Carnegie Supernova Project-II: Near-infrared Spectroscopic Diversity of Type II Supernovae

Type II supernovae (SNe II) are classified by the presence of Balmer-series hydrogen lines in their optical spectra (Minkowski 1941). They are believed to result from the explosion of massive stars (>8M_⊙) that have retained a significant portion of their hydrogen envelope. Pre-explosion images of SN II locations suggest that the majority of SNe II are the result of red supergiant (RSG) explosions (e.g., Smartt et al. 2004; Maund et al. 2005; Smartt 2009, 2015; Van Dyk 2017; Van Dyk et al. 2019).

Historically, SNe II have been divided into subclasses based on the shape of their light curves (Barbon et al. 1979). In this work the more modern nomenclature for SNe II photometric classifications is used, with SNe exhibiting fast declining light curves classified as SNe IIL, and those with slow declining light curves classified as SNe IIP. The majority of SN II V-band light curves evolve with an initial rise to maximum followed by a steep decline before settling onto a more gentle decline, commonly referred to as the plateau. During this plateau phase, hydrogen recombination drives the light curve until a sharp decline onto the radioactive tail (Woosley & Iben 1987). One photometric and two spectroscopic classes were added within the SN II population: SNe IIb, SNe IIn, and SN 1987A-like, respectively (Arnett et al. 1989; Schlegel 1990; Filippenko et al. 1993). For the remainder of this paper, we will use the SNe II designation to refer to the collection of SNe that previously have been classified as SNe IIL or SNe IIP, as the other previously mentioned sub-types (IIb, IIn, and SN 1987A-like) are not included in the data analyzed.

It has been suggested that fast declining SNe II have a smaller amount of hydrogen in their envelopes than slow declining SNe II (Popov 1993; Faran et al. 2014; Gutiérrez et al. 2014; Moriya et al. 2016). However, it is still not known whether there are progenitor differences of explosion scenarios that separate the slow and fast declining SNe II into two distinct groups (Barbon et al. 1979; Patat et al. 1993). Furthermore, recent publications have shown that there is no discrete photometric separation between these SNe (e.g., Anderson et al. 2014; Sanders et al. 2015; Galbany et al. 2016; Valenti et al. 2016; Pessi et al. 2019); although see Arcavi et al. (2012) for a possible separation.

The amount of SN II optical data obtained has greatly increased in the past decade with the focus turning toward larger samples to examine their spectroscopic and photometric diversity (Anderson et al. 2014; Faran et al. 2014; Gutiérrez et al. 2014, 2017a, 2017b; Galbany et al. 2016; Rubin & Gal-Yam 2016; Valenti et al. 2016; Faran et al. 2018). Large photometric studies have defined parameters useful for characterizing the diversity of SN II optical light curves. They found that brighter SNe II decline more quickly at every phase, have shorter plateau phases, and higher ${}^{56}$ Ni masses. This is significant because it further supports that faster declining SNe II are the result of explosions with lower hydrogen envelope masses, which causes their shorter plateau duration. Gutiérrez et al. (2017a), using optical spectra, found that SNe II span a continuous range in equivalent widths and velocities.

SNe II in the near-infrared (NIR) have yet to be explored in detail. However, there are well-known advantages to observing spectroscopically in the NIR, namely, the variety of strong lines present, less line blending when compared to the optical, and a lower optical depth revealing the core at earlier times (Meikle et al. 1993). Furthermore, at late times, carbon monoxide (CO) is sometimes observed (Spyromilio et al. 1988, 2001; Rho et al. 2018).

Spectroscopically, the most well-studied SN II in the NIR is SN 1987A (Bouchet et al. 1987; Oliva et al. 1987; Catchpole et al. 1988; Elias et al. 1988; Rank et al. 1988; Whitelock et al. 1988; Catchpole et al. 1989; Meikle et al. 1989; Sharp & Hoeflich 1990; Bouchet et al. 1991; Danziger et al. 1991; Meikle et al. 1991, 1993) with over 30 NIR spectra of SN 1987A available. SN 1987A exploded in the Large Magellanic Cloud, which allowed for extensive spectroscopic and photometric coverage still being obtained today. SN 1987A was a peculiar SN II, which most likely originated from a blue supergiant progenitor (Podsiadlowski 1992). This peculiarity led to the definition of an additional SN II subclass, SN 1987A-like, which are SNe that have long rising light curves.

