GALEX DETECTION OF SHOCK BREAKOUT IN TYPE IIP SUPERNOVA PS1-13arp: IMPLICATIONS FOR THE PROGENITOR STAR WIND

The Pan-STARRS1 system is a high-etendue wide-field imaging system, designed for dedicated survey observations. The system is installed on the peak of Haleakala on the island of Maui in the Hawaiian island chain. A description of the Pan-STARRS1 system, both hardware and software, is provided by Kaiser et al. (2010). The Pan-STARRS1 optical design (Hodapp et al. 2004) uses a 1.8 m diameter f/4.4 primary mirror and a 0.9 m secondary, and a 3.3 deg diameter field of view. The Pan-STARRS1 imager (Tonry & Onaka 2009) comprises a total of sixty 4800 × 4800 pixel detectors, with 10 μm pixels that subtend 0.258 arcsec.

The PS1 observations were obtained through a set of five broadband filters, which we have designated as g_P1, r_P1, i_P1, z_P1, and y_P1. Although the filter system for Pan-STARRS1 has much in common with that used in previous surveys, such as SDSS (Abazajian et al. 2009), there are important differences. Further information on the passband shapes is described in Stubbs et al. (2010). Photometry is in the "natural" Pan-STARRS1 system, $m=-2.5{\rm log} (flux)+m^{\prime}$, with a single zero-point adjustment m' made in each band to conform to the AB magnitude scale (Tonry et al. 2012).

Images obtained by the Pan-STARRS1 system are processed through the Image Processing Pipeline (IPP; Magnier 2006) on a computer cluster at the Maui High Performance Computer Center. The pipeline runs the images through a succession of stages, including flat-fielding ("de-trending"), a flux-conserving warping to a sky-based image plane (with a plate scale of 0.25 arcsec pixel^-1), masking and artifact removal, and object detection and photometry. Mask and variance arrays are carried forward at each stage of the IPP processing. Photometric and astrometric measurements performed by the IPP system are described in Magnier (2007) and Magnier et al. (2008), respectively.

PS1 has two extragalactic survey modes: the 3π survey observes the entire sky above a declination of −30 deg in five bands with a sparse cadence (a few epochs per month per filter), and the MDS is a deeper survey of 10 fields distributed across the sky in five filters with a daily cadence. This paper uses images and photometry from the PS1 MDS. Observations of three to five MD fields are taken each night and the filters are cycled through in the following pattern: g_P1 and r_P1 in the same night (dark time), followed by i_P1 and z_P1 on the subsequent second and third nights, respectively. Around full moon only y_P1 data are taken. A nightly stacked image is created using a variance-weighted scheme from eight dithered exposures, allowing for the removal of defects like cosmic rays and satellite streaks, and resulting in a 5σ depth of ∼23 mag in the g_P1, r_P1, i_P1, and z_P1 bands, and ∼22 mag in the y_P1 band in the AB system.

The nightly MD stacked images are processed through a frame-subtraction analysis using the photpipe pipeline (Rest et al. 2005), and significant excursions in the difference images are detected using a modified version of DoPHOT (Schechter et al. 1993). Reference templates are constructed for the MD fields from two epochs obtained at the beginning of the observing season. Spectroscopically confirmed SNe are later reprocessed using deep template images, constructed from all nightly stacks with good seeing in the seasons that the SN did not explode, with the routine transphot (Rest et al. 2014), which implements the HOTPANTS image differencing algorithm (Becker et al. 2004), then point-spread function (PSF) photometry is applied using DAOPHOT (Stetson 1987). The weighted average position of the SN is calculated, and forced photometry is then applied in every difference image at that position. The PS1 MDS SN search has yielded hundreds of spectroscopically confirmed SNe, including SNe Ia for cosmology studies (Rest et al. 2014), core-collapse SNe, including 76 SNe IIP (Sanders et al. 2014), as well as several rare, superluminous SNe at high redshift (Chomiuk et al. 2011; Berger et al. 2012; Lunnan et al. 2013).

We conducted an extended survey with GALEX in the NUV during 2012 December through 2013 April of four PS1 MD fields, in order to perform a high-cadence (1 day) time domain survey and to complete GALEX coverage started with the GALEX Time Domain Survey (Gezari et al. 2013) for 9 out of the 10 PS1 MD fields. The FUV detector on GALEX was no longer operational after 2009 May. The one-day cadence of the GALEX observations was chosen to better probe the cooling envelope emission of RSG stars, which has a characteristic timescale of two days (Nakar & Sari 2010; Rabinak & Waxman 2011) and was just barely resolved by the two-day cadence of the GALEX Time Domain Survey for the SN IIP 2010aq (Gezari et al. 2010). The exposure times for each GALEX visit averaged 1.1 ks, resulting in a 3σ depth of 23.5 mag in the NUV in the AB system. Here we present the joint GALEX and PS1 discovery of SN IIP PS1-13arp, for which we catch the onset of a luminous UV burst with GALEX that precedes its standard optical behavior observed by PS1.

