HOST-GALAXY PROPERTIES OF 32 LOW-REDSHIFT SUPERLUMINOUS SUPERNOVAE
FROM THE PALOMAR TRANSIENT FACTORY
D. A. Perley
1,2
, R. M. Quimby
3,4
,L.Yan
5
, P. M. Vreeswijk
6
, A. De Cia
6,7
, R. Lunnan
2
, A. Gal-Yam
6
, O. Yaron
6
,
A. V. Filippenko
8
, M. L. Graham
8
, R. Laher
9
, and P. E. Nugent
8,10
1
Dark Cosmology Centre, Niels Bohr Institute, University of Copenhagen, Juliane Maries Vej 30, DK-2100 København Ø, Denmark; dperley@dark-cosmology.dk
2
Department of Astronomy, California Institute of Technology, MC 249-17, 1200 East California Blvd., Pasadena, CA 91125, USA
3
Department of Astronomy, San Diego State University, San Diego, CA 92182, USA
4
Kavli IPMU (WPI), UTIAS, The University of Tokyo, Kashiwa, Chiba 277-8583, Japan
5
Infrared Processing and Analysis Center, California Institute of Technology, Pasadena, CA 91125, USA
6
Department of Particle Physics and Astrophysics, Faculty of Physics, The Weizmann Institute of Science, Rehovot 76100, Israel
7
European Southern Observatory, Karl-Schwarzschild-Strasse 2, D-85748 Garching bei München, Germany
8
Department of Astronomy, University of California, Berkeley, CA 94720-3411, USA
9
Spitzer Science Center, California Institute of Technology, MC 314-6, Pasadena, CA 91125, USA
10
Lawrence Berkeley National Laboratory, 1 Cyclotron Rd., Berkeley, CA 94720, USA
Received 2016 April 27; revised 2016 July 13; accepted 2016 July 19; published 2016 October 3
ABSTRACT
We present ultraviolet through near-infrared photometry and spectroscopy of the host galaxies of all superluminous
supernovae (SLSNe) discovered by the Palomar Transient Factory prior to 2013 and derive measurements of their
luminosities, star formation rates, stellar masses, and gas-phase metallicities. We find that TypeI (hydrogen-poor)
SLSNe (SLSNe I) are found almost exclusively in low-mass (
*
<´
MM210
9
) and metal-poor ( 12+log
10
[O/
H]
<
8.4
) galaxies. We compare the mass and metallicity distributions of our sample to nearby galaxy catalogs in
detail and conclude that the rate of SLSNe I as a fraction of all SNe is heavily suppressed in galaxies with
metallicities
Z0.5
. Extremely low metallicities are not required and indeed provide no further increase in the
relative SLSN rate. Several SLSN I hosts are undergoing vigorous starbursts, but this may simply be a side effect
of metallicity dependence: dwarf galaxies tend to have bursty star formation histories. TypeII (hydrogen-rich)
SLSNe (SLSNe II) are found over the entire range of galaxy masses and metallicities, and their integrated
properties do not suggest a strong preference for (or against) low-mass/low-metallicity galaxies. Two hosts exhibit
unusual properties: PTF 10uhf is an SLSN I in a massive, luminous infrared galaxy at redshift z=0.29, while PTF
10tpz is an SLSN II located in the nucleus of an early-type host at z=0.04.
Key words: galaxies: abundances – galaxies: dwarf – galaxies: photometry – supernovae: general
Supporting material: machine-readable tables
1. INTRODUCTION
The recently discovered observational class of “super-
luminous” supernovae (SLSNe) has complicated what was
once a fairly straightforward view of the fates of massive stars
in the local universe, in which all stars above
M8
were
thought to explode via a common mechanism of iron core
collapse (see, e.g., Filippenko 1997, for a review). SLSNe have
characteristic peak visual absolute magnitudes between −21
and −22.5 (
~
L10 ;
10
Gal-Yam 2012; Nicholl et al. 2015),
making them much more luminous than typical core-collapse
supernovae (CCSNe), which peak between −15 and −18 mag
(
~
L10 10 ;
89
–
Richardson et al. 2002; Li et al. 2011.) Most
SLSNe also evolve much more slowly and have higher peak
temperatures than ordinary CCSNe, and the time-integrated
bolometric radiative output of an SLSN may reach 10
51
erg,
exceeding a typical CCSN by 2–3 orders of magnitude. This
points to a much larger progenitor mass and may require a
fundamentally different explosion mechanism.
