Observations of a Fast-expanding and UV-bright Type Ia Supernova SN 2013gs

Our photometric observations of SN 2013gs began on 2013 November 30 (one day after discovery) and lasted for about four months. The optical photometry is mainly obtained with the 0.8 m Tsinghua-NAOC reflecting Telescope (TNT¹⁴ ) located at the NAOC Xinglong Observatory. A 1340 × 1300 pixel back-illuminated CCD, with a field of view of 115 × 112 (pixel size ∼052 pixel⁻¹) is mounted on the Cassegrain focus of the telescope (Huang et al. 2012). Some data were also obtained with the Yunnan Faint Object Spectrograph and Camera (YFOSC) mounted on the Li-Jiang 2.4 m telescope of Yunnan Astronomical Observatories (LJT), the 2.0 m Liverpool Telescope (LT)+IO:O and the Asiago Faint Object Spectrograph & Camera (AFOSC) on the 1.82 m Copernico telescope.¹⁵ The UV observations were triggered immediately after the discovery, with a cadence of 2 days, using the Ultra-Violet/Optical Telescope (UVOT; Roming et al. 2005) on the Swift spacecraft (Gehrels et al. 2004). The UV photometry was taken from the Swift Optical/Ultraviolet Supernova Archive¹⁶ (Brown et al. 2014). The reduction is based on that of Brown et al. (2009), including subtraction of the host galaxy count rates and uses the revised UV zero-points and time-dependent sensitivity from Breeveld et al. (2011).

As SN 2013gs is located near the center of UGC 5066, the contamination of the host galaxy cannot be neglected when measuring the SN flux. Therefore, image subtraction was applied before performing photometric measurements with template images taken about one year after the explosion. The SN image is first registered to the template image, and the former is then scaled using the foreground stars to the same point-spread function as the latter. The TNT instrumental magnitudes of the SN and local standard stars are finally obtained with an ad hoc pipeline adapted for the TNT data (based on the IRAF¹⁷ DAOPHOT package; Stetson 1987). The LT and AFOSC data were obtained using a pyhton-based pipeline¹⁸ (Cappellaro et al. 2015).

The UBVRI instrumental magnitudes of TNT and YFOSC were transformed to the standard Johnson UBV (Johnson et al. 1966) and Kron–Cousins RI (Cousins 1981) systems by observing a series of standard stars (Landolt 1992) on photometric nights. The BV instrumental magnitudes of LT and AFOSC are calibrated with the Landolt standard stars obtained in eight photometric nights and the URI magnitudes are converted from the Sloan catalogs using the relation of Chonis & Gaskell (2008). The calibrated magnitudes of local standard stars are listed in Table 1. The final calibrated magnitude of SN 2013gs is transformed from the instrumental magnitudes by these local standard stars and presented in Table 2. The error bars account for the uncertainties due to photon noise and photometric system calibration.

A total of 22 low-resolution optical spectra were obtained with different telescopes and instruments including the AFOSC on the 1.82 m Copernico telescope, Cassegrain spectrograph and BAO Faint Object Spectrograph and Camera mounted on the 2.16 m telescope at Xinglong Observatory and YFOSC on LJT of Yunnan Astronomical Observatories. The journal of spectroscopic observations is listed in Table 3. In order to reduce the contamination from the host galaxy, the local flux next to SN 2013gs was used as background to carefully extract the SN flux, which was calibrated with standard stars observed on the same night at similar air masses as the SN. The spectra were corrected for the atmospheric extinction using the extinction curves of local observatories, and telluric lines were removed using spectra of standard stars.

The UBVRI-band light curves of SN 2013gs are shown in Figure 2. These multiband observations were used to derive the light-curve parameters such as peak magnitudes and the luminosity decline rate Δm₁₅(B). From the polynomial fit, we find that SN 2013gs reached the B-band maximum on JD 2,456,640.2 ± 0.2 (2013 December 13.7 UT) with B_max = 15.53 ± 0.02 mag and the post-maximum decline rate Δm₁₅(B) = 1.00 ± 0.05 mag. From the B- and V-band light curves, we derived B_max − V_max = 0.17 ± 0.03 mag, which suggests a non-negligible reddening in the direction of SN 2013gs.

