EP-FXT: Einstein Probe Follow-up X-ray Telescope

• EP-FXT is a narrow-field, Wolter-I X-ray telescope designed to rapidly follow up transients detected by WXT, refining source positions from arcminutes to arcseconds.
• The instrument evolved from a lobster-eye concept to a two-module, pn-CCD system offering broad energy coverage (0.3–10 keV), high-resolution spectroscopy, and robust timing capabilities.
• EP-FXT’s low background and wide 1° field of view also support extended X-ray studies, making it effective for both transient events and diffuse emissions in galaxy clusters.

The Einstein Probe Follow-up X-ray Telescope, usually abbreviated FXT and often referred to as EP-FXT to distinguish it from the mission itself, is the narrow-field X-ray instrument on board Einstein Probe. Within the mission architecture it serves the pointed follow-up role after discovery by the Wide-field X-ray Telescope (WXT): refining positions, obtaining deeper X-ray measurements, and supporting rapid multi-wavelength identification of newly detected transients. A common source of confusion is that the 2015 small-mission concept described FXT as a micro-pore-optics lobster-eye telescope, whereas the later mission paper and in-flight literature describe the operational instrument as a two-module Wolter-I pn-CCD system; the instrument’s encyclopedic description therefore requires attention to both the mission-concept phase and the realized in-flight configuration (Yuan et al., 2015, Yuan et al., 2022, Xiao et al., 11 Jul 2025).

1. Mission role and operational concept

Einstein Probe was designed around a two-step workflow. WXT surveys the sky in soft X-rays over a very large instantaneous field of view, while FXT performs pointed follow-up of newly discovered events and targets of opportunity. The 2015 mission paper states that EP would carry “a survey instrument Wide-field X-ray Telescope (WXT) with a large instantaneous Field-of-View (FoV, $60^\circ \times60^\circ$) and a narrow-field ($1^\circ \times1^\circ$) Follow-up X-ray Telescope (FXT), as well as a fast alert downlink system,” and describes FXT as the instrument used to “perform immediate follow-up observations of newly discovered transients” (Yuan et al., 2015). The later mission paper preserves the same functional division, but frames FXT as the instrument that improves WXT source positions from about $\sim 1$ arcmin to several arcseconds while providing spectra, light curves, and images (Yuan et al., 2022).

The operational sequence in the mission literature is explicit. WXT surveys; onboard processing detects and classifies a new transient; the spacecraft slews; FXT begins pointed follow-up; and WXT continues to monitor the new sky region centered on the transient position (Yuan et al., 2015, Yuan et al., 2022). The 2022 mission design states that FXT can normally begin pointed follow-up within 4 minutes after onboard trigger, while transient alert downlink latency is about 10 minutes or so (Yuan et al., 2022). The earlier 2015 concept stressed that alerts should reach the ground “within one minute or so,” illustrating that rapid public alerting was central from the beginning (Yuan et al., 2015).

The scientific targets of this follow-up chain are correspondingly broad. The mission papers associate FXT-enabled follow-up with tidal disruption events, otherwise quiescent black-hole flares, electromagnetic counterparts of gravitational-wave transients, high-redshift GRBs, supernova shock breakouts, X-ray flashes, low-luminosity GRBs, X-ray rich GRBs, GRB precursors, magnetars, stellar coronal flares, classical novae, supergiant fast X-ray transients, AGN and blazar outbursts, and compact-object transients such as X-ray binaries (Yuan et al., 2015).

2. From concept study to operational instrument

The instrument history is not static. The 2015 EP concept paper described FXT as an MPO lobster-eye telescope using the same micro-pore-optics focusing technology as WXT, with a single module, a $1^\circ \times 1^\circ$ field of view, a 1.4 m focal length, a $6\times6$ MPO tiling, a CCD focal-plane detector, a bandpass of 0.5–4.0 keV, angular resolution $<5$ arcmin FWHM, effective area $60~\mathrm{cm^2}$ at 1 keV in the central focus, and sensitivity $\sim 3\times10^{-12}~\mathrm{erg~s^{-1}~cm^{-2}}$ at 1000 s (Yuan et al., 2015). By contrast, the 2022 mission paper and the subsequent in-flight literature describe FXT as a two-unit Wolter-I telescope with pn-CCD detectors, substantially broader energy coverage, and arcsecond-class localization capability (Yuan et al., 2022, Xiao et al., 11 Jul 2025).

