The Exceptionally Luminous Type II-L SN 2008es
A. A. Miller, R. Chornock, D. A. Perley, M. Ganeshalingam, W. Li, N. R. Butler, J. S. Bloom, N. Smith, M. Modjaz, D. Poznanski, A. V. Filippenko, C. V. Griffith, J. H. Shiode, J. M. Silverman
aa r X i v : . [ a s t r o - ph ] S e p Draft version August 15, 2018
Preprint typeset using L A TEX style emulateapj v. 08/13/06
THE EXCEPTIONALLY LUMINOUS TYPE II-LINEAR SUPERNOVA 2008ES
A. A. Miller , R. Chornock , D. A. Perley , M. Ganeshalingam , W. Li , N. R. Butler , J. S. Bloom , N.Smith , M. Modjaz , D. Poznanski , A. V. Filippenko , C. V. Griffith , J. H. Shiode , and J. M. Silverman Draft version August 15, 2018
ABSTRACTWe report on our early photometric and spectroscopic observations of the extremely luminous TypeII supernova (SN) 2008es. With an observed peak optical magnitude of m V = 17 . z = 0 . M V = − .
3, making it the second mostluminous SN ever observed. The photometric evolution of SN 2008es exhibits a fast decline rate( ∼ − ), similar to the extremely luminous Type II-L SN 2005ap. We show that SN 2008esspectroscopically resembles the luminous Type II-L SN 1979C. Although the spectra of SN 2008eslack the narrow and intermediate-width line emission typically associated with the interaction of aSN with the circumstellar medium of its progenitor star, we argue that the extreme luminosity ofSN 2008es is powered via strong interaction with a dense, optically thick circumstellar medium. Theintegrated bolometric luminosity of SN 2008es yields a total radiated energy at ultraviolet and opticalwavelengths of & ergs. Finally, we examine the apparently anomalous rate at which the TexasSupernova Search has discovered rare kinds of supernovae, including the five most luminous supernovaeobserved to date, and find that their results are consistent with those of other modern SN searches. Subject headings: supernovae: general — supernovae: individual (SN 2008es) INTRODUCTION
Wide-field synoptic optical imaging surveys are con-tinuing to probe the parameter space of time-variablephenomena with increasing depth and temporal cover-age (e.g., Becker et al. 2004; Morales-Rueda et al. 2006;Bramich et al. 2008), unveiling a variety of transientsranging from the common (Rau et al. 2008) to the unex-plained (e.g., Barbary et al. 2008). Untargeted (“blind”)synoptic wide-field imaging surveys, such as the TexasSupernova Search (TSS; Quimby 2006) conducted withthe ROTSE-III 0.45-m telescope (Akerlof et al. 2003),have uncovered a large number of rare transients, includ-ing the four most luminous supernovae (SNe) observedto date: SN 2005ap (Quimby et al. 2007), SN 2008am(Yuan et al. 2008a), SN 2006gy (Ofek et al. 2007; Smithet al. 2007, 2008b), and SN 2006tf (Smith et al. 2008a).Observations of these very luminous events are startingto allow the detailed physical study of the extrema incore-collapse SNe. They appear to be powered in partby their interaction with a highly dense circumstellarmedium (CSM; see Smith et al. 2008a, and referencestherein), though other possibilities have been advanced.Clearly, the discovery of more such events would allowan exploration of the variety of the phenomenology asrelated to the diversity of progenitors and CSM.Recently, the TSS discovered yet another luminoustransient on 2008 Apr. 26.23 (UT dates are usedthroughout this paper), which they suggested was a vari-able active galactic nucleus at a redshift z = 1 .
02 (Yuanet al. 2008b). Gezari & Halpern (2008) then hypoth-esized that the transient was a flare from the tidal dis-ruption of a star by a supermassive black hole. Milleret al. (2008) first identified SN 2008es as potentially Department of Astronomy, University of California, Berkeley,CA 94720-3411. Email: [email protected] GLAST Fellow. Sloan Research Fellow. an extremely luminous Type II SN (see also Gezariet al. 2008a), and we later definitively confirmed thiswith further spectroscopic observations (Chornock et al.2008a); the event was assigned the name SN 2008es bythe IAU (Chornock et al. 2008b). It is located at α =11 h m s , δ = +54 o ′ ′′ (J2000.0).Here we present our analysis of SN 2008es, whichis classified as a Type II-Linear (II-L) SN based onthe observed linear (in mag) decline in the photomet-ric light curve (Barbon, Ciatti, & Rosino 1979; Doggett& Branch 1985). At z = 0.213, SN 2008es has a peakoptical magnitude of M V = − .
