Xtend: Wide-Field Soft-X-ray Telescope
- Xtend is a wide-field soft-X-ray imaging telescope aboard XRISM that maps extended celestial sources using a large 38′×38′ field of view.
- It integrates an X-ray Mirror Assembly and a Soft X-ray Imager to provide low-background, moderate spectral resolution imaging critical for diffuse emission studies.
- Xtend complements the high-resolution Resolve microcalorimeter by trading calorimetric resolution for broader coverage, supporting both targeted and survey-style observations.
Xtend is the wide-field soft-X-ray imaging telescope on board the X-Ray Imaging and Spectroscopy Mission (XRISM). It comprises the X-ray Mirror Assembly (XMA) and the Soft X-ray Imager (SXI), covers the $0.4$–$13$ keV band, and provides a field of view reported as either or in the instrument literature. Within XRISM, Xtend is the wide-field complement to the narrow-field, high-resolution Resolve microcalorimeter: it supplies imaging spectroscopy over a large solid angle, contextual imaging around Resolve targets, stray-light monitoring, and stand-alone observations of extended and time-variable sources (Mori et al., 2023, Mori et al., 2024, Uchida et al., 26 Mar 2025).
1. System definition and mission role
Xtend is one of the two telescopes onboard XRISM, launched in September 2023, and was developed in the context of XRISM as a recovery mission for Hitomi (Mori et al., 2024, Noda et al., 12 Feb 2025). Its stated science role is to recover and extend the soft-X-ray imaging capability associated with cluster studies, supernova remnants, AGN, and extended hot plasmas, while also providing wide-field support for Resolve. The required use cases include mapping extended sources, guiding Resolve observations, and accumulating survey-style data with moderate spectral resolution (Mori et al., 2023, Suzuki et al., 2024).
A recurring point of terminology is that Xtend denotes the telescope system, whereas SXI denotes the CCD focal-plane camera installed at the focus of the Xtend optics. The system-level combination of XMA and SXI is central to the published performance claims: the mirror establishes the collecting area, PSF, FoV, and vignetting behavior, while the CCD subsystem determines the sampling, event recognition, energy response, CTI behavior, and operational modes (Mori et al., 2024, Noda et al., 12 Feb 2025).
Relative to Resolve, Xtend trades calorimetric resolution for solid angle and grasp. Resolve covers a or central field, whereas Xtend spans the full Moon on the sky and covers almost the same energy band. This division of labor is explicit in the instrument requirements: Xtend was required to support wide-field spectroscopy and stray-light monitoring around Resolve targets, while also enabling its own imaging and spectroscopic science (Noda et al., 12 Feb 2025).
2. Optical and detector architecture
The XMA is described as a thin-foil-nested, conically approximated Wolter-I optic with a focal length of $5.6$ m (Mori et al., 2023, Mori et al., 2024). Published descriptions report 203 nested foils per module, Au-coated reflecting surfaces, and a pre-collimator for stray-light reduction. The mirror was designed to support an on-axis HPD of approximately $1.3'$–$1.7'$ and an on-axis FWHM of approximately $7.2''$–$13$0, depending on the performance summary and calibration context (Mori et al., 2023, Mori et al., 2024, Uchida et al., 26 Mar 2025).
At the focal plane, SXI uses four P-channel, back-illuminated CCDs arranged in a $13$1 array, each with a $13$2m-thick depletion layer (Mori et al., 2023, Mori et al., 2024, Noda et al., 12 Feb 2025). The CCDs employ frame-transfer architecture with a $13$3 mm $13$4 mm imaging area and $13$5 physical pixels of $13$6m square per device (Noda et al., 12 Feb 2025). Default on-chip $13$7 binning produces $13$8m logical pixels. Most performance papers describe a $13$9 logical array, while an initial-operations summary reports a 0 logical format for the full-window mode; the published record therefore contains multiple instrument-format descriptions (Suzuki et al., 2024, Mori et al., 2024, Noda et al., 12 Feb 2025).
The CCD design incorporates a 1 nm Al optical blocking layer, implemented as two 2 nm films, plus an additional buried Al layer around the device edges (Noda et al., 12 Feb 2025). These changes were introduced specifically to eliminate the light-leak artifacts encountered on Hitomi. The same redesign cycle added a notch implant in the transfer channel, which confines signal packets and improves radiation tolerance; flight-model tests report approximately 3 higher radiation tolerance than in un-notched devices (Noda et al., 12 Feb 2025). Charge injection is executed every 160 physical rows, equivalently every 80 logical rows, to mitigate CTI (Mori et al., 2023, Suzuki et al., 2024).
