Institution(s): 1. Laboratoire d'Astrophysique de Marseille
In the COSMOS field, we have built a sample of 22 000 Lyman break galaxies. About 100 of them are individually detected with Herschel / SPIRE but the vast majority of them are not. To get some crucial information of the dust properties and the total star formation rate of these galaxies, we have separated those galaxies in several bins of UV luminosity (L_UV), stellar mass (M_star) and UV slope beta_UV. Then, we have stacked the LBGs and obtained detections for most of them. The resulting FIR spectral energy distributions (SEDs) are pretty nice and we have added the information in the UV/optical/NIR/MIR/radio to obtain a full understanding and modeling of the SED. The modeling of these observed SEDs with our code CIGALE provides us with crucial but well-constrained physical parameters. This data also allows us to estimate the evolution of the shape UV dust attenuation law. We will try to extrapolate and predict what can we learn from this information to constrain the expected observations at very high redshifts with JWST and WISH.