Zoom In Zoom Out Reset image size

The unfiltered light curve obtained by the 0.6 m Schmidt telescope is also reported in Figure 2 and the magnitudes are listed in Table 3. The unfiltered magnitude was calibrated by the R-band magnitude from the Positions and Proper Motions Star Catalog Extended (Röser et al. 2008). The large error bars are mainly attributed to the systematic errors between the unfiltered and R-band magnitudes (Li et al. 2003). The unfiltered light curves reached a peak of 15.01 ± 0.20 mag on JD 2,456,641.7 ± 1.2, about 1.5 days after B_max, while the time of first light will be estimated in Section 3.4.

In Figure 3, we compared the optical light curves of SN 2013gs with other SNe Ia consistent with the SN sample used in the UV comparison, including SNe 2002bo (Δm₁₅(B) = 1.15; Benetti et al. 2004; Krisciunas et al. 2004), 2006X (Δm₁₅(B) = 1.17; Wang et al. 2008), 2007gi (Δm₁₅(B) = 1.31; Zhang et al. 2010), 2009ig (Δm₁₅(B) = 0.89; Foley et al. 2012), and ASASSN-14lp (Δm₁₅(B) = 0.80; Shappee et al. 2016), 2003du (Δm₁₅(B) = 1.02; Stanishev et al. 2007), 2005cf (Δm₁₅(B) = 1.05; Wang et al. 2009b). The B-band light curve of ASASSN-14lp is approximated from the g-band magnitude through the relation of Chonis & Gaskell (2008). The first five SNe Ia have higher photospheric velocities and can be put into the HV subclass, while the latter two belong to the Normal subclass. The light curves of the comparison sample are normalized to the peaks of SN 2013gs. One can see that SN 2013gs and the comparison SNe Ia have similar light-curve shapes near the maximum brightness. In the early phase, SN 2013gs seems to have a rising rate similar to that of SN 2003du but slower than those of SN 2002bo, SN 2005cf, and SN 2006X. This is evident in the UBVI bands but not in the R. In the early nebular phase, the evolution of the light curve becomes more complicated. In the B band, the later-time decline rate β was estimated to be 1.39 ± 0.11 mag (100 days)⁻¹, which is slightly smaller than those of SN 2003du (1.65 ± 0.02 mag (100 days)⁻¹) and SN 2005cf (1.62 ± 0.07) mag (100 days)⁻¹ as shown in Figure 3(b). The decay rates in other bands are also calculated and listed in Table 5. Although the RI-band light curves of SN 2013gs are similar to the comparison SNe Ia, its second bump/shoulder feature (at around t ≈ 25 days) is weaker than the comparison SNe Ia (in particular the HV SNe Ia).

Zoom In Zoom Out Reset image size

The UV and optical light curves of SN 2013gs obtained by Gehrels Swift-UVOT are also shown in Figure 2 and listed in Table 4. SN 2013gs was not detected in the uvm2 band (centered at ∼2200 Å) presumably due to the large reddening bump near 2200 Å (Cardelli et al. 1989). In Figure 4, we compared the UV (uvw1 and uvw2) light curves of SN 2013gs with SN 2005cf (Δm₁₅(B) = 1.05; Pastorello et al. 2007; Wang et al. 2009b), SN 2006X (Δm₁₅(B) = 1.17; Brown et al. 2009), SN 2009ig (Δm₁₅(B) = 0.89; Foley et al. 2012), SN 2011fe (Δm₁₅(B) = 1.18; Brown et al. 2012b; Zhang et al. 2016), SN 2012fr (Δm₁₅(B) = 0.80; Zhang et al. 2014; Contreras et al. 2018), and ASASSN-14lp (Δm₁₅(B) = 0.80; Shappee et al. 2016). The uvw1- and uvw2-band light curves of SN 2013gs span from t = −11.5 days to t = +6.5 days relative to the B-band maximum. The UV light curves of the comparison SNe Ia have been normalized to their B-band maxima and shifted to match the UV peaks of SN 2013gs. The uvw1-band light curve of SN 2013gs shows close resemblance to SN 2009ig around the maximum light. In comparison, SN 2006X shows a slower post-maximum decline rate while SN 2011fe declines more rapidly than other SNe Ia. The uvw2-band light curve of SN 2013gs is similar to that of SN 2009ig and SN 2011fe, though it has larger photometric errors. The UV light curves of SN 2013gs are similar to those of SN 2009ig. The light-curve parameters of other bands are listed in Table 5.