The instrument requirements tie the optical and detector subsystems together. Xtend was required to achieve at least 4 FoV, but in practice the delivered FoV is reported as 5 or 6; the required effective area is 7 cm8 at 9 keV and 0 cm1 at 2 keV; the required energy resolution is 3 eV FWHM at Mn K4 at beginning of life and 5 eV at end of life (Noda et al., 12 Feb 2025). Ground and in-flight reports state that these requirements were met (Mori et al., 2024, Noda et al., 12 Feb 2025).
3. Readout modes, thermal control, and CTI mitigation
Xtend is operated in several readout modes designed to balance FoV, time resolution, live-time fraction, and pile-up tolerance. The full-window mode reads the entire array with a frame exposure of 6 s and a live-time fraction of 7. The 8-window mode uses a 9 s frame exposure and a live-time fraction of 0. The 1-window mode with burst option reduces the frame exposure to 2 s, with photon arrival time stamps accurate to 3 s within each 4 s cycle, at a live-time fraction of 5 (Yoneyama et al., 25 Nov 2025). Initial operations verified full-window, 6-window, burst, and 7-window-plus-burst operation, together with time-tagged and automated command sequences for South Atlantic Anomaly passages and dark-level handling (Suzuki et al., 2024).
Thermal control is provided by a single-stage Stirling cooler with fine heater control. The nominal CCD operating temperature is 8C, with demonstrated operation down to 9C (Mori et al., 2023, Mori et al., 2024). Spacecraft thermal-vacuum tests and in-orbit commissioning showed stable operation at the design temperature, with reported stability of 0C in TVAC conditions and 1C peak-to-peak, or control within 2C, during early operations (Suzuki et al., 2024, Mori et al., 2024). Pre-flight thermal-vacuum tests yielded Mn K3 resolutions of approximately 4–5 eV at 6C, consistent with subsystem tests (Mori et al., 2024).
CTI calibration is a major aspect of the Xtend CCD system. Experimental studies identified at least three trap populations with characteristic time constants of approximately 1 pixel, 10 pixels, and 100 pixels, respectively (Kanemaru et al., 2020). The shortest-time-constant traps dominate immediate trailing charge, while deeper traps contribute more to cumulative CTI. The published model distinguishes four transfer phases—7, 8, 9, and $5.6$0—and parameterizes the pulse-height degradation as a product of per-phase losses. A key empirical result is that the fast transfer in the imaging area is approximately $5.6$1–$5.6$2 worse than the fast transfer in the storage area, despite identical clock periods (Kanemaru et al., 2020). With charge injection enabled, the apparent flux dependence of CTI becomes negligible within the tested range, and the instrument is intended to use charge injection in all science modes (Kanemaru et al., 2020).
Operational experience also refined several details after launch. The split threshold was kept at 25 channels after balancing energy resolution against good-event ratio; bad columns were masked to suppress pseudo-events; and the charge-injection row pattern was shifted on 2024-03-10 after the original pattern overlapped the nominal aim point and caused approximately $5.6$3 photon loss for point sources (Suzuki et al., 2024). These adjustments illustrate that the delivered in-orbit behavior is not only a consequence of hardware design, but also of command sequencing, calibration, and event-screening policy.
4. In-orbit performance and calibration status
The in-orbit performance literature reports that Xtend meets or exceeds its primary engineering requirements. First-light observations demonstrated the full FoV on the galaxy cluster Abell 2319, with all CCDs active except for known bad columns and with the onboard $5.6$4Fe calibration spots visible in the corners (Mori et al., 2024). On-axis image quality in orbit is reported as HPD $5.6$5 or $5.6$6 and FWHM $5.6$7 or $5.6$8, depending on the calibration summary. Published examples state that Xtend resolves structures down to approximately $5.6$9 and localizes point sources to $1.3'$0 (Mori et al., 2024, Uchida et al., 26 Mar 2025).
