Quiescent Low-Mass X-ray Binaries (qLMXBs)

• qLMXBs are transient neutron-star binaries in a quiescent state characterized by low accretion and soft, thermal X-ray spectra.
• Analysis using realistic hydrogen atmosphere models and radius inference techniques provides key constraints on neutron star masses, radii, and the dense-matter equation of state.
• Observations in globular clusters, including variability and hard-tail components, offer insights into residual accretion and systematic uncertainties in qLMXB studies.

Quiescent low-mass X-ray binaries (qLMXBs) are transient neutron-star low-mass X-ray binaries observed in their low-luminosity quiescent state, typically lasting months to years between brighter accretion outbursts. In this state, the accretion flow is largely shut down, so the X-ray emission is expected to come primarily from the neutron-star surface and atmosphere rather than from the disk. That observational geometry makes qLMXBs a central class for neutron-star structure studies: the thermal spectrum can be used to infer an allowed region in the mass–radius plane and thereby constrain the dense-matter equation of state (EoS), particularly when high-sensitivity imaging spectroscopy from Chandra and XMM-Newton is available (Bhattacharyya, 2010).

1. Definition, quiescent state, and phenomenology

Within the broader LMXB taxonomy, quiescence is a low-accretion state of transient systems rather than a separate evolutionary class. Transient systems show outbursts separated by quiescent intervals; in quiescence the mass accretion rate onto the compact object is very low or nearly zero, the accretion disk is weak or absent as a major X-ray emitter, and for neutron-star systems thermal emission from the stellar surface can become detectable (Bahramian et al., 2022). In the qLMXB literature summarized here, the term is used primarily for neutron-star binaries whose quiescent spectra are soft and atmosphere-like.

Observed qLMXB luminosities span the low-luminosity regime characteristic of quiescence. In globular-cluster searches, candidate qLMXBs are commonly discussed at $L_{\rm X}\sim10^{32}$–$10^{33}$ (Guillot et al., 2011), while broader reviews note that quiescent LMXBs can reach $L_X \sim 10^{30}-10^{33}\,\mathrm{erg\,s^{-1}}$ (Bahramian et al., 2022). Their spectra often contain two components: a soft thermal component from the neutron-star surface and, in many systems, an additional hard tail commonly modeled as a power law (Bahramian et al., 2015). That dual phenomenology is central to the class: some qLMXBs are overwhelmingly thermal, whereas others retain a substantial non-thermal contribution even in quiescence.

The standard physical interpretation ties the thermal component to deep crustal heating. During outburst, accreted matter compresses the crust and deposits heat; during quiescence that heat is reradiated through a thin atmosphere. This framework motivates the use of qLMXBs as probes of the stellar surface, the crust–core thermal state, and ultimately the EoS (Bhattacharyya, 2010).

2. Thermal spectrum formation and atmosphere modeling

The basic observational picture is that the neutron-star atmosphere radiates approximately as a thermal component. If the emission is treated as a blackbody-like surface signal, the observed bolometric flux and fitted temperature are related to the apparent radius at infinity by

$$R_{\infty} = \left(F_{\infty}/\sigma T^4_{\infty}\right)^{1/2}d,$$

where $F_{\infty}$ is the observed bolometric flux, $T_{\infty}$ is the fitted temperature, $d$ is the source distance, and $\sigma$ is the Stefan–Boltzmann constant (Bhattacharyya, 2010). Because the atmosphere is not a perfect blackbody and the surface is gravitationally redshifted, the observed and physical quantities are linked through

$T_{\rm BB} = T_{\infty}(1+z)/f,$

$R_{\rm BB} = R_{\infty} f^2/(1+z),$

with

$10^{33}$0

for a non-spinning star, where $10^{33}$1 is the color correction factor (Bhattacharyya, 2010).

In practice, qLMXB spectra are usually not fit with a pure blackbody. Realistic hydrogen atmosphere models are standard because they better represent the emergent spectrum and usually yield more reliable radii than naive blackbody fits (Bhattacharyya, 2010). Analyses of U24 in NGC 6397, Aql X-1, and the 47 Tuc sources X7 and X5 all use non-magnetic hydrogen-atmosphere models such as nsatmos, often with the emitting fraction fixed to unity to represent whole-surface emission [(Guillot et al., 2010); (Zelati et al., 2013); (Bogdanov et al., 2016)].

