X-ray Spectroscopy & Timing (XSPECT)

• XSPECT is defined as the integrated measurement of X-ray energy spectra and photon arrival times, enabling detailed study of spectral components and transient phenomena.
• The methodology leverages high effective area detectors and precise event-level timing to capture variability from microseconds to years, essential for compact object and accretion studies.
• Applications span probing black-hole and neutron-star physics to transient astrophysics, as demonstrated by instruments like XPoSat's XSPECT and complementary mission designs.

X-ray SPECtroscopy and Timing (XSPECT) denotes the integrated acquisition and interpretation of X-ray energy spectra and photon arrival times, and in current literature it appears both as a general observational paradigm and as the formal name of the soft X-ray spectrometer aboard India’s XPoSat mission. In both senses, the defining objective is the same: to couple energy-resolved measurements with precise time tagging and, when possible, long-baseline monitoring, so that spectral components, variability processes, and transient phenomena can be studied on timescales ranging from microseconds to years (Ray et al., 2018, V et al., 26 May 2025).

1. Terminology and conceptual scope

In mission-concept literature, XSPECT is used explicitly to mean X-ray spectroscopy and timing rather than the XSPEC spectral-fitting software. LOFT states this distinction directly, and STROBE-X adopts the same combined spectroscopy-and-timing framing for broadband, high-throughput studies of compact objects, accretion flows, and transients (Belloni et al., 2012, Ray et al., 2019).

The term is also used as an instrument name. On XPoSat, XSPECT is the soft X-ray spectrometer/timer co-aligned with the POLIX polarimeter and configured for long-duration spectro-temporal monitoring in the 0.8–15 keV band (Chatterjee et al., 11 Jun 2025). A related usage appears in methodological work, where “X-ray spectroscopy and timing” is treated as an analysis strategy rather than only a hardware configuration; “time-decomposed spectroscopy,” for example, uses temporal behavior to separate thermal and nonthermal flare components before forward spectral modeling (Setterberg et al., 29 Apr 2025).

Taken together, these usages define XSPECT as a class of measurements in which the photon’s energy and arrival time are jointly primary observables. A plausible implication is that XSPECT is best understood as an observational regime spanning instrumentation, calibration, and inference methodology rather than as a single mission architecture.

2. Instrumental realizations and mission concepts

Mission studies have converged on a recurring XSPECT architecture: large effective area, event-level timing, moderate to high spectral resolution, low background, and some combination of pointed and wide-field capability. STROBE-X realizes this with a soft-band concentrator array, a hard-band large-area collimated detector, and a coded-mask wide-field monitor; LOFT uses a square-meter-class collimated Large Area Detector plus a coded-mask Wide Field Monitor; EPE emphasizes non-dispersive microcalorimetry with sub-ms timing; XPoSat XSPECT implements a smaller-area but operational soft-band spectrometer optimized for long dwell times (Ray et al., 2018, Belloni et al., 2012, Garcia et al., 2011, Chatterjee et al., 11 Jun 2025).

Realization	Band and timing	Defining characteristics
STROBE-X	0.2–30 keV; XRCA better than 100 ns RMS, LAD maximum drift ≈7 μs	XRCA + LAD + WFM; order-of-magnitude effective-area gains over NICER and RXTE; WFM sensitivity ≈20× the RXTE ASM
LOFT	2–80 keV; 10 μs resolution	LAD peak effective area ~10 m² at 8 keV; WFM observes >1/3 of the sky at once
EPE	0.2–10 keV; sub-ms timing	XMS microcalorimeter with ΔE ≈ 2.5 eV FWHM and $A_{\mathrm{eff}} \approx 5{,}432~\mathrm{cm}^2$ at 1 keV
XPoSat XSPECT	0.8–15 keV; 1 ms resolution	16 SCDs, long integrations driven by POLIX, pile-up fraction <1% up to ~60 Crab

STROBE-X is the most explicit statement of the XSPECT paradigm. Its XRCA covers 0.2–12 keV with CCD-level energy resolution of 85–175 eV FWHM and timing better than 100 ns RMS, while the LAD covers 2–30 keV nominally, with dynamic range to 80 keV, total effective area ≈5 m² at 8 keV, and count rates of ≈120,000–156,000 counts s$^{-1}$ for Crab-like sources, depending on formulation within the mission studies (Ray et al., 2018, Ray et al., 2019). LOFT anticipated many of the same design principles: micro-pore collimation, silicon drift detectors, CCD-like resolution down to 200 eV FWHM at 6 keV, and an event throughput of ≈240,000 counts s$^{-1}$ at 1 Crab (Belloni et al., 2012). EPE represents a different branch of XSPECT, replacing silicon-timing detectors with a TES microcalorimeter array and retaining sub-ms timing by distributing high flux across many pixels (Garcia et al., 2011).

