Interacting Supernovae and Supernova Impostors. SN 2009ip, is this the end?

0.94 MB·English

by  A. Pastorello

Draftversion October 15,2012
PreprinttypesetusingLATEXstyleemulateapjv.12/16/11
INTERACTING SUPERNOVAE AND SUPERNOVA IMPOSTORS.
I. SN 2009IP, IS THIS THE END?
A. Pastorello1⋆, E. Cappellaro1, C. Inserra2, S. J. Smartt2, G. Pignata3, S. Benetti1, S. Valenti4,5, M. Fraser2,
K. Taka´ts3,6, I. Arcavi7, S. Benitez8, M. T. Botticella9, J. Brimacombe10, F. Bufano3, F. Cellier-Holzem11, M.
T. Costado12, G. Cupani13, N. Elias-Rosa14, M. Ergon15, J. P. U. Fynbo16, M. Hamuy17, A. Harutyunyan18, K. M.
Ivarson19, E. Kankare20, R. Kotak2, A. P. LaCluyze19, K. Maguire21, S. Mattila22, J. Maza17, M. McCrum2, M.
Miluzio23, H. U. Norgaard-Nielsen16, M. C. Nysewander19, P. Ochner1, Y.-C. Pan21, M. L. Pumo1, D. E.
2 Reichart19, S. Taubenberger8, L. Tomasella1, M. Turatto1, and D. Wright2
1 Draft versionOctober 15, 2012
0
2 ABSTRACT
t We report the results of a 3 year-long dedicated monitoring campaign of a restless Luminous Blue
c Variable (LBV) in NGC 7259. The object, named SN 2009ip, was observed photometrically and
O
spectroscopically in the optical and near-infrared domains. We monitored a number of erupting
episodesinthe pastfew years,andincreasedthe densityofourobservationsduringeruptiveepisodes.
2
In this paper we present the full historical data set from 2009-2012 with multi-wavelength dense
1
coverageofthetwohighluminosityeventsbetweenAugust-September2012. Weconstructbolometric

light curves and measure the total luminosities of these eruptive or explosive events. We label them
]
R the 2012a event (lasting ∼50 days) with a peak of 3× 1041ergs−1, and the 2012b event (14 day
S rise time, still ongoing) with a peak of 8×1042 ergs−1. The latter event has reached an absolute
. R-band magnitude of about -18, comparable in brightness and luminosity to that of a core-collapse
h supernova (SN). Our historical monitoring has detected high-velocity spectral features (∼13000km
p
s−1) in September 2011, one year before the current SN-like event. This suggests that the detection
-
o of such high velocity outflows cannot, conclusively, point to a core-collapse SN origin. We suggest
r thattheinitialpeakinthe2012aeventwasunlikelytobeduetoafaintcore-collapseSN.Wepropose
t
s that the high intrinsic luminosity of the latest peak, the variability history of SN 2009ip, and the
a detection of broad spectral lines indicative of high-velocity ejecta are consistent with a pulsational
[ pair-instability event, in which the star may have survived the last outburst. The question of the

survival of the LBV progenitor star and its future fate remain open issues, only to be answered with
1
v future monitoring of this historically unique explosion.
8 Subject headings: supernovae: general — supernovae: individual (SN 2009ip), supernovae: individual
6 (SN 2000ch)
5
3
.
0 1INAF-Osservatorio Astronomico di Padova, Vicolo
1
dell’Osservatorio5,35122Padova,Italy
2 ⋆[email protected]
1 2Astrophysics Research Centre, School of Mathematics and
: Physics, Queen’s University Belfast, Belfast BT7 1NN, United
v Kingdom
i 3Departamento de Ciencias Fisicas, Universidad Andres
X Bello,Avda. Republica252,Santiago, Chile
4Las Cumbres Observatory Global Telescope Network, Inc.
r
a SantaBarbara,CA93117,USA
5Department of Physics, University of California Santa UAB,08193Bellaterra,Spain
Barbara,SantaBarbara,CA93106-9530, USA 15The Oskar Klein Centre, Department of Astronomy,
6Department of Optics & Quantum Electronics, University AlbaNova,Stockholm University,10691Stockholm,Sweden
ofSzeged, D´omt´er9,Szeged, H-6720,Hungary 16DarkCosmologyCentre,NielsBohrInstitute, Copenhagen
7Department of Particle Physics and Astrophysics, The University, Juliane Maries Vej 30, 2100 Copenhagen O, Den-
WeizmannInstitute ofScience, Rehovot76100, Israel mark
8Max-Planck-Institut fu¨r Astrophysik, Karl-Schwarzschild- 17Departamento de Astronom´ıa, Universidad de Chile,
Str. 1,85741Garching,Germany Casilla36-D,Santiago, Chile
9INAF-Osservatorio Astronomico di Capodimonte, Salita 18Telescopio Nazionale Galileo, Fundaci´on Galileo Galilei -
Moiariello16,I-80131Napoli,Italy INAF,RamblaJos´eAnaFern´andezP´erez,7,38712Bren˜aBaja,
10Coral Towers Observatory, Coral Towers, Esplanade, TF,Spain
Cairns4870, Australia 19University of North Carolina at Chapel Hill, Campus Box
11Laboratoire de Physique Nucl´eaire et des Hautes nergie, 3255,Chapel Hill,NC27599-3255, USA
Universit´e Pierre et Marie Curie Paris 6, Universit´e Paris 20Tuorla Observatory, Department of Physics and Astron-
Diderot, Paris 7, CNRS-IN2P3, 4 place Jussieu, F-75252 Paris omy,UniversityofTurku,Piikkio¨,21500,Finland
Cedex05,France 21Department of Physics (Astrophysics), University of
12Instituto de Astrof´ısica de Andaluc´ıa, CSIC, Apdo 3004, Oxford,DWB,KebleRoad,OxfordOX13RH,UitedKingdom
18080, Granada,Spain 22Finnish Centre for Astronomy with ESO (FINCA), Uni-
13INAF - Osservatorio Astronomico di Trieste, via Tiepolo versityofTurku,V¨ais¨al¨antie20,FI-21500, Piikkio¨,Finland
11,I-34143Trieste,Italy 23Department of Astronomy, Padova University, Vicolo
14Institut de Ciencies de l’Espai (IEEC-CSIC), Campus dellOsservatorio3,I-35122, Padova,Italy