Most previously published SNe II NIR spectra are of a single SN across a small number of epochs (Benetti et al. 2001; Hamuy et al. 2001; Elmhamdi et al. 2003; Pozzo et al. 2006; Pastorello et al. 2009; Maguire et al. 2010; Fraser et al. 2011; Tomasella et al. 2013; Dall'Ora et al. 2014; Jerkstrand et al. 2014; Morokuma et al. 2014; Takáts et al. 2014, 2015; Valenti et al. 2015, 2016; Rho et al. 2018; Bostroem et al. 2019; Szalai et al. 2019). These works have been able to characterize the general spectral line evolution from a couple days past explosion through the nebular phase. The SN IIP SN 1999em was particularly well studied allowing line identification and analysis (Hamuy et al. 2001; Elmhamdi et al. 2003), however, SNe II are diverse and thus, SNe II in the NIR have not been fully explored.

By modeling the NIR and optical spectra of SNe II, Takáts et al. (2014) found that the 0.98–1.12 μm region contains a high-velocity (HV) helium feature on the blue side of the Paschen gamma (P_γ) hydrogen absorption. This feature, previously unconfirmed, mirrors the velocity of the HV hydrogen feature claimed in the optical by Gutiérrez et al. (2017a).

In this paper we present the largest SNe II NIR data set published to date. The data were obtained between 2011 to 2015 as a part of the Carnegie Supernova Project-II (CSP-II; e.g., Phillips et al. 2019); CSP-II was a National Science Foundation-supported program to study Type Ia SNe as distance indicators, which aimed to improve upon CSP-I (Hamuy et al. 2006) by observing objects from untargeted searches detected in the Hubble flow. CSP-II also worked to obtain a large sample of SN Ia NIR spectra for studies in cosmology and in their explosion physics (Hsiao et al. 2019). Furthermore, NIR spectra of all types of nearby SNe were obtained, as the current sample is small and they are crucial for understanding the origins of these explosions.

The sections of this paper are outlined as follows. Section 2 is an overview of the data sample, including the observation and reduction techniques used. In Section 3, we describe the process used for measuring various photometric and spectroscopic properties. In Section 4, we outline the NIR spectral features and their evolution over time. In Section 5, properties of the NIR hydrogen features are discussed. Section 6 describes the observed NIR spectral dichotomy. In Section 7, we present the results of a principal component analysis (PCA) on the CSP-II SNe II data set and corresponding NIR spectral templates. The discussion of results and conclusions are in Sections 8 and 9, respectively.

The CSP-II sample contains 92 NIR spectra of 32 SNe II. Figure 1 shows the phase of each NIR spectrum, as well as the distribution of the number of spectra taken per SN. NIR spectra were obtained with the Magellan Baade Telescope, equipped with the Folded-port Infrared Echellette (FIRE; Simcoe et al. 2013), and with the NASA Infrared Telescope Facility (IRTF), equipped with SpeX (Rayner et al. 2003). The spectra were reduced and corrected for telluric absorption following the procedures outlined in Hsiao et al. (2019). Spectra obtained after 300 days are not included in this sample because they are well into the nebular phase, giving a total of 81 NIR spectra from 30 SNe II. See Table 1 for a list of all SNe used, Table 2 for a log of the observations, and Figures 2–4 for all the spectroscopic data. Spectra can be downloaded from the Web.²² The high throughput of FIRE allows us to recover enough counts in the telluric regions to enable telluric corrections, such as the Pα feature, to a precision of 10% or better in 70% of our sample. 11 SNe II have multi-epoch observations, and 14 have well-sampled light curves. Photometry was obtained with the Las Campanas Observatory Swope and du Pont telescopes. The observing strategy and technique for the photometric data are described in Phillips et al. (2019).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The sample ranges from redshift 0.002 to 0.037, with a mean of 0.013 and median of 0.009. Table 3 lists the redshift and host galaxy for each SN II within the sample. Host information obtained from the NASA/IPAC Extragalatic Database (NED),²³ provides us with distance estimates to each SN II. Table 4 lists both redshift-independent and redshift-dependent distances. We use the redshift-independent distance to calculate absolute luminosity whenever possible. Our sample contains no SNe IIn, SNe IIb, or SN 1987A-like objects.

