Recent work based on a global measurement of the ICM properties find evidence for an increase of the iron abundance in galaxy clusters with temperature around 2-4 keV up to a value about 3 times larger than that typical of very hot clusters. We have started a study of the metal distribution in these objects from the sample of Baumgartner et al. (2005), aiming at resolving spatially the metal content of the ICM. We report here on a 42ks XMM observation of the first object of the sample, the cluster Abell 2028. The XMM observation reveals a complex structure of the cluster over scale of 300 kpc, showing an interaction between two sub-clusters in cometary-like configurations. At the leading edges of the two substructures cold fronts have been detected. The core of the main subcluster is likely hosting a cool corona. We show that a one-component fit for this region returns a biased high metallicity. This inverse iron bias is due to the behavior of the fitting code in shaping the Fe-L complex. In presence of a multi-temperature structure of the ICM, the best-fit metallicity is artificially higher when the projected spectrum is modeled with a single temperature component and it is not related to the presence of both Fe-L and Fe-K emission lines in the spectrum. After accounting for the bias, the overall abundance of the cluster is consistent with the one typical of hotter, more massive clusters. We caution the interpretation of high abundances inferred when fitting a single thermal component to spectra derived from relatively large apertures in 3-4 keV clusters, because the inverse iron bias can be present. Most of the inferences trying to relate high abundances in 3-4 keV clusters to fundamental physical processes will likely have to be revised.
