DES13S2cmm: The First Superluminous Supernova from the Dark Energy Survey

1.3 MB·English

by  A. Papadopoulos

Mon.Not.R.Astron.Soc.000,000–000(0000) Printed2February2015 (MNLATEXstylefilev2.2)
DES13S2cmm: The First Superluminous Supernova from the Dark
Energy Survey
5
1 A. Papadopoulos1⋆, C. B. D’Andrea1, M. Sullivan2, R. C. Nichol1, K. Barbary3, R.
0
2 Biswas4, P. J. Brown5, R. A. Covarrubias6,7 D. A. Finley8, J. A. Fischer9, R. J. Foley7,10,

n D. Goldstein11,12, R. R. Gupta4, R. Kessler13,14, E. Kovacs4, S. E. Kuhlmann4, C.
a
J Lidman15, M. March9, P. E. Nugent11,12, M. Sako9, R. C. Smith16, H. Spinka4, W.

0 Wester8, T. M. C. Abbott16, F. Abdalla17, S. S. Allam6,18, M. Banerji17, J. P. Bernstein4,
3
R. A. Bernstein19, A. Carnero20,21, L. N. da Costa20,21, D. L. DePoy5, S. Desai22,23,

]
E H. T. Diehl8, T. Eifler24, A. E. Evrard25,26, B. Flaugher8, J. A. Frieman8,13, D. Gerdes25,
H D. Gruen27,28, K. Honscheid29, D. James16, K. Kuehn15, N. Kuropatkin8, O. Lahav17,
.
h M. A. G. Maia20,21, M. Makler30, J. L. Marshall5, K. W. Merritt8, C. J. Miller25,26, R.
p
- Miquel31,32, R. Ogando20,21, A. A. Plazas33 N. A. Roe12, A. K. Romer34, E. Rykoff35, E.
o
r Sanchez36, B. X. Santiago21,37, V. Scarpine8, M. Schubnell25, I. Sevilla35, M. Soares-
t
s Santos8, E. Suchyta29, M. Swanson6, G. Tarle25, J. Thaler10, D. L.Tucker8, R. H.
a
[ Wechsler38, J. Zuntz39

2
v
2 2February2015
3
2
7
ABSTRACT
0
We present DES13S2cmm, the first spectroscopically-confirmed superluminous supernova
.
1 (SLSN)fromtheDarkEnergySurvey(DES).Webrieflydiscussthedataandsearchalgorithm
0 usedtofindthiseventinthefirstyearofDESoperations,andoutlinethespectroscopicdata
5 obtained from the EuropeanSouthern Observatory (ESO) Very Large Telescope to confirm
1 itsredshift(z = 0.663±0.001basedonthehost-galaxyemissionlines)andlikelyspectral
:
v type (type I). Using this redshift, we find Mpeak = −21.05+0.10 for the peak, rest-frame
U −0.09
i U-bandabsolutemagnitude,andfindDES13S2cmmtobelocatedinafaint,lowmetallicity
X
(sub-solar),lowstellar-masshostgalaxy(log(M/M⊙) = 9.3±0.3);consistentwithwhatis
r
seenforotherSLSNe-I.WecomparethebolometriclightcurveofDES13S2cmmtofourteen
a
similarly well-observed SLSNe-I in the literature and find it possesses one of the slowest
decliningtails(beyond+30daysrestframepastpeak),andisthefaintestatpeak.Moreover,
we find the bolometric light curves of all SLSNe-I studied herein possess a dispersion of
only 0.2–0.3 magnitudesbetween +25 and +30 days after peak (rest frame) dependingon
redshiftrangestudied;thiscouldbeimportantfor‘standardising’suchsupernovae,asisdone
withthemorecommontypeIa.WefitthebolometriclightcurveofDES13S2cmmwithtwo
competingmodels for SLSNe-I – the radioactivedecay of 56Ni, and a magnetar – and find
that while the magnetar is formally a better fit, neither model providesa compellingmatch
tothedata.Althoughweareunabletoconclusivelydifferentiatebetweenthesetwophysical
modelsforthisparticularSLSN-I,furtherDESobservationsofmoreSLSNe-Ishouldbreak
thisdegeneracy,especiallyifthelightcurvesofSLSNe-Icanbeobservedbeyond100daysin
therestframeofthesupernova.
Keywords: surveys-stars:supernovae:general-stars:supernovae:DES13S2cmm
1 INTRODUCTION
Thelastfiveyearshaveseentheemergenceofanewclassofultra-
⋆ E-mail:[email protected]
bright stellar explosions: superluminous supernovae (SLSNe; for
2 A. Papadopouloset al.
areview seeGal-Yam2012), some 50timesbrighter than classi- 15atz > 1.5(Cookeetal.2012),perhapstrackingtheincreased
cal supernova (SN) types. Data on these rare and extreme events cosmic star-formation rate and/or decreasing cosmic metallicity:
arestillsparse,withonly∼50SLSNdetectionsreportedinthelit- SLSNe are preferentially found in faint, low-metallicity galaxies
erature.Theseeventshavegenerallybeenpoorlystudiedwithin- (Neilletal.2011;Chenetal.2013;Lunnanetal.2014).Moreover,
completeimagingandspectroscopy,leavingmanyaspectsoftheir thepeakofaSLSN’senergyoutputislocatedintheUVregionof
observationalcharacteristics,andtheirphysicalnature,unknown. theelectromagneticspectrum,whichisredshiftedtoopticalwave-
Yet these SLSNe could play a key role in many diverse ar- lengthsathighredshift.
eas of astrophysics: tracing the evolution of massive stars, driv- In this paper, we outline our first search for SLSNe in the
ingfeedbackinlowmassgalaxiesathighredshift,andproviding Dark Energy Survey (DES). In Section 2, we discuss details of
potentialline-of-sightprobesofinterstellarmediumtotheirhigh- DESand our preliminarysearch, whileinSection3, wedescribe
redshifthosts(Bergeretal.2012).SLSNehavealsobeendetected thefirstSLSNdetected inDESand provide detailsof itsproper-
toz ∼ 4(Cookeetal.2012),farbeyond thereachof thecurrent ties, including our spectroscopic confirmation. In Section 4, we
best cosmological probe, type Ia supernovae (SNe Ia). If SLSNe compare our SLSN to other events in the literature and explore
can be standardised (e.g. Inserra&Smartt 2014), asisdone with the possible power sources for our event. We conclude in Sec-
SNeIa(Tripp1998;Phillips1993;Riessetal.1996),thenanew tion5.Throughoutthispaper,weassumeaflatΛ-dominatedcos-
eraofSNcosmologywouldbepossible,withthepotentialtoaccu- mologywithΩ =0.28andH =70kms−1Mpc−1,consistent
M 0
ratelymaptheexpansionrateoftheuniversefarintotheepochof withrecentcosmologicalmeasurements(e.g.Andersonetal.2014;
deceleration(Delubacetal.2014). Betouleetal.2014).
