SN 2008iy: An Unusual Type IIn Supernova with an Enduring 400 Day Rise Time

1.3 MB·English

by  A. A. Miller

Mon.Not.R.Astron.Soc.000,1–15(2009) Printed27January2011 (MNLATEXstylefilev2.2)
SN 2008iy: An Unusual Type IIn Supernova with an
Enduring 400 Day Rise Time
A. A. Miller1(cid:63), J. M. Silverman1, N. R. Butler1, J. S. Bloom1†, R. Chornock1,2,
A. V. Filippenko1, M. Ganeshalingam1, C. R. Klein1, W. Li1, P. E. Nugent3,
N. Smith1, and T. N. Steele1.
1Department of Astronomy, University of California, Berkeley, CA 94720-3411, USA.
1
2Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA 02138, USA.
1
3Lawrence Berkeley National Laboratory, 1 Cyclotron Road, Berkeley, CA 94720, USA.
0
2

n
a
J
ABSTRACT
6
2
WepresentspectroscopicandphotometricobservationsoftheTypeIInsupernova

] (SN)2008iy.SN2008iyshowedanunprecedentedlylongrisetimeof∼400days,making
E
it the first known SN to take significantly longer than 100 days to reach peak optical
H luminosity.ThepeakabsolutemagnitudeofSN2008iywasM ≈−19.1mag,andthe
r
. total radiated energy over the first ∼700 days was ∼2 ×1050 erg. Spectroscopically,
h
SN 2008iy is very similar to the Type IIn SN 1988Z at late times, and, like SN 1988Z,
p
it is a luminous X-ray source (both supernovae had an X-ray luminosity L > 1041
- X
o erg s−1). SN 2008iy has a growing near-infrared excess at late times similar to several
r other SNe IIn. The Hα emission-line profile of SN 2008iy shows a narrow P Cygni
t
s absorptioncomponent,implyingapre-SNwindspeedof∼100kms−1.Wearguethat
a the luminosity of SN 2008iy is powered via the interaction of the SN ejecta with a
[
dense,clumpycircumstellarmedium.The∼400dayrisetimecanbeunderstoodifthe
2 number density of clumps increases with distance over a radius ∼1.7 ×1016 cm from
v theprogenitor.Thisscenarioispossibleiftheprogenitorexperiencedanepisodicphase
9 of enhanced mass loss < 1 century prior to explosion or if the progenitor wind speed
1 increasedduringthedecadesbeforecorecollapse.Wefavourtheformerscenario,which
7 is reminiscent of the eruptive mass-loss episodes observed for luminous blue variable
4
(LBV)stars.Theprogenitorwindspeedandincreasedmass-lossratesserveasfurther
.
1 evidencethatatleastsome,andperhapsall,TypeIInsupernovaeexperienceLBV-like
1 eruptions shortly before core collapse. We also discuss the host galaxy of SN 2008iy,
9 a subluminous dwarf galaxy, and offer a few reasons why the recent suggestion that
0 unusual,luminoussupernovaepreferentiallyoccurindwarfgalaxiesmaybetheresult
:
v of observational biases.
i
X Key words:
supernovae: general — supernovae: individual (SN 2008iy, SN 1988Z) — stars: mass-
r
a loss — circumstellar matter
1 INTRODUCTION each new discovery serves to bracket our understanding of
the physical origin of well-established classes of SNe and
The recent development of synoptic, wide-field imaging has
does, in principle, demand an increased clarity in our un-
revealed an unexpected diversity of transient phenomena.
derstanding of the post-main-sequence evolution of massive
Onesuchexampleisthediscoveryofanewsubclassofvery
stars.
luminoussupernovae(VLSNe;e.g.,Ofeketal. 2007;Smith
et al. 2007; Quimby et al. 2007b). While events of this
nature are rare (Miller et al. 2009a; Quimby et al. 2009),
Yet another unusual transient was discovered by
(cid:63) E-mail:[email protected]. the Catalina Real-Time Transient Survey (CRTS; Drake
† SloanResearchFellow. et al. 2009a), which they announced via an ATel as
(cid:13)c 2009RAS
2 Miller et al.
CSS080928:160837+041626 on 2008 Oct. 07 UT1 (Drake et
Table 1.DeepSkyObservationsofSN2008iy.
al. 2008). The transient was classified as a Type IIn SN
with a spectrum taken on 2009 Mar. 27 (Mahabal et al. date maga σmagb
2009; see Schlegel 1996 for a definition of the SN IIn sub-
(MJD)
class,andFilippenko1997forareviewofthespectralprop-
erties of SNe). The SN was later given the IAU designation 54618.35 17.37 0.12
SN 2008iy (Catelan et al. 2009). Mahabal et al. (2009) 54621.35 17.38 0.12
54623.33 17.35 0.12
noted that the transient was present on CRTS images dat-
54626.32 17.34 0.12
ing back to 2007 Sep. 13; however, it went undetected by
54627.36 17.37 0.12
theCRTSautomatedtransientdetectionsoftwareuntil2008
54628.29 17.37 0.12
becauseitwasblendedwithanon-saturated,nearby(∼11(cid:48)(cid:48)
54629.30 17.36 0.12
separation) star. 54638.21 17.34 0.12
Here, we present our observations and analysis of 54651.20 17.28 0.12
SN2008iy,whichpeakedaround2008Oct.29(Catelanetal.
2009) and had a rise time of ∼400 days. This implies that
SN2008iytooklongertoreachpeakopticalbrightnessthan a Observedvalue;notcorrectedforGalacticextinction.
any other known SN. Type II SN rise times are typically (cid:46) b The calibration uncertainty dominates over the statistical
1week(e.g.,SNe2004etand2006bp,Lietal.2005;Quimby uncertaintywithvaluesof∼0.12and∼0.01mag,respectively.
et al. 2007a; see also Patat et al. 1993; Li et al. 2010a),
andhaveneverpreviouslybeenobservedtorise(cid:38)100days,
2007 Sep. 13 the SN was between 18.3 and 19.0 mag, while
let alone 400, making SN 2008iy another rare example of
on2008Feb.19theSNwasobservedbetween17.9and18.3
the possible outcomes for the end of the stellar life cycle.
mag.
In addition to an extreme rise time, SN 2008iy is of great
Near-infrared (NIR) observations of SN 2008iy were
interestbecausetheuniquecircumstellarmedium(CSM)in
conductedwiththe1.3-mPetersAutomatedInfraredImag-
whichitexplodedmayprovidealinktoverylong-livedSNe,
ing Telescope (PAIRITEL; Bloom et al. 2006) starting on
suchasSN1988Z,andthusprovidecluesintothenatureof
2009 Apr. 13. PAIRITEL observes simultaneously in the J,
their progenitors.
H,andK bands.Observationswerescheduledandexecuted
This paper is organised as follows. Section 2 presents s
viaaroboticsystem,andthedatawerereducedbyanauto-
theobservations.ThedataareanalysedinSection3,andthe
mated pipeline (Bloom et al. 2006). The SN flux was mea-
results are discussed in Section 4. We give our conclusions,
sured via aperture photometry using SExtractor (Bertin &
aswellaspredictionsforthefuturebehaviourofSN2008iy,
Arnouts 1996), calibrated against the Two Micron All Sky
in Section 5.
Survey(2MASSSkrutskieetal. 2006).Filteredphotometry
ofSN2008iyisshowninFig.2andsummarisedinTable2.
