IXPE Observations of Cygnus X-2

• The paper presents the first definitive X-ray polarization measurement from a weakly magnetized neutron star in Cygnus X-2, emphasizing the dominant role of a highly polarized reflection component.
• Multi-instrument observations (IXPE, NICER, NuSTAR, INTEGRAL) and full Stokes analysis were employed to constrain the geometry of the accretion flows through broadband spectral modeling.
• Broadband spectral fits quantified flux fractions as ~74% disk, 16% Comptonized, and 10% reflection, with the polarization angle aligning with the radio jet to underscore the reflection signature.

The Imaging X-ray Polarimetry Explorer (IXPE) observation of Cygnus X-2 provides the first definitive measurement of X-ray polarization from a weakly magnetized neutron-star low-mass X-ray binary (LMXB). Cygnus X-2, classified as a Z-source, was targeted in the soft X-ray band during a coordinated, multi-instrument campaign including IXPE, NICER, INTEGRAL, and NuSTAR. These observations address fundamental questions about the geometry of accretion flows and the dominant emission mechanisms near neutron stars by leveraging polarimetric sensitivity and broad-band spectral coverage (Farinelli et al., 2022, Liu et al., 4 Jan 2026).

1. Observational Campaign and Data Processing

The IXPE campaign observed Cygnus X-2 in the 2–8 keV band with a net exposure of 93.2 ks, accompanied quasi-simultaneously by NuSTAR (3–40 keV), NICER (1–10 keV), and INTEGRAL (JEM-X: 3–30 keV, ISGRI: 25–200 keV). Data recorded included the full Stokes I, Q, U parameters for polarization, and comprehensive spectral data enabling model decomposition (Farinelli et al., 2022, Liu et al., 4 Jan 2026).

• IXPE data were processed with ixpeobssim and calibrated using CALDB releases. Events were extracted from a 100″ radius region; background was negligible (<0.2%).
• NICER spectra were processed above 1.5 keV to exclude mirror-coating residuals.
• NuSTAR analysis employed standard bright-source filtering.
• INTEGRAL data used established spectral extraction frameworks.

Diagnostic diagrams (hardness–intensity and color–color) and the detection of a 7 Hz quasi-periodic oscillation confirmed the source was in the Normal Branch of the Z track during all observations (Liu et al., 4 Jan 2026).

2. Broadband Spectral Modeling

Simultaneous spectral fitting utilized both phenomenological and physically motivated models:

• Model Components:
  • tbabs (interstellar absorption),
  • diskbb (multi-color blackbody disk; $kT_{\text{in}} \approx 1$ keV),
  • comptt (unsaturated Comptonization, $kT_{\text{e}}=3$ keV, $\tau \approx 4$),
  • Gaussian/excess for Fe K$\alpha$ line ($E_{\text{ga}} = 6.67 \pm 0.05$ keV, $\sigma_{\text{ga}} = 0.29^{+0.08}_{-0.06}$ keV, EW $\approx 40 \pm 10$ eV).
• Relativistic Reflection:
  • Subsequent analyses implemented relativistically blurred reflection models (relconv*reflionx_bb, or relxillNS) to self-consistently model both the Fe K$\alpha$ complex and the Compton hump. Reflection fraction in the IXPE band was constrained as $R = F_{\text{refl}}/F_{\text{total}} \approx 0.10$, indicating the reflected emission comprises $\sim$10% of the total 2–8 keV photon flux (Liu et al., 4 Jan 2026).

Model fits quantified the fractional contributions as $\sim$74% (disk), 16% (Comptonized), and 10% (reflection) in the soft X-ray band, with parameter values consistent across both classic and relativistic model frameworks.

