Supergiant X-ray Binaries (SgXBs)

• Supergiant X-ray binaries are systems where a neutron star accretes from the dense, line-driven wind of an evolved OB supergiant, producing persistent or transient X-ray emission.
• Observations and simulations reveal a continuum between classical systems and supergiant fast X-ray transients, with variations driven by wind structure, orbital geometry, and accretion regimes.
• Multiwavelength diagnostics and hydrodynamic models show that factors like wind clumping, magnetospheric gating, and local absorption critically modulate luminosity and X-ray variability.

Supergiant X-ray binaries (SgXBs; often written SGXBs in the literature) are high-mass X-ray binaries in which a compact object, usually a neutron star, accretes from the line-driven wind of an evolved O/B supergiant companion. In the classical form, the X-ray output is comparatively persistent and wind-fed; in the transient form represented by supergiant fast X-ray transients (SFXTs), the same basic stellar configuration produces much lower average luminosity, much larger dynamic range, and short bright flares. Contemporary work increasingly treats SgXBs and SFXTs not as completely disjoint families but as a coupled phenomenology controlled by donor-wind structure, orbital geometry, local absorption, neutron-star spin and magnetic field, and transitions between accretion regimes (Bodaghee et al., 2011, Bozzo et al., 2014).

1. Definition, discovery, and class structure

SgXBs are one of the three principal Galactic HMXB subclasses alongside Be/X-ray binaries and SFXTs. Their defining stellar configuration is a compact object embedded in the dense wind of an evolved OB supergiant. Relative to Be/X-ray binaries, which are associated with main-sequence Be donors and typically transient outbursts near periastron, classical SgXBs are characterized by steady wind-fed accretion and generally persistent X-ray emission, with variability driven by inhomogeneities in the donor wind rather than only by orbital triggering (Bodaghee et al., 2011).

The modern prominence of SgXBs is an INTEGRAL-era result. In the pre-INTEGRAL 2000 catalog there were 54 BeHMXBs and 7 sgHMXBs; by the 2006 catalog, 32% of firmly identified HMXBs were sgHMXBs, and the sgHMXB/HMXB ratio had increased by about a factor of 6. INTEGRAL also effectively tripled the total number of Galactic sgHMXBs and revealed two especially important observational groupings within the supergiant population: heavily obscured systems and SFXTs (Chaty, 2010).

This observational expansion immediately complicated any rigid taxonomy. Classical persistent systems typically vary by factors of order $10$–50 or $\sim 20$–50, whereas SFXTs show flares lasting a few hours, dynamic ranges of $\sim 10^3$–$10^5$, hard-X-ray duty cycles of $0.05\%$ to $3$–$4\%$, intermediate states around $10^{33}$–$10^{34}\,\mathrm{erg\,s^{-1}}$, and quiescent states around $10^{32}\,\mathrm{erg\,s^{-1}}$ (Sidoli, 2011, Romano, 2015). Yet several studies identify intermediate objects rather than a clean dichotomy. IGR J18483-0311 behaves as an intermediate SFXT, and IGR J00370+6122 was explicitly presented as a system whose properties fit neither established subgroup, reinforcing a continuum interpretation rather than a strict binary classification (González-Galán, 2015).

Their Galactic distribution is consistent with youth. HMXBs are spatially clustered with OB star-forming complexes; for separations $\sim 20$0, the HMXB–OB cross-correlation reaches $\sim 20$1–$\sim 20$2 with the Peebles estimator and up to $\sim 20$3 with the Landy–Szalay estimator, while at $\sim 20$4 the observed and random distributions become statistically compatible. This is consistent with SgXBs remaining close to recent massive-star formation sites (Bodaghee et al., 2011).

2. Donor stars, stellar winds, and the wind-fed engine

The donor stars in SgXBs are evolved OB supergiants, typically luminosity class I or II, launching dense radiatively driven winds that provide the accretion reservoir. A line-driven wind is usually described with a $\sim 20$5-law,

$\sim 20$6

and the continuity relation

$\sim 20$7

In wind-fed accretion, the characteristic capture scale is

$\sim 20$8

with capture fraction

$\sim 20$9

and the corresponding direct-accretion luminosity

$\sim 10^3$0

Because $\sim 10^3$1, moderate changes in wind speed can shift the accretion flow between qualitatively different regimes (Gimenez-Garcia et al., 2016).

