SN 2014J at M82: I. A middle-class type Ia supernova by all spectroscopic metrics

Four spectra (on nights Jan 22^nd, Feb 03^rd, 11^th and 13^th) were obtained using the Intermediate Dispersion Spectrograph (IDS) mounted at the Cassegrain focus of the INT. Two detectors were used depending on the set up in different nights. On Jan 22^nd the EEV10 detector was used, which has a pixel size of 13.5 $\mu$m and a spatial dispersion of 0.40^′′ pix^-1. The observation was performed with the R1200R grating, with a spectral dispersion of 0.47 Å  pix^-1, and a 1^′′-width slit. On the other three nights the REDPLUS2 detector, with a pixel size of 15 $\mu$m and a spatial dispersion of 0.44^′′ pix^-1, was used. Observations were performed with three different gratings: R1200Y (spectral dispersion of 0.53 Å  pix^-1) and the 1.3^′′ slit; R300V (spectral dispersion of 2.06 Å  pix^-1) with the 1.2^′′ slit; and R400V (spectral dispersion of 1.55 Å  pix^-1) with the 8^′′ slit.

The remaining spectra were taken at the 4.2m WHT using either the Intermediate dispersion Spectrograph and Imaging System (ISIS) or the Auxiliary-port CAMera (ACAM).

ISIS is mounted at the Cassegrain focus and consists of a dual-beam spectrograph with two independent arms which provide simultaneous observations in blue and red bands of the spectrum. The blue arm incorporates an EEV12 detector, with a pixel size of 13.5 $\mu$m, and spatial dispersion of 0.20^′′ pix^-1. The gratings used were: R600B (spectral dispersion of 0.45 Å pix^-1), R300B (spectral dispersion of 0.86 Å  pix^-1) and R158B (spectral dispersion of 1.62 Å pix^-1). Slit width was 1^′′ in all cases. The red arm has a REDPLUS detector, with a pixel size of 15.0 $\mu$m and a spatial dispersion of 0.22^′′ pix^-1. In this arm the configurations used have the following specifications: R600R (spectral dispersion 0.49 Å  pix^-1); R316R (spectral dispersion 0.93 Å  pix^-1); and R158R (spectral dispersion 1.81 Å  pix^-1). Slit width was also 1^′′ in all cases.

ACAM is mounted permanently at a folded-Cassegrain focus, and it provides fixed-format low-resolution spectroscopy. All spectra were taken with the 400-line grism V400, which covers a spectral range from $\sim$ 3500 to $\sim$ 9400 Å  ($\sim$ 3.3 Å pix^-1), and it has a spatial dispersion of 0.25^′′ pix^-1. Slit width was 1^′′ for all cases except one (Jan 28^th) when the 2.0^′′ wide slit was used.

All spectra have been reduced using standard IRAF¹¹1Image Reduction Analysis Facility, distributed by the National Optical Astronomy Observatories (NOAO), which is operated by AURA Inc., under cooperative agreement with NSF routines, including debiasing and flat-fielding. The spectra have been calibrated in wavelength taking observations of internal arc lamps installed in the Acquisition and Guiding (A&G) Box using the same instrumental configuration as for the science observations. The following standard stars were observed for the flux calibration: HD93521, BD08 2015, Grw+70 5824, G191-B2B, Feige 34, BD+75 325, PG0934+554 belonging to Massey et al. (1988), Oke (1990) and Stone (1977) catalogues; and HD109995, all included in the standard database of IRAF. The routines used for these purposes were: identify, reidentify, fitcoords and transform (for calibration in wavelength); and standard, sensfunc and calibrate (for flux calibration). In Figure 2 we show examples of three 2-dimensional spectra from different instruments at both telescopes, where the wavelength coverage, strong emission lines, and sky contamination can be seen. Figure 3 shows a composite of the fully-reduced observed spectra, where all have been shifted to the rest frame by the recession velocity of M82 ($v$ = 203 km s^-1). The complete set of spectroscopy is available electronically¹¹1https://github.com/lgalbany/SN2014J or can be downloaded from the Weizmann interactive supernova data repository (WISeREP²²2http://wiserep.weizmann.ac.il; Yaron & Gal-Yam 2012).

