Résumé
Context. The stellar evolution code Modules for Experiments in Stellar Astrophysics (MESA) is public and is widely used by the community. It includes the possibility of taking several non-standard processes such as atomic diffusion into account. Even if the effect of gravitational settling is considered a standard ingredient in stellar modelling today, this is not the case for radiative accelerations. The specific treatment of atomic diffusion along with the radiative accelerations has never been compared with other stellar evolution codes. Benchmarking these codes is important because improved accuracy is required in order to analyse data from present and future space missions, such as the Kepler, Transiting Exoplanet Survey Satellite, and PLAnetary Transits and Oscillations of stars missions. Aims. The aim of this paper is to compare MESA models including atomic diffusion (with radiative accelerations) with models computed with the Montreal/Montpellier stellar evolution code and with the Code d'Evolution Stellaire Adaptatif et Modulaire (CESTAM). Additionally, we assess the impact of some MESA options related to atomic diffusion. Methods. We calculated atomic diffusion, including radiative accelerations, following the abundance profiles of 14 elements with MESA models. This was then compared with 1.1 and 1.4 M-circle dot models computed with the Montreal/Montpellier and CESTAM codes. Various tests of MESA options for atomic diffusion were also carried out by varying only one of them at a time. Results. We find that the abundance profiles of the considered elements in the MESA models compare rather well with the models computed with the two other codes when atomic diffusion options are carefully set. We also show that some options in MESA are crucial for a proper treatment of atomic diffusion.