3, among SNe secondonly to SN 2005ap. Aside from the extreme luminos-ity, SN 2008es is of great interest since detailed ultra-violet (UV) through infrared (IR) observations providea unique opportunity to study the mass-loss propertiesof an evolved post-main sequence massive star via itsinteraction with the surrounding dense CSM. A similaranalysis of SN 2008es has been presented by Gezari etal. (2008b).The outline of this paper is as follows. We presentour observations in §
2, and the photometric and spectro-scopic analyses of this and public (NASA) data in § §
4, respectively. A discussion is given in §
5, and our con-clusions are summarized in §
6. Throughout this paperwe adopt a concordance cosmology of H = 70 km s − Mpc − , Ω M = 0 .
3, and Ω Λ = 0 . OBSERVATIONS
Here we present our ground-based optical and near-infrared (NIR) photometry and optical spectroscopy,along with space-based
Swift
UV, optical, and X-ray ob-servations. NIR photometry of SN 2008es was obtainedsimultaneously in J , H , and K s with the Peters Auto-mated Infrared Imaging Telescope (PAIRITEL; Bloomet al. 2006) beginning 2008 May 16. To improve thephotometric signal-to-noise ratio (S/N), we stacked im-ages over multiple nights. For the K s images the S/N Miller et al. TABLE 1PAIRITEL Observations of SN 2008es t mida Obs. window b Filter Exp. time Mag c (day) (day) (sec)4602.24 2.06 J 2895.91 17.74 ± ± ± ± ± ± ± ± ± ± ± ± Note . — PAIRITEL observations were stacked overmultiple epochs to increase the S/N. a Mid-point between the first and last exposures in a singlestacked image, reported as JD − b Time between the first and last exposures in a singlestacked image. c Observed value; not corrected for Galactic extinction. of the SN remained low, despite the stacks made overmultiple epochs, and therefore we do not include thesedata in our subsequent analysis. Aperture photometry,using a custom pipeline, was used to measure the J and H photometry of the isolated SN calibrated to the 2MASScatalog (see Bloom et al. 2008). The resulting lightcurves are presented in Figure 1. The final PAIRITELphotometry is reported in Table 1.Optical photometry of SN 2008es was obtained in BVRI with the Katzman Automatic Imaging Telescope(KAIT; Filippenko et al. 2001) and the 1-m Nickeltelescope at Lick Observatory beginning 2008 May 30.Point-spread function (PSF)-fitting photometry was per-formed on the SN and several comparison stars using the
IRAF/DAOPHOT package (Stetson 1987) and transformedinto the Johnson-Cousins system. Calibrations for thefield were obtained with the Nickel telescope on threephotometric nights. The final photometry from KAITand the Nickel telescope is given in Tables 2 and 3, re-spectively.Additional optical photometry was obtained in
U BVRI on 2008 June 02 and
BVRI on 2008 Aug. 05with PFCam on the 3-m Shane telescope at Lick Ob-servatory. The data were reduced using standard tech-niques and aperture photometry was used to extract theSN flux.
BVRI calibrations were done using the samestars as those used for the Nickel and KAIT observations.Calibrations for the U band were not obtained with theNickel telescope. Therefore, we convert the Sloan Digi-tal Sky Survey (SDSS) colors of stars in the field to the U band with the color transformations of Jester et al.(2005), and use these stars for our U -band calibration.The final PFCam photometry is reported in Table 4.The Swift satellite observed SN 2008es during 13epochs between 2008 May 14 and Aug. 21. Wedownloaded the Ultraviolet/Optical Telescope (UVOT;Roming et al. 2005) data from the
Swift data archiveand analyzed the Level 2 sky image data in U , B , and V according to the photometry calibration and recipe by Liet al. (2006). The Swift
UV filters (
U V W U V M
1, and
U V W
2) were reduced following Poole et al. (2008). The
TABLE 2KAIT Observations of SN 2008es t obsa Filter Exp. time Mag b (day) (sec)4620.72 B 360.00 18.36 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± a Exposure mid-point, reported asJD − b Observed value; not corrected for Galac-tic extinction. final
Swift
UVOT photometry is reported in Table 5.Simultaneous observations of SN 2008es occurred withthe
Swift
X-ray Telescope (XRT; Burrows et al. 2005),for which we confirm a nondetection of X-ray emission(see also Gezari & Halpern 2008). To place a limit onthe source flux we assume a power-law spectrum witha photon index Γ = 2, absorbed by elements associatedwith Galactic H I . We took an extraction region of ra-dius 64 pixels ( ∼ ′ ) and fit the PSF model around thecentroid of the optical emission. We obtain a 3 σ limitingflux of 9 × − erg cm − s − in the 0.3–10 keV bandfor a total exposure of 54.7 ks. This represents an upperlimit to the X-ray luminosity of SN 2008es of ∼ . × ergs s − .We obtained spectra of SN 2008es on 2008 May 16.3,2008 May 29.3, and 2008 July 7.3 using the Kast spectro-uminous Type II-L SN 2008es 3 Time (rest-frame days) m a g UVO T PF C a m N i c k e l KA I T P A I R I TEL R O T S E -III b UVW2UVM2 - 1.0UVW1- 2.2U - 3.0B - 5.0V - 6.0R - 7.0I - 7.8unfiltered - 7.0J - 8.5H - 9.0
