NEAR-INFRARED PHOTOMETRY OF THE TYPE IIn SN 2005ip: THE CASE FOR DUST CONDENSATION

SN 2005ip was discovered in NGC 2906 on UT 2005 November 5 (Boles et al. 2005). NGC 2906 lies at a distance of 30 Mpc, given a host galaxy recession velocity of 2140 km s⁻¹ (de Vaucouleurs et al. 1991) and H₀ = 72 km s⁻¹ Mpc⁻¹ (Freedman et al. 2001). Early optical spectra suggested the discovery occurred a few weeks following the explosion (Modjaz et al. 2005). The presence of H in the spectrum led to the classification of the supernova as a Type II event (Modjaz et al. 2005). Late spectra of SN 2005ip showed the development of a narrow Hα line that is characteristic of Type IIn supernovae (Smith et al. 2009). Various observers accumulated optical photometry of SN 2005ip over the first 100 days (Teamo 2007¹). The supernova was brightest, V ∼ 14.8, at the time of the first observation. The fact that all data were obtained post-peak does not allow for the explosion date to be constrained. An unfiltered R-band light curve beginning on day ∼ 10 post-discovery from Smith et al. 2009) shows a linear magnitude decline through day ∼ 150 post-discovery followed by luminosity plateau at R ∼ 17.8 that begins no later than day ∼ 300 and persists throughout the duration of our observations. Observations from Swift on day 461 reveal a magnitude V ∼ 18.5 and a 0.2–10 keV X-ray luminosity of 1.6 ± 0.3 × 10⁴⁰ erg s⁻¹ (Immler & Pooley 2007).

Our observations of SN 2005ip consist of J-, H-, and K_s-band images spanning from 2005 December through 2008 June, days 7-935 following discovery. All observations were made with FanCam, a 1024 × 1024 HAWAII-I HgCdTe imaging system on the University of Virginia's 31 inch telescope at Fan Mountain, just outside of Charlottesville, VA (S. Kanneganti et al. 2008, in preparation). Each epoch consists of 15 min of integration in JHK_s bands, which have detection limits at the 10σ level of 0.066, 0.098, and 0.156 mJy (or 18.5, 17.5, and 16.5 mag), respectively. Individual exposures are sky background limited and have an integration time of either 30 or 60 s. Flat-field frames are composed of dusk and dawn sky observations. We employed standard near-infrared data reduction techniques in IRAF. The relatively small galaxy size made it possible to fit it into a single array quadrant. Empty quadrants were efficiently utilized as sky exposures. Data were taken with the galaxy placed in each quadrant and each quadrant was reduced separately. Ultimately, all reduced quadrants were coadded. Figure 2(a) shows a typical reduced J-band image.

Zoom In Zoom Out Reset image size

Template subtraction minimizes photometric confusion from the underlying galaxy. While there are no existing FanCam observations of the galaxy prior to SN 2005ip, IRAF's DAOPHOT point-spread function (PSF) and SUBSTAR packages provide the means to create and subtract an average PSF at the supernova location for each epoch (Figure 2(b)). A coaddition of each epoch yields a master template that is deeper and smoother than any individual epoch. The subtraction of the template from each epoch leaves only the supernova (Figure 2(c)). To ensure a clean subtraction, individual and template PSFs are scaled to match in shape and intensity using IRAF's PSFMATCH package. We performed photometry with IRAF's APPHOT package, implementing an 8 pixel radius aperture and a 3 pixel wide annulus with an 8 pixel inner radius. For magnitude calibration, the supernova is compared to three Two Micron All Sky Survey (2MASS) reference stars located in the eastern part of Figure 2.

This template subtraction method yields results similar to, but less noisy than, other techniques, such as aperture photometry and PSF photometry without template subtraction. Table 1 lists the JHK_s photometry and Figure 3 plots the light curve. The K_s-band magnitude remains fairly constant at ⩽ 14.2 mag (a flux of ⩾ 1.3 mJy) for the duration of the observations, even as the J band declines. A notable "kink" occurs at day ∼ 200, although we do not speculate on the cause in this paper. Table 2 lists the near-infrared color evolution. Figure 4 shows the color to quickly redden between ∼ 50 and 200 days. This trend becomes more gradual and continues throughout the last observation epoch.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Figure 5 shows the near-infrared fluxes. Matching the H − K_s colors to the Planck function, presuming a dust emission/absorption efficiency, Q(λ), establishes a near-infrared dust temperature. The flux of a single dust particle of radius a at a temperature T_d and wavelength λ is given by

