On the Progenitor and Supernova of the SN 2002cx-like Supernova 2008ge
Ryan J. Foley, Armin Rest, Maximilian Stritzinger, Giuliano Pignata, Joseph P. Anderson, Mario Hamuy, Nidia I. Morrell, Mark M. Phillips, Francisco Salgado
aa r X i v : . [ a s t r o - ph . C O ] S e p Draft version November 6, 2018
Preprint typeset using L A TEX style emulateapj v. 11/10/09
ON THE PROGENITOR AND SUPERNOVA OF THE SN 2002CX-LIKE SUPERNOVA 2008GE
Ryan J. Foley , Armin Rest , Maximilian Stritzinger , Giuliano Pignata , Joseph P. Anderson Mario Hamuy , Nidia I. Morrell , Mark M. Phillips , and Francisco Salgado Draft version November 6, 2018
ABSTRACTWe present observations of supernova (SN) 2008ge, which is spectroscopically similar to the peculiarSN 2002cx, and its pre-explosion site that indicate that its progenitor was probably a white dwarf.NGC 1527, the host galaxy of SN 2008ge, is an S0 galaxy with no evidence of star formation or massivestars. Astrometrically matching late-time imaging of SN 2008ge to pre-explosion
HST imaging, weconstrain the luminosity of the progenitor star. Since SN 2008ge has no indication of hydrogen orhelium in its spectrum, its progenitor must have lost its outer layers before exploding, requiringthat it be a white dwarf, a Wolf-Rayet star, or a lower-mass star in a binary system. Observationsof the host galaxy show no signs of individual massive stars, star clusters, or H II regions at theSN position or anywhere else, making a Wolf-Rayet progenitor unlikely. Late-time spectroscopy ofSN 2008ge show strong [Fe II ] lines with large velocity widths compared to other members of this classat similar epochs. These previously unseen features indicate that a significant amount of the SN ejectais Fe (presumably the result of radioactive decay of Ni generated in the SN), further supporting athermonuclear explosion. Placing the observations of SN 2008ge in the context of observations ofother objects in the class of SN, we suggest that the progenitor was most likely a white dwarf.
Subject headings: astrometry — stars: evolution — supernovae: general — supernovae: individual(SN 2008ge) INTRODUCTION
In the last decade, a new class of supernovae (SNe) hasbeen discovered. The class, named after its first mem-ber, SN 2002cx (Li et al. 2003), has several observationalsimilarities to typical SNe Ia, but also has several dis-tinguishing properties: low luminosity for its light-curveshape (e.g., Li et al. 2003), a lack of a second maximumin the NIR bands (e.g., Li et al. 2003), low photosphericvelocities (e.g., Li et al. 2003), late-time spectra domi-nated by narrow permitted Fe II (e.g., Jha et al. 2006),and a host-galaxy morphology distribution highly skewedto late-type galaxies (Foley et al. 2009; hereafter F09;Valenti et al. 2009). Additionally, a single member ofthe class, SN 2007J, displays strong He I in its spectrum(F09).A recent member of this class, SN 2008ha, had botha very low photospheric velocity at maximum brightness This paper includes data gathered with the 6.5 meter Mag-ellan telescope at Las Campanas Observatory, Chile. Based on observations obtained at the Gemini Observa-tory, Cerro Pachon, Chile (Gemini Programs GS-2008B-Q-32,GS-2008B-Q-56, and GS-2009A-Q-17). Harvard-Smithsonian Center for Astrophysics, 60 GardenStreet, Cambridge, MA 02138, USA. Clay Fellow. Electronic address [email protected] . Department of Physics, Harvard University, 17 OxfordStreet, Cambridge, MA 02138, USA. Dark Cosmology Centre, Niels Bohr Institute, University ofCopenhagen, Juliane Maries Vej 30, 2100 Copenhagen Ø, Den-mark. Carnegie Observatories, Las Campanas Observatory, Casilla601, La Serena, Chile. Department of Astronomy, The Oskar Klein Centre, Stock-holm University, 10691 Stockholm, Sweden. Departamento de Ciencias Fisicas, Universidad AndresBello, Avda. Republica 252, Santiago, Chile. Departamento de Astronom´ıa, Universidad de Chile, San-tiago, Chile. (about 4000 km s − ; Foley et al. 2010a) and extremelylow luminosity ( M V = − . M ⊙ main-sequencemass CO star undergoing a significant amount of fall-back (Moriya et al. 