An impostor among us II: Progenitor, environment, and modelling of AT 2016jbu
S. J. Brennan, M. Fraser, J. Johansson, A. Pastorello, R. Kotak, H. F. Stevance, T. -W. Chen, J. J. Eldridge, S. Bose, P. J. Brown, E. Callis, R. Cartier, M. Dennefeld, Subo Dong, P. Duffy, N. Elias-Rosa, G. Hosseinzadeh, E. Hsiao, H. Kuncarayakti, A. Martin-Carrillo, B. Monard, G. Pignata, D. Sand, B. J. Shappee, S. J. Smartt, B. E. Tucker, L. Wyrzykowski, H. Abbot, S. Benetti, S. Blondin, Ping Chen, J. Bento, A. Delgado, L. Galbany, M. Gromadzki, C. P. Gutiérrez, L. Hanlon, D. L. Harrison, D. Hiramatsu, S. T. Hodgkin, T. W. -S. Holoien, D. A. Howell, C. Inserra, E. Kankare, S. Kozlowski, K. Maguire, T. E. Müller-Bravo, C. McCully, P. Meintjes, N. Morrell, M. Nicholl, D. O'Neill, P. Pietrukowicz, R. Poleski, J. L. Prieto, A. Rau, D. E. Reichart, T. Schweyer, M. Shahbandeh, J. Skowron, J. Sollerman, I. Sosz?yski, M. D. Stritzinger, M. Szyma?ski, L. Tartaglia, A. Udalski, K. Ulaczyk, D. R. Young, M. van Leeuwen, B. van Soelen
MMNRAS , 1–24 (2021) Preprint 3 March 2021 Compiled using MNRAS L A TEX style file v3.0
An impostor among us II:Progenitor, environment, and modelling of AT 2016jbu
S. J. Brennan (cid:63) , M. Fraser , J. Johansson , A. Pastorello , R. Kotak H. F. Stevance , T. -W. Chen , , J. J. Eldridge , S. Bose , , P. J. Brown , E. Callis ,R. Cartier , M. Dennefeld , Subo Dong , P. Duffy , N. Elias-Rosa , , G. Hosseinzadeh ,E. Hsiao , H. Kuncarayakti , , A. Martin-Carrillo , B. Monard , G. Pignata , ,D. Sand , B. J. Shappee , S. J. Smartt , B. E. Tucker , , , L. Wyrzykowski ,H. Abbot , S. Benetti , J. Bento , S. Blondin , , Ping Chen , A. Delgado , , L. Galbany ,M. Gromadzki , C. P. Guti´errez , , L. Hanlon , D. L. Harrison , , D. Hiramatsu , ,S. T. Hodgkin , T. W. -S. Holoien , D. A. Howell , , C. Inserra , E. Kankare ,S. Koz(cid:32)lowski , T. E. M¨uller-Bravo , K. Maguire , C. McCully , , P. Meintjes ,N. Morrell , M. Nicholl , , D. O’Neill , P. Pietrukowicz , R. Poleski , J. L. Prieto , ,A. Rau , D. E. Reichart , T. Schweyer , , M. Shahbandeh , J. Skowron , J. Sollerman ,I. Soszy´nski , M. D. Stritzinger , M. Szyma´nski , L. Tartaglia , A. Udalski ,K. Ulaczyk , , D. R. Young , M. van Leeuwen , B. van Soelen The authors’ affiliations are shown in Appendix A.
ABSTRACT
In the second of two papers on the peculiar interacting transient AT 2016jbu, we present the bolometric lightcurve,identification and analysis of the progenitor candidate, as well as preliminary modelling to help elucidate the natureof this event. We identify the progenitor candidate for AT 2016jbu in quiescence, and find it to be consistent with a ∼
20 M (cid:12) yellow hypergiant surrounded by a dusty circumstellar shell. We see evidence for significant photometricvariability in the progenitor, as well as strong H α emission consistent with pre-existing circumstellar material. Theage of the resolved stellar population surrounding AT 2016jbu, as well as integral-field unit spectra of the regionsupport a progenitor age of >
16 Myr, again consistent with a progenitor mass of ∼
20 M (cid:12) . Through a joint analysisof the velocity evolution of AT 2016jbu, and the photospheric radius inferred from the bolometric lightcurve, wefind that the transient is consistent with two successive outbursts or explosions. The first outburst ejected a shell ofmaterial with velocity 650 km s − , while the second more energetic event ejected material at 4500 km s − . Whetherthe latter is the core-collapse of the progenitor remains uncertain, as the required ejecta mass is relatively low (fewtenths of M (cid:12) ). We also place a restrictive upper limit on the ejected Ni mass of < (cid:12) . Using the BPASS code, we explore a wide range of possible progenitor systems, and find that the majority of these are in binaries, someof which are undergoing mass transfer or common envelope evolution immediately prior to explosion. Finally, we usethe
SNEC code to demonstrate that the low-energy explosion of some of these systems together with sufficient CSMcan reproduce the overall morphology of the lightcurve of AT 2016jbu.
Key words: circumstellar matter – stars: massive – supernovae: general – supernovae: individual: AT 2016jbu
This is the second of two papers on the interacting tran-sient AT 2016jbu (Gaia16cfr). We report photometric andspectroscopic observations in Paper I and present an in- (cid:63)
Contact e-mail: [email protected] depth comparison of AT 2016jbu and SN 2009ip-like tran-sients which include SN 2009ip (Fraser et al. 2013a; Grahamet al. 2014), SN 2015bh (Elias-Rosa et al. 2016; Th¨one et al.2017), LSQ13zm (Tartaglia et al. 2016a), SN 2013gc (Regui-tti et al. 2019) and SN 2016bdu Pastorello et al. (2018). Thework presented here will focus on the progenitor candidate, © a r X i v : . [ a s t r o - ph . H E ] M a r S. J. Brennan et al. its environment as well as modelling and interpretation of thespectral and photometric evolution.In Paper I we showed that AT 2016jbu had a history ofvariability, reaching a peak r -band absolute magnitude of M r ∼ − . Event A and
Event B . The lightcurvereaches an absolute magnitude of M V ∼ − . ∼
200 days similar tothat seen in the Type IIn SN 1996al (Benetti et al. 2016).AT 2016jbu shows a smooth evolution of the H α emissionprofile, changing from a P Cygni profile, typically seen inType II supernova (SN) spectra which show strong, singu-lar peaked, hydrogen emission lines (Kiewe et al. 2012; Tad-dia et al. 2015), to a double-peaked emission profile whichpersists until late times, indicating complex, H-rich, circum-stellar material (CSM). AT 2016jbu and SN 2009ip-like ob-jects show strong similarities in late time spectra with strongCa ii , He i and H emission lines as well as a lack of anyemission from explosively nucleosynthesised material such as[O i ] λλ , i ] λ ∼ (cid:12) LBV from pre-explosion images (Smith et al. 2010;Foley et al. 2011). However, this was measured in a singleband only, which would be strongly affected by flux in H α .For AT 2016jbu, we will show that the bright contributionof H α in F350LP gives similarly misleading Spectral EnergyDistribution (SED) fitting results. An LBV as the direct pro-genitor for a Type IIn SN also contradicts current stellarevolutionary theory, which suggests an LBV has just beganhelium-core burning and does not have the massive Fe corenecessary to explode as a core-collapse supernova (CCSN)(Heger et al. 2003; Humphreys et al. 2016, 2017).The the nature of SN 2009ip-like transients is much morecontentious. On one hand, there is evidence that these aregenuine core-collapse supernovae (CCSNe), the progenitorwas destroyed and the transient will fade after CSM interac-tion finishes (Smith et al. 2014a; Pastorello et al. 2013, 2019a;Graham et al. 2014; Smith & Mauerhan 2012). On the otherhand, some suggest these may be non-terminal events (Fraseret al. 2013a, 2015; Margutti et al. 2014; Graham et al. 2017),and SN 2009ip-like events are a result of either pulsational-pair instabilities (Woosley et al. 2007; Marchant et al. 2019),binary interaction (Pastorello et al. 2019a; Kashi et al. 2013),merging of massive stars (Soker & Kashi 2013) or instabilitiesassociated with rapid rotation close to the ΩΓ-limit (Maeder& Meynet 2000).As a follow-up to Paper I, we continue the discussion onAT 2016jbu using the data previously presented, focusing onthe progenitor and its local environment as well as examin-ing the controversial topic of the powering mechanism behindSN 2009ip-like events. For consistency with Paper I, and tocompare to previous work by Kilpatrick et al. (2018) (here-after referred to as K18), we take the distance modulus forNGC 2442 to be 31 . ± .
06 mag. This corresponds to a distance of 20 . ± .
58 Mpc and adopt a redshift z=0.00489from the H I Parkes All Sky Survey (HIPASS) (Wong et al.2006). The foreground extinction towards NGC 2442 is takento be A V = 0 .
556 mag (Schlafly & Finkbeiner 2011) via theNASA Extragalactic Database (NED; ). We do not correctfor any host galaxy or circumstellar extinction, however notethat the blue colours seen in the spectra of AT 2016jbu do notpoint towards significant reddening by additional dust (fur-ther discussed in Sect. 4.2 and Sect. 2). We take the V -bandmaximum at Event B (as determined through a polynomialfit) as our reference epoch (MJD 57784 . ± .
5; 2017 Jan 30).Significant lightcurve features will use the same naming con-vention as in Paper I for specific points in the lightcurve;
Rise, Decline, Plateau, Knee, Ankle .In Sect. 5 we investigate the CSM environment aroundAT 2016jbu and using photometry presented in Paper I re-construct the bolometric evolution of
Event A and
Event B up until the seasonal gap (+140 days), which we discuss inSect. 5.1. The progenitor of AT 2016jbu is discussed in Sect. 2using pre-explosion as well as late time imaging from the
Hubble Space Telescope ( HST ). This presence of pre-existingdust is discussed in AT 2016jbu using SED fitting as well as dusty modelling in Sect. 3. Using
HST and Very Large Tele-scope (VLT) + Multi Unit Spectroscopic Explorer (MUSE)observations, we investigate the surrounding stellar popula-tion and environment in Sect. 4. The powering mechanism be-hind AT 2016jbu is discussed in Sect. 6. In Sect. 7.1, the mostlikely progenitor for AT 2016jbu is examined. AT 2016jbuand most SN 2009ip-like transients display a high degreeof asymmetry, most likely due to a complex CSM environ-ment, and this is expanded upon in Sect. 7.2. Finally, we willaddress the explosion scenario for AT 2016jbu and perhapsother SN 2009ip-like transients, focusing on a CCSN scenarioin Sect. 7.3, and an explosion in a binary system in Sect. 7.4.
The progenitor of AT 2016jbu were discussed by K18, whosuggest that it was consistent with an F8 type star of ∼
18 M (cid:12) from an optical SED fit, although circumstellar extinctionplaces this as a lower bound.The progenitor of SN 2009ip is thought to be a LBV typestar with a much larger mass of ∼ (cid:12) (Smith et al.2010; Foley et al. 2011). This inference was based on pre-explosion imaging and detections were made in a single pho-tometric band that covers H α . We caution the validity of thismeasurement as the pre-explosion history of SN 2009ip showsmultiple, strong outbursts that likely result in CSM interac-tion which will result in a strong contamination from H α tobroadband photometry.It is a topic of controversy whether LBV stars can bethe direct progenitors of Type IIn SNe (see discussion byDwarkadas 2011, and references therein). Theory places clas-sical LBVs in a post-main sequence phase rather than a finalpre-SN phase. Stellar models require that an LBV star ejectsits H envelope, becomes a WR star and then explodes asa hydrogen-deficient SN (Smith 2017; Groh 2017). However,the direct detection of LBV-like progenitors for some SNe has https://ned.ipac.caltech.edu/ MNRAS , 1–24 (2021) rogenitor, environment, and modelling of AT 2016jbu challenged this picture (e.g. Gal-Yam et al. 2007; Gal-Yam &Leonard 2009).There is a wealth of pre-explosion images of NGC 2442 andin this section we explore this data to identify and charac-terise the progenitor of AT 2016jbu. Here we are specificallyconcerned with the quiescent (or apparently quiescent) pro-genitor which can only be identified in deep, high resolutiondata. NGC 2442 was observed with the
Hubble Space Telescope ( HST ) on a number of occasions both prior to and afterthe discovery of AT 2016jbu using the Advanced Camera forSurveys (ACS) and both the UV-Visible and IR channelsof the Wide Field Camera 3 (WFC3/UVIS and WFC3/IR).We retrieved all images where the image footprint coveredthe site of AT 2016jbu from the Mikulski Archive for SpaceTelescopes (MAST ), these data are listed in Table. 1. In allcases, science-ready reduced images were downloaded. Withthe exception of the late-time ACS images taken in 2019, allanalysis was performed on frames that have been already cor-rected for charge transfer efficiency losses at the pixel level(i.e. drc/flc files). For the 2019 ACS images corrections forcharge transfer efficiency were applied to the measured pho-tometry.In order to locate a progenitor candidate for AT 2016jbu,we aligned the F814W -filter image taken in 2017, whenthe transient was bright, to the ACS+
F814W image from2006, approximately ten years prior to discovery. Using 20point sources common to both frames and within 20 (cid:48)(cid:48) ofAT 2016jbu, we derive a transformation between the pixelcoordinates with an root mean square (rms) scatter of only12 milliarcseconds (pixel scale ∼ (cid:48)(cid:48) /pixel). A brightsource is clearly visible at the position, and we identify this asthe progenitor candidate. The progenitor candidate is shownin Fig. 1, and is the same source as was identified by K18.We performed Point Spread Photometry (PSF) fittingon all HST images using the November 2019 release ofthe dolphot package (Dolphin 2000), with the instrument-specific ACS and WFC3 modules. In all cases, we performedphotometry following the instrument-specific recommenda-tions of the dolphot handbook regarding choice of aperturesize. The WFC3 images were taken at two distinct point-ings, and each set were analysed separately, otherwise eachcontiguous set of imaging with a particular instrument werephotometered together, using a single deep drizzled imageas a reference frame for source detection. Examination of theresidual images after fitting and subtracting a PSF to sourcesin the field revealed no systematic residuals, indicating sat-isfactory fits in all cases. We show the HST photometry forAT 2016jbu in Fig. 2.We find that the photometry reported by K18 is fainterthan what we measure, with a difference of ∼ . F350LP . We compared our measured
F350LP magnitudesand those of K18 to the values reported in the Hubble Source mastweb.stsci.edu/ http://americano.dolphinsim.com/dolphot/dolphot.pdf Catalog (HSC; Whitmore et al. 2016). As the magnitudes re-ported in the HSC are in the AB mag system, we appliedthe conversion from AB to Vega mag before comparing toour photometry. The HSC
F350LP magnitudes are consistentwith what we report here, and we also see the same variabil-ity for the progenitor candidate. The cause of the differencebetween our photometry and that of K18 hence remains un-known.We note that the broad-band photometry from
HST ismore than likely affected by the strong emission in H α . InFig. 3 we show the throughput of the HST filters compared toa late-phase spectrum of AT 2016jbu. The long-pass
F350LP filter will contain flux from H α . Fortuitously, H α falls in thelow-throughput red wing of the F555W filter, where it willhave negligible effect. To verify this, we used synphot (Lim2020) to perform synthetic
F555W -filter photometry on the+271 d spectrum of AT 2016jbu, and on the same spectrumwhere H α has been excised. The latter returns a magnitudethat is only 0.05 mag fainter than the former, and so the F555W filter is not significantly affected by line emission.The progenitor is relatively red, bright, and shows signif-icant variability over timescales of ∼ weeks. Correcting forforeground extinction, in 2006 the progenitor candidate hadan absolute magnitude in F814W = − . ± .
