Fast X-ray Transients (FXTs)

• FXTs are short-lived X-ray events defined by rapid onset, brief duration, and high flux contrast, spanning both Galactic and extragalactic sources.
• Their discovery relies on diverse instruments and bandpasses, with surveys like Einstein Probe and NuSTAR identifying distinct subpopulations.
• Multiwavelength follow-ups and host galaxy analyses are critical for distinguishing progenitor channels, from GRB afterglows to tidal disruption events.

Searching arXiv for recent and foundational FXT papers to ground the article. Fast X-ray Transients (FXTs) are short-lived X-ray outbursts whose defining observables are rapid onset, brief duration, and large flux contrast relative to any persistent emission. In the modern extragalactic literature, FXTs are usually soft X-ray flashes in the $0.3$–$10$ keV band lasting from $\sim 10^{1}$ to $10^{4}$ s, or from a few hundred seconds to a few hours; in broader survey usage, the same label has also been applied operationally to transients longer than 1 s and shorter than 1 day. The term therefore denotes an observational class rather than a single progenitor channel. Across the literature, FXTs encompass Galactic stellar flares, wind-fed high-mass X-ray binary flares, supergiant fast X-ray transients (SFXTs), gamma-ray bursts and X-ray flashes, supernova shock breakouts, compact-object merger transients, and tidal disruption events (Gillanders et al., 2024, Eappachen et al., 2023, Zand et al., 18 Dec 2025).

1. Terminology, taxonomic scope, and historical usage

The literature uses “FXT” in two partially overlapping senses. In older wide-field survey work, FXTs are a mixed population defined primarily by duration and detectability. The BeppoSAX-WFC catalog, for example, defines FXTs as transient X-ray events with durations longer than 1 s and shorter than 1 day, explicitly including GRBs, XRFs, stellar flares, and HMXB flares while excluding type-I and type-II X-ray bursts from Galactic neutron stars (Zand et al., 18 Dec 2025). In contemporary time-domain astrophysics, by contrast, the term often refers more specifically to extragalactic soft X-ray flashes, first identified in archival imaging data and now routinely discovered in real time by Einstein Probe (Srivastav et al., 2024).

This dual usage has direct taxonomic consequences. In the BeppoSAX-WFC sample, the duration versus spectral-hardness distribution is bimodal, with a division around $\sim 1$ ksec: the short FXTs are identified as GRBs and XRFs, whereas the long FXTs are identified as nearby stellar flares and X-ray binaries (Zand et al., 18 Dec 2025). In the extragalactic literature, however, the key distinction is less between Galactic and non-Galactic fast transients than between multiple extragalactic progenitor channels whose observables overlap in duration but differ in luminosity, spectrum, host environment, and multiwavelength behavior (Eappachen et al., 2023, Hoof et al., 25 Jun 2026).

A related terminological complication is the place of SFXTs. SFXTs are a well-established Galactic subclass of fast X-ray transients, but they are physically high-mass X-ray binaries rather than the cosmological soft X-ray flashes emphasized by recent Einstein Probe work. Their inclusion under the FXT umbrella is historically correct, yet they are best treated as a distinct subfield with its own accretion physics and diagnostic criteria (Romano et al., 2012, Sidoli, 2011).

2. Discovery channels and instrumental regimes

The observational history of FXTs is strongly instrument-dependent. Galactic SFXTs were unveiled mainly by INTEGRAL/IBIS, whose wide field of view and hard X-ray sensitivity were suited to catching brief bright flares from wind-fed binaries, even though persistent or quiescent emission was usually below its sensitivity threshold (Sidoli et al., 2010). Swift then transformed SFXT studies by combining BAT triggers, rapid slewing, and XRT follow-up, enabling both prompt spectroscopy and long-term monitoring across outburst, intermediate, and quiescent states (Romano et al., 2011).

The extragalactic FXT field developed differently. Before Einstein Probe, most known events were recovered from archival Chandra, XMM-Newton, and eROSITA data, so prompt counterpart identification was generally impossible (Gillanders et al., 2024). Einstein Probe altered this regime by continuously surveying the soft X-ray sky in the $0.5$–$4$ keV band. EP-WXT has an instantaneous field of view of about $3600\ \mathrm{deg}^2$, localizes FXTs initially to about a 3 arcmin radius, and EP-FXT refines positions to roughly $10$–$30\arcsec$, which is the critical step that makes rapid optical and radio follow-up feasible (Srivastav et al., 2024, Aryan et al., 29 Apr 2025).

This change in observing mode is already visible at the sample level. In the first year of operation, EP reported 72 high signal-to-noise FXTs, and follow-up programs were able to search for optical counterparts for a large fraction of them in near real time (Aryan et al., 29 Apr 2025). The mission has therefore shifted the field from retrospective source recovery to prospective population building, with prompt localization now central to physical interpretation rather than ancillary to it (Gillanders et al., 2024).