For the most prominent features listed above we measured their expansion velocity ($v$) and pseudo-equivalent width ($pW$) when the data allowed. The depth of the feature ($d$) has been also measured for the Si II $\lambda$5972 and Si II $\lambda$6355 absorptions (See Figure 4 for a representation of these three parameters). Their measurement has been performed as follows. First, two adjacent continuum regions 15 Å wide around the feature were selected. These were used to perform a bootstrapping method using the 225 different combinations of 2 points, one from the blue and one from the red regions. For each repetition, a segment representing the pseudo-continuum was drawn and used to normalize the observed spectrum. In the normalized spectrum, a Gaussian fit was performed to determine the wavelength ($\lambda_{e}$) at which the minimum of the feature fell. For each of the 225 repetitions, we store $\lambda_{e}$, the depth $d$ from the pseudo-continuum to the normalized spectrum at that wavelength, and the integral of the feature on the normalized spectrum ($pW$) from the pseudo-continuum. Moreover, we kept all individual uncertainties coming from the instrumental flux errors. Finally, we averaged the 225 measurements and uncertainties of $\lambda_{e}$, $d$, and $pW$, and used the standard deviation of their distributions as a systematic uncertainty, which has been added in quadrature to the instrumental error. Velocities are derived from the shift of the average minimum of the feature with respect the expected rest-frame wavelength, via the relativistic Doppler formula,

where $\lambda_{0}$ is the rest-frame wavelength of the corresponding feature, and $\Delta\lambda$ is the difference between the measured wavelength $\lambda_{e}$ and $\lambda_{0}$. In Tables 2 and 3 we present the resulting measurements for $v$ and $pW$, respectively.

The reddening modifies the shape of the spectrum reducing the flux more strongly in bluer wavelengths and producing an effect on spectral features. Although the minima of the spectral features is not going to be strongly affected (thus the $v$), both the slope of the pseudo-continuum and its deepness would be affected. Nordin et al. (2011) and Garavini et al. (2007) studied the effect of reddening in the uncertainties of $pW$ measurement and concluded that for values of $E(B-V)$ lower than 0.3 the difference is lower than 5%, but for more extinguished SN it could be important, which is the case of SN 2014J. To avoid a systematic error from reddening, we de?redden all spectra as described above before measuring line parameters.

Several spectral indicators have been used in the literature to study the properties of SN Ia and interpret their heterogeneities.

Nugent et al. (1995) defined the fractional depth of the Si II $\lambda$5972 trough the Si II $\lambda$6355 absorptions in the near-maximum light spectrum, $R({\rm Si\leavevmode\nobreak\ II})=\frac{d_{5972}}{d_{6355}}$, and showed that it correlates well with the absolute magnitude at peak, which in turn correlates with the brightness decline rate through the Phillips (1993) relation: the more luminous the SN Ia, the lower the $R({\rm Si\leavevmode\nobreak\ II})$ at maximum light and the slower the brightness decline. Both photometric and spectroscopic heterogeneities are attributed to differences in the effective temperature, which depend on total amount of ⁵⁶Ni produced in the explosion and the kinetic energy. The former can be estimated from the brightness decay in the bolometric light-curve, and the latter from the expansion velocity of the ejecta. However, the lack of correlation between the $R({\rm Si\leavevmode\nobreak\ II})$ and the photospheric velocity (deduced from Si II $\lambda$6355 blueshift) pointed out that only one-parameter cannot account for the SN Ia spectroscopic diversity (Hatano et al., 2000). In fact, the current widely spread SN Ia photometric standardization needs two parameters (stretch and color, Guy et al. 2007; Conley et al. 2008; Jha et al. 2007) to reduce the scatter on the peak magnitudes, and even a third parameter accounting for the environment has been proposed (Lampeitl et al., 2010; Sullivan et al., 2010; Betoule et al., 2014; Moreno-Raya et al., 2015).