3. XSPECT on XPoSat

XPoSat, launched on 1 January 2024 into a 650 km, 6° inclination orbit, carries two co-aligned instruments: POLIX and XSPECT. Because POLIX requires long integrations, observations last from days to weeks, and science exposures occur on the night side of each orbit; the resulting day-averaged source-observation duty cycle is ~20%. XSPECT exploits this operational pattern for long-term spectro-temporal monitoring of bright sources (V et al., 26 May 2025, Chatterjee et al., 11 Jun 2025).

The instrument uses sixteen second-generation Swept Charge Devices (CCD236, e2v) arranged in four quads. One SCD is blocked by a 500 μm Ta sheet to monitor the local particle background. The operating band is 0.8–15 keV, readout is at 100 kHz, event timing resolution is 1 ms, and event times are transformed to UTC using an onboard GPS receiver and then barycenter-corrected for timing analysis. Two square collimator classes are used; post-launch calibration refined their fields of view to 1.95°×1.95° and 2.89°×2.89°, with open area fractions of 59.5% and 73.5% and FOV-averaged alignment correction factors of 88.2% and 94.5%, respectively (Chatterjee et al., 11 Jun 2025).

The optical train includes aluminized polyimide light-blocking filters, while the response model folds together device geometry, collimator transmission, alignment corrections, detector quantum efficiency, filter transmission, and the single-event fraction. On-axis open aperture area before multiplication by energy-dependent transmission and efficiency is ~39.9 cm². Ground and in-flight calibrations further tuned the effective device depth to 57 μm and the top dead-layer thickness to 952 nm (V et al., 26 May 2025, Chatterjee et al., 11 Jun 2025).

A key operational property is brightness tolerance. XSPECT’s sweeping SCD readout makes it effectively pile-up free for extreme soft X-ray fluxes by the standards of conventional CCD spectrometers; the in-flight calibration paper gives a pile-up fraction <1% up to ~60 Crab, while instrument-design work emphasizes $>2000$ counts s$^{-1}$ without photon pile-up (Chatterjee et al., 11 Jun 2025, V et al., 26 May 2025).

4. Timing, spectroscopy, and analysis formalisms

Much of XSPECT reduces to count-statistics and Fourier analysis. STROBE-X expresses the expected counts in an exposure $t$ as

$C = t \int F(E)\,A_{\mathrm{eff}}(E)\,dE,$

with Poisson-limited sensitivity approximated by

$\mathrm{SNR} \approx \frac{S}{\sqrt{S+B}}.$

For timing analyses, the Nyquist scale follows $f_{\mathrm{N}} = 1/(2\Delta t)$, and the Fourier resolution is $\delta f = 1/T_{\mathrm{obs}}$. In the STROBE-X design this corresponds to $^{-1}$0 for XRCA and $^{-1}$1 for LAD (Ray et al., 2018).

Power spectral analysis is standardized. The Leahy normalization is written as

$^{-1}$2

and the QPO quality factor as

$^{-1}$3

Cross-spectral timing, lag–energy spectra, coherence, and phase-coherent pulse timing are therefore natural components of XSPECT rather than secondary products (Ray et al., 2018, Wilson-Hodge et al., 2017).

Absolute time reconstruction is equally central. XRISM’s timing system provides an end-to-end example of XSPECT timing infrastructure: GPS receiver $^{-1}$4 Satellite Management Unit time indicator $^{-1}$5 SpaceWire distribution $^{-1}$6 instrument LOCAL_TIME $^{-1}$7 ground time assignment $^{-1}$8 barycentric correction. The mission requirement was 1.0 ms absolute timing accuracy, split equally between Resolve and the bus+ground components; in orbit, the bus+ground jitter was found to be below $^{-1}$9, and the Crab main pulse aligned with NICER and NuSTAR within $^{-1}$0 (Terada et al., 18 Mar 2025).

XSPECT has also been generalized into an inference method. In time-decomposed spectroscopy, the observed signal is modeled as

$^{-1}$1

and per-energy coefficients are estimated by weighted linear regression onto temporal pseudobases. In two solar-flare case studies, the nonthermal energies were constrained by more than an order of magnitude and by two orders of magnitude relative to traditional fits (Setterberg et al., 29 Apr 2025). This suggests that XSPECT can be defined by the joint use of spectral and temporal structure even when the separation is performed algorithmically rather than instrumentally.

5. Scientific domains and demonstrated applications

Mission concepts place black-hole accretion, neutron-star dense matter, and transient astrophysics at the center of XSPECT. STROBE-X states that pulse-profile modeling for ≈20 pulsars and burst oscillation sources can yield ≈5% $^{-1}$2–$^{-1}$3 constraints per source, while LAD+XRCA can measure reverberation lags in X-ray binaries in 1–10 Hz bands with ≈20 μs precision per bin in 1 ks and enable AGN reverberation mapping for a 2 mCrab AGN in 100 ks (Ray et al., 2018). LOFT simulations similarly emphasize rapid QPO and Fe-K studies, including the detection of twin QPOs at 188 and 268 Hz in 1 ks at 1 Crab where RXTE/PCA required 54 ks for a ~3–4σ detection (Belloni et al., 2012).