SLSNehavebeendividedintothreepossibletypes:SLSN-II,
SLSN-IandSLSN-R(seeFig.1ofGal-Yam2012).SLSNe-IIshow
signs of interactions with CSM via narrow hydrogen lines (e.g., 2 THEDARKENERGYSURVEY
Ofeketal.2007;Smithetal.2007),andthusmaysimplyrepresent
thebrightendofacontinuumofTypeIInSNe(althoughthisisnot The Dark Energy Survey (DES; Flaugher 2005) is a new imag-
wellestablished). ingsurveyofthesouthernskyfocusedonobtainingaccuratecon-
SLSN-I are spectroscopically classified as hydrogen free straints on the equation-of-state of dark energy. Observations are
(Quimbyetal. 2011), and are possibly related to Type Ic SNe at carriedoutusingtheDarkEnergyCamera(DECam;Flaugheretal.
latetimes(Pastorelloetal.2010),butnormalmethodsofpowering 2012; Diehl&FortheDarkEnergySurveyCollaboration 2012)
such SNe(e.g., theradioactive decay of 56Ni or energy fromthe on the 4-metre Blanco Telescope at the Cerro Tololo Inter-
gravitationalcollapseofamassivestar)donotappeartobeableto AmericanObservatory(CTIO)inChile.DECamwassuccessfully
simultaneously reproduce their extreme brightness, slowly-rising installedandcommissionedin2012,andpossessesa3deg2 field-
light curves, and decay rates (Inserraetal. 2013). Many alterna- of-viewwith62fullydepleted,red-sensitiveCCDs.
tivemodelshavebeenproposed:theinjectionsofenergyintoaSN
ejecta via the spin down of a young magnetar (Kasen&Bildsten
2.1 DESSupernovaSurvey
2010; Woosley 2010; Inserraetal. 2013); interaction of the SN
ejecta with a massive (3-5M⊙) C/O-rich circumstellar material DES conducted science verification (SV) observations from
(CSM; e.g., Blinnikov&Sorokina 2010); or collisions between November2012toFebruary2013,andthenbeganfullscienceop-
high-velocity shells generated from a pulsational pair-instability erations in mid-August 2013, with year one running until Febru-
event (Woosleyetal. 2007). These explanations are still actively ary2014.DESisperformingtwosurveysinparallel:awide-field,
debatedintheliterature. multi-colour (grizY) survey of 5000deg2 for the study of clus-
SLSNe-R are rare and characterised by possessing ex- ters of galaxies, weak lensing and large scale structure; and a
tremelylong, slow-declining light curves (> 200days inthe rest deeper, cadenced multi-colour (griz) search for SNe. In this pa-
frame).SLSNe-Roriginallyappearedconsistentwiththedeathof perwefocusonthisDESSNSurvey,asoutlinedinBernsteinetal.
&100M⊙ starsviathepairinstabilitymechanism(Gal-Yametal. (2012),whichsurveys30deg2over10DECamfields(two‘deep’,
2009). However, new observations – and the lack of significant eight ‘shallow’) located in the XMM-LSS, ELAIS-S, CDFS and
spectral differences between SLSN-I and SLSN-R – have chal- ‘Stripe82’regions.AsofNovember2014,theDESSNSurveyhas
lengedthenotionthattheseclassesaretrulydistinct(Nicholletal. already discovered over a thousand SN Ia candidates, with more
2013) and maybe better described together as Type Ic SLSNe than70spectroscopicclassificationstodate.
(Inserraetal.2013). This rolling search for SNe Ia is also ideal for discovering
Insummary,theoriginofthepowersourceforSLSNe(ofall othertransientphenomenaincludingSLSNe.Usingthelightcurve
types)remainsunclear,withthepossibilitythatfurthersourcesof ofSNLS-06D4eu(aSLSN-I; Howelletal.2013),wecalculatethat
energymaybeviable.Amoredetailedunderstanding requiresan DEScoulddetectsuchaneventtoz=2.5giventhelimitingmag-
increase in both the quantity and quality of the data obtained on nitudesfortheDESdeepfields.Forexample,assumingavolumet-
theseevents.However,findingmoreSLSNe(ofanytype)withex- ricrateof32Gpc−3yr−1fromQuimbyetal.(2013)andcorrect-
istingtransientsearchesischallenging,especiallyinthelocaluni- ingfortimedilation,weestimateDESshouldfindapproximately
versewheretheratesareonly≃10−4thatofthecore-collapseSN 3.5SLSN-IeventsperDESseason(fivemonths)inthedeepfields
rate(Quimbyetal.2013;McCrumetal.2014)thusmakingithard andmanymoreintheshallowfields(althoughatalowerredshift).
to assemble large samples of events over reasonable timescales, Thisbasicpredictionisuncertainbecause:i)Wehavenotcorrected
even from large surveys. For example, McCrumetal. (2014) de- for‘edgeeffects’thatwillreducethenumberofSLSNewithsig-
tected only 10 SLSNcandidates over 0.3 6 z 6 1.4 within the nificant light-curve coverage; ii) The volumetricratesthemselves
firstyearofthePanSTARRS1MediumDeepSurvey. may evolve withredshift (e.g., Cookeetal. 2012) and have large
Searches for SLSNe at higher redshifts may be more prof- uncertainties(McCrumetal.2014);andiii)Wehavenotincluded
itable as the rates of SLSNe appear to rise by a factor of ∼10- anycorrectionsforsurveycompleteness.
DES13S2cmm 3
2.2 SelectingSLSNecandidates 3.1 LightCurve
WebeganoursearchforSLSNeusingthedataproductsfromthe InFig.2wepresentthemulti-colourlightcurveforDES13S2cmm
DESSNSurvey. Allimagingdatawerede-trended and co-added constructedfromthefirstyearofDESobservations. Wehavenot
using a standard photometric reduction pipeline at the National includedupperlimitsfromtheearlier,pre-explosionepochsavail-
Center for Supercomputing Applications (NCSA) using the DES able from the DES SV period in 2012-2013, as these data were
DataManagementsystem(DESDM;Mohretal.2012;Desaietal. taken hundreds of days prior to the data shown in Fig. 2 and are
2012),producingapproximately30‘searchimages’perfield(inall already used to create the reference template image for this field
filters)overthedurationofthefive-monthDESseason.Weperform usedinthedifferenceimagingdetectionoftheSNe.
difference imaging on each of these search images, using deeper We use the photometric pipeline from Sullivanetal. (2011)
templateimagesforeachfieldcreatedfromtheco-additionofsev- (alsoused inMaguireetal. 2012; Ofeketal. 2013) toreduce the
eralepochsofdataobtainedduringtheSVperiodinlate2012.Be- DESphotometricdataforDES13S2cmm.Thispipelineisbasedon
foredifferencing,thesearchandtemplateimageswereconvolved differenceimaging,subtractingadeep,goodseeing,pre-explosion
tothesamepointspreadfunction(PSF). reference image from each frame the SN is present in. The pho-
ObjectswereselectedfromthedifferenceimagesusingSEX- tometryisthenmeasuredfromthedifferencedimageusingaPSF-
TRACTOR (Bertin&Arnouts 1996) v2.18.10, and previously un- fittingmethod,withthePSFdeterminedfromnearbyfieldstarsin
knowntransientcandidateswereidentifiedandexaminedusingvi- theunsubtractedimage.ThisaveragePSFisthenfitattheposition
sual inspection (or ‘scanning’) by DES team members. Over the oftheSNevent,weightingeachpixelaccordingtoPoissonstatis-
course of the first year of DES approximately 25,000 new tran- tics,yieldingaflux,andfluxerror,fortheSN.
