Fourteen Months of Observations of the Possible Super-Chandrasekhar Mass Type Ia Supernova 2009dc
Jeffrey M. Silverman, Mohan Ganeshalingam, Weidong Li, Alexei V. Filippenko, Adam A. Miller, Dovi Poznanski
MMon. Not. R. Astron. Soc. , 1–30 (2010) Printed 2 November 2018 (MN L A TEX style file v2.2)
Fourteen Months of Observations of the PossibleSuper-Chandrasekhar Mass Type Ia Supernova 2009dc
Jeffrey M. Silverman, , (cid:63) Mohan Ganeshalingam, Weidong Li, Alexei V. Filippenko, Adam A. Miller, and Dovi Poznanski Department of Astronomy, University of California, Berkeley, CA 94720-3411, USA Marc J. Staley Fellow
Accepted . Received ; in original form
ABSTRACT
In this paper, we present and analyse optical photometry and spectra of the extremelyluminous and slowly evolving Type Ia supernova (SN Ia) 2009dc, and offer evidencethat it is a super-Chandrasekhar mass (SC) SN Ia and thus had a SC white dwarf (WD)progenitor. Optical spectra of SN 2007if, a similar object, are also shown. SN 2009dchad one of the most slowly evolving light curves ever observed for a SN Ia, with a risetime of ∼
23 days and ∆ m ( B ) = 0 .
72 mag. We calculate a lower limit to the peakbolometric luminosity of ∼ . × erg s − , though the actual value is likely almost40% larger. Optical spectra of SN 2009dc and SN 2007if obtained near maximumbrightness exhibit strong C II features (indicative of a significant amount of unburnedmaterial), and the post-maximum spectra are dominated by iron-group elements. Allof our spectra of SN 2009dc and SN 2007if also show low expansion velocities. However,we see no strong evidence in SN 2009dc for a velocity “plateau” near maximum lightlike the one seen in SN 2007if (Scalzo et al. 2010). The high luminosity and lowexpansion velocities of SN 2009dc lead us to derive a possible WD progenitor massof more than 2 M (cid:12) and a Ni mass of about 1.4–1.7 M (cid:12) . We propose that the hostgalaxy of SN 2009dc underwent a gravitational interaction with a neighboring galaxyin the relatively recent past. This may have led to a sudden burst of star formationwhich could have produced the SC WD progenitor of SN 2009dc and likely turnedthe neighboring galaxy into a “post-starburst galaxy.” No published model seems tomatch the extreme values observed in SN 2009dc, but simulations do show that suchmassive progenitors can exist (likely as a result of the merger of two WDs) and canpossibly explode as SC SNe Ia.
Key words: supernovae: general – supernovae: individual (SN 2009dc, SN 2007if,SN 2003fg, SN 2006gz)
Type Ia supernovae (SNe Ia) are differentiated from othertypes of SNe by the absence of hydrogen and the presenceof broad absorption from Si II λ (cid:63) E-mail: [email protected] companion star (e.g., Whelan & Iben 1973) or the mergerof two degenerate objects (e.g., Iben & Tutukov 1984; Web-bink 1984). However, the nature of the companion and thedetails of the explosion itself are both still quite uncertain.The cosmological utility of SNe Ia comes from the factthat they are standardizable candles (i.e., their luminositiesat peak can be calibrated). This naively seems reasonablesince SNe Ia should all have the same amount of fuel andthe same trigger point: they should all explode when a WDnearly reaches the Chandrasekhar mass of ∼ (cid:12) .In 2003, SNLS-03D3bb (also known as SN 2003fg) wasdiscovered (Howell et al. 2006) and was shown to be a SN Iathat was overluminous by about a factor of 2, and had aslowly declining light curve, quite low expansion velocities,and unburned material present in near-maximum light spec- c (cid:13) a r X i v : . [ a s t r o - ph . H E ] J u l Silverman, et al. tra. This last observation implies that a layer of carbon andoxygen from the progenitor existed on top of the burned sil-icon layer. In SNe Ia, carbon is usually extremely weak orcompletely absent (even at very early times) in optical andinfrared (IR) spectra, although it has been (sometimes ten-tatively) identified in a few other cases (e.g., Branch et al.2003; Marion et al. 2006; Thomas et al. 2007; Foley et al.2010). Howell et al. (2006) suggest that all of the odditiesseen in SN 2003fg could be explained if its progenitor WDhad a mass greater then the canonical upper limit for WDsof ∼ (cid:12) , a so-called “super-Chandrasekhar mass” (SC)WD.Hicken et al. (2007) then presented data on SN 2006gzwhich shared some of the strange properties of SN 2003fg,and they concluded that SN 2006gz must have come from aWD merger leading to a SC SN Ia. Recently, Scalzo et al.(2010) published observations of SN 2007if which is yet an-other example of this emerging class of possible SC SNe Ia.They not only observed low expansion velocities, but theyalso saw a plateau in the expansion velocity near maximum-brightness which they interpret as evidence for the SN ejectarunning into a shell of material. The accurately determinedtotal mass of the SN 2007if system is well above the Chan-drasekhar mass.Recently, it was pointed out that there might be an-other member of this SC SN Ia class, SN 2009dc (Haru-tyunyan et al. 2009; Marion et al. 2009; Yamanaka et al.2009). SN 2009dc was discovered 15 . (cid:48)(cid:48) . (cid:48)(cid:48) § ∼ α J2000 = 15 h m . s δ J2000 = +25 ◦ (cid:48) . (cid:48)(cid:48)
0; the SN and its host are shown inFigure 1. No object was visible in our data at the positionof the SN on 2009 Mar. 28 to a limiting magnitude of ∼ II λ µ m. They again note some similarities to (as well asdifferences from) SN 2006gz, as well as similarities with thepossible SC SN Ia 2003fg (Howell et al. 2006).Yamanaka et al. (2009) presented early-time opticaland near-IR observations of SN 2009dc and showed thatit did indeed share many of the properties seen in bothSN 2003fg and SN 2006gz, suggesting that it too is a possi-ble SC SN Ia. Furthermore, Tanaka et al. (2010) publishedspectropolarimetry of SN 2009dc which indicated that theexplosion was quite spherically symmetric.In this paper we present and analyse our own opticalphotometric and spectroscopic data for SN 2009dc (as wellas spectra of SN 2007if) with the goal of more definitivelyanswering the question of whether SN 2009dc and SN 2007ifwere truly SC SNe Ia. We show some of the most precisedata on the rise time of a possible SC SN Ia ever published, ′ Figure 1.
KAIT image of SN 2009dc and its host galaxy,UGC 10064. The field of view is 6.7 (cid:48) × (cid:48) . SN 2009dc is labeledalong with the comparison stars used for differential photometry.The scale of the image is marked in the top right; north is upand east is to left. The SN is sufficiently far from its host galaxythat template subtraction was not performed when conductingphotometry. as well as some of the latest-time photometric and spectralobservations of a member of this class. In § § § §
5, we summarize our conclusions and ruminateabout the future.
Observations of SN 2009dc began on 2009 Apr. 17, aboutone week before maximum B -band brightness, in BVRI fil-ters using the 0.76-m Katzman Automatic Imaging Tele-scope (KAIT; Filippenko et al. 2001) and the 1-m Nickeltelescope, both at Lick Observatory. We continued to fol-low SN 2009dc for over 5 months until 2009 Sep. 26, whenit reached the western limit of both telescopes in the earlyevening. We obtained late-time gVRI images of SN 2009dcusing the dual-arm Low-Resolution Imaging Spectrometer(LRIS; Oke et al. 1995) with the 10-m Keck I telescopeon 6 Feb. 2010 (281 days past maximum),
BRI images us-ing the DEIMOS spectrograph (Faber et al. 2003) mountedon the 10-m Keck II telescope on 12 June 2010 (403 dayspast maximum), and V -band images again using LRIS on13 June 2010.Our optical photometry is complemented with datataken from the UltraViolet Optical Telescope (UVOT) on c (cid:13)000
BRI images us-ing the DEIMOS spectrograph (Faber et al. 2003) mountedon the 10-m Keck II telescope on 12 June 2010 (403 dayspast maximum), and V -band images again using LRIS on13 June 2010.Our optical photometry is complemented with datataken from the UltraViolet Optical Telescope (UVOT) on c (cid:13)000 , 1–30 N 2009dc, Super-Chandrasekhar Mass? the Swift
Observatory in the U , B , V , UV W UV M
2, and
UV W
Swift archives.Our early-time optical images were reduced usinga pipeline developed for KAIT and Nickel data (Gane-shalingam et al. 2010a). The images were bias subtractedand flatfielded at the telescope. Given the distance ofSN 2009dc from the host galaxy, we did not find it necessaryto attempt galaxy subtraction. We performed differentialphotometry using point-spread function (PSF) fitting pho-tometry on SN 2009dc and several comparison stars in thefield with the DAOPHOT package in IRAF. The outputinstrumental magnitudes were calibrated to the standardJohnson BV and Cousins RI system using colour terms de-rived from many photometric nights. The comparison starswere calibrated against Landolt (1992) standard stars using13 photometric epochs to achieve a root-mean square (rms)of < .
01 mag. The comparison stars are identified in Fig-ure 1 with corresponding photometry in Table 1. To estimatethe uncertainty in our photometry pipeline, we add artifi-cial stars with the same magnitude and PSF as the SN toeach data image at positions of similar background bright-ness; the rms of doing this procedure 20 times is taken asthe uncertainty. The photometric uncertainty is added inquadrature to the calibration error to produce our final un-certainty, adopting an error floor of 0.03 mag for B and I and 0.02 mag for V and R to account for slight systematicdifferences between the KAIT and Nickel photometry. Ourfinal photometry of SN 2009dc is presented in Table 2.Late-time data obtained at the Keck I and Keck II tele-scopes were bias subtracted and flatfielded using standardimaging techniques. Differential photometry was performedusing PSF fitting photometry on the SN and comparisonstars that were not saturated, but also detected in our cal-ibration images obtained with the Nickel telescope. Cali-brations for the g band were obtained using the transfor-mations presented by Jester et al. (2005). In cases whereall of the field stars from our Nickel calibration were satu-rated, calibrations for fainter stars in the field were obtainedfrom the Sloan Digital Sky Survey (SDSS) and transformedinto BVRI using transformations for stars from Jester et al.(2005). Colour-term corrections were not applied. We in-clude a systematic error of 0.03 mag in all bands. Our finallate-time photometry of SN 2009dc is presented in Table 3,which also includes an epoch of photometry obtained withthe Nickel.We downloaded the Level-2 UVOT data from the
Swift archive. Images taken during the same pointing were regis-tered and stacked to produce deeper images. We performedaperture photometry using the recipes prescribed by Li et al.(2006) for the optical data and Poole et al. (2008) for the UVdata. We modified the U -band data to be in the Johnson-Cousins system using the colour corrections found in Li et al.(2006). In general, the UVOT B and V photometry is ingood agreement with the photometry from our ground-based IRAF: The Image Reduction and Analysis Facility is dis-tributed by the National Optical Astronomy Observatory, whichis operated by the Association of Universities for Research in As-tronomy (AURA) under cooperative agreement with the NationalScience Foundation (NSF).
Table 5.
Journal of Spectroscopic Observations of SN 2009dcUT Date Age a Range (˚A) Airmass b Exp (s)2009 Apr. 18.5 − c
52 3400–10200 1.56 250/200 d c
281 3500–10200 1.19 3 × (630/600) da Rest-frame days relative to the date of B -band brightness max-imum, 2009 Apr. 25.4 (see § b Airmass at midpoint of exposure. c These observations used LRIS (Oke et al. 1995) on the 10-mKeck I telescope. The others used the Kast spectrograph on theLick 3-m Shane telescope (Miller & Stone 1993). d The blue side was exposed longer than the red side due to therelatively long readout time of the red-side CCD in LRIS. telescopes to within ± .
05 mag. The UVOT B -band data aresystemically brighter, which could possibly be attributed togalaxy light falling within our aperture. Our results are givenin Table 4.We did not follow the photometric behaviour ofSN 2007if, another SC SN Ia candidate, but we do presentspectroscopic observations (see § B -band max-imum brightness and 0.07416 as the redshift ( z ) of SN 2007if,both taken from Scalzo et al. (2010). Beginning about a week before maximum brightness, opticalspectra of SN 2009dc were obtained mainly using the dual-arm Kast spectrograph (Miller & Stone 1993) on the Lick3-m Shane telescope. Two spectra were also obtained us-ing LRIS on Keck I. Our last spectral observation occurred281 days after B -band maximum.The Kast spectra all used a 2 (cid:48)(cid:48) wide slit, a 600/4310grism on the blue side, and a 300/7500 grating on the redside, yielding full-width at half-maximum (FWHM) resolu-tions of ∼ ∼
10 ˚A, respectively. The LRIS spectrumwas obtained with a 1 (cid:48)(cid:48) slit, a 600/4000 grism on the blueside, and a 400/8500 grating on the red side, resulting inFWHM resolutions of ∼ ∼ ∼ ∼ (cid:48)(cid:48) -wide slit was aligned along the parallacticangle to reduce differential light losses. Table 6 summarizesthe spectral data on SN 2007if presented here. We also notethat our last spectrum of SN 2007if (from 2007 Dec. 13) c (cid:13) , 1–30 Silverman, et al.
Table 1.
Comparison Stars for the SN 2009dc FieldStar α J2000 δ J2000 B (mag) V (mag) R (mag) I (mag) N calib SN 15:51:12.12 +25:42:28.01 15:51:00.40 +25:44:00.1 18.769 (010) 17.418 (006) 16.503 (005) 15.744 (011) 92 15:50:59.16 +25:43:04.7 17.501 (008) 16.957 (006) 16.608 (004) 16.236 (008) 123 15:51:00.72 +25:43:02.5 18.126 (009) 16.816 (004) 15.940 (004) 15.228 (008) 104 15:51:16.96 +25:42:40.6 18.671 (006) 17.805 (005) 17.249 (006) 16.734 (011) 115 15:51:11.83 +25:41:48.3 16.661 (007) 15.872 (004) 15.406 (003) 14.984 (009) 136 15:51:13.64 +25:41:20.5 15.383 (007) 14.471 (004) 13.935 (003) 13.460 (009) 137 15:51:01.38 +25:41:11.2 15.892 (007) 15.200 (005) 14.808 (003) 14.420 (010) 138 15:51:17.44 +25:39:59.6 17.156 (006) 16.534 (004) 16.174 (004) 15.790 (008) 131 σ uncertainties are in parentheses, in units of 0.001 mag. Table 6.
Journal of Spectroscopic Observations of SN 2007ifUT Date Age a Range (˚A) Airmass b Exp. (s)2007 Oct. 15.4 37 3300–9200 1.03 12002007 Nov. 12.2 63 3300–9150 1.49 12002007 Dec. 13.2 c
92 3300–9150 1.01 1200 a Rest-frame days relative to the date of B -band maximum bright-ness, 2007 Sep. 5.4 (Scalzo et al. 2010). b Airmass at midpoint of exposure. c This observation used a slightly higher resolution grism for theblue side. was taken under nonideal observing conditions (clouds werepresent and the atmospheric seeing was poor), and thus itsspectrophotometric accuracy is not as good as that of theother observations.All spectra were reduced using standard techniques(e.g., Foley et al. 2003). Routine CCD processing and spec-trum extraction were completed with IRAF, and the datawere extracted with the optimal algorithm of Horne (1986).We obtained the wavelength scale from low-order polyno-mial fits to calibration-lamp spectra. Small wavelength shiftswere then applied to the data after cross-correlating a tem-plate sky to the night-sky lines that were extracted withthe SN. Using our own IDL routines, we fit spectrophoto-metric standard-star spectra to the data in order to fluxcalibrate our spectra and to remove telluric lines (Wade &Horne 1988; Matheson et al. 2000). Information regardingboth our photometric and spectroscopic data (such as ob-serving conditions, instrument, reducer, etc.) was obtainedfrom our SN database (SNDB). The SNDB uses the popu-lar open-source software stack known as LAMP: the Linuxoperating system, the Apache webserver, the MySQL rela-tional database management system, and the PHP server-side scripting language (see Silverman et al. 2010, for furtherdetails).