2 Pastorello et al.
1. INTRODUCTION
Luminous Blue Variables (LBVs) are among the
most luminous and massive stars found in late-type
galaxies. In a few cases, these stars have been ob-
served to produce major eruptions that mimic a gen-
uine supernova (SN) explosion. For this reason, they
gained the label of SN impostors (Van Dyk et al. 2000).
The discrimination between SN impostors (i.e. LBV-
type eruptions) and type IIn SNe can be ambiguous
(see e.g. the SN 2011ht-like objects, Roming et al.
2012; Mauerhan et al. 2012a; Humphreys et al. 2012;
Kankare et al. 2012; Dessart et al. 2009; Chugai et al.
2004).
LBVs have been widely studied in the Milky
Way, Local Group galaxies and beyond (e.g.
Humphreys & Davidson 1994; Humphreys et al. 1999;
Maund et al. 2006; Smith et al. 2011b). They have
high mass-loss rates and frequently show what is known
as S-Doradus variability during which mass-loss is
enhanced, possibly due to temperature changes and
ionization balance of atomic species that drive the wind
(Vink & de Koter 2002). Giant eruptions have been
observed during which several solar masses of material
can be ejected and the intrinsic stellar luminosity
Fig.1.—SN2009ipinNGC7259,andreferencestarsinthehost
increases substantially. The physical mechanism that galaxyfield.
triggers these giant eruptions is still unknown. Based
on analysis of SN data, a link between some LBVs and first mentioned the possibility that SN 2009ip exploded
SNe IInhasbeenproposed(seee.g.Kotak & Vink 2006; asarealcore-collapseSN26.High-cadenceopticalimag-
Smith & Owocki 2006; Smith et al. 2007; Trundle et al. ing in the R and I bands showing the strong September
2008, 2009). There is one case (SN 2005gl) in which a 2012 re-brightening has been presented by Prieto et al.
likely LBV has been observed to explode as luminous (2012b). We also note that no (Margutti et al. 2012a,b;
SNe IIn (Gal-Yam et al. 2007; Gal-Yam & Leonard Chandra & Soderberg 2012; Hancock et al. 2012) or
2009), and one other case (SN 2010jl) for which there is marginal (Campana 2012; Margutti & Soderberg 2012)
a plausible argument for a massive progenitor star of a X-ray and radio detections of SN 2009ip have been re-
type IIn SN (M > 30 M⊙, Smith et al. 2011a). 25 ported so far.
In an exciting turn of events, a well observed LBV in In this paper we present observations of the LBV
the spiral galaxy NGC 7259 (designated as SN 2009ip known as SN 2009ip in NGC 7259 over a period of 3yrs
during a giant outburst in 2009) has recently been including: i) data showing erratic variability starting
proposed to have now exploded as a core-collapse SN from August 2009, when the object closely resembled
(Mauerhan et al. 2012b, and references therein). The NGC 3432-LBV1 (aka SN 2000ch, Wagner et al. 2004;
object was first discovered on August 26, 2009 by the Pastorello et al. 2010), a SN impostor that experienced
CHASESNSearch(Maza et al.2009)asafainttransient multipleenergeticoutbursts. OurdataofSN2009ipalso
at≈ 17.9mag,andwaslaterclassifiedasa SN impostor includeobservationsofrepeatedoutburstsduringthepe-
by a number of teams (Miller et al. 2009; Li et al. 2009; riodMayto October2011whichhavenot been reported
Berger et al.2009). The natureofSN2009ipwaswidely todate; ii)therecentevolutionoftheLBVasaputative
discussed in Smith et al. (2010, 2011b) and Foley et al. SN.
(2011). Through the analysis of pre-outburst archival
HST images these studies providedrobust evidence that 2. OBSERVATIONS
the progenitor was a very massive star (MZAMS > 60 Three years ago, after the first announcement of the
M⊙) that experienced repeated eruptions typical of the discoveryofatransientinNGC7259(Maza et al.2009),
LBV phase. we initiated an extensive spectroscopic and photometric
Subsequent re-brightenings were announced by the monitoring campaign in the optical bands using a num-
Catalina Real-Time Survey team on October 1, 2010 ber of telescopes available to our collaboration. After
(Drake et al. 2010) and, very recently, on July 24, 2012 about 100 days, the follow-up strategy was relaxed and
(Drake et al. 2012), which were first labeled as new the photometric monitoring was limited to the R band.