Zoom In Zoom Out Reset image size

The Galactic reddening toward SN 2013gs is A_V^Gal = 0.063 mag from the dust map of Schlegel et al. (1998) and 0.053 mag from Schlafly & Finkbeiner (2011). We adopted the average value of 0.058 mag, corresponding to $E(B-V)$ = 0.019 mag using the standard extinction coefficient R_V = 3.1. The reddening from the host galaxy can be estimated through the relation between Δm₁₅(B) and B_max − V_max (see Phillips et al. 1999; Wang et al. 2009b). The comparison gives a host reddening of E(B − V) = 0.21 mag for SN 2013gs. With the ΔC₁₂ method (Wang et al. 2005), we estimated E(B − V) = 0.22 mag. The average value of E(B − V)_host = 0.22 ± 0.05 mag is adopted for this paper. The interstellar Na i D doublet lines from the host galaxy are not discerned clearly in the low-resolution spectra of SN 2013gs, hence they cannot be used to estimate independently the host galaxy reddening estimate. The extinction at NUV band varied slightly with different R_V. SN 2013gs is not a very HV Ia like 2006X, so we prefer to use R_V = 3.1 to estimate the host galaxy extinction.

Figure 5 presents the reddening-corrected color curves of SN 2013gs. The color curves of SN 2013gs are similar to those of the comparison SNe Ia before and around the maximum light. As shown in Figures 5(a) and (b), SN 2013gs reached its bluest U − B and B − V colors about 7 days before the B-band maximum, while they reach the reddest values about 3–4 weeks after the maximum light. We also note that SN 2002bo, SN 2006X, and SN 2007gi reach their red peaks after maximum about one week earlier than normal SNe Ia (i.e., SN 2003du and SN 2005cf), suggestive of a possible correlation of this difference with ejecta velocity. After t ∼ 1 months, the color curves of those SNe Ia are characterized by more heterogeneous evolution. In particular, SN 2013gs is bluer than the comparison SNe Ia in U − B and B − V, while it is redder in the V − R and V − I colors (Figures 5(c) and (d)). Another interesting feature about SN 2013gs is that its V − R color seems to be bluer than the other SNe Ia at very early phases (i.e., at t < −10 days).

Zoom In Zoom Out Reset image size

The UV color curves of SN 2013gs and the comparison sample are shown in Figure 6. The heterogeneity of the UV − V colors is larger than that of the optical colors, due to the significant change of the photospheric temperature during the explosion and/or a larger line blanketing in the UV domain. The ${uvw}1-V$ color curve has sort of a "V" shape, with the bluest color reached about 5 days before the B-band maximum. Such an "V"-shape feature is also mentioned by Milne et al. (2010) and is similarly seen in the U − B and B − V color curves. We note that SN 2013gs has one of the bluest ${uvw}1-V$ before the maximum light, similar to SN 2011fe. SN 2011fe also shows very strong UV emission (Zhang et al. 2016) and belongs to the "NUV-blue" group of SNe Ia defined by Milne et al. (2013). In ${uvw}1-V$, SN 2013gs is bluer than SN 2009ig by ∼0.2–0.4 mag, though these two SNe Ia have similar shapes in their UV light curves. SN 2013gs has the reddest ${uvw}2-V$ color among all SNe Ia of our sample for the earliest data point. This deviation may be due to the large uncertainties associated with the red-leak corrections (Brown et al. 2010, 2015). Since SN 2013gs can be associated to both the HV and NUV-blue subclasses of SNe Ia, it apparently becomes an exception for the tendency that NUV-blue events belong to a subset of the larger low-velocity gradient group (Milne et al. 2013).

