Ultraviolet Detection of the Binary Companion to the Type IIb SN 2001ig
Stuart D. Ryder, Schuyler D. Van Dyk, Ori D. Fox, Emmanouil Zapartas, Selma E. de Mink, Nathan Smith, Emily Brunsden, K. Azalee Bostroem, Alexei V. Filippenko, Isaac Shivvers, WeiKang Zheng
aa r X i v : . [ a s t r o - ph . S R ] J a n Draft version January 17, 2018
Preprint typeset using L A TEX style emulateapj v. 01/23/15
ULTRAVIOLET DETECTION OF THE BINARY COMPANION TO THE TYPE IIb SN 2001ig
Stuart D. Ryder , Schuyler D. Van Dyk , Ori D. Fox , Emmanouil Zapartas , Selma E. de Mink ,Nathan Smith , Emily Brunsden , K. Azalee Bostroem , Alexei V. Filippenko , Isaac Shivvers , and WeiKangZheng Draft version January 17, 2018
ABSTRACTWe present
HST /WFC3 ultraviolet imaging in the F275W and F336W bands of the Type IIb SN 2001igat an age of more than 14 years. A clear point source is detected at the site of the explosion having m F275W = 25 . ± .
10 and m F336W = 25 . ± .
13 mag. Despite weak constraints on both thedistance to the host galaxy NGC 7424 and the line-of-sight reddening to the supernova, this sourcematches the characteristics of an early B-type main sequence star having 19 , < T eff < ,
000 Kand log( L bol /L ⊙ ) = 3 . ± .
14. A BPASS v2.1 binary evolution model, with primary and secondarymasses of 13 M ⊙ and 9 M ⊙ respectively, is found to resemble simultaneously in the Hertzsprung-Russell diagram both the observed location of this surviving companion, and the primary star evolu-tionary endpoints for other Type IIb supernovae. This same model exhibits highly variable late-stagemass loss, as expected from the behavior of the radio light curves. A Gemini/GMOS optical spectrumat an age of 6 years reveals a narrow He ii λ Keywords: binaries: close – binaries: general – stars: evolution – stars: massive – supernovae: general– supernovae: individual (SN 2001ig) INTRODUCTION
Core-collapse supernovae (CCSNe) can result whenmassive stars exhaust their available fuel and thecores collapse under their own weight, thereby releas-ing enough potential energy to eject their outer lay-ers (Bethe et al. 1979; Woosley et al. 2002). Stripped-envelope SNe (SESNe) are a subset of CCSNe with pro-genitor stars that have lost most or all of their outerhydrogen envelope, and in some cases even their heliumenvelopes (e.g., Filippenko 1997).Type IIb supernovae (SNe) are distinguished bytheir initially strong H lines that fade away over thecourse of weeks to months (Filippenko 1988, 1997;Filippenko et al. 1993; Gal-Yam 2016). This moderateamount of observed H is usually interpreted as an in-termediate case between the H-rich SNe IIP/IIL andH-poor SNe Ib/c, presumably reflecting an increase in [email protected] Australian Astronomical Observatory, 105 Delhi Rd, NorthRyde, NSW 2113, Australia. Caltech/IPAC, Mailcode 100-22, Pasadena, CA 91125, USA. Space Telescope Science Institute, 3700 San Martin Drive,Baltimore, MD 21218, USA. Anton Pannekoek Institute for Astronomy, University ofAmsterdam, Science Park 904, 1098 XH, Amsterdam, TheNetherlands. Steward Observatory, University of Arizona, Tucson, AZ85721, USA. Department of Physics, University of York, Heslington, YorkYO10 5DD, United Kingdom. Department of Physics, University of California, Davis, CA95616, USA. Department of Astronomy, University of California, Berke-ley, CA 94720-3411, USA Miller Senior Fellow, Miller Institute for Basic Research inScience, University of California, Berkeley, CA 94720, USA stripping of the stellar envelope from IIP/L → IIb → Ib → Ic. Recent evidence, however, appears to demon-strate that SNe IIb are spectroscopically distinct fromSNe Ib/c at all epochs (Liu et al. 2016). Further indi-cations that SNe IIb form a separate channel from SNeIIP/IIL come from their low ejecta masses (Lyman et al.2016), together with radiative transfer models of theirspectra (Dessart et al. 2012).At least two scenarios can account for the strippingof the progenitor star’s envelope prior to its eventualdemise in a SN explosion (for both