LBV-type eruptions (e.g. Foley et al. 2012). However,
from the detection of high-velocity spectral features on 26 We note that after the Smith & Mauerhan communica-
September 15 and 16, 2012 Smith & Mauerhan (2012a) tion there has been a proliferation of electronic telegrams on
this transient, with different interpretations on its nature - SN
vs. SN impostor - (Marguttietal. 2012a; Martinetal. 2012a;
25 We note that eruptions of Wolf-Rayet stars producing im- Brimacombe 2012; Marguttietal. 2012b; Smith&Mauerhan
postors with a luminosity similar to that of an LBV outburst 2012b; Leonardetal. 2012; Burgasseretal. 2012; Vinkoetal.
have later on been observed to explode as He-rich Ibn SNe 2012; Prietoetal. 2012a; Martinetal. 2012b; Galletal. 2012;
(Pastorelloetal.2007,2008a;Foleyetal.2007)orhybridIIn/Ibn Bohlsen 2012), although it is quite clear that most authors now
events (Pastorelloetal.2008b;Smithetal.2012). favortheSNexplosionscenario.
SN 2009ip, is this the end? 3
-18
-17
-16
-15
-14
-13
-12
-11
-10
-9
-8
0 200 400 600 800 1000 1200 1400
Days (from the discovery)
8
10
12
14
16
18
20
22
24
26
150 160 170 180 190 200 210 220 230
JD-2456000
16
18
20
22
60 70 80 90 100 110 120 130 140 150 160 170 180
JD-2455000
Fig.2.—Top: R-bandabsolutelightcurveofSN2009ip(bluediamonds)comparedwiththoseoftheimpostorNGC3432-LBV1(yellow
circles), the debated SN/impostor 1961V (photographic plate magnitudes, magenta dot-dashed line) and the historical visual light curve
of η Carinae during the period 1842-1845 (revised by Smith&Frew 2011, red dotted line). The cyan diamonds represent CRTS V-band
measurements(seealsoDrakeetal.2010,2012). ThedatashowingNGC3432-LBV1duringtheperiod2008-2012arefromPastorelloetal.
(2010), plus additional recent unpublished observations (see Appendix). The epoch 0of the η Carinaelight curveis year 1842.213 (UT).
TheerraticphotometricvariabilityisacommonpropertyofmajoreruptionsofLBVs. Middle: Ultra-violet/optical/near-infraredapparent
lightcurvesofthetransientfromAug8,2012,2weeksbeforethepublicationoftheannouncementofanewre-brighteningfromDrakeetal.
(2012). Shiftsof∆U=+0.27,∆B=+0.018and∆V=-0.042have beenappliedtotheu,b,vSwift/UVOTmagnitudes ofSN 2009ipto
matchtheU,B,VJohnson photometry. Theshiftshavebeencomputed afteracomparisonofthemagnitudes ofthe referencestarsinthe
SNfieldinthe twophotometric systems. Bottom: BVRIlightcurves oftheimpostorSN2009ipduringthe first3months fromthefirst
everdetection in2009(Mazaetal.2009).
4 Pastorello et al.
fromAugust 2009and spanning a periodof more than3
yearsisshowninFigure2(Toppanel)alongwiththatof
2005cs
asimilarevent,NGC3432-LBV1(Pastorello et al.2010),
43.5
1998S the debated transient (SN or impostor) 1961V (photo-
2009ip graphic mags, Bertola 1963, 1964, 1965, 1967) and the
43.0 revised visual light curve of the Giant Eruption of η
Carinae in 1842-1845 (see Smith et al. 2011b, and ref-
erencestherein). Thesamedistancemodulus(µ =31.55
]
1−42.5 2012b mag) and interstellar extinction (AR = 0.051) adopted
g s event by Smith et al. (2010) and Mauerhan et al. (2012b) for
r SN 2009ip have been used in the absolute R-band light
e
[)42.0 2012a curve of Figure 2. The erratic light curves of all these
L event transients show similar features. SN 2009ip experienced
(
g afewintenseeruptivephases,includingthoseonAugust-
o
l41.5 September 2009and fromMay to October 2011,charac-
terizedby a sequence of sharpluminosity peaks followed
by rapid magnitude declines; the multi-band light curve
41.0 of the 2009 event covering a period of about 3 months
is shown in the bottom panel. The 2009 eruptive phase
presents the erratic evolutiontypicalof on LBV-type gi-
40.5 anteruption,andisverysimilartothoseobservedinthe
40 60 80 100 120 Giant Eruption of η Carinae and in NGC 3432-LBV1.
days [after JD 2456100]
Other re-brightenings were registered by CRTS (to
magnitudes V∼ 17 on Jul 15,2010,and V∼ 17.7 on Sep
29, 2010, Drake et al. 2010, shown as cyan diamonds in
Fig.3.—BolometriclightcurveofSN2009ipfromAugusttoOc- Figure 2, top). Older records(before August 2009)from
tober2012(showingboththe2012aand2012bevents), compared
the CRTS archive29 and from Smith et al. (2010) have
with the bolometric light curves of the faint type IIP SN 2005cs
(Brownetal. 2007; Pastorelloetal. 2006, 2009) and the type never registered the transient at a magnitude brighter
IIn/IILSN1998S(Liuetal.200;Fassiaetal.2000;Gerardyetal. than about ∼ 20.4. These new data are more compre-
2002;Pozzoetal.2004). ThelightcurvesofSNe2005csand1998S hensive, and reveal a recent variability history for SN
areshowninanarbitrarytemporalscaletowellmatchrespectively
2009ipwhichismorecomplexthanonecaninferfromthe
the2012aand2012beruptiveevents ofSN2009ip.
schematic light curve representation of Mauerhan et al.
