A Bright Ultraviolet Excess in the Transitional 02es-like Type Ia Supernova 2019yvq
J. Burke, D. A. Howell, S. K. Sarbadhicary, D. J. Sand, R. C. Amaro, D. Hiramatsu, C. McCully, C. Pellegrino, J. E. Andrews, P. J. Brown, Koichi Itagaki, M. Shahbandeh, K. A. Bostroem, L. Chomiuk, E. Y. Hsiao, Nathan Smith, S. Valenti
DDraft version January 19, 2021
Typeset using L A TEX twocolumn style in AASTeX62
A Bright Ultraviolet Excess in the Transitional 02es-like Type Ia Supernova 2019yvq
J. Burke,
1, 2
D. A. Howell,
1, 2
S. K. Sarbadhicary, D. J. Sand, R. C. Amaro, D. Hiramatsu,
1, 2
C. McCully,
1, 2
C. Pellegrino,
1, 2
J. E. Andrews, P. J. Brown,
5, 6
Koichi Itagaki ( 板 垣 公 一 ), M. Shahbandeh, K. A. Bostroem, L. Chomiuk, E. Y. Hsiao, Nathan Smith, and S. Valenti Department of Physics, University of California, Santa Barbara, CA 93106-9530, USA Las Cumbres Observatory, 6740 Cortona Dr, Suite 102, Goleta, CA 93117-5575, USA Center for Data Intensive and Time Domain Astronomy, Department of Physics and Astronomy, Michigan State University, EastLansing, MI 48824 Steward Observatory, University of Arizona, 933 North Cherry Avenue, Tucson, AZ 85721-0065, USA Department of Physics and Astronomy, Texas A&M University, 4242 TAMU, College Station, TX 77843, USA George P. and Cynthia Woods Mitchell Institute for Fundamental Physics & Astronomy Itagaki Astronomical Observatory, Yamagata 990-2492, Japan Department of Physics, Florida State University, Tallahassee, FL 32306, USA Department of Physics and Astronomy, University of California, 1 Shields Avenue, Davis, CA 95616-5270, USA (Received Soon; Revised After that; Accepted After that)
Submitted to ApJABSTRACTWe present photometric and spectroscopic observations of the nearby Type Ia SN 2019yvq, from itsdiscovery ∼ ∼
100 days after its peak brightness. This SN exhibits severalunusual features, most notably an extremely bright UV excess seen within ∼ Swift
UV data, this early excess outshines its “peak” brightness, making this object moreextreme than other SNe with early UV/blue excesses (e.g. iPTF14atg and SN 2017cbv). In addition,it was underluminous ( M B = − . m ( B ) = 1 . . × − M (cid:12) /yr on the mass-loss rate from a symbiotic progenitor, which does not exclude a red giant or mainsequence companion. Ultimately we find that no one model can accurately replicate all aspects of thedataset, and further we find that the ubiquity of early excesses in 02es-like SNe Ia requires a progenitorsystem that is capable of producing isotropic UV flux, ruling out some models for this class of objects. Keywords: supernovae: individual (SN 2019yvq) – supernovae: general INTRODUCTIONDespite the fact that Type Ia supernovae (SNe) wereused as standardizable candles to discover the acceler-ating expansion of the universe and constrain its energycontent (Riess et al. 1998; Perlmutter et al. 1999), open
Corresponding author: J. [email protected] questions remain about their progenitor systems. TheSNe themselves are understood to be the thermonuclearexplosions of carbon/oxygen white dwarfs (WDs) (Hoyle& Fowler 1960), but beyond that there are large uncer-tainties about both the progenitor system(s) and explo-sion mechanism(s).Many possible progenitor systems have been theo-rized. The two broad classes are the single-degeneratechannel (Whelan & Iben 1973), where the WD accretesmatter slowly from a nondegenerate companion, and a r X i v : . [ a s t r o - ph . H E ] J a n Burke et al. the double-degenerate channel (Iben & Tutukov 1984),where the source of the extra matter needed to ignitethe WD is a second WD. Within these two broad chan-nels exist many specific and sometimes exotic scenar-ios, e.g. dynamically driven double-degenerate double-detonation systems (Shen et al. 2018) or rotating super-Chandrasekhar mass WD progenitors (Yoon & Langer2005). For reviews, see Howell (2011), Wang & Han(2012), and Maoz et al. (2014).Kasen (2010) predicted an observational signaturethat could distinguish between the single- and double-degenerate cases. If the donor star were nondegener-ate then the SN ejecta will run into it and get shock-heated. The shock-heated ejecta would then emit anexcess of UV/blue light which could be detected in theSN’s early-time lightcurve. The strength of this signa-ture is dependent on the companion’s size and separa-tion, the velocity of the ejecta, and the viewing angleof the event. Kasen (2010) predicted that the viewingangle effect alone would make this early blue excess vis-ible in only 10% of SNe Ia which explode through thissingle-degenerate channel.Following the publication of Kasen (2010), manyrolling supernova searches were examined for evidenceof the effect in the optical and UV (Hayden et al. 2010;Bianco et al. 2011; Ganeshalingam et al. 2011; Tucker2011). These found no evidence for the predicted shockwith a red giant companion. Brown et al. (2012a) alsoexcluded red giant companions from a smaller sampleof SNe Ia with constraining UV data. The early op-tical observations of SN 2011fe were additionally ableto place extremely tight constraints on optical and UVshock emission from the companion (Nugent et al. 2011;Brown et al. 2012b).Early blue excesses have since been seen in a smallnumber of SNe, most notably SN 2012cg (Marion et al.2016), iPTF14atg (Cao et al. 2015), iPTF16abc (Milleret al. 2018), and SN 2017cbv (Hosseinzadeh et al. 2017).The proliferation of transient surveys has allowed formuch more consistent and thorough followup of youngSNe (e.g. Yao et al. 2019). This in turn has revealeda wide range of early behaviors including varying earlycolor evolution (Bulla et al. 2020; Stritzinger et al. 2018;Brown et al. 2017, 2018) and a range of (sometimes bro-ken) power laws which describe their rising lightcurves(Olling et al. 2015; Miller et al. 2018, 2020a; Li et al.2019; Shappee et al. 2019; Dimitriadis et al. 2019).A number of progenitor scenarios can reproduce somerange of these observed properties, including explosionswhich vary the degree of nickel mixing in the explod-ing WD (Piro & Morozova 2016) leading to a rangeof early colors, and models of sub-Chandrasekhar mass WDs detonated by the ignition of a surface layer of He(Polin et al. 2019a) leading to a wide range of absolutemagnitudes and colors.In this paper we present early-time photometry andspectroscopy of the Type Ia SN 2019yvq, a SN discov-ered in late 2019 which displays a rare, and unusuallystrong, blue bump at early times. The object displaysother unusual behavior, including extremely broad andhigh-velocity Si II at peak and strong nebular [Fe II] and[Ca II]. Its unique combination of characteristics makeit an excellent stress-test for several models of SNe Ia.Multiple papers have already been written about thisobject (Miller et al. 2020b; Siebert et al. 2020; Tuckeret al. 2020), which we reference throughout, as this workagrees with prior findings in some respects and disagreesin others.In Section 2 we describe the object’s discovery andthe observational followup by Las Cumbres Observa-tory, which obtained data presented here for the firsttime, and the
Swift space telescope. In Section 3 wediscuss interesting features of the dataset, and we com-pare specifically to 02es-like SNe Ia in Section 4. InSection 5 we compare our data to models from Kasen(2010) and Polin et al. (2019a) and discuss the difficultyof finding a single model that reproduces all featuresof our dataset. In Section 6 we discuss constraints onthe progenitor system as indicated by radio observationsfrom the Karl G. Jansky Very Large Array. We discussimplications of the event and its properties in Section 7.We conclude in Section 8. DISCOVERY & OBSERVATIONS2.1.
Discovery
SN 2019yvq was discovered by Koichi Itagaki (Itagaki2019) on 2019 December 28.74 UT using a Celestron 14inch telescope at an unfiltered magnitude of 16.7. Anondetection of the same field, using an identical setup,was found the night before (2019 December 27.72 UT),with a limiting unfiltered magnitude of ∼ he Transitional 02es-like SN 2019yvq -20-10 0 10 20 30 40 50 60 70 80 90 100Days from B peak81012141618202224 A pp a r e n t m ag n i t ud e UVW2-8UVM2-7UVW1-5U-2B+0g+1.5Unfiltered+3V+4r+5i+6Spectralepochs -24-22-20-18-16-14-12-10 A b s o l u t e m ag n i t ud e LCOSwift ZTFItagaki
Figure 1.