For each SN II, the explosion date is assumed to be the midpoint between the last non-detection and discovery, while the uncertainty in the explosion date is taken as half of the range between the last non-detection and the discovery date, a method adopted by Anderson et al. (2014) and Gutiérrez et al. (2017a). Spectral matching techniques are employed in order to constrain the explosion date using the Supernova Identification (SNID) code for each SN in the sample, even those with well constrained, less than 20 days before discovery, last non-detection (Blondin & Tonry 2007). The phases obtained using SNID match those from the last non-detection method, without any obvious bias toward last non-detection or discovery. It is assumed that there is a flat prior in the phase distribution. SNID does not have NIR templates, so all spectral matching was done in the optical with previously published data. Using SNID with updated SN II templates has been shown to be a valid method to constrain the explosion epoch (e.g., Anderson et al. 2014; Gutiérrez et al. 2017a). When using SNID, the explosion date is taken as the average of the best matched spectra and the range is taken as the error. Of the 30 SNe II, 13 have well-defined last non-detections, 8 have explosion dates constrained with SNID, 4 have poorly constrained last non-detections, and 5 have no optical data and no available last non-detection date.

3.1. Photometric Measurements

3.2. Spectroscopic Measurements

SNe II as a whole are generally similar spectroscopically in the NIR. Using the spectral line identifications from Branch (1987) and Meikle et al. (1989), we identify the features and evolution over time of the SNe II within our sample; see Figure 5.

Zoom In Zoom Out Reset image size

4.1. Early Phase

4.2. Plateau Phase

As SNe II settle into the plateau phase ∼20 days past explosion, more features appear (see the middle spectrum in Figure 5). The Ca ii triplet is the strongest feature at this phase. It initially just shows an absorption feature and develops a stronger P Cygni profile as time passes. It is comprised of a blend of three Ca II lines: two close together and one more prominent, located at λλ0.854 μm, 0.866 μm, and 0.892 μm, respectively. The development of this feature is not uniform, with many SNe II showing different line strengths.

The S i λ0.922 μm feature shows up in the NIR around 30 days post explosion. The S i P Cygni profile strengthens over time in all SNe II.

Sr ii λ1.004 and λ1.092 μm are both blended with H i features and cannot be resolved. Sr ii λ1.033 μm appears in all SNe II but with a varied evolution, with many SNe not showing Sr ii until after the plateau phase ends; however, there are SNe that show strong Sr ii absorption from Sr ii λ1.033 μm during the plateau. For these cases, the feature can be seen as early as 20 days past explosion, increasing in strength over time and becoming a dominant feature in the SN spectrum.

P_δ λ1.005 μm is one of the weaker hydrogen lines detectable in SNe II NIR spectra. Blended with the P_δ feature is Sr ii λ1.004 μm that causes stronger line blending than is seen in the other hydrogen features. The strength of the P Cygni profile does not increase as quickly as other hydrogen features.

P_γ λ1.094 μm is highly blended with He i λ1.083 μm and Sr ii λ1.092 μm. It has a very weak absorption, that is only noticeable in most SNe toward the end of the plateau as a notch on the red side of the He i absorption component. The emission in this region is attributed mostly to P_γ and evolves similarly to the other H I features.

During the plateau, the most prominent absorption arises from He i λ1.083 μm and has other possible contributions from Sr ii λ1.092 μm and P_γ λ1.095 μm. Since other Sr ii transitions in this region are weak, it is unlikely that this absorption is dominated by Sr ii. He i is excited nonthermally and would only be seen if the ⁵⁶Ni was located close to the He region in the ejecta (Graham 1988). We note that this feature is not expected to have a significant contribution from H i because no strong H i absorption is present in the other Paschen series lines. However, the emission of this P Cygni profile is most likely to have contribution from P_γ, as strong emission features are seen in other Paschen series lines.

