Early and Late-Time Observations of SN 2008ha: Additional Constraints for the Progenitor and Explosion
Ryan J. Foley, Peter J. Brown, Armin Rest, Peter J. Challis, Robert P. Kirshner, W. Michael Wood-Vasey
aa r X i v : . [ a s t r o - ph . S R ] D ec Draft version October 31, 2018
Preprint typeset using L A TEX style emulateapj v. 12/01/06
EARLY AND LATE-TIME OBSERVATIONS OF SN 2008HA: ADDITIONAL CONSTRAINTS FOR THEPROGENITOR AND EXPLOSION
Ryan J. Foley , Peter J. Brown , Armin Rest , Peter J. Challis , Robert P. Kirshner , andW. Michael Wood-Vasey Draft version October 31, 2018
ABSTRACTWe present a new maximum-light optical spectrum of the the extremely low luminosity and excep-tionally low energy Type Ia supernova (SN Ia) 2008ha, obtained one week before the earliest publishedspectrum. Previous observations of SN 2008ha were unable to distinguish between a massive star andwhite dwarf origin for the SN. The new maximum-light spectrum, obtained one week before theearliest previously published spectrum, unambiguously shows features corresponding to intermediatemass elements, including silicon, sulfur, and carbon. Although strong silicon features are seen insome core-collapse SNe, sulfur features, which are a signature of carbon/oxygen burning, have alwaysbeen observed to be weak in such events. It is therefore likely that SN 2008ha was the result of athermonuclear explosion of a carbon-oxygen white dwarf. Carbon features at maximum light showthat unburned material is present to significant depths in the SN ejecta, strengthening the case thatSN 2008ha was a failed deflagration. We also present late-time imaging and spectroscopy that areconsistent with this scenario.
Subject headings: supernovae: general — supernovae: individual(SN 2008ha) INTRODUCTION
Supernova (SN) 2008ha has the lowest peak luminosityand ejecta velocity of any SN Ia yet observed (Foley et al.2009; Valenti et al. 2009, hereafter Paper I and V09, re-spectively). It peaked at M V = − . ∼ − ,and had a rise time of ∼
10 days; these values are ∼ − slower, and 10 daysshorter than that of normal SNe Ia, respectively. To-gether, these observations indicate a very low kinetic en-ergy of ∼ × ergs and an ejecta mass of 0.15 M ⊙ (Paper I).Although SN 2008ha resembles SN 2002cx, the proto-type of a class of peculiar SNe Ia with lower than normalluminosity and ejecta velocity (see Jha et al. 2006 for areview of the class), the similar energetics and spectra ofSN 2008ha and some peculiar core-collapse SNe led V09to suggest that the progenitor of SN 2008ha was a mas-sive star. In Paper I we considered several progenitorand explosion models: (1) the collapse of a massive starto a black hole with most of the star “falling back”; (2)a runaway nuclear reaction caused by electron captureon a white dwarf (WD); (3) the failed deflagration of aWD; and (4) nuclear burning of a massive He shell onthe surface of a WD in an AM CVn system (a “SN .Ia”;Bildsten et al. 2007).Delicate balancing of the energetics and the detectionof a SN 2002cx-like object in an S0 galaxy makes thefallback model unfavorable for SN 2008ha and the class, Harvard-Smithsonian Center for Astrophysics, 60 GardenStreet, Cambridge, MA 02138. Clay Fellow. Electronic address [email protected] . Pennsylvania State University, Department of Astronomy &Astrophysics, University Park, PA 16802. Department of Physics, Harvard University, 17 Oxford Street,Cambridge, MA 02138 Department of Physics and Astronomy, 100 Allen Hall, 3941O’Hara St, University of Pittsburgh, Pittsburgh, PA 15260. respectively. The electron-capture scenario predicts com-plete burning to Ni with no intermediate-mass elements(IMEs). The observations of SN 2008ha and other mem-bers of the SN 2002cx-like class are all consistent witha deflagration (with SN 2008ha being a failed deflagra-tion; Paper I). The .Ia model fits SN 2008ha, but cannot currently reproduce the luminosity required for othermembers of the class.Nucleosynthetic models provides additional informa-tion to distinguish between these models. Explosive C/Oburning produces a significant amount of S while Heburning produces more Ca (see Perets et al. 2009 for arecent examination of explosions with different fractionsof each process). Furthermore, it is expected that S iscontained in the inner layers of a core-collapse explosion,while it is predominantly in the outer layers of a ther-monuclear explosion (e.g., Thielemann et al. 1991). V09argued that the lack of S in their first spectrum was mostconsistent with a massive star origin. A strong detectionof S at early times would give support for C/O burningand a WD progenitor.The SN 2008ha spectra presented in Paper I and V09spanned the phases of 6.5 to 68.1 days past maximumlight in the B band (2008 November 12.7 UT; Paper I;UT dates will be used throughout this paper). Here wepresent two spectra which extend this range in both di-rections, with phases of − B maximum. We also present late-time photometry whichconstrains the amount of generated Ni. These observa-tions, including the detection of strong S lines in themaximum-light spectrum, further constrain the possi-ble progenitor and explosion models for SN 2008ha, andhighly favor a WD progenitor. OBSERVATIONS AND DATA REDUCTION
