EP260321a: X-ray SBO of SN 2026gzf

• The paper identifies EP260321a as a fast, soft X-ray transient representing the shock breakout of broad-lined Type Ic supernova SN 2026gzf.
• It employs coordinated X-ray and optical follow-up to precisely constrain explosion timing, ejecta geometry, and circumstellar material profiles.
• The analysis reveals model-dependent breakout scales and suggests eruptive mass loss in the progenitor, bridging diverse physical processes in stellar death.

Searching arXiv for papers on EP260321a and SN 2026gzf to ground the article in the current literature. EP260321a is a fast, soft X-ray transient detected by the Einstein Probe on 2026-03-21 and identified with the onset of the broad-lined Type Ic supernova SN 2026gzf at redshift $z \approx 0.0343$–0.0345. In the current literature, it is primarily interpreted as a supernova shock breakout (SBO) from a stripped-envelope progenitor, most likely a Wolf–Rayet-like system, with the X-ray flash, the first-day blue optical excess, and the subsequent radioactive supernova all tracing different stages of the same terminal explosion. The event is notable because it combines prompt soft X-ray detection, rapid optical follow-up, archival evidence for pre-explosion variability, and later radio and polarimetric constraints, thereby connecting late-stage mass loss, circumstellar interaction, explosion geometry, and SN Ic-BL phenomenology in a single object (Chen et al., 8 Jun 2026, Yuan et al., 8 Jun 2026, Rastinejad et al., 8 Jun 2026).

1. Discovery, localization, and identification

EP260321a was discovered by the Einstein Probe Wide-field X-ray Telescope (WXT), with follow-up by the Follow-up X-ray Telescope (FXT). Published analyses use slightly different temporal reference points. One report gives the WXT trigger at 2026-03-21 12:23:07 UTC ($T_0$), while another gives the real-time detection at 12:30:18 UTC and, from retrospective analysis of the prior survey exposure, defines the onset as $T_0=$ 2026-03-21 12:16:08 UTC (MJD 61120.511). These differences reflect different operational or analysis definitions rather than disagreement about the event itself (Chen et al., 8 Jun 2026, Yuan et al., 8 Jun 2026).

Rapid optical follow-up was decisive. Lulin Observatory began imaging at $T_0+1.25$ h in $r$ and $i$, finding a blue variable source at

$$\alpha = 09^\mathrm{h}59^\mathrm{m}42^\mathrm{s}.88,\qquad \delta = +00^\circ25'06''.3 \ (\mathrm{J2000}),$$

coincident with the WXT/FXT localization and projected near a galaxy at $z=0.0345$. The source brightened by $\sim 0.5$–0.7 mag in $r$ during the first $T_0$0 h, and spectroscopy obtained within days showed broad SN-like features at the galaxy redshift, establishing the counterpart as the broad-lined Type Ic supernova SN 2026gzf rather than a Galactic variable (Chen et al., 8 Jun 2026).

The positional association is strong. Accurate astrometry using DECam, Subaru/HSC, SDSS, Pan-STARRS, and MUSE-based analyses places the SN, the archival blue compact source, and the pre-explosion variable source at the same location to within $T_0$1 in one study, with a random foreground interloper probability $T_0$2–$T_0$3. The literature therefore treats EP260321a and SN 2026gzf as the same physical event: the X-ray transient is the earliest detected high-energy signature of the supernova explosion (Chen et al., 8 Jun 2026, Martin-Carrillo et al., 8 Jun 2026).

2. X-ray phenomenology and the shock-breakout interpretation

The X-ray emission is uniformly described as unusually soft and thermal. Depending on the extraction and fitting procedure, the WXT and FXT spectra are well fit by absorbed blackbodies with $T_0$4 eV and $T_0$5 eV, or $T_0$6 eV and $T_0$7 eV, with intrinsic absorption $T_0$8. No significant non-thermal power-law tail is required, and several papers explicitly emphasize the absence of a GRB-like prompt component or an X-ray afterglow-like tail (Chen et al., 8 Jun 2026, Wen et al., 17 Jun 2026, Yuan et al., 8 Jun 2026).

