This paper started out as a fishing expedition. For the last few years, I have observed Type Ia supernovae (SNe Ia) with the Hubble Space Telescope (HST) when they are >500 days past maximum light. In parallel, Adam Riess has been using HST to calibrate the SN Ia distance ladder by observing variable stars in nearby SN host galaxies. Most of the SNe in these galaxies were years old, but some happened to still be visible. Since he knew about my interest in late-time observations of SNe Ia, he invited me to see what I could do with one of the SNe in his data, SN 2017erp. The data for SN 2017erp would only range between ~200 and 500 days, but all of my previous observations were in the optical, so I was curious to see whether the SN would behave similarly in the near-infrared F160W (H) filter.

It did not. Instead of declining in brightness, the light curve stayed pretty flat for nearly 300 days. I collected four more SNe Ia from Adam’s various HST programs (Figure 1). One of them - SN 2018gv, observed at similar phases as SN 2017erp - showed the same F160W plateau.

Based on sparse data, three previous studies also noted hints of a plateau. However, each of those SNe had been observed at different phases, and none of the studies had compared their observations to previous ones. Luckily, SNe 2017erp and 2018gv spanned the entire length of the plateau. Taken together with the other three SNe from our study, and as many SNe as I could find in the literature, the full story of the plateau emerged (Figure 2). SNe Ia transition onto a plateau phase in the near-infrared at ~150 days past maximum. The plateau lasts for nearly a year and, at ~500 days, the SNe fall off the plateau into a second decline phase.

While discovering the plateau was a fishing expedition, analyzing it was a collaborative effort. Russell Ryan reduced an HST spectrum of SN 2017erp at ~600 days that, when compared to similar spectra of the SN acquired by Kate Maguire, indicated that the transition off the plateau could be caused by a change in the ionization state of the SN ejecta (Figure 3). Matt Nicholl reduced spectra of SN 2014J that showed that the plateau was not observed in the K band. Arturo Avelino measured the light curve properties of the SNe in our sample, from which we noticed correlations between the features of the plateau and the luminosities of the SNe. And Luke Shingles, Ivo Seitenzahl, and Robert Fisher provided the theoretical analysis of the plateau, noting that it could be due to scattering of photons from the UV to longer wavelengths.

Do all types of SNe Ia exhibit this plateau? Are SNe Ia distributed along the plateau in a continuous fashion or are there “bright” and “dim” plateau SNe? Are there other possible explanations for the existence of the plateau? We hope that these, and many other questions, will be sparked by our paper.