The Ep-Liso correlation: A new diagnostic tool for kilonova transients

TL;DR

This work introduces the Ep–Liso correlation as a new diagnostic tool for kilonova transients by analyzing time-resolved spectra of AT2017gfo. By extracting the blackbody component's peak energy $E_{ m p}$ and isotropic luminosity $L_{ m iso}$ across epochs, they identify a two-phase evolution: an initial power-law Ep–Liso track for $t_{ m gw} \lesssim 2.5$ d followed by Ep saturating near $\sim 1$ eV as the photosphere recedes. They validate the diagnostic by comparing to a large POSSIS radiative-transfer model grid, showing good agreement and demonstrating that Ep–Liso can constrain ejecta properties such as mass, velocity, and Ye, while highlighting the need for finer grids and spectroscopic follow-up. The findings suggest broader applicability to kilonovae and hint at a possible link to gamma-ray burst radiative regimes, motivating future multi-event studies to exploit this correlation for post-merger physics.

Abstract

The AT2017gfo kilonova transient remains a unique multi-messenger event thanks to its proximity (z=0.00987) and the possibility to investigate time-resolved spectra, providing evidence of r-process nucleosynthesis. The kilonova signal was extensively studied in the spectral and time domains, giving key insights into the chemical composition and physical properties of the ejecta. Here, we report the discovery of a novel correlation between two fundamental observables: the peak energy of the EF_E spectrum, Ep, and the isotropic-equivalent luminosity, Liso. In particular, we show that up to about 2.5 days after the merger, the AT2017gfo spectrum evolves according to: log10[Ep/eV] = -0.13 (+0.02/-0.02) + 0.62 (+0.02/-0.02) * log10[Liso/(1e41 erg/s)] (68% C.L.) while in subsequent epochs, Ep remains almost constant with Liso, flattening around 1 eV. Exploiting simulations from a state-of-the-art radiative transfer code, we demonstrate that our kilonova model inherently predicts this peculiar correlation, suggesting a new diagnostic tool for comparing observables against simulations. Future kilonova observations will provide additional insight into the physics behind the Ep-Liso correlation.

Figures (7)

• Figure 1: Spectrum of VLT/X-Shooter for AT2017gfo at epoch $t_{\rm gw}\xspace=1.43$ days in $E\,F_E\,$ units along with the best-fit model and the associated residuals (in units of $\sigma$).
• Figure 2: Spectra of AT2017gfo as observed by VLT/X-Shooter in $EF_E$ units (erg cm$^{-2}$ s$^{-1}$). The epochs not observed with VLT/X-Shooter (see Table 1) are not shown to enhance the visualisation of the temporal evolution of $E_{\rm p}\xspace$. The subrelativistic blackbody continuum extracted from the grbjet component of the best-fit model is superimposed on the observational data, and the corresponding $E_{\rm p}\xspace$ value is shown with a triangle for each spectral epoch.
• Figure 3: The $E_{\rm p}\xspace-L_{\rm iso}\xspace$ correlation for the blackbody component in the AT2017gfo spectra. Temporal epochs range from 0.49 to 7.40 days after the BNS merger, and the estimated uncertainties on both quantities are of the order of symbol size (see Table \ref{['tab:bf_epliso']}).
• Figure 4: Comparison between the observed AT2017gfo spectrum at epoch $t_{\rm gw}\xspace =1.43$ days and the synthetic POSSIS spectrum obtained with the fit_epliso diagnostic.
• Figure 5: Comparison between the $E_{\rm p}\xspace-L_{\rm iso}\xspace$ correlation obtained from the blackbody component of the AT2017gfo spectra and the three different POSSIS best-fit models, namely fit_phot, fit_epliso, and fit_comb. Table \ref{['tab:bf_possis']} reports the POSSIS best-fit parameters.