BE05 studied the evolution of the $R({\rm Si\leavevmode\nobreak\ II})$ ratio and proposed that SN Ia could be well separated in two groups depending on their value at maximum: (i) High velocity gradient (HVG) SN Ia, with high $R({\rm Si\leavevmode\nobreak\ II})$ values right after explosion and which decreases monotonically to $R({\rm Si\leavevmode\nobreak\ II})\lesssim$ 0.2 around the epoch of maximum brightness; and (ii) Low Velocity Gradient (LVG) SN Ia show either no evolution or increasing $R({\rm Si\leavevmode\nobreak\ II})$ values from explosion up to the peak brightness, when they tend to have higher $R({\rm Si\leavevmode\nobreak\ II})$ values than HVG SN Ia. Additionally, by using a cluster analysis with other photometric and spectral properties, they defined a third group of underluminous SN Ia (FAINT) which actually have $R({\rm Si\leavevmode\nobreak\ II})>$ 0.4 at the epoch of maximum brightness. This separation was even clearer when comparing the Si II $\lambda$6355 velocity gradient ($\dot{v}_{10}$, from the epoch of maximum brightness up to 10 days post-maximum) with the $\Delta$m₁₅ decline rate parameter. While FAINT SN Ia show higher $\Delta$m₁₅ values than the other two groups, LVG and HVG are disentangled by their velocity gradient ($\langle\dot{v}_{10}\rangle_{\rm HVG}$ = 97 $\pm$ 16, and $\langle\dot{v}_{10}\rangle_{\rm LVG}$ = 37 $\pm$ 18 km s^-1 d^-1 in the original BE05 sample).

Finally, this separation was also interpreted as differences in the mechanism responsible for the explosion (while HVG SN Ia could be produced by delayed-detonations, LVG SN would be the result of deflagrations), different heavy element mixing in the structure of the WD (more efficient for HVG SN Ia, and less in LVG SN Ia), or differences in the viewing angle assuming asymmetric explosions (two SNe Ia that are physically identical in three dimensions could be classified differently just because they are seen from different directions; Maeda et al. 2010). There is a remarkable continuity in the $R({\rm Si\leavevmode\nobreak\ II})$ parameter at maximum light, enabling the presence of extreme, peculiar, and intermediate class objects.

BR06 proposed a different classification by constructing a diagram from the pseudo-equivalent widths of the same Si II features used by Nugent et al. (1995) measured at maximum light. Although the spectral features used in BE05 and BR06 diagrams are the same, the depth of the feature used in the BE05 and the shape, width, and strength with respect the adjacent continuum used in the BR06 are not showing exactly the same information. Based on the position of the SN Ia in that diagram and on the actual appearance of the Si II $\lambda$6355 feature, BR06 distinguished four different groups: (i) Shallow-silicon SN Ia (SS) show small $pW$ values in the two features ($\lesssim$ 70 Å for Si II $\lambda$6355 and $\lesssim$ 25 Å for Si II $\lambda$5972); (ii) Core-normal SN Ia (CN) have similar $pW$ and shape of the $\lambda$6355 absorption, and higher $pW$ Si II $\lambda$5972 values than shallow silicon SN Ia (up to 105 Å). Differences were due to lower temperatures than the former; (iii) Broad-line SN Ia (BL) have even higher Si II $\lambda$6355 $pW$ values ($\gtrsim$ 105 Å), and show broader and deeper absorptions than those of the core-normal SN Ia; and (iv) Cool SN Ia (CL), which have higher $pW$ for both features, specially higher Si II $\lambda$5972 $pW$ compared to the other groups ($\gtrsim$ 30 Å). As in the BE05 diagram, the intermediate regions between groups are populated with intermediate objects, showing a sequential continuity in the spectral properties for SN Ia.