sientcandidateswereselectedinthismanner,includingmanydata WecalibratethefluxmeasurementsofDES13S2cmmtoaset
artefactsinadditiontolegitimateSN-likeobjects.Amachinelearn- oftertiarystandardstarsproducedbytheDEScollaboration(Wyatt
ingalgorithm(Goldsteinetal.,inpreparation)wasusedtoimprove etal.,inpreparation),intendedtobeintheABphotometricsystem
ourefficiencyofselectingrealtransients.Transientcandidateswere (Oke&Gunn1983).ThephotometryshowninFig.2hasalsobeen
photometricallyclassifiedusingthePhotometricSNIDentification checked against the standard DES detection photometry, used to
software (PSNID;Sakoetal. 2008, 2011) todetermine the likely find and classify all candidates for follow-up, and found to be in
SNtypebasedonfitstoaseriesofSNtemplates.TheseSNcan- goodagreement.Thelight-curvedataarepresentedinTable1.
didateswerestoredinadatabaseandprioritizedfortheirspectro- We correct for Galactic extinction using the maps of
scopicfollow-up. Schlafly&Finkbeiner(2011),whichestimateE(B−V)=0.028
During the first season of DES, PSNID only included tem- at the location of DES13S2cmm, leading to extinction values of
platesforthenormalSNtypesIb/c,IIandIa,soafullyautomated 0.123,0.070,0.051and0.039magnitudesintheDESgriz filters
classificationofSLSNcandidateswasnotpossible.Therefore,for respectively.Wedonotincludethesecorrectionsinthevaluesre-
thisfirstseason of DES,wesearched for candidate SLSNeusing ported in Table 1, but do correct for extinction in all figures and
thefollowingcriteria:i)Atleast onemonthof multi-colour data, analysisherein,includingFigure2.
i.e.,typicallyfivetosixdetections(S/N > 3.5)ineachofgriz;
ii) A low PSNID fit probability to any of the standard SN sub-
3.2 Spectroscopy
classes;iii)Locatedgreaterthanonepixelfromthecentroidofthe
hostgalaxy(toeliminateAGN);andiv)Peakobservedbrightness On 2013 November 12, we requested Director’s Discretionary
nofainterthanonemagnitudebelowthatofitshostgalaxy.These Time at the European Southern Observatory (ESO) Very Large
broadcriteriawillbesatisfiedbymostSLSNe,whilealsohelping Telescope (VLT) to observe DES13S2cmm, and were awarded
toeliminatemanyofthepossiblecontaminatingsources. target-of-opportunity time within days. The closeness of the ob-
Using this methodology, we selected 10 candidate SLSNe ject to the moon and instrument scheduling issues conspired
overthecourseofthefirstseason.Wesecuredauseablespectrum to delay observations until 2013 November 21, at which point
foroneofourSLSNcandidates,DES13S2cmm,whichisdiscussed DES13S2cmm was approximately 30 days after peak brightness
indetailinSection3,whilenoneoftheothercandidateswerespec- (restframe).AspectrumwasobtainedwiththeFOcalReducerand
troscopicallyconfirmed.Weareattemptingtoobtainspectroscopic lowdispersionSpectrograph(FORS2Appenzelleretal.1998)us-
redshifts from the host galaxies of these other SLSN candidates, ingtheGRIS 300I+11 grism,theOG590order blocker,anda1′′
thoughthiswillbechallengingintheseveralcaseswherenohost slit,withanexposuretimeof3600s(3×1200s).Thisconfiguration
isdetectedinourtemplateimages. providedaneffectivewavelengthcoverageof5950A˚ to9400A˚.
The reduction of the FORS2 data followed standard proce-
dures using the Image Reduction and Analysis Facility1 environ-
3 DES13S2CMM ment (v2.16), using the pipeline described in Ellisetal. (2008).
Thispipeline includesanoptimal two-dimensional (2D)sky sub-
In Fig. 1 we show DES riz imaging data at the position of
tractiontechnique asoutlined inKelson (2003), subtracting a2D
DES13S2cmm:pre-explosion,post-explosion,andtheirdifference
image. DES13S2cmm is located at RA(2000) = 02h42m32s.82, skyframeconstructedfromasub-pixelsamplingofthebackground
Dec.(2000)=−01◦21′30′.′1,andwasfirstdetectedon2013Au- spectrum and a knowledge of the wavelength distortions deter-
minedfrom2Darccomparison frames.Theextracted1D spectra
gust27intheDESSN‘S2’field(a‘shallow’field)locatedinthe
werethenscaledtothesamefluxlevel, thehost-galaxy emission
Stripe82region.DEStransientnamesareformattedfollowingthe
conventionDESYYFFaaaa,whereYYarethelasttwodigitsofthe
yearinwhichtheobservingseasonbegan(13),FFisthetwochar- 1 TheImageReductionandAnalysisFacility(IRAF)isdistributedbythe
acterfieldname(S2),andthefinalcharacters(allletters,maximum NationalOpticalAstronomyObservatories,whichareoperatedbytheAs-
of4)providearunningcandidateidentification,uniquewithinan sociationofUniversitiesforResearchinAstronomy,Inc.,undercooperative
observingseason,asistraditionalinSNastronomy(cmm). agreementwiththeNationalScienceFoundation.
4 A. Papadopouloset al.
Figure1.Colourimages(riz)ofthefieldsurroundingDES13S2cmm.Left:Deeptemplateimagecreatedfromtheco-additionofseveralepochsofdataobtained
duringtheDESScienceVerificationperiodinlate2012;thelikelyhostgalaxyisindicatedbythearrow.Centre:Thesearchimagetakenclosetomaximum
light(28-Sept-2013)withthepositionofDES13S2cmmindicated.Right:Thedifference ofthetwopreviousimages,clearlyidentifying DES13S2cmmat
center.ThisimagedemonstratesthequalityofdifferenceimagesusedfordiscoveryandmonitoringduringtheDESobservingseason.
1-Sep-2013 1-Oct-2013 1-Nov-2013 1-Dec-2013 1-Jan-2014 1-Feb-2014
DES z-2.0
S
DES i -1.0
DES r -0.5
20 DES g
)
B
A 21
(
e
d
u
nit
g 22
a
M
nt
e
r
a 23
p
p
A
24
25
-50 0 50 100 150
Modified Julian Day - 56559.2 (days)
Figure2.TheobservedABmagnitudelightcurveofDES13S2cmminthefourDESSNsearchfilters(griz)asafunctionoftheobservedphase.‘S’denotes
theepochwhentheESOVLTspectrumwasobtained. Theerror-barsrepresent1-σuncertainties, andthearrowsymbolsare3-σ upperlimits.Thephase
relativetopeakwasdefinedusingtheobserved-framer-band.Wehaveartificiallyoffset(inmagnitudes)ther,iandzbanddataforclarity.
