SN 2011fe: A Laboratory for Testing Models of Type Ia Supernovae

Laura ChomiukA,B,C

A Department of Physics and Astronomy, Michigan State University, East Lansing, MI 48824 B National Radio Astronomy Observatory, P.O. Box O, Socorro, NM 87801 C Email: [email protected]

Abstract: SN 2011fe is the nearest supernova of Type Ia (SN Ia) discovered in the modern multi- wavelength telescope era, and it also represents the earliest discovery of a SN Ia to date. As a normal SN Ia, SN 2011fe provides an excellent opportunity to decipher long-standing puzzles about the nature of SNe Ia. In this review, we summarize the extensive suite of panchromatic data on SN 2011fe, and gather interpretations of these data to answer four key questions: 1) What explodes in a SN Ia? 2) How does it explode? 3) What is the progenitor of SN 2011fe? and 4) How accurate are SNe Ia as standardizeable candles? Most aspects of SN 2011fe are consistent with the canonical picture of a massive CO white dwarf undergoing a deflagration-to-detonation transition. However, there is minimal evidence for a non-degenerate companion star, so SN 2011fe may have marked the merger of two white dwarfs. Keywords: supernovae: general — supernovae: individual (SN 2011fe) — white dwarfs — novae, cataclysmic variables

1 Introduction 2 SN 2011fe: A normal SN Ia

The multi-band light curve of SN 2011fe, measured in Discovered on 2011 August 24 by the Palomar Tran- exquisite detail at UV through IR wavelengths, is typ- sient Factory, SN 2011fe1 was announced as a Type ical of SNe Ia (Figure 2; Vink´oet al. 2012; Brown Ia supernova (SN Ia) remarkably soon after explosion et al. 2012; Richmond & Smith 2012; Munari et al. (just 31 hours; Nugent et al. 2011a,b). SN 2011fe is 2013; Pereira et al. 2013). In the B-band, the light nearby, located in the well-studied galaxy M101 at a curve declines by ∆m15 = 1.1 mag in 15 days (Rich- distance of ∼7 Mpc (Figure 1; Shappee & Stanek 2011; mond & Smith 2012; Pereira et al. 2013). To power the 56 Lee & Jang 2012). As the earliest and nearest SN Ia light curve of SN 2011fe, ∼0.5 M of Ni is required discovered in the modern multi-wavelength telescope (Nugent et al. 2011b; Bloom et al. 2012; Pereira et al. era, SN 2011fe presents a unique opportunity to test 2013). This 56Ni mass is quite typical for SNe Ia (How- models and seek answers to long-standing questions ell et al. 2009). about SNe Ia. In addition, time-resolved optical spectroscopy shows SN 2011fe to be a spectroscopically-normal Such a testbed has been sorely needed. SNe Ia SN Ia (Parrent et al. 2012; Pereira et al. 2013; Maz- are widely used by cosmologists to measure the ex- zali et al. 2013). SN 2011fe can be classified as “core pansion parameters of the Universe, and led to the normal” in the schemes of Benetti et al. (2005) and Nobel Prize-winning discovery of dark energy (Riess Branch et al. (2006). et al. 1998; Perlmutter et al. 1999). However, impor- In all relevant details, SN 2011fe appears to be a tant unknowns remain that stymie the use of SNe Ia normal SN Ia. It has, therefore, been taken to be as precise cosmological tools. The progenitor systems representative of its class. Conclusions reached for of SNe Ia are poorly understood, and the explosion SN 2011fe may be extrapolated to SNe Ia generally, mechanism itself is elusive (reviews by Hillebrandt & but we must be careful in this extrapolation, remem- Niemeyer 2000; Livio 2001; Howell 2011; Wang & Han bering that SN 2011fe is only one object. A number of arXiv:1307.2721v1 [astro-ph.HE] 10 Jul 2013 2012 contain more details). recent studies have shown that SNe Ia are diverse, and This paper reviews the significant body of work that differences in their observational properties may already published on SN 2011fe as of mid-2013. I focus imply real variety in progenitor systems and explosion on five questions: (i) Is SN 2011fe a normal SN Ia? (ii) mechanisms (e.g., Foley et al. 2012; Wang et al. 2013) What exploded in SN 2011fe? (iii) How did it explode? . (iv) What is the progenitor of SN 2011fe? and (v) How accurately can we use SNe Ia as standard candles? A concise summary follows in Section 6. 3 What exploded in SN 2011fe?

While it has long been thought that a SN Ia marks 1The source was originally dubbed PTF 11kly. the complete disruption of a white dwarf (e.g., Hoyle &