We present our final optical
BVRI light curves of SN 2009dcin Figure 2. We include for comparison a “normal” Type IaSN 2005cf (Wang et al. 2009b), another SC SN Ia candidateSN 2006gz (Hicken et al. 2007), the “standard overlumi- nous” Type Ia SN 1991T (Lira et al. 1998), and the peculiarSN 2002cx-like SN 2005hk (Phillips et al. 2007).The light curves of SN 2009dc have many featureswhich are noticeably distinct from most other SNe Ia. Thelight curves are much broader than those of spectroscopi-cally normal SNe Ia such as SN 2005cf. Even in comparisonto the overluminous SN 1991T, SN 2009dc evolves signifi-cantly more slowly.The light curve of SN 2006gz presentedby Hicken et al. (2007) provides an good match, althoughSN 2009dc appears to have a slower rise time and a slowerdecline.The absence of a prominent second maximum in the R and I bands is particularly striking. The secondarymaximum in I has been attributed to the propagation ofan ionization front through iron-group elements (IGEs) asthey transition from being doubly ionized to singly ionized(Kasen 2006). Observationally, the strength of the secondarymaximum has been found to correlate with peak luminos-ity, powered by Ni decay, with luminous SNe exhibiting moreprominent secondary peaks (Hamuy et al. 1996; Nobili et al.2005). However, this feature is weak in SN 2009dc, eventhough it is more luminous than a typical SN Ia. Kasen(2006) found that significant mixing of Ni in the compo-sition of a SN Ia can lead to a blending of the two I -bandpeaks, possibly explaining why a secondary peak is weak inSN 2009dc.After correcting for Milky Way (MW) extinction (fromthe dust maps of Schlegel et al. 1998), we fit a polynomialto the B -band light curve to find that SN 2009dc peakedon JD 2 , , . ± . B max =15 . ± .
04 mag. These values are consistent with thosederived by Yamanaka et al. (2009).We also measure ∆ m ( B ), the decline in flux frommaximum light to 15 days past maximum in the B band.Correcting for time dilation, we find ∆ m ( B ) = 0 . ± .
03 mag, making SN 2009dc one of the most slowly evolv-ing SNe ever discovered and comparable to other SC SN Iacandidates in the literature. Interestingly, Yamanaka et al.(2009) measure ∆ m ( B ) = 0 . ± .
03 mag. Comparingthe two photometric datasets, we find discrepancies betweenmeasurements of local standard stars of ∼ m ( B ).The host galaxy of SN 2009dc is part of the Lick Obser- c (cid:13) , 1–30 N 2009dc, Super-Chandrasekhar Mass? Table 2.
Early-time
BVRI photometry of SN 2009dcJD B (mag) V (mag) R (mag) I (mag) Telescope2454925.95 · · · · · · · · · KAIT2454938.98 15.670 (030) 15.638 (020) 15.596 (020) 15.718 (030) KAIT2454939.95 15.549 (030) 15.541 (020) 15.500 (020) 15.577 (030) Nickel2454940.92 15.474 (030) 15.472 (030) 15.430 (030) 15.524 (030) Nickel2454940.93 15.511 (030) 15.495 (020) 15.475 (020) 15.608 (030) KAIT2454942.95 15.455 (030) 15.409 (020) 15.376 (020) 15.529 (030) KAIT2454946.97 15.383 (030) 15.327 (020) 15.272 (020) 15.450 (030) KAIT2454948.87 15.369 (030) 15.350 (020) 15.288 (020) 15.454 (030) KAIT2454951.92 15.449 (030) 15.386 (020) 15.311 (020) 15.449 (030) KAIT2454957.97 15.770 (030) 15.535 (020) 15.423 (020) 15.537 (030) KAIT2454959.85 15.930 (030) 15.610 (020) 15.464 (020) 15.554 (030) KAIT2454960.92 16.004 (030) 15.591 (020) 15.476 (020) 15.593 (030) KAIT2454962.93 16.137 (036) 15.656 (024) 15.543 (026) 15.633 (084) KAIT2454964.90 16.322 (030) 15.762 (020) 15.584 (020) 15.568 (030) KAIT2454966.93 16.528 (030) 15.846 (020) 15.642 (020) 15.562 (030) KAIT2454966.95 16.478 (030) 15.839 (020) 15.611 (020) 15.481 (030) Nickel2454968.89 16.677 (030) 15.926 (020) 15.675 (020) 15.565 (030) KAIT2454970.88 · · · · · ·
KAIT2454971.92 16.880 (030) 16.054 (020) 15.701 (024) 15.500 (030) Nickel2454972.89 16.972 (030) 16.106 (020) 15.748 (020) 15.579 (030) KAIT2454974.88 17.154 (030) 16.202 (020) 15.799 (020) 15.557 (030) KAIT2454975.87 17.188 (030) 16.239 (020) 15.797 (020) 15.526 (030) Nickel2454977.90 17.384 (030) 16.334 (020) 15.874 (020) 15.659 (030) KAIT2454981.89 17.650 (047) 16.488 (020) 16.009 (020) 15.707 (030) KAIT2454989.83 18.045 (078) 16.893 (027) 16.364 (020) 15.979 (034) KAIT2454989.87 17.891 (062) 16.855 (030) 16.333 (030) 15.951 (030) Nickel2454993.85 18.128 (080) 16.943 (041) 16.480 (021) 16.167 (030) KAIT2454993.90 18.032 (042) 16.960 (022) 16.483 (020) 16.093 (030) Nickel2454999.81 18.188 (057) 17.083 (026) 16.696 (020) · · ·
KAIT2454999.83 18.153 (037) 17.102 (020) 16.675 (027) 16.315 (030) Nickel2455004.78 18.208 (033) 17.180 (033) 16.835 (028) 16.478 (034) Nickel2455007.84 18.265 (030) 17.289 (022) 16.913 (027) 16.540 (031) Nickel2455009.80 18.254 (089) 17.277 (029) 16.944 (022) 16.677 (039) KAIT2455014.83 18.449 (084) 17.344 (057) 17.151 (021) 16.761 (038) KAIT2455015.83 18.370 (046) 17.427 (022) 17.132 (031) 16.794 (042) Nickel2455019.85 18.377 (106) 17.432 (043) 17.247 (040) 16.937 (036) Nickel2455022.74 18.271 (097) 17.581 (047) 17.295 (025) 16.890 (050) KAIT2455025.83 18.585 (049) 17.607 (069) 17.431 (059) 17.176 (053) Nickel2455027.69 18.502 (054) 17.625 (041) 17.439 (035) 17.243 (036) KAIT2455032.69 18.512 (089) 17.770 (043) 17.570 (020) 17.343 (036) KAIT2455032.85 18.607 (032) 17.726 (020) 17.588 (025) 17.317 (043) Nickel2455034.73 18.678 (030) 17.779 (036) 17.643 (028) 17.425 (046) Nickel2455037.69 18.686 (053) 17.805 (045) 17.725 (023) 17.455 (035) KAIT2455040.77 18.756 (030) 17.902 (023) 17.800 (021) 17.504 (030) Nickel2455042.68 18.646 (062) 18.060 (077) 17.908 (066) 17.690 (111) KAIT2455042.77 18.844 (060) 17.912 (037) 17.886 (035) 17.586 (037) Nickel2455044.73 18.788 (041) 17.971 (043) 17.946 (028) 17.666 (053) Nickel2455047.68 19.035 (151) 17.965 (055) 18.007 (116) 17.646 (057) KAIT2455047.73 18.900 (142) 18.075 (076) 18.026 (049) 17.825 (119) Nickel2455052.68 18.992 (115) 18.208 (050) 18.178 (040) 17.860 (140) KAIT2455054.77 · · · · · · · · · · · ·
KAIT2455082.65 19.163 (089) 18.789 (088) 19.011 (188) 18.618 (143) KAIT2455087.64 · · · · · · · · ·
KAIT2455090.68 19.444 (052) 18.765 (037) 19.219 (091) 18.800 (081) Nickel2455093.66 19.528 (078) 18.820 (043) 19.176 (086) 18.867 (144) Nickel2455100.66 19.665 (061) 18.954 (095) 19.441 (151) · · ·
Nickel1 σ uncertainties are in parentheses, in units of 0.001 mag.c (cid:13) , 1–30 Silverman, et al.
Table 3.
Late-time photometry of SN 2009dcJD Phase a Telescope Filter Exposure (s) Mag σ B
600 21.868 0.268250 Nickel V
360 21.016 0.3202455233.10 281 Keck/LRIS g
180 21.894 0.050281 Keck/LRIS V
180 21.988 0.042281 Keck/LRIS R
60 22.600 0.084281 Keck/LRIS I
120 21.483 0.0602455359.05 403 Keck/DEIMOS B
360 25.010 0.120403 Keck/DEIMOS R
450 24.987 0.143403 Keck/DEIMOS I
450 23.746 0.1852455360.10 404 Keck/LRIS V 540 24.834 0.152 a Rest-frame days relative to the date of B -band maximum brightness. Table 4.
UVOT Photometry of SN 2009dcJD Filter Mag σ JD Filter Mag σ U U U U U U B B B B B V V V V V vatory Supernova Search (LOSS; Li et al. 2000; Filippenkoet al. 2001), allowing us to put strict constraints on the risetime of SN 2009dc. Our first detection of SN 2009dc is froman unfiltered image on 2009 Apr. 4, when the SN is seen at R = 18 . ± .
18 mag, indicating a rise time >
21 days. Thisfirst detection is included in Figure 2 as part of our R -banddata (Li et al. 2003, show that KAIT unfiltered data approx-imate the R band). In an image taken on 2009 Mar. 28, thereis no detection of SN 2009dc down to R ≈ . ± . ± .
17 days using 105 SDSS-II SNe Ia,with slowly declining SNe tending to have shorter rise times.The rise time of SN 2009dc is significantly longer than theaverage SN Ia in their sample, despite being a slowly declin-ing SN, and therefore does not follow the trend found fornormal SNe Ia in their sample. Riess et al. (1999), on theother hand, found an average rise time of 19 . ± . m ( B ) = 1 . m (15) B < .
95 mag, making anextrapolation to ∆ m ( B ) = 0 .
72 mag to compare with therise time of SN 2009dc unreliable.The cosmological application of SNe Ia as precise dis-tance indicators relies on being able to standardize theirluminosity. Phillips (1993) showed that ∆ m ( B ) is well cor- related with luminosity at peak brightness for most SNe Ia,the so-called “Phillips relation”. In Figure 3, we plot abso-lute V magnitude as a function of ∆ m ( B ) for 71 SNe Iawith z Virgo infall > .
01 from the LOSS photometry database(Ganeshalingam et al. 2010b) and SN 2009dc to determinewhether SN 2009dc follows the Phillips relation. Host-galaxyextinction for the sample of 71 SNe Ia is derived usingMLCS2k2.v006 with the glosz prior, which assumes that thelate-time ( B − V ) t =+35 d colour is indicative of the host-galaxy extinction (Jha et al. 2007). For SN 2009dc, thelate-time colour evolution differs significantly from that ofnormal SNe Ia; thus, we instead adopt E ( B − V ) host =0 . R V = 3 . § H = 70 km s − Mpc − . The best-fit line to SNe with0 . < ∆ m ( B ) < . σ of the relationshaded in grey. SN 2009dc is clearly an overluminous outlierin comparison to SNe Ia having similar values of ∆ m ( B ).Events similar to SN 2009dc cannot be standardized usingcurrent distance-fitting techniques which rely on parameter-izations of light-curve shape (Howell et al. 2006).We measure the late-time decay of the light curves usinga linear least-squares fit to data in the range ∼ . ± .
04 mag per c (cid:13) , 1–30 N 2009dc, Super-Chandrasekhar Mass? B max M ag n i t ud e + C o n s t a n t BVRI U uvw1uvm2uvw2 Figure 2.
Optical and ultraviolet (UV) light curves of SN 2009dcfrom KAIT and the 1-m Lick Nickel telescope. For comparisonwe also plot the optical light curves of the overluminous Type IaSN 1991T, the normal SN 2005cf, the peculiar SN 2002cx-likeSN 2005hk, and another SC SN Ia candidate SN 2006gz. Thelight curves of each SN are shifted so that t = 0 days correspondsto the time of B max and reach the same peak magnitude for agiven filter. The evolution of SN 2009dc is atypical compared toother SNe Ia. SN 2009dc evolves much more slowly and exhibitsonly a muted secondary maximum in the I band. We includein our R -band light curve an early detection of SN 2009dc ( redopen circle ) in an unfiltered image from the Lick Observatory SNSearch (LOSS), about 3 weeks before maximum brightness. Thedata sources for the other SNe are cited in the text.
100 days in B , 1 . ± .
02 mag per 100 days in V , 2 . ± .
03 mag per 100 days in R , and 2 . ± .
04 mag per 100 daysin I . Leibundgut (2000) find typical decay rates for SNe Ia of1.4 mag per 100 days in B , 2.8 mag per 100 days in V , and4.2 mag per 100 days in I . SN 2009dc shows significantlyslower decline rates in V and I .Late-time photometry of SN 2009dc (Table 3) takenwith the Nickel telescope shows only marginal detectionsin B and V and upper limits in R and I , while our Keckimages, 281 days past maximum, give clear detections in gVRI . Figure 4 shows the late-time behaviour of SN 2009dcin comparison to late-time photometry of SN 2003du, a typ-ical SN Ia, taken from Stanishev et al. (2007), and the lineardecay rates found using data ∼ SN 2009dc !" ! ( m %) * B +(,-./0 ! &% ! &! ! %1 ! %$ M V ( , - ./ !" ! ( m %) * B +(,-./0 ! &% ! &! ! %1 ! %$ M V ( , - ./ Figure 3. M V as a function of ∆ m ( B ) for 71 SNe Iawith z Virgo infall > .
01 from the LOSS photometry archive( filled circles , Ganeshalingam et al. 2010b), compared withSN 2009dc ( filled star ). Host-galaxy extinction is derived usingMLCS2k2.v006 (Jha et al. 2007), except in the case of SN 2009dcwhere we adopt E ( B − V ) host = 0 . R V = 3 . § H = 70 km s − Mpc − ,Ω m = 0 .
27, and Ω Λ = 0 .
73 (Spergel et al. 2007). The best-fit lineto the data, excluding SN 2009dc, in the range 0 . < ∆ m ( B ) < . σ shaded in grey. SN 2009dc is clearly an outlier that does notfollow the luminosity-width relation. decay rates, at 281 days past maximum SN 2009dc is fainterby ∼ B and V , brighter by ∼ R ,and ∼ I . In comparison to SN 2003du,SN 2009dc is within ∼ BV R and ∼ I . We caution that ourinterpolated values for SN 2003du suffer from a rather largegap in the light curve between 225 days and 366 days af-ter maximum. While the late-time behaviour of SN 2009dcdoes not match the linear decay of SN 2003du exactly inall bands, our detections indicate that it has not undergonean unexpected drop in luminosity, and it is consistent withthe late-time light curve being powered by Co decay (see § ∼
400 days past maximum show onlymarginal detections in all bands, indicating that the SN isstill active despite the points lying below interpolated valuesfrom SN 2003du. The detections in B and V are clearlybelow what we would expect from SN 2003du, while R and I are not too far below expectations. We address possiblereasons for the drop in flux in § The colour evolution in B − V , V − R , and R − I forSN 2009dc in comparison to SNe 2006gz, 2005cf, 1991T, c (cid:13) , 1–30 Silverman, et al. M ag n i t ud e B V B max M ag n i t ud e R B max I Figure 4.
The late-time decay of SN 2009dc from 50 to about400 days past maximum light compared to that of SN 2003du( solid line ), a typical SN Ia. The light curves are shifted suchthat t = 0 days corresponds to the time of B max , and a constanthas been added to each SN 2003du light curve to match the peakmagnitude of SN 2009dc. Filled circles are KAIT and Nickel data,whereas open squares are LRIS and DEIMOS data. Upper limitsare marked with arrows. The LRIS g -band point is plotted in the B -band light curve panel. We fit the decay in BVRI , plotted asa dashed line, using a linear least-squares fit to our well-sampleddata between 50–150 days past maximum, before the SN becameprojected too close to the Sun. Late-time photometry obtainedwith LRIS at 281 days past maximum indicates that the flux in B and V is below what is expected from linear extrapolations,but mostly agrees with expectations from SN 2003du. However,the flux in R and I is larger than what is expected compared tothe extrapolations. Our R -band detection falls on the comparisonlight curve, while the I -band detection is brighter than what isexpected. SN 2009dc is detected in all bands at ∼
400 days pastmaximum, indicating that the SN is still active despite the pointslying below interpolated values from SN 2003du. The data forSN 2003du were taken from Stanishev et al. (2007). and 2005hk is displayed in Figure 5. All SNe have beencorrected for MW extinction using reddening derived fromthe dust maps of Schlegel et al. (1998). We have also cor-rected for host-galaxy extinction, assuming R V = 3 .