A strong He i absorption profile with a weak emission component is present in the early optical spectra of SNe II (Gutiérrez et al. 2017a). Thus, in the NIR we do not expect He i to contribute a large amount of flux in emission to this feature. It is also possible for the emission component to be shifted blueward by the absorption of H i and Sr ii. At 50 days past explosion, this P Cygni profile consists of a blend comprised of He i, H i, and Sr ii, with He i λ1.083 μm being the dominant contributor in absorption. Thus, we will refer to this absorption as He i for the remainder of this work. At later times, as the photosphere recedes to the inner hydrogen-rich region, we expect heavier elements to appear. Hence, late contributions from C i and Si i are likely.

Around 50 days post explosion, a dichotomy in the region around He i λ1.083 μm appears. Figure 6 shows two SNe that represent the dichotomy. SNe with more prominent He i absorption show no other lines in this region until later phases when a small absorption, comprised of P_γ and Sr ii λ1.092 μm, appears on the red side of the He i absorption. SNe with a shallower He i absorption show other features that are not seen in SNe, which exhibit a deeper He i. The P_γ/Sr ii absorption appears earlier in these SNe and an additional absorption appears on the blue side of He i, hereafter feature A (Figure 6).

Zoom In Zoom Out Reset image size

Feature A, if present, appears before the plateau and is similar to the He i λ1.083 μm transition at these early times. During the plateau, feature A does not significantly increase in strength compared to the other absorption features in the spectrum and becomes less prominent compared to He i absorption in the same region.

O i λ1.129 μm appears on the red side of the P_γ emission toward the end of the plateau phase. Si i λ1.203 μm is also seen during this period and strengthens over time.

P_β λ1.282 μm is seen at the beginning of the plateau and exhibits a symmetric P Cygni profile, which becomes dominated by its emission feature with very little absorption. Toward the end of the plateau, the P_β line profile seems to widen, likely due to the presence of a Si i multiplet in the region.

P_α λ1.875 μm is seen early during the plateau, appearing predominately in emission. It may form earlier, along with the other Paschen series lines; however, it is located in a band of strong telluric absorption making it difficult to study for most spectra. Looking at SNe II with high signal-to-noise ratios (S/N) in this region, we see that P_α increases in strength over time. In the same band of telluric absorption, P λ0.954 μm and the Brackett series hydrogen line B_δ λ1.944 μm appear around 30 days post explosion as weak emission features. These lines originate from the same upper level and thus both grow in strength, with neither getting particularly strong compared to the other hydrogen emission features seen.

B_γ λ2.165 μm shows up around the same time as P_α and cannot be resolved if the S/N of the spectrum is low, e.g., in some of the SN 2013gd spectra. The absorption of B_γ is weak enough that it can only be seen in the highest S/N spectra within the sample. The emission is also weak when compared to the Paschen series.

4.3. Radioactive Decay Tail

After the plateau phase ends, we observe more metal lines forming, mostly in emission, and the emission component of all lines that formed during the plateau strengthen.

[Fe ii] λ1.279 μm, Mg i λ1.503 μm, and [Fe i] λ1.980 μm emerge during the radioactive decay tail as weak emission features. Some SNe, e.g., SN 2013gd, do not show significant Mg i emission at any time. He i λ2.058 μm appears as a weak absorption and does not appear in all SNe II. This He i λ2.058 μm transition should be highly correlated with the He i λ1.083 μm transition.

CO begins to appear after the plateau. The first CO overtone appears largely as emission features around 2.3 μm. The earliest detection in this data set is found in the SN 2013by spectrum taken 95 days post explosion. The CO feature becomes stronger over time, appearing in more SNe II around 120 days post explosion. We find CO in four SNe II within our sample, appearing in most spectra taken after 120 days post explosion.

Late time spectra with possible CO detections are shown in Figure 7. Modeling and further analysis of CO will be performed in future work.

Zoom In Zoom Out Reset image size

The NIR spectral features were measured as outlined in Section 3.2 and are presented below. Figures 8 and 9 show the velocity and pEW evolution for the CSP-II sample over time, respectively. Figures 8 and 9 present the data split by s₂ in order to search for any relationships between photometric and NIR spectral properties.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

He i λ1.083 μm strengthens over time until nebular phases and often has the highest absorption pEW, increasing from 10 Å to ∼50–100 Å in the NIR spectrum. There is a dichotomy seen in the He i pEW, SNe well above 50 Å and those below 50 (Figure 9 top right panel). The velocity of He i absorption decreases over time, similar to the other spectral features, going from ∼11,000 km s⁻¹ at early times to ∼4000 km s⁻¹ after the plateau phase ends.