1 2 Publications of the Astronomical Society of Australia

Figure 1: Images of M101 obtained on three successive nights, from left to right: 2011 Aug 23.2, Aug g 24.2, andFigure Aug 25.2 1: PTF UT. The-band green image arrow sequence points of the to field SN 2011fe, of Messier which 101 was showing not detected the appearance in the of first image but subsequentlySN 2011fe. brightened From left to dramatically. right, images are Figure from August from 23.22,Nugent 24.17, et al. and (2011b), 25.16 UT. reprinted The supernova by permission from Macmillanwas not detected Publishers on the Ltd: first Nature, night to copyright a 3-s limiting 2011. magnitude of 21.5, was discovered at magni- tude 17.35, and increased by a factor of 10 in brightness to mag 14.86 the following night. The supernova peaked at magnitudeVink´oet al.:9.9, Distance making to M101 it the from fifth SN 2011fe brightest supernova in the past century. ⇠ PTF is a wide-field optical experiment designed to systematically explore the variable sky on a 9.5 9.5 Table 5. SALT2 best-fitting parameters 22, 23 10 variety of timeB scales, with 10 one particularV focus theon very knowing early detection the precise of SNe time of. theDiscoveries explosion. This is 10.5 10.5 11 11 estimated to be UT 2011 August 23 16:29±20 minutes 11.5 such as this one have been 11.5 made possible by coupling real-timeParameter computational Value tools to extensive 12 12 24, 25 t (JD) 2,455,815.505 (0.047) 12.5 astronomical follow-up 12.5 observations . by Nugent0 et al. (2011b). They derive this time by 13 13 m$ (mag) 9.959 (0.027) 13.5 13.5 fittingB a power law to the early optical light curve, 14 14 x1 0.296 (0.049) Observed magnitude 14.5 Observed magnitude 14.5 followingc the expectation−0.030 (0.018) that the SN luminosity L will -20 -10 0 10 20 30 40 50 -20 -10 0 10 20 30 40 50 − increaseµ0(G10) as the (mag) area 29.034 of the (0.078) optically-thick photosphere, Days from B-maximum Days from B-maximum 2 producingµ0(K09) the (mag) relation 29.068L (0.062)∝ t . This simple model fits 9.5 10 µ (final) (mag) 29.05 (0.08) 10 R 10.5 I the photometry0 very well over the first four days (see 10.5 11 11 11.5 Notes. ErrorsFigures are given 3 and in parentheses. 4). The final distance modulus (last 11.5 12 12 row) is the average of the G10 and K09 estimates. 12.5 12.5 The modeling of the shock breakout is aided by the 13 13 13.5 13.5 serendipitous availability of optical imaging of M101 14 14 Observed magnitude 14.5 Observed magnitude 14.5 Contrarythat hadto MLCS been,the obtainedSALT2 model a mere does four not explicitly hours after in- Nugent -20 -10 0 10 20 30 40 50 -20 -10 0 10 20 30 40 50 Days from B-maximum Days from B-maximum clude theet distance al.’s estimated or the distance time modulus, of explosion, thus, it must but bebefore de- the SN rived fromwas the actually fitting parameters. discovered We followed (Bloom two et slightly al. 2012). dif- These Figure 2: BVRI LightSALT2 curves for SN 2011fe mea- Fig. 7. The fitting of LCs to the observed data of ferent proceduresdata, obtained for this: thewith one The presented Open by University’s Guy et al. (2010) 0.4-m tele- suredSN 2011fe. in the Johnson-Cousins system (Vega magni-(G10) and an independent realization given by Kessler et al. scope, yield a robust non-detection at$ this epoch. The tudes). Best-fit SN Ia templates from SALT2 are(2009) (K09). Starting from the fitting parameters mB, x1 and c, first detection of SN 2011fe was made 11 hours after plotted as black lines. Figure from Vink´oet al.the distance modulus µ0 in the G10 calibration can be obtained 3.3.1. MLCS2k2 as Nugent et al.’s estimated explosion time. (2012), reproduced with permission c ESO. $ The faint optical flux at very early times places The fitting of Eq. 2 was performed by using a simple, self- mBB = mB 0.008( 0.005) x1 + 0.013( 0.004) χ2 strong− constraints± · on the shock± breakout signal, imply- developed -minimization code, which scans through the al- CC = 0.997( 0.097) C + 0.002( 0.009) x1 + 0.035( 0.008) lowed parameter space with a given step and finds the lowest ing that± the exploding· ± star was· compact,± with a radius 2 s = 0.107( 0.006) x1 + 0.991( 0.006) 56 χ within this range. The fitted parameters were the moment of R? . ±0.02 R· (Figure± 3). From the measured Ni FowlerB-maximum 1960), (t ), little the V direct-band extinction proof existedA ,theLC-parameter until SN 2011fe.∆ µ0 = mBB MB + a (s 1) b CC, (5) 0 V mass,− we know· − the− star· was &0.5 M . Only degen- and the distance modulus µ0,withstepsofδt0 = 0.1, δAV = 0.01, where weerate have stars adopted satisfyMB = these19.218 mass0.032, anda = 1 radius.295 0.112 constraints. δ∆ = 0.01 and δµ0 = 0.01, respectively. At the expense of longer − ± ± 2 and b =Thus3.181 the0.131 exploded (G10). body in SN 2011fe must have been 3.1computation Radius time, this ofapproach the maps exploded the entire χ hypersurface star ± and finds the absolute minimum in the given parameter volume. Thea K09 neutron calibration star applies or white a simpler dwarf. formula: There are no plau- TheWe radius have of fixed an the exploding reddening-law star (Rparameter?) can as beR estimatedV = 3.1 sible$ mechanisms for producing SN Ia-like thermonu- µ0 = mB M0 + α x1 β c, (6) throughappropriate the for phenomenonMilky Way reddening, of shock although breakout: several rece whennt clear− yields· from− · a neutron star (e.g., Jaikumar et al. theresults SN suggest shock that emerges some high-velocity from the surface SNe Ia can of the be bet- star,where M2007).0 = 19 Radius.157 0.025, constraintsα = 0.121 therefore0.027 and provideβ = 2.63 strong ev- ter modeled with significantly lower R (Wang et al., 2009; . − ± ± ± it produces a distinctive photometricV signature. Shock0 22 haveidence been adopted that SN from 2011fe K09. marked the explosion of a white Foley & Kasen, 2011). Since SN 2011fe suffered from only mi- The uncertainties in the formulae above were taken into ac- breakoutnor reddening is expected and most of to it appear is due to as Milky an early-timeWay dust (see excess be- count bydwarf a Monte-Carlo (Bloom et technique: al. 2012). we calculated 10,000 dif- inlow), the it lightis more curve, appropriate with tothe adopt luminosity the galactic reddening and durationlaw. ferent realizationsSeveral of recentthe above papers parameters suggest by adding that Gauss this initialian anal- ofNevertheless, the excess because scaling of the with low the reddening, radius the of value the of explodingRV has randomysis numbers may having be too standard simplistic deviations (Piro equal 2012; to the Pirouncer- & Nakar starnegligible (Rabinak effect on& theWaxman final distance. 2011; Kasen 2010; Piro et al.tainties2012, above,2013; to the mean Mazzali values et of al. all parameters,2013). Since and dertheived bolometric 2010).The best-fitting MLCS2k2 model LCs are plotted together µ0 fromluminosity each randomized of a SN set of Ia parameters is powered applying by 56Ni Eq. at 5 and early times, with the data in Fig. 5. The final parameters are given in Table 4. 6. Then the average56 and the standard deviation of the resulting The shock breakout constraint depends crucially2 if the Ni is not mixed into the outermost regions of The 1σ uncertainties were estimated from the contour of ∆χ = sample of µ0 values are adopted as the SALT2 estimate for the 1correspondingto68%confidenceinterval.Fig.6showsthe distance modulus and its uncertainty. Table 5 lists the best-fitting map of the the χ2 hypersurface and the shape of the contours parameters and errors. The final SALT2 distance modulus were around the minimum for the two key parameters ∆ and µ0.Itis obtained as an unweigthed average of the two values from the seen that µ0 is strongly correlated with ∆,whichisthemajor G10 and K09 calibrations. source of the relatively large uncertainty δµ = 0.07 mag, despite the very good fitting quality. Note that the best-fitting MLCS2k2 template LC corresponds 4. Discussion to ∆m15(B) = 1.06 0.06, which is slightly lower than the value The application of MLCS2k2 and SALT2 LC-fitters resulted in derived by Richmond± & Smith (2012) (1.21 0.03), but more distance moduli of µ0(MLCS2k2) = 29.21 0.07 and µ0(SALT2) similar to the one given by Tammann & Reindl± (2011) (1.10 ± ± = 29.05 0.08, respectively. 0.05). Again, all these parameters are consistent with the finding It is± seen that there is a 2σ disagreement between these that SN 2011fe has a nearly perfect fiducial SN Ia LC. two values. Taking into account∼ that these distance moduli were obtained by fitting the same, homogeneous, densely sampled, 3.3.2. SALT2 high-quality photometric data of a nearby SN, and both codes provided excellent fitting quality, this ∆µ0 = 0.16 mag disagree- The SALT2 fitting was computed by running the code as de- ment is rather discouraging. Note that the difference exists de- scribed on the SALT2 website. The fitted parameters provided spite correcting the results from both codes to the same Hubble- $ 1 1 by the code are: mB (rest-frame B-magnitude at maximum), x1 constant, H0 = 73 km s− Mpc− (see above). (stretch) and c (color). Note that SALT2 restricts the fitting to Asimilardifference of µ0(SALT2) µ0(MLCS2k2) = 0.2 data obtained no later than 40 days after maximum. mag was found by Kessler et al. (2009)− for the “Nearby SNe”±

6 journals.cambridge.org/pas 3 4 Piro, A. L. and Nakar, E. The Astrophysical Journal Letters,744:L17(5pp),2012January10 Bloom et al.

Table 1 Primary Radius and Density Constraints

a Mp Rp,max Avg. ρp Lshock,max Tshock,max 3 1 (M )(R )(gcm− )(ergs− )(K) " " Shock breakout—Rabinak et al. (2011)b 0.5... 0.022 6.29 104 4.50 1039 6024 × × 1.0... 0.020 1.77 105 4.48 1039 6043 × × 1.4... 0.019 2.93 105 4.47 1039 6053 × × 3.0... 0.017 9.16 105 4.46 1039 6075 × × Ejecta heating secondary—Kasen (2010)c 0.5... 0.038 1.26 104 3.94 1039 8048 × × 1.0... 0.027 7.05 104 3.96 1039 7388 × × 1.4... 0.023 1.58 105 4.00 1039 7103 × × 3.0... 0.017 9.29 105 4.18 1039 6531 × × Shock breakout—Piro et al. (2010)d 0.5... 0.016 1.73 105 5.27 1039 12110 × × Figure 2. Absolute g-band magnitude vs. time since explosion in three 1.0... 0.019 2.07 105 5.26 1039 12091 × × theoretical models for the early-time shock-heated evolution of Type Ia SNe. 1.4... 0.021 2.26 105 5.26 1039 12082 Figure 3: Early-time optical light curve for FigureIG 4: The limits on a× shock breakout× signature Shown is 4 hr, 5σ non-detection discussed in Section 2.2 and the first two 3.0...F .3.—Comparisonoftheearly 0.025 2.75g-band105 data (Nugent5.25 et10 al.39 2011) and a12062 SNdetections 2011fe from plotted PTF (Nugent as et black al. 2011 dots.). The black The line light shows curvethe L t2 andnon-detection R depend upper limit on (Bloom the estimated et al.× 2012) to theoretica explosion×llightcurvesfrom date. In radioactive? heating (dashed curves) and shock-heated cooling (solid curves) radioactive-heating behavior seen in later-time2 PTF data, consistent with∝ the is wellFIG fit.2.—AsummaryofthefitstoSN2011fe.Thetoppanelshowsthe with a simple L t power law (dot- bothcalculatedNotes. panels, according the to Piro measured et al. (2010). early-time This does not inc lightlude the curve suppres- of non-detection.inferred bolometric For the Kasen luminosity (2010) (filled companion∝ circles) interaction and fit bolo model,metricR luminos-denotes ted black line). Three variations of shock break- siona of the shock-heated cooling (or “drop out”) that occurs when the dif- the separationity (solid distance curve), which between is composed the two stars, of L and56 (dashed the light curve) curve and is shownLtail (dotted for an SN5 2011feσ limit assuming is shown the 4 hr as non-detection filled circles (see the and text) and an shock upper- opacity κ fusion wave2 moves1 into ideal gas dominated material (Rabinaketal.2012). = outobservercurve). models aligned The with middle are the plotted panel collision shows axis, as the colorwhich blue temperature, produces lines the(Rabinak and brightest thebottompanel observed & The0.2 top cm panelg− is. roughly the explosion time inferred from a t2 extrapolation. shows M = L /!. limitb arrow. The shock breakout signature is a luminosity. 56 56 TheAssumes bottom panelfp assumes0.05 and thatE the51/M explosionc 1. Extremum occurred 0.5days values earlier, of fp (0.03–0.13) for Waxman 2011; Kasen 2010; Piro et al. 2010), and sold black line,= and the SN= light curve, powered whichchange theR radiusp,max constraintby no more is a than factor 20% of 1. from9larger. that given. Fixing E51 1yields are showndays. However, for different assuming radii slightly of the different exploding power-laws star. pro- 56 = byan RNi,p,max about is a 50% dashed smaller line. at M The0.5 M topand about panel two shows times larger at (given the temperature) was consistent with the non-detection. contrast, T changes with explosion= time" because an explo- Assumingduces fits that with thesimilar explosion quality and time results can in explosion be deter- times theM model3.0 Mc . of Bloom et al. (2012, as in Figure 3), Figure 2 shows the results for a selection of different analytical−0.22 sion= further" in the past implies more expansion at any given minedthat from vary bythe about simple1day.Sincetheoretically power-law fit (Nugentv ett al. is c The radius derived is the separation distance and the limit derived is models,the preferred assuming velocityE51/M≈ profilec 1(constantexplosiveyieldper we consider JD 2455796∝ .6tobe assumingtime and thus an a explosion smaller Tc.Thismeansthatanadditionalcon- date of UT 2011 August 23 2011b), the non-detection= was obtained four hours assuming the brightest possible viewing angle. The radius limit comes from unitthe mass), mostf likelyp 0 explosion.05, κ0.2 time1, with and an a uncertainty variety of of values roughl ofy 16:30straintthe requirement and on the constraining explosion that primary time size R could? must. 0 be.02 smallermade R via than. The a the temperature semimajor bottom axis of = = measurement, although this requires detailed spectral model- afterRp.ThePIRATEobservationat4hristhemostconstraining explosion0.5day.Thisisactuallyverysimilar(within0 and implies a size R? . .01days)ofthe.02 R panelthe binary. is for an explosion date of UT 2011 August