1, usingthe following reported values of E ( B − V ) host : 0.13 mag forSN 1991T (Lira et al. 1998), 0.10 mag for SN 2005cf (Wanget al. 2009b), 0.09 mag for SN 2005hk (Phillips et al. 2007),0.18 mag for SN 2006gz (Hicken et al. 2007), and 0.10 magfor SN 2009dc.The colour curves indicate that SN 2009dc was a partic-ularly blue SN Ia even compared to the prototypical over-luminous SN 1991T. In B − V , SN 2009dc remains bluerthan SNe 1991T, 2005cf, and 2005hk until ∼
50 days pastmaximum light. The colour curves for SN 2009dc are mostsimilar to those of SN 2006gz, another SC SN Ia candidate.At t >
50 days, the B − V colour of SN 2009dc becomesredder than that of SN 1991T and SN 2005cf. B − V [ m ag ] −0.6−0.4−0.20.00.20.40.6 V − R [ m ag ] B max −0.4−0.20.00.20.4 R − I [ m ag ] Figure 5. B − V (top), V − R (middle), and R − I (bottom)colour curves of SN 2009dc. Plotted for comparison are the colourcurves of SNe 1991T, 2005cf, 2005hk, and 2006gz. All curves havebeen corrected for MW reddening and host-galaxy extinction.We have not included errors in derived host-galaxy extinction forSN 2009dc which will systematically shift curves in one direction.At early times, SN 2009dc is bluer than SNe 1991T, 2005cf, and2005hk. The evolution most clearly resembles that of SN 2006gz.We plot the Lira law as a solid line in the range 35 < t <
85 days.SN 2009dc shows a slower red to blue colour evolution comparedto the Lira-Phillips relation, especially after t = 70 days. Thedata sources are cited in the text. Lira (1996) showed that SNe with low host-galaxy ex-tinction had similar B − V colour evolution between t = +30to +90 days independent of light-curve shape (the “Lira-Phillips relation”). Hicken et al. (2007) used this relationshipto derive the host-galaxy extinction for SN 2006gz. However,a comparison between the slope of the Lira-Phillips rela-tion and the B − V colour evolution of SN 2009dc showsdisagreement. Adopting the relation derived using 6 low-reddening SNe Ia by Phillips et al. (1999) and fitting fora constant E ( B − V ) host between t = +35 and +90 days,we find E ( B − V ) host = 0 . ± .
04 mag, with χ = 41for 15 degrees of freedom. If we restrict our fit to be-tween t = +35 and +70 days, we find a better fit with E ( B − V ) host = 0 . ± .
04 mag; χ = 5 for 7 degrees offreedom. We include the fit to the Lira-Phillips relation inFigure 5. c (cid:13)000
04 mag; χ = 5 for 7 degrees offreedom. We include the fit to the Lira-Phillips relation inFigure 5. c (cid:13)000 , 1–30 N 2009dc, Super-Chandrasekhar Mass? On 2009 Apr. 16.22, ∼ B -band maximum,Harutyunyan et al. (2009) obtained a spectrum of SN 2009dcwhich showed “prominent C II lines and a blue continuum”,and they noted that the spectrum resembled pre-maximumspectra of SN 2006gz (Hicken et al. 2007), but with a muchlower expansion velocity. We obtained a spectrum ∼ II and S II lines, and appearsto also contain O I , all of which are usually found in near-maximum spectra of SNe Ia. However, there are also two ap-parent C II lines (labeled in Fig. 6 and discussed in § II features at 14 days before maximum, by 8 daysbefore maximum they have almost completely disappeared(whereas SN 2009dc has strong C II at 7 days before max-imum). In addition, it is apparent from Figure 6 that theexpansion velocity of SN 2006gz is much larger than thatof SN 2009dc (based on the Si II λ § II features are not nearly as prominent in SN 2003fg as they arein SN 2009dc. It is possible that this is due to the fact thatthe spectrum of SN 2003fg was taken 9 days later (relativeto maximum light) than our spectrum of SN 2009dc. II The tell-tale spectroscopic signature of a SN Ia is broadabsorption from Si II λ is interesting aboutthe Si II feature in SN 2009dc is its extremely low velocity.The minimum of the Si II λ − about 9 days be-fore maximum brightness (Harutyunyan et al. 2009), andwe measure the feature to be blueshifted by ∼ − in our spectrum taken 7 days before maximum. Yamanakaet al. (2009) report it to be blueshifted by 8000 km s − in their spectrum 3 days before maximum, decreasing to6000 km s − by 25 days after maximum.These velocities are much lower than the average photo-spheric velocity near maximum for SNe Ia of ∼ − (Wang et al. 2009a). In fact, according to Wang et al. (2009a), the Si II velocities seen in SN 2009dc are approxi-mately 8 σ below the average. Interestingly, these velocities are similar to those oftwo other possible SC SNe Ia, SN 2003fg and SN 2007if(Howell et al. 2006; Scalzo et al. 2010, respectively). How-ever, two other objects that were claimed to be possi-ble SC SNe Ia show much more normal Si II velocities.Hicken et al. (2007) report expansion velocities near max-imum light of about 11,000–13,000 km s − for SN 2006gz.SN 2004gu, which was compared to SN 2006gz by Contr-eras et al. (2010), also shows a normal photospheric velocityof ∼ − at 5 days before maximum (this work)slowing to ∼ − at 3 days past maximum. Figure 4 of Yamanaka et al. (2009) and Figure 4 ofScalzo et al. (2010) both present the velocity evolution ofthe Si II λ ∼ − at t ≈ − II λ − (see § ∼
74 km s − day − , which is over twice as large as that foundfor SN 2007if (34 km s − day − ; Scalzo et al. 2010). Further-more, combining our measurements with all previously pub-lished values of the expansion velocity of SN 2009dc (Haru-tyunyan et al. 2009; Yamanaka et al. 2009; Tanaka et al.2010), we see no strong evidence for a velocity “plateau”near maximum light like the one seen in SN 2007if (Scalzoet al. 2010).We measure the equivalent width (EW) of the Si II λ ∼
40 ˚A in our spectrum of SN 2009dcobtained 7 days before maximum. This is similar to the EWof the same line in SN 2006gz at similar epochs (Hickenet al. 2007), and both are well below the average Si II EWfor normal SNe Ia (though they are comparable to the over-luminous SN 1991T-like SNe Ia, e.g., Hachinger et al. 2006;Wang et al. 2009a). II As ubiquitous as Si II is in the spectra of SNe Ia, C II isnearly as rare. In a few cases, mainly at very early times,weak C II has been detected in optical spectra of SNe Ia(e.g., Branch et al. 2003; Thomas et al. 2007; Tanaka et al.2008; Foley et al. 2010). Candidate SC SNe Ia, on the otherhand, exhibit strong C II in their near-maximum spectra(Howell et al. 2006; Hicken et al. 2007; Scalzo et al. 2010).In our spectrum of SN 2009dc obtained 7 days beforemaximum brightness, we detect absorption from C II λ λ II λ − in our pre-maximum spectrum. The expansionvelocity of this feature is ∼ − at 3 days before Note that they consider SNe Ia with photospheric velocitiesgreater than 3 σ above the average to be “high-velocity SNe Ia.” This second velocity has a large uncertainty since it was mea-sured by approximating the minimum of the Si II absorption fea-ture by eye from Figure 3 of Contreras et al. (2010).c (cid:13) , 1–30 Silverman, et al. l og ( F λ ) + c on s t a n t
91T (−8)04gu (−5)06gz (−14)06gz (−8) 09dc (−7)03fg (+2) 05hk (−5)05cf (−2)S II Si II C II C II O I
Figure 6.
Our pre-maximum spectrum of SN 2009dc (and various comparison SNe) with a few major lines identified and days relativeto maximum light indicated for each spectrum (in parentheses). From top to bottom, the SNe are the “standard overluminous” Type IaSN 1991T (Filippenko et al. 1992), a possible SC SN Ia SN 2004gu (Contreras et al. 2010, though this spectrum is from our own database),another possible SC SN Ia SN 2006gz (Hicken et al. 2007) at two different epochs, SN 2009dc (this work), yet another possible SC SN IaSN 2003fg (Howell et al. 2006), the SN 2002cx-like peculiar SN 2005hk (Chornock et al. 2006), and finally the “standard normal” Type IaSN 2005cf (Wang et al. 2009b). All spectra have had their host-galaxy recession velocities removed and have been dereddened accordingto the values presented in their respective references. Note that Si II, S II, and O I are often found in near-maximum spectra of SNe Ia,but C II is not. maximum, slowing to about 7000 km s − at 3 days aftermaximum, and ∼ − by 6 days after maximum(Yamanaka et al. 2009; Tanaka et al. 2010). By about 18days past maximum this feature seems to have completelydisappeared, though it is difficult to be sure given the lowS/N spectrum seen in Figure 3 of Yamanaka et al. (2009).However, it is clear from the same figure that by 25 daysafter maximum, the feature is most definitely gone. It is alsoundetected in our spectrum obtained 35 days past maximum(see § II λ ∼ − , which is in good agreement with thevelocity of the C II λ ∼ − (see Fig. 1 of Tanaka et al.2010, though the feature is not marked), but it has alsolikely disappeared by 18 days past maximum (Yamanakaet al. 2009, Fig. 3) and is certainly gone by 35 days pastmaximum (see § II , the velocities seen in SN 2009dc once againmatch those of SN 2003fg at ∼ II λ ∼ − ) is clearly detected. On the other hand,SN 2007if appears to have both the C II λ λ II λ II λ c (cid:13)000
Our pre-maximum spectrum of SN 2009dc (and various comparison SNe) with a few major lines identified and days relativeto maximum light indicated for each spectrum (in parentheses). From top to bottom, the SNe are the “standard overluminous” Type IaSN 1991T (Filippenko et al. 1992), a possible SC SN Ia SN 2004gu (Contreras et al. 2010, though this spectrum is from our own database),another possible SC SN Ia SN 2006gz (Hicken et al. 2007) at two different epochs, SN 2009dc (this work), yet another possible SC SN IaSN 2003fg (Howell et al. 2006), the SN 2002cx-like peculiar SN 2005hk (Chornock et al. 2006), and finally the “standard normal” Type IaSN 2005cf (Wang et al. 2009b). All spectra have had their host-galaxy recession velocities removed and have been dereddened accordingto the values presented in their respective references. Note that Si II, S II, and O I are often found in near-maximum spectra of SNe Ia,but C II is not. maximum, slowing to about 7000 km s − at 3 days aftermaximum, and ∼ − by 6 days after maximum(Yamanaka et al. 2009; Tanaka et al. 2010). By about 18days past maximum this feature seems to have completelydisappeared, though it is difficult to be sure given the lowS/N spectrum seen in Figure 3 of Yamanaka et al. (2009).However, it is clear from the same figure that by 25 daysafter maximum, the feature is most definitely gone. It is alsoundetected in our spectrum obtained 35 days past maximum(see § II λ ∼ − , which is in good agreement with thevelocity of the C II λ ∼ − (see Fig. 1 of Tanaka et al.2010, though the feature is not marked), but it has alsolikely disappeared by 18 days past maximum (Yamanakaet al. 2009, Fig. 3) and is certainly gone by 35 days pastmaximum (see § II , the velocities seen in SN 2009dc once againmatch those of SN 2003fg at ∼ II λ ∼ − ) is clearly detected. On the other hand,SN 2007if appears to have both the C II λ λ II λ II λ c (cid:13)000 , 1–30 N 2009dc, Super-Chandrasekhar Mass? sion velocities near 15,500 km s − . However, both featuresare effectively undetectable 4 days later (Hicken et al. 2007).We also note that neither line is seen in spectra of SN 2004guat 5 days before maximum (see Fig. 6, although this spec-trum has low S/N) or at 3 days after maximum (Contreraset al. 2010). For SN 2009dc, Marion et al. (2009) reported no de-tectable absorption from C II λ do see evidence for absorptionfrom this transition, as well as absorption from C II λ § II λ λ ∼
19 ˚A and ∼
13 ˚A, respectively. Hicken et al. (2007) re-port that the C II λ II lines than typical SNe Ia. For example, one of the strongestC II features ever seen in a normal SN Ia was in SN 2006D,where C II λ II In our pre-maximum spectrum (as well as most of our post-maximum spectra; see § II H&K. The narrow componentsare at the redshift of SN 2009dc and its host galaxy. At allepochs these lines are just at our instrumental resolutionlimit; thus, calculating an accurate FWHM is nearly im-possible. In addition, the H&K lines occur in a part of thespectrum where numerous broad absorption features blendtogether, so defining a local continuum is difficult, exacer-bating the problem of measuring reliable line strengths. De-spite all this, we attempt to measure the strength of thisnarrow feature. We calculate that the Ca II H&K lines ap-pear to have FWHM ranging from about 100–300 km s − .However, the line strengths are consistent with no changeduring these epochs, given our rather large uncertainties.Some SNe Ia exhibit narrow, time-variable Na I D ab-sorption lines with unchanging, narrow Ca II H&K ab-sorption (e.g., Simon et al. 2009). However, time-variableCa II H&K absorption has not been seen. SN 2009dc doesshow weak, narrow absorption from Na I D, but it is fartoo weak to determine if variability in the line strengths ex-ists (see § II H&K absorption in SN 2009dc, but suchvariability should be searched for in other, future SC SNe Ia.The narrow Ca II H&K may arise from calcium-rich in-terstellar material (ISM) in the host galaxy of SN 2009dc,UGC 10064, and in fact this feature is strong in a spec-trum of the core of the galaxy (Abazajian et al. 2009).However, since SN 2009dc is relatively far from the host’snucleus, this seems somewhat unlikely. The alternative is The C II λ that the calcium-rich material which gives rise to the strongCa II H&K line is in close proximity to the SN site. Thismeans that SN 2009dc likely exploded in a region of calcium-rich ISM, or that the progenitor star of SN 2009dc itself hadcalcium-rich circumstellar material. However, which of thesetwo possibilities best reflects the true situation and how thismay affect the SN itself are beyond the scope of this paper.Another Ca II feature commonly seen in SNe Ia is thebroad Ca II near-IR triplet. However, we see no hint of thisline in our spectrum obtained 7 days before maximum. Onthe other hand, Tanaka et al. (2010) clearly detect this fea-ture near 8400 ˚A in their spectropolarimetric observationstaken ∼ ∼ . II λ To determine which other species are present in our pre-maximum spectrum of SN 2009dc, we use the spectrum-synthesis code SYNOW (Fisher et al. 1997). SYNOW isa parameterized resonance-scattering code which allows forthe adjustment of optical depths, temperatures, and veloci-ties in order to help identify spectral features seen in SNe.Before fitting, we deredden our pre-maximum spec-trum of SN 2009dc using E ( B − V ) MW = 0 .
070 mag, E ( B − V ) host = 0 . § R V = 3 .
1, and the red-dening curve of Cardelli et al. (1989). We also remove thehost-galaxy recession velocity ( cz = 6300 ±
300 km s − , asdetermined from the narrow Ca II H&K absorption in ourspectrum) before comparing with the output of SYNOW.Our derived SYNOW fit is compared to our actual pre-maximum spectrum in Figure 7. Also shown are the spectralfeatures from each of the individual species that were used inthe final fit. Our SYNOW fit faithfully reproduces the vastmajority of the features seen in our pre-maximum spectrumof SN 2009dc, but a few features are not matched exactly(specifically, the ones near 3700 ˚A, 4100 ˚A, and 4600 ˚A).In addition to the species mentioned above (Si II , C II ,and weak Ca II ), we detect O I , S II , Mg II , Si III , Si IV ,Fe II , Co II , and Co III . All of these ions (except C II , asmentioned previously) are often found in early-time spectraof SNe Ia, though O I may be weak in some overluminousSNe Ia (e.g., Filippenko 1997). The detection of C II in thespectra of SN 2009dc (see § − , with maximum velocities of 10,000–15,000 km s − depending on the ion. This value matchesour derived velocities from the Si II and C II absorption fea-tures (see § § c (cid:13) , 1–30 Silverman, et al. S ca l e d F λ + C on s t a n t O ISi IIC IICa IIS IIMg IISi IIISi IVFe IICo IICo IIIsumSN 2009dc (−7 d)
Figure 7.