The spectroscopic properties of SLSNe are diverse: they
include events showing strong hydrogen emission throughout
their observed evolution, events that show no hydrogen lines at
any epoch, and intermediate cases of weak and/or transient
hydrogen emission. Mirroring the classification scheme for
ordinary SNe, SLSNe are classified as Type I (no hydrogen
observed) or Type II (hydrogen observed); see Gal-Yam (2012)
for a review of SLSN classes and properties.
Events showing narrow or intermediate-width hydrogen
lines in their spectra (all of which are Type II by definition, and
which represent the majority of events in this class) are
simplest to accommodate physically, since the existence of
these lines is direct evidence of interaction between SN ejecta
and a dense surrounding medium (e.g., Moriya et al. 2013).
This process permits the bulk kinetic energy of the outflow to
be tapped and converted to electromagnetic radiation, helping
to explain the large radiative output of these events and easing
the fundamental energy requirements. Indeed, “ordinary” Type
IIn SNe (SNe IIn) are the most luminous class of CCSNe and
are thought to occur when SN ejecta collide with shells of
material from previous eruptions (e.g., Schlegel 1990; Filip-
penko 1997; Kiewe et al. 2012). The underlying mechanism in
SLSNe IIn is presumably directly analogous. Nevertheless, the
amount of kinetic energy that must be converted to radiation in
order to accommodate these events requires an extremely
massive circumstellar envelope and therefore a very large
initial mass (Smith et al. 2010; Chatzopoulos et al. 2013),
possibly within the range at which other evolutionary channels
beyond ordinary core collapse may become relevant.
Events lacking narrow hydrogen lines (including all SLSNe
I, but also any SLSNe II whose Balmer lines are broad and/or
weak) represent an even greater challenge for progenitor
models, since it is not clear whether interaction with a
circumstellar medium is available to ease the radiative energy
The Astrophysical Journal, 830:13 (31pp), 2016 October 10 doi:10.3847/0004-637X/830/1/13
© 2016. The American Astronomical Society. All rights reserved.
1
requirements. It is possible that the ejecta are interacting with a
dense, hydrogen-poor shell of previously ejected material (e.g.,
Chevalier & Irwin 2011; Ginzburg & Balberg 2012), although
this would imply a very large initial mass and again point
toward the possibility of exotic evolutionary or explosion
channels. Furthermore, the absence of observed narrow lines
from other elements is surprising. For SLSNe I, the stellar
progenitor must also rid itself of its hydrogen envelope during
its lifetime and yet retain sufficient mass at the time of death to
produce an explosion with
»
E
10
K
51
erg ejecting
MM10
ej
worth of heavy elements, a challenge for stellar
evolutionary theory.
While it is possible that either or both classes of transient
may simply constitute extrema of ordinary stellar evolution and
explode via core collapse, the remarkable observational
properties of SLSNe have sparked renewed interest in more
exotic explosion mechanisms. One well-established model of
particular theoretical interest is the pair-instability supernova
(PISN), an explosion produced when the temperature required
to maintain hydrostatic equilibrium in the core of a star
becomes so high that photons disintegrate into particle pairs
and the star collapses (Rakavy & Shaviv 1967; Barkat
et al. 1967). Such events should originate from the most
massive stars (
>
MM300 ;
init
Yoshida & Umeda 2011) and
are expected to produce enormous quantities of radioactive
nickel that could easily power an SLSN; at least one well-
known SLSN I (SN 2007bi) has been interpreted with
reasonable success within this model (Gal-Yam et al. 2009;
but see Dessart et al. 2012). In a variant on the pair-instability
mechanism, the pulsational PISN, the massive star undergoes
several incomplete pair-instability episodes, leading to a series
of envelope-shedding eruptions before the final explosion
(Woosley et al. 2007; Waldman 2008), naturally providing
both an intrinsically very energetic explosion and shells of
material for it to interact with. Problematically, however,
classical pair-instability models lead to very large masses of
56
Ni; this decays to
56
Co, whose much slower decay to
56
Fe
should produce a luminous exponential decay phase in the late-
time light curve. While evolution consistent with this has been
seen in a handful of cases (referred to as “Type R” SLSNe by
Gal-Yam 2012), the majority fade too fast to be explained by
this mechanism. High-mass, noninteracting core-collapse
models (e.g., Yoshida et al. 2014) also share this problem.