±non-detection by Nugent et al. (2011). In the bottom panel we ingd that is outside the scope of this work (see the discussion4/3 fordata the point, exploding which limits star.Rp Figure! 0.02 R from.Table Bloom1 summarizes et al. 23Using 4:30, their yielding Equations a(35) 12-hour and (36) but long corrected dark by a factor phase of 7− and(L)and present the velocity data along with our" best-fit velocity evo- of tmin1/3 in Piro & Nakar 2012). (2012),the detailed reproduced radius constraints by permission under different of the assumptions AAS. of 7− (Teff ) to fix the improper scalings. lutions. Open symbols indicate data that were not used for the a larger radius constraint R? . 0.04 R . Figure progenitor mass and under the different models. 56 fit because they are near peak where the velocity profile is not from3.2. PiroRadius & Constraints Nakar (2012), and Shallowest reproducedNi for by SN permis- 2011fe The expressions we have used for the early luminosity hold expected to be a power law. sionUsingTemperature–radius of the the authors. data from.Nugentetal. Non-detections (2011) of a quiescent and a non- counter- only underFor any the given assumption explosion that time radiation we can energylook for dominates the 56Ni that in detectionpart in Hubble7hrs Space earlier, Telescope Bloometal.(HST)imagingyieldaspecific (2012) argued that the post-shock ejecta. In fact, the diffusion wave will eventually ≈ produces the observed luminosity. The results from fitting the theluminosity progenitor (Lν of)constraintatcertainopticalfrequencies.With SN 2011fe had a radius ! 0.02R by the ejecta, it will take some time for light to diffuse ! recedephotometric into higher observations density regions are presented of ejecta in where Figure gas 2. pressure In this usingthe assumption models of of shock-heated a spectrum of cooling the primary, (Piro these et al. limits 2010; can be out. The diffusion time could lead to a “dark phase” dominates.particular The case luminosity we use the is time then of expected the non-detection to drop for suddenly; the ex- Rabinakturned intoet al. a 2012). limit on But the this bolometric assumed luminosity that the time (L of). Li ex- (2011) betweenRabinakplosion et explosion al. time, (2011 which and)showthat,forconstantopacity,thetime is optical sufficiently rise, close lasting to our a preferred few hours time explodedplosion could star be in accurately SN 2011fe. determined from extrapolating t2 56 considered mostly spectra of an unseen secondary, using model toof thisdaysthat drop thedepending qualitativeis proportional on features the to radialR arep,whicheffectivelylimitsthe unchanged. profile of In theNi top (Piro panel backinput in spectra time. As of emphasized red giants in to Piro derive & NakarL constraints. (2012), this For is a high &minimal Nakarwe compare progenitor 2012, 2013).the radius inferred that bolometric we are capable lightcurve of probing. (filled circl Fromes) not generally a robust method for finding the explosion time to the model fit (solid curve). We also plot the contributions 3.2(seeeffective also Abundances §4), temperature so it is worth primary, revisiting in here the the we radius consider exploded constrain a blackbodytfor theirSupplementing expression for tdrop lightwe curves find this with minimal spectroscopic radius to be in- as the input spectrum and solve for the bolometric luminosity from direct heating L56 (dashed curve) and the diffusive tail arangeofexplosiontimes.star formationL (dotted about curve). the The velocity direct0 evolution heating.66 0.56 component0. of15 the is photo- larger at andIn Figure effective 3 we radius plot the using early the data Stefan–Boltzmann and non-detection law upper (see also tail Rp,min 0.013 t4hrE51− Mc f0.05 R , (2) spherelate can times help and≈ accountthe diffusive for tail this is stronger effect. at Piro early" & times Nakar when Early-timelimitLiu foret al. SN optical2011 2011fefor spectra for a similar two differentshow analysis). significant explosion We perform carbon times. Theand a similar 56 (2012)wherethet use4hrNi spectroscopic ist/ less4hr.Thevalueof abundant. measurements Nevertheless,Rmin isL from justtail is smaller Parrentnever more than oxygentheoreticalanalysis features with curves the at includeChandra a range radioactiveX-ray of velocities non-detection, heating (dashed(7 000–30 convolving curves) 000 dif- = 6 etour al. limitsthan (2012) a factoron toRp of refinedetermined two greater the explosion in than TableL56.Thismeansthatatleast1 date,suggestingthatthe of SN 2011fe andferent shock-heated−1 input blackbody cooling spectra (solid curves). to find a The radius first limit. thing At to 10 K, km s ) (Nugent56 et al. 2011b; Parrent et al. 2012;3 backwardbreakdownvery roughly, to of UT radiation the 2011 observed energyAugust bolometric domination 23 02:30 luminosity (with is not can a likely con-be use tod notefor is example, that Ni the cannot limits always (1σ ,2 beσ present,and3 atσ the)are1 earliest.2 times10− R , 56 56 Pereira et al.3 2013; Mazzali et3 al. 2013). Neutral× car- " undermineto infer our theresults. distribution of Ni, and that Ni must be present, and1.5 still10 produce− R ,and1 the observed.8 10− lightcurves.R . In the bottom servative uncertainly of 0.5 day). This explosion time bon is also observed in IR56 spectra (Hsiao et al. 2013). Theatearly least inphotometry some amount, of SN at 2011fe the depths also that tightly are constrainsprobed by the panelIn× we Figure had to3" cut, we off show the ×Ni these for times primary-star" earlier than constraints 0.9days as a is 14earliest hours emission. earlier than that of Nugent et al. (2011b). SNafter 2011fe explosion is certainly in order to not not alone overpredict amongst the g-band SNe upper Ia in Usingnature aof different a possible spectroscopic companion star. data The set interaction and modeling of the SN function of effective temperature and average density. Primary ejecta withThe middle a companion panel of star Figure produces 2 shows emission the inferred which color depends tem- showinglimitstars reported with carbon average in Bloom in densityits et spectrum, al. (2012).less than although(Inρ the top10 it4 panelgcm isthe no 3 and strategy,perature MazzaliT ,andthebottompanelshowsthemassof et al. (2013) find an explosion time56Ni best-studied56Ni cut-off is example. needed.) This In impliesrecent that years, forp earlier a rash explosi of stud-on− linearly on thec separation distance (Kasen 2010). This emission effective temperatures larger than 106 K(at=ρ 1012 gcm 3) thatabove is more the thandiffusion 33 wavehours depth, before given Nugent by et al.’s esti- times there is a sharp cut-off in the 56Ni distributionp near the − will be anisotropic and vary by a factor of 10 depending on iesare have excluded. found C ii in many SNe Ia spectra,= provided mate: UT 2011 August 22 07:00. ∼ thatdepth observations that generates are the luminosity obtained of early the first in the detected explosion light. the orientation. AssumingM56( thet)= companionL56(t)/!(t). star in Roche-lobe(14) 56 These results highlight the uncertainties in con- (e.g.,This is Parrent not unexpected et al. 2011; since FolatelliNi probably et does al. 2012; not extend Silver- to overflow,This issuch roughly that its independent radius is ! of1/ the2 of explosion the separation time because distance, it the very3. surface COMPARISONS and the earliest TO PRIMARY emission will CANDIDATES be due to the strainingand thatis just the the set observer’s by radius the bolometric ofviewing the exploded luminosity angle is unfavorable star at any with given (such shock time. that In mandiffusive & Filippenko tail. In §3.5 2012). we further Nugent discuss et what al. (2011b) depth in andthe breakoutthe light curve models. is fainter An earlier by a factor explosion of 10 time from and its maximum) longer ParrentAccepting et al. 0 (2012).5 M interpretas a conservative the presence lowerof limit C forin the " darkour data phase restrict translate the companion into less star stringent radius to radiusRc ! 0 lim-.1 R . early-timeprimary mass, spectra low-mass of SN 2011femain-sequence as evidence stars, that brown car- dwarfs, itsUnless from the early-time time since explosion non-detections for the PIRATE (Figure data 4). is vastly If " bonand and planets oxygen are arenot viable.the unburnt In Figure remains3, we of show the the ex- main SNunderestimated 2011fe has a(by 24-hour" day), long this apparently dark phase, excludes then the Roche-lobe pho- plodedsequence star. of They stably conclude H-burning that stars the star with that mass exploded 0.5, 1, 1.4, tometryoverflowing presented red giant by and Bloom main-sequence et al. (2012) companions only limits to high asand SN 2011fe 3 M .Thehydrogenmainsequence,shownusingsolar- was likely a CO white dwarf, as commonly significance. metallicity" isochrones from Marigo et al. (2008), is excluded as R? to .0.1 R (Piro & Nakar 2012). As illustrated expected for SNe Ia (e.g., Livio 2001). in Figure 3 of Bloom et al. (2012), this less strin- Mazzali et al. (2013) use observations of the gent limit could accommodate unusual non-degenerate 3fastest-moving outermost layer of ejecta (v > 19 400 stars, such as carbon or perhaps helium stars, as the km s−1) in SN 2011fe to estimate the metallicity of 4 Publications of the Astronomical Society of Australia The Astrophysical Journal Letters,752:L26(7pp),2012June20 Parrent et al.