A complementary hard-band perspective is supplied by NuSTAR. A systematic archival search over 204 Ms of $10$0–$10$1 keV exposure yielded five candidate FXTs, four of them spectrally hard, indicating that the hard-X-ray transient sky contains a subset not well represented in soft-band surveys (Brightman et al., 20 Jan 2026). The instrumental lesson is that FXT phenomenology is not invariant across bandpass: survey energy range materially shapes which physical channels enter the discovered sample.

3. Observational phenomenology

In the modern extragalactic sense, FXTs are typically soft X-ray events lasting from a few hundred seconds to several hours, or approximately $10$2–$10$3 s in the $10$4–$10$5 keV band (Eappachen et al., 2023, Gillanders et al., 2024). Their peak isotropic X-ray luminosities are typically $10$6, and their total isotropic energies are roughly $10$7 (Biswas et al., 13 Jun 2025). Within that broad range, different survey-defined subsamples occupy distinct regions of parameter space.

Archival XMM-Newton FXTs with secure or candidate host associations tend to occupy the lower-luminosity part of the distribution. Host redshifts in one XMM sample lie at $10$8, implying peak X-ray luminosities of $10$9–$\sim 10^{1}$0 and projected offsets of $\sim 10^{1}$1–$\sim 10^{1}$2 kpc (Eappachen et al., 2023). In contrast, Einstein Probe FXTs with known redshifts are described as systematically more luminous, with $\sim 10^{1}$3 and redshifts extending from about $\sim 10^{1}$4 to $\sim 10^{1}$5 (Srivastav et al., 2024). A plausible implication is that archival Chandra/XMM samples and EP real-time samples are not sampling the same region of transient phase space.

Light-curve morphology is likewise diverse. Some FXTs exhibit a fast rise, a plateau lasting $\sim 10^{1}$6–$\sim 10^{1}$7 ks, and a steep decay close to $\sim 10^{1}$8, as in the Chandra candidate magnetar-powered events XRT 170901, XRT 030511, and XRT 110919 (Lin et al., 2022). Others show afterglow-like power-law decays, optical rebrightening, or late supernova components, as in EP240414a (Srivastav et al., 2024). Still others are identified only through their prompt X-ray flash, with no contemporaneous optical counterpart found to deep limits (Eappachen et al., 2023, Eappachen et al., 2023).

Spectrally, soft-band FXTs and hard-band FXTs need not resemble one another. The NuSTAR candidates include four events with power-law indices $\sim 10^{1}$9, a marked contrast with the softer extragalactic FXTs usually found below 10 keV (Brightman et al., 20 Jan 2026). This suggests that “FXT” is best regarded as a phenomenological envelope spanning multiple spectral classes rather than a single soft transient family.

4. Supergiant Fast X-ray Transients as a Galactic FXT subclass

SFXTs are high-mass X-ray binaries containing a compact object, usually inferred to be a neutron star, accreting from the wind of a blue OB supergiant companion (Romano et al., 2012). They are among the best-studied Galactic realizations of fast X-ray transient behavior and were one of the major outcomes of INTEGRAL Galactic plane monitoring (Sidoli, 2011).

Their defining phenomenology consists of bright hard X-ray flares lasting a few minutes to a few hours, embedded in broader outburst episodes that can last a few days (Sidoli, 2011). Peak luminosities are typically $10^{4}$0–$10^{4}$1, while quiescent luminosities are around $10^{4}$2 and the more frequent intermediate state lies at $10^{4}$3–$10^{4}$4 (Romano et al., 2011, Romano et al., 2010). A Swift-based uniform analysis further states that SFXT outbursts can reach peak luminosities of $10^{4}$5, with quiescent luminosities as low as $10^{4}$6–$10^{4}$7 (P. et al., 2022). Their dynamic range is therefore extreme, spanning 3–5 orders of magnitude in many descriptions and up to six orders of magnitude in soft X-rays in the Swift trigger analysis (Romano et al., 2012, P. et al., 2022).

Swift monitoring changed the physical picture of SFXTs by showing that they are not simply “off” between rare flares. Bright outbursts occupy only about $10^{4}$8–$10^{4}$9 of the time, inactivity duty cycles lie between $\sim 1$0 and $\sim 1$1, and low-flux activity can persist for up to weeks after the brightest phase (Romano et al., 2012). The most probable $\sim 1$2–$\sim 1$3 keV flux outside bright outburst is $\sim 1$4–$\sim 1$5, corresponding to luminosities of about $\sim 1$6–$\sim 1$7, which indicates prolonged intermediate-state accretion rather than binary on/off behavior (Romano et al., 2012).