06 and an F W − F W colour of 1.13 ± F N magnitude, whichcovers H α , is much brighter than would be expected. This in-dicates that even ten years before the eruption or explosionof AT 2016jbu its progenitor was characterised by strong H α emission.In early 2016, between seven and ten months prior to thestart of Event A , NGC 2442 was observed repeatedly withWFC3 in
F350LP , F555W , F814W and
F160W . This datasetgives us a unique insight into the variability of the quies-cent progenitor prior to explosion. We see that even in qui-escence (arbitrarily defined as when the progenitor is fainterthan mag ∼ -10), the progenitor displays strong variability. Inparticular in the best-sampled F350LP lightcurve, the pro-genitor varies in brightness by 1.9 mag in only 20 days. Asdiscussed by K18, such rapid variability is hard to explain (al-though there is some similarity to the fast variability seen inthe pre-explosion lightcurve of SN 2009ip; Pastorello et al.2013). While it is impossible to know if the variability isperiodic on the basis of the short time coverage availablefor AT 2016jbu, if it is periodic then the apparent period isaround 45 days (found via a low order polynomial fit to the
F350LP lightcurve).The variability seen in
F350LP in early 2016 is also seenin other bands, which appear to track the same overall pat-tern of brightening and fading. Fig. 2 shows the colour evo-lution of
F350LP-F555W , F350LP-F814W , F350LP-F160W .In all cases (with the exception of the earliest
F350LP-F160W colour, which is likely due to a spurious
F350LP magnitude)we see a relatively minor colour change over three months. Infact, it is possible that the apparent small shift towards bluercolours is simply due to H α growing stronger, which wouldcause the F350LP magnitude to appear brighter, rather thanany change in the continuum temperature.At late times the progenitor candidate for AT 2016jbu isstill present. In 2019, over two years since the epoch of max-
MNRAS , 1–24 (2021)
S. J. Brennan et al.
Table 1.
Observational log for all
HST images covering the site of AT 2016jbu. Measured photometry (in the Vega-mag system) forAT 2016jbu is also reported. Phase is in rest frame days relative to
Event B maximum light (MJD 57784.4)Date Phase (d) Instrument Filter Exposure (s) Mag (err)2006-10-20 − . ×
395 24.999 (0.037)- - - F658N 3 ×
450 21.207 (0.024)- - - F814W 3 ×
400 23.447 (0.019)2016-01-21 − . ×
420 23.625 (0.017)- - WFC3/IR F160W 2 ×
503 20.726 (0.003)2016-01-31 − . ×
420 22.215 (0.026)- - - F555W 2 ×
488 22.645 (0.002)2016-02-08 − . ×
420 22.134 (0.001)- - WFC3/IR F160W 2 ×
503 19.570 (0.005)2016-02-17 − . ×
420 23.108 (0.012)- - - F814W 2 ×
488 22.287 (0.003)2016-02-23 − . ×
420 23.212 (0.022)2016-02-28 − . ×
420 23.985 (0.022)- - - F555W 2 ×
488 24.399 (0.004)2016-03-04 − . ×
420 22.729 (0.022)- - WFC3/IR F160W 2 ×
503 20.224 (0.011)2016-03-10 − . ×
420 22.690 (0.037)- - - F814W 2 ×
488 21.967 (0.022)2016-03-15 − . ×
420 22.868 (0.016)- - WFC3/IR F160W 2 ×
503 20.323 (0.014)2016-03-21 − . ×
420 23.400 (0.013)- - - F555W 2 ×
488 23.962 (0.012)2016-03-30 − . ×
420 23.775 (0.006)- - WFC3/IR F160W 2 ×
503 21.301 (0.020)2016-04-09 − . ×
420 23.767 (0.006)- - - F814W 1 ×
488 23.079 (0.035)2019-03-21 +776.7 WFC3/UVIS1 F555W 320,390 23.882 (0.025)- - - F814W 2 ×
390 23.239 (0.032)2019-03-31 +787.0 ACS/WFC F814W 4 ×
614 23.529 (0.014)
Pre-explosion
AT2016jbu
Late-time
Figure 1. (cid:48)(cid:48) × (cid:48)(cid:48) cutouts of all HST images centered on the progenitor candidate for AT 2016jbu. Columns are ordered in wavelengthfrom left to right.MNRAS000
Figure 1. (cid:48)(cid:48) × (cid:48)(cid:48) cutouts of all HST images centered on the progenitor candidate for AT 2016jbu. Columns are ordered in wavelengthfrom left to right.MNRAS000 , 1–24 (2021) rogenitor, environment, and modelling of AT 2016jbu M a g n i t u d e C o l o u r + o ff s e t
380 370 360 350 340 330 320 310 300 290Days until V-Band Maximum 780 790 0.10.20.30.40.50.6[ F350LP-F814W ]+0[ F350LP-F555W ]+1[ F350LP-F160W ]-2
Figure 2.
Foreground extinction corrected
HST lightcurves of AT 2016jbu and its progenitor are shown in the top panel. We also includea DECam r -band detection at −
352 d as a red filled circles with error bars. Error bars for
HST measurements are smaller than the pointsizes. The horizontal line is to guide the eye in comparing the late time ( ∼ +2 year) and pre-outburst ( ∼ −
10 year)
F814W magnitudes.We also plot the
F350LP and
F160W lightcurves with a line to help guide the eye. Colour curves, corrected for foreground reddening,are shown in the bottom panel. Colours as offset for legibility by the amounts stated in the legend. Å )0.00.20.40.60.81.0 T r a n s m i ss i o n F435W F555W F350LP F814W
Figure 3.
HST filters used for pre-explosion lightcurve, comparedto the +271 d spectrum. Only
F350LP covers the strong H α emis-sion seen at this epoch. imum light, a source is found at approximately the same F814W magnitude as was seen in 2006. It is unlikely thatthis source is a compact cluster, as the pre-explosion pho-tometric variability can only be explained if a single star iscontributing most of the flux. Moreover, we compared the2006
F814W and 2019
F814W images, and find that the po-sition of the source is consistent to within 17 mas betweenthe two epochs. This implies that the same source is likelydominating the emission at both epochs, and if there is an un-derlying cluster it must be much fainter than the progenitorsource.
In order to determine the luminosity and effective tempera-ture of the progenitor of AT 2016jbu, we consider the WFC3photometry taken in early 2016. As a first step, we normalizeout the variability seen over this period so that we can buildan SED from photometry taken in different filters at differ-ent epochs. To do this, we fit a linear function to the colourcurves of our
HST observations. We disregard the first epochfor the
F350LP − F160W colour (which is significantly red-der than the other epochs); this measurement is unreliableas the progenitor was affected by bad pixels in two of thethree individual exposures. We then use the fitted functionsto interpolate or extrapolate the magnitude of AT 2016jbuin
F555W , F814W or F160W as necessary. Finally, we shiftthe SEDs up or down in magnitude so that they all have thesame
F814W magnitude. The resulting normalised progeni-tor SEDs can be seen in Fig. 4.In order to determine a progenitor temperature from theobserved SED, we compare to MARCS stellar atmospheremodels (Gustafsson et al. 2008). We used the pysynphot package to perform synthetic photometry on the surfacefluxes of the models and hence calculate their magnitude ineach of the
F555W , F814W and
F160W filters. We shiftedeach model so that it matches the
F814W absolute magni-tude of the progenitor. In the lower panel of Fig. 4 we compareto the spherically-symmetric MARCS models for 15M (cid:12) redsupergiants (RSGs; log(g)=0) at solar metallicity. While wecan see that the models provide a reasonable agreement, itis clear that the warmest model (at 4500 K) is still too redto match the
F555-F814W colour of the progenitor, implyingthat the progenitor is hotter than this. Conversely, the 4000 Kmodel provides a good match to the
F814W-F160W coloursof the progenitor. As the 15 M (cid:12) super-giant models cover
MNRAS , 1–24 (2021)
S. J. Brennan et al. Å ]1312111098 V e g a m a g , Z=Z Å ]1312111098 V e g a m a g
15 M , log(g)=+0.0, = 2 km~s , Z=Z Figure 4.
HST
SEDs for AT 2016jbu based on the early 2016WFC3 imaging are shown in black. All SEDs have been shifted sothat their
F814W magnitudes match, as discussed in the text. The
F350LP filter magnitudes have not been included in the SED asthey are strongly affected by H α emission. We also plot a numberof SEDs derived from MARCS models. In the lower panel we showthe 15 M (cid:12)
MARCS models appropriate to cool red supergiants.As this model grid does not extend above 4500 K, we also plot aset of 5 M (cid:12) models with slightly higher log(g) in the upper panel.All models have been shifted so that they match the
F814W filtermagnitude of the progenitor, and we can see that while the coolermodels can match the NIR part of the SED, hotter temperaturesare required to match the optical. a relatively small temperature range, we also explored the5 M (cid:12) spherically symmetric MARCS models at log(g)=1.0which span a broader range (upper panel in Fig. 4). We findthat a 5000 K model can reproduce the optical colours of theprogenitor, while the NIR is better matched with a cooler4000 K model.While AT 2016jbu does not appear to suffer from highlevels of circumstellar extinction around maximum light, wecannot exclude the possibility that the progenitor coloursare caused by close-in CSM dust that was subsequently de-stroyed. To explore this possibility, we used the dusty (Ivezic& Elitzur 1997) code to calculate observed SEDs for a grid ofprogenitor models allowing for different levels of CSM dust. dusty solves for radiation transport within a dusty medium.Since a dust-enshrouded progenitor could be hotter than the range of temperatures covered by the
MARCS modelgrid, we used the
PHOENIX models (Husser et al. 2013)as our input spectra. The PHOENIX models cover the tem-perature range from 6000–12000 K in 200 K increments, andhave log(g) between 1 and 2 dex.
MARCS models coveringa temperature range from 2600–7000 K in 100 K incrementsand log(g) between 1 and 2 dex were also tested as inputto dusty . These models were then processed by dusty , as-suming spherically symmetric dust comprised of 50 per centsilicates and 50 per cent amorphous carbon. The dust den-sity followed a r − distribution, with a radial extent vary-ing between 1.5 and 20 times the inner radius of the dustshell. The dust mass is parameterized in terms of the opticaldepth in V -band, τ V , which varied between 0 and 5. For eachtemperature and dust combination, we calculated synthetic F W − F W and F W − F W colours, and com-pared to the foreground extinction corrected colours of theAT 2016jbu progenitor. In Fig. 5 we plot all models that havecolours within 0.1 mag of the progenitor. We find that we areable to match the progenitor colours with models with tem-peratures of between 10 . and 10 . K, for a circumstellardust shell with optical depth τ V between 1.6 and 2.6.We calculated a luminosity for each of these models by in-tegrating over its spectrum, and find that the progenitor hada luminosity log( L ) between 4.7 and 5.1 dex (depending ontemperature and extinction). Comparing to the BPASS sin-gle star evolutionary tracks at Solar metallicity in Fig. 5, wefind that these correspond to approximately the luminosityof a 17–22 M (cid:12) star with a temperature of between 6,000 and8,000 K as it crosses the HR diagram to become a RSG.
We present a SED model fitted to our −
11 day dataset inFig. 6. We fit at this phase as it has the broadest wavelengthcoverage without the need for interpolation. We fit two black-body models to the photometric points; one representing ahot photosphere, and the second fitted to the IR excess seenin H , K , W
1, and W
2. A single blackbody does not fit obser-vations seen at −
11 d before maximum. Allowing for a secondcooler blackbody at a larger radius gives a model that fits thedata well.This additional blackbody is consistent with warm dustymaterial at a distance of 170 AU and a temperature of T BB ∼ . × L (cid:12) . The hot blackbody has a radius of 36 AU,a temperature of T BB ∼ . × L (cid:12) and represents R BB at this time. Wefind a dust mass of M dust ≈ . × − M (cid:12) (Using eq.1from Foley et al. 2011). In comparison Smith et al. (2014b)finds a lower dust mass of (3 − × − M (cid:12) for SN 2009ip.Additionally we note a similarity to the SED for SN 2009ippresented in Margutti et al. (2014). The IR excess may becaused by thermal radiation of pre-existing dust in the CSMre-heated by an eruption at the beginning of Event B , i.e. anIR echo.We can compute the radius within which any dust will be http://phoenix.astro.physik.uni-goettingen.de MNRAS , 1–24 (2021) rogenitor, environment, and modelling of AT 2016jbu Log ( T eff [K])4.04.55.05.56.06.57.07.58.08.59.09.5 L o g ( L / L ) M M M M M M M M M M M M M M AT 2016jbu(K18)SN 2015bh(Boian & Groh 2017) Car (1890) Car (Today) Car (GE)SN 2009ip(Smith et al. 2010)IRC+10420 (Klochkova 2019) S D o r a d u s I n s t a b ili t y S t r i p L B V s i n o u t b u r s t PHOENIX MARCS3.73.83.94.04.1
Log ( T eff [K])4.64.74.84.95.05.15.25.35.4 L o g ( L / L ) M M M M M M AT 2016jbu(K18)IRC+10420 (Klochkova 2019) 1.691.902.112.322.53
Figure 5.
Hertzsprung–Russell (HR) diagram showing single star evolutionary tracks from
BPASS (Eldridge et al. 2017; Stanway &Eldridge 2018). We include SN 2009ip at log(
L/L (cid:12) ) = 5 . T eff ) = 3 .
92 (Smith et al. 2010; Foley et al. 2011), as well asSN 2015bh (Boian & Groh 2018) and IRC+10420 (Klochkova et al. 2016). η Car is plotted (red triangles) at several phases given inparentheses (Prieto et al. 2014). We include the progenitor estimates for AT 2016jbu from K18, found from
HST pre-explosion imagingas a green star marked with an arrow for clarity. We highlight the
Yellow-Void between 7000 K and 10000 K (de Jager 1998). The purplepoints show the evolution of AT 2016jbu in outburst (viz. L BB and R BB from Fig. 11). The transient evolves in a clockwise directione.g. Event A begins at log( T eff ) ≈ .