(b)

(a)

Figure 3. SymbolsFigure are 5: for The clarity distribution within the plot and of ionsdifferent in ions the are ejecta represented of SN by different2011fe. colors, The as larger labeled backgroundin panel (b). (a) Based panel on (a) our time plots series the of synthetic SYNAPPS spectra, vmin estimates of each ion are plotted vs. time. The dashed line represents the location of our best modeled photosphere. (b) Here we reproduce panel (a), butminimum include the velocity Doppler widths measured of our fits. for Color a given intensities ion are as mapped a function to the normalized of time. sum Points of a single are color-codedabsorption line component to different for each ions ion in every fit. Therefore,as panel shown (b) only in the depicts round where panellight has (b). been scattered Panel (b)by a particularshows the ion. Duevelocity to telluric range features, observed we have onlyfor includeddominant our fitted ions. Doppler The widths for 1 1 the high velocitycircular O i component. white dotted The numbered lines rings mark (dashed-white) radial increments are in increments of 5 of 000 5 kkm km s− . s− , spanning ejecta velocities of 0 to 1 (A color version30 000 of this km figure s− is. available Figure in from the online Parrent journal.) et al. (2012), reproduced by permission of the AAS.

C II 6578 O I 7774 the progenitor. Most of this material is carbon, but 3.3 Mass of the exploded star the remaining 2% of the mass should represent1.2d the heavier elements in the progenitor white dwarf. By SN Ia models would be strongly constrained if we could modeling an Fe-group absorption feature at ∼4800 A˚ determine whether white dwarfs-16.4d must reach the Chan- in conjunction with HST UV spectroscopy, Mazzali1.5d drasekhar mass to explode, or if sub-Chandrasekhar explosions are common. Unfortunately, estimates et al. find a metallicity of ∼0.25–0.5 Z for the2.6d out- ermost ejecta. Foley & Kirshner (2013) also use UV of the ejected mass are challenging-16.1d to achieve at spectroscopy to argue that the progenitor of SN 2011fe the necessary accuracy (e.g., Mazzali et al. 1997; Stritzinger et al. 2006). Uncertainties of <15% are had sub-solar metallicity. M101’s gas-phase metallic-4.5d -15.0d ity, measured at the galactocentric radius of SN 2011fe, needed to distinguish between Chandrasekhar and sub- is ∼0.5 Z (Stoll et al. 2011)—consistent with5.1d esti- Chandrasekhar models, an extremely difficult task -13.1d mates for SN 2011fe’s progenitor system. However, given the diversity of elements, densities, and ionic it is important to keep in mind that SNe Ia show7.2d a states in the ejecta of SNe Ia. Mazzali et al. (2013) estimate-12.5d ∼1.1 M of mate- range of delays between the formation of the progeni- −1 tor system and explosion (Maoz & Mannucci 2012);7.5d it rial is ejected at speeds >4 500 km s in SN 2011fe; would not be surprising if there was an offset between this determination is a model-dependent-10.4d lower limit, the metallicity of SN 2011fe and the current gas-phase8.3d as it does not account for-10.0d the slowest moving mate- rial. Future work on nebular spectra may lead to a metallicity in the region. 10.4d Scaled Flux + Constant more complete census of the-9.3d ejecta mass in SN 2011fe (e.g., Stehle et al. 2005). In the meantime, the 56Ni 15.5d mass places a secure lower limit-7.2d on the ejecta mass in The metallicity of the progenitor may be important16.5d SN 2011fe, Mej > 0.5 M . -2.0d in shaping a SN Ia, perhaps affecting the yield of 56Ni 19.5d -1.0d (Timmes et al. 2003; Jackson et al. 2010) and also de- +1.9d termining the observed spectral energy distribution20.6d in 4 How did it explode?+3.1d+3.0d the rest-frame UV, where high-redshift observations of SNe Ia commonly take -20000 place (e.g., -10000 H¨oflich et0 al. 1998; The-20000 volume and-10000 quality0 of data available on SN 2011fe Lentz et al. 2000; MaguireDoppler et al. 2012).Velocity The [km/s] measure- enableDoppler us to Velocity explore [km/s] the details of the white dwarf’s ments in SN 2011fe are an important data point for destruction. What is the relative significance of sub- Figure 4. “Snap-shots” of unburned material, i.e., spectroscopic signatures of unburned material as best traced by C ii λ6578 and possibly O i λ7774. On the left, we testing the predicted effects of metallicity on observed sonic deflagration and super-sonic detonation fronts? compare our SYNAPPS fits (red line) to the observations (black line) that show a feature near 6300 Å. At 1.2 days after texpl the C ii absorption minimum corresponds 1 to a DopplerSNe velocity Ia. of 16,000 km s− , at the upper reaches of photospheric velocities.Might On the right, theexplosion we plot the same have but been instead triggered for the O i byλ7774 a detona- absorption feature. Our fits (blue line) place O i in both photospheric and higher velocity regions. The gray band highlights the early evolution of the O i feature and the symbol indicates the location of telluric features. ⊕ (A color version of this figure is available in the online journal.)

5 journals.cambridge.org/pas 5

Figure 6: Top-Left Panel: Density distribution of material in a delayed detonation model 100 s after explosion. Top-right panel: Density distribution in a white dwarf merger model 100 s after explosion. Bottom Panels: Optical spectra of SN 2011fe during four epochs spanning 6–27 days after explosion (one epoch per row). Observed spectra are shown in red, and compared with model spectra for the delayed detonation scenario (left column) and the white dwarf merger (right column). Black lines represent models averaged over all viewing angles, while grey lines show model spectra from 25 different viewing angles. Figure from R¨opke et al. (2012), reproduced by permission of the AAS.

