THE 1999aa-LIKE TYPE Ia SUPERNOVA IPTF14BDN IN THE ULTRAVIOLET AND OPTICAL

We began a Swift UVOT photometric monitoring program on 2014 May 27 and determined that the target had become bright enough for spectroscopic observations to begin on 2014 June 05. We obtained UVOT photometry in the six photometric bandpasses from phases −18 days through +45 days relative to B-band maximum light and was analyzed using the reduction method of the Swift Optical/Ultraviolet Supernova Archive (Brown et al. 2014a). We removed host galaxy light from the photometry in all six bandpasses using template images obtained one year after the initial observations. The reduced photometry is presented in Table 1.

Six epochs of spectra were observed between phases −9 through +18 days using both the UV and V grisms operating in clocked mode. The UVOT spectra snapshots were extracted, wavelength calibrated and flux calibrated using UVOTPY (Kuin 2014; Kuin et al. 2015) and were then coadded for each epoch using variance weighting. Our extraction process for the UV grism images utilized galaxy template images observed on 2015 May 19 and June 4 to remove light from the host galaxy and other contaminant sources near the SN (Smitka et al. submitted). This resulted in the removal of a feature near 2500 Å in the −6 and −3 day spectra which we attribute to a faint zero order diffraction source in the grism images. Details of the spectroscopic observations are presented in Table 2.⁷

We observed optical photometry and spectrophotometry with the Lijiang 2.4 m telescope and the Yunnan Faint Object Spectrograph and Camera (YFOSC; Zhang et al. 2014). Photometric observations were made in the UBVRI bandpasses between the dates of 2014 May 29 and July 24. We performed background subtraction of the host galaxy light for all filters using template observations gathered on 2015 March 17. To estimate the residual brightness of iPTF14bdn in the template images we used the UBVRI photometry of SN 1991T (Lira et al. 1998), which extends ∼400 days past maximum in all bands. By assuming that iPTF14bdn behaved similarly to SN 1991T in all bands, we found that we should expect the B band to contain the brightest residual source in which the SN is ∼22 mag (∼0.001 of the maximum-light flux). The lack of extinction is suggestive that the SN exploded in a low-dust environment and that we should not expect light echoes to be present in the template images. Following template subtraction, we made photometric measurements using aperture photometry. Spectroscopic observations were made on the dates shown in Table 2 spanning phases of −15 through +40 days relative to B-band maximum light.⁸

Additional photometric measurements in the UBVRI bandpasses were made using the 80 cm Tsinghua-NAOC (National Astronomical Observatories of China) Telescope (TNT; Wang et al. 2008) between the dates of 2014 June 18 and June 28. These data were reduced and processed using the same methods as the YFOSC data.

Light curves of iPTF14bdn are displayed in Figure 1. Visual inspection of the light curves shows very slow declining behavior and an I-band secondary peak about 22 days after B-band maximum. To calculate light-curve parameters for the UVOT photometry, we used a version of the SNooPy light-curve fitting algorithm (Burns et al. 2011) that had been modified by Chris Burns to accept the UVOT photometric bandpasses and calculate S-corrections (Suntzeff 2000) onto the Carnegie Supernova Project photometric system for the u, B, and V bandpasses (Hamuy et al. 2006; Contreras et al. 2010; Stritzinger et al. 2011). Using this we calculate B-band maximum light to occur at MJD = 56822.5 ± 0.3 and Δm₁₅(B) = 0.84 ± 0.05 mag. This slow declining behavior is characteristic of the 1991T-like and 1999aa-like classes of objects and does not provide adequate information for distinguishing between these two categorizations. Light-curve parameters for the ground-based Lijiang 2.4 m and TNT photometry were calculated using smooth interpolating spline fits to the data. Our calculated light-curve parameters are presented in Table 3.

Zoom In Zoom Out Reset image size

The color evolution of iPTF14bdn in uvm2–uvw1 and uvw1–V are presented in Figure 2. In both colors iPTF14bdn is bluer than normal at the earliest epochs. Its early time evolution is toward redder colors in contrast to normal SNe Ia at this epoch. Around five days prior to B-band maximum we observe iPTF14bdn to transition to normal SNe Ia evolution in both colors.