Due to its unpredictable behavior, we kept up a mon- (2012b).
itoring campaign of this object during the following 3 DuringJuly-August2012anewre-brighteningwasan-
years. nounced by Drake et al. (2012, cyan diamonds in Figure
Aftertherecentre-brighteningofSN2009ipannounced 2, top). This event was then followed by a strong un-
bytheCatalinaReal-TimeSurveyteamonJuly24,2012 precedentedburst(startingaroundSeptember23)which
(Drake et al.2012),weintensifiedourobservingcadence isabout30times moreluminousthanthepreviousoscil-
and secured multi-color photometry and spectroscopy lations. ThisSN-likeriseinluminositywillbeextensively
from the optical to the near-IR domains. In addition, discussedlaterinthispaper. ThephasebetweenAugust
SWIFT optical and ultra-violet observations have been to October 2012 is shown in the middle panel of Figure
triggered (PIs: R. Margutti and P. W. A. Roming) and 2, with panchromaticlightcurvesof SN2009ipobtained
included in our analysis27, particularly to give a wide with the ground-based telescopes and the SWIFT satel-
wavelengthbolometric light curve of the 2012 eruptions. lite.
Figure 3 shows the bolometric light curve of SN
2.1. Photometry 2009ip from the August 2012 re-brightening announced
by Drake et al. (2012) to the current epoch. It appears
SN 2009ipis located close to a red (R = 18.05± 0.04,
to show 2 distinct phases: a broader (and fainter) ear-
R-I = 0.72 ± 0.05) foreground star, in a remote posi-
lier peak (that we will label as “2012a event” for sim-
tion North-East of the host galaxy (Figure 1). Our pho-
plicity), that ends around September 23 and reaches
tometric measurements were performed using the PSF-
a luminosity of 3 × 1041 erg s−1, and a fast-rising,
fitting technique, with the simultaneous fit of the tran-
higher luminosity second peak (“2012b event”) with a
sientandthenearbystar. Anumberofreferencestarsin
the SN field were calibrated using observations of stan- maximum at about 8 × 1042 erg s−1. Mauerhan et al.
dard fields from the catalog of Landolt (1992), and used
see their Table 2) show a large and significant discrepancy with
to improve the photometric calibration of SN 2009ip in
ourphotometry. Inparticular,thephotometry ofSN 2009ipfrom
non-photometric nights. The final photometry of the Mauerhanetal. (2012b) is too faint by more than 2 mags in the
transient and the magnitudes of the reference stars are Bbandand0.3intheIband. TheR-banddataareinreasonable
listed in Appendix28. agreementtotheorderofafewhundredthsofamagnitude. Hence
theaverageB-RcolorcomputedwiththeMauerhanetal. photom-
The R-bandabsolutelightcurveofSN2009ipstarting etry is ∼ 2.5, vs. B-R ∼ 0.5-0.7 that is calculated with our data.
We are confident that our calibration is correct because we used
27 Independent measurements using the images from the same standardphotometriccalibrationproceduresbasedontheLandolt
datasetwerepublishedinMauerhanetal.(2012b). catalog,andourphotometryprovidesbluecolorsthatareinagree-
28 WefindexcellentagreementwiththeCRTSandPrietoetal. mentwiththeverybluespectralcontinuum ofthistransient.
(2012b) photometry, while the data of Mauerhanetal. (2012b, 29 http://nesssi.cacr.caltech.edu/catalina/current.html
SN 2009ip, is this the end? 5
2000 3000 4000 5000 6000 7000 8000 9000
Fig.4.—Sequence ofspectra of the LBV inNGC 7259, obtained fromSeptember 2009 toSeptember 2011. Allspectra areinthe host
galaxywavelength frame.
(2012b)notedthatthemaximumluminosityofthe2012a luminosity, more rapidly in the ultra-violet and the blue
event is consistent with the luminosity of a faint SN IIP optical bands. We will see in Section 3 that the 2012a
(Pastorello et al. 2004), although with a faster evolving and 2012b sequence of events may have an alternative
light curve. Along with spectral similarities, this led explanation.