UV and optical extinction-corrected photometryof SN 2019yvq. As discussed in Section 3.1 we adopt E ( B − V ) host = 0 .
052 throughout our analysis, in addition to E ( B − V ) Milky Way = 0 . SED Machine on the Palomar 60-in telescope taken on2020 January 12.36 further confirmed that SN 2019yvqis a SN Ia. We have downloaded these spectra from the Transient Name Server (TNS) and incorporated theminto our analysis.SN 2019yvq is located at right ascension 12 h m s . ◦ (cid:48) (cid:48)(cid:48) . z =0.00908 (Rothberg & Joseph 2006,retrieved via NED ). NGC 4441 is an SAB0-type galaxy,and is clearly undergoing a merger event as can be seenin deep images from the DESI Legacy Imaging Sur-vey (Dey et al. 2019). A surface brightness fluctua-tion (SBF) distance to NGC 4441 suggests D ≈
20 Mpc(Tonry et al. 2001), although the disturbed nature ofthe host likely affects this measurement. The Hubble-flow distance is D ≈
40 Mpc, which is in agreement withthe distance modulus calculated in Miller et al. (2020b).Both to be consistent with Siebert et al. (2020) andTucker et al. (2020), and because using the SBF distancevalue would further decrease the object’s already lowluminosity, we adopt the distance modulus from Milleret al. (2020b) throughout this work ( µ = 33 . ± . D = 42 . ± . E ( B − V )=0.017 mag using the Schlafly& Finkbeiner (2011) calibration of the Schlegel et al.(1998) dust maps. 2.2. Photometry
Figure 1 displays our full photometric dataset.An intense
UBVgri follow-up campaign was under-taken using the 1-m telescopes of Las Cumbres Obser-vatory (LCO; Brown et al. 2013). Data were reducedusing lcogtsnpipe (Valenti et al. 2016) by perform-ing PSF-fitting photometry. Zeropoints for images inthe
UBV filters were calculated from Landolt standardfields (Landolt 1992) taken on the same night by thesame telescope. Likewise, zeropoints for images in the gri filter set were calculated by using Sloan magnitudesof stars in the same field as the object (SDSS Collabo-ration et al. 2017).Observations from the
Neil Gehrels Swift Observatory ( Swift ; Gehrels et al. 2004) and the Ultra-Violet Op-tical Telescope (UVOT; Roming et al. 2005) were ob-tained under GI Program 1518168 and reduced usingthe pipeline associated with the
Swift
Optical Ultravi-olet Supernovae Archive (SOUSA; Brown et al. 2014)and the zeropoints of Breeveld et al. (2010). The tem-poral sensitivity changes were corrected for using the https://wis-tns.weizmann.ac.il/ http://ned.ipac.caltech.edu/ http://legacysurvey.org/viewer Burke et al. N o r m a li ze d F λ + o ff s e t -14.15-12.251.774.657.7717.62-12.94-10.71-2.4214.49 S i II ⊕ O I C II C a II FLOYDSHOWPol SPRATSEDMBCSpec l og ( F λ ) + o ff s e t ⊕ ⊕ C I N o r m a li ze d F λ + o ff s e t ⊕ [ C a II ] BCSpecBCSpecMMTBlueChannelBinospec
Figure 2.
The top left and right-hand panels indicate the optical spectral evolution of SN 2019yvq, separated into panelspurely for readability. The bottom left panel shows the IR spectrum at ∼ B -band maximum are included as labels on each spectrum. The wavelengths of spectral featuresare marked with dashed lines, corresponding to their approximate velocity which they have at maximum light to guide the eyein tracking their velocity evolution. Telluric features are marked with ⊕ . The primary source for spectra was the FLOYDSinstrument at Las Cumbres (black spectra), but a number of other spectra (detailed in Sections 2.1 and 2.3) are included aswell. The final three spectra have been binned by a factor of 5, for clarity. . Template observations from 2012were used to subtract the host galaxy count rates fromthe UVW2 , UVM2 , and
UVW1 filters.In addition to the Las Cumbres and
Swift photometricdata, we have also obtained unfiltered photometry takenwith the Itagaki Astronomical Observatory’s Celestron14-inch telescope in the days after discovery, includingthe nondetection taken the day prior to SN 2019yvq’sdiscovery. https://heasarc.gsfc.nasa.gov/docs/heasarc/caldb/swift/docs/uvot/uvotcaldb throughput 06.pdf We gather g and r band data from the public ZTFdata stream using the MARS transient broker , andpresent the near-peak data in Figure 1 as comparison.2.3. Spectroscopy
Figure 2 displays our full spectroscopic dataset.A sequence of optical spectra were taken primarilywith the FLOYDS spectrograph mounted on Las Cum-bres Observatory’s 2-m telescope on Haleakala, HI, andwere reduced as described in Valenti et al. (2014). https://mars.lco.global/ he Transitional 02es-like SN 2019yvq Method E ( B − V ) σ E ( B − V ) M B Na ID 0.052 +0 . − . -18.41Lira Law 0.268 0.043 -19.29 SNooPy . ± .
060 (sys) -19.60
SNooPy (no i ) 0.445 0 . ± .
060 (sys) -20.02SALT2 0.347 0.015 -19.62SALT2 (no i ) 0.631 0.019 -20.78MLCS2k2 0.252 0.0036 -19.23MLCS2k2 (no i ) 0.279 0.0038 -19.34 Table 1.
Range of extinction values and peak absolute mag-nitudes computed using different methods and SN Ia fittingprograms. SALT2 and MLCS2k2 fits were done using the sncosmo package and Lira Law fits were done with a fixedslope, as discussed in the text. We adopt the Na ID extinc-tion value throughout our analysis.
Additional optical spectroscopy was obtained with the2.3-m Bok telescope and the B&C spectrograph us-ing both the 300 line/mm grating and a higher resolu-tion 1200/mm line grating. We also obtained an MMTmedium resolution (1200 l/mm) spectrum on 2020-02-18 11:27 UTC using the Blue Channel spectrograph(Schmidt et al. 1989). These data were reduced usingstandard IRAF tasks. We use the Na ID doublet in thehigh resolution data as one method of estimating hostgalaxy extinction from cold gas as discussed in Section3.2.3.Finally, a near-infrared spectrum of SN 2019yvq wastaken on 2020 Jan 20 (UT) with SpeX (Rayner et al.2003) on the NASA Infrared Telescope Facility in cross-dispersed ‘SXD’ mode, providing wavelength coveragefrom ∼ µ m; these data were reduced in a stan-dard way, as described in Hsiao et al. (2019).All new data are made publicly available on the Weiz-mann Interactive Supernova Repository (Yaron & Gal-Yam 2012). DATA ANALYSIS3.1.
Lightcurve and Color Evolution Analysis
The lightcurve of SN 2019yvq is presented in Figure1. The most striking feature of this lightcurve is thestrong wavelength-dependent excess of the first epoch,seen in data from Las Cumbres, ZTF, and
Swift . Wenote especially the excess in the mid-UV
Swift filters,where the magnitude during the initial bump is brighterthan the “peak” magnitude. This is even more extremethan other objects with an observed mid-UV excess atearly times such as SN 2012cg (Marion et al. 2016) and https://wiserep.weizmann.ac.il/ t V B − V ( M il ky W a y c o rr ec t e d ) SN 2019yvqLira Law:Slope=-0.0118Best-fit:Slope=-0.0074 ± E ( B − V ) = 0 . ± . Figure 3.
Comparisons of the B − V color evolution ofSN 2019yvq (black) to the Lira Law (pink). The best-fit line(dashed) to the appropriate SN 2019yvq data has a slope2 . σ away from the expected slope. Fixing the slope (solidline) is one method of measuring the host extinction, re-ported in Table 1. Following the convention of Phillips et al.(1999), data are plotted relative to t V (days from V -bandmaximum). iPTF14atg (Cao et al. 2015). We also note that SN2017cbv (Hosseinzadeh et al. 2017), the SN Ia with themost clearly resolved early optical blue bump, displayedonly a moderate excess in the UVW1 , UVM2 , or
UVW2 bands compared to what is expected from companionshock interaction models (as shown in Figure 3 of thatpaper), although its UV colors are still quite blue com-pared to other normal SNe Ia (Brown et al. 2017).Different methods of estimating the extinction due tothe host galaxy of SN 2019yvq yielded significantly dif-ferent results, as summarized in Table 1. For all fits wefixed R V, host = 3 . B − V color evolution of many SNe Ia issimilar between 30 and 90 days after V maximum, andcan be fit with a line described by Equation 1 of thatpaper. That expected linear color evolution is shownin pink in Figure 3. E ( B − V ) can then be measured Burke et al. UV W - U UV M - U UV W - U -10 U - B B - V g - r B - i r - i SN 2017cbvSN 2011fe iPTF14atgSN 2019yvq GroundSwift Companion shockingDouble detonation
Figure 4.