Figure3.Thereduced VLTFORS2two-dimensional spectrum forDES13S2cmm.Thekeynebularemissionlinesusedtoderive theredshiftofthehost
galaxy,andthereforesupernova,arelabelled.
DES13S2cmm 5
Table1.LightcurvedataforDES13S2cmmusedinFig.2.Weprovidetheobservedcalendardate,modifiedJulianDay(MJD),observedphaserelativeto
peakMJD(definedrelativetopeakfluxinther-band),andfluxes(notcorrectedforGalacticextinction)with1-σuncertainties intheDESgrizpassbands.
ThesefluxescanbeconvertedtoABmagnitudesusingmag =−2.5log (f)+31.
AB 10
Date MJD Phase fg fr fi fz
(days)
30-AUG-2013 56534.3 -25 1138±77 1705±94 1730±181 1913±231
3-SEP-2013 56538.3 -21 1112±103 2063±129 2234±186 1845±288
8-SEP-2013 56543.3 -16 1369±104 2348±148 2285±198 2285±277
12-SEP-2013 56547.2 -12 1373±130 2329±167 2433±207 2376±284
15-SEP-2013 56550.2 -9 ± 2691±413 2818±380 2193±389
24-SEP-2013 56559.2 0 2022±388 3727±363 3183±346 3548±295
28-SEP-2013 56563.2 4 1668±97 3688±133 3400±187 3488±241
2-OCT-2013 56567.2 8 1805±141 3666±166 3742±209 3169±261
10-OCT-2013 56575.2 16 1370±108 3285±137 3636±192 3182±230
14-OCT-2013 56579.1 20 1260±243 3182±223 3482±259 2663±295
25-OCT-2013 56590.3 31 886±346 1782±319 2924±277 3163±352
29-OCT-2013 56594.1 35 667±146 1782±187 2398±267 2987±366
6-NOV-2013 56602.1 43 344±103 1698±132 2326±185 2433±257
10-NOV-2013 56606.1 47 803±426 1338±418 2633±510 1817±634
13-NOV-2013 56609.1 50 -312±276 970±252 2672±281 2726±426
20-NOV-2013 56616.1 57 347±195 1855±239 1848±307 1739±331
1-DEC-2013 56627.0 68 184±107 1306±129 2032±176 2322±216
9-DEC-2013 56635.0 76 259±208 1322±178 1766±225 2318±304
19-DEC-2013 56645.1 86 302±226 1305±211 1697±230 1808±248
23-DEC-2013 56649.1 90 261±136 956±157 1667±248 2038±276
27-DEC-2013 56653.1 94 41±156 1295±174 1613±201 1930±248
3-JAN-2014 56660.0 101 252±119 1000±132 1736±196 1993±261
10-JAN-2014 56667.1 108 310±343 1144±364 1609±408 1586±387
18-JAN-2014 56675.0 116 -75±236 717±220 1290±261 1863±297
25-JAN-2014 56682.0 123 306±158 992±188 1382±215 1254±270
30-JAN-2014 56687.0 128 345±188 523±179 1518±217 1404±261
6-FEB-2014 56694.0 135 -346±359 1459±376 1602±465 2443±461
linesinterpolatedover,andthespectracombinedusingaweighted spectra of SNe to a library of template SN spectra (via χ2 min-
meanandtheuncertaintiesintheextractedspectra. imisation)whileaccountingforthepossibilityofhost-galaxycon-
In Fig 3 we present our reduced 2D FORS2 spectrum, tamination in the data. We added as template spectra of SLSNe-
wherewavelength increases totheright along thehorizontal axis I (PTF09atu, PTF09cnd, PTF10nmn) and SLSN-R (SN2007bi)
and the vertical axis is distance along the slit. We highlight to SUPERFIT to facilitate our classification. We found that the
the obvious narrow nebular emission lines in the host galaxy highest-rankedfitstoDES13S2cmmweretheseSLSNetemplates,
of DES13S2cmm, specifically [OII] λ3727A˚, Hβ λ4861A˚, and preferred over templates of normal SN types (e.g., SN Ia). The
[OIII] λ4959,5007A˚. We assume these lines originate from the SLSNe-ItemplateswereslightlybetterfitstoDES13S2cmmthan
underlyinghostgalaxyastheirspatialextentalongtheverticalaxis SN2007bi (SLSN-R), although this may be due to not having a
isgreaterthantheobservedtraceofthemainsupernovaspectrum. template for SN2007bi at a similar phase to our spectrum for
Thisobservation does not exclude the possibility that such emis- DES13S2cmm.
sion lines come from the supernova, but it is difficult to address WeshowourspectrumofDES13S2cmminFig.4,alongside
thisissuewithourlowresolution,andlowsignal-to-noise,data.We spectrafromotherwell-studiedSLSNeintheliteratureatasimilar
notetheobservedemissionlinesarenotsignificantlybroadenedor phaseintheirlightcurves(Quimbyetal.2011;Inserraetal.2013;
asymmetrical,aswitnessedintypeIInsupernovae. Gal-Yametal.2009).ThereisadeficitofspectraforSLSNeinthe
Weseenoevidenceofanyfurtheremissionlines,fromtheSN literatureattheselatephasesforarobustcomparison.However,the
and/orhostgalaxy.Therefore,itisunlikelyDES13S2cmmisatype broad resemblance between the spectra of PTF10hgi, SN2011kf,
IIsupernova(normalorsuperluminous). SN2011ke and DES13S2cmm, all taken at approximately +30
Usingthesenebularemissionlines,wedeterminearedshiftof dayspastpeak,suggestsDES13S2cmmisasimilarobjecttothese
z = 0.663±0.001 forthehost galaxyofDES13S2cmm,where other SLSNe-I.Thespectrumof DES13S2cmmisalsosimilarto
theuncertaintyisderivedfromthewavelengthdispersionbetween the spectrum of SN2007bi (a SLSN-R in Gal-Yametal. 2009),
individualemissionlines.Thismeasurementisconsistentwiththe although the differences between Type I and Type R SLSNe re-
redshiftoftheSNbasedonidentificationsofthebroadabsorption mainsunclear,andeitherclassificationwouldbeinterestinggiven
featuresseeninthespectrum.Astheredshiftderivedfromthehost- the sparseness of the data. We also show in Fig. 4 the spectrum
galaxyspectrumissignificantlymoreprecise,henceforthweadopt ofSN2007if,a‘super-Chandrasekhar’ SNIa(Scalzoetal.2010),
thisvalueastheredshifttoDES13S2cmm. an over-luminous (M = −20.4) and likely thermonuclear SN
V
ThespectralclassificationforDES13S2cmmisbasedonthe locatedinafainthostgalaxy.Thiscomparisonclearlyshowsthat
SUPERFITprogram(Howelletal.2005),whichcomparesobserved DES13S2cmmisunlikelytobeasuper-ChandrasekharSNIa.
