A Long-duration Luminous Type IIn Supernova KISS15s: Strong Recombination Lines from the Inhomogeneous Ejecta–CSM Interaction Region and Hot Dust Emission from Newly Formed Dust^*

KISS15s was discovered on 2015 September 18.78 UT (MJD = 57283.78) as a $g\sim 19.6$ mag transient (Figure 1) during the KISS (Morokuma et al. 2014), using the 1.05 m KisoSchmidt Telescope located at KisoObservatory in Japan; the telescope is equipped with a 22 × 22 WFC (the KWFC; Sako et al. 2012). KISS had performed g-band observations for selected sky regions within the Sloan Digital Sky Survey (SDSS) Legacy Survey region (York et al. 2000), and searched for optical transients through the point-spread function (PSF)–matched image subtraction technique, using SDSS g-band mosaic images as the reference frames (see Morokuma et al. 2014, for details of the KISS pipeline). Re-examination of the previous KWFC images revealed KISS15s detection in the image obtained on 2015 September 11.63 UT (MJD = 57276.63), although the PSF profile of KISS15s was partially affected by bad pixels in this first detection image. After the discovery of KISS15s, we carried out follow-up imaging observations with the Kiso/KWFC for g-, r-, and i-bands.

Zoom In Zoom Out Reset image size

We noticed that KISS15s is identical to an optical transient, PS15bva, independently discovered by the Pan-STARRS Survey for Transients (Kaiser et al. 2010).⁹ The discovery date of PS15bva by Pan-STARRS is 2015 August 30 (MJD = 57264.56),¹⁰ 12 days earlier than the first detection by KISS. No spectroscopic follow-up observation for KISS15s = PS15bva has been reported. This work presents the first spectroscopic classification of this object.

KISS15s is located at α = 03^h08^m31640, δ = −00°50'0555 (J2000)¹¹ , 053 west, and 305 north of the center of the star-forming galaxy SDSS J030831.67-005008.6,¹² at a redshift of z = 0.03782 measured by the SDSS. Although the SDSS database assigns a redshift of z = 0.03794 for a point-source object on the northeast side of SDSS J030831.67-005008.6 (Figure 1), this object is probably a foreground Galactic star with an SDSS spectrum contaminated by emission lines from SDSS J030831.67-005008.6. Because KISS15s is located on the edge of the extended structure of SDSS J030831.67-005008.6, we identify SDSS J030831.67-005008.6 as the host galaxy of KISS15s. Follow-up spectroscopy for KISS15s (described in Section 2.2) confirms that the redshift of KISS15s is consistent with the redshift of this galaxy. A luminosity distance (d_L) and distance modulus (DM) of the host galaxy corrected for the Virgo infall + Great Attractor + Shapley supercluster local flow is d_L = 156 Mpc and DM = m − M = 35.97 mag, respectively.¹³ The angular-spatial size scale is 0.720 kpc/''; thus, the galactocentric offset of KISS15s is ∼2.23 kpc. The Galactic extinction toward KISS15s was obtained from the NASA/IPAC Extragalactic Database (NED) as E(B − V) = 0.053 mag (${R}_{V}=3.1;$ Fitzpatrick 1999; Schlafly & Finkbeiner 2011); ${A}_{u},{A}_{g},{A}_{r},{A}_{i},{A}_{z},{A}_{J},{A}_{H}$, and A_K = 0.255, 0.199, 0.137, 0.102, 0.076, 0.043, 0.027, and 0.018 mag, respectively. The Galactic extinction coefficients of WISE W1- and W2-bands are ${A}_{W1}=0.011$ and ${A}_{W2}\,=0.007$ mag, according to the Fitzpatrick (1999) extinction curve.

The KISS data reduction pipeline only provides rough estimates of the g-band magnitude of the transients. To extract the g-, r-, and i-band photometric measurements of KISS15s, we reanalyzed the KISS imaging data, starting from the raw FITS files. Each of the KISS Kiso/KWFC images was obtained with an exposure time of 180 s without dithering. After applying an overscan, bias, and pixel-flat and illumination-flat corrections (see Kokubo 2016, for details of the KWFC data reduction), the images were processed by SExtractor (Bertin & Arnouts 1996) to produce aperture photometry catalogs of field stars around KISS15s. For each image, the aperture diameter size was set to twice the seeing full width at half maximum (FWHM) size to maximize the signal-to-noise ratio (S/N) of the photometric measurements and minimize the effects of systematic centering errors (e.g., Mighell 1999). Then the photometric zero-point of each image was determined by taking a 3σ-clipping weighted average of the differences between the instrumental magnitudes in the SExtractor catalog and the SDSS magnitudes of field stars retrieved from the SDSS Stripe 82 database (see the appendix of Annis et al. 2014).

Then template matching and image subtraction were applied to the KWFC images, using WCSRemap and High-order Transform of PSF and Template Subtraction (HOTPANTS) version 5.1.10 software, written by A. Becker¹⁴ (Becker 2015). SDSS Stripe 82 deep coadd images, constructed by Annis et al. (2014) from all imaging data obtained on or before 2005 December 1 (York et al. 2000; Frieman et al. 2008), were used as pre-SN template (=reference) images. The SDSS Stripe 82 coadd images are much deeper and have higher resolution than KWFC images (a 50% completeness limit for point sources of 23.6, 24.6, 24.2, 23.7, and 22.3 mag for the u-, g-, r-, i-, and z-bands, respectively, with a median seeing size of ∼11; Annis et al. 2014). WCSRemap was used to produce astrometrically remapped reference images that aligned exactly with individual KWFC images. Then HOTPANTS was used to align the PSF and flux scale of the reference image with those of the KWFC image and to subtract the reference image from the KWFC image. SExtractor aperture photometry for KISS15s was applied to the subtracted KWFC images. It should be noted that because the host galaxy light is almost completely removed from the subtracted KWFC images, the photometric measurements of KISS15s derived here, as described, are free from contamination of the underlying host galaxy flux.

Finally, photometric measurements obtained with the same filter each day were combined by taking a weighted mean. Table 1 presents the g-, r-, and i-band measurements for KISS15s from the Kiso/KWFC image subtraction photometry; Figure 2 shows the light curves corrected for Galactic extinction.

Zoom In Zoom Out Reset image size

SkyMapper is a fully automated 1.3 m optical telescope at Siding Spring Observatory in Australia (Scalzo et al. 2017). First Data Release (DR1.1) of the SkyMapper Southern Survey (Wolf et al. 2018)¹⁵ contains u-, g-, r-, i-, and z-band imaging data at the position of KISS15s obtained on MJD = 56899.78 (2014 August 30), 56902.77 (2014 September 2), 57234.78 (2015 July 31), and 57262.75 (2015 August 28).

First, we searched for cataloged photometry data at the position of KISS15s in the SkyMapper photometry table (dr1.master), and found that the SkyMapper pipeline detected KISS15s in r-, i-, and z-band images on 2015 August 28 (MJD = 57262.75), named SMSS 030831.61-005005.8. This means that the SkyMapper data provide optical photometric measurements at earlier epochs than the Kiso/KWFC first detection. The PSF magnitudes reported in the photometry table are r = 18.225 ± 0.041 mag, i = 18.462 ± 0.110 mag, and z = 18.125 ± 0.053 mag. It should be noted that the SkyMapper images were obtained with short exposure times to achieve 10σ magnitude limits of ∼18.0 mag for all wavelength bands (Wolf et al. 2018); thus, the photometric errors are larger than those for Kiso/KWFC photometry, although KISS15s is brighter at the epoch of the SkyMapper observation.

To further check whether KISS15s was detected in much earlier SkyMapper images and to carry out difference image photometry, we downloaded 600'' × 600'' SkyMapper cut-out images centered on KISS15s. HOTPANTS image subtraction was applied to the SkyMapper images, with SDSS Stripe 82 coadd images as a reference. Then, using the Python module Photutils (Bradley et al. 2017), we applied forced aperture photometry for the difference images with the circular aperture centered on the sky coordinate of KISS15s. Magnitude zero points were calculated by comparing measured instrumental magnitudes of field stars around KISS15s with the SDSS magnitudes. Finally, photometric measurements of each filter obtained on the same day were binned by taking weighted averages.

Table 2 reports the >3σ detection SkyMapper photometry data points for KISS15s. We found that KISS15s was barely detected in r-, i-, and z-band images on 2015 July 31 (Figure 3). Only a single exposure per each filter was obtained at this epoch, in which the exposure times were 40, 5, 5, 10, and 20 s for u-, g-, r-, i-, and z-bands, respectively. KISS15s was not detected in u- and g-band images obtained on the same day, giving magnitude 5σ lower limits of u > 18.09 mag and g > 17.88 mag. The SkyMapper's first detection date is earlier than the first detection by KISS survey data by 42 days; thus, SkyMapper provides the earliest optical photometry data of KISS15s.