tion on the white dwarf’s surface? The distribution of below the white dwarf’s surface. Dredge-up of 56Ni newly-synthesized elements within the ejecta can con- to this height may present a challenge to standard de- strain the explosion mechanism of SNe Ia. layed detonation models: these posit ignition at many points and result in relatively symmetric explosions. Parrent et al. (2012) obtain a time series of op- The observed distribution of 56Ni requires strong mix- tical spectra and fit them using the software package ing, as might be provided by an asymmetric deflagra- SYNAPPS in order to identify the ions contributing to tion ignition in a delayed detonation scenario (Maeda each spectrum. They measure variations in velocity et al. 2010) or bubbles seen in models of gravitationally of each ion’s features with time, and map the veloc- confined detonations (Meakin et al. 2009). However, ity range of each ion in Figure 5, ranging from 5 000 such highly asymmetric models generally conflict with to 30 000 km s−1. Figure 5 shows that the ejecta observations of SNe Ia (Blondin et al. 2011) and with of SN 2011fe are well mixed, with Si, Ca, Fe, and O spectropolarimetric observations of SN 2011fe (Smith present throughout much of the ejecta. et al. 2011, see below for more discussion). Alterna- As discussed in Section 3.1, the early-time light tively, a double detonation scenario (where a He-rich curve contains information about the radial distri- shell detonates on the surface of the white dwarf and bution of newly-synthesized 56Ni. By modeling the drives a shock inward, inducing nuclear burning of the light curves and velocity evolution of SN 2011fe, Piro entire white dwarf) might also explain the presence of (2012) and Piro & Nakar (2012) find that 56Ni must be Fe-group elements at the outer edges of the ejecta (Piro present in the outer ejecta, constituting a mass frac- & Nakar 2012). However, double detonation models −2 tion of a few percent at a mass depth of just 10 M 6 Publications of the Astronomical Society of Australia also struggle to match the observed spectra of SNe Ia spectropolarimetry of SN 2011fe is difficult, the obser- (Kromer et al. 2010; Woosley & Kasen 2011). Recent vations hint at some small departures from spherical modeling of the standard delayed-detonation scenario symmetry. can yield 56Ni at large radii along some lines of sight Most constraints therefore imply that SN 2011fe is (?); more work is required to determine if such models consistent with a mildly asymmetric delayed detona- can explain the observations of SN 2011fe. tion model. In the future, the late-time light curve Mazzali et al. (2013) compare a time series of of SN 2011fe (&4 years after explosion) may distin- UV+optical spectra with SN Ia explosion models. guish between explosion models (R¨opke et al. 2012). They consider a pure deflagration model in the form The amount of radioactive 55Fe in the ejecta (half life: of the famous benchmark W7 (Nomoto et al. 1984), 2.75 yr) scales with the central density of the exploded which has a steep density profile at the outermost radii white dwarf. Therefore, the delayed detonation of a and very little mass expanding at the highest veloci- Chandrasekhar-mass white dwarf should be brighter at ties. This model produces good fits to optical spectra, late times than the merger or two sub-Chandrasekhar but it overpredicts the flux in the UV—more material white dwarfs. However, it will be challenging to infer is required at large velocity, above the photosphere, in the bolometric luminosity solely from optical photom- order to absorb this light. The W7 model is contrasted etry (McClelland et al. 2013), and the effect of 55Fe with a delayed detonation model (Iwamoto et al. 1999), will need to be carefully disentangled from the possi- which has significantly more material expanding at ble late-time contribution of the puffed-up companion high velocities (>16 000 km s−1). However, this model star (see Section 5; Shappee et al. 2013a). predicts larger blueshifts to the UV Fe-group features than observed. Therefore, Mazzali et al. (2013) com- pose a hybrid model with an outer density profile of 5 What is the progenitor of intermediate steepness between the pure-deflagration and delayed detonation models. This hybrid essen- SN 2011fe? tially corresponds to a weak delayed detonation and provides a better fit to the optical+UV spectra. It is generally thought that, in order to explode as a SN Ia, a CO white dwarf must be destabilized by Observations of SN 2011fe are also compared with mass transfer from a binary companion. However, the two different three-dimensional explosion models by nature of the companion—whether a main sequence, R¨opke et al. (2012). One model is for a delayed detona- subgiant, or giant star in a “single-degenerate” bi- tion of a Chandrasekhar-mass white dwarf, while the nary, or another white dwarf in a “double-degenerate” other represents a violent merger of two WDs (1.1 M binary—remains unknown. Similar uncertainties exist + 0.9 M ). Both models produce the right amount of for the nature of the mass transfer, whether gradual 56Ni to match the light curves of SN 2011fe, and both accretion or a sudden merger. models can fit the spectra of SN 2011fe reasonably well (Figure 6). The delayed detonation model matches the early time spectra better, while the merger provides a 5.1 Constraints on the companion significantly better fit at later times. In both mod- els, the predicted spectra are blue-shifted relative to star the data on SN 2011fe; this discrepancy might be re- Li et al. (2011) study deep pre-explosion HST imaging solved by increasing the oxygen-to-carbon ratio in the of the site of SN 2011fe in M101. No source is present progenitor white dwarf (assumed in the R¨opke et al. at the location of SN 2011fe, ruling out bright binary study to be 1:1). companions like most red giants (Figure 7). These are Each model of R¨opke et al. (2012) provides a range by far the deepest pre-explosion limits placed on the of spectra, varying with viewing angle because the progenitor of a SN Ia; others have been factors of >60 model explosions are not spherically symmetric (grey less sensitive and also yielded non-detections (Maoz & lines in Figure 6). The merger is significantly more Mannucci 2008). Still, Li et al. cannot exclude main asymmetric than the delayed detonation (top row of sequence or subgiant companions of .4 M . Figure 6), so the spectra predicted from the merger If the SN shock plows over a non-degenerate com- model cover a wider swath of possible observations. panion star, this interaction is expected to produce an Future work is needed to test if the relatively asym- early-time blue “bump” in the UV/optical light curve metric explosions predicted by white dwarf mergers (Kasen 2010). The amplitude of this bump depends on are inconsistent with spectropolarimetric observations the binary separation and viewing angle. Brown et al. of SNe Ia, which constrain the geometry of the ejecta (2012) find no such feature in Swift/UVOT photome- (Wang & Wheeler 2008). try of SN 2011fe (see also Bloom et al. 2012; Figure 3), The spectropolarimetric observations of Smith and constrain the binary separation to ∼few ×1011 cm et al. (2011) find that SN 2011fe is polarized at a level (∼0.01 AU). This constraint also rules out red giant of only ∼0.2–0.4%. However, compared with the con- companions, and, assuming mass transfer by Roche tinuum and other spectral lines, the strong Si iiλ6755 Lobe overflow, it implies a mass of .1 M for a po- feature has different time-dependent polarization prop- tential main sequence companion. However, the dark erties. Smith et al. propose a geometric model wherein phase predicted by Piro & Nakar (2012) and discussed the continuum photosphere is an ellipse elongated in in Section 4 may soften these constraints by a factor the polar direction, with a Si-rich “belt” stretching of several; future work is needed to self-consistently along the equator. While a unique interpretation of the model the dark phase and the ejecta’s interaction with journals.cambridge.org/pas 7

ray trapping in the ejecta, which is responsible for pow- ering the Hα emission. Considerable uncertainties re- main in predicting the Hα luminosity associated with a given mass of entrained hydrogen, but late-time Hα observations hold promise for constraining the com- panions of SNe Ia. A final test of the companion to SN 2011fe is pos- sible from late-time observations of the light curve. A non-degenerate companion should expand and grow in luminosity after being shocked by the supernova blast wave; it is expected to remain a factor of 10−103 over- luminous for ∼103 − 104 yr (Shappee et al. 2013a). The signature of such a puffed-up companion should be visible in SN 2011fe &3.5 years after explosion, but could at first be confused with variations in radioac- tive yields from the SN itself (R¨opke et al. 2012). Very late time measurements will effectively search for such an altered companion at the site of SN 2011fe.

