The Progenitor and Early Evolution of the Type IIb SN 2016gkg

Type IIb supernovae (SNe IIb) are likely the result of the core collapse of massive stars that have lost most of their hydrogen envelope prior to explosion. At early times, and through maximum light, SNe IIb show hydrogen features typical of Type II SNe, which later give way to He i absorption lines similar to those observed in SNe Ib (Filippenko et al. 1994).

Several SNe IIb caught soon after explosion exhibited "double-peaked" light curves. These have been interpreted as the signature of shock breakout cooling of a progenitor star core surrounded by extended, low-mass material (e.g., Bersten et al. 2012; Nakar & Piro 2014). Well-studied examples of this phenomenon include SNe 1993J (e.g., Richmond et al. 1994), 2011dh (e.g., Arcavi et al. 2011; Ergon et al. 2014), 2011fu (Kumar et al. 2013), and 2013df (Morales-Garoffolo et al. 2014; Van Dyk et al. 2014). Progenitor constraints from deep pre- and post-explosion imaging (e.g., Aldering et al. 1994; Maund & Smartt 2009 for SN 1993J) have revealed a picture of Type IIb's originating from yellow massive stars (${M}_{\mathrm{ZAMS}}=12-16$ M${}_{\odot }$) in binary systems.

Here, we present the first month of evolution of the Type IIb SN 2016gkg, progenitor constraints from pre-explosion Hubble Space Telescope (HST) imaging and its early temperature evolution. SN 2016gkg was discovered by V. Buso on 2016 September 20.19 UT,¹⁹ 784 south and 491 west from the nucleus of NGC 613. It was confirmed via photometry (Nicholls et al. 2016; Tonry et al. 2016) and typed as a young Type II SN (Jha et al. 2016). After a fast decline (Chen et al. 2016), its light curve began to rise again toward a second maximum. Progenitor constraints from HST pre-imaging at the location of SN 2016gkg were also reported (Kilpatrick et al. 2016), and we will discuss these further below. We adopt a distance of $26\,\mathrm{Mpc}$, based on a Tully–Fisher measurement ($m-M=32.1\pm 0.4\,\mathrm{mag}$; Tully et al. 2009). However, since Tully–Fisher measurements for NGC 613 range from $\sim 20\ \mathrm{to}\ 30\,\mathrm{Mpc}$, we will also discuss the implications of a lower host distance ($20\,\mathrm{Mpc}$). We assume ${A}_{V}=0.053\,\mathrm{mag}$ for the foreground Galactic extinction (Schlafly & Finkbeiner 2011).

2.1. Photometry

Imaging data were processed as follows. ${UBVgri}$ frames from the Las Cumbres Observatory global telescope network (LCO; Brown et al. 2013) were reduced using lcogtsnpipe (e.g., Valenti et al. 2016). UBV RI data from the Public ESO Spectroscopic Survey of Transient Objects (PESSTO²⁰ ) using the ESO Faint Object Spectrograph and Camera v2 (EFOSC2) were reduced using the PESSTO pipeline (Smartt et al. 2015). ${gri}$ data were also obtained by the Nordic Optical Telescope (NOT) Unbiased Transient Survey (NUTS²¹ ) with the Andalucia Faint Object Spectrograph and Camera (ALFOSC) on the NOT and reduced using a dedicated pipeline (SNOoPY; Cappellaro 2004). Filterless data from the $D\lt 40\,\mathrm{Mpc}$ (DLT40) SN search using the PROMPT5 telescope (Reichart et al. 2005) were calibrated to APASS r-band.²² Early Swift data (PIs: Brown, Drout) were reduced following the prescriptions of Brown et al. (2009), using the zeropoints from Breeveld et al. (2010). Data from the Asteroid Terrestrial-impact Last Alert System (ATLAS) telescope system (Tonry 2011) were also used in our early-time light curve (Arcavi et al. 2016).

The light curves are presented in Figure 1, including publicly available early-time photometry from ASAS-SN and other sources (Nicholls et al. 2016).