We have obtained two new spectra of SN 2008ha. Theearly-time spectrum was obtained on 2008 November 11(1 day before B maximum) with the Hobby-Eberly Tele- Foley et al.scope using the Low Resolution Spectrograph (Hill et al.1998). The spectrum is the combination of four 450-second exposures. The late-time spectrum was obtainedon 2009 July 1/2 (231 days after B maximum) withGemini-North using GMOS in nod and shuffle (N&S)mode (Hook et al. 2004). The GMOS spectrum is thecombination of six 1280-second (after accounting for nodtime) exposures, producing 12 spectra with effective ex-posure times of 640 seconds and a total exposure timeof 7680 seconds. The GMOS spectra were obtained withthree slightly different central wavelengths to compensatefor chip gaps. Five of the six exposures were obtained onJuly 1, with the remaining exposure obtained on July 2.Standard CCD processing and spectrum extractionwere accomplished with IRAF . The GMOS data werereduced using the Gemini IRAF N&S package. Low-order polynomial fits to calibration-lamp spectra wereused to establish the wavelength scale, and small adjust-ments derived from night-sky lines in the object frameswere applied. We employed our own IDL routines to fluxcalibrate the data and remove telluric lines using thewell-exposed continua of spectrophotometric standards(Wade & Horne 1988; Foley et al. 2003, 2006).Two epochs of r -band images were obtained on 2009June 23 and 2009 July 1/2 with LDSS3 on the Magel-lan Clay telescope and with GMOS on the Gemini-Northtelescope, respectively. These images were reduced usingstandard techniques. The flux of the SN was determinedby fitting a PSF profile in the flattened images using theDoPHOT photometry package (Schechter et al. 1993).Since all of our images presumably have some SN flux,difference imaging is not yet possible. The presence ofan underlying H II region causes our flux measurementsto be upper limits to the true SN flux. We convert the r -band instrumental magnitudes into R -band magnitudesusing the R -band catalog magnitudes of local standardstars (Paper I). No instrumental color corrections wereapplied. SPECTRAL ANALYSIS
Maximum-Light Spectrum
The maximum-light spectrum (shown in Figure 1) con-sists almost exclusively of IMEs with low ejecta veloci-ties. The minima of the Si II λ I λ ∼ − , respec-tively. The Si II feature was much weaker a week later(the first spectrum from Paper I) and could not be di-rectly measured, but the characteristic photospheric ve-locity of ∼ − for the 6 day spectrum is approx-imately half that of Si II in the maximum-light spectrum.Similarly the O I velocity is 2.5 times larger at maximumbrightness than it is at t = 6 days.The identification of IMEs is confirmed by fitting thespectrum with the SN spectrum-synthesis code SYNOW(Fisher et al. 1997). Although SYNOW has a simple,parametric approach to creating synthetic spectra, it isa powerful tool to aid line identifications (see Paper I IRAF: the Image Reduction and Analysis Facility is distributedby the National Optical Astronomy Observatory, which is operatedby the Association of Universities for Research in Astronomy, Inc.under cooperative agreement with the National Science Founda-tion. S ca l e d f λ + C on s t a n t Fe IIS IISi IIFe IIIS IIS IINa DSi IIO ISi IIC II C II O IMg II? O I/Ca II
SN 2008hat = −1 d
Fig. 1.—