6 A. Papadopouloset al.
2
SN2011kf
(SLSN−I)
z=0.245
t∼32
1 DES13S2cmm
z=0.663
t=34.9
0
)
λ
(f PTF10hgi
ux (SLSN−I)
e fl zt==03.51.010
ativ −1 SN2011ke
Rel (SLSN−I)
z=0.143
t=31.2
−2 SN2007bi
(SLSN−R)
z=0.128
t∼56
−3
SN2007if
(SN Ia−SuperCh)
−4 t∼38
2500 3000 3500 4000 4500 5000 5500
Rest wavelength (Angstroms)
Figure4. TheVLTFORS2spectrum ofDES13S2cmm(second from top)compared toother SLSNeofsimilar phase intheir light curves. Fromtop to
bottom,wealsoshowSN2011kf(Inserraetal.2013),PTF10hgi(Quimbyetal.2011),SN2011ke(Inserraetal.2013)andSN2007bi(classifiedasTypeR
inGal-Yametal.2009);asdiscussedinSection3.2,SUPERFITidentifies theseSLSNeassomeofthebest-fitting spectraltemplates toDES13S2cmm.In
eachcase,thelightgreyshowstherawdataandthesolidlineisaSavitsky-Golaysmoothedversionofthespectrum.ThespectraarelabelledwiththeSN
name,SLSNtype(exceptDES13S2cmm),redshift,andthephasewhenthespectrumwastaken.Weshowthespectrumofthe‘super-Chandrasekhar’SNIa
SN2007ifasdiscussedinScalzoetal.(2010).Notethatthespectrumfor2007bi,whileatalaterphasethantheotherdata,istheearliestspectralobservation
ofthisSN.
3.3 Host-galaxyproperties brighterthanMpeak =−21(typicalforSLSNe; Gal-Yam2012).
U
Since the start of the first DES observing season the photomet-
ThehostgalaxyofDES13S2cmmisdetectedinourreferenceim- riccatalogderivedfromSVdatahasimproved, andpresentlythe
agesthatcontainnoSNlight.Using SEXTRACTORindual-image DESDM photo-z for the host of DES13S2cmm has improved to
mode and the i-band image as the detection image, we measure z =0.71±0.06,whichisinbetteragreementwithourmea-
host-galaxymagnitudesofg = 24.24±0.13,r = 23.65±0.10, suprheodtospectroscopicredshift(z=0.663±0.001).
i=23.31±0.10andz =23.26±0.17(ABMAG AUTOmagni-
UsingtheDEShost-galaxyphotometryandthespectroscopic
tudes).Thelargephotometricerrorsareduetotherelativelyshal-
redshift from VLT, we estimate that the stellar mass of the host
lowdepth,andpoorseeing,ofthetemplateimagefromtheDESSV
data–wecannotusetheDESfirst-yeardataforthehostphotometry galaxy is log(M/M⊙) = 9.3 ± 0.3 using stellar population
models from PEGASE.2 (Fioc&Rocca-Volmerange 1997), and
asallepochsofthisdatacontainsomelightfromDES13S2cmm.
The estimated photometric redshift of the host galaxy at the log(M/M⊙) = 9.0±0.3 using thestellar population templates
from Maraston (2005) (both corrected for Milky Way extinction
time of discovery was z = 0.86 ± 0.13 (68 percentile er-
photo andfixingtheredshiftofthetemplatestoz = 0.663).Suchalow
ror). This photometric redshift was derived by the DESDM neu-
stellar-mass host galaxy is consistent with the findings for other
ral network photo-z module code, with uncertainties estimated
SLSNe(Neilletal.2011;Chenetal.2013;Lunnanetal.2014).
from the nearest neighbour method, using data taken during SV
(Sa´nchezetal. 2014). Photometric redshifts (when available) are Weestimatethehost-galaxy metallicityfromour VLTspec-
used when selecting SLSN candidates for spectroscopic follow- trum using the double-valued metallicity indicator R , defined
23
up, and for DES13S2cmm this redshift implied the event was as ([OII] λ3727 + [OIII] λλ4959,5007)/Hβ. Fluxes and un-
DES13S2cmm 7
certainties are derived individually for each emission line from thedefinitionof aSLSNeinGal-Yam(2012).Thiswasachieved
fitsof a gaussian plus a first-order polynomial, with the assump- by integrating the best-fitting black-body spectra through a stan-
tion that there is no contamination of the galaxy emission lines dardU-band filterresponse function (onthe Vegasystem) toob-
from the SN. We compute metallicities using the calibrations of taintherest-frameabsolutemagnitude inU-band (M ).Wefind
U
Kobulnicky&Kewley (2004) and McGaugh (1991), and use the the 2013 September 24 epoch (t = 0 relative phase in Table 1)
formulae derived in Kewley&Ellison (2008) to convert these givesthelargest(peak) absolute U-bandmagnitude ofMpeak =
U
metallicities into the calibration of Kewley&Dopita (2002) – a −21.05+0.10, which is consistent with the SLSNe threshold of
−0.09
stepthatallowsfordirectcomparisonbetweendifferentmethods. Mpeak <−21inGal-Yam(2012).
U
We note that since Hα and NII λ6584 are redshifted to wave- We check this result using a model-independent estimate
lengths beyond our spectral coverage, we cannot determine the based onthe observed r-band peak magnitude. At theredshift of
branch of the R23 function that the metallicity lies on. However, DES13S2cmm,theobserver-framerbandmapsintotherestframe
thederivedvalueforR23isclosetoitstheoreticalmaximum,which as a synthetic filter with an effective wavelength of ≈ 3800A˚,
minimizesthedifferenceinmetallicityestimatesbetweenthetwo slightlyredwardofthestandardU-band.Definingasyntheticfilter
branches. assuch means thek-correction isindependent of theSN spectral
Assuming the lower (upper) branch, we find a metallic- energydistribution,andsinceweareintheABsystemthecorrec-
ity (12 + log[O/H]) of 8.30 (8.38) from McGaugh (1991) and tionissimply2.5log(1+z).Weapplythisk-correctiontotheob-
8.30 (8.42) from Kobulnicky&Kewley (2004), which is consis- servedr-bandpeakmagnitude,aswellascorrectionsforGalactic
tent with the median metallicity of 8.35 found by Lunnanetal. extinctionanddistancemodulus,toyieldanestimatedpeakabso-
(2014) for a sample of 31 SLSNe host galaxies. The total un- lutemagnitudeof−20.47(AB),or−20.74(Vega)inoursynthetic
certainty is dominated by the calibration (∼ 0.15 dex), while filter (where we have computed the AB to Vega conversion for
the statistical uncertainty and the scatter induced by the conver- ourfilter).Whilethisisfainterthanourblackbody-basedestimate,
siontoKewley&Dopita(2002)arecomparativelynegligible.We Vegamagnitudesaresensitivetotheexactshapeofthefilterinthis
thus find a host-galaxy abundance ratio that is distinctly sub- region due to the jump in Vega from 3700-3900A˚; thus even the
solar (8.69; Asplundetal. 2009), though not as extremely low smalldifferenceineffectivewavelengthbetweenU bandandour
as has been seen for other SLSNe, such as SN2010gx (12 + syntheticfiltercanresultinsignificantoffsets.Theabsolutemagni-
log[O/H]=7.5;Chenetal.2013). tudeinoursyntheticfilterfromthebest-fittingblackbodymethod
yieldsM = −20.59,provingthatourmodel-independent
synthetic
estimateisconsistentwithourresultsfromtheblackbodyfitting.