The temporal characterization likewise depends on definition. The initial WXT exposure of $T_0$9 s is a lower limit because it was interrupted by the autonomous slew to FXT, but a more complete reconstruction gives a peak at $T_0=$0 s after onset and a $T_0=$1 duration of $T_0=$2 s in 0.3–10 keV. The reconstructed light curve rises nearly linearly to $T_0=$3 s and then decays smoothly and exponentially, without significant substructure (Chen et al., 8 Jun 2026, Yuan et al., 8 Jun 2026).

Reported energetics span a modest range because the papers quote different bands and estimators. One set of analyses gives an average or peak luminosity of order $T_0=$4; another quotes a peak isotropic-equivalent luminosity

$T_0=$5

and a total radiated energy

$T_0=$6

All studies agree, however, that EP260321a is among the softest and intrinsically dimmest extragalactic fast X-ray transients detected by Einstein Probe (Rastinejad et al., 8 Jun 2026, Yuan et al., 8 Jun 2026, Wen et al., 17 Jun 2026).

The core physical interpretation is SBO. The relevant condition is the usual radiation-mediated criterion

$T_0=$7

with $T_0=$8 sub-relativistic. The observed thermal spectrum with $T_0=$9–0.16 keV and the absence of a non-thermal tail are taken to imply $T_0+1.25$0, or in one detailed fit $T_0+1.25$1 (Yuan et al., 8 Jun 2026).

What remains unsettled is the breakout radius and the detailed CSM topology. Several papers argue that the duration is too long for breakout at a bare Wolf–Rayet surface and infer breakout in dense CSM at

$T_0+1.25$2

By contrast, one multi-wavelength analysis derives a much more compact SBO scale,

$T_0+1.25$3

with $T_0+1.25$4, and places the optical CSM at larger radii. This leaves an active interpretive split between a single compact dense shell at $T_0+1.25$5 and a stratified configuration involving an inner SBO layer plus a separate outer interaction zone (Chen et al., 8 Jun 2026, Yuan et al., 8 Jun 2026, Rastinejad et al., 8 Jun 2026).

3. SN 2026gzf as a broad-lined Type Ic supernova

Spectroscopy from VLT/X-shooter, VLT/FORS2, NOT/ALFOSC, WiFeS, GMOS, SOAR, SALT, HET, and DESI consistently classifies SN 2026gzf as a Type Ic broad-line supernova. The spectra closely resemble SN 2006aj in several studies, with broad Fe-, Si-, O-, and Ca-dominated features and no discernible H or He lines in the classification sense, although one TARDIS analysis finds modest residual helium at high velocity and detects He I $T_0+1.25$6 at late times (Chen et al., 8 Jun 2026, Rastinejad et al., 8 Jun 2026, Wen et al., 17 Jun 2026).

The photometric evolution combines a very early blue component with a more standard Ic-BL peak. The first day is characterized by a luminous blue excess, with colours as blue as $T_0+1.25$7 mag and an early rise in $T_0+1.25$8 of $T_0+1.25$9 mag over 4.7 h in the Lulin data. At later times the SN reaches a normal Ic-BL peak, reported as $r$0 to $r$1 mag or $r$2 mag, with the $r$3-band maximum at $r$4–15 d after the X-ray trigger and decline rates such as $r$5 mag (Chen et al., 8 Jun 2026, Rastinejad et al., 8 Jun 2026, O'Connor et al., 8 Jun 2026).

The ejecta velocities are high even within the Ic-BL class. Measurements based on Fe II $r$6, Si II $r$7, and Ca II give values of $r$8–30,000 km s$r$9 in broad terms, with early Fe II reaching $i$0–37,700 km s$i$1 and Si II around 29,000 km s$i$2. One TARDIS model for day 16.5 gives $i$3 and $i$4 (Rastinejad et al., 8 Jun 2026, O'Connor et al., 8 Jun 2026, Wen et al., 17 Jun 2026).

