On the Origin of SN 2016hil—A Type II Supernova in the Remote Outskirts of an Elliptical Host

SN 2016hil was detected using the 48 inch telescope at Palomar Observatory (P48), on 2016 October 22 (UT dates are used throughout this paper) at 07:55 (JD 2457683.801), in the r band (20.29 ± 0.12 mag) and was observed in the g band (20.34 ± 0.13 mag) 40 minutes later. The source was at α = 01^h10^m2475, $\delta =+14^\circ 12^{\prime} 15\buildrel{\prime\prime}\over{.} 5$ (J2000). The last non-detection was 9 days before the explosion down to a 3σ limit of 20.89 mag in the r band, although an earlier marginal 3σ detection in the r band was later identified (see Section 2.3).

In Figure 1, we present the unusual location of the event: on the outskirts of the elliptical galaxy SDSS J011024.51+141238.7. This galaxy is observed at a redshift z = 0.06079, consistent with the redshift derived from the Hα emission line in the spectra of SN 2016hil (see Section 2.2).

Zoom In Zoom Out Reset image size

We collected four optical spectra during a period of 40 days when SN 2016hil was visible. On 2016 October 26, the first spectrum of the SN was obtained using the Double Beam Spectrograph (DBSP; Oke & Gunn 1982) mounted on the Palomar 200 inch Hale telescope (P200). The gratings of 600/4000 and 316/7500 were used for the blue and red cameras, respectively, with the D55 dichroic. The data were reduced using standard procedures, including bias and flatfield corrections, one-dimensional spectral extraction, wavelength calibration with comparison lamps, and flux calibration using observations of standard stars observed during the same night and at approximately similar airmasses to the SN.

Three additional spectra were obtained with the Low Resolution Imaging Spectrometer (LRIS; Oke et al. 1995) on the 10 m Keck I telescope. The gratings of 300/3400 and 300/8500 were used for the blue and red cameras, respectively, with the 560 dichroic. The data were reduced using the LRIS automated reduction pipeline (LPipe) (Perley 2019), and are made available to the public on WISeREP (Yaron & Gal-Yam 2012). In Table 1 we report our spectral observation log and in Figure 2 we present the full set of spectra. They exhibit broad hydrogen emission features that evolve rapidly throughout the observation period, based on which SN 2016hil is classified as an SN II.

Zoom In Zoom Out Reset image size

After the detection, follow-up observations were made using the Spectral Energy Distribution Machine (SEDM; Blagorodnova et al. 2018) mounted at the 60 inch telescope at Palomar Observatory (P60), in addition to routine monitoring with the P48. Photometry was acquired with the SDSS g and Mould-R bands for the P48 images, and with the SDSS gri bands for the P60 images. Mould-R was then converted to the SDSS r band using the Lupton color equations (2005).¹⁰ Since SN 2016hil is located on a simple background, we chose to use aperture photometry in order to extract source fluxes. This was done by designing custom apertures and annuli with the MATLAB Astronomy & Astrophysics Toolbox¹¹ (Ofek 2014). We calibrated zero-points with SDSS stars. Using the images from the days previous to first detection, we summed the non-detection fluxes and derived summed non-detection limits to constrain the shape of the light curve before peak brightness (not including the flux from the marginal 3σ detection at t = −10 days). We repeated this procedures for the epochs after rebrightening observed at t ≈ 32 days, when poor weather conditions at Palomar prevented further photometric observations.

Moreover, we obtained approximate photometry synthesized from the Keck/LRIS spectrum taken 37 days after the first detection. To acquire some estimate for the systematic error involved in synthetic photometry, we compared the scatter of the synthetic photometry acquired from the earlier spectra to the linear interpolation of the light curve in the relevant filter. For the i-band filter for which no such data exist, we took the error to be the mean of the uncertainty in the gr bands.

We corrected for Galactic extinction using the NASA/IPAC Extragalactic Database¹² (NED), which cites a value of ${A}_{V}=0.121$ mag for this line of sight based on Schlafly & Finkbeiner (2011).