2010); however, stars with a main-sequence mass of M . M ⊙ are not expected to un-dergo fallback (Fryer 1999). The maximum-light spec-trum of SN 2008ha suggested that C/O burning occurredduring the explosion, pointing further to a WD progeni-tor (Foley et al. 2010a).The best way to determine the progenitor system forthese events would be to unambiguously detect the pro-genitor (or donor) star of a SN in pre-explosion imag-ing. This technique has been very successful at detectingnearby core-collapse SN progenitors (see Smartt 2009,for a review), but has rarely been attempted for SNe Ia(Maoz & Mannucci 2008; Nelemans et al. 2008). For thistype of study, the following data are required: a high-resolution (to separate individual stars), relatively deep(to detect the progenitor star) pre-explosion image of theSN site and a SN image that is deep enough to havegood detections of several stars in common with the pre-explosion image. A relative astrometric solution is ob-tained for the two images, and then the SN position canbe precisely determined on the pre-explosion image. Po-tential progenitor stars can then be identified and theirproperties studied, or alternatively if no stars are consis-tent with the SN position, limits can be placed on the Foley et al.properties of the progenitor star. With the appropriateobservations, one can place interesting constraints on theprogenitor properties of SN 2002cx-like SNe.SN 2008ge was discovered by the CHilean Auto-mated Supernova sEarch CHASE (Pignata et al. 2009)on 2008 October 8.27 (UT dates are used through-out this paper) at mag 12.8 (Pignata et al. 2008b) inNGC 1527, an S0 galaxy with a recession velocity of1212 km s − . Using the surface brightness fluctuationmethod, Tonry et al. (2001) determined that the dis-tance modulus for NGC 1527 is µ = 31 . ± .
22 mag.Using the new Cepheid zero point of Freedman et al.(2001), this corresponds to a corrected distance mod-ulus of µ = 31 . ± .
22 mag, which we will adopt as thedistance modulus for SN 2008ge.Despite discovering SN 2008ge on 2008 October 8.27,Pignata et al. (2008b) also report detecting SN 2008geon previous images, with the earliest detection on 2008September 8.23 (at mag 12.4) and the last non-detectionon 2008 August 21.24. Our light curve (presented inSection 2.2) shows that SN 2008ge peaked in V around2008 September 17.We determined from a spectrum obtained on 2008October 15 that SN 2008ge was similar to the pe-culiar SN 2002cx 23 d past maximum brightness(Stritzinger et al. 2008). This is consistent with our V -band phase of 28 d on that date. Given the sparse pre-maximum data, the rise time should be between 9 – 27 d.NGC 1527 has been twice imaged by HST before thedetection of SN 2008ge. In this paper, we present obser-vations of SN 2008ge and its host galaxy, determine theproperties of potential progenitor stars at the location ofSN 2008ge, and examine the implications of those prop-erties in the context of the progenitors for SN 2002cx-likeobjects. OBSERVATIONS AND DATA REDUCTION
Identification and HST Photometry of theProgenitor Site
NGC 1527 was observed with
HST /WFPC1 on1993 October 20 (Program 4904; PI Illingworth) and
HST /WFPC2 on 1995 January 6 (Program 5446; PIIllingworth). The WFPC1 observation was a single 300 sexposure with the F555W (roughly V ) filter, while theWFPC2 observation included two exposures of 80 s eachwith the F606W (a broad V ) filter.To determine whether the progenitor of SN 2008geis detected in the archival HST observations, we per-formed differential astrometry using optical observationsof the SN. Observations of SN 2008ge were obtainedwith LDSS3 with the Magellan Clay telescope on2009 October 16. The images were reduced using theSMSN pipeline as described in Rest et al. (2005) andMiknaitis et al. (2007).The astrometry was performed with the IRAF tasks geomap and geotran using 29 objects in common withthe HST /WFPC2 images resulting in an astrometric rms IRAF: the Image Reduction and Analysis Facility is dis-tributed by the National Optical Astronomy Observatory, which isoperated by the Association of Universities for Research in Astron-omy, Inc. under cooperative agreement with the National ScienceFoundation. of σ LDSS3 → HST = 45 mas in each coordinate. The sameprocedure was performed with the
