X-ray Polarimetry-Timing

• X-ray polarimetry-timing is the analysis of rapid changes in linear polarization degree and angle, adding geometric insights to timing and spectroscopy.
• It employs phase-resolved and Fourier-domain techniques on event-driven data from instruments like Gas Pixel Detectors to capture dynamic polarization signals.
• Recent missions such as IXPE and eXTP demonstrate its power to constrain models of pulsars, magnetars, and black-hole systems through combined timing and polarimetric studies.

X-ray polarimetry-timing is the characterization of rapid variability in the linear polarization degree $p$ and polarization angle $\psi$ of X-ray emission from compact objects. It extends standard polarimetry and spectral-timing by tracking how the Stokes vector changes with time, spin phase, orbital phase, or Fourier frequency, thereby adding geometric and causal information to spectroscopy and flux variability. In its modern form, the field is built around event-driven X-ray polarimeters—especially photoelectric Gas Pixel Detectors (GPDs)—that measure event time, energy, position, and photoelectron azimuth, and around simultaneous timing/spectroscopy instruments that provide high-S/N reference light curves and ephemerides (Ingram, 2022, Bellazzini et al., 2010, Zhang et al., 2018).

1. Scope and historical development

Historically, X-ray polarimetry was limited by instrumental sensitivity. Earlier measurements relied on Bragg diffraction at $45^\circ$ and Compton scattering at $90^\circ$, and the only unambiguous early detection was the Crab Nebula. Those approaches were constrained by narrow energy bands, higher thresholds, low sensitivity for most sources, and, in many cases, a requirement for rotation around the beam axis. The development of imaging photoelectric polarimeters changed this situation by allowing the photoelectron track itself to encode polarization, with simultaneous imaging, moderate spectroscopy, and high-rate timing in a non-dispersive detector that does not require rotation (Bellazzini et al., 2010).

This instrumental shift is what made polarimetry-timing operationally meaningful. IXPE was designed to perform imaging, timing, and energy-resolved polarimetry in the 2–8 keV band, and the launch of IXPE marked the point at which the field became observational rather than primarily predictive (Soffitta et al., 2021). The same period also produced mission concepts explicitly organized around simultaneous timing and polarimetry. The enhanced X-ray Timing and Polarimetry mission, eXTP, was conceived to deliver simultaneous high-throughput timing, spectroscopy, and X-ray polarimetry for dense matter, strong-field gravity, and QED in extreme magnetic fields (Zhang et al., 2018). The X-ray Polarization Probe concept extended the same logic to true broadband spectro-polarimetry across 0.2–60 keV with microsecond event tagging (Jahoda et al., 2019).

The conceptual motivation is straightforward: spectral-timing alone often leaves geometric degeneracies unresolved. Polarimetry-timing adds observables that respond directly to scattering geometry, magnetic topology, and relativistic transport. In black-hole X-ray binaries this targets precessing coronae, reflection, and reverberation; in pulsars and magnetars it targets rotating-vector behavior, magnetospheric scattering, and vacuum birefringence; in jets it targets highly polarized synchrotron components whose flux may be subdominant while their polarization signature is not (Ingram, 2022).

2. Polarimetric observables and statistical foundations

For photoelectric polarimeters, the measured azimuthal distribution of reconstructed event angles is the basic observable. A common form is

$N(\phi)=A\left[1+\mu\,p\,\cos 2(\phi-\psi)\right],$

where $\mu$ is the modulation factor for a 100% polarized beam, $p$ is polarization degree, and $\psi$ is polarization angle (Soffitta et al., 2021). In the eXTP PFA description, the fully polarized response is also written as

$N(\phi)=A+B\cos^2(\phi-\phi_0),$

with

$\mu=\frac{\max-\min}{\max+\min}=\frac{B}{2A+B}.$

Measured PFA modulation factors include 38% at 3 keV and 57% at 6 keV, with laboratory modulation $\psi$0 at 6.14 keV (Zhang et al., 2018).

The event-based Stokes formalism provides the standard estimator set. With event azimuths $\psi$1 and optional weights $\psi$2,

$\psi$3

The corresponding polarization estimators are

$\psi$4

with instrument response handled through $\psi$5 calibration and, where needed, energy-dependent weighting (Zhang et al., 2018). IXPE uses the same Stokes structure but with pipeline-specific calibration corrections, including pixel equalization, gain uniformity and charging corrections, temperature and GEM gain corrections, and event-by-event spurious modulation subtraction in Stokes space (Soffitta et al., 2021).