S-corrections (Stritzinger et al. 2002) were estimated for the appropriate filters at the times of the spectra, and by then linearly interpolating the trend for different epochs. This became significant (up to ∼0.3 mag) for the P60 r-band photometry since at the redshift of SN 2016hil the evolving Hα feature is at the boundary of the filter (see Figure 2). Absolute magnitude light curves are thus plotted separately.

Table 2 reports the measured gri magnitudes for the Palomar data, as well as the late-time photometry from Keck in Section 2.4. The gri S-corrected light curve is presented in Figure 3. Photometry is made available on WISeREP. Although the S/N is low, we suggest that the light curve of SN 2016hil has a double peak, which can also be corroborated by the lower panel of Figure 3.

Zoom In Zoom Out Reset image size

On the nights of 2017 June 24 (t = 246 days) and 2018 December 1 (t = 771 days) we obtained simultaneous r and g photometry of SN 2016hil with Keck/LRIS. The 2017 June data consist of four dithered exposures totaling 1290 s in g and 1200 s in r. The 2018 December data consist of eight dithered exposures totaling 2598 s in g and 2400 s in r. These data were processed following standard techniques for CCD reductions using LPipe.

In order to eliminate contamination by residual light from the nearby galaxy and from surrounding sources, aperture photometry was performed manually: background and background noise were estimated by establishing an elliptical contour of the host and extending it to reach the location of the event. A series of custom apertures (with a radius of 127) were then constructed along this contour, and the background flux was measured with adjustments for any additional flux gradient. The manual measurements were performed in SAOImageDS9 (Joye & Mandel 2003). The photometric zero-points were acquired using unsaturated stars in the field and by comparing them to the converted SDSS catalog filters as discussed in Section 2.3. Extinction was treated as discussed in Section 2.3, and no S-corrections were applied. In the first epoch, there were faint and marginally significant detections of the transient in r and g separately. In order to boost the significance of the detection, r and g images were summed, after manual cross-astrometry was performed using the Graphical Astronomy and Image Analysis Tool (GAIA; Draper et al. 2014). This resulted in a >3σ detection in the summed image. In Figure 4 a comparison between both epochs in the synthetic R+g band is made, demonstrating the presence of a transient in the first epoch and its absence in the later epoch.

Zoom In Zoom Out Reset image size

SN 2016hil has peculiar photometric properties for a spectroscopically regular SN II. These usually present a plateau light curve (IIP) or a linearly declining light curve (IIL). The light curve of SN 2016hil is thus unusual, presenting an apparent double peak in the gri bands, as can be corroborated from the lower panel of Figure 3. Although not consistent with a plateau or a linear decline, the photometry is quite noisy. It remains to be seen whether these peculiarities will repeat in similar events in the future. For the rest of the paper, we assume the double peak of the light curve is real. However, none of our main conclusions change if this is not the case.

Unusual for a spectroscopically normal SN II, a double-peaked light curve is more characteristic of Type IIb events (see, e.g., Arcavi 2017 for discussion). The spectroscopic features of SN 2016hil, however, exclude the SN IIb classification, since there are no strong helium signatures and prominent presence of hydrogen persists throughout the spectral evolution. In SNe IIb, double-peaked light curves have been suggested to be the result of a peculiar density structure Bersten et al. (2012). Nakar & Piro (2014) show that such a light curve can be produced by a compact core surrounded by an extended low-mass envelope. In some cases the double-peaked structure is attributed to binary interaction (as shown, for example, by Benvenuto et al. 2013). On the other hand, Sapir & Waxman (2017) claim that such a density structure is not necessary to produce a double peak, which can be produced by a standard progenitor star. In such double-peaked light curves, the first peak is generally thought to be powered by shock cooling, and the second peak is powered by the radioactive decay of ⁵⁶Ni. The double-peaked light curve of SN 2016hil seems to indicate that the event had a progenitor different than the red supergiant expected for a standard SN II (Smartt 2009 and references therein).

We estimated the bolometric light curve of SN 2016hil to see if it is consistent with a radioactively powered light curve, and acquire limits on the corresponding ⁵⁶Ni mass. The bolometric correction was estimated from g magnitudes and the g-r color, using a quadratic fit to the color based on a sample of SNe II as described by Lyman et al. (2014). Since the color evolution was observed to be linear over the entire period of observations, but color was not available for all epochs, we fit a linear trend and used this fit to compute the bolometric correction for all times where a measurement in either g or r was available [including the late-time (t = 37 days) synthetic photometry point]. The bolometric correction as calculated appears in Table 2. Using the bolometric luminosity, the integrated bolometric energy output of the SN is estimated to be (7.9 ± 3.5) × 10⁴⁸ erg.

We assume that all late-time luminosity is due to ⁵⁶Ni decay, and use the following model for ${}^{56}\mathrm{Ni}{\to }^{56}\mathrm{Fe}$ decay (see Katz et al. 2013; Nakar et al. 2016; Wygoda et al. 2019). At early times, all γ-rays produced in the decay are scattered and deposit their energy in the ejecta. At late times, only a fraction ${f}_{\gamma }\approx {t}_{0}^{2}/{t}^{2}$ of the γ-rays deposit their energy in the ejecta, where t₀ is the γ-ray escape time. A common interpolation for the intermediate times is ${f}_{\gamma }\approx (1-{e}^{-{t}_{0}^{2}/{t}^{2}})$, which captures the correct limits at late and early times. Using this, the total energy output produced by ⁵⁶Ni decay is given by

$$\begin{eqnarray*}\begin{array}{lcl}{Q}_{\mathrm{Ni}}(t) & = & \displaystyle \frac{{M}_{{}^{56}\mathrm{Ni}}}{{M}_{\odot }}{f}_{\mathrm{dep}}\cdot (6.45\ {e}^{-\tfrac{t}{8.8\mathrm{days}}}+1.44\ {e}^{-\tfrac{t}{111.3\mathrm{days}}})\\ & & \times {10}^{43}\,\mathrm{erg}\,{{\rm{s}}}^{-1},\end{array}\end{eqnarray*}$$

where f_dep = (0.97f_γ + 0.03) is the total fraction of deposited energy due to the radioactive decay, including the energy deposited by positrons. Using this expression, we can place a lower bound on the total ⁵⁶Ni mass at late times by assuming that all the luminosity at 246 days is due to ${}^{56}\mathrm{Co}{\to }^{56}\mathrm{Fe}$ decay, and that f_γ = 1. This gives a lower bound of ${M}_{{}^{56}\mathrm{Ni}}\geqslant 0.012\,{M}_{\odot }$. Alternatively, we can compute the ${M}_{{}^{56}\mathrm{Ni}}$ for a given t₀.

We can further note that since ${\int }_{0}^{t}{Q}_{\mathrm{Ni}}t^{\prime} {dt}^{\prime} \leqslant {\int }_{0}^{t}{L}_{\mathrm{bol}}t^{\prime} {dt}^{\prime}$ for all times (${\int }_{0}^{t}L(t^{\prime} )t^{\prime} {dt}^{\prime}$ is a conserved quantity, accounting for adiabatic losses), we can place an upper bound on the ⁵⁶Ni mass for a given t₀, using the ⁵⁶Ni mass required to power the entire light curve:

$$\begin{eqnarray*}&&\begin{array}{l}\displaystyle \frac{{M}_{{}^{56}\mathrm{Ni}}}{{M}_{\odot }}\\ \leqslant \displaystyle \frac{{\int }_{0}^{t}{L}_{\mathrm{bol}}t^{\prime} {dt}^{\prime} }{{\int }_{0}^{t}{f}_{\mathrm{dep}}\cdot (6.45\ {e}^{-\tfrac{t^{\prime} }{8.8\mathrm{days}}}+1.44\ {e}^{-\tfrac{t^{\prime} }{111.3\mathrm{days}}})\cdot {10}^{43}\,\mathrm{erg}\,{{\rm{s}}}^{-1}t^{\prime} {dt}^{\prime} }.\end{array}\end{eqnarray*}$$

Here the late-time observations are assumed to have the minimal value of f_dep, (corresponding to the maximal amount of ⁵⁶Ni producing the late-time luminosity), while agreeing with the rest of the light-curve data. This gives an upper limit of ${M}_{{}^{56}\mathrm{Ni}}\leqslant 0.07\,{M}_{\odot }$, above which the ⁵⁶Ni mass as measured from late times will not agree with the integrated luminosity. We note that this upper limit is somewhat dependent on the starting point of the integration, but will not change our results or conclusions significantly. For example, changing the explosion time to 5 days earlier than the first photometry point would increase the upper limit by 30% to 0.09 M_⊙, which is still well within the typical range for SNe II.