That whole-surface assumption is physically motivated in quiescence. The magnetic field is usually low, so there is no strong reason for the emission to come from a small hot spot; the whole surface is expected to radiate more uniformly (Bhattacharyya, 2010). The apparent radius measured by atmosphere fitting is then connected to the physical radius through the standard redshift relation

$10^{33}$2

which is the radius “seen at infinity” (Guillot et al., 2010).

Atmosphere composition remains a critical branch point. Hydrogen atmospheres are the standard assumption, but helium atmospheres are plausible in some systems and generally imply larger inferred radii and masses for the same observed spectrum [(Bogdanov et al., 2016); (Lattimer et al., 2013)]. This suggests that spectral modeling is never purely geometric; it is inseparable from assumptions about composition, redshift, and the origin of any high-energy excess.

3. Radius inference and equation-of-state constraints

Because the apparent thermal emission depends on both $10^{33}$3 and $10^{33}$4 through redshift and atmosphere physics, a measured quiescent spectrum defines an allowed region in the mass–radius plane. This is especially powerful in globular clusters, where the distance is relatively well known, often at the $10^{33}$5–10% level, removing one of the largest systematics in radius inference (Guillot et al., 2014). Radius scales directly with distance, so cluster membership is a major observational advantage (Bhattacharyya, 2010).

Individual systems have yielded some of the cleanest spectroscopic radius constraints. For U24 in NGC 6397, simultaneous fitting of five Chandra observations with nsatmos gave $10^{33}$6 km for $10^{33}$7, $10^{33}$8 eV, and $10^{33}$9 km; the source showed no significant aperiodic or periodic variability and no substantial hard tail, reinforcing the whole-surface thermal interpretation (Guillot et al., 2010). For 47 Tuc X7 and X5, improved low-pile-up Chandra spectroscopy yielded $L_X \sim 10^{30}-10^{33}\,\mathrm{erg\,s^{-1}}$0 km for X7 and $L_X \sim 10^{30}-10^{33}\,\mathrm{erg\,s^{-1}}$1 km for X5, both assuming hydrogen atmospheres and $L_X \sim 10^{30}-10^{33}\,\mathrm{erg\,s^{-1}}$2 (Bogdanov et al., 2016).

Population-level analyses have pushed the method further by combining sources. One common-radius MCMC analysis of five globular-cluster qLMXBs found $L_X \sim 10^{30}-10^{33}\,\mathrm{erg\,s^{-1}}$3 km at 90% confidence under conservative assumptions about distance, absorption, and possible hard tails (Guillot et al., 2013). A later simultaneous MCMC fit to six qLMXBs reproduced a constant-radius result of $L_X \sim 10^{30}-10^{33}\,\mathrm{erg\,s^{-1}}$4 km at 90% confidence, and $L_X \sim 10^{30}-10^{33}\,\mathrm{erg\,s^{-1}}$5 km when causality was imposed (Guillot et al., 2014).

These radius measurements have been used to test candidate EoSs directly. In the EOS-constrained analysis of six qLMXBs, the models WFF1, AP4, MPA1, PAL1, MS0, and three representative chiral effective field theory EoSs were compared by forcing each source to lie on a specific mass–radius curve. PAL1 and MS0 were rejected at the 99% confidence level, while the other tested EoSs were not excluded at $L_X \sim 10^{30}-10^{33}\,\mathrm{erg\,s^{-1}}$6 certainty (Guillot et al., 2014). More broadly, the 47 Tuc X5/X7 study concluded that, when combined with other spectroscopic measurements, the preferred radii are $L_X \sim 10^{30}-10^{33}\,\mathrm{erg\,s^{-1}}$7 km around a canonical mass of $L_X \sim 10^{30}-10^{33}\,\mathrm{erg\,s^{-1}}$8, favoring a somewhat softer dense-matter EOS than many purely nucleonic models tuned to low-density nuclear data (Bogdanov et al., 2016).

Interpretation, however, depends on the adopted framework. A softer EoS generally predicts smaller radii for a given mass, while a stiffer EoS allows larger radii; qLMXB measurements are therefore fundamentally radius-sensitive, and systematic shifts in the inferred emitting area can move the favored EoS class (Bhattacharyya, 2010).

