MEASURING EJECTA VELOCITY IMPROVES TYPE Ia SUPERNOVA DISTANCES

We use a sample of 121 spectroscopically normal Type Ia supernovae (SNe Ia) to show that their intrinsic color is correlated with their ejecta velocity, as measured from the blueshift of the Si II lambda 6355 feature near maximum brightness, nu(Si) (II). The SN Ia sample was originally used by Wang et al. to show that the relationship between color excess and peak magnitude, which in the absence of intrinsic color differences describes a reddening law, was different for two subsamples split by nu(Si) (II) (defined as Normal and High Velocity). We verify this result, but find that the two subsamples have the same reddening law when extremely reddened events (E(B - V) > 0.35 mag) are excluded. We also show that (1) the High-Velocity subsample is offset by similar to 0.06 mag to the red from the Normal subsample in the (B-max - V-max)-M-V plane, (2) the B-max - V-max cumulative distribution functions of the two subsamples have nearly identical shapes, but the High-Velocity subsample is offset by similar to 0.07 mag to the red in B-max - V-max, and (3) the bluest High-Velocity SNe Ia are similar to 0.10 mag redder than the bluest Normal SNe Ia. Together, this evidence indicates a difference in intrinsic color for the subsamples. Accounting for this intrinsic color difference reduces the scatter in Hubble residuals from 0.190 mag to 0.130 mag for SNe Ia with AV less than or similar to 0.7 mag. The scatter can be further reduced to 0.109 mag by exclusively using SNe Ia from the Normal subsample. Additionally, this result can at least partially explain the anomalously low values of R-V found in large SN Ia samples. We explain the correlation between ejecta velocity and color as increased line blanketing in the High-Velocity SNe Ia, causing them to become redder. We discuss some implications of this result, and stress the importance of spectroscopy for future SN Ia cosmology surveys, with particular focus on the design of WFIRST.