SNe IIb and Ib/csubclasses). Stellar winds accompanied by extreme oreruptive mass loss can shed significant amounts of gasover the lifetime of massive stars, (e.g., Heger et al. 2003;Smith & Owocki 2006); however the mass loss rates forthese tend to be overestimated (Smith 2014). Alterna-tively (or additionally), interaction with a massive binarycompanion can transport the H-rich outer layers into thecircumstellar medium (CSM; e.g., Podsiadlowski et al.1993; Eldridge et al. 2008; Claeys et al. 2011; Hirai et al.2014). These two scenarios allow for a large range ofpotential progenitor systems and thus of possible com-panions at the moment of explosion, as summarized byZapartas et al. (2017a). In scenarios that include a sur-viving binary companion, most models predict that thissurvivor should be a relatively unevolved, hot main se-quence star (Eldridge et al. 2015).Identification of the progenitor system can provideimportant constraints on the mass loss scenarios andtheoretical evolution of massive stars. While SNe IIbconstitute only ∼
10% of local CCSNe (Smith et al.2011; Shivvers et al. 2017), they are one of the best-studied subclasses, including several progenitor detec-tions. SN 1993J in M81 attracted considerable interest inpart owing to its proximity (3.6 Mpc), which enabled thedetection of not just the progenitor star (Aldering et al.1994; Van Dyk et al. 2002) but also a putative earlyB-type supergiant companion star (Maund et al. 2004;Fox et al. 2014). Progenitor stars have also been identi-fied in pre-explosion images of the Type IIb SNe 2008ax(Crockett et al. 2008; Folatelli et al. 2015), 2011dh(Maund et al. 2011; Van Dyk et al. 2011, 2013), 2013df(Van Dyk et al. 2014), and 2016gkg (Tartaglia et al.2017; Kilpatrick et al. 2017), while searches for a binarycompanion to SN 2011dh are inconclusive (Folatelli et al.2014; Maund et al. 2015). The relatively low initialmasses inferred from these SNe IIb progenitors, whichoverlap with the progenitor mass range of SNe IIP(Smartt 2009), further complicate the hypothesis thatSNe IIb form a transition from normal SNe IIP/IIL toSNe Ibc with increasing initial mass and/or mass lossrate.The mass loss and progenitor properties can also beconstrained indirectly. The interaction of the expandingSN blast wave with a pre-existing CSM consisting of ma-terial shed by the stellar progenitor gives rise to multi-wavelength emission, including X-ray inverse Comptonscattering and radio continuum synchrotron radiation.Multi-frequency radio light curves on timescales of hoursto years enable the reconstruction of the progenitor star’smass loss history centuries into the past. With a mea-sured or assumed wind speed, the mass loss rate can becalculated as a function of time to narrow down poten-tial progenitors (Weiler et al. 2002; Horesh et al. 2013;Smith 2014).The Type IIb SN 2001ig was discovered visuallyby Evans et al. (2001) on 2001 December 10.43 UTin the outskirts of the nearby late-type spiral galaxyNGC 7424, but received surprisingly little optical studydespite reaching 12 th magnitude (Bembrick et al. 2002).Silverman et al. (2009) published 12 optical spectra,Maund et al. (2007) presented 3 epochs of optical spec-tropolarimetry, and Ryder et al. (2006) showed a late-time spectrum, all taken within the first year after explo-sion. Silverman et al. (2009), Maund et al. (2007), andShivvers et al. (2013) pointed out both the observed sim-ilarities and differences between SN 2001ig and SN 1993J.Shivvers et al. (2013) further compared SN 2001ig in thenebular phase with SN 2011dh. Had there been an opti-cal light curve for SN 2001ig, particularly at early times,it would have been evident from the timing of the de-cline from the initial cooling phase whether SN 2001igwas more like SN 1993J, with its extended-envelope pro-genitor, or SN 2011dh, arising from a somewhat morecompact star.SN 2001ig was detected in X-rays with Chandra (Schlegel & Ryder 2002), yet the most complete cover-age of all was at radio wavelengths. Following the ini-tial detection within 5 days of discovery, Ryder et al.