GRB 230307A Formed No Dust or Was Not a Binary Neutron Star Merger

TL;DR

This study reevaluates the infrared emission of GRB 230307A using JWST spectra at about 29 and 61 days after the burst. By replacing a simple blackbody with a physically motivated dust spectral-energy distribution and performing MCMC fits, the authors test carbonaceous, silicate, and iron dust in a two-component (warm and cold) model to explain the IR continuum and the 2.1 μm feature previously attributed to Te and linked to r-process nucleosynthesis. They find cold-dust masses of order a few ×10^{-3} M_⊙ across compositions, with warm components ~10^{-5} M_⊙, and derive corresponding C, Si, Mg, O, and Fe mass fractions in the ejecta; these abundances are inconsistent with r-process–enriched kilonova yields expected from NS mergers. Additionally, the late-time Ni-56 constraint ($<2.6 imes10^{-3}$ M_⊙) and the Te III line challenge show that no KN model can simultaneously provide the required dust and heavy-element production, casting doubt on the KN interpretation and suggesting the continuum may not arise from dust or that GRB 230307A did not originate from a binary compact-object merger.

Abstract

We present a new analysis of the JWST infrared spectra of GRB 230307A (AT 2023vfi), a long gamma-ray burst (GRB) with an infrared excess and spectral lines suggestive of significant heavy $r$-process production. The spectra, taken 29 and 61~days after the GRB trigger, have blackbody-like continua with $T_{\rm eff} \approx 550$ K and an emission line near $2.1$ $μ$m, previously attributed to [Te III]. This line identification has been used as evidence for an $r$-process-powered kilonova (KN), despite no KN model producing a blackbody-like spectrum at late times. Such an infrared continuum could be emitted by newly formed dust, and we model the thermal emission to infer dust properties, including composition and mass. We find that the emission requires at least 3--$6 \times 10^{-3}$~M$_{\odot}$ of carbon or silicate dust, which is inconsistent with $r$-process yields expected from a neutron star merger. Alternatively, the continuum could be from $2\times 10^{-3}$~M$_{\odot}$ of metallic iron dust, which is mildly consistent (at 3$σ$) with KN models, but such dust is unlikely to form in the expanding ejecta. GRB 230307A's low late-time luminosity also constrains the amount of radioactive $^{56}$Ni produced to $<2.6 \times 10^{-3}$~M$_{\odot}$ (3$σ$). No KN model can simultaneously form the necessary dust for the infrared continuum and heavy elements for the [Te III] line. We conclude that the blackbody continuum is not due to dust emission, or GRB 230307A did not originate from a binary compact-object merger.

Figures (2)

• Figure 1: JWST/NIRSpec spectra of AT 2023vfi taken 29 (top) and 61 days (bottom) after the GRB trigger, as reduced by GillSmartt2025. Overplotted in burnt red are the best-fit models for each spectrum. Those spectra are a combination of a power-law component (green dotted curve), two carbonaceous dust components (gold dashed and purple dot-dashed curves), and Gaussian profiles (not shown). The temperatures of the dust components are listed on the plot: 336 and 626 K for the +29-day spectrum and 261 and 444 K for the +61-day spectrum. The earlier spectrum has three Gaussian components, two centered near 2.1 $$m and one centered near 4.2 $$m . The later spectrum has only a single Gaussian component, centered near 2.1 $$m . While fitting the +61-day spectrum, the data in the grey region ($< 1.7$$$m ) were not used.
• Figure 2: MCMC corner plot showing the two-component dust model fits to the blackbody-like emission in AT 2023vfi, using carbonaceous dust (grain size 0.1 $$m). The model fits the 29-day and 61-day data simultaneously, assuming fixed dust masses between the two epochs. This corner plot corresponds to the fit shown in Figure \ref{['fig:mcmcfit']}, with the afterglow and emission features fixed to the values described in GillSmartt2025.