Some, and perhaps all, SLSNe may therefore require yet
another mechanism to reenergize the ejecta. If interaction and
radioactive decay are excluded as possibilities, the only
remaining power source capable of meeting the energy
requirements is the compact object itself, the so-called “central
engine.” The most popular central-engine model invokes a
spinning-down highly magnetic neutron star (“magnetar”) that
energizes the SN by winds and X-ray radiation from inside
(Mazzali et al. 2006; Kasen & Bildsten 2010; Woosley 2010;
Inserra et al. 2013; Metzger et al. 2015). Alternative central-
engine models include jet feedback from fallback accretion
onto the central neutron star or black hole (Gilkis et al. 2015;
Soker 2016).
The host-galaxy environments of SLSNe provide strong
constraints on progenitor models. For example, simple, single-
star pair-instability models predict that PISNe should be
produced only by stars with very low initial metallicity (Langer
et al. 2007). If this model is correct, these explosions should
not form in metal-rich environments. The energy-injection
model involves a rapidly rotating central engine similar in
nature to the central engine of gamma-ray bursts (GRBs; e.g.,
Usov 1992); if this model explains some or all SLSNe, then it
would be reasonable to expect similarities between the hosts of
SLSNe and the hosts of long-duration GRBs, which are
observed to avoid high-metallicity galaxies and occur pre-
dominantly at low to intermediate metallicity in the local
universe (Stanek et al. 2006; Modjaz et al. 2008; Graham &
Fruchter 2013; Krühler et al. 2015; Japelj et al. 2016; Perley
et al. 2016). Other models invoke dynamical interactions and
stellar mergers in dense environments (Pan et al. 2012; van den
Heuvel & Portegies Zwart 2013), which would favor
particularly intense starbursts. In any case, regardless of the
underlying theoretical model, the degree of similarity or
dissimilarity between the hosts of Type I versus Type II events
(or between subclasses of these events, or between these events
and other classes of SNe) might help establish whether these
explosions are closely related or fundamentally different.
The very fact that SLSNe were discovered only in the past
decade provides evidence that the sites of SLSNe might differ
from those of ordinary CCSNe. Prior to about 2005, all major
nearby SN searches—most notably, the Lick Observatory
Supernova Search (LOSS) with the Katzman Automatic
Imaging Telescope (KAIT; Filippenko et al. 2001)—were
targeted surveys, using small-field-of-view cameras to periodi-
cally image the positions of known galaxies. For reasons of
efficiency, nearby and relatively high mass galaxies were
preferentially targeted, rendering these searches insensitive to
transients that might occur preferentially or exclusively in
smaller systems ( unless discovered in the background).
However, starting about 10 yr ago, a number of wide-field
untargeted optical surveys began operation, providing the
capability to search much larger volumes of space in an
unbiased manner; these include the Texas Supernova Search
(which discovered the first widely recognized SLSNe, SN
2005ap and SN 2006gy), the Catalina Real-Time Survey
(Drake et al. 2009), the Palomar Transient Factory (PTF; Law
et al. 2009), Pan-STARRS (Kaiser et al. 2002), La Silla Quest
(Hadjiyska et al. 2012), SkyMapper (Keller et al. 2007), the
Dark Energy Survey (Dark Energy Survey Collaboration 2016),
and the All-Sky Automated Survey for Supernovae (ASAS-SN;
Shappee et al. 2014). A large fraction of the SLSNe reported by
these surveys originate from very faint galaxies (Neill
et al. 2011), undetected in pre-explosion images such as the
Sloan Digital Sky Survey (SDSS
). While in part this reflects the
great distances at which SLSNe are discovered, more detailed
analysis of SLSN host-galaxy samples suggests that they differ
intrinsically from the host populations of more ordinary SNe in
various ways: low masses and metallicities are typical (Chen
et al. 2013; Lunnan et al. 2013, 2014; Angus et al. 2016), and
galaxies with exceptionally strong emission lines are remark-
ably frequent ( Leloudas et al. 2015).
