Probing the origin of the kilonova candidate GRB 230307A: analysis of host galaxy and offset

TL;DR

GRB 230307A’s kilonova candidate lies ~40 kpc from a disk-like host at $z=0.0647$, prompting tests of disk-formed BNSs with natal kicks versus a globular-cluster origin. Using JWST photometry, MUSE kinematics, and Bayesian mass modeling with an NFW halo, the study shows the globular-cluster channel is extremely unlikely, while a disk-origin scenario remains possible only under finely-tuned natal kicks and long delay times, as supported by SEVN population-synthesis comparisons. The work provides precise host morphology and mass-assembly constraints, demonstrates the power of combining high-resolution imaging with kinematic data, and highlights the diverse environments in which compact binary mergers can occur. This has implications for interpreting kilonova offsets, constraining BNS kick distributions, and guiding future multi-wavelength follow-ups of such extreme events.

Abstract

We investigate the host galaxy of the long gamma-ray burst GRB 230307A, which is associated with a kilonova candidate likely produced by a binary neutron-star (BNS) merger. The transient occurred at a projected offset of $\sim 40$ kpc from its spiral-galaxy host. We consider two explanations for this large distance: (i) NSs that merge inside a remote globular cluster, or (ii) a BNS that formed in the disk whose orbit was strongly modified by the NS natal kicks. Using JWST data and comparisons with known globular clusters, we show that a globular-cluster origin is extremely unlikely, ruling out case (i). Considering case (ii), using JWST and MUSE data, we derive the host galaxy morphology, stellar mass, estimate the atomic gas (HI+He) contribution, and the host rotation curve. Assuming an NFW halo and applying Bayesian inference, we obtain a mass model for the host galaxy. From this model, we compute the time required for a disk-formed BNS, with a given natal kick, to reach the observed offset while marginalizing over uncertainties and over the initial position in the disk. We compare these results with BNS-merger simulations from the SEVN population-synthesis code combined with PARSEC stellar evolutionary tracks, which provide the coalescence time and kick velocity for each realization. The two approaches have an overlap in the kick-time diagram, but only 0.1\% of the simulated systems fall within the 2$σ$ region of the galaxy mass model. This indicates that a disk origin is possible, but requires fine-tuned conditions for the kilonova to occur at such a large distance from the host galaxy.

Figures (9)

• Figure 1: A color-magnitude diagram of a sample of globular clusters from Blakeslee12 in $I$- and $H$-band compared with limits on a counterpart to GRB 230307A in F070W and F150W. The red line corresponds to our joint detection threshold to point sources in F070W and F150W, with the grey region containing detected and the white region undetected point sources. We rule out nearly all of the globular cluster luminosity function, implying that it is extremely unlikely that the progenitor of GRB 230307A formed in situ in a globular cluster.
• Figure 2: Upper: Surface brightness profile obtained extracting the galaxy light in elliptical bins, following the galaxy axis ratio, of increasing radius, shown as blue dots with error bars. The black dashed line represents the 1D fit of these data, assuming an exponential profile. The red line shows the output of the 2D GALFIT for an exponential profile, which leads to a larger disk scale length ($h$) than the previous case. Bottom: GALFIT fit: left image, middle model, and right residuals.
• Figure 3: Left: Velocity map of the H$$ emission (see Sec. \ref{['sec:kinematics']}). The solid black line represents the path of virtual extractions. A clear rotation pattern can be discerned in this map. Right: The rotation curve extracted from the map on the left with error bars indicating the $1$ interval of uncertainty.
• Figure 3: Standard MFMTK output, in the 356 band. The top row shows the galaxy image, first box, and the masked regions (in white). The light green contour shows the Petrosian region, where the parameters are measured. The recovered parameters are shown below. The second image shows a single Sérsic fit. Both the 1-D and 2-D fit are tabled. The 2-D recovered Sérsic index is $n\simeq 1.3$. In the third box is presented the residual image (image - model), which clearly shows asymmetry in the spiral arms, suggesting it could be a lopsided galaxy. The fourth and fifth panel show the asymmetry map and the smoothness. The other four boxes show the surface brightness profile, the polar map and the second derivative of the galaxy light profile, i.e. the curvature Lucatelli that is consistent with a disk galaxy ($k=0$).
• Figure 4: Left. Marginalized 2D and 1D sections of the posterior distribution for the fitted parameters. In the 2D regions, the contours indicate the 1$$ and 2$$ credible regions, and the black dot marks the mode of each marginalized case. Right. The best-fit (MAP) rotation curve with a NFW dark matter halo ($_) = 1.10, \; \Delta i = 0.01, \; \log c = 0.75, \; \log (r_{200}/\hbox{kpc}) = 2.64$).