This evolution can be summarized compactly.

Aspect	2015 concept description	Later mission / in-flight description
Optical concept	MPO lobster-eye	Wolter-I nested mirrors
Modules	1	2 co-aligned units, FXT-A and FXT-B
Focal length	1.4 m	1.6 m
Detector	CCD	pn-CCD
Energy range	0.5–4.0 keV	0.3–10 keV
Field of view	$1^\circ \times 1^\circ$	$1^\circ \times 1^\circ$
Angular performance	$1^\circ \times1^\circ$0 arcmin FWHM	30 arcsec HPD design; later papers quote 20–24 arcsec or 22 arcsec HPD
Effective area	$1^\circ \times1^\circ$1 at 1 keV	$1^\circ \times1^\circ$2 at 1.25 keV for 2 units

In the operational configuration, each FXT unit is a short-focal-length Wolter-I telescope with 54 nested gold-coated nickel shells and a focal length of 1.6 m. The two mirror assemblies are contributions from ESA and MPE, the detectors are pn-CCDs developed/provided by MPE, and the instrument development is led by IHEP, CAS with major international contributions (Yuan et al., 2022). The focal-plane detector is a frame-transfer, back-illuminated, fully depleted pn-CCD with a $1^\circ \times1^\circ$3 pixel imaging area and $1^\circ \times1^\circ$4 pixels, operating over 0.3–10 keV (Yuan et al., 2022).

The 2022 mission paper quotes the core design values as: field of view $1^\circ \times1^\circ$5, effective area $1^\circ \times1^\circ$6 at 1.25 keV for the two units, angular resolution 30 arcsec HPD, energy resolution 120 eV at 1.25 keV FWHM, and source locating uncertainty 4 arcsec $1^\circ \times1^\circ$7 (Yuan et al., 2022). Qualification measurements cited there reported 21.0 ± 0.3 arcsec HPD at 1.49 keV for one mirror assembly and detector energy resolution of 97 eV at 1.25 keV, indicating that at least some qualification-model measurements were better than the nominal goals (Yuan et al., 2022). Later event papers variously quote angular resolution of 30 arcsec, 20–24 arcsec half-power diameter, or 22 arcsec HPD, depending on context and analysis pipeline (Xiao et al., 11 Jul 2025, Zhang et al., 2024, Wen et al., 16 Jun 2026).

3. Detectors, observing modes, and background calibration

FXT’s pn-CCD system is coupled to a six-position filter wheel comprising thin, medium, open, closed, hole, and $1^\circ \times1^\circ$8Fe calibration positions. The detectors are operated at $1^\circ \times1^\circ$9C using helium pulse-tube refrigerators, with qualification-model tests giving temperature stabilization of $\sim 1$0 (Yuan et al., 2022). The readout modes are operationally important: full-frame mode has a 50 ms readout time, partial-window mode 2 ms, and timing mode $\sim 1$1s per row; the bright-source limits for pile-up fraction $\sim 1$2 are 10 mCrab, 200 mCrab, and 5 Crab, respectively (Yuan et al., 2022).

Pre-launch modeling treated FXT background as the sum of instrumental background and field-of-view background. The Geant4 study predicted an instrumental background of $\sim 1$3 in the imaging area over 0.5–10 keV, with the cosmic X-ray background dominating below 2 keV through the optics and particle/off-axis instrumental contributions dominating above 2 keV. For a 25-minute pointed observation it predicted sensitivity of several $\sim 1$4Crab, in the order of $\sim 1$5, in 0.5–2 keV and several tens of $\sim 1$6Crab, in the order of $\sim 1$7, in 2–10 keV; the paper states that this sensitivity becomes worse by a factor of $\sim 1$8 if additional 10% systematic uncertainty of background subtraction is included (Zhang et al., 2021).

