Physics of 1 keV line in X-ray binaries

X-ray binaries (XRBs) often exhibit spectral residuals in the 0.5 to 2 keV range, known as the ``1 keV residual/1 keV feature", with variable centroid and intensity across different systems. Yet a comprehensive scientific explanation of the variability of the 1 keV feature has remained largely elusive. In this paper, we explain for the first time the origin and variability of the 1 keV feature in XRBs using the spectral synthesis code \textsc{Cloudy}. We constructed line blends for the emission and absorption lines and study the variability of these blends with ionization parameters, temperature, and column density. We conducted a sample study involving five XRBs including two ultraluminous X-ray sources (ULXs): NGC 247 ULX-1, NGC 1313 X-1, a binary X-ray pulsar : Hercules X-1, and two typical low-mass X-ray binaries (LMXBs): Cygnus X-2, and Serpens X-1. Our analysis establishes a self-consistent framework explaining the variability of the 1 keV spectral feature, attributing its diversity to differences in spectral energy distribution, ionization parameter, temperature, column density, and disk reflection properties. This framework provides a comprehensive explanation for the observed 1 keV feature across these diverse XRB systems, offering insights into the underlying physical mechanisms at play.

💡 Research Summary

The authors address a long‑standing puzzle in X‑ray binary (XRB) spectroscopy: the ubiquitous “1 keV residual” that appears as a broad excess or dip between roughly 0.5 keV and 2 keV. While this feature has been reported in a wide variety of systems—from ultraluminous X‑ray sources (ULXs) to accreting pulsars and low‑mass X‑ray binaries—the centroid energy and strength have been observed to vary both between sources and over time within a single source, and a unified physical explanation has been lacking.

To solve this problem the paper combines high‑resolution X‑ray observations with detailed photo‑ionization and collisional plasma modeling using the spectral synthesis code Cloudy. Five representative objects are selected: the ULXs NGC 1313 X‑1 and NGC 247 ULX‑1, the pulsar Hercules X‑1, and the LMXBs Cygnus X‑2 and Serpens X‑1. For each source the authors reduce XMM‑Newton RGS or NICER data, construct source‑specific spectral energy distributions (SEDs) and adopt hydrogen densities appropriate for the environment (10¹⁰ cm⁻³ for ULXs, 10¹⁵ cm⁻³ for LMXBs).

A key methodological innovation is the definition of line blends rather than fitting individual Gaussian components. In Cloudy the total intensity of all emission lines in a given energy interval is summed to form an “Em blend” (0.6–1.4 keV). This blend is further split into “Em left” (0.6–1.0 keV) and “Em right” (1.0–1.4 keV) to track centroid shifts. Analogous absorption blends (Abs left, Abs right, Abs blend) and reflection blends (Reflect left, Reflect right, Reflect blend) are also constructed for the 0.5–2.0 keV band. By varying the ionization parameter (ξ), hydrogen column density (N_H), and temperature (for collisional models), the authors map how the relative strengths of these blends change.

The results show that photo‑ionized equilibrium (PIE) models naturally reproduce the observed variability: increasing ξ moves the dominant emission from the lower‑energy side to the higher‑energy side of the blend, shifting the apparent centroid toward ~1.1 keV, while higher N_H enhances both left‑ and right‑hand absorption, creating the symmetric dips often seen in ULXs. Collisional ionization equilibrium (CIE) models, relevant for hotter plasma (10⁶–10⁷ K), also generate a right‑hand dominated blend because Fe L‑shell and Ne X lines become stronger at higher temperatures. Reflection from a dense accretion‑disk surface (n > 10¹⁵ cm⁻³) adds a low‑energy component (Fe L, O VIII) that boosts Em left, explaining why some LMXBs (e.g., Serpens X‑1) show a centroid near 0.8 keV.

System‑by‑system fits yield the following best‑fit parameters:

• NGC 1313 X‑1: ξ ≈ 10³ erg cm s⁻¹, N_H ≈ 10²³ cm⁻², Em right dominant → centroid ≈ 1.1 keV.
• NGC 247 ULX‑1: ξ varies between 10² and 10⁴, producing time‑dependent shifts of the centroid between 0.9 and 1.2 keV, matching the observed variability.
• Hercules X‑1: strong magnetic field and disk precession enhance the reflection blend; a modest Abs right component creates a shallow dip just above 1 keV.
• Cygnus X‑2: high column (≈ 10²⁴ cm⁻²) and moderate ξ (≈ 10²) lead to a pure PIE emission blend with little reflection, reproducing the strong, relatively narrow 1 keV excess seen in NICER data.
• Serpens X‑1: low N_H but high disk density yields a prominent reflection blend, shifting the centroid to ≈ 0.8 keV.

Overall, the study demonstrates that the “1 keV feature” is not a single atomic line but a composite of many emission, absorption, and reflected lines whose relative contributions are governed by the source’s SED, ionization state, column density, temperature, and disk geometry. The line‑blend approach preserves the intrinsic line profile, avoids the oversimplifications of Gaussian fitting, and provides a physically motivated framework that can be directly applied to upcoming high‑resolution missions such as XRISM and Athena. The authors conclude that this unified model explains the diversity of the 1 keV residual across a broad class of X‑ray binaries and offers a robust tool for future spectroscopic diagnostics.