76 and log(
L/L (cid:12) ) ≈ .
7, and
Event B at Log ( L/L (cid:12) ) ≈ . T eff ) ≈ .
85. We include theoutput of our dusty modelling for AT 2016jbu using
PHOENIX models (multi-coloured squares) and
MARCS models (multi-colouredtriangles). The colour of each point corresponds to its optical depth ( τ ν ) which is provided on the colourbar on the right. We include aninset of the region around the progenitor in the top left of the plot. evaporated/vaporised at the phase of our SED fitting. Theradius of this dust-free cavity is given by: R c = (cid:115) L SN π σ T evap (cid:104) Q (cid:105) (1)where R c is the cavity radius, L SN is the luminosity ofthe transient, taken to be 1 . × L (cid:12) , σ is the Stefan-Boltzmann constant and (cid:104) Q (cid:105) is the averaged value ofthe dust emissivity. Assuming radiation is absorbed withefficiency ∼ unity by the dust, we find a cavity radius of ∼
245 AU for graphite grains ( T evap = 1900 K ) and ∼
400 AUfor silicate grains ( T evap = 1500 K ). Both values are signif- icantly larger than what we find from our warm blackbodyradius ( ∼
170 AU).A dust destruction radius larger than the blackbody radiusof our putative warm dust component appears at first glanceto be inconsistent. To ameliorate this we suggest that thedust may not be homogeneously distributed, and could bein either optically clumps or an aspherical region that pro-vides some shielding from evaporation. Over time, we expectthat the dust is further heated and destroyed during the riseto
Event B maximum. We find that by maximum bright-ness that this additional blackbody component is no longerneeded, suggesting that the dust causing this NIR excess hasbeen destroyed.
MNRAS , 1–24 (2021)
S. J. Brennan et al. Wavelength [ Å ]10 F l u x D e n s i t y [ e r g s c m Å ] RVBirgH JKzuW1W2
Figure 6.
SED fit of AT 2016jbu at -11 days before
Event B maximum. Extinction corrected photometry are grouped to 1 daybins and weighted averaged. Flux errors are given as standard de-viation of bins. Horizontal error bars represent approximate filterband-pass. Hot blackbody is given in blue, the cooler blackbody inred, and the black line is compound model. We note a similaritythe SED for SN 2009ip presented in Margutti et al. (2014).
As discussed in Paper I there are
Spitzer +IRAC observa-tions of the progenitor site of AT 2016jbu, which show tenta-tive detections in 2003 and 2018. Using 2003 Spitzer/IRACand 2016
HST /F160W observations, K18 find fits consistentwith a compact dusty CSM component with mass M dust ≈ . × − M (cid:12) at 72 AU. This may represent a dusty shellthat is later seen as our 170 AU warm blackbody. However,due to the time frame between Spitzer/ HST observationsthere are large uncertainties on dust parameters from K18.Fitting Spitzer data only gives a slightly higher M dust valueof ∼ − M (cid:12) at 120 AU. Due to the erratic variability seenin AT 2016jbu it is uncertain as to whether these dust shellsare the same, as AT 2016jbu may have a stratified CSM en-vironment resulting from successive outbursts.Although there is strong evidence for pre-existing dust, wedo not see any signature for newly formed dust in the envi-ronment around AT 2016jbu (Meikle et al. 2007; Smith et al.2008; Smith 2011). We see no NIR excess in late time J and K bands in late time photometry nor an IR excess evident inspectra. Furthermore, there is no blue-shift in the core emis-sion component in H α (Paper I), which is another indicatorof newly formed dust. Along with direct detections of progenitors, analysis of theresolved stellar population in the vicinity of a SN has alsobeen used to infer the progenitor age and hence initial mass(Gogarten et al. 2009; Maund 2017; Williams et al. 2018). Anadvantage to this technique is that it will not be affected byany peculiar evolutionary history or variability of the progen-itor that may cause it to appear less or more massive than ittruly is. On the other hand, using the environment around aSN is an indirect proxy for the progenitor age, and is pred-icated on the assumption that the local stellar populationis coeval. This method is also complicated by possible con-tamination from other stellar populations from multiple starformation episodes.
In order to study the population in the vicinity of AT 2016jbuwe require sources to be matched between different filter im-ages. While this is straightforward for bright sources suchas the progenitor of AT 2016jbu it is more challenging forfainter or blended sources, especially when images have dif-ferent pixel scales or orientations. We hence re-ran the pho-tometry on a subset of the
HST images (
F435W , F658N and
F814W from 2006 Oct 20;
F350LP and
F555W from 2016Jan 31), using a single drizzled ACS
F814W image as thereference image for all filters.We chose a projected radius of 150 pc (1.48 (cid:48)(cid:48) ) aroundAT 2016jbu as a compromise between identifying sufficientstars to be able to constrain the population age and ensur-ing we are still sampling a local population that is plausiblycoeval with the progenitor. We also create a less restrictivecatalog of sources within a projected distance of 300 pc fromAT 2016jbu. After applying cuts to select only sources witha point source PSF, dolphot detects 84 sources at S/N > PARSEC isochrones (Marigo et al. 2017; Bressanet al. 2012) in three different filter/colour combinations. Weuse the most recent version of the PARSEC models (version1.2S; Chen et al. 2015), and for the purposes of the compar-ison we have applied our foreground reddening and distancemodulus to the
PARSEC models.The progenitor of AT 2016jbu clearly stands out fromthe local population, both in terms of its bright apparentmagnitude and unusual colours. The colour of AT 2016jbushould not be compared to these isochrones; not only willthe F LP filter be strongly affected by H α emission, butas the various filter combinations plotted do not come fromcontemporaneous data, the variability seen in the progeni-tor will significantly affect the apparent colour. Turning tothe 150 pc population, it is clear that no source is found tobe brighter than the log( τ life ) = 7 . AgeWizard and
BPASS models (Eldridge et al.2017; Stevance et al. 2020a,b), we obtain a probability dis-tribution for the age of the resolved stellar population within150 pc around AT 2016jbu (see Fig. 8). The 90 percent confi-dence interval is found to be 10 . -10 . yrs. Additionally, wecan ascertain that the neighbouring population of AT 2016jbuis older than 10 Myrs (5 Myrs) with over 95 (99.8) percentconfidence.Therefore, there is no evidence for a very young environ-ment which would be expected for a 60 (or even 150) M (cid:12) progenitor as proposed for SN 2009ip and η Car (Smith et al.2010; Foley et al. 2011).
MUSE -ing on the local environment
We further investigate the nature of AT 2016jbu by lookingat its local environment in Integral Field Unit (IFU) data. http://stev.oapd.inaf.it/cgi-bin/cmd MNRAS000
We further investigate the nature of AT 2016jbu by lookingat its local environment in Integral Field Unit (IFU) data. http://stev.oapd.inaf.it/cgi-bin/cmd MNRAS000 , 1–24 (2021) rogenitor, environment, and modelling of AT 2016jbu F W F W yrs 10 yrs 10 yrs 10 yrs 1 0 1 2F350LP-F814W21222324252627282930 F L P Figure 7.
Color Magnitude Diagram (CMD) of the stellar population around the site of AT 2016jbu. We show three different colourcombinations each with
PARSEC isochrones with population ages (log( τ life )) given in the upper legend. Blue squares are point sourceswithin 150 pc and green triangles are within 300 pc. The progenitor of AT 2016jbu from the early 2016 HST observations is given as ared star in each panel.
Figure 8.
Probability distribution of the age of the 150 pc stellarneighbourhood of AT 2016jbu. The 90 percent confidence intervalis highlighted in grey
AT 2016jbu was observed on 2017 Dec 2 (+303 d) usingthe VLT equipped with the MUSE instrument in Wide FieldMode. The date cube was obtained as part of a survey ofSN late-time spectra in conjunction with the AMUSING sur-vey of SN environments (Galbany et al. 2016; Kuncarayaktiet al. 2020). We downloaded the pre-calibrated data cubesfrom the ESO archive and present our data analysis for theenvironment around AT 2016jbu in Fig. 9.We fit for spectral features at each spaxel using a Gaussianemission profile with a linear pseudo continuum over a smallwavelength range. For measuring the ratio of H α and H β forthe extinction map we constrain the ratio of the two emissionlines such that Hα/Hβ ≤ .
85 (Case B recombination). Toexclude the effects of AT 2016jbu on the analysis, we exclude any pixel within 3 (cid:48)(cid:48) of AT 2016jbu. We do not account for anystellar absorption effects and as such, values here are lowerlimits. For completeness we include the extracted spectrumof AT 2016jbu in Fig. 10.We show the extinction map across the field of view (FOV)using the method in Dom´ınguez et al. (2013), measured usingthe Balmer decrement. A proxy for the star formation rate(SFR) is measured using L Hα (Kennicutt 1998). L Hα was cor-rected for extinction using the Balmer decrement (Vale Asariet al. 2020). We also plot a metallicity map using the metal-licity indicators given by Dopita et al. (2016).Fig. 9 does not include the core of the host galaxy, nor thesouthern arm. AT 2016jbu is located north of the southerndistorted spiral arm of NGC 2442 and is still clearly presentin NGC 2442 almost a year after maximum as seen in thewhite light image constructed from the datacube. The FOV(1 (cid:48) × (cid:48) ) does however include the location of SN 1999ga (Pa-storello et al. 2009) as well as a luminous region in the cen-ter frame. This “ Super-Bubble ” has been noted by previousauthors (Pancoast et al. 2010), and is seen in the irregularkinematic pattern seen in the center of the FOV. This is aspherical-looking area within the diffuse region to the south-west of the nuclear region, with a diameter of ∼ MNRAS , 1–24 (2021) S. J. Brennan et al. h m s s s -69°32'48"33'00"12"24"36" RA D e c
500 pc 500 pc 500 pc500 pc 500 pc F [ e r g Å c m s ] E ( B - V ) V e l o c i t y [ K m s ] + L o g ( O / H )[ D o p i t a ] S F R [ M y r ] Figure 9.
IFU analysis of the environment of AT 2016jbu. The spectral cube was corrected for Milky Way extinction and redshift.Observations were taken on 2017 Dec 2 (+303 d). Data is orientated such that north is up and east is to the left. Included in each panel isa horizontal scale bar showing 500 pc. We include a white light image (5000–7000˚A) (top left), an extinction map (top middle) based onDom´ınguez et al. (2013), a velocity field plot from H α corrected for recessional velocity (top right), star formation rate based on Kennicutt(1998) (bottom left) and a metallicity map (bottom right) based on Dopita et al. (2016). The location of AT 2016jbu is marked with ared circle of radius 3 (cid:48)(cid:48) . We also include the location of SN 1999ga as a square to the south west of AT 2016jbu. Data is not shown whereEW < (cid:48)(cid:48) of AT 2016jbu. O INa D [Fe II][O I] [O I] [Ca II]Fe IFe IIFe II Fe II Fe IIFe IIFe IIFe II Ca NIRHe I Paschen4500 5000 5500 6000 6500 7000 7500 8000 8500 9000 9500
Wavelength [ Å ]1.52.02.5 F [ e r g s c m Å ] O I O I O I O I He IHe IHe IHe IHe I H N I H Figure 10.
Extracted spectrum of AT 2016jbu from VLT+MUSE. Spectrum was extracted using a 1 (cid:48)(cid:48) aperture at the transient positionand corrected for redshift and Galactic extinction. We marks strong emission features in red, forbidden transition lines are marked inblue. The transient appears relatively blue even at ∼ +10 months, a possible sign of ongoing interaction. by Ryder et al. (2001); Pancoast et al. (2010). Another is rampressure stripping acting in a south to north direction, whichmay explain the massive H i cloud to the north of NGC 2442(Ryder et al. 2001).As noted by previous authors (Pancoast et al. 2010; Ryderet al. 2001; Pastorello et al. 2008), this Super-Bubble regionis in the vicinity of both AT 2016jbu and SN 1999ga. Thisregion shows a high SFR and is bright in B-band, both signsof massive star formation. High SFR is linked with a high SN rate (Botticella et al. 2012) and it is a fair assumption thatthe general location of this
Super-Bubble is likely to host CC-SNe, as is obvious from SN 1999ga. The top middle panel inFig. 9 maps the extinction across the FOV using the BalmerDecrement (Dom´ınguez et al. 2013). We find a value for thelocal extinction ( E B − V < .