Among these surveys, the PTF has been the most prolific
discoverer of SLSNe: the sample of 32 SLSNe discovered in
2009–2012 that we present here (Section 2) is comparable in
size to the sample of publicly released SLSNe from all other
surveys combined. Furthermore, all of these events occurred at
relatively low redshifts (
<
z
0.51
), so in all cases the SN and
host are relatively accessible to comprehensive follow-up
observations. In contrast, Pan-STARRS, the next most prolific
individual survey with a published SLSN sample, has
2
The Astrophysical Journal, 830:13 (31pp), 2016 October 10 Perley et al.
discovered most of its SLSNe at significantly greater distances
(
<<z
0
.5 1.6
from the sample of Lunnan et al. 2014).
In complementary papers we will be presenting the entire
suite of observations of the PTF SLSN sample, including
details of the discovery and sample selection, spectroscopic
properties (Leloudas et al. 2016; R. Quimby et al. 2016, in
preparation), and multiband light curves (A. De Cia et al. 2016,
in preparation). In this work we present observations of the host
galaxies of these events from an extensive ground- and space-
based campaign, effectively doubling the sample of well-
studied SLSN hosts and providing the first large, homo-
geneous, single-survey sample in the local universe.
The paper is organized as follows. In Section 2 we overview
the operations in PTF leading to successful discovery and
classification of SLSNe and outline the selection of our sample.
Our observations are described in Section 3, including
ultraviolet (UV), optical, and near-infrared (NIR) photometry
and spectroscopy from Keck and Palomar supplemented by
Hubble Space Telescope (HST) and Spitzer imaging; we also
summarize our analysis techniques used to provide measure-
ments of physical parameters such as mass, star formation rate
(SFR), and metallicity using these observations. The host
galaxies are discussed on an individual basis in Section 4.In
Section 5 we examine our SLSN sample as an ensemble and
compare the physical properties of the population against those
of volume-limited star-forming field-galaxy samples. We
discuss our results and their implications in Sections 6 and 7.
2. SAMPLE SELECTION
2.1. PTF Discovery of Supernovae
The current public literature sample of SLSNe and SLSN
hosts (see, e.g., Leloudas et al. 2015; Angus et al. 2016) is
combined from a large variety of different surveys, each of
which contributes only a few events to the overall total.
11
Many
of these discoveries were based on archival reanalysis of earlier
surveys (such as SDSS; Leloudas et al. 2012) to recover events
that were not recognized to be SLSNe at the time.
Consequently, the existing sample of low-z SLSNe is quite
heterogeneous in construction, and the biases that may affect
the nature of the cataloged population are nontrivial. In
contrast, the PTF sample we present here was discovered by
a single survey using a single camera and telescope and via
(typically) the same group of scientists and follow-up
resources. Nevertheless, PTF is a complex effort, and its
cadence, motivations, and emphasis have varied substantially
since its inception, so the sample presented here is subject to its
own biases and incompletenesses. Discussion of possible
biases related to these factors will be presented in Section 6.1.
We provide a brief summary of the survey and its operations
below.
The PTF is a synoptic optical survey using the 48-inch
Oschin Schmidt Telescope (P48) at the Palomar Observatory
near San Diego, California, and a 7.2 deg
2
camera (Rahmer
et al. 2008). Observations of the sky are acquired every night
during clear weather, except within a few days of full moon
each month when Hα survey observations are performed. PTF
operated between 2009 and 2012, and although the facility is
continuing operations as the intermediate Palomar Transient
Factory (iPTF) until the end of 2016, this paper exclusively
addresses events discovered during the original 4 yr period.