PS1-13arp was identified as an optical transient in the PS1 field MD06 by photpipe on UT 2013 April 26 at R.A. 12:18:25.108, decl. +46:37:01.30 (J2000), with a spatial offset of 1.577 arcsec east and 0.275 arcsec north of its host galaxy (SDSS J121824.98+463700.5). Figure 1 shows the PS1 difference image detection in the iband, as well as a color image of the deep reference template with only the host galaxy flux. The SN was also detected as a variable source by GALEX during monitoring of the field PS_NGC4258_MOS107 at the 8.3σ level on UT 2013 April 15. The host galaxy is detected in the NUV and measured from the four epochs before the SN is detected by the GALEX pipeline to have ${\rm NUV}=21.20\pm 0.05$ mag, measured from a 6 arcsec radius circular aperture and including a correction for the total enclosed energy of −0.23 mag (Morrissey et al. 2007). Table 1 gives the SN photometry from GALEX and PS1.

Zoom In Zoom Out Reset image size

Table 1.  GALEX and PS1 Photometry

Filter	Date (MJD)	ϕ (days)a	magb	1σ
NUV	56393.07	−2.9	<23.69	⋯
NUV	56394.03	−2.1	<23.70	⋯
NUV	56395.05	−1.2	<23.63	⋯
NUV	56396.02	−0.4	<23.65	⋯
NUV	56397.04	0.5	21.58	0.06
NUV	56398.00	1.3	21.60	0.06
NUV	56399.02	2.2	22.14	0.10
NUV	56399.98	3.0	22.40	0.07
NUV	56401.01	3.9	22.05	0.15
NUV	56401.96	4.7	23.35	0.12
${{g}_{{\rm P}1}}$	56325.60	−60.8	<23.16	⋯
${{g}_{{\rm P}1}}$	56328.60	−58.2	<23.57	⋯
${{g}_{{\rm P}1}}$	56331.60	−55.6	<23.90	⋯
${{g}_{{\rm P}1}}$	56336.60	−51.3	<23.84	⋯
${{g}_{{\rm P}1}}$	56354.60	−35.9	<23.25	⋯
${{g}_{{\rm P}1}}$	56357.60	−33.3	<23.82	⋯
${{g}_{{\rm P}1}}$	56384.50	−10.3	<23.35	⋯
${{g}_{{\rm P}1}}$	56387.50	−7.7	<23.55	⋯
${{g}_{{\rm P}1}}$	56390.40	−5.2	<23.48	⋯
${{g}_{{\rm P}1}}$	56398.50	1.7	22.34	0.13
${{g}_{{\rm P}1}}$	56401.30	4.1	22.41	0.23
${{g}_{{\rm P}1}}$	56416.40	17.1	22.69	0.18
${{g}_{{\rm P}1}}$	56421.30	21.3	22.98	0.17
${{g}_{{\rm P}1}}$	56427.30	26.4	23.22	0.26
${{g}_{{\rm P}1}}$	56430.40	29.1	24.44	1.13
${{g}_{{\rm P}1}}$	56443.30	40.1	23.41	0.22
${{g}_{{\rm P}1}}$	56451.30	47.0	24.03	0.52
${{g}_{{\rm P}1}}$	56460.30	54.7	24.09	0.46
${{g}_{{\rm P}1}}$	56478.30	70.1	26.19	3.25
${{r}_{{\rm P}1}}$	56325.60	−60.8	<23.18	⋯
${{r}_{{\rm P}1}}$	56328.60	−58.2	<23.37	⋯
${{r}_{{\rm P}1}}$	56331.60	−55.6	<23.40	⋯
${{r}_{{\rm P}1}}$	56332.60	−54.8	<23.51	⋯
${{r}_{{\rm P}1}}$	56336.60	−51.3	<23.41	⋯
${{r}_{{\rm P}1}}$	56354.60	−35.9	<23.17	⋯
${{r}_{{\rm P}1}}$	56357.60	−33.3	<23.24	⋯
${{r}_{{\rm P}1}}$	56371.50	−21.4	<23.31	⋯
${{r}_{{\rm P}1}}$	56384.50	−10.3	<23.26	⋯
${{r}_{{\rm P}1}}$	56387.50	−7.7	<23.14	⋯
${{r}_{{\rm P}1}}$	56390.40	−5.2	<23.05	⋯
${{r}_{{\rm P}1}}$	56398.50	1.7	23.12	0.34
${{r}_{{\rm P}1}}$	56401.30	4.1	22.71	0.22
${{r}_{{\rm P}1}}$	56416.40	17.1	22.48	0.19
${{r}_{{\rm P}1}}$	56420.40	20.5	22.43	0.14
${{r}_{{\rm P}1}}$	56427.30	26.4	22.36	0.16
