AT2025ulz and S250818k: zooming in with the Hubble Space Telescope

TL;DR

This work analyzes AT2025ulz, a transient in the S250818k GW localization, to test whether it is a kilonova or a supernova. By combining GTC optical imaging/spectroscopy with HST ultraviolet/optical/infrared observations and employing a Gaussian-process–emulated two-component kilonova model, the authors assess the transient’s ejecta properties and host environment. They find that while the early optical evolution superficially resembles a kilonova, the very early detections, blue-to-red color at 4.8 days ($F336W-F160W \approx 1.4$ mag), large inferred ejecta masses ($m_d \approx 0.05\,M_\odot$, $m_w \approx 0.08\,M_\odot$, $m_{ej} \approx 0.13\,M_\odot$), and a low GW chirp mass ($\mathcal{M}_{c} \sim 0.1-0.87\,M_\odot$) argue against a binary neutron star kilonova origin. Spectroscopy reveals Balmer-line P-Cygni features with velocities $\sim 10^4$–$1.6\times10^4$ km s$^{-1}$, consistent with a Type II SN, and HST imaging shows aspiral, star-forming host, further supporting a stellar progenitor. The results underscore the difficulty of identifying kilonovae in GW follow-ups and highlight the critical role of host morphology and late-time spectroscopy in robust classification.

Abstract

AT2025ulz is an optical/near-infrared transient discovered during follow-up of the candidate gravitational wave (GW) event S250818k. Its young age ($\lesssim$1 d), rapid decline and strong color evolution over the first 48 hr classify it as a potential kilonova candidate. In this work, we present the results of our observing campaign, carried out with the Gran Telescopio Canarias (GTC) and the Hubble Space Telescope (HST). Although the early time evolution of AT2025ulz resembles some aspects of a kilonova, its rapid onset ($\sim$3 hr after the GW trigger) and luminosity (a factor of $\sim5$ brighter than AT2017gfo in $g$-band) are difficult to reproduce. Only a small subset of our kilonova models matches its multi-color light curve, and the inferred ejecta mass is uncomfortably large given the low chirp mass ($\lesssim\!0.87\!$ M$_{\odot}$) of the GW candidate. HST observations place the transient within a nearby ($z=0.08489$) spiral galaxy with on-going star-formation and measure a color ($F336W-F160W\!\approx\!1.4$ mag) that is too blue to match with a kilonova. Our data support the classification of AT2025ulz as a supernova, initially undergoing a shock-cooling phase and later entering its photospheric phase, and spectroscopically identified via its broad absorption features.

Figures (6)

• Figure 1: Source-frame chirp mass ($\mathcal{M}_c$) v.s. luminosity distance ($D_{\rm L}$) for events detected during the O4a observing run. Gray circles represent candidates of the GWTC-4.0 catalogue LVK2025GWTC4Update with $\mathrm{FAR} \leq 1~\mathrm{yr^{-1}}$ and $P_{\mathrm{astro}} > 0.5$, where the circle size scales with the 90 % credible sky-localization area ($\Delta\Omega$).
• Figure 2: Probability map of the candidate GW event S250818k. A second lobe is located in the southern hemisphere and is not shown. The position of AT2025ulz is marked by the star. A zoom-in view of the field from the Hubble Space Telescope observations is displayed in the right panel, oriented with north up and east to the left.
• Figure 3: GTC spectra (a) and multi-color light curve (b) of AT2025ulz, corrected for Galactic extinction. a. The observed spectra are plotted as semi-transparent lines, with smoothed solid lines for plotting purposes and dashed lines for uncertainties. Both spectra contain significant, though different, levels of host-galaxy contribution. The second-epoch spectrum, which is less affected by the host, has been rescaled by a factor of two for comparison. Emission and absorption lines are marked by the red and black dashed lines, respectively. The P-Cygni profiles of H$\alpha$ and H$\beta$ are highlighted in a deeper color shade. Wavelengths affected by telluric absorption and sky emission are marked by gray and pink areas, respectively Lord1992. b. Circles and downward triangles denote detections and upper limits, respectively. Filled symbols indicate our observations (Table \ref{['tab:gtc']}), while open symbols correspond to supplementary data (Table \ref{['tab:gcn']} and Gillanders2025). The data are scaled for plotting purposes. The smoothed light curves of AT2017gfo, redshifted to $z = 0.08489$ and further shifted by $\Delta$m $= -1.8$ ($g$), $-1.5$ ($r$), $-1.3$ ($i$), $-1$ ($z$), $-0.2$ ($H$), are shown as dotted lines for comparison with AT2025ulz. The similar trend of evolution is highlighted by the solid lines.
• Figure 4: Posteriors of parameters (a) and related multi-wavelength light curve with viewing angle $\theta=0$ deg (b) of kilonova model. Data used for the kilonova model fit are plotted as filled circles. Early ($\approx$3-4 hr) detections of AT2025ulz (white circles) cannot be reproduced by our kilonova models.
• Figure 5: Temporal evolution of the score of AT2025ulz being a kilonova with respect to AT2017gfo at the same time.