Sensitivity is usually summarized by the Minimum Detectable Polarization. In rate form, the standard 99% confidence expression is

$\psi$6

where $\psi$7 and $\psi$8 are source and background count rates and $\psi$9 is exposure time (Zhang et al., 2018). In the background-negligible limit,

$45^\circ$0

showing the familiar dependence on modulation factor, effective area, source flux, and exposure (Zhang et al., 2018). For a Crab-like spectrum in the eXTP PFA 2–8 keV band, the spectrum-weighted mean modulation is $45^\circ$1, and $45^\circ$2 ks yields $45^\circ$3 at 99% confidence (Zhang et al., 2018).

Uncertainty propagation also motivates the timing strategy. In the GPD literature, a background-free sample with $45^\circ$4 events gives

$45^\circ$5

so finer time or phase binning immediately drives up the required count rate (Bellazzini et al., 2010). This is why direct $45^\circ$6 and $45^\circ$7 light curves are practical only for bright sources or long bins, and why Fourier-domain methods became central for stochastic variability (Ingram, 2022).

3. Instrument platforms and mission architectures

The current observational baseline is IXPE. It uses three co-aligned 4 m focal-length mirror modules, each focusing onto one GPD-based Detector Unit, and delivers imaging X-ray polarimetry in 2–8 keV. Its timing system is built around GPS-derived PPS and on-board time counters, with 1 μs timing resolution and 1–2 μs accuracy. Instrument-level dead time is 1.1 ms at 2.69 keV and 1.2 ms at 6.4 keV, and the observed Crab rate is 150 c/s in 2–8 keV. Calibrated modulation factors are $45^\circ$8 at 2.69 keV and $45^\circ$9 at 6.40 keV, while instrument-level spurious modulation is $90^\circ$0 at $90^\circ$1 keV and $90^\circ$2 at 5.89 keV (Soffitta et al., 2021).

eXTP generalizes this into a simultaneous timing–spectroscopy–polarimetry observatory. In the 2018 mission overview, the payload comprised LAD, SFA, PFA, and WFM, with LAD and SFA both providing 10 μs time resolution and PFA providing imaging polarimetry in 2–8 keV (Zhang et al., 2018). The LAD was specified with 3.4 m² effective area at 8 keV, $90^\circ$3 counts s$90^\circ$4 for a Crab-like source, dead time $90^\circ$5 at 1 Crab, and background systematics constrained to $90^\circ$6 over ks timescales (Zhang et al., 2018). The PFA was described as four identical focusing telescopes with a total effective area of 915 cm² at 2 keV, 495 cm² at 3 keV, 216 cm² at 4 keV, and 46 cm² at 6 keV, using GPDs with total 2–8 keV background $90^\circ$7 counts s$90^\circ$8 in the source aperture (Zhang et al., 2018). The 2025 approved baseline preserves the same scientific logic, but specifies an SFA/PFA/W2C architecture in which SFA-T provides $90^\circ$9 μs resolution and $N(\phi)=A\left[1+\mu\,p\,\cos 2(\phi-\psi)\right],$0 μs absolute timing accuracy, while PFA specifies $N(\phi)=A\left[1+\mu\,p\,\cos 2(\phi-\psi)\right],$1 μs event time resolution, dead time $N(\phi)=A\left[1+\mu\,p\,\cos 2(\phi-\psi)\right],$2 at 1 Crab, and $N(\phi)=A\left[1+\mu\,p\,\cos 2(\phi-\psi)\right],$3 for a 1 mCrab source in $N(\phi)=A\left[1+\mu\,p\,\cos 2(\phi-\psi)\right],$4 s (Zhang et al., 9 Jun 2025).

XPP pushes the architecture toward simultaneous broadband component separation. Its concept uses three grazing-incidence telescopes and simultaneous LEP, MEP, and HEP polarimeters spanning 0.2–60 keV, with $N(\phi)=A\left[1+\mu\,p\,\cos 2(\phi-\psi)\right],$5 μs event time-tagging and $N(\phi)=A\left[1+\mu\,p\,\cos 2(\phi-\psi)\right],$6 energy resolution “at all wavelengths.” It aims for $N(\phi)=A\left[1+\mu\,p\,\cos 2(\phi-\psi)\right],$7 MDP in $N(\phi)=A\left[1+\mu\,p\,\cos 2(\phi-\psi)\right],$8 s for a 1 mCrab source, and explicitly targets time-resolved polarimetry from milliseconds to hours (Jahoda et al., 2019).

Detector electronics are now a limiting element rather than a conceptual one. The XPOL-III CMOS ASIC was developed for next-generation GPD throughput, reducing per-event dead time from $N(\phi)=A\left[1+\mu\,p\,\cos 2(\phi-\psi)\right],$9 ms in XPOL-I to $\mu$0 μs at $\mu$1 keV under nominal test conditions, with $\mu$2 at 5.2 keV and spurious modulation $\mu$3 at 5.9 keV. This directly improves phase-resolved polarimetry, transient response, and Fourier timing by lowering pile-up and dead-time distortions (Minuti et al., 2022).

