POISSON project III. Investigating the evolution of the mass accretion rate

Context. As part of the Protostellar Optical-Infrared Spectral Survey On NTT (POISSON) project, we present the results of the analysis of low-resolution near-IR spectroscopic data (0.9-2.4 mu m) of two samples of young stellar objects in the Lupus (52 objects) and Serpens (17 objects) star-forming clouds, with masses in the range of 0.1 to 2.0 M-circle dot and ages spanning from 10(5) to a few 10(7) yr. Aims. After determining the accretion parameters of the targets by analysing their Hi near-IR emission features, we added the results from the Lupus and Serpens clouds to those from previous regions (investigated in POISSON with the same methodology) to obtain a final catalogue (143 objects) of mass accretion rate values ((M) over dot(acc)) derived in a homogeneous and consistent fashion. Our final goal is to analyse how (M) over dot(acc) correlates with the stellar mass (M-*) and how it evolves in time in the whole POISSON sample. Methods. We derived the accretion luminosity (L-acc) and (M) over dot(acc) for Lupus and Serpens objects from the Br gamma (Pa beta in a few cases) line by using relevant empirical relationships available in the literature that connect the HI line luminosity and L-acc. To minimise the biases that arise from adopting literature data that are based on different evolutionary models and also for self-consistency, we re-derived mass and age for each source of the POISSON samples using the same set of evolutionary tracks. Results. We observe a correlation. (M) over dot(acc) similar to M-*(2.2) between mass accretion rate and stellar mass, similarly to what has previously been observed in several star-forming regions. We find that the time variation of (M) over dot(acc) is roughly consistent with the expected evolution of the accretion rate in viscous disks, with an asymptotic decay that behaves as t(-1.6). However, (M) over dot(acc) values are characterised by a large scatter at similar ages and are on average higher than the predictions of viscous models. Conclusions. Although part of the scattering may be related to systematics due to the employed empirical relationship and to uncertainties on the single measurements, the general distribution and decay trend of the (M) over dot(acc) points are real. These findings might be indicative of a large variation in the initial mass of the disks, of fairly different viscous laws among disks, of varying accretion regimes, and of other mechanisms that add to the dissipation of the disks, such as photo-evaporation.