${{r}_{{\rm P}1}}$	56430.40	29.1	22.61	0.21
${{r}_{{\rm P}1}}$	56443.30	40.1	23.08	0.32
${{r}_{{\rm P}1}}$	56450.30	46.1	24.65	1.86
${{r}_{{\rm P}1}}$	56460.30	54.7	23.16	0.30
${{r}_{{\rm P}1}}$	56478.30	70.1	23.14	0.34
${{i}_{{\rm P}1}}$	56323.60	−62.5	<22.79	⋯
${{i}_{{\rm P}1}}$	56326.60	−59.9	<23.21	⋯
${{i}_{{\rm P}1}}$	56332.60	−54.8	<23.34	⋯
${{i}_{{\rm P}1}}$	56334.50	−53.1	<23.22	⋯
${{i}_{{\rm P}1}}$	56337.60	−50.5	<23.29	⋯
${{i}_{{\rm P}1}}$	56352.50	−37.7	<22.78	⋯
${{i}_{{\rm P}1}}$	56355.50	−35.1	<23.29	⋯
${{i}_{{\rm P}1}}$	56358.60	−32.5	<23.15	⋯
${{i}_{{\rm P}1}}$	56369.40	−23.2	<23.05	⋯
${{i}_{{\rm P}1}}$	56388.50	−6.8	<22.96	⋯
${{i}_{{\rm P}1}}$	56391.40	−4.4	<23.16	⋯
${{i}_{{\rm P}1}}$	56393.30	−2.7	<23.23	⋯
${{i}_{{\rm P}1}}$	56396.50	0.0	<22.96	⋯
${{i}_{{\rm P}1}}$	56399.40	2.5	23.05	0.29
${{i}_{{\rm P}1}}$	56403.40	5.9	22.62	0.20
${{i}_{{\rm P}1}}$	56411.40	12.8	22.27	0.35
${{i}_{{\rm P}1}}$	56414.40	15.4	22.26	0.17
${{i}_{{\rm P}1}}$	56417.40	17.9	22.39	0.18
${{i}_{{\rm P}1}}$	56420.40	20.5	22.39	0.15
${{i}_{{\rm P}1}}$	56422.30	22.1	22.32	0.16
${{i}_{{\rm P}1}}$	56432.30	30.7	22.47	0.18
${{i}_{{\rm P}1}}$	56444.30	41.0	22.44	0.17
${{i}_{{\rm P}1}}$	56447.30	43.6	22.45	0.17
${{i}_{{\rm P}1}}$	56450.30	46.1	22.52	0.25
${{i}_{{\rm P}1}}$	56452.30	47.8	22.50	0.19
${{i}_{{\rm P}1}}$	56455.30	50.4	22.53	0.18
${{i}_{{\rm P}1}}$	56458.30	53.0	22.49	0.21
${{i}_{{\rm P}1}}$	56462.30	56.4	22.71	0.21
${{i}_{{\rm P}1}}$	56479.30	71.0	22.74	0.21
${{z}_{{\rm P}1}}$	56324.60	−61.6	<22.64	⋯
${{z}_{{\rm P}1}}$	56338.60	−49.6	<22.71	⋯
${{z}_{{\rm P}1}}$	56348.60	−41.1	<22.47	⋯
${{z}_{{\rm P}1}}$	56351.50	−38.6	<22.62	⋯
${{z}_{{\rm P}1}}$	56353.50	−36.9	<22.56	⋯
${{z}_{{\rm P}1}}$	56356.50	−34.3	<22.76	⋯
${{z}_{{\rm P}1}}$	56359.60	−31.6	<22.61	⋯
${{z}_{{\rm P}1}}$	56364.50	−27.4	<22.61	⋯
${{z}_{{\rm P}1}}$	56370.40	−22.4	<22.73	⋯
${{z}_{{\rm P}1}}$	56383.50	−11.1	<22.63	⋯
${{z}_{{\rm P}1}}$	56386.30	−8.7	<22.67	⋯
${{z}_{{\rm P}1}}$	56389.50	−6.0	<22.59	⋯
${{z}_{{\rm P}1}}$	56392.30	−3.6	<22.67	⋯
${{z}_{{\rm P}1}}$	56394.30	−1.9	<22.48	⋯
${{z}_{{\rm P}1}}$	56397.40	0.8	<22.59	⋯
${{z}_{{\rm P}1}}$	56400.40	3.4	23.17	0.63
${{z}_{{\rm P}1}}$	56402.30	5.0	22.11	0.27
${{z}_{{\rm P}1}}$	56404.30	6.7	22.40	0.29
${{z}_{{\rm P}1}}$	56410.40	11.9	22.60	0.36
${{z}_{{\rm P}1}}$	56412.40	13.6	22.48	0.32
${{z}_{{\rm P}1}}$	56415.40	16.2	22.29	0.25
${{z}_{{\rm P}1}}$	56421.40	21.4	22.51	0.30
${{z}_{{\rm P}1}}$	56423.40	23.1	22.10	0.22
${{z}_{{\rm P}1}}$	56429.30	28.1	22.32	0.27
${{z}_{{\rm P}1}}$	56451.30	47.0	22.42	0.30
${{z}_{{\rm P}1}}$	56453.30	48.7	22.95	0.64
${{z}_{{\rm P}1}}$	56456.30	51.3	22.31	0.27
${{z}_{{\rm P}1}}$	56459.30	53.8	22.55	0.31
${{z}_{{\rm P}1}}$	56461.30	55.6	22.61	0.36
${{z}_{{\rm P}1}}$	56463.30	57.3	22.66	0.35
${{z}_{{\rm P}1}}$	56480.30	71.8	22.00	0.35

