LATE-TIME CIRCUMSTELLAR INTERACTION IN A SPITZER SELECTED SAMPLE OF TYPE IIn SUPERNOVAE

Warm Spitzer/IRAC (Fazio et al. 2004) obtained a second epoch of data for the 10 SNe IIn discovered to have late-time IR emission by Fox et al. (2011) and the well-studied Type IIn SNe 2002bu, 2003lo, and 2010jl (Program PID 80023). Table 1 lists the observational details. The Spitzer Heritage Archive⁴ provided access to the Post Basic Calibrated Data, which are already fully coadded and calibrated. The background flux in most of the SN host galaxies is bright and exhibits rapid spatial variations. Although template subtraction is a commonly used technique to minimize photometric confusion from the underlying galaxy, no pre-SN observations exist. Instead, photometry was performed using the DAOPHOT point-spread function (PSF) photometry package in IRAF (see Figure 1).⁵

Zoom In Zoom Out Reset image size

WISE mapped the entire sky during 2010 at 3.4, 4.6, 12, and 22 μm (Wright et al. 2010). WISE therefore included the same SNe that were observed by Spitzer. The NASA/IPAC Infrared Science Archive⁶ provided fully calibrated images of each field. Like the Spitzer data described above, photometry was performed using the DAOPHOT PSF photometry package in IRAF. Pixel fluxes were converted from MJy sr⁻¹ to mJy according to the Explanatory Supplement to the WISE All-Sky Data Release Products, which discusses the pixel size, zero points, and aperture correction in detail. Poor resolution and sensitivity resulted in nondetections for all SNe in the 12 and 22 μm bands, so we do not include that photometry here. Table 2 lists the fluxes and upper limits.

Tables 3 and 4 and Figures 2 and 3 summarize the optical photometry and spectra. Data were obtained with the dual-arm Low Resolution Imaging Spectrometer (LRIS; Oke et al. 1995) mounted on the 10 m Keck I telescope and the DEep Imaging Multi-Object Spectrograph (DEIMOS; Faber et al. 2003) mounted on the 10 m Keck II telescope. Keck/LRIS spectra were obtained using the 600/4000 or 400/3400 grisms on the blue side and the 400/8500 grating on the red side, along with a 1'' wide slit. This resulted in a wavelength coverage of 3200–9200 Å and a typical resolution of 5–7 Å. Keck/DEIMOS spectra were obtained using the 1200/7500 grating, along with a 1'' wide slit. This resulted in a wavelength coverage of 4750–7400 Å and a typical resolution of 3 Å. Most observations had the slit aligned along the parallactic angle to minimize differential light losses (Filippenko 1982); moreover, LRIS is equipped with an atmospheric dispersion corrector.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The photometric images were reduced using standard CCD processing techniques in IDL and Python, utilizing online astrometry programs SExtractor and SWarp⁷ (D. Perley 2012, private communication). Photometry was performed using the DAOPHOT PSF photometry package in IRAF. Calibration was performed using field stars with reported fluxes in the Sloan Digital Sky Survey Data Release 9 Catalogue (Ahn et al. 2012). The spectra were reduced using standard techniques (e.g., Foley et al. 2003; Silverman et al. 2012). Routine CCD processing and spectrum extraction were completed with IRAF, and the data were extracted with the optimal algorithm of Horne (1986). We obtained the wavelength scale from low-order polynomial fits to calibration-lamp spectra. Small wavelength shifts were then applied to the data after cross-correlating a template sky to an extracted night-sky spectrum. Using our own IDL routines, we fit a spectrophotometric standard-star spectrum to the data in order to flux calibrate the SN and to remove telluric absorption lines (Wade & Horne 1988; Matheson et al. 2000).

Assuming only thermal emission, the mid-IR photometry provides a strong constraint on the dust temperature, mass, and thus IR luminosity. Assuming optically thin dust with mass M_d and particle radius a, at a distance d from the observer, thermally emitting at a single equilibrium temperature T_d, the flux can be written as (e.g., Hildebrand 1983)

$$\begin{equation} F_\nu = \frac{M_{\rm d} B_\nu (T_{\rm d}) \kappa _\nu (a)}{d^2}, \end{equation} \tag{ 1 }$$

where B_ν(T_d) is the Planck blackbody function and κ_ν(a) is the dust mass absorption coefficient.

For simple dust populations of a single size composed entirely of either silicate or graphite, the IDL MPFIT function (Markwardt 2009) finds the best fit (see Figure 1) of Equation (1) by varying M_d and T_d to minimize the value of χ². The absorption coefficients, κ, are given in Figure 4 of Fox et al. (2010). With only two mid-IR fluxes, our fits are limited to a single component. Table 5 lists the best-fit parameters for graphite grains of size a = 0.1 μm (found to be the best-fit value for these SNe by Fox et al. 2011). Figure 2 plots the corresponding IR luminosity evolution for each SN, as well as the luminosities derived from R-band and X-ray observations.