Zoom In Zoom Out Reset image size

On SkyMapper's second detection date, 2015 August 28, KISS15s was detected in r-, i-, and z-bands, but not in u- and g-bands, giving magnitude 5σ lower limits of u > 18.69 mag and g > 18.98 mag. At this epoch, two exposures were obtained per each filter; the total exposure times were 80, 10, 10, 20, and 40 s for u-, g-, r-, i-, and z-bands, respectively. The last non-detection by SkyMapper observations was 2014 September 2, roughly 1 yr before the first detection date of 2015 July 31.

SkyMapper r-, i-, and z-band photometry data of KISS15s are shown in Figure 2, along with Kiso/KWFC data.

The Pan-STARRS 3π Survey detected KISS15s as of 2015 August 30 (referred to as PS15bva in the Pan-STARRS Survey for Transients database). Given that single-exposure Pan-STARRS images are not publicly available, we retrieved Pan-STARRS photometry data of KISS15s directly from the Open Supernova Catalog (Guillochon et al. 2017). Transient photometry was performed on difference images, using reference images created by stacking frames in each band between 2010 and 2012.¹⁶ KISS15s was imaged in i- and w-bands, in which the i-band is similar to that of the SDSS system and the w-band is approximately a g+r+i-band (Tonry et al. 2012). Table 3 lists the Pan-STARRS i- and w-band photometry data of KISS15s, where multiple photometry data obtained within a day are binned by taking the weighted average. The i- and w-band light curves are shown in Figure 2. In the figure, the i-band magnitude is corrected for the Galactic extinction of ${A}_{i}=0.102$ mag; however the w-band magnitude is not corrected, because the extinction estimate for the w-band is uncertain.

We obtained single epoch g-, r-, and i-band photometry for KISS15s with the Mayall 4 m Telescope in KOSMOS imaging mode (Martini et al. 2014) on 2017 July 25 UT (MJD = 57959.5). KOSMOS was used with a 2k4k E2V charge-coupled device (CCD) with a pixel scale of 0292 pixel⁻¹. The exposure time for each band was 180 s. The seeing condition on the observing night was ∼5 pixel = 15.

Mayall/KOSMOS images were analyzed in the same way as the Kiso/KWFC data, using HOTPANTS and g-, r-, and i-band references created from SDSS Stripe 82 coadd data. Aperture photometry for the reference-subtracted images was performed using Photutils. The Mayall/KOSMOS image subtraction photometry of KISS15s is summarized in Table 4. The Mayall/KOSMOS g- and i-band data are well below the power-law extrapolation of the Kiso/KWFC and SkyMapper photometry data (Figure 2). The significant drop in optical continuum emission indicates that the heating sources of the continuum emission region, which may be the innermost part of KISS15s, lose most of their kinetic energy (see Section 3.1.1 for details).

Mayall/KOSMOS r-band photometry shows an excess compared to g- and i-band data, and has a consistent brightness with the last Kiso/KWFC r-band observation. The apparent r-band excess relative to the g- and i-bands is probably due to flux contamination from a persistently strong Hα emission line (Section 2.2). It should be noted that the Mayall/KOSMOS r-band filter transmission is relatively enhanced at wavelengths around ${\lambda }_{\mathrm{obs}}\sim 6800$ Å compared to that of the original SDSS r-band filter (adopted by the Kiso/KWFC). The enhanced Hα emission line EW at later epochs (Section 2.2) and the enhanced filter response can explain the r-band excess. We note that a similar r-band excess relative to the other bands due to the contamination of the strong Hα emission line was also observed in luminous type IIn SN 2010jl (Fransson et al. 2014).

KISS15s was serendipitously detected in g-, r-, and z-band images obtained through the DECam Legacy Survey (DECaLS DR7; Dey et al. 2018).¹⁷ DECaLS uses a 3 deg² FoV prime focus optical imager DECam, which is mounted on the Blanco 4 m telescope at Cerro Tololo Inter-American Observatory and includes 62 2k4k CCDs at 0263 pixel⁻¹ resolution. The g-, r-, and z-band images of KISS15s were obtained on 2017 November 11–12, 2017 September 17, and 2015 November 28, respectively. Two images were obtained per each filter at detector positions of S28 and N29 (see e.g., Shaw 2015); the exposure times of g-, r-, z-band images were 80 and 166 s, 65 and 58 s, and 85 and 88 s, respectively. The seeing FWHMs were ∼3.3–4.6 pixels, corresponding to 09−12.

We downloaded the images and associated inverse-variance maps from the DECam Legacy Survey web page. The DECaLS images were analyzed in the same way as Kiso/KWFC data, using HOTPANTS and g-, r-, and z-band references created from SDSS Stripe 82 data. Aperture photometry for the reference-subtracted images was performed using Photutils. The DECaLS image subtraction photometry of KISS15s is summarized in Table 5, where a weighted average of the two photometric measurements for each filter is reported. As shown in Figure 2, the DECaLS g-band measurement confirms a faster rate of decline at the late epochs (≳600 days since discovery) compared to the slower temporal evolution at the early epochs.

As with Mayall/KOSMOS r-band photometry (Section 2.1.4), DECaLS r-band photometry also shows an excess, compared to g- and i-band Mayall/KOSMOS data in late epochs. The DECam r-band filter transmission is relatively enhanced at wavelengths around ${\lambda }_{\mathrm{obs}}\sim 6800$ Å compared to the original SDSS r-band filter. Thus, the apparent r-band excess of the DECaLS measurement is probably due to flux contamination from the strong Hα emission line.

The NEOWISE reactivation mission has been scanning the entire sky, using a space-based infrared telescope in the MIR wavelength range, specifically W1 (3.4 μm) and W2 (4.6 μm) bands, since late 2013 as an extension of the original mission, the WISE All-Sky Survey (Wright et al. 2010; Mainzer et al. 2014). The NEOWISE survey visits each sky region roughly twice per year (each visit consists of 9–16 exposures) and thus is useful for examining the long-term MIR light curve evolution of astronomical transients. In this work, we used the NEOWISE 2018 Data Release, which includes all W1- and W2-band images and associated photometry catalogs acquired before the end of 2017.

First, we searched for a point source at the position of KISS15s in the NEOWISE-R Single Exposure (L1b) Source table available at the NASA/IPAC Infrared Science Archive, and found that KISS15s was detected in a NEOWISE W1-band single exposure image obtained on 2015 August 5 (MJD = 57239.47; frame id = 62453b177). KISS15s has continuously been detected since the first detection both in the W1- and W2-bands. The first detection of KISS15s by NEOWISE was 1 month prior to the first detection by Kiso/KWFC (Section 2.1.1), and ∼7 days after the detection by SkyMapper (Section 2.1.2).

Then we examined whether KISS15s appeared in WISE/NEOWISE images by creating a coadded image for each visit for each band, using an online version of the WISE/NEOWISE Coadder ICORE (Masci & Fowler 2009; Masci 2013).¹⁸ Images of 003 × 003 (at 10 pixel⁻¹) coadd intensity and standard deviation at the position of KISS15s were created for each band for each NEOWISE visit (∼10 exposures per visit). Images with a qual_frame>=5, a moon separation angle >20°, and ssa_sep (distance from South Atlantic Anomaly edge) > 0° were used as input. A simple area weighting was used for its interpolation, converting the native input pixel scale of 275 pixel⁻¹ to the output pixel scale of 10 pixel⁻¹ (Section 7 of Masci 2013). Aside from the NEOWISE data, W1- and W2-band images obtained during the epochs of the WISE all-sky survey (in 2010–2011; ∼40 exposures) were coadded into W1- and W2-band referenced images, respectively; then the reference images were subtracted from the single visit coadd images to create difference images. The residual background in the difference image was estimated by taking a median of the pixel fluxes, and the estimated residual background was subtracted from the difference image.

Figure 4 shows the WISE/NEOWISE reference and difference images at the position centered on KISS15s. No signal was detected in the difference images before the NEOWISE visit on 2015 August 5. Since then, KISS15s has appeared continuously as a point source. We can see that the W2-band fluxes were only barely detected in 2015 August (at the ∼3.43σ level). Aperture photometry was applied to the difference images using Photutils, in which a fixed standard circular aperture with 825 radius centered on the sky coordinate of KISS15s was used. The same aperture photometry was also applied to the associated variance images to estimate the photometric errors of the aperture fluxes. To account for the spatial correlation of pixel noise, the photometric errors of the aperture fluxes derived from the photometry of the variance map were inflated by a factor of 2.75 (equating to the ratio of the input to output pixel scale; section 13 of Masci 2013). To determine the magnitude zero points, first we downloaded 03 × 03 ALLWISE W1- and W2-band images around KISS15s and measured the circular aperture instrumental magnitudes of field stars. The mean difference between the instrumental and Vega magnitudes of field stars cataloged in the ALLWISE point source catalog was calculated as a magnitude zero-point for each of the W1- and W2-bands. The magnitude offsets between Vega and AB magnitude systems were taken from Jarrett et al. (2011) as ${m}_{W1,\mathrm{AB}}={m}_{W1,\mathrm{Vega}}+2.699$ mag and ${m}_{W2,\mathrm{AB}}={m}_{W2,\mathrm{Vega}}\,+3.339$ mag.