(2004) presented almost 2 years worth of data from theAustralia Telescope Compact Array (ATCA) and theVery Large Array (VLA) at frequencies between 1.4 and22.5 GHz. Fitting the multi-frequency light curves en-abled them to infer a mass loss rate from the progenitorof ∼ × − M ⊙ yr − , for a wind velocity of 10 km s − .There are no pre-explosion images of NGC 7424 of suf-ficient depth or resolution to enable a direct identifica- tion of the progenitor to SN 2001ig. The radio lightcurves, however, exhibited regular modulations with aperiod of ∼
150 days between peaks, which Ryder et al.(2004) attributed to “sculpting” of the CSM by a binarycompanion. Ryder et al. (2006) imaged the location ofSN 2001ig with the Gemini Multi-Object Spectrograph(GMOS) on the Gemini South 8 m telescope and identi-fied an optical counterpart in g ′ and r ′ data, consistentwith the presence of a surviving supergiant companionof spectral type late-B through late-F, depending on theextinction assumed.Soderberg et al. (2006) drew attention to several sim-ilarities between the radio light curves of SN 2001igand those of the broad-lined Type Ic SN 2003bg, in-cluding the strength of the modulations and their pe-riod. Since the likelihood of these two distinct SN sub-types having almost the same binary progenitor systemproperties (orbital period, mass ratio) that would giverise to such similar CSM structures must be extremelysmall, Soderberg et al. (2006) favored instead a scenarioin which both SN 2001ig and SN 2003bg have singleWolf-Rayet star progenitors that undergo quasi-periodicepisodes of enhanced mass loss. Kotak & Vink (2006)suggested that S Doradus-type variations, as seen in Lu-minous Blue Variables (LBVs), could be one such mecha-nism. Ben-Ami et al. (2015) analyzed early-time ultravi-olet (UV) spectra obtained with the Hubble Space Tele-scope ( HST ) of four SNe IIb, including SN 2001ig. Incontrast to SNe 1993J, 2011dh, and 2013df, the UV spec-tra of SN 2001ig showed a quite weak continuum, andstrong reverse-fluorescence features more akin to thoseof Type Ia SNe, consistent with a high radioactive Nimass and a compact progenitor object.As part of
HST program GO-14075 (PI: O. Fox), weused the Wide Field Camera 3 (WFC3) to search for sur-viving binary companions to nearby SESNe in the UV.In Zapartas et al. (2017b) we placed deep upper limitson the mass of a binary companion to the broad-linedType Ic SN 2002ap, and used binary population synthe-sis models to rule out the 40% of SESN channels thatwould have resulted in a surviving main sequence com-panion more massive than this. Section 2.1 presentsour new
HST /WFC3 UV observations of SN 2001ig,while Section 2.2 describes previously unpublished Gem-ini South/GMOS spectroscopy from 2007. We outlineour photometric results in Section 3.1 and our spectro-scopic results in Section 3.2. Our interpretation of thesein terms of a binary progenitor system is in Section 4 andour conclusions are summarized in Section 5. We notethat NGC 7424 has also recently served as host to theType II SN 2017bzb (Morrell et al. 2017). OBSERVATIONS
HST
Imaging
We imaged the site of SN 2001ig with WFC3/UVISon 2016 April 28 UT (14.4 yr after explosion) in bandsF275W and F336W, with total exposures of 8694 s and2920 s, respectively. The exposures were line-dithered,to improve image quality and resolution and to miti-gate against hot pixels and cosmic ray hits, and post-flashed to help mitigate against charge-transfer efficiency(CTE) losses. The images had been processed throughthe standard pipeline at the Space Telescope Science In-
Figure 1. (Left) : A portion of the
HST