A central issue is the origin of the first-day optical/UV emission. Hydrodynamic STELLA calculations and semi-analytic MOSFiT fits show that the earliest data are more than a magnitude brighter than standard $i$5-powered models and cannot be reproduced by shock cooling of a compact envelope alone. Several groups therefore favor an additional power source, most commonly ejecta–CSM interaction. A different analysis argues that optical data at $i$6 s imply an emitting region expanding at $i$7 and interprets that component as a subrelativistic cocoon from a choked jet. These are not identical models, but both reject a purely radioactive origin for the earliest blue excess (Chen et al., 8 Jun 2026, Rastinejad et al., 8 Jun 2026, Yuan et al., 8 Jun 2026).

The inferred explosion parameters depend strongly on model choice. Published estimates range from $i$8 and $i$9 in a MOSFiT fit excluding the earliest interaction-dominated data, through $$\alpha = 09^\mathrm{h}59^\mathrm{m}42^\mathrm{s}.88,\qquad \delta = +00^\circ25'06''.3 \ (\mathrm{J2000}),$$0 and $$\alpha = 09^\mathrm{h}59^\mathrm{m}42^\mathrm{s}.88,\qquad \delta = +00^\circ25'06''.3 \ (\mathrm{J2000}),$$1 in a pure radioactive TransFit model with extreme mixing, to $$\alpha = 09^\mathrm{h}59^\mathrm{m}42^\mathrm{s}.88,\qquad \delta = +00^\circ25'06''.3 \ (\mathrm{J2000}),$$2 and $$\alpha = 09^\mathrm{h}59^\mathrm{m}42^\mathrm{s}.88,\qquad \delta = +00^\circ25'06''.3 \ (\mathrm{J2000}),$$3 in a CSM+SN model tied to a dense shell at $$\alpha = 09^\mathrm{h}59^\mathrm{m}42^\mathrm{s}.88,\qquad \delta = +00^\circ25'06''.3 \ (\mathrm{J2000}),$$4 AU. This spread is itself informative: the underlying SN is clearly Ic-BL, but the partition of luminosity between radioactive diffusion, interaction, and possibly engine-related components remains model-dependent (Chen et al., 8 Jun 2026, Yuan et al., 8 Jun 2026, Rastinejad et al., 8 Jun 2026, Martin-Carrillo et al., 8 Jun 2026).

4. Circumstellar material and pre-explosion mass loss

Circumstellar structure is central to every current interpretation of EP260321a. In one hydrodynamic framework, the progenitor is a $$\alpha = 09^\mathrm{h}59^\mathrm{m}42^\mathrm{s}.88,\qquad \delta = +00^\circ25'06''.3 \ (\mathrm{J2000}),$$5 C+O star with $$\alpha = 09^\mathrm{h}59^\mathrm{m}42^\mathrm{s}.88,\qquad \delta = +00^\circ25'06''.3 \ (\mathrm{J2000}),$$6 cm surrounded by dense wind-like CSM

$$\alpha = 09^\mathrm{h}59^\mathrm{m}42^\mathrm{s}.88,\qquad \delta = +00^\circ25'06''.3 \ (\mathrm{J2000}),$$7

extending to

$$\alpha = 09^\mathrm{h}59^\mathrm{m}42^\mathrm{s}.88,\qquad \delta = +00^\circ25'06''.3 \ (\mathrm{J2000}),$$8

for a total

$$\alpha = 09^\mathrm{h}59^\mathrm{m}42^\mathrm{s}.88,\qquad \delta = +00^\circ25'06''.3 \ (\mathrm{J2000}),$$9

In that picture, the X-ray flash is the SBO at the CSM outer edge and the first-day blue excess is the optical continuation of the same ejecta–CSM interaction (Chen et al., 8 Jun 2026).

Other studies infer different CSM masses and radii. A thermal SBO analysis models the breakout in a thin shell with $z=0.0345$0, $z=0.0345$1, density at breakout $z=0.0345$2, and a participating mass

$z=0.0345$3

implying ejection within about a month before core collapse. By contrast, the failed-jet study adopts a denser shell with

$z=0.0345$4

inner radius fixed at

$z=0.0345$5

and $z=0.0345$6. The Redback hybrid model instead finds an interaction region beginning at

$z=0.0345$7

with total

$z=0.0345$8

but only

$z=0.0345$9

in the optically thick part that dominates diffusion. Taken together, the literature does not yet converge on a unique CSM profile, but it does converge on the presence of dense, proximate material that is atypical of a steady Wolf–Rayet wind (Yuan et al., 8 Jun 2026, Martin-Carrillo et al., 8 Jun 2026, Rastinejad et al., 8 Jun 2026).