We note that the peak magnitude (Vega) for DES13S2cmm
4 RESULTS classifiesitasaSLSN,provided themagnitude isintrinsictothe
SNandnotenhancedbyothereffects.Visualinspectionoftheim-
4.1 Bolometriclight-curveofDES13S2cmm
ages in Fig. 1 (including the surrounding areas) shows little evi-
We show in Fig. 5 the bolometric light curve of DES13S2cmm. dence for strong gravitational lensing which could affect the ob-
This single light curve was constructed by fitting a single black- servedbrightness of thisSN(wenotethat thehost galaxywould
body spectrum to the DES multi-colour data (requiring measure- havetobelensedaswell).Wecannotconclusivelyruleoutthepos-
ments in a minimum of three photometric bandpasses for a fit), sibilityofstronglensing,asinthecaseofPS1-10afx(Quimbyetal.
on each epoch in the light curve (Table 1). In detail, we redshift 2014), but as discussed above the spectrum and extended light
black-bodyspectratotheobserverframeandintegratethesespec- curveof DES13S2cmmareinconsistent withanormalSN,while
trathroughtheresponsefunctionsoftheDESfilters.Wethendeter- the probability of such a strong lensing event in our DESdata is
minethebest-fittingparametersforablackbodytotheextinction- lessthantheprobabilityofdiscoveringaSLSN,basedonthelens-
correctedphotometryandintegratethisspectrumintherestframe ingstatisticsgiveninQuimbyetal.(2014),andthelowerredshift
toobtainthebolometricluminosityperepoch.Thisprocedureben- of DES13S2cmmand smallersearch areaof DES.Moreover, the
efitsfromthehomogeneityoftheDESdataaswepossessapprox- observed r − i peak colour of DES13S2cmm is consistent with
imatelyequally-spacedepochsacrossthewholelightcurve–typi- the expected colours of other unlensed SNe (including SLSNe)
callyinfourDESpassbands–duetoourrollingsearch.Theerrors showninFigure4of Quimbyetal.(2014),i.e.,DES13S2cmmis
onthebolometricluminositieswerecalculatedfromthefittingun- located below the thick black line in this plot. Therefore, we do
certaintiesontheblack-bodyspectra. notconsidergravitationallensingfurtherinthispaper,buthigher-
We visually checked our best-fit blackbody curves for each resolutionimagingdatashouldbeobtainedforDES13S2cmm,af-
epochoftheDES13S2cmmlightcurveagainsttheobservedgriz tertheSNeventhasvanished,toprovidebetterconstraintsonpos-
photometryandfoundthemtobereasonablewithintheerrors.The siblestronglensingasdiscussedinQuimbyetal.(2014).
largestdiscrepancieswereseenintheg-bandwheresomeepochs
were below the fitted curves, possibly due to UV absorption in
theunderlyingspectralenergydistributionsrelativetoablackbody
4.2 Comparisonofbolometriclight-curves
curve.Inserraetal.(2013)comparedthepredictedUVflux(from
ablackbodycurve)fortwoSLSNewithSWIFTUVobservations, For comparison with DES13S2cmm, we also show in Fig. 5
and found the fluxes were consistent within the error. Therefore the bolometric light curves for fourteen SLSNe-I (or Ic, as
we make no additional correction to the derived bolometric light discussed in Inserraetal. 2013) in the literature: PS1-10ky.
curve,butnotethatquantifiablesystematicerrorsinthebolometric PS1-10awh (Chomiuketal. 2011) and PS1-10bzj (Lunnanetal.
luminosityrequirespectraltemplates,whichinturnrequiresmore 2013);SN2010gx(Pastorelloetal.2010);PTF10hgi,SN2011ke,
spectraldataofSLSNethaniscurrentlypubliclyavailable. PTF11rks, SN2011kf and SN2012il (Inserraetal. 2013);
Using the bolometric light curve of DES13S2cmm, we can SNLS06D4eu and SNLS07D2bv (Howelletal. 2013); and
determine the peak luminosity of the event and compare it to LSQ12dlf, SSS120810 and SN2013dg (Nicholletal. 2014).
8 A. Papadopouloset al.
46 DES13S2cmm [z = 0.66]
PS1-10bzj [z = 0.64]
SLSNe-I PS1-10awh [z = 0.90]
PS1-10ky [z = 0.95]
SN2010gx [z = 0.23]
SN2013dg [z = 0.26]
45 SSS120810 [z = 0.15]
LSQ12dlf [z = 0.25]
SN2012il [z = 0.17]
s)] SN2011kf [z = 0.24]
g/ PTF11rks [z = 0.19]
er SN2011ke [z = 0.14]
(bol 44 SPNTFL1S00h7gDi2bv [[zz == 01..150] ]
L SNLS06D4eu [z = 1.5]
[0
1
g
o
l
43
56Co decay
42
46

SLSNe-R
DES13S2cmm [z = 0.66]
SN2007bi [z = 0.12]
45 PTF12dam [z = 0.10]
PS1-11ap [z = 0.52]
)]
s
g/
r
e
(ol 44
b
L
[0
1
g
o
l
43
56Co decay
SNe type Ia
42
-20 0 20 40 60 80 100
Rest frame phase (days)
Figure5.ThebolometriclightcurveofDES13S2cmm(redcircles)comparedtootherSLSNeintheliterature(seetextforexplanation).Thesebolometriclight
curveswereconstructedbyfittingasingleblack-bodyspectrumtothemulticolourdataavailableperepoch(requiringatleastthreephotometricmeasurements
perepoch),andthenbyintegratingtheblack-bodyspectralenergydistributiontoproducethebolometricluminosityperrest-frameepoch(describedindetail
inSection4.2).Weshowstatisticalerrorsontheblackbodyfitstoeachepochoftheselightcurves,butinmostcasesthiserrorissmallerthantheplotting
symbol.Westressthattheseerrorsdonotaccount forsystematicuncertainties involved inthebolometric calculation (e.g.,assumingablackbody forthe
spectralenergydistributionofSLSNe,especiallyfortheUVpartoftherest-framespectralenergydistribution).Thephaserelativetopeak(y-axis)forthe
literature SLSNeistakendirectly fromthepublishedindividual studies andnotrecalculated herein.Thedashedlines showninbothpanels representthe
expected56Codecayrateatlatetimes.Inthelowerpanel,weshowthebolometriclightcurveexpectedforatypicalTypeIaSNe(solidline)toagainillustrate
theextraordinarybrightness,andextendedduration,oftheseSLSNe
We exclude from this figure, and our analysis, other SLSNe-I PTF09atu, PTF09cnd, and PTF09cwl (Quimbyetal. 2011);
that do not possess data for at least three passbands for each SCP06F6 (Barbaryetal. 2009); and SN2006oz (Leloudasetal.
epoch (defined as being taken within a 24 hour window) over a 2012).