The in-orbit spectral resolution is reported as $1.3'$1–$1.3'$2 eV FWHM at $1.3'$3–$1.3'$4 keV, consistent with ground measurements and comfortably inside the mission requirement (Mori et al., 2024, Uchida et al., 26 Mar 2025). This resolution is sufficient to separate He-like and H-like Fe K$1.3'$5 lines, a point emphasized in the in-orbit performance summary (Uchida et al., 26 Mar 2025). Cross-calibration observations of 3C 273 with other major X-ray observatories yielded on-axis effective areas of approximately $1.3'$6 cm$1.3'$7 at $1.3'$8 keV and $1.3'$9 cm$1.7'$0 at $1.7'$1 keV, matching pre-launch expectations and ground tests within the reported uncertainties (Uchida et al., 26 Mar 2025). Earlier ground calibration summaries quoted approximately $1.7'$2 cm$1.7'$3 at $1.7'$4 keV and $1.7'$5 cm$1.7'$6 at $1.7'$7 keV (Mori et al., 2024).
Background performance is a defining feature of Xtend. The mission requirements specified NXB below $1.7'$8 counts keV$1.7'$9 s$7.2''$0 arcmin$7.2''$1 cm$7.2''$2 in $7.2''$3–$7.2''$4 keV, and early in-orbit operations reported that this requirement was met (Suzuki et al., 2024, Noda et al., 12 Feb 2025). The in-orbit performance paper quotes an NXB level of $7.2''$5 counts s$7.2''$6 keV$7.2''$7 arcmin$7.2''$8 cm$7.2''$9 at $13$00 keV, while an earlier status paper describes a continuum of approximately $13$01 counts s$13$02 arcmin$13$03 keV$13$04 at the same energy and emphasizes orbital stability to $13$05 over one month (Mori et al., 2024, Uchida et al., 26 Mar 2025). Both accounts converge on the same instrumental conclusion: Xtend has a low, stable, and relatively line-sparse particle background in low-Earth orbit.
Several known instrumental issues from Hitomi were explicitly addressed in Xtend and verified in orbit. Light leakage was suppressed by sealing panel openings, applying low-reflectivity coatings, and thickening the Al optical-blocking layer; in-orbit checks found no significant light-leak events (Uchida et al., 26 Mar 2025). Crosstalk events caused by capacitive coupling were mitigated through threshold retuning and pipeline masking, with pseudo-crosstalk events reported as suppressed by more than $13$06 above $13$07 keV (Uchida et al., 26 Mar 2025). Health monitoring through overclocking data, calibration-source spectra, and day-Earth observations indicates stable readout noise, CTI growth rates of approximately $13$08–$13$09 yr$13$10 per transfer, and negligible contamination even one year after launch (Uchida et al., 26 Mar 2025).
A widely used systems metric for diffuse-emission work is the grasp, reported as $13$11 cm$13$12 deg$13$13 at $13$14 keV (Uchida et al., 26 Mar 2025). This value, together with the low NXB and wide FoV, underlies many of the instrument’s extended-source use cases.
5. Analysis methodology, extended-emission sensitivity, and pile-up treatment
Xtend’s published analysis workflows for faint diffuse emission rely on the combination of wide FoV, low NXB, and explicit instrumental modeling. In the V4641 Sgr study, the data-reduction sequence consisted of the standard XRISM pipeline with HEAsoft 6.32 and CALDB v20240815, good-time filtering to exclude Earth eclipse, SAA, and limb intervals, removal of flickering pixels, production of a raw $13$15–$13$16 keV image, subtraction of a particle-background map derived from night-Earth data and rescaled to the $13$17–$13$18 keV count rate, and application of a vignetting correction derived from day-Earth flat-field data (Suzuki et al., 2024). This sequence is representative of how Xtend’s low-background design is operationalized for diffuse-source work.
For imaging analysis, the PSF at $13$19–$13$20 keV was simulated with the xrtraytrace ray-tracing tool and calibrated in orbit, with a stated systematic uncertainty in the PSF tail of less than $13$21 inside $13$22 (Suzuki et al., 2024). In that same analysis, the extended component was modeled with a Gaussian-like radial surface-brightness profile,
$13$23
and the imaging significance was assessed either by an F-test or through annular-bin excess significance relative to PSF plus background (Suzuki et al., 2024). For spectral work, RMFs were generated with xtdrmf, ARFs with xaarfgen under a uniform-sky assumption, and source-plus-background spectra were fitted simultaneously with sky, NXB, and extended-emission components (Suzuki et al., 2024). These details matter because Xtend’s moderate angular resolution is offset by a calibration strategy specifically tailored to low-surface-brightness analyses.
