Optical and Near-infrared Observations of the Nearby SN Ia 2017cbv

The raw optical images were collected by the Yale SMARTS team and reduced through a data pipeline via the NOAO IRAF package.²⁶ The reduction included the subtraction of an overscan region, a zero frame, and division by a normalized dome flat.²⁷ The reduced images were downloaded from the SMARTS ftp site. Then, cosmic rays were detected and removed using Laplacian cosmic-ray identification (van Dokkum 2001).²⁸ The astrometric calibration for the optical images were carried out using Astrometry.net (Lang et al. 2010).

As SN 2017cbv is located far away (160'') from the center of its host galaxy, light contamination from the host is negligible (see Figure 1). Aperture and point-spread function (PSF) photometry were performed on the optical images of SN 2017cbv.

Zoom In Zoom Out Reset image size

The photometry of the stars in the field was not completely consistent when comparing Carnegie Supernova Project (CSP; Burns et al. 2020) and ANDICam images. We found an uncorrected illumination pattern in the ANDICam images, which is more evident when combining a large number of images. Unfortunately, there are no available flat-field images to make a proper correction, so instead, we opted to use only stars in the neighborhood of the SN to produce relative photometry. In detail, we used two bright and isolated stars in the close neighborhood of SN 2017cbv to measure differential photometry (we did not detect time-series variability for the two selected stars during our observing window). The CSP also observed the field in the ugriBV bands (Burns et al. 2020) and provided us with their calibration in order to derive the BVRI values. We measured differential photometry using PSF and aperture photometry relative to star 2 (using the PSFEx and SEP tools, respectively; Bertin & Arnouts 1996; Bertin 2011; Barbary 2018), as star 1 was saturated in all CSP ri-band images but not on our ANDICam images. We used the PSF photometry, as it was also used in the NIR images. We took the aperture photometry as a sanity check, showing a 0.02 mag systematic deviation, which was added to the final error budget as a systematic error.

Due to the illumination problem (see above), we did not calibrate the stars in the SN field; instead, we used calibrations from CSP photometry (Burns et al. 2020). The BV-band calibrations of ANDICam in the natural system were transformed from CSP BV bands via its color term coefficients of B − V in Table 2. The ANDICam R- and I-band calibrations in the natural system were derived in a similar way, and the standard magnitudes of the CSP RI bands were measured from the average values of the ri bands following the Sloan Digital Sky Survey (SDSS) transformation equations (Jester et al. 2005; Lupton et al. 2005; Jordi et al. 2006). The BVRI magnitudes of star 2 are listed in Table 3.

In spite of having a calibrated set of local-sequence stars, we still need color terms to transform instrumental photometry into the standard system. The color terms of ANDICam were determined from images of the standard field Rubin 149 following the equations

$$\begin{eqnarray}&&B=b+{Z}_{B}-{k}_{B}\times X+{C}_{B}\times (B-V),\end{eqnarray} \tag{ 1 }$$

$$\begin{eqnarray}&&V=v+{Z}_{V}-{k}_{V}\times X+{C}_{V}\times (B-V),\end{eqnarray} \tag{ 2 }$$

$$\begin{eqnarray}&&R=r+{Z}_{R}-{k}_{R}\times X+{C}_{R}\times (V-R),\end{eqnarray} \tag{ 3 }$$

$$\begin{eqnarray}&&I=i+{Z}_{I}-{k}_{I}\times X+{C}_{I}\times (V-I),\end{eqnarray} \tag{ 4 }$$

where bvri are the instrumental magnitudes; BVRI are the standard magnitudes; Z_B, Z_V, Z_R, and Z_I are the zero-point magnitudes; k_B, k_V, k_R, and k_I are the extinction coefficients; X is the airmass; and C_B, C_V, C_R, and C_I are the color terms. We adopted the same extinction coefficients from CTIO's calibration pages in Table 2. The Rubin 149 field was observed for 60 photometric nights, and the color term parameters in Table 2 were determined by calibrating the Landolt standard stars in Rubin 149.

We compared the BVRI-band photometry of SN 2017cbv observed with ANDICam to the data of Wee et al. (2018), the CSP II program (Burns et al. 2020), and the 1 m data taken with the Las Cumbres Observatory Supernova Key Project and Global Supernova Project (Hosseinzadeh et al. 2017c); see Figure 2. Our BV-band photometry is consistent with CSP II and Hosseinzadeh et al. (2017c) within 0.05 mag, while our BVRI-band light curves are systematically fainter by 0.101 ± 0.042, 0.156 ± 0.027, 0.108 ± 0.018, and 0.152 ± 0.023 mag relative to the same filter in Wee et al. (2018). We double-checked our photometry by employing two independent methods simultaneously (aperture and PSF photometry), while only aperture photometry was used in Wee et al. (2018). Our aperture photometry is corrected by the aperture size. The aperture correction was measured on bright isolated stars between 15'' and 7'' and applied to 7'' aperture photometry. This is the same method used to measure standard stars by Landolt (2009). Our calibration involved independent observations of local-sequence stars taken from the CSP II project, while Wee et al. (2018) took observations of standard star fields for calibration and target fields with the same instrument. Our BV-band photometry is most consistent with that from CSP II (Burns et al. 2020) and Hosseinzadeh et al. (2017c).

Zoom In Zoom Out Reset image size

The ANDICam NIR raw images have been binned on-site at CTIO using an IRAF script and uploaded to the Yale Repository. For the binned NIR images, we first applied flat-field correction and cosmic-ray rejection. Then, a sky frame was subtracted from each image using neighboring images; if the three dithered science images (a, b, and c) were taken one by one with the same exposure time, the differences of two neighboring images (a-b, b-a, and c-b) were taken as the sky-subtracted images. We measured the dither offsets relative to the first frame by picking up a bright source on the dithered science images using skycat²⁹ and took the measured dither offsets as the initial values for all science frames. Stars were extracted on the image after the initial dither offset correction (Bertin & Arnouts 1996), and their pixel positions were matched to get additional shifts. The images with corrections for the dither offset- (initial + extra) corrected frames were then combined to create a coadded image, which can be used to perform photometry simultaneously via source extraction and photometry (SEP; Barbary 2018) and the PSF Extractor (PSFEx; Bertin & Arnouts 1996; Bertin 2011) after astrometric calibration. For the NIR images, WCS information was added manually to one reference image, and then all other images were aligned with the reference using the Python module Astroalign (Beroiz et al. 2020).

As we did for the optical images, we performed differential photometry of SN 2017cbv relative to neighbor stars 1 and 3 (∼3 mag fainter than 1; see Figure 1). We did not take star 2 as the local star because it is too close to the image's edge. We found that PSF photometry performs better than aperture photometry. We tested this by comparing the variance of the difference between stars 1 and 3, which is smaller for the PSF than aperture photometry in all filters. We also noted an intrinsic dispersion of the order of 0.025 mag for the PSF difference distribution, which we account for as an instrumental noise component, and add it to the photometry error budget. As star 1 is brighter and there is no time-series variability during our observing period, we finally adopted the differential PSF photometry of SN 2017cbv relative only to star 1.