Ground-based optical observations of SN 2008iy were
2 OBSERVATIONS obtained using the 1.0-m Nickel telescope located at Lick
Observatory (Mt. Hamilton, CA, USA) starting on 2009
2.1 Photometry
Apr. 18. BVRI photometry from the Nickel was measured
The field of SN 2008iy, which is located at α = using the DAOphot package (Stetson 1987) in IRAF4, and
16h08m37.27s,δ=+04◦16(cid:48)26.5(cid:48)(cid:48)(J2000),wasimagedmulti- transformedintotheJohnson–Cousinssystem.Calibrations
pletimesbythePalomarQuestsurvey,andthosedatahave for the field were obtained on ten photometric nights with
beenreprocessedaspartoftheDeepSkyproject2 (DS;Nu- the Nickel telescope. The Nickel photometry is shown in
gent2009).Thebestconstraintsontheexplosiondatecome Fig. 2 and summarised in Table 3.
from DS imaging: in a coadd of two images from 2007 Jul. SN 2008iy was observed by the space-based Swift ob-
05, we do not detect the SN down to a 3σ limit of i>20.9 servatory on 2009 Apr. 15 and 16. Swift observed the SN
mag.DSimagesarebestapproximatedbytheSloanDigital simultaneously with both the X-ray Telescope (XRT; Bur-
SkySurvey(SDSS;Adelman-McCarthyetal. 2008)i-band rows et al. 2005) and the Ultraviolet/Optical Telescope
filter. The field of SN 2008iy has well-calibrated SDSS pho- (UVOT; Roming et al. 2005). We downloaded the Level-2
tometry which we use to calibrate the DS images. The ob- UVOT images, and measured the U, B, and V photometry
served DS magnitudes are shown in Fig. 1 and summarised usingtherecipeofLietal. (2006).TheUVfilters(UVM1,
in Table 1. UVW1, and UVW2) were calibrated using the method of
ToconstraintherisetimeofSN2008iywevisuallyesti- Poole et al. (2008). Our final UVOT photometry is sum-
mate the possible range of magnitudes for the SN from im- marised in Table 4. The XRT observed the field for a to-
agesontheCRTSwebsite3 basedonacomparisontoSDSS tal of 10.71 ks. We extract the 0.3–10.0 keV counts from
images.Aspreviouslymentioned,SN2008iyisblendedwith an extraction region of 64 pixels (∼ 2.5(cid:48)), where we fit the
a nearby star and so the automated photometry produced point-spread function model (see Butler & Kocevski 2007)
by the CRTS does not detect the SN on these epochs. On atthecentroidoftheX-rayemission.FromthenativeXRT
1 UT dates are used throughout this paper unless otherwise 4 IRAF is distributed by the National Optical Astronomy Ob-
noted. servatory, which is operated by the Association of Universities
2 http://supernova.lbl.gov/∼nugent/deepsky.html. forResearchinAstronomy(AURA)undercooperativeagreement
3 http://crts.caltech.edu/. withtheNationalScienceFoundation.
(cid:13)c 2009RAS,MNRAS000,1–15
The 400 Day Rise of SN 2008iy 3
17 (a)
18
(a)
g (b)
a
m (b)
t SN 1988Z
n 19
e
r
a
p
p
a
20 SN 1999em Deep Sky i-band (c)
Nickel I-band
Catelan et al. r-band
CRTS visual mag
21 D(Sc )limit
0 200 400 600
Days since discovery (rest frame)
Figure 1. Left: Apparent optical light curve of SN 2008iy showing the long rise time of the SN, including data from Deep Sky (this
work), the Nickel telescope (this work), Catelan et al. (2009), and the CRTS website. Note: conservatively, we take the start time of
theSN(t=0)tocoincidewiththefirstepochwhereCRTSdetectsthesource;however,thetrueexplosiondateislikelypriorto2007
Sep. 13, between the last Deep Sky non-detection and the first CRTS detection. Notice that the light curve is very broad, and almost
symmetric about the peak. For comparison we show the light curve of the standard Type II-P SN 1999em (data from Leonard et al.
2002) and the Type IIn SN 1988Z, which shows many similarities to SN 2008iy at late times (see Section 3.2; SN 1988Z data from
Turattoetal. 1993),astheywouldhaveappearedattheredshiftofSN2008iy.Right:ThreepanelswithimagesofthefieldofSN2008iy.
Eachimageis∼2.5(cid:48)×4(cid:48),withnorthupandeasttotheleft.BlackcrosshairsmarkthelocationoftheSN.Thebottomimage,marked
(c),showsthelastnon-detectionfromtheDeepSkydata.Themiddle(b)andtop(a)imagesshowtheCRTSdetectionsonday0and
day153,respectively;SN2008iycanclearlybeseenineachofthem.TheSNwasnotflaggedasatransientineitherthemiddleortop
imagebytheCRTSsoftware.
astrometry the centroid of the X-ray emission, located at α side of the double-arm (blue+red) LRIS system. The spec-
= 16h08m37.23s, δ = +04◦16(cid:48)26.7(cid:48)(cid:48) (J2000), with a radial tra were reduced and calibrated using standard procedures
uncertainty of 5(cid:48)(cid:48) (90% confidence interval), coincides with (e.g., Matheson et al. 2000). Clouds were present on the
theopticalpositionofSN2008iy,suggestingthattheX-rays night of 2009 Apr. 25, and the observing setup from that
arefromtheSN.Inanastrometricfitrelativeto2MASSwe night did not include any wavelength overlap between the
measure the position of the star ∼11(cid:48)(cid:48) from the SN to be red and blue arms of the Kast spectrograph; thus, the ab-
α = 16h08m36.98s, δ = +04◦16(cid:48)16.4(cid:48)(cid:48) (J2000), with a root- solute flux calibration on this night is less certain than on
mean square scatter of 0.19(cid:48)(cid:48) in α and 0.18(cid:48)(cid:48) in δ for our the other nights. For the Kast spectra, we observed the SN
astrometric solution. This star is therefore well outside the with the slit placed at the parallactic angle, so the relative
90% confidence interval for the location of the X-ray emis- spectralshapesshouldbeaccurate(Filippenko1982).LRIS
sion. Furthermore, there is no catalogued X-ray source at is equipped with an atmospheric dispersion corrector; nev-
the position of the SN in the ROSAT catalogue (Voges et ertheless, we observed the SN at the parallactic angle on
al. 1999), also suggesting that the star located ∼11(cid:48)(cid:48) from this night as well. In Fig. 3 we show the spectral sequence
the SN is not the source of X-ray emission. We measure a ofSN2008iy,andasummaryofourobservationsisgivenin
background-subtracted0.3−10.0keVcountrateof(19±6) Table 5.
×10−4 counts s−1.
2.3 Host Galaxy
2.2 Spectroscopy
There is no catalogued redshift for the host galaxy of
Spectra of SN 2008iy were obtained on 2009 Apr. 18, 2009 SN 2008iy. To determine the redshift of the SN we exam-
Apr. 25, and 2009 Jul. 23 with the Kast spectrograph on inedtheLRISspectrumforthepresenceofnarrowemission
theLick3-mShanetelescope(Miller&Stone1993).Anad- lines. Based on a fit to the [O III] λλ4959, 5007 doublet we
ditional spectrum was obtained on 2009 Sep. 22 with the adoptaheliocentricredshiftofz=0.0411forSN2008iy(not
Low-Resolution Imaging Spectrometer (LRIS; Oke et al. corrected for peculiar motions or Virgo infall). This corre-
1995)onthe10-mKeckItelescope.Toimprovethespectral sponds to a luminosity distance of d =179 Mpc (H =71
L 0
resolution at the location of the Hα emission feature from km s−1Mpc−1, Ω = 0.27, Ω = 0.73), which we adopt
M Λ
SN2008iy,weusedthe1200linesmm−1 gratingonthered throughout the remainder of this paper.