PTF employs both R- and g-band filters, but prior to 2013 the
large majority of the survey was conducted in R, and all of the
SLSNe presented here were discovered in R.
The survey discovers far more transient events than can be
observed spectroscopically: the PTF database reports 19,595
likely transients discovered in 2009–2012, of which only 2131
(11%) have secure classifications. Human oversight is
necessary at several stages in the process to choose
astrophysically real and scientifically interesting targets for
follow-up observations. All objects found by the automated
detection and verification pipelines (Brink et al. 2013) are
screened by human scanners to confirm their astrophysical
nature and rule out nontransient false positives (cosmic rays,
poor subtractions, asteroids, and variable sources). At the time
of discovery, the scanner may choose to nominate an object for
follow-up spectroscopy. Objects may also be nominated later,
as further data are collected. Weather permitting, these are then
targeted at the next observing run (usually 1–2 runs occur
monthly during dark time). Spectra are reduced within a few
days of being acquired, and a preliminary classification is
established either visually or via standard classification
routines; events with unclear or ambiguous classifications
are flagged for reobservation with higher signal-to-noise ratio
(S/N) or at later epochs. Classifications are revisited at later
times once all data are in hand.
2.2. Definition and Identification of SLSNe
The class of SLSNe is necessarily defined via a combination of
photometric and spectroscopic qualities: to qualify, an event must
be clearly an SN (usually implying the detection of broad features
in the spectrum, as well as an SN-like light curve and color
evolution) and also must be much more luminous than ordinary
SNe (it must be “super” luminous). Beyond this there is no
standard definition of what observables are required to establish
what is or is not an SLSN. An absolute magnitude limit of
<-M 2
1
at peak was adopted by Gal-Yam (2012), but this
choice is empirical and somewhat arbitrary (it is also wavelength
dependent). Furthermore, several SNe with properties very
similar to those of SLSNe I in particular (in terms of colors,
light curves, spectra, and total radiative output) do not quite reach
this luminosity, while a small number of SNe that are probably
not related to massive stars at all (specifically the Type Ia-CSM
SNe; Silverman et al. 2013) occasionally do surpass it.
The task of defining SLSNe in a physically meaningful way,
as well as the isolation of all events within PTF satisfying that
definition, is therefore quite complicated. A detailed analysis of
this topic will be deferred to the upcoming dedicated works of
R. Quimby et al. (2016, in preparation), A. De Cia et al. (2016,
in preparation), and G. Leloudas et al. (2016, in preparation)
including a presentation of all spectra and light curves. For the
purposes of this paper, we establish our own working definition
of SLSNe in the PTF sample as follows.
We require, at minimum, an absolute magnitude of
<-M 20.0
R
at peak to consider inclusion of an event in our
sample. This guarantees that every event in our sample is
indeed very luminous and eliminates the vast majority of
ordinary SNe in the PTF sample. Circumstellar interaction is
capable of significantly boosting the luminosity of all types of
SNe (Ofek et al. 2014); indeed, SNe IIn have in particular been
known since the 1980s to exceed this threshold on occasion
11
The relatively large Pan-STARRS sample of 15 events presented by
McCrum et al. (2015) and Lunnan et al. (2014) is an exception, but it probes a
higher and more difficult-to-study redshift range.
3
The Astrophysical Journal, 830:13 (31pp), 2016 October 10 Perley et al.
(Richardson et al. 2002). We therefore apply a more stringent
cut if narrow hydrogen lines are present, requiring
<-M 20.5
R
at peak. (A few events were discovered after peak brightness,
and one is heavily extinguished by host-galaxy dust. In these
cases peak magnitudes require extrapolations or corrections;
see Section 2.4.)
Many of the most luminous transient candidates identified by
PTF turn out to not be SNe: active galactic nuclei (AGNs) and
quasi-stellar objects (QSOs) are particularly common. Most
such objects can be easily eliminated from consideration based
on their past or continued variability or via spectroscopy;
alternatively, an off-nuclear location or a smoothly rising and
falling light curve with blue-to-red spectral evolution usually
provides good evidence that a transient is an SN and not an
AGN. Even so, SLSNe II can look spectroscopically similar at
certain phases to narrow-line AGNs (as can normal SNe IIn;
e.g., Filippenko 1989), and in cases where photometric and
spectroscopic coverage of the SN is poor it is not always easy
to completely rule out an AGN flare. For two events in our
sample (PTF 09uy and PTF 11dsf), we favor an SLSN
interpretation but note that an AGN has not been fully
eliminated (see also the discussion of PTF 10tpz in Section 6.3).