45) within 500 pc of AT 2016jbuwith a similar value seen across the FOV. The top right panelin Fig. 9 gives the velocity dispersion across the FOV. The lo-cation of AT 2016jbu lies in an area moving at ∼ −
100 km s − MNRAS000
100 km s − MNRAS000 , 1–24 (2021) rogenitor, environment, and modelling of AT 2016jbu (image corrected for red-shift: z = 0.00489). The bottom leftpanel shows a pseudo SFR based on the extinction correctedH α emission (Kennicutt 1998). The figure shows two brightregions of star formation, which is clear from the white lightimage. AT 2016jbu is situated on the outskirts of a moder-ate star-forming region, ∼ − M (cid:12) yr − . SN 1999ga lies onthe edge of the brighter star forming region. We include ametallicity map (bottom right panel) following the metallic-ity indicators from Dopita et al. (2016). The full FOV yieldsan approximately solar environment, with the median metal-licity across the field as 8.66 dex ( Z ≈ . The bolometric lightcurve for AT 2016jbu is computed using ugiz , UBV R , JHK , Gaia G , W W Swift +UVOT
UVW2 , UVM2 , UVW1 , U , B , and V .All calculations were carried out using Superbol (Version1.7; Nicholl 2018).Effective wavelengths were taken from Fukugita et al.(1996) and zeropoint flux energies were taken from Tonryet al. (2018), while Superbol was modified to also handle ourWISE data. Extinction values in each filter were computedusing the York Extinction Solver (McCall 2004). All magni-tudes were converted to F λ , and interpolated where necessaryto account for epochs without specific filter coverage, taking r -band as the reference filter. Black body fitting is performedfor photometric bands that are centered on λ > F λ usingthe trapezoidal rule between 0.2 and 4.5 µ m ( UVW2 to W2 ).We present the results of our blackbody fitting in Fig. 11.AT 2016jbu is an interacting transient showing strong emis-sion lines. Interpreting the blackbody evolution of photom-etry alone may be misleading, due to the uncertainty asto whether the photometry is continuum-dominated or line-dominated. For completeness we investigate blackbody fitsfrom our optical spectra. A black body function was fittedto the optical spectra presented in Paper I while excludingstrong emission features and only fitting for λ > ∼ +125 d. After thistime, our observations become strongly line-dominated andblackbody fitting becomes unreliable. In Fig. 11 we show the blackbody luminosity ( L BB ) , radius( R BB ), and temperature ( T BB ) fits from Superbol . In orderto understand this evolution, we try here to connect this tothe velocities seen in the spectra.The H emission for AT 2016jbu shows two distinct absorp-tion components (see Paper I). The first component is seenin a P Cygni profile that is present up until ∼ Event B maximum. The second component is present for ∼ https://github.com/mnicholl/superbol is common in SN 2009ip-like transients. The presence of tworegions of material with different velocities lends credenceto the idea that the interaction with some material during,or prior to, Event A is the main source of energy input for
Event B (Fraser et al. 2013a; Th¨one et al. 2017; Elias-Rosaet al. 2016; Benetti et al. 2016). Fitting a P Cygni absorptionprofile gives a maximum velocity of ∼ −
850 km s − , with abulk velocity of ∼ −
600 km s − for the slower absorptionfeature seen in the Balmer lines. We refer to this materialas Shell 1 . The higher velocity absorption (which we refer toas
Shell 2 ) has a maximum velocity from the blue edge ofthe line of ∼ − − with the bulk of the materialat ∼ − − . Using these velocities, we attempt toconstrain ejection/collision times.We assume a dense shell causing the P Cygni absorp-tion was ejected at ∼ −
90 d. This is the first epoch whereAT 2016jbu is detected in VLT + FOcal Reducer/low dis-persion Spectrograph 2 (FORS2) imaging and is most likelyassociated with the start of
Event A . Assuming unimpededexpansion, we include the distance travelled by this material(
Shell 1 ) in Fig. 11 as a orange band.The ejection epoch for the material causing the second highvelocity absorption component is open to debate. There is noevidence for this additional absorption in optical spectra at −
24 d and it is only seen on −
15 d. Under the presumptionthat we do not see this shell of material (
Shell 2 ) until it in-teracts with the pre-existing material or until it is no longerocculted by an existing photosphere, we find that a shell mov-ing at ∼ − for ∼ R BB ∼ . × cm. We include the distance travelled by Shell 2 in Fig. 11 as a blue band. We can constrain the ejec-tion date of this HV material to ∼
21 days before maximumlight with the collision date (when
Shell 2 catches up to
Shell1 ) at ∼
19 days before maximum light.We draw attention to the blackbody evolution over theperiod −
19 d to −
13 d. During this timeframe we see aninflection between the decline of
Event A and the rise of
EventB . Although we have low-cadence coverage during
Event A (Paper I), the distance travelled by
Shell 1 (orange band)follows R BB quite well during Event A . R BB then contractsslightly beginning around −
30 d to a minimum at −
19 d. At ∼ −
19 d R BB then begins to increase at a velocity similar tothe velocity profile of Shell 2 . This implies that the blackbodyradius now follows this material, which is likely
Shell 2 withadditional material swept-up from
Shell 1 . We initially find T BB at ∼ ∼ −
30 d. T BB evolves exponentially from 6000 K at −
30 d to 12000 Kat −
12 d. After
Event B maximum (marked as
Decline in Fig11), T BB cools to ∼ Knee epoch and slightlyincrease to ∼ Ankle epoch.It is important to note that we see both components inspectra during the first month of
Event B . Additionally, theFWHM and velocity offset does not significantly evolve dur-ing the first few months (see Paper I). We suggest that
Shell1 is highly asymmetrical with high density material alongequatorial regions and low density along polar regions. Thisis motivated by the spectral evolution of the H α profile, theevolution of R BB and the degenerate appearance of the H α emission lines in SN 2009ip-like objects, see Paper I for fur-ther discussion. If Shell 2 is spherically symmetric, some ma-terial of
Shell 2 would not interact with
Shell 1 and expandfreely along the polar regions.
MNRAS , 1–24 (2021) S. J. Brennan et al. L BB [ e r g s ] R i s e D e c li n e P l a t e a u K n ee A n k l e T BB [ K ]
50 25 0 25 50 75 100 125Days since V-band Maximum10 R BB [ c m ] P-Cygni+BroadEmission SecondAbsorption R e d E m i ss i o n B l u e E m i ss i o n Shell Shell Figure 11.
Blackbody luminosity (top panel), temperature (middle panel), and radius (lower panel) of AT 2016jbu calculated using
SuperBol (Nicholl 2018). The orange shaded region shows the linear distance travelled by the slower moving material
Shell 1 causingthe P Cygni absorption. The orange shaded region is the same for the faster moving material
Shell 2 . The lower and upper bounds foreach band are bulk and max velocities respectively. In the second and third panel, we include T BB and R BB fits to our optical spectra(Paper I). We include approximate epochs where specific H α features emerge in the R BB panel. We include vertical dashed lines indicatingsignificant phases in the lightcurve evolution. In the lower panel we include approximate epochs where specific features are seen in H α , asdiscussed in the text. We include labels indicating when certain spectral compo-nents appear in the H α in Fig. 11, bottom panel. We see thatthe HV blue absorption feature coincides with the evolutionof Shell 2 ; this absorption is clearly seen at −
18 d and isdetected until +5 d with fitting-model dependent tentativedetections up to +10 d. This second absorption componentappears during the rise in R BB during Event B and van-ishes when R BB reaches its maximum at ∼ +7 d. At ∼ +9 d R BB remains at a constant value and we see the emergenceof a broad, red shoulder emission in H α at ∼ − ,FWHM ∼ − . This may follow material expandingat ∼ − , a receding photosphere or both. Severaldays later the blue emission feature appears in H α and re-mains until late times. At +18d this blue emission is centeredat ∼ − − with FWHM ∼ − .Photons from the interaction site between Shell 1 and
Shell2 may be diffusing outwards at this epoch. We see that thered shoulder emission only appears after R BB reaches its max-imum values, shortly followed by the blue shoulder emissiona week later. This leads to our conclusion that Shell 1 ispartially asymmetric and when
Shell 2 collides with it,
Shell1 is partially engulfed. The interaction between these twoshells then becomes apparent at ∼ +7 d when the asymmetric emission features are clearly seen in H α . A disk or torus-likeenvironment can explain such a trend.The R BB peaks at 1 . × cm, at ∼ EventB maximum and remains roughly constant until the
Knee phase. Thereafter, there is a drop of ∼ − × cm/dayuntil the beginning of the Ankle phase. R BB remains roughlyconstant at ∼ . × cm up until the seasonal gap beginsat +140 d. This epoch coincides with a narrowing of bothred and blue emission features and an increase in Equiva-lent Width (EW) of both components. This may represent atime when opacities drop significantly and there is less pho-ton scattering. Using this collision scenario as the dominantenergy input for this transient, we will explore the necessaryenergy budget in Sect. 6. Using the evolution of R BB we canbetter understand the nature of the explosion of AT 2016jbu,and we will further discuss this in Sect. 7.2. The nature of the energy input of AT 2016jbu and SN 2009ip-like transients is debated. If AT 2016jbu is indeed a CCSNthen this energy comes from an imploding iron core. The earlylightcurve is powered by the fast moving SN ejecta material.
MNRAS000
MNRAS000 , 1–24 (2021) rogenitor, environment, and modelling of AT 2016jbu Ejecta interacting with a dense CSM can power the lightcurvefor many years (see Fraser 2020, and references therein). Ifthe transient is a CCSN, after the ejecta expands and cools,the late time lightcurve is powered by the radioactive decayof Ni. We discuss the possible presence of Ni in Sect. 6.1.If AT 2016jbu is a SN, then it is spectroscopically classedas Type IIn, meaning we see strong signs of interaction witha dense, slow-moving CSM. This is clear from the narrowline emission at early times, and the evolution of asymmetricfeatures in H α . We discuss the energy input from ejecta/CSMinteraction in Sect. 6.2 Ni mass
A product of CCSNe is explosively synthesised radioactive Ni, whose decay can power the late time lightcurve of H-rich supernovae after the hydrogen ejecta have fully recom-bined and any additional interaction has stopped. Anderson(2019) find that for H-rich, Type II SNe, the median valuefor the amount of Ni synthesised is 0.032 M (cid:12) . We show ourattempt to fit for a nickel decay tail in Fig. 12 (green dashedline). We find that the pseudo-bolometric lightcurve shows adecay that is consistent with that of radioactive nickel decayduring the
Ankle stage.Determining an explosion epoch for AT 2016jbu is con-tentious. The transient is clearly detected at -90 d inVLT+FORS2 imaging. We determine in Sect. 5.1 that a sec-ond eruption (that may represent a genuine CCSN) occurredat ∼ -21 d. Using eq. 6 from Nadyozhin (2003), and taking theexplosion epoch as ∼ -90d, we find a value of M Ni ≤ .
033 M (cid:12) and taking ∼ −
21 d we find a value of M Ni ≤ .
016 M (cid:12) . Fol-lowing the arguments made in Sect. 5.1 we will take the latterexplosion date as the more plausible motivated by the appar-ent second eruption at −
21 d, indicating a potential CCSN.We will return to this point in Sect. 7.1.This limit on Ni is consistent with other SN 2009ip-liketransients. However, it is clear that during this time thereis still on-going CSM-interaction, as demonstrated from themulti-component H α profile in Paper I, and as such, this valueshould be considered a conservative upper limit, assumingany Ni is produced at all.
A previously explored scenario for the double-peakedlightcurve of SN 2009ip-like objects is
Event A representsa low energy eruption from the progenitor star and
Event B is powered by the interaction between the ejecta from thiseruption, and some pre-existing CSM that was ejected in thepreceding year (Mauerhan et al. 2013a; Fraser et al. 2013a;Th¨one et al. 2017). We measure the radiated energy releasedfrom
Event A ( −
90 d to −
21 d) as 3 . × erg and the en-ergy from Event B ( −
21 d to +450 d) as ∼ . × erg.Fraser et al. (2013a) find a similar value for SN 2009ip at ∼ . × erg.If we assume that Event A is a symmetric explosion we canapproximate it using an
Arnett Model (Arnett & Chevalier1996). Taking the diffusion timescale for a photon to be t d ≈ L D , where D is a diffusion coefficient with D ≈ λc = cρ κ .Assuming that Event A corresponds to the adiabatic expan- sion of a photosphere, and assuming L ≈ R , we can describethe diffusion timescale as: τ d = (cid:18) κ M ej π c v ej (cid:19) / , (2)by substituting in R ≈ τ d × v sh and ρ ≈ M ej π R , where M ej and v ej are the ejecta mass and velocity respectively, κ is theopacity of the ejecta, and c is the speed of light. We take therise time in r -band of Event A to be similar to the diffusiontime, and we get a value of ∼
60 days. We assume the P Cygniminima follows this dense material ejected prior to, or during,the beginning of
Event A , as suggested by Th¨one et al. (2017)for SN 2015bh. Using Eq. 2 and taking v ej ≈
700 km s − and assuming a mean opacity of κ = 0.34 cm g − (assum-ing e − scattering dominates in the H-rich ejecta) we find M ej for the Event A is ∼ (cid:12) giving a kinetic energy of ∼ . × erg. This value is a factor 10 less than is requiredto power Event B .This is a crude approximation as the assumptions madewill not reflect the morphology of the
Event A eruption, asdiscussed in Sect. 5.1, and as such this value is an upper limit.However, it is unlikely that an eruptive outburst around thebeginning of
Event A interacting with pre-existing CSM isthe dominant power source for
Event B as the energy budgetalone is off by approximately two orders of magnitude.Assuming free expansion, the constrained ejection times,and velocities for our multiple shell models given in Sect. 5.1,the beginning of
Event B coincides with material from bothshells being at the same location R BB ≈ . × cm(Fig. 11). This suggests that Event B is powered from thecollision at ∼ −
19 d of
Shell 2 which interacts with the slowermoving material ejected at the beginning of
Event A ( Shell1 ).It is difficult to measure the mass of
Shell 2 . If we as-sume that
Event B is powered solely by CSM interactions,we calculate an upper limit of M ej ∼ .
37 M (cid:12) travellingat ∼ − can account for the energy seen, whileallowing for an extremely low porosity (or overlapping sur-face area between ejecta and CSM) of 10%. This value willchange depending on the opening angle of Shell 1 , as exploredin disk interaction models (Vlasis et al. 2016; Suzuki et al.2019; Kurf¨urst et al. 2020). Even with this conservative esti-mate, our values of M ej are much lower than those seen inCCSNe or η Car. However, extremely low porosity (e.g. 1%)would allow for a few M (cid:12) of ejected material if we assume noinput to the lightcurve from radioactive decay.Although observed after peak luminosity, both SN 2013Land SN 2010jl showed a plateau phase after maximum light(Ofek et al. 2014; Taddia et al. 2020). This trend is dis-cussed by Chevalier & Irwin (2011); SN ejecta interactingwith a dense mass loss region can form a plateau in lumi-nosity lasting the duration of the shock interaction, and end-ing when the entire interaction material is shocked. As thephoton mean free path increases with the geometric expan-sion of the CSM, the innermost regions of the interaction arerevealed. This was suggested to explain the double-peakedspectral profiles of SN 2010ij (Ofek et al. 2014), SN 2013L(Taddia et al. 2020), and iPTF14hls (Andrews & Smith 2018;Sollerman et al. 2019; Moriya et al. 2020) at late times. Weuse the emergence of the blue emission feature and the de-crease of the peak velocity offset as a proxy for the shock
MNRAS , 1–24 (2021) S. J. Brennan et al.
100 50 0 50 100 150 200 250 300 350 400 450 500 550 600Days since V-band maximum10 L b o l [ e r g s ] R i s e D e c li n e P l a t e a u K n ee A n k l e
40 60 80 100 12010 L b o l [ L ] Figure 12.
Pseudo-bolometric luminosity of AT 2016jbu using
SuperBol (Nicholl 2018). We include the luminosity shock function (Eq.4; solid red line) and a radioactive decay tail fit (green solid line). Both functions are extrapolate until the end of observations (dashedlines). Both function are fitted to the post-ankle stage and we include a zoom-in of this area in the top right. We find a Ni mass is0.016 M (cid:12) (assuming the SN explosion date as −
21 d) and ˙ M is 0.05 M (cid:12) yr − for Eq. 4. front. We discuss the evolution of this feature in Paper I. Wefit a declining power law function to the peak velocity of theblue emission from +20 to +120 days which is well fit by: v blue ( t ) ≈ (1375 ± × (cid:18) t d (cid:19) − . ± . km s − , (3)Both red and blue emission components follow Eq. 3 well (thered component has a different normalisation constant) up un-til the seasonal gap (+140 days). After that both componentsmaintain at a higher velocity and coast at ∼ ± − upuntil the end of our spectroscopic observations (+575 days),see Paper I. Under the assumption of steady-state mass loss,the luminosity from CSM-shock interaction can be describedby: L sh = (cid:15)
12 ˙ Mv wind v ej , (4)where L sh is the luminosity from CSM-ejecta interaction, (cid:15) is the conversion efficiency from kinetic to thermal energy(taken to be 0.5), v ej is the ejecta velocity, which is set toEq. 3, and v wind is the wind velocity. We fit Eq. 4 to ourbolometric lightcurve during the period from the Knee stageup until the beginning of the seasonal gap. Fitting to thistime-frame gives an upper limit for ˙ M ≈ .