(cid:13)c 2009RAS,MNRAS000,1–15
4 Miller et al.
?
ant] Ca II H% H$[O III] He II H#[O III] [N II] [Fe X][O I] H" [Ca II] [O II] ? O I O I Ca II
nst 100.000 e I e I e I e I e I e I e I
o H H H H H H H
c
) * d +560 (x 110)
−1
Å
−1 d +567 (x 10)
−2 s 1.000
m
c
g d +652
r
e
−15 d +711 (x 0.1)
0
(1 0.010
F !
[
g
o SN 1988Z
l day +504 (x 0.01)
0.000
3000 4000 5000 6000 7000 8000 9000
rest wavelength (Å)
Figure 3. Spectral sequence of SN 2008iy (shown in black) and a single, late-time spectrum of SN 1988Z from our spectral database
(shown in red), which was observed 504 days after discovery. To the left of the spectra we give the time since discovery, in rest-frame
days,atwhicheachspectrumwastaken.ProminentspectralfeaturesofSN2008iyarelabelledatthetopofthefigure.Thespectrashow
littleevolutionbetweendays560and711.FeIImultiplets42,48,and49canalsobeseen,thoughwehavenotlabelledthoselines.The
spectrahavebeencorrected forGalacticreddening(E(B−V)=0.065mag; Schlegel, Finkbeiner &Davis1998).Wehave assumedno
reddeningintheSNhostgalaxy(seeSection2.3).
Table 2.PAIRITELObservationsofSN2008iy. Table 3.NickelObservationsofSN2008iy.
tmida J magb H magb Ks magb date B maga V maga Rmaga I maga
(MJD) (Vega) (Vega) (Vega) (MJD) (Vega) (Vega) (Vega) (Vega)
54934.42 16.59±0.05 16.00±0.05 14.98±0.05 54939.52 19.06±0.01 18.59±0.03 17.86±0.01 17.87±0.03
54937.43 16.57±0.06 15.99±0.06 15.05±0.05 54940.46 19.09±0.03 18.56±0.02 17.86±0.02 17.89±0.03
54941.43 16.57±0.05 16.04±0.08 15.06±0.08 54971.37 19.35±0.06 18.71±0.04 17.99±0.04 17.98±0.11
54944.40 16.70±0.05 15.98±0.07 14.98±0.07 54975.36 19.25±0.03 18.75±0.02 17.96±0.02 17.92±0.07
54948.40 16.59±0.05 15.87±0.06 14.98±0.06 54979.46 19.23±0.03 18.71±0.02 17.97±0.02 ...
54951.39 16.55±0.04 15.96±0.06 15.02±0.06 54999.35 19.36±0.05 18.79±0.03 18.07±0.03 18.12±0.08
54954.35 16.70±0.04 15.88±0.05 15.00±0.08 55004.37 19.40±0.05 18.84±0.03 18.07±0.02 18.13±0.11
54957.37 16.63±0.05 15.91±0.05 14.99±0.10 55007.43 19.56±0.11 18.84±0.10 18.02±0.04 18.37±0.13
54959.37 16.68±0.05 15.89±0.06 14.97±0.07 55015.36 19.43±0.11 18.95±0.06 18.10±0.04 18.26±0.08
54961.44 16.69±0.06 15.99±0.08 14.83±0.12 55019.30 19.58±0.09 19.02±0.06 18.19±0.03 18.21±0.07
54970.30 16.77±0.08 15.98±0.12 ... 55025.36 19.66±0.24 18.66±0.10 18.16±0.08 18.18±0.12
54975.28 16.66±0.07 15.86±0.05 14.87±0.06 55034.26 19.52±0.02 19.02±0.03 18.19±0.02 18.23±0.11
54981.35 16.72±0.04 15.90±0.05 14.93±0.08 55040.30 19.56±0.05 18.97±0.03 18.18±0.03 18.30±0.09
54986.29 16.61±0.05 15.85±0.06 15.06±0.12 55042.24 19.60±0.07 19.11±0.06 18.21±0.03 18.31±0.11
54990.30 16.68±0.08 16.00±0.11 14.59±0.07 55047.27 19.43±0.11 19.21±0.10 18.22±0.04 18.24±0.20
54994.30 16.69±0.05 15.89±0.04 14.84±0.06 55071.22 19.66±0.04 19.17±0.05 18.23±0.03 18.34±0.09
54998.31 16.67±0.05 15.92±0.07 14.90±0.07 55077.23 20.07±0.24 19.33±0.17 18.29±0.03 18.58±0.17
55003.18 16.86±0.07 15.97±0.05 14.96±0.11 55100.15 19.75±0.07 19.25±0.08 18.35±0.03 18.49±0.06
55030.20 16.92±0.19 16.01±0.15 ...
55041.18 16.78±0.07 15.94±0.08 14.82±0.10
55091.10 17.08±0.09 15.94±0.07 14.74±0.08 a Observedvalue;notcorrectedforGalacticextinction.
a Midpoint between the first and last exposures in a single The reddening of SN 2008iy by the host galaxy is un-
stackedimage. certain.InourKastspectrawedonotdetectanyabsorption
b Observedvalue;notcorrectedforGalacticextinction. from Na I D, and therefore we adopt Aλ,host =0 mag.
Weconfirmthepossibledetectionofthehostgalaxyby
Catelan et al. (2009) in a stack of SDSS g, r, and i-band
images of the field. When the g + r + i stack is calibrated
(cid:13)c 2009RAS,MNRAS000,1–15
The 400 Day Rise of SN 2008iy 5
15 (GALEX; Martin et al. 2005). On 2004 May 17, the field
Ks+0.5 was imaged as part of the all-sky imaging survey (AIS);
the host was not detected to a 5σ limiting magnitude of
16 H
FUV (cid:38)20.2andNUV (cid:38)20.5mag.Thefieldwasreimaged
as part of the medium imaging survey on 2008 Jun. 05/06,
mag 17 J and SN 2008iy was detected at FUV =21.175±0.094 mag
ent I-0.7 and NUV =19.929±0.036 mag.
ar 18
p
p R
a
19
V 3 RESULTS
B
20 3.1 Photometric Analysis
550 600 650 700 750 At z = 0.0411, the absolute peak magnitude of SN 2008iy
Days since discovery (rest frame) was a relatively modest Mr ≈ −19.1 mag. This places
SN 2008iy well below the most luminous SNe IIn, such as
Figure 2.FilteredphotometryincludingBVRI fromtheNickel SN 2006gy (MR = −21.7 mag; Ofek et al. 2007; Smith
telescopeandJ,H,andKs fromPAIRITEL,showingthedecline et al. 2007) and SN 2008fz (MV = −22.3 mag; Drake
rateofSN2008iybetweendays∼560and715.Tohelpguidethe et al. 2009b), and more in the range of the well-studied
eye,weillustratethelinearfittoeachbandusedtodeterminethe SN IIn 1988Z (M (cid:46) −18.9 mag; Turatto et al. 1993;
R
photometric decline rates (see text). Note that the single epoch
Stathakis & Sadler 1991). Assuming no bolometric correc-
ofUVOTB andV-bandobservationswemeasured,whichagree
tion,thetotalintegratedopticaloutputfromSN2008iydur-
to < 1σ with Nickel data taken < 2 days later, is not shown in
ingthefirst∼700daysafterdiscoveryis∼2×1050 erg.Be-
thisfigure.
tween day 550 and 720 the SN declines in the optical and
the J band, while it actually gets brighter in the H and
K bands. During this time, the linear decay rates are as
Table 4.UVOTObservationsofSN2008iy. s
follows: β = 0.51 mag (100 day)−1, β = 0.50 mag (100
B V
tmida filter magb σmag ddaayy))−−11,, ββR==00.3.304mmaagg(1(01000dadya)y−)1−,1β, βI==−00..4033mmaagg((110000
(MJD) (Vega) J H
day)−1, and β = −0.19 mag (100 day)−1. These decline
Ks
54936.31 UVW2 19.14 0.06 rates, all slower than the expected rate of decline for ra-
54936.32 UVM2 18.95 0.08 dioactive56Co,0.98mag(100day)−1,stronglysuggestthat
54936.32 UVW1 18.53 0.07 theSNisstillbeingpoweredbyCSMinteraction∼700days
54936.31 U 18.60 0.06
after explosion.