Zoom In Zoom Out Reset image size

The bottom panel of Figure 2 shows the W1- and W2-band difference image photometry light curves of KISS15s after 2015 August 5 (Table 6). Interestingly, the W1- and W2-band MIR fluxes increased over time from 2015 to 2017, as opposed to the optical light curves that have steadily decreased since discovery (Figure 2). The late-time excess emission at IR wavelengths observed in SNe IIn are commonly interpreted as new dust formation in the CDS region, or as dust IR echoes from the CSM heated by X-ray and ultraviolet (UV) radiation from the ejecta-CSM interaction (e.g., Gerardy et al. 2002; Fox et al. 2010, 2011; Stritzinger et al. 2012; Fransson et al. 2014; Sarangi et al. 2018; Szalai et al. 2018). Actually, the W1 − W2 band color of KISS15s during the late phase (W1 − W2 ≃ 0 AB mag) indicates that this IR emission component has a high color temperature of T_dust ∼ 1200 K, which is close to the sublimation temperature of graphite/silicate dust grains (e.g., Fransson et al. 2014). More detailed analyses of the IR light curves are presented in Section 3.1.2.

We performed J-, H-, and K_s-band (1.24, 1.66, and 2.16 μm, respectively) imaging observations for KISS15s on 2017 December 21–22 UT (MJD = 58108.9 at mid exposure) with the Nayuta telescope, using the Nishi-Harima infrared camera (NIC). NIC is composed of three 1024 × 1024 pixel HAWAII arrays; each of the three arms has a 27 × 27 field of view, with a pixel scale of 016 pixel⁻¹, and is capable of obtaining J-, H-, and K_s-band images simultaneously. KISS15s and five relatively brighter field stars were imaged simultaneously on each image for differential photometry. Most of the images were obtained with a 10-point circular dithering pattern (radius: 10''), using an exposure time of 150 s per frame and adopting Fowler sampling with eight readouts at the beginning and end of the exposure (readout noise ∼10 e⁻¹). The total on-source exposure time was 215 minutes. All images obtained during the run were shifted and added to generate a final image using SWARP (Bertin et al. 2002).

We performed two-dimensional fitting of the KISS15s and the host galaxy to derive the NIR magnitudes of KISS15s. We first modeled the PSFs of the coadded J-, H-, and K_s-band images by fitting three Gaussians to a simultaneously imaged nearby field star, 2MASS J03083221-0050320. Then, each of the NIR images of KISS15s + foreground star + host galaxy were fitted with a model composed of two PSFs for KISS15s, the foreground star, and one inclined exponential disk for the host galaxy. The magnitude zero points of the NIR images were determined by comparing the 2MASS point source magnitudes (Skrutskie et al. 2006) and PSF-fitting instrumental magnitude measurements for five nearby field stars simultaneously imaged with KISS15s. The Vega magnitudes were converted into AB magnitudes using offsets between AB and Vega magnitude systems of 0.91, 1.39, and 1.85 mag for J-, H-, and K_s-bands, respectively (Blanton & Roweis 2007). The PSF fitting was performed using IDL routine MPFIT2DFUN,¹⁹ developed by Craig B. Markwardt; multi-component model fitting was performed using GALFIT (Version 3.0.5; Peng et al. 2002).

Figure 5 shows the best-fitting GALFIT model of J-, H-, and K_s-band images. The PSF magnitudes of KISS15s are summarized in Table 7. KISS15s is brighter in the longer wavelength bands, suggesting that the NIR bands are dominated by the same host dust component revealed by NEOWISE W1- and W2-band photometry.

Zoom In Zoom Out Reset image size

Following the Kiso/KWFC discovery of KISS15s on 2015 September 18.78 UT, we carried out optical spectroscopy measurements for rapid classification of the transient on 2015 September 19.7 UT (MJD = 57284.7) with the 2.0 m Nayuta telescope, using an optical spectrograph, LISS (Hashiba et al. 2014). LISS employs a back-illuminated fully depleted-type Hamamatsu 2k1k CCD. We used a grism with a linear dispersion of 12.4 Å pixel⁻¹ and a 20 width long-slit to obtain a very-low-resolution optical spectrum for KISS15s. The observed-frame wavelength range was 4000–10,000 Å, and the total on-source exposure time was 65 minutes. The spatial scale of LISS is 0244 pixel⁻¹, and the instrumental broadening is estimated to be σ_λ ∼ 30 Å. The slit direction was set to a position angle (PA) of 1585 to align the slit spatial direction to the semimajor axis of the host galaxy.²⁰ The spectra were extracted using an aperture size of 244.

The spectrophotometric standard star G191-B2B was observed for the flux calibration. Because the atmospheric extinction coefficients (mag/airmass values as a function of wavelength) at NHAO are not well constrained, we did not apply the airmass correction for the KISS15s spectrum. The airmass values during the observations were 1.320–1.248 for KISS15s and 1.530–1.518 for G191-B2B; thus, the bluest part of the KISS15s spectrum may be overestimated by ∼0.1 mag relative to the reddest part. Data reduction, including overscan, bias, flat correction, cosmic ray rejections with the use of L.A.Cosmic (lacos_spec; van Dokkum 2001), and wavelength and flux calibrations, were carried out using IRAF. The LISS spectrum was corrected for telluric absorption features in the wavelength ranges of 6860–6890 Å and 7600–7630 Å (O₂ bands) by dividing the KISS15s spectrum by the normalized combined spectrum of the standard star. The spectrum was corrected for Galactic extinction using the Fitzpatrick (1999) extinction curve.

Finally, the absolute flux of the LISS spectrum was scaled to match the Galactic extinction-corrected i-band magnitude of KISS15s obtained from broadband photometry. We chose the i-band as the reference wavelength band, because the i-band is less affected by the Hα emission line and blue continuum emission from the star-forming host galaxy than the g- and r-bands. To perform the spectrophotometry, we first interpolated the i-band magnitude of KISS15s at the epoch of Nayuya/LISS spectroscopy from the i-band broken-line light-curve model (Figure 2; see Section 3.1.1 for details). Then the filter-convolved i-band magnitude of the raw LISS spectrum was calculated using Speclite.²¹ The spectrophotometric scaling factor was calculated from the difference between the broadband and spectroscopic i-band magnitudes.

Figure 6 presents the Nayuta/LISS spectrum of KISS15s. Based on the redshift information (z = 0.03782) of the host galaxy SDSS J030831.67-005008.6 taken from the SDSS database, a strong, resolved broad emission line at the observed wavelength of λ_obs ∼ 6800 Å can readily be identified as the Hα emission line related to KISS15s. The Hβ emission line, He i emission line (λ5876), and Ca ii IR triplet (λλ8498, 8542, 8662) are also barely detectable in the Nayuta/LISS spectrum with a consistent redshift. Thus, the Nayuta/LISS spectrum confirms that KISS15s actually belongs to the galaxy SDSS J030831.67-005008.6. The broad Hα line does not show any P Cygni features, as is the case for other SNe IIn.

Zoom In Zoom Out Reset image size

At later epochs, we obtained three low-resolution spectra and two high-resolution spectra for KISS15s using the DIS mounted on the ARC3.5 meter telescope at the Apache Point Observatory (APO). DIS is a dual channel imaging spectrograph, composed of a Marconi CCD42-20-0-310 on the blue side and an E2V CCD42-20-1-D21 on the red side.²² Table 8 lists the log of the ARC3.5 m/DIS observations of KISS15s.

Table 8.  Log of Nayuta/LISS and ARC3.5 m/DIS Long-slit Spectroscopy of KISS15s

MJD	Date	Inst.-Grating	Slit	CCD Binning	σinst
				(spatial × dispersion)	(km s−1)
57284.7	2015 Sep 19	LISS-very low	20	2 × 2	1490
57358.2	2015 Dec 2	DIS-B400/R300	20	1 × 1	141.6
57417.1	2016 Jan 30	DIS-B400/R300	15	2 × 1	111.6
57665.4	2016 Oct 4	DIS-B400/R300	20	1 × 1	551.9 (b), 148.6 (n)
57430.1	2016 Feb 12	DIS-B1200/R1200	09	1 × 1	20.8
57665.4	2016 Oct 4	DIS-B1200/R1200	09	1 × 1	150.4 (b), 18.4 (n)

Note. The slit direction is set to PA = 1585. The velocity dispersions of the instrumental line profiles σ_inst are evaluated by fitting Gaussian functions to the arc lamp spectra at λ_obs = 6900–7100 Å. The instrumental line profiles on 2016 October 4 have two components: a narrow core ("n") and a broad wing ("b"), which may be due to defocusing of the spectrograph. For the spectra obtained on 2016 October 4, we assume σ_inst = σ_inst(b) (Section A).

Download table as:  ASCIITypeset image

The first epoch ARC3.5 m/DIS low-resolution optical spectrum of KISS15s was obtained on 2015 December 2 UT (MJD = 57358.2). A standard low-resolution grating setup of B400/R300 for blue/red channels, 1 × 1 on-chip binning, and a 20 long-slit were used. The spatial scales were 040 pixel⁻¹ and 042 pixel⁻¹ for the blue and red channels, respectively. The total exposure time was 45 minutes (three exposures of 15 minutes each), and the slit direction was set to PA = 1585. A spectrophotometric standard star GD50 was also observed on the same day for the flux calibration.