Pile-up is the main limiting effect for bright sources. A dedicated pile-up simulator for XRISM/Xtend was built on ComptonSoft and Geant4, with charge-cloud formation, diffusion, frame formation, and grade reconstruction treated in Monte Carlo (Yoneyama et al., 25 Nov 2025). The pile-up fraction is defined as
$13$24
and, for a single pixel with Poisson arrival rate $13$25 and frame time $13$26, the pile-up probability is
$13$27
Under a Crab-spectrum assumption for a point source, the reported $13$28 pile-up limits are $13$29 counts s$13$30 in full-window mode, $13$31 counts s$13$32 in $13$33-window mode, and $13$34 counts s$13$35 in $13$36-window mode with burst option (Yoneyama et al., 25 Nov 2025). For a diffuse source in full-window mode, the reported $13$37 threshold is $13$38 counts s$13$39 arcmin$13$40, corresponding to approximately $13$41 erg s$13$42 cm$13$43 arcmin$13$44 in $13$45–$13$46 keV (Yoneyama et al., 25 Nov 2025). This quantitatively defines the regime in which Xtend can operate as a high-grasp wide-field imager without significant flux distortion.
6. Scientific applications and demonstrated results
Xtend’s science program spans both support for Resolve and stand-alone imaging spectroscopy. The in-orbit performance summary emphasizes its role in identifying contaminating point sources and foreground or background structure outside Resolve’s FoV, and notes that Xtend’s soft-band coverage compensated for Resolve during early mission phases (Uchida et al., 26 Mar 2025). The same report identifies first-light cluster and SNR observations, reduced-pile-up window modes for bright rapidly varying sources, transient searches, and tiling-mode follow-up of gravitational-wave and high-energy-neutrino alerts as core applications (Uchida et al., 26 Mar 2025). This suggests that Xtend’s operational niche is defined less by single-point-source acuity than by a combination of grasp, FoV, background control, and flexible timing modes.
A major demonstration of its extended-source capability is the detection of diffuse X-ray emission around the PeVatron microquasar V4641 Sgr (Suzuki et al., 2024). Xtend detected the emission with significance greater than $13$47 in imaging and greater than $13$48 in spectral analysis. The radial extent was fitted with $13$49 arcmin, corresponding to $13$50 pc at a distance of $13$51 kpc, and the extracted annular spectrum yielded a non-thermal fit with $13$52 cm$13$53, photon index $13$54, surface brightness $13$55 erg s$13$56 cm$13$57 arcmin$13$58, and integrated $13$59–$13$60 keV flux $13$61 erg s$13$62 cm$13$63 (Suzuki et al., 2024). The interpretation offered in that work is that the observed scale can be matched either by an enhanced magnetic field of approximately $13$64G or by a suppressed diffusion coefficient of approximately $13$65 cm$13$66 s$13$67 at $13$68 TeV. More generally, the paper argues that Xtend’s very large FoV and well-characterized NXB enable the first detection of faint extended X-ray halos around Galactic PeVatrons (Suzuki et al., 2024).
Xtend has also been used in an unconventional observing configuration: day-Earth occultations that capture solar-flare X-rays reflected in the Earth’s atmosphere (Suzuki et al., 5 Sep 2025). Over roughly one year of data, Xtend measured abundances of Mg, Si, S, Ar, Ca, and Fe during M- and X-class flares and found an inverse-FIP effect consistent with Suzaku-based results. The large effective area and FoV were reported to permit abundance tracking in several X-class flare loops on timescales of a few 100 s, and the neutral or low-ionized Fe-K$13$69 equivalent width was found to show an anti-correlation with hard-X-ray flux with best-fit power-law slope $13$70 (Suzuki et al., 5 Sep 2025). Although this is not a primary astrophysical use case of the telescope design, it demonstrates that Xtend’s calibration, throughput, and cadence are sufficiently stable to support quantitative spectroscopy even in a scattering-dominated geometry.
A common misconception is to treat Xtend primarily as an auxiliary instrument to Resolve. The published in-orbit and science-performance record does not support that reduction. Xtend was designed to complement Resolve, but the literature also presents it as a stand-alone wide-field imaging spectrometer with low and stable background, large grasp, multiple readout modes, and demonstrated sensitivity to extended halos, flare-reflection spectra, diffuse backgrounds, transient phenomena, and other contexts in which solid angle and background systematics are more decisive than sub-eV spectroscopy (Uchida et al., 26 Mar 2025).