The top eleven spectra show the constituent species in our SYNOW fit. The second from bottom spectrum is the sum ofthe top eleven spectra. The bottom spectrum is our pre-maximum spectrum of SN 2009dc (dereddened by E ( B − V ) MW = 0 .
070 magand E ( B − V ) host = 0 . R V = 3 . cz = 6300 ±
300 km s − ). Finally, weremove the underlying 20,000 K blackbody continuum from all spectra shown. each other, it is not surprising that these parameters are sim-ilar to those found in a SYNOW fit of SN 2003fg (Howellet al. 2006, Supplementary Information). Our fit requires aslightly higher photospheric velocity (they use 8000 km s − ),but this seems reasonable since our spectrum was obtained7 days before maximum and the spectrum of SN 2003fg fitby Howell et al. (2006) is from ∼ after maximum.All ions in our SYNOW fit had an excitation temper-ature around 10,000 K except C II , which was raised to20,000 K in order to get the relative line strengths to matchthe data. This is lower than what was calculated by How-ell et al. (2006) for SN 2003fg. They required an excitationtemperature for C II of 35,000 K to match their spectrum,while we obtain a temperature that is less than 60% of thatvalue.Finally, we require an underlying blackbody tempera-ture of 20,000 K in order to reproduce the overall continuumshape of SN 2009dc. This is hot for a SN Ia, even at early times, and is indicative of the production of a large amountof Ni (Nugent et al. 1995). Blackbody temperatures inthe range of 10,000–15,000 K seem to be more common fornormal SNe Ia (e.g., Patat et al. 1996), with the overlumi-nous Type Ia SN 1991T having temperatures at the highend of that range (Mazzali et al. 1995). Howell et al. (2006)used a blackbody temperature of 9000 K to fit SN 2003fg,which does not seem abnormally high for a SN Ia 2 dayspast maximum. Even though this value is a bit lower thanour derived blackbody temperature, it is not too surprisingsince our spectrum of SN 2009dc was obtained about 9 daysearlier (relative to maximum light) than their spectrum ofSN 2003fg.
The rest of our spectral data of SN 2009dc (i.e., all of ourpost-maximum spectra) are presented in Figure 8. Note that c (cid:13) , 1–30 N 2009dc, Super-Chandrasekhar Mass? during our observation on 2009 Aug. 14.2 (day 109), therewas smoke and ash in the sky throughout the night (from afire near the observatory), and thus the continuum shape isless accurate (due to the nonstandard extinction caused bythe smoke) than in the other observations. Our day 35 (2009 May 31.3) spectrum of SN 2009dc is shownin Figure 9, along with spectra of the peculiar SN 2002cx(Li et al. 2003), another possible SC SN Ia SN 2007if (Scalzoet al. 2010, though the displayed spectrum is from our owndatabase), and the “standard normal” Type Ia SN 2005cf(Wang et al. 2009b). While there are some spectroscopicsimilarities between SN 2009dc and SN 2005cf at this epoch,there are obvious differences as well. Most notably, thereare many more medium-width features in the spectrum ofSN 2009dc compared with SN 2005cf, and the features thatdo clearly match are at much lower velocities in SN 2009dc.This is the same as what was seen in the pre-maximumspectra ( § II H&K feature as wellas the Ca II near-IR triplet are much stronger in SN 2005cfthan they are in SN 2009dc, again consistent with the pre-maximum spectrum of SN 2009dc ( § II fea-tures are detected in both of these objects, there are still afew medium-width features in SN 2009dc that are not seenin SN 2007if. In addition, the velocities of some (but notall) of the features in SN 2007if are larger than those inSN 2009dc by ∼ − . However, SN 2007if appearsto match SN 2005cf well at this epoch, even though its ex-pansion velocity is still not as large as that of SN 2005cf,and SN 2007if is missing the Si II λ § II ,although Ca II (as mentioned above), Co II , and Na I arealso present. There is evidence for O I λ I λ II and Co II transitions can account for theseparts of the spectrum without the need for oxygen. Possibledetections of Ti II (Li et al. 2003) and Cr II (Branch et al.2004) have been claimed for SN 2002cx, but it is extremelydifficult to disentangle the individual contributions of eachof these IGEs in SN 2009dc.The one clear difference between the spectrum ofSN 2009dc and that of SN 2002cx (and SN 2005hk as well)is that SN 2009dc has a strong absorption feature centredat 6250 ˚A that is not seen in either of these other pecu-liar SNe Ia. This line is also not present in our spectrumof SN 2007if from a similar epoch. In Figure 10 we presenta SYNOW fit to our spectrum of SN 2009dc from 35 days past maximum based originally on a fit to the spectrum ofSN 2002cx from 25 days past maximum (Li et al. 2010).Both SYNOW fits reproduce the majority of the strongestfeatures in SN 2009dc, and the only difference between thetwo is the addition of Si II at an expansion velocity and ex-citation temperature similar to other ions in the fit. Basedon the comparison of our two fits, we attribute this myste-rious absorption in SN 2009dc to Si II λ ∼ − ). Note that the spectrum of SN 2005cffrom 28 days past maximum, shown in Figure 9, has strongSi II λ ∼ − , nearly twice the velocity seenin SN 2009dc. We discuss this feature further in § Our day 52 (2009 Jun. 17.5) and day 64 (2009 Jun. 29.3)spectra of SN 2009dc are shown in Figure 11, along withspectra of SN 2002cx (Li et al. 2003) and SN 2007if (Scalzoet al. 2010, but the displayed spectrum is from our owndatabase). Note that SN 2009dc does not evolve much spec-troscopically between days 52 and 64, though there are afew differences which will be discussed in § − .As mentioned above, the Si II λ § II H&K features aresimilar in strength, emission from the Ca II near-IR tripletis weaker in SN 2009dc and has a complex profile with mul-tiple peaks. We discuss this feature further in the followingsection. Figure 8 shows all of our post-maximum spectra ofSN 2009dc. Overall, the spectra do not change very muchfrom 35 to 109 days past maximum (see § c (cid:13) , 1–30 Silverman, et al. l og ( F λ ) + c on s t a n t +35+52+64+80+87+92+109+281SN 2009dcCa II H&K Ca II IR tripletSi II Figure 8.
Our post-maximum spectra of SN 2009dc, with days relative to maximum light indicated for each spectrum. Importantfeatures discussed in the text are labeled. All spectra have been dereddened by E ( B − V ) MW = 0 .
070 mag and E ( B − V ) host = 0 . R V = 3 .
1. We have also removed the host-galaxy recession velocity ( cz = 6300 ±
300 km s − ). of the broad emission near 5400 ˚A and in the ranges 6500–7200 ˚A, 7800–8300 ˚A, and (to a lesser extent) 8800–9200 ˚A.The first two of the ranges mainly fall into the R and I bands, respectively. The emission from these features, whichis strong through about 1 month past maximum and thenbegins to decrease with time, is possibly responsible for theplateau-like broad peak in the R and I bands near maximumbrightness and their slow late-time decline (as compared tomore normal SNe Ia, see Fig. 2).Li et al. (2003) also saw these features in SN 2002cx,and their temporal evolution and the effect they had on thenear-maximum R and I -band light curves is nearly identicalto what we find in SN 2009dc. However, Li et al. (2003) wereunable to determine which ions were responsible for some ofthe features in these ranges in SN 2002cx, and we are like-wise unsure of their origin. The detailed spectral modelingrequired to definitively determine their identity is beyondthe scope of this paper.As noted previously, we detect strong Si II λ II λ II λ II λ II line at late times inSN 2009dc is possibly due to its large ejecta mass. In § Ni massesever produced by a SN Ia, and thus one of the largest totalejecta masses as well. The massive ejecta may imply thata greater than average amount of silicon was synthesisedin the explosion of SN 2009dc, which could give rise to therather long-lived Si II λ c (cid:13) , 1–30 N 2009dc, Super-Chandrasekhar Mass? l og ( F λ ) + c on s t a n t SN 2002cx (+20, +25)SN 2009dc (+35)SN 2007if (+37)SN 2005cf (+28)Ca II H&K Ca II IR tripletSi II
Figure 9.
Our spectrum of SN 2009dc 35 days past maximum and a few comparison SNe, with days relative to maximum light indicatedfor each spectrum (in parentheses). From top to bottom, the SNe are the peculiar SN 2002cx (Li et al. 2003, where we have combinedtheir spectra from 20 and 25 days past maximum), SN 2009dc (this work), another possible SC SN Ia SN 2007if (Scalzo et al. 2010,though the spectrum shown is from our own database), and the “standard normal” Type Ia SN 2005cf (Wang et al. 2009b). Importantfeatures discussed in the text are labeled. All spectra have had their host-galaxy recession velocities removed and have been dereddenedaccording to the values presented in their respective references. clumpy distribution, and thus the longevity of the observedSi II λ II near-IR triplet is weaker in SN 2009dc and has a complexprofile with multiple peaks, even though the Ca II H&K fea-tures are similar in strength. A SYNOW fit indicates that weare actually resolving the individual components of the Ca II near-IR triplet along with shallow absorption possibly fromMg I and another, unidentified ion. The Ca II and Mg I fea-tures are blueshifted by ∼ − , which matches fairlywell with our previously calculated post-maximum expan-sion velocities (see § II near-IR triplet, Mg I ,and probably another ion whose identity we are unable todefinitively determine. On 2010 Feb. 6.6 (281 days after maximum), we obtainedour final spectrum of SN 2009dc, shown in Figure 12 alongwith late-time spectra of the normal Type Ia SN 1998bu (Jhaet al. 2006), SN 2006gz (Maeda et al. 2009), another normalType Ia SN 1990N (G´omez & L´opez 1998, via the onlineSUSPECT database ), and SN 2002cx (Jha et al. 2006).At over 9 months past maximum light, SN 2009dc ap-pears to have completely transitioned to the nebular phase.The SN ejecta are now optically thin and the spectrumshows various emission features. The broad peaks in therange 3800–5500 ˚A are usually attributed to blends of var-ious features from [Fe II ] and [Fe III ] (e.g., Mazzali et al.1998). These blends appear slightly weaker in SN 2009dc as http://suspect.nhn.ou.edu/ ∼ suspectc (cid:13) , 1–30 Silverman, et al. S ca l e d F λ SN 2009dc (+35)with Si II S ca l e d F λ without Si IIwith Si IISN 2009dc (+35) Figure 10.
The top panel shows our spectrum of SN 2009dc from 35 days past maximum ( solid line ) overplotted with our best SYNOWfit which includes Si II ( dotted line ). The bottom panel is the same as the top panel but zoomed in and centred on the absorption featurein SN 2009dc near 6250 ˚A; it also shows the same SYNOW fit but with no Si II ( dashed line ). The SYNOW fit with
Si II appears toreproduce the data better, and thus we attribute this absorption to Si II λ ∼ − ). We have dereddenedour data by E ( B − V ) MW = 0 .
070 mag and E ( B − V ) host = 0 . R V = 3 .
1. We have also removed the host-galaxy recessionvelocity ( cz = 6300 ±
300 km s − ) from the data. compared to the normal Type Ia SNe 1998bu and 1990Nat similar epochs. Interestingly, Maeda et al. (2009) foundthat SN 2006gz completely lacks these features, though theirspectrum has a low S/N.The Ca II near-IR triplet is detected in both of the nor-mal SNe Ia presented in Figure 12 as well as in SN 2009dcand possibly SN 2006gz (though the spectrum is noisy atthese wavelengths). There is essentially no emission in ei-ther SN 2009dc or SN 2006gz from ∼ ∼ III ] emission has been identified innormal SNe Ia (Kuchner et al. 1994; Ruiz-Lapuente et al.1995). The ratio of emission from the strongest [Fe
III ] fea-ture (4589–4805 ˚A) to emission from the strongest [Co
III ]feature (5801–5995 ˚A) has been measured in a handful ofSNe Ia at late times; it increases with both phase relative tomaximum and ejecta temperature (Kuchner et al. 1994).Using our spectra of SNe 1998bu and 1990N shown in Figure 12, we calculate ratios of about 5.6 and 7.7, respec-tively. These fit nicely with the other data displayed in Fig-ure 2 of Kuchner et al. (1994). However, it is nearly impos-sible to calculate this ratio for SN 2009dc and SN 2006gz,since they appear to lack almost any putative cobalt emis-sion at this epoch (and the cobalt and iron lines in spectraat earlier epochs are still significantly blended). We attemptto measure the tiny amount of [Co
III ] emission above thebackground level from 5801 ˚A to 5995 ˚A and calculate aFe/Co ratio of about 27. This value, which is likely an un-derestimate, is already well above that of any other SN Iaat this phase presented by Kuchner et al. (1994). It is alsoabove the dotted lines in Figure 2 of Kuchner et al. (1994),implying that the ejecta temperature of SN 2009dc may be (cid:38) ,
000 K. Even though this is broadly consistent with ourrelatively large value of 20,000 K for the blackbody temper-ature of SN 2009dc before maximum ( § c (cid:13) , 1–30 N 2009dc, Super-Chandrasekhar Mass? l og ( F λ ) + c on s t a n t SN 2002cx (+56)SN 2009dc (+52)SN 2009dc (+64)SN 2007if (+63)Ca II H&K Ca II IR tripletSi II
Figure 11.
Our spectra of SN 2009dc about 52 and 64 days past maximum, and a few comparison SNe with days relative to maximumlight indicated for each spectrum (in parentheses). The top SN is the peculiar SN 2002cx (Li et al. 2003), the next two spectra areSN 2009dc (this work), and the bottom spectrum is another possible SC SN Ia SN 2007if (Scalzo et al. 2010, though the displayedspectrum is from our own database). Important features discussed in the text are labeled. All spectra have had their host-galaxyrecession velocities removed and have been dereddened according to the values presented in their respective references. extremely hot for a normal SN Ia 9 months past maximumbrightness and the shape of the spectrum at this epoch doesnot seem to support such a high temperature.The most well-defined spectral feature in our day 281spectrum of SN 2009dc is the broad, double-peaked line cen-tred near 7200 ˚A, which was also the strongest feature de-tected in a spectrum of SN 2006gz from 338 days past max-imum (Maeda et al. 2009). In late-time spectra of SNe Iathis is attributed to blends of various [Fe II ] lines, whereas incore-collapse SNe it is associated with [Ca II ] λλ II ] and [Ca II ] lines (Maeda et al. 2009). The shape ofthe profile in SN 2009dc matches roughly that of SN 1998bu(though the central dip in SN 2009dc is deeper), but differssomewhat significantly from that of SN 1990N. Shapes simi-lar to the one seen in SN 2009dc have, however, been seen byMaeda et al. (2009), who present synthetic late-time spectra of SNe Ia. Their models with more massive progenitors seemto lead to the feature near 7200 ˚A becoming strong relativeto the complex of lines in the range 3800–5500 ˚A, as seen inSN 2009dc. As already mentioned numerous times, SN 2009dc sharessome interesting properties with both SN 2002cx andSN 2005hk, which are members of a class of peculiar SNe Ia(Filippenko 2003; Li et al. 2003; Jha et al. 2006). These so-called “SN 2002cx-like” SNe are characterized by low expan-sion velocities and low peak luminosities, slow late-time de-clines in R , and Fe III features dominating early-time spectra(Jha et al. 2006). Many of these criteria hold for SN 2009dcas well, but the very obvious difference is that the peak lu- c (cid:13) , 1–30 Silverman, et al. l og ( F λ ) + c on s t a n t Figure 12.