WFC3/UVIS F275Wimage from 2016 obtained as part of GO-14075; the exact site ofSN 2001ig is indicated with tick marks. (Right) : The same, but inF336W. Some cosmic ray hits have not been completely removedfrom the image in the right panel. North is up, east to left, in bothpanels. stitute (STScI) before we obtained them from the Mikul-ski Archive for Space Telescopes (MAST). Specifically,the individual flt frames had been corrected for CTElosses. We then ran these corrected flc frames throughthe routine AstroDrizzle within PYRAF, to create finalimage mosaics in each band.We identified the SN on the WFC3 images by astromet-rically aligning the F336W image mosaic to the GMOS g ′ image obtained in 2004 under very good seeing condi-tions (0 . ′′ . ′′
45 FWHM; Ryder et al. 2006). Using theeleven stars in common between the two images as as-trometric fiducials, we identify a source corresponding tothe location of SN 2001ig with a rms uncertainty of 0.32UVIS pixel (12.7 milliarcsec). The location of the sourcein both the F275W and F336W images is indicated inFigure 1.We measured photometry for this detected source byinputting the individual flc frames into Dolphot (Dolphin2000). We ran this routine with parameters set to Fit-Sky=3, RAper=8, and InterpPSFlib=1, using the Tiny-Tim model point-spread functions (PSFs). By runningthe frames first through AstroDrizzle, cosmic ray hitsin the frames had also been flagged, which is impor-tant for accurate aperture correction. As a result, wefound for the identified source m F275W = 25 . ± .
10 and m F336W = 25 . ± .
13 mag. These are robust detectionsof a point-like source at the position of SN 2001ig withsignal-to-noise ratios of ∼
12 and 9, respectively. The ob-ject identifier in the Dolphot output was equal to 1 forboth bands, and the sharpness parameter was quite low: − .
011 and − .
007 for F275W and F336W, respectively.
GMOS Spectroscopy
Deep optical spectroscopy of the site of SN 2001ig wasobtained with the Gemini South Telescope using GMOS(Hook et al. 2004) as part of program GS-2006B-Q-11(PI: S. Ryder). A total of 5 hours on-source integra-tion was obtained over the course of two nights in 2007,July 18 and November 6 UT, in photometric IQ20 (see-ing < . ′′ ) conditions. The B600 grating was used witha 0 . ′′ R ∼ gmos tasks in V1.10 of the gemini package within iraf . A master bias frame constructed by averaging with3 σ clipping a series of bias frames was subtracted from allraw images. Images of a Cu-Ar lamp spectrum were used Figure 2.
Spectra from Gemini/GMOS in 2007 of the site ofSN 2001ig (black), and the nebulosity adjacent to this (red), withthe latter displaced vertically to assist comparison. Major emissionfeatures are marked. Note the presence of He ii λ to wavelength calibrate the science images and straighten(rectify) them along the spatial dimension, while im-ages of a quartz-halogen lamp spectrum helped correctfor sensitivity variations within and between the originale2v CCDs.The two-dimensional datasets from each night were re-duced separately, then registered spatially and in wave-length space before being coadded. By comparing con-tinuum sources in the resultant spectral image with fieldstars visible in GMOS images from 2004 (Ryder et al.2006, their Figure 1), the rows containing emission fromSN 2001ig could be identified. The flux within an aper-ture 0 . ′′ ANALYSIS
Photometric Properties of the Companion
The total reddening towards SN 2001ig is uncertain,but we expect it to be relatively low. The use of quitea narrow slit in order to minimize background contam-ination of the SN 2001ig spectrum over several hoursof integration (and thus a large range in slit parallac-tic angle traversed) precludes the use of the Balmerdecrement to estimate the extinction toward adjacentH ii regions because of slit losses . The contributionfrom the Galactic foreground extinction is A V = 0 . . . concluded that the contribution from the host galaxy waslikely no greater than the Galactic contribution; whilealso noting that Maund et al. (2007) did require someadditional reddening, beyond the Galactic component,to account for the observed optical polarization of theSN signal. We therefore assume that the host reddeningis essentially equivalent to the Galactic reddening; thetotal extinction to the SN site is thus A V = 0 .