HST /WFPC1 images,which despite its smaller field of view, and therefore only5 stars in common, we obtained an astrometric rms of σ GB → HST = 20 mas.The SN is very close to the bright nucleus of NGC 1527.At the time of our Magellan observation, it had faded tothe point where it was difficult to directly measure itsposition. To provide an accurate SN position, we sub-tracted the south-west section of the galaxy from thenorth-east section of the galaxy (rotated and flippedabout the galaxy center), producing a relatively cleandifference image (shown in Figure 1). The SN is clearlydetected in our stacked photometry with position ( α , δ ) = (04:08:24.689, − HST imagesare shown in Figure 1. In both cases, we find no clearprogenitor coincident with objects in the archival
HST images. The WFPC2 image is 0.45 mag deeper than theWFPC1 image, and the filters are very similar. There-fore, for the remainder of this study, we focus on theWFPC2 image.Since no progenitor star was detected, we can use the
HST /WFPC2 image to place limits on the brightnessof the progenitor. The surface-brightness profile is ex-tremely smooth, and we expect that any bright starwould be easily detected. Using dophot , we were ableto detect nearly all astrophysical objects to a signal-to-noise ratio (S/N) of ∼ dophot on that image to recover the artificial stars. We findthat at the SN position and positions with similar back-grounds that we recover >
90% of all objects to
S/N = 3.We then calculated the brightness that would yield a par-ticular S/N. For
S/N = 3 (5), we have a limiting magni-tude of m F W = 24 . M V = − . − . Supernova Photometry
Photometry of SN 2008ge was obtained by the 0.41 mPanchromatic Robotic Optical Monitoring and Polarime-try Telescope (PROMPT Reichart et al. 2005) in the Lu-minance filter, which is a passband filter with wavelengthrange from ∼ ∼ V band we used the S-correction technique(Stritzinger et al. 2002), but following the prescriptionof Pignata et al. (2008a). We convert to the V bandbecause its effective wavelength is closest to the Lumi-nance filter. Since the spectroscopic follow up beganaround 20 days after maximum light, to correct the earlyphotometric points we used the spectra of SN 2005hk(Phillips et al. 2007). The two objects are both spectro-scopically similar to SN 2002cx at later times (see Fig-ure 4) and the correction agrees quite well for the phasesN 2008ge: Progenitor and Supernova 3 Figure 1.
LDSS3 r -band (left), HST /WFPC2 F606W (center) and
HST /WFPC F555W (right) images at the position of SN 2008geobtained 1 year after and 13 and 15 years before explosion, respectively. The images are 20 ′′ × ′′ , and North is up and East is left.The LDSS3, HST /WFPC, and
HST /WFPC2 images have a pixel scale of 0.188, 0.044, and 0.1 ′′ pixel − , respectively. The LDSS3 imagehas been flipped about the center of the host galaxy and subtracted from the original image to create a difference image. The position ofSN 2008ge is marked by the black circles whose radius corresponds to 1 ′′ for the LDSS3 image and 10 σ uncertainty in the position for the HST images. where the spectroscopic sequences overlap. This makesus confident that the early-time corrections are appro-priate.We present our V -band light curve of SN 2008ge inTable 1 and Figure 2. The SN peaked at V ≈ . V B (Phillips et al.2007). At our adopted distance modulus (and assumingno host-galaxy extinction; see Section 2.3), SN 2008gepeaked at M V ≈ − . ∼ M V = − .
49 mag) and 2005hk ( M V = − .
08 mag;Phillips et al. 2007). SN 2008ge declines slower than ei-ther SN 2002cx or SN 2005hk in V . SN 2008ge has∆ m ( V ) ≈ .
34 mag (the relatively large photomet-ric uncertainty and single data point before maximumpropagates to a large uncertainty in the measurement),which is much smaller than the measured value for eitherSN 2002cx (0 . ± .
06 mag) or 2005hk (0 . ± .