These classifications will be further investigated and discussed
by G. Leloudas et al. (2016, in preparation).
Tidal disruption events (TDEs) represent another, less
frequently observed class of phenomena associated with
accretion onto supermassive black holes. These typically
exhibit peak magnitudes around −19 but can occasionally be
brighter than −20 (Arcavi et al. 2014). The spectroscopic and
photometric properties of TDEs and SLSNe are usually distinct
—and while ambiguous cases can arise especially at the high-
luminosity end (Chornock et al. 2014; Leloudas et al. 2017;
G. Duggan et al. 2016, in preparation; see also the last
paragraph of Section 7), these are particularly rare, and we
identify no such cases within the 4 yr PTF sample
covered here.
The most luminous SNe Ia exceed −20 mag but are easily
identified spectroscopically. Type Ia-CSM SNe can be even
brighter (occasionally, even
<
-21
mag), but as photospheric
SNIa features are still evident, these can similarly be identified
spectroscopically (Silverman et al. 2013). In the course of this
analysis we identified several new SNeIa-CSM within the PTF
sample that will be reported in separate work.
Other types of luminous transients are also known to exist
whose connection to SNe is not yet clear, in particular the fast-
rising transients of Arcavi et al. (2016) and Drout et al. ( 2014).
With the exception of the single event already identified by
Arcavi et al. (2016),wefind no further members of these
classes in our sample.
In total, 32 events satisfy all of the above criteria and
constitute the PTF SLSN sample. All show behavior
characteristic of SNe, including broad spectral lines, evolu-
tionary timescales of months, and blue-to-red spectral evolution
in cases where multiband data are available.
2.3. Subclassification of SLSNe
SLSNe within the sample are then subcategorized spectro-
scopically as Type “I” or “II” based on the absence or presence
(respectively) of hydrogen in their spectra. While in principle
this is a straightforward distinction, it conceals some complex-
ity. For example, a few SLSNe show no hydrogen in any of
their early-time spectra but then develop broad hydrogen lines
at late times (Gezari et al. 2009; Miller et al. 2009; Benetti
et al. 2014; Yan et al. 2015). Even for events that do exhibit
hydrogen emission in all existing spectra, this emission can
sometimes be relatively weak and/or exhibit no narrow
component. While a strictly literal interpretation would classify
these events as Type II, some of them may quite plausibly be
physically more closely related to SNe I (or represent an
intermediate case or another class entirely; see also Inserra
et al. 2016).
In spite of these occasional ambiguities, the “I” versus “II”
distinction is sufficient for the vast majority of events in our
sample: nearly all events without hydrogen at the time of
discovery never show hydrogen in any follow-up spectra, and
nearly all events with hydrogen exhibit strong emission lines at
all phases including a narrow component. The possible
exceptions include PTF 10aagc (Type I with ambiguous, weak,
late-time broad hydrogen), PTF 10uhf (Type I, but with a
possible faint signature of broad Balmer emission that is
difficult to disentangle from the host [N
II] emission), and PTF
12gwu (Type II, but the hydrogen lines are much weaker than
in the rest of our sample and no obvious narrow component is
present). For this paper, we maintain the initial, conservative
classifications of these events from the presence or absence of
unambiguous hydrogen in their discovery spectra.
Among the SLSNe I, we denote a small number of events
(three) as belonging to the subclass of long-lived “R”-types,
which show exponentially declining late-time light curves
consistent with radioactive decay and which have been
suggested to be examples of PISNe (Gal-Yam et al. 2009;
Gal-Yam 2012), although this interpretation is contested by
other authors (e.g., Dessart et al. 2012; Nicholl et al. 2013;
Jerkstrand et al. 2016). We will generally refer to them as Type
“I-R.” With only three events, we do not have sufficient sample
size to perform a statistically robust comparison between the
host properties of these events and the more rapidly declining
SNeI, but as we observe no strong distinction between the host
properties of these events and other SLSNe I in our sample
(and other authors have reported similar results; e.g., Lunnan
et al. 2014; Leloudas et al. 2015), we will generally treat all
SLSNe I together in our analysis regardless of their light-curve
properties.