$$\begin{equation} F_{\lambda,T}=Q(\lambda)B_\lambda (T), \end{equation} \tag{ 1 }$$

where B_λ(T) is the Planck function. The emission/absorption efficiency is given as

$$\begin{equation} Q = \left(\frac{a}{\lambda }\right)^n,\quad\quad {\rm for}\;\quad a\ll \lambda \end{equation} \tag{ 2 }$$

$$\begin{equation} Q = 1,\qquad {\rm for}\quad a\gg \lambda \end{equation} \tag{ 3 }$$

where n is an integer. For large grain sizes (e.g., a ⩾ 1 μm), n = 0 because the dust acts as a blackbody. For smaller grains (i.e., a ≪ 1 μm), n > 0. We consider the cases n = 0, 1, and 2 to account for both large grains that are associated with condensing dust (Pozzo et al. 2004) and smaller grains that are associated with pre-existing optically thin dust, which is characterized by λ⁻¹ and λ⁻² emission/absorption efficiencies (Dwek et al. 1983).

Zoom In Zoom Out Reset image size

Figure 5 plots the corresponding Planck curves over the original fluxes. The fits exclude the J-band fluxes because of supernova photospheric flux and line emission contamination, especially at early times when the photospheric temperature is ∼ 10,000 K. At these early times, the H-band flux may also contain some contribution from the supernova photosphere. By later times, however, we presume the photosphere has faded sufficiently so that the warm dust dominates the flux. With only 2 data points for each epoch there is no clear "best" fit, so it is necessary to consider all cases.

Overall, the Planck peaks trend toward longer wavelengths/cooler temperatures. Table 3 lists the H − K_s color temperatures for each emission/absorption efficiency at each epoch. Figure 6(a) plots the color temperature evolution. In all cases, the fading supernova photosphere contributes to the dramatic temperature decline over the first ∼ 150 days. Although the exact point at which the dust temperature begins to dominate is unknown, we presume the photospheric component to decline sufficiently below the dust temperature after ∼ 150 days, at which point the temperature begins a slow, steady decline. While the chosen efficiency can alter the derived dust temperatures at early times by a few hundred Kelvin, these temperatures are less than or equal to the dust vaporization temperature (∼1400–1600 K). Over the next two years, the temperatures drop to ∼ 1000 K. Considering the observations are made at near-infrared wavelengths, they probe the warmest and most luminous grains. Cooler dust may also be present, but its emission peaks at wavelengths longer than those observed here.

Zoom In Zoom Out Reset image size

Table 3. Temperatures and Luminosities of SN 2005ip

Epoch (d)	Tbb (K)	$T_{\lambda ^{-1}}$ (K)	$T_{\lambda ^{-2}}$ (K)	Lbb (1041 erg s−1)	$L_{\lambda ^{-1}}$ (1041 erg s−1)	$L_{\lambda ^{-2}}$ (1041 erg s−1)	Lradioactive (1041 erg s−1)
7	...	...	...	...	...	...	42.09
8	...	...	...	...	...	...	42.09
63	1969(89)	1547(56)	1279(39)	41.43(0.03)	41.38(0.03)	41.34(0.03)	41.86
83	1809(47)	1449(31)	1213(22)	41.53(0.01)	41.48(0.02)	41.43(0.02)	41.80
93	1765(44)	1422(29)	1194(20)	41.57(0.01)	41.50(0.01)	41.46(0.01)	41.75
120	1667(38)	1359(26)	1149(18)	41.59(0.01)	41.53(0.01)	41.49(0.01)	41.64
151	1599(39)	1314(27)	1118(19)	41.57(0.01)	41.50(0.01)	41.46(0.01)	41.53
171	1582(45)	1303(31)	1110(22)	41.56(0.02)	41.50(0.02)	41.46(0.02)	41.45
186	1561(38)	1289(26)	1100(19)	41.60(0.01)	41.53(0.01)	41.49(0.01)	41.39
200	1538(38)	1273(26)	1088(19)	41.59(0.01)	41.53(0.01)	41.49(0.02)	41.34
351	...	...	...	...	...	...	40.75
432	1345(39)	1140(28)	990(21)	41.53(0.02)	41.48(0.02)	41.42(0.02)	40.43
480	1390(41)	1172(29)	1013(22)	41.53(0.02)	41.46(0.02)	41.41(0.02)	40.25
548	1386(48)	1169(34)	1012(26)	41.50(0.02)	41.43(0.02)	41.38(0.02)	39.98
688	1186(42)	1024(31)	902(24)	41.59(0.02)	41.50(0.02)	41.43(0.02)	39.43
722	1229(45)	1056(34)	926(26)	41.52(0.02)	41.43(0.02)	41.38(0.02)	39.31
767	1163(45)	1007(33)	888(26)	41.53(0.02)	41.45(0.02)	41.39(0.02)	39.14
831	1093(59)	954(45)	847(35)	41.55(0.03)	41.46(0.03)	41.39(0.04)	38.89
852	1108(48)	966(36)	856(28)	41.56(0.03)	41.46(0.03)	41.39(0.03)	38.82
895	1098(46)	958(35)	850(28)	41.56(0.02)	41.46(0.03)	41.39(0.03)	38.67
935	991(64)	876(50)	785(40)	41.63(0.04)	41.52(0.04)	41.43(0.04)	38.52

