2008ApJ...681..462D

We use the HST ACS imaging of the two GOODS fields during Cycles 11, 12, and 13 to derive the Type Ia supernova rate in four redshift intervals in the range 0.2<z<1.8. Compared to our previous results based on Cycle 11 observations only, we have increased the coverage of the search by a factor 2.7, making the total area searched equivalent to 1 deg\ul 2. The sample now consists of 56 Type Ia supernovae. The rates we derive are consistent with our results based on the Cycle 11 observations. In particular, the small number of supernovae detected at z>1.4 supports our previous result that there is a drop in the Type Ia supernova rate at high redshift, suggesting a long time delay between the formation of the progenitor star and the explosion of the supernova. If described by a simple one-parameter model, we find a characteristic delay time of 2-3 Gyr. However, a number of recent results based on, e.g., low-redshift supernova samples and supernova host galaxy properties suggest that the supernova delay-time distribution is bimodal. In this model a major fraction of the Type Ia supernova rate are ``prompt'' and follow the star formation rate, while a smaller fraction of the rate have a long delay time, making this channel proportional to mass. While our results are fully consistent with the bimodal model at low redshifts, the low rate we find at z>1.4 appears to contradict these results. Models that correct for star formation hidden by dust may explain at least part of the differences. Here we discuss this possibility together with other ways to reconcile data and models.