The second epoch ARC3.5 m/DIS low-resolution spectrum was obtained on 2016 January 30 UT (MJD = 57417.1), using the same instrumental setup as the first observation, except for the use of a 15 width long-slit and 2 × 1 binning. The observation may have been affected to some extent by cirrus clouds.

The third epoch ARC3.5 m/DIS spectrum was obtained on 2016 February 12 UT (MJD = 57430.1) using a high-spectral resolution mode: 09 width long-slit, 1 × 1 binning, a no-order sorting filter, and B1200/R1200 gratings. The B1200 and R1200 grisms were tilted in their housings so that the target wavelengths (5040 Å and 6800 Å, respectively) were located toward the center of the detector. As summarized in Table 8, the instrumental broadening of the high-resolution mode was evaluated to be σ_inst ∼ 20 km s⁻¹ in the Hα emission wavelength region of KISS15s. The total exposure time was about 74 minutes, and the slit direction was set to PA = 1585. The spectrophotometric standard star GD50 was observed on the same day for flux calibration.

The fourth epoch ARC3.5 m/DIS spectrum was obtained on 2016 October 4 (MJD = 57665.4), using both low- and high-spectral resolution modes and the same instrumental configurations as the previous observations. The total exposure times were 60 and 66.7 minutes for the low- and high-resolution spectroscopy mode, respectively; however, cloudy weather conditions prevented us from achieving the planned S/N objective.

Data reduction of ARC3.5 m/DIS spectra was carried out using IRAF. Simple aperture extraction was performed, in which the extraction aperture was set to ∼24 to minimize flux contamination from the host galaxy. The relative airmass correction was applied using the APO atmospheric extinction coefficients compiled by J. Davenport.²³ All spectra were corrected for Galactic extinction using the Fitzpatrick (1999) extinction curve, under the assumption of R_V = 3.1. The sensitivity functions derived from the data of the spectrophotometric standard stars obtained on the first (low-resolution) and third (high-resolution) DIS observation epochs were used for the flux calibration. Note that due to the differences in the grism tilt angles between observations runs, the fourth epoch blue-channel DIS spectrum could not be accurately calibrated. For each DIS spectrum, we corrected for telluric absorption features in the wavelength ranges of 6860–6890 and 7600–7630 Å (O₂ bands) by dividing the KISS15s spectrum by a continuum-normalized spectrophotometric standard star spectrum.

Finally, the absolute flux scales of DIS spectra were calibrated by spectrophotometry. Because the i-band is less affected by blue continuum emission from the star-forming host galaxy (Section 4.1) and the strong emission lines compared to g and r-bands, we used the broken-line i-band light-curve model (Figure 2; see Section 3.1.1) as the reference magnitude for spectrophotometry measurements. High-resolution mode spectra did not cover a sufficient wavelength ranges; thus, we applied spectrophotometry only to low-resolution mode spectra. Specifically, the observed spectra were convolved with the SDSS i-band filter transmission curve using Speclite; the spectrophotometric calibration factor for each spectrum was derived from the ratio of the i-band power-law fitted flux and the filter-convolved flux.

Figure 6 shows low-resolution ARC3.5 m/DIS spectra. At early epochs, the DIS spectra showed consistent spectral features with the lower-resolution Nayuta/LISS spectrum. The broad Hα, Hβ, and Ca ii IR triplet emission lines are clearly visible. In addition, an intermediate-velocity width component can be identified on top of the broad Hα emission line in both DIS and Nayuta/LISS spectra. The broad component (v_FWHM ∼ 14,000 km s⁻¹) had weakened considerably by 2016 October 4; however, the intermediate component (v_FWHM ∼ 2000 km s⁻¹) retained its high luminosity during the observations.

Several narrow emission lines (e.g., [O ii]λ3727, Hβ, [O iii]λλ4959,5007, and Hα) are also clearly visible in the DIS spectra. As discussed in Section 3.2.1, these narrow components are probably due to the emission from extended H ii regions in the host galaxy and thus are not directly related to the SN explosion of KISS15s. As for the narrow component, the different slit widths and variable seeing conditions during our observations were believed to be the main causes of the variation in the narrow line flux among DIS spectra. As discussed in detail in Section A, the DIS red-arm spectrum obtained on 2016 October 4 has complex instrumental broadening profiles; this may have an effect on the apparent weakness of the narrow emission line in Figure 6.

In Figure 6, blackbody spectra with temperatures of ${T}_{\mathrm{BB}}=4000$ and 6000 K are also shown for comparison. It is clear that the observed optical spectra of KISS15s cannot be explained by a single temperature blackbody spectrum. This may imply that the optical light from KISS15s is affected by the host galaxy's dust extinction, as discussed in Section 2.3.

Figure 7 presents high-resolution ARC3.5 m/DIS spectra. Although we obtained the high-resolution spectra for the purpose of detecting a narrow P Cygni feature due to the CSM moving at the wind velocity (expected to be v_w ∼ 10–100 km s⁻¹; Miller et al. 2010), we found no P Cygni feature in the Hα emission line profile of KISS15s. Therefore, it was impossible to determine the wind velocity v_w of the progenitor of KISS15s from these spectroscopic observations. As mentioned previously, the DIS red-arm spectrum obtained on 2016 October 4 suffers from complex instrumental broadening (Section A). Thus, we interpreted the apparent line profile variation in the narrow emission line component between high-resolution spectra on 2016 February 12 and October 4 as artificial.

Zoom In Zoom Out Reset image size

The multi-component Hα emission line profile of KISS15s revealed by the ARC3.5 m/DIS confirms that this object is an SN IIn with strong ejecta-CSM interaction. Particularly, the long-duration optical continuum light curves, IR emission excess, and complex spectral properties of KISS15s all suggest that KISS15s belongs to the "1988Z-like" subclass of SNe IIn (e.g., Stritzinger et al. 2012; Taddia et al. 2013; Smith et al. 2017).

The g- and i-band measurements at early epochs (MJD < 57800) revealed that the optical light curves of KISS15s showed a slow decline for ∼500–600 days since the first detection by SkyMapper at MJD = 57234 (Figure 2). After that, KISS15s experienced a rapid decline in luminosity, as discussed in Section 2.1.4. Although the r- and z-bands are contaminated by strong Hα and Ca ii IR triplet emission, respectively, the g- and i-band fluxes are dominated by optical continuum emission.

If the g-band luminosity light curve at MJD < 57800 is assumed to be a power law,

$$\begin{eqnarray}&&{L}_{g}\propto {t}^{\alpha },\end{eqnarray} \tag{ 2 }$$

the observed magnitude can be expressed as

$$\begin{eqnarray}&&g={c}_{g}-2.5\alpha \mathrm{log}(\mathrm{MJD}-{\mathrm{MJD}}_{0}),\end{eqnarray} \tag{ 3 }$$

where α is the power-law index in Equation (2), c_g is the g-band magnitude at $\mathrm{MJD}={\mathrm{MJD}}_{0}+1$, and MJD₀ is a model parameter corresponding to the SN explosion date. By fitting Equation (3) to the Galactic extinction-corrected g-band light curve of KISS15s at MJD < 57800, we obtain ${c}_{g}={17.29}_{+0.50}^{-0.76}$ mag, $\alpha =-{0.45}_{+0.08}^{-0.11}$, and ${\mathrm{MJD}}_{0}={57209.2}_{-36.1}^{+22.5}$. MJD₀ = 57209.2 corresponds to 2015 July 6, which implies that the explosion date of KISS15s is probably ∼1 month before the first detection by SkyMapper on MJD = 57234. The best-fit power-law model is shown in Figure 2.

Next, to evaluate the temporal evolution of the g − i color, we constructed broken-line light-curve models for the g- and i-bands as follows. First, the light curve was divided into three parts (MJD < 57450, 57450 < MJD < 57800, and 57800 < MJD), and a linear regression line was fitted separately to each. At MJD < 57450, linear regression lines for the g- and i-band magnitude light curves were calculated as the light-curve model. At 57450 < MJD < 57800, the g-band light curve was modeled by a linear regression line of g-band data between MJD = 57400 and 57800. In the same way, the g-band light curve at 57800 < MJD was modeled by a linear regression line of the two late-time g-band measurements from Mayall/KOSMOS and Blanco/DECam. Given that i-band data are too sparse to directly estimate g − i color at MJD > 57450, we simply assumed a constant g − i color fixed to an average value evaluated from the data between MJD = 57420 and 57450. The Galactic extinction-corrected g − i color at 57420 < MJD < 57450 was evaluated as $g-i=0.67\pm 0.03\,\mathrm{mag};$ we assumed that this color remained constant at $\mathrm{MJD}\gt 57450$:

$$\begin{eqnarray}&&g-i=0.67\pm 0.03\,\mathrm{mag}\ \ \ \ \ (\mathrm{MJD}\gt 57450).\end{eqnarray} \tag{ 4 }$$

The best-fit linear light-curve models are shown in Figure 2. The r-band shows a clear excess relative to the g- and i-band light curves, particularly in late epochs (>600 days since discovery), indicating a stronger flux contribution from the Hα emission in the r-band at later epochs. With the g- and i-band broken-line light-curve models, the g − i color corrected for the host galaxy extinction of $E{(B-V)}_{\mathrm{host}}=0.6$ mag and the corresponding g − i color temperature were calculated; the g − i blackbody color as a function of blackbody temperature was calculated using Speclite and assuming a source redshift of z = 0.03782. Then the temperatures of the g − i colors of KISS15s were searched. Figure 9 shows the g − i color temperature as a function of time, along with the uncertainty due to broken-line light-curve modeling. On the assumption of a SMC-like host galaxy extinction of $E{(B-V)}_{\mathrm{host}}=0.6$ mag (Section 2.3), the g − i color at $\mathrm{MJD}\gt 57450$ becomes $g-i=-0.36\pm 0.03$ mag, and the g − i color temperature is

$$\begin{eqnarray}&&{T}_{\mathrm{BB},\mathrm{opt}}={\rm{13,000}}\,{\rm{K}}\pm 600\,{\rm{K}}\ \ \ \ \ (\mathrm{MJD}\ \gt \ 57450),\end{eqnarray} \tag{ 5 }$$

as shown in Figure 9.