Optical spectrum of SN 2008ha at t = − B maximum (solid black line) and a SYNOW model fit(red dashed line). The spectrum is dominated by IMEs. Spec-tra of SNe 1999aa, 2004aw, 2005hk, 2006gz, and 2007gr are shownfor comparison (each shifted by selected amounts to give the bestmatch to the ejecta velocity of SN 2008ha). for details). In the early-time spectrum, SYNOW unam-biguously identifies the presence of O, Na, Si, S, Ca, andFe, all common features in maximum-light SN Ia spec-tra, as well C, which is rarely seen. Although strong Si II is the hallmark feature of SNe Ia, there are examples ofSNe Ic with strong Si features (Taubenberger et al. 2006;Valenti et al. 2008); however, those objects did not havestrong S II features.The SYNOW fit of SN 2008ha is similar to thoseof the − − II for SN 2008ha (Chornock et al. 2006 did fit theSN 2005hk spectrum with C III , but the identificationwas ambiguous). SN 2008ha appears to have weaker Felines than the other objects, signaling a lower Ni yieldrelative to that of IMEs.We compare the maximum-light spectra of SN 2008hato SNe Ia and SNe Ic finding that SN 2008ha mostclosely resembles a SN Ia. Our comparison spectra(see Figure 1) include SN 2005hk, a SN Ia similar toSN 2002cx (Phillips et al. 2007), SN 1999aa, a slightlyover-luminous SN Ia (Garavini et al. 2004), SN 2006gz, aluminous SN Ia with C features in its pre-maximum spec-tra (Hicken et al. 2007), SN 2004aw, a SN Ic that wasoriginally classified as a SN Ia (its true nature was onlyrealized with nebular spectra; Taubenberger et al. 2006),and SN 2007gr, a well-observed SN Ic with a spectrumsomewhat similar to SN 2004aw (Valenti et al. 2008).The maximum-light spectrum of SN 2008ha is generallysimilar to all five comparison spectra after correcting forarly/Late Observations of SN 2008ha 3velocity differences. But the SNe Ic have weaker Si II than that seen in SN 2008ha and lack obvious S II lines.The SNe Ia (and particularly SN 2006gz) have spectrathat are more similar to the spectrum of SN 2008ha,strengthening the classification of SN Ia for SN 2008ha;however, the spectra still have differences, and it is diffi-cult to definitively place SN 2008ha in this class.The spectrum of SN 2008ha unambiguously containsthe signature of C in its ejecta, showing strong ab-sorption corresponding to C II λ II λ Late-Time Spectrum
The late-time ( t = 231 days) spectrum of SN 2008hais shown in Figure 2 along with the 227-day spectrumof SN 2002cx. We compensate for the contamination ofan underlying H II region by subtracting a linear contin-uum fit from each SN spectrum and compare the residualspectra.The lower panels of Figure 2 present the continuum-subtracted spectrum of SN 2008ha on an expanded wave-length scale. The low S/N of the spectrum prevents usfrom detecting any individual feature in a SN 2002cx-likespectrum other than possibly [Ca II ] λλ II NIR triplet (8498, 8542, and 8662 ˚A). The late-timespectrum of SN 2008ha exhibits no obvious emission from[O I ] (which is expected for SNe Ic and predicted for de-flagration models; Gamezo et al. 2003), [Ca II ], or Ca II .In fact, there appears to be absorption at the positionthat we would expect Ca II emission. As the observa-tions were performed in N&S mode, there was no localbackground subtraction performed and these features arenot an artifact of having SN light in a background region.This wavelength range has significant night sky emission,and imperfect sky-line subtraction may cause such fea-tures. Regardless, there are no definitive detections ofstrong, broad features corresponding to Ca II .Between t = 56 and 227 days, the velocity width of the[Ca II ] feature in SN 2002cx decreased from FWHM ≈ − to ∼
900 km s − . In the 62 day spectrumof SN 2008ha, [Ca II ] had FWHM ≈
900 km s − . Ifthe width of the feature decreases at the same rate inboth object, we expect the feature to have FWHM ≈
350 km s − in the late-time spectrum of SN 2008ha. Allnarrow lines have a similar redshift and velocity width( ∼
200 km s − FWHM, equivalent to the instrumentalresolution). It is possible that the velocity width is belowour instrumental resolution, and in that case we couldmistake lines from the SN as lines from the H II region(such as [O I ]). There are no intermediate-width emis-sion lines (such as H or He) that one would expect ifthe SN ejecta were interacting with a dense circumstel-lar medium associated with a massive star progenitor. Rest Wavelength (Å) R e l a ti v e f λ Fig. 2.— ( top panel ): The extracted 231 day optical spectrumof SN 2008ha (solid black line), which is contaminated by an un-derlying H II region. A linear fit to the continuum is also shown(red dotted line). ( second panel ): The continuum-subtracted spec-tra of SNe 2008ha (black line) and 2002cx ( t = 227 days; red line;Jha et al. 2006). ( bottom panels ): The continuum-subtracted spec-trum of SN 2008ha from above is reproduced on an enlarged scalewith finer resolution. The red and blue lines are the continuum-subtracted 227 day spectrum of SN 2002cx and 62 day spectrumof SN 2008ha (Paper I). The ratio of [N II ] λ α lines from the H II region is 0.046, corresponding to a metallicity (as de-termined by the N2 method of Pettini & Pagel 2004) of12+log(0 / H) = 8 . ± .
