Time Dilation from Spectral Feature Age Measurements of Type IA Supernovae.

Abstract

We have developed a quantitative, empirical method for estimating the age of Type la Supernovae (SNe la) from a single spectral epoch. The technique examines the goodness of fit of spectral features as a function of the temporal evolution of a large database of SNe la spectral features. When a SN la spectrum with good signal-to-noise ratio over the rest frame range 3800 to 6800 Â is available, the precision of a spectral feature age (SFA) is (lo-)~ 1.4 days. SFA estimates are made for two spectral epochs of SN 1996bj (z = 0.574) to measure the rate of aging at high redshift. In the 10.05 days which elapsed between spectral observations, SN 1996bj aged 3.35± 3.2 days, consistent with the 6.38 days of aging expected in an expanding Universe and inconsistent with no time dilation at the 96.4% confidence level. The precision to which individual features constrain the supernova age has implications for the source of inhomogeneities among SNe la. © 1997 American Astronomical Society. [S0004-6256(97)00808-X] 1. TYPE la SUPERNOVAE AS CLOCKS A beguiling prediction of an expanding Universe is that distant objects will appear to age at a slower rate than nearby ones. Type la Supernovae (SNe la) provide extragalactic clocks of unparalleled precision which are sufficiently luminous to reveal this remarkable phenomenon. While a few doubt that expansion alone causes cosmological redshifts (e.g., Arp 1987, 1994; Arp et al 1990; Narlikar & Arp 1993), the conventional interpretation has only modest experimental verification (Sandage & Perelmuter 1991). Initial suggestions that time dilation might be seen in the photometric history of SNe la (Wilson 1939; Rust 1974) have been confirmed with recent observations of high red-shift SN la light curves (Leibundgut et al. 1996; Goldhaber et al. 1997). Specifically, Leibundgut et al. (1996) demonstrated that the broad light curve of SN la 1995K (z =0.479) was consistent with those of nearby SNe la when dilated by a factor (1 + z) as prescribed by cosmological expansion. Yet recent work has shown that there is an intrinsic variation in the breadth of SN la light curves which is related to the peak luminosity of the supernova (Phillips 1993; Riess et al. 1995, 1996; Hamuy et al. 1995, 1996a,b). The sense of the correlation is that more luminous SNe la give rise to broader light curves. A skeptic might postulate that within the large volume searched at high redshift, we are naturally selecting the intrinsically brightest and broadest light curves ever seen. Although observations of the SN la spectra can be used to eliminate such objects (Nugent et al