There is a clear correspondence between both diagrams detailed above (see e.g. Parrent et al. 2014). BR06 CL SN Ia correspond to the BE05 FAINT SN Ia, which is to be expected, since both temperature and luminosity are controlled mainly by the ⁵⁶Ni mass. BR06 BL SN Ia correspond to the BE05 HVG SN Ia. This also makes sense, because broad Si II $\lambda$6355 absorption requires high Si II optical depth over a substantial velocity range, which makes it possible for the absorption minimum to shift appreciably with time. BL SN Ia may have thicker silicon layers than the core-normal SN Ia. Both BR06 CN and SS SN Ia correspond to the BE05 LVG SN Ia. The lower velocity range over which Si II has a high optical depth permits only a smaller shift in the absorption minimum with time.

Lately, WA09 proposed a different diagram between the pseudo-equivalent width of the Si II $\lambda$6355 absorption and the velocity of the same feature near $B$ band maximum light, instead of the $pW$ Si II $\lambda$5972 absorption used in BR06 diagram. They distinguished between normal SN Ia and high-velocity (HV), which showed a linear trend in the direction of higher velocity for higher $pW$, and also defined two other groups: underluminous SN Ia (which corresponds to 91bg-like SN Ia) with lower $pW$ and lower velocities at maximum, and overluminous SN Ia (corresponding to 91T-like SN Ia) with lower velocities but similar $pW$ than normal SN Ia. This description has one-to-one correspondence with the BR06 groups.

This scheme summarizes the spectroscopic diversity discovered so far: a decreasing temperature sequence from 1991T-like (SS) to normal (CN) and 1991bg-like SN Ia (CL), plus the high-velocity SN (BL) as a branch from the normal SN Ia group.

In Figure 5 we show the two diagnostic diagrams by BE05 and BR06. In this and in the following Figures, SN 2011fe has also been shown as a reference for comparisons. It is also a nearby object which shows few signs of extinction ($E(B-V)\sim 0.03$, accounting for both MW and host galaxy reddening, Mazzali et al. 2014). Spectra has been downloaded from WISeREP and the same spectral parameters have been measured in an homogeneous way for this work.³³3Although, as stated in the text, SN 2011fe spectra have been download from WISeREP, the actual sources that published the data used in this paper are Parrent et al. (2012), Pereira et al. (2013), Mazzali et al. (2014), and Maguire et al. (2014).

The BE05 diagram have been filled with the original objects from BE05 and Altavilla et al. (2009). HVG SN Ia (in purple) show higher values of $R({\rm Si\leavevmode\nobreak\ II})$ in the pre-maximum spectra which decrease near maximum light, while for LVG SN Ia the $R({\rm Si\leavevmode\nobreak\ II})$ parameter evolves increasing monotonically or show no evolution at all. Compared to Si II $\lambda$6355, the Si II $\lambda$5972 absorption is produced by a transition with a higher excitation energy, so $R({\rm Si\leavevmode\nobreak\ II})$ should increase in strength for higher temperatures (Nugent et al., 1995). This has been interpreted by BE05 as HVG SN having cooler temperatures at the line-forming regions that increase approaching maximum, while LVG SN, on the other hand, have high temperatures already well before maximum. Nugent et al. (1995) suggested the reason: at lower temperatures the blanketing from Fe II and Co II increase the apparent strength of Si II $\lambda$5972, at higher temperatures Fe III and Co III wash out the feature. Moreover, Nugent et al. (1995) showed a correlation between $R({\rm Si\leavevmode\nobreak\ II})$ at maximum and the peak absolute magnitude, in the direction of brighter SN having lower $R({\rm Si\leavevmode\nobreak\ II})$ values.