Fig. 1.—
Observed UV-optical-NIR light curves of SN 2008es.The data have not been corrected for Galactic or host-galaxy ex-tinction. We include ROTSE-IIIb unfiltered observations from theliterature (open symbols; Gezari et al. 2008b), as well as ouroptical-NIR observations (filled symbols; KAIT, Nickel, PFCam,and PAIRITEL) and space-based UVOT observations from
Swift (filled triangles). We adopt the discovery date of SN 2008es, 2008Apr. 26.23, as “day 0” for this SN. Low-order polynomial fits toeach band have been overplotted to help guide the eye. graph on the Lick 3-m telescope (Miller & Stone 1993).Additional spectra were obtained on 2008 June 7.4 and2008 Aug. 3.3 using the Low Resolution Imaging Spec-trometer on the Keck I 10-m telescope (Oke et al. 1995)and on 2008 June 21.2 and 2008 June 23.2 using the R.C. Spectrograph on the Kitt Peak 4-m telescope, follow-ing the approval of our request for Kitt Peak Director’sDiscretionary Time. The spectra were extracted and cal-ibrated following standard procedures (e.g., Matheson etal. 2000). Clouds were present during several of theobservations, making the absolute flux scales unreliable.The spectrograph slit was placed at the parallactic an-gle, so the relative spectral shapes should be accurate(Filippenko 1982), with the exception of the Kitt Peakspectra, which have a small amount of second-order lightcontamination at wavelengths longward of ∼ TABLE 3Nickel Observations of SN 2008es t obsa Filter Exp. time Mag b (day) (sec)4627.75 B 360.00 18.56 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± a Exposure mid-point, reported asJD − b Observed value; not corrected for Galac-tic extinction.
TABLE 4PFCam Observations of SN 2008es t obsa Filter Exp. time Mag b (day) (sec)4619.81 U 1230.00 17.73 ± ± ± ± ± ± ± ± ± a Exposure mid-point, reported as JD-2450000. b Observed value; not corrected for Galac-tic extinction.
The two Kitt Peak spectra show little evolution in thetwo days that separate them and have been combinedto increase the S/N (day 33 in Figure 2). A log of ourspectroscopic observations is presented in Table 6.We searched our spectra for the possible presence ofnarrow lines and were unable to positively identify any,either in emission or absorption. Therefore, without adetection of the host galaxy, we determine the redshiftof SN 2008es directly from the SN spectrum. As the SN Miller et al.
TABLE 5UVOT Observations of SN 2008es t obsa Filter Exp. time Mag b (day) (sec)4600.75 UVW2 1585.70 17.02 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± a Exposure mid-point, reported asJD − b Observed value; not corrected forGalactic extinction. L og (f λ × c on s t a n t ) SN 2005ap(−3 d)
SN 2008es
Ages relativeto maximum lightin V band (in rest−frame days)+3+13+21+33+45+68
Fig. 2.—
Spectral evolution of SN 2008es. Spectra of SN 2008es(in color) are labeled with their ages in the rest frame of the su-pernova ( z = 0 . V -band maximum onMay 13.3. The spectra become progressively redder as the SN agesand broad P-Cygni spectral features become more prominent withtime. By +68 d, a broad emission feature of H α is clearly present.The top spectrum (in black) is the earliest spectrum of SN 2005ap,the most luminous observed SN (Quimby et al. 2007). aged, broad P-Cygni spectral features appeared, includ-ing an emission feature near 7900 ˚A that we identify asH α near a redshift of 0.2. Other spectral features areconsistent with a Type II SN at about that redshift.To get a more accurate redshift, we identified two sim-ilar reference spectra of the Type II-L SNe 1979C and1980K from the literature (Branch et al. 1981; Uomoto& Kirshner 1986). These spectral comparisons are shownin the bottom panel of Figure 3. We used the Super-Nova IDentification code of Blondin & Tonry (2007) tocross-correlate the day 68 spectrum of SN 2008es withthe two reference spectra and derived a weighted-average z = 0 . ± . α emissionline in the day 68 spectrum, which yields a flux centroidof z = 0 . α emission lines of a similarwidth at late times. PHOTOMETRIC RESULTS
Our dataset provides excellent broad-band coverage ofSN 2008es from the UV to the NIR, which allows us touminous Type II-L SN 2008es 5
TABLE 6Log of Spectroscopic Observations
Age a UT Date Instrument b Range Exp. Time Seeing Airmass Photometric?(days) (˚A) (s) ( ′′ ) (y/n)3 2008-05-16.345 Kast 3300–8830 1500 2.4 1.3 y13 2008-05-29.253 Kast 3300–8820 1800 2.3 1.1 n21 2008-06-07.399 LRIS 3200–9230 1200 1.6 2.0 y33 2008-06-21.194 RC 4200–9280 1800 2.9 1.3 y34 2008-06-23.189 RC 4200–9280 2400 1.8 1.3 y45 2008-07-07.252 Kast 3300–9280 4200 2.5 1.6 y68 2008-08-03.254 LRIS 3500–9190 877 0.9 2.0 n a Age in rest-frame days relative to the observed V -band maximum on 2008 May 13.3. b Kast = Kast spectrograph on Lick 3-m telescope. LRIS = Low Resolution Imaging Spectrom-eter on Keck-I 10-m telescope. RC = R. C. Spectrograph on Kitt Peak 4-m telescope. F l ux (f λ ) + c on s t a n t SN 2008es(+21 d)
