Light Travel Time Effects in Kilonova Models

Abstract

The extremely rapid evolution of kilonovae results in spectra that change on an hourly basis. These spectra are key to understanding the processes occurring within the event, but this rapid evolution is an unfamiliar domain compared to other explosive transient events, such as supernovae. In particular, the most obvious P Cygni feature in the spectra of AT2017gfo -- commonly attributed to strontium -- possesses an emission component that emerges after, and ultimately outlives, its associated absorption dip. This delay is theorised to arise from reverberation effects, wherein photons emitted earlier in the kilonova's evolution are scattered before reaching the observer, causing them to be detected at later times. We aim to examine how the finite speed of light -- and therefore the light travel time to an observer -- contributes to the shape and evolution of spectral features in kilonovae. Using a simple model, and tracking the length of the journey photons undertake to an observer, we are able to test the necessity of accounting for this time delay effect when modelling kilonovae. In periods where the photospheric temperature is rapidly evolving, we show spectra synthesised using a time independent approach are visually distinct from those where these time delay effects are accounted for. Therefore, in rapidly evolving events such as kilonovae, time dependence must be taken into account.

Figures (8)

• Figure 1: Spectra generated using only packets released from the photosphere at $t_{\rm ph} = 1.4 \; \rm{days}$ post-merger in simulation time. Packets are binned according to the time, $t_{\rm det}$, when they reach a hypothetical observer, with each line overlaid on each other. The observer is positioned 40 Mpc from the kilonova, matching the distance to AT2017gfo Smartt2017. The bins are 1 hr wide and centred on the days listed. A reference 6400 K blackbody, chosen to match the peak of the plotted lines, is shown in grey as a comparison aid. For clarity, time bins with low overall counts are grouped together. The rest wavelength of the Srii line is shown with a grey, dashed, vertical line.
• Figure 2: Spectra generated for a three hour $t_{\rm det}$ window centred on the 1.4 day observations of AT2017gfo, located at a distance of 40 Mpc. The colours correspond to the time of release of packets from the photosphere, $t_{\rm ph}$, with the coloured areas illustrating how many packets released from the photosphere in that time frame are contributing to the spectrum. Each region is stacked to show contributions of the net flux. The grey region consists of time frames with low total numbers of packets, which have been combined and plotted together for clarity. The dashed, grey line denotes the rest wavelength of the Srii line.
• Figure 3: Spectra generated under conditions chosen to match the 1.4 day observations of AT2017gfo. Top panel shows the total flux emitted in the 3 hour window centred on the 1.4 day observation (teal), which is then subdivided into separate lines showing the scattered (pink) and unscattered (orange) photons. Middle panel shows the ratio of the photon flux that was scattered to the total number of photons in each wavelength bin. The bottom panel shows the average difference between $t_{\rm ph}$ and $t_{\rm det}$ of the scattered photons (pink) in each wavelength bin. This is shown relative to this difference calculated for all photons (teal) with a constant, $\Gamma$, added to bring this baseline to 0 on the y axis. The dashed, grey line denotes the rest wavelength of the Srii line.
• Figure 4: As for \ref{['fig:1Day']}, but instead centred around the 2.4 day observational time frame (upper panel) and the 4.4 day observational time frame (lower panel).
• Figure 5: As for \ref{['fig:1Day3panel']}, with the left column centred around the 2.4 day observational time frame, and the right column centred around 4.4 days.