AT 2024qfm: a luminous fast blue optical transient at a redshift of z = 0.2267 identified by Lasair-ZTF

Abstract: Luminous fast blue optical transients (LFBOTs) emit from x-ray to radio wavelengths, epitomised by the discovery of AT 2018cow in a host galaxy at 65 Mpc. In the following eight years eleven more have been found, at redshifts $0.075 \lesssim z \lesssim0.34$, plus one identified retrospectively from 2016. Here we present the discovery of AT 2024qfm, classified as an LFBOT in a host galaxy at $z = 0.2267 \pm 0.0002$. Its ultraviolet-to-optical luminosity and rapid 13 day fade closely match AT 2018cow. We describe how the transient was identified in the Zwicky Transient Facility alert stream using a custom filter in the Lasair broker that flags flux gradients over time. Another LFBOT candidate was identified with the same methodology (AT 2024kth). The physical origin of LFBOTs remains debated with no firm consensus, and further progress requires more discoveries, host-galaxy characterisation, and multi-wavelength analysis to constrain theory. We discuss this discovery in the context of Rubin Observatory's Legacy Survey of Space and Time (LSST), whose sensitivity will increase the effective LFBOT survey volume tenfold relative to ZTF, out to $z \lesssim 0.6$, and show that our FastFinder filter could recover such events. We highlight the challenge of detecting their fast evolution with sufficiently low latency to trigger multi-wavelength follow-up that can constrain theoretical models.

• The paper presents AT 2024qfm as a spectroscopically confirmed LFBOT at z = 0.2267, identified by Lasair’s Fastfinder using rapid fading and blue-colour alerts, with a peak absolute magnitude of about −21.0.
• Multi-band optical and ultraviolet observations show a rise lasting only a few days, a decline of roughly 0.3 mag per day, persistent g−r < −0.3 colours, and no secondary radioactive peak, closely matching AT 2018cow.
• The discovery of AT 2024qfm and an unconfirmed companion candidate suggests existing LFBOT rates may be underestimated by up to 66%, while LSST could expand the search volume tenfold if rapid classification and follow-up are achieved.

Discovery via the Lasair Fastfinder annotator

AT 2024qfm (ZTF24aaxhxhf) was first detected in the ZTF public alert stream on 24 July 2024, with a non-detection 0.98 days earlier. While the ALeRCE broker independently reported it as a young supernova candidate, the authors flagged it through a custom "Fastfinder" annotator running on the Lasair broker, which identifies transients with rapid photometric gradients. The annotator triggered on an early decline of $\mathrm{d}g/\mathrm{d}t = 0.37 \pm 0.14$ mag d$^{-1}$ with comparable behaviour in $r$-band and a blue colour. This discovery path is significant beyond the single object: Fastfinder also flagged AT 2024kth, another LFBOT candidate at $z_{\rm phot} = 0.16 \pm 0.06$ with $M_g = -20.9 \pm 0.9$ mag and a decline of $-0.26 \pm 0.04$ mag d$^{-1}$, though that event lacks spectroscopic confirmation. Taken together, the two candidates would raise LFBOT rate estimates derived from only three ZTF events by 66%, illustrating how much the census depends on search methodology.

Photometric and spectroscopic characterisation

Multi-instrument follow-up — Pan-STARRS1/2 imaging, LOT and LT photometry, forced photometry from ZTF and ATLAS, and ten epochs of Swift UVOT data spanning roughly 8–13 days after first detection — constrains the rise above half-maximum to a few days and the decline to $\mathrm{d}g/\mathrm{d}t \approx 0.3$ mag d$^{-1}$, essentially unchanged from the discovery estimate. The transient remained bluer than $g-r \approx -0.3$ mag throughout, and no secondary radioactive-powered peak appears in any band, consistent with established LFBOT behaviour.

Gemini/GMOS spectroscopy beginning about five days after $^{-1}$0-band peak, plus a NOT/ALFOSC spectrum 1.7 days later, show a blue, largely featureless continuum with no broad transient-intrinsic features; the only lines are narrow host-galaxy nebular emission. These yield a redshift of $^{-1}$1, consistent with SDSS DR15 ($^{-1}$2) and Legacy Survey DR9 ($^{-1}$3) photometric estimates. At this distance the peak magnitude is $^{-1}$4. The host emission lines extend to the transient's position at a projected radius of 4.0 kpc, where they show a velocity offset of $^{-1}$5 km s$^{-1}$6, compatible with rotation for a baryonic mass near $^{-1}$7.