The velocity of the P_β absorption at early and late times is nearly uniform among the sample; however, during the plateau there is a split in the velocities between fast and slow decliners, ∼7500 km s⁻¹ for fast decliners and ∼4500 km s⁻¹ for slow decliners. The absorption pEW increases from ∼1–2 Å to 10 Å during the plateau. The emission pEW increases to over 100 Å during the plateau, becoming the second strongest emission behind P_α during this period. The emission velocities and absorption/emission pEWs show no correlation with the photometric subclass.

The absorption and emission equivalent widths of the P_δ P Cygni profile do not increase as quickly as other hydrogen features. With velocities around 11,000 km s⁻¹, P_δ often has the lowest velocities of all present hydrogen features, especially in fast declining SNe where the P_β velocities are higher, at certain epochs, than in slow declining SNe. The P_δ velocities are likely influenced by the Sr ii λ1.092 μm blend, causing them to be lower than expected. The P_δ absorption pEW evolves much like P_β going from ∼1–2 Å to 10 Å during the plateau. The emission pEW is uniform among the sample, evolving linearly over time from ∼1–2 Å to 50 Å at 120 days past explosion.

Feature A has the most inconsistent evolution, in velocity and pEW, compared to other NIR features (Figure 10). This feature has been previously identified as either Si i λ1.033 μm or HV He i λ1.083 μm (Takáts et al. 2014). The absence of other metal lines in the spectra at similar times suggests that feature A is not Si i. Furthermore, the timing of the feature also suggests it is not Si as the photosphere is not expected to be in the metal-rich region or the inner region of the hydrogen-rich envelope at these early epochs. Therefore, we conclude that feature A is most likely HV He i. If present in an SN, it appears early, for example 7 days past explosion in SN 2013gd, and does not change significantly over time in pEW or velocity (Figure 11). The early onset of the feature suggests that it was produced in the outermost layer of the envelope. Unlike the photospheric He i absorption, feature A does not appear at the same wavelength in each SN, if it appears at all. It has a velocity spread between objects of ∼4000 km s⁻¹ at similar epochs. This velocity spread assumes the feature is He i.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

NIR spectra of SNe II are generally uniform showing the same features with similar time evolution between SNe. However, they exhibit a dichotomy of properties that emerges around 50 days post explosion and lasts until the end of the SN plateau phase. This is most prominent in the region around He i λ1.083 μm; a comparison between the two groups is shown schematically in Figure 6. Therefore, we classified our sample of SNe II based on this dichotomy and describe the differences between the two groups (weak and strong) below.

We define weak SNe II as those with He i λ1.083 μm absorption pEW less than 50 Å after 50 days from explosion, and strong SNe II with He i pEW greater than 50 Å. This quantitative division was arbitrarily chosen based on Figure 9. There are no SNe with intermediate pEW (50–75 Å) in our sample (histogram in the top right panel of Figure 9 shows the pEW of each SN interpolated to 75 days). Figure 11 shows all the SNe in the sample that could be classified as weak or strong. Although the pEW of He i was used to classify the two groups, several other differences between the two groups are also observed.

Weak SNe show an accompanying absorption feature, feature A, on the blue side of He i, and strong SNe do not. As discussed in Section 4.2, feature A is most likely HV He i 1.083 μm. Feature A always shows up before photospheric He i 1.083 μm, consistent with the interpretation that feature A is HV He i. Weak SNe show earlier notches arising from P_γ/Sr ii absorption on the blue side of the He i/H i emission, ∼20 days past explosion. Strong SNe show the P_γ/Sr ii absorption at ∼40 days. Weak SNe tend to exhibit strong Sr II features in the 1.0–1.1 μm region sooner, ∼20 days. Strong SNe may not show Sr II at all, e.g., SN 2013hj. Weak SNe have a more layered velocity structure and decline in velocity more quickly than strong SNe (Figure 8). CO emission can be seen in strong SNe before 100 days past explosion. Emission from the first CO overtone appears at later times, past 120 days, in weak SNe, if seen at all.