Zoom In Zoom Out Reset image size

The explosion time of SNe Ia can be inferred from the very early rise of its light curves. Some properties of progenitor systems, such as the radius of the exploding star (Piro & Nakar 2013) or binary interaction (Kasen 2010), can also be constrained by the early time evolution of the light curves. The first detection of SN 2013gs was made at t = −13.86 days from the B-band maximum light and an upper limit to ∼19.0 mag was found at about t = −17.80 days before the B-band maximum. Assuming that SNe Ia expand like homogenous fireballs at very early times, the rise time t_r, depending primarily on the opacity, could be calculated by the t² model (L ∝ t²; Arnett 1982; Riess et al. 1999). Using the data from the first detection to about 8 days before the B-band maximum and a t² model, we constrain the best-fit bolometric rise time as 18.72 ± 0.18 days. This rise time is larger than those of SN 2009ig (17.13 ± 0.07 days; Foley et al. 2012), SN 2011fe (17.64 ± 0.01 days; Zhang et al. 2016), SN 2013dy (∼17.8 days; Zheng et al. 2013), ASASSN-14lp (16.94 ± 0.10 days; Shappee et al. 2016), the average Sloan Digital Sky Survey SN Ia (17.38 ± 0.17 days in B band; Hayden et al. 2010), and the average low-redshift SNe Ia (17.44 ± 0.39 days; Strovink 2007). Considering a general form of the fireball model, the index of the power law can be a free parameter (L ∝ tⁿ). For the unfiltered light curve as shown in Figure 2, we obtain a best-fit rise time of 16.89 ± 0.88 days and the corresponding index is 1.47 ± 0.26. This smaller index is consistent with the value given by Piro & Nakar (2013) with 1.5. For the model fit with free index, we obtain a rise time of 16.28 ± 1.25 in U, 15.81 ± 0.91 in B, 17.69 ± 1.36 in V, 17.21 ± 3.25 in R, and 18.76 ± 3.03 days in I, respectively. However, the first-light time derived from the above analysis may have great uncertainty due to the lack of early time photometry.

The host galaxy of SN 2013gs has a distance modulus of m − M = 34.29 ± 0.15 mag (Mould et al. 2000). Adopting the reddening discussed above, the peak absolute magnitudes of SN 2013gs are M_U = −19.62 ± 0.15 mag, M_B = −19.29 ± 0.15 mag, M_V = −19.24 ± 0.15 mag, M_R = −19.17 ± 0.15 mag, and M_I = −18.90 ± 0.15 mag. They are similar to those of typical SNe Ia (e.g., M_V = −19.27 mag from Wang et al. 2006). The UV-band peak absolute magnitudes are also calculated as M_uvw1 = −18.86 ± 0.18 mag and M_uvw2 = −17.87 ± 0.22 mag.

We build the quasi-bolometric light curve of SN 2013gs based on the UV and optical photometry to estimate the peak bolometric luminosity and nickel mass produced in the explosion. The bolometric light curves of SN 2013gs are mostly determined by the optical photometry. The early UV contribution is derived from the Gehrels Swift-UVOT observations. The ratio between UV (∼1500–3000 Å) and optical emission (∼3000–9000 Å) is about 0.15 in the rising phase and declined to 0.12 at around the B-band maximum light larger than those of SN 2005cf and SN 2011fe (∼10%; Wang et al. 2009b; Zhang et al. 2016) at similar phases. We did not observe SN 2013gs in the near-infrared (NIR) bands, and a contribution of ∼5% (derived from SN 2005cf at around the maximum light) is thus applied to the NIR correction of bolometric luminosity. For SN 2013gs, the peak bolometric luminosity is estimated to be 1.82 (±0.16) × 10⁴³ erg s⁻¹, which is comparable to that of other SNe Ia. The error includes uncertainties in distance modulus, photometry, reddening correction, and flux correction beyond the optical window.

Using the Arnett law (Arnett 1982), we can estimate the mass of radioactive ⁵⁶Ni from the maximum-light luminosity using the following relation:

$$\begin{eqnarray}&&{L}_{\max }=(6.45\times {e}^{\tfrac{-{t}_{r}}{8.8}}+1.45\times {e}^{\tfrac{-{t}_{r}}{111.3}})\,{M}_{\mathrm{Ni}}\times {10}^{43}\,\mathrm{erg}\,{{\rm{s}}}^{-1},\end{eqnarray} \tag{ 1 }$$

where t_r is the rise time of the bolometric light curve, and ${M}_{\mathrm{Ni}}$ is the mass of ⁵⁶Ni in unit of solar mass (Stritzinger & Leibundgut 2005). For SN 2013gs, t_r is taken as 16.89 ± 0.88 days from the analysis in the above subsection. Inserting the parameters of t_r and L_max into the above equation, we derive a nickel mass of 0.83 ± 0.10 M_⊙. This value is within the range of normal SNe Ia.