The implied mass-loss rates are correspondingly extreme. For $\sim 0.5$0 within $\sim 0.5$1 cm and $\sim 0.5$2, one study derives

$\sim 0.5$3

over the final $\sim 0.5$4 d. The failed-jet analysis, using Ho’s scaling, gives

$\sim 0.5$5

for its denser shell. Both values are orders of magnitude above steady line-driven WR winds and are therefore interpreted as eruptive, dynamical mass loss rather than stationary outflow (Chen et al., 8 Jun 2026, Martin-Carrillo et al., 8 Jun 2026).

The strongest evidence for a longer pre-SN history comes from archival imaging. Pan-STARRS $\sim 0.5$6-band data spanning $\sim 0.5$7 yr show a persistent blue compact source at the SN position, significant long-term variability with no periodicity, and brightening from $\sim 0.5$8 mag ($\sim 0.5$9) over $r$0 to $r$1 yr to $r$2 mag ($r$3) in the final $r$4 yr, corresponding to a luminosity increase by a factor $r$5. Major flares are reported at MJDs 57779.5, 60021.3, 60315.5, 60696.5, 60761.3, and 61053.5; for $r$6, these would map to shells at $r$7 cm (Chen et al., 8 Jun 2026).

Those timescales are then linked to late nuclear burning. One interpretation associates the decade-long variability and final 3 yr brightening with oxygen-burning-driven instabilities and suggests an additional silicon-burning episode in the final weeks, unobserved directly, as the source of the innermost shell responsible for EP260321a. A plausible implication is that SN 2026gzf records both a years-long build-up of outer CSM and a last-weeks inner ejection episode, although the exact radial decomposition remains model dependent (Chen et al., 8 Jun 2026, Yuan et al., 8 Jun 2026).

5. Geometry, jets, and asymmetry

Polarimetry provides the clearest direct constraint on ejecta geometry. Imaging polarimetry at +4.6 d and spectropolarimetry at +16.5 d relative to the X-ray onset show negligible ISP and very low continuum polarization. At day 16.5, the continuum values are

$r$8

corresponding to $r$9 and an inferred photospheric axial ratio of only $T_0$00 for Thomson-scattering atmospheres with $T_0$01. The outer ejecta are therefore close to spherical at early times (Wen et al., 17 Jun 2026).

The line polarization is more structured. Ca II NIR3 shows a peak polarization above $T_0$02, with two principal velocity components: a dominant component spanning $T_0$03–40,000 km s$T_0$04 and a secondary detached component extending above $T_0$05 km s$T_0$06. On the $T_0$07–$T_0$08 plane, the main Ca II feature follows a dominant axis consistent with axisymmetry, while the secondary component departs laterally from that axis, implying non-axisymmetric substructure or clumping in the outermost ejecta. A 3D Monte-Carlo calculation using cone-like regions of enhanced Ca II opacity finds that a viewing angle of $T_0$09 from the primary symmetry axis can plausibly reproduce the Ca II flux and polarization profiles (Wen et al., 17 Jun 2026).

This geometrical picture matters for the engine question. Despite the globally spherical continuum, several independent data sets disfavor a successful classical jet. No GRB was detected. The EP X-rays are thermal rather than afterglow-like. Radio monitoring over days to weeks produces either upper limits or, in one campaign, only a faint source interpreted as star formation rather than SN synchrotron emission. One multi-wavelength study rules out an on-axis jet with isotropic-equivalent kinetic energy $T_0$10 erg for densities $T_0$11; another finds that if a successful relativistic jet existed it must satisfy $T_0$12 and $T_0$13 erg for $T_0$14; a third concludes that an on-axis jet is excluded and that even very faint EP250304a-like jets require $T_0$15, while more typical GRB-like jets require $T_0$16 (Rastinejad et al., 8 Jun 2026, O'Connor et al., 8 Jun 2026, Martin-Carrillo et al., 8 Jun 2026).