We found systematic differences for the nightly zero-points when calibrating the three standard fields (RU149, P9144, and LHS2397a) during photometric nights. So we took the JHK_s photometry of star 1 from the Two Micron All Sky Survey (2MASS; Cutri et al. 2003a, 2003b; Skrutskie et al. 2006) to avoid a calibration problem. Its Y-band magnitude was derived from the relationship between Y − J and J − H colors in Hodgkin et al. (2009). The YJHK_s magnitudes of star 1 are listed in Table 4.

We also compare the YJHK_s-band photometry with that from Wee et al. (2018) in Figure 3. The YJK_s-bands are consistent, the differences being −0.063 ± 0.040, −0.014 ± 0.049, and −0.035 ± 0.078 mag, respectively. For the H band, we found a systematic difference of −0.101 ± 0.032 mag.

Zoom In Zoom Out Reset image size

Six NIR spectra were taken with FLAMINGOS-2 (Eikenberry et al. 2006) on the Gemini South 8.2 m telescope at the Gemini South Observatory, including the earliest one at only 2.30 days past explosion. The FLAMINGOS-2 spectra were acquired in long-slit mode with the JH grism and filter in place, along with a slit width of 072. This setup yielded wavelength coverage of 1.0–1.8 μm and a resolution of R ∼ 1000. The long-slit spectra were acquired at the parallactic angle with a standard ABBA pattern and reduced in a standard way using the F2 PyRAF package provided by Gemini Observatory. The XTELLCOR pipeline was used to perform telluric corrections and flux calibrations. More details of the data reduction process can be found in Brown et al. (2019) and Hsiao et al. (2019).

Two NIR spectra were provided by the ePESSTO collaboration (Smartt et al. 2015) with the SOFI instrument (Moorwood et al. 1998) on the 3.6 m NTT at La Silla Observatory. The SOFI spectra were observed with blue and red grisms with a slit width of 10, wavelength coverage of 0.9–2.5 μm, and a resolution of R ∼ 500. The conventional ABBA nod-along-the-slit technique was adopted. For each SOFI spectrum, a Vega-like or solar analog was also observed for flux calibration. The SOFI spectra were reduced by performing the following steps: cross-talk correction, flat-field correction, wavelength calibration, sky subtraction, spectral extraction, telluric absorption correction, and flux calibration (Smartt et al. 2015).

Five NIR spectra were observed with FIRE (Simcoe et al. 2013) on the 6.5 m Magellan Baade telescope at Las Campanas Observatory. The FIRE spectra were acquired with the high-throughput prism mode with a slit of 06, wavelength coverage of 0.8–2.5 μm, and a similar resolution as SOFI. For each epoch, the conventional ABBA nod-along-the-slit technique and the sampling-up-the-ramp readout mode were used. Meanwhile, an A0V star was observed for telluric correction close in time, angular distance, and airmass to our science target (Vacca et al. 2003; Hsiao et al. 2015, 2019). The IDL pipeline firehose (Simcoe et al. 2013) was specifically developed to reduce FIRE spectra.

One SpeX (Rayner et al. 2003) spectrum was taken with the 3.0 m NASA IRTF at the summit of Maunakea. The SpeX spectrum was obtained in cross-dispersed mode with a slit of 05, yielding a wavelength range of 0.8–2.5 μm (Hsiao et al. 2019). The data were reduced with the IDL code Spextool (Cushing et al. 2004), which is designed for handling the SpeX data. The flux calibration process for the SpeX spectrograph is similar to that of other devices, such as FIRE, SOFI, and FlAMINGOS-2.

Figure 4 shows the NIR spectra obtained for SN 2017cbv, covering the phases from −17.6 to +49.4 days relative to ${t}_{B}^{\max }$. The early NIR spectra are dominated by electron scattering with a well-defined photosphere. Thus, the first three early spectra in Figure 4 show relatively featureless blue continua. Later, spectral features develop at 1.05, 1.25, and 1.65 μm when the SN is close to B-band maximum (Wheeler et al. 1998). The most prominent features at those epochs are associated with intermediate-mass species, i.e., O i, Mg ii, and Si iii (see also Figure 2 in Hsiao et al. 2013). Specifically, the strong and relatively isolated absorption feature at 1.05 μm was identified as Mg ii λ1.0927 μm by Wheeler et al. (1998), Hamuy et al. (2002a, 2002b) in SN 1999ee, Gall et al. (2012) in SN 2005ef, and Hsiao et al. (2013) in SN 2011fe. The emission feature at 1.25 μm was identified as Si iii by Hsiao et al. (2013) and Fe iii by Hamuy et al. (2002b) and Rudy et al. (2002). A strong feature around 1.6 μm was identified as Fe iii by Hsiao et al. (2013), Mg ii/Si ii/Co ii by Marion et al. (2009), and Si ii by Wheeler et al. (1998) and Gall et al. (2012). At longer wavelengths, the spectrum is also featureless, except for a weak emission feature around 2.05 μm perhaps due to Si iii (Wheeler et al. 1998).

Zoom In Zoom Out Reset image size

Figure 5 displays the comparison between SNe 2017cbv and 2011fe at −17.6, −11.5, −2.6, and +4.4 days relative to the B-band maximum (Hsiao et al. 2013). Both spectra are matched very well, except for the Mg ii λ1.0927 μm absorption feature, which is very weak in SN 2017cbv at t = −17.6 and −11.5 days.

Zoom In Zoom Out Reset image size

Two weeks past maximum, the spectra showed dramatic evolution dominated by strong emission/absorption features. The Mg ii λ1.0927 μm absorption feature disappeared, and the most remarkable features were the strong and wide peaks around 1.5–1.7 μm, which can be attributed to blends of Co ii, Fe ii, and Ni ii (Wheeler et al. 1998). Meanwhile, new peaks at around 2.2 and 2.4 μm appeared and gradually developed that are mainly attributed to iron-group element Co ii (Wheeler et al. 1998). Figure 6 shows the postmaximum comparisons of SNe 2017cbv, 2011fe (Hsiao et al. 2013), 2012fr (Hsiao et al. 2019), and 2014J (Sand et al. 2016) at comparable phases. They are very consistent with each other at +16.3, +34.4, +36.3, +36.5, and +49.4 days after maximum light. Two spectra were taken at +49.4 days on May 17, one by FIRE and the other by SOFI. We adopted the spectrum observed by FIRE, as the colors from the FIRE spectrum are closer to the NIR-band photometry and the S/N is higher in Figure 4.

Zoom In Zoom Out Reset image size

A product of explosive carbon burning, Mg ii is a sensitive probe of the location of the inner edge of carbon burning (Wheeler et al. 1998; Marion et al. 2001, 2009; Höflich et al. 2002; Hsiao et al. 2013) in velocity space. The Mg ii λ1.0927 μm line is expected to be observed with decreasing velocity in the early spectral evolution and then to remain at an almost constant velocity when the photosphere has receded below the inner edge of the Mg ii distribution. We can only measure the absorption minimum of the Mg ii λ1.0927 μm line at t = −11.5, −6.6, −2.6, and +4.4 days, with an almost constant velocity of ∼10,500 km s⁻¹. This suggests that its photosphere had receded below the inner edge of magnesium, according to the analysis by Wheeler et al. (1998). Alternatively, Meikle et al. (1996) interpreted the constant velocity of Mg ii as a detached feature.