Zoom In Zoom Out Reset image size

The slow time evolutions of the optical light curves suggest that the CSM interaction is the dominant source of the optical luminosity of KISS15s (rather than the radioactive decay of ⁵⁶Ni and ⁵⁶Co), similar to other SNe IIn. The rapid decline in luminosity in late epochs (>600 days since discovery) implies that the volume of the continuum emission region begins to suddenly decrease at this epoch. Figure 10 compares the absolute magnitude optical light curves of KISS15s with other 1988Z-like SNe IIn.²⁵ In Figure 10, the magnitudes are corrected for the putative host galaxy extinction of $E{(B-V)}_{\mathrm{host}}=0.6$ mag. The long-duration optical light curve of KISS15s is a common characteristic of the SN 1988Z-like subclass of SN IIn (Stritzinger et al. 2012; Smith 2017; Smith et al. 2017). Figure 10 shows that after host galaxy extinction correction, KISS15s is one of the most luminous and longest-duration SN IIn among the well-observed samples of SNe IIn. The 1988Z-like SNe IIn share remarkably similar early- and late-phase photometric and spectroscopic properties at the optical wavelengths; in fact, KISS15s shows emission line profiles similar to SN 1988Z, the prototype of the 1988Z-like SNe IIn (see Figure 8). The sudden drop in the continuum emission at the late epoch probed by g- and i-band photometry of KISS15s may correspond to a similar luminosity drop observed in SN 2010jl at ∼320 days since discovery. In the case of SN 2010jl, the sudden luminosity drop has been interpreted such that the shocked shell escapes the dense (bipolar) CSM at the time of the transition (Fransson et al. 2014; Moriya 2014); the same scenario may be the case for the optical luminosity drop observed in KISS15s.

Zoom In Zoom Out Reset image size

The 3–4 μm IR emission of SNe IIn can readily be identified as hot dust thermal emission (Smith et al. 2009; Fox et al. 2010; Maeda et al. 2013; Gall et al. 2014; Szalai et al. 2018). Graphite/silicate dust grains sublimate at temperatures of T_sub = 1400–1800 K (e.g., Guhathakurta & Draine 1989; Baskin & Laor 2018), and the 3–4 μm IR emission corresponds to dust thermal emission close to the dust sublimation temperature. Figure 11 compares the W1- and W2-band light curves of KISS15s, with several years of Spitzer/IRAC band 1 (3.6 μm) and band 2 (4.5 μm) light curves for other SN 1988Z-like SNe IIn (SN 2005ip, SN 2006jd, and SN 2010jl) taken from the literature. Compared to the Spitzer/IRAC sample of SNe IIn, NEOWISE data of KISS15s enable the temporal evolution of the IR emission to be followed from the very early epochs (∼7 days after the first detection). The peak IR luminosity of KISS15s is similar to those of other SN 1988Z-like SNe IIn, suggesting that the IR emission mechanism in KISS15s is identical to them.

Zoom In Zoom Out Reset image size

As for SN 2010jl, although the significant IR excess above the extrapolated UV-optical emission at <1 yr since discovery is most likely due to the IR echo of the initial flash of the SN by preexisting dust in the CSM (Sarangi et al. 2018), the late-time IR emission associated with dust-induced increasingly blueshifted line profiles are due to new dust formation in the CDS (Andrews et al. 2011; Maeda et al. 2013; Gall et al. 2014; Szalai et al. 2018). As discussed later, the first W1- and W2-band data points of KISS15s (at 7.5 days since discovery) are consistent with the Rayleigh–Jeans tail of the optical blackbody emission, as evidenced by the blue W1 − W2 color. We found that the W1- and W2-band dust thermal emission light curves of KISS15s increase smoothly, without any early-phase excess at <1 yr, as opposed to SN 2010jl. The smooth IR light curves of KISS15s indicate that the early-phase IR echo is unimportant for KISS15s and that the IR emission is mainly from newly formed dust in the CDS, suggested as the origin of hot dust thermal emissions of other SNe IIn (e.g., Fox et al. 2010, 2011; Maeda et al. 2013; Gall et al. 2014; Fransson et al. 2014).

Figure 12 shows the temporal evolution of the W1 − W2 color temperature of the IR emission component of KISS15s at 177.5, 373.1, 539.6, and 737.2 days since discovery, where the W1 − W2 blackbody colors of various temperatures were calculated using Speclite, assuming a source redshift of z = 0.03782. Here, we also assumed that the flux contribution from the Rayleigh–Jeans tail of the optical blackbody emission is not significant at these epochs, and the observed W1 − W2 color was directly used to calculate the color temperature. The W1 − W2 color of KISS15s shows little time variation, and the color temperature remains almost constant during the observations at ∼1200 K. The weighted average of the W1 − W2 color temperature is

$$\begin{eqnarray}&&{T}_{\mathrm{BB},\mathrm{IR}}=1190\,{\rm{K}}\,(\pm 60\,{\rm{K}}).\end{eqnarray} \tag{ 6 }$$

It should be noted that dust grains formed inside of SN ejecta observed in type II-P SNe generally have dust temperatures of at most several hundreds of Kelvin (Szalai et al. 2018); thus, the persistently high dust temperature of KISS15s can definitely be attributed to the newly formed dust grains in the CDS, as observed in other SNe IIn. The estimated dust temperature is slightly lower than the sublimation temperature of silicate and graphite grains (${T}_{\mathrm{sub}}$ = 1400 and 1800 K, respectively; e.g., Baskin & Laor 2018), indicating that both of these dust grain species are likely contributors to the IR continuum emission of KISS15s.

Zoom In Zoom Out Reset image size

In contrast to the increasing IR light curves, the optical light curves of KISS15s decrease over time during the observations. Figure 13 presents the g-, r-, W1-, and W2-band optical-to-IR broadband SED of KISS15s at the epochs of NEOWISE W1- and W2-band observations, where the optical data points at g- and i-bands are interpolated by the broken-line light-curve models shown in Figures 2 and 10. The photometry data in Figure 13 are corrected for the host galaxy extinction of $E{(B-V)}_{\mathrm{host}}=0.6$ mag. In Figure 13, the best-fit optical + IR blackbody model spectrum for the SED of each epoch is also shown, where the time-dependent optical blackbody temperature is from Figure 9; the IR blackbody is assumed to have a constant temperature of ${T}_{\mathrm{BB},\mathrm{IR}}=1190\,{\rm{K}}$ and is scaled to the W2-band magnitudes. As for the continuum emission, the IR emission increases as the optical emission fades, and the 3–4 μm IR luminosity is comparable to the optical luminosity at about 2 yr after the discovery. This observation is consistent with the suggestion that new dust formation in the SN environment is enhanced several hundreds of days after the SN explosion (e.g., Gall et al. 2014; Sarangi et al. 2018). Figure 13 shows that after the correction of $E{(B-V)}_{\mathrm{host}}\,=0.6$ mag, the earliest-phase IR luminosities (7.5 days since discovery) can fully be attributed to Rayleigh–Jeans tail emission in the optical blackbody emission of ${T}_{\mathrm{BB}}=7400$ K (see Figure 9). This provides additional evidence that KISS15s is heavily reddened by SN-related/unrelated dust in the host galaxy. In addition, the non-detection of the IR blackbody component at the earliest phase (7.5 days since discovery) excludes the possibility of a large IR flux contribution from IR echoes of the SN shock breakout radiation by preexisting dust in close proximity to KISS15s at this epoch.