A detailed donor-wind comparison between the SFXT IGR J17544-2619 and the classical SgXB Vela X-1 illustrates the point. The derived donor types are O9I for IGR J17544-2619 and B0.5Iae for Vela X-1. Their terminal velocities differ strongly: $\sim 10^3$2 in IGR J17544-2619 and $\sim 10^3$3 in Vela X-1, with local wind velocities at the neutron-star orbit of about $\sim 10^3$4 and $\sim 10^3$5, respectively. In the Bozzo et al. accretion-regime framework adopted there, the lower wind speed and long pulsar spin of Vela X-1 favor persistent direct accretion, whereas the faster wind in IGR J17544-2619, combined with a short spin, favors inhibited accretion or a propeller-like regime (Gimenez-Garcia et al., 2016).

Wind structure is intrinsically non-stationary. In classical SgXBs, clumps are typically described as density contrasts of about a factor of $\sim 10^3$6, velocity contrasts of a factor of a few, and sizes up to $\sim 10^3$7. Such structure produces strong short-timescale variations in local density and velocity while leaving the long-term average luminosity near that expected for a structured but otherwise ordinary supergiant wind (Bozzo et al., 2014).

Donor properties may also differ systematically from isolated supergiants. In a four-source SGXB thesis sample, the donor atmospheres were reported to have higher projected rotational velocity, higher He abundance, and higher N/C ratio than isolated supergiants, while H$\sim 10^3$8 variations were dominated by intrinsic wind processes rather than clean orbital modulation (González-Galán, 2015).

The donor stars are also optically variable on short timescales. TESS photometry of 14 supergiant-companion HMXBs showed fast aperiodic variability in all sources, with a periodogram consisting of low-frequency flattening, a red-noise component at $\sim 10^3$9–10 d$10^5$0, and white noise at high frequency. The SGXB/SFXT donors show higher low-frequency amplitude and lower characteristic frequencies than persistent Be/X-ray binaries, and—unlike Be systems—an almost total lack of coherent high-frequency signals. This suggests that much of the rapid optical variability is intrinsic to the evolved supergiant donor rather than generated by the accretion flow (Filippantonio et al., 2024).

3. Luminosity states, duty cycles, and the classical–SFXT continuum

Classical SgXBs typically occupy the $10^5$1–$10^5$2 regime and show variability by factors of $10^5$3–50 on timescales of hundreds to thousands of seconds. SFXTs spend most of their time in much fainter states, around $10^5$4–$10^5$5, punctuated by short outbursts lasting a few hours and reaching luminosities comparable to the persistent luminosity of classical systems (Bozzo et al., 2014, Bozzo et al., 2014).

Long-term Swift/XRT monitoring transformed this phenomenology. The bright outbursts that originally defined SFXTs are only a small fraction of their activity; most of their life is spent in intermediate accretion states at $10^5$6–$10^5$7. This implies that SFXTs are not simply “off” between outbursts. The Swift project measured inactivity duty cycles for several SFXTs and classical controls, and established that sources such as IGR J08408-4503, IGR J16328-4726, IGR J17544-2619, and XTE J1739-302 spend a large fraction of time below the XRT detection threshold for 900 s observations, whereas systems reclassified as classical, such as IGR J16465-4507 and IGR J16493-4348, show much smaller inactivity duty cycles and single-peaked luminosity distributions (Romano, 2015).

A particularly powerful diagnostic is the cumulative luminosity distribution. In soft X-rays, classical SgXBs show a single knee around $10^5$8–$10^5$9, whereas SFXT distributions are shifted downward by factors of $0.05\%$0–100. Intermediate SFXTs such as IGR J18483-0311 and IGR J17354-3255 show similar shapes but with knees around $0.05\%$1; the extreme prototypes IGR J17544-2619, XTE J1739-302, and IGR J08408-4503 are shifted still lower and often show plateaus or multiple knees (Bozzo et al., 2014, Bozzo et al., 2014).

Case studies emphasize that the boundary is phenomenological rather than absolute. In Suzaku data, the classical SGXB IGR J16207-5129 showed a prolonged $0.05\%$2 ks attenuation episode, tentatively interpreted as an eclipse, while the prototype SFXT IGR J17391-3021 = XTE J1739-302 was observed not in a giant outburst but in a low-activity state with three weak flares only about a factor of 5 above pre-flare emission. Combining Suzaku with Swift monitoring showed that this low-activity state constitutes $0.05\%$3 of all observations in the 0.5–10 keV band, making it the most common emission phase for that source (Bodaghee et al., 2011).