Unburned carbon provides the most direct diagnostic of the primordial material from the progenitor. The weak-absorption carbon feature was mainly detected from early optical spectra via C ii λ0.6580 μm (Thomas et al. 2007; Scalzo et al. 2010; Silverman et al. 2011; Silverman & Filippenko 2012; Taubenberger et al. 2011; Thomas et al. 2011; Zheng et al. 2013). Alternatively, NIR C i λ1.0693 μm can be taken as a superior carbon tracer compared with the optical C ii λ0.6580 μm, as C i can be detected at maximum light (e.g., Höflich et al. 2002). Hosseinzadeh et al. (2017c) detected a strong C ii λ0.6580 μm feature at t = −19 days, similar to SN 2013dy at t = −16 days (Zheng et al. 2013), although the carbon feature disappeared by day −13 for SN 2017cbv. Very strong carbon C ii was also seen in iPTF 16abc (Miller et al. 2018). We tried to detect the C i from our NIR spectra, and we saw a notch close to λ1.0693 μm near 1.03 μm taken on 2017 April 2, or t ∼ 4.4 days after B-band maximum. We applied the automated spectrum synthesis code SYNAPPS (Thomas et al. 2011) to identify C i λ1.0693 μm, as shown in Figure 7. The blueshift of the C i line (green dashed line in Figure 7) was observed at 11,000 km s⁻¹ at t ∼ 4.4 days after B-band maximum. The velocity of the unburned carbon in the NIR spectrum was consistent with the velocity of Mg ii λ1.0927 μm at the same epoch in Section 2.3.3. If the detected C i is real, SN 2017cbv could be the second case to support the hypothesis that a change in the ionization condition occurs as the temperature cools, indicating that the signature of C ii λ0.6580 μm appears in the very early phase before B-band maximum and C i λ1.0693 μm appears later, i.e., around maximum, as similarly reported by Hsiao et al. (2013) for SN 2011fe. Note that the detection of C i λ1.0693 μm in Figure 7 is interpreted by SYNAPPS (Thomas et al. 2011), which is independent of the very early optical detection of C ii λ0.6580 μm (Hosseinzadeh et al. 2017c).

Zoom In Zoom Out Reset image size

Maeda et al. (2014) emphasized that the Paβ in postmaximum NIR spectra can provide a powerful diagnostic of the presence of unbound hydrogen-rich matter expelled from a companion. Hydrodynamic and radiative transfer models in Maeda et al. (2014) found that the postmaximum Paβ is easily observed, and this feature grows stronger at ∼1–2 months after maximum, covering a range of viewing angles between the observer, SN, and companion star. Sand et al. (2016) tried to identify the Paβ line for the nearby Type Ia SN 2014J. They found no evidence for the presence of Paβ emission after comparing the observed spectra around Paβ λ1.282 μm with the red giant scenario corresponding to 0.3, 0.1, and 0.03 M_⊙ of hydrogen for the boundary cases: θ = 0° and 180° (Maeda et al. 2014). Thus, Sand et al. (2016) gave a rough hydrogen mass upper limit of 0.1 M_⊙ for all SN–companion star orientations and claimed that it was not distinguishable between the scenario with hydrogen masses of 0.03 M_⊙ and observations. We have compared our postmaximum NIR spectra of SN 2017cbv with those of SNe 2011fe, 2012fr, and 2014J at several phases of t = +16.3, +34.4, +36.3, +36.5, and +49.4 days, as shown in Figure 8. We can see that our five postmaximum spectra are well matched with those of SN 2014J, and both have comparable signal-to-noise ratios in Figure 8. No Paβ lines were detected from our five postmaximum NIR spectra by visual inspection, and this yields a similar hydrogen mass limit of less than 0.1 M_⊙ from the companion star of SN 2017cbv, although the limit depends on the viewing angles. Analysis of the Paβ line of SN 2017cbv using the same NIR spectrum at 34 days after maximum was performed in Hosseinzadeh et al. (2017c), and they drew similar conclusions. Nondetection of H_α from nebular spectroscopy gave an even lower hydrogen mass limit (Sand et al. 2018).

Zoom In Zoom Out Reset image size

Figure 9 shows the BVRIYJHK_s-band light curves of SN 2017cbv from our observations (also see Appendix Table 11 for optical and Appendix Table 12 for NIR data). The light curves were sampled during the period t ∼ −16 to +125 days relative to B-band maximum, making SN 2017cbv one of the best-observed SNe Ia in the optical and NIR bands simultaneously. The morphology of the light curves resembles that of normal SNe Ia, showing a shoulder in the R band and a pronounced secondary maximum in the I and NIR bands. The NIR light curves of SN 2017cbv reached their first peak ∼4 days earlier than the B-band curve, consistent with the statistical analysis of an SN Ia sample (Dhawan et al. 2015).

Zoom In Zoom Out Reset image size

Gaussian process regression (GPR) was applied via the Python module Scikit-learn (Pedregosa et al. 2011) to estimate the light-curve shape parameter Δm₁₅(B) = 0.990 ± 0.013 mag and the peak time of the B-band light curve ${t}_{B}^{max}\,=57,840.87\pm 0.10$ MJD, which are used throughout this paper. The value Δm₁₅(B) is the B-band magnitude difference between the peak B_t=0 = 11.710 ± 0.006 mag and 15 days after B_t=15 = 12.700 ± 0.011 mag. The GPR is a nonparametric, Bayesian approach to regression in the area of machine learning (Rasmussen 2006). Similarly, we implemented GPR to fit the phases and maximum peak magnitudes of the VRIYJHK_s bands. When possible, we also use the same GPR to deduce the phases and peak magnitudes of the secondary maximum and the minimum magnitude between the two maxima for each of the IYJHK_s-band light curves. We follow the nomenclature adopted by Biscardi et al. (2012) to parameterize the light curves of SN 2017cbv. For the X band, at the phases of t1(X), t2(X), and t0(X) relative to B maximum, the first maximum m1(X), the secondary maximum m2(X), and a minimum m0(X) between the two maxima are reached. The uncertainties of the phases were measured using a jackknife procedure. For example, we select N points around the maximum t1 and then start a loop, taking one data point out and fitting with the GPR to the rest of the N − 1 data points. An array of N measurements of t1 is obtained once the loop is completed. Then we do statistics with this array to get the standard deviation σ_t1, and the final uncertainty of t1 will be ${\sigma }_{{\text{}}t1}\times \sqrt{N}$. We repeat the above process to obtain the uncertainties of t1, t2, and t0 for each band. The light-curve parameters m1, m2, and m0 and their times relative to t_Bmax (t1, t2, and t0), as well as the decay rate β between 40 and 90 days, are tabulated in Tables 5 and 6. As shown in Elias et al. (1981), the IYJHK_s light curves show the first maximum within −2 to −5 days of the B-band maximum, which is consistent with the results for SNe Ia with Δm₁₅ < 1.8 (Folatelli et al. 2010) and in agreement with the statistical studies in the YJH bands (Dhawan et al. 2015). Recent studies suggested that the timing of the i-band maximum indicates the physical state of the SN Ia explosion (González-Gaitán et al. 2014; Ashall et al. 2020). Thus, the time of i-band maximum can be used to subclassify Ia into normal, 91T-like, 03fg-like, 91bg-like, and 02cx-like objects (Figure 3 and Table 1 of Ashall et al. 2020), of which, SN 2017cbv locates in the normal subtype.