In the optical, a small notch on the blue of the H_α absorption was interpreted as HV H i, after 30 days post explosion, and its evolution is well-studied (Gutiérrez et al. 2017a). Following Takáts et al. (2014), we compare the evolution of these HV features in Figure 12 to determine if the features are formed in the same region. In our sample only one SN II shows both HV features: SN 2012aw. However, this does not mean that the HV H i is absent in all other SNe II, as it could be mixed into H_α as suggested by Chugai et al. (2007). H i is also easier to excite than He i, which could explain the higher incidence of HV H i in the optical. It is possible for an SN to show a feature that looks similar to HV H i and no NIR HV counterpart, e.g., SN 2013ej. The absorption on the blue side of H_α in the optical may also be due to Si ii, as suggested by Valenti et al. (2014) for SN 2013ej. When these HV components appear in the same SNe, their velocities match, e.g., SN 2012aw with HV H i and HV He i both around 14,000 km s⁻¹.

Zoom In Zoom Out Reset image size

The NIR spectroscopic classifications of weak and strong are found to have a one-to-one correspondence with the IIP and IIL subclasses, based on the decline rate s₂. For the SNe in our sample where we can determine a photometric classification (five slow decliners and four fast decliners), all slow decliners are weak and all fast decliners are strong. When we include all of the previously published data (see Table 5) that can be spectroscopically classified, either by pEW measurements or by the line profiles in the region around He i λ1.083 μm (eight slow decliners and two fast decliners), we find only one exception to the rule, SN 2012A (Tomasella et al. 2013). Looking at NIR spectra, the strong dichotomy suggests that fast declining and slow declining SNe II are two distinct groups of objects, at least phenomenologically; whereas when viewed in the optical, these objects appear to have a continuous range of photometric and spectroscopic properties (Anderson et al. 2014; Gutiérrez et al. 2017a, 2017b; Pessi et al. 2019). A Silverman model (Silverman 1981) was used to create s₂ probability density functions (PDFs) for both the CSP-II and Gutiérrez et al. (2017b) photometric samples. A Silverman model assumes that each point can be represented by a Gaussian profile. These profiles can then be added together and normalized to create a PDF of the data set. A PDF was made using the error from each s₂ measurement; however, for the majority of measurements this error is small compared to the s₂ value and produces a PDF that overestimates the number of measurement modes. Thus, each point in the Silverman model is assigned the same kernel density, which was chosen using the critical width function of Silverman (1981). The resulting PDFs of the two samples are similar despite the small numbers of the CSP-II sample, suggesting that the CSP-II sample is not photometrically biased.

To further test the validity of the dichotomy observed in the NIR spectra of SNe II, we performed a PCA on a selection of data within our sample. PCA has been applied in a variety of astronomical research previously (e.g., Suzuki 2006; Hsiao et al. 2007) to recognize patterns in the data and as a tool for creating spectral templates. PCA reduces the dimensionality of a multidimensional data set, such that a large fraction of the variance in the data is captured in a few principal components.

We require that each input spectrum for PCA have a S/N of at least 50 in the telluric regions, 1.35–1.45 μm and 1.80–2.00 μm, as PCA is susceptible to noise spikes. The zeroth principal component makes up ∼50% of the total variance of the spectrum, as shown in Figure 13. This component describes the slope, line strength, and velocity over time of the SNe II, suggesting these spectral properties are all correlated. There is a strong correlation between the projections of observed spectra onto the zeroth principle component and phase, as expected, because the line strengths of most features increase with time. The first component picks out ∼31% of the variance and shows further color and calcium changes when the region around He i λ1.083 μm is uniform. The second component makes up ∼6% of the total variance and is of particular interest as it picks out the difference in this region between weak and strong SNe II. The dichotomy of the region around He i λ1.083 μm motivated splitting the sample into two templates and further supports the two distinct groups outlined in Section 6.