05 mag).We caution that our V band may still be slightly con-taminated by th R band, and could partially explain thewidth of the V -band light curve. Spectroscopy
Seven low-resolution spectra of SN 2008ge were ob-tained at the Las Campanas Observatory with the Boller& Chivens spectrograph on the 2.5 m du Pont tele-scope, the IMACS spectrograph (Dressler et al. 2006) onthe Magellan Baade 6.5 m telescope, and the LDSS3spectrograph on the Magellan Clay 6.5 m telescope.An additional three spectra were also procured withthe GMOS spectrograph (Hook et al. 2005) on the 8 mGemini-South telescope. A single high-resolution ( R ≈ , Table 1
CHASE V -BandPhotometry ofSN 2008geJD V a (mag)2454717.7 13.95 (06)2454726.4 13.82 (12)2454734.8 13.91 (08)2454741.7 14.08 (06)2454747.8 14.40 (05)2454754.7 14.66 (06)2454760.6 14.84 (07)2454781.6 15.23 (05)2454787.6 15.34 (06)2454800.7 15.53 (07)2454817.6 15.92 (06)2454838.6 16.48 (08)2454842.6 16.45 (07)2454847.6 16.54 (08) Note . — Uncertain-ties (units of 0.01 mag)are given in parentheses. a V -band measurementsare transformed fromobservations observedin the Luminance filter. ble 2.Standard two-dimensional image processing and spec-trum extraction was accomplished with IRAF. Low-orderpolynomial fits to calibration-lamp spectra were derivedfrom night-sky lines in the object frames were applied.Flux calibration was applied to the extracted spectra us-ing observations of spectrophotometric standards usuallyobserved during the same night as the SN. Telluric fea-tures were removed from all of the spectra taken at LasCampanas using high S/N spectra of telluric standards.Telluric corrections were not performed on the Geminispectra.The spectroscopic sequence of SN 2008ge is presentedin Figure 3, and Figure 4 contains a comparison be-tween SN 2008ge and SN 2002cx. Although the phases inour comparison are not perfectly matched, and some of Foley et al. A pp a r e n t B r i gh t n e ss ( m a g ) SN 2008geV
SN 2002cxSN 2005hk
Figure 2. V -band light curves of SNe 2002cx (dashed line),2005hk (solid line), and 2008ge (blue points). The light curvesof SNe 2002cx and 2005hk have been shifted to match the lightcurve of SN 2008ge at peak. Table 2
Log of Spectral ObservationsTelescope / ExposurePhase a UT Date Instrument (s)40.4 2008 Oct. 27.3 du Pont/B&C 3 × × × × × × × a Days since V maximum, 2008 Sep. 16.9 (JD 2,454,726.4). the features appear to be slightly broader in SN 2008ge,the strong resemblance between the two objects is ir-refutable. The first spectrum of SN 2008ge is dominatedby Fe-group elements, with additional strong lines fromNa D and Ca II . We refer the reader to Branch et al.(2004), which performed a detailed study of the spectraof SN 2002cx.No signatures of H α nor He I is found in the spectralseries of SN 2008ge, indicating that the progenitor waslikely an evolved star. Our high S/N, high-dispersion( R ≈ , α emission from thehost galaxy. We measure a 3 σ upper limit of 0.01 ˚A forthe equivalent width of the Na D2 lines suggesting thatthere is little to no host-galaxy extinction. There is alsono sign of star formation at the position of the SN norany significant amount of hydrogen in the circumstellarenvironment. R e l a ti v e f λ + C on s t a n t +40 d+46 d+48 d+67 d+81 d+114 d+145 d+160 d+191 d+225 d Figure 3.
Optical spectra of SNe 2008ge. Phases relative to max-imum brightness are marked.
NGC 1527 is undetected in IRAS, with relatively strictlimits of < . < . × − M ⊙ year − . Additionally,NGC 1527 is not detected in H I
21 cm with a flux limitof < . × − Jy (Huchtmeier 1989). Both of theseobservations place significant constraints on the SFR ofthe host galaxy. RESULTS
Limits on Progenitor System
In Figure 5, we present the non-rotating standardmass-loss evolutionary tracks for stars with Z = 0 . Z/Z ⊙ , respec-tively) of the Geneva group (Lejeune & Schaerer 2001).We compare these tracks to the upper limit of M V = − . M init . M ⊙ . Forhigh metallicity, the limit is lower with M init . M ⊙ . Ifthe progenitor was on the horizontal or red-giant branch,then there are stricter limits of M init . M ⊙ for boththe low and high metallicity cases.As discussed in Section 2.3, there is no indication ofhydrogen in the spectrum of SN 2008ge, so the progeni-tor star was almost certainly not on the main sequence,horizontal branch, or red-giant branch star (althoughN 2008ge: Progenitor and Supernova 5 R e l a ti v e f λ SN 2008ge +40 dSN 2002cx +25 d
Cr II Fe IINa IFe II Co II Fe II Co IICa II
SN 2008ge +225 dSN 2008ge +67 dSN 2002cx +227 d
Figure 4.
