On the diversity of super-luminous supernovae: ejected mass as the dominant factor

We assemble a sample of 24 hydrogen-poor super-luminous supernovae (SLSNe). Parameterizing the light curve shape through rise and decline timescales shows that the two are highly correlated. Magnetar-powered models can reproduce the correlation, with the diversity in rise and decline rates driven by the diffusion timescale. Circumstellar interaction models can exhibit a similar rise-decline relation, but only for a narrow range of densities, which may be problematic for these models. We find that SLSNe are approximately 3.5 magnitudes brighter and have light curves 3 times broader than SNe Ibc, but that the intrinsic shapes are similar. There are a number of SLSNe with particularly broad light curves, possibly indicating two progenitor channels, but statistical tests do not cleanly separate two populations. The general spectral evolution is also presented. Velocities measured from Fe II are similar for SLSNe and SNe Ibc, suggesting that diffusion time differences are dominated by mass or opacity. Flat velocity evolution in most SLSNe suggests a dense shell of ejecta. If opacities in SLSNe are similar to other SNe Ibc, the average ejected mass is higher by a factor 2-3. Assuming $\kappa=0.1\,$cm$^2\,$g$^{-1}$, we estimate a mean (median) SLSN ejecta mass of 10$\,$M$_\odot$ (6$\,$M$_\odot$), with a range of 3-30$\,$M$_\odot$. Doubling the assumed opacity brings the masses closer to normal SNe Ibc, but with a high-mass tail. The most probable mechanism for generating SLSNe seems to be the core-collapse of a very massive hydrogen-poor star, forming a millisecond magnetar.