The interpretation of the putative engine differs across papers. One study explicitly favors a jet choked in a dense circumstellar shell, arguing that the combination of $T_0$17 erg, initial velocities of $T_0$18 km s$T_0$19, lack of a cocoon-like early optical bump, and radio non-detections is more naturally explained by a failed jet than by a successful breakout or a purely neutrino-driven explosion. Another study instead interprets the earliest optical emission as a subrelativistic cocoon from a choked jet, while still rejecting a GRB origin for the X-rays. By contrast, the polarimetry paper is more conservative, concluding only that engine-driven asymmetry and extended-CSM SBO can coexist without producing a classical GRB (Martin-Carrillo et al., 8 Jun 2026, Yuan et al., 8 Jun 2026, Wen et al., 17 Jun 2026).

6. Host environment, rates, and broader significance

All published studies place SN 2026gzf in a low-mass, star-forming host environment, but the detailed metallicity values depend strongly on aperture, calibration, and whether one is measuring the global host, the blue knot, or the immediate SN site. Reported global stellar masses range from $T_0$20 to $T_0$21, while local compact-source estimates include $T_0$22 and even $T_0$23 for the knot alone. The corresponding SFR estimates likewise span $T_0$24, $T_0$25, $T_0$26, and $T_0$27, but the qualitative picture is consistent: EP260321a/SN 2026gzf occurred in an actively star-forming dwarf or blue knot environment typical of Ic-BL and long-GRB host samples (Chen et al., 8 Jun 2026, Rastinejad et al., 8 Jun 2026, Martin-Carrillo et al., 8 Jun 2026).

Metallicity is more contentious but central. One set of local measurements gives $T_0$28 for the blue compact source and $T_0$29 globally. Another reports $T_0$30 at the SN site and describes it as lower than any previously known Ic-BL SN site. Yet another derives $T_0$31–7.95 at the explosion site, while a global-host analysis using $T_0$32 yields $T_0$33. The common thread is that the environment is not metal-rich; the divergence reflects methodology and spatial scale rather than a simple contradiction (Chen et al., 8 Jun 2026, Martin-Carrillo et al., 8 Jun 2026, O'Connor et al., 8 Jun 2026, Rastinejad et al., 8 Jun 2026).

The event also has rate implications. Using the ZTF Bright Transient Survey Ic-BL rate and assuming all Type Ic-BL SNe produce EP260321a-like SBOs, one study infers an Einstein Probe detection rate of $T_0$34–$T_0$35, which it states is inconsistent at the 90% confidence level with the current EP detection rate of such events. That paper therefore suggests that most Ic-BL SNe probably produce less luminous X-ray SBO signals than EP260321a. A different estimate, based on a single detection within the EP sensitivity volume, gives a volumetric rate

$T_0$36

and notes compatibility, within large uncertainties, with the ASAS-SN Ic-BL rate. The present sample size is too small to resolve that tension decisively (Rastinejad et al., 8 Jun 2026, Martin-Carrillo et al., 8 Jun 2026).

The broader significance of EP260321a lies in temporal anchoring. Because SBO occurs within minutes to $T_0$37 s of core collapse rather than days before optical peak, the X-ray onset constrains the explosion time to within less than an hour. That precision is directly relevant to targeted neutrino and gravitational-wave searches, and it also enables unusually stringent separation of SBO, cocoon, CSM-interaction, and radioactive components in the electromagnetic data (Yuan et al., 8 Jun 2026).

In aggregate, EP260321a occupies a distinctive position among stripped-envelope explosions. It has been described as the first Type Ic-BL SN with a definitive X-ray SBO, one of the softest and dimmest extragalactic EP fast X-ray transients, and the faintest shock breakout yet associated with a broad-lined supernova. The current literature agrees on the identification of a thermal SBO linked to SN 2026gzf and on the existence of unusually important nearby CSM; it remains divided, however, on whether the innermost breakout occurred in a $T_0$38 shell or a more compact layer, and on how much of the earliest optical luminosity should be assigned to interaction versus a choked-jet cocoon. That combination of consensus and unresolved structure is precisely what makes EP260321a a benchmark event for the physics of stripped massive-star death (Rastinejad et al., 8 Jun 2026, Wen et al., 17 Jun 2026, O'Connor et al., 8 Jun 2026).