Zoom In Zoom Out Reset image size

2.2. Spectroscopy

Spectra were obtained from multiple sources, reduced in a standard manner. The classification spectrum (Jha et al. 2016) was obtained using the Southern African Large Telescope with the Robert Stobie Spectrograph and reduced using pysalt (Crawford et al. 2010). Optical spectra from PESSTO were obtained with EFOSC2, while NIR spectra were taken with the Son OF ISAAC camera (SOFI; Moorwood et al. 1998); all of these data were reduced as in Smartt et al. (2015). FLOYDS optical spectra from LCOGT were reduced as in Valenti et al. (2014). Optical spectra from ALFOSC and the NUTS collaboration were reduced using their dedicated pipeline (FOSCGUI²³ ). Gemini south data using the NIR spectrograph FLAMINGOS-2 (${JH}$ grism, $1.0-1.8\,\mu {\rm{m}}$, and $R\approx 1000$) were reduced using the F2 pyraf package. The final sequence is presented in Figure 2. Spectra will be released through the Weizmann Interactive Supernova data REPository (Yaron & Gal-Yam 2012).

Zoom In Zoom Out Reset image size

2.3. HST Pre-explosion Imaging

3.1. Light Curves

3.2. Spectra

The spectral fluxes were adjusted using photometry at similar epochs, and corrected for Galactic foreground extinction and redshift of the host (${v}_{\odot }=1481\pm 5$ km s⁻¹; Koribalski et al. 2004). At early phases (i.e., $\mathrm{phase}\lesssim 2.7\,\mathrm{days}$) spectra are dominated by a blue, almost featureless continuum, while P-Cygni line profiles become evident as the temperature of the ejecta decreases.

Fitting a blackbody to the spectral continuum, we infer a temperature of $\simeq 10,000\,{\rm{K}}$ at $+1.70\,\mathrm{days}$, rapidly decreasing to $\simeq 7000\,{\rm{K}}$ at $\sim +5\,\mathrm{days}$. We also estimated the host-galaxy reddening at the position of SN 2016gkg using the equivalent width of the Na ID doublet in an XSHOOTER spectrum ($\mathrm{EW}({\rm{D}}1)=0.26\pm 0.02\,\mathring{\rm A}$, $\mathrm{EW}({\rm{D}}2)=0.42\pm 0.02\,\mathring{\rm A}$; see Figure 2(d)). Based on the correlation with the color excess (e.g., Poznanski et al. 2012), we obtain $E{(B-V)}_{\mathrm{NGC}613}\ \lesssim 0.15\,\mathrm{mag}$. Including this contribution, we get a slightly different temperature evolution: from $\simeq 13,000\,{\rm{K}}$ to $\simeq 7900\,{\rm{K}}$ in the first $\sim 5\,\mathrm{days}$ of the spectroscopic evolution.

Hα is the most prominent line. From ∼$+14\,\mathrm{days}$ a second component appears, most likely He i at 6678 Å, while Hα becomes more evident. In Figure 3(b), we report the evolution of the expansion velocities inferred from the position of the minima of the P-Cygni profile. We infer expansion velocities for H i, declining from $\sim \mathrm{16,500}$ km s⁻¹ at $+1.70\,\mathrm{days}$ to $\sim \mathrm{12,200}$ km s⁻¹ at $\sim +21\,\mathrm{days}$.

Zoom In Zoom Out Reset image size

At $\sim +3.93\,\mathrm{days}$ the He i line at 5876 Å becomes visible, with an expansion velocity decreasing from $\sim \mathrm{16,500}$ to $\sim 8800$ km s⁻¹ in $\sim 21\,\mathrm{days}$. The presence of helium in the early spectra of SN 2016gkg is also confirmed by our NIR spectra (Figure 2(b)). Our $\sim +4\,\mathrm{day}$ and $\sim 11\,\mathrm{day}$ spectra both show a blue continuum, with prominent Paschen lines in emission. The emission at $\sim 10750$Å is most likely a blend of $\mathrm{Pa}\gamma$ and He i at 10830 Å; the bluest absorption in the P-Cygni profile is consistent with the He i expansion velocity inferred from the optical spectra. From the shallow He i (20581 Å) feature visible at $\sim +11\mathrm{days}$ and $\sim +15\,\mathrm{days}$ we infer an expansion velocity of $\sim \mathrm{10,000}$ km s⁻¹, in agreement with our optical spectra.