$R({\rm Si\leavevmode\nobreak\ II})$ measurements for both SN 2014J and SN 2011fe are listed in Table 4. SN 2011fe has $R({\rm Si\leavevmode\nobreak\ II})$ values higher than SN 2014J during the whole period shown in the left panel of Figure 5. In the pre-maximum phase this can be interpreted as SN 2011fe having lower photospheric temperature due to more line blanketing by Fe and Co, and in the near maximum phase peaking at fainter absolute magnitude. Although the reported values in the literature point to similar or even slightly brighter $M_{B}$ for SN 2011fe than for SN 2014J (14J: -19.26 $\pm$ 0.26 mag Kawabata et al. 2014, -19.19 $\pm$ 0.10 Marion et al. 2015; 11fe: -19.45 $\pm$ 0.08 Tammann & Reindl 2011, -19.21 $\pm$ 0.15 Richmond & Smith 2012) the $\Delta$m₁₅ values for SN 2011fe reported are slightly higher than for 2014J.

According to the relation between the $B$-band peak absolute magnitude and the $\Delta$m₁₅(B) parameter, which accounts for the brightness decline in the $B$-band fifteen days after maximum (Phillips, 1993; Phillips et al., 1999), these two SN lay perfectly on top of the bulk relation of normal SN Ia, well far from the subluminous SN Ia-91bg, and the overluminous super-Chandra SN Ia, as shown in Figure 6. This supports the classification of SN 2014J as a normal SN Ia from a photometric point of view, allowing its use for cosmological analyses.

Due to the lack of very early $R({\rm Si\leavevmode\nobreak\ II})$ measurements for SN 2014J no definitive classification can be done using BE05 diagram, although it seems to tentatively follow other LVG SN Ia behavior, and definitely have lower $R({\rm Si\leavevmode\nobreak\ II})$ value at maximum than SN 2011fe. However, the near maximum light $pW$ measurements allowed to position 2014J in the BR06 diagram (See right panel in Figure 5). The BR06 diagram has been populated with data from the Center for Astrophysics Supernova Program (CfA, Blondin et al., 2012) and from the Carnegie Supernova Project (CSP, Folatelli et al., 2013) to define the regions covered by each of the four groups. Although SN 2011fe falls within the CN sector (corresponding to LVG in BE05 diagram) and SN 2014J are within the BL region (which corresponds to HVG in BE05), both are located close to the border defined between the two groups. The resulting BR06 classification is in agreement with what we found in BE05: while both objects tend to follow the behaviors of their corresponding groups, they both seem to be extreme objects of their classes and similar to each other.

As a cross-check for the classification provided by these diagrams, in Figure 7 we show the corresponding diagram presented by WA09. Since SN 2014J also shows higher velocities than SN 2011fe, it is also positioned within the BL group, but in the border with the CN region.

Figure 8 shows the evolution of the expansion velocities of Si II, S II and Ca II features. In the background, we plotted in each panel the evolution of the velocities of the CSP SN Ia presented in Folatelli et al. (2013) colored by the spectroscopic group defined in BR06 diagram. The overall expansion velocities of all features are higher for SN 2014J with respect SN 2011fe. We note that in all panels, while SN 2011fe is on top of other CN SN Ia velocity evolution, SN 2014J is on the bottom end of the BL group, which stresses it being a BL SN, but very close to the CN group, and is in agreement to what we found in the spectroscopic diagrams. For the Si II $\lambda$6355 absorption we calculated the $\dot{v}_{10}$ for both SN. SN 2014J shows a velocity gradient of -58.4 $\pm$ 7.3 km s^-1 d^-1, and for SN 2011fe we found -56.0 $\pm$ 4.8 km s^-1 d^-1, in agreement with Pereira et al. (2013) who found -59.6 $\pm$ 3.2 km s^-1 d^-1. Although SN 2014J’s $\dot{v}_{10}$ is faster than the value found for SN 2011fe, both values are within the quoted errors.

