Mass Composition from 3 EeV to 100 EeV using the Depth of the Maximum of Air-Shower Profiles Estimated with Deep Learning using Surface Detector Data of the Pierre Auger Observatory

We present a new analysis for estimating the depth of the maximum of air-shower profiles, X_max, to investigate the evolution of the ultra-high-energy cosmic ray mass composition from 3 to 100 EeV. We use a recently developed deep-learning-based technique for the reconstruction of X_max from the data of the surface detector of the Pierre Auger Observatory. To avoid systematic uncertainties arising from hadronic interaction models in the simulation of surface detector data, we calibrate the new reconstruction technique with observations of the fluorescence detector. Using the novel analysis, we have a 10-fold increase of statistics at E > 5 EeV with respect to fluorescence detector data. We are able, for the first time, to study the evolution of the mean and standard deviation of the X_max distributions up to 100 EeV. We find an excellent agreement with fluorescence observations and confirm the increase of the mean logarithmic mass hln(A)i and a decrease of the X_max fluctuations with energy. The X_max measurement at the highest — so far inaccessible — energies is consistent with a pure mass composition and a mean logarithmic mass of around ∼ 3 (estimated using the Sibyll 2.3d and the EPOS-LHC hadronic interaction models).