To support the classification of SN 2016gkg as a Type IIb SN, we use the Gelato (Harutyunyan et al. 2008) comparison tool on our $\sim +19\,\mathrm{day}$ spectrum (see Figure 2(c)), finding good matches with Type IIb SNe (SNe 1993J, 2008aq, and 2011dh).

A detailed analysis, including the results from the complete spectroscopic follow-up campaign, will be presented in a forthcoming paper.

3.3. Progenitor Properties

We obtained adaptive optics imaging of SN 2016gkg with the Very Large Telescope + Nasmyth Adaptive Optics System Near-infrared Imager and Spectrograph (NaCo) on 2016 October 2.3 UT. Ks imaging of $4500\,{\rm{s}}$ ($75\times 60\,{\rm{s}}$) was taken with the S54 camera, using SN 2016gkg itself as a natural guide star. Sky-subtracted images were aligned and co-added to produce a single deep post-explosion image.

Pre- and post-explosion HST and NaCo images were aligned using seven common reference sources. All reference sources lie on one side of the SN position, leading to some extrapolation of the geometric transformation. A transformation between the two set of pixel coordinates was derived allowing for translation, rotation, and a single magnification factor. Since so few sources were used for the alignment, we are susceptible to individual influential data points. Hence, we used a jack-knife sampling technique to test the effects of excluding either one or two reference sources from the fit. We used the derived transformation in each case to determine a position for the SN on the pre-explosion F814W image. The resulting position lies between two sources: Source A, at pixel 586.4, 789.4 and Source B, located at 585.9, 787.3 (see Figure 4).

Zoom In Zoom Out Reset image size

The positions of A and B were measured using the dolphot package, while their uncertainties (∼0.5 WFPC2 pixels) were determined through Monte Carlo simulations. Source A is consistent with the $1\sigma$ error on the transformed SN position (including the rms errors from the geometric transformations, and the uncertainties estimated from jack-knife sampling), while B lies $\sim 1.5\sigma$ distant. Both sources are hence viable progenitor candidates for SN 2016gkg.

Both hstphot and dolphot (Dolphin 2000) were used for the photometry. hstphot only detects Source A, with magnitudes in the VEGAMAG system $F450W=23.75\pm 0.19$, $F606W\ =23.56\pm 0.10$, and $F814W=23.17\pm 0.11\,\mathrm{mag}$. The source is well fitted with a PSF, but has a sharpness parameter ($-0.31$) that is barely consistent with point sources (between $-0.3$ and $+0.3$), possibly implying that Source A is slightly extended. dolphot was run with standard parameters on both images in each filter simultaneously and detects both sources. We find $F450W=23.60\pm 0.14$, $F606W=23.72\pm 0.08$, and $F814W=23.25\pm 0.14\,\mathrm{mag}$ for A and $F450W\ =24.52\pm 0.37$, $F606W=24.57\pm 0.15$, and $F814W\ =24.13\pm 0.29\,\mathrm{mag}$ for B. Sharpness and χ² are consistent with a single point source for both. For Source A, we adopt the magnitudes from HSTPhot, as this is a fully optimized tool specifically developed for WFPC2. We find different values for the progenitor photometry to Kilpatrick et al. (2016) (after converting their photometry from STMAG to VEGAMAG: $F405W=23.42$, $F606W=23.10$, $F814W=23.32\,\mathrm{mag}$), and so we performed additional checks. The progenitor is in the Hubble Source Catalog²⁴ with magnitude $F450W=23.85\pm 0.08\,\mathrm{mag}$ (converted to VEGAMAG from ABmag) and $F606W=23.34\pm 0.05\,\mathrm{mag}$, which is marginally brighter than the $23.56\pm 0.10\,\mathrm{mag}$ we find. We also compared the $F606W$ magnitudes of a set of sources on the WF4 chip in the HSC and from our own measurements with HSTPhot, and found no systematic offset between the two.