HFQPO studies are a canonical XSPECT application because they require both large count rates and phase-resolved spectroscopy. For the 166 Hz QPO in GRS 1915+105, GR hotspot calculations predict that the hardness peak trails the flux peak by ~0.2 in phase. A high count-rate mission like LOFT would make it possible to get a QPO phase for each photon, enabling QPO-phase-resolved spectral analysis and, in simultaneous campaigns, phase-resolved polarimetry (Beheshtipour et al., 2016).

Operational studies show the same logic at work in source-specific analyses. Broadband timing and spectroscopy of MAXI J1816−195 measured a decay-phase spin-up of $^{-1}$4 and, from the combination of $^{-1}$5 and contemporaneous flux, inferred $^{-1}$6 (Li et al., 2023). In SMC X-2, joint NuSTAR+NICER XSPECT resolved a cyclotron line at 31 keV, found bulk Comptonization to dominate thermal Comptonization, and linked a moderate anti-correlation of cyclotron energy with flux to super-critical accretion (Tobrej et al., 7 Jan 2025).

Phase-resolved XSPECT can also isolate short-lived structures that disappear in averaged spectra. In X Persei, spectra were produced in four phase bins per 837 s pulse, at intervals as short as 210 s, yielding ~1200 spectra. Transient Fe K$^{-1}$7 emission was detected in only ~8–9% of phase-resolved spectra, and the combined timing–spectral analysis was used to infer a dense, compact Be disk with $^{-1}$8–3.3 and $^{-1}$9 (Sanjurjo-Ferrín et al., 8 Jun 2026).

On XPoSat, XSPECT has already been applied to thermonuclear-burst phenomenology. In 4U 1608-52, it tracked a superburst with a rise of ~13.6 minutes, exponential decay time $>2000$0 h, and persistent-emission suppression characterized by $>2000$1 throughout; in both 4U 1608-52 and Aql X-1, persistent spectra were fitted with either blackbody+disk blackbody or an optically thick Comptonized disk, and the Comptonized model showed a harder-when-brighter trend (Chatterjee et al., 15 May 2026). In Sco X-1, XSPECT resolved the Z-track in the 0.8–15 keV band and showed that the Z-track is driven by changes in the optical depth of the corona, the Comptonization flux, the disk flux, and the inner disk temperature, while detecting no QPOs in any branch (Prakash et al., 3 Oct 2025).

6. Calibration, operations, and limitations

XSPECT depends on calibrated event-mode data products. STROBE-X is designed to downlink “all events all the time” from XRCA, LAD, and WFM, at ≈300 Mbps and ≈540 Gb/day average, with low-Earth orbit operations at ≈550 km and ≈10° inclination, slewing up to ≈15° min$>2000$2, and S-band transient alerts in <10 s (Ray et al., 2018). XRISM demonstrates the same principle in operational form: event-level FITS timing products are reconstructed from spacecraft time plus LOCAL_TIME and then barycenter-corrected with HEASoft tools (Terada et al., 18 Mar 2025).

XPoSat XSPECT provides a fully specified Level-0 to Level-2 chain. Raw payload data are separated into events and housekeeping, instrument-frame times are transformed to UTC, temperature-dependent gains are applied, standard GTIs are generated, and single-event light curves and spectra are produced per FOV. Blank-sky data exceeding 1 Ms support “static” background files, while a “hybrid” method uses the saturation count rate to scale the Galactic Cosmic Ray component during long observations. With average background levels, the 5σ sensitivity is ~0.6 mCrab in 10 ks for 15 detectors (Chatterjee et al., 11 Jun 2025).

Calibration caveats are mission-specific but recurrent. XSPECT on XPoSat requires explicit treatment of residuals in the ~1.2–1.8 keV range for very bright sources; the calibration paper recommends adding ~3% systematics in this band rather than introducing ad hoc Gaussian components (Chatterjee et al., 11 Jun 2025). In the Aql X-1 and 4U 1608-52 campaign, spectra were restricted to 0.8–11 keV because source spectra were background-dominated above ~11 keV, and eclipse-only observing produced gaps of ~60–70 minutes between on-source intervals (Chatterjee et al., 15 May 2026). XRISM, although meeting its timing budget, retains a practical phase-resolution floor from LOCAL_TIME quantization, budgeted at 132 μs for Hp/Mp events and increasing to 242 μs for Ls events (Terada et al., 18 Mar 2025).

A persistent misconception is terminological: XSPECT denotes X-ray spectroscopy and timing science, whereas XSPEC is the spectral-fitting package commonly used to analyze XSPECT data (Belloni et al., 2012). In practice, the two are closely linked: XSPECT missions and analyses produce event lists, calibrated spectra, light curves, and response matrices, and these products are then modeled with established timing and spectral software environments. The resulting framework is not a single instrument class but a mature observational program in which timing precision, spectral response, and count-rate capability are designed and interpreted jointly.