Zoom In Zoom Out Reset image size

Cosmological studies using both Type II and Type Ia SNe are pushing into the NIR to minimize the effect of dust extinction (e.g., Rodríguez et al. 2019), and representative spectral templates of all types of SNe are crucial for future SN cosmological experiments. The two spectral templates are split by classification of weak and strong He i λ1.083 μm absorption and are created by polynomial fitting to the projections over time. The resulting strong and weak templates are plotted, with different colors representing differents epochs, in Figure 14. We conclude that slow declining SNe II (conventionally IIP) are best represented by the weak spectral template while the strong template better fits fast declining SNe II (conventionally IIL). Our template spectra can be found on the Web.²⁴

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Due to the homogeneity of the hydrogen-rich envelope, the difference in spectral features between fast and slow declining SNe II most likely comes from explosion energy and temperature changes (Hoeflich 1988; Takeda 1990; Duschinger et al. 1995; Gerardy et al. 2000). The light curves are powered by the energy stored by the hydrodynamical shell during the early phase, the heating by radioactive decay of ${}^{56}\mathrm{Co}\to {}^{56}\mathrm{Ni}\to {}^{56}\mathrm{Fe}$, and the recombination energy. During the plateau phase, the opacity drops by several orders of magnitude from the outer recombined layers above the photosphere to the inner ionized layers. The release of stored energy and the recombination of hydrogen powers the plateau phase (Chieffi et al. 2003). The luminosity drops when the recombination front reaches the inner edge of the hydrogen-rich layers. The total energy released and the duration of the plateau decreases with the total hydrogen mass and, eventually, marks the transition from slow declining SNe II with long plateaus to fast declining SN II; however, explosion energy likely plays a significant role (Popov 1993). With increasing hydrogen mass, the C/O-rich layers are exposed later. This may explain why in weak SNe II we see later emission from the first overtone of the CO band (Gerardy et al. 2000; Smith et al. 2001; Matsuura 2017; Banerjee et al. 2018), pointing to a mass sequence.

Progenitors of slow declining SNe II have been interpreted to have a lower zero-age main-sequence mass than those of fast declining SNe II, as they have retained much of their hydrogen envelope (Heger et al. 2000). Alternatively, Morozova et al. (2017, 2018) suggest that SNe IIL may be the result of an RSG surrounded by dense circumstellar material. If SNe IIL have lost some of their hydrogen envelope, the process in which they lose this mass is uncertain. We cannot make a distinction in this work between SNe that have lost some of their hydrogen envelope by wind or a common envelope phase.

The longer lifetime of presumably lower mass progenitors of the slower declining SNe II leading up to the explosion allows more s-process elements, such as Sr, to be produced, providing a possible explanation for the Sr II lines observed mostly in the SNe II weak class (Limongi et al. 2000). However, we see no correlation with spectral type and progenitor mass, determined using pre-explosion images, for those SNe within our sample and the sample of Smartt (2015), possibly due to the large uncertainty in mass.

The mixing of ⁵⁶Ni to HVs could explain the pEW difference between SNe II weak and strong. The presence of He i requires high energy nonthermal photons close to the He layer (Graham 1988). This can only be achieved by gamma-rays produced from the radioactive decay of ⁵⁶Ni.

Multiple scenarios exist to explain the HV features in the spectral weak class. The most likely explanation is from thermal excitation produced by a reverse shock as a result of the interaction between the SN ejecta and wind (Chugai et al. 2007), in which H and He are excited by X-rays, with the HV H possibly mixed into H_α and HV He present in the NIR. Alternatively, Dessart & Hillier (2008) suggest a radiation transport effect in layers with certain density slopes where hot, high-density inner layers and low-density outer layers with long recombination times are separated by a region of low excitation common to all SNe II. However, this theory does not explain why these weak features are not seen in fast declining SNe II. Moreover, the models of Dessart & Hillier (2008) predict He i showing up first and vanishing after ∼20 days, in contradiction to the observations. Thus, we believe the HV features seen are powered by a shock.

We have presented 81 NIR spectra of 30 SNe II observed between 2011 and 2015 as part of the CSP-II. The spectra range between 1 and 150 days post explosion. Using V-band light curves, photometric properties were measured for 14 SNe within the sample. The evolution of NIR spectral features was outlined and for the most dominant features; pEW and velocities were systematically measured.