Zoom In Zoom Out Reset image size

The spectra of iPTF14bdn shown in Figure 3 represent the first published UV time-series spectra of a 1999aa-like SN Ia. Upon visual inspection we observe a weak Ca ii H & K feature near 3800 Å at early times and its subsequent evolution toward normal SN Ia levels starting around maximum light. This characteristic was first observed by Filippenko et al. (1992) in SN 1991T and later by Garavini et al. (2004) in SN 1999aa. We see a lack of strong Si ii 6355 Å features prior to maximum light and its subsequent development days after maximum light, also characteristic of 1991T-like and 1999aa-like objects (Filippenko et al. 1992; Garavini et al. 2004).

Zoom In Zoom Out Reset image size

To classify iPTF14bdn we compared the B-band photometric light-curve parameters and −6 day spectrum to those of SNe Ia subclasses. We find that Δm₁₅(B) = 0.84 mag of iPTF14bdn is consistent with that of SN 1999aa with Δm₁₅(B) = 0.85 mag (Jha 2002) and SN 1991T with Δm₁₅(B) = 0.95 mag (Lira et al. 1998). A spectral comparison of iPTF14bdn at −6 days to SNe 1999aa, 1991T and a normal SNe Ia is shown in Figure 4. We conclude that in the optical iPTF14bdn is a clone of SN 1999aa at this epoch.

Zoom In Zoom Out Reset image size

Using abundance tomography modeling, Sasdelli et al. (2014) determined that the optical spectra of SN 1991T at early times displayed several highly ionized features not associated with normal SNe Ia, notably strong Fe iii, Co iii, Si iii, and the lack of strong Si ii and Ca ii H & K features. In the week prior to maximum light, the spectra develop more normal Si ii features and then evolve to be indistinguishable from normal SNe Ia by one week after maximum light. Garavini et al. (2004) cites similar findings for SN 1999aa and notes that the resemblance to normal spectra is seen to occur at slightly earlier epochs than SN 1991T.

In the UV, we observe an overall similar trend from deviancy toward normalcy upon comparison to SN 2011fe in Figure 5 (Mazzali et al. 2014). The Ca ii H & K feature near 3800 Å is initially very weak at more than a week prior to maximum light and grows progressively more normal with time, becoming indistinguishable between +5 and +9 days. This is indicative of a highly energized atmosphere prior to maximum light.

Zoom In Zoom Out Reset image size

In the 2800–3200 Å range, we see much less UV absorption than is normal at 9 days prior to maximum light. This bright feature is shown to evolve to normal flux levels by 6 days prior to maximum light. Leading up to maximum light we see a significant deficiency of flux in the 3000–3500 Å range. The models of SN 1991T of Sasdelli et al. (2014) suggest that this strong absorption feature is attributable to a combination of Fe iii, Co ii, and Co iii indicating a highly energized atmosphere. By 5 days following maximum light we observe the evolution blueward of 3500 Å to be largely indistinguishable from SN 2011fe.

We observe no major differences between SN 2011fe and iPTF14bdn blueward of 2800 Å. In this region both SNe appear to evolve similarly, indicating that line blanketing is playing a major role in suppressing flux.

Based on a thorough analysis of the optical spectra of SN 2012fr, Childress et al. (2013) concluded that 2012fr was not a 1991T-like or 1999aa-like SN with high velocity features superimposed, but was rather best classified as a very luminous, slow declining normal SN (Δm₁₅(B) = 0.80 mag). This conclusion was based on the strong presence and high velocity of a Si ii 6355 Å feature and a lack of strong Fe signatures characteristic of 1991T-like SNe. In a separate analysis, Zhang et al. (2014) presented Swift UVOT and optical photometry and optical spectra of SN 2012fr. Their analysis found Δm₁₅(B) = 0.85 mag and concluded that SN 2012fr is likely a member of a subset of 1991T-like SNe Ia whose differences can be the result of viewing angle and asymmetric ejecta. In Figure 6 we present a comparison of UVOT UV-grism spectra of SN 2012fr and iPTF14bdn. We extracted and coadded the spectra of SN 2012fr using the same methods as iPTF14bdn. No galaxy template images were used in the extraction of these spectra because no signs of contamination are present. A composite maximum light UVOT spectrum of this SN was first presented by Brown et al. (2014c) and the full UVOT spectral series is presented here for the first time.

Zoom In Zoom Out Reset image size

Redward of 4000 Å we note that the spectra appear very similar at all epochs for which we make comparisions. We see weaker Ca ii H & K absorption in iPTF14bdn than is present in SN 2012fr, as Childress et al. (2013) reported for 1991T-like SNe.