Table 5.  Light-curve Parameters of SN 2017cbv

Filter	m(${t}_{B}^{\max })$	t1a	m1	t0	m0	t2	m2	AMilky Way
	(mag)	(days)	(mag)	(days)	(mag)	(days)	(mag)	(mag)
B	11.710 ± 0.006	0.00 ± 0.10	11.710 ± 0.006	⋯	⋯	⋯	⋯	0.615
V	11.643 ± 0.007	1.24 ± 0.10	11.637 ± 0.007	⋯	⋯	⋯	⋯	0.453
R	11.607 ± 0.008	0.66 ± 0.17	11.605 ± 0.008	⋯	⋯	⋯	⋯	0.358
I	11.840 ± 0.008	−2.99 ± 0.14	11.793 ± 0.007	15.32 ± 0.22	12.472 ± 0.012	26.98 ± 0.11	12.312 ± 0.012	0.256
Y	⋯	−4.74 ± 0.09	12.050 ± 0.020	⋯	⋯	⋯	⋯	0.175
J	12.004 ± 0.018	−3.57 ± 0.19	11.883 ± 0.015	16.74 ± 0.77	13.613 ± 0.024	31.66 ± 0.69	13.206 ± 0.018	0.122
H	12.180 ± 0.018	−4.72 ± 0.13	12.027 ± 0.016	7.79 ± 0.20	12.343 ± 0.016	25.96 ± 0.29	12.085 ± 0.016	0.078
Ks	11.938 ± 0.017	−3.07 ± 0.27	11.877 ± 0.015	12.77 ± 0.35	12.252 ± 0.017	25.33 ± 0.44	12.064 ± 0.016	0.052

Note.

^aThe t1, t2, and t0 are phases of the first maximum m1, the secondary maximum m2, and the minimum m0 between the two maxima relative to B maximum.

Download table as:  ASCIITypeset image

Dhawan et al. (2015) also studied the YJH-band light curves of 91 SNe Ia from the literature and made an extensive statistical analysis for NIR light-curve shape parameters, i.e., t1, t0, and t2, and late-time decay β in these three bands. We found that the phases t1, t0, and t2 for SN 2017cbv in Table 5 for each band are consistent with values reported by Dhawan et al. (2015). As predicted by Kasen (2006), a larger Ni mass of SN leads to higher temperatures and thus a later $2\to 1$ recombination wave occurring around 7000 K, which tends to delay the secondary maximum. The epoch of secondary maximum t2 was slightly delayed for SN 2017cbv, perhaps indicating a larger Ni mass, when compared with SN 2011fe, which shows an earlier secondary maximum. The SN 2011fe has Δm₁₅ = 1.18 mag (Zhang et al. 2016), which also indicated a lower Ni mass.

After the (secondary) maximum, the light curves of SNe Ia usually show a linear decline in magnitude. We also calculated the decay rate β for SN 2017cbv in the BVRIYJHK_s bands during the phases 40 days < t < 90 days. The values of the decay rate in different bands are listed in Table 6. Our optical decline rates β are consistent with those reviewed by Leibundgut (2000). In the early nebular phase, the NIR-band light curves of SN 2017cbv are found to have faster decay rates than the optical ones, consistent with the results reported by Dhawan et al. (2015). As noted by Dhawan et al. (2015), SNe Ia tend to have similar late-time decay rates in the NIR bands. At late time, the SN gradually becomes transparent to the γ rays produced by radioactive decays. Similar late-time decay rates perhaps suggest a similar internal structure of the explosions, which is also in agreement with the predictions for Chandrasekhar-mass models (Woosley et al. 2007) that produce different nickel masses but with similar radial distributions of iron-group elements.

The Milky Way extinction in the BVRIJHK_s bands toward SN 2017cbv was derived from the dust maps of Schlafly & Finkbeiner (2011) in individual bandpasses, CTIO B, V, R, and I and 2MASS J, H, and K_s, which are similar to the ANDICam filters used in the paper. The extinction values are listed in the last column of Table 5.³⁰ The Galactic reddening toward SN 2017cbv is E(B − V) = 0.162 mag (Schlafly & Finkbeiner 2011).

The Y-band extinction A_Y = 0.175 was estimated by considering the mean R_V-dependent extinction law, and we refer to Equations (1), (2a), and (2b) in Cardelli et al. (1989),

$$\begin{eqnarray}&&\langle A(\lambda )/{A}_{V}\rangle =a(x)+b(x)/{R}_{V},\end{eqnarray} \tag{ 5 }$$

$$\begin{eqnarray}&&a(x)=0.574{x}^{1.61},\end{eqnarray} \tag{ 6 }$$

$$\begin{eqnarray}&&b(x)=-0.527{x}^{1.61},\end{eqnarray} \tag{ 7 }$$

where x is the reciprocal of the Y-band central wavelength at λ = 1.03 μm (Hillenbrand et al. 2002), the ratio of total to selective extinction is R_V = 3.1, and the V-band extinction toward SN 2017cbv is A_V = 0.453 (Schlafly & Finkbeiner 2011).

The Na i D lines in the high-resolution spectrum provide an independent measurement of both the Milky Way and host reddening. Burns et al. (2020) published one high-resolution spectrum of SN 2017cbv using the Magellan Inamori Kyocera Echelle (MIKE; Bernstein et al. 2003) in their Figure 5. The Na i D lines can be seen toward the Milky Way, but no Na i D lines are seen toward the host galaxy. A higher Milky Way reddening of E(B − V) = 0.23 ± 0.16 mag was obtained based on the equivalent width (EW) of the Na i D lines (Burns et al. 2020). This may further suggest that SN 2017cbv suffers some Milky Way reddening but negligible host reddening. Ferretti et al. (2017) also published five high-resolution spectra of SN 2017cbv with the Ultraviolet and Visual Echelle Spectrograph (UVES; Dekker et al. 2000) and found low values of EW for the Na i (D1 and D2) lines, consistent with negligible host reddening.

Figure 10 shows comparisons of the optical light curves of SN 2017cbv with those of well-observed normal SNe Ia, i.e., SN 2001el (Δm₁₅ = 1.15 mag; Krisciunas et al. 2003), SN 2002dj (Δm₁₅ = 1.08 mag; Pignata et al. 2008), SN 2003du (Δm₁₅ =1.02 mag; Stanishev et al. 2007), SN 2004S (Δm₁₅ = 1.10 mag; Krisciunas et al. 2007), SN 2005cf (Δm₁₅ = 1.07 mag; Wang et al. 2009b), SN 2011fe (Δm₁₅ = 1.18 mag; Zhang et al. 2016), SN 2012cg (Δm₁₅ = 0.86 mag; Marion et al. 2016), SN 2012fr (Δm₁₅ = 0.82 mag; Zhang et al. 2014; Contreras et al. 2018), and SN 2014J (Δm₁₅ = 1.08 mag; Foley et al. 2014; Marion et al. 2015; Srivastav et al. 2016; Li et al. 2019a). The comparison sample includes all available normal SNe Ia that have been well observed in the optical/NIR bands and have similar light-curve shapes as SN 2017cbv. It can be seen that the near-maximum-light curves of SN 2017cbv are very similar to the comparison sample. The late-time decay rate during the interval t = 40–90 days after the peak denoted as β here was estimated for the BVRIYJHK_s bands,³¹ and the corresponding values are listed in Table 6. The BVRI-band late-time decay rates β of SN 2017cbv appear relatively slower than or similar to those of the corresponding rates of the comparison sample. More details can be seen in Table 6 and Figure 10. During 20–90 days after B maximum, the B magnitude of SN 2017cbv falls in between that of SN 2012fr and SN 2011fe; this suggests that these SNe Ia have different ⁵⁶Co hard-gamma-ray escaping ratios from the ejecta.