5.2 Constraints on the circumbi- nary medium A red giant progenitor is also ruled out by searches for circumbinary material using radio and X-ray ob- servations (Horesh et al. 2012; Chomiuk et al. 2012; Figure 7: A Hertzprung-Russell diagram showing Figure 2 Progenitor system constraints in a Hertzsprung-Russell (H- Margutti et al. 2012). The interaction between a su- | R)the diagram limits compared on to a some companion proposed single-degenerate star to progeni- SN 2011fe, de- pernova blastwave and the circumstellar medium accel- tors. The thick yellow line is the 2s limit in V-band absolute magnitude (MV ) againstrived effective from temperature pre-explosion at the SN location (seeHST text) fromimaging. a combina- The pa- erates particles to relativistic speeds and amplifies the tion of the four HST filters, weighted using synthetic colours of redshifted stellarrameter spectra at space solar metallicity above for that the temperature yellow and luminosity line class. is ruled out, magnetic field in the shock front, producing radio syn- A more conservative limit comes from taking the single filter that most con- strainsexcluding the stellar type most and luminosity red class; giants shown is as the 2 thes limit assuming companions to chrotron emission (Chevalier 1982, 1998). These same theSN adopted 2011fe. distance The modulus stellar7,8 of 29.04 mainmag (middle sequence light yellow curve) is plotted as relativistic electrons also up-scatter photons radiated with a total uncertainty of 0.23 mag (top/bottom light yellow curve). Depicted area theblack theoretical line estimates and (He-star giant channel branches18) and observed for candidate stars of vari- by the supernova itself, producing inverse Compton systems (V445 Pup 21, RS Oph 20, U Sco 22,29, and T CrB 20). Also plotted radiation at X-ray wavelengths (Chevalier & Frans- areous theoretical masses evolutionary are tracks plotted (from 1 Myr as to 13colored Gyr) of isolated dots stars(key for in bot- atom range of left). masses for Severalsolar metallicity; famous note that the limits candidates on the progenitor for single- son 2006). While these signals are often observed in mass of SN 2011fe under the supersolar metallicity assumption are similar to nearby core-collapse SNe (Weiler et al. 2002; Soder- thosedegenerate represented here. SN The greyIa curve progenitors at top is the limit inferred are alsofrom HST plotted as analysis of SN 2006dd, representative of the other nearby SN Ia progenitor berg et al. 2006), no SN Ia has ever been detected limitsshaded (see Supplementary grey regions: Information). Forrecurrent the helium-star novae channel, bolo- with red gi- metric luminosity corrections to the V band are adopted based on effective at radio or X-ray wavelengths, implying that SNe Ia temperature 30. The foreground Galactic and M101 extinction due to dust ant companions9 (RS Oph and T CrB), a recurrent is negligible and taken to be AV = 0 mag here. Had a source at the 2.0s do not explode in dense environments (Panagia et al. photometricnova with level been a detected main-sequence in the HST images at the companion precise location of (U Sco), 2006; Immler et al. 2006; Hancock et al. 2011; Russell the SN, we would have been able to rule out the null hypothesis of no signif- icantand progenitor a He with nova 95% confidence. (V445 As such,Pup). we use Figure the 2s photometric from Li et al. & Immler 2012). SN 2011fe is no exception, with mul- uncertainties(2011). in quotingReprinted the brightness by limits permission on the progenitor system. from Macmillan tiple epochs of non-detections in deep radio and X-ray Publishers Ltd: Nature, copyright 2011. data. With SN 2011fe, we can place the most stringent 6 limits to date on the circumbinary environment around a SN Ia. Assuming the circumbinary material is dis- a companion. tributed in a wind profile (ρ = M˙ r−2, where M˙ 4πvw Shappee et al. (2013b) also place constraints on and vw are the mass-loss rate and velocity of the the companion star to SN 2011fe by searching for Hα wind), Chomiuk et al. (2012) use deep radio limits emission in a nebular spectrum nine months after ex- from the Karl G. Jansky Very Large Array to find ˙ −10 −1 plosion. If the companion to a SN Ia is non-degenerate, that M . 6 × 10 M yr in the surroundings −1 ∼0.1–0.2 M of hydrogen is predicted to be swept of SN 2011fe, for vw = 100 km s . Assuming a from the companion and entrained in the low-velocity uniform density medium, they find its density must −3 ejecta (Marietta et al. 2000; Pan et al. 2012; Liu et al. be nCSM . 6 cm . These limits on the circumbi- 2012b). Once the ejecta become optically thin, the nary medium not only rule out a red giant companion hydrogen-rich material should be observable as Hα for SN 2011fe, but also exclude optically-thick accre- emission (Mattila et al. 2005). Studies of previous tion winds and non-conservative mass transfer during SNe Ia constrained the entrained H to .0.01 M Roche Lobe overflow (Chomiuk et al. 2012). The en- (Leonard 2007), but Shappee et al. (2013b) place an vironment around SN 2011fe is extremely low-density. order-of-magnitude stronger limit in SN 2011fe, .0.001 Horesh et al. (2012) caution that radio limits on the M . If this result holds, it would essentially exclude circumbinary density depend on poorly-understood all non-degenerate secondaries and require a double- microphysical parameters governing the efficiency of degenerate model for SN 2011fe. However, more theo- particle acceleration and magnetic field amplification. retical work is needed to thoroughly explore gamma- Margutti et al. (2012) use X-ray observations from 8 Publications of the Astronomical Society of Australia

Chandra and Swift to place constraints on the cir- vae. The white dwarf will radiate thermal emission cumbinary medium which are less model-dependent of ∼few ×105 K, with a spectral energy distribution than the radio limits, because they do not depend peaking in the far-UV to soft X-ray. Liu et al. (2012a) on an assumed magnetic field strength. While the X- and Nielsen et al. (2012) search deep pre-explosion X- ray limits on circumbinary density are somewhat less ray imaging of the site of SN 2011fe, looking for evi- stringent than the radio constraints, it is worth noting dence of such a super-soft X-ray source, and emerge that the sensitivity of X-ray observations to circum- with non-detections. While this constraint rules out stellar material scales with the bolometric luminosity many known super-soft sources, it is not stringent of the supernova. The deep Chandra observation on enough to exclude those with lower luminosities or SN 2011fe was obtained just three days after discovery, cooler temperatures; for example, a source like the significantly before light curve peak. If instead Chan- persistent super-soft source Cal 83 remains viable (Liu dra had observed at optical maximum, the X-ray con- et al. 2012a). straints on circumbinary material around SN 2011fe One piece of evidence suggesting a single- would be more stringent than the radio limits. degenerate system is that the fastest-moving ejecta A search for circumstellar dust carried out by Jo- (>19 400 km s−1) in SN 2011fe are almost exclusively hansson et al. (2013) uses imaging from Herschel at composed of carbon (98% by mass; Mazzali et al. 70 µm and 160 µm. Both pre- and post-explosion 2013). Mazzali et al. interpret these outermost ejecta imaging yield non-detections, constraining the dust as the ashes of the material accreted onto the white −3 mass in the vicinity of SN 2011fe to . 7×10 M dwarf before the SN. If SN 2011fe marked the merger (assuming a dust temperature of 500 K). of two CO white dwarfs, a significant fraction of this A clean circumbinary environment is also found material should be oxygen—but the O fraction is small by Patat et al. (2013), who study optical absorption and strongly constrained by the observed O i feature lines along the line of sight to SN 2011fe. In a few at 7774 A.˚ They conclude that the dominance of C in SNe Ia, time-variable Na i D absorption has been ob- highest-velocity ejecta is support for H-rich accretion served and attributed to the presence of circumbinary onto the white dwarf, because H will fuse to C on the material that is ionized by the SN and then recom- outskirts of a SN Ia, but the ejecta will expand before bines (Patat et al. 2007; Simon et al. 2009). Patat significant amounts of C can subsequently fuse to O. et al. (2013) obtain multi-epoch high-resolution spec- The outer ejecta might also be consistent with the ac- troscopy of SN 2011fe and find no evidence for time cretion of helium under special circumstances, but in variability in the Na i D profile, again implying a lack most conditions He should burn explosively up to the of significant circumbinary material. Fe-peak. A relatively exotic strategy for “hiding” the non- degenerate companion of a SN Ia was proposed by 5.3 Constraints on the accretion Justham (2011) and Di Stefano et al. (2011) and history dubbed the “spin-up/spin-down” model. A white dwarf accreting from a non-degenerate companion may The above-described constraints rule out many single- reach significant rotational speeds by conservation of degenerate progenitors, and are often cited as evi- angular momentum, and centripetal force will help dence that SN 2011fe was the product of a white dwarf support the white dwarf and prevent it from explod- merger. However, the constraints can not conclusively ing as a SN Ia. Upon the cessation of mass transfer exclude a main-sequence or sub-giant donor of reason- (presumably due to the evolution of the companion), ably low mass, 1–2 M , transferring material via . the white dwarf will begin to spin down, and after Roche lobe overflow. a delay, it will finally explode as a SN Ia. The spin- At low mass transfer rates, such a system might down time is uncertain and potentially highly variable, look like the recurrent nova U Sco (Figure 7), which ∼103 −1010 yr. This delay may provide sufficient time ejects 10−6 M every ∼10 years in a nova explosion for the evolved companion to lose any remaining H- and harbors a white dwarf near the Chandrasekhar rich envelope and contract to a small and unobtrusive mass (Thoroughgood et al. 2001; Diaz et al. 2010; radius. While this model can reconcile SN 2011fe to a Schaefer 2010; although the white dwarf in U Sco is range of single-degenerate progenitor systems (Hachisu likely composed of ONe, rather than CO; Mason 2011). et al. 2012), it is highly speculative. Accreting white Li et al. (2011) collect a time-series of pre-explosion dwarfs are observed to spin at significantly lower rates photometry at the site of SN 2011fe to search for no- than predicted by simple conservation of momentum vae preceding the supernova. All epochs yield non- (Sion 1999), and the models of spinning white dwarfs detections, and they estimate a ∼60% chance that a remain preliminary (e.g., Yoon & Langer 2005). nova would have been detected if it had erupted in the five years prior to SN 2011fe. There have also been suggestions in the literature that nova shells around SNe Ia should produce time-variable NaD absorption 6 How accurate are SNe Ia as features (Patat et al. 2011); Patat et al. (2013) find no standardizeable candles? evidence of a nova-like shell surrounding SN 2011fe. In a progenitor system with a main sequence com- Because the distance to M101 is relatively well known panion transferring mass at higher rates, steady burn- and because SN 2011fe is so well-studied, it consti- ing of hydrogen is expected on the white dwarf sur- tutes a test of SNe Ia as standardizeable candles. face, instead of unstable burning in the form of no- Matheson et al. (2012) collect distance measurements 1. How accurate are SNe Ia journals.cambridge.org/pasas standardizeable candles?9

M101 (non-SN) distance estimates span 0.4 dex! SALT2 SN distances MLCS2K2 have errors ~0.16 dex!