Mauerhan et al.(2012b)to suggestthatthe 2012aevent We also remark that none of the comparison objects
was the true core-collapse SN event of the LBV star. in Figure 2 shows the regular, SN-like light curve that
We confirm that the bolometric luminosity of the 2012a characterized SN 2009ip during the 2012b event. This
event is similar to SN 2005cs, as one can note from Fig- late photometric evolution combined with the bright lu-
ure3. Thesubsequentfasterrisetothesecondpeak(the minous peak (MR ≈ -18) may support the claim that at
2012bevent)presentsaneventightersimilaritywiththat least during the 2012b event SN 2009ip has finally ex-
of the type IIn/IIL SN 1998S.The 2012beventwas pro- plodedasarealsupernova. WenotethatthecolorofSN
posedbyMauerhan et al.(2012b)tobeduetostrongSN 2009ip at the light curve peak (on October 6, 2012) is
ejecta-CSM interaction. We measure a 2 week long rise- U-V≈-1mag,significantlybluer thanthatofthe 2012a
time, reaching a peak apparent magnitude of B = 13.80 event at maximum (U-V ≈ -0.5 mag). At the pre-burst
(R = 13.65) on October 6, 2012 and then it declines in minimum of September 23, the U-V color was instead
6 Pastorello et al.
2000 3000 4000 5000 6000 7000 8000 9000
Fig.5.— Sequence of spectra obtained between August and September 2012, including those of the putative SN explosion. A higher
resolutionXShooterspectrumobtainedonSeptember24,2011,i.e. beforethe2012re-brightening,isalsoshownatthetopofthesequence
(greencolor). Allspectraareallinthehostgalaxywavelength frame.
significantly redder, i.e. ≈ 0 mag. The spectra relative to the 2009 outburst reported in
Figure 4 are all dominated by prominent Balmer lines
2.2. Spectroscopy
with a complex profile. The weak absorption features
Optical and near-infrared spectra of SN 2009ip (Fig- indicate that the bulk of the ejected material is mov-
ures 4, 5, 6 and 7; the log of observations is in Ap- ing with a velocity of 2900 ± 700 km s−1, but the blue
pendix) were collected using the 8.2-m Very Large Tele- edgeofthe isolatedHβ absorptionsuggeststhe presence
scope (VLT) UT1 (+ FORS) and UT2 (+XShooter) at of fast-moving material which is expanding at a velocity
the Cerro ParanalObservatory (ESO Chile), the 3.58-m of about 5000-6000 km s−1. The Hα emission compo-
ESO-NTT(+EFOSC2andSOFI)attheLaSillaObser- nent in September 2009 has a Lorentzian profile with
vatory (ESO Chile), the 8.2-m Gemini South Telescope a FWHM velocity of about 700-800 km s−1, which in-
(with GMOS) in CerroPacho´n(Chile), the 3.58-mTele- creases to about 1100-1200 km s−1 during the period
scopioNazionaleGalileo(TNG,equippedwithLRS),the October-November,2009 (when the object was receding
4.2-mWilliamHerschelTelescope(WHT,withISIS)and to a more quiescent stage).
the 2.56-m Nordic Optical Telescope (+ ALFOSC) lo- Afterafurtheroutburst(September2010)reportedby
cated in La Palma (Canary Islands, Spain).
SN 2009ip, is this the end? 7
Drake et al. (2010), a spectrum obtained on October 6, for the 2012a event. The cores of the absorptions of
2010showsSN2009ipatasimilarstageastheNovember the Balmer features indicate expansion velocity of the
24, 2009 spectrum, i.e. with the star again quiescent. ejected materialof∼ 5000-6000km s−1 (4200± 500 km
TheFWHMvelocityoftheLorentzianHαcomponentin s−1 from the Fe II lines), but the blue edge of the wings
this phase is still around 1300 km s−1. The September still reach to much higher velocities (about 13800 km
2, 2011 VLT spectrum reported at the bottom of Figure s−1). Figure 6 (top) shows a comparison of SN 2009ip
4 shows SN 2009ip to be back to a dormant stage, and at 3 representative epochs (September 24, 2011; August
the FWHM velocity of the Lorentzian Hα component is 8 and September 28, 2012) with a spectrum of NGC
about 940 km s−1. 3432-LBV1 in outburst (April 24 2009). The high ve-
Figure 5 shows the spectra of the transient during locity P-Cygni absorption (in the Balmer lines) is cer-
the period August-October 2012, compared with a VLT tainly stronger in the 2012a event than we observed in
spectrum obtained on September 24, 2011 (green line), 2011 and in NGC3432-LBV1 in outburst, but we illus-
during another outburst episode. In the September 24, trate here that the detection of high velocity gas is not
2011 spectrum, the FWHM velocity of Hα, which still only restricted to core-collapse SNe. Similar high ve-
hasaLorentzianprofile,hasslightlydecreasedtoaround locity edges are clearly detected in SN 2009ip in 2011
790 km s−1, and other Balmer lines clearly show very (13800km s−1) and in NGC3432-LBV1(∼9000km s−1,
broad absorption components, with a blue edge that in- Pastorello et al. 2010). We will discuss the implications
dicates that there is material moving with a velocity as of this in Section 3.