^aRest-frame days since the estimated time of shock breakout on MJD 56396.5. ^bAB magnitudes, not corrected for Galactic extinction, and with the host galaxy flux removed.

Download table as:  ASCIITypeset images: 1 2

We obtained spectroscopy of the UV/optical transient with Gemini/GMOS on UT 2013 May 4. We obtained three exposures with a 1.0 arcsec slit and an R400 grating of 1200 s each, resulting in a spectral resolution of ∼4 Å with a dither along the slit and a change in the grating central wavelength in order to avoid bad columns and to cover the gap in wavelength coverage for the grating. The spectrum was overscan-corrected, bias-subtracted, flat-fielded, wavelength-calibrated, and flux-calibrated using standard IRAF routines. Observations of a standard star (Feige 34) were used for flux calibration and corrections for telluric absorption. The 2D spectrum displays strong, spatially extended emission lines (Hα, [N ii] $\lambda \lambda 6583,6548$, [S ii]$\lambda \lambda 6716,6731$) indicative of star formation in the host galaxy, from which we measure a redshift of $z=0.1665\pm 0.001$.

We fitted the rest-frame host galaxy broadband spectral energy distribution (SED) from GALEX FUV and NUV (Martin et al. 2005), PS1 g_P1, r_P1, i_P1, z_P1, y_P1 deep stacks with u-band coverage from CFHT (S. Heinis et al. 2015, in preparation), and infrared from WISE at 3.4 and 4.6 μm (Wright et al. 2010) with the code CIGALE (Noll et al. 2009) using version v0.3, implemented in Python. CIGALE combines a UV–optical stellar SED with a dust component emitting in the infrared, and ensures a full energy balance between the dust absorbed stellar emission and its re-emission in the IR. The fit requires both an old stellar population with an exponentially declining star formation rate with $\tau =6.7\pm 2$ Gyr as well as a recent stellar population with a moderate star formation rate of $3.6\pm 1.2\;{{M}_{\odot }}$ yr⁻¹, with a total stellar mass of ${\rm log} \left( M/{{M}_{\odot }} \right)=10.4\pm 0.1$. Note that this mass is consistent with the gas-phase metallicity that we estimate from the ratio of the strong nebular emission lines in the host galaxy spectrum. We measure ${\rm log} ([{\rm NII}]\lambda 6584/H\alpha )=-0.37\pm 0.06$, which corresponds to a metallicity of ${\rm log} ({\rm O}/{\rm H})+12\approx 9$ (Kewley & Dopita 2002), and is in good agreement with the mass–metallicity relation for star-forming galaxies (Tremonti et al. 2004).