Calibration is the limiting factor!

Distance to M101 uncertain to ~8%!

(Matheson et al. 2012, Vinko et al. 2012)! Figure 8: Distance moduli measured to M101 with a range of techniques and 1σ error bars. Black lines are pre-SN 2011fe estimates which use Cepheids, TRGB, or PNLF methods. Red lines use near-IR light curves of SN 2011fe to estimate the distance, and represent fits to a variety of calibrations (Matheson et al. 2012). Blue lines use BVRI light curves of SN 2011fe with the fitters SALT2 and MLCS2k2 (Vink´oet al. 2012). Figure modified from Matheson et al. (2012) to include data from Vink´oet al. (2012); reproduced by permission of the AAS.

to M101 obtained independent of SN 2011fe, using similar procedure as Vink´oet al. (2012), but use JHKs Cepheid variable stars (Freedman et al. 2001; Macri light curves of SN 2011fe to compare various SN Ia et al. 2001; Saha et al. 2006; Shappee & Stanek 2011), calibrations. Plotted in red in Figure 8 are distance the tip of the red giant branch (TRGB; Sakai et al. moduli calculated using the H-band peak apparent 2004; Rizzi et al. 2007; Shappee & Stanek 2011; also brightness of SN 2011fe and six different calibrations Lee & Jang 2012), and the planetary nebula luminos- of near-IR light curves (Krisciunas et al. 2004; Wood- ity function (PNLF; Feldmeier et al. 1996). These es- Vasey et al. 2008; Mandel et al. 2009; Folatelli et al. timates span distance moduli of 29.04–29.53 (Figure 2010; Burns et al. 2011; Kattner et al. 2012; assuming −1 −1 8) with a standard deviation of 0.18 mag. H0 = 72 km s Mpc ). The calibrations yield dis- The light curves of SN 2011fe can be used to in- tance moduli ranging from 28.93 mag to 29.17 mag. dependently estimate the distance to its host galaxy. Each measurement has a significant uncertainty asso- Vink´o et al. (2012) observe optical light curves in ciated with it, ∼0.16 mag, due to the scatter in the BVRI and apply two often-used light curve fitters to data used to develop the calibration. The distribution estimate the distance to SN 2011fe: MLCS2k2 (Jha of distance moduli calculated from these calibrations et al. 2007) and SALT2 (Figure 2; Guy et al. 2007, has a standard deviation of 0.12 mag, consistent with 2010). The best-fit MLSCS2k2 template returns a dis- the error in a single calibration. tance modulus of 29.21±0.07 mag, while SALT2 yields Vink´oet al. (2012) estimate that the error on the 29.05 ± 0.08 mag (Figure 8); both assume H0 = 73 km distance to M101, using estimates from both Cepheids s−1 Mpc−1. The errors in distance moduli are domi- and SN 2011fe, remains at ∼8%—-a rather large un- nated by degeneracies in the template fits, not by noise certainty, given that M101 is one of the best-studied in the data. The difference in distance modulus mea- nearby galaxies. Still, distance estimates to SN 2011fe sured from these two calibrations is consistent with agree with recent Cepheid and TRGB distance deter- the scatter in each calibration measured from a larger minations to M101 within 1σ of quoted systematic er- sample of SNe Ia (∼0.15 mag; Kessler et al. 2009). rors on these calibrations. These results imply that The scatter observed between SN Ia light curves the systematic errors in standard candle calibrations is smaller in the near-IR than in the optical (Phillips are well-estimated and large offsets do not exist in zero 2012). Therefore, Matheson et al. (2012) carry out a points. 10 Publications of the Astronomical Society of Australia

7 Conclusions viduals to Populations” Conference, which took place September 2012 in Garching, Germany. Their kind • What exploded in SN 2011fe? A carbon/oxygen invitation for a review talk on SN 2011fe was the in- white dwarf of sub-solar metallicity. spiration for this article. She is also grateful to Ken • How did it explode? Most data are consistent Shen, Fernando Patat, Alicia Soderberg, Max Moe, with a slightly asymmetric delayed detonation, but and Dean Townsley for helpful discussions, Stuart Ry- other scenarios might also fit, if studied in more de- der and Bryan Gaensler for their patience and work, tail. and an anonymous referee for useful comments. • What is the progenitor of SN 2011fe? Despite much deeper searches than in any preceding SN Ia, very little evidence for a non-degenerate companion is References found in SN 2011fe. Small corners of single-degenerate parameter space remain viable, but the data imply Benetti, S., et al. 2005, ApJ, 623, 1011 that SN 2011fe may have been the merger of two white Blondin, S., Kasen, D., R¨opke, F. K., Kirshner, R. P., dwarfs. It is important to remember that SN 2011fe is & Mandel, K. S. 2011, MNRAS, 417, 1280 just one supernova, and the class of SNe Ia may be diverse. Bloom, J. S., et al. 2012, ApJ, 744, L17 • How accurate are SNe Ia as standardizeable candles? Different calibrations of SN Ia light curves, Branch, D., et al. 2006, PASP, 118, 560 when applied to SN 2011fe, yield a range of distances to M101 with a standard deviation of 11%. These agree Brown, P. J., et al. 2012, ApJ, 753, 22 with Cepheid and TRGB distances to M101 at the 1σ Burns, C. R., et al. 2011, AJ, 141, 19 level. Because of its early discovery, proximity, and nor- Chevalier, R. A. 1982, ApJ, 259, 302 malcy, SN 2011fe constitutes a unique opportunity for detailed study of a SN Ia. It is likely to be a decade —. 1998, ApJ, 499, 810 or more before the next similarly bright and nearby SN Ia explodes, and in the meantime theorists should Chevalier, R. A., & Fransson, C. 2006, ApJ, 651, 381 continue to develop models and revisit the exquisite data collected for SN 2011fe, with the goal of further Chomiuk, L., et al. 2012, ApJ, 750, 164 constraining its progenitor system and explosion mech- Di Stefano, R., Voss, R., & Claeys, J. S. W. 2011, ApJ, anism. For example, additional work is needed to ac- 738, L1 curately predict the Hα luminosity expected from a single-degenerate SN Ia in the nebular phase. Spec- Diaz, M. P., Williams, R. E., Luna, G. J., Moraes, M., tra of SN 2011fe, spanning just one day after explosion & Takeda, L. 2010, AJ, 140, 1860 to the late nebular phase and the UV to the IR, are a rich observational resource which have just begun Feldmeier, J. J., Ciardullo, R., & Jacoby, G. H. 1996, to be tapped. Papers modeling the spectra have, to ApJ, 461, L25 date, considered only a small handful of specific explo- sion models; future work should more thoroughly ex- Folatelli, G., et al. 2010, AJ, 139, 120 plore the parameter space of plausible explosion mech- —. 2012, ApJ, 745, 74 anisms, analyze the uniqueness of predicted observ- ables from different models, and consider all observ- Foley, R. J., & Kirshner, R. P. 2013, ApJ, 769, L1 ables when comparing with models. Late-time photometry on SN 2011fe should be pur- Foley, R. J., et al. 2012, ApJ, 752, 101 sued for years to come, with goals of constraining yields of radioactive isotopes and searching for a puffed-up Freedman, W. L., et al. 2001, ApJ, 553, 47 companion star. Guy, J., et al. 2007, A&A, 466, 11 Continued efforts to compare observations of SN 2011fe with theory will ensure a solid groundwork —. 2010, A&A, 523, A7 for interpreting the large samples of SNe Ia to be ob- tained with LSST. When the next nearby bright SN Ia Hachisu, I., Kato, M., & Nomoto, K. 2012, ApJ, 756, explodes, we will be in an even better position to test L4 models of SNe Ia, armed with the next generation of time-domain telescopes like LSST, ASKAP, and LOFT Hancock, P. P., Gaensler, B. M., & Murphy, T. 2011, and a polished theoretical framework. ApJ, 735, L35 Hillebrandt, W., & Niemeyer, J. C. 2000, ARAA, 38, Acknowledgments 191 H¨oflich, P., Wheeler, J. C., & Thielemann, F. K. 1998, L. C. is a Jansky Fellow of the National Radio As- ApJ, 495, 617 tronomy Observatory. She thanks the organizers of the “Supernovae Illuminating the Universe: From Indi- Horesh, A., et al. 2012, ApJ, 746, 21 journals.cambridge.org/pas 11