high as 12500 km s−1 already at this epoch (see also These broad absorptions disappear at the time of the
Figure 6). This is the highest velocity outflow that has 2012b event, in September 28 and October 4 spectra
been detected in an LBV-like eruption of any sort and (Figures 5 & 6, top), when the luminosity of SN 2009ip
indicates that high velocities are observed without core- reaches the unprecedented maximum. At these times,
collapseorthe catastrophicdestructionofthe star. This the spectra are very similar to those of many type IIn
hasimportantconsequencesfortheinterpretationofhigh SNe (e.g. SN 1999el, Di Carlo et al. 2002), with the
velocity ejecta as evidence for the core-collapse mecha- H lines presenting a narrow emission component with
nism in the 2012a event. We subsequently obtained an a FWHM velocity of about 290 km s−1 and very broad
NTT spectrum on August 8, 2012 (JD = 2456148.91, wings (∼3600 km s−1). Similar velocities are measured
i.e. 10 days before the new outburst - the 2012a event - in the He I lines, which arenow more prominent than in
wasannouncedbyDrake et al.2012). Thebroadabsorp- past spectra, whilst the Fe II lines are no longer visible.
tion features were present also atthis epoch, and indeed The spectrum of SN 2009ipobtained on25 September
were stronger than in the September 24, 2011 spectrum 2009 with VLT-UT1 equipped with FORS2 has a very
(Figure 5). The minimum of the broad absorption com- high signal-to-noise ratio. This gives us the opportunity
ponents of the Balmer lines has a core at 8600 ± 400 to identify the most important lines in the spectrum of
km s−1, with a blue wing extending up to 14000 km SN 2009ip (Figure 6, bottom). The spectrum is domi-
s−1, while the Lorentzianemission survives at a FWHM natedby strongBalmerandPaschenlines ofH, showing
velocity of about 1380 km s−1. The presence of these weak and narrow(2850± 490km s−1) P-Cygniprofiles.
components was observed in September 15 and 16, 2012 Weak He I lines (being the 5876˚A feature blended with
spectrabySmith & Mauerhan(2012a), andthis wasthe
Na I D 5890-5896 ˚A) and a number of Fe II multiplet
criticalmeasurementthatledtheauthorstoproposethat
lines are also detected. We note that in the September
the LBV had exploded as a core-collapse SN, i.e. that
28, 2012 spectrum of SN 2009ip (during the 2012b out-
the 2012a event was due to stellar core-collapse and an
burst, Figure 6), the spectral properties are quite simi-
explosion with fairly low kinetic energy like SN 2005cs.
lar to those observed in the afore-mentioned VLT spec-
Ourspectra collectedbetweenAugust18 andSeptem-
trum,althoughthedetectionofFeIIlinesisnotobvious.
ber 5, 2012 show little evolution: the H features show
Mostoftheselinesarealsovisibleinthespectrumofthe
prominent P-Cygni profiles, with deep minima at 8000-
9000 km s−1 and edges possibly extending to 14000- impostor NGC3432-LBV1 shown as a comparison, with
15000 km s−1. The Hα narrow emission component quite similar velocity of the narrow components (≤ 650
still has a FWHM velocity of 800 ± 100 km s−1, while kms−1). NarrowOIandCaIIlinesarerelativelypromi-
nent in SN 2009ip, while they were not unambiguously
the highest resolution spectra allow us to measure the
detectedinNGC3432-LBV1(althoughthismightbedue
FWHM velocity from the clearly detected narrow Fe II
emissions (multiplet 42) to be about 240 ± 20 km s−1. to the lower signal-to-noise spectrum).
A sequence of near-infrared spectra of SN 2009ip is
As highlighted by Mauerhan et al. (2012b), the spec-
showninFigure7. Thecontinuumisalwaysquitebluein
tra from September 10 to 23 (2012a event) do closely
these spectra. The strongest lines are detected as broad
resemble those of type II SNe (the similarity with early
features with P-Cygni profiles, and narrower emissions
spectraoftheunder-luminoustypeIIPSN2005csshown
superimposed to the broad components. The broad P-
in their Figure 2 is remarkable). Both H and Fe II lines
Cygnicomponentsbecomemoreevidentwithtimeandin
nowshowbroadP-Cygniprofileswithaprominentbroad
the September 23,2012spectrum(atthe time ofthe on-
emissioncomponent. Howeverwenowpresentspectraof
setofthe2012beruption)theydominateoverthenarrow
the 2012a event covering a period from August 8, 2012
lines. We identify Br γ at 2165 nm, Pa α (that is barely
to September 23, 2012 (47 days), and we do not observe
visible in the middle of the telluric absorption around
the typical evolution of a type II SN over this period.
1875nm),Paβat1282nmandPaγat1094nm,blended
In particular, 15-20 days after explosion, type II-P SNe
with He I 1083 nm. The September 23, 2012 spectrum,
developthe strong,broadnear-infraredCa II triplet fea-
in particular, shows a broad Pa β with FWHM veloc-
ture (Pastorello et al. 2006), but we don’t observe this
8 Pastorello et al.
2009ip
Aug 8, 2012
2009ip
Sep 28, 2012
2009ip
Sep 24, 2011
2009ip
Nov 25, 2009
4000 5000 6000 7000 8000 9000
Fig.6.— Top: comparison of spectra of SN 2009ip at 3 representative epochs (24 September 2011, and during the 2012a and 2012b
events)withaspectruminoutburstofNGC3432-LBV1. Theverticaldashedgreenlinesmarkthepositionofthehighestvelocityedgesof
the Hβ components in the 2 objects. Bottom: lineidentification in the optical spectrum of SN 2009ip obtained on September 25, 2009
(VLT+XShooter), andcomparisonwithaspectrumofNGC3432-LBV1inoutburst.