Download table as:  ASCIITypeset image

In contrast to the decreasing dust temperature, the dust luminosity plateaus early on and does not decline throughout the extent of the observations. Assuming a distance of 30 Mpc, integrating the Planck curves over all wavelengths yields a lower limit to the bolometric dust luminosity, shown in Figure 6(b) and listed in Table 3. Again, this calculation does not account for cooler dust that could contribute at longer wavelengths. Throughout the entire set of observations, the dust luminosity remains at ∼ 3 × 10⁴¹ erg s⁻¹.

Comparing SN 2005ip photometry to prior, well studied events can help expose the dust heating mechanism. We find three well sampled supernovae with similar bright near-infrared late time emission: SN 1998S (Fassia et al. 2000; Pozzo et al. 2004), SN 1995N (Gerardy et al. 2002), and SN 2006jc (Sakon et al. 2007; Smith et al. 2008; Mattila et al. 2008; Nozawa et al. 2008; Di Carlo et al. 2008). The associated spectra of these supernovae reveal further details about the surrounding environment and the nature of the infrared emission.

Type IIn SN 1998S is most comparable to SN 2005ip, with similar luminosities and temperatures, although a decline in the K-band luminosity is observed as early as day 333 (Fassia et al. 2000; Pooley et al. 2002; Pozzo et al. 2004). Pozzo et al. (2004) conclude dust condensation in the cool, dense shell behind the reverse shock is responsible for the late time luminosity. They do not consider the forward shock given the increased likelihood for dust condensation in the post-shocked ejecta. All observations are consistent with this scenario. The observed optical line profiles likely arise from the low-ionization gas in the cooling shock (Chevalier & Fransson 1994). The coincidence of the line velocities with the velocity of the dust shell (derived from the estimated shell radius from a blackbody approximation versus the time since the explosion) suggests a physical relationship between the two. The condensing dust also explains the fading central and redshifted Hα peaks with respect to the blueshifted peak observed by Gerardy et al. (2002). Fassia et al. (2001) observe CO emission to develop at the same time the absorption is observed. Finally, the pure blackbody fits of the dust (n = 0) are more suggestive of dust condensing in optically thick lumps than pre-existing, optically thin dust.

Type IIn SN 1995N also shows comparable temperatures and luminosities to SN 2005ip (Fox et al. 2000; Gerardy et al. 2002; Pastorello et al. 2005). Similar to SN 2005ip, the luminosity plateau extends all the way through day ∼1000, after which a decline was observed out to day 2413 (Gerardy et al. 2002). Gerardy et al. (2002) do not consider dust condensation, but instead suggest the dust is heated by either an IR echo or by the X-rays produced in direct circumstellar interaction heating of pre-existing dust. While the IR echo model adequately describes the photometric evolution, the implied dust cavity radius, created via heating and vaporization by the peak supernova luminosity, requires an unusually bright peak luminosity. Instead, the authors suggest an event prior to the core collapse could have cleared out the large dust cavity required in this instance. Alternatively, the authors' circumstellar medium interaction model does not require such a large dust cavity. The pre-existing dust is heated by X-ray and UV emissions produced by the much more slowly evolving shock interactions with a dense circumstellar medium.