We extract a 1D SN spectrum by first extracting a 1D spectrum using a 5 pixel aperture centered on the SN (including SN and galaxy light), and then extract a 1D spectrum with an aperture centered on a region 5 pixels away from the SN center, with only galaxy light. We isolate the SN spectrum by subtracting the 1D galaxy spectrum from the 1D SN+galaxy spectrum. We coadded the first two exposures, which had the highest SN signal. We show the coadded SN spectrum in Figure 2, in comparison to the spectrum of SN 2006bp at a similar phase since shock breakout from Quimby et al. (2007). We detect the P Cygni absorption features of Hβ, Fe ii $\lambda 5169$, Na i $\lambda 5892$, and Hα. We fit the strong broad Hα line with a Gaussian with an FWHM = 7500 ± 1200 km s⁻¹ (shown in Figure 3). The line appears mostly symmetric, and blueshifted by 2200±100 km s⁻¹ from the galaxy's narrow Hα line. We only see weak evidence of blueshifted absorption, as one would expect in a P Cygni profile. The weak blueward absorption is consistent with a young SN IIP, which typically do not develop a strong P Cygni Balmer line shape until ∼3 weeks after explosion (Quimby et al. 2007; Dessart et al. 2008). However, the weak P Cygni absorption could also reveal unusual conditions in the SN envelope or circumstellar environment. Unfortunately, without a later epoch of spectroscopy, we can only speculate as to whether PS1-13arp would have developed a stronger P Cygni line profile at late times.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

While a burst of neutrinos marks the time of core collapse, the SN shock breakout is the first electromagnetic signature of an SN explosion. The SN shock "breaks out" when the radiative diffusion timescale,

$$\begin{eqnarray*}&&{{t}_{{\rm d}}}\sim \frac{3{{(\delta R)}^{2}}}{\lambda c}=\frac{3\kappa \rho {{(\delta R)}^{2}}}{c}\approx \frac{3\tau {{H}_{\rho }}}{c},\end{eqnarray*}$$

where $\delta R$ is the depth of the shock, $\lambda =1/\kappa \rho$ is the photon's mean free path, κ is the opacity, τ is the optical depth, and ${{H}_{\rho }}$ is the atmospheric scale height, is shorter than the shock travel time, ${{t}_{s}}=\delta R/{{v}_{s}}$ (Arnett 1996). For a standard RSG envelope density and radial structure, the duration of the radiative precursor is expected to be ∼1000 s (Klein & Chevalier 1978; Matzner & McKee 1999). However, the shock breakout pulse will be prolonged up to ∼1 hr due to light-travel time effects in the largest RSGs (Levesque et al. 2005), or up to days in the presence of a dense circumstellar wind (Ofek et al. 2010; Chevalier & Irwin 2011; Moriya et al. 2011).

The rise time and duration of the UV burst in PS1-13arp are a factor of greater than ∼50 longer than both the radiative diffusion time at shock breakout from Matzner & McKee (1999),

$$\begin{eqnarray*}&&{{t}_{{\rm sbo}}}=860{{k}^{-0.58}}{{\left( {{\rho }_{1}}/{{\rho }_{*}} \right)}^{-0.28}}E_{51}^{-0.79}M_{15}^{0.21}R_{500}^{2.16}\;{\rm s},\end{eqnarray*}$$

where R₅₀₀ is the stellar radius in units of 500 ${{R}_{\odot }}$, M₁₀ is the stellar mass in units of 10 ${{M}_{\odot }}$, and E₅₁ is the energy of the explosion in units of 10⁵¹ erg s⁻¹, and the light-crossing time of an RSG,

$$\begin{eqnarray*}&&{{t}_{lc}}={{R}_{\star }}/c=1160{{R}_{500}}\;{\rm s}.\end{eqnarray*}$$

Even assuming an envelope density a factor of 1000 less dense than the standard stellar structure models of Woosley & Heger (2007), as was done in Schawinski et al. (2008), one can increase ${{t}_{{\rm sbo}}}$ only by a factor of ∼7. However, such an extended, tenuous structure is better attributed to a circumstellar shell of material rather than the stellar envelope itself (Falk & Arnett 1977). The long duration of the UV burst suggests that either the shock is "breaking out" in a wind around the progenitor star, or it is associated with the post-shock-breakout cooling envelope emission previously detected in the UV in SNe IIP with GALEX (Gezari et al. 2008, 2010; Schawinski et al. 2008; Ganot et al. 2014). In the following sections, we address both scenarios.