06 mag.We further assume a Cardelli et al. (1989) reddening lawwith R V = 3 . − .
25) and reddened by theamount assumed above. We find that the photometryis consistent with an early B-type star with effectivetemperature T eff =19,000–22,000 K. The correspondingbrightness in V of these models is then 27.11–27.35 mag.The distance to the host galaxy NGC 7424 is notwell determined. However, it likely sits somewhere be-tween 10.9 (Böker et al. 2002) and 11.5 Mpc (Tully 1988;Soria et al. 2006), depending on the value of the Hubbleconstant assumed, with a corresponding distance modu-lus of 30.19–30.30 mag. If the detected source is a mainsequence star, its absolute magnitude is then in the range M V = − .
25 to − .
90. For the above range in T eff , atthe assumed metallicity, the V -band bolometric correc-tion is − .
82 to − .
16 mag (Paxton et al. 2011, 2013,2015; Choi et al. 2016). This would imply that the bolo-metric magnitude is M bol = − . ± .
35 mag, whichfor M bol ( ⊙ ) = 4 .
74 mag, corresponds to a bolometricluminosity log( L bol /L ⊙ ) = 3 . ± . T eff and L bol ona Hertzsprung-Russell diagram (HRD). For comparisonwe show a MIST single-star evolutionary track at initialmass 9 M ⊙ at metallicity [Fe/H] = − .
25 (Paxton et al.2011, 2013, 2015; Choi et al. 2016); the track is shown asa solid line up to the data point and then extrapolated asa dotted line for the remainder of the track. The locusof the detected source agrees with a star at this massnearing the terminal-age main sequence (TAMS).We have found 24 binary evolution models (out of atotal of 12678 models generated at this metallicity) fromBPASS version 2.1 (Eldridge et al. 2017) for which thesecondary star tracks place them within the uncertaintiesof the detected object at the time their primaries explode.In Figure 3 we also show as green circles the endpointsof the tracks of the corresponding model primaries, i.e.when carbon burning ends, and core-collapse is immi-nent. However, in nearly all of these cases the primarystar endpoint is either cooler (appearing as a red super-giant) or hotter than expected for SN IIb progenitors in-cluding SN 1993J and SN 2011dh (to which SN 2001ig’searly spectral evolution is most similar: Ryder et al.2006; Shivvers et al. 2013), or even SN 2008ax. This http://bpass.auckland.ac.nz/. is evident also in Fig. 21 of Eldridge et al. (2017) whichshows the BPASS models for Type IIb SNe favoring blueor red supergiant progenitors over yellow supergiants.We found one BPASS model having a secondary ofinitial mass 9 M ⊙ (similar to what we infer for the de-tected source), a primary of initial mass 13 M ⊙ (seeFigure 3), and an initial orbital period of 400 days. Al-though the agreement is not perfect, the endpoint ofthis model has the secondary on the HRD at a locusnot far ( ∼ σ ) from that of the detected object; whilethe track of the 13 M ⊙ primary ends at a T eff and L bol that is not too dissimilar from the locus of the progeni-tor of SN 2011dh, which had an initial mass 10–19 M ⊙ (Van Dyk et al. 2011; Maund et al. 2011)); or that forSN 1993J (Van Dyk et al. 2002), whose progenitor masswas in the range 13–22 M ⊙ .This particular BPASS model terminates with a pri-mary core mass of ∼ M ⊙ and with ∼ ⊙ of Hremaining. These numbers are quite similar to thoseyielded by the independent binary evolution models ofOuchi & Maeda (2017) for a secondary-to-primary massratio of 0 . < q < .