majorityof theirlight curves. Thiscriterion,sodefined toensure
well-constrained bolometric light-curve fits, excludes namely The full bolometric light curves for these SLSNe-I were
calculated using the same methodology as described above for
DES13S2cmm 9
DES13S2cmm (i.e. using the published photometric data and re- Both Fig. 5 and Inserra&Smartt (2014) suggest some stan-
quiring three or more passbands per epoch). As a check, our dardisation of SLSNe light curves is possible but further obser-
bolometric light curves for PTF10hgi, SN2011ke, PTF11rks, vations and analyses are required to find the best approach. For
SN2011kf and SN2012il are similar to those published in example,thelightcurveofDES13S2cmmdoesnotappeartofol-
Inserraetal.(2013),withanyoffsetsduetodifferencesintheinte- low the correlation found by Inserra&Smartt (2014) as it has a
grationoftheblackbodycurve;weintegrateoverallwavelengths, faint peak magnitude, relativetoother SLSNe,but isalso slowly
while Inserraetal. (2013) integrate over the wavelength range of declining towards +30 days after peak. A direct comparison to
their observed optical passbands (we are able to reproduce their Inserra&Smartt(2014)isdifficultasourstudyusesdifferentsam-
luminosities if we follow their procedure). Our bolometric light- pleselectioncriteria(only9oftheir16SLSNeareincludedhere
curves for PS1-10ky. PS1-10awh, and PS1-10bzj are also similar because of our requirement to have three passbands per epoch),
tothosepublishedintheliterature. whileInserra&Smartt(2014)estimateacommon syntheticpeak
InFig.5,thephaserelativetopeakforDES13S2cmmwasde- magnitudesforalltheirSLSNeusingtheirowntime-evolvingspec-
finedasthepeakoftheobserver-framer-band,whichoccurredon tral template. A more detailed comparison will be possible with
MJD56563.2,whileforotherSLSNe-Iwesimplyusedthephase furtherDESSLSNe.
relativetopeakasreportedbytheindividual literaturestudies.In
thecaseofSNLS06D4euandSNLS07D2bv,Howelletal.(2013)
reportedaphasebasedontheircalculationofthebolometriclight 4.3 PowersourceofDES13S2cmm
curves,whichweuseherein,butnotetheremaybegreateruncer-
taintyinthedefinitionofthetimeofpeakfortheselightcurves. ThedetailsofthepowersourceofSLSN-Iremainunclearandare
muchdebated(seeSection1).Hereweinvestigatethetwopopu-
The top panel of Fig. 5 shows that DES13S2cmm is one of
larexplanationsintheliteratureforSLSNe-I:radioactivedecayof
the faintest SLSNe-I around peak compared to the other SLSNe-
56Ni,andenergydepositionfromaMagnetar.
I in the literature, and possesses one of the slowest declining
tails(beyondapproximately+30rest-frame)oftheSLSNestudied
herein.InthebottompanelofFig.5,wecompareDES13S2cmm
to possible Type R SLSNe (to be consistent with the classifica- 4.3.1 Radioactive56Nimodel
tionschemeofGal-Yam2012)andfindthelate-timelightcurveof
As can been seen in Fig. 5, the decline rate of DES13S2cmm
DES13S2cmmissimilar toboth PS1-11ap(McCrumetal. 2014)
is approximately that expected from the decay of 56Co (dashed
and SN2007bi (Gal-Yametal. 2009), highlighting the possible
line). This is suggestive of the SN energy source being the pro-
overlap between these two SLSNe types, as discussed in Sec-
ductionandsubsequentdecayoflargequantitiesof56Niproduced
tion3.2andInserraetal.(2013).
in the explosion. To test this theory we fit the bolometric light
Fig.5alsodemonstratesthatthefifteenSLSNe-Ilightcurves
curveofDES13S2cmmtoamodelof56Ni,followingtheprescrip-
(top panel) have similar luminosities at approximately +25 days
tion of Arnett (1982), which is the approximation of an homol-
(rest frame) past peak, close to the inflection point in the light
ogously expanding ejecta (Chatzopoulosetal. 2009; Inserraetal.
curveswhentheirextendedtailsbegintoappear.Wecalculatethe 2013)wheretheluminosityatanygiventime(inergs−1)isgiven
dispersion, as a function of relative phase, for the SLSNe-I light
by
curves shown in Fig.5, linearly interpolating between large gaps
in each light curve toprovide an evenly sampled set of data. We L(t)=M e−(t/τm)2 ǫ t 2u e(u/τm)2e−u/τNidu
findaminimumdispersionofσ(log10Lbol)=0.11–equivalentto Ni (cid:20) NiZ0 τm2
0.29magnitudes–around+30dayspastpeak(restframe),derived
fromthenineSLSNeinFig.5constrainingthisphase.Thisdisper- +ǫ t 2u e(u/τm)2 e−u/τNi−e−u/τCo du δ , (1)
CoZ τ2 (cid:18) (cid:19) (cid:21) γ
sioninthelight-curvescanbedecreasedtoσ(log L )=0.083, 0 m
10 bol
or 0.20 magnitudes, at approximately +25 days past peak (rest whereuisthe(time)integration variable.Theenergyproduction
frame)ifweexcludethefourhighestredshiftSLSNe-Iabovez≃1 rateanddecayratefor56Niareǫ =3.9×1010 ergs−1g−1and
Ni
(SNLS06D4eu,SNLS07D2bv,PS1-10ky.PS1-10awh).Thebolo- τ = 8.8days,whilefor56Cothesevaluesareǫ = 6.8×109
Ni Co
metric luminosities of these SLSNe-I are more difficult to esti- ergs−1g−1 andτ = 111.3days.Theamountof56Niproduced
Co
mate because of the uncertainties associated with modelling the in the explosion is M . Parameterizing the rise-timeof the light
Ni
UV part of the spectrum by assuming a simplyblackbody curve. curve,τ (fromEqns18and22ofArnett1982)isthegeometric
m
Therecould also beevolution inthe population withredshift. As meanofthediffusionandexpansiontimescales,andisgivenby
commented inQuimbyetal.(2011), suchlight curvecharacteris-
tics could point towards a possible ‘standardisation’ of SLSNe-I, τ =1.05 κ 1/2 Me3j 1/4 (2)
aswithSNeIa,leadingtotheiruseashigh-redshift‘standardcan- m (cid:16)βc(cid:17) (cid:16) E (cid:17)
dles’forcosmologicalstudies(Kingetal.2014).
HereE istheexplosionenergy,M isthetotalamountofejected
ej
Recently, Inserra&Smartt (2014) has proposed a ‘standard- mass, κ is the optical opacity (assumed to be 0.1 cm2 g−1 and
isation’ of the peak magnitude of 16 SLSNe using a stretch-
constantthroughout,asinInserraetal.(2013)),andβisaconstant
luminositycorrectionsimilartothe∆m15 relationshipofPhillips withvalue≈13.7(Arnett1982).
(1993)forSNeIa.TheyfindthatSLSNethatarebrighteratpeak
Weuseδ todenotethegamma-raydepositionfunction:the
γ
haveslowerdeclininglightcurveoverthefirst30daysafterpeak
efficiencywithwhichgamma-raysaretrappedwithintheSNejecta.