The cumulative-distribution and duty-cycle work leads to an important correction of a common misconception: SFXTs are not merely classical SgXBs with rarer bright flares. They are systematically underluminous on average and spend much of their time in low-level accretion states that are invisible to wide-field hard-X-ray imagers (Bozzo et al., 2014).

4. Diagnostics of the local circumsource environment

The local environment of the compact object is commonly diagnosed through absorption, fluorescent iron lines, and multiwavelength wind tracers. One population-level result is that SGXBs and Be/X-ray binaries segregate in $0.05\%$4-based diagrams, with SFXTs bridging the gap. Across HMXBs, the $0.05\%$5-versus-orbital-period plane shows a weak anti-correlation with Spearman coefficient $0.05\%$6 and a $0.05\%$7 probability of chance occurrence, while the $0.05\%$8-versus-spin-period plane shows a weak positive correlation with $0.05\%$9 and about a $3$0 probability of chance occurrence. SGXBs occupy the high-$3$1 side of both spaces, consistent with dense supergiant winds and long pulsation periods in wind-fed systems (Bodaghee et al., 2011).

Outside eclipses, classical SgXBs also show systematically larger Fe K$3$2 equivalent widths than SFXTs. In a sample of 55 classical-SgXB and 21 SFXT observations from XMM-Newton and Suzaku, classical systems were found to be both more absorbed and more luminous, and to show stronger fluorescence lines. The same study confirmed a correlation between Fe K$3$3 equivalent width and absorbing column, but found no evidence for the previously claimed anti-correlation between Fe-line equivalent width and luminosity. The favored interpretation is not that SFXTs necessarily launch fundamentally different winds at the stellar surface, but that inhibited accretion in SFXTs keeps the source too faint to photoionize and brake the local wind efficiently; the local flow therefore remains faster and less dense, producing lower $3$4 and weaker fluorescence (Pradhan et al., 2017).

Longer-wavelength observations now provide an outer-wind counterpart to these inner-wind X-ray diagnostics. A twelve-source ALMA/NOEMA/VLA survey of neutron-star + OB-supergiant systems found 9/12 millimeter detections, including all four SFXTs in the sample, but only two radio detections. All systems with well-constrained SEDs showed inverted spectra with $3$5–0.8 in $3$6, consistent with free-free emission from the donor wind rather than a jet. The thermal-wind scaling was written as

$3$7

In that sample, SFXTs were systematically fainter at 100 GHz than prototypical classical SgXBs, and this mm faintness tracked their weaker Fe K$3$8 lines, suggesting less dense large-scale winds in SFXTs (Eijnden et al., 6 Aug 2025).

Multiwavelength diagnostics can also reveal extreme or unusual supergiant-hosting systems. CXO 174528.79-290942.8 (XID 6592) was argued to host a red supergiant donor and possibly to behave like an SFXT, with $3$9 mag W1 variability on few-hour timescales and a proposed hidden flare of order $4\%$0 lasting $4\%$1–$4\%$2 s. If confirmed, it would be the first red-supergiant SFXT and only the second known X-ray red-supergiant binary (Gottlieb et al., 2020).

5. Hydrodynamics, accretion-regime transitions, and disk formation

Hydrodynamic work has substantially revised the naive “captured clump = observed flare” picture. Three-dimensional simulations that inject realistic line-driven-instability wind structure into the accretion flow show that clumps are strongly processed by the detached bow shock and wake before reaching the neutron star. The bow shock acts as a variability filter: it slows, deforms, and mixes incoming overdensities, introduces a time lag, and decouples instantaneous obscuration from instantaneous accretion. In those calculations, the intrinsic mass-accretion rate varies only by factors of about $4\%$3 in a close configuration and $4\%$4 in a wider configuration, less than the factor of a few tens to 100 often observed in classical SgXBs. Likewise, simulated column-density excursions from individual clumps are too small to explain the strongest Vela X-1 absorption events. The implication is that clumps are necessary but not sufficient (Mellah et al., 2017).