Our spectrum of SN 2009dc about 281 days past maximum, and a few comparison SNe with days relative to maximum lightindicated for each spectrum (in parentheses). The spectra from top to bottom are the normal Type Ia SN 1998bu (Jha et al. 2006),SN 2009dc, SN 2006gz (Maeda et al. 2009), SN 1990N (via SUSPECT), and the peculiar Type Ia SN 2002cx (Jha et al. 2006). Majorspectral features are labeled. All spectra have had their host-galaxy recession velocities removed and have been dereddened according tothe values presented in their respective references. minosity of SN 2009dc is much larger than average (see § § R and I -band light curves of SN 2009dc.This lack of a secondary maximum has been observed inSN 2002cx-like objects (Li et al. 2003; Phillips et al. 2007).However, as seen in Figure 2, SN 2005hk (a SN 2002cx-likeobject) evolves much faster than SN 2009dc in all bandsand proves to be a poor match photometrically. The colourindices of SN 2009dc and SN 2005hk are again a poor matchas shown in Figure 5.Spectroscopically, SN 2009dc does not really resembleSN 2002cx-like objects before maximum (see Fig. 6), norabout 9 months after maximum (see Fig. 12). At these lat-est times, the spectrum of SN 2009dc consists of broademission from forbidden iron lines while the spectrum ofSN 2002cx is made up of many narrow permitted lines ofFe II , Na II , Ca II , and (tentatively) O I (Jha et al. 2006). This is supported by the fact that the narrow spectral fea-tures of SN 2002cx do not simply appear to correspond toresolved versions of the blends seen in SN 2009dc.Despite all of these differences, we have shown thatat 1–2 months past maximum brightness the spectra ofSN 2009dc and SN 2002cx are surprisingly similar. In Fig-ure 9 we have combined the day 20 and 25 spectra ofSN 2002cx from Li et al. (2003) in order to improve theS/N and the wavelength coverage. Not only are the overallshapes of these two spectra nearly identical, but the width,strength, and existence of almost all of the individual spec-tral features match extremely well (with one significant ex-ception, the Si II λ § ∼ c (cid:13)000
Our spectrum of SN 2009dc about 281 days past maximum, and a few comparison SNe with days relative to maximum lightindicated for each spectrum (in parentheses). The spectra from top to bottom are the normal Type Ia SN 1998bu (Jha et al. 2006),SN 2009dc, SN 2006gz (Maeda et al. 2009), SN 1990N (via SUSPECT), and the peculiar Type Ia SN 2002cx (Jha et al. 2006). Majorspectral features are labeled. All spectra have had their host-galaxy recession velocities removed and have been dereddened according tothe values presented in their respective references. minosity of SN 2009dc is much larger than average (see § § R and I -band light curves of SN 2009dc.This lack of a secondary maximum has been observed inSN 2002cx-like objects (Li et al. 2003; Phillips et al. 2007).However, as seen in Figure 2, SN 2005hk (a SN 2002cx-likeobject) evolves much faster than SN 2009dc in all bandsand proves to be a poor match photometrically. The colourindices of SN 2009dc and SN 2005hk are again a poor matchas shown in Figure 5.Spectroscopically, SN 2009dc does not really resembleSN 2002cx-like objects before maximum (see Fig. 6), norabout 9 months after maximum (see Fig. 12). At these lat-est times, the spectrum of SN 2009dc consists of broademission from forbidden iron lines while the spectrum ofSN 2002cx is made up of many narrow permitted lines ofFe II , Na II , Ca II , and (tentatively) O I (Jha et al. 2006). This is supported by the fact that the narrow spectral fea-tures of SN 2002cx do not simply appear to correspond toresolved versions of the blends seen in SN 2009dc.Despite all of these differences, we have shown thatat 1–2 months past maximum brightness the spectra ofSN 2009dc and SN 2002cx are surprisingly similar. In Fig-ure 9 we have combined the day 20 and 25 spectra ofSN 2002cx from Li et al. (2003) in order to improve theS/N and the wavelength coverage. Not only are the overallshapes of these two spectra nearly identical, but the width,strength, and existence of almost all of the individual spec-tral features match extremely well (with one significant ex-ception, the Si II λ § ∼ c (cid:13)000 , 1–30 N 2009dc, Super-Chandrasekhar Mass? − for SN 2002cx, and Phillips et al. (2007) de-termine expansion velocities in nearly the same range forSN 2005hk. For SN 2009dc we also derive velocities in thisrange for a few ions (O I , Ca II , Fe II , and Si II ) with sig-nificant uncertainty since not only are many of these fea-tures blended, but the line identifications themselves are notdefinitive. These values also match well with the expansionvelocity of ∼ − derived by Yamanaka et al. (2009)from their spectrum of SN 2009dc from 25 days past maxi-mum.Both Li et al. (2003) and Branch et al. (2004) deter-mine an expansion velocity for SN 2002cx at 56 days pastmaximum of ∼ − based on the Fe II λ − for SN 2005hk at 55 dayspast maximum (also using the Fe II λ II feature, we determine an expansion velocity of ∼ − for our spectra of SN 2009dc from 52 and64 days past maximum. If we use the Si II λ − depending on the fea-ture (Li et al. 2003; Branch et al. 2004), whereas they areabout equal to those of SN 2005hk (Phillips et al. 2007).In conclusion, SN 2009dc and SN 2002cx-like objectshave quite different photometric properties but are nearlyidentical spectroscopically only at 1–2 months past maxi-mum brightness. The expansion velocity, the blackbody tem-perature, and the chemical composition match extremelywell at these epochs, but the evolution of these observablesfrom pre-maximum to nearly a year after maximum is quitedifferent between SN 2009dc and SN 2002cx-like objects. In Figure 13 we present our late-time spectra of anotherpossible SC SN Ia, SN 2007if. As noted earlier, the spectraof SN 2007if generally look similar to those of SN 2009dc,SN 2002cx, and even SN 2005cf at comparable epochs. Themost significant spectral differences between SN 2007if andSN 2009dc are, as stated previously, that the expansion ve-locities seen in SN 2007if are ∼ − larger than thosein SN 2009dc and SN 2002cx at comparable phases. Thislarger velocity is likely causing some of the medium-widthfeatures seen in SN 2009dc and SN 2002cx to be “smoothedout”, and hence we do not detect many of these features inSN 2007if. It also has velocities which are smaller than thoseseen in SN 2005cf, but otherwise the spectra match well.We have also stated previously that SN 2007if does notshow the strong Si II λ II near-IR triplet of SN 2007if shows the complexstructure seen in SN 2009dc.The main aspects of the spectral evolution of SN 2007ifover our three epochs is the disappearance of broad emis-sion near 5500 ˚A and in the ranges 6700–7100 ˚A and possi-bly 7800–8100 ˚A (though our data are noisy in this second range of wavelengths). These match nearly exactly with thespectral features of SN 2009dc that were seen to decrease instrength over similar epochs. As mentioned above, we detect weak (but noticeable), nar-row absorption from Na I D in all of our spectra ofSN 2009dc, both at zero redshift and at the recession ve-locity of the SN itself ( cz = 6300 ±
300 km s − ). Eventhough there are clear notches in all of our spectra at thecorrect wavelengths for Na I D in the MW and in the hostof SN 2009dc (UGC 10064), they are not broader than ourspectral resolution.We attempt to measure an EW of both absorption fea-tures in each of our observations, and we calculate averagevalues of ∼ . ± . ∼ . ± . I D absorption into extinction, but a few (mainlyempirical) relations do exist (e.g., Barbon et al. 1990; Mu-nari & Zwitter 1997; Turatto et al. 2003). Using these con-versions, we determine a range of values for E ( B − V ) bothin the MW (0 . (cid:46) E ( B − V ) MW (cid:46) .
18 mag) and inUGC 10064 (0 . (cid:46) E ( B − V ) host (cid:46) . E ( B − V ) to the value from the Galac-tic dust maps of Schlegel et al. (1998). Their maps give E ( B − V ) MW = 0 .
070 mag for the position of SN 2009dc,slightly lower than our range of values. However, since theaccuracy of the Schlegel et al. (1998) dust maps is almost cer-tainly higher than that of our EW measurements, we adopt E ( B − V ) MW = 0 .
070 mag for our analysis.The extinction from UGC 10064 is a bit more com-plicated to determine accurately. If the Lira-Phillips rela-tion holds for SN 2009dc, then we expect an extinction of E ( B − V ) host ≈ .
26 mag (see § I D is directly propor-tional to E ( B − V ), as Tanaka et al. (2010) do, then wederive E ( B − V ) host ≈ (cid:0) . / . (cid:1) × . ≈ . E ( B − V ) host = 0 . SN 2009dc occurred ∼ (cid:48)(cid:48) away from the centre of the S0galaxy UGC 10064. We adopt the luminosity distance ( d L )as the distance to the SN, again assuming a ΛCDM Universewith H = 70 km s − Mpc − , Ω m = 0 .
27, and Ω Λ = 0 . The third wavelength range which possibly decreased instrength in SN 2009dc, 7800–8100 ˚A, is outside of our spectralcoverage of SN 2007if.c (cid:13) , 1–30 Silverman, et al. l og ( F λ ) + c on s t a n t +37+63+92SN 2007if Figure 13.
Our spectra of SN 2007if with days relative to maximum light indicated in each case. We have removed the host-galaxyrecession velocity ( cz = 22,200 ±
250 km s − , Scalzo et al. 2010). (Spergel et al. 2007) and z = 0 . ± . cz = 300 km s − toallow for a peculiar velocity induced by the gravitationalinteraction of neighboring galaxies. At d L = 97 ± ∼
13 kpc from thecentre of UGC 10064. A useful galactic scale is the ra-dius containing 50% of the flux (Petrosian 1976), which forUGC 10064 is about 4 . (cid:48)(cid:48) ± . (cid:48)(cid:48)
21 in the R band (Abaza-jian et al. 2009). This puts SN 2009dc about 5.7 Petrosianradii (in projection) from the centre of its host. For compar-ison, SN 2003fg was found to be projected 0.9 kpc from thecore of its low-mass star-forming host (Howell et al. 2006)and SN 2007if was discovered effectively directly on top ofits low-mass and low-metallicity host (Scalzo et al. 2010).The host of SN 2006gz is an Scd galaxy, and the SN wasprojected 14.4 kpc from the nucleus (Hicken et al. 2007),while SN 2004gu was projected 2.21 kpc from its host’s nu-cleus (Monard et al. 2004). It seems that host-galaxy typeand distance from the nucleus vary widely among candidate SC SN Ia (though the distance comparison is difficult givenprojection effects). However, SN 2003fg, SN 2007if, and pos-sibly SN 2009dc (see below) were all associated with regionsof recent star formation.Wegner & Grogin (2008) inspected an optical spectrumof the core of UGC 10064 and derived an age of 4 . ± . Z /H] = 0 . ± .
07, and a chemical abundance[ α /Fe] = 0 . ± .
09. This value for metallicity is about 1 σ above the average found for 26 S0/E void galaxies by Wegner& Grogin (2008), whereas the chemical abundance is justabout the same as their average value. Optical spectra ofthe nucleus of UGC 10064 show no strong emission lines(Wegner & Grogin 2008; Abazajian et al. 2009), implyingvery little ongoing star formation ( (cid:46) . (cid:12) yr − ) in thecore of the galaxy.However, recent deep H I surveys of S0 galaxies haveshown that the majority of noncluster early-type galaxiescontain substantial reservoirs (10 –10 M (cid:12) ) of neutral gasthat extend to large galactocentric radii (Morganti et al.2006), whose morphology and kinematics suggest external c (cid:13) , 1–30 N 2009dc, Super-Chandrasekhar Mass? origins. On the other hand, in the majority of these galaxiesthe column density of this gas is too low to support starformation at large radii, and thus the stellar populationsthroughout the galaxy are old.Interestingly, in a fraction of the early-type galaxies richin H I , a dynamical feature such as a prominent bar can cre-ate relatively dense ring-like accumulations of neutral gas atlarge radii (e.g., Schinnerer & Scoville 2002; Weijmans et al.2008). In these regions of enhanced H I , low-level star forma-tion ( ∼ (cid:12) yr − ) is observed (Jeong et al. 2007; Shapiroet al. 2010). Since SN 2009dc exploded in the outskirts ofan S0 galaxy, it seems difficult to say with certainty whetherSN 2009dc came from an old or young stellar population.In the Supplementary Information of Howell et al.(2006), the age of the host galaxy of SN 2003fg is esti-mated to be ∼
700 Myr (with significant uncertainty) andthe star-formation rate to be 1 . +0 . − M (cid:12) yr − , averagedover 0.5 Gyr. This star-formation rate is nearly an order ofmagnitude greater that the value derived by Wegner & Gro-gin (2008) for the core of UGC 10064 as well as the residualstar-formation rate seen in rings of enhanced neutral gas atlarge galactocentric radii (Jeong et al. 2007; Shapiro et al.2010). SC SNe Ia are expected to be found preferentially inyoung stellar populations (e.g., Howell et al. 2006; Chen &Li 2009, and references therein), but they are not necessarilyrestricted to them. If SC SNe Ia are the result of the mergerof two degenerate objects, then the merger timescale can beset by the emission of gravitational waves (e.g., Iben & Tu-tukov 1984). Depending on the particular orbital parametersof the binary, there may be a significant delay between theformation of the constituent degenerate objects and theireventual merger. Thus, SC SNe Ia could occur in older stel-lar populations.Even though there is a significant difference betweenthe average star-formation rates near the sites of SN 2009dcand SN 2003fg, there is still a possibility that they bothcome from relatively young stellar populations. As men-tioned above, an age of ∼
700 Myr was calculated for theenvironment of SN 2003fg (Howell et al. 2006, Supplemen-tary Information), while an age of ∼ core of the host of SN 2009dc (Wegner & Grogin2008). The fact that SN 2009dc is a few effective radii awayfrom the nucleus of its host, coupled with the observationsof low levels of residual star formation in other early-typegalaxies at such galactocentric radii, imply that it is possi-ble that SN 2009dc also came from a relatively small, butyoung, stellar population on the edge of UGC 10064, andthat the age calculated for the core of the galaxy is not ac-tually representative of the local environment of SN 2009dc.Furthermore, in the SDSS Data Release 7, an ex-tended, blue source appears 99 . (cid:48)(cid:48) Swift /UVOT images of the SN 2009dc field(shown in Fig. 14) indicate that UGC 10063 is significantlybluer than UGC 10064. The irregularly shaped UGC 10063is barely visible in the KAIT R -band image to the northwestof SN 2009dc in Figure 1. Stacked UVOT images show thatUGC 10063 increases in prominence in progressively bluerbands. As seen in Figure 14, UGC 10063 seems to have a g Uuvm2 uvw2
1’ 1’UGC 10063 2009dc UGC 10063 2009dc
Figure 14.
Keck g -band image and stacked UVOT U , uvm uvw (cid:48) × (cid:48) , and theremaining panels are 4.67 (cid:48) × (cid:48) ; north is up and east is to theleft in all four images. UGC 10063 increases in prominence inprogressively bluer bands, and is much bluer than UGC 10064(compare its appearance here with that in Fig. 1). It also seemsto have a tidal tail extending toward UGC 10064, pointed roughlyin the direction of the site of SN 2009dc. tidal tail extending toward UGC 10064, pointed roughly atthe site of SN 2009dc.On 2010 Feb. 7 we obtained a spectrum of UGC 10063using LRIS on the Keck I telescope (with a 1 (cid:48)(cid:48) -wide slit, a600/4000 grism on the blue side, and a 400/8500 gratingon the red side, resulting in FWHM resolutions of ∼ ∼ α is undetected). StrongBalmer lines and the lack of obvious emission lines are thehallmarks of so-called “post-starburst” galaxies in which vig-orous star-formation activity ended between ∼
50 Myr and1.5 Gyr ago (Poggianti et al. 2009, and references therein).This is consistent with UGC 10063 being a SBdm galaxy(de Vaucouleurs et al. 1991). From the seven Balmer ab-sorption lines marked in Figure 15, we calculate the redshiftof UGC 10063 to be 0 . ± . I line ( z = 0 . ± . II H&K absorption).Based on the spectrum of UGC 10063 and its positionrelative to UGC 10064, we propose that these two galax-ies had a close encounter in the relatively recent past. IfUGC 10063 passed near the nucleus of UGC 10064, gasfrom UGC 10063 could have been stripped off and becomegravitationally bound to UGC 10064 (which has been seenin both simulations and observations, e.g., Barnes & Hern- c (cid:13) , 1–30 Silverman, et al. l og ( F λ ) UGC 10063(SDSS J155108.42+254321.3)
Balmer series at z = 0.021
Figure 15.