Color evolution of SN 2019yvq compared with other SNe Ia. We assume an explosion epoch of SN 2019yvq derivedfrom the best-fit companion shocking model, and the two sets of model colors plotted are the best-fit models described in Section5. We note again the extremely strong early blue color in every filter combination besides r − i . by fitting a line with the same slope to the color data,and finding the linear offset needed to deredden the fitto the expected Lira Law values. Using this methodwe measure E ( B − V ) = 0 . ± .
043 for SN 2019yvq.However, the B − V color evolution of SN 2019yvq hasa best-fit slope 2 . σ away from the slope predicted bythe Lira Law. The shallower slope of SN 2019yvq is notunprecedented (see e.g. F¨orster et al. 2013), but does cast doubt on the E ( B − V ) value obtained from theLira Law comparison.We also attempted to fit the BVgri data from LasCumbres using the
SNooPy software package (Burnset al. 2011). We obtained the extinction value by com-paring to
EBV model , which required a high extinctionvalue (0.342) to match the data. similar to the findingsin Miller et al. (2020b). The fits start at a phase of -10 he Transitional 02es-like SN 2019yvq i maximum, sowe also performed fits which excluded those data.In contrast to normal SNe Ia, SN 2019yvq lacks astrong secondary NIR peak, although Tucker et al.(2020) do find evidence of a weak secondary NIR max-imum in both the ZTF i -band data and the TESSlightcurve. We take this very weak secondary NIR peakas one of several pieces of evidence that the object isintrinsically underluminous compared to normal SNe Ia(see Section 4).We repeated this process on the U BV gri
Las Cum-bres data using the SALT2 (Guy et al. 2007) andMLCS2k2 (Jha et al. 2007) fitting packages, accessedthrough SNCosmo (Barbary et al. 2016) with an added
CCM89Dust component to measure E ( B − V ). We ex-clude the first three epochs of data, to reduce biases fromattempting to fit the early blue excess. The fits weregenerally poor: in order to achieve a χ of less than2 on the best fits (MLCS2k2, no i band), we requireda systematic error of more than three times the aver-age flux error to be added in quadrature at each point.In general the fits again overpredicted the secondary i -band peak. Values for the SNooPy and SNCosmo fits arereported in Table 1.The fact that different methods of estimating E ( B − V ) led to such a wide range of extinction values, andthe fact that methods which relied on fitting to SN Iatemplates resulted in generally poor fits, led us to con-clude that SN 2019yvq is an inherently peculiar SN Ia.We therefore adopt the extinction value obtained fromfitting the Na ID lines, E ( B − V ) = 0 . +0 . − . (seeSection 3.2.3 for methodology). This value, while sig-nificantly lower than other possible values, results inan underluminous peak absolute magnitude, which isconsistent with SN 2019yvq’s weak secondary IR max-imum and high lightcurve decline rate. Additionally, itis consistent with the value calculated in Miller et al.(2020b) ( E ( B − V ) host ≈ . B data to obtain standard lightcurve parameters. We findthat SN 2019yvq reached its peak apparent magnitudeof B max = 15 . ± .
03 ( M B = − . ± .
1) on MJD58862 . ± .
4, with ∆ m ( B ) = 1 . ± .
10. We notethat this ∆ m is lower than the value inferred by Miller et al. (2020b) from the g lightcurve and used in Siebertet al. (2020).The color evolution of SN 2019yvq is presented in Fig-ure 4. The Swift data for all objects were extinction-corrected using the method of Brown et al. (2010) (Ta-ble 1). We note that SN 2019yvq becomes rapidly red-der in all optical colors (besides r − i ) over the first fivedays. In ( B − V ) and ( g − r ) especially, it is much redderthan typical SNe Ia such as SN 2011fe (data from Zhanget al. 2016) and more closely mirrors the evolution ofiPTF14atg. iPTF14atg was also an underluminous SNIa with a strong early UV excess (Cao et al. 2015), andbelonged to the 02es-like subclass, whose namesake isdescribed in Ganeshalingam et al. (2012). As discussedin Section 4, we classify SN 2019yvq as a transitional02es-like.In terms of Swift
UV colors, SN 2019yvq stands outeven more compared to typical SNe Ia, and is (cid:38)
U V W − U ) at ∼ M B versus ∆ m ( B ) relation ofPhillips (1993), populated with a large sample of nearbySNe Ia (see Figure 14 from Parrent et al. 2014, with orig-inal data from Blondin et al. 2012; Folatelli et al. 2012;Pakmor et al. 2013). When we include the “blue” and“red” sample of early SN Ia of Stritzinger et al. (2018)(hereafter S18), we see the tendency of early blue objectsto be slower declining and slightly brighter than the redsample. SN 2019yvq notably stands out from the “early-blue” sample with its much higher decline-rate. In thisparameter space it is closer to another transitional 02es-like, SN 2006bt (the orange star in Figure 5), althoughstill well-separated from that object.3.2. Spectral Analysis
We show the spectral evolution of SN 2019yvq in Fig-ure 2, from roughly −
14 to +117 days with respect to B -band maximum. Using the Supernova IDentificationsoftware package ( SNID ; Blondin & Tonry 2007) on theFLOYDS spectrum taken at +1.8 d with respect to B -band maximum we find that all reasonable matches cor-respond to normal SN Ia. In particular, the spectrum iswell matched to SN 2002bo near maximum light exceptin the region of ∼ β , H α and [N II] emission; upon investigation, Burke et al. ∆ m ( B ) -20.5-20.0-19.5-19.0-18.5-18.0 M B ( p e a k ) SN 2019yvqSN 2006btEarly red (S18)Early blue (S18)
50 100 150 200pEW Si II 6355 (Å)0102030405060 p E W S i II ( Å ) BLSSCNCL
10 12 14 16Si II Velocity (1000 km/s)-20.0-19.5-19.0-18.5-18.0 M B ( p e a k ) Polin et al. (2019) modelsZheng et al. (2018) sample
Figure 5.
Demographic properties of SN 2019yvq (black star in each plot). We note that SN 2019yvq is at the edge of normalparameter space in several respects, and is well-separated from the early blue objects of S18. It is instead closer to (althoughstill substantially different from) the transitional 02es-like SN 2006bt (orange star in each plot).
Left : Luminosity decline raterelation for SNe Ia, with the gray background points coming from the union of samples presented by several groups (Blondinet al. 2012; Folatelli et al. 2012). The orange polygon and data points replicate the sample of 02es-like SNe Ia in Taubenberger(2017), with the transitional SN 2006bt represented by the orange star in each plot. In blue and red we show the early SN Iasample presented by S18, split by their early light curve colors. Out of the S18 sample, we have adjusted the absolute magnitudeof SN 2017cbv to match the distance of D = 12 . Center : The location of SN 2019yvq (blackstar) in the Branch diagram (Branch et al. 2006), which groups SNe Ia as broad line (BL), shallow silicon (SS), core normal(CN), or cool (CL) based on the pseudo-equivalent widths of two Si II features. The background sample is the same as the leftpanel, and the only other 02es-like (in orange) in Blondin et al. (2012) is SN 2002es itself.
Right : A replica of the plot fromPolin et al. (2019a) comparing 0 .
01 M (cid:12)
He shell double detonation models to a sample of SNe Ia from Zheng et al. (2018), withvelocities measured at peak. The prototype object SN 2002es has a Si II velocity which is too low (5890 km s − ) to fit in theaxis range of these plots. we believe this emission is from the host galaxy due toslight mis-centering of the SN within the slit.3.2.1. Velocities and Spectral Classification
We measure a Si II λ − near maximum light, as well as pseudo-equivalent width(peW) values of 169 ˚A and 20 ˚A for the Si II λ λ λ (cid:38) − near max). To putSN 2019yvq in the context of the standard Branch clas-sification scheme (Branch et al. 2006), we plot it alongwith a larger sample of SNe Ia (Blondin et al. 2012) inthe center panel of Figure 5. Here SN 2019yvq is clearlya Broad Lined (BL) SN Ia, with a very deep and broadSi II λ λ B -band magnitude in the rightpanel of Figure 5. This plot is largely a reproduction ofFigure 11 in Polin et al. (2019a), with the grey datapoints originating from the SNe Ia sample of Zhenget al. (2018); the blue and red sample of S18 and SN2006bt are plotted as well. As discussed by Polin et al.(2019a), two groups of SNe Ia are apparent in the plot:one that is tightly clumped at v ≈ ,
500 km s − and M B ≈ − . M (cid:12) He shell double detonation models. Itis clear that SN 2019yvq is not well-matched by eitherpopulation, and a model with different He shell mass isneeded to replicate its position, as is found in Section5.2. 3.2.2.