8, initial orbital period 200 < P <
600 days, and a low efficiency ( f ≈
0) of mass accre-tion from the primary onto the secondary. The blueloop in Fig. 3 is similar to that seen in the models ofYoon et al. (2017) for Type IIb interacting binary pro-genitors with periods of a few hundred days, and massescloser to 10 M ⊙ than 20 M ⊙ . These primaries leave lowmass helium-rich remnants (roughly 2–4 M ⊙ ) that un-dergo a second mass transfer stage when they swell upagain during helium shell burning. Building on the con-vention of Dewi et al. (2002) they refer to this as “CaseEBB/LBB” mass transfer.The primary in this BPASS model also appears toexperience a sudden increase in mass loss in the last ∼ ∼ Figure 3.
Hertzsprung-Russell diagram showing the locus of thepoint source detected at the site of SN 2001ig (solid symbol).Its properties are inferred from comparison of the observed
HST ultraviolet photometry with stellar atmosphere models for mainsequence stars at metallicity [Fe/H] = − .
25 (Castelli & Kurucz2003). Shown for comparison in red is a single-star evolu-tionary track at 9 M ⊙ for this same metallicity from MIST.Also shown are the locations of the secondaries (green crosses)which are consistent with the detected point source (to withinthe uncertainties) of 24 BPASS v2.1 binary evolution models,as well as the corresponding endpoints of the model primaries(green open circles). For reference, the loci of the progeni-tors of SN 1993J (Aldering et al. 1994; Van Dyk et al. 2002), SN2011dh (Maund et al. 2011; Van Dyk et al. 2011), and SN 2008ax(Folatelli et al. 2015) are shown in magenta (dotted lines). Addi-tionally, for comparison in blue is a BPASS binary evolution modelwith a primary (dashed-dotted line) and secondary (long-dashedline) of initial masses 13 and 9 M ⊙ , respectively, and an initialperiod of 400 days; the terminus of the primary track is indicatedwith a star. the HR diagram; in both cases these are then scaled ap-propriately for the metallicity of the star. Similarly theType IIb binary progenitor models of Yoon et al. (2017)use the MESA code (Paxton et al. 2011), and the so-called “Dutch” scheme for mass loss which is a hybrid ofthese and other mass loss prescriptions (Glebbeek et al.2009). In none of these models however is the massloss rate tied directly to a specific core- or shell-burningphase. Spectroscopic Properties of the Companion
Figure 2 compares the extracted spectrum from thelocation of SN 2001ig with that of the adjacent nebulos-ity. The two are virtually identical, indicating some de-gree of probable foreground and/or background contam-ination of the SN+companion optical spectrum by thisnebulosity. The one notable difference between them,however, is the clear detection (30 σ ) of emission fromHe ii λ − ) has a flux ratio relative to thenearby H β line (thus independent of the reddening as-sumed) of 0 . ± . ii λ http://mesa.sourceforge.net. signpost of the hottest stars, in particular, Wolf-Rayet(WR) stars, albeit as a much broader feature owing tothe fast, dense stellar winds (Crowther 2007). Neb-ular He ii λ HST imaging to argue against a WR progenitor in thatparticular case.Such weak and narrow He ii λ α and He i lines, rather than He ii λ ii λ ii λ DISCUSSION
An Innocent Bystander?
While the object detected in the WFC3 images couldsimply be a chance alignment of an unrelated fore-ground/background source, this seems extremely un-likely for two reasons. First, the somewhat sparse dis-tribution of detectable UV sources in Figure 1 makesthis a rather low statistical probability; and second, it isas unlikely that this contaminating source just happensto show rare, narrow He ii λ A V ∼ A V = 0 mag) spectral type. Our HST
UV imaging taken12 years later now indicates the companion is more likelyto be a main sequence B star with quite low extinction( A V = 0 .