54937.49 UVW2 19.12 0.04
Forcomparisonpurposes,inadditiontothelightcurve
54937.85 B 19.04 0.06
of SN 2008iy, Fig. 1 shows the light curve of the standard
54937.85 V 18.65 0.08
TypeII-PSN1999em(Leonardetal. 2002)asitwouldhave
appearedattheredshiftofSN2008iy.Aswellasillustrating
theverylongrisetimeofSN2008iy,thiscomparisonshows
a Midpoint between the first and last exposures in a single
that SN 2008iy was fairly luminous at the time of discov-
stackedimage.
b Observedvalue;notcorrectedforGalacticextinction. ery. Thus, despite the sparse sampling over the first ∼200
days,thislargeluminositysuggeststhattheearlydetections
are not related to a pre-SN outburst, as was observed for
to the SDSS r band, we find that the host has r = 22.75 ± SN2006jc(Foleyetal. 2007;Pastorelloetal. 2007).CRTS
0.15±0.08magina2.2(cid:48)(cid:48) diameteraperture,wherethetwo images indicate that the SN was rising over the first ∼400
errorsrepresentthestatisticalandcalibrationuncertainties, days(Mahabaletal. 2009),whichprovidesfurtherevidence
respectively. againstapre-SNoutburst.Fig.1alsoshowsthelightcurve
As first noted by Catelan et al. (2009), the field of of the long lived, interacting, Type IIn SN 1988Z (Turatto
SN 2008iy was observed by the Galaxy Evolution Explorer et al. 1993). The explosion date of SN 1988Z is not well
constrained (see Section 4.3), so we shift the first epoch of
detection to day 0. Notice the similarity of the decline rate
Table 5.Logofspectroscopicobservations.
of SN 2008iy and SN 1988Z around day ∼600; these SNe
Epocha UTDate Telescope/ Exposure also have very similar late-time spectra (see Section 4.2).
We were unable to fit the UV through NIR spec-
Instrument (s)
tral energy distribution (SED) of SN 2008iy to a single-
560 2009-04-18.458 Shane3-m/Kast 1500 temperature blackbody model. This is not surprising, as
567 2009-04-25.479 Shane3-m/Kast 1500
SNe II are dominated by emission lines at late times, and
652 2009-07-23.274 Shane3-m/Kast 1800
single-temperatureblackbodymodelstypicallyapplyonlyto
711 2009-09-22.236 KeckI10-m/LRIS 600
youngSNeII(see,e.g.,Filippenko1997).Directintegration
of the UV−NIR SED shows that the bolometric correction,
relative to the I band, is a factor of ∼3 in luminosity, cor-
a Definedasrest-framedaysrelativetoday0,2007Sep.13. respondingto∼1.2mag,onday∼560.Somefeaturesstand
(cid:13)c 2009RAS,MNRAS000,1–15
6 Miller et al.
out from the SED: there is a strong R-band excess rela- We can estimate the density of the unshocked emit-
tive to the other optical bands, with V −R=0.8 mag and ting material based on the relative intensities of the three
R−I =−0.2magonday715.TheredV −Rcolourrelative [O III] lines mentioned above. Note that [O III] λ4363 is
toablueR−I colourcanbeattributedtotheHαemission only resolved in our day 711 Keck spectrum, so the follow-
withlargeequivalentwidth.Thisemissionalsoaccountsfor ingestimateofthedensityisatthatepochonly.Followinga
therelativelyslowdecayrateintheRband,ascomparedto removaloftheunderlyingcontinuum,wefitthreeGaussian
the other optical bands. There is also a NIR excess relative profiles to [O III] λ4363 and [O III] λλ4959, 5007, and find
to the I-band flux. In fact, this excess increases with time, that R = I[λ4959+λ5007]/I[λ4363] ≈ 1.7. As mentioned
I−K ≈2.8magonday∼560andI−K ≈3.7magonday above, the strong presence of [O III] λ4363 indicates a high
∼710,whichisindicativeofthegrowingimportanceofdust densityfortheemittingmaterial.Infact,withR≈1.7,the
in the emission from SN 2008iy. electron density, n , must be > 106 cm−3 regardless of the
e
A NIR excess at late times has been observed in many temperature of the emitting gas. Typically, [O III] emission
SNe IIn (e.g., Gerardy et al. 2002), and it is the result of comes from photoionised regions with T = 16,000–20,000,
either new dust formation (e.g., SN 2005ip; Smith et al. in which case n ≈ 107 cm−3 (see Fig. 11 in Filippenko &
e
2009; Fox et al. 2009) or the presence of a NIR light echo Halpern1984).Notethatthishigh-densitymaterialislikely
from preexisting dust (Dwek 1983), or both. To distinguish only present within clumps in the CSM (see Section 4.2),
between these two possibilities, which are virtually identi- andthatinterclumpportionsoftheCSMhavealowerden-
cal from photometry alone, requires well-sampled optical sity.
spectra (see Section 3.2.2) since line profiles are expected
to change with time if new dust is being formed. On day
3.2.1 The Hα Profile
706, corresponding to our last PAIRITEL observation, the
H−Ks colour of SN 2008iy was ∼1.2 mag. Assuming that AsseeninFig.3,theHαemissionfeaturedominatesoverall
the dust radiates as a perfect blackbody, this colour corre- theotherlines.Inthehigh-resolutionKeckspectrum,wesee
sponds to a dust temperature of Tdust ≈ 1320 K, while evidenceforthreedistinctemissionfeatures:abroadcompo-
theKs-bandmeasurementcorrespondstoadustluminosity nent (FWHM ≈ 4500 km s−1), an intermediate component
LNIR ≈ 4.8×1042 erg s−1, assuming no bolometric cor- (FWHM≈1650kms−1),andamarginallyresolved,narrow
rection as we cannot constrain the emission in the mid-IR. PCygnifeature(FWHM≈75kms−1),asshowninFig.4.