In Figure 5(a), we show the positions of A and B in a color–color diagram, with respect to the theoretical locus of supergiants (SGs) derived using synphot from ATLAS9 models (Castelli & Kurucz 2004): while their colors are slightly offset from the models, they are broadly consistent with a yellow or blue supergiant.

Zoom In Zoom Out Reset image size

We did not correct the magnitudes for foreground Galactic (${A}_{V}=0.05\,\mathrm{mag}$) or host extinction, since it is small compared to the uncertainty on distance, and will have minimal effect on the bolometric correction (see also Figure 5(a)). In addition, we cannot exclude the presence of significant circumstellar dust around the progenitor, which would then be destroyed during the shock breakout (e.g., Fraser et al. 2012).

The large uncertainty on the distance of NGC 613 plays an important role in the identification of a plausible progenitor for SN 2016gkg. A variety of distances are in the literature, spanning from ∼19.6 to $\sim 31.7\,\mathrm{Mpc}$ ($\mu =31.46\pm 0.80$ and $\mu =32.51\pm 0.47\,\mathrm{mag}$, respectively). In Figure 5(b), we show the position of A in the Hertzsprung–Russell diagram with respect to a set of evolutionary tracks computed with the Cambridge stars models (Eldridge & Tout 2004). Adopting a distance modulus $\mu =32.11\pm 0.38\,\mathrm{mag}$, a solar metallicity, the recommended values of temperature and surface gravity for F5I stars (${T}_{\mathrm{eff}}=6900\,{\rm{K}}$ and $\mathrm{log}g=+1.44\,\mathrm{dex}$; Schmidt-Kaler 1982), and assuming a $0\,\mathrm{mag}$ bolometric correction (see, e.g., Girardi et al. 2008), we obtain ${M}_{\mathrm{bol}}=-8.55\pm 0.40,\mathrm{mag}$. Although A and B sit close to the 26 M${}_{\odot }$ and 15 M${}_{\odot }$ tracks, respectively, none of them is a plausible SN progenitor, since they correspond to stars during the He core burning phase (see Smartt et al. 2009). Alternatively, the yellow color could be explained by additional stripping experienced by the progenitor from a stellar companion (see also Kilpatrick et al. 2016).

Hence, we compare their luminosities with the endpoint of the evolutionary tracks (dashed lines in Figure 5(b)). Source A has $\mathrm{log}L/{L}_{\odot }=5.32\,\mathrm{dex}$, corresponding to the $19-20$ M${}_{\odot }$ tracks, while B falls in the region where the $12-13$ M${}_{\odot }$ tracks end, with $\mathrm{log}L/{L}_{\odot }=4.91\,\mathrm{dex}$. These luminosity and temperature estimates give radii of $\sim 320$ R${}_{\odot }$ and 200 R${}_{\odot }$ for A and B, respectively. If A is the progenitor star of SN 2016gkg, this would imply a yellow SG more luminous than the progenitor of SN 1987A (White & Malin 1987; Walborn et al. 1989). On the other hand, adopting $\mu =31.46\pm 0.80$ ($D=19.6\,\mathrm{Mpc}$) for NGC 613, we obtain ${M}_{\mathrm{bol},A}=-7.90\pm 0.81\,\mathrm{mag}$ and ${M}_{\mathrm{bol},B}=-6.89\ \pm 0.81\,\mathrm{mag}$, corresponding to $\mathrm{log}{L}_{A}/{L}_{\odot }\simeq 5.06\,\mathrm{dex}$ and $\mathrm{log}{L}_{B}/{L}_{\odot }\simeq 4.65\,\mathrm{dex}$. This would imply lower masses and radii for both Sources A and B ($\sim 15$ M${}_{\odot }$, with $R\simeq 240$ R${}_{\odot }$ and $\sim 9$ M${}_{\odot }$, with $R\simeq 150$ R${}_{\odot }$, respectively). This analysis suggests a larger radius for the progenitor of SN 2016gkg, arguing against the predictions of the Rabinak & Waxman (2011) model.