SN 1979C(+13 d) ⊕ H α He IH β F l ux (f λ ) + c on s t a n t SN 2008es(+68 d)
SN 1979C(+43 d) SN 1980K(+19 d) ⊕⊕ H γ H β Fe II He I/Na I H α Fig. 3.—
Spectral comparisons of SN 2008es to two SNe II-L. Allspectra are labeled with their respective ages in rest-frame days rel-ative to maximum light, and prominent spectral features are iden-tified. The zero point of the flux scale is accurate for the SN 2008esspectra (in blue), while the comparison spectra (in black) are ver-tically offset. The top panel shows a comparison at an early epochof SN 2008es to SN 1979C from 1979 Apr. 28 (Branch et al. 1981).The bottom panel shows a comparison at a later epoch to SN 1979Cfrom 1979 May 28 (Branch et al. 1981) and SN 1980K (spectrafrom 1980 Nov. 15 and 17 combined; Uomoto & Kirshner 1986).Telluric absorption bands in the SNe 1979C and 1980K spectra aremarked with a ⊕ symbol. model changes in the spectral energy distribution (SED).In order to sample each photometric band onto a singleset of common epochs, we fit low-order polynomials toeach light curve, which we then interpolate onto a com-mon grid. NIR observations were only included on or T ( K ) T BB R BB R a d i u s ( c m ) l og L B o l ( e r g s - ) Blackbody Fit -18-19-20-21-22-23-24-25 A b s o l u t e M B o l ( m a g ) Direct Integration
Fig. 4.—
Photospheric and bolometric luminosity evolution ofSN 2008es. Top: temperature evolution of SN 2008es based on BBfits (open circles) and inferred radius (open triangles). The errorbars on the radius are likely to be slightly overestimated becausethe errors on the BB temperature and luminosity are correlated.Bottom: bolometric luminosity of SN 2008es derived via two in-dependent methods, BB modeling (closed squares) and integrationof the total UV+optical (+NIR where available) flux (open dia-monds). The two methods agree to within . around epochs where we detected the SN.We create an SED at each of the common epochs andfit a single-component blackbody (BB) to the data fol-lowing the procedure described by Modjaz et al. (2008).Prior to the SED/BB fits we add a systematic term tothe uncertainty in the photometric measurement in eachband. This term is added in quadrature to the statisticaluncertainty, and for the NIR ( JH ) is equal to 2%, in theoptical ( BVRI ) we adopt 3%, while for U band we adopt10%, and the adopted UV ( U V W U V M U V W
2) sys-tematic uncertainty is 20%. Across all epochs of the BBfits the total χ per degree of freedom is 45 . /
65, whichsuggests that our systematic terms may be slightly over-estimated. We assume no host-galaxy extinction (for fur-ther details see § E ( B − V ) = 0.011mag (Schlegel et al. 1998). From the SED/BB fits wederive the temperature, radius, and luminosity of the SNas a function of time, as shown in Figure 4. The temper-ature and luminosity decrease while the radius increaseswith time, as expected for an expanding and cooling SNphotosphere. Miller et al. λ eff (Å)23222120191817 M a gn it ud e ( A B ) λ eff,rest (Å) 10100 F ν ( µ J y ) eff = 11359 ± 318 K χ /dof = 5.13 / 7H J I R V B U UVW1 UVW2UVM265.4 daysT eff = 6900 ± 218 K χ /dof = 2.81 / 4 I R V B U UVW1 UVW2UVM2 Fig. 5.—
SEDs from SN 2008es corrected for Galactic extinction.We show two representative SEDs of SN 2008es at ∼
27 and ∼ UV W
UV M
In order to determine the time of V -band maximumlight, which at the redshift of this SN roughly corre-sponds to rest-frame B , we convert the early time g , r ,and i ′ photometry taken with the Palomar 1.5-m tele-scope from Gezari et al. (2008b) to V RI using thecolor equations from Jester et al. (2005). Following aquadratic fit to these data, we find that the observed V -band maximum occurred on ∼ ∼
15 rest-frame days after discovery. The observed peak is V =17.8 mag, which corresponds to M V = − . Swift filters,
U V W
U V W
U V M
2. Thisexcess was also observed in the Type II-Plateau (II-P)SN 2006bp, where it was attributed to complex line blan-keting by Fe-peak elements (Immler et al. 2007). Theblue excess is readily identified in the UV color curves(e.g.,
U V W U V W
1) of SN 2008es, which evolve to-ward the red until ∼
50 rest-frame days after discovery,at which point the
U V W U V W