03 (stat). This is consistent withthat of a nearby H II region (Paper I) and indicative ofno significant H emission from circumstellar interaction. PHOTOMETRIC ANALYSIS
We reproduce the early-time light curves of SN 2008hafrom Paper I and V09 in Figure 3 along with our twolate-time photometric upper limits. Our late-time imag-ing resulted in R -band measurements of 21 . ± .
05 and21 . ± .
05 mag 222 and 231 days after B maximum, re-spectively. However, these images have had no templatesubtraction, so the photometry reflects the brightness ofthe combination of the SN and the underlying H II re-gion, and therefore, these numbers are upper limits onthe brightness of the SN. The differences in two measure-ments may indicate real fading between the two epochs,but can also be explained as systematic differences in theinstrument/filter responses.Paper I found that the light curves of SNe 2005hk and2008ha were well matched if the light curve of SN 2005hkwas “stretched” by a factor of 0.73. In Figure 3, we com- Foley et al. R B a nd A pp a r e n t B r i gh t n e ss ( m a g ) C o D eca y Fig. 3.— R -band light curve of SN 2008ha. The circles andsquares are from Paper I and V09, respectively. The upper limitsare from our new photometry. The red dotted line indicates thedecay expected from generating 10 − M ⊙ of Ni. The solid anddashed black lines is the light curve of SN 2005hk (Phillips et al.2007; Sahu et al. 2008) shifted to match the peak brightness ofSN 2008ha. The dashed line has been scaled to match the widthof SN 2008ha (see Paper I for details), while the solid line has notbeen modified. pare the R -band light curve of SN 2008ha to that ofSN 2005hk (Phillips et al. 2007; Sahu et al. 2008) bothwith and without this stretching. The stretched lightcurve is a good approximation of the early-time behav-ior because the SEDs are similar. But the light curvesexhibit different decline rates because the SNe had dif-ferent opacities, amounts of ejecta/ Ni, and kinetic en-ergies. The ejecta of SN 2002cx-like objects appear tobe optically thick at very late times (Jha et al. 2006;Sahu et al. 2008), so opacity effects may still dominateat t = 250 days. Regardless, the SN 2005hk light curvesshould give an indication of the expected decay rate atlate times if SN 2008ha evolves in a similar fashion.If the peak luminosity is powered by Ni decay, thenthe luminosity at late times should be powered by Co(the decay product of Ni) decay. The brightness atlate times is therefore directly related to the Ni mass.Assuming the relationship between Ni mass and bolo-metric luminosity (Sutherland & Wheeler 1984), a so-lar SED for bolometric corrections, and full γ -ray trap-ping, the late-time photometry places a limit of M Ni . − M ⊙ (see Figure 3). This value is also consistentwith the brightness of SN 2008ha at t ≈
60 days. Con-sidering the uncertainty of the SED, this is consistentwith the estimate from the early-time light curve of(3 . ± . × − M ⊙ . DISCUSSION AND CONCLUSIONS
The maximum-light spectrum provides key informa-tion for understanding the nature of SN 2008ha. Thespectrum has strong lines from IMEs, including Si and S. Nucleosynthetic models suggest that strong S featuresare indicative of C/O burning (e.g., Perets et al. 2009).Furthermore, no core-collapse SN has been observed tohave strong S lines at maximum light, and having thesefeatures in the maximum-light spectrum indicates that anon-negligible amount of S is in the outer ejecta, suggest-ing a WD progenitor (e.g., Thielemann et al. 1991). Oneof the points adduced by V09 in favor of the core-collapseinterpretation of this event was the absence of S II andthe weak Si II seen in their earliest spectrum, 8 days aftermaximum brightness. The new data presented here showthat S II is present and Si II is significantly stronger thanin those data. Although peculiar abundances in the outerlayers of a massive star may be the cause of the S lines,the data currently favor the idea of a WD progenitor.SN 2008ha shares many similarities (luminosity, spec-tral features, etc) with SN 2005E, a low-luminosity SN Ibwith strong Ca lines at late times (Paper I; Perets et al.2009. There have been several SN 2005E-like objectsdiscovered, but they exist predominantly in early-typegalaxies, in contrast to the