Superluminous Supernovae


Superluminous supernovae are hydrogen-rich (SLSNe-II), or hydrogen-poor (SLSNe-I), explosions so bright that they require a power source beyond that of traditional supernovae. SLSNe-I rise to a peak over 20-90 days, then decline over a timescale roughly twice as long. At early times they have a blue continuum, peaking in the ultraviolet, have temperatures in excess of 14,000 K, and show ionized lines of carbon and oxygen out of thermodynamic equilibrium. As the supernovae cool, their spectra start to resemble SNe Ic, though with a time delay. They also favor environments with metallicities half solar or lower. Modeling indicates that they are explosions of stripped carbon oxygen stellar cores, similar to, but sometimes more massive than the progenitors of SNe Ic. SLSNe-I similar to SN 2007bi have broader lightcurves, and seemingly more massive progenitors. Some have proposed that these are pair instability supernovae, but in general the supernovae rise too quickly for this model. Most SLSNe-I show no signs of interaction, and instead seem to be powered by a central engine. The magnetar spin-down model has been the most successful at reproducing the lightcurves and peak luminosity of SLSNe, though it may not be unique. Most SLSNe-II seem to be powered by interaction of these SNe with circumstellar material, as in SNe IIn. However, there are a handful of hybrid cases, or SLSNe-II, with weak or little interaction, which may be related to SLSNe-I.


Superluminous supernovae (SLSNe), as the name attests, are supernovae that are brighter than usual. As a result of the overly broad name, the category is a catch-all describing several classes of supernovae – some with hydrogen, some without, some interacting, some probably not. A few authors have defined SLSNe as those brighter than \(M=-21\) at peak, though this arbitrary cut could leave out related physical phenomena. Instead, I define SLSNe as luminous SNe which cannot be explained by the power sources fueling traditional (Types I and II) supernovae: radioactive decay from a moderate amount of elements synthesized in the explosion, the energy deposited by a shock unbinding the star, or interaction with moderate but obvious amounts of circumstellar material (CSM) previously lost by the supernova progenitor or a companion.

This last point creates a gray area. Should Type IIn supernovae count as SLSNe? Type IIn supernovae are those with a strong blue continuum at early times, and narrow and intermediate width hydrogen emission lines at some points in their spectroscopic evolution. They are thought to be the collapse of massive stars whose ejecta shock CSM. On one hand, they have been recognized as a class since the 1980s, a large and diverse one, and the source of their luminosity is not a mystery. On the other, some SNe IIn are so bright that they have been considered SLSNe [e.g. SN 2006gy (Smith et al., 2007; Ofek et al., 2007), which reached a peak absolute magnitude of \(-22\)]. A complicating factor is that interaction should be considered as a possible power source for SLSNe, whether or not the spectra show narrow lines. Here I compromise – I will generally not include clear SNe IIn, as their power source is not a mystery. However, I will mention a few extraordinary cases where appropriate, and discuss interaction as a possible power source.

In the late 90s a few unusually luminous supernovae were discovered — SN 1997cy (Germany 2000, \(M=-20\)) was interacting, SN 1999as (Knop 1999) was a Type Ic supernova which peaked well above the typical absolute magnitude, and 1999bd (Nugent et al., 1999) was an early and relatively unrecognized SLSN-II. However these events were largely explained away as either something possibly related to hypernovae (in the case of SN 1999as), or a SN IIn (SN 1997cy and SN 1999bd).

The history of SLSNe starts in earnest with two discoveries by Robert Quimby and the Texas Supernova Search who first established SNe with individual luminosities so extreme that they clearly represented a crisis to typical models. SN 2006gy (Smith et al., 2007), a SN IIn, was brighter than \(M=-21\) for about 100 days, and radiated more than \(10^{51}\) ergs. SN 2005ap (discovered first, but published second in Quimby et al., 2007) was what we now call a SLSN-I, discovered at z=0.283. It peaked at an unfiltered absolute magnitude of -22, unheard-of at the time, and showed a band of five O II features, which is today recognized as a signature of the class.

This was followed by SN 2008es, another superluminous Type II discovered by ROTSE-IIIb in a dwarf galaxy at z=0.205 (Gezari et al., 2008; Miller et al., 2008). It had a lightcurve similar to a SN II with a linear decline (SN II-L), but was more than an order of magnitude more luminous, reaching \(M_{V}=-22.2\).

Meanwhile, Barbary et al. (2008) described SCP06F6, a transient in the Supernova Cosmology Project’s cluster supernova search, which had an unknown redshift, three mysterious absorption lines, an extraordinarily long rise and decline, and slowly evolving spectra. The Supernova Legacy Survey identified two similar supernovae, which were talked about at conferences, but unpublished until their status as SLSNe was determined (Howell et al., 2013). The status of all these events was made clear with the discovery by Quimby et al. (2011) of events in the Palomar Transient Factory (PTF) sample that had O II lines like SN 2005ap, yet a redshift high enough to reveal restframe UV lines like the SCP06F6 and SNLS events. They also found weak host galaxy Mg II lines in an SCP06F6 spectrum placing it at \(z=1.189\), thereby cementing its status as a SLSN.

After people knew what to look for, SLSNe were found in many surveys. Pastorello et al. (2010) presented SN 2010gx from Pan-STARRS1 (PS1; aka CSS100313: 112547-084941, and PTF10cwr). They established that these hydrogen poor SLSNe started to look like SNe Ic at later times, albeit with a time delay. More were found in PS1 (e.g. Chomiuk et al., 2011; Berger et al., 2012). And a SN from SDSS-II established the first solid evidence for a bump in the lightcurve at early times (Leloudas et al., 2012).

Parallel to this, SN 2007bi was discovered (Gal-Yam et al., 2009), a superluminous supernova with a long, slow decline matching the decay of \({}^{56}\)Co, and a nebular spectrum interpreted as showing evidence for a large synthesized mass of \({}^{56}\)Ni. Thus the SN was interpreted as the first evidence for a Pair Instability Supernova (PISN; Barkat et al., 1967; Rakavy et al., 1967).

This early work was reviewed by Gal-Yam (2012), who divided SLSNe into three classes: SLSNe-I (those without hydrogen; §\ref{sec:hpoor}), SLSNe-II (those with hydrogen; §\ref{sec:hrich}), SLSNe-R (where R stands for radioactive, i.e. those like SN 2007bi). I keep all three classes, but rather than use the ‘R’ category (which relies on possibly incorrect theoretical interpretation), I call the latter SN 2007bi-likes (§\ref{sec:2007bi}). In addition, I’ll mention intermediate events, including a new class of fast rising luminous transients (§\ref{sec:intermediate}).