U V M U V W . − , which is slightly faster than the V -band(roughly rest-frame B ) rate of 0.036 mag d − . This rateis also slightly slower than the rest-frame B -band de-cay of both SN 1979C (0.046 mag d − ; Panagia et al. 1980) and SN 1980K (0.055 mag d − ; Barbon, Ciatti, &Rosino 1982). Furthermore, integrating the bolometriclight curve from 15 to 83 rest-frame days after discoveryyields a total radiated energy of ∼ × ergs, compa-rable to the canonical 10 ergs deposited into the kineticenergy of a SN. If we assume the same bolometric correc-tion factor to the observed V -band light curve both pre-and post-maximum, and we include the early data fromGezari et al. (2008b), we find the total radiated energyof SN 2008es over the first 83 d after discovery is ∼ × ergs.The three Swift
UVOT observations taken more than80 d after the discovery of SN 2008es show tentative ev-idence for a significant reduction in the SN photomet-ric decline rate (see Figure 1). We note, however, thatthe late-time UVOT measurements have significant er-rors, and these observations may be consistent with theearly trend seen in the light curve. Gezari et al. (2008b)suggested that these observations were evidence that theSN luminosity made a transition to being powered by ra-dioactive heating from Co. With only two observationsof this decline, we caution against prematurely identify-ing the Co decay tail, and instead argue that the ob-servations are inconclusive at this time. For instance,it would be possible to mimic the behavior of Co de-cay with a late-time light echo, or interaction with anextended CSM. We note that SN 2006gy also showedevidence for a “flattening” several months after explo-sion (Smith et al. 2007), but further observations takenthe following year indicated that this was not clearly the Co tail (Smith et al. 2008b). Our last optical spectrum(Fig. 2) also shows no evidence that the ejecta were be-coming optically thin on that date, as would be expectedif SN 2008es were transitioning to the radioactive decaytail. Late-time photometry will be necessary to conclu-sively identify if and when the light curve of SN 2008esmade a transition to being powered by Co decay, andthus allow a measurement of the amount of Ni synthe-sized in the explosion. SPECTROSCOPY
The SN 2008es spectral sequence plotted in Figure 2is labeled with ages (in the rest frame) relative to theobserved V -band maximum in order to facilitate com-parison with SNe 1979C and 1980K in Figure 3. Ourfirst two spectra of SN 2008es (at +3 and +13 d rela-tive to maximum light) show a smooth and featurelessblue continuum with no identifiable spectral features. Inparticular, we do not detect an emission feature near5650 ˚A (4660 ˚A in the rest frame) seen in earlier spectraof SN 2008es taken between 2008 May 1 and 2008 May8 (Yuan et al. 2008b; Gezari et al. 2008b). Gezari etal. (2008b) attribute this feature solely to He II λ II λ III /N III λ − III , N
III , and O
III with an expansion velocityof about 20,000 km s − .Our next spectrum, at +21 d, is noticeably redderand is the first to show strong spectral features. BothSN 1979C at a similar epoch and SN 2008es show a bluecontinuum with weak, low-contrast lines of H and He I lines present mostly in absorption (Fig. 3). However, H α is present only in emission and is weak in SN 2008es. TheH α line in SN 2008es has a full width at half-maximumintensity (FWHM) of 10,000 km s − and an equivalentwidth of only 22 ˚A. (Both of these values have large un-certainties due to difficulties in defining the continuumfor a line with such a low amplitude.) One differencebetween the two objects is the lower apparent velocitiesin SN 2008es. The H β and He I λ − . In SN 2008es, these values are 6000 and 5700km s − , respectively, although we caution that the ex-act values are correlated with the assumed value for theredshift.Over the next two months, the spectra of SN 2008esplotted in Figure 2 became redder, reflecting the cool-ing photospheric temperature evolution discussed above.In addition, the spectral features first seen in the day+21 spectrum gradually became more prominent. TheSN features are still muted in amplitude relative to thoseexpected in a normal Type II SN. This may be an exam-ple of the “top-lighting” effect described by Branch et al.(2000), where continuum emission from interaction withCSM illuminates the SN ejecta from above and results ina rescaling of the amplitudes of spectral features.By the time of our day +68 spectrum, the P-Cygnispectral features due to H Balmer lines, Na I , andFe II become more prominent and the overall appearancestarts to resemble that of normal SNe II. The H α profilelacks an absorption component, which may be commonto SNe II-L and not SNe II-P (e.g., Schlegel 1996, Fil-ippenko 1997). The broad H α emission extends (at zerointensity) from − − , with a FWHMof 9500 km s − . This velocity width is intermediate be-tween that of the SNe 1979C and 1980K spectra (FWHM ≈ ≈ − , respectively).Unlike in most core-collapse SNe, the velocity of theminimum of the H β line increased over time, from 6000km s − at day 21, to 8700 km s − at day 68, as shownin the bottom panel of Figure 6. The exact values of thevelocity depend directly on the assumed redshift, butthe trend is independent of those uncertainties. Unfor-tunately, the other absorption lines are mostly blended(except for H α , which does not show an absorptioncomponent), so we cannot isolate the velocity trend inother spectral features without spectral modeling. Thetop panel of Figure 6 shows the evolution of the He I λ I λ I dominatesthe blend, and as the ejecta cool at later epochs Na I should dominate, resulting in an ∼