In Figure 8, we compare the spectra of SN 2013gs with well-observed SNe Ia with similar Δm₁₅(B) at representative epochs (t ∼ −10, 0, +1 week and +1 month since the B maximum). The sample includes SN 2002bo (Benetti et al. 2004), SN 2003du (Stanishev et al. 2007), SN 2004dt (Altavilla et al. 2007; Silverman et al. 2012), SN 2005cf (Wang et al. 2009b), and SN 2009ig (Foley et al. 2012). All spectra are corrected for the reddening of the Milky Way and host galaxies.

Zoom In Zoom Out Reset image size

The earliest spectrum of SN 2013gs taken at t ≈ −12 days (Figure 8(a)) shows singly ionized lines of intermediate-mass elements like Si, S, Mg, and Ca. The O i λ7773 is not detected in the spectrum of SN 2013gs at any epoch. Compared to other HV SNe Ia, the absorption lines of Si ii λ6355 and W-shaped S ii features are relatively weak in SN 2013gs. The absorption at about 5800 Å due to Si ii λ5972 is barely visible in SN 2013gs, while it is prominent in SNe 2002bo at comparable phases. The Fe ii/iii absorption (∼5000 Å) is shallower than some HV SNe Ia such as SN 2002bo. SN 2013gs shows a similar Ca NIR triplet to that of SN 2004dt, but the latter has a clear O i absorption, an important spectral diagnostic for constraining the explosion mechanism of SNe Ia (Zhao et al. 2016). At around the maximum brightness (Figure 8(b)), SN 2013gs is quite similar to SN 2002bo, but its spectral features are weaker, especially for the Ca ii NIR triplet. We measure R(Si ii) (Nugent et al. 1995), the line ratio of Si ii λ5972 and Si ii λ6355, to be 0.08 ± 0.02 for SN 2013gs. This smaller value of R(Si ii) suggests that SN 2013gs has a higher photospheric temperature. It is expected that the spectral energy distribution of SN shifts toward shorter wavelengths with higher photospheric temperatures. The higher temperature can also keep the iron-group elements at higher ionization states, which finally results in a lower degree of line blanketing in the UV region (Sauer et al. 2008). This could produce stronger UV emission for SN 2013gs. By one week after maximum light (Figure 8(c)), the W-shaped profile of S ii line almost disappeared in SN 2013gs and other HV SNe Ia, while this feature is still visible in SN 2005cf. At t ∼ 1 month (Figure 8(d)), the spectral profiles of all SNe Ia become quite similar to each other, although the absorptions of the Ca ii NIR triplet still shows some diversity in strength.

Figure 9 shows the expansion velocities of SN 2013gs inferred from the Si ii 6355 and Ca ii NIR triplet absorptions, along with those obtained for SNe 2002bo, 2003du, 2005cf, 2006X, 2007gi, 2009ig, and 2011fe (Silverman et al. 2015; Zhao et al. 2015). The Si ii velocity evolution of SN 2013gs is very close to that of SN 2002bo and SN 2009ig, lying between the fast-expanding objects like SN 2006X and the normal SNe Ia like SN 2005cf and SN 2011fe. The profiles of Si ii λ6355 from t = −12 days to t = +5 days are symmetric and can be fitted well with a single-Gaussian function. No HV features are observed in the spectral sequence. Assuming that the expansion velocity of the photosphere decreases in an exponential way, we estimate the Si ii velocity of SN 2013gs at peak as v_si = 12,800 ± 400 km s⁻¹ at around the maximum light, with a velocity gradient $\dot{v}$ = 101 ± 10 km s⁻¹. These values are noticeably larger than those of normal SNe Ia, putting SN 2013gs into the HV or HVG subclasses of SNe Ia according to the classification schemes proposed by Wang et al. (2009b) and Benetti et al. (2005). The velocities inferred from the S ii, and Ca ii are higher than Normal SNe Ia by 1000 km s⁻¹, 2000 km s⁻¹, respectively. At t = −7 days, the absorption profiles of Ca ii H and K are asymmetric, probably as a consequence of contamination from Si iiλ3860. It is hard to distinguish the HV component and measure its strength. The absorption line almost keeps a Gaussian profile and varies slowly since t = −5 days. The Ca ii NIR triplet display a triangle-shaped profile in the spectra of SN 2013gs before t = −5 days. The substructures of Ca ii λ8498, λ8542, and λ8662 absorptions are blended. In analogy with Si ii λ6355 and Ca H and K, the profile of Ca ii NIR triplet evolves to become symmetric after t = −5 days.

Zoom In Zoom Out Reset image size