Interestingly, Gerardy et al. (2002) rule out direct collisional heating of pre-existing dust by the hot, post-shocked electrons. The authors argue dust temperatures of 800–1000 K require prohibitively high gas densities (n_g ≳ 10²¹ cm⁻³ for T_e = 10⁷ K). However, we find these densities are not required in this scenario. For a = 0.1 μm silicate grains and a gas temperature T_e = 10⁷ K, Dwek et al. (2008) show in their Figure 1 that for dust temperatures of ∼ 800 K, gas densities are only n_g ∼ 10⁵ cm⁻³. These densities are not prohibitively high and are, in fact, close to what would be expected for typical progenitor mass loss rates (∼10⁻⁴–10⁻⁵ M_☉ yr⁻¹) and wind velocities (∼10 kms⁻¹) at radii of ∼10¹⁷ cm.

The authors also argue that collisional heating does not predict the observed drop in the infrared luminosity before its X-ray counterpart. We find, however, that the sputtering lifetime, τ_sput, of the dust grains is short enough that there should not necessarily be a correlation between the infrared and X-ray luminosities. The sputtering lifetime at gas temperatures above ∼ 10⁶ K given by (Dwek & Arendt 1992) is

$$\begin{equation} \tau _{\rm sput} ({\rm yr}) \approx 1\times 10^6 \frac{a({\mu}m)}{n_g({\rm cm^{-3}})}. \end{equation} \tag{ 4 }$$

For the grain size (a = 0.1 μm) and gas densities (n_g ∼ 10⁵ cm⁻³) above, τ_sput ≈ 1 year, which is shorter than the observed light plateau in SN 1995N. In such a short period of time, infrared emission from collisionally heated dust will decrease as the larger dust grains are destroyed. These effects could explain the lack of correlation between the infrared and X-ray luminosities. Nonetheless, collisional heating is still unlikely in this case. Draine (1981) shows that collisionally heated grains radiate only ∼12% of the shock power before the grains are sputtered, thereby limiting the contribution of infrared emission from collisionally heated dust.

SN 2006jc also shows an extended period of increased late time infrared emission. Unlike the previous two examples, SN 2006jc is a Type Ib event with a dense circumstellar medium that is likely associated with a LBV-like outburst originating from the progenitor at −730 days (Yamaoka et al. 2006; Pastorello et al. 2007). Between days 51 and 75, the supernova undergoes a brightening at near-infrared wavelengths (Arkharov et al. 2006; Di Carlo et al. 2008). On day 75, Smith et al. (2008) measured a luminosity L ∼ 2.5 × 10⁴¹ ergs⁻¹ and temperature T ∼ 1600 K. The temperature stayed constant for only ∼ 1 month and then cooled. Throughout this time period, X-rays indicate interaction of the supernova shock with the circumstellar shell created by the previous LBV outburst (Immler et al. 2008). The supernova continued to remain bright at infrared wavelengths through day 493, which was the last observation epoch (Mattila et al. 2008; Di Carlo et al. 2008).

While warm dust is the widely accepted source of the late time near-infrared emission, the origin of the dust is debated. Smith et al. (2008) argue the dust is newly formed based on (1) the development of a red/IR continuum consistent with the temperature of newly formed graphite grains or slightly hotter silicate grains, (2) the coincidental fading of the redshifted wings of the narrow He I emission lines, and (3) a simultaneous decline in the expected optical flux (Foley et al. 2007) and brightening in near-infrared. Sakon et al. (2007) and Nozawa et al. (2008) propose the dust condenses in the freely expanding ejecta. Both Smith et al. (2008) and Mattila et al. (2008) reject this scenario because it cannot explain the observed spectral line reddening. Instead, these authors favor dust condensation in the cool, dense shell created as the supernova shock interacts with the circumstellar medium. Smith et al. (2008) favor dust condensation in the reverse shock because of the observed carbon-rich features that are consistent with enriched ejecta encountering the reverse shock. In contrast, Mattila et al. (2008) conclude the reverse shock would not be radiative and, therefore, not conducive to dust formation. Instead, the authors argue for dust condensation in the cool, dense shell behind the forward shock.