4. Timing-domain methodologies

The most direct form of polarimetry-timing is phase-resolved analysis. For a time bin or phase bin $\mu$4,

$\mu$5

with

$\mu$6

This is well matched to coherent pulsations, burst oscillations, and orbital phase studies (Bellazzini et al., 2010). IXPE and eXTP both exploit this regime, and the PFA’s sub-ms or microsecond event tagging was explicitly described as enabling phase-resolved Stokes analysis across pulsar cycles or QPO phases (Zhang et al., 2018).

Stochastic variability requires a different treatment. Direct $\mu$7 and $\mu$8 light curves become statistically unstable in short bins, and phase-folding is inappropriate for signals whose phase wanders. The Fourier approach introduced for fast stochastic X-ray polarimetry-timing solves this by treating polarization variability exactly as a cross-spectral problem. In the modulation-angle formulation,

$\mu$9

and cross-correlating with a reference light curve $p$0 gives

$p$1

The real and imaginary parts versus $p$2 recover the Fourier-domain polarization content at each frequency (Ingram, 2022).

The original formulation argued that this method should permit detection of quasi-periodic swings in polarization angle predicted by Lense–Thirring precession, provided the mean polarization degree is greater than $p$3–$p$4 (Ingram et al., 2017). A subsequent implementation on real IXPE data introduced event-level $p$5, $p$6, and $p$7 time series, detector-to-sky angle recovery, per-event spurious-polarization correction, and dead-time mitigation through independent detector references and co-spectra. On RX J0440.9+4431 and Her X-1, the technique recovered the known polarization variability signal already seen in phase-folding analyses, thereby verifying that Fourier-domain polarimetry-timing can be applied to real IXPE observations (Ewing et al., 21 Jul 2025).

The practical workflow is now relatively well defined: barycentric correction, GTI filtering, dead-time accounting, event-angle calibration, construction of $p$8, $p$9, and $\psi$0 time series or modulation-angle-selected light curves, Fourier transforms per segment, and cross-spectral averaging across segments or narrow frequency bands (Ingram, 2022). Dead time remains a critical issue. For IXPE, a recommended mitigation is to use co-spectra between independent detector modules and to form subject/reference bands across different GPDs, which suppresses dead-time-correlated noise and artificial lags (Ingram, 2022, Ewing et al., 21 Jul 2025).

5. Astrophysical applications and empirical demonstrations

The scientific scope is broad because polarization variability is geometry-sensitive in regimes where intensity alone is not. For neutron stars, pulse-profile modeling combined with phase-resolved polarimetry can break degeneracies in hotspot latitude, beaming, and line-of-sight inclination, and was explicitly highlighted in the eXTP science case for equation-of-state inference (Zhang et al., 2018). For magnetars, vacuum birefringence predicts high, phase-dependent polarization with energy-dependent mode conversion, making $\psi$1 and $\psi$2 direct tests of QED in $\psi$3–$\psi$4 G fields (Zhang et al., 2018). For black-hole systems, precessing inner flows, reflection, and reverberation are expected to imprint QPO-phase modulation of $\psi$5 and weaker but detectable modulation of $\psi$6 (Ingram, 2022).

Observationally, IXPE has already shown that phase-resolved polarimetry can constrain compact-object geometry in detail. In EXO 2030+375, phase-averaged analysis returned a low polarization degree of 0%–3%, while phase-resolved analysis showed variation in the range 2%–7%. Fitting the rotating vector model gave a magnetic obliquity of $\psi$7 and a pulsar inclination of $\psi$8, leading to an interpretation in which the magnetic axis swings close to the observer line of sight and the observed behavior is shaped by complex accreting geometry, magnetic multipoles with asymmetric topology, and gravitational light bending (Malacaria et al., 2023).

Orbital-phase polarimetry extends the same logic to X-ray binaries. In GS 1826-238, phase-resolved IXPE polarimetry was modeled as scattering of a largely unpolarized compact-source beam off the companion star. Joint fitting of the phase-dependent Stokes curves yielded $\psi$9, $N(\phi)=A+B\cos^2(\phi-\phi_0),$0, $N(\phi)=A+B\cos^2(\phi-\phi_0),$1, and $N(\phi)=A+B\cos^2(\phi-\phi_0),$2, while also showing that inclination recovery becomes biased near a critical inclination of $N(\phi)=A+B\cos^2(\phi-\phi_0),$3. The analysis demonstrated that orbital scattering can, in principle, constrain inclination even when optical methods are difficult, although the present dataset remained limited by the small scattered fraction and weak modulation (Rankin et al., 2023).