Zoom In Zoom Out Reset image size

Figure 14 presents the IR broadband SED from Nayuta/NIC J-, H-, and K_s-band data on 2017 December 21–22 (874.9 days since discovery) and from NEOWISE W1- and W2-band data on 2017 August 5–6 (737.2 days since discovery). Although these data were not strictly simultaneously obtained, we can conclude that the overall IR SED shape is consistent with the ${T}_{\mathrm{BB},\mathrm{IR}}=1190$ K blackbody SED. The perfect blackbody IR SED shape of KISS15s suggests that the typical radius a of the dust grains responsible for the IR continuum is much larger than the observed IR wavelengths ($2\pi a\gtrsim {\lambda }_{\mathrm{rest}};$ i.e., $a\gtrsim 0.3\mbox{--}0.6$μm). The J-band photometry shows a clear excess from the IR blackbody SED, suggesting that the short IR wavelengths are affected by the flux contamination from unmodeled host galaxy components, the Rayleigh–Jeans tail of the optical blackbody of KISS15s (Figure 13), and also from strong emission lines in the J-band (He i, Paγ, and Paβ).

Zoom In Zoom Out Reset image size

The constant temperature and the increasing luminosity indicate that the surface area of the dust thermal emission region is expanding. Figure 15 shows the integrated IR blackbody luminosity as a function of observation epoch, ${L}_{\mathrm{BB},\mathrm{IR}}(t)$, calculated from the W2-band magnitudes of KISS15s by assuming a constant blackbody temperature of ${T}_{\mathrm{BB},\mathrm{IR}}=1190$ K. If we assume that the dust distribution is approximated as a homogeneous thin shell, the blackbody radius can be defined as (Fox et al. 2010, 2011)

$$\begin{eqnarray}\begin{array}{rcl}{r}_{\mathrm{BB},\mathrm{IR}}(t) & = & {\left(\displaystyle \frac{{L}_{\mathrm{BB},\mathrm{IR}}(t)}{4\pi {\sigma }_{\mathrm{sb}}{T}_{\mathrm{BB},\mathrm{IR}}^{4}}\right)}^{1/2}\\ & = & 14.4\,{\rm{lt \mbox{-} day}}\times {\left(\displaystyle \frac{{L}_{\mathrm{BB},\mathrm{IR}}(t)}{2\times {10}^{42}\mathrm{erg}{{\rm{s}}}^{-1}}\right)}^{1/2}\\ & & \times \,{\left(\displaystyle \frac{{T}_{\mathrm{BB},\mathrm{IR}}}{1190{\rm{K}}}\right)}^{-2},\end{array}\end{eqnarray} \tag{ 7 }$$

where σ_sb is Stefan–Boltzmann's constant and r_BB,IR(t) is increasing over time. The numerical calculation results of ${r}_{\mathrm{BB},\mathrm{IR}}(t)$ are shown in the middle panel of Figure 15. Note that the actual dust radius can be a few times larger than r_BB,IR if the dust distribution is inhomogeneous. The bottom panel of Figure 15 shows the expansion velocities of the IR blackbody radius, v_dust. Two definitions of the velocities are possible: one is

$$\begin{eqnarray}&&{v}_{\mathrm{dust}}=\displaystyle \frac{{r}_{\mathrm{BB},\mathrm{IR}}(t)}{t}=\displaystyle \frac{{r}_{\mathrm{BB},\mathrm{IR}}(\mathrm{MJD})}{\mathrm{MJD}-{\mathrm{MJD}}_{0}},\end{eqnarray} \tag{ 8 }$$

and the other is

$$\begin{eqnarray}&&{v}_{\mathrm{dust}}=\displaystyle \frac{{{dr}}_{\mathrm{BB},\mathrm{IR}}(t)}{{dt}}\simeq \displaystyle \frac{{\rm{\Delta }}{r}_{\mathrm{BB},\mathrm{IR}}(\mathrm{MJD})}{{\rm{\Delta }}\mathrm{MJD}},\end{eqnarray} \tag{ 9 }$$

where the former represents the hypothetical linear velocity required for the dust shell to reach ${r}_{\mathrm{BB},\mathrm{IR}}(t)$ at a given t, and the latter is the dust shell velocity between two adjacent epochs. We assume an explosion date of ${\mathrm{MJD}}_{0}=57209.2$ (Section 3.1.1) with Equation (8).

Zoom In Zoom Out Reset image size

It should be noted that the derived ${r}_{\mathrm{BB},\mathrm{IR}}(t)$ is more than an order of magnitude smaller than the light-crossing distance since the SN explosion of KISS15s (MJD₀ ∼ 57209.2; Section 3.1.1). This suggests that the IR echo model is not the main source of the late-phase NIR excess emission. The estimated dust shell velocity v_dust is consistent with the ejecta and/or shock front velocities inferred from the velocity width of broad/intermediate Hα emission line components (see Sections 4.3 and 4.4), suggesting that the dust grains are located in the proximity of the ejecta-CSM interaction region or CDS. These results, the late-time IR excess, the hot dust temperature, and the inferred expansion velocity, are consistent with the scenario that the hot dust thermal emission of KISS15s is due to newly formed dust grains in the CDS.

To examine whether the dust grains can really be formed against the strong UV-optical radiation field around the SN, we can compare the derived dust shell size estimate with a dust sublimation radius r_sub, within which the dust grains are completely sublimated due to intense UV-optical radiation from the SN. The dust sublimation radius is given as (Waxman & Draine 2000; van Velzen et al. 2016)

$$\begin{eqnarray}&&{r}_{\mathrm{sub}}={\left(\displaystyle \frac{{L}_{\mathrm{abs}}}{16\pi {\sigma }_{\mathrm{sb}}{T}_{\mathrm{sub}}^{4}}\displaystyle \frac{\langle {Q}_{\mathrm{UV}}\rangle }{\langle {Q}_{\mathrm{IR}}\rangle }\right)}^{1/2},\end{eqnarray} \tag{ 10 }$$

where $\langle {Q}_{\mathrm{UV}}\rangle$ and $\langle {Q}_{\mathrm{IR}}\rangle$ are the Planck-averaged UV-optical absorption and IR emission efficiencies of the dust grains, respectively, and L_abs is the SN UV-optical luminosity absorbed by the dust. Because graphite grains have a higher sublimation temperature than silicate grains, here we consider the case of the graphite sublimation radius defined by ${T}_{\mathrm{sub}}=1800$ K. As described previously, we can approximate $\langle {Q}_{\mathrm{UV}}\rangle \sim \langle {Q}_{\mathrm{IR}}\rangle \sim 1$ on the assumption of a large dust grain size (e.g., Fox et al. 2010). We can expect that the UV-optical luminosity of KISS15s is always, at most, on the order of ${L}_{\mathrm{abs}}\sim {L}_{\mathrm{BB},\mathrm{opt}}\sim {10}^{43}\,\mathrm{erg}\,{{\rm{s}}}^{-1}$ (Figure 13). Thus,

$$\begin{eqnarray}&&{r}_{\mathrm{sub}}=7.1\,\mathrm{lt} \mbox{-} \mathrm{day}\,{\left(\displaystyle \frac{{L}_{\mathrm{BB},\mathrm{opt}}}{{10}^{43}\mathrm{erg}{{\rm{s}}}^{-1}}\right)}^{1/2}{\left(\displaystyle \frac{{T}_{\mathrm{sub}}}{1800{\rm{K}}}\right)}^{-2}.\end{eqnarray} \tag{ 11 }$$

The dust sublimation radius of 7.1 light days is compared to the blackbody radius r_BB,IR in the middle panel of Figure 15. The dust shell radius is larger than the dust sublimation radius (${r}_{\mathrm{BB},\mathrm{IR}}\gt {r}_{\mathrm{sub}}$), which indicates that the dust grains can indeed exist at the inferred location of r_BB,IR.

On the assumption of single-size dust grains and that the dust grains are mostly optically thin to IR emission, the total dust mass M_d required to produce the observed IR emission can be estimated as (e.g., Smith & Gehrz 2005)

$$\begin{eqnarray}&&{M}_{d}=\left(\displaystyle \frac{a\rho }{3{\sigma }_{\mathrm{sb}}\langle {Q}_{\mathrm{IR}}\rangle {T}_{d}^{4}}\right){L}_{\mathrm{BB},\mathrm{IR}},\end{eqnarray} \tag{ 12 }$$

where ρ is the mass density of the dust grain. Substituting the mass density of a graphite grain (ρ = 2.25 g cm⁻³), a grain size of a = 0.5 μm, a dust temperature of ${T}_{d}=1190$ K, ${L}_{\mathrm{BB},\mathrm{IR}}=2\times {10}^{42}$ erg s⁻¹ (Figure 15), and $\langle {Q}_{\mathrm{IR}}\rangle \sim 1$ into Equation (12), we obtain

$$\begin{eqnarray}\begin{array}{rcl}{M}_{d} & \simeq & 3\times {10}^{-4}\,{M}_{\odot }\left(\displaystyle \frac{a}{0.5\,\mu {\rm{m}}}\right){\left(\displaystyle \frac{{T}_{d}}{1190{\rm{K}}}\right)}^{-4}\\ & & \times \,\left(\displaystyle \frac{{L}_{\mathrm{BB},\mathrm{IR}}}{2\times {10}^{42}\,\mathrm{erg}\,{{\rm{s}}}^{-1}}\right).\end{array}\end{eqnarray} \tag{ 13 }$$

It should be noted that the estimated total mass of the newly formed large dust grains is consistent with that observed in SN 2010jl (Gall et al. 2014). The derived value sets a lower limit of the total dust mass if the dust IR emission is self-shielded. The dust grains radiating ∼3 μm IR emission will undergo rapid radiative cooling. Therefore, KISS15s may currently harbor a cool dust component of a mass of $\gg 3\times {10}^{-4}\,{M}_{\odot }$, which may be detected by future follow-up observations at MIR and longer wavelengths.