800 km s − redwardshift of the rest wavelength. After taking into accountthat redward shift, the top panel of Figure 6 also showssome weak evidence for a blueward shift of the absorptionminimum to ∼ − .The only supernova known to us to show increasing ab-sorption velocities over time is the peculiar SN Ib 2005bf. F l ux + c on s t a n t +21+33+45+68 He I λ λ −15 −10 −5 0 5 10Velocity (1000 km s −1 )0.40.60.81.01.2 F l ux + c on s t a n t +21+33+45+68 H β Fig. 6.—
Velocity evolution of absorption minima. The spectraare labeled with dates after maximum light in the same manner asin Figure 2. In the bottom panel the apparent blueshift of the H β absorption minimum increases over time. A vertical dashed lineat − − marks the velocity of the absorption minimumon day 21 to guide the eye. In the top panel, the evolution of theHe I λ I λ λ − − marks the absorption minimum on day 21 to guide the eye. The trend of increasing absorption velocities was onlyvisible in the three He I lines, but not in other lines suchas Ca II H&K (Modjaz 2007). Tominaga et al. (2005)explained the effect as being due to progressive outwardexcitation of He I by radioactive Ni in the interior asthe ejecta expanded and the density decreased, an effectthat seems to be of little relevance to SN 2008es. A morelikely possibility is that blending with some unidentifiedline is shifting the velocity of the apparent absorptionminimum. At late times H β dominates its region of thespectrum, but at early times He II λ β ab-sorption profile. Another scenario is that some unusual,but as yet unidentified, radiative transfer effect is affect-ing the wavelength of the apparent absorption minimumin SN 2008es. For example, if the optical depth in an op-tically thick line increases sufficiently rapidly with time(e.g., as the ejecta cool and recombine), then the appar-ent absorption minimum could move blueward (Jeffery& Branch 1990). None of these suggestions explain whysuch an effect would be present only in SN 2008es andnot in normal SNe. DISCUSSION
Miller et al. V , R ) Fig. 7.—
Rest-frame brightness evolution of five of the mostluminous known SNe. For SN 2008es we derive the absolute mag-nitude using UVOT and the KAIT/Nickel V -band magnitudes(filled green squares) and the KAIT/Nickel R -band magnitudes(red squares), assuming that day 0 is the discovery date. Early g -band photometry from Gezari et al. (2008b), converted to V -band using the color equations of Jester et al. (2005), is also shown(open green squares). The light curve for SN 2005ap (orange) isderived from unfiltered photometry in Quimby et al. (2007), andSN 2006gy (gray circles) is also unfiltered from Smith et al. (2007). R -band photometry of SN 2006tf (blue circles) is from Smith et al.(2008a), and r / r ′ photometry of SN 2005gj (dotted line) is fromPrieto et al. (2007). The light curve for SN 2006tf is shifted by+16 d from that in Smith et al. (2008a); since the explosion dateis not known, we chose to align its time of peak luminosity withthose of SNe 2005ap and 2008es. The Physical Nature of SN 2008es
The photometric evolution of SN 2008es (see Fig. 7)is much faster than that of other very luminous SNe likeSNe 2006gy (Smith et al. 2007; Ofek et al. 2007), 2006tf(Smith et al. 2008a), and 2005gj (Prieto et al. 2007),and its spectrum also betrays no evidence for the strongCSM interaction seen in these other SNe IIn, in theform of narrow lines from the CSM or intermediate-widthH α from the post-shock shell. Indeed, considering bothits photometric and spectroscopic evolution, we suggestthat SN 2008es is most like the overluminous SN 2005ap(Quimby et al. 2007) and thereby most closely resemblesa Type II-L that is 4–5 mag more luminous than typicalSNe II-L (Richardson et al. 2002).To power the tremendous luminosity of SN 2008eswith radioactive decay would require an initial Nimass of ∼