Given the large uncertainties, we have no arguments to favor a particular distance for NGC 613, and we cannot exclude any of the mentioned values of the mass for the candidate progenitor. Also, given the SN position in the HST images, we lack the astrometric accuracy to exclude either of the two sources as a progenitor for SN 2016gkg, rule out the presence of a stellar companion (see Kilpatrick et al. 2016) or even exclude the possibility that the sources we have discussed are in fact star clusters. An improved distance to NGC 613 will address the first issue, while late-time imaging will clarify direct progenitor constraints.

The spectra of SN 2016gkg closely resemble those of other Type IIb SNe, in particular SNe 1993J and 2011dh. He i lines, the classical features for this class of transients, are visible since $+8.82\,\mathrm{days}$, becoming stronger with time.

Our photometric analysis reveals double-peaked light curves in all optical and UV bands, with a relatively fast decline in $g$- and $B$- bands, faster than in SN 1993J, but comparable to SN 2011dh.

Deep HST pre-explosion images, and the temperature evolution within the first $\sim 5\,\mathrm{days}$ from explosion, help to constrain the physical properties of the progenitor. Following Rabinak & Waxman (2011), we obtain a radius consistent with a $\sim 48-124$ R${}_{\odot }$ supergiant. The analysis of the archival HST images revealed the presence of two sources at the position of SN 2016gkg (Sources A and B). Source A seems to correspond to a yellow supergiant star with an initial mass and radius in the range $15\mbox{--}20$ M${}_{\odot }$  and $240\mbox{--}320$ R${}_{\odot }$, while B corresponds to a $9\mbox{--}13$ M${}_{\odot }$ star with $R\simeq 150\mbox{--}200$ R${}_{\odot }$. These radii are larger than those obtained adopting the Rabinak & Waxman (2011) formalism. Like in SN 2011dh, this might be due to the progenitor structure required to produce the double-peaked light curve observed in Type IIb SNe (i.e., the presence of an extended, $R\simeq {10}^{13}\,\mathrm{cm}$, low-mass, $M\leqslant 0.1$ M${}_{\odot }$ envelope becoming transparent after a few days of expansion; see Bersten et al. 2012; Nakar & Piro 2014).

Given the large uncertainties on the distance of NGC 613, and the unfavorable position in the HST images, we cannot rule out either of the two sources as the possible progenitor star, as well as the presence of a stellar companion. Deep late-time post-explosion HST images centered on the position of SN 2016gkg will provide the final word on the nature of the progenitor system.

Based on data from: ESO as part of PESSTO (197.D-1075.191.D-0935) and under programme 097.D-0762(A). Gemini Observatory under GS-2016B-Q-22 (PI: Sand). The Nordic Optical Telescope of the Instituto de Astrofísica de Canarias. The European Organisation for Astronomical Research in the southern hemisphere under ESO programme 198.A-0915(A). FOSCGUI is a graphic user interface aimed at extracting SN spectroscopy and photometry obtained with FOSC-like instruments, developed by E. Cappellaro. D.J.S. acknowledges NSF grants AST-1412504 and AST-1517649. S.J.S. acknowledges EU/FP7-ERC grant 29122. M.F. acknowledges the support of a Royal Society—Science Foundation Ireland University Research Fellowship. L.G. was supported in part by the US National Science Foundation under grant AST-1311862. M.S. acknowledges support from EU/FP7-ERC grant number 615929. K.M. acknowledges support from STFC through an Ernest Rutherford Fellowship. D.A.H., C.M., and G.H. are funded by NSF AST-1313484. Support for I.A. was provided by NASA through the Einstein Fellowship Program, grant PF6-170148. N.E.R. acknowledges financial support by the 1994 PRIN-INAF 2014 (project "Transient universe: unveiling new types of stellar explosions with PESSTO") and by MIUR PRIN 2010-2011, "The dark universe and the cosmic evolution of baryons: from current surveys to Euclid." M.S. acknowledges support from the Danish Agency for Science and Technology and Innovation via a Sapere Aude Level 2 grant, the Villum foundation, and the Instrumentcenter for Danish Astrophysics (IDA).