In the cooling envelope scenario, the GALEX observations did not catch the brief radiative precursor from shock breakout, but rather the UV burst is from the adiabatically expanding shock-heated SN ejecta. As the hot envelope expands and cools, the peak energy of the thermal emission shifts from the soft X-rays into the UV when $3kT=h{{\nu }_{{\rm NUV}}}(1+z)$, where ${{\nu }_{{\rm NUV}}}=2316$ Å (or T = 2.09 eV), causing a peak in the NUV light curve. The rate of cooling depends sensitively on the radius of the progenitor, since the smaller the progenitor, the more energy is lost to adiabatic expansion. Using the analytical model for the cooling envelope emission for an RSG, the largest likely SN II progenitor, from Nakar & Sari (2010), this occurs at a time

$$\begin{eqnarray*}&&{{t}_{{\rm peak},{\rm NUV}}}=1.9\;M_{15}^{-0.23}R_{500}^{0.68}E_{51}^{0.20}{\rm days}.\end{eqnarray*}$$

The rise time in PS1-13arp is constrained from the time between the last NUV non-detection and the first NUV detection to be $\lt 0.87$ days in the SN rest frame, or from the fitted time of shock breakout from Section 3 to be $\lt 0.5\pm 0.3$ days. In order for this UV rise time to be compatible with PS1-13arp, it would require $M_{15}^{-0.23}R_{500}^{0.68}E_{51}^{0.20}\lt 0.46$. The UV peak can occur much sooner for a smaller star, such as a BSG (${{t}_{{\rm peak}}}\approx 0.54$ days). However, the late-time optical light curve of PS1-13arp is incompatible with such a compact progenitor. For a BSG with a typical radius of a factor of 10 smaller than an RSG, one would expect a redder and shorter-duration plateau (Kleiser et al. 2011; Dessart et al. 2013), which is inconsistent with the excellent fit of the plateau of SN PS-13arp to the SN 2006bp model.

Figure 6 shows the evolution of the UV/optical spectral SED of PS1-13arp over time, with NUV fluxes linearly interpolated to the time of the PS1 observations. Unfortunately, we do not have a strong constraint on the temperature of the UV burst. We can only apply a lower limit from the z-band non-detection of $T\gt 2.2\times {{10}^{4}}$ K. Given that the next epoch, at a phase (ϕ) of 1.7 rest-frame days since our estimated time of shock breakout, is hotter, with a least-squares fitted temperature of $T=(2.8\pm 0.3)\times {{10}^{4}}$ K, we adopt this as our lower limit to the temperature of the UV burst at $\phi =0.8$ days, since the SN should be cooling with time. The NUV flux density, corrected for Galactic extinction, then corresponds to a peak bolometric luminosity of ${{L}_{{\rm bol}}}\gt 1.7\times {{10}^{43}}$ erg s⁻¹, which is an order of magnitude larger than expected for the cooling envelope at the time of the NUV peak from Nakar & Sari (2010):

Zoom In Zoom Out Reset image size

$$\begin{eqnarray*}&&{{L}_{{\rm RSG},{\rm peak}}}=2.70\times {{10}^{42}}M_{15}^{-0.87}{{R}_{500}}E_{51}^{0.96}{\rm erg}\;{{{\rm s}}^{-1}}.\end{eqnarray*}$$

The observed NUV absolute magnitude of PS1-13arp is also brighter than the corresponding peak absolute magnitude of this model of ${{M}_{{\rm NUV}}}\sim -16$ mag. For a BSG, the luminosity at the time of the NUV peak is even lower (${{L}_{{\rm BSG},{\rm peak}}}\sim 8\times {{10}^{41}}$ erg s⁻¹). Figure 7 compares the early UV/optical light curve of PS1-13arp with the RSG SN shock breakout and cooling envelope emission model from Nakar & Sari (2010), scaled to match the late-time SN 2006bp model fit. While the observations and model are in good agreement close to the cooling envelope model peak and later ($t\gt 2$ days), there is a clear prolonged UV excess above the model at earlier times. We conclude that the peak of the observed UV burst is more luminous than expected for the cooling envelope emission following shock breakout, and thus we now investigate whether or not the UV burst could be from the shock breakout pulse itself but prolonged in time due to the presence of a circumstellar wind.