The forward shock will eventually destroy the dust via sputtering, resulting in a declining mid-IR component. For a spherically symmetric dust shell, the shock will reach the dust shell at a time t ≳ r_bb/v_s, where v_s is the forward-shock velocity and r_bb is the blackbody radius, which is derived from the IR luminosity in Table 5 assuming an optically thick shell. This equation places a lower limit on the timescale over which dust destruction can occur. For example, the forward shock may decelerate with time. Also, an optically thin dust shell composed of smaller grains emitting with modified blackbodies requires a larger shell radius to generate the observed flux. Both of these effects increase the time over which it takes the forward shock to reach the dust shell.

For each SN detected by Spitzer, Figure 4 plots the blackbody radius and dust temperature. The diagonal arrows signify that these values are upper limits given the limitations of the single-component fit in Figure 1. While the numbers vary slightly, a blackbody radius r_bb ≲ 0.1 lt-yr and a shock velocity v_s ≈ 3000 km s⁻¹ will result in a mid-IR plateau that continues for roughly t ≲ 10.5 yr. Once the shock reaches the shell, the luminosity decline will occur on timescales consistent with the dust-shell radius (t_decline = 2r_bb/c ≈ 0.2 yr), as the dust destruction on the near side of the shell will be observed before dust destruction on the far side.

Zoom In Zoom Out Reset image size

We search for evidence of dust destruction between the first and second epochs of Spitzer observations. Figure 2 shows that SNe 2005gn and 2008J are not detected on the second epoch of observations, which occurred at 6.1 and 3.7 yr post-explosion, respectively. Assuming a 3000 km s⁻¹ shock, these results place upper limits on the dust-shell radii of r_d < 0.06 and 0.036 lt-yr. A slower shock would result in a proportionally smaller shell radius. While still detected, SN 2005cp also shows evidence of fading nearly 6.2 yr post-explosion. The horizontal arrows in Figure 4 illustrate both the lower and upper limits on the dust-shell radius derived for SNe 2005gn and 2008J.

For the scenario in which the dust shell is heated by continuous radiation generated by ongoing CSM interaction, an optical and/or X-ray counterpart should exist. Equation (2) calculates this luminosity as a function of the observed dust temperature, T_d, and shell radius, r_d. Along with the observed dust temperatures and radii, Figure 4 plots the theoretical dust temperature as a function of shell radius for contours of constant luminosity. In most cases, the observed dust parameters are consistent with optical and/or X-ray luminosities 10⁸ L_☉ ≲ L_opt/X ≲ 10⁹ L_☉.

Figure 2 plots the observed optical and X-ray luminosities for each SN. For SNe 2008gm and 2008en, the optical and/or X-ray luminosities are fairly consistent with the contours in Figure 4. The optical and/or X-ray luminosities may seem a little low compared to the expected values given by the dust parameters in Figure 4, but we stress that the observed dust parameters are only upper limits due to the limitations of a single-component fit. Furthermore, even in an optically thin scenario (i.e., τ < 1), 50% of the optical flux can be absorbed by the dust. Overall, the optical/X-ray fluxes are sufficient to power the mid-IR emission. Although Figure 2 excludes previously published optical data, observations by Smith et al. (2009b), Stritzinger et al. (2012), and Zhang et al. (2012) yield similar conclusions for SNe 2005ip, 2006jd, and 2010jl, respectively. Due to the high background flux from their underlying galaxies, upper limits for SNe 2005cp, 2007rt, and 2008cg (see Table 3) do not rule out the presence of high-energy emission from CSM interaction. No observations were acquired of SN 2006qq.

SN 2010jl does stand out, however, with a relatively large dust temperature for the derived blackbody radius. Andrews et al. (2011) derive similar parameters from Spitzer observations obtained on day 90 post-explosion and suggest that these characteristics are more consistent with an IR echo powered by the initial SN flash. Optical and X-ray observations, however, reveal clear signatures of CSM interaction. Continued monitoring will ultimately constrain the dust-shell size and heating source.

Although Spitzer did not detect SNe 2005gn and 2008J, the dust shell upper limits (0.6 and 0.036 lt-yr, respectively) derived in Section 3.1 offer valuable constraints on their possible heating mechanism. Assuming dust temperatures (737 K and 704 K) derived from the first epoch of Spitzer observations by Fox et al. (2011), Figure 4 shows that the dust never required an optical and/or X-ray luminosity greater than 10^8.5 L_☉. While we do not have late-time optical observations of these two SNe, such limits are consistent with the other optical and X-ray observations.