05 M (cid:12) yr − , if weassume an LBV wind with v wind ≈
250 km s − (we find asimilar value for v wind from our earliest H α profile). Setting v wind ≈
700 km s − , the value of the P Cygni minima, weobtain ˙ M ≈ .
14 M (cid:12) yr − .We base the above calculations on the assumption thatthe luminosity between +70 d to +140 d is shock-CSM inter-action dominated, with no other major contributing energysource i.e. no major contribution from radioactive decay. IfAT 2016jbu is surrounded by a dense, disk-like CSM and theplane of the disk is somewhat face on, the assumption thatthis phase is interaction dominated is motivated by models(e.g. Fig. 11 from Vlasis et al. 2016). These models show asimilar lightcurve shape to AT 2016jbu, including a tail re-sembling radioactive Ni decay at ∼ +80 days past maximumbrightness (these models assume no Ni). Symmetric ejecta and disk interaction models show that the energy input at the
Knee stage is dominated by this ejecta-disk interaction. Wewill return to the possibility of a disk-like CSM in Sect.7.2.After the seasonal gap (+140 days), the velocity of thered/blue emission does not follow Eq. 3 and the bolometricluminosity does not follow Eq. 4. At this point the lightcurvehas increased in brightness, which is clearly seen in Fig. 12.However by ∼
400 d, L bol fades below the extrapolated valuefrom Eq. 4.After the seasonal gap both red and blue emission lineshave similar FWHM, ∼ − , with the red emissionhaving a slightly larger width, but converging to the FWHMof the blue component at ∼
400 d. If the red/blue emissionfollows the shock interaction, this suggest an increased veloc-ity of the shock front. Conserving mass flux in the shock wehave ρ v = ρ v where subscript 1,2 represent the post-and pre- shock regions respectively. If the shock transversesto a lower density CSM environment this can account for theincreased velocity seen. This might indicate that the shockhas now reached a lower density environment, perhaps cre-ated by the series of outbursts in the years prior. However,it is not obvious how interaction with a less dense regionof CSM would account for the increased luminosity as wellas the increased strength of He i emission lines (also seen inSN 1996al; Benetti et al. 2016) at this time (Paper I). In the following section we will discuss the nature ofAT 2016jbu. There is much debate as to the natureSN 2009ip-like objects (Pastorello et al. 2008; Smith &Mauerhan 2012; Fraser et al. 2013a; Smith et al. 2014a;Margutti et al. 2014; Graham et al. 2014; Pastorello et al.2019a). Any scenario for AT 2016jbu or SN 2009ip-like tran-sients needs to account for all of the following points: Outbursts reaching an absolute magnitude of M r ∼− ± MNRAS000
400 d. If the red/blue emissionfollows the shock interaction, this suggest an increased veloc-ity of the shock front. Conserving mass flux in the shock wehave ρ v = ρ v where subscript 1,2 represent the post-and pre- shock regions respectively. If the shock transversesto a lower density CSM environment this can account for theincreased velocity seen. This might indicate that the shockhas now reached a lower density environment, perhaps cre-ated by the series of outbursts in the years prior. However,it is not obvious how interaction with a less dense regionof CSM would account for the increased luminosity as wellas the increased strength of He i emission lines (also seen inSN 1996al; Benetti et al. 2016) at this time (Paper I). In the following section we will discuss the nature ofAT 2016jbu. There is much debate as to the natureSN 2009ip-like objects (Pastorello et al. 2008; Smith &Mauerhan 2012; Fraser et al. 2013a; Smith et al. 2014a;Margutti et al. 2014; Graham et al. 2014; Pastorello et al.2019a). Any scenario for AT 2016jbu or SN 2009ip-like tran-sients needs to account for all of the following points: Outbursts reaching an absolute magnitude of M r ∼− ± MNRAS000 , 1–24 (2021) rogenitor, environment, and modelling of AT 2016jbu A faint event, reaching an absolute magnitude of M r ∼− ± An second event a few weeks later, reaching an abso-lute magnitude of M r ∼ − . ± . ∼ ,
000 km s − . Ejected Ni mass of (cid:46) (cid:12) . No directly observed synthesized material, either fromexplosive nucleosynthesis or late-stage stellar evolution.A possible addition to this list is double-peaked emissionlines. This is seen in the majority of SN 2009ip-like transientsalthough, ironically, not SN 2009ip itself.We address the probable progenitor in Sect. 7.1. Using ourhigh cadence multi-chromatic photometry presented in PaperI, and the bolometric evolution from Sect. 5, we will presenta likely explosion model and circumstellar (CS) geometry forAT 2016jbu, that can be extrapolated to other SN 2009ip-liketransients in Sect. 7.2. We will discuss the validity of a CCSNscenario in Sect. 7.3, and the possibility of the progenitorbeing in an interacting binary system in Sect. 7.4
The events of SN 2009ip-like transients may represent a criti-cal step in the late time evolution of massive stars. A dramaticincrease in luminosity allows for supper-Eddington winds andhigh mass-loss rates, however the mechanism resulting inthese outburst is unknown. Observations of shock features inthe Homunculus Nebula around η Car may even point to ex-plosive mass loss. Furthermore, in the classical picture, LBVsshould not be SN progenitors as they have just transitionedto the He-core burning stage.It is generally thought that SN 2009ip-like transients arisefrom massive stars (Foley et al. 2011; Pastorello et al. 2013;Fraser et al. 2013a; Smith et al. 2014b; Fraser et al. 2015;Smith et al. 2016b; Elias-Rosa et al. 2016; Pastorello et al.2019a). The progenitor of SN 2009ip is thought to be a 60–80 M (cid:12)
LBV from pre-explosion images (Smith et al. 2010;Foley et al. 2011). However, this was measured in a singleband only, which will be heavily contaminated by H α flux.As shown in Fig. 4, the bright contribution of H α in F350LP will provide misleading SED fitting results. While LBVs ex-perience erratic mass loss as they undergo a short transitionfrom O-Type to the WR stars, AT 2016jbu appears to be toolow mass for this scenario, with a progenitor mass of only ∼
20 M (cid:12) (Sect. 2.2). We note that this relatively low masswas found while taking into account the effect of H α emis-sion on the SED.The progenitor mass for AT 2016jbu is the most secure forany SN 2009ip-like transient in literature, as it is based ona broad optical to NIR SED. From our SED fitting to theearly 2016 HST data, we find the color of the progenitor isconsistent with a yellow hyper-giant. Using dusty modellingand matching the output spectra to these colour values wefind values for L and T which are consistent with a mass ofsingle star of 17–22 M (cid:12) . Moreover, the local environment,which can be be assumed to be composed of a similar stellarpopulation, demonstrates that we can effectively rule out avery young population (expected for a 60–80 M (cid:12) star).In order to explore the progenitor further, we turn to a gridof stellar models created with the
BPASS code. The
BPASS stellar model library contains the time varying properties of over 250,000 star systems for a grid of initial parameters and apopulation containing a realistic fraction of binary and singlestar systems (Eldridge et al. 2017; Stanway & Eldridge 2018).Using hoki (Stevance et al. 2020a), we searched for modelsmatching the observed temperature and luminosity of theprogenitor of AT 2016jbu, considering both the possibilityof a terminal core-collapse supernova and of a non-terminalevent.For the CCSN (non-terminal explosion) scenario we find792 (3328) matching stellar models, and only 3 (5) of thesemodels correspond to single star systems.The ZAMS and final mass distributions, as well as theevolutionary tracks for both interpretations, are presentedin Fig. 13. The ZAMS mass distribution of the CCSN pro-genitor models is noticeably lower and narrower than thatcorresponding to the non-terminal explosion matches. We canevaluate the mean and standard deviation for the two scenar-ios: M = 12 . (cid:12) , σ = 1 . (cid:12) and M = 18 . (cid:12) , σ = 13 M (cid:12) for CCSN and non-terminal explosion cases, respectively.The most salient point is that very massive stellar pro-genitors (e.g. classical LBVs) are confidently excluded. Ad-ditionally, the non-terminal explosion scenario requires moremassive (and hence shorter-lived) progenitors compared tothe CCSN interpretation. We can quantify this further bycalculating the mean lifetime of our matching models. Wefind log( τ life ) = 7 . +0 . − . yrs and log( τ life ) = 7 . +0 . − . yrs forthe CCSN and non-terminal explosion scenarios, respectively.Comparing these lifetimes to the age constraints placed onthe neighbourhood of AT 2016jbu in Section 4, the CCSN sce-nario is preferred over the non-terminal explosion, althoughthe latter cannot be excluded.An interesting problem to solve with the CCSN scenario isthat of the presence and geometry of the CSM, as discussed inSect. 5. The LBV-type winds invoked in Sect. 6.2 do not applyto lower mass progenitors; indeed we find an average mass-loss rate over the last 1 Myr of log( ˙M) = − . +0 . − . M (cid:12) yr − and log( ˙M) = − . +0 . − . M (cid:12) yr − for the CCSN and non-terminal scenario, respectively.One can sustain a dense CSM even with a low mass lossrate provided the wind velocity is sufficiently small. Usinglog( ˙M)/ v wind as a proxy for wind density, we compare theaverage ratio found in our models to that assumed in Sect 6.2.We find that for both sets of progenitor models ˙M/ v wind ≈ − , compared to a value of ≈ − found for AT 2016jbu.Thus, we can confidently assert that steady winds are notable to create the CSM observed in AT 2016jbu.The alternative is episodic mass loss resulting from Rochelobe overflow or common envelope evolution (CEE). We ex-amined the CCSN progenitor models found in BPASS andfind that 117 models are in a CEE phase at the time of CCSNexplosion; furthermore we find another 163 undergoing masstransfer. Similarly, for the non-terminal models we find that286 models are in the CEE phase and 275 are undergoingmass transfer at the point where they match the observedL and T of the AT 2016jbu progenitor. Consequently, the
BPASS models reveal that the peculiar combination of prop-erties and environment of AT 2016jbu can be explained bybinary interactions. https://github.com/HeloiseS/hoki MNRAS , 1–24 (2021) S. J. Brennan et al.
Figure 13.
Mass distributions (ZAMS and final) and evolutionary tracks of
BPASS models matching the L and T derived from dusty modelling. Upper (lower) panel shows the CCSN (Non-terminal) scenario. Each evolutionary track is plotted at a low transparency andtherefore the lighter the tracks, the rarer they are in our matches. We mark the search area in T and L from Sect. 2 in each HR diagram.
A radially confined, dense, disk-like CS environment has beensuggested for SN 2009ip-like transients (Smith et al. 2014a;Margutti et al. 2014; Levesque et al. 2014; Margutti et al.2014; Fraser et al. 2015; Benetti et al. 2016; Tartaglia et al.2016a; Pastorello et al. 2018; Andrews & Smith 2018) aswell as other Type IIn SNe (van Dyk et al. 1993; Benetti2000; Stritzinger et al. 2012; Benetti et al. 2016; Andrewset al. 2017; Nyholm et al. 2017) and super-luminous super-novae (SLSNe) (Metzger 2010; Vlasis et al. 2016). Double-peaked line profiles are signs of a CS disk or torus like en-vironment. This is similar to the presence of double-peakedH α (and other emission lines) originating from an accretiondisks in active galactic nuclei (e.g. Shapovalova et al. 2004)as well as double-peaked emission from Be/shell stars (e.g.Andrillat et al. 1986) although their formation and poweringmechanism are extremely different. We show in Paper I thatAT 2016jbu and other SN 2009ip-like objects show a degree of degeneracy in the appearance of their H α profiles, whichmay be explained with a simple viewing angle effect.We suggest that AT 2016jbu has a torus-like structure, suchas has been suggested for η Car (see review by Smith 2009)and SN 2009ip (Levesque et al. 2014; Margutti et al. 2014;Reilly et al. 2017), and a significant portion, if not all, ofthe explosion energy is a result of a ejecta-torus interaction,which dominates around a month after maximum light. Weare motivated by the persistent double-peaked profile seen inBalmer lines as well as the shape of the lightcurve.Recently, several groups have modelled the interaction ofejecta with aspherical CSM (Vlasis et al. 2016; McDowellet al. 2018; Suzuki et al. 2019; Kurf¨urst et al. 2020; Nagaoet al. 2020). Vlasis et al. (2016) has modeled the lightcurveevolution of a spherically symmetric eject colliding with adisk-like CSM. We find that similarities between these modelsand AT 2016jbu. One important feature is after ∼ +80daysthese models seem to follow a decay similar to that expectedfrom Ni. The energy source at this time is solely poweredfrom CSM interaction and not from radioactive decay. How-
MNRAS000
MNRAS000 , 1–24 (2021) rogenitor, environment, and modelling of AT 2016jbu * Not to scaleBeginning of Event A Event B
Beginning of
Plateau stageAnkle
Stage onwardsPrior to/beginning of
Event A -31 d+0 d+18 d+345 d Shell 1 EjectedShell 1 and 2 collide Ejecta/Disk Interaction Dominated Shell 2 becomes partially transparentInternal heating from interactionPhotons from interaction site escape freelyShell 1 / DiskShell 2 / Symmetric EjectaRed and Blue shoulder Emission Shell 1 / Disk Shell 2 / Symmetric EjectaProgenitor
Figure 14.
Proposed geometry and explosion scenario for AT 2016jbu. This diagram illustrates the discussion in Sect. 5.1 and Sect. 6.2.Left panels show a simplified illustration of the CS environment around the progenitor. Middle column shows H α profile at correspondingepochs. The upper right panel shows the bolometric luminosity and the lower right shows the blackbody radius. We include the distancetravelled by Shell 1 (blue shaded region) and
Shell 2 (blue shaded region).