3. Polarization Measurement and Stokes Analysis

X-ray polarization was derived via model-independent PCUBE analysis and through parameterized spectro-polarimetric fits:

• Net Polarization:
  • PCUBE analysis: $P = 1.4 \pm 0.3$\%, $PA = 132^\circ \pm 6^\circ$ (1$\sigma$).
  • XSPEC fit (polconst*tbabs*(diskbb+comptt+gaussian)): $P = 1.8 \pm 0.3$\%, $\phi = 140^\circ \pm 4^\circ$.
  • Both values align closely with the radio jet position angle ($\sim$140°).
• Component-wise Polarization:
  • The Comptonized emission (comptt) was previously inferred to show $P_{\text{comptt}} = 4.0^{+2.0}_{-1.7}\%$, $\phi_{\text{comptt}} = 133^\circ \pm 11^\circ$ (Farinelli et al., 2022).
  • However, with increased spectral model fidelity and independent flux decomposition, recent work finds that for $f_{\text{refl}} \sim 0.10$ and $P_{\text{comptt}} \lesssim 0.5\%$, the intrinsic reflection polarization must be $P_{\text{refl}} \approx 13$–30\% (Liu et al., 4 Jan 2026).

$$P_{\text{tot}} = \frac{f_{\text{refl}} P_{\text{refl}} + f_{\text{comp}} P_{\text{comp}}}{f_{\text{refl}} + f_{\text{comp}}}$$

For $P_{\text{comp}} \simeq 0$, this implies $$P_{\text{refl}} \approx P_{\text{tot}} / f_{\text{refl}} \approx 14$$–20\%.

4. Physical Interpretation and Accretion Geometry

The polarization properties decisively constrain the geometry of the X-ray-emitting regions in Cygnus X-2:

• In a canonical scattering-dominated, optically thick disk atmosphere at $i \sim 60^\circ$, theoretical expectations are for a few-percent polarization perpendicular to the jet. Instead, IXPE finds a polarization degree of $1.4$–$1.8$\% with a polarization angle aligned with the known radio jet, inconsistent with a disk origin as the dominant polarized source (Farinelli et al., 2022, Liu et al., 4 Jan 2026).
• The observed polarization axis and amplitude for the total emission can be matched if the reflected component from the inner accretion disk is highly polarized ($\sim$20\%) and constitutes about 10% of the observed flux. The Comptonization region, likely a boundary layer or spreading layer, is predicted to produce $P \lesssim 0.5$\%, contributing negligibly to the net observed signal (Liu et al., 4 Jan 2026).
• Reflection models imply disk inclination angles dependent on density and ionization: standard models suggest $i < 10^\circ$, but higher-density/NS-optimized models find $i \approx 20^\circ$–$40^\circ$, reducing tension with orbital inclinations and observed X-ray dips.

The low to moderate disk inclination required by the observed degree of polarization is consistent with that inferred from other observables, and the polarization orientation traces the system’s jet axis.

5. Role of Reflection and Component Degeneracy

A critical result emerging from these campaigns is the degeneracy between the Comptonized and reflected spectral components in both flux and polarization:

Spectral Component	Flux Fraction (2–8 keV)	Expected $P$
Disk Blackbody	$\sim$74%	$\lesssim$3.3% upper limit (Farinelli et al., 2022)
Comptonized (comptt)	$\sim$16%	$\lesssim$0.5% (Liu et al., 4 Jan 2026); up to 4% if prior models used
Reflection	$\sim$10%	13–30% (Liu et al., 4 Jan 2026)

• The observed $P_{\text{tot}} \sim 1.4$–$1.8$\% cannot be explained by boundary layer or disk wind scattering models alone, which produce $P \lesssim 2\%$.
• The requirement for a highly polarized reflection component is robust even accounting for model systematics and degeneracies. The disk-reflected fraction, though modest in terms of flux, dominates the observed polarization.

A plausible implication is that reflection is essential to the X-ray polarization of neutron star LMXBs in high-luminosity, soft states.

6. Broader Context and Future Directions

IXPE’s observations of Cygnus X-2 mark the transition to polarimetry-informed X-ray accretion geometry studies:

• The data confirm the long-suspected importance of boundary/spreading layers but highlight the necessity of accounting for reflected emission when interpreting polarization signatures.
• The geometric alignment of polarization with the radio jet—present in both IXPE and historical OSO-8 results—provides a robust tracer of system orientation and helps connect accretion and outflow phenomena across wavelengths.
• Outstanding questions include the detailed energy dependence of polarization from reflection, phase/branch variability, and the universality of reflection-dominated polarization signals in LMXBs.
• Future campaigns targeting other Z and atoll sources, as well as time-resolved and energy-resolved spectro-polarimetry, are essential to fully map the population-level properties of accreting neutron star systems and to validate/refine the reflection-dominated emission models (Farinelli et al., 2022, Liu et al., 4 Jan 2026).