2.4. Sample Properties
The sample is summarized in Table 1. In total, we present 14
events of Type II and 18 events of Type I, only three of the
latter being Type R. The large majority of these objects have
not been previously presented in the literature.
Two of the SLSNe are noteworthy from the point of view of
sample selection. PTF 10vwg is in a crowded low-Galactic-
latitude field and, while detected in PTF survey images, was
not identified as a transient candidate until it was discovered in
the background of an LOSS/KAIT image (Kodros et al. 2010)
—so it is not truly a PTF object. (Stellar confusion and high
foreground extinction also introduce severe complications in
characterizing the host.) PTF 10tpz is near the nucleus of an
early-type galaxy, and the SN spectrum is highly reddened
owing to host extinction. Without an extinction correction it
would not be superluminous, but depending on the (highly
uncertain)
host column, it is probably close to or above our
threshold. Were it not for the known example of SN 2006gy,
which occurred under similar circumstances (e.g., Ofek et al.
2007; Smith et al. 2007), this event would likely not have been
4
The Astrophysical Journal, 830:13 (31pp), 2016 October 10 Perley et al.
categorized as an SLSN. In any case, the considerations
involving its discovery are quite different than for other PTF
events (a much smaller effective detection volume because of
the large extinction, plus more complex issues involving host
subtraction and AGN contamination), so it should not be
treated with statistical weight equal to the others. We will
include the hosts of these events in our plots and analysis where
possible, but we emphasize that they would be excluded from
any attempt to produce a statistically uniform sample from
these data.
3. OBSERVATIONS
3.1. Ground-based Imaging
The SLSNe in the sample above were targeted for late-time
imaging in a variety of wavebands spanning the near-UV to the
NIR. Several host galaxies were bright enough to be well
recovered in SDSS (at least in the gri filters),sowe
downloaded the processed survey images from the SDSS
archive (Alam et al. 2015). For fainter hosts and for other filters
we observed with other facilities: the Palomar 60-inch (P60)
telescope imaging camera (Cenko et al. 2006), the Large
Format Camera or the Wide-Field Infrared Camera ( WIRC) on
the Palomar 5-m Hale telescope, or the Low-Resolution
Imaging Spectrometer (LRIS; Oke et al. 1995) or the Multi-
Object Spectrometer for Infrared Exploration (MOSFIRE;
McLean et al. 2012) on the Keck I 10 m telescope.
All ground-based images used for the host-galaxy spectral
energy distribution (SED) analysis were acquired either pre-
explosion or long after the SN peak time (at least 2 yr, typically
3–4yr). Nevertheless, SLSNe exhibit a range of light-curve
behaviors and decay times, and there is no guarantee based on
the time difference alone that the SN is not contributing light.
We use various checks appropriate to the situation to rule out
significant SN contributions: (1) direct confirmation based on a
reference-subtracted image that the SN was much fainter than
the host well prior to the observation in question, (2) verifying
nonvariability between widely spaced epochs, (3) absence of
any SN-like broad features in high-S/N contemporaneous (
or
earlier) spectra, or detection of features whose contribution to
Table 1
Superluminous Supernovae from PTF
PTF ID α
a
δ
a
Class
b
z
M
,
V peak
c
t
peak
c
-
E
BV
d
Notes
09as 12:59:15.864 +27:16:40.58 I 0.1867 −20.8 2009 Mar 24 0.008 L
09uy 12:43:55.771 +74:41:07.58 II 0.3145 −21.2 2009 Jul 08 0.020 L
09atu 16:30:24.553 +23:38:25.43 I 0.5015 −22.5 2009 Aug 18 0.042 L
09cnd 16:12:08.839 +51:29:16.01 I 0.2584 −23 2009 Sep 10 0.021 L
09cwl 14:49:10.08 +29:25:11.4 I 0.3499 −22.5 2009 Aug 07 0.014 =SN 2009jh
10bfz 12:54:41.288 +15:24:17.08 I 0.1701 −20.9 2010 Jan 31 0.018 L
10bjp 10:06:34.30 +67:59:19.0 I 0.3584 −21.4 2010 Feb 16 0.055 L
10cwr 11:25:46.73 −08:49:41.9 I 0.2297 −21.8 2010 Mar 21 0.035 =SN 2010gx
10fel 16:27:31.103 +51:21:43.45 II 0.2356
<
-20.