Event A begins with the eruption of a massive shell of materialthat is either an oblate spheroid or a radially confined disk. This originates from the progenitor system (given as a green filled circle).The eruption and expansion of this
Shell 1 is seen in L bol and R BB , both peaking at ∼ −
27 d. As discussed in Sect. 5.1, at ∼ −
21 d,
Shell 2 / symmetric ejecta (blue) collide and partially engulf the disk. L bol and R BB follow the expansion of the opaque Shell 2 at avelocity seen in the HV material seen in H α . Both L bol and R BB peak at ∼ Shell 2 then becomes optically thin and the photopsherebegins to move inwards in velocity space. There is a linear decay in R BB until ∼ +22 d or the beginning of the plateau stage. Photonsoriginating from the interaction site between Shell 1 and
Shell 2 begin to diffuse outwards, as
Shell 2 has become partially transparent.This coincides with the metamorphosis of the blue HV absorption to an emission profile. Up until the seasonal gap, these emission linesdecay in velocity with an index ∼ . R BB plateaus at ∼ +25 d due to effective internal heating from the site of interaction. At ∼ +45 d,the Knee stage drops in L bol and R BB with R BB at a slightly higher value when compared to the beginning of Event B . Both red andblue emission lines narrow at this stage which may signify any intervening material is now completely optically thin and any escapingphotons undergo minimal scattering. The dominant source of energy is now shock interaction in the disk. We include a possible survivingprogenitor in the final two panels. ever, these models cannot explain the increased brightnessin AT 2016jbu after the seasonal gap, although this likelyreflects a clumpy CSM and would require fine-tuning of theCSM density profile.Models by Kurf¨urst et al. (2020) have modeled ejecta in-teraction with a disk-like CSM for a range of viewing angles(Model A and fig. 12 in Kurf¨urst et al. 2020) demonstratinga clear viewing angle degeneracy, with looking down throughthe polar direction showing the greatest “ double-peaked ”-nessand looking through the equator of the disk showing theleast (i.e singularly peaked emission lines). This can natu-rally explain the variations in H α appearance found amongstSN 2009ip-like transients (Paper I). In this context SN 2009ipis viewed down through the polar regions and AT 2016jbu would be viewed through a significant amount of disk mate-rial.A disk-like CSM has been proposed for SN 2009ip toexplain the photometric and spectroscopic features seen(Levesque et al. 2014; Margutti et al. 2014; Reilly et al. 2017).Reilly et al. demonstrates using spectropolarimetric observa-tions that a flat, edge-on or inclined CSM can be fitted toSN 2009ip. However, it has been found that solutions fromsuch toy models are extremely degenerate, even for a smallnumber of parameters, and that incorrect solutions can resultin acceptable fits to observed data (Stevance 2019; Stevanceet al. 2019). Using complex geometries, without a full explo-ration of parameter space, as done in Reilly et al. (2017),or multiple absorbing regions (as we suggest in this work), MNRAS , 1–24 (2021) S. J. Brennan et al. would result in a greater number of free parameters and agreater number of degeneracies.However, our interpretation of a disk-like geometry forAT 2016jbu is motivated solely by the photometric and spec-troscopic evolution, and is not affected by the uncertaintiesin the interpretation of spectropolarimetric observations forSN 2009ip. We provide a illustration of such a geometry ap-plied to AT 2016jbu in Fig. 14.
The main point of controversy is whether AT 2016jbu andSN 2009ip-like transients are indeed CCSNe; meaning theprogenitor has been destroyed and the transient will eventu-ally decay following a radioactive decay tail. This begs thequestion; If these are indeed CCSNe, when did core-collapseoccur?
SN 2009ip-like transients display two broad, luminousevents, rather than the singularly peaked lightcurve, typi-cally associated with SNe. Mauerhan et al. (2013b) suggestthat
Event A is a CCSN and
Event B is a result of ejecta in-teracting with a dense CSM. In this scenario, with respect toAT 2016jbu, the duration of Event A ( ∼
60 days) is the timeneeded for this ejecta to reach the inner edge of the CSM.This scenario would be consistent with the early evolutionof R BB expanding at ∼
700 km s − ; however this velocityis implausibly slow for SN ejecta. More problematic still, at −
21 d we see an increase in velocity where R BB expands at ∼ − . In the case of core-collapse, we hence regardit as more plausible that Event B is the terminal explosion ofthe progenitor, where the ejecta interacts with a non-terminaloutburst that was ejected at ∼
700 km s − around the startof Event A . This scenario is also reinforced by the rise time( ∼
17 days) and peak magnitude ( M V ∼ − . EventB (Nyholm et al. 2020).We find a low value of Ni of (cid:46) (cid:12) for AT 2016jbu,consistent with other SN 2009ip-like transients. During thistime there is also strong on-going CSM interaction so thisvalue is an upper limit. Paper I demonstrates the absence ofspectral lines normally associated with the nebular phase ofCCSNe. Such a low Ni mass would be unusual for a normalCCSN, although an exception would be a faint electron cap-ture SN or a sub-luminous Fe CCSN from a star with a ZAMSmass of around 8 – 10 M (cid:12) . However, we find the mass of theAT 2016jbu progenitor to be significantly larger than thatexpected for an ECSN progenitor (Doherty et al. 2017). Afinal possibility that can explain such a low Ni mass (if thisis a CCSN) is significant fallback onto a compact remnant(Zampieri et al. 1998).Benetti et al. (2016) suggested that the low ejecta mass( ∼ (cid:12) ), modest explosion energy ( ∼ × erg), and low Ni ( < .
018 M (cid:12) ) of SN 1996al is consistent with significantfallback onto a 7–8 M (cid:12) black hole. This naturally explainsthe apparent absence of metal rich ejecta, as this material,found in the inner layers of the progenitor, has fallen intothe black hole. However, while SN 1996al appears similar toAT 2016jbu (see Paper I), a lack of observational data meansit is unknown whether this even had an
Event A , and wecannot be certain it is a member of the SN 2009ip-like family.Some challenges remain for the fallback scenario. A lowmetallicity environment is required, so that the progenitorstar has retained much of its ZAMS mass (e.g. Heger et al. 2003). This is hence an appealing scenario for SN 2009ip,due to its remote location ( ∼ ∼
20 M (cid:12) progenitor will loose a significant fraction of itsmass before exploding.We see from Fig. 9 that AT 2016jbu is located near amoderately star-forming region that is likely to host CCSNe,as seen from SN 1999ga. On the contrary, SN 2009ip is lo-cated on the outskirts of its host spiral galaxy, NGC 7259,at a galactocentric radius of ∼ KickedMass Gainers in a binary star system.For the progenitor to travel ∼ (cid:12) star, a binary companion may berequired to provide an additional source of fuel after the starshave been ejected (Smith et al. 2016b).However, the biggest difficulty with AT 2016jbu as a CCSNis that it is in stark contrast to the predictions of stellarevolutionary models. A 20 M (cid:12) star is expected to end itslife as a RSG which undergoes Fe core-collapse (Heger et al.2003). From our dusty modelling in Sect. 2, we find that theprogenitor of AT 2016jbu is not situated at the end of anysingle star evolutionary track. This suggests that the pro-genitor is not sufficiently evolved to undergo core-collapse.Our conclusion in Sect. 2 also suggests that the progenitor ofAT 2016jbu is not a RSG but rather a YHG. We also notethat if AT 2016jbu is indeed a CCSN, it is more appropriateto compare to the luminosity of the progenitor to the ter-minal luminosity of the models (typically corresponding tothe end of core He-burning), in which case we find that itmust have been a 12–16 M (cid:12) star. One must caution howeverthat if the progenitor of AT 2016jbu was in a binary, thenthe expectations from single star evolution can be drasticallyaltered.A tantalising hint of a surviving progenitor is AT 2016jbureturning to its pre-explosion magnitude in 2019, as shownin Fig. 2. However, this detection may be serendipitous andfurther late time monitoring will be needed to confirm anysurviving progenitor. Several authors have suggested that SN 2009ip-like transientsare a result of binary interaction (Smith et al. 2014b; Kashiet al. 2013; Soker & Kashi 2013; Smith et al. 2018a) as wellas some other SN Impostors e.g. SN 2000ch (Pastorello et al.2010; Smith 2011; Clark et al. 2013). A multiple stellar sce-nario has been suggested for η Car (e.g. Smith et al. 2018b;Hirai et al. 2020), wherein the η Car system consisted of atriple star system with interaction between the trio form-ing the eventual Homunculus Nebula around a binary sys-tem that we see today. Mass transfer within a binary systemcould naturally explain an asymmetric CSM environment,
MNRAS000
MNRAS000 , 1–24 (2021) rogenitor, environment, and modelling of AT 2016jbu which we interpret as a circumstellar/circumbinary disk forAT 2016jbu (see Okamoto et al. 2009, and references thereinon CG Carinae), although the arguments made in this papercan also be applied to a bipolar outflow.Binary merger events have recently been associated withRed Novae (RNe) and the more extreme, Luminous Red No-vae (LRNe) (see review by Pastorello et al. 2019b, and ref-erences therein). These transients typically fall into the classof Gap Transients (Kasliwal et al. 2012; Pastorello & Fraser2019) and are amongst the most powerful stellar cataclysms.LRNe span a wide range of absolute magnitudes, from − −
15 mag (Pastorello & Fraser 2019), and show a wide rangeof lightcurve shape and duration. These events typically con-sist of a double-peaked lightcurve; the first peak is relativelyblue at B-V ∼ ∼ ∼
100 km s − ), and weak Balmer lines(Pastorello et al. 2019b).The physical interpretation of LRNe is debated. The pro-genitors of LRNe are likely massive contact binaries, and thedoubled peaked lightcurve is a consequence of a stellar mergerplus a Common Envelope Ejection (CEE) (Smith et al. 2016a;Metzger & Pejcha 2017; Pastorello et al. 2019b). Pastorelloet al. (2019b) suggest that there may be a continuum span-ning between RNe to LRNe, with the possibility that thisrange can reach to brighter magnitudes (most likely causedby higher mass systems). SN 2009ip-like events may be somecombination of a binary merger where the system consists ofa relatively massive primary where the stars undergo a Com-mon Envelope (CE) phase followed by a massive eruption.AT 2016jbu and SN 2009ip-like transients show a commonpeak magnitude and shape (i.e. Event B appear to be simi-lar among SN 2009ip-like events). We do not have adequatecolour information for the peak of
Event A for AT 2016jbuhowever,
Event B has a colour of B-V ∼
0, which is compa-rable to that seen in LRNe in their first peak. AT 2016jbunever gets redder than B − V of (cid:46) ∼ B − V value of ∼ (cid:12) ), most of which is moving at lessthan 5000 km s − . They further predict the remnant of theirmergerburst will be a hot red giant star that will becomeapparent years after the transient fades, as is commonly as-sociated with RNe and LRNe (e.g Pastorello et al. 2019b).Kashi et al. (2013) discuss a similar explosion mechanism tothe scenario we discuss in Sect. 5.1 and conclude the double-peaked lightcurve of SN 2009ip may be explained by to twosuccessive outburst, separated by ∼
20 days caused by peri-astron passages of the binary system.It is appealing to conclude that AT 2016jbu is the result ofa coalescing binary. This can naturally explain the historicvariability, double-peaked lightcurve, and (inferred) asym-metric CS environment. Metzger & Pejcha (2017) proposedthat LRNe can be well modeled by a single symmetric erup- tion in an asymmetric CSM environment. This asymmetricCSM is fueled by mass transfer within the binary over manyorbits preceding the double-peaked event. The first peak ofLRNe can be comfortably powered via cooling envelope emis-sion from fast moving ejecta. Radiative shocks from the colli-sion of this ejecta with material in the equatorial plane thenpower the second peak. This would be inconsistent with ourproposed “ catch-up ” scenario for AT 2016jbu, although itcannot be ruled out conclusively.We can speculate that the events prior to
Event B inAT 2016jbu and SN 2009ip-like events are similar to LRNe,including as mass transfer / Roche Lobe Overflow (RLOF)seen in the decade leading up to
Event A , and a merger/CEEpowering
Event A itself. To explain the homogeneity of
EventB , the merging of the binary system must cause a violent (andpossibly terminal) eruption.Each SN 2009ip-like transient remains relatively blue fora long period of time, unlike what we see in LRNe, whichis likely a sign of continued interaction. If we assume thatSN 2009ip-like transients are indeed an upscaled version ofLRNe then this continued interaction at late times may re-flect a more massive progenitor than is commonly associatedwith LRNe. In this scenario would expect a surviving star tobecome visible after this interaction has abated.
SNEC
To further explore the plausibility of the progenitor matching
BPASS models from Sect. 7.1, we exploded a small subsetof these with the SuperNova Explosion Code (
SNEC , Mo-rozova et al. 2015). The full details of how
BPASS modelsare exported and exploded within
SNEC can be found inEldridge et al. (2019). The key addition to using the progen-itor model structure is to add on a CSM component aroundthe star. Here we use the values derived in Sect. 5.1 of aterminal wind velocity of 250 km s − and a mass-loss rate of0.05 M (cid:12) yr − . For each of the input stellar models we usean explosion energy of 10 . ergs, 0.016 M (cid:12) of Ni, and aninner mass cut at 1.4 M (cid:12) with nickel mixing out to 2 M (cid:12) .The resultant simulated bolometric lightcurves are shown inFig. 15.Our models are able to reproduce the magnitude of thepeak luminosity, although exact matching of the lightcurvepost-peak is difficult. Phases of the swept up wind becomingtransparent, followed by the ejecta can be seen as the suddendrop-offs in Fig. 15. We find that the width of the
Event B peak is dependent to some degree to how the density of thewind varies with distance from the star. The figure showsthe resultant models where we assume ρ wind ( r ) ∝ r − . . Wefound that the shallower the density gradient the wider thepeak and a best match is found with an exponent n = − . (cid:12) .Some models which have experienced a merger during theirbinary evolution have a higher ejecta mass do not matchthe lightcurve, being less luminous or evolving more slowly.Achieving an exact match between the models and observedlightcurve would require significant fine-tuning of the detailsof the CSM around the star, in terms of density profile, windvelocity, and the details of the wind acceleration. An exactmatch may also be impossible given the spherically symmet-ric assumptions of SNEC . However, we take the reasonably
MNRAS , 1–24 (2021) S. J. Brennan et al.
20 0 20 40 60 80 100 120Days since Event B maximum40.040.541.041.542.042.543.0 L o g ( L [ e r g / s ]) AT 2016jbu (UVOIR)SN 2009ip (Optical) M = 12.0, logP = 2.8 M = 10.0, logP = 2.6 M = 9.5, logP = 2.8 M = 17.0, logP = 3.0 M = 9.5, logP = 2.8 M = 13.0, logP = 0.6 M = 11.0, logP = 2.6 M = 9.0, logP = 2.8 Figure 15.
Diagram showing the observed lightcurve andlightcurves simulated by
SNEC from progenitors which match thepre-explosion constraints. All include the circumstellar medium asdescribed earlier. The list of progenitor information is shown in Ta-ble 2. We include our pseudo-bolometric for AT 2016jbu in blackand the optical pseudo-bolometric lightcurve for SN 2009ip (Fraseret al. 2013b; Pastorello et al. 2013).
Table 2.