Search for Unburned Carbon
The presence of unburned carbon in SN Ia spectra ispotentially a powerful discriminant between explosionmodels. Chandrasekhar-mass delayed detonation explo-sions predict complete carbon burning for normal-bright he Transitional 02es-like SN 2019yvq λ λ ∼ λ λ B -band maximum, as it has been sug-gested that the C I λ µ m line is a good tracer ofunburned carbon. No C I line is apparent, but this spec-trum is later than ideal since this feature is most visiblearound maximum light (e.g. Hsiao et al. 2013, 2019).Detailed modeling is necessary to completely rule outany subtle carbon feature, but this is beyond the scopeof the current work.In conclusion, we can make no definitive claim aboutthe presence of either C II λ λ µ m,partially due to low signal to noise data, although wecan rule out the strong carbon seen in previous SNeIa with blue light curve excesses. This lack of strongcarbon is in broad agreement with expectations fromsub-Chandrasekhar helium shell detonation models (e.g.Polin et al. 2019a), which we explore further in ourmodel comparisons below.3.2.3. Medium Resolution Spectra and Na ID
The Na ID doublet is often used to estimate hostgalaxy extinction in nearby SNe (e.g. Poznanski et al.2012), although the correlation between host extinctionand Na ID equivalent width has a large scatter (e.g.Galbany et al. 2019). Although the diffuse interstellar N o r m a li ze d F λ -19.5-19.0-18.5-18.0-17.5 M B Figure 6.
Nebular spectra of SNe Ia focusing on the [CaII], [Fe II], [Ni II] line complex. This feature is strongestin the nebular spectra of underluminous SNe Ia, and is thesubject of thorough modeling in Siebert et al. (2020) for a+153d Keck spectrum of SN 2019yvq. The legend displaysthe shortened SN name (e.g. SN2019yvq → B maximum. Spectra have been normal-ized to have identical mean fluxes over their full wavelengthrange ( ∼ band at 5780˚A has been shown to be a superior tracerof host extinction (Phillips et al. 2013), we do not de-tect the line in our medium resolution Bok spectrum.The Na ID doublet at the redshift of SN2019yvq’s host( z =0.00908) is clearly visible in our medium resolutionBok B&C spectrum ( R ≈ E ( B − V ) host, Na ID = 0 . +0 . − . mag. Asdiscussed in Section 3.1, this is the host extinction valuewe use throughout the paper.3.2.4. Nebular spectra of SN 2019yvq
The nebular spectra of SNe Ia can provide an indepen-dent way to differentiate between progenitor systems,0
Burke et al. since different progenitors and explosion channels shouldhave different nebular signatures.The violent merger of two WDs should result in neb-ular [O I] due to its ejection at low velocities (Pakmoret al. 2012), although this has only been seen in the neb-ular spectra of the 02es-like SN 2010lp (Taubenbergeret al. 2013) and is not present in the nebular spectra ofSN 2019yvq.The double-detonation scenario should only partiallyburn the core, leaving strong Ca signatures (Polin et al.2019b). SN 2019yvq does display nebular [Ca II] whichis intermediate in strength between typical- and low-luminosity SNe Ia, as shown in Figure 6.Lastly, the companion interaction scenario should pro-duce H and He emission from the swept-up material(Boty´anszki et al. 2018; Dessart et al. 2020), althoughthis is seen in an extremely limited number of cases(Kollmeier et al. 2019; Prieto et al. 2020). We use thenebular spectra of SN 2019yvq to measure limits on theluminosity and mass of swept-up H and He, followingthe methodology of Sand et al. (2019) and referencestherein. To briefly summarize, we first smooth the spec-trum on a scale much larger than the expected widthof an H α feature. We then subtract off the smoothedspectrum and search for any excess flux in the resid-uals, assuming an expected width of FWHM ≈ − (22 ˚A) for the line width and a potential off-set from the rest wavelength of up to ∼ − as well. Following Equation 1 from Boty´anszki et al.(2018), we then estimate the mass of the stripped ma-terial, after predicting the luminosity of SN 2019yvqat +200 days. For the nebular spectrum taken +106days past maximum, M H < . × − M (cid:12) and M He < . × − M (cid:12) (using the He I λ M H < . × − M (cid:12) and M He < . × − M (cid:12) .With access to a higher signal-to-noise spectrum, Siebertet al. (2020) place even stricter limits on the amountof swept-up He and He: M H < . × − M (cid:12) and M He < . × − M (cid:12) .The combination of the presence of [Ca II] and a lackof narrow hydrogen emission is consistent with a double-detonation progenitor system, which is what is inferredby Siebert et al. (2020). Despite these limits, we cannotunequivocally claim that SN 2019yvq is a double detona-tion event due to discrepancies in best-fit models of pho-tospheric photometry and nebular spectroscopy. Ourconclusion in this regard is in agreement with Tuckeret al. (2020) and Miller et al. (2020b), and is discussedin more detail in Section 5.2. COMPARISONS TO SN 2002ES
Parameter 02es-like SNe Ia SN 2019yvq M B -17.6 – -18.1 -18.41∆ m ( B ) 1.1 – 1.3 1.36Rise time (days) 19 – 20 18.7( B − V ) max v Si II (km s − ) 6000 – 10000 14400Ti II at peak Yes Yesnebular [Fe II] and [Ca II] Yes Yes Table 2.
Comparisons between SN 2019yvq and 02es-likeSNe Ia. Parameter ranges for 02es-like SNe Ia are taken fromTaubenberger (2017) and are intended to be approximate,reflecting the small sample size and diversity of this subclass.
SN 2019yvq shares some characteristics with 02es-likeSNe Ia, and could be considered an 02es-like depend-ing on how broad a definition of that subclass is taken.We classify it as a transitional 02es-like. Although thisterm has not previously been used in the literature to de-scribe any objects, it accurately reflects the nature of SN2019yvq. Table 2 summarizes various photometric andspectroscopic signatures of 02es-like SNe Ia, taken fromTaubenberger (2017). See Ganeshalingam et al. (2012)for a study of the eponymous SN 2002es, and Tauben-berger (2017) and White et al. (2015) for reviews of thissubclass.SN 2019yvq is at the edge of what could be consid-ered 02es-like in several respects. Its peak brightnessand lightcurve width are on the edge of the class, asseen in the left panel of Figure 5. Like 02es-like SNeIa, SN 2019yvq also displays an almost nonexistent sec-ondary IR maximum and red colors after its initial blueexcess (see Figure 4 and its similarity to the 02es-likeiPTF14atg).Spectroscopically there are both similarities and ob-vious differences, as highlighted in Figure 7. The peakspectrum of SN 2019yvq is most similar to SN 2002bo,which also displayed deep Si II 6355 and had a similar SiII line ratio. SN 2002bo had a more typical peak lumi-nosity for SNe Ia ( M B = − .
41, Benetti et al. 2004).SN 2019yvq’s Si II velocity and line ratio make it anoutlier compared to other 02es-like SNe Ia, since thesespectral features would normally indicate an energeticand luminous event. Figure 7 also includes for com-parison SN 2006bt, which displayed Si II 6355 whichwas higher-velocity and broader than typical SNe Ia,but weaker and lower-velocity than SN 2019yvq. Wewould also classify SN 2006bt as a transitional 02es-like(in agreement with Taubenberger 2017), and we refer to he Transitional 02es-like SN 2019yvq N o r m a li ze d d e r e dd e n e d F λ + o ff s e t SNe Ia Peak Spectra Comparison -19.5-19.0-18.5-18.0-17.5-17.0 M B Figure 7.