The observed optical and X-ray luminosities are consistent with the physical scenario described above, but the source of the emission is not immediately clear. For example, the emission may originate from ongoing CSM interaction, but it may also be associated with an underlying H ii region. To identify distinct signatures of CSM interaction, we obtained ground-based optical spectra of the four SNe detected at visual wavelengths (SNe 2005ip, 2006jd, 2008en, and 2008gm) and SN 2005cp, as summarized in Table 4 and Figure 3. While SN 2005ip exhibits clear signs of CSM interaction via the high-ionization Fe lines and broadened Hα, the evidence in the other SNe is more ambiguous.

Unlike the LRIS spectra, the moderate-resolution DEIMOS spectra resolve down to <100 km s⁻¹ and can differentiate between H ii regions and CSM interaction, which typically has post-shock velocities >100 km s⁻¹. Figure 5 illustrates these differences with the Hα line. SNe 2005ip and 2006jd both have distinct, relatively broad (>1000 km s⁻¹) features associated with the forward shock and a narrower (∼100–200 km s⁻¹) component likely associated with the pre-shocked CSM (Fox et al. 2011). The line-profile asymmetry (a more prominent blue side) of the broader component is likely due to dust absorption (Smith et al. 2009b; Stritzinger et al. 2012). The broad feature in SN 2008en is less obvious, particularly in the LRIS spectrum. The moderate-resolution DEIMOS spectrum, however, resolves the two components. The LRIS spectrum of SN 2008gm also does not rule out a component of ∼500 km s⁻¹. The DEIMOS spectra of SNe 2005cp and 2008gm, however, show only a narrow (<100 km s⁻¹) line suggestive of an H ii region.

Zoom In Zoom Out Reset image size

The mid-IR traces the characteristics of the CSM at the dust-shell radius. Assuming a dust-to-gas mass ratio expected in the H-rich envelope of a massive star, Z_d = M_d/M_g ≈ 0.01, the dust-shell mass can be tied to the progenitor mass-loss rate,

$$\begin{equation} \dot{M} = \frac{M_{\rm d}}{Z_{\rm d} \Delta r} v_{w} \end{equation} \tag{ 3 }$$

$$\begin{equation} = \frac{3}{4} \left (\frac{M_{\rm d}}{M_{\odot }}\right) \left (\frac{v_{w}}{120 \,\rm km \,s^{-1}}\right) \left (\frac{0.05\, \rm lt{\hbox{-}}yr}{r}\right) \left (\frac{r}{\Delta r}\right) M_{\odot }\, {\rm yr}^{-1}, \end{equation} \tag{ 4 }$$

for a progenitor wind speed v_w. The relatively narrow lines of SNe IIn originate in the slow pre-shocked CSM and therefore correspond to the progenitor wind speed. For these SNe, the narrow-line profiles have an FWHM intensity in the range ∼100–500 km s⁻¹ (Fox et al. 2011). Assuming a thin shell, Δr/r = 1/10, Table 6 lists the associated mass-loss rate for each SN.

Table 6. Mass-loss Rates

SN	$\dot{M}$ (IR)	$\dot{M}$ (opt/X)
	(M☉ yr−1)	(M☉ yr−1)
2005cp	3.4 × 10−2	⋅⋅⋅
2005ip	8.3 × 10−2	1.7 × 10−2
2006jd	1.7 × 10−1	2.2 × 10−3
2006qq	1.2 × 10−1	⋅⋅⋅
2007rt	1.0 × 10−1	⋅⋅⋅
2008cg	6.1 × 10−2	⋅⋅⋅
2008en	7.8 × 10−2	1.1 × 10−3
2008gm	2.8 × 10−3	⋅⋅⋅
2010jl	8.8 × 10−2	1.2 × 10−4

Download table as:  ASCIITypeset image

By contrast, the optical and/or X-ray observations trace the CSM characteristics at the inner radii. Assuming a steady mass loss, the rate can be written as a function of the optical/X-ray luminosity, progenitor wind speed, and shock velocity (e.g., Chugai & Danziger 1994; Smith et al. 2009b):

$$\begin{equation} \dot{M} = \frac{2 v_w}{\epsilon v_{s}^3}L_{\rm opt/X}, \end{equation} \tag{ 5 }$$

$$\begin{eqnarray} & = & 2.1 \times 10^{-4} \left (\frac{L_{\rm opt/X}}{3 \times 10^{41}\, {\rm erg s^{-1}}}\right) \left (\frac{\epsilon }{0.5}\right)^{-1} \nonumber \\ & & \times \left (\frac{v_w}{120\, {\rm km s^{-1}}\,}\right) \left (\frac{v_s}{10^4\, {\rm km s^{-1}}\,}\right)^{-3}\, M_\odot \, {\rm yr}^{-1}, \end{eqnarray} \tag{ 6 }$$

where < 1 is the efficiency of converting shock kinetic energy into visual light. While the conversion efficiency varies greatly depending on shock speed and wind density, we assume ≈ 0.5, acknowledging that this value may be high. Table 6 also lists the mass-loss rates for the SNe at the inner radii based on the optical/X-ray luminosities in Table 3.