06 mag; Section 3.1). We found that the ( g ′ − r ′ )color and ( u ′ − g ′ ) blue limit of the 2004 source wereinconsistent with either a pure H ii region or the nebu-lar spectrum of SN 1993J at a similar age. It is quitelikely that the source spectrum will have evolved since2004, fading in the manner extrapolated for SN 2011dhby Maund et al. (2015), but perhaps rebrightening ow-ing to ongoing or resumed CSM interaction as hintedat by the emergence of He ii λ Binary Evolutionary Channel for the SN 2001igProgenitor
The presence of a surviving stellar companion wouldimply that the progenitor star of SN 2001ig was in abinary system. Indeed, if that is the case, mass transferonto the binary companion of the progenitor would haveplayed an important role in the removal of almost all ofits H-rich envelope, thus leading to a Type IIb SN event.Zapartas et al. (2017a) provide theoretical predictionsfor the binary companions of all SESNe (includingSNe IIb and SNe Ib/c). They find that in the major-ity of scenarios a binary companion star is expected atthe moment of explosion. In most cases the compan-ion still resides on the main sequence, since it evolveson a longer timescale owing to its lower initial mass. Infact, our inferred mass of ∼ M ⊙ for the companionof SN 2001ig coincides with the broad peak of the pre-dicted mass distribution of main sequence companions ofall SESNe shown in Zapartas et al. (2017a).Binary channels that originate from short-period sys-tems (about 1 . log P (days) .
3) have been suggestedto result in compact SN IIb progenitors (Yoon et al.2010; Stancliffe & Eldridge 2009; Bersten et al. 2012;Benvenuto et al. 2013; Folatelli et al. 2015; Yoon et al.2017), with a thin H envelope of mass . . M ⊙ (e.g.,Yoon et al. 2017). In low-metallicity environments, weakstellar winds allow the thin remaining envelope to still bepresent when the progenitor explodes. The progenitors ofthese compact SNe IIb stay on the blue part of the HRDor are expected to end their lives as yellow supergiants(YSGs), as in the case of direct progenitor detections ofSN 2011dh and SN 2013df. The BPASS model shown inFigure 3 follows such a scenario, having an initial periodlog P = 2 . M ⊙ of H remaining on theprogenitor at the moment of explosion.Alternatively, wider initial orbits lead to more mas-sive H envelopes (e.g., Yoon et al. 2017). Progeni-tors with an extended low-mass H envelope of ∼ M ⊙ (Podsiadlowski et al. 1993; Woosley et al. 1994;Elmhamdi et al. 2006; Claeys et al. 2011) are expectedto stay on the red part of the HRD. Claeys et al.(2011) find that extended SN IIb progenitors orig-inate from almost equal-mass systems with q = M accretor /M donor & . log P (days) . .
3. Although they assume conser-vative mass transfer as their standard assumption, theyalso explored binary evolution for different mass-transferefficiencies to determine the uncertainty in how muchmass is accreted by the companion star.In the case of low efficiency, as expected in systems with wide orbits (e.g., Schneider et al. 2015), the inferredcompanion mass of ∼ M ⊙ is relatively close to its birthmass. At the same time, the progenitor star initiallyshould be somewhat more massive, since q & .