This luminosity is very large, though upon making similar The main panel of Fig. 4 shows a fit to the Hα profile (red
assumptions Gerardy et al. (2002) found NIR luminosities line)whichincludesabroad(FWHM≈4500kms−1)andan
>1042ergs−1forSNe1995Nand1997abatlatetimes.Ger- intermediate(FWHM≈1650kms−1)component.Panel(a)
ardy et al. also found that the NIR luminosity was roughly inFig.4showsaclose-upviewofthenarrowemissionafter
anorderofmagnitudegreaterthantheX-rayemissionfrom thebroadandintermediatecomponentshavebeenremoved
SN 1995N, which is also the case for SN 2008iy (see Sec- usingasplinefit.WefitthesplinetotheobservedHαprofile
tion 3.3). The H −Ks colour of SN 2008iy is bluer than from−1000kms−1to1000kms−1aftermaskingtheregion
thoseoftheGerardyetal.sampleatasimilarepoch,which between −200 km s−1 and 200 km s−1. A narrow absorp-
may be a result of the long rise of SN 2008iy or a contri- tion minimum such as this, located at ∼−100 km s−1, has
bution to the NIR emission from the SN in addition to the beenseeninanumberoftypicalSNeIIn(e.g.,SNe1997ab,
dust. Future NIR observations, as the underlying SN light 1997eg, and 1998S, Salamanca, Terlevich & Tenorio-Tagle
continues to fade, will place more stringent constraints on 2002;Salamanca2000;Fassiaetal. 2001),andittracesthe
the dust near SN 2008iy. outflowvelocityoftheprogenitorwind.Wenotethatwhile
thenarrowabsorptionisonlymarginallyresolved,aPCygni
profile with characteristic speeds of ∼10 km s−1, typical of
3.2 Spectroscopic Analysis
red supergiants (RSGs), would go completely unresolved in
Our spectra of SN 2008iy at ages >560 days are very simi- the Keck spectrum. Therefore, this P Cygni profile must
lar to those of an unpublished spectrum, from our spectral be associated with an outflow velocity that is >10 km s−1,
database, of the Type IIn SN 1988Z taken at a compara- which will have important consequences for the progenitor
ble epoch, as shown in Fig. 3. The general features resem- (see Section 4.3).
ble those of several other SNe IIn (Filippenko 1997): there It is interesting that the broad plus intermediate fit
are prominent Balmer and He I emission lines, most no- overestimates the Hα flux out to about −450 km s−1. In
tably the dominant Hα emission, which primarily feature panel (b) of Fig. 4 we show the residual flux, centred on
intermediate-width emission components with full width at Hα,followingthesubtractionofourtwo-Gaussianfittothe
half-maximumintensity(FWHM)∼1650kms−1inthecase broad plus intermediate emission. The blue velocity at zero
of SN 2008iy. The higher-resolution Keck spectrum reveals intensity (BVZI) extends to roughly −450 km s−1, which
anumberofnarrow,marginallyresolved(FWHM(cid:46)170km means that the true wind speed from the progenitor, as
s−1)emissionlines,includingHα,Hβ,[OIII]λλ4363,4959, traced by the absorbing gas moving directly along the line
5007,andHeIλ7065.Theonlyhigh-ionisationlinewedetect of sight toward the SN, may be >100 km s−1 and possibly
is [Fe X] λ6375. The relative lack of narrow forbidden lines as high as ∼450 km s−1. If this feature represents a true
andthelowintensityratioof[OIII]λλ4959,5007to[OIII] lack of emission, it would be evidence for a two-component
λ4363 suggest a large electron density for the ejecta (Filip- CSM: one with velocity v ≈ 100 km s−1 and the other
w,1
penko&Halpern1984;Filippenko1989;Stathakis&Sadler withBVZI≈v ≈450kms−1.Similarfeatureswereseen
w,2
1991).Therelativespectralfeaturesdonotshowstrongevo- in SN 1998S (Fassia et al. 2001), which were modelled by
lution between days 560 and 711. Chugaietal. (2002)tobeaslowwindacceleratedtohigher
(cid:13)c 2009RAS,MNRAS000,1–15
The 400 Day Rise of SN 2008iy 7
4
−15−2−1−1F (10 erg cm s Å)!123 SHN" , 2d0a0y8 i7y11 −−−000000000.........202442024 ((ba−))500 0 500 −15−2−1−1g [ F (10 erg cm s Å)]!01..10 SHN" 2008iy He I 6678! dddaaayyy 567651021 He I 7065!
e I lo
H
0
−4000 −2000 0 2000 4000 6000 −20000 −10000 0 10000 20000
velocity (km s−1) velocity (km s−1)
Figure4.DetailedviewoftheHαprofilefromtheKeckspectrum Figure 5.EvolutionoftheHαprofileofSN2008iybetweenday
takenonday711.Mainpanel:Theblacklineshowstheobserved 560andday711.Theprofileremainslargelyunchanged,withthe
Hα profile, while the smooth red line illustrates a two-Gaussian exceptionofavariablespectralfeaturebetween∼200kms−1and
fit(intermediate,FWHM≈1650kms−1;broad,FWHM≈4500 20,000kms−1,fromday652today711.Similarspectralfeatures
kms−1)totheprofile.Residualemissionandabsorptionfroma havebeenobservedinotherinteractingSNe(seetext).Emission
narrowcomponenttotheprofilecaneasilybeseen.Thelocation linesofHeIλ6678andHeIλ7065arelabelled.
of He I λ6678 has been marked, and the residuals from an im-
perfecttelluricabsorptioncorrectionhavebeenmarkedbythe⊕
symbol.Inset(a):Close-upviewoftheresidual,narrowPCygni the spectrum from day 567, as it was taken under cloudy
component.Thesmoothredlineshowsafittothefeaturethatin- conditions and the overall flux scaling is uncertain, but we
cludesaGaussianinemissionandaGaussianinabsorption(both do not expect significant spectral evolution between days
GaussianshaveFWHM≈75kms−1).Theabsorptionminimum
560 and 567. The decline in the Hα luminosity is accompa-
of the narrow absorption is located at −100 km s−1. Note that
nied by a rise in the equivalent width over this same time
for this fit we removed the broad plus intermediate components
period. For the narrow emission on day 711, we measure
using a spline fit over the region from −1000 km s−1 to 1000
km s−1(see text). Inset (b): Residuals following the subtraction L(Hα) = (2.8±0.6)×1039 erg s−1, though we note that
ofthetwo-Gaussianbroadplusintermediatefitfromtheobserved thisvalueislikelyunderestimatedbecausetheblueemission
Hαprofile.Theabsorptionminimumstilloccursat−100kms−1, wing is probably being partially absorbed. The FWHM of
buttheBVZIextendsto∼−450kms−1,meaningtheprogenitor theintermediatecomponentis∼1650kms−1 atallepochs,
windspeedmaybe>100kms−1 (seetext). whiletheFWHMofthebroadcomponentdropsfrom∼5800
kms−1 onday560,to∼5200kms−1 onday650,to∼4400
km s−1 on day 711. We attribute this change in the broad
velocitiesbyradiationfromtheSNphotosphere, and inSN
component to rapid evolution of the spectrum redward of
2006gy, which Smith et al. (2010) argued was the result
Hα between days 560 and 711, as shown in Fig. 5. The na-
of a CSM shell that had been ejected from the progenitor
ture of this feature is not currently understood: it could
in a pre-SN eruption. These two scenarios predict different
potentially be an absorption feature, or it could be related
behaviour as the SN evolves.
toalackofemissionfromsomeunidentifiedspecies.Similar
In the radiatively accelerated scenario, as shown by
spectralevolutionredwardofHαhasbeenseenintheinter-
Chugaietal. (2002)forSN1998S,thesecond,fastercompo-
actingSNe2005gjand2006gy(Prietoetal. 2007;Smithet
nent to the wind has a negative velocity gradient. Observa-
al. 2010). We do not associate this feature with Hα, as the
tionally,thiseffectmanifestsitselfasaBVZIforthesecond, red velocity at zero intensity corresponds to velocities of (cid:38)
fasterCSMcomponentthatdecreaseswithtime,asobserved 22,000 km s−1, and there are no other emission lines that
in SN 1998S (Fassia et al. 2001). In the shell-ejection sce-
show evidence for velocities this large.