4. Variability, hard tails, and residual accretion

A central misconception about qLMXBs is that quiescent emission is necessarily static. Monitoring of Aql X-1 demonstrates the opposite. Weekly Swift/XRT observations during 2012 March 15 to November 9 found the source to be highly variable in quiescence, with $L_X \sim 10^{30}-10^{33}\,\mathrm{erg\,s^{-1}}$9, $$R_{\infty} = \left(F_{\infty}/\sigma T^4_{\infty}\right)^{1/2}d,$$0, and two flares reaching $$R_{\infty} = \left(F_{\infty}/\sigma T^4_{\infty}\right)^{1/2}d,$$1 in 0.3–10 keV (Zelati et al., 2013). The quiescent spectra required a soft thermal component below $$R_{\infty} = \left(F_{\infty}/\sigma T^4_{\infty}\right)^{1/2}d,$$2 keV plus a hard power-law tail above $$R_{\infty} = \left(F_{\infty}/\sigma T^4_{\infty}\right)^{1/2}d,$$3 keV, and the statistically preferred global fit was one in which only the power-law normalization varied, by about a factor of 80, while $$R_{\infty} = \left(F_{\infty}/\sigma T^4_{\infty}\right)^{1/2}d,$$4, $$R_{\infty} = \left(F_{\infty}/\sigma T^4_{\infty}\right)^{1/2}d,$$5, and $$R_{\infty} = \left(F_{\infty}/\sigma T^4_{\infty}\right)^{1/2}d,$$6 remained tied (Zelati et al., 2013).

At the same time, decade-long monitoring of 12 qLMXBs in three globular clusters found no evidence of thermal variability in 10 cases. The allowed temperature variations were below 11% for the seven qLMXBs without detectable power-law components and below 20% for three others that showed non-thermal emission; only NGC 6440 CX 1 and Terzan 5 CX 12 showed marginal evidence for thermal changes (Bahramian et al., 2015). This contrast indicates that the class is heterogeneous: some sources are dominated by a stable thermal floor consistent with core-heated emission, whereas others display obvious low-level accretion phenomenology.

Hard-spectrum qLMXBs illustrate the boundary cases. Swift J1749.4-2807, observed in quiescence at $$R_{\infty} = \left(F_{\infty}/\sigma T^4_{\infty}\right)^{1/2}d,$$7, showed three eclipses consistent with the outburst ephemeris, but its 0.5–10 keV spectrum was best fit by a simple power law with $$R_{\infty} = \left(F_{\infty}/\sigma T^4_{\infty}\right)^{1/2}d,$$8 and $$R_{\infty} = \left(F_{\infty}/\sigma T^4_{\infty}\right)^{1/2}d,$$9; a pure hydrogen atmosphere model did not fit adequately, and the upper limit on any thermal component was $F_{\infty}$0 keV with $F_{\infty}$1 (Degenaar et al., 2012). SAX J2224.9+5421 provides another instructive case: source 4, the preferred quiescent counterpart, has a soft atmosphere-like spectrum with $F_{\infty}$2 eV and $F_{\infty}$3, while source 1 remains a harder alternative with $F_{\infty}$4 and $F_{\infty}$5, suggestive of strong magnetic-field effects or low-level accretion (Degenaar et al., 2014).

These examples matter methodologically. Non-thermal tails plausibly associated with residual accretion can bias thermal fits and inferred radii; conversely, the absence of obvious variability or a hard tail strengthens the interpretation of quiescent emission as whole-surface thermal radiation [(Bhattacharyya, 2010); (Bahramian et al., 2015)].

5. Globular clusters, source identification, and population studies

Globular clusters are the canonical qLMXB laboratories because their distances are independently constrained and their dense stellar environments produce rich compact-binary populations. Their disadvantages are equally important: crowding and source confusion can complicate spectral interpretation, especially for hard tails in cluster cores [(Bhattacharyya, 2010); (Guillot et al., 2011)].

The candidate qLMXB XMMU J180916$F_{\infty}$6255425 in NGC 6553 illustrates both the promise and the complexity. Its spectrum is statistically consistent with a neutron-star H-atmosphere model at the cluster distance, with a combined EPIC fit yielding $F_{\infty}$7 km, $F_{\infty}$8 eV, and a required power-law tail with $F_{\infty}$9 contributing $T_{\infty}$0 of the total flux (Guillot et al., 2011). Archival Chandra data showed the same source position and a consistent thermal spectrum, but also revealed low-significance nearby possible sources, leading to the conclusion that the power-law excess is most likely due to unresolved neighboring sources rather than intrinsic emission (Guillot et al., 2011).