An optical spectrum of UGC 10063 (SDSS J155108.42+254321.3). The inset is a zoom-in of the full spectrum. NarrowBalmer-series absorption (H β through H θ ) is marked, and these features yield z = 0 . ± . α is undetected at an expected wavelength of ∼ α emission from residual H II regions fills inthe absorption line. quist 1991; Falc´on-Barroso et al. 2004). This newly acquiredgas (some of which could be at large galactocentric radii)would have likely undergone a burst of star formation soonafter the two galaxies had their closest encounter (Mihos &Hernquist 1994; Barnes & Hernquist 1996), and this mightbe how and when the progenitor of SN 2009dc formed. BothSN 2007if and SN 2003fg were also found in hosts with recentstar formation (Scalzo et al. 2010); in fact, SN 2003fg wasdiscovered in a tidal feature (Howell et al. 2006) reminiscentof SN 2009dc and UGC 10063. To estimate the bolometric luminosity of SN 2009dc, we fol-low the method outlined by Howell et al. (2009). Using spec-tra of SN 2009dc presented here and spectra of the similar SC SN Ia candidate SN 2007if from Scalzo et al. (2010),we take each simultaneous epoch of
BVRI photometry andwarp the spectrum closest in phase to match the photom-etry using a smooth third-order spline. We then integratethe warped spectrum over wavelengths of 4000–8800 ˚A (i.e.,from the blue edge of the B -band filter to the red edge of I -band filter.) We convert the flux into a luminosity usingthe luminosity distance ( d L = 97 ± § BVRI only accounts for ∼
60% of the total luminosity in the case of SN 2005cf,a normal SN Ia. They also found that an additional 20% isemitted in the UV and U band, and the final 20% in thenear-IR. We test to see if the tabulated corrections for theUV and U band are reasonable for SN 2009dc by first calcu-lating the bolometric luminosity using optical and UV pho- c (cid:13) , 1–30 N 2009dc, Super-Chandrasekhar Mass? t =20.5 d t =22.8 d ! "! : ( ;< ( ’ ! : ! B: (C(!4!(.%15!!=>?(@9A ! B: (C(!4 ! B: (C(!46(.%15!!"?D Figure 16.
Bolometric light curve of SN 2009dc assuming variouslevels of extinction, in comparison with that of SN 2005cf. Thecurve plotted with empty circles assumes no host-galaxy extinc-tion, the half-filled circles assume E ( B − V ) host = 0 . R V = 3 .
1, and the filled circles assume E ( B − V ) host = 0 . R V = 3 .
1. For comparison, we include the bolometric lightcurve of SN 2005cf as a dashed line (Wang et al. 2009b). Weindicate our calculated range of values of t +1 / (i.e., the num-ber of days after maximum bolometric luminosity when the bolo-metric luminosity has decreased to half its maximum value) forSN 2009dc within the shaded grey region bounded by dotted lines.The lower bound of t +1 / = 20 . E ( B − V ) host = 0 . t +1 / = 22 . tometry at t = 0 days (using our optical ground-based pho-tometry combined with UVOT photometry of SN 2009dc).We find that our results are consistent with a correctionof 20%. Yamanaka et al. (2009) found that their near-IRdata for SN 2009dc represented ∼
20% of the bolometric lu-minosity. With this, we cautiously adopt a 40% correctionfor our bolometric luminosity using our well-sampled
BVRI photometry. Overall, SN 2005cf is a much different SN thanSN 2009dc, so we include a systematic uncertainty of 20%for our bolometric luminosity.Figure 16 shows our final bolometric light curves forthe various levels of host-galaxy extinction we have adoptedthroughout this paper. We also plot the bolometric lightcurve of SN 2005cf for comparison.If we use our fiducial nonzero value for the host-galaxyreddening, E ( B − V ) host = 0 . E ( B − V ) host = 0 . M V = − . ± . L bol = (3 . ± . × erg s − ) and M V = − . ± .
10 mag ( L bol = (7 . ± . × erg s − ), respectively. We note that assumingno host-galaxy extinction we find M V = − . ± .
10 mag( L bol = (2 . ± . × erg s − ). A summary of theseresults can be found in Table 7. Our range of values of L bol for SN 2009dc matches quite well that of Yamanaka et al. Table 7.
Bolometric luminosity for various reddenings E ( B − V ) host L bol,max M V, max (mag) (10 erg s − ) (mag)0.0 2 . ± . − . ± . . ± . − . ± . . ± . − . ± . (2009): (2 . . × erg s − . The most significant differ-ence between the two analyses comes from using a differentrange of values for E ( B − V ) host and R V . Ni Masses and Energetics Ni Masses from Arnett’s Law
The vast majority of the bolometric luminosity of SNe Iacomes from the decay of Ni to Co (and subsequently thedecay of Co to Fe), so we can calculate how much Niwas created in SN 2009dc by using “Arnett’s law”, whichasserts that the luminosity of a SN Ia at maximum light isproportional to the instantaneous rate of radioactive decayof Ni (e.g., Arnett 1982; Stritzinger & Leibundgut 2005).Arnett’s law is often written M Ni = L bol α ˙ S ( t R ) , (1)where M Ni is the mass of Ni present in the ejecta, L bol isthe bolometric luminosity at maximum light, α is the ratioof bolometric to radioactive luminosities (which is of orderunity), and ˙ S ( t R ) is the radioactive luminosity per mass of Ni as a function of the rise time, t R .For SN 2009dc, we have tight constraints on the risetime from our photometric data and, as described in § ± S ( t R ) from Stritzinger & Leibundgut (2005), weget ˙ S ( t R ) ≈ (1 . ± . × erg s − M (cid:12)− . Stritzinger& Leibundgut (2005) find α = 1 . ± .
2, which we use forour analysis, though we note that this is an upper limit(which leads to a lower limit on the Ni mass) since any Ni that is above the photosphere will not contribute to theSN luminosity.Using these values, along with our most likely estimateof the bolometric luminosity mentioned above, we calculatethat SN 2009dc synthesised about 1 . ± . (cid:12) of Ni. Ifwe adopt the bolometric luminosity assuming no host-galaxyreddening and our largest reasonable host-galaxy reddening,along with our calculated rise time and α = 1 .
2, we get ∼ (cid:12) and ∼ (cid:12) of Ni, respectively. Ni Masses from Late-Time Photometry
The detection of SN 2009dc in late-time images at levelssimilar to those of normal SNe Ia offers a stark contrast toSN 2006gz, another SC SN Ia candidate. Late-time pho-tometry obtained by Maeda et al. (2009) ∼
360 days af-ter maximum found that SN 2006gz had faded significantlyfaster than normal SNe Ia, casting doubt on the amountof Ni synthesised in the explosion. The authors constrain M Ni (cid:46) . (cid:12) assuming that the positrons are nearly com-pletely trapped in the ejecta. This is at odds with the ∼ (cid:12) c (cid:13) , 1–30 Silverman, et al. of Ni calculated to power the light curve at early times(Hicken et al. 2007). To reconcile this difference in nickelmass, it was suggested that perhaps SN 2006gz was a highlyasymmetric explosion, and due to the exact viewing anglemore Ni was inferred at early times than was actuallypresent (Maeda et al. 2009). However, SN 2009dc appears tobe spherically symmetric at early times (Tanaka et al. 2010).Furthermore, Maeda et al. (2009) posit that dust might haveformed in the ejecta of SN 2006gz by this time, which wouldreprocess much of the optical light to longer wavelengths.However, we do not see strong evidence for dust formationin SN 2009dc since the spectral features all appear to evolvesymmetrically.To constrain the nickel mass produced by SN 2009dcfrom late-time photometry, we write out the SN luminos-ity as a function of time (e.g., Clocchiatti & Wheeler 1997;Maeda et al. 2003, 2009), L ( t ) = M Ni (cid:16) ε γ (1 − e − τ ( t ) ) + f e + ε e + (cid:17) e − t/ . , (2)where t is the number of days from the explosion date, M Ni is the initial amount of Ni synthesised, ε γ = 6 . × erg s − g − , τ ( t ) is the time-dependent optical depth to γ -rays, f e + is the fraction of positrons trapped in the ejecta, ε e + = 2 . × erg s − g − , and 111.3 days is the e -foldingtime of Co decay. Evaluating τ ( t ) requires defining a modelwith estimates for the kinetic energy of the explosion andthe total mass of the white dwarf.We can place constraints on the Ni mass by only con-sidering the luminosity powered by the trapped positrons;to do this we set τ ( t ) = 0 in Eqn. (2). Using only the opti-cal contribution to the bolometric luminosity over the range4000–8000 ˚A from our Keck data taken 281 days past max-imum (which is 304 days after explosion), we calculate the Ni mass for a number of reddening and f e + values whichcan be found in Table 8. Our calculation of the bolometricluminosity does not include a bolometric correction for anypossible contribution to the bolometric luminosity in theNIR. For our nominal values of E ( B − V ) host = 0 . R V = 3 .
1, and f e + = 1, we calculate a Ni mass of1 . ± . (cid:12) , consistent with the Ni mass derived fromArnett’s law and our data at B max ( § Ni mass using our epoch ofphotometry from 403 days past maximum (which is 426 daysafter explosion). However, without a spectrum to model thespectral energy distribution at this phase, we had to useour last spectrum taken 281 days past B max , which couldlead to an inaccurate measure of the bolometric luminosity.Furthermore, the detection of the SN was marginal in allbands. Using our nominal values of E ( B − V ) host = 0 . R V = 3 .
1, and f e + = 1, we calculate a Ni mass of0 . ± .
06 M (cid:12) , clearly much lower than previous estimatesof the Ni mass. We summarise Ni masses for a range ofparameters in Table 8.The drop in derived Ni mass and expected luminos-ity could indicate the formation of dust between t = 281and 426 days past maximum light, although this cannot be Note that this value was derived by assuming a high value forthe host-galaxy reddening of SN 2006gz, which might have beenan overestimate.
Table 8.
Ni mass estimates from late-time photometryDays past f e + E ( B − V ) host L abol 56 NiExplosion (mag) (10 erg s − ) (M (cid:12) )304 0.75 0 3 . ± . . ± . . ± . . ± . . . ± . . ± . . . ± . . ± . . . ± . . ± . . . ± . . ± . . ± .
05 0 . ± . . ± .
05 0 . ± . . . ± .
07 0 . ± . . . ± .
07 0 . ± . . . ± .
12 0 . ± . . . ± .
12 0 . ± . a Bolometric luminosity only contains contribution over the range4000–8800 ˚A. substantiated without IR data or a relatively high-S/N spec-trum. The unexpected drop could also be an indication thatthe positron trapping fraction at this phase is significantlylower than the nominal value of f e + = 1. To produce the ∼ (cid:12) of Ni found using our photometry from 281 dayspast maximum, we would need f e + ≈ .
25. SN 2009dc couldhave undergone an “IR catastrophe” which would causemost of the emission to escape at IR wavelengths (as op-posed to the optical; Axelrod 1988) as the temperature ofthe SN cools to ∼ Ni mass could suggestthat the late-time light curve of SN 2009dc is not poweredby the decay of Co. This conclusion, however, is at oddswith the Ni mass derived from our epoch of photometryfrom t = 281 days past maximum light. Without near-IRdata or a spectrum at this epoch, it is difficult to reconcilethese differences. We again caution that the spectral energydistribution is ill constrained at this phase. Using the technique developed by Howell et al. (2006), we at-tempt to combine our observations of SN 2009dc with prop-erties of WDs and SNe Ia in order to calculate the mass ofthe progenitor WD as well as the amount of Ni producedin the explosion. We note that this method is independentof the one used in the previous section which utilitized thelate-time photometry of SN 2009dc. As will be mentionedbelow, there are a few assumptions and simplifications thatgo into this analysis, and thus the final values should prob-ably not be taken literally. However, the sense of the rela-tionships and trends these calculations indicate should bereliable and will show that SN 2009dc likely had a SC WDprogenitor.The ejecta of SNe Ia derive their kinetic energy ( E K )from the energy released during explosive nucleosynthesis( E n ), but the ejecta must first overcome the binding energy( E b ) of the WD progenitor (Branch 1992). Therefore, we can c (cid:13)000
25. SN 2009dc couldhave undergone an “IR catastrophe” which would causemost of the emission to escape at IR wavelengths (as op-posed to the optical; Axelrod 1988) as the temperature ofthe SN cools to ∼ Ni mass could suggestthat the late-time light curve of SN 2009dc is not poweredby the decay of Co. This conclusion, however, is at oddswith the Ni mass derived from our epoch of photometryfrom t = 281 days past maximum light. Without near-IRdata or a spectrum at this epoch, it is difficult to reconcilethese differences. We again caution that the spectral energydistribution is ill constrained at this phase. Using the technique developed by Howell et al. (2006), we at-tempt to combine our observations of SN 2009dc with prop-erties of WDs and SNe Ia in order to calculate the mass ofthe progenitor WD as well as the amount of Ni producedin the explosion. We note that this method is independentof the one used in the previous section which utilitized thelate-time photometry of SN 2009dc. As will be mentionedbelow, there are a few assumptions and simplifications thatgo into this analysis, and thus the final values should prob-ably not be taken literally. However, the sense of the rela-tionships and trends these calculations indicate should bereliable and will show that SN 2009dc likely had a SC WDprogenitor.The ejecta of SNe Ia derive their kinetic energy ( E K )from the energy released during explosive nucleosynthesis( E n ), but the ejecta must first overcome the binding energy( E b ) of the WD progenitor (Branch 1992). Therefore, we can c (cid:13)000 , 1–30 N 2009dc, Super-Chandrasekhar Mass? write E K = M WD v E n − E b , (3)where M WD is the mass of the WD just before explosion and v is a representative velocity of the ejecta. A single value forvelocity only relates to the kinetic energy of the SN averagedover the entire ejecta, and thus there has been disagreementas to which actual value should be used for a given SN.Previous studies have used the velocity of the Si II λ Ni (Nomoto et al. 1984; Khokhlov et al. 1993), thetotal mass of the WD can be related to the nickel mass by M Ni = 0 . M WD f IGE , (4)where f IGE is the fractional composition of IGEs in the SNejecta. However, the exact amount of IGEs that are madeup of Ni may not be the same for all SNe Ia and likelydependent on other, external factors such as metallicity (e.g.,Howell et al. 2009, and references therein).Furthermore, Branch (1992) found that fusing equalparts carbon and oxygen all the way to iron yields ε Fe =1 . × erg M (cid:12)− , and burning the same material onlyup to silicon liberates about 76% of that energy. Thus wecan write E n = ε Fe M WD ( f IGE + 0 . f IME ) , (5)where f IME is the fractional composition of IMEs in theejecta.Finally, we define the fractional composition of un-burned carbon in the SN ejecta as f C . Based on our defi-nitions we can write1 = f C + f IME + f IGE . (6)Putting all of this together, we relate various properties ofthe progenitor WD to properties of the SN ejecta it createdusing v ≈ ε Fe (cid:18) . M Ni M WD − . f C + 0 . (cid:19) − E b M WD . (7)However, given the assumptions and simplifications thatwent into this “equation”, we stress that it should be usedmore as an illustration of how the different parameters re-late to each other and less as a tool to actually calculateWD and Ni masses.Values of E b range from about 0 . × erg for a WDwith a mass of 1.4 M (cid:12) to about 1 . × erg for a WDwith a mass of 2 M (cid:12) and a central density of 4 × g cm − (Yoon & Langer 2005). We perform separate calculationsfor each pair of these values of E b and M WD . Tanaka et al.(2010) measure the Si II λ − in their spectrum of SN 2009dc obtained6 days past maximum, while we find the same feature to beblueshifted by 5000–6000 km s − in our spectra taken 35,52, and 64 days past maximum. Since v has been evaluated This parameter does not appear in the equation for E n ; it rep-resents the amount of unburned material, and thus does not con-tribute to the energy released by nucleosynthesis. at 10 and 40 days past maximum for similar analyses pre-viously (Benetti et al. 2005; Howell et al. 2006), we will usea range of velocities from 7000 km s − to 5500 km s − forour analysis. Finally, since there is clear indication of un-burned material in the ejecta of SN 2009dc (see § f C must be nonzero. However, based on modelsof SN Ia ejecta (e.g., Nomoto et al. 1984) and the relativescarcity of carbon detections in SNe Ia (Marion et al. 2006;Thomas et al. 2007), we will only use values in the range0.05–0.3 for f C .With such large uncertainties and so many parameters,we will clearly calculate a huge range of Ni masses, and werealize that a few assumptions and model-dependent valueshave come into the derivation. However, this is still a usefulline of reasoning in our attempt to determine both the massof the WD progenitor and the amount of Ni synthesisedby SN 2009dc. We therefore forge ahead and attempt toeliminate at least some of the most extreme situations.Almost all of parameter space is ruled out when we usea 1.4 M (cid:12)
WD (and its associated binding energy), since wecalculate negative nickel masses for these cases. The largest Ni mass we can reasonably derive for a 1.4 M (cid:12)
WD (using f C = 0 . v = 7000 km s − ) is 0.06 M (cid:12) , nearly an orderof magnitude smaller than the amount of nickel produced bya normal SN Ia (e.g., Nomoto et al. 1984; Kasen et al. 2008).We have also shown that SN 2009dc is much more luminousthan the average SN Ia. Therefore, if it did have a 1.4 M (cid:12) progenitor, then we would need to invoke a large, hithertounknown energy source to power its light curve, which seemshighly unlikely.The situation changes when we use a 2 M (cid:12) WD (witha central density of 4 × g cm − ), put forth as a possibleSC SN Ia progenitor by Yoon & Langer (2005). Again, asignificant part of parameter space is eliminated, since atlow values of f C we calculate negative (or extremely small)nickel masses. Assuming that f IME must be (cid:38) . ∼ (cid:12) . In order to match our nominal Nimass range of 1.4–1.7 M (cid:12) and satisfy our constraints on thefractional composition of elements mentioned previously, wemust resort to using a WD progenitor more massive than2 M (cid:12) .Again, the final values calculated above are almost cer-tainly not the true answer. However, the range of plausiblevalues presented, as well as the relationships between theparameters of the SN and its WD progenitor, yet again in-dicate that the progenitor of SN 2009dc was likely a SC WD.The various analyses presented above seem to stronglyfavor the conclusion that the progenitor of SN 2009dc was aSC WD with a mass of probably greater than ∼ (cid:12) . Theseanalyses also indicate that SN 2009dc most likely produced1.4–1.7 M (cid:12) of Ni (assuming our nominal value for thehost-galaxy reddening and our peak bolometric luminosity).More than 1 M (cid:12) of Ni was almost certainly created bySN 2009dc, and the actual value could be as high as ∼ (cid:12) .This matches well with the conclusions of Yamanaka et al.(2009), who calculate a similar range of nickel masses forSN 2009dc (1.2–1.8 M (cid:12) ). Part of the difference can be ac-counted for by the fact that the two studies use differentvalues of host-galaxy reddening (causing the derived bolo-metric luminosities to differ somewhat; see § c (cid:13) , 1–30 Silverman, et al. the majority of the difference in derived Ni masses comesfrom the assumed rise time of SN 2009dc. Yamanaka et al.(2009) adopt a rise time of 20 days based on comparisons totypical SNe Ia and SN 2006gz, while we use a rise time of23 days based on our pre-maximum photometry. The longerrise time used in our study leads to a larger derived nickelmass for SN 2009dc.SN 2003fg had a derived progenitor mass of ∼ (cid:12) and produced ∼ (cid:12) of Ni (Howell et al. 2006), andthese values are quite similar to those we calculate forSN 2009dc. SN 2006gz was estimated to have produced 1–1.2 M (cid:12) of Ni, which is on the low end of our range forSN 2009dc, and it was also claimed to have a SC WD pro-genitor (though no attempt was made to further constrainthe progenitor mass; Hicken et al. 2007).