Spectra of SN 2008ge at +40 (black; top panel), +67(blue; bottom panel), and +225 d (black; bottom panel) relative to V maximum. Comparison spectra of SN 2002cx (+25 and +227 drelative to B maximum in the top and bottom panels, respectively)are shown in red. For the well-sampled SN 2005hk, its light curvespeaked in V B (Phillips et al. 2007). Line identificationsfrom Branch et al. (2004) have been marked. a potential binary companion could be). In Figure 5,we also plot example Wolf-Rayet stars (Crowther 2007).The most luminous of these stars, which correspond tolate-type WN Wolf-Rayet stars, are ruled out by our 3- σ limit. However, earlier WN and WC/WO stars can notbe directly ruled out by the current observations.NGC 1527 is an S0 galaxy with no indication of starformation in its spectrum, no indication of a dust lane orother star formation (Phillips et al. 1996; Figure 1), andno detection from IRAS or in H I , placing strong con-straints on the SFR of < . × − M ⊙ year − . Fittinga single-stellar population (Jimenez et al. 2003) to thehost-galaxy spectrum (obtained at the same time as theSN spectrum), we find a good fit with a single 9.5 Gyrpopulation (see Foley et al. 2010b for details of the fit-ting). Additionally, there is no indication of any emissionlines in the host-galaxy spectrum, placing strong con-straints on the presence of massive stars or star clusters.All evidence indicates that there is not any significantpopulation of massive stars in NGC 1527. The surface-brightness profile is extremely smooth, giving further evi-dence of a lack of individual massive stars or star clusters M V ( m a g ) l og ( L / L O • )
120 85 60 40 25 20 15 12 60 40 25 20 15 12
Figure 5.
Temperature-Magnitude diagram displaying theGeneva group stellar evolution tracks (Lejeune & Schaerer 2001)for Z = 0 .
001 (dashed curves) and 0.1 (solid curves). The initialmass for each track is labeled. The black and green curves repre-sent initial masses that are consistent with our magnitude limit ifthe star exploded on the main sequence. The blue and red curvesrepresent initial masses that are consistent with our magnitudelimit if the star exploded on the main sequence or at a later stageof evolution. The black dots represent Wolf-Rayet stars (Crowther2007) corrected from v to V using the relationship of Breysacher(1986). The black solid (dotted) horizontal line represents our 3 σ (5 σ ) limiting magnitude for the progenitor of SN 2008ge. in the galaxy. For a star brighter than our detection limitto be masked by the galaxy, there would need to be largefluctuations to the surface-brightness profile, which donot exist. Unique Phase Spectroscopy
Very few SN 2002cx-like objects have had their spec-tra published, and only SNe 2002cx (Li et al. 2003),2005hk (Phillips et al. 2007; Sahu et al. 2008), 2007qd(McClelland et al. 2010) and 2008ha (F09; Foley et al.2010a; Valenti et al. 2009) have had their light curvespublished. As a result, only these four objects havespectra with phase information. Li et al. (2003) pub-lished spectra with phases ranging from − B maximum. Jha et al. (2006) presentedspectra of SN 2002cx at +227 and +277 d relative to B maximum. Phillips et al. (2007) presented spectraof SN 2005hk that covered − B maximum while Sahu et al. (2008) extended the spec-troscopic coverage by presenting spectra at +228 and Foley et al.+377 d. SN 2007qd only has four published spectra withthe earliest and latest being at +3 and +15 d, respec-tively. Although SN 2008ha had some spectroscopic dif-ferences from SNe 2002cx and 2005hk, spectra rangingfrom − B maximum have also beenpublished (F09; Foley et al. 2010a; Valenti et al. 2009).However, no spectra of this class have been publishedthat cover the phases 68 – 226 d relative to B maximum.Our spectroscopic coverage of SN 2008ge fills in thisgap with 6 spectra in this range. As seen in Figure 3,during this time the spectra remain relatively similar.This is not unexpected since the spectra of SN 2002cx at+56 and +227 d were very similar to each other exceptfor the line widths (Jha et al. 2006). The main changein the spectral evolution of SN 2008ge is that the featureat ∼ II ]; see Section 3.3)increases in strength with time.Although it is important to see a SN 2002cx-like ob-ject transition from early to late times, SN 2008ge maynot behave in a typical manner for this class. UnlikeSNe 2002cx and 2005hk, the widths of the features donot significantly decrease (to separate into narrow, pri-marily Fe II , features) in SN 2008ge with time. Late-Time Spectroscopy
At late times, the spectra of SNe 2002cx and 2005hkwere dominated by permitted Fe II lines, with additionalfeatures from Na I , Ca II , [Ca II ], and possibly [O I ](Jha et al. 2006; Sahu et al. 2008). Jha et al. (2006) andSahu et al. (2008) both suggested that these objects mayhave emission from [Fe II ] λλ , I λ t = 227 d (see Figure 4), but SN 2008gehas much broader lines. SNe 2002cx and 2005hk had ex-tremely low-velocity features ( ∼
500 km s − ) at these latetimes. Clearly, SN 2008ge does not share this character-istic. Fitting a Gaussian to the relatively isolated featureat 5900 ˚A, which is likely Na D (but could possibly be[Co III ] λ − .In addition to the line widths, the main difference be-tween the late-time spectra of SNe 2002cx and 2008geis the strong emission near 7300 ˚A for SN 2008ge. Onemight expect this to be [Ca II ] λλ II ] λλ II ] λλ II ] λ R e l a ti v e f λ SN 2008ge +225 dSN 2002cx +227 dSN 2005P R e s i du a l f λ [Fe II][Ca II] Ca II Figure 6.