Zoom In Zoom Out Reset image size

Table 6.  The Decay Rate β of the Comparison Sample in the BVRI Bands

Name	βBa	βV	βR	βI
	(mag(100 days)−1)	(mag(100 days)−1)	(mag(100 days)−1)	(mag(100 days)−1)
SN 2017cbv	1.442 ± 0.057	2.555 ± 0.033	3.139 ± 0.037	4.029 ± 0.049
SN 2005cf	1.663 ± 0.050	2.639 ± 0.036	3.215 ± 0.035	4.381 ± 0.061
SN 2011fe	1.397 ± 0.004	2.762 ± 0.005	3.249 ± 0.023	4.236 ± 0.047
SN 2012fr	1.639 ± 0.015	2.683 ± 0.031	⋯	⋯
SN 2014J	1.334 ± 0.024	2.852 ± 0.042	3.355 ± 0.064	4.248 ± 0.099
SN 2001el	1.493 ± 0.097	2.638 ± 0.072	3.221 ± 0.094	4.244 ± 0.059
SN 2004S	1.608 ± 0.114	2.704 ± 0.107	3.185 ± 0.094	3.773 ± 0.064
SN 2003du	1.707 ± 0.033	2.626 ± 0.043	3.094 ± 0.066	4.577 ± 0.127
	βYa	βJ	βH	${\beta }_{{K}_{s}}$
SN 2017cbv	5.288 ± 0.053	6.033 ± 0.121	4.092 ± 0.046	4.034 ± 0.173
SN 2012fr	5.345 ± 0.041	6.167 ± 0.110	4.133 ± 0.039	⋯

Note.

^aThe late-time decay rate β of the light curve during the interval t = 40–90 days relative to ${t}_{B}^{\max }$.

Download table as:  ASCIITypeset image

In Figure 11, the YJHK_s-band light curves of SN 2017cbv are compared with those of SNe 2001el, 2002dj, 2003du, 2004S, 2005cf, 2011fe, 2012cg, 2012fr, and 2014J. The overall light curves of SN 2017cbv in the YJHK_s bands resemble the comparison SNe in Figure 11 and Table 6. One can see that their secondary maximum features show some differences, and these variations might be related to the progenitor metallicity, the concentration of iron-group elements, and the abundance stratification in SNe Ia (Kasen 2006). After t > 90 days, the light-curve decay slope in the NIR bands becomes less steep; this could be influenced by the Co ii, like mid-IR Co ii λ10.5 μm time-series variation, flattening out at about day 90–100 (Telesco et al. 2015).

Zoom In Zoom Out Reset image size

A comparison of several well-observed SNe is shown in Figures 12–14. All photometry has been corrected for reddening in the Galaxy and the host galaxies by the values from the corresponding published papers, except for SN 2017cbv, for which only the Galactic extinction was corrected. The optical color curves of SN 2017cbv and the comparison sample (B − V, V − R, and V − I) are presented in Figure 12. At very early phases, t < −10 days with respect to B-band maximum, the colors of SN 2017cbv are much bluer than the comparison SNe (also see Hosseinzadeh et al. 2017c; Stritzinger et al. 2018; Bulla et al. 2020). The B − V color of SN 2017cbv stays flat at about −0.05 mag soon after explosion and then slowly becomes bluer until day −5. Also, it is the bluest SN in colors B − V, V − R, and V − I until day −11. The V − R color is even bluest among all comparison SNe Ia until 10 days after B maximum. The blue colors seen in the early light curves of some SNe Ia have been interpreted as interactions between SN ejecta and a companion star, serving as evidence in favor of the SD scenario (Brown et al. 2012; Marion et al. 2016; Hosseinzadeh et al. 2017c; Dimitriadis et al. 2019a). After B-maximum light, the color evolution of SN 2017cbv matches well with the comparison sample and the Lira–Phillips relation (blue solid line; Phillips et al. 1999) at about 30 days past maximum. For a nearby SN Ia sample in the Lira law regime, Förster et al. (2013) claimed that the B − V slope −0.013 mag day⁻¹ can be used to classify the faster and slower decliners. Faster decliners (<−0.013 mag day⁻¹) have a higher EW of Na i D lines, redder colors, and a lower R_V reddening law at maximum light, suggesting the presence of circumstellar material (Wang et al. 2009a, 2013, 2019; Förster et al. 2013), while slower decliners are the opposite. The slope of SN 2017cbv in the Lira phase is −0.010 mag day⁻¹; thus, it is a slower decliner. This is consistent with the very small EW measurements of Na i absorption lines for SN 2017cbv (Ferretti et al. 2017) and blue B − V color at maximum light.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

We also compare the V − NIR color evolution (V − J, V − H, and V − K_s) of SN 2017cbv and the comparison SNe Ia in Figure 13. We can see that the V − NIR color evolution of SN 2017cbv matches with the comparison sample, especially the well-observed SN 2011fe. The study by Burns et al. (2014) derived empirical relations of intrinsic colors V − J and V − H at maximum light relative to the light-curve shape parameter s_BV in their Table 2. According to the relations for their low-reddening sample (LRS), and assuming an s_BV = 1.11 for SN 2017cbv (Burns et al. 2020), we obtained intrinsic colors (V − J)_host^max = −0.61 ± 0.08 and $(V-H{)}_{\mathrm{host}}^{\max }=-0.85\pm 0.09$ mag. Meanwhile, Table 5 gives the observed colors $(V-J{)}_{\mathrm{host}}^{\max }\,=-0.60\pm 0.02$ and $(V-H{)}_{\mathrm{host}}^{\max }=-0.80\pm 0.02$ mag at maximum epochs. Thus, we can derive $E(V-J{)}_{\mathrm{host}}^{\max }=0.01\,\pm 0.08$ and $E(V-H{)}_{\mathrm{host}}^{\max }=0.05\pm 0.09$ mag. This indicates that SN 2017cbv suffers neglected host reddening.

The J − H and H − K_s color evolutions of SN 2017cbv and the comparison SNe Ia are shown in Figure 14. Overplotted are the phases of the first maximum t1 = −3.6 days, the secondary maximum t2 = 31.7 days, and the minimum between the two maxima t0 = 16.7 days in the J band relative to B-band maximum, as shown by the vertical dashed lines. As seen from the top panel of Figure 14, the J − H color illustrates a pronounced evolution after t1. The flux in the J band decreases obviously with respect to the H band shortly after t1, probably due to the lack of emission features around 1.2 μm (Spyromilio et al. 1994; Hoflich et al. 1995; Wheeler et al. 1998). This trend holds until t2, when the Fe ii λ1.25 μm emission line forms. Then, the J − H color becomes redder again as a result of a faster decline rate in the J band. The bottom panel of Figure 14 displays the H − K_s plot, and SN 2017cbv matches well with the comparison SNe.