primarily late-type hosts ofSN 2002cx-like objects (SN 2008ha was discovered in andwarf irregular galaxy), and suggestive that they haveold progenitors. Perets et al. (2009) found that by chang-ing the amount of He and C/O burning, the ratio of S-to-Ca in the ejecta of a SN can be manipulated. SNe 2005Eand 2008ha may have similar progenitors and/or ex-plosion mechanisms. The different host-galaxy popula-tions of these classes may indicate that progenitor ageor metallicity has an affect on the resulting explosion,similar to SN 1991T and SN 1991bg-like objects.To derive the previous estimates of the ejecta massand kinetic energy, the velocity at 6 days past maxi-mum brightness was extrapolated to the time of maxi-mum assuming a velocity gradient similar to that of anormal SN Ia. With these new data, the extrapolation isnot necessary, and the systematic uncertainty related tothese measurements can be reduced. The velocity of themaximum-light spectrum is twice that of the adoptedvalue from Paper I, increasing the kinetic energy andejecta mass by factors of 8 ( E ∝ v ) and 2 ( M ∝ v ),respectively. We therefore revise our estimates of the ki-netic energy to be 1 . × ergs and M ej = 0 . M ⊙ . Wenote that this analysis assumes that the composition andopacity of the ejecta of SN 2008ha are similar to those ofa normal SN Ia, which may cause systematic uncertaintyof order a factor of two (see Paper I for details).The late-time photometry is consistent with the pro-duction of a few times 10 − M ⊙ of Ni, similar to esti-mates from the early-time light curve (Paper I; V09).The late-time spectrum shows that there are no ex-tremely strong emission lines from the SN; however, therelatively low S/N spectrum places weak limits on suchfeatures.The strong C lines in the maximum-light spectrumindicates that there is unburned material far into theejecta, consistent with a deflagration (Gamezo et al.2003). The low ejecta and Ni mass are consistent witha failed deflagration of a WD that did not disrupt theprogenitor (Paper I), but are perhaps also explained bya sub-Chandrasekhar mass or more exotic WD explosion.Pre-maximum data is critical for understanding thenature of SN 2008ha-like objects. Although our obser-vations are all consistent with a failed deflagration ofarly/Late Observations of SN 2008ha 5a WD, we can not completely rule out other models. Ifthe progenitor of SN 2008ha was a very massive star, onemight expect a brightening in X-rays or radio if the pro-genitor had a wind or pre-explosion outbursts. Futureearly-time X-ray and radio observations will help con-strain the nature of similar events. Eventually we willdetect a SN 2008ha-like object with deep pre-imagingdata and either detect or highly constrain the propertiesof the progenitor.R.J.F. is supported by a Clay Fellowship. R.J.F. wouldlike to thank G. Narayan and B. Stalder for managingthe imaging data, and the Magellan and Gemini staffs forbeing extremely accommodating. Discussions during theKITP conference “Stellar Death and Supernovae” werebeneficial to this study. Supernova research at Harvardis supported by NSF grant AST09–07903.Based in part on observations obtained at the Gem-ini Observatory, which is operated by the Association ofUniversities for Research in Astronomy, Inc., under a co-operative agreement with the US National Science Foun- dation on behalf of the Gemini partnership: the NSF(United States), the Science and Technology FacilitiesCouncil (United Kingdom), the National Research Coun-cil (Canada), CONICYT (Chile), the Australian Re-search Council (Australia), Minist´erio da Ciˆencia e Tec-nologia (Brazil) and Ministerio de Ciencia, Tecnolog´ıa eInnovaci´on Productiva (Argentina). The HET is a jointproject of the University of Texas at Austin, the Penn-sylvania State University, Stanford University, Ludwig-Maximilians-Universit¨at M¨unchen, and Georg-August-Universit¨at G¨ottingen. The HET is named in honor ofits principal benefactors, William P. Hobby and RobertE. Eberly. The Marcario LRS is named for Mike Mar-cario of High Lonesome Optics who fabricated severaloptics for the instrument but died before its completion.The LRS is a joint project of the HET partnership andthe Instituto de Astronom´ıa de la Universidad NacionalAut´onoma de M´exico.
Facilities: