OGLE-2013-SN-079: A LONELY SUPERNOVA CONSISTENT WITH A HELIUM SHELL DETONATION

On October 16.21 ut, we took PESSTO spectra of the two elliptical galaxies, probably interacting, at coordinates α = 00^h35^m0447, δ = −67^o41'027 (triangle in Figure 1) and α = 00^h35^m0521, δ = −67^o41'147 (square in Figure 1), finding redshifts z = 0.074 and z = 0.072, respectively. The second is also listed in the NASA/IPAC Extragalactic Database as 2MASXJ00350521-6741147 at redshift z = 0.071. If one of the two galaxies is the host, the transient would be at a projected distance of ∼49.9 kpc (∼35'') from the first galaxy, or ∼40.2 kpc (∼30'') from the second.

Zoom In Zoom Out Reset image size

In our deepest images (VrI), we performed aperture photometry on the possible host (the closest). We found that the radius containing half of the light of the galaxy is R_1/2 ∼ 28 kpc, similar to the value R_1/2 ∼ 30 kpc measured using the V magnitude of the galaxy and a Sérsic index of n = 4 (Graham 2013, and reference therein). Thus OGLE13-079 is likely located in the extreme outskirts of the host, in the halo region. Based on this identification, we adopt z = 0.07 as the redshift for OGLE13-079 throughout the following.

The host galaxy type and the remote location of OGLE13-079 would argue against a massive star origin of this explosion (Hamuy et al. 1996; Anderson & James 2009). The offset distribution of 520 SNe published by Kasliwal et al. (2012) supports this idea. Massive stars could reach large distances as a consequence of tidal stripping during galaxy interaction but they should then be in the intracluster environment of tidal tails. We do not observe such tails in our data.

We note that the host and location of OGLE13-079 are different from the bright and fast objects SNe 2002bj and 2010X, which both occurred close to the galaxy nuclei (Poznanski et al. 2010; Kasliwal et al. 2010). On the other hand, Ca-rich objects like SN2005E were found far from the nucleus (Kasliwal et al. 2012; Yuan et al. 2013; Lyman et al. 2014). OGLE13-079 shows the largest projected distance from its host nucleus among these types of objects and it is comparable only to PTF09dav (Sullivan et al. 2011).

Optical and near infrared (NIR) images were reduced (trimmed, bias subtracted, and flat-fielded) using the PESSTO (Smartt et al. 2014), the OGLE (Wyrzykowski et al. 2014), and Faulkes Telescope pipelines. Photometric zero-points and color terms were computed using observations of standard fields (VI in Vega and grz in AB system). We then calibrated the magnitudes of a local stellar sequence shown in Figure 1. The average magnitudes of the local-sequence stars were used to calibrate the photometric zero-points in non-photometric nights. The JHK photometry was calibrated to the Two Micron All Sky Survey system using the same local sequence stars. Our optical and NIR photometric measurements were performed using the point-spread function (PSF) fitting technique. Differences between passbands were taken into account (applying the S-correction; Stritzinger et al. 2002; Pignata et al. 2004).

All spectra (Table 2) were reduced and calibrated in the standard fashion (including trimming, overscan, bias correction, and flat-fielding) using standard routines within iraf. The final flux calibration was checked by comparing the integrated spectral flux, transmitted through standard Sloan or Bessell filters, with our photometry. We applied a multiplicative factor when necessary, and the resulting flux calibration is accurate to within 0.2 mag.

OGLE13-079 was observed during the rising phase only in the I-band but subsequently followed in a range of optical filters until it disappeared beyond our detection threshold in late November (Figure 2). The non-detection four days before the first observation places a strong constraint on the explosion epoch, allowing us to define the rise time and the initial shape of the light curve. We estimate the explosion to have occurred at MJD 56553.5 ± 2, hence ∼11 days before I-band maximum. The post-peak decline has the same shape in each band with the exception of the g band, in which OGLE13-079 fades by 1.5 mag over five days, in contrast to an average decrease of 0.48 mag in the other bands. The slope in the VrIz bands is fairly constant until the last detection, with the exception of a shoulder in the I band at ∼15 days after peak that is reminiscent of the secondary maximum in type Ia (Kasen & Plewa 2007).

Zoom In Zoom Out Reset image size

The OGLE13-079 magnitudes are comparable to SN2010X but ∼1.4 mag fainter than SN2002bj and ∼1.4 mag brighter than SN2005E. Indeed, it is brighter than the most luminous (M_R = −16.4 mag for PTF09dav, Kasliwal et al. 2012) of the Ca-rich transients. However, the post-peak decline of OGLE13-079 of Δm₁₅(r) ∼ 1.2 is similar to that of SN2005E (Δm₁₅(R) ∼ 1.2) and slower than Δm₁₅(R) ∼ 2.4 and ∼2.3 of SN2002bj and SN2010X, respectively. Thus, while the OGLE13-079 r-band peak magnitude is similar to those of the fast evolving objects, the decline is comparable with the slower ones (Figure 2).

To compare OGLE13-079 data with theoretical predictions from explosion models (see Section 3.3), a bolometric light curve of OGLE13-079 was constructed using similar methods to Inserra et al. (2013a, 2013b) and is shown in Figure 2. We initially built a gVrIz pseudo-bolometric light curve and added a u contribution by assuming u − g ≈ 0, which is consistent with the synthetic photometry retrieved from our early spectra. We added a NIR contribution assuming V − J ≈ 1.8 to be constant throughout the evolution and adopting NIR colors (J − H and J − K) similar to the average of SNe Ia (Contreras et al. 2010; Friedman et al. 2014). Clearly, the light curve constructed via these steps has substantial uncertainties, therefore, we also construct a robust lower limit on the bolometric light curve by integrating only over the measured flux in gVrIz. We placed an upper limit on the bolometric light curve by using the best-fit blackbody curve of the spectral energy distribution retrieved through the red optical bands (rIz) by integrating from near ultraviolet to K-band.

The peak luminosity L_bol ≈ 1.4 × 10⁴² erg s⁻¹ is roughly a factor ten and two less than in normal and faint (SN1991bg-like) SNe Ia, respectively.

We compared our photometric and spectroscopic data with those predicted for the detonation of a 0.20 M_☉ He layer around a 0.60 M_☉ CO core presented by Shen et al. (2010, hereafter 0.6+0.2 M_☉). We also compare to angle-averaged predictions from 2D models computed by Sim et al. (2012): specifically, we chose one of their helium shell detonations (HeD-S, which has a 0.59 M_☉ CO core plus 0.21 M_☉ He layer) and one of their double detonations (CSDD-L, which has a 0.59 M_☉ CO core plus 0.21 M_☉ He layer). These were chosen because they give the best match to OGLE13-079 absolute magnitude.

Of the models we considered, the best overall fit to the data is found with the CSDD-L model. However, the match is not perfect and, with the exception of the I band, the model is always too faint. We note that a good match of the I-band evolution is also achieved by the HeD-S model but this model has a brighter and wider peak than the data or the CSDD-L model. The 0.6+0.2 M_☉ model of Shen et al. is fainter and has a faster decline compared to the HeD model, ∼8-10d, depending on the band. This is likely a consequence of differences in the model ejecta structure and ionization state, which lead to differing degrees of line blanketing in the blue and reprocessing of the UV/blue light by heavy elements. The 0.6+0.2 M_☉ model does produce an I-band shoulder, although ∼10 days earlier than observed. We note that the rise time of OGLE13-079 is similar to those predicted by the models considered, while the "late" (>15d) decline has a comparable slope to the Sim et al. the models. It is also noticeable that SNe 2002bj and 2010X are too rapidly fading compared to the models, while SN2005E is too dim.

The best match of the bolometric light curve is also found with the CSDD-L model, which is able to adequately fit the data from −6d to 16d from maximum. After that, OGLE13-079 declines more slowly for the next 10 days (15 ≲ d ≲ 25) but then settles onto a decline slope similar to the model. The HeD-S model is between the limits of the bolometric light curves (gray area in Figure 2), but the fit generally appears poorer both around peak and at later times compared to the CSDD-L model. The Shen et al. model also fits the data reasonably well, although in the bolometric light curve of OGLE13-079 we do not observe any double-peaked behavior. Rise time and post maximum evolution differences between data and models are roughly similar to the single bands previously shown.

In the top right panel of Figure 2 we compare the V − R color evolution of OGLE13-079 with those predicted by the models. Since we did not have R-band observations, we transformed r to R magnitude using snake (a python code for K-correction and magnitude conversion; C. Inserra, in preparation). Although the light curves are more similar to the CSDD-L model (see above), the V − R behavior is closer to the HeD-S model with a V − R ∼ 1.1 after 10 days since maximum. We note that the V − R evolution is also comparable to CSDD-L model but shifted by ∼0.3 to bluer values. Both the Sim et al. and Shen et al. models of pure helium detonation have similar behavior, although the Sim et al. models are systematically redder.¹⁹

In Figure 3 we show OGLE13-079 with three other well studied transients of similar luminosity, SNe 2002bj, 2005E, and 2010X, at similar epochs. The spectra of SN2002bj are quite different from those of OGLE13-079: SN2002bj has a much bluer continuum and does not show the strong Ti lines (around 5300 Å and 5900 Å). SN2005E has a similar red color to OGLE13-079  but also does not show the two prominent Ti lines. In contrast, SN2005E shows a strong Ca ii NIR triplet feature, forbidden [Ca ii] 7291,7323 Å, and unambiguous He i lines. These led Perets et al. (2010) to identify the object as a Type Ib and a Ca-rich transient. We do not see forbidden [Ca ii] or any He i lines in OGLE13-079 spectra at comparable phases. OGLE13-079 has a similar color to SN2010X but the spectral features are again different. The spectra are particularly distinct between 5500 Å and 7500 Å where Na i, O i and Mg i are not present in OGLE13-079. SN2010X also shows a prominent Ca ii NIR triplet, which is similar in strength to those of other Ca-rich events but is not visible in the OGLE13-079 spectrum with comparable intensity. We note that the OGLE13-079 Ca ii NIR triplet velocity is ∼4000 km/s slower than those of SNe 2005E and 2010X in the same phase. We conclude that OGLE13-079 does not closely resemble any of these objects in its spectral features or colors.

In Figure 3 we compare the OGLE13-079 spectrum with Sim et al. and Shen et al. models at similar epochs (∼18 days from explosion). The flux computed from the model spectra has been scaled to match that of OGLE13-079. All lines and their absorption and emission strength are well reproduced by the Shen et al. 0.6+0.2 M_☉ model. The model is slightly bluer, as also highlighted by the color evolution. The CSDD-L model does a reasonably good job in reproducing the spectrum, although we note that it does not provide a good match to the two noteworthy lines at 5300 Å and 5900 Å, which we have attributed to Ti ii. These differences between the models could be due to the distribution of Fe-group elements in the ejecta, which are key to shaping the spectra (see Figure 6 of Sim et al. 2012), and generally to the treatment of ionization. The comparison between the 0.6+0.2 and HeD-S models (similar masses but different densities) shows that the Sim et al. models fit the color better but the Shen et al. model fit the Ti ii lines quite well. The existing models do not predict strong He lines, however, we note that further studies that include treatments of non-thermal excitation are required to better investigate the formation of He lines in the optical (Hachinger et al. 2012). We note that the majority of model lines are either not observed or have different strength (e.g., Ca ii) in the other transients mentioned in Section 4.1. Although OGLE13-079 light curves could, in principle, be fitted with a massive star explosion it would be difficult to reproduce an oxygen-free and titanium-dominated spectral sequence.