Figure 15(a) shows the CMD presented by Wang et al. (2003, 2006). In order to estimate the CMAGIC color excess E_BV(B − V), we use the quantity as defined in Wang et al. (2003),

$$\begin{eqnarray}&&\epsilon (B-V)=({B}_{\max }-{B}_{{BV}})/{\beta }_{{BV}},\end{eqnarray} \tag{ 8 }$$

where B_max is the B-band maximum, and β_BV and B_BV denote the slope and the value for the intercept at (B − V) = 0 for the linear region from 5 to 27 days after the B-band maximum with the CMAGIC relation (Wang et al. 2003):

$$\begin{eqnarray}&&B={B}_{{BV}}+{\beta }_{{BV}}(B-V).\end{eqnarray} \tag{ 9 }$$

The CMAGIC color excess for a reddening-free SN ₀ also depends on the light-curve shape parameter Δm₁₅(B), and the linear relationship between them was derived from a low-extinction sample of SNe Ia (Phillips et al. 1999; Wang et al. 2003):

$$\begin{eqnarray}&&{\epsilon }_{0}=(-0.118\pm 0.013)+(0.249\pm 0.043)({\rm{\Delta }}{m}_{15}(B)-1.1).\end{eqnarray} \tag{ 10 }$$

The final color excesses of SNe Ia can be measured based on the equation

$$\begin{eqnarray}&&{E}_{{BV}}(B-V)=\epsilon (B-V)-{\epsilon }_{0}.\end{eqnarray} \tag{ 11 }$$

We fit the B magnitude and B − V color of SNe 2017cbv and 2011fe between 5 and 27 days, and we derived E_BV(B − V)_11fe = 0.030 ± 0.044 and E_BV(B − V)_17cbv = 0.173 ± 0.029 mag, which includes both the contribution from the Milky Way and the SN's host galaxy. We can take SN 2011fe as a reference with no dust extinction from the Milky Way and its host galaxy (Nugent et al. 2011; Johansson et al. 2013; Patat et al. 2013). The SN 2017cbv has negligible host extinction, E(B − V)_host = 0.011 ± 0.029 mag, after considering the Milky Way extinction toward SN 2017cbv (Schlafly & Finkbeiner 2011).

The distance of the two SNe can also be derived based on the CMAGIC method by comparing the observed B_BV and the absolute value ${M}_{{BV}}^{B}$ from the empirical relation between the absolute magnitude ${M}_{{BV}}^{B}$ versus Δm₁₅ in Figure 10 of Wang et al. (2003). Table 3 of Wang et al. (2003) gives the fit result to the absolute magnitude ${M}_{{BV}}^{B}$ versus ${\rm{\Delta }}{m}_{15}$ relation for R_B = 3.3, expressed as

$$\begin{eqnarray}&&{M}_{{BV}}^{B}=(-19.35\pm 0.02)+(0.60\pm 0.08)({\rm{\Delta }}{m}_{15}(B)-1.1).\end{eqnarray} \tag{ 12 }$$

The extinction correction A_BV is given by Equation 3(a) of Wang et al. (2003), expressed as

$$\begin{eqnarray}&&{A}_{{BV}}=({R}_{B}-{\beta }_{{BV}})E(B-V),\end{eqnarray} \tag{ 13 }$$

where R_B = 3.3, β_BV = 2.250 ± 0.030, and E(B − V)_17cbv(host) = 0.011 ± 0.029 mag. The observable value B_BV is the color–magnitude intercept, which is calculated from the intercept of the typical color B − V = 0.6 mag reported by Wang et al. (2003). To minimize the covariance of the distance estimates to the slope β_BV, Wang et al. (2003) defined the following equation to measure B_BV as the standard color–magnitude intercept:

$$\begin{eqnarray}&&{B}_{{BV}}={B}_{{BV}0.6}-1.164.\end{eqnarray} \tag{ 14 }$$

Equation (14) yields B_BV = 11.648 ± 0.030 mag for SN 2017cbv. Thus, we obtained the distance modulus μ = 30.58 ± 0.05 mag (D = 13.1 ± 0.3 Mpc) after applying ${M}_{{BV}}^{B}$, A_BV, and B_BV in Equation (17) in Section 3.4.1. Based on the same method, we estimated the distance modulus μ = 29.14 ± 0.10 mag for SN 2011fe, which is consistent with the Cepheid distance modulus of SN 2011fe (Shappee & Stanek 2011).

Wang et al. (2003) also found that B magnitudes and the various colors (i.e., B − R, B − I) are linearly related. Thanks to our well-observed optical and NIR photometric observations, we found that the B magnitude versus B − K_s color and V magnitude versus V − K_s color are also linearly related in the phase ranges of 5 days ≤ t ≤ 30 days and t ≤ −5 days. The two linear relations have an intersection point for SN 2017cbv (red dashed lines) and SN 2011fe (black dashed lines), as shown in panels (d) and (j) of Figure 15. If we assume that SN 2017cbv and SN 2011fe are intrinsically the same, matching their corresponding colors should give us the comparable interstellar dust extinctions. There are no first-principle physical insights to help us determine which points on the CMDs are most representative of the intrinsic color of the SN Ia population. To explore the various possibilities, we tested the color excesses E(B − K_s) and E(V − K_s) by matching the reddest color, bluest color, and intersection point between SN 2017cbv and SN 2011fe.

We plot the B versus B − K_s diagram in panels (d)–(i) of Figure 15. Panels (d)–(f) are the observations without extinction corrections of the Milky Way and the host, only corrections of the Milky Way, and corrections of the Milky Way and its host by matching with the intersection point of two dashed lines (black for SN 2011fe and red for SN 2017cbv). One line is the linear fit to the phase interval 5 days ≤ t ≤ 30 days relative to B-band maximum, and the other is the linear fit to the phase range t ≤ −5 days. Thus, we obtained E(B − K_s)_host = −0.05 ± 0.05 mag in panel (f).

Figure 15(g) is the same as Figure 15(d), except that the solid lines are the interpolation values to B and K_s in the phase range of −15 days ≤ t ≤ 60 days after B-band maximum. When matching the bluest color of SN 2017cbv with that of SN 2011fe in panel (h), we obtained E(B − K_s)_host = −0.13 ± 0.05 mag. While matching the reddest color of SN 2017cbv with that of SN 2011fe in panel (i), we obtained E(B − K_s)_host = −0.27 ± 0.05 mag.

Based on these four different assumptions on the uniformity of the intrinsic color, we obtained four different values of E(B − K_s). Among them, matching the reddest color yields the most negative values of E(B − K_s) = −0.27 ± 0.05 mag. This value is inconsistent with the CMAGIC estimates. It would imply a significant overcorrection of the Milky Way reddening if the reddest color is used as the reference point for extinction correction. It is more likely that the difference at the reddest color is intrinsic to the SNe, and, contrary to what has been employed in the Lira–Phillips relation, the late-time color cannot be used as reliable estimates of extinction, at least for these two very well observed SNe.