Timing studies place SFXTs within, but not trivially inside, the wider HMXB hierarchy. Measured orbital periods span from about $\sim 1$8 days to $\sim 1$9 days, and spin periods are known for several systems, including AX J1841.0−0536 at $0.5$0 s and IGR J16418−4532 at $0.5$1 s (Sidoli, 2011). Proposed mechanisms include clumpy supergiant winds, orbital eccentricity or anisotropic wind geometry, and centrifugal or magnetic gating (Sidoli, 2013). The Swift BAT diagnostic study adds a phenomenological discriminator: BAT-triggered SFXTs are uniquely image triggers, are very long in BAT terms with $0.5$2–$0.5$3 s, are faint at higher BAT energies, and show a soft-XRT decline of at least three orders of magnitude within a day, usually returning toward quiescence within $0.5$4–$0.5$5 days (P. et al., 2022).

5. Extragalactic progenitor channels

The extragalactic FXT population is explicitly treated as heterogeneous. Recurrently proposed channels include supernova shock breakout, binary neutron star mergers and their magnetar remnants, white dwarf–intermediate mass black hole tidal disruption events, long-GRB-related emission, cocoon or mildly relativistic jet emission, and jetted TDEs (Eappachen et al., 2023, Eappachen et al., 4 Nov 2025).

Supernova shock breakout remains viable for a subset of low-luminosity archival events. In the XMM-Newton host study, XRT 110621 is identified as consistent with an SN SBO because its host redshift is $0.5$6, its inferred peak X-ray luminosity is $0.5$7, and its host is a low-mass star-forming galaxy (Eappachen et al., 2023). In the broader host-galaxy study of eight FXTs, two events are described as consistent with a supernova shock breakout, while several others are too luminous or too offset for that interpretation to be favored (Hoof et al., 25 Jun 2026).

Binary neutron star merger models are motivated both phenomenologically and by specific events. The Chandra transients XRT 170901, XRT 030511, and XRT 110919 show a fast rise, a $0.5$8–$0.5$9 ks plateau at $4$0, and a steep decay near $4$1, closely resembling CDF-S XT2 and motivating a magnetar-powered post-merger interpretation (Lin et al., 2022). XRT 210423, for which no transient optical counterpart was found to $4$2 AB mag and several faint host candidates were identified, is likewise argued to be best explained by a binary neutron star merger (Eappachen et al., 2023).

GRB-related origins have become especially prominent in the Einstein Probe era. EP240315a yielded the first optical and radio counterpart to a distant FXT, with a spectroscopic redshift $4$3; its optical and radio luminosities are most consistent with either a long GRB or a relativistic TDE, though the paper states that the data favor a GRB (Gillanders et al., 2024). EP240414a showed an early optical component consistent with a GRB afterglow, a rebrightening interpreted as refreshed shocks, and a later component compatible with an emerging supernova, leading to the proposal that EP FXT discoveries may predominantly result from high-redshift GRBs and appear distinct from the previously discovered lower-redshift, lower-luminosity FXT population (Srivastav et al., 2024). EP241107a further strengthens this connection: its optical and radio follow-up, plus afterglow modeling with $4$4, $4$5, and $4$6, are interpreted as indicating a GRB origin (Eappachen et al., 4 Nov 2025).

TDE channels remain viable in multiple cases. White dwarf–intermediate mass black hole tidal disruption events are retained as plausible explanations for several archival FXTs in host-based analyses (Eappachen et al., 2023, Hoof et al., 25 Jun 2026). EP240315a also remains consistent with a relativistic jetted TDE at the level of X-ray and radio luminosity, although the early rest-frame UV detections are stated to be more naturally GRB-like (Gillanders et al., 2024). The present consensus is therefore not convergence on a single engine, but convergence on a multichannel interpretation.

6. Host galaxies, offsets, and multiwavelength counterparts

Host-galaxy work is central because it fixes redshift, converts flux to luminosity, and supplies environmental diagnostics such as stellar mass, star-formation rate, metallicity, stellar population age, and offset (Hoof et al., 25 Jun 2026). The standard methodology combines positional coincidence, chance-alignment probability, projected offset, and spectral-energy-distribution fitting, often with BAGPIPES under an exponentially declining star-formation history (Eappachen et al., 2023, Hoof et al., 25 Jun 2026).

In the XMM-Newton host study, candidate hosts lie at redshifts $4$7, with inferred peak X-ray luminosities of $4$8–$4$9 and projected offsets of $3600\ \mathrm{deg}^2$0–$3600\ \mathrm{deg}^2$1 kpc (Eappachen et al., 2023). In the later eight-source host survey, deeper ground-based imaging and spectroscopy identified or re-identified candidate hosts for several events, and XRT191223 was reclassified as a Galactic stellar flare because its counterpart is consistent with an M1 V star and its $3600\ \mathrm{deg}^2$2 is consistent within $3600\ \mathrm{deg}^2$3 with a stellar flare interpretation (Hoof et al., 25 Jun 2026). These studies show that host assignment is not merely an epilogue to transient discovery; it can remove contaminants from the extragalactic sample altogether.