nario, the wind exhibits a positive velocity gradient as the
ejectedshellisfreelyexpanding,whichisthoughttobethe
resultofanexplosive(∼1049−1050erg)mass-lossevent(see
3.2.2 He I Emission
e.g.,Chugaietal. 2004).ThisscenarioleadstoaBVZIthat
increases with time, as was observed for SN 2006gy (Smith In addition to the prominent Balmer emission lines,
et al. 2010). With only a single high-resolution spectrum, SN 2008iy exhibits several He I lines in emission, includ-
weareunabletoprobetheevolutionofthisfeature,butwe ing λλ3820, 4026, 4471, 5016, 5875, 6678, and 7065. None
note that the velocity of a radiatively accelerated CSM is oftheselinesshowevidenceforabroadcomponent,andthe
proportionaltot−2(Fransson,Lundqvist&Chevalier1996), intermediate component, FWHM ≈ 1100–1400 km s−1, is
so this mechanism is unlikely to be significant at late times slightly narrower than that observed for Hα. As previously
(> a few hundred days). mentioned, He I λ7065 has a narrow, marginally resolved
As does the overall continuum, the Hα luminosity de- emissioncomponentinadditiontotheintermediatecompo-
clines over the course of our observations. In Table 6 we nent.
summarise the observed properties of the Hα profile from A systematic, increasing blueshift in the He I profiles
our spectra on days 560, 652, and 711. We do not include has been argued as evidence for dust formation in the cool,
(cid:13)c 2009RAS,MNRAS000,1–15
8 Miller et al.
Table 6.Hαlinewidths,luminosity,equivalentwidth,andratiotoHβ.
Epocha Br.FWHM Int.FWHM Nar.FWHM L(Hα)b EWc Hα/Hβ
(kms−1) (kms−1) (kms−1) (ergs−1) (˚A)
560 5774 1641 ... 7.86×1041 2122 9.49
652 5162 1685 ... 6.10×1041 2386 12.22
711 4405 1639 78 6.50×1041 2786 12.54
a Definedasrest-framedaysrelativetoday0,2007Sep.13.
b Uncertainty∼±10%.
c Uncertainty∼±10%.
denseshellformedinthepost-shockgasoftheTypeIInSNe 3.1), implying that radioactivity contributes minimally to
1998S and 2005ip (Pozzo et al. 2004; Smith et al. 2009). the light curve of SN 2008iy. Furthermore, the SN ejecta
Newly formed dust in the post-shock region absorbs light will suffer considerable adiabatic losses during the 400 day
from the receding ejecta, resulting in a suppression of flux riseofthelightcurve,suggestingthatinteractionistheonly
fromtheredsideoftheemissionline.UsingHeIλ7065,the viable mechanism for powering SN 2008iy over this entire
strongest He line in our spectra, we searched for changes time.
in the red wing of the profile. The Kast spectra from days Following our assumption of interaction-driven lumi-
560and567areverynoisy,butinthehighersignal-to-noise nosity, the expelled gas from the progenitor overtaken by
ratio spectra from days 652 and 711 we see no evidence for the shock at each epoch can be written as (see Chugai &
a change in the red wing of the emission profile. This hints Danziger 1994)
that the observed NIR excess (see Section 3.1) is the result
2 v
of a NIR echo, rather than newly formed dust, though we M˙ = L w , (1)
ψ V3
caution that with only two spectra separated by ∼50 days SN
the lack of change in the He I profile does not constitute where M˙ is the mass-loss rate of the progenitor, ψ is an
definitive evidence for this. efficiency factor describing the conversion of kinetic energy
intoradiation,ListheluminosityoftheSN,v isthewind
w
speedoftheprogenitor,andV isthevelocityoftheblast
SN
3.3 X-ray Analysis
wave overrunning the CSM. We adopt ψ = 0.5, but this
TodeterminetheX-rayluminosityofSN2008iyweassume numberislikelyanoverestimategiventheopticallythinna-
a thermal plasma spectrum with temperature 0.6 keV (see, tureoftheemission(seeSection4.2).TheCSMwindisonly
e.g., Immler & Kuntz 2005), which has been absorbed by a probed by our higher-resolution spectrum from day 711, so
Galacticcolumndensityofn(H)=4.8×1020cm−2(Kalberla we adopt vw = 100 km s−1 based on the narrow P Cygni
et al. 2005). Using the 0.3–10 keV XRT count rate (see absorption seen in Hα, though this may not be valid at all
Section2.1)andPIMMS5,wemeasuretheunabsorbedflux, timesbetweendays0and700(seebelow).WeadoptVSN =
assuming no absorption within the host, to be (6.1 ± 1.9) 5000 km s−1 based on the typical widths of the broad Hα
×10−14 ergcm−2 s−1.AttheadopteddistanceofSN2008iy emission, which should trace the speed of the blast wave.
this corresponds to a luminosity of L =(2.4±0.8)×1041 Unlike our assumption of a constant wind speed, the adop-
X
erg s−1. Note that Fox et al. (2000) fit the late-time (> 3 tion of a constant-velocity blast wave at all times should at
yr)X-rayspectraofSN1995Nwithalargertemperatureof the very least be valid after the peak, at which point the
9.1keV,whichwouldcorrespondtoafluxandluminosityof CSM density is decreasing with increasing radius. Hydro-
(9.1±2.9)×10−14 ergcm−2 s−1 and(3.5±1.1)×1041 erg dynamic modelling of SNe IIn demonstrates that the blast
s−1,respectively,werewetoadoptthisvalueforSN2008iy. waveisquicklydeceleratedtoaroughlyconstantexpansion
Similar luminosities (L > 1041 erg s−1) are observed for speed as the ejecta sweep up successively more CSM (see,
X
otherinteractingSNemorethan1yearpostexplosion(e.g., e.g., Chugai et al. 2004). This scenario may not apply to
SNe 1988Z and 1995N; Fox et al. 2000). SN2008iy,however,becausethe∼400dayrisetimeimplies
thattheCSMdensitymayhavebeenincreasingwithradius,
inwhichcasetheblastwavemayhavebeencontinuallyde-
celerated during the rise.
4 DISCUSSION
FollowingtheaboveassumptionswecanuseEquation1
4.1 Mass Loss in the SN 2008iy Progenitor to determine the mass loss traced by the optical luminosity
at a number of different epochs of interest. At each epoch
Weassumethatthecontinuumemissiontracedbythelight
we determine the radius, R=V t ≈ 5000 km s−1 t ,
curve of SN 2008iy is powered solely by CSM interaction. SN SN SN
andtheluminositybasedonthelightcurveshowninFig.1.