As with SN 2011fe, we see less absorption between 2800–3200 Å in iPTF14bdn around nine days prior to maximum light. We note that the shapes of the features in this region are more similar here than was the case with SN 2011fe. Continuing to maximum light we also again observe a deficiency of flux in the 3000–3500 Å range in iPTF14bdn, where we now note that the overall shape of these absorption features are more similar than was the case with SN 2011fe. Again, we notice no significant differences blueward of 2800 Å.

The synthetic spectra synthesis program SYN++ (Thomas et al. 2011) is derived from SYNOW (Fisher et al. 1997). It computes synthetic spectra of SNe Ia in the photospheric phase by modeling the radiative transfer of photons emitted by a sharp photosphere below an expanding ejecta atmosphere. Spectra lines are assumed to form by resonance scattering using the Sobolev approximation of a moving atmosphere (Sobolev 1960; Jeffery 1989). The SYNAPPS code automates a χ² minimization routine which identifies the combination of SYN++ model atmosphere parameters that provide the best matching synthetic spectrum to an observed spectrum. The SYN++ and SYNAPPS fitting parameters include the photospheric temperature and velocity, and individual ion temperatures, velocities, ionization states, and optical depths. The program does not distinguish between stable and radioactive atomic species and does not simulate isotopic line shifts. We choose to focus only on line identifications made by the SYN++ and SYNAPPS programs and their analogous features in our observed spectra. It is noteworthy that SYNAPPS is not an ab initio SNe Ia modeling code and so does not offer as much insight into the SN progenitor system and explosion mechanism as other methods (Woosley & Kasen 2011). Thus, we encourage additional analysis of the data we present here using other modeling techniques.

To investigate the atomic structure of the SN ejecta we performed SYNAPPS model fits to the spectra near −10 and +10 days relative to B-band maximum light. Timely computation of the model fits required the use of a high performace computer cluster.⁹ For this purpose the UVOT and optical spectra at the epochs nearest these phases were smoothed, normalized, and stitched together. The model parameters assumed that all ions were attached to the photosphere and equal weight was placed on all wavelengths of the spectra. Individual ionic contributions were generated using SYN++ and the best fit parameters computed by SYNAPPS. The resulting model fits with the ion line strengths are presented in Figure 7. The model fits included all ions shown and additional contributions from Mg ii, Na i, and O i, which are not shown.

Zoom In Zoom Out Reset image size

In the model fit of the −10 day spectrum we find evidence that singly and doubly ionized iron group elements dominate the line forming regions. The two strongest features in the optical near 4250 and 4910 Å are attributed to Fe iii absorption. Ni iii is also identified in the optical by shallow features near 4700 and 5350 Å. Both of these ions also display significant absorption blueward of 3500 Å. Co iii is shown to have little to no absorption redward of 4000 Å while contributing significantly blueward of 3300 Å. Fe ii and Co ii are shown to produce less absorption than their doubly ionized counterparts while Ni ii dominates blueward of 4000 Å. Si iii and S iii are shown to be more prominent than their singly ionized counterparts. Very weak Si ii absorption is observed near 6100 Å. A Ca ii H & K absorption feature was required by the model to fit the observed absorption feature near 3700 Å. This Ca ii feature is weaker than that of a normal SN Ia at this early epoch and when present with strong doubly ionized iron group elements is considered to be the hallmark of a 1999aa-like SN Ia as opposed to a 1991T-like SN Ia (Garavini et al. 2004).

At +10 days our model shows the iron group elements to be predominantly in the first ionization state. While Fe iii still shows weak absorption features, Fe ii now dominates in both the optical and UV. We find that absorption blueward of 3000 Å is predominantly the result of Fe ii with Ni ii contributing to a lesser degree. Our model now finds no evidence of Co iii and Ni iii and instead find significant absorption from Co ii and Ni ii. The transition from doubly ionized iron group elements to singly ionized states is shown to now suppress the early-time feature near 3000 Å. In particular, the strong Fe ii absorption feature that has developed at this wavelength suggests that it is the ion primarily responsible for the evolution of this feature. Si ii is now prominent and comparable in strength to normal SNe Ia with no Si iii features evident. Ca ii is now prominent near 3700 and 8200 Å and comparable in strength to SN 2011fe.