An estimate of a bolometric luminosity light curve can be obtained from blackbody fitting for optical and IR data points, as described in Sections 3.1.1 and 3.1.2. Because the IR blackbody component of KISS15s can naturally be attributed to the newly formed dust in the ejecta-CSM interaction region (rather than IR echoes from preexisting dust), the bolometric luminosity is estimated to be the sum of optical and IR blackbody components (Fransson et al. 2014). Here we define the pseudo-bolometric luminosity as the sum of the optical and IR blackbody luminosities, ${L}_{\mathrm{bol}}={L}_{\mathrm{BB},\mathrm{opt}}+{L}_{\mathrm{BB},\mathrm{IR}}$, which can be calculated from the best-fit blackbody models shown in Figure 13.

Figure 16 shows the temporal evolution of the optical and IR blackbody luminosities and the pseudo-bolometric luminosity of KISS15s. The power-law index of the optical luminosity is ${L}_{\mathrm{BB},\mathrm{opt}}\propto {t}^{-0.23}$ on the assumption of ${\mathrm{MJD}}_{0}=57209.2$ (Section 3.1.1). The relative luminosity contribution of the IR luminosity relative to the optical luminosity becomes stronger at later epochs; consequently, the optical + IR pseudo-bolometric luminosity shows slower temporal evolution compared to the optical luminosity. The power-law index of the pseudo-bolometric luminosity light curve is given by

$$\begin{eqnarray}&&{L}_{\mathrm{bol}}\propto {t}^{-0.16}\end{eqnarray} \tag{ 14 }$$

at <600 days since discovery.

Zoom In Zoom Out Reset image size

After the rapid decline in the optical luminosity at >600 days, the pseudo-bolometric luminosity accordingly decrease suddenly, as the IR luminosity remains nearly constant at these epochs. This behavior of the bolometric luminosity evolution indicates that the sudden decrease in optical luminosity is not due to dust extinction by newly formed dust; instead, this behavior supports a scenario in which the intrinsic weakening of the ejecta-CSM interaction causes accelerated fading of the optical luminosity, as described in Section 3.1.1 (see Maeda et al. 2013; Moriya 2014).

The peak luminosity of KISS15s is estimated to be ${L}_{\mathrm{peak}}\sim 1.0\times {10}^{43}\,\mathrm{erg}\,{{\rm{s}}}^{-1}$, and the total radiated energy at the first 600 days is roughly $\sim 4.0\times {10}^{50}\,\mathrm{erg}$. This estimate indicates that KISS15s is one of the most luminous SNe IIn discovered in the local universe so far, comparable to the luminous (${L}_{\mathrm{peak}}\sim 3.0\times {10}^{43}\,\mathrm{erg}\,{{\rm{s}}}^{-1}$) type IIn SN 2010jl (Fransson et al. 2014). In the case of SN 2010jl, if the observed luminosity is assumed to be solely powered by ⁵⁶Ni → ⁵⁶Co → ⁵⁶Fe radioactive decay, it requires an unreasonably large ⁵⁶Ni mass of ${M}_{{56}_{\mathrm{Ni}}}\sim 3.4\,{M}_{\odot }$ (see Moriya et al. 2013, for details). From the same discussion, the required ⁵⁶Ni mass to account for the luminosity observed in KISS15s is ${M}_{{56}_{\mathrm{Ni}}}\sim 1.1\,{M}_{\odot }$. Although the ⁵⁶Ni mass production over $1\,{M}_{\odot }$ may be possible with the SNe of massive stars (e.g., Umeda & Nomoto 2008), a more natural explanation for the high luminosity, as well as the persistent light curves even after the decay time of ⁵⁶Co (${\tau }_{\mathrm{Co}}=111.3$ days; Nadyozhin 1994), is that KISS15s is powered by the strong ejecta-CSM interaction, as is the case for other luminous SNe IIn.

Figure 17 compares the spectral profiles of Hα and Hβ emission lines of the DIS low-resolution spectrum obtained on 2015 December 2. It is clear that the Hβ and Hα emission lines have nearly identical spectral profiles, and both of them have broad (${v}_{\mathrm{FWHM}}\,\sim$ 14,000 km s⁻¹), intermediate (${v}_{\mathrm{FWHM}}\,\sim$ 2000 km s⁻¹), and narrow (${v}_{\mathrm{FWHM}}\lt 100$ km s⁻¹) components. It should be noted that the He i emission lines and the Hγ emission line also have similar line profiles (Figure 8), although the He i emission line luminosities are much weaker than the Hβ and Hα emission lines. As shown in Figure 18 (see also Figure 8), the optical spectral properties of KISS15s are similar to type IIn SN 1988Z and other SN 1988Z-like subclasses of SN IIn (see, e.g., Stathakis & Sadler 1991; Pastorello et al. 2002; Miller et al. 2010; Stritzinger et al. 2012).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

To quantitatively evaluate the line widths and line intensities of the Hα emission, we fitted multi-component Gaussian models to Nayuta/LISS and ARC3.5 m/DIS spectra in the spectral region around the Hα emission line. ${\chi }^{2}$ minimization was performed using the Python version of MPFIT (Markwardt 2009).²⁶ Nayuta/LISS and ARC3.5 m/DIS low-resolution spectra were fitted in an observed-frame wavelength range of λ_obs = 5860–7440 Å, and the ARC3.5 m/DIS high-resolution spectra were fitted at λ_obs = 6240–7310 Å.

First, the DIS low- and high-resolution spectra were fitted by three Gaussians (broad, intermediate, and narrow) and a single power-law model. The Nayuta/LISS very-low-resolution spectrum was fitted by two Gaussians and a single power-law model, because the spectral resolution of Nayuta/LISS is lower than the velocity width of the intermediate component; thus, the narrow component is mixed with the intermediate component. The measurement uncertainties of the model parameters were evaluated by 10,000 trials of Monte Carlo resampling, where 10,000 mock spectra were generated by adding Gaussian flux noise to the original spectrum using the calculated flux density errors. Figure 19 shows the profiles of the Hα emission of KISS15s. The best-fit model spectra and the fitting residuals are also shown in the same figure, and the best-fit model parameters are tabulated in Table 9. The fitted models mostly explain the observed emission line profiles, except for a blueshift excess component at v ∼ −5000 km s⁻¹, as indicated in Figure 17.

Zoom In Zoom Out Reset image size

Next, we performed another model fitting to the Hα profiles by adding an additional Gaussian component to account for the blueshift excess component. Figure 20 shows the same Hα emission line profiles as Figure 19, but fitted by the refined model with the additional Gaussian blueshift excess component. The best-fit parameters of the four Gaussians model are tabulated in Table 10. The best-fit model shown in Figure 20 reasonably reproduces the observed line profiles, and the fitting residuals no longer show any signature of remaining spectral components. We also tried to model the line profiles with additional Gaussian absorption of the broad component, instead of the additional Gaussian emission component, but no reasonable fitting was achieved. This analysis suggests that the apparent asymmetry of the entire Hα line profile of KISS15s is not due to absorption of broad or intermediate emission lines by opaque gases or dusts located in the ejecta-CSM interaction region (as suggested for other SNe IIn by, e.g., Smith et al. 2009; Fox et al. 2011; Smith et al. 2012b; Taddia et al. 2013; Fransson et al. 2014; Chevalier & Fransson 2017). Instead, the observed line profile of KISS15s is most likely due to the presence of an additional blueshifted symmetric intermediate Gaussian emission component with respect to the symmetric narrow, intermediate, and broad Gaussians, which may be related to the inhomogeneity of the CSM distributions (Smith et al. 2015; Andrews et al. 2016). There is no evidence of Lorenzian-like broad wing components in the observed line profiles of KISS15s, suggesting that the electron scattering in optically thick ejecta-CSM regions is not relevant in KISS15s (see, e.g., Smith et al. 2010; Fransson et al. 2014; Borish et al. 2015; Huang & Chevalier 2018). Considering the better fitting for the observed spectra compared to the three Gaussians model, here we focus on the fitting results from the "additional Gaussian model" or the four Gaussians model (Table 10 and Figure 20).