Zoom In Zoom Out Reset image size

The "shock breakout" is defined as when the radiation from the SN shock is able to escape in front of the shock, i.e., when the radiative diffusion time is shorter than the shock travel time (see Section 4.1). For a bare star, this occurs in the outer envelope of the star (not actually at its surface). However, when the star is enshrouded in a dense wind, the shock breakout is delayed and occurs outside the surface of the star in the circumstellar wind. The theoretical literature still refers to this phenomenon as the shock breakout (Balberg & Loeb 2011; Chevalier & Irwin 2011; Ofek et al. 2013), even though the energy deposited from the shock is diffusing out (escaping) into the wind instead of the stellar envelope. A radiation-dominated shock in a uniform medium has a characteristic thickness in units of optical depth of ${{\tau }_{s}}\approx c/{{v}_{s}}$. The shock will breakout in the wind if ${{\tau }_{w}}\gt {{\tau }_{s}}$, where ${{\tau }_{w}}$ is the thickness of the shock traveling through the wind in units of optical depth. Here we assume the wind is from steady mass loss from the progenitor star $\dot{M}$, resulting in a density distribution ${{\rho }_{w}}=\dot{M}/4\pi {{r}^{2}}{{v}_{w}}\equiv D{{r}^{-2}}$, where $D\equiv 5.0\times {{10}^{16}}\left( {{{\dot{M}}}_{0.01}}/{{v}_{w,10}} \right)$ g cm⁻¹ and ${{\dot{M}}_{0.01}}=\dot{M}/(0.01{{M}_{\odot }}$ yr⁻¹), and we assume the slow wind velocity characteristic of an RSG, ${{v}_{w,10}}={{v}_{w}}/(10$ km s⁻¹). Chevalier & Irwin (2011) approximate the optical depth of a shock wave traveling through the wind as ${{\tau }_{w}}\approx R_{d}^{-1}\kappa D$, where R_d is the depth of the shock. Hence, shock breakout occurs when ${{\tau }_{w}}\approx c/{{v}_{s}}$, such that ${{R}_{d}}=\kappa D{{v}_{s}}/c$. The luminosity of the breakout radiation is given by ${{L}_{{\rm bo}}}\approx {{E}_{{\rm rad}}}/{{t}_{{\rm bo}}}$, where ${{E}_{{\rm rad}}}$ is the radiative energy in the breakout shell and ${{t}_{{\rm bo}}}={{R}_{d}}/{{v}_{s}}=\kappa D/c=6.6k\left( {{{\dot{M}}}_{0.01}}/{{v}_{w,10}} \right)$ days, where ${{t}_{{\rm bo}}}$ is the time of shock breakout and $k=\kappa /(0.34$ cm² g⁻¹). This is given in Chevalier & Irwin (2011) as ${{L}_{{\rm bo}}}\approx {{E}_{{\rm rad}}}/{{t}_{{\rm bo}}}$, and thus

$$\begin{eqnarray*}&&{{L}_{{\rm bo}}}=6.0\times {{10}^{43}}{{k}^{-0.6}}E_{51}^{1.2}M_{15}^{-0.6}{{\left( {{{\dot{M}}}_{0.01}}/{{v}_{w,10}} \right)}^{-0.2}}\;{\rm erg}\;{{{\rm s}}^{-1}},\end{eqnarray*}$$

with a corresponding temperature of

$$\begin{eqnarray*}&&{{T}_{{\rm rad}}}=1.1\times {{10}^{5}}{{k}^{-0.25}}{{\left( {{t}_{d,1}} \right)}^{-0.25}}K,\end{eqnarray*}$$

where ${{t}_{d,1}}={{t}_{d}}/(1\;{\rm day})$. Given the observed rise time of ${{t}_{{\rm rise}}}\lesssim 0.87$ days in the SN rest frame and the weak dependence of T on t_d, we assume a shock breakout radiation temperature of $T\approx {{10}^{5}}$ K, which is allowed by our observed SED at the UV peak based on the z-band upper limit during the UV burst. Given our observed Galactic-extinction corrected luminosity density in the NUV of ${{L}_{{{\nu }_{e}}=(1+z){{\nu }_{o}}}}={{f}_{{{\nu }_{o}}}}4\pi d_{{\rm L}}^{2}/(1+z)=6.25\times {{10}^{27}}$ erg s⁻¹ Hz⁻¹ and assuming radiation from an optically thick blackbody, for which ${{L}_{\nu }}=\left( 2\pi h/{{c}^{2}} \right){{\nu }^{3}}/\left( {{e}^{h\nu /kT}}-1 \right)4\pi R_{{\rm bb}}^{2}$, one gets ${{R}_{{\rm bb}}}=5.8\times {{10}^{13}}$ cm and a bolometric luminosity of $2.4\times {{10}^{44}}$ erg s⁻¹ (a factor of 14 greater than the observed lower limit). The inferred radius is close to the upper range of radii measured for RSGs ($2-6\times {{10}^{13}}$ cm; Levesque et al. 2005) and is potentially situated outside the stellar envelope in the circumstellar wind. Given the rise time of $\lesssim 0.87$ days, this radius then corresponds to a shock velocity of ${{v}_{s}}\gtrsim 7\times {{10}^{3}}$ km s⁻¹, similar to the velocity width of the detected broad Hα line.