In any case, the temperature remains constant for about one month at early times. Smith et al. (2008) argue that if there is a sufficient amount of circumstellar material, then grains will continue to form at the vaporization temperature. To produce the observed luminosity of L = (2–3) × 10⁴¹ erg s⁻¹ by cooling through ∼ 1600 K, the particle condensation rate is given as $\dot{N}=L/\epsilon$, where is the thermal energy per particle. For dust grains, the thermal energy per particle is given as = CΔT, where C is the dust heat capacity (Draine & Li 2001) and ΔT is 1600 if we assume the dust cools immediately and contributes only to the instantaneous luminosity. In this case, we calculate that the dust must condense on the order of ∼ 1000 M_☉ day⁻¹. This unrealistic mass condensation rate indicates that there must be an additional heating source to power the observed infrared luminosity. Given the optical light curve, which is powered by radioactivity, quickly declines (Foley et al. 2007) at about the same time as Arkharov et al. (2006) observe near-infrared brightening suggests that the newly formed dust is absorbing the optical emission from the central supernova. In this case, the likely dust heating source is the photospheric emission from radioactivity. Mattila et al. (2008) similarly argue that the cooling of the newly formed grains is insufficient to account for the observed near-infrared luminosity and conclude a similar heating source for their dust condensation scenario.

Warm dust surrounding SN 2005ip is the likely source of the near-infrared excess. As discussed in Section 1, there are several possible explanations for the large late time K_s-band luminosity: dust condensation in the ejecta, collisional heating in circumstellar medium shock interaction, an infrared echo, and dust condensation in the cool, dense shell behind either the forward or reverse shocks. This section briefly explores each possibility in order to distinguish between the different scenarios. SNe 1998S, 1995N, and 2006jc provide important points of reference.

3.3.1. Dust Condensation in the Unshocked Ejecta

3.3.2. Collisional Heating

As the forward shock expands into the circumstellar medium, the hot post-shocked electrons may collisionally heat pre-existing dust grains produced by the progenitor wind. To heat the dust to the observed temperatures (T ≈ 1400–1600 K), Figure 1 of Dwek et al. (2008) shows that the required gas densities are high, (n ≳ 10⁶ cm⁻³), assuming post-shock gas temperatures on order of 10⁷ K. The sputtering effects for such high gas densities would significantly limit the contribution of infrared emission from collisionally heated dust throughout the duration of the observations.

A stronger argument against this scenario is the inconsistency in the time lag between the supernova and the observed infrared emission. The supernova peak luminosity vaporizes all dust grains within a radius r_v that, for a dust emission/absorption efficiency Q_λ ∝ λ⁻¹, is proportional to the SN's peak luminosity L_peak (Dwek 1983),

$$\begin{equation} r_v\propto \left(\frac{L_{\rm peak}}{T_d^5}\right)^{0.5}. \end{equation} \tag{ 5 }$$

For a supernova with a peak UV-optical bolometric luminosity of 1 × 10¹⁰ L_☉ with an exponential decline rate time scale of 25 days (L_t = 1.0 × 10¹⁰e^−t/25d L_☉), Dwek (1983, 1985) shows r_v ≈ 6 × 10¹⁶ cm (0.06 light years) for carbon-rich grains (T_evap = 1900 K) and 3 × 10¹⁷ cm (0.31 light years) for oxygen-rich grains (T_evap = 1500 K). While the very early evolution of SN 2005ip is not well known, an early spectrum of SN 2005ip (M. Modjaz 2008, private communication) and R-band photometry (Smith et al. 2009) indicate an observed luminosity of ∼0.2 × 10¹⁰ L_☉ and an exponential time scale of ∼ 35 days. Considering the supernova was not detected until a few weeks past maximum, the actual peak is most likely greater by a factor of a few, which is comparable to the supernova peak luminosity used by Dwek (1983) as described above.

The size of this dust cavity implies a time lag between the initial explosion and the onset of infrared emission equal to the shock propagation time across the dust-free radius. Using the minimum of the Hα absorption line, Modjaz et al. (2005) measure the supernova expansion velocity to be 1.5 × 10⁹ cm s⁻¹ (0.05c) at an early age. This speed refers to the photosphere and there is presumably higher velocity gas, but there is also likely to be deceleration resulting from the interaction. At this speed the shock must travel for ∼ 450 days before encountering the dust, assuming the dust cavity radius for carbon-rich grains given above, and ∼ 2300 days for the oxygen-rich grains. Figure 4, however, shows a considerably shorter time delay before a detectable infrared excess.

3.3.3. IR Echo

3.3.4. Dust Condensation in Post-Shocked Gas