Host galaxy properties

Bagpipes SED fitting with a non-parametric continuity star-formation history gives:

Property	Value
$^{-1}$8	$^{-1}$9
SFR	$r$0
$r$1	$r$2
$r$3	$r$4 mag
Mass-weighted age	$r$5 Gyr

The Balmer decrement implies $r$6, $r$7 erg s$r$8, and an SFR of 2.1 $r$9, broadly agreeing with the SED fit. The host sits toward the massive end of known LFBOT hosts but is not an outlier; notably, it hosts the oldest stellar population among known LFBOT hosts with relatively low dust content, suggesting a more evolved stellar population than is typical for the class.

Comparison with other LFBOTs

In rest-frame $z_{\rm phot} = 0.16 \pm 0.06$0 absolute magnitudes and colours, AT 2024qfm closely tracks both AT 2018cow and AT 2020xnd, implying similar effective temperatures and emitting-region sizes and velocities. Its spectrum matches a contemporaneous-phase AT 2018cow spectrum almost exactly. The authors are explicit that the UV/optical data alone do not discriminate between circumstellar shock models, central-engine accretion power, or black hole–Wolf-Rayet mergers; X-ray and radio analysis is deferred to companion work. The projected host offset is intermediate between the small (~2 kpc) offsets of CSS161010, AT 2018cow, AT 2018lug, AT 2020mrf and AT 2020xnd and the larger offsets of AT 2022tsd and especially AT 2023fhn, reinforcing that LFBOT offsets span a wider range than the earliest examples suggested.

Limitations and open questions

Several caveats bear directly on the results. No template subtraction was applied to the Swift UVOT data, so host contamination is a concern where transient flux is comparable to host limits. The absolute-magnitude comparison to AT 2018cow and AT 2024wpp uses no colour or $z_{\rm phot} = 0.16 \pm 0.06$1-correction beyond foreground extinction, while other objects receive approximate corrections. The claimed 66% upward revision of LFBOT rates rests on one unconfirmed candidate (AT 2024kth) lacking spectroscopy. Most fundamentally, the paper concedes that optical data cannot distinguish between the leading physical models, and that the class's physical origin remains unsettled.

Outlook for LSST

The paper quantifies LSST's prospects: with WFD single-visit limits of $z_{\rm phot} = 0.16 \pm 0.06$2 and $z_{\rm phot} = 0.16 \pm 0.06$3, measuring a ~0.4 mag d$z_{\rm phot} = 0.16 \pm 0.06$4 decline over the typical ~5-day per-band cadence requires observed peaks of $z_{\rm phot} = 0.16 \pm 0.06$5 and $z_{\rm phot} = 0.16 \pm 0.06$6, implying detection out to $z_{\rm phot} = 0.16 \pm 0.06$7 ($z_{\rm phot} = 0.16 \pm 0.06$8) or $z_{\rm phot} = 0.16 \pm 0.06$9 ($M_g = -20.9 \pm 0.9$0) for $M_g = -20.9 \pm 0.9$1 events. This opens a co-moving volume roughly ten times larger than ZTF's, potentially raising the discovery rate from just over one per year to one per month. However, the practical latency problem is real: AT 2024qfm took five days from ZTF detection to spectroscopic confirmation, due to the need for a second epoch, manual review, and Gemini trigger latency. The authors propose exploiting intra-night third exposures (a 6-hour baseline could reveal a ~0.1 mag fade) as a possible mitigation, though whether cadence choices will permit sufficiently rapid classification for multi-wavelength follow-up remains an open question set by survey strategy.

Conclusion

AT 2024qfm adds a spectroscopically confirmed LFBOT at $M_g = -20.9 \pm 0.9$2 whose properties are nearly indistinguishable from AT 2018cow, discovered through a broker-based fast-transient filter rather than traditional classification pipelines. Its massive, old host extends the diversity of LFBOT environments, and the paired discovery of AT 2024kth suggests current rate estimates may be conservative. The paper's principal contribution is methodological: it demonstrates that alert-broker annotators can identify LFBOTs in real time, while quantifying the latency and cadence constraints that will govern whether Rubin/LSST can convert its tenfold volume advantage into timely multi-wavelength constraints on the LFBOT engine.