Similar to the previous section, we also plot V versus V − K_s in panels (j)–(o) of Figure 15. By matching with the intersection point of the two lines for SNe 2017cbv and 2011fe, we obtained E(V − K_s)_host = −0.00 ± 0.05 mag in panel (l). By matching with the bluest and reddest colors, we obtained E(V − K_s)_host = −0.03 ± 0.05 mag in panel (n) and E(V − K_s)_host = −0.08 ± 0.05 mag in panel (o), respectively.

We note further that the intrinsic colors B − K_s and V − K_s of SN 2017cbv are bluer than the expected value of SN 2011fe when matching them with the intersection point, the bluest color, and the reddest color: panels (f) and (l), (h) and (n), and (i) and (o) in Figure 15, respectively. We assumed no host extinction for SN 2011fe based on the reddening analysis and distance determination on our CMAGIC diagram (in Section 3.2.1) and other work by Nugent et al. (2011), Johansson et al. (2013), and Patat et al. (2013).

We also note that the B − V color of SN 2017cbv in panel (c) is bluer than that of SN 2011fe by 0.18 ± 0.07 mag around t ∼ 32 days; at this phase, the two SNe have the reddest color. This suggests that SN 2017cbv and SN 2011fe are different in the B and/or V bands at some late phases. When we matched the reddest colors B − K_s of SN 2017cbv with SN 2011fe, the estimated color excess E(B − K_s) of the host in panel (i) is different from the early-phase estimates (matching with the bluest color or intersection point) by 2σ–3σ in panels (f) and (h). In contrast, the color excess E(V − K_s) of the two SNe by matching their reddest color in panel (o) is comparable with the measurements by matching the bluest color in panel (n) and intersection point in panel (l). This may further suggest that the B magnitudes between SN 2017cbv and SN 2011fe are more different than the V magnitudes at these late phases, although the B and V magnitudes of SN 2017cbv are brighter than those of SN 2011fe after B-band maximum in Figure 10, where their maxima are matched.

Phillips et al. (1999) compiled a set of unobscured SNe Ia and derived a relation between their intrinsic pseudocolor (B_max − V_max)₀ and their decay parameter Δm₁₅(B):

$$\begin{eqnarray}\begin{array}{rcl}{({B}_{\max }-{V}_{\max })}_{0} & = & -0.070(\pm 0.012)+0.114(\pm 0.037)\\ & & \times \ ({\rm{\Delta }}{m}_{15}(B)-1.1).\end{array}\end{eqnarray} \tag{ 15 }$$

This relation allows one to estimate the host reddening suffered by any normal SN Ia by just comparing their measured pseudocolor and intrinsic estimate. After correcting by Milky Way reddening, SN 2017cbv shows a pseudocolor (B_max − V_max) that is even bluer than the expected value from Phillips et al.'s (1999) relation, or E(B − V)_host = −0.006 ± 0.016 mag. This value corresponds to an E(B − K_s) of about −0.019 mag, which is inconsistent with the value estimated by matching the late-time CMD in B versus B − K_s, suggesting again that the late-time colors of SNe Ia can be substantially different.

The top panel of Figure 12 shows the B − V color evolution curve, and the unreddened Lira–Phillips loci is overplotted with a blue solid line. The following relation was derived to describe the intrinsic B − V color evolution in the phase interval 30 days ≤ t_V ≤ 90 days (Lira 1996; Phillips et al. 1999):

$$\begin{eqnarray}&&{(B-V)}_{0}=0.725-0.0118({t}_{V}-60).\end{eqnarray} \tag{ 16 }$$

Applying the relation to SN 2017cbv, we obtain a host extinction of E(B − V)_host = 0.000 ± 0.037 mag.

Wang et al. (2003) derived the CMAGIC relation for SNe Ia over the phase interval 5 days ≤ t_B ≤ 27 days after B-band maximum. We applied the CMAGIC relation to SN 2017cbv in Section 3.2, and we derived the host extinction E(B − V)_host = 0.011 ± 0.029 mag.

By averaging the host extinction of SN 2017cbv based on the above three methods, we obtained E(B − V)_host = 0.002 ± 0.009 mag. The low host galaxy reddening is consistent with the fact that the SN exploded at the outskirts of NGC 5643 and no narrow Na i D absorption lines were detected in the low-resolution spectra, even in the MIKE spectrum (Burns et al. 2020). Ferretti et al. (2017) published five high-resolution spectra of SN 2017cbv and found values of EW for Na i (D1 and D2) at the lower end of the empirical relation between strength of Na i D absorption versus reddening (Poznanski et al. 2012), consistent with zero reddening. Thus, we assume no host galaxy reddening for SN 2017cbv in our study.

The effects of extinction are considerably reduced in the JHK_s bands, and it seems there are relatively constant peak magnitudes in the NIR bands (Meikle 2000; Krisciunas et al. 2004a, 2004b, 2007). The SNe Ia have a more uniform peak luminosity in the NIR bands (Krisciunas et al. 2004a; Wood-Vasey et al. 2008; Folatelli et al. 2010; Matheson et al. 2012; Phillips 2012; Avelino et al. 2019). The well-sampled NIR photometry of SN 2017cbv can be used to determine the distance modulus toward NGC 5643. For each band, the apparent maximum magnitudes m1 and the magnitudes at ${t}_{B}^{\max }$ are listed in the third column of Table 7 and the fourth and second columns, respectively, of Table 5. For each case, we used the following formula to derive the distance modulus μ:

$$\begin{eqnarray}\begin{array}{rcl}\mu & = & m-M+5\mathrm{log}({H}_{0}/72)-{A}_{\mathrm{Milky}\mathrm{Way}}-{A}_{\mathrm{Host}}\\ & & +\ {S}_{\mathrm{Correction}}+{K}_{\mathrm{Correction}}.\end{array}\end{eqnarray} \tag{ 17 }$$

• 1.  
  Here m stands for the apparent magnitudes, M represents the absolute NIR magnitudes from these calibration sources (Krisciunas et al. 2004a; Mandel et al. 2009; Wood-Vasey et al. 2008; Folatelli et al. 2010; Burns et al. 2011; Kattner et al. 2012), and all assume a Hubble constant H₀ = 72 km s⁻¹ Mpc⁻¹ (Freedman et al. 2001; Spergel et al. 2007).
• 2.  
  The Milky Way extinction A_{Milky Way} toward SN 2017cbv was adopted with A_J = 0.122, A_H = 0.078, and ${A}_{{K}_{s}}$ = 0.052 mag (Schlafly & Finkbeiner 2011).
• 3.  
  The host extinction A_Host of SN 2017cbv was assumed to be zero based on the analysis presented in Section 3.3.
• 4.  
  The S_Correction was applied between our JHK_s-band magnitudes on the 2MASS system and the CSP-calibrated magnitude (Contreras et al. 2010) for the calibration sources (Folatelli et al. 2010; Burns et al. 2011; Kattner et al. 2012). They are S_Correction(J) = 0.005, S_Correction(H) = −0.038, and S_Correction(K_s) = 0.009 mag, which are added to the CSP-calibrated magnitudes. The remaining calibration sources have been calibrated to the 2MASS system (Persson et al. 1998), and no S_Correction is necessary.
• 5.  
  No K_Correction has been applied to our photometry due to the close distance of SN 2017cbv.