ity of about 6200 km s−1 and a prominent blue-shifted 3. REAL SUPERNOVA OR SUPERNOVA
absorption of Pa γ + He I 1083 nm with an expansion IMPOSTOR?
velocity of about 10000 km s−1, as obtained from the
SN 2009ipis a remarkable object for a number of rea-
position of the broad absorption minimum. The nar-
sons: i) it experienced a series of energetic outbursts
row He I 1083 nm line, which was marginally detectable
since 2009, when the transient reached absolute peak
in previous spectra, is now clearly visible, and is well
magnitudesbetween-14and-15;ii)thespectralfeatures
separated from Pa γ. The narrow Paschen lines have
revealthepresenceofejectedmaterialatveryhighveloc-
Lorentzian profiles with a FWHM velocity of about 400 ities (several × 103 km s−1); iii)the progenitorstar was
kms−1, whilstthenarrowHe Iλ1083nmappearsto be
slightlybroader(∼800-1000km s−1) andwith aroughly o-1b0s.e0rv±ed0.t3o)abnedexwtarsempreolypolusemditnooubseianmqausiseisvceenLcBeV(M(>V 6=0
Gaussian profile.
M⊙, Foley et al. 2011; Smith et al. 2010); iv) finally, in
September 2012 the star displayed a further, exception-
SN 2009ip, is this the end? 9
800 1000 1200 1400 1600 1800 2000 2200
wavelength (nm)
Fig. 7.—Sequenceofnear-infraredspectraofSN2009ipobtainedfromAugusttoSeptember2012. TheXShooterspectrumofSeptember
24,2011isalsoshowningreen.
allyluminousoutburst(the2012bevent,withMV ≈-18, Thepresenceofveryfastmaterial(∼13000kms−1)in
Mauerhan et al. 2012b, and references therein), suggest- SN 2009ip already almost 1 year before the putative SN
ing that the LBV may have experienced a core-collapse explosion (i.e. in the 24th September 2011 spectrum),
SN explosion. The luminosity during that event, and its and also in NGC 3432-LBV1 (∼ 9000 km s−1) suggests
similarity to SNe IIn spectra are possibly the strongest that these LBV related eruptions could quite feasibly
indicators that a core-collapse SN has occurred, more be linked with the 2012a event. The highest velocity in
so than the broad lines of the spectra during the 2012a the Homunculus Nebula surrounding η Carinae reaches
pre-cursor event. 3500−6000kms−1(Smith2008). TypicalLBVeruptions
The complex, erratic 2009-2012 light curve of SN arediscussedintermsofextremestellarwinds drivenby
2009ip (Section 2.1) indicates that the LBV entered a the super-Eddington luminosity of the star. However,
very active variability phase resembling those of the un- these winds are expected to have velocities of the order
usual NGC 3432-LBV1 or η Carinae during the Giant of a few × 102 km s−1 (e.g. Smith 2008). The detec-
Eruption of the 19th century. In the case of NGC 3432- tionofthis high-velocitygasinsomeLBVoutbursts(in-
LBV1, multiple eruptions on short time scales (about cluding the afore-mentioned events) suggests that these
200-220days) have been proposed to be the result of vi- episodes probably originate in explosions deeper in the
olent pulses of a very massive star (possibly via the pul- star, perhaps in the core. These release an energy that
sational pair-instability mechanism) that is approaching maycompetewiththoseofweakSNe(e.g. faintSNeIIP,
theendofitslife,presumablywiththecore-collapse. Al- such as SN 1999br, Pastorello et al. 2004), producing a
ternatively, the pulses may be regulated by the passage blastwavethatallowsthe startoexpelmassiveportions
of a companion star to the periastron (Pastorello et al. of the envelope (Smith 2008). All of this is expected to
2010)30.
the modulated, quasi-periodic light curve of NGC 3432-LBV1
30 The presence of a companion was proposed to explain (Pastorelloetal.2010).
10 Pastorello et al.
producetransientsthatcloselymimictheenergyandthe explosion. What triggers these ejections is still unclear,
overallpropertiesofarealSNexplodinginadenseCSM but the very high progenitor mass (Smith et al. 2010;
(type IIn). Foley et al.2011)indicatesthattheeventsmaybesigna-
tures of pulsational pair-instability (Barkat et al. 1967;
3.1. No core-collapse SN during the 2012a event Woosley et al. 2007)31.
One of the most remarkable findings inferred from the The Woosley et al.(2007) modelof a pulsationalpair-
instability SN suggests that colliding shells of material
analysisoftheAugustandearlySeptemberspectraofSN
can dissipate most of the relative kinetic energy as ra-
2009ip (during the 2012a event) is that the bulk of the
ejected material has extremely high expansion velocities diation. One solar mass of material moving at 8000km
(8000-9000kms−1,withedgesextendingupto14000km s−1 hasakineticenergyofmorethan1050erg,enoughto
s−1, see Section 2.2, and Mauerhan et al. 2012b). This, power the measured bolometric light curve of the 2012b
event shown in Figure 3. As SN 2009ip has experienced
and the striking similarity between the early September
multiple mass ejections, perhaps even more than those
spectra of SN 2009ipand those of the weak type IIP SN
we have detected due to possible gaps in the observa-
2005cs(Pastorello et al.2006,2009)ledMauerhan et al.