The parameters of the
BPASS models exploded with
SNEC . Values of M with an “*” indicate black hole companions. M M M final M CO /M (cid:12) /M (cid:12) log( P i / days) /M (cid:12) /M (cid:12)
17 11.9 3 5.9 4.013 10.4 0.6 18.0 3.09.5 2.85 2.8 2.8 1.89.5 3.8 2.8 2.8 1.89 2.7 2.8 2.6 1.712 3.2* 2.8 3.3 2.210 3.2* 2.6 2.5 1.711 4.0* 2.6 2.8 1.9 close match between the model and observed lightcurves toindicate that a subset of the
BPASS models can explainAT 2016jbu. Intriguingly, the low CO core mass of severalof the progenitor models suggest an explosion close to theelectron-capture regime where lower nickel masses and ex-plosion energies would be expected. In this paper, we have investigated the progenitor and en-vironment of AT 2016jbu as well as modelling the transientitself. If AT 2016jbu is a single star, we find that the pro-genitor is consistent with a ∼
20 M (cid:12) progenitor (e.g. fig. 4in Smartt et al. 2009), with a color consistent with a YHG.Modelling of circumstellar dust using dusty gives a lumi-nosity and temperature of the progenitor similar to knownYHGs. We show that the local environment around the pro-genitor of AT 2016jbu is consistent with a CCSN from a pro- genitor with ZAMS mass ∼
20 M (cid:12) ; as the stellar populationhas an age log( τ life ) = 7 . +0 . − . yrs. We confidently rule outthe possibility that the progenitor of AT 2016jbu is an LBVof 50–80 M (cid:12) , as has been proposed for SN 2009ip (Smithet al. 2010; Foley et al. 2011).We find that the Event A/B light curve can be modelled bytwo shells of material, with
Shell 1 travelling at ∼
700 km s − and Shell 2 travelling at ∼ − . Event B is pow-ered by a “catch-up” scenario, where the faster material from
Shell 2 collides with material ejected at, or prior to,
EventA . The spectroscopic and photometric evolution is consistentwith a spherically symmetric ejecta colliding with, and tem-porarily engulfing, a disk-like CSM. This interaction is thedominant energy source after ∼ ∼
200 days,AT 2016jbu shows increased interaction, likely reflecting aclumpy CSM.AT 2016jbu shows only tentative evidence for core-collapse.We find a upper limit of Ni of (cid:46) .
016 M (cid:12) but with strongon-going CSM interaction at this time, the real value of Niis probably much lower (if any at all). Almost 1.5 years af-ter maximum brightness, AT 2016jbu lacks signs of explo-sively nucleosynthesised material or emission from the burn-ing products of late time stellar evolution.We explore the possibility that AT 2016jbu is the resultof a binary system. We compare our progenitor models withan extensive group of
BPASS models, exploring both CCSNand non-terminal events. We find that matching models haveM
ZAMS (cid:46)
25 M (cid:12) . Steady state mass loss due to the pro-genitor wind is unable to produce the CSM density neces-sary to power the lightcurve and episodic mass loss may berequired. Using
SNEC we find that a low explosion energy(5 . × ergs) with a small ejecta mass ( ∼ (cid:12) ) can com-fortably power AT 2016jbu (assuming spherical symmetry).AT 2016jbu show qualitative similarities to LRNe, such asa history of outbursts and a double-peaked lightcurve. How-ever AT 2016jbu and other SN 2009ip-like events are ∼ BPASS and
SNEC results show that the progenitor ofAT 2016jbu may have a final core mass in the electron captureregime. This is an interesting possibility as it may explain theuniformity seen in SN 2009ip-like transients.It appears that there is not a simple explanation for thesetransients. Following
Hickam’s dictum , a low energy SN (pos-sibly an ECSN) within a binary system with a disk-likeCSM can account for the rise and peak of
Event B , low Ni, continued CSM interaction, and unique spectral fea-tures of AT 2016jbu. Additional binary interaction might ex-plain
Event A e.g. due to a merger. Detailed modelling ofthis proposed scenario is beyond the scope of this paper andfuture work will involve exploring these scenarios in a non-symmetric setting.The true nature of AT 2016jbu (and other SN 2009ip-liketransients) remains elusive. Perhaps the ultimate answer willcome if or when very late time observations reveal a surviv-ing progenitor. To date, no conclusive evidence exists as towhether these transients destroy their progenitor. However,one must account for the possibility that if the progenitorsurvived, it may be obscured by a significant amount of dust.Deep images covering the full SED will hence be required toconfidently rule out surviving, but dust-enshrouded, star. Tothis end, future observations with the upcoming
James Webb
MNRAS000
MNRAS000 , 1–24 (2021) rogenitor, environment, and modelling of AT 2016jbu Space Telescope will be essential. Alongside this, deep opticalimaging from the Vera C. Rubin Observatory may capturesimilar pre-explosion variability in the years/decades priorto future SN 2009ip-like events, perhaps even allowing for acountdown timer before these events.
DATA AVAILABILITY
The photometric and spectroscopic data underlying this ar-ticle are available as described in Paper I. The BPASS mod-els are available at https://bpass.auckland.ac.nz/ , whileHST data are available at the Mikulski Archive for SpaceTelescopes athttps://archive.stsci.edu . ACKNOWLEDGEMENTS
S. J. Brennan acknowledges support from Science Founda-tion Ireland and the Royal Society (RS-EA/3471). M.F issupported by a Royal Society - Science Foundation IrelandUniversity Research Fellowship. T.M.B was funded by theCONICYT PFCHA / DOCTORADOBECAS CHILE/2017-72180113. T.W.C acknowledges the EU Funding under MarieSk(cid:32)lodowska-Curie grant H2020-MSCA-IF-2018-842471, andthanks to Thomas Kr¨uhler for GROND data reduction.M.N is supported by a Royal Astronomical Society Re-search Fellowship. B.J.S is supported by NSF grants AST-1908952, AST-1920392, AST-1911074, and NASA award80NSSC19K1717. M.S is supported by generous grants fromVillum FONDEN (13261,28021) and by a project grant(8021-00170B) from the Independent Research Fund Den-mark. L.H acknowledges support for Watcher from ScienceFoundation Ireland grant 07/RFP/PHYF295. Time domainresearch by D.J.S. is supported by NSF grants AST-1821987,1813466, & 1908972, and by the Heising-Simons Founda-tion under grant , a community-developed core Python package for Astronomy (AstropyCollaboration et al. 2013; Price-Whelan et al. 2018). Thisresearch made use of data provided by Astrometry.net .Parts of this research were supported by the AustralianResearch Council Centre of Excellence for All Sky Astro-physics in 3 Dimensions (ASTRO 3D), through projectnumber CE170100013. This research has made use of theNASA/IPAC Extragalactic Database (NED), which is op-erated by the Jet Propulsion Laboratory, California Institute https://astrometry.net/use.html of Technology, under contract with the National Aeronau-tics and Space Administration. We acknowledge TelescopeAccess Program (TAP) funded by the NAOC, CAS, andthe Special Fund for Astronomy from the Ministry of Fi-nance. This work was partially supported from Polish NCNgrants: Harmonia No. 2018/06/M/ST9/00311 and Daina No.2017/27/L/ST9/03221. This work made use of v2.2.1 of theBinary Population and Spectral Synthesis (BPASS) modelsas described in Eldridge et al. (2017) and Stanway & El-dridge (2018). This research is based on observations madewith the NASA/ESA Hubble Space Telescope obtained fromthe Space Telescope Science Institute, which is operated bythe Association of Universities for Research in Astronomy,Inc., under NASA contract NAS 5-26555. These observa-tions are associated with program 15645. Observations werealso obtained from the Hubble Legacy Archive, which is acollaboration between the Space Telescope Science Institute(STScI/NASA), the Space Telescope European Coordinat-ing Facility (ST-ECF/ESAC/ESA) and the Canadian As-tronomy Data Centre (CADC/NRC/CSA). This research hasmade use of the SVO Filter Profile Service supported fromthe Spanish MINECO through grant AYA2017-84089. APPENDIX A: AUTHOR AFFILIATIONS School of Physics, O’Brien Centre for Science North,University College Dublin, Belfield, Dublin 4, Ireland The Oskar Klein Centre, Department of Physics, AlbaNova,Stockholm University, SE-106 91 Stockholm, Sweden INAF-Osservatorio Astronomico di Padova, Vicolodell’Osservatorio 5, I-35122 Padova, Italy Department of Physics and Astronomy, University ofTurku, FI-20014, Turku, Finland The Department of Physics, The University of Auckland,Private Bag 92019, Auckland, New Zealand The Oskar Klein Centre, Department of Astronomy,AlbaNova, Stockholm University, SE-106 91 Stockholm,Sweden Max-Planck-Institut f¨ur Extraterrestrische Physik,Giessenbachstraße 1, 85748 Garching, Germany Department of Astronomy, The Ohio State University, 140W. 18th Avenue, Columbus, OH 43210, USA Center for Cosmology and AstroParticle Physics (CCAPP),The Ohio State University, 191 W. Woodruff Avenue, Colum-bus, OH 43210, USA Department of Physics and Astronomy, Texas A&MUniversity, 4242 TAMU, College Station, TX 77843, USA Cerro Tololo Inter-American Observatory, NSF’s NationalOptical-Infrared Astronomy Research Laboratory, Casilla603, La Serena, Chile Institut d’Astrophysique de Paris (IAP), CNRS & Sor-bonne Universite, France Kavli Institute for Astronomy and Astrophysics, PekingUniversity, Yi He Yuan Road 5, Hai Dian District, Beijing100871, China NINAF-Osservatorio Astronomico di Padova, Vicolodell’Osservatorio 5, I-35122 Padova, Italy Institute of Space Sciences (ICE, CSIC), Campus UAB, http://svo2.cab.inta-csic.es/theory/fps/ MNRAS , 1–24 (2021) S. J. Brennan et al.
Carrer de Can Magrans s/n, 08193 Barcelona, Spain Center for Astrophysics | Harvard & Smithsonian, 60Garden Street, Cambridge, MA 02138, USA Department of Physics, Florida State University, 77Chieftan Way, Tallahassee, FL 32306, USA Tuorla Observatory, Department of Physics and Astron-omy, FI-20014 University of Turku, Finland. Finnish Centre for Astronomy with ESO (FINCA),FI-20014 University of Turku, Finland CBA Kleinkaroo, Calitzdorp, South Africa Departamento de Ciencias F´ısicas, Universidad AndresBello, Avda. Republica 252, Santiago, 8320000, Chile Millennium Institute of Astrophysics, Santiago, Chile Department of Astronomy/Steward Observatory, 933North Cherry Avenue, Rm. N204, Tucson, AZ 85721-0065,USA Institute for Astronomy, University of Hawai’i, 2680Woodlawn Drive, Honolulu, HI 96822, USA Astrophysics Research Centre, School of Maths andPhysics, Queen’s University Belfast, Belfast BT7 1NN, UK Mt Stromlo Observatory, The Research School of Astron-omy and Astrophysics, Australian National University, ACT2601, Australia National Centre for the Public Awareness of Science, Aus-tralian National University, Canberra, ACT 2611, Australia The ARC Centre of Excellence for All-Sky Astrophysicsin 3 Dimension (ASTRO 3D), Australia Astronomical Observatory, University of Warsaw, Al.Ujazdowskie 4, 00-478 Warszawa, Poland Unidad Mixta Internacional Franco-Chilena de As-tronom´ıa, CNRS/INSU UMI 3386 and Instituto de As-trof´ısica, Pontificia Universidad Cat´olica de Chile, Santiago,Chile Aix Marseille Univ, CNRS, CNES, LAM, Marseille,France Institute of Astronomy, Madingley Road, Cambridge,CB30HA, UK RHEA Group for ESA, European Space AstronomyCentre (ESAC-ESA), Madrid, Spain Departamento de F´ısica Te´orica y del Cosmos, Universi-dad de Granada, E-18071 Granada, Spain Tuorla Observatory, Department of Physics and Astron-omy, FI-20014 University of Turku, Finland Kavli Institute for Cosmology, Institute of Astronomy,Madingley Road, Cambridge, CB3 0HA, UK Las Cumbres Observatory, 6740 Cortona Drive, Suite 102,Goleta, CA 93117-5575, USA Department of Physics, University of California, SantaBarbara, CA 93106-9530, USA The Observatories of the Carnegie Institution for Science,813 Santa Barbara St., Pasadena, CA 91101, USA School of Physics & Astronomy, Cardiff University,Queens Buildings, The Parade, Cardiff, CF24 3AA, UK School of Physics and Astronomy, University of Southamp-ton, Southampton, Hampshire, SO17 1BJ, UK School of Physics, Trinity College Dublin, The Universityof Dublin, Dublin 2, Ireland Department of Physics, University of the Free State, POBox 339, Bloemfontein 9300, South Africa Carnegie Observatories, Las Campanas Observatory,Colina El Pino, Casilla 601, Chile Birmingham Institute for Gravitational Wave Astronomy and School of Physics and Astronomy, University of Birm-ingham, Birmingham B15 2TT, UK Institute for Astronomy, University of Edinburgh, RoyalObservatory, Blackford Hill, EH9 3HJ, UK Nucleo de Astronomıa de la Facultad de Ingenierıa yCiencias, Universidad Diego Portales, Av. Ej ´ ercito 441,Santiago, Chile Department of Physics and Astronomy University ofNorth Carolina at Chapel Hill Chapel Hill, NC 27599, USA Department of Physics, Florida State University, 77Chieftain Way, Tallahassee, FL 32306-4350, USA Department of Physics and Astronomy, Aarhus University,Ny Munkegade, DK-8000 Aarhus C, Denmark Department of Physics, University of Warwick, CoventryCV4 7 AL, UK Astrophysics Research Centre, School of Mathematics andPhysics, Queen‘s University Belfast, Belfast BT7 1NN, UK