Large projected offsets are not uncommon. XRT 210423 has one candidate host within the 1" X-ray uncertainty region, but also two additional candidate hosts at offsets of $3600\ \mathrm{deg}^2$4 kpc and $3600\ \mathrm{deg}^2$5 kpc, respectively, and the merger interpretation is partly motivated by considering all prospective hosts jointly (Eappachen et al., 2023). EP240414a is located at a projected radial separation of 27 kpc from its likely host galaxy at $3600\ \mathrm{deg}^2$6, with a chance-coincidence probability $3600\ \mathrm{deg}^2$7 (Srivastav et al., 2024). More generally, the eight-source host study reports that a two-dimensional KS comparison of host-normalized offsets is consistent with SGRB offsets ($3600\ \mathrm{deg}^2$8) and inconsistent with the compared LGRB, Ic-BL SN, and core-collapse SN samples ($3600\ \mathrm{deg}^2$9) (Hoof et al., 25 Jun 2026).

Prompt optical and radio follow-up now provides a second major axis of discrimination. For older XMM FXTs, forced photometry in Pan-STARRS and ATLAS found no contemporaneous optical counterparts, and the quoted limits were generally not deep enough to exclude many plausible optical transients at the inferred host distances (Eappachen et al., 2023). By contrast, EP discoveries are beginning to yield a structured counterpart phenomenology. In the first year of EP operation, Lulin follow-up of 42 observable FXTs detected optical counterparts for 12; these counterparts were generally faint ($10$0 mag) and declined rapidly ($10$1 mag per day), while 11 of the 42 FXTs showed direct evidence of GRB association through significant temporal and spatial overlap (Aryan et al., 29 Apr 2025). Radio detections of EP240315a and EP241107a further anchor at least part of the EP sample in the relativistic-jet domain (Gillanders et al., 2024, Eappachen et al., 4 Nov 2025).

7. Population structure and outstanding problems

A central result of the recent literature is that FXTs are unlikely to be a single population. The host-galaxy analysis of eight FXTs concludes that Chandra- and XMM-Newton-detected FXTs likely arise from a variety of origins, whereas many Einstein Probe FXTs appear consistent with a collapsar scenario (Hoof et al., 25 Jun 2026). The optical-counterpart study of EP FXTs reaches a related conclusion from a different angle: a significant fraction of EP-discovered FXTs are associated with relativistic jets, either GRBs or jetted TDEs, while the non-detection of optical counterparts for many other EP events suggests a “dark FXT” subset analogous to dark GRBs (Aryan et al., 29 Apr 2025).

The population-level distinction between archival and EP samples is now a major organizing idea. Historical Chandra/XMM samples are mostly lower-redshift and lower-luminosity, roughly spanning a few $10$2 to $10$3 and generally $10$4, whereas EP-detected FXTs with redshifts have $10$5 and extend from $10$6 to $10$7 (Srivastav et al., 2024). In the eight-source host study, a two-dimensional KS test gave $10$8 or smaller when comparing their sample with the EP sample, strongly disfavoring a shared parent distribution (Hoof et al., 25 Jun 2026). This suggests that “FXT” may denote overlapping populations selected into prominence by different survey bandpasses and triggering modes.

Event rates sharpen the same point. One recent rate comparison gives the observed FXT volumetric rate as $10$9 and concludes that millisecond magnetars are unlikely to be the dominant progenitors, contributing at most about $30\arcsec$0 of the observed population (Biswas et al., 13 Jun 2025). The implication is not that magnetar-powered FXTs do not occur, but that even if events such as CDF-S XT2 analogs are real, they cannot account for the class as a whole.

A further unresolved issue is the relation between soft-band and hard-band FXTs. NuSTAR’s five hard-band candidates, with luminosities of $30\arcsec$1 to $30\arcsec$2 for plausible host associations and volumetric rates of $30\arcsec$3–$30\arcsec$4, most resemble low-luminosity GRBs and would often have been too faint below 10 keV to be detected by Einstein Probe in a 1000 s exposure (Brightman et al., 20 Jan 2026). This indicates that band-limited samples can encode distinct physical mixtures even when nominal durations overlap.

The immediate frontier is therefore comparative rather than purely discovery-driven. Prompt soft X-ray alerts, refined positions, deep host-galaxy work, and rapid optical and radio follow-up are now the decisive tools for separating SN shock breakout, merger-powered magnetar, GRB-afterglow, cocoon, and TDE interpretations on an event-by-event basis. The literature already supports a stable broad conclusion: FXTs are a heterogeneous family of fast X-ray phenomena whose observed properties depend jointly on progenitor physics, host environment, viewing geometry, and discovery bandpass.