In Figure 8 we show the −10 and +10 day best-fit model spectra divided by the best-fit blackbody emission profiles as calculated by SYNAPPS and SYN++. The best-fit model photospheric temperatures were 9200 and 8800 K for the −10 and +10 day spectra respectively. This provides insight into the total effective opacity of the line forming region in the model as a function of wavelength and is useful for comparison to other models (Hoeflich et al. 1993). In the −10 day model near 3000 Å we see evidence for significant transmission of the underlying blackbody flux. The proportion of transmitted flux is greatly reduced in the +10 day model.

Zoom In Zoom Out Reset image size

In Figure 9 we compare the rest-frame of iPTF14bdn at 6 days prior to B-band maximum light to comparable epoch spectra of SN 1991T (Mazzali et al. 1995), two high-z 91T-like SNe Ia from the ESSENCE Project (Miknaitis et al. 2007; Foley et al. 2009) and SN 1999aa (Garavini et al. 2004). We observe iPTF14bdn to closely resemble SN 1999aa at this epoch. Within the regions of spectral overlap we find an overall good agreement among the spectra with the exception of the range 3500–4000 Å, where we observe heterogeniety. Sasdelli et al. (2014) performed abundance tomography modeling of the spectra of SN 1991T within this wavelength range and determined that the ions which made the strongest contributions to SN 1991T at this epoch were Co iii, S iii, or Si iii with minimal contribution from Ca ii H & K. Garavini et al. (2004) defines the presence of a Ca H & K absorption feature in this region to be the primary distinctive characteristic between 1999aa-like and 1991T-like SNe Ia. Among all of these SNe the strengths of this absorption feature is much weaker than that of a normal SNe Ia.

Zoom In Zoom Out Reset image size

Among the SNe plotted we observe a continuum of absorption strengths within this wavelength range. We computed SYNAPPS fits of the iPTF14bdn, SN 1991T, and the ESSENCE SNe spectra using the same ions as our previous analysis and found that the varying strength of the absorption feature in this range corresponds to a varying contribution of Ca ii H & K. The SYNAPPS fits and the relative strengths of the Ca ii absorptions are shown in Figure 10. This can be indicative that 1991T-like and 1999aa-like events are part of a continuous distribution of physical parameters.

Zoom In Zoom Out Reset image size

It is noteworthy that of the five SNe we analyzed d093 of the ESSENCE project displays the strongest absorption feature in this wavelength range. The classifications of Foley et al. (2009) grouped 1991T-like and 1999aa-like SNe Ia into a single category due to their spectroscopic similarities. The absorption feature between 3500–4000 Å and its attribution to Ca ii presented here suggest that this SN is most appropriately classified as a 1999aa-like event.

The analysis of SNe Ia absolute magnitudes in the UVOT bandpasses of Brown et al. (2010) included two 1991T-like SNe Ia: 2007S and 2007cq, and no 1999aa-like SNe. In this work it was shown in the uvm2 (${\lambda }_{\mathrm{eff}\_\mathrm{SNe}\_\mathrm{Ia}}=2360$ Å) filter the absolute peak magnitude of 2007cq was ∼2 mag brighter than normal SNe Ia. The data of 2007S (Milne et al. 2010) shows uvw1 peaking 2.4 days earlier than the B-band maximum.

UVOT photometry has previously been presented for the 1991T-like SNe 2007S, 2007cq, and 2011dn (Brown et al. 2010; Milne et al. 2010, 2013; Brown et al. 2014b). UV colors are observed for 2007cq at ∼ −7 days to +21 days and 2011dn for ∼−5 days to ∼+12 days. Milne et al. (2010) found SN 2007S to have a high reddening value of E(B−V) = 0.53 and did not present data in the uvm2 band, and thus we opt to not use this SN as a comparison object in this case. In the (${uvw}1\;-\;v$) color the three SNe trace the evolution of normal SNe Ia at epochs later than ∼5 days. In (${uvm}2\;-\;{uvw}1$) iPTF14bdn and SN 2011dn show normal evolution later than ∼5 days while SN 2007cq is ∼1 mag bluer. For both colors it is only in the case of iPTF14bdn that the earliest epoch photometric observations exist and show bluer than normal colors and color evolution toward redder values contrasting normal evolution.