Comparisons of SNe Ia peak spectra over a wide range of luminosities. Although the spectrum of SN 2019yvq is quitesimilar to SN 2002bo (a more typical luminosity SN Ia), its primary difference is in the ∼ Foley et al. (2010) for a thorough study of this unusualobject.02es-like SNe Ia are also characterized by Ti II atpeak, which is seen in lower luminosity SNe Ia like SN1991bg (see Figure 7). We note that the spectra of SN2019yvq and SN 2002bo are quite dissimilar bluewardsof ∼ Burke et al. SAB0 -15.8 Swift YesiPTF14atg E-S0 -15.5 Swift YesiPTF14dpk Starburst -16.3 R MaybePTF10acdh · · · -14.5 R UnknownPTF10ujn · · · -10.7 R UnknownPTF10bvr E ?? R UnknownSN 2002es S0 -7.3 B UnknownSN 1999bh Sb 0.6 B UnknownSN 2006bt , S0/a -2.6 B UnknownPTF10ops , SAa? -6.6 B UnknownSN 2010lp SAb -7 B Unknown
Table 3.
A literature sample of known 02es-like SNe Ia.iPTF14atg is the only other 02es-like observed in blue filtersas early as SN 2019yvq, and it also displays a UV excess.iPTF14dpk displayed a sharp rise from its last non-detection,and its first detection is high relative to a power law rise.PTF10ops is either ∼
148 kpc offset from the spiral galaxySDSS J214737.86+055309.3, or in a very faint satellite galaxyof it. Sources: 1: this work; 2: Cao et al. (2015); 3: Caoet al. (2016); 4: White et al. (2015); 5: Ganeshalingam et al.(2012); 6: Taubenberger (2017); 7: Foley et al. (2010); 8:Maguire et al. (2011).
Table 3 lists all known 02es-like SNe Ia, including SN2019yvq. The three SNe which were detected the earli-est all display unusual lightcurve properties. iPTF14atg(Cao et al. 2015) has already been discussed as a primeexample of an early UV excess. The early lightcurve ofiPTF14dpk (Cao et al. 2016) differed from iPTF14atg,as it rose more than 1.8 magnitudes/day between its lastnon-detection and earliest detection (in R , the only ob-served band at that epoch). Cao et al. (2016) take thisas evidence of a dark phase, a time period after the ex-plosion where the energy generated by radioactive decayhas not yet reached the photosphere (i.e. the explosionhas occurred but is not yet visible). The lightcurve alsodeclined between the first and second epochs, althoughCao et al. (2016) attribute this to scatter consistent withthe errors and not a physical dimming. The paper con-cludes that the lightcurve of iPTF14dpk is consistentwith the ejecta-companion interaction scenario but seenfrom an unfavorable viewing angle.The fact that the three 02es-like SNe Ia which have theearliest observations all display extremely unusual, butconsistent, lightcurve properties could be evidence thatthey all arise from identical progenitor systems, but thesample of such well-observed events will need to be ex-panded beyond its current limited numbers to make this statement with statistical confidence. But even with thesmall sample size we can say that the companion-ejectainteraction models, which predict a strong UV excess ∼
10% of the time due to viewing angle constraints, areunlikely to be the source of 02es-like SNe Ia if two ofthe three SNe observed at the right epochs display suchan excess with certainty, and the third displays a poten-tial weak excess. We discuss these implications more inSection 7. MODEL COMPARISONSWe compare SN 2019yvq to two main classes of modelswhich are capable of producing early blue bumps: com-panion shocking models from Kasen (2010) and doubledetonation sub-Chandrasekhar mass models from Polinet al. (2019a). Our best-fit models in these two cate-gories are included in Figure 8. We also discuss com-parisons to models with varying Ni distributions. Noone model reproduces all features of the dataset, so wediscuss their benefits and shortcomings.5.1.
Companion Shocking
As discussed in the introduction, Kasen (2010) pre-dicted that an early blue/UV excess could be seen inthe lightcurves of SNe Ia when the ejecta collide with anondegenerate companion and gets shock-heated. Thisexcess arising from companion shocking would only bevisible within a few days of the explosion, and wouldonly be seen for ∼
10% of SNe Ia due to viewing angleeffects.Hosseinzadeh et al. (2017) previously used these mod-els to fit the lightcurve of SN 2017cbv. As described inthat paper, they require a total of eight parameters togenerate fits: (1) the explosion epoch t , (2) the com-panion separation a , (3) a factor involving the ejectamass and speed ( x ∝ M v ), (4) the time of maximum t max , (5) the lightcurve stretch s , (6) and (7) factors onthe r and i flux of the SiFTO template (Conley et al.2008) r r and r i , and (8) a factor on the U shock flux r U .We make use of lightcurve fitting (Hosseinzadeh2019) to fit these models, which uses a Markov ChainMonte Carlo routine based on the emcee package(Foreman-Mackey et al. 2013) to generate fits. The mod-els consist of two components: a blackbody flux com-ponent and a SiFTO template which can be stretchedand scaled. We extend the blackbody component of themodel to include the early UVW2 , UVM2 , and
UVW1Swift data, since the first two epochs were taken in aregime where the SN flux was dominated by the earlyexcess.Fits struggled to converge until the following stepswere taken: (1) we put a tight prior on the explosion he Transitional 02es-like SN 2019yvq -22-20-18 -16-14-12 A b s o l u t e m ag n i t ud e + o ff s e t UVW2 - 4UVM2 - 3UVW1 - 2 U - 1B+0g+1 Unfiltered+2V+2 r+3i+4LCO + early Swift dataCompanion shockingDouble detonation
Figure 8.
Comparisons between the Las Cumbres and early
Swift data for SN 2019yvq and two different models. Thenon-detection and first detection from Itagaki are includedin black. Shown in the dashed line is the best-fit compan-ion shocking model from Kasen (2010). The parameters forthis model are in Table 4 (see Section 5.1 for more detail).The SN template used to generate the companion shockingmodel did not extend into the mid-UV, so only the black-body flux component is shown for the
Swift filters. The dot-ted line is the best-fit double detonation model from Polinet al. (2019a): a 0.95 M (cid:12)
WD progenitor with 0.055 M (cid:12) ofHe (see Section 5.2 for more detail). epoch and enforced adherence to the non-detection fromItagaki Astronomical Observatory, and (2) we extendedthe multiplicative factor on the U shock flux to include Swift data due to the strength of the excess in thosebands as well. The parameters for our best-fit model arelisted in Table 4, along with the corresponding best-fitmodel for SN 2017cbv from Hosseinzadeh et al. (2017).The most significant of these is the r U factor: Hos-seinzadeh et al. (2017) find that the U shock flux for SN 2019yvq SN 2017cbv t (MJD) 58844.3 ± a (R (cid:12) ) 52 +6 − M M Ch (cid:0) v − (cid:1) . ± .
03 3 . ± . t max (MJD) 58863 . ± .
08 57840.2 s . ± .
007 1.04 r r . ± .
006 0.95 r i . +0 . − . r U . ± .
04 0.61
Table 4.
Comparisons between the best-fit parametersof the Kasen (2010) companion shocking models for SN2019yvq (this work) and SN 2017cbv (Hosseinzadeh et al.2017). Parameters: time of explosion ( t ), companion sepa-ration ( a ), a parameter involving the ejecta mass and velocity( ∝ Mv ), time of peak ( t max ), lightcurve stretch ( s ), factorson the r and i flux in the SiFTO template ( r r , r i ), and a fluxfactor on the U though UV W r U ). models describing SN 2017cbv must be scaled by a fac-tor of 0.61. There are several possible explanations forthis, including assumptions of spherical symmetry andblackbody SEDs, or the effects of line blanketing fromiron group elements (IGEs) causing the UV/blue flux tobe overestimated.However, we do not find that the U (and U V W U V M U V W
2) shock flux needs to be scaled down tomatch the data. Instead the best-fit model has a UVflux enhancement of about 27%. An increase of thisamount is unsurprising: the analytic expressions for theblackbody luminosity used in lightcurve fitting andderived from Kasen (2010) replicate the numerical mod-els of companion-ejecta interaction seen at a viewingangle of approximately 30 ◦ (see Figure 2 of that pa-per). Explosions with smaller viewing angles result inhigher observed luminosities, up to about 0.25 dex (afactor of 1.8) brighter for a perfectly aligned scenario.Although our model does not include the viewing angleas a parameter, better-aligned explosions can generatethe required shock flux enhancement.The other notably discrepant parameter between thetwo fits is the parameter involving mass and velocity. Itis worth noting that the relevant velocity is not exactlythe ejecta velocity, rather it is the transition velocity be-tween different power laws in the density profile for themodeled ejecta. Assuming M Ch of ejecta, the value ofthis parameter for SN 2017cbv corresponds to a veloc-ity of about 12000 km s − . Using the same assumption,the value for SN 2019yvq corresponds to a transitionvelocity of about 7000 km s − .The best-fit companion separation (52 R (cid:12) ) impliesa companion radius of ∼
20 R (cid:12) , assuming Roche lobe4
Burke et al. overflow (Eggleton 1983). This stellar radius does notexclude most main sequence stars, and the separationlies towards the extreme of the expected distribution formain sequence donor stars, based on binary populationsynthesis models (Liu et al. 2015).Miller et al. (2020b) also use the Kasen (2010) modelsto fit their data, although with a different methodol-ogy. They fit only shock-dominated data (within ∼ ± (cid:12) and an explosion date of58845 . ± .
