2020A&A...634A..11D

Context. The most massive galaxies (M_stellar≥10¹¹M_☉) in the local Universe are characterized by a bulge-dominated morphology and old stellar populations, in addition to being confined to a tight mass-size relation. Identifying their main components can provide insights into their formation mechanisms and subsequent mass assembly. Aims. Taking advantage of Hubble Space Telescope (HST) CANDELS data, we analyze the lowest redshift (z<0.5) massive galaxies in the H and I band in order to disentangle their structural constituents and study possible faint non-axisymmetric features. Methods. Our final sample consists of 17 massive galaxies. Due to the excellent HST spatial resolution for intermediate redshift objects, they are hard to model by purely automatic parametric fitting algorithms. We performed careful single and double (bulge-disk decompositions) Sersic fits to their galaxy surface brightness profiles. We compare the model color profiles with the observed ones and also derive multi-component global effective radii attempting to obtain a better interpretation of the mass-size relation. Additionally, we test the robustness of our measured structural parameters via simulations. Results. We find that the Sersic index does not offer a good proxy for the visual morphological type for our sample of massive galaxies. Our derived multi-component effective radii give a better description of the size of our sample galaxies than those inferred from single Sersic models with GALFIT. Our galaxy population lies on the scatter of the local mass-size relation, indicating that these massive galaxies have not experienced a significant growth in size since z∼0.5. Interestingly, the few outliers are late-type galaxies, indicating that spheroids must reach the local mass-size relation earlier. For most of our sample galaxies, both single- and multi-component Sersic models withGALFIT show substantial systematic deviations from the observed surface brightness profiles in the outskirts. These residuals may be partly due to several factors, namely a nonoptimal data reduction for low surface brightness features or the existence of prominent stellar haloes for massive galaxies, or they could also arise from conceptual shortcomings of parametric 2D image decomposition tools. They consequently propagate into galaxy color profiles. This is a significant obstacle to the exploration of the structural evolution of galaxies, which calls for a critical assessment and refinement of existing surface photometry techniques.