Figure 9 shows the evolution of the pseudo-equivalent widths of the eight strongest features measured in this work. The average evolution (and the 1$\sigma$ deviation) of the $pW$ in each feature for the four groups defined by BR06 obtained by measuring the $pW$ from the CSP SN Ia sample are also shown for reference. In all panels the BL and CN strips are mostly overlapping, making difficult any association of the studied objects to any of these groups. The differences are clearer in the evolution of the SS and CL groups. CL SN Ia show higher $pW$ Mg II, and $pW$ Si II $\lambda$5972, and lower $pW$ S II W values, while SS have lower $pW$ in all features except in the S II W absorption.

SN 2014J seems to be more associated to the BL class if one accounts for the behavior of the Ca II H&K $pW$. The early higher values are characteristic of this group compared to the lower values that both the CN group and SN 2011fe show. This is confirmed by the lower $pW$ values found post-maximum for SN 2014J and the BL group in the S II W feature, and the slightly higher $pW$ values for the Si II $\lambda$4130, SI II $\lambda$5972, Fe II, and Ca II IR features. For Mg II we only were able to measure the $pW$ up to $\sim$10 days past-maximum light, and no conclusions can be made from this early phase. Finally, the Si II $\lambda$6355 $pW$ evolution of both SN 2011fe and SN 2014J is very similar, although the value at maximum light is higher for 14J than for 11fe. In general, BL covers the SN 2014J evolution, and SN 2011fe $pW$s evolution follow the CN strip. However, they both are very similar, which underlines the proximity of the two objects within the two groups.

Figure 10 in its top panel shows the existing correlation between $R({\rm Si\leavevmode\nobreak\ II})$ and $\Delta$m₁₅(B). Since $R({\rm Si\leavevmode\nobreak\ II})$ traces the temperature, and $\Delta$m₁₅ the brightness, the scatter the LVG objects introduces in the tight correlation between HVG and FAINT groups, can be interpreted as a need for a different physical parameter besides temperature to explain the heterogeneity of SN Ia, as discussed in BE05. Both SN 2011fe and 2014J follow the linear trend and, in this diagram SN 2014J is closer to the LVG SN Ia behavior.

We showed in Figure 8 that both SN Ia have similar velocity gradients. Their $\Delta$m₁₅ (1.06 $\pm$ 0.06 for 2014J, averaging the values reported by Tsvetkov et al. 2014; Ashall et al. 2014; Marion et al. 2015; Kawabata et al. 2014, and 1.11 $\pm$ 0.07 for SN 2011fe, from Tammann & Reindl 2011; Richmond & Smith 2012; McClelland et al. 2013; Pereira et al. 2013) are similar as well, so in the bottom panel of Figure 10, they are positioned in the overlapping region between the LVG and the HVG groups. Here we include both the BE05 objects and CSP objects. They seem to agree well with the regions defined by their classifications. FAINT SN Ia have large $\Delta$m₁₅ and $\dot{v}$ values, while the other two groups have lower $\Delta$m₁₅ and are separated by their $\dot{v}$. Here, the expansion velocity gradient, $\dot{v}$, seems to be weakly correlated with $\Delta$m₁₅(B) for LVG and FAINT groups, while HVG SN Ia are separate from LVG by their larger $\dot{v}$ values.

In all diagnostics, SN 2014J follows the characteristics of BL/HVG SN Ia, being near the boundaries between CN/LVG subclasses. SN 2011fe is in the opposite situation falling in the CN/LVG regions in all parameter spaces, but close to the BL/HVG sector. Looking at most of the parameters studied in this work, the differences between SN 2011fe and 2014J are within the reported uncertainties. The subclasses in all diagnostics are not distinct in the sense that they are not separated by regions. However, SN Ia appear to be better described as a continuum than distinct subclasses, and SN 2011fe and 2014J being that close in the border of their subclasses is further evidence that the subclassifications are artificial.