04 (MJD). This companion separation isseveral times smaller than our best-fit value (Table 4),and the explosion date is more than 1.5 days after ours.Since their explosion date is in fact almost two hours af-ter the initial detection from Itagaki, we are unsurprisedby the disagreement in companion separations.As a final remark on the best-fit parameters in Table4, we note that SN 2019yvq and SN 2017cbv have similarrise times (18.7 days and 18.2 days, respectively). Thesevalues are quite typical for SNe Ia – Firth et al. (2015a)find an average rise time of 18 . ± .
54 days in a sampleof 18 well-sampled objects.Although lightcurve fitting generates modellightcurves and not spectra, we reproduce the spec-tral effects of this model by taking a spectrum of SN2011fe at a similar epoch to our earliest spectrum anddiluting it with a blackbody of the predicted size andtemperature. The effects of this blackbody dilution areshown in Figure 9, where it can be seen that they doa qualitatively good job replicating the early spectrumof SN 2019yvq (in black), with its blue continuum andweak features. Further, quantitatively fitting for thebest-fit temperature needed to reproduce the strengthof spectral features (keeping the radius the same as pre-dicted by the fits) results in a temperature only about350 K higher than predicted by the models. These twotemperatures being consistent with each other providesindependent confirmation of the validity of the compan-ion shocking models.Companion shocking models can produce a wide rangeof early blue bumps depending on the companion sepa-ration, size, and viewing angle (see Figures 2 and 3 ofKasen 2010). While the fits for SN 2019yvq are not per-fect, notably underpredicting the strength of the declineto the second epoch of
Swift data, they both closely re-produce the wavelength-dependent behavior of the earlyexcess and predict a temperature closely aligned withwhat is expected by diluting an early spectrum withblackbody flux. 5.2.
Double Detonation N o r m a li ze d d e r e dd e n e d F λ + o ff s e t Figure 9.
Our earliest spectrum of SN 2019yvq (black line)compared to a spectrum of SN 2011fe at a comparable epoch.Epochs listed with respect to days from B -band maximum.The magenta line represents the SN 2011fe spectrum dilutedby a 8794 K blackbody, the temperature predicted at thatepoch by our best-fit companion shocking models. Allowingthe temperature of the blackbody to vary and comparingto the the SN 2019yvq with a χ ν test, we obtain a best-fittemperature of about 350 K higher (yellow line). The greenline represents the spectrum at the same epoch (measuredfrom explosion) from the best-fit double detonation model. As described in detail in Polin et al. (2019a), the ex-plosion mechanism of these models consists of the igni-tion of a surface layer of He which then detonates theunderlying C/O WD. We compared observations of SN2019yvq with double-detonation models which had WDmasses between 0.6 and 1.3 M (cid:12) and He shell massesbetween 0.01 and 0.1 M (cid:12) .We measure the overall best-fit model in our grid bydoing a simple reduced χ comparison between eachmodel and the U BV gri photometry. We fix the explo-sion epoch to be the same used in the best-fit companionshocking model, as described in Section 5.1. Normallyone would infer an explosion epoch from a power-law fitto the rising data (e.g. Ganeshalingam et al. 2011; Firthet al. 2015b) however in this case these fits were verypoorly constrained. This was primarily due to a limited he Transitional 02es-like SN 2019yvq (cid:12)
WDwith a 0.055 M (cid:12) layer of He. This model is shown asthe dotted line in the photometry of Figure 8 and thecolor evolution of Figure 4, and the spectrum from thismodel matching the epoch of our earliest SN 2019yvqspectrum is shown in Figure 9. Although most of thisspectrum is a blue continuum with weak features, ingeneral agreement with the observations, we find thatit predicts much stronger features in the ∼ ∼ ∼ U data. Thedrop after the early excess is also stronger in all bandsthan is seen in the data, and the models predict a “redbump” which is not seen in the data (see Figure 4).Additionally, all reasonably well-fitting models in thegrid predict a U decline that is steeper than observed.In the case of the best-fit model, it is steeper than theobserved decline-rate by more than a factor of two (inmagnitudes per day).There are also several advantages to double detona-tion models which match the observed data: a lack ofC in the spectra, a weak secondary IR maximum, anda blue/UV excess at roughly the right epochs are somepoints of agreement.Both Miller et al. (2020b) and Siebert et al. (2020)use the models from Polin et al. (2019a) to fit differ-ent aspects of SN 2019yvq’s dataset. Fitting to the gri ZTF photometry in addition to some
Swift data over ap-proximately the same epochs shown here, Miller et al.(2020b) find a best-fit model consisting of a 0.92 M (cid:12)
WD with a 0.04 M (cid:12)
He shell. Their results are similarto what is presented here: general agreement on somecounts (early blue excess), and diagreement on others(difficulty fitting bluer filters).Siebert et al. (2020) extend the best-fit model of Milleret al. (2020b) into the nebular phase, and show that thebest-fit model based on photospheric photometry is apoor match for nebular spectroscopy, overpredicting thestrength of the [Ca II] and [Fe II] feature by a factorof several. Instead, to match the nebular spectra theyfind a best-fit model consisting of a 1.1 M (cid:12)
WD with a0.05 M (cid:12)
He shell. This nebular model is in turn a poormatch to the photospheric photometry, overpredictingthe bluer bands by more than a magnitude and greatlyunderpredicting the strength of the early excess in opti-cal bands. We find it difficult to reconcile this discrepancy, andcannot definitively claim that SN 2019yvq is the result ofa double-detonation, despite the several points in favorof these models as listed above.5.3.
Nickel Distributions
Photometry
Variations in Ni distributions in the WD progenitorare also known to produce a range of SN Ia behavior(e.g. Piro & Morozova 2016; Magee et al. 2020).Using the same methodology described in Section 5.2,we look for best-fit models from the grid of 255 modelsprovided by Magee et al. (2020). These models makeuse of the radiative transfer code
TURTLS (Magee et al.2018) and vary the density profiles, Ni masses, kineticenergy, and degree of Ni mixing to produce a range oflightcurves up to +25 days from the explosion.Fitting the
UBVgri
Las Cumbres lightcurve, we findthe best-fit model is
EXP Ni0.8 KE0.50 P4.4 . This hasan exponential density profile, 0.8 M (cid:12) of Ni, and a ki-netic energy of 0.50 foe. The last element of the modelname (
P4.4 ) describes the scaling parameter which de-termines the Ni distribution, and indicates the Ni is com-paratively mixed through the ejecta.However, while this model does as well as the othertwo classes of models we have discussed at fitting the risetime and peak absolute magnitude, it contains no earlyexcess. The authors note in Magee et al. (2020) thatalthough they can fit a majority of SNe in their sample,the remaining objects have an early excess which themodels cannot replicate. Since we consider the early UVexcess to be the most unique feature of this SN, the mostdifficult and interesting aspect to model, and potentiallythe biggest clue to what the progenitor system is, we donot include this best-fit model in Figure 8.The same authors also released a set of models usinga similar methodology capable of reproducing early ex-cesses due to clumps of Ni in the outer ejecta (Magee &Maguire 2020). However, since these models were basedon SN 2017cbv and SN 2018oh data and both these SNehad typical peak luminosities unlike the underluminousSN 2019yvq, we do not include them as comparisons.Additionally, these models display early red bumps sim-ilar to those seen in the double detonation models, whichare not seen in our data (see Figure 4).5.3.2.
Spectroscopy
In addition to the above photometric modeling, wealso utilize
Tardis (Kerzendorf & Sim 2014) to exam-ine the spectroscopic effects of varying Ni distributionsand photospheric velocities. A full exploration of theseeffects are outside the scope of this paper, but we reportinitial observations here.6
Burke et al.