The in-flight background paper then replaced simulation with calibration from Performance Verification and filter-wheel-closed data. It finds that the instrumental backgrounds of FXT-A and FXT-B are consistent with each other, with an average rate of $\sim 1$9 at 0.5–10 keV for each module, nearly uniform across detector pixels apart from a row-dependent increase of $1^\circ \times 1^\circ$0, and strongly modulated by geomagnetic position (Zhang et al., 1 Jul 2025). Within a single observation the 0.5–10 keV background can vary by a factor of $1^\circ \times 1^\circ$1–8 over an orbit after normal good-time screening, while long-term evolution shows a periodic variation associated with orbital precession at $1^\circ \times 1^\circ$2 days and amplitude $1^\circ \times 1^\circ$3 (Zhang et al., 1 Jul 2025).

A distinctive feature of FXT full-frame operation is the simultaneous recording of events in both the imaging area (IMG) and the frame-store area (FSA). The in-flight background paper exploits the linear correlation between FSA and IMG rates to build an empirical IMG background model, implemented in the full-frame background-estimation tool fxtbkggen, available from FXT CALDB Version 1.20 (Zhang et al., 1 Jul 2025). The same paper notes that the pre-launch simulation underestimated the measured in-orbit background by about 23%, rounded to $1^\circ \times 1^\circ$4 in the abstract and conclusion (Zhang et al., 1 Jul 2025).

4. Localization and analysis workflow in practice

The practical value of FXT is most visible in localization. A one-year optical follow-up study of EP-discovered fast X-ray transients states that WXT typically provides localization with error circles of radius $1^\circ \times 1^\circ$5, whereas EP-FXT improves this to error circles having radii between 10 and 30 arcsec, and identifies that improvement as the reason narrow-field optical facilities can search efficiently for multi-wavelength counterparts (Aryan et al., 29 Apr 2025).

Case studies show that this is not merely a design aspiration. For EP240408a, WXT localized the transient to a 3 arcmin radius uncertainty at 90% confidence, while the first FXT observation detected an X-ray source inside that circle and improved the localization to 5 arcsec (90% c.l.); the same observation provided high-quality 0.5–10 keV spectroscopy and later a deep nondetection in a second visit (Zhang et al., 2024). For EP J115415.8−501810, FXT improved the localization from the WXT discovery uncertainty of 2.1 arcmin to about 10 arcsec, refined the source position, detected a coherent X-ray modulation, and constrained the spectrum well enough to identify strong local absorption and hot optically thin thermal emission (Xiao et al., 11 Jul 2025). For EP241107a, one follow-up paper states that EP-FXT began follow-up about five minutes after the WXT trigger and detected an X-ray source with an uncertainty of about $1^\circ \times 1^\circ$6, enabling confident association with the optical and radio counterparts (Eappachen et al., 4 Nov 2025).

The software stack used in event papers is correspondingly mature. Standard reduction is done with FXT Data Analysis Software, most commonly through fxtchain, which performs particle identification, pulse-invariant calculation, bad/hot pixel flagging, good-time screening, spectrum and light-curve extraction, and response generation; barycentric correction is carried out with fxtbary in timing analyses (Xiao et al., 11 Jul 2025, Zhang et al., 2024). Extended-source studies use the same family of tools, together with fxtarfgen, fxtrmfgen, and xselect, and background treatment may incorporate fxtbkggen plus image reprojection and point-source masking (Wen et al., 16 Jun 2026, Zheng et al., 10 Jul 2025).

The observing mode matters for analysis. In the intermediate-polar study EP J115415.8−501810, the single FXT observation was taken in full-frame mode, and the authors emphasize that full-frame mode preserves full imaging capability and provides 50 ms time resolution, in principle enabling searches for periodicities down to the sub-second regime (Xiao et al., 11 Jul 2025). This is a useful corrective to the misconception that FXT is only a localization camera: the in-flight papers use it as an imaging, timing, and spectroscopy instrument.

5. Transient astrophysics enabled by EP-FXT

Mission-level science cases in the proposal literature presented FXT as the post-discovery instrument for tidal disruption events, black-hole flares, gravitational-wave counterparts, high-redshift GRBs, supernova shock breakouts, X-ray flashes, low-luminosity GRBs, magnetars, novae, stellar flares, blazar outbursts, and other variable sources (Yuan et al., 2015). In-flight papers now show how that role is realized.