In black-hole X-ray binaries, the joint interpretation of timing, spectroscopy, and polarimetry is already producing state-dependent geometric diagnostics. A quasi-simultaneous IXPE+NICER+NuSTAR+AstroSat study of eleven systems reported significant polarization degrees in the 2–8 keV band for multiple hard and intermediate states and found a positive correlation between polarization degree and the Comptonized photon fraction $N(\phi)=A+B\cos^2(\phi-\phi_0),$4, together with an anti-correlation with the disc-to-Comptonized flux ratio $N(\phi)=A+B\cos^2(\phi-\phi_0),$5. Type-C QPOs coincided with large $N(\phi)=A+B\cos^2(\phi-\phi_0),$6, hard spectra, and higher polarization degree, while softer, disc-dominated states showed weak or absent polarization (Majumder et al., 4 Jun 2025). This strongly suggests that, in those sources, the polarized component is controlled by Comptonizing geometry rather than by the thermal disc alone.

6. Systematics, performance boundaries, and future directions

Polarimetry-timing is limited as much by systematics as by photon statistics. At low true polarization, the measured $N(\phi)=A+B\cos^2(\phi-\phi_0),$7 is Rice-biased, short bins are non-Gaussian, and direct $N(\phi)=A+B\cos^2(\phi-\phi_0),$8 or $N(\phi)=A+B\cos^2(\phi-\phi_0),$9 light curves become unstable; this is why current best practice is to work with $\mu=\frac{\max-\min}{\max+\min}=\frac{B}{2A+B}.$0 and $\mu=\frac{\max-\min}{\max+\min}=\frac{B}{2A+B}.$1 or with modulation-angle cross-spectra rather than with naive short-bin polarization curves (Ingram, 2022). Energy dependence in $\mu=\frac{\max-\min}{\max+\min}=\frac{B}{2A+B}.$2 must be calibrated and either folded into weights or handled through response matrices. Background is usually effectively unpolarized for focused imaging instruments, but it still dilutes $\mu=\frac{\max-\min}{\max+\min}=\frac{B}{2A+B}.$3 if source flux varies (Ingram, 2022).

Instrumental modulation floors remain fundamental. The GPD literature reported residual modulation for an unpolarized Fe-55 source of $\mu=\frac{\max-\min}{\max+\min}=\frac{B}{2A+B}.$4 at 5.9 keV, with intrinsic symmetry keeping systematics $\mu=\frac{\max-\min}{\max+\min}=\frac{B}{2A+B}.$5 (Bellazzini et al., 2010). IXPE ground calibration measured instrument-level spurious modulation of $\mu=\frac{\max-\min}{\max+\min}=\frac{B}{2A+B}.$6 at $\mu=\frac{\max-\min}{\max+\min}=\frac{B}{2A+B}.$7 keV and $\mu=\frac{\max-\min}{\max+\min}=\frac{B}{2A+B}.$8 at 5.89 keV, together with polarization-angle systematics below a degree (Soffitta et al., 2021). For the eXTP PFA, systematic residual modulation for unpolarized sources is described as controllable below $\mu=\frac{\max-\min}{\max+\min}=\frac{B}{2A+B}.$9, but the same mission documentation also emphasizes that such a residual floor sets a hard limit on the detection of very low polarization degrees regardless of flux or exposure (Zhang et al., 2018).

The technology trajectory is nevertheless favorable. XPOL-III shows that next-generation GPD systems can preserve polarimetric, spectral, imaging, and timing capability while reducing event dead time by at least a factor of seven, which is directly relevant for bright-source phase-resolved work and for preserving Fourier power at high frequencies (Minuti et al., 2022). At higher energies, a wide-field triple-GEM time projection chamber with optical readout has already demonstrated reconstructed electrons in the 10–60 keV range, angular resolutions as good as $\psi$00, and inferred modulation factors up to 0.9, indicating that photoelectric-effect polarimetry is not restricted to the soft band (Fiorina et al., 30 Oct 2025).

The mission outlook is correspondingly expansive. eXTP’s approved baseline couples microsecond timing, focusing polarimetry, and rapid transient triggering in a single observatory, while XPP proposes simultaneous LEP/MEP/HEP coverage from 0.2 to 60 keV with $\psi$01 μs time-tagging (Zhang et al., 9 Jun 2025, Jahoda et al., 2019). The successful implementation of Fourier-domain stochastic polarimetry-timing on real IXPE data suggests that these larger-area missions will not merely improve phase-resolved polarimetry for pulsars and orbital studies, but should also make QPO polarimetry, polarization-resolved propagation lags, and polarized reverberation mapping observationally routine (Ewing et al., 21 Jul 2025).