We start with a base model, which consists of an earlySN 2011fe spectrum identical to the one used in Heringeret al. (2017) at an epoch of +5 . v inner boundary (photospheric velocity) of this modelis 12 ,
400 km/s. We then alter the Ni distribution andphotospheric velocity of this model in an attempt toreplicate the SN 2019yvq.Our perturbations were unsuccessful at reproducingthe earliest spectrum, but we note observable effects ofaltering the Ni distribution. Adopting a uniform Ni dis-tribution for the outer ejecta with a mass fraction of0.19 (replicating the most mixed model of Piro & Moro-zova 2016), we note that the red wings of the Si II 6355and O I 7774 lines become asymmetrically broader, andthat the Ca NIR triplet drastically reduces in strength.Artificially introducing a mass of Ni in the outermostportions of the ejecta ( > ,
000 km/s) weakens the MgII complex and other features blueward of ∼ ,
000 km s − , whichis significantly higher than the default value of 12 , − . Miller et al. (2020b) find velocities of as highas 25 ,
000 km s − are necessary to fit their earliest spec-trum, but since the maximum velocity in the Tardis model is 24 ,
000 km s − this is unreachable for us. Wedo note that at high photospheric velocities, such as18 ,
000 to 20 ,
000 km s − , the strengths of most spectro-scopic features begin to match the weak values of ourearliest spectrum and the spectrum begins to be dom-inated by a blue continuum. However, as also pointedout by Miller et al. (2020b), Tardis has a photosphericboundary which is not wavelength-dependent inside ofwhich is a quasi-blackbody. Because our
Tardis mod-els have a limited velocity range, increasing the model’sphotospheric velocity thus increases the percentage ofthe model’s mass which acts as a blackbody and ef-fectively dilutes the spectral features from the tenuousouter layers with a strong blackbody component. Black-body dilution is also a signature of the companion shock-ing models, and is shown in Figure 9. The blackbodytemperature predicted by the companion shocking mod-els is also thousands of Kelvin hotter than the photo-spheric temperatures
Tardis calculates for this velocityrange (between 6,000 and 7,000 K).Miller et al. (2020b) use additional Ni distributionmodels based on Magee & Maguire (2020) and find that the predicted spectra have strong line blanketing blue-ward of ∼ i -band flux.Since unusual Ni distributions result in spectral fea-tures absent in the observed spectra, and since high pho-tospheric velocities replicate the effects of the companioninteraction scenario, we do not include these spectra inour comparisons. PROGENITOR CONSTRAINTS FROM RADIOOBSERVATIONSRadio emission is a sensitive probe of circumstellarmedium (CSM) of the progenitor. The CSM is pol-luted by mass-loss from the progenitor in the pre-SNstage, and interaction of the SN ejecta with this CSM ac-celerates electrons to relativistic energies and amplifiesthe ambient magnetic field, producing synchrotron ra-dio emission (Chevalier 1982, 1984, 1998). Simple mod-els of radio emission have provided constraints on theCSM environment and progenitor properties for bothcore-collapse (e.g. Ryder et al. 2004; Soderberg et al.2006; Chevalier & Fransson 2006; Weiler et al. 2007;Salas et al. 2013) and SNe Ia (Panagia et al. 2006;Chomiuk et al. 2016). Radio emission is yet to be de-tected from a SN Ia , but non-detections have providedstringent constraints on progenitor scenarios (Chomiuket al. 2016), particularly for nearby events like SN 2011fe(Horesh et al. 2012; Chomiuk et al. 2012) and SN 2014J(P´erez-Torres et al. 2014).Radio observation of SN 2019yvq was obtained withthe Karl G. Jansky Very Large Array (VLA) on 2020Jan 26, 11:39:53, which is within 29.77 days of t (de-rived in Section 2.2). The observation block was 1-hrlong, with 38.23 mins time-on-source for SN 2019yvq.Observations were taken in X-band (8–12 GHz) in the D-configuration of the VLA (DDT: 19B-346, PI: S. Sarbad-hicary). The observations were obtained in wide-bandcontinuum mode, yielding 4 GHz of bandwidth sampledby 32 spectral windows, each 128 MHz wide sampledby 1 MHz-wide channels with two polarizations. Weused 3C286 as our flux and bandpass calibrator, andJ1313+6735 as our phase calibrator. Data were cali-brated with the VLA CASA calibration pipeline (ver-sion 5.6.2-2) . The pipeline consists of a collection ofalgorithms that automatically loads the raw data intoa CASA measurement set (MS) format, flags corrupteddata (e.g. due to antenna shadowing, channel edges, ra-dio frequency interference or RFI), applies various cor-rections (e.g. antenna position, atmospheric opacity) https://science.nrao.edu/facilities/vla/data-processing/pipeline he Transitional 02es-like SN 2019yvq tclean in CASA. We used multi-term, multi-frequencysynthesis as our deconvolution algorithm (set with deconvolver=‘mtmfs’ in tclean ), which performs de-convolution on a Taylor-series expansion of the wide-band spectral data in order to minimize frequency-dependent artifacts (Rau & Cornwell 2011). We set nterms=2 which uses the first two Taylor terms to cre-ate images of intensity (Stokes-I) and spectral index.The SN is offset ∼ (cid:48)(cid:48) from the bright central radio nu-cleus of the galaxy, and as a result the emission at theSN site is dominated by sidelobes from the nucleus forthe typical resolution ∼ . (cid:48)(cid:48) expected in X-band imagesin D-configuration. For this reason, we only imaged the10-12 GHz bandwidth with tclean , excluded visibil-ity data from baselines shorter than 6 k λ , and appliedBriggs-weighting on the remaining visibility data withthe parameter robust=0 . This provided just enoughangular resolution and source sensitivity at the SN siteto determine if any radio emission separate from thenucleus is associated with the SN site.No radio source was detected at the site of SN 2019yvqin the cleaned, deconvolved 11-GHz image with a syn-thesized beam of 5 . (cid:48)(cid:48) × . (cid:48)(cid:48) . The flux at the exactlocation of the SN is − µ Jy. Using the AIPS task
IMEAN , we obtain an RMS of 11 . µ Jy per beam, whichtranslates to a 3 σ . × ergs/s/Hz, assuming a distance of 42.5 Mpc.The 3 σ upper limit can shed some light on the CSMaround 2019yvq similar to the methodology in Chomiuket al. (2012) and Chomiuk et al. (2016). Using theChevalier (1982) model of a CSM characterized by ρ =˙ M / πr v w (where ρ is density in gm/cm , ˙ M is themass-loss rate from the progenitor, r is the distancefrom progenitor and v w is wind velocity), we obtain anupper limit of (4 . × − M (cid:12) /yr on the mass-loss rate from a symbiotic progenitor (involving a red-giant companion, assuming v w =10 km/s). The range ofmass-loss rates reflect the uncertainty in the parameter (cid:15) b , the fraction of shock energy shared by the amplifiedmagnetic field, with typical values in the range 0.01-0.1 for SNe (Chomiuk et al. 2012). These limits areshown in Figure 10. Chomiuk et al. (2016) measuredthe mean mass-loss rate in symbiotic progenitors in theMilky Way to be log ( ˙ M ) = − . ± .
03 M (cid:12) /yr (ass-suming v w = 100 km/s), so our measurement does notexclude the possibility of a red-giant companion. Sce-narios involving accretion from a main-sequence com-panion accompanied by steady nuclear burning are alsonot excluded by our limit (Chomiuk et al. 2012). -9 -8 -7 -6 Mass loss rate [M fl /yr]10 W i nd v e l o c i t y [ k m / s ] SymbioticprogenitorsOptically-thickaccretion windsNovashellsQuiescencebetweennovae OuterLagrangianlosses
Wind Model (Chevalier 1982) † b = 0 . † b = 0 . Figure 10.