There is a strong dichotomy in the NIR spectral features of SNe II, and it is the most prominent in the features around He i λ1.083 μm. Thus, we presented spectral classifications based on the strength of this feature: strong and weak SNe II. Characteristics of the two classes are outlined as follows:

• 1.  
  Weak SNe show an accompanying absorption feature on the blue side of the He i λ1.083 μm, feature A, which we interpret as an HV component of the same He i line, and strong SNe do not. The HV He i line always has an earlier onset than the photospheric He i λ1.083 μm absorption feature.
• 2.  
  Weak SNe show earlier absorption, ∼20 days past explosion, from P_γ/Sr ii than strong SNe. Strong SNe show this feature at ∼40 days.
• 3.  
  Weak SNe more often exhibit Sr ii features in the 1.0–1.1 μm region.
• 4.  
  Strong SNe can form the first overtone of CO at earlier times, less than 100 days. Weak SNe have later formation of CO, past 120 days.

We found that these spectral classifications of strong and weak SNe correspond to the plateau decline rate: slow declining SNe II (IIP) are weak and fast declining SNe II (IIL) are strong. This a somewhat surprising result given the previous research showing a continuum in optical light curves and spectroscopic features. The HV He i feature seen is most likely due to a reverse shock, as explained in Chugai et al. (2007). However, this does not explain the dichotomy. PCA was performed on the spectra, which further confirmed the observed dichotomy that represented ∼6% of the spectroscopic variance. Finally, using PCA, two spectral templates for SNe II were created, SNe weak and strong. These templates are crucial for cosmological studies using SNe II.

The authors would like to thank the anonymous referee for their comments. The work of the CSP-II has been generously supported by the National Science Foundation (NSF) under grants AST-1008343, AST-1613426, AST-1613455, and AST-1613472. The CSP-II was also supported in part by the Danish Agency for Science and Technology and Innovation through a Sapere Aude Level 2 grant.

We would like to thank Michael Cushing for his work on Spextool in processing low-resolution data. P.H. acknowledges support by the NSF grant 1715133. Research by D.J.S. is supported by NSF grants AST-1821987, AST-1821967, AST-1813708, and AST-1813466. CPG acknowledges support from EU/FP7-ERC grant no. [615929]. Support for J.L.P. is provided in part by FONDECYT through the grant 1191038 and by the Ministry of Economy, Development, and Tourism's Millennium Science Initiative through grant IC120009, awarded to The Millennium Institute of Astrophysics, MAS.

Software: firehose (Simcoe et al. 2013), xtellcor (Vacca et al. 2003), Spextool (Cushing et al. 2004), SNID (Blondin & Tonry 2007).

For early epochs and the plateau phase, a four-parameter fit is applied to determine the decline rates (s₁ and s₂), the transition epoch between the two decline rates (t_trans), and the magnitude offset of the light curve. Fitting is also performed with multiple linear three parameter functions in order to determine if the light curve has two distinct early time decline phases characterized by multiple slopes. Bayesian Information Criterion (Schwarz 1978) was used to determine the goodness of fit between the three- and four-parameter models. A linear fit is also applied to determine the radioactive decay tail slope, s₃, if sufficient photometric data past the plateau exists. The midpoint time of transition, t_PT, between the plateau and linear decline phases is obtained through fitting the light curves with an eight-parameter fit function from Olivares et al. (2010),

$$\begin{eqnarray}&&F(t)=\displaystyle \frac{-{a}_{0}}{1+{e}^{\left(t-{t}_{\mathrm{PT}}\right)/{w}_{0}}}+{p}_{0}\left(t-{t}_{\mathrm{PT}}\right)+{m}_{0}-{{Pe}}^{-{\left(\tfrac{t-Q}{R}\right)}^{2}},\end{eqnarray} \tag{ 1 }$$

where t_PT corresponds to the midpoint of transition between plateau and radioactive tail, and m₀ is the zero-point magnitude at the transition time. The first term is a Fermi–Dirac function that describes the transition between plateau and linear decline phases. The second term and zero-point magnitude, m₀, describe the radioactive tail of the light curve and offset the fit, respectively. The last term is a Gaussian that describes the shape of the light curve during the plateau. Figures 15 and 16 and show the CSP-II V-band light curves, as well as other light curves from the literature. Results of the fitting are listed in Table 6.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size