7, andwill naturally form a He core of 3–4 M ⊙ , consistent withthe ejecta mass of ∼ M ⊙ inferred by Silverman et al.(2009). Thus, in the scenario of an extended SN IIbdiscussed here, the small difference in the evolutionarytimescales owing to the similarity of the mass, as wellas the possible absence of significant mass accretion ontothe companion, results in only a limited rejuvenation ofthe companion (Hellings 1983, 1984). This could explainthe fact that the detected companion appears to lie closeto the TAMS, as seen in Figure 3.A further consequence of possible non-conservativemass transfer is that SN IIb progenitors are expectedto have significant CSM around them just before the ex-plosion, corresponding to mass loss rates on the orderof 10 − to 10 − M ⊙ yr − , when averaged over the fi-nal 1000 years before the SN explosion (Ouchi & Maeda2017), which is comparable to that inferred from the ra-dio observations for SN 2001ig (Ryder et al. 2004). TheCSM produced by the binary interaction could be thecause of the observed He ii λ SUMMARY AND CONCLUSIONS
We have obtained late-time UV imaging (over 14 yearspast explosion) of the site of SN 2001ig and identified intwo separate filters a point source at the known locationof the SN. Allowing for the uncertainties in distance andextinction toward the host galaxy, we find this source tobe consistent with an early B-type main sequence starwith T eff =19,000–22,000 K and a bolometric luminositylog( L bol /L ⊙ ) = 3 . ± .
14. We show that the evolu-tionary track of a BPASS model with a 9 M ⊙ secondarystar passes near the TAMS on the HRD at about thesame time that its 13 M ⊙ primary companion reachesa location on the HRD similar to those observed for theprogenitors of SN 2011dh and SN 1993J. The growingnumber of surviving companions found in SNe IIb, cou-pled with their relatively low progenitor masses, weak-ens the case for massive single stars such as LBVs beingthe progenitors of most SNe IIb (Soderberg et al. 2006;Kotak & Vink 2006; Groh et al. 2013).Although the progenitor star of SN 2001ig was neveridentified directly, we believe our detection of what isalmost certainly the surviving companion in this modelmakes a strong case for a binary interaction scenario forSN 2001ig, leading to a partially-stripped envelope anddense CSM. A ground-based optical spectroscopic com-parison of the location of SN 2001ig with its neighbor-hood at an age of almost 6 years reveals narrow He ii λ HST -based optical spectral energy distribu-tion. These authors pointed out that the UV emissionfrom SN 2011dh, interpreted by Folatelli et al. (2014) asevidence for a binary companion, could instead be associ-ated with such a potentially protracted CSM interaction,and recommended further monitoring at optical wave-lengths with
HST once this still relatively young SESNhas faded enough to distinguish between these two pos-sibilities. Similarly, additional optical imaging with
HST in the future will be necessary to determine whether theUV emission we have detected at the site of SN 2001ig isproduced entirely by a surviving hot companion, or has acontribution from long-term, low-level CSM interaction.
ACKNOWLEDGMENTS
This work is based in part on observations made withthe NASA/ESA
Hubble Space Telescope , obtained at theSpace Telescope Science Institute (STScI), which is op-erated by the Association of Universities for Research inAstronomy, Inc., under NASA contract NAS 5-26555.Support was provided by NASA through grants GO-14075 and AR-14295 from STScI. It is also based inpart on observations obtained at the Gemini Observa-tory, which is operated by the Association of Universi-ties for Research in Astronomy, Inc., under a coopera-tive agreement with the NSF on behalf of the Geminipartnership: the National Science Foundation (UnitedStates), the National Research Council (Canada), CON-ICYT (Chile), Ministerio de Ciencia, Tecnología e Inno-vación Productiva (Argentina), and Ministério da Ciên-cia, Tecnologia e Inovação (Brazil). We thank the refereefor their suggestions, and are grateful to J. J. Eldridge fordiscussions regarding the BPASS models. AVF’s groupis also grateful for generous financial assistance from theChristopher R. Redlich Fund, the TABASGO Founda-tion, NSF grant AST-1211916, and the Miller Institutefor Basic Research in Science (U.C. Berkeley). EZ issupported by a grant of the Netherlands Research Schoolfor Astronomy (NOVA). SdM acknowledges support by aMarie Sklodowska-Curie Action (H2020 MSCA-IF-2014,project BinCosmos, id 661502).REFERENCES