Howell, D. A. 2011, Nature Communications, 2 Mandel, K. S., Wood-Vasey, W. M., Friedman, A. S., & Kirshner, R. P. 2009, ApJ, 704, 629 Howell, D. A., et al. 2009, ApJ, 691, 661 Maoz, D., & Mannucci, F. 2008, MNRAS, 388, 421 Hoyle, F., & Fowler, W. A. 1960, ApJ, 132, 565 —. 2012, PASA, 29, 447 Hsiao, E. Y., et al. 2013, ApJ, 766, 72 Margutti, R., et al. 2012, ApJ, 751, 134 Immler, S., et al. 2006, ApJ, 648, L119 Marietta, E., Burrows, A., & Fryxell, B. 2000, ApJS, Iwamoto, K., Brachwitz, F., Nomoto, K., Kishimoto, 128, 615 N., Umeda, H., Hix, W. R., & Thielemann, F.-K. 1999, ApJS, 125, 439 Mason, E. 2011, A&A, 532, L11

Jackson, A. P., Calder, A. C., Townsley, D. M., Matheson, T., et al. 2012, ApJ, 754, 19 Chamulak, D. A., Brown, E. F., & Timmes, F. X. 2010, ApJ, 720, 99 Mattila, S., Lundqvist, P., Sollerman, J., Kozma, C., Baron, E., Fransson, C., Leibundgut, B., & Nomoto, Jaikumar, P., Meyer, B. S., Otsuki, K., & Ouyed, R. K. 2005, A&A, 443, 649 2007, A&A, 471, 227 Mazzali, P., et al. 2013, arXiv 1305.2356 Jha, S., Riess, A. G., & Kirshner, R. P. 2007, ApJ, 659, 122 Mazzali, P. A., Chugai, N., Turatto, M., Lucy, L. B., Danziger, I. J., Cappellaro, E., della Valle, M., & Johansson, J., Amanullah, R., & Goobar, A. 2013, Benetti, S. 1997, MNRAS, 284, 151 MNRAS, 431, L43 McClelland, C. M., Garnavich, P. M., Milne, P. A., Justham, S. 2011, ApJ, 730, L34 Shappee, B. J., & Pogge, R. W. 2013, ApJ, 767, 119

Kasen, D. 2010, ApJ, 708, 1025 Meakin, C. A., Seitenzahl, I., Townsley, D., Jordan, IV, G. C., Truran, J., & Lamb, D. 2009, ApJ, 693, Kattner, S., et al. 2012, PASP, 124, 114 1188

Kessler, R., et al. 2009, ApJS, 185, 32 Munari, U., Henden, A., Belligoli, R., Castellani, F., Cherini, G., Righetti, G. L., & Vagnozzi, A. 2013, Krisciunas, K., Phillips, M. M., & Suntzeff, N. B. 2004, NewA, 20, 30 ApJ, 602, L81 Nielsen, M. T. B., Voss, R., & Nelemans, G. 2012, Kromer, M., Sim, S. A., Fink, M., R¨opke, F. K., MNRAS, 426, 2668 Seitenzahl, I. R., & Hillebrandt, W. 2010, ApJ, 719, 1067 Nomoto, K., Thielemann, F.-K., & Yokoi, K. 1984, ApJ, 286, 644 Lee, M. G., & Jang, I. S. 2012, ApJ, 760, L14 Nugent, P., Sullivan, M., Bersier, D., Howell, D. A., Lentz, E. J., Baron, E., Branch, D., Hauschildt, P. H., Thomas, R., & James, P. 2011a, ATel, 3581 & Nugent, P. E. 2000, ApJ, 530, 966 Nugent, P. E., et al. 2011b, Nature, 480, 344 Leonard, D. C. 2007, ApJ, 670, 1275 Pan, K.-C., Ricker, P. M., & Taam, R. E. 2012, ApJ, Li, W., et al. 2011, Nature, 480, 348 750, 151 Liu, J., Di Stefano, R., Wang, T., & Moe, M. 2012a, Panagia, N., Van Dyk, S. D., Weiler, K. W., Sramek, ApJ, 749, 141 R. A., Stockdale, C. J., & Murata, K. P. 2006, ApJ, Liu, Z. W., Pakmor, R., R¨opke, F. K., Edelmann, P., 646, 369 Wang, B., Kromer, M., Hillebrandt, W., & Han, Parrent, J. T., et al. 2012, ApJ, 752, L26 Z. W. 2012b, A&A, 548, A2 —. 2011, ApJ, 732, 30 Livio, M. 2001, in Supernovae and Gamma-Ray Bursts: the Greatest Explosions since the Big Bang, Patat, F., et al. 2007, Science, 317, 924 ed. M. Livio, N. Panagia, & K. Sahu, 334 Patat, F., Chugai, N. N., Podsiadlowski, P., Mason, Macri, L. M., et al. 2001, ApJ, 549, 721 E., Melo, C., & Pasquini, L. 2011, A&A, 530, A63

Maeda, K., R¨opke, F. K., Fink, M., Hillebrandt, W., Patat, F., et al. 2013, A&A, 549, A62 Travaglio, C., & Thielemann, F.-K. 2010, ApJ, 712, 624 Pereira, R., et al. 2013, arXiv 1302.1292

Maguire, K., et al. 2012, MNRAS, 426, 2359 Perlmutter, S., et al. 1999, ApJ, 517, 565 12 Publications of the Astronomical Society of Australia

Phillips, M. M. 2012, PASA, 29, 434 Timmes, F. X., Brown, E. F., & Truran, J. W. 2003, ApJ, 590, L83 Piro, A. L. 2012, ApJ, 759, 83 Vink´o,J., et al. 2012, A&A, 546, A12 Piro, A. L., Chang, P., & Weinberg, N. N. 2010, ApJ, 708, 598 Wang, B., & Han, Z. 2012, NewAR, 56, 122

Piro, A. L., & Nakar, E. 2012, arXiv 1211.6438 Wang, L., & Wheeler, J. C. 2008, ARAA, 46, 433

—. 2013, ApJ, 769, 67 Wang, X., Wang, L., Filippenko, A. V., Zhang, T., & Zhao, X. 2013, Science, 340, 170 Rabinak, I., & Waxman, E. 2011, ApJ, 728, 63 Weiler, K. W., Panagia, N., Montes, M. J., & Sramek, Richmond, M. W., & Smith, H. A. 2012, JAAVSO, 40, R. A. 2002, ARAA, 40, 387 872 Wood-Vasey, W. M., et al. 2008, ApJ, 689, 377 Riess, A. G., et al. 1998, AJ, 116, 1009 Woosley, S. E., & Kasen, D. 2011, ApJ, 734, 38 Rizzi, L., Tully, R. B., Makarov, D., Makarova, L., Dolphin, A. E., Sakai, S., & Shaya, E. J. 2007, ApJ, Yoon, S.-C., & Langer, N. 2005, A&A, 435, 967 661, 815

R¨opke, F. K., et al. 2012, ApJ, 750, L19

Russell, B. R., & Immler, S. 2012, ApJ, 748, L29

Saha, A., Thim, F., Tammann, G. A., Reindl, B., & Sandage, A. 2006, ApJS, 165, 108

Sakai, S., Ferrarese, L., Kennicutt, Jr., R. C., & Saha, A. 2004, ApJ, 608, 42

Schaefer, B. E. 2010, ApJS, 187, 275

Shappee, B. J., Kochanek, C. S., & Stanek, K. Z. 2013a, ApJ, 765, 150

Shappee, B. J., & Stanek, K. Z. 2011, ApJ, 733, 124

Shappee, B. J., Stanek, K. Z., Pogge, R. W., & Gar- navich, P. M. 2013b, ApJ, 762, L5

Silverman, J. M., & Filippenko, A. V. 2012, MNRAS, 425, 1917

Simon, J. D., et al. 2009, ApJ, 702, 1157

Sion, E. M. 1999, PASP, 111, 532

Smith, P. S., Williams, G. G., Smith, N., Milne, P. A., Jannuzi, B. T., & Green, E. M. 2011, arXiv 1111.6626

Soderberg, A. M., Chevalier, R. A., Kulkarni, S. R., & Frail, D. A. 2006, ApJ, 651, 1005

Stehle, M., Mazzali, P. A., Benetti, S., & Hillebrandt, W. 2005, MNRAS, 360, 1231

Stoll, R., Shappee, B., & Stanek, K. Z. 2011, ATel, 3588

Stritzinger, M., Leibundgut, B., Walch, S., & Con- tardo, G. 2006, A&A, 450, 241

Thoroughgood, T. D., Dhillon, V. S., Littlefair, S. P., Marsh, T. R., & Smith, D. A. 2001, MNRAS, 327, 1323