X-ray Luminosity–Temperature Relation

Updated 6 December 2025

The X-ray luminosity–temperature relation is a scaling law that links the thermal energy and radiative output of galaxy clusters, delineating both gravitational and non-gravitational influences.
Empirical studies show slopes ranging from 2.0 to 3.5, where deviations from the self-similar model highlight the roles of AGN feedback, radiative cooling, and core selection effects.
This relation is critical for cosmological applications by converting observable X-ray properties into mass proxies, thus refining dark energy and structure growth measurements.

The X-ray luminosity–temperature (Lₓ–T) relation is the cornerstone empirical scaling describing the connection between thermal properties and radiative output in hot, diffuse baryonic atmospheres of galaxy clusters, groups, and massive galaxies. It encodes both fundamental gravitational scaling and the integrated effect of non-gravitational physics—most notably radiative cooling, AGN feedback, and gas preheating—on the thermodynamic structure of the intracluster medium (ICM). Modern measurements leverage large, well-characterized X-ray cluster samples to extract the form, normalization, intrinsic scatter, redshift evolution, and dependencies on system relaxation and mass scale, with rigorous attention to selection biases and astrophysical systematics.

1. Theoretical Foundations and Self-Similar Baseline

The canonical self-similar model for cluster formation, rooted in the physics of spherical collapse and virial equilibrium, predicts X-ray bolometric luminosity to follow $L_X \propto T^2 E(z)$ for clusters dominated by thermal bremsstrahlung emission, where $E(z)$ represents the Hubble parameter evolution (Fujita et al., 2019). This baseline is derived under the assumptions of scale-free gravitational heating, hydrostatic equilibrium, and a single characteristic density set by the critical density at formation. However, the hierarchical nature of structure formation introduces a concentration–mass dependence; clusters with lower mass (and hence earlier formation epochs) are more centrally concentrated. Accounting for the mass dependence of halo structure and the fundamental plane of cluster scaling relations, the true baseline slope becomes shallower, $L_X \propto T^{1.6-1.8}$ , rather than the canonical $T^2$ (Fujita et al., 2019).

2. Empirical Lₓ–T Relation: Local and High-z Clusters

Observed cluster samples show $L_X$ – $T$ scaling exponents consistently in the range $2.5-3.5$ for bolometric luminosity, with substantial evidence for a positive or mildly evolving normalization at higher redshift and pronounced steepening relative to the self-similar expectation (Mittal et al., 2011, Ebrahimpour et al., 2018, Giles et al., 2015, Babyk et al., 2013). For example, the XXL Survey analysis yields $L_X \propto T^{3.08\pm0.15} E(z)^{1.64\pm0.77}$ (bolometric) with an intrinsic scatter of $\sim0.47$ dex (Giles et al., 2015), while the XMM Cluster Survey finds $L_X \propto T^{3.09\pm0.08} E(z)^{1+\gamma}$ with $\gamma\approx0.58$ (consistent with self-similar at $1\sigma$ ) (Ebrahimpour et al., 2018). In Chandra high-z samples ($0.4 < z < 1.4$), $L_X \propto (1+z)^{1.5\pm0.23} T^{2.55\pm0.07}$ , establishing clearly stronger luminosities at fixed temperature with increasing redshift (Babyk et al., 2013).

Nevertheless, unbiased samples targeting low X-ray surface-brightness clusters or those selected independently of X-ray properties produce slopes closer to $2.0$ (Andreon et al., 2022), highlighting the significant role of selection effects and sample composition on fitted exponents.

Table: Representative Lₓ–T Power-Law Fit Parameters

Sample/Survey	Slope ( $\alpha$ )	Evolution	Intrinsic Scatter (dex)	arXiv ID
XXL-100-GC	$3.08 \pm 0.15$	$E(z)^{1.64\pm0.77}$	$0.47$	(Giles et al., 2015)
XCS+redMaPPer	$3.09 \pm 0.08$	$E(z)^{1+\gamma}$	$0.32$	(Ebrahimpour et al., 2018)
XUCS (unbiased)	$2.02 \pm 0.22$	negligible	$0.23$	(Andreon et al., 2022)
HIFLUGCS (all)	$2.94 \pm 0.16$	—	$0.45$	(Mittal et al., 2011)
400d (low-mass)	$3.29 \pm 0.33$	—	$0.51$	(Zou et al., 2016)
Chandra ( $z>0.4$ )	$2.55 \pm 0.07$	$(1+z)^{1.50\pm0.23}$	$0.20$–$0.30$	(Babyk et al., 2013)
XCLASS-redMaPPer	$3.03 \pm 0.26$	$E(z)^1$	$0.44$	(Molham et al., 2020)

3. Physical Interpretation: Non-Gravitational Heating, Cooling, and Feedback

The steepened $L_X$ – $T$ relation relative to the self-similar slope is robustly interpreted as evidence for non-gravitational energy input into the ICM. Mechanisms include:

AGN Feedback: AGN-driven jets inflate cavities and deposit kinetic energy, especially in low-mass systems, suppressing central gas density and X-ray emissivity.
Radiative Cooling: Core-dominated cooling can boost $L_X$ at fixed $T$ in strong cool-core clusters, driving a higher slope in these subsamples (Mittal et al., 2011).
Mergers and Shocks: Unrelaxed clusters or those experiencing recent mergers can display even steeper $L_X$ – $T$ scalings due to entropy redistribution (Maughan et al., 2011).
Preheating: Early preheating raises the entropy floor, particularly in group-scale halos, suppressing $L_X$ at lower $T$ (Hilton et al., 2012, Andreon et al., 2010).

Hydrodynamical simulations including efficient early feedback reproduce both the steep slope and normalization, while AGN feedback recipes that inject energy tied to ongoing star formation or only at late times tend to underpredict the observed effect (Hilton et al., 2012).

4. Impact of Sample Selection, Core-Excised Measurements, and Intrinsic Scatter

The measured $L_X$ – $T$ slope and normalization are highly sensitive to sample construction and treatment of cluster cores. X-ray flux-limited or extent-selected samples are subject to Malmquist/Eddington bias, which boosts the recovered slope and artificially inflates normalization if not rigorously corrected (Giles et al., 2015, Molham et al., 2020). Core-excised measurements systematically reduce the scatter and can reduce the slope toward self-similar values in massive, relaxed clusters ( $\alpha\sim1.9$ ) (Maughan et al., 2011). Systems with strong cool cores have the steepest slopes ( $\alpha>3$ ), while non-cool-core clusters and unbiased samples converge to shallower slopes (Mittal et al., 2011, Andreon et al., 2022). Intrinsic scatter at fixed $T$ is typically $0.2$–$0.5$ dex, with core excision often reducing the scatter by $\sim$ 25% (Mittal et al., 2011, Zou et al., 2016).

5. Evolution and Redshift Dependence

Redshift evolution of the normalization and shape of the $L_X$ – $T$ relation encodes the history of non-gravitational processes. The majority of observational studies, once bias-corrected, find a positive evolution of normalization—clusters of fixed $T$ are more luminous at higher $z$ —with evolution factors $E(z)^{\,\gamma}$ where $\gamma\sim1$ –1.6 is consistent with or exceeds self-similar predictions (Giles et al., 2015, Babyk et al., 2013). However, unbiased optically-selected high-z clusters can appear fainter than predicted by pure self-similar or high-feedback models, implying a suppression of $L_X$ at fixed $T$ potentially due to early entropy injection (Andreon et al., 2010). The XMM Cluster Survey finds the normalization evolving as $(1+z)^{-1.5\pm0.5}$ , suggesting slower-than-self-similar evolution, consistent with high-z entropy being set early and maintained thereafter (Hilton et al., 2012).

6. Group/Cluster Regime Transition and Early-Type Galaxies

At lower temperatures ( $T<2$ keV), corresponding to galaxy groups, theory and some observations predict a possible break or shallower slope, possibly reflecting the increasing dominance of line cooling, nonthermal pressure and AGN heating (Voit et al., 2017, Babyk et al., 2018). For early-type galaxies, the $L_X$ – $T$ relation is even steeper, $L_X \propto T^{4.5}$ , indicating that feedback processes dominate completely over gravitational heating in setting the gas phase structure (Babyk et al., 2018).

A change-point or broken power-law with a transition near $2$ keV has been statistically tested, but current large surveys (e.g., XCS+redMaPPer) show only modest preference for such a model; the data are well described by a single steep power-law across $1Ebrahimpour et al., 2018).

7. Cosmological and Astrophysical Applications

The $L_X$ – $T$ relation is integral to the use of galaxy clusters as cosmological probes, enabling robust mass-observable calibrations for cluster abundance studies, and for cosmological parameter inference involving growth of structure and dark energy (Giles et al., 2015). Systematic uncertainties in normalization, slope, redshift evolution, and scatter propagate directly into cosmological constraints. Moreover, the $L_X$ – $T$ normalization provides a cosmology-sensitive test when mapped across the sky; recent studies have detected statistically significant ( $>3\sigma$ ) anisotropies in the $L_X$ – $T$ normalization, raising the possibility of large-scale cosmological anomalies or unrecognized systematics (Migkas et al., 2017, Migkas et al., 2020).

Concordance between observed $L_X$ – $T$ relations and the precipitation-limited feedback model supports a universal role for AGN-regulated thermal balance in all massive halos, although detailed departures at group/galaxy scale and the incidence of multiphase phenomena retain sensitivity to the microphysics of feedback and cooling (Voit et al., 2017).