One important use case is compact-binary characterization. In EP J115415.8−501810, FXT transformed a WXT transient alert into a physically interpretable data set: a 3093 s full-frame observation yielded a 231 s modulation in the 0.3–2 keV band, spectroscopy consistent with tbabs × tbpcf × bremss, a partial-covering absorption column $1^\circ \times 1^\circ$7, covering fraction $1^\circ \times 1^\circ$8, and $1^\circ \times 1^\circ$9 keV, leading to classification as an intermediate polar (Xiao et al., 11 Jul 2025). In EP J005146.9−730930, regular EP-FXT monitoring over about three months established the rise, peak, and decay of a type II Be/X-ray binary outburst, resolved the source from nearby SXP 138, and revealed spectral hardening toward higher luminosity; XMM-Newton later supplied the decisive $6\times6$0 s pulsation measurement (Haberl et al., 23 Jul 2025).

Another use case is the characterization of EP fast X-ray transients. For EP240408a, FXT localized the source to 5 arcsec, delivered a high-S/N joint FXT+NICER spectrum consistent with an absorbed power law with $6\times6$1 and $6\times6$2, and contributed an early plateau measurement plus a late deep nondetection; the event remained inconsistent with known classes such as jetted TDEs, GRBs, X-ray binaries, and fast blue optical transients, so its nature was left open (Zhang et al., 2024). For EP240315a and EP240414a, the literature uses EP-FXT follow-up as the step that converted WXT alerts into securely localized optical/radio/SN counterparts, ultimately linking some EP FXTs to high-redshift GRBs, broad-lined Type Ic supernovae, or more complex jet-driven stellar explosions (Gillanders et al., 2024, Dalen et al., 2024).

These individual case studies sit within a larger survey stream. A population analysis of EP-discovered fast X-ray transients is based on WXT-selected events rather than on an FXT-only sample, but it clarifies the class of sources that FXT is asked to localize and characterize: the paper finds significant luminosity evolution $6\times6$3, a broken local luminosity function with break luminosity $6\times6$4, and a local volumetric rate of about $6\times6$5, arguing that much of the EP FXT population may be linked to collapsars and low-luminosity or gamma-ray-suppressed long GRBs (Guo et al., 15 Oct 2025).

6. Extended-source applications and scientific position

Although designed as a follow-up telescope, EP-FXT has also emerged as a capable instrument for diffuse X-ray astrophysics. This development is driven by the combination of a $6\times6$6 field of view and low particle background, repeatedly highlighted in the cluster papers.

In the Virgo Cluster, a deep 295 ks EP-FXT data set revealed a giant sloshing spiral connecting the northwest and southeast cold fronts, while spatially resolved spectroscopy over a field of approximately 28.5 arcmin produced two-dimensional maps of temperature, metallicity, pseudo-pressure, and pseudo-entropy. The paper explicitly states that the large field of view and low particle background are what make FXT well suited to this kind of low-brightness diffuse emission, and it notes that the particle background is about one-fifth that of eROSITA (Zheng et al., 10 Jul 2025). In A3571, a 40 ks Performance Verification observation extended measurements beyond $6\times6$7 in imaging, found northern and southern surface-brightness excesses without clear shocks or cold fronts, and argued for sloshing motions triggered by an off-center minor merger, again presenting EP-FXT as especially capable for cluster outskirts because of wide field and very low particle background (Zheng et al., 8 Jan 2026). In WHY J0501+01, combined survey and target-of-opportunity FXT data were used to measure $6\times6$8 keV and $6\times6$9, classify the system as a disturbed cluster, and connect its intracluster-medium state to a rare sextuple-merging brightest cluster galaxy (Wen et al., 16 Jun 2026).

These results suggest that EP-FXT’s scientific identity is now broader than the narrow label “follow-up telescope” might imply. Direct evidence from the mission and in-flight literature still places transient localization and characterization at the center of its design (Yuan et al., 2022), but the operational record shows a second niche: wide-field, low-background imaging spectroscopy of extended, low-surface-brightness systems such as nearby galaxy clusters (Zheng et al., 10 Jul 2025, Zheng et al., 8 Jan 2026). In both regimes, the same properties recur—moderate angular resolution, broad 0.3–10 keV coverage, pn-CCD spectroscopy, and stable background modeling—and together they define EP-FXT as the precision X-ray instrument within the Einstein Probe ecosystem.