pi$ solid angle, mean temperature at last interaction, and line of sight velocity at last interaction. For these figures, we use 32x32 direction bins, uniformly spaced in azimuthal angle (horizontal) and cosine of the polar angle (vertical) to give the same solid angle in each bin.
• Figure 3: Spectra for polar and equatorial viewing directions for the 3D AD1 and 3D AD2 models at 1.1 days. The height of each wavelength point is colored according to the emitting species of the last interactions of the emerging radiation packets. The area under the horizontal axis shows the distribution of frequencies (colored by absorbing/scattering ion) just prior to the last interactions of the emerging packets. The 11 most-significant ions are separately colored, while the "Other" group combines many smaller contributions from other ions.
• Figure 4: Time series of spectra in the polar direction of the 3D AD2 model compared to reddening and redshift corrected spectra of AT2017gfo Pian:2017keSmartt:2017kw. The area under the spectra have been coloured by the emitting species of the last interactions of the emerging packets. The times of the artis and AT2017gfo spectra intentionally do not match.
• Figure 5: Spherically-averaged spectra at 1.1 days for the 3D AD1 (solid blue), 3D AD2 (solid orange), 1D AD1 (dashed blue), 1D AD2 (dashed orange) models.