The Chevalier & Irwin (2011) model places two independent constraints on the mass-loss rate ($\dot{M}$). The first is from the timescale of the event. The observed rise time corresponds to $\dot{M}\lesssim 1.3\times {{10}^{-3}}\left( {{v}_{w,10}}/k \right){{t}_{d,1}}{{M}_{\odot }}$ yr⁻¹. The duration of the shock breakout signal in the Chevalier & Irwin (2011) wind model is on the order of the rise time to peak. Thus, the rise time in PS1-13arp cannot be much shorter than the duration of the UV burst of ≳0.82 days in the SN rest frame. Thus, we favor a mass-loss rate closer to ${{10}^{-3}}\;{{M}_{\odot }}$ yr⁻¹. The observed luminosity corresponds to $\dot{M}\sim 1\times {{10}^{-5}}\left( {{v}_{w,10}} \right){{k}^{-3}}E_{51}^{6}M_{15}^{-3}\;{{M}_{\odot }}$ yr⁻¹. Given the strong sensitivity of the peak luminosity to E₅₁, this could be easily reconciled with the mass-loss rate estimated from the timing of the event by adopting an explosion energy of ${{E}_{51}}\sim 1.5$. Figure 8 plots the parameter space for peak bolometric luminosity versus rise time to peak in the NUV for shock breakout in a wind and cooling envelope emission for BSGs and RSGs with a range of ${{E}_{51}}=(0.5-1.5)\times {{10}^{51}}$ erg, $R=250-850\;{{R}_{\odot }}$ for the RSG and $R=25-85$ for the BSG, and $M=10-25\;{{M}_{\odot }}$. The observed peak luminosity of PS1-13arp is well above that expected for the cooling envelope phase, and is within the range of luminosities and timescales for a burst powered by shock breakout in a wind.

Zoom In Zoom Out Reset image size

The upper limit on $\dot{M}$ from the observed rise time is consistent with the lack of excess thermal radiation in the optical bands at later times. Moriya et al. (2011) model the light curves of SNe IIP as a function of $\dot{M}$, and find that for $\dot{M}\gt {{10}^{-3}}{{M}_{\odot }}$ yr⁻¹ there is an observable excess in the UV and optical bands due to the conversion of the kinetic energy of the ejecta into thermal energy via interaction with the CSM. The mass-loss rates in RSGs inferred from spectroscopic studies range from ${{10}^{-7}}\;{\rm to}\;{{10}^{-4}}\dot{M}$ yr⁻¹ (van Loon et al. 2005; Mauron & Josselin 2011). Thus, while the inferred mass-loss rate from the luminosity and duration of the shock breakout is on the high end of the observed distribution, these studies are sensitive to the time-averaged mass-loss rate and do not sample short-lived, episodic enhanced mass-loss that may occur in pre-supernova RSGs (Meynet et al. 2014), perhaps as a result of pulsation-driven superwinds (Yoon & Cantiello 2010). Unfortunately, we do not have a sufficient signal-to-noise ratio (S/N) in the spectrum of PS1-13arp to look for a high-velocity absorption feature or "notch," predicted to be a signature of the excitation of ejecta by X-rays produced by circumstellar interaction (Chugai et al. 2007).

X-ray and radio observations of SNe IIP also put conservative upper limits on the mass-loss rates of pre-SN RSGs to $\lt {{10}^{-6}}-{{10}^{-5}}\;{{M}_{\odot }}$ yr⁻¹ from the lack of thermal bremsstrahlung emission or synchrotron radiation from the interaction of the SN shock wave with the CSM (Chevalier et al. 2006; Dwarkadas 2014). Radio and X-ray emission from SN IIP 1999em were interpreted as the result of circumstellar interaction with a stellar wind of $\dot{M}\sim 2\times {{10}^{-6}}\;{{M}_{\odot }}$ yr⁻¹ (Pooley et al. 2002). Mass-loss rates should scale with the RSG luminosity, and thus with stellar mass. Sanders et al. (2014) were not able to put a direct constraint on the progenitor mass of PS1-13arp from the comparison of the bolometric light curve to hydrodynamic models; however, they did infer an ejected nickel mass from a comparison of the late-time bolometric light curve of PS1-13arp to SN 1987 A of ${{M}_{{\rm Ni}}}=0.34_{-0.22}^{+0.20}{{M}_{\odot }}$. This amount of nickel is on the high-mass tail of the distribution for their SNe IIP sample, and could be indicative of a more massive progenitor RSG for PS1-13arp. Another clue is the weak P Cygni absorption in the Hα profile of PS1-13arp at $t\sim 17$ days. Weak P Cygni absorption has been found in brighter than normal SNe IIP and is explained by lower envelope masses and higher mass-loss rates (Gutiérrez et al. 2014).