Any theoretical model which is postulated to explainSN 2009dc, with or without a SC WD, must be able to re-produce the observed peculiarities for which we have verytight constraints: ( a ) high luminosity even when assumingno host-galaxy reddening, ( b ) relatively long light-curve risetime, ( c ) relatively slow photometric decline and late-timespectroscopic evolution, ( d ) the presence of carbon in spec-tra near maximum brightness, ( e ) the presence of silicon inspectra as late as a few months past maximum brightness, ( f )IGEs dominating the spectra at late times, and ( g ) mostlyspherically symmetric ejecta near maximum with possibleclumpy layers of IMEs (Tanaka et al. 2010).Below we consider both SC and non-SC models, all ofwhich involve the thermonuclear explosion of a WD. How-ever, the similarities between SN 2009dc and SN 2002cx andits brethren (see § Two- and three-dimensional models of differentially rotat-ing massive WDs have been presented in the literature,and calculations show that SC WDs with masses as largeas 2 M (cid:12) are possible by accretion from a nondegeneratecompanion, while masses up to ∼ (cid:12) are possible froma double-degenerate merger (Yoon & Langer 2005; Yoonet al. 2007, respectively). More recent studies of double-degenerate mergers have even shown possible SN Ia progen-itors with total masses approaching 2.4 M (cid:12) (Greggio 2010).This is encouraging for the case of SN 2009dc, since our en-ergetics arguments imply that its progenitor is likely greaterthan ∼ (cid:12) . However, calculations by Piro (2008) suggestthat differential rotation is unlikely for WDs accreting froma nondegenerate companion, and thus their masses cannotexceed the Chandrasekhar mass by more than a few percent.Yoon & Langer (2005) also consider how much Niwould be produced by a SN Ia whose progenitor is ∼ (cid:12) ,calculating values of 0.4–1.3 M (cid:12) . This large range does en-compass the lower end of our range of Ni for SN 2009dc. However, only specific kinetic energies (i.e., kinetic ener-gies per unit mass) which are lower than what we find forSN 2009dc by a factor of ∼ II and Ca II features and thelack thereof in two different C II features. Thus, we con-clude that the models of Pfannes et al. (2010) do not seemto reproduce the observations of SN 2009dc.Models of SC WDs and their evolution in close binarysystems with nondegenerate companions can also be foundin the literature (Chen & Li 2009). These evolutionary cal-culations include the effects of varying the orbital period ofthe binary systems, the metallicity and mass-transfer rateof the binary companion, and (most importantly) the massof the WD. Chen & Li (2009) find that WDs with initialmasses of ∼ (cid:12) , under the right mass-transfer condi-tions, can accrete up to masses of about 1.4–1.8 M (cid:12) beforeexploding as a SN Ia (though they find that most of theseWDs explode with masses not much above 1.4 M (cid:12) ). Thus itseems somewhat unlikely that these models can explain theprogenitor we suggest for SN 2009dc; even if the occasionalWD can accrete up to 1.8 M (cid:12) , this is at the lowest end ofour range of WD progenitor masses.Spherically symmetric, one-dimensional radiationtransport calculations for normal and SC WDs have beencarried out by Maeda & Iwamoto (2009). They find thatbased on the light-curve shapes, photospheric velocities,peak bolometric luminosities, and peak effective temper-atures, SN 2006gz likely came from a SC WD whileSN 2003fg did not. Since the observables of SN 2003fgused for their analysis have very similar values to thoseof SN 2009dc, it would seem that their models imply thatSN 2009dc is also not a SC SN Ia. Maeda & Iwamoto (2009)point out that SN 2003fg (and thus probably SN 2009dc aswell) could be a SC SN Ia if the progenitor star were highlyaspherical, but again this seems unlikely for SN 2009dcfrom the spectropolarimetric data (Tanaka et al. 2010).The primary independent variable used in the calcu-lations of Maeda & Iwamoto (2009) is t +1 / , the numberof days after maximum bolometric luminosity when the Maeda & Iwamoto (2009) claim that most of the emission likelyshifted to the near-IR and mid-IR at late times in order to explainthe relatively faint late-time observations of Maeda et al. (2009).c (cid:13) , 1–30
N 2009dc, Super-Chandrasekhar Mass? bolometric luminosity has decreased to half its maximumvalue (Contardo et al. 2000). For SN 2006gz they measure t +1 / = 18 days from the publicly available photometry,but for SN 2003fg they convert the stretch value publishedby Howell et al. (2006) to ∆ m and then to a t +1 / valueof 13.5 days (adopting an empirical linear fit to ∆ m and t +1 / values from Contardo et al. 2000). Using their conver-sions between ∆ m and t +1 / , we calculate t +1 / ≈
14 days(which, unsurprisingly, is nearly the same as the value calcu-lated for SN 2003fg by Maeda & Iwamoto 2009). However,when we measure t +1 / directly from our light curve, weget a minimum value of about 20.5 days (when we use ourmaximum plausible host-galaxy reddening), which is nearly50% larger! Furthermore, if one converts the ∆ m valueof SN 2006gz to t +1 / using the linear fit, one again finds t +1 / ≈
14 days, suggesting that the linear conversion de-rived by Maeda & Iwamoto (2009) is not accurate for suchlow values of ∆ m . This is not wholly unexpected since the∆ m values of SNe 2006gz, 2003fg, and 2009dc (or t +1 / as measured from light curves) are all below (or above) thevalues used to derive the linear fit (Contardo et al. 2000).In addition, the conversion from stretch to ∆ m used byMaeda & Iwamoto (2009) for SN 2003fg did not take intoaccount the fact that the definition of stretch has evolvedas new SN Ia light-curve templates have been constructed(e.g., Conley et al. 2008).We can now compare SN 2009dc to the analysis ofMaeda & Iwamoto (2009) using the actual values of t +1 / ≈ . . t +1 / value, as well as the extremely low photospheric veloc-ity of SN 2009dc, none of their models (using normal or SCWDs) appears to be viable. Their “normal WD” model with M WD = 1 .
39 M (cid:12) and M Ni = 0 . (cid:12) accounts for the lowvelocity with large t +1 / , but it underpredicts L bol by a fac-tor of a few. Some of the SC WD models ( M WD (cid:62) . (cid:12) and M Ni = 1 . (cid:12) ) of Maeda & Iwamoto (2009) can al-most account for the large values of L bol and t +1 / seen inSN 2009dc, but they then overpredict the photospheric ve-locity of SN 2009dc by a factor of 2 or so (with the modelshaving the highest mass WDs coming the closest to match-ing the observed velocities).While no single model of Maeda & Iwamoto (2009)clearly reproduces our observations of SN 2009dc, its prop-erties seem to be on the outskirts of the parameter spaceexplored by their analysis. This hints at the possibility thatmore extreme WD and Ni masses may be required tomatch the observational data of SN 2009dc. In addition, itis possible that multi-dimensional analyses are required totruly capture the underlying physics of a SC WD explosion.Furthermore, a few synthetic late-time spectra ofSC SNe Ia have been presented in the literature (e.g.,Maeda et al. 2009). Their models take as inputs a WDprogenitor mass, a WD central density, and mass frac-tions of burning products, and then use a one-dimensionalMonte Carlo radiation transport code, along with ioniza-tion/recombination equilibrium and rate equations, to cal-culate synthetic light curves and spectra. Both of theirSC SN Ia models roughly match our SN 2009dc photom-etry. In addition, their late-time spectra are similar to ourday 281 spectrum of SN 2009dc. Maeda et al. (2009) notethat as they increase their models’ progenitor mass, the[Fe II ]/[Ca II ] feature near 7200 ˚A becomes stronger rela- tive to the blends of [Fe II ] and [Fe III ] near 3800–5500 ˚A.This is seen in SN 2009dc (as compared to SNe Ia with morenormal peak luminosity), and it further supports our findingthat the progenitor of SN 2009dc was a SC WD. However,it should be noted that the SC SN Ia models of Maeda et al.(2009) only use WDs with masses of 2 and 3 M (cid:12) , and Nimasses of only 1 M (cid:12) , which is on the low end of our rangeof calculated values for the Ni yield of SN 2009dc.
A number of models which employ Chandrasekhar-massWDs as the progenitors of super-luminous SNe Ia (ones thatwe would consider possibly SC SNe Ia) have been proposed(Hillebrandt et al. 2007; Sim et al. 2007; Kasen et al. 2008,2009). These models usually invoke the off-centre ignitionof a normal WD progenitor which leads to nuclear burn-ing (and thus Ni production) that is peaked away fromthe centre of the WD. This off-centre nickel blob would in-crease the observed luminosity if the blob were offset fromthe centre of the WD toward the observer, with the maxi-mum effect occurring when it is offset directly along the lineof sight. These viewing angles also lead to the fastest light-curve rise times in such simulations (Sim et al. 2007; Kasenet al. 2008).Once again, we can compare our observed values forSN 2009dc to the models. Since the nickel blob is offset fromthe centre, asphericity is introduced into the explosion byconstruction (Kasen et al. 2008). If there is in fact a blobof Ni in SN 2009dc and it is offset from the progenitor’scentre directly along our line of sight (thus maximizing themeasured luminosity), then perhaps there is still azimuthalsymmetry in the explosion. This may account for the lowlevels of continuum polarization measured by Tanaka et al.(2010), as well as the higher levels of IME line polarization,but it seems tenuous at best.Observing along the axis of the offset blob of Ni, Hille-brandt et al. (2007) produce light curves that peak at a bolo-metric magnitude of M bol ≈ − . M B ≈ −
20 mag, andKasen et al. (2009) claim models with luminosities as highas 2 . × erg s − . These brightest magnitudes and lu-minosities that can be obtained by off-centre explosions arenearly equal to our lower limits for SN 2009dc (assuming nohost-galaxy reddening), and are likely below the true values.Furthermore, we note that the maximum amount of Niobtained from these explosions is 0.9–1.1 M (cid:12) (Hillebrandtet al. 2007; Kasen et al. 2009), which is once again at thelowest end of our range of calculated values for SN 2009dc.Finally, as mentioned above, the viewing angles thatmaximize the observed peak magnitudes also minimize therise time. We find a relatively long rise time of 23 ± must be >
21 days. This is significantly longerthan the rise times for these viewing angles as derived fromthe models ( ∼
12 days and ∼
18 days; Hillebrandt et al. 2007;Kasen et al. 2008, respectively). Thus, it seems that none ofthese models which include a Chandrasekhar-mass WD isviable for SN 2009dc.It should be mentioned, however, that these modelsmostly assume expansion velocities of “normal” SNe Ia (e.g.Sim et al. 2007) and we have shown that the expansion veloc- c (cid:13) , 1–30 Silverman, et al. ity of SN 2009dc is significantly lower. This is evident in thesynthetic spectra derived from these models; from all view-ing angles, they resemble early-time normal SN Ia spectramuch more than the spectra of SN 2009dc near maximumbrightness (Kasen et al. 2008). Thus, perhaps it is not toosurprising that these specific examples of non-SC models donot match the observations of SN 2009dc.
In this paper we have presented and analysed optical pho-tometry and spectra of SN 2009dc and SN 2007if, both ofwhich are possibly SC SNe Ia. Our photometric and spectraldata on SN 2009dc constitute one of the richest datasets everpublished on a SC SN Ia candidate. Our well-sampled lightcurve follows SN 2009dc from about 1 week before maximumbrightness until about 5 months past maximum, and showsthat SN 2009dc is one of the slowest photometrically evolv-ing SNe Ia ever observed. We derive a rise time of 23 daysand ∆ m ( B ) = 0 .