The half life of 56Ni is τ ≈ 6.0 days, much too short to
We do not adopt the bolometric correction determined in
powerthe>400dayriseofSN2008iy,whilethephotometric
Section 3.1 for two reasons: (i) it is unclear whether this
decay rates are too slow to be powered by 56Co (see Sec.
samecorrectionisvalidatearlytimes,and(ii)emissionfrom
a NIR echo should not be incorporated into Equation 1, as
5 http://heasarc.nasa.gov/Tools/w3pimms.html. that contribution to the total luminosity is not directly the
(cid:13)c 2009RAS,MNRAS000,1–15
The 400 Day Rise of SN 2008iy 9
resultofCSMinteraction.Thus,theestimatesbelowofthe coefficient for Hα, hν is the energy of an Hα photon, x is
23
luminosity do not reflect any emission from dust. At early thedegreeofHionisation,X isthehydrogenmassfraction,
times, day ∼153, L ≈ 6.8×108 L , which corresponds to andN isAvogadro’snumber.Wecannotconstrainx,X,or
(cid:12) A
M˙ ≈ 1.3 × 10−2 M yr−1 at R ≈ 6.6×1015 cm, while at r , so we conservatively adopt x=1, X =1, and r (cid:29)r ,
(cid:12) 2 2 1
the time of peak L≈1.2×109 L , R≈1.7×1016 cm, and allofwhichhavetheeffectofminimisingw.AssumingCase
(cid:12)
M˙ ≈2.3×10−2 M yr−1.Onday563,coincidentwithour B Hα recombination, α =8.64×10−14 cm3 s−1, which is
(cid:12) 32
X-rayobservationsofSN2008iy theopticalluminositywas appropriate assuming that the narrow emission comes from
L≈7×108L ,whichcorrespondstoR≈2.4×1016cmand photoionised gas. Substituting the narrow Hα luminosity
(cid:12)
M˙ ≈ 1.3 × 10−2 M yr−1. On day 714, roughly coincident (L(Hα)=2.8×1039 ergs−1)andr =V t =3.0×1016
(cid:12) 1 SN SN
with our high-resolution Keck spectrum, L ≈ 4×108 L , cm into Equation 3, we find that w ≈ 1017 g cm−1. With
(cid:12)
R≈3.0×1016 cm, and M˙ ≈ 0.8 × 10−2 M yr−1. v = 100 km s−1 on day 711, this corresponds to M˙ ≈
(cid:12) w
These results are quite remarkable; they suggest that 1.6×10−2 M yr−1.Atasimilarepoch(seeabove),wefind
(cid:12)
at a time t ≈ V t /v ≈ 5000 km s−1 × 400 day/100 M˙ ≈ 0.8×10−2 M yr−1 based on the optical luminosity.
SN SN w (cid:12)
km s−1 ≈ 55 years prior to the explosion of SN 2008iy, the Again,theagreementbetweenthesetwomethodstowithin
progenitorunderwentaperiodofheightenedmassloss.This a factor of ∼2 is reasonable given the uncertainties in our
enhancedmass-lossperiodwasthenfollowedbyaperiodof adopted parameters.
decreasing mass loss leading up to the time of the SN ex-
plosion.Whilethequantitativeresultsforthemasslossdis-
4.2 Late-Time Emission and the Similarity to SN
cussed above are sensitive to our adopted quantities, most
1988Z
specifically the progenitor wind speed and the speed of the
SN blast wave, the fact remains that the continuum lumi- To explain the late-time (i.e., post peak) emission from
nosity did increase over a period of ∼400 days after the SN 2008iy we adopt a model that is virtually identical to
SNexplosion.RearrangementofEquation1showsthatthis thatdevelopedbyChugai&Danziger(1994)forSN1988Z:
luminosity increase is proportional to the wind-density pa- theCSMcontainsopticallythickclumpsinadditiontoarar-
rameter, w = M˙/vw. Thus, w must have increased over a efiedwindbetweentheseclumps(see,e.g.,Fig.1ofChugai
distance R ≈VSNtSN ≈1.7×1016 cm from the progenitor, &Danziger1994).AstheSNejectasweepthroughtheCSM,
regardlessofthetruewindspeedoverthatdistance.Anin- they drive a fast-moving forward shock into the rarefied
creasing value of w means that during roughly the century wind.Thisshockedmaterialformsacooldenseshellbetween
prior to the SN, there was a significant change in the wind the forward and reverse shocks, and gives rise to the broad
properties of the progenitor. emissioncomponentseeninHα(FWHM≈5000kms−1 for
Using our X-ray detection of SN 2008iy we have an al- SN2008iy).Atthesametimeaslowershockisbeingdriven
ternative method to probe the mass-loss history of the pro- intothedenseclumps,whichleadstotheintermediate-width
genitoronday∼563.FollowingImmler&Kuntz(2005),the emission (FWHM ≈ 1650 km s−1 in the case of SN 2008iy)
X-ray luminosity may be written as seeninthespectra.ThenarrowPCygniprofileresultsfrom
4 (cid:18)M˙ (cid:19)2 the pre-shock photoionised wind.
L = Λ(T) (V t)−1, (2) The observed similarities between SN 2008iy and
X (πm)2 v SN
w SN 1988Z provide further evidence that comparable mod-
els are appropriate for the two SNe. In addition to hav-
where L is the X-ray luminosity, m is the mean mass per
X
particle(m=2.1×10−24gforaH+Heplasma),Λ(T)isthe ing similar spectra at late times (see Fig. 3), SNe 1988Z
coolingfunctionofaplasmaheatedtotemperatureT,M˙ is and 2008iy have large X-ray luminosities (both SNe have
L (cid:38)1041 ergs−1;Fabian&Terlevich1996,thiswork),and
themass-lossrateoftheprogenitor,v isthewindspeedof X
the progenitor (∼100 km s−1, see abowve), V is the speed similar late-time decline rates (∼0.3−0.5 mag (100 day)−1
SN
of the SN blast wave (∼ 5000 km s−1, see above), and t is intheoptical;Turattoetal. 1993,thiswork),whichinboth
casesareslowerthantheexpectedbolometricdeclineofra-
the time since explosion, ∼563 days. Assuming a tempera-
ture T = 107 K, appropriate for an optically thin thermal dioactive 56Co. Estimates for the mass-loss rate from the
progenitor of SN 1988Z vary by roughly an order of mag-
plasma (Immler & Kuntz 2005), the effective cooling func-
tionisΛ(T)=3×10−23 ergcm3 s−1.TheX-rayluminosity nitude. From X-ray observations Schlegel & Petre (2006)
onday563wasL ≈2.4×1041 ergs−1.Substitutingthese find M˙1988Z ≈10−3 M(cid:12) yr−1, while Williams et al. (2002)
values into EquatXion 2, we find M˙ ≈ 0.7×10−2 M yr−1, estimate M˙1988Z ≈ 10−4 M(cid:12) yr−1 based on radio observa-
(cid:12)
tions, and the models of the optical emission by Chugai &
whichshowsagreementwiththevaluederivedfromtheopti-
calcontinuum,∼1.3×10−2 M yr−1,giventheuncertainty Danziger (1994) yield M˙1988Z ≈ 7×10−4 M(cid:12) yr−1. While
(cid:12)
theseestimatesare1–2ordersofmagnitudelessthanthose
in a number of the assumptions we have adopted.
for SN 2008iy, we note that in each of the above estimates
Anadditionalestimateofthemass-lossrateofthepro-
for M˙ a wind speed of 10 km s−1 was adopted for the
genitorcomesfromthenarrowHαemissionseeninourKeck 1988Z
progenitorofSN1988Z.IfSN1988Zhadaprogenitorwind
spectrum on day 711. Chugai & Danziger (2003) show that
speed closer to ∼100 km s−1, a very reasonable possibility
the Hα emission from the unshocked CSM is related to w:
giventhewindspeedofSN2008iyandothersimilaritiesbe-
1 (cid:16) r (cid:17)
L(Hα)= α hν (xXwN )2 1− 1 , (3) tween the two SNe, this would result in an increase in the
4πr 32 23 A r
1 2 estimated mass-loss for the progenitor by a factor of ∼10,
wherer istheinnerradiuscorrespondingtothepositionof therebybringingtheestimatesforthetwoSNeintoaccord.