For the SNe with both IR and optical observations, Table 6 highlights that the dust shells correspond to a period of higher mass loss and densities. By contrast, the CSM that generates the optical and/or X-ray emission corresponds to a period of lower mass-loss rates. The progenitor mass loss was variable, not continuous. The rates given in Table 6 (e.g., $\dot{M}\approx 10^{-1}\, {\rm to}\, 10^{-3}$ M_☉ yr⁻¹) are consistent with previous comparisons of SNe IIn to the episodic dense winds observed in some massive stars (e.g., Fransson et al. 2002; Smith et al. 2009a, 2011; Kiewe et al. 2012). In almost every case, the dust-shell radius and progenitor wind speeds suggest outbursts that occurred on the order of tens to hundreds of years prior to the SN explosion.

While stellar evolution models are still limited in their ability to achieve such substantial mass loss, the nature of the CSM wind can constrain the progenitor mass and explosion physics (e.g., Balberg & Loeb 2011). The observations in this paper present only a single snapshot of the CSM environment, but a number of theoretical models can generate the entire multi-wavelength evolution as a function of the progenitor mass-loss history. For example, the early (<100 days) optical and X-ray evolution can be described as a function of the immediate CSM density and explosion energy (e.g., Chevalier & Irwin 2011, 2012; Moriya et al. 2011; Moriya & Tominaga 2012). Furthermore, Ofek et al. (2012) and Svirski et al. (2012) show that for a sufficiently dense CSM, the X-rays will escape only months to years after the SN peaks in the visible. The escape of the X-rays (sometimes referred to as a circumstellar breakout) will differ observationally for varying progenitor mass-loss rates (Chevalier & Irwin 2011). To fully understand the mass-loss history of the progenitor therefore requires well-sampled multi-wavelength observations.

Figure 6 plots the multi-wavelength data available for individual SNe in our sample, including data from Figure 2 and pre-existing optical and X-ray photometry for SNe 2005ip, 2006jd, 2007rt, and 2010jl (Immler & Pooley 2007; Smith et al. 2009b; Fox et al. 2011; Chandra et al. 2012a, 2012b; Zhang et al. 2012; Stritzinger et al. 2012). While fitting the theoretical models (such as those by Moriya & Tominaga 2012 and Svirski et al. 2012) is beyond the scope of this paper, Figure 6 serves as a potentially useful template for theorists (e.g., Pan et al. 2013).

Zoom In Zoom Out Reset image size

Furthermore, the figure highlights the diversity among SNe IIn at various wavelengths. For example, the mid-IR luminosities are similar and serve as useful indicators for shock interaction and optical/X-ray emission. The optical and X-ray evolution differ significantly, suggestive of different progenitor mass-loss characteristics and, perhaps, different progenitors. In the case of SN 2010jl, the absorbed luminosity increases by a factor of two more than 1 yr after the visible maximum, revealing the first case of an X-ray breakout from a dense CSM (Chandra et al. 2012b). By contrast, SN 2006jd shows an X-ray plateau but no sign of a shock breakout, while SN 2005ip has insufficient X-ray observations. The optical light curve of SN 2010jl is quite luminous (>10⁹ L_☉), but fades throughout the extent of the observations, whereas SN 2006jd peaks at ∼500 days and SN 2005ip plateaus through more than 1000 days. In the case of SN 2006jd, a large X-ray luminosity implies a high density, but a small amount of photoabsorption in the spectrum implies a low density (Chandra et al. 2012a). These conflicting measurements suggest an asymmetric CSM. The slow evolution of the unabsorbed emission further suggests a progenitor wind-density profile that does not follow the standard r⁻² model. In the case of SN 2005ip, the late-time optical luminosity is over a magnitude larger than the X-ray luminosity. SN 2005ip does, however, exhibit very high-ionization coronal lines in the visible spectrum. Smith et al. (2009b) explain these characteristics with a lower density CSM interspersed with denser clumps. The shock interaction with the dense clumps generates X-rays, but with less intensity than in SN 2006jd. Since the clumps have a small filling factor, the X-rays can escape and photoionize the dense, pre-existing CSM at larger radii.