72 mag, which are two of the most ex-treme values for these parameters ever seen in a SN Ia. As-suming no host-galaxy reddening, we derive a peak bolo-metric luminosity of about 2 . × erg s − , though this isalmost certainly an underestimate since we observe strongevidence for at lease some host reddening. Using our nonzerovalues for E ( B − V ) host , the peak bolometric luminosity in-creases by about 40%–200%.Spectroscopically, SN 2009dc also evolves relativelyslowly. Strong C II absorption features (which are rarelyobserved in SNe Ia) are seen in the spectra near maximumbrightness, implying a significant amount of unburned fuelfrom the progenitor WD in the outer layers of the SN ejecta.Si II absorption also appears in our spectra of SN 2009dcand remains visible even 2 months past maximum. Ourpost-maximum spectra are dominated by a forest of IGEfeatures and, interestingly, resemble spectra of the peculiarSN Ia 2002cx. Finally, the spectra of SN 2009dc all show verylow expansion velocities at all layers (i.e., unburned carbon,IMEs, and IGEs) as compared to other SNe Ia. This may beexplained by a massive WD progenitor which consequentlyhas a large binding energy. Even though the expansion ve-locities are small, we see no strong evidence in SN 2009dc fora velocity “plateau” near maximum light like the one seenin SN 2007if (Scalzo et al. 2010).Using various luminosity and energy arguments, we cal-culate that the progenitor of SN 2009dc is possibly a SC WDwith a mass greater than ∼ (cid:12) , and that at least ∼ (cid:12) of Ni was likely formed in the explosion (though the mostprobable value is in the range 1.4–1.7 M (cid:12) ). These valuesare larger than (or about as large as) those calculated forany other SN Ia ever observed. We propose that the hostgalaxy of SN 2009dc underwent a gravitational interactionwith a nearby galaxy (UGC 10063) in the relatively recentpast, and that this could have induced a sudden burst ofstar formation which may have given rise to the progeni-tor of SN 2009dc and turned UGC 10063 into the “post-starburst” galaxy that we observe today. We also compareour observed quantities for SN 2009dc to theoretical models,and while no model seems to match or explain every aspectof SN 2009dc, simulations show that SC WDs with massesnear what we calculate for the progenitor of SN 2009dc can possibly form, likely from the merger of two WDs. Further-more, models of extremely luminous SNe Ia which employa Chandrasekhar-mass WD progenitor cannot explain ourobservations of SN 2009dc.Thus, taking all of these extreme values into account,we conclude that SN 2009dc is very likely a SC SN Ia. Asmentioned previously, many of the observed peculiarities ofSN 2009dc are also seen in SN 2003fg and SN 2007if. There-fore, we concur with Howell et al. (2006) and Scalzo et al.(2010) that both SN 2003fg and SN 2007if (respectively) arealso probably SC SNe Ia. However, given their fairly normalexpansion velocities and relative weakness (or even absence)of C II features near maximum brightness, it seems thatSN 2006gz and SN 2004gu are less likely to be SC SNe Ia.New large transient searches such as Pan-STARRS(Kaiser et al. 2002) and the Palomar Transient Factory (Rauet al. 2009; Law et al. 2009) will probably find many SC orother super-luminous SNe Ia in the near future. Since itseems that they cannot be standardized in the same way asmost SNe Ia, they will need to be handled separately or ig-nored in cosmological surveys which will use large numbersof SNe Ia. However, the simulations of Chen & Li (2009)show that donor stars with lower metallicities (e.g., Popula-tion II stars) are less likely to form WDs with masses greaterthan 1.7 M (cid:12) than higher metallicity stars. Thus, it is pos-sible that contamination levels from SC SNe Ia, which arealready rare at low redshifts (i.e., average metallicity), maybe relatively small in medium or high-redshift surveys. ACKNOWLEDGMENTS
We thank P. F. Hopkins, K. M. Sandstrom, andK. L. Shapiro for useful discussions regarding host galax-ies. We are especially grateful to D. A. Howell for thenear-maximum spectrum of SN 2003fg (and, as the ref-eree, for many useful comments), K. Maeda for the late-time spectrum of SN 2006gz, R. A. Scalzo and P. E. Nu-gent for information regarding SN 2007if, and M. Yamanakafor early-time comparison photometry of SN 2009dc. Wealso thank I. Arcavi, S. B. Cenko, J. Choi, B. E. Cobb,R. J. Foley, C. V. Griffith, M. T. Kandrashoff, M. Kislak,I. K. W. Kleiser, J. Leja, M. Modjaz, A. J. L. Morton,J. Rex, T. N. Steele, P. Thrasher, and X. Wang for theirassistance with some of the observations and data reduc-tion, as well as J. Kong, N. Lee, and E. Miller for theirhelp in improving and maintaining the SNDB. We are grate-ful to the staffs at the Lick and Keck Observatories fortheir support. Some of the data presented herein were ob-tained at the W. M. Keck Observatory, which is oper-ated as a scientific partnership among the California Insti-tute of Technology, the University of California, and theNational Aeronautics and Space Administration (NASA);the observatory was made possible by the generous finan-cial support of the W. M. Keck Foundation. The authorswish to recognize and acknowledge the very significant cul-tural role and reverence that the summit of Mauna Keahas always had within the indigenous Hawaiian commu-nity; we are most fortunate to have the opportunity toconduct observations from this mountain. This researchhas made use of the NASA/IPAC Extragalactic Database(NED) which is operated by the Jet Propulsion Labora- c (cid:13) , 1–30 N 2009dc, Super-Chandrasekhar Mass? tory, California Institute of Technology, under contract withNASA. A.V.F.’s group is supported by the NSF grant AST–0908886, DOE grants DE–FC02–06ER41453 (SciDAC) andDE–FG02–08ER41563, NASA/ Swift grant NNX09AL08G,and the TABASGO Foundation. KAIT and its ongoing op-eration were made possible by donations from Sun Microsys-tems, Inc., the Hewlett-Packard Company, AutoScope Cor-poration, Lick Observatory, the NSF, the University of Cal-ifornia, the Sylvia & Jim Katzman Foundation, and theTABASGO Foundation. J.M.S. is grateful to Marc J. Staleyfor a Graduate Fellowship.
REFERENCES
Abazajian K. N., et al., 2009, ApJS, 182, 543Amanullah R., et al., 2010, ApJ, 716, 712Arnett W. D., 1982, ApJ, 253, 785Axelrod T. S., 1988, in K. Nomoto ed., Atmospheric Di-agnostics of Stellar Evolution. Lecture Notes in Physics.p. 375Barbon R., Benetti S., Rosino L., Cappellaro E., TurattoM., 1990, A&A, 237, 79Barnes J. E., Hernquist L. E., 1991, ApJ, 370, L65Barnes J. E., Hernquist L. E., 1996, ApJ, 471, 115Benetti S., et al., 2005, ApJ, 623, 1011Branch D., 1992, ApJ, 392, 35Branch D., Baron E., Thomas R. C., Kasen D., Li W.,Filippenko A. V., 2004, PASP, 116, 903Branch D., et al., 2003, AJ, 126, 1489Cardelli J. A., Clayton G. C., Mathis J. S., 1989, ApJ, 345,245Chen W., Li X., 2009, ApJ, 702, 686Chornock R., Filippenko A. V., 2008, AJ, 136, 2227Chornock R., Filippenko A. V., Branch D., Foley R. J., JhaS., Li W., 2006, PASP, 118, 722Clocchiatti A., Wheeler J. C., 1997, ApJ, 491, 375Conley A., et al., 2008, ApJ, 681, 482Contardo G., Leibundgut B., Vacca W. D., 2000, A&A,359, 876Contreras C., et al., 2010, AJ, 139, 519de Vaucouleurs G., et al., 1991, Third Reference Catalogueof Bright Galaxies. Springer, New YorkFaber S. M., et al., 2003, in Iye M., Moorwood A. F. M.,eds, Proceedings of the Society of Photo-Optical Instru-mentation Engineers (SPIE) Conference. p. 1657Falc´on-Barroso J., et al., 2004, MNRAS, 350, 35Filippenko A. V., 1982, PASP, 94, 715Filippenko A. V., 1997, ARA&A, 35, 309Filippenko A. V., 2003, in Hillebrandt W., Leibundgut B.,eds, From Twilight to Highlight The Physics of Super-novae. Springer-Verlag, Berlin, p. 171Filippenko A. V., et al., 1992, ApJ, 384, L15Filippenko A. V., Li W. D., Treffers R. R., Modjaz M.,2001, in Paczy´nski B., Chen W. P., Lemme C., eds, Small-Telescope Astronomy on Global Scales. Astron. Soc. Pac.,San Francisco, p. 121Fisher A., Branch D., Nugent P., Baron E., 1997, ApJ, 481,L89Foley R. J., et al., 2003, PASP, 115, 1220Foley R. J., et al., 2010, ApJ, 708, L61Ganeshalingam M., et al., 2010a, submitted Ganeshalingam M., et al., 2010b, in prep.G´omez G., L´opez R., 1998, AJ, 115, 1096Greggio L., 2010, MNRAS, p. 906Hachinger S., Mazzali P. A., Benetti S., 2006, MNRAS,370, 299Hamuy M., Phillips M. M., Suntzeff N. B., Schommer R. A.,Maza J., Smith R. C., Lira P., Aviles R., 1996, AJ, 112,2438Harutyunyan A., Elias-Rosa N., Benetti S., 2009, CBET,1768, 1Hayden B. T., et al., 2010, ApJ, 712, 350Hicken M., et al., 2009, ApJ, 700, 1097Hicken M., Garnavich P. M., Prieto J. L., Blondin S., De-Poy D. L., Kirshner R. P., Parrent J., 2007, ApJ, 669,L17Hillebrandt W., Sim S. A., R¨opke F. K., 2007, A&A, 465,L17Horne K., 1986, PASP, 98, 609Howell D. A., et al., 2006, Nature, 443, 308Howell D. A., et al., 2009, ApJ, 691, 661Iben Jr. I., Tutukov A. V., 1984, ApJS, 54, 335Jeong H., Bureau M., Yi S. K., Krajnovi´c D., Davies R. L.,2007, MNRAS, 376, 1021Jester S., et al., 2005, AJ, 130, 873Jha S., Branch D., Chornock R., Foley R. J., Li W., SwiftB. J., Casebeer D., Filippenko A. V., 2006, AJ, 132, 189Jha S., Riess A. G., Kirshner R. P., 2007, ApJ, 659, 122Kaiser N., et al., 2002, in Tyson J. A., Wolff S., eds, Pro-ceedings of the Society of Photo-Optical InstrumentationEngineers (SPIE) Conference. p. 154Kasen D., 2006, ApJ, 649, 939Kasen D., R¨opke F. K., Woosley S. E., 2009, Nature, 460,869Kasen D., Thomas R. C., R¨opke F., Woosley S. E., 2008,Journal of Physics Conference Series, 125, 012007Kessler R., et al., 2009, ApJS, 185, 32Khokhlov A., Mueller E., H¨oflich P., 1993, A&A, 270, 223Kowalski M., et al., 2008, ApJ, 686, 749Kuchner M. J., Kirshner R. P., Pinto P. A., Leibundgut B.,1994, ApJ, 426, L89Landolt A. U., 1992, AJ, 104, 340Law N. M., et al., 2009, PASP, 121, 1395Leibundgut B., 2000, A&A Rev., 10, 179Leonard D. C., Li W., Filippenko A. V., Foley R. J.,Chornock R., 2005, ApJ, 632, 450Li W., et al., 2003, PASP, 115, 453Li W., et al., 2010, in prep.Li W., Filippenko A. V., Chornock R., Jha S., 2003, PASP,115, 844Li W., Jha S., Filippenko A. V., Bloom J. S., Pooley D.,Foley R. J., Perley D. A., 2006, PASP, 118, 37Li W. D., et al., 2000, in Holt S. S., Zhang W. W., eds,Cosmic Explosions. American Institute of Physics, NewYork, p. 103Lira P., et al., 1998, AJ, 115, 234Lira P., 1996, Master’s thesis, Univ. ChileMaeda K., Iwamoto K., 2009, MNRAS, 394, 239Maeda K., Kawabata K., Li W., Tanaka M., Mazzali P. A.,Hattori T., Nomoto K., Filippenko A. V., 2009, ApJ, 690,1745Maeda K., Mazzali P. A., Deng J., Nomoto K., Yoshii Y.,Tomita H., Kobayashi Y., 2003, ApJ, 593, 931 c (cid:13) , 1–30 Silverman, et al.
Marion G. H., H¨oflich P., Wheeler J. C., Robinson E. L.,Gerardy C. L., Vacca W. D., 2006, ApJ, 645, 1392Marion H., Garnavich P., Challis P., Calkins M., Peters W.,2009, CBET, 1776, 1Matheson T., Filippenko A. V., Ho L. C., Barth A. J.,Leonard D. C., 2000, AJ, 120, 1499Mazzali P. A., Cappellaro E., Danziger I. J., Turatto M.,Benetti S., 1998, ApJ, 499, L49Mazzali P. A., Chugai N., Turatto M., Lucy L. B., DanzigerI. J., Cappellaro E., Della Valle M., Benetti S., 1997, MN-RAS, 284, 151Mazzali P. A., Danziger I. J., Turatto M., 1995, A&A, 297,509Mihos J. C., Hernquist L., 1994, ApJ, 431, L9Miller J. S., Stone R. P. S., 1993, Lick Obs. Tech. Rep. 66.Santa Cruz: Lick Obs.Monard L. A. G., Quimby R., Gerardy C., H¨oflich P.,Wheeler J. C., Chen Y., Smith H. J., Bauer A., 2004,IAU Circ., 8454, 1Morganti R., et al., 2006, MNRAS, 371, 157Munari U., Zwitter T., 1997, A&A, 318, 269Nobili S., et al., 2005, A&A, 437, 789Nomoto K., Thielemann F., Yokoi K., 1984, ApJ, 286, 644Nugent P., Phillips M., Baron E., Branch D., HauschildtP., 1995, ApJ, 455, L147Oke J. B., et al., 1995, PASP, 107, 375Patat F., Benetti S., Cappellaro E., Danziger I. J., DellaValle M., Mazzali P. A., Turatto M., 1996, MNRAS, 278,111Perlmutter S., et al., 1999, ApJ, 517, 565Petrosian V., 1976, ApJ, 209, L1Pfannes J. M. M., Niemeyer J. C., Schmidt W., 2010, A&A,509, 75Phillips M. M., 1993, ApJ, 413, L105Phillips M. M., et al., 2007, PASP, 119, 360Phillips M. M., Lira P., Suntzeff N. B., Schommer R. A.,Hamuy M., Maza J., 1999, AJ, 118, 1766Piro A. L., 2008, ApJ, 679, 616Poggianti B. M., et al., 2009, ApJ, 693, 112Poole T. S., et al., 2008, MNRAS, 383, 627Puckett T., Moore R., Newton J., Orff T., 2009, CBET,1762, 1Rau A., et al., 2009, PASP, 121, 1334Riess A. G., et al., 1998, AJ, 116, 1009Riess A. G., et al., 1999, AJ, 118, 2675Ruiz-Lapuente P., Kirshner R. P., Phillips M. M., ChallisP. M., Schmidt B. P., Filippenko A. V., Wheeler J. C.,1995, ApJ, 439, 60Scalzo R. A., et al., 2010, ApJ, 713, 1073Schinnerer E., Scoville N., 2002, ApJ, 577, L103Schlegel D. J., Finkbeiner D. P., Davis M., 1998, ApJ, 500,525Schneider S. E., et al., 1990, ApJS, 72, 245Shapiro K. L., et al., 2010, MNRAS, p. 92Silverman J. M., et al., 2010, in prep.Sim S. A., Sauer D. N., R¨opke F. K., Hillebrandt W., 2007,MNRAS, 378, 2Simon J. D., et al., 2009, ApJ, 702, 1157Spergel D. N., et al., 2007, ApJS, 170, 377Stanishev V., et al., 2007, A&A, 469, 645Stritzinger M., Leibundgut B., 2005, A&A, 431, 423Tanaka M., et al., 2008, ApJ, 677, 448 Tanaka M., et al., 2010, ApJ, 714, 1209Thomas R. C., et al., 2007, ApJ, 654, L53Turatto M., Benetti S., Cappellaro E., 2003, in HillebrandtW., Leibundgut B., eds, From Twilight to Highlight: ThePhysics of Supernovae. Springer, Berlin, p. 200Valenti S., et al., 2009, Nature, 459, 674Wade R. A., Horne K., 1988, ApJ, 324, 411Wang X., et al., 2009a, ApJ, 699, L139Wang X., et al., 2009b, ApJ, 697, 380Webbink R. F., 1984, ApJ, 277, 355Wegner G., Grogin N. A., 2008, AJ, 136, 1Weijmans A., Krajnovi´c D., van de Ven G., OosterlooT. A., Morganti R., de Zeeuw P. T., 2008, MNRAS, 383,1343Whelan J., Iben Jr. I., 1973, ApJ, 186, 1007Yamanaka M., et al., 2009, ApJ, 707, L118Yoon S., Langer N., 2005, A&A, 435, 967Yoon S., Podsiadlowski P., Rosswog S., 2007, MNRAS, 380,933 c (cid:13)000