High redshift x-ray galaxy clusters

In cosmology, the validation of cluster formation and evolution models requires accurate knowledge of the relationship between cluster luminosity (L_X) and gas temperature (T_X) and of its changes over given redshift ranges. As powerful x-ray emitters, galaxy clusters offer a unique insight into the luminosity-temperature relation. But one problem that must be addressed is how to relate cluster x-ray gas temperature or luminosity to the quantities predicted by cosmological models, usually mass.

In 1986, Kaiser proposed a simple self-similar model that took into account gravitational processes such as shocks and adiabatic compression during cluster formation.¹ This model is, however, being questioned since the observed slope of the L_X–T_X relation is much steeper (~3) than that predicted by the model. Such deviation from self-similarity is taken as evidence that simple gravitational collapse is not the only process that governs the heating of baryons. Furthermore, it suggests that non-gravitational processes occurring before or during cluster formation contribute to the cluster energy budget in a non-negligible way. Any observed change in the L_X–T_X scaling law at high redshift with respect to low redshift should allow us to distinguish between the different scenarios proposed for heating schemes and cooling efficiency.² In this context, our goal is to investigate the L_X–T_X relation of the high redshift x-ray clusters observed with the European Space Agency x-ray satellites Chandra and XMM-Newton.

Previous galaxy cluster studies performed with x-ray telescopes such as Einstein, ASCA, Beppo-SAX, or ROSAT attributed the total x-ray emission from the clusters to thermal bremsstrahlung emission arising from the thin hot gas that fills the inter-cluster regions. The advantage of using observatories such as Chandra and XMM-Newton is that they allow resolution of x-ray point sources projected into the cluster and the subtraction of their non-thermal emission from the diffuse thermal emission of the cluster.

We recently investigated the impact of these point sources on the physical properties of the x-rays.³ Biases inherent to several measurements (T_X, L_X, L_X–T_X relation) were quantified using a sample of 18 high-z (0.25<z<1.01) clusters from the Chandra archive. The analysis showed that the point sources projected into the cluster's extended emission considerably affect cluster temperature and luminosity estimates (up to 13% and 17% respectively). These percentages become even larger for clusters with z>0.7, where temperature and luminosity increase up to 24% and 22%, respectively. These results suggest that point sources should be removed to correctly estimate the cluster properties. However, the inclusion of the point sources does not impact significantly on the slope and normalization of the L_X–T_X relationship. This is because the correction applied to T_X and L_X produces a moderate shift for each cluster in the L_X–T_X plane, almost parallel to the best-fit of the “correct” L_X–T_X relation. Our study will be useful for the future analysis of archival cluster data and future x-ray surveys, specifically those which do not benefit from the spatial resolution provided by Chandra.

We have also presented results on the evolution of the “correct” L_X–T_X relation.⁴ We used an archival sample of 39 high redshift (0.25 < z <1.3) objects, either observed by Chandra or XMM–Newton, to re-examine the L_X–T_X relation for high-z galaxy clusters for comparison with that of nearby clusters^5,6 and theoretical models. Our analysis confirms previous results:^7–9 the slope of the relation is steeper (α=3) than expected from the self–similar model prediction (α=2). The evolution of the L_X–T_X relation with respect to the local universe goes in the direction of high-z clusters being more luminous than the local ones by a factor ≈ 2 at any given temperature (see Figure 1). This result provides insight into a new type of L_X–T_X evolution. The redshift dependence of the relation in the full redshift range (i.e. 0 ≤ z ≤ 1.3) cannot be described by the self-similar model, nor by any power law of the form (1+z)^A where A is the evolution parameter. It seems that a strong evolution, similar or stronger than the self-similar, is required by the L_X–T_X relation from z = 0 to z ∼ 0.3, followed by a much weaker, if any, evolution at higher redshift. This weaker evolution would be compatible with the increasing importance of non-gravitational effects at high redshift (see Figure 2) in the structure formation process.¹⁰

Figure 1. The L_X-T_X relation for an archival sample of 39 distant clusters.⁴ The solid line indicates the best power law fit to the data. Dotted and dashed lines refer to nearby cluster samples.

Figure 2. Ratio between L_X and T_X^α, where α is the temperature dependence of the L_X-T_X relation as described by Voigt for three non-gravitational models,² versus 1+z. The dashed lines indicate the redshift dependence of the models, considering only the 39 distant cluster sample. The dot-dashed lines represent the models normalized to the nearby cluster samples at z=0. The three models are in good agreement with the data for the 0.3 < z < 1.3 redshift range (dashed lines). When the entire range, i.e. 0 < z < 1.3, is taken into account (dot-dashed lines), none of the models are valid. z: redshift.

The “two-step evolution” observed when one considers the entire 0–1.3 redshift range is an unexpected result that requires confirmation. To correctly analyze the evolution of the L_X–T_X relation, the first priority would be to use a more statistically significant sample of nearby clusters (z < 0.25). A handful of clusters not contaminated by cooling cores and at z < 0.1 could indeed shed light on this issue. These considerations should be kept in mind for future generation satellites to provide them with a larger field of view and improved resolution and sensitivity.