10 M ⊙ (see, e.g., Smith et al. 2007). This Ni mass would very likely need to be generated in apair-instability explosion (Barkat, Rakavy, & Sack 1967;Bond, Arnett & Carr 1984), but this Ni mass seemsproblematic given that it would be larger than the mod-est envelope mass indicated by the relatively fast riseand decay time (see below). In addition, the photomet-ric decline of 0.042 mag d − is faster than that of Co,0.0098 mag d − , making radioactive heating unlikely asthe dominant source of the energy. Despite the lack of a Type IIn spectrum, the mostlikely interpretation seems to be that the high luminos-ity of SN 2008es is the result of converting shock energyinto visual light. This can be accomplished, in principle,if the shock kinetic energy is thermalized throughout amassive envelope, like a normal SN II-P or II-L, but witha much larger initial radius of (2–3) × cm (based onthe peak luminosity and the evolution in Fig. 4). Werethe initial radius much smaller than this it would provedifficult to convert ∼ ergs of kinetic energy to ∼ ergs of radiation because there would be significant adi-abatic losses. The apparent temperature evolution from ∼ ∼
66 d (Fig. 4),reminiscent of other normal SNe II, suggests that therecombination photosphere is receding through a cool-ing envelope in SN 2008es. In this scenario, the usualnarrow/intermediate-width H α emission that is taken asa signpost of CSM interaction might be avoided if theCSM shell is initially very opaque, and if the shock en-counters no further CSM at larger radii (see Smith et al.2008a).Since the required initial radius exceeds that of thelargest known red supergiants (see Smith et al. 2001)by a factor of 20–30, it requires the envelope to be anunbound, opaque CSM shell ejected prior to the SN ex-plosion instead of a traditional bound stellar envelope.Similar models were suggested for SN 2005ap (Quimbyet al. 2007), SN 2006gy (Smith & McCray 2007), andSN 2006tf (Smith et al. 2008a), implying CSM envelopemasses of 0.6, ∼
10, and 18 M ⊙ , respectively. The corre-sponding CSM mass for SN 2008es would be roughly ∼ ⊙ in this scenario, because its evolution and expansionspeeds are slower than those of SN 2005ap. This CSMmass of ∼ ⊙ allows the observed H α Doppler veloc-ities to remain faster than in SNe 2006gy and 2006tf,where the heavy CSM shells decelerated the shocks toonly 4000 and 2000 km s − , respectively (Smith et al.2007; Smith et al. 2008a). The smaller CSM mass forSN 2008es relative to SNe 2006gy and 2006tf would alsoexplain the comparatively short rise and decay observedfor SN 2008es, as the light diffusion time for this SN ismuch shorter than in SNe 2006gy and 2006tf (Smith &McCray 2007). The putative envelope ejection precedingSN 2008es must have occurred 10–100 yr prior to the ex-plosion (for an unknown progenitor wind speed of v CSM = 10–100 km s − ), indicating a progenitor mass-loss rateof order 0.05–0.5 M ⊙ yr − . This is much larger than anysteady stellar wind (see Smith et al. 2007, and refer-ences therein), providing another case of impulsive massejection in the decades immediately preceding some SNe.An alternative analysis of SN 2008es, submitted forpublication shortly after this paper was made availableelectronically, also suggests that the extreme luminos-ity is powered via interaction with CSM (Gezari et al.2008b). However, the initial version of Gezari et al.(2008b) argues against an unbound CSM shell ejectionprior to the SN explosion, and instead argues that theCSM is created by a dense progenitor wind. The Host of SN 2008es
Currently there is no conclusive detection of the hostgalaxy of SN 2008es, so its metallicity and correspond-ing implications for the pre-SN evolution are not known.From SDSS DR6 images (Adelman-McCarthy et al.uminous Type II-L SN 2008es 92008), we derive a 3 σ upper limit of m r ′ > M V > − L ∗ galaxy and could be compara-ble to or fainter than the Small Magellanic Cloud (with M V = − . .Without a definitive detection of the host galaxy weassume that there is no host extinction. This assump-tion is further supported by both the low luminosity ofthe host galaxy and our BB fits (see §
3) that show ex-cellent agreement with our UV measurements, where theeffects of host extinction would be especially prominent.In analogy with the underluminous hosts of SNe 2006tf(Smith et al. 2008a) and 2005ap (Quimby et al. 2007),deep imaging after the SN has faded may uncover thehost galaxy nearly coincident with the SN position, whichwould allow further constraints to be placed on extinctionin the host. These three extreme SNe and their underlu-minous hosts may be hinting that very luminous SNe IIpreferentially occur in low-luminosity host galaxies.
Rates of Extremely Luminous SN 2008es-likeEvents
Over the past three years the TSS has successfullyfound tens of SNe with a surprising rate of unusualobjects. The list of such SNe includes SN 2005ap,SN 2006gy, SN 2006tf, SN 2008am, and now SN 2008es.However, this apparent high anomaly rate is probably theresult of combining a huge survey volume with an intrin-sically rare class of objects. Following Quimby (2008),we compare the volume probed by the TSS for SNe Ia,and bright core-collapse SNe. The highest redshift of the ∼
30 SNe Ia found by the TSS is z ≈ .
1, while SN 2005apwas found at z ≈ .