In Figure 5 we present the bolometric luminosity plotted together with the two limiting cases for the energy deposition due to ⁵⁶Ni decay ${M}_{{}^{56}\mathrm{Ni}}=0.07\,{M}_{\odot }$, t₀ = 100 days and ${M}_{{}^{56}\mathrm{Ni}}=0.012\,{M}_{\odot }$, ${t}_{0}\to \infty$. We can thus conclude that the late-time photometry of SN 2016hil can provide a ⁵⁶Ni content consistent with the second peak of the light curve being powered by ⁵⁶Ni decay.

Zoom In Zoom Out Reset image size

As can be seen in Figure 2, spectra of SN 2016hil display a strong presence of hydrogen, but few other features were identified. SN 2016hil can thus be classified as a spectroscopically regular SN II. The absorption minima of the P-Cygni profile of the Hα feature correspond to expansion velocities of ∼5000 km ${{\rm{s}}}^{-1}$ throughout the spectral evolution. Across all spectra, this Hα absorption minimum is weak relative to those of other Blamer features. This is more characteristic of an SN IIL than of an SN IIP (see, e.g., Figure 18 in Arcavi 2017). In all spectra, there are no indications of narrow host emission lines, which could serve as indicators of SF. In the spectrum taken 11 days after detection, an unidentified broad emission feature appears near 6300 Å. It is not seen in the spectrum taken 2 days earlier, probably owing to the low S/N. The lack of other features seems to indicate a low metallicity, which is expected from a low-luminosity host galaxy (i.e., according to the mass-Z relation; Tremonti et al. 2004). However, since a metallicity gradient is present in many galaxies (e.g., Sánchez et al. 2014) the low metallicity of the event could also be consistent with the environment in the outskirts of the main host galaxy.

In a sample by Taddia et al. (2016), the strength of the Fe ii λ 5018 feature was used to determine the metallicity according to the method of Dessart et al. (2014). In Figure 6 we put the spectra of SN 2016hil in context with such SNe, including PTF10gxi and PTF12ftc, for which the metallicity was determined to be Z = 0.4Z_⊙. The fact that the Fe ii λ 5018 feature is visible in the spectra of both SNe, and not in any of the spectra of SN 2016hil, suggests that it has a similar or lower metallicity content (e.g., Anderson et al. 2016, 2018).

Zoom In Zoom Out Reset image size

The continua of the spectra were fitted to blackbody emission. This was done by iteratively fitting a continuum, subtracting it, removing outliers, and refitting the remaining data, until the temperature converges. In all spectra, the temperature was found to be close to 7000 K, without a clear trend in time. Uncertainties were estimated using 68% confidence bounds, not accounting for systematic errors. The fitted temperatures and their corresponding uncertainties appear in Table 3.

Identifying the host of SN 2016hil with certainty is crucial for putting this event in context. Our initial association of SN 2016hil with the galaxy SDSS J011024.51+141238.7 is primarily due to SN 2016hil having a redshift consistent with that of the nearby galaxy. We compared the host spectrum, acquired from the SDSS Science Archive Server (SAS), to templates of various galaxy types (Kinney et al. 1996). It is most consistent with being an elliptical galaxy.