Top panel : Late-time spectra of SNe 2002cx (red; t =227 d; Jha et al. 2006), 2005P (blue; Jha et al. 2006), and 2008ge(black; t = 220 d). The SN 2005P spectrum was obtained on 2005May 11, but no light curve has been published for this object. Bottom panel : Residual spectra created by subtracting the spectraof SNe 2002cx and 2005P from SN 2008ge (red and blue curves,respectively). The strong residuals corresponding to [Fe II ] λλ II ] λλ II NIR triplet aremarked. additional underlying lines in the spectra of SNe 2002cxand 2008ge may be different. In addition to the positiveresiduals from [Fe II ], there is a strong negative residualfrom the Ca II NIR triplet.Since the line widths of SNe 2002cx and 2008ge are verydifferent at these times, we also compare SN 2008ge toanother member of the class, SN 2005P. No light curves ofSN 2005P have been published, so the phase informationis not precise. SN 2005P was discovered on 2005 Jan 21by Burket & Li (2005) with the last non-detection from2004 July 8. A spectrum obtained on 2005 May 11 andpresented by Jha et al. (2006) is reproduced in Figure 6.Assuming that SN 2005P peaked at most 10 d after dis-covery, the spectrum has a phase of 100 ≤ t ≤
307 d.Although the phase of the spectrum might not be per-fectly consistent with that of SN 2008ge, it is somewhatsimilar (and presumably it did not evolve quickly at theselate times).The spectra of SNe 2005P and 2008ge are very similar.Although SN 2005P has strong emission at the wave-lengths corresponding to the [Fe II ] features, they arenot as strong as in SN 2008ge. Additionally, the cen-troid of the red peak is offset from that of SN 2008ge.To further examine these differences, we produce a resid-ual spectrum in the same manner as above and presentN 2008ge: Progenitor and Supernova 7the result in the bottom panel of Figure 6. The resid-ual spectrum has strong positive residuals – but not asstrong as for SN 2002cx – at the wavelengths of [Fe II ],suggesting that the SN 2008ge has stronger [Fe II ] emis-sion than SN 2005P, which in turn has stronger emissionthan SN 2002cx. The residual spectrum also has a strongnegative residual at the Ca II NIR triplet.From the residual spectrum, it is clear that the offsetin the centroid of the red peak of the 7300 ˚A featurefor SN 2005P and 2008ge is the result of weaker [Fe II ]emission and stronger [Ca II ] λλ II ] lines are easily identified as strong negativeresiduals in the residual spectrum.Nebular lines have recently been used to probe theasymmetry of SNe Ia (Maeda et al. 2010). In partic-ular, the [Fe II ] λ II ] λ II ] λ II ] λλ II ] λλ ∼ II ] λ II ] λ II ] λ − , respectively.Maeda et al. (2010) found offsets between about − − . SNe 2002cx and 2005hk both hadvelocity offsets (as measured by the [Ca II ] lines) of ∼ +300 km s − (Sahu et al. 2008). SN 2005hk had rel-atively low polarization at maximum light, indicating asmall asymmetry (Chornock et al. 2006). It is unclearif the shift of forbidden lines in SN 2008ge is indicativeof asymmetry, and furthermore, if such an asymmetrytranslates into a different late-time spectrum than othermembers of the class. DISCUSSION & CONCLUSIONS
The Progenitor of SN 2008ge
The progenitors of SN 2002cx-like objects have re-cently been a subject of debate (F09; Foley et al.2010a; Valenti et al. 2009). Fortunately, the position ofSN 2008ge, which we have shown to be a spectroscopicmember of the class, was imaged by
HST before the starexploded. Since SN 2008ge is a very nearby object, theseimages constrain the luminosity, and therefore mass, ofthe progenitor star and any possible binary companion.We have pinpointed the location of the SN in pre-explosion images that indicate that the progenitor star(or system) had M V ≥ − . σ limit), correspond-ing to a mass limit of M init . M ⊙ for a star on themain sequence and M init . M ⊙ for horizontal or red-giant branch star. The same limits apply to any potentialbinary companion.Since there is no indication of H or He in the spec-trum of SN 2008ge, its progenitor was likely a highlyevolved star such as a WD or Wolf-Rayet star. Our lim-its are only able to rule out the most-luminous Wolf-Rayet stars, corresponding to stars with initial