tionalcoverage(Figure2),itisplausiblethereareshells,
(2012b) to conclude that SN 2009ip had likely exploded
or clumps of slower moving gas that will slow the fast
as a faint, 56Ni-poor core-collapse SN during the Au-
ejecta of 2009ip during the 2012a episode.
gust re-brightening episode. The fact that we can ob-
As discussed in Mauerhan et al. (2012b), there are no
serve features from the SN ejecta inside an extended
known line-drivenwind mechanism or continuum driven
and dense CSM is explained with a non homogeneous,
wind mechanism for driving material off the stellar sur-
possibly clumpy distribution of the material lost by the
face at the high velocities observed. The energy to pro-
LBVinpulsationsprecedingthe explosion. While thisis
vide & 1050erg per solar mass ejected must presumably
plausible, we would caution that the detection of high-
come from a core-related event.
velocity ejecta cannot be regardedas a conclusive proof,
There is also some consistency in the velocity of the
because very high velocity material was also observed
material ejected during the 2012a event and the radius
in NGC 3432-LBV1 (Pastorello et al. 2010, where the
of the emitting region in the 2012b episode. The 2012a
broad wing of the Hβ absorption extended to 9000 km
s−1), during an eruption of a known SN impostor. eventlastsapproximately50days,duringwhichthebulk
of material starts at 8000-9000km s−1 on 5 September
The core-collapse SN scenario proposed by
2012, slowing to 5000-6000km s−1 after about 10 days.
Mauerhan et al. (2012b) is unlikely, since there is
a number of observables that require a rather ad-hoc The fast ejecta likely travelled around 5 ×104R⊙, be-
fore impacting on a surrounding shell and causing the
combination of events: i) the high-velocity absorption
dramatic increase in luminosity in the rise to the 2012b
wings measured in the spectra obtained after the an-
lightcurvepeak. Ifthekineticenergyoftheshellisthen
nouncement of the 2012a outburst episode (Drake et al.
converted into radiative energy, one would expect that
2012) are actually similar to those we have seen in the
September 24, 2011 and August 8, 2012 spectra, which an emitting sphere of radius 5×104R⊙ at a black-body
temperatureofaround10000KwouldemitatL≃afew
raises the question whether and in case when the SN ×1043ergs−1. This crude luminosity estimate is of the
explosion occurred; ii) the peak absolute magnitude
same order of magnitude to that we see in Figure 3.
(MR ∼-15)andtheevolutionarytimescalesofthe2012a
The pulsational pair-instability SN model requires a
event are consistent with those of previous eruptive
episodes (in particular the 2009 event, see Figure 2); star of initial mass to be in the range 95-130 M⊙. The
standard mass-loss prescriptions for such massive stars
finally iii) it is not trivial to explain how an extremely
has to be relaxed so that in the final stages the star
massive LBV (M>60 M⊙, likely with MZAMS ≥ 90-100 should retain enough mass to enhance the core temper-
M⊙) can explode as a weak type II SN: we may need ature to cause the pair-instability. The progenitor has
to invoke sub-sequent eruptions to explain the events
before July-August 2012, and subsequently a fall-back beenestimatedto havemorethan 60M⊙, implying that
has retained most of its envelope. This is supported by
core-collapse SN with formation of a black hole.
theevidencethatbroadhydrogenfeaturesaredetectedin
In the Mauerhan et al. (2012b) interpretation, the
all the ejection episodes (Smith et al. 2010; Foley et al.
2012beventisfairlysimplyexplainedascore-collapseSN
2011, see also Section 2.2). An interesting measurement
ejecta-CSMinteraction. Howeveritisalsoplausiblethat
would be the metallicity at the distance of SN 2009ip
the 2012a event was an eruptive phase, and the 2012b
fromthe hostgalaxynucleus (about 4kpc) to determine
luminosity comes from the actual core-collapse, similar
ifitissignificantlymetalpoor. Atthecurrentstage,only
to what is assumed to occur in IIn SNe, or even from
a statistical approach is possible to estimate the local
the collision of material ejected in the previous eruption
oxygen abundance. Adopting the host galaxy distance
with pre-existing CSM.
and reddening of Smith et al. (2010), the host galaxy
has an absolute B-band magnitude of -17.9. Following
3.2. SN 2009ip, a pulsational pair-instability event
Thedetectionofhighvelocityejecta(12500kms−1)on 31 The pulsational pair-instability scenario discussed by
September 24, 2011 indicates that the star has managed Woosleyetal. (2007) is applicable for stars with main-sequence
to eject material at velocities that we would normally massesintherange95-130M⊙. Thisisapparentlyabovethemass
associate with a SN explosion. It is very unlikely that proposed for the precursor of SN 2009ip. We note, however, that
the absolute magnitude of the LBV progenitor of SN 2009ip (see
the core collapsed at this point (see Section 3.1), which
e.g. Figure3inFoleyetal.2011)isalsoconsistentwithevolution-
implies that the high velocity material has been ejected arytracksofstellarmassesthataremuchhigherthan60M⊙,that
in the 2012a event without invoking a core-collapse SN hastoberegardedasalowermasslimit.