Idealized Bondi–Hoyle–Lyttleton calculations in the regime most relevant to wind-fed neutron-star SgXBs—high Mach number, small effective accretor size, weak upstream gradients—further sharpen this point. Without upstream gradients, the 3D flow is stable. With finite gradients, three regimes appear: stable flow, turbulent unstable flow without a disk, and turbulent flow with a disk-like structure. Most observed SgXBs with reasonably determined parameters fall into the turbulent, diskless regime; OAO 1657-415 is the notable system most likely to host a persistent disk or disk-like structure. Thus instability does not generically imply classical disk formation (Xu et al., 2019).

These results motivate accretion-regulation models beyond clump-only variability. Two broad families dominate the literature. In magnetic or centrifugal gating models, the relative ordering of the accretion, magnetospheric, and corotation radii determines whether matter reaches the neutron star; modest changes in density or velocity can then produce large luminosity swings. In quasi-spherical settling accretion, a hot shell above the magnetosphere throttles inflow, reducing the luminosity by factors of $4\%$5 in the radiative-cooling regime and $4\%$6 in the Compton-cooling regime, while bright flares are triggered by sporadic magnetic reconnection with magnetized wind material (Bozzo et al., 2014).

Neither framework is decisively established. Magnetar-strength gating is weakened by the reported $4\%$7 keV cyclotron line in IGR J17544-2619, implying $4\%$8 rather than $4\%$9, while settling-accretion explanations often require systematic wind differences between SFXTs and classical SgXBs that are not straightforwardly seen in donor spectroscopy. A plausible synthesis is that the observed phenomenology emerges from structured winds, hydrodynamic filtering, and source-dependent magnetospheric regulation rather than from any single ingredient alone (Bozzo et al., 2014).

6. Geometry, evolution, and open problems

One influential geometrical unification picture organizes the supergiant population by orbital scale relative to the dense inner wind. In that scheme, classical and obscured sgHMXBs have small, roughly circular orbits within a dense inner wind or cocoon; classical SFXTs have larger or more eccentric orbits that sample a lower-density environment for most of the orbit; intermediate systems cross the transition. A characteristic radius of about $10^{33}$0 is used to separate an inner, high-clump-density region from a more dilute outer flow (Chaty, 2010).

The geometrical picture is useful but incomplete. A direct four-source SGXB thesis found that eccentricity does not correlate simply with the X-ray-flux differences among classical SGXBs, SFXTs, and intermediate systems, again supporting a continuum rather than a one-parameter taxonomy. The same work treated IGR J00370+6122 explicitly as a system between the standard subgroups (González-Galán, 2015).

Evolutionary models introduce an additional possibility: some SGXBs may sustain Roche-lobe overflow on a nuclear timescale rather than being purely wind-fed. Binary calculations with donors possessing a steep subsurface H/He gradient show that SGXBs can maintain stable mass transfer even at donor-to-accretor mass ratios exceeding 20. In a representative neutron-star model, a $10^{33}$1 donor with a $10^{33}$2 accretor sustained $10^{33}$3 for $10^{33}$4 yr while the donor radius shrank from $10^{33}$5 to $10^{33}$6 and the orbital period from 9.1 d to 1.4 d. These models were proposed as a way to explain the large Galactic SGXB population and the near absence of SGXBs in the SMC (Quast et al., 2019).

Magnetic structure in the donor wind may also matter. Spectropolarimetry of the SFXT IGR J11215-5952 yielded longitudinal magnetic-field measurements of $10^{33}$7 and $10^{33}$8, implying a donor surface field strong enough to magnetically confine the wind and plausibly create an equatorial enhancement intersected by the neutron star during its periodic outbursts. This supports models in which magnetized wind structure, not only neutron-star magnetospheric physics, helps set the transient behavior (Hubrig et al., 2018).

Several controversies therefore remain active. One concerns whether the decisive difference between SFXTs and classical SgXBs lies primarily in the companion wind or in the accretion regime; X-ray absorption and Fe-line studies favor the latter, whereas recent mm results indicate that large-scale wind density may still differ systematically between subclasses (Pradhan et al., 2017, Eijnden et al., 6 Aug 2025). Another concerns the physical status of “intermediate” systems, some of which behave more like reclassified classical SgXBs in cumulative-luminosity space, while others seem to bridge the classical and transient phenomenologies (Bozzo et al., 2014). A third concerns how donor variability, wind clumping, photoionization wakes, magnetospheric gating, and possible transient disk formation combine in individual sources. The field increasingly converges on a multiscale picture: SgXB behavior is set simultaneously by donor-star physics, wind hydrodynamics, orbital geometry, and compact-object boundary conditions.