masses of & M ⊙ (Crowther 2007, and references therein), whilethe minimum initial mass to reach a Wolf-Rayet stageappears to be ∼ M ⊙ . Observations of the host galaxy also present indi-rect constraints on the progenitor. The spectrum ofthe galaxy has no emission lines, and is well fit by asingle-stellar population model with an age of 9.5 Gyr.The surface-brightness profile of the galaxy is extremelysmooth, indicating that there are no exceptionally lu-minous stars or large star clusters near the position ofSN 2008ge. Limits from IRAS place the SFR below7 . × − M ⊙ year − . Wolf-Rayet stars are very young,and they are generally associated with high SFRs andspatially coincident with star clusters and H II regions(e.g., Hadfield et al. 2005) that should have been de-tected in the HST images if present. We see no indica-tion of (1) narrow emission lines in the SN spectrum, (2)narrow emission lines in the host-galaxy spectrum, (3)far-infrared emission from the host galaxy, (4) H I
21 cmemission from the host galaxy, (5) any luminous sourceat the position of the SN, or (6) any luminous sourcewithin the
HST field of view. This evidence greatly con-strains the environment of the progenitor of SN 2008geand is highly suggestive that it was not a massive star ofany sort.The pre-explosion imaging combined with a lack of hy-drogen in the SN spectrum rules out all stars except forsome Wolf-Rayet stars, WDs, and relatively low-massbinary stars. The additional observations of the host-galaxy make a Wolf-Rayet progenitor unlikely. A binarysystem where one star has transferred its hydrogen (andpossibly helium) envelope to its companion before ex-ploding (possibly through electron capture) is possible.A WD progenitor is also consistent with all observations.But if the progenitor of SN 2008ge was a WD, then weexpect any binary companion to have a mass less thanthe maximum mass that still allows WD formation (oth-erwise the progenitor of SN 2008ge would have explodedas a SN before reaching the WD stage). From open clus-ters, we see that some stars with M init = 6 . M ⊙ will be-come WDs (Ferrario et al. 2005). For all stages of stellarevolution, a 6 . M ⊙ star would be below our detectionlimit; therefore, this is also consistent with our observa-tions. SN 2008ge: The Supernova
Unfortunately, SN 2008ge was detected long after max-imum brightness, precluding detailed early-time obser-vations. Luckily, we were able to recover the SN on ourpre-detection images. With these images, we constructeda light curve that included maximum light.SN 2008ge peaked at M V ≈ − . V -band light curve de-clines very slowly, with ∆ m ( V ) ≈ .
34 mag. The sim-ilar peak magnitudes and ejecta velocity of SNe 2002cxand 2008ge and drastically different decline rates can beexplained if both objects generated a similar amount of Ni, but SN 2008ge had more massive ejecta. More de-tailed modeling, which is beyond the scope of this paper,is required to determine the exact physical parameters ofthese objects.The late-time spectra of SNe provide a glimpse at theinner regions of the SN ejecta. Although most SNe be-come “nebular” by 200 d after maximum, this is clearlynot the case for SNe 2002cx and 2005hk, which hadP-Cygni lines at this time, indicating a photosphere Foley et al.(Jha et al. 2006; Sahu et al. 2008). The late-time spec-trum of SN 2008ge does not clearly show any P-Cygnilines. However, the overall shapes of the spectra ofSNe 2002cx and 2008ge are very similar, and it is likelythat the composition of their ejecta are very similar, butSN 2008ge simply has a larger velocity.The nebular spectra of SNe Ia are dominated by forbid-den Fe transitions while SNe Ic are dominated by Mg I ],[O I ], [Ca II ], and Ca II features. The detection of strong[Fe II ] emission at late times is a further indication thatSN 2008ge generated a significant amount of Ni (whicheventually decayed to Fe).There is a significant offset in one of the forbidden Felines that can not be explained by simple galactic motion.This offset may be indicative of an asymmetric explosion(at least in the core) that could perhaps explain the spec-tral differences between SN 2008ge and other membersof this class at late times.