Zoom In Zoom Out Reset image size

Table 10.  Best-fit Gaussian Model for the Hα Line Profile with a Blueshift Excess Component

Parameter Name	LISS	DIS Low 1	DIS Low 2	DIS Low 3	DIS High 1	DIS High 2
Date	2015 Sep 19	2015 Dec 2	2016 Jan 30	2016 Oct 4	2016 Feb 12	2016 Oct 4
Days Since Discovery	50.7 days	124.2 days	183.1 days	431.4 days	196.1 days	431.4 days
Norm. (10−17 erg s−1 cm−2 Å−1)	9.72 ± 0.07	5.20 ± 0.01	3.61 ± 0.02	2.52 ± 0.02	2.90 ± 0.06	0.70 ± 0.03
Index	0.36 ± 0.07	1.13 ± 0.03	0.72 ± 0.08	0.00 ± 0.11	−0.06 ± 0.24	0.47 ± 0.54
σobs,b (km s−1)	6427.16 ± 158.91	6076.60 ± 44.98	5922.66 ± 100.59	4355.02 ± 191.32	6526.60 ± 198.63	4201.36 ± 283.85
λobs,b (Å)	6795.94 ± 3.11	6795.87 ± 1.17	6802.42 ± 2.18	6798.58 ± 2.94	6804.27 ± 3.14	6811.12 ± 5.70
Fluxb (10−17 erg s−1 cm−2)	3685.78 ± 107.11	2174.53 ± 22.13	1669.75 ± 33.36	796.34 ± 34.03	1482.86 ± 44.84	330.64 ± 24.83
σobs,i (km s−1)	1420.75 ± 84.34	971.30 ± 12.54	1012.55 ± 15.51	998.11 ± 14.29	1070.06 ± 13.36	904.97 ± 18.84
λobs,i (Å)	6816.41 ± 1.48	6813.15 ± 0.19	6805.61 ± 0.28	6804.06 ± 0.23	6802.67 ± 0.24	6803.46 ± 0.39
Fluxi (10−17 erg s−1 cm−2)	677.10 ± 62.69	726.85 ± 9.45	853.24 ± 15.79	1223.74 ± 26.65	951.19 ± 16.34	417.99 ± 12.97
σobs,n (km s−1)	⋯	140.43 ± 4.06	121.09 ± 4.59	140.88 ± 49.27	47.20 ± 1.30	102.97 ± 14.33
${\lambda }_{\mathrm{obs},n}$ (Å)	⋯	6812.98 ± 0.08	6810.72 ± 0.10	6813.57 ± 1.56	6809.51 ± 0.03	6812.56 ± 0.17
Fluxn (10−17 erg s−1 cm−2)	⋯	120.79 ± 3.83	109.74 ± 4.45	14.78 ± 5.01	99.93 ± 2.36	29.34 ± 3.20
${\sigma }_{\mathrm{obs},\mathrm{add}}$ (km s−1)	⋯	1405.69 ± 79.96	1069.12 ± 57.61	762.90 ± 123.95	1145.75 ± 49.37	749.03 ± 399.18
${\lambda }_{\mathrm{obs},\mathrm{add}}$ (Å)	6675.12 ± 9.46	6679.37 ± 1.90	6692.49 ± 1.30	6714.40 ± 2.89	6695.31 ± 0.96	6714.74 ± 5.04
Fluxadd (10−17 erg s−1 cm−2)	95.13 ± 35.07	108.08 ± 9.93	158.10 ± 12.25	44.83 ± 10.84	215.18 ± 12.51	14.89 ± 8.61

Note. Same as Table 9. Subscript "'add" denotes the additional blueshifted component. The intermediate and blueshifted components in the LISS spectrum are unresolved and assumed to have the same velocity width.

Download table as:  ASCIITypeset image

The narrow emission line component is partially resolved in the DIS high-resolution spectrum obtained at MJD = 57430.1 ("DIS high 1" in Tables 9 and 10). The rest-frame velocity FWHM (${\mathrm{FWHM}}_{\mathrm{rest}}$) can be estimated from the observed-frame line width (σ_obs), the instrumental broadening width (σ_inst), and the source redshift z as

$$\begin{eqnarray}&&{\mathrm{FWHM}}_{\mathrm{rest}}=2\sqrt{2\mathrm{ln}2}\displaystyle \frac{\sqrt{{\sigma }_{\mathrm{obs}}^{2}-{\sigma }_{\mathrm{inst}}^{2}}}{1+z}.\end{eqnarray} \tag{ 15 }$$

Substituting ${\sigma }_{\mathrm{obs},n}=47.2$ km s⁻¹ and ${\sigma }_{\mathrm{inst}}=20.8$ km s⁻¹ taken from Tables 10 and 8, respectively, to Equation (15), the rest-frame velocity FWHM of the narrow component of KISS15s is evaluated as ${\mathrm{FWHM}}_{\mathrm{rest},n}=114.8$ km s⁻¹. It should also be noted that the Hα line flux of the narrow component is ∼10⁻¹⁵ erg s⁻¹ cm², which is roughly in agreement with the host galaxy Hα line flux measured from the SDSS spectrum (see Section 4.1 for details). Therefore, although it is still possible that there is a flux contribution to the narrow line from the unshocked CSM photoionized by strong radiation from KISS15s, we can assume that the observed narrow emission line component predominantly comes from the foreground/background H ii regions in the host galaxy that may not directly be related to the SN explosion event of KISS15s. We put a rough upper limit on the velocity width of the undetected narrow emission line component from the CSM formed by the progenitor's stellar wind as ${\mathrm{FWHM}}_{w}\lesssim 100$ km s⁻¹, which means that ${v}_{w}\lesssim 100$ km s⁻¹ on the assumption of ${v}_{w}\,\simeq {\mathrm{FWHM}}_{w}$ (e.g., Taddia et al. 2015; see Section 4.4 for further discussion).

Figure 21 shows the temporal evolution of the rest-frame velocity width FWHM (${\mathrm{FWHM}}_{\mathrm{rest}};$ Equation (15)) and the velocity shift relative to the systemic redshift of z = 0.03782 of the decomposed Hα emission line components of KISS15s. The velocity width of the broad component gradually decreases as a function of time from FWHM ∼ 14,000 km s⁻¹ to ∼10,000 km s⁻¹, and the velocity shift only shows weak variability. It is interesting to note that the velocity widths of the intermediate and blueshift excess components are consistent with each other, suggesting that the two components are produced by the same emission mechanism. The line-of-sight velocity of the two intermediate-width components are, however, very different: the velocity shift of the intermediate component decreases from ∼0 km s⁻¹ to about −500 km s⁻¹; on the other hand, that of the blueshift component increases from about −6000 km s⁻¹ to about −4500 km s⁻¹. The coexistence of these two emission line components implies that the emission region has a spherically asymmetric geometry, which may be related to the inhomogeneous structure of the CSM produced by the intrinsic non-spherically symmetric progenitor's stellar wind (see Section 4.3 for further discussion; Borish et al. 2015; Smith 2017, and references therein).

Zoom In Zoom Out Reset image size

Figure 22 presents the light curves of the decomposed Hα emission line luminosities. The broad emission line luminosity monotonically decreases, as with the power-law luminosity evolution of the optical continuum light curve. The similarity of the luminosity evolution between the broad Hα line and optical continuum implies that the broad emission line component and optical continuum emission are powered by the same energy source, which may be the ingoing and/or outgoing intense radiation from the ejecta-CSM interaction region (e.g., Chugai & Danziger 1994; Chevalier & Fransson 2017). The intermediate and blueshifted components show different temporal evolutions with each other, as well as with the broad component. The luminosity of the intermediate component increases during the observations, which may imply that the volume of the ejecta-CSM interaction region is increasing. On the other hand, the blueshift component becomes weaker at the last epoch of the spectroscopic observation, suggesting that this component is produced from a separate region from where the intermediate component emerges.

Zoom In Zoom Out Reset image size

In addition to He iλ5876, the He iλ7065 emission line is clearly visible in the KISS15s spectra (Figure 8), and it seems to have at least two spectral components with intermediate (∼2000 km s⁻¹) and narrow (<300 km s⁻¹) line widths. The broad emission bump at wavelengths of ${\lambda }_{\mathrm{rest}}\sim 8600$ Å is probably due to a blend of broad O iλ8446.

KISS15s shows prominent broad emission line features of the Ca ii IR triplet (λλ8498, 8542, 8662), as is also observed in several other SNe IIn, such as SN 1988Z (e.g., Turatto et al. 1993). Blended Ca ii IR triplet emission shows a temporal evolution similar to that of the Hα emission line, although the detailed spectral decomposition is difficult due to insufficient spectral coverages of our data and blending of multiple broad/intermediate lines of Ca ii and probably O iλ8446 (e.g., Jencson et al. 2016). The strong, broad/intermediate Ca ii triplet emission with no associated P Cygni absorption feature must be emitted by the same region as that with the strong Hα emission line (i.e., the CDS and/or photoionized ejecta; Chugai & Danziger 1994; Dessart et al. 2016).

The origin of the broad bump at the wavelengths of λ_rest ∼ 7300 Å in KISS15s spectra is unclear, but it may be due to a blend of broad lines of He i and Ca ii and several forbidden lines (e.g., Graham et al. 2014; Smith et al. 2017). The flux excess in the wavelength range of ${\lambda }_{\mathrm{rest}}\lt 5700$ Å generally observed in spectra of SNe IIn, which is composed of a blend of Fe ii and several other emission lines (Stathakis & Sadler 1991; Stritzinger et al. 2012), is not observed or is significantly weaker in KISS15s compared to SN 1988Z (Figure 8).