5
<2010 Apr 03 0.017 L
10heh 12:48:52.05 +13:26:24.5 II 0.3379 −21.2 2010 Jun 03 0.024 L
10hgi 16:37:47.074 +06:12:31.83 I 0.0987 −20.3 2010 Jun 20 0.074 =SN 2010md
10jwd 16:43:43.325 +44:31:43.8 II 0.477 −21.4 2010 Jul 02 0.012 L
10nmn 15:50:02.809 −07:24:42.38 I-R 0.1237 −20.5 2010 Jul 07 0.138 L
10qaf 23:35:42.887 +10:46:32.57 II 0.2836 −21.6 2010 Aug 05 0.070 L
10qwu 16:51:10.572 +28:18:07.62 II 0.2258 −21.0 2010 Aug 21 0.040 L
10scc 23:28:10.495 +28:38:31.10 II 0.242 −21.5 2010 Aug 26 0.093 L
10tpz 21:58:31.74 −15:33:02.6 II 0.0395
- 19
2010 Sep 02 0.041 Heavily extinguished
10uhf 16:52:46.696 +47:36:21.76 I 0.2882 −22 2010 Sep 18 0.018 Possible very weak Hα?
10vqv 03:03:06.859 −01:32:35.42 I 0.4518 − 22.5 2010 Oct 13 0.061 L
10vwg 18:59:32.881 +19:24:25.74 I-R 0.1901 −21 2010 Sep 07 0.467 =SN 2010hy; KAIT/LOSS discovery
10yyc 04:39:17.297 −00:20:54.5 II 0.2147 −21 2010 Nov 13 0.041 L
10aagc 09:39:56.923 +21:43:17.09 I 0.206 −20.4 2010 Oct 04 0.022 Late-time hydrogen lines?
11dij 13:50:57.798 +26:16:42.44 I 0.1428 −21.5 2011 Apr 28 0.011 =SN 2011ke
11dsf 16:11:33.55 +40:18:03.5 II 0.3848 −22.1 2011 May 27 0.009 L
11hrq 00:51:47.22 −26:25:10.0 I 0.057
<
-20
<2011 Jul 11 0.012 L
11rks 01:39:45.528 +29:55:27.43 I 0.1924 −21.1 2012 Jan 11 0.038 L
12dam 14:24:46.228 +46:13:48.64 I-R 0.1073 −21.5 2012 Jun 12 0.100 L
12epg 12:55:36.596 +35:37:35.79 II 0.3422 −21.3 2012 May 30 0.015 L
12gwu 15:02:32.876 +08:03:49.47 II 0.275 −21.4 2012 Jul 25 0.033 Hydrogen lines very weak
12mkp 08:28:35.092 +65:10:55.60 II 0.153 −21.0 2013 Jan 25 0.046 L
12mue 03:18:51.072 −11:49:13.55 II 0.2787 −21.4 2012 Dec 21 0.062 L
12mxx 22:30:16.728 +27:58:22.01 I 0.3296 −22.5 2012 Dec 10 0.041 L
Notes.
a
Supernova position (J2000).
b
Supernova classification.
c
Approximate peak visual magnitude of the supernova and corresponding UT date. More refined measurements will be presented by A. De Cia et al. (2016, in
preparation).
d
Galactic (foreground) selective extinction in magnitudes; from Schlafly & Finkbeiner (2011).
(This table is available in machine-readable form.)
5
The Astrophysical Journal, 830:13 (31pp), 2016 October 10 Perley et al.