REFERENCES
Anderson J. P., 2019, Astron. Astrophys., 628, A7Andrews J. E., Smith N., 2018, Mon. Not. R. Astron. Soc., 477,74Andrews J. E., Smith N., McCully C., Fox O. D., Valenti S., HowellD. A., 2017, Mon. Not. R. Astron. Soc., 471, 4047Andrillat Y., Jaschek M., Jaschek C., 1986, A&AS, 65, 1Arnett D., Chevalier R., 1996, Physics Today - PHYS TODAY, 49Astropy Collaboration et al., 2013, Astron. Astrophys., 558Benetti S., 2000, Mem. Soc. Astron. Italiana, 71, 323Benetti S., et al., 2016, Mon. Not. R. Astron. Soc., 456, 3296Boian I., Groh J. H., 2018, Astron. Astrophys., 617, A115Botticella M. T., Smartt S. J., Kennicutt R. C. J., Cappellaro E.,Sereno M., Lee J. C., 2012, Memorie della Societa AstronomicaItaliana Supplementi, 19, 158Bressan A., Marigo P., Girardi L., Salasnich B., Dal Cero C.,Rubele S., Nanni A., 2012, Mon. Not. R. Astron. Soc., 427,127Chen Y., Bressan A., Girardi L., Marigo P., Kong X., Lanza A.,2015, Mon. Not. R. Astron. Soc., 452, 1068Chevalier R. A., Irwin C. M., 2011, Astrophys. J., 729, L6Clark J. S., Bartlett E. S., Coe M. J., Dorda R., Haberl F., LambJ. B., Negueruela I., Udalski A., 2013, Astron. Astrophys., 560,A10Doherty C. L., Gil-Pons P., Siess L., Lattanzio J. C., 2017, Publ.Astron. Soc. Australia, 34, e056Dolphin A. E., 2000, Publ. Astron. Soc. Pacif., 112, 1383Dom´ınguez A., et al., 2013, Astrophys. J., 763, 145Dopita M. A., Kewley L. J., Sutherland R. S., Nicholls D. C., 2016,Ap&SS, 361, 61Drilling J. S., Landolt A. U., 2000, in Allen’s Astrophysical Quan-tities. Springer-Verlag New York, p. 381Dwarkadas V. V., 2011, Mon. Not. R. Astron. Soc., 412, 1639Eldridge J. J., Stanway E. R., Xiao L., McClelland L. A. S., TaylorG., Ng M., Greis S. M. L., Bray J. C., 2017, Publ. Astron. Soc.Australia, 34, e058Eldridge J. J., Guo N. Y., Rodrigues N., Stanway E. R., Xiao L.,2019, Publ. Astron. Soc. Australia, 36, e041Elias-Rosa N., et al., 2016, Mon. Not. R. Astron. Soc., 463, 3894Foley R. J., Berger E., Fox O., Levesque E. M., Challis P. J., IvansI. I., Rhoads J. E., Soderberg A. M., 2011, Astrophys. J., 732,32Fraser M., 2020, Royal Society Open Science, 7Fraser M., et al., 2013a, Mon. Not. R. Astron. Soc., 433, 1312Fraser M., et al., 2013b, Astrophys. J., 779, L8MNRAS000
Anderson J. P., 2019, Astron. Astrophys., 628, A7Andrews J. E., Smith N., 2018, Mon. Not. R. Astron. Soc., 477,74Andrews J. E., Smith N., McCully C., Fox O. D., Valenti S., HowellD. A., 2017, Mon. Not. R. Astron. Soc., 471, 4047Andrillat Y., Jaschek M., Jaschek C., 1986, A&AS, 65, 1Arnett D., Chevalier R., 1996, Physics Today - PHYS TODAY, 49Astropy Collaboration et al., 2013, Astron. Astrophys., 558Benetti S., 2000, Mem. Soc. Astron. Italiana, 71, 323Benetti S., et al., 2016, Mon. Not. R. Astron. Soc., 456, 3296Boian I., Groh J. H., 2018, Astron. Astrophys., 617, A115Botticella M. T., Smartt S. J., Kennicutt R. C. J., Cappellaro E.,Sereno M., Lee J. C., 2012, Memorie della Societa AstronomicaItaliana Supplementi, 19, 158Bressan A., Marigo P., Girardi L., Salasnich B., Dal Cero C.,Rubele S., Nanni A., 2012, Mon. Not. R. Astron. Soc., 427,127Chen Y., Bressan A., Girardi L., Marigo P., Kong X., Lanza A.,2015, Mon. Not. R. Astron. Soc., 452, 1068Chevalier R. A., Irwin C. M., 2011, Astrophys. J., 729, L6Clark J. S., Bartlett E. S., Coe M. J., Dorda R., Haberl F., LambJ. B., Negueruela I., Udalski A., 2013, Astron. Astrophys., 560,A10Doherty C. L., Gil-Pons P., Siess L., Lattanzio J. C., 2017, Publ.Astron. Soc. Australia, 34, e056Dolphin A. E., 2000, Publ. Astron. Soc. Pacif., 112, 1383Dom´ınguez A., et al., 2013, Astrophys. J., 763, 145Dopita M. A., Kewley L. J., Sutherland R. S., Nicholls D. C., 2016,Ap&SS, 361, 61Drilling J. S., Landolt A. U., 2000, in Allen’s Astrophysical Quan-tities. Springer-Verlag New York, p. 381Dwarkadas V. V., 2011, Mon. Not. R. Astron. Soc., 412, 1639Eldridge J. J., Stanway E. R., Xiao L., McClelland L. A. S., TaylorG., Ng M., Greis S. M. L., Bray J. C., 2017, Publ. Astron. Soc.Australia, 34, e058Eldridge J. J., Guo N. Y., Rodrigues N., Stanway E. R., Xiao L.,2019, Publ. Astron. Soc. Australia, 36, e041Elias-Rosa N., et al., 2016, Mon. Not. R. Astron. Soc., 463, 3894Foley R. J., Berger E., Fox O., Levesque E. M., Challis P. J., IvansI. I., Rhoads J. E., Soderberg A. M., 2011, Astrophys. J., 732,32Fraser M., 2020, Royal Society Open Science, 7Fraser M., et al., 2013a, Mon. Not. R. Astron. Soc., 433, 1312Fraser M., et al., 2013b, Astrophys. J., 779, L8MNRAS000 , 1–24 (2021) rogenitor, environment, and modelling of AT 2016jbu Fraser M., et al., 2015, Mon. Not. R. Astron. Soc., 453, 3886Fukugita M., Ichikawa T., Gunn J. E., Doi M., Shimasaku K.,Schneider D. P., 1996, Astron. J., 111, 1748Gal-Yam A., Leonard D. C., 2009, Nature, 458, 865Gal-Yam A., et al., 2007, Astrophys. J., 656, 372Galbany L., et al., 2016, Mon. Not. R. Astron. Soc., 455, 4087Gogarten S. M., Dalcanton J. J., Murphy J. W., Williams B. F.,Gilbert K., Dolphin A., 2009, Astrophys. J., 703, 300Graham M. L., et al., 2014, Astrophys. J., 787, 163Graham M. L., et al., 2017, Mon. Not. R. Astron. Soc., 469, 1559Groh J. H., 2017, Philosophical Transactions of the Royal Societyof London Series A, 375, 20170219Gustafsson B., Edvardsson B., Eriksson K., Jørgensen U. G., Nord-lund ˚A., Plez B., 2008, Astron. Astrophys., 486, 951Heger A., Fryer C. L., Woosley S. E., Langer N., Hartmann D. H.,2003, Astrophys. J., 591, 288Hirai R., Podsiadlowski P., Owocki S. P., Schneider F. R. N., SmithN., 2020, arXiv e-prints, p. arXiv:2011.12434Humphreys R. M., Weis K., Davidson K., Gordon M. S., 2016,Astrophys. J., 825, 64Humphreys R. M., Davidson K., Hahn D., Martin J. C., Weis K.,2017, Astrophys. J., 844, 40Husser T. O., Wende-von Berg S., Dreizler S., Homeier D., ReinersA., Barman T., Hauschildt P. H., 2013, Astron. Astrophys.,553, A6Ivezic Z., Elitzur M., 1997, Mon. Not. R. Astron. Soc., 287, 799Kashi A., Soker N., Moskovitz N., 2013, Mon. Not. R. Astron. Soc.,436, 2484Kasliwal M. M., et al., 2012, Astrophys. J., 755, 161Kennicutt Robert C. J., 1998, Annu. Rev. Astron. Astrophys., 36,189Kiewe M., et al., 2012, Astrophys. J., 744, 10Kilpatrick C. D., et al., 2018, Mon. Not. R. Astron. Soc., 473, 4805Klochkova V. G., Chentsov E. L., Miroshnichenko A. S., PanchukV. E., Yushkin M. V., 2016, Mon. Not. R. Astron. Soc., 459,4183Kuncarayakti H., et al., 2020, Astrophys. J., 902, 139Kurf¨urst P., Pejcha O., Krtiˇcka J., 2020, Astron. Astrophys., 642,A214Levesque E. M., Stringfellow G. S., Ginsburg A. G., Bally J.,Keeney B. A., 2014, Astron. J., 147, 23Lim P. L., 2020, synphot, doi:10.5281/zenodo.3971036, https://doi.org/10.5281/zenodo.3971036
Maeder A., Meynet G., 2000, Annu. Rev. Astron. Astrophys., 38,143Marchant P., Renzo M., Farmer R., Pappas K. M. W., Taam R. E.,de Mink S. E., Kalogera V., 2019, Astrophys. J., 882, 36Margutti R., et al., 2014, Astrophys. J., 780, 21Marigo P., et al., 2017, Astrophys. J., 835, 77Mauerhan J. C., et al., 2013a, Mon. Not. R. Astron. Soc., 430,1801Mauerhan J. C., et al., 2013b, Mon. Not. R. Astron. Soc., 431,2599Maund J. R., 2017, Mon. Not. R. Astron. Soc., 469, 2202McCall M. L., 2004, Astron. J., 128, 2144McDowell A. T., Duffell P. C., Kasen D., 2018, Astrophys. J., 856,29Meikle W. P. S., et al., 2007, Astrophys. J., 665, 608Metzger B. D., 2010, Mon. Not. R. Astron. Soc., 409, 284Metzger B. D., Pejcha O., 2017, Mon. Not. R. Astron. Soc., 471,3200Moriya T. J., Mazzali P. A., Pian E., 2020, Mon. Not. R. Astron.Soc., 491, 1384Morozova V., Piro A. L., Renzo M., Ott C. D., Clausen D., CouchS. M., Ellis J., Roberts L. F., 2015, Astrophys. J., 814, 63Nadyozhin D. K., 2003, Mon. Not. R. Astron. Soc., 346, 97Nagao T., Maeda K., Ouchi R., 2020, Mon. Not. R. Astron. Soc.,497, 5395 Nicholl M., 2018, Research Notes of the American AstronomicalSociety, 2, 230Nyholm A., et al., 2017, Astron. Astrophys., 605, A6Nyholm A., et al., 2020, Astron. Astrophys., 637, A73Ofek E. O., et al., 2014, Astrophys. J., 781, 42Okamoto Y. K., et al., 2009, Astrophys. J., 706, 665Pancoast A., Sajina A., Lacy M., Noriega-Crespo A., Rho J., 2010,Astrophys. J., 723, 530Pastorello A., Fraser M., 2019, Nature Astronomy, 3, 676Pastorello A., et al., 2008, Mon. Not. R. Astron. Soc., 389, 113Pastorello A., et al., 2009, Astron. Astrophys., 500, 1013Pastorello A., et al., 2010, Mon. Not. R. Astron. Soc., 408, 181Pastorello A., et al., 2013, Astrophys. J., 767, 1Pastorello A., et al., 2018, Mon. Not. R. Astron. Soc., 474, 197Pastorello A., et al., 2019a, Astron. Astrophys., 628, A93Pastorello A., et al., 2019b, Astron. Astrophys., 630, A75Price-Whelan A. M., et al., 2018, Astron. J., 156Prieto J. L., et al., 2014, Astrophys. J., 787, L8Reguitti A., et al., 2019, Mon. Not. R. Astron. Soc., 482, 2750Reilly E., Maund J. R., Baade D., Wheeler J. C., H¨oflich P., Spy-romilio J., Patat F., Wang L., 2017, Mon. Not. R. Astron. Soc.,470, 1491Ryder S. D., et al., 2001, Astrophys. J., 555, 232Schlafly E. F., Finkbeiner D. P., 2011, Astrophys. J., 737, 103Shapovalova A. I., et al., 2004, Astron. Astrophys., 422, 925Smartt S. J., Eldridge J. J., Crockett R. M., Maund J. R., 2009,Mon. Not. R. Astron. Soc., 395, 1409Smith N., 2009, arXiv e-prints, p. arXiv:0906.2204Smith N., 2011, Mon. Not. R. Astron. Soc., 415, 2020Smith N., 2017, Philosophical Transactions of the Royal Society ofLondon Series A, 375, 20160268Smith N., Mauerhan J., 2012, The Astronomer’s Telegram, 4412,1Smith N., Tombleson R., 2015, Mon. Not. R. Astron. Soc., 447,598Smith N., Foley R. J., Filippenko A. V., 2008, Astrophys. J., 680,568Smith N., et al., 2010, Astron. J., 139, 1451Smith N., Mauerhan J. C., Prieto J. L., 2014a, Mon. Not. R. As-tron. Soc., 438, 1191Smith N., Mauerhan J. C., Prieto J. L., 2014b, Mon. Not. R. As-tron. Soc., 438, 1191Smith N., et al., 2016a, Mon. Not. R. Astron. Soc., 458, 950Smith N., Andrews J. E., Mauerhan J. C., 2016b, Mon. Not. R.Astron. Soc., 463, 2904Smith N., et al., 2018a, Mon. Not. R. Astron. Soc., 480, 1466Smith N., et al., 2018b, Mon. Not. R. Astron. Soc., 480, 1466Soker N., Kashi A., 2013, Astrophys. J., 764, L6Sollerman J., et al., 2019, Astron. Astrophys., 621, A30Stanway E. R., Eldridge J. J., 2018, Mon. Not. R. Astron. Soc.,479, 75Stevance H. F., 2019, arXiv e-prints, p. arXiv:1906.07184Stevance H. F., et al., 2019, Mon. Not. R. Astron. Soc., 485, 102Stevance H., Eldridge J., Stanway E., 2020a, The Journal of OpenSource Software, 5, 1987Stevance H. F., Eldridge J. J., McLeod A., Stanway E. R., ChrimesA. A., 2020b, Mon. Not. R. Astron. Soc., 498, 1347Stritzinger M., et al., 2012, Astrophys. J., 756, 173Suzuki A., Moriya T. J., Takiwaki T., 2019, Astrophys. J., 887,249Taddia F., et al., 2015, Astron. Astrophys., 580, A131Taddia F., et al., 2020, Astron. Astrophys., 638, A92Tartaglia L., et al., 2016a, Mon. Not. R. Astron. Soc., 459, 1039Tartaglia L., et al., 2016b, Astrophys. J., 823, L23Th¨one C. C., et al., 2017, Astron. Astrophys., 599, A129Tonry J. L., et al., 2018, Publ. Astron. Soc. Pacif., 130, 064505Vale Asari N., et al., 2020, Monthly Notices of the Royal Astro-nomical Society, 498, 4205–4221 MNRAS , 1–24 (2021) S. J. Brennan et al.
Vlasis A., Dessart L., Audit E., 2016, Mon. Not. R. Astron. Soc.,458, 1253Whitmore B. C., et al., 2016, Astron. J., 151, 134Williams B. F., Hillis T. J., Murphy J. W., Gilbert K., DalcantonJ. J., Dolphin A. E., 2018, Astrophys. J., 860, 39Wong O. I., et al., 2006, Mon. Not. R. Astron. Soc., 371, 1855Woosley S. E., Blinnikov S., Heger A., 2007, Nature, 450, 390Zampieri L., Shapiro S. L., Colpi M., 1998, Astrophys. J., 502,L149de Jager C., 1998, A&ARv, 8, 145van Dyk S. D., Weiler K. W., Sramek R. A., Panagia N., 1993,Astrophys. J., 419, L69This paper has been typeset from a TEX/L A TEX file prepared bythe author.MNRAS000