SN 2008ge in the Context of the Class ofSN 2002cx-like Objects
Valenti et al. (2009) suggested that SN 2002cx-like ob-jects, and particularly SN 2008ha, have massive progen-itors. F09 also presented massive progenitors as one ofmany possibilities for SN 2002cx-like objects in general,and SN 2008ha in particular (although Foley et al. 2010ashowed that SN 2008ha likely had a WD progenitor).Since it was unlikely that SN 2008ge had a Wolf-Rayetprogenitor and since fallback SNe require a star with M init & M ⊙ (Fryer 1999), we can rule out the fall-back scenario. The electron capture of a star in a bi-nary system that has lost its outer envelopes to a binarycompanion is still viable for SN 2008ge; however, otherobservations make this an unlikely path for the entireclass of SN 2002cx-like objects (F09). Of all models pre-sented by F09, only one is consistent with all observa-tions: a deflagration of a WD, where some SN 2002cx-like events are possibly a full deflagration of a Chan-drasekhar mass WD, some are possibly full deflagrationsof a sub-Chandrasekhar mass WD, and some are a par-tial deflagration of a WD that does not fully disruptthe star. Recent numerical models of sub-ChandrasekharWD detonations have been successful at producing lightcurves and spectra somewhat similar to normal SNe Ia(Fink et al. 2010; Pakmor et al. 2010; Sim et al. 2010;van Kerkwijk et al. 2010). These models only exploreprogenitors with relatively large total mass, and addi-tional models at lower total mass may reproduce the fea-tures of this class. (In fact, as the total mass decreases,the models predict that the SN will fall off of the Phillipsrelationship similar to SN 2002cx-like objects; Sim et al.2010.) Whatever is the correct model, it must explain thestrong [Fe II ] lines in the late-time spectra of SN 2008ge,which suggests that a significant fraction of its ejectais Fe, likely the result of radioactive decay of Ni, al-though a significant portion of the Fe may come from Fe or even possibly directly synthesized Fe.The detection of He (but not H) in SN 2007J (F09) isa further constraint for the progenitors of these objects,indicating that a significant amount of He is present inat least some of the progenitor systems. The He maycome from the WD itself or from a binary companion,requiring that the companion be a He star.The host-galaxy morphology distribution of this class of objects is heavily skewed to late-type galaxies, withonly SN 2008ge hosted by an early-type galaxy (F09).Although this distribution is consistent with SN 1991T-like SNe Ia (F09), which must have WD progenitors, itis an indication that a significant number of progenitorsystems must be relatively young.With the constraint on the progenitor of SN 2008ge, wehave a crucial limit on the progenitors of the SN 2002cxsubclass of SNe. Future observations of nearby SNe withdeep pre-explosion imaging may further constrain theprogenitors of these objects.
Facilities:
Du Pont (B&C), Gemini:South (GMOS),HST (WFPC1, WFPC2), Magellan:Baade (IMACS),Magellan:Clay (LDSS3, MIKE), PROMPTR.J.F. is supported by a Clay Fellowship. G.P.acknowledges support by the Proyecto FONDECYT11090421 and from Comit´e Mixto ESO-Gobierno deChile. G.P. and M.H. acknowledge support from theMillennium Center for Supernova Science through grantP06-045-F funded by “Programa Bicentenario de Cien-cia y Tecnolog´ıa de CONICYT”, “Programa IniciativaCient´ıfica Milenio de MIDEPLAN” and partial supportfrom Centro de Astrof´ısica FONDAP 15010003 and byFondecyt through grant 1060808 from the Center ofExcellence in Astrophysics and Associated Technologies(PFB 06).We are indebted to the staffs at the Las Campanas andGemini Observatories for their dedicated services. Weappreciate conversations with E. Berger, R. Chornock,W. High, and B. Stalder about this object. Insightfulcomments from D. Branch were very helpful.This material is based upon work supported by theNational Science Foundation (NSF) under grant AST–0306969. The Dark Cosmology Centre is funded by theDanish NSF.Based on observations made with the NASA/ESAHubble Space Telescope, and obtained from the Hub-ble Legacy Archive, which is a collaboration between theSpace Telescope Science Institute (STScI/NASA), theSpace Telescope European Coordinating Facility (ST-ECF/ESA) and the Canadian Astronomy Data Centre(CADC/NRC/CSA). Based in part on observations ob-tained at the Gemini Observatory, which is operated bythe Association of Universities for Research in Astron-omy, Inc., under a cooperative agreement with the USNational Science Foundation on behalf of the Geminipartnership: the NSF (United States), the Science andTechnology Facilities Council (United Kingdom), the Na-tional Research Council (Canada), CONICYT (Chile),the Australian Research Council (Australia), Minist´erioda Ciˆencia e Tecnologia (Brazil) and Ministerio de Cien-cia, Tecnolog´ıa e Innovaci´on Productiva (Argentina).This research has made use of the NASA/IPAC Extra-galactic Database (NED) which is operated by the JetPropulsion Laboratory, California Institute of Technol-ogy, under contract with the National Aeronautics andSpace Administration.
REFERENCESAlard, C. 2000, A&AS, 144, 363