Limits (in gray) for the mass loss rate of theprogenitor of SN 2019yvq from its VLA observations, follow-ing the model of Chevalier (1982), shown for typical rangeof values of (cid:15) b which parameterizes the fraction of shock en-ergy in the amplified post-shock magnetic field in radio lightcurve models. These observations can rule out some symbi-otic progenitor systems, but they do not exclude red giantcompanions or other methods of mass loss.7. DISCUSSIONSN 2019yvq is an unusual event in many respects. Ithas: a strong early UV flash; red colors besides the earlyflash; relatively faint peak luminosity, a moderately highdecline rate, and a weak secondary IR maximum; broad,high-velocity Si II 6355 paired with both weak Si II 5972and Ti II at peak; and nebular [Ca II] and [Fe II]. Thesepaint a conflicting picture, with some aspects pointing toa low-energy explosion (low luminosity, weak secondaryIR maximum, nebular [Ca II], peak Ti II) and otherspointing to a high-energy event (Si II velocity and lineratio). Due to several characteristics it shares, or almostshares, with low-luminosity 02es-like SNe Ia, we classifyit as a transitional member of that subclass (see Table2 and the rest of Section 4).This object being a transitional 02es-like has two ma-jor implications.The first is the confirmation that transitional 02es-like SNe Ia can exist. This has precedent in the ob-ject SN 2006bt (Foley et al. 2010; Ganeshalingam et al.2010), which can be considered a transitional memberof this class (Taubenberger 2017) despite its high veloc-ities (12,500 km s − at 3 days before maximum) andrelatively bright luminosity ( M B, peak ∼ −
19, with un-8
Burke et al. certain reddening correction). This object is included inboth Figure 5 (orange star) and Figure 7 for compari-son. However, SN 2019yvq is by no means a clone of SN2006bt as it lies in extremely sparsely populated regionsof parameter space in several respects (see Figure 5, alsoFigure 2 of Tucker et al. 2020). On the Phillips relationSN 2019yvq has similar parameters to SN 2012Z, buton the Branch diagram SN 2019yvq is most similar toSN 2002bo. SNe 2002bo and 2012Z are substantiallydifferent SNe. A transitional 02es-like SN that not onlyshares characteristics with both these SNe but is alsodistinct from another transitional member of its subclasssupports evidence that there is a continuum of eventsbetween normal SNe Ia and 02es-likes. Assuming a con-tinuum of events instead of discrete subclasses, this alsosuggests that 02es-like SNe do not arise from progenitorsystems which are distinct from the systems of normalSNe Ia.The second major implication comes from the factthat the three 02es-like SNe Ia with very early data (SN2019yvq, iPTF14atg, and iPTF14dpk) all display un-usual early-time lightcurves (see Section 4 and Table 3).Of these, the two with
Swift data at these early epochsdisplay the two strongest early UV flashes in SNe Ia.iPTF14dpk unfortunately only has R -band photometry,and while at first glance its first data point appears in-dicative of an early excess, Cao et al. (2016) say that thiswould require an extreme explosion energy and wouldlead to higher velocities than are observed. The lack ofmulti-band photometry makes us hesitant to accept thatconclusion incontrovertibly. According to Kasen (2010),if such early excesses are due to companion–ejecta shockinteraction they should only be seen in ∼
10% of eventswith such early data. Instead, for 02es-like SNe Ia, theyare seen in two (or three) of the three early events. Thisis unlikely – even with the current small sample size, theodds of so many early excesses are somewhere between 1in 100 and 1 in 1000. And as discussed in Section 5.2, thediscrepancies between photospheric and nebular best-fitmodels make us hesitant to claim that SN 2019yvq is adouble detonation event either, even though those mod-els can produce early UV excesses. We are left consid-ering progenitor scenarios which could produce an earlyexcess which is both fit relatively successfully by shockinteraction models but is not viewing angle-dependent.In addition to models which have already been dis-cussed (double detonations and varied Ni distributions,see Sections 5.2 and 5.3.1), there are a few possibilitiesfor progenitor systems configured in such a way to pro-duce more isotropic shocks. One option lies in the ac-cretion disks which form as the (primary) WD accretesmatter. Levanon & Soker (2019) model the exquisitely sampled early bump seen in the K2 data of SN 2018ohas the interaction of the SN ejecta with what they re-fer to as “Disk-Originated Matter,” since accretion diskscould also give rise to bipolar jets. The addition of anaccretion disk and jets would more easily account for theubiquity of early excesses since these components can beseen more isotropically. Piro & Morozova (2016), in ad-dition to modeling the degree of Ni mixing in WD pro-genitors, also investigate the effects of a more generaldistribution of CSM. These models can produce earlyexcesses which occur on a range of timescales and inten-sities, depending on the total amount of external mat-ter in the CSM and its density scaling. In particularthey can produce early bumps which only last ∼ CONCLUSIONS & SUMMARYWe have discussed the discovery and follow-up obser-vations of SN 2019yvq, a nearby SN Ia with a rare andunusually strong excess in its early lightcurve, in ad-dition to several other uncommon features. This earlyexcess is most pronounced in the UV, where the objectis brighter during the excess than during the epochs ofits optical peak.This object is one of a very limited number of SNeIa with early UV/blue excess, and it demonstrates aneven stronger excess than other objects in the sample.SN 2019yvq deviates significantly from SNe Ia that areblue at early times but otherwise normal. Instead itshares some, but not all, features of the 02es-like SNIa subclass, including a low peak luminosity, red color,moderately high decline rate, Ti II at peak, and neb-ular [Ca II] and [Fe II]. We classify SN 2019yvq as atransitional member of the 02es-like subclass.Although models which simulate WD double detona-tion and ejecta–companion shock interaction can cre-ate lightcurves with excess flux at early times, we findthat no one model can accurately reproduce all unusual he Transitional 02es-like SN 2019yvq i -band ZTF data,post-maximum TESS data, and a Keck NIRES spec-trum) and, like us, are unable to satisfactorily explainevery aspect of the SN 2019yvq dataset. As in Siebertet al. (2020) we also find strong [Ca II] and [Fe II] emis-sions in the nebular spectra of SN 2019yvq in additionto strong limits on the amount of swept-up H and He,but we do not take this as exclusive evidence of a doubledetonation explosion.Two other 02es-like SNe Ia also display unusual earlylightcurves (iPTF14atg and iPTF14dpk). The devia-tions from a power-law rise in all 02es-like SNe Ia withsufficiently early data makes us further doubt that theearly UV excess seen in SN 2019yvq arises from ejecta–companion shock interaction, as viewing angle effectsdictate that such excesses should only be seen in ∼ ∼ lightcurve fitting . We also thank E. Heringer forproviding the Tardis models from Heringer et al. (2017), and R. Cartier for providing the syn++ mod-els from Cartier et al. (2017).J.B., D.A.H., D.H., C.M., and C.P. are supported byNSF grants AST-1313484 and AST-1911225, as well asby NASA grant 80NSSC19kf1639.S.K.S. and L.C. are supported by NSF grant AST-1907790.Time domain research by D.J.S. is supported by NSFgrants AST-1821987, 1813466, & 1908972, and by theHeising-Simons Foundation under grant
Tardis , a community-developed software package for spectral synthesis in su-pernovae (Kerzendorf & Sim 2014; Kerzendorf et al.2019). The development of
Tardis received supportfrom the Google Summer of Code initiative and fromESA’s Summer of Code in Space program.
Tardis makes extensive use of Astropy and PyNE.
Facilities:
Las Cumbres Observatory (Sinistro),FTN (FLOYDS), Bok (B&C Spectrograph), MMT(Blue Channel spectrograph), IRTF (SpeX), Swift(UVOT), VLA
Software: astropy (Astropy Collaboration et al.2013; The Astropy Collaboration et al. 2018),
SNooPy (Burns et al. 2011),
Tardis (Kerzendorf et al. 2019),sncosmo (Barbary et al. 2016), SALT2 (Guy et al. 2007),MLCS2k2 (Jha et al. 2007), lightcurve fitting (Hos-seinzadeh 2019), emceee (Foreman-Mackey et al. 2013)REFERENCES
Astropy Collaboration, Robitaille, T. P., Tollerud, E. J.,et al. 2013, A&A, 558, A33Barbary, K., rbiswas4, Goldstein, D., et al. 2016,doi:10.5281/zenodo.168220Bellm, E. C., Kulkarni, S. R., Graham, M. J., et al. 2019,PASP, 131, 018002Benetti, S., Meikle, P., Stehle, M., et al. 2004, MNRAS,348, 261Bianco, F. B., Howell, D. A., Sullivan, M., et al. 2011, ApJ,741, 20 Blondin, S., & Tonry, J. L. 2007, ApJ, 666, 1024Blondin, S., Matheson, T., Kirshner, R. P., et al. 2012, AJ,143, 126Boty´anszki, J., Kasen, D., & Plewa, T. 2018, ApJL, 852, L6Branch, D., Dang, L. C., Hall, N., et al. 2006, PASP, 118,560Breeveld, A. A., Curran, P. A., Hoversten, E. A., et al.2010, MNRAS, 406, 1687Brown, P. J., Breeveld, A. A., Holland, S., Kuin, P., &Pritchard, T. 2014, Ap&SS, 354, 89 Burke et al.