The photometric evidence for increased UV flux at early phases is in agreement with our observation of the 2800–3200 Å feature in the earliest UV spectra of iPTF14bdn. It is likely that this feature is the source of the greater than normal uvw1 luminosity of the 1991T-like SN measured by Brown et al. (2010) and in the photometry presented here for a 1999aa-like SN. This feature is also likely to contribute to the observed high luminosity in the uvm2 band, which is moderately sensitive to this wavelength regime. Our SYNAPPS analysis of this epoch's spectra suggests that the origin of this greater than normal luminosity is largely due to the presence of doubly ionized iron group elements which provide a low opacity window at these early epochs. For 1999aa-like SNe we conclude that the UVOT photometric filters appear able to trace the effect of temperature evolution of the iron group elements, most notably the progression of Fe iii cooling to Fe ii. Due to their spectroscopic similarities, we also expect this to be true for 1991T-like SNe Ia as well.

In Section 3.7 we presented evidence of Fe iii, Co ii, Co iii, Ni ii, and Ni iii in the −10 day UV-optical spectrum of iPTF14bdn. By +10 days we see an evolution toward Fe ii, Fe iii, Co ii, and Ni ii. If we assume a model of expanding ejecta shells above a radiating photosphere these results suggest a continuous distribution of iron group elements in the ejecta. In this scenario, the higher than normal temperatures at early times can be explained by the radioactive decay of greater than normal Ni ion abundances in the outer layers. The possible physical origin of this behavior can be an explosion mechanism that produces significant mixing of newly synthesized iron group elements into the outer layers.

Secondary maxima in infrared light curves are often observed in SNe Ia. Models of the physical origins of these features have suggested that the effect is dependent upon the mass and distribution of ⁵⁶Ni in the ejecta, and upon the recombination of doubly ionized iron group ions (Hoeflich et al. 1995; Kasen 2006). The analysis of Jack et al. (2015) identified the secondary maximum in the I-band to be the result of Fe ii emission from high excitation transitions following recombination of doubly ionized elements to singly ionized states. Their conclusion was that the I-band captures a transient Fe ii emission feature at ∼7500 Å which temporarily increases the in-band flux. Evidence supporting this conclusion was presented using optical spectra of the normal SNe 2014J and 2011fe just before and during the secondary maximum and PHOENIX models at similar epochs. They did not make any investigation of the presence of this feature in peculiar SNe Ia.

The photometry of iPTF14bdn in Figure 1 displays a secondary maximum in the I-band occurring around 20 days after B-band maximum light. Our identification in Section 3.7 of Fe iii recombining to Fe ii between −10 and +10 days, the persistent presence of Fe iii at +10 days, and the well documented similarities of normal, 1991T-like and 1999aa-like SNe at these epochs suggest that the secondary maximum may be caused by the same feature as Jack et al. (2015). Our spectroscopic sample enables us to investigate the occurrence of this phenomenon in our data set by making a comparison of the optical spectral features just prior to and during the I-band secondary maximum. In Figure 11 we compare the optical spectra of June 24 and July 5, at phases +10 and +21 days, respectively, in the region of the Lijiang 2.4 m's YFOSC I-band sensitivity. The combined transmission efficiency of the YFOSC I-band filter and CCD quantum efficiency is shown as a dotted line to highlight the spectral region to which our photometric observations are most sensitive. We show that the photometric system used is most sensitive to ∼7700 Å and is thus well suited for detecting the 7500 Å Fe ii emission noted by Jack et al. (2015).

Zoom In Zoom Out Reset image size

In iPTF14bdn we observe a similar increase in flux around ∼7500 Å as Jack et al. (2015) at the epoch of I-band secondary maximum. This confirms that the same physical processes of the normal SNe 2014J and 2011fe are likely at work in this 1999aa-like SN. We conclude that the source of the I-band secondary maximum in iPTF14bdn is likely attributable to the same Fe ii emission as in normal SNe.

We note that here the comparison spectra come from epochs of roughly equal I-band magnitude, the earlier epoch being about 5 days prior to the I-band trough minimum. This is in slight contrast to those of Jack et al. (2015), whose pre-secondary maximum spectra are timed closer to the trough minimum and lower fluxes. Thus, we expect our flux differences to be less stark, as is the case.

Prompted by our detection of Fe iii recombining to Fe ii between −10 and +10 days, we investigated the possibility of detecting the same Fe ii emission feature in our optical spectra of −13, −11, and −5 days by comparing to SN 2011fe (Pereira et al. 2013; Mazzali et al. 2014). In all cases we found no evidence for Fe ii emission around ∼7500 Å.