1
theSNblastwave,r istheouterradiusrelatedtothefast- Based on the late-time emission, SN 2008iy seems to
2
moving forward shock, α is the effective recombination belongtothegroupofSNeIInthatexhibitaslowevolution
32
(cid:13)c 2009RAS,MNRAS000,1–15
10 Miller et al.
sustained by long-lived ((cid:38) 1 decade) CSM interaction. In centurypriortoexplosion,thiscouldresultinadensitypro-
addition to SN 1988Z, other members of this group include file that peaks ∼1.7 ×1016 cm from the progenitor. In the
SN1986J(Rupenetal. 1987),SN1995N(Foxetal. 2000; clumpy-wind scenario described in Section 4.2, this would
Fransson et al. 2002), and possibly also the VLSN 2003ma mean that the ejecta are expanding into a wind where the
(Rest et al. 2009). Both SN 1986J and SN 1995N exhibit numberdensityofclumpsisincreasingwithradius.Thus,as
evidence for a clumpy progenitor wind, like SN 2008iy and moreandmoreclumpsareovertakenbytheejecta,thecon-
SN1988Z.Chugai(1993)modelledtheX-rayemissionfrom tinuum luminosity continues to rise, until after ∼400 days
SN 1986J as the interaction between the SN ejecta and a when the ejecta have reached ∼1.7 ×1016 cm, the number
clumpy wind (though note that Houck et al. 1998 prefer a density of clumps begins to decline and so does the optical
modelwithasmoothCSM,buttheyconcludethattheycan- luminosity.Alternatively,thelongrisecouldresultfromthe
notruleouttheclumpymodel).Theevidenceforaclumpy ejectaexpandingintoanonsphericalwind,suchasabipolar
windfromtheprogenitorofSN1995Ncomesfromboththe outflow,thoughthishypothesiswouldneedtobeexamined
optical spectra (Fransson et al. 2002) and the X-ray emis- with detailed hydrodynamical models.
sion(Zampierietal. 2005).Also,theopticaldeclineofboth The inferred mass-loss rate for SN 2008iy, M˙ ≈ 10−2
SNe1986Jand1995Nisveryslow:SN1986Jdeclinedby(cid:46) M yr−1,issimilartothatforthefirstLBV,PCygni,dur-
(cid:12)
1magintheopticalbetween1994and2003(Milisavljevicet ing its great eruption (M˙ ≈10−2 M yr−1 during the
PCyg (cid:12)
al. 2008), while the V-band decline of SN 1995N was only 1600 AD eruption; Smith & Hartigan 2006). Furthermore,
2.2 mag between 1998 and 2003 (Zampieri et al. 2005). theterminalwindspeedinthePCygninebulais185kms−1
The duration of the interaction means that the large (Lamersetal. 1996;Najarro,Hillier&Stahl1997),whichis
mass-loss rates from the respective progenitors must have similartotheobservedBVZIforthenarrowblueabsorption
been sustained for at least ∼ 100 yr prior to core collapse seen in SN 2008iy, corresponding to v ≈160–450 km s−1.
w
aftertheconservativeassumptionthatV =1000kms−1, We illustrate these comparisons to show that derived prop-
SN
andv =100kms−1.Typicallytheblastwavecontinuesto ertiesoftheprogenitorwindofSN2008iyaresimilartothe
w
expandat>1000kms−1(SN1988Zhadabroadcomponent giant eruption of a Galactic LBV; we are not insinuating
that remained nearly constant at ∼2000 km s−1 from day that the great outburst from P Cygni is a direct analogue
∼1500 to 3000; Aretxaga et al. 1999), which would make to the proposed LBV-like eruption from SN 2008iy. Such
thistimeperiod>100yr.Theseverylong-livedSNeIInare eruptions are not expected from RSGs.
therefore connected in that their progenitors experienced FurtherevidencethattheprogenitorofSN2008iycould
lengthy periods with high mass-loss rates. This stands in nothavehadaRSGprogenitorcomesfromtheobservednar-
starkcontrasttotheTypeIInSN1994W,whichhadalight row P Cygni Hα profile. RSGs have typical wind speeds of
curvewithanabruptdrop∼100dayspostexplosion(Soller- ∼10 km s−1, with extreme RSG winds reaching 40 km s−1
man,Cumming&Lundqvist1998).Themasslossfromthe (see Smith et al. 2007). The observed 100 km s−1 wind
progenitor of SN 1994W has been modelled to occur in a speed from SN 2008iy is more characteristic of LBVs (e.g.,
short ((cid:46)1 yr), violent episode (Chugai et al. 2004), and P Cygni, see above) or the escape velocity from a blue su-
the sudden drop in the light curve occurs because the SN pergiant. Similar ∼100 km s−1 P Cygni profiles have been
ejecta have overtaken the dense CSM, thereby halting any seeninanumberofSNeIIn,suchasSNe1997ab(Salamanca
interaction luminosity. 2000),1997eg(Salamanca,Terlevich&Tenorio-Tagle2002),
1998S(Fassiaetal. 2001),and2007rt(Trundleetal. 2009),
implyingthattheprogenitorsofeachoftheseSNemayhave
4.3 Origin of the 400 Day Rise Time and been in a similar state shortly before core collapse.
Implications for the Progenitor The dense CSM and large luminosities associated with
many SNe IIn have led a number of authors to suggest
With a model to account for the late-time emission from
that at least some SNe IIn are associated with progenitors
SN2008iy,wearestillleftwiththepuzzleofexplainingthe
thatexperiencedLBV-likemasslossshortlybeforecorecol-
400 day rise time.6 This rise in the optical is significantly
lapse (see e.g., Chu et al. 1999; Salamanca 2000; Chugai
longerthanthatseeninanyotherknownSN.Aspreviously
& Danziger 2003; Chugai et al. 2004). This possible con-
discussed, this scenario is possible if the progenitor under-
nection was considerably strengthened following the direct
went a phase of enhanced mass loss ∼55 years prior to the
identification of the progenitor of the normal Type IIn SN
SNexplosion.Theopticalluminositytracesthewind-density
2005glonarchivalHubbleSpaceTelescopeimages(Gal-Yam
parameter,w,meaningthatduringthedecadespriortoex-
et al. 2007; Gal-Yam & Leonard 2009). While the diagnos-
plosioneither(i)themass-lossratedeclined,or(ii)thewind
tics above are not ubiquitous for all SNe IIn7, many of the
speed increased, or (iii) both.
signatures (high-density CSM, episodic mass loss with M˙
Episodic periods of enhanced mass loss, sometimes in
≈10−2 M yr−1,and∼100kms−1 progenitorwindspeed)
the form of a shell ejection, have been observed for numer- (cid:12)
are shared with SN 2008iy, suggesting that it too had an
ous LBVs (see Humphreys & Davidson 1994), though the
LBV-like progenitor.
underlying physics of these eruptions is not currently well
As mentioned above, an alternative way to generate
understood(seeSmith&Owocki2006).Iftheprogenitorof
SN2008iyhadagianteruption(similartothatofLBVs)(cid:46)1 a wind-density parameter that is increasing with distance
6 We have no spectra of the SN while it was still on the rise, 7 Notethatthelow-resolutionspectrographstypicallyemployed
thuswehavenowayofknowingiftheSNunderwentsignificant to observe SNe lack sufficient resolution to resolve narrow
spectralevolutionpriortoday560. (FWHM(cid:46)150kms−1)absorptionlines.
(cid:13)c 2009RAS,MNRAS000,1–15