Multiwavelength studies refine classification further. Simultaneous Chandra and HST observations of 47 Tuc found three certain qLMXBs (X5, X7, W37) and two likely qLMXBs (X4, W17). X5 and W37 show X-ray eclipses, X7 is extremely stable and dominated by a soft atmosphere component, X4 shows correlated X-ray and optical/NUV variability consistent with changes in accretion rate, and W17 has a power-law-dominated X-ray spectrum plus a newly discovered blue, variable NUV counterpart (Berg et al., 2024). The same study showed that qLMXBs in 47 Tuc have larger X-ray-to-optical flux ratios than cataclysmic variables, and argued that the ratio of X-ray flux to excess blue optical flux is a particularly effective discriminator: at

$T_{\infty}$1

only qLMXBs are found, whereas at lower values the class overlap with CVs increases (Berg et al., 2024).

At the population scale, Terzan 5 currently provides the richest cluster sample. Analysis of 737 ks of Chandra data identified 22 candidate qLMXBs, more than in any other Galactic globular cluster. The qLMXB identification was based on X-ray colors and absorbed neutron-star-atmosphere-plus-power-law fitting; more than 50% of the qLMXB sources have neutron-star thermal components contributing over 80% of the total luminosity, and no thermally dominated qLMXBs were seen below $T_{\infty}$2, though more power-law-dominated objects could have been missed (Kumawat et al., 4 Aug 2025). The qLMXB radial distribution is consistent with a population of mass $T_{\infty}$3, supporting the interpretation of a centrally concentrated neutron-star-binary population shaped by cluster dynamics (Kumawat et al., 4 Aug 2025).

6. Systematics, cooling physics, and contested assumptions

The principal systematic uncertainties in qLMXB inference are distance, interstellar absorption, atmosphere composition, residual accretion, and nonuniform surface emission. Because $T_{\infty}$4, even modest distance errors propagate directly into the inferred radius. Interstellar absorption removes a substantial part of the soft X-ray spectrum and increases degeneracies among temperature, absorption, and emitting area. Atmosphere composition is another major uncertainty: although hydrogen is the standard assumption, the actual composition may differ, and the inferred color correction and spectral shape depend on composition (Bhattacharyya, 2010).

Several studies show that these are not secondary details. A Bayesian reanalysis of five qLMXBs found that independent absorption estimates are strongly preferred over self-consistent spectral-fit $T_{\infty}$5 values used in one earlier framework, with a Bayes factor of $T_{\infty}$6 for H10 versus G13 $T_{\infty}$7; the same study also found that models allowing either hydrogen or helium atmospheres are favored over models requiring hydrogen for all five stars, with H+He / H $T_{\infty}$8 (Lattimer et al., 2013). In the 47 Tuc X7/X5 analysis, even $T_{\infty}$9 pile-up materially changed the mass–radius confidence contours, and the authors argued on astrophysical grounds that a helium atmosphere for X7 is unlikely despite being formally allowed by the X-ray data (Bogdanov et al., 2016).

Quiescent emission is also a probe of neutron-star thermal relaxation after outburst. Numerical cooling calculations for MXB 1659-29, KS 1731-260, EXO 0748-676, XTE J1701-462, and IGR J17480-2446 showed that MXB 1659-29, KS 1731-260, and EXO 0748-676 can be described within a deep crustal cooling scenario, whereas XTE J1701-462 and IGR J17480-2446 require models beyond standard crustal cooling, including residual accretion during quiescence, additional shallow heating, or thermal insulation from a buried magnetic field (Turlione et al., 2013). For EXO 0748-676 specifically, simultaneous UV and X-ray work found no clear UV–X-ray correlation and $d$0, much higher than in low-level accretors such as Cen X-4 and Aql X-1; this favored crust-cooling thermal emission over ongoing low-level accretion as the dominant origin of the quiescent X-ray evolution (Parikh et al., 2021).

A further contested assumption concerns the interpretation of radius posteriors under causality. In one combined qLMXB analysis, imposing causality did not shift the fitted radius posterior, but the authors emphasized that if a neutron star of mass $d$1 were observed, then radii below 11 km would be ruled out under causality, effectively falsifying the small-radius interpretation used in that analysis (Guillot et al., 2014). This underscores a broader point: qLMXB results do not stand alone, but interact with mass measurements, atmosphere composition constraints, and instrumental systematics.

The resulting consensus is neither that qLMXBs are systematics-free nor that they are too uncertain to be useful. Rather, they are promising precisely because quiescence can provide a comparatively clean view of the stellar surface, but only when the major systematics are controlled. Their constraints become most powerful when combined with independent distance estimates, better atmosphere models, and complementary measurements from bursts, timing, spectroscopy, and orbital dynamics (Bhattacharyya, 2010).