3. The comoving volume scanned bythe TSS is therefore ∼
23 times bigger for SN 2008es-likeobjects than for SNe Ia. Of the five bright objects foundby the TSS, only two, SN 2005ap and SN 2008es, canbe considered together as a class. Thus, the comparativerate is about 30 / × ≈
350 times smaller.The KAIT SN search (Filippenko et al. 2001) has dis-covered about 400 SNe Ia in the past 10 years in tar-geted nearby galaxies, hence the expected number ofSN 2008es-like objects is of order 1. This is consistentwith the nondetection of any such SN.Using the light curve of SN 2008es and the accumu-lated observations of about 10 ,
000 spiral galaxies in theKAIT sample which were observed regularly during thepast ten years, we find an upper limit on the rate (perunit K -band luminosity) of such SNe to be about 1 / § A galaxy ∼ ′′ to the NE of the SN position is unlikely to bethe host itself even if found to be at a similar redshift as SN 2008es;this would require significant massive star formation at a projectedphysical distance of ∼
31 kpc. and therefore we cannot directly compare them to theTSS. We conclude that the number of luminous Type II-L SNe, as discovered by the TSS, does not appear to bethe result of a statistical fluke, and predict many suchdetections with future wide-field synoptic surveys. CONCLUSIONS
We have reported on our early observations ofSN 2008es, which at a peak optical magnitude of M V = − . − ,indicates that the radioactive decay of Co is likely notthe dominant source of energy for this SN. Integration ofthe bolometric light curve of SN 2008es yields a total ra-diated energy output of & ergs. The optical spectraof SN 2008es resemble those of the luminous SN 1979C,but with an unexplained increase in the velocity of theH β absorption minimum over time.We also examined the rate of discovery of extremelyluminous SNe by the Texas Supernova Search and findthat their discovery of the five most luminous observedSNe in the past four years is probably not a fluke; severalmore such detections are expected in the coming years.Finally, what behavior can we expect from SN 2008esat late times, roughly 1 yr or more after discovery? Re-gardless of whether the peak luminosity was powered byradioactive decay or optically thick CSM interaction (seeSmith et al. 2008b), a SN can be powered by strongCSM interaction at late times if the progenitor had asufficiently high mass-loss rate in the centuries beforeexploding. We have seen examples of both: SN 2006tfhad strong H α emission indicative of ongoing CSM in-teraction at late times (Smith et al. 2008a), whereasSN 2006gy did not (Smith et al. 2008b). SN 1979C wasless luminous at peak than those two SNe, but it hasbeen studied for three decades because its late-time CSMinteraction is powering ongoing emission in the radio, op-tical, and X-rays (Weiler et al. 1981; Fesen & Matonick1993; Immler, Pietsch, & Aschenbach 1998). With suchan extraordinarily high peak luminosity, a late-time IRecho such as that seen in SN 2006gy (Smith et al. 2008b)is also likely if SN 2008es has dust waiting at a radiusof ∼ Nishould be evident in the late-time decline rate.We thank the anonymous referee for comments thathelped improve this paper. We are grateful to B. Jan-nuzi for approving Kitt Peak DD observations, and toDiane Harmer and David L. Summers for carrying outthe observations. M. Malkan accommodated a small tele-scope time trade, while N. Joubert and B. Macomber as-sisted with some of the observations. P. Nugent kindlychecked for pre-discovery images of SN 2008es from Deep-Sky. A.A.M. is supported by a UC Berkeley Chancel-lor’s Fellowship. M.M. is supported by a Miller Insti-tute research fellowship. N.R.B. and D.A.P. are par-tially supported by a SciDAC grant from the Depart-ment of Energy. J.S.B.’s group is partially supportedby NASA/
Swift grant
Swift
Guest Investigator Grant
Swift data archive. Some of the data presented hereinwere obtained at the W. M. Keck Observatory, which isoperated as a scientific partnership among the Califor-nia Institute of Technology, the University of California,and NASA; the Observatory was made possible by thegenerous financial support of the W. M. Keck Founda-tion. The authors wish to recognize and acknowledge thevery significant cultural role and reverence that the sum-mit of Mauna Kea has always had within the indigenousHawaiian community; we are most fortunate to have theopportunity to conduct observations from this mountain.data archive. Some of the data presented hereinwere obtained at the W. M. Keck Observatory, which isoperated as a scientific partnership among the Califor-nia Institute of Technology, the University of California,and NASA; the Observatory was made possible by thegenerous financial support of the W. M. Keck Founda-tion. The authors wish to recognize and acknowledge thevery significant cultural role and reverence that the sum-mit of Mauna Kea has always had within the indigenousHawaiian community; we are most fortunate to have theopportunity to conduct observations from this mountain.