2009ApJ...691..661H

Timmes et al. found that metallicity variations could theoretically account for a 25% variation in the mass of ⁵⁶Ni synthesized in Type Ia supernovae (SNe Ia), and thus account for a large fraction of the scatter in observed SN Ia luminosities. Higher-metallicity progenitors are more neutron rich, producing more stable burning products relative to radioactive ⁵⁶Ni. We develop a new method for estimating bolometric luminosity and ⁵⁶Ni yield in SNe Ia and use it to test the theory with data from the Supernova Legacy Survey. We find that the average ⁵⁶Ni yield does drop in SNe Ia from high-metallicity environments, but the theory can only account for 7%-10% of the dispersion in SN Ia ⁵⁶Ni mass, and thus luminosity. This is because the effect is dominant at metallicities significantly above solar, whereas we find that SN hosts have predominantly subsolar or only moderately above-solar metallicities. We also show that allowing for changes in O/Fe with the metallicity [Fe/H] does not have a major effect on the theoretical prediction of Timmes et al., so long as one is using the O/H as the independent variable. Age may have a greater effect than metallicity–we find that the luminosity-weighted age of the host galaxy is correlated with ⁵⁶Ni yield, and thus more massive progenitors give rise to more luminous explosions. This is hard to understand if most SNe Ia explode when the primaries reach the Chandrasekhar mass. Finally, we test the findings of Gallagher et al. that the residuals of SNe Ia from the Hubble diagram are correlated with host galaxy metallicity, and we find no such correlation.