To put this host in context of the general population of host galaxies of SNe II, we compare its photometric properties to the host galaxies of the (i)PTF CCSN sample (S. Schulze et al. 2019, in preparation). This homogeneous sample consists of 503 SNe II, detected between the beginning of 2009 and the beginning of 2017. We retrieved archival images of the host galaxy from Galaxy Evolution Explorer (GALEX) Data Release (DR) 8/9 (Martin et al. 2005), SDSS DR9 (Ahn et al. 2012), PS1 DR1 (Chambers et al. 2016), the Two-Micron All Sky Survey (2MASS; Skrutskie et al. 2006), and the unWISE (Lang 2014) images from the NEOWISE (Meisner et al. 2017) Reactivation Year 3. Furthermore, we use the matched-aperture photometry software package Lambda Adaptive Multi-Band Deblending Algorithm in R (LAMBDAR; Wright et al. 2016) that is based on a photometry software package developed by Bourne et al. (2012) and tools which will be presented by S. Schulze et al. (2019, in preparation). The photometry was either calibrated against zero-points (GALEX, PS1, SDSS, and NeoWISE) or against a set of stars (2MASS). The resulting photometry is summarized in Table 4.

As for the (i)PTF CCSN host-galaxy sample, we model the spectral energy distribution (SED) of the host with the software package LePhare¹³ version 2.2 (Arnouts et al. 1999; Ilbert et al. 2006) and standard assumptions (Bruzual & Charlot 2003 stellar population-synthesis models with the Chabrier initial mass function Chabrier (2003), an exponentially declining star formation history and the Calzetti et al. (2000) attenuation curve).

Figure 7 shows the observed SED. It is best described by a galaxy, dominated by an old stellar population, with a large stellar mass content of $\mathrm{log}(M/{M}_{\odot })={10.88}_{-0.07}^{+0.56}$ and a low SF rate (SFR) of $\mathrm{SFR}={0.20}_{-0.07}^{+0.49}\ {M}_{\odot }\ \,{\mathrm{yr}}^{-1}$. The age of the stellar population and the large mass corroborate the conclusion from the SDSS spectrum that this is indeed an elliptical galaxy. The low but non-negligible SFR is not in conflict with this interpretation. Schawinski et al. (2007) showed that ∼30% of a volume-limited sample of luminous E galaxies exhibited signs of recent SF. Furthermore, such SF is expected to trace the stellar population of the galaxy, and is not likely to extend to the outskirts.

Zoom In Zoom Out Reset image size

Another option could be that SN 2016hil occurred in a faint satellite of the main host, where there is still SF activity. As can be seen in Figure 4, the relatively deep Keck/LRIS images reveal no obvious dwarf galaxy or star-forming region at the location of SN 2016hil. Using the low-S/N (2σ) flux detected in the t = 771 days epoch in the r and g bands, we attempt to constrain the galaxy mass and SFR of a possible dwarf satellite host. We repeated the SED fitting process using the r and g photometry. The results constrain the presence of a potential dwarf host such that $\mathrm{log}(M/{M}_{\odot })={7.27}_{-0.24}^{+0.43}$, and SFR $\,\leqslant \,0.01\ {M}_{\odot }\ {\mathrm{yr}}^{-1}$. As the SED is based only on g and r photometry, the mass estimate should be regarded as an upper limit. The SFR upper limit is corroborated using a limit derived from the GALEX/MIS non-detection in the FUV band, according to the procedure outlined in Kennicutt (1998). Assuming the limiting magnitude of GALEX/MIS (∼23.5 mag), we acquire a limit of SFR $\leqslant \,0.022\ {M}_{\odot }\ {\mathrm{yr}}^{-1}$, in agreement with the modeling results.

To put both host-galaxy candidates in the context of the general population of SN II host galaxies, we compare their masses and absolute magnitudes to those of the SN II hosts from the (i)PTF survey (Figure 8; values taken from Schulze et al. 2019, in preparation). Both candidate host galaxies have extreme values for an SN II host. The elliptical galaxy is among the most luminous and the most massive galaxies in the sample. At the other extreme, the potential dwarf galaxy cospatial with the SN site would be the least luminous host in the SN II (i)PTF sample with M_g = −10.1 ± 0.73 mag, at least 2 mag fainter than the next faintest host. Combined with the mass limit of ${10}^{7.3}\,{M}_{\odot }$, this puts this object at the very low end of mass and luminosity functions of star-forming galaxies.

Zoom In Zoom Out Reset image size