According to Equation (17), the distance modulus to SN 2017cbv ranges from 30.11 ± 0.11 mag (D = 10.5 ± 0.5 Mpc) to 30.41 ± 0.19 mag (D = 12.1 ± 1.1 Mpc), shown in Figure 16 and listed in Table 7. Note that the uncertainty of each case in Table 7 is dominated by the calibration of the absolute NIR peak magnitude. We note that the same absolute calibrations were applied to the JHK_s magnitudes of SN 2011fe by Matheson et al. (2012), yielding a dispersion of 0.31 mag, very similar to the case of SN 2017cbv.

Zoom In Zoom Out Reset image size

SNooPy is a well-established light-curve fitting method to generate template light curves in the CSP natural system and derive distances to SNe Ia (Burns et al. 2011). Applying SNooPy to our BVRIYJHK_s-band light curve, we obtained a distance modulus of μ = 30.46 ± 0.08 mag (Burns et al. 2014), or distance D = 12.4 ± 0.5 Mpc. This distance estimate should be an independent measurement of CSP (Burns et al. 2020), as we have independent data. When we calibrated the optical using the CSP calibration (Burns et al. 2020), we established some correlation there.

Comparing the measurement of the color–magnitude intercept parameter B_BV to its absolute value in Table 3 of Wang et al. (2003), we obtained μ = 30.58 ± 0.05 mag for SN 2017cbv (D = 13.1 ± 0.3 Mpc).

The CMDs (B versus B − V, B versus B − K_s, and V versus V − K_s in Section 3.2) in Figure 15 give Δμ_17cbv(B) = 1.10 mag relative to SN 2011fe. Assuming a Cepheid distance μ_11fe = 29.04 ± 0.19 mag for SN 2011fe (Shappee & Stanek 2011), we obtained a distance modulus μ_17cbv = 30.14 ± 0.19 mag for SN 2017cbv (D = 10.7 ± 0.9 Mpc).

Burns et al. (2020) applied three methods to the photometric data of SN 2013aa to estimate its distance modulus. They are μ = 30.46 ± 0.08 mag from SNooPy fitting (Burns et al. 2011, 2014), μ = 30.56 ± 0.04 mag from the MLCS2k2 fitter (Jha et al. 2007), and μ = 30.62 ± 0.04 mag from the SALT2 algorithm (Guy et al. 2007), respectively (assuming H₀ = 72 km s⁻¹ Mpc⁻¹). The adopted three methods yield an average estimate of SN 2013aa μ = 30.55 ± 0.08 mag. The SN 2013aa also exploded in the same galaxy with SN 2017cbv, which provides an independent distance determination of SN 2017cbv. The SN 2013aa was discovered by the Backyard Observatory Supernova Survey (BOSS) on 2013 February 13 (Parker et al. 2013) and classified as an SN Ia (Parrent et al. 2013). The SN 2013aa is 74'' west and 180'' south of the core of the host galaxy NGC 5643 (Graham et al. 2017).

In summary, we measured the distance of SN 2017cbv with three methods: NIR absolute calibration, SNooPy fitting, and the CMAGIC diagram. These derived values in Figure 16 are consistent with the results (μ = 30.45 ± 0.09 mag, D = 12.3 ± 0.5 Mpc; Sand et al. 2018) made via the MLCS2K2 fitter (Jha et al. 2007) using independent data and the value μ = 31.14 ± 0.40 mag (D = 16.9 ± 3.1 Mpc) determined from the Tully–Fisher method (Bottinelli et al. 1985; Tully & Fisher 1988). Individually, the distance moduli of SN 2017cbv from NIR absolute calibration (μ = 30.11 ± 0.11 to 30.41 ± 0.19 mag) are smaller than the light-curve template fitters with SNooPy for our data and MLCS2k2 for independent data (Sand et al. 2018), consistent within 2.6σ. The NIR absolute calibration values are also smaller than the CMAGIC diagram, consistent within 2.5σ, except for the smallest value (μ = 30.11 ± 0.11 mag) in the H band, calibrated with Mandel et al. (2009).

Another SN Ia, SN 2013aa, exploded in NGC 5643 and provided an independent distance to NGC 5643 (Burns et al. 2020), consistent with our measurements. Going forward, we adopt our CMAGIC results for the distance to this galaxy (μ = 30.58 ± 0.05 mag) to estimate the quasi-bolometric luminosity, and we compare with theoretical models in the following section.

In calculation of the bolometric luminosity, we adopt the CMAGIC distance modulus of SN 2017cbv, μ = 30.58 ± 0.05 mag, which is close to the external independent distance determination of SN 2013aa (Burns et al. 2020) exploded in the same galaxy. This gives a peak luminosity L_peak = 1.48 ± 0.07 × 10⁴³ erg s⁻¹ and a synthesized nickel mass M_Ni = 0.73 ± 0.03 M_⊙, which is used throughout this paper.

Kasen (2006) attributed the secondary maximum of NIR emission to the ionization evolution of iron-group elements in the ejecta and predicted that the secondary maximum should be delayed for larger Fe masses, indicating that the phase of the secondary maximum should be a function of the nickel mass in the explosion. This was investigated by recent studies (Biscardi et al. 2012; Jack et al. 2012; Dhawan et al. 2015, 2016). Dhawan et al. (2016) studied the correlation between the phase of the secondary maximum t2 and the bolometric peak luminosity and established the L_peak versus t2 relations for the YJH-band light curves. The well-sampled optical and NIR observations of SN 2017cbv presented in this paper offer a good chance to test the above-derived nickel mass using the reddening-free and distance-independent method. The time t2 was measured to be 31.66 ± 0.69 and 25.96 ± 0.29 days from the J- and H-band photometry, and the corresponding peak bolometric luminosities are 1.25 ± 0.17 and 1.11 ± 0.27 × 10⁴³ erg s⁻¹, yielding 0.62 ± 0.08 and 0.55 ± 0.13 M_⊙ nickel masses, respectively. The corresponding value derived from the Y-band data is 0.66 ± 0.10 M_⊙ if we adopt its corresponding secondary maximum of 33.83 ± 0.45 days according to Wee et al. (2018).

We measured the nickel mass of SN 2017cbv using two methodologies that employ the quasi-bolometric light curve and the phases of the secondary maximum in the NIR bands. Those derived values of the nickel mass are listed in Tables 9 and 10, which are consistent with each other within 2σ, and comparable to those of normal SNe Ia such as SN 2005cf (M_Ni = 0.78 M_⊙; Wang et al. 2009b), SN 2011fe (M_Ni ∼ 0.57 M_⊙; Zhang et al. 2016), SN 2012cg (M_Ni ∼ 0.72 M_⊙; Chakradhari et al. 2019), and SN 2012fr (M_Ni ∼ 0.60 M_⊙; Contreras et al. 2018). Our measured nickel mass is matched with some M_ch DDT models and sub-M_ch double detonation models discussed in Section 4, while the BV-band light curves of SN 2017cbv are matched with M_ch DDT model 25 with a nickel mass of 0.6 M_⊙ after B maximum.