(rest frame) and, correcting for this correlation, the dispersion of
For thisfunction wefollow Arnett (1982), which isalso used by
thepeakmagnitudeofSLSNecanbereducedtoσ ≃ 0.2.Thisis
Inserraetal. (2013) and uses the deposition function defined in
similarinsizetothedispersionbetween+25to+30dayspastpeak
Colgateetal.(1980),
seeninFig.5andisonlyafactoroftwoworsethanthedispersion
obtainedforstandardisedSNeIa. δ =G[1+2G(1−G)(1−0.75G)], (3)
γ
10 A. Papadopouloset al.
77
Table2.Best-fittingparametersforavarietyof56Nimodelfitstothebolo-
metriclightcurveforDES13S2cmm.fNi,max isthemaximumfractionof Magnetar
56Niallowed intheejecta foreachfit;forDES13S2cmmthebest-fitting 66 fNi,max=1.0
modelalwaysmaximizesthisvalue.Ejectaand56Nimassesaregivenin fNi,max=0.5
unitsofsolarmasses.Thereare22degreesoffreedomforeachfit. g/s) 55 fNi,max=0.3
er
fNi,max E(1051erg) Mej MNi t0(MJD) χ2 43y (10 44
0.30 31.88 14.77 4.43 56508.8 81.5 osit
n
0.40 14.83 10.59 4.23 56510.4 76.0 mi 33
u
0.50 8.22 8.21 4.10 56511.5 71.5 L
0.60 5.08 6.68 4.01 56512.4 67.9
0.70 3.38 5.62 3.93 56513.1 65.0 22
0.80 2.38 4.84 3.88 56513.7 62.7
0.90 1.74 4.25 3.83 56514.2 60.9 11
1.00 1.32 3.79 3.79 56514.6 59.4 30 K) 12
e (1 10
atur 8
whereG≡τγ/(τγ+1.6),withthe‘opticaldepth’forgamma-rays mper 6
approximatelygivenby Te 4
Herevej =τγ ≈10E(cid:16)0kκ./13(cid:17)M(cid:16)e4jττam2Nn2id(cid:17)i(cid:16)svineju5(n.05i.t31s×o+f1tc/0mτ10Nsi−)21.(cid:17)W. enoteth(4a)t 15adius (10 cm) 102468
the depositiopn function used in Chatzopoulosetal. (2009) has a R 0
functionalformthatissimilartothatusedhere,butintheirapprox- -20 0 20 40 60 80 100
imationτ onlydeviatesfrom≈ 1atmuchlaterepochs,resulting Rest Frame Epoch (Days)
γ
inalight-curvedecay ratemirroring56Coandlargelyinsensitive Figure6.Upperpanel:BolometriclightcurveofDES13S2cmm,withthe
tochangesintheexplosionenergyorejectamass. best-fitmagnetarmodel(greensolidline)andthree56Nimodelswithdif-
We thus have four parameters in our model: the explosion ferent f (blue lines) overplotted. The magnetar model is a better
Ni,max
epoch(t ),theenergyoftheexplosion(E),thetotalejectedmass matchtothedatathanthebest-fit56Ni,andhasasignificantly betterχ2
0
(Mej), and the amount of ejected mass which is 56Ni (MNi). We than the models with a more realistic fNi,max (see text). However both
define the fraction of the ejecta mass that is in 56Ni as f = modelshavedifficultyreproducingthepeakluminosity,thepost-peakde-
Ni
cline,andthelate-timeflattening.MiddleandBottompanels:Temperature
M /M .Wefitourmodeltothedatawithavarietyofupperlimits
Ni ej andradiusevolutionofthebest-fittingblackbodytoeachphotometricepoch
onf ,varyingfrom0.3to1.0.However, wefollowInserraetal.
Ni whichisusedtoconstructthebolometriclightcurve.
(2013)inassumingaphysicallymotivatedupperlimitonf of0.5,
Ni
whichtheybaseontheprevalence ofintermediate-masselements
intheirspectraoflower-zSLSNe,andthecalculationsof56Nipro- basicmodelthatcouldyielddifferentlight-curveshapes.Forexam-
ductionincore-collapseSNebyUmeda&Nomoto(2008).Were- ple,ithasbeennotedthatatwo-componentmodelthatisdenserin
portthebest-fittingparametersandgoodnessoffitforthesemodels theinnercomponentwouldproduceaquickerflatteningpost-peak
inTable2aswellasshowtherangeoff modelsinFigure6to (Maedaetal.2003).
Ni
illustratethescaleofuncertaintiesinthetheoreticalmodelling.
Wefindthatthe56Nimodelisnotagoodfittothebolometric
4.3.2 Magnetarmodel
lightcurve,asthebest-fittingmodelhasaχ2/dof =2.7(χ2of59
for22degreesoffreedom),andisphysicallyunrealistic(fNi =1). As an alternative to the radioactive decay of 56Ni, we also fit
AscanbeseeninTable2,themodelpredictionforMNiisrelatively ourderivedbolometriclightcurveforDES13S2cmmtoamagne-
robust(3.8−4.4M⊙),asthisparameterisprimarilyconstrainedby tar model, which proposes that SLSNe are powered by the rapid
thepeakluminosityofthelightcurve.However,tomatchthelate spinning-down of a neutron star with an extreme magnetic field
time (t > 30 days) flattening of the light curve a large amount (Woosley 2010; Kasen&Bildsten 2010). In this model the input
of ejecta mass or low explosion energy (see Eqn. 2) is required energyisdefinedbytheinitialspinperiod(P ,inmilliseconds)
ms
to increase the diffusion time-scale, which simultaneously makes and magnetic field strength (B , in units of 1014 Gauss) of the
14
thefittothepost-peakdeclinepoorwhilepredictingamuchlonger magnetar, and the rise-time parameter (τ , Eqn. 2) reflects the
m
risetimethanisseen.Wealsonotethattheseparametersarepoorly combined effect of the explosion energy (E), the ejected mass
constrainedindividuallyasEandMej arecorrelated(Equation2), (Mej),andtheopacity(κ).Weusethesemi-analyticalmodelout-
andthisdegeneracycannotbebrokenwithoutthepresenceofhigh- linedinInserraetal.(2013).Theluminosityofthemagnetarmodel
qualityspectraldata(Mazzalietal.2013). attimetisgivenby(inergs−1)
Wedonotprovideuncertaintiesonthebestfitparametersfor
each fNi model in Table 2. Thisis because the formal uncertain- L(t)=4.9×1046e−(t/τm)2δ t 2u e(u/τm)2 B124Pm−s4 du,
tiesonparameters(suchasMNi)aresmallerforeachfitthanthey Z0 τm2 (1+u/τp)2
(5)
arebetweenfitswithdifferentconstraintsonf .Thisisduetothe
Ni
simplicityofthe56Nimodel;wedonotconsiderhereinNimixing, where u is the (time) integration variable, τ is the spin-down
p
non-standarddensityprofiles,norotherpossiblevariationsinthis timescale,τ = 4.7B−2P2 days,andδ isthedepositionfunc-
p 14 ms