FROM SHOCK BREAKOUT TO PEAK AND BEYOND: EXTENSIVE PANCHROMATIC OBSERVATIONS OF THE TYPE Ib SUPERNOVA 2008D ASSOCIATED WITH SWIFT X-RAY TRANSIENT 080109

The X-ray transient (Berger & Soderberg 2008) rises in a time 50 ± 30 s, and has a duration T₉₀ = 470 ± 30 s or Δt_FWHM = 80 ± 30 s. Consistent with Soderberg et al. (2008b), we adopt t₀ = 2008-01-09 13:32:49 (± 5 s) as the time of onset of the XRT, i.e., the start of the Swift/XRT observations.

In Figure 1, we plot the X-ray light curve of XRT 080109/SN 2008D, after adequately subtracting the other three X-ray field sources as resolved by Chandra observations (Pooley & Soderberg 2008; see below). Prior to 1 ks but after the peak time, the X-ray luminosity (L_X) decay can be fit by a broken power law in time with late-time index α = −3.4 ± 0.6 (where L_X ∝ t^α), while data taken Δt≲ 300 s are best described by a decaying power law with α = −0.8 ± 0.2. For comparison, we plot in Figure 1 the X-ray light curve of XRF 060218/SN 2006aj, the most recent GRB/XRF-SN (Campana et al. 2006; Mirabal et al. 2006; Modjaz et al. 2006; Pian et al. 2006; Sollerman et al. 2006), which was discovered by Swift (Campana et al. 2006; Butler et al. 2006). While the expression "X-ray flash" (XRF, see Heise et al. 2001) is used in the literature when referring to the high-energy event 060218, the transient 060218 was qualitatively different from classical X-ray flashes, in that it likely exhibited a shock breakout component (Campana et al. 2006; Waxman et al. 2007). In Section 9, we discuss in detail the properties of the peculiar XRF 060218 and XRT 080109 compared to other classical XRFs.

Zoom In Zoom Out Reset image size

The peak X-ray luminosity of XRT 080109 is ∼700 times lower than that of XRF 060218 associated with SN 2006aj. The fluence of the X-ray transient is about 3.0 × 10⁻⁸${\rm erg\, cm^{\scriptscriptstyle -2}}$ in the 2–10 keV band (3.7 × 10⁻⁸ ${\rm erg\, cm^{\scriptscriptstyle -2}}$ in the 2–18 keV band), where we have adopted the count-rate conversion for the pure power-law spectral fit (see Section 2.2) of 4.0 × 10⁻¹¹ ${\rm erg \,cm^{\scriptscriptstyle -2}\, s^{\scriptscriptstyle -1}}$ per count s⁻¹ which corresponds to 4.3 × 10⁴² ${\rm erg\, s^{\scriptscriptstyle -1}}$ per count s⁻¹ for the adopted distance to XRT 080109 of D = 31 Mpc.

Chandra observations (Pooley & Soderberg 2008; Soderberg et al. 2008b) have shown that three bright field sources lie near the X-ray transient and likely contribute substantial counts at late times even for small Swift/XRT extraction regions. These sources (X1, X2, and X3; see Pooley & Soderberg 2008) are 38, 14, and 26 (9.0, 3.2, and 6.1 XRT pixels), respectively, from the X-ray transient. Using the Chandra spectral fitting results and the Swift/XRT point-spread function (PSF) from the calibration database, we determine a field source contamination count rate of 2 × 10⁻³ (6.9 × 10⁻⁴) counts s⁻¹ for the 16 (4.4) pixel extraction region.

Correcting to the true source flux level using the PSF model shows that the flux from XRT 080109 (Figure 1) is dominated by the field source flux for any observations that start 5 ksec after t₀ (i.e., the data point at 16 ksec) and is likely completely overwhelmed by the field source flux after 1 day. We therefore use the 16 pixel extraction region for the temporal/spectral analysis prior to 5 ksecand the 4.4 pixel extraction region to measure or limit the flux at later times.

We determine the maximum-likelihood X-ray centroid position and frame offset relative to Sloan Digital Sky Survey (SDSS; Δα = 08 and Δδ = 24, using five X-ray/optical source matches) as described by Butler (2007b). Our best-fit XRT position is α = 09^h09^m30.78^s, δ = +33°08'201 (±16, J2000, 90% conf.), which is more uncertain than the optical position of SN 2008D with position end figures 30.65^s and 203. These positions are consistent with those of Soderberg et al. (2008b). The X-ray light curve of XRT 080109 is given in Table 1.

Table 1. X-ray Light Curve of XRT 080109

Δt (s)	δt (s)	Luminosity Lx (1042 ${\rm erg\, s^{\scriptscriptstyle -1}}$)	$\sigma _{(L_{ {\rm X}})}$ (1042 ${\rm erg\, s^{\scriptscriptstyle -1}}$)
11.29	11.28	3.60	1.06
27.59	5.02	15.38	4.53
37.62	5.02	13.65	4.34
47.65	5.02	12.43	4.14
55.17	2.51	28.02	8.67
60.19	2.51	28.02	8.67
65.20	2.51	38.47	10.11
71.47	3.76	18.49	5.78
78.99	3.76	20.32	6.04
86.51	3.76	21.68	6.29
92.78	2.51	25.26	8.28
99.05	3.76	16.84	5.52
109.08	6.27	13.88	3.92
119.11	3.76	20.62	6.04
125.38	2.51	30.36	9.06
131.65	3.76	17.03	5.50
137.91	2.51	25.70	8.23
149.20	8.78	8.04	2.48
164.24	6.27	14.51	3.91
175.52	5.01	14.37	4.34
186.81	6.27	10.17	3.30
200.60	7.52	11.18	3.14
215.64	7.52	8.57	2.75
228.18	5.01	12.71	4.13
244.46	11.28	5.68	1.83
263.28	7.52	9.53	2.89
278.32	7.52	9.34	2.89
294.62	8.78	7.34	2.36
317.19	13.79	4.64	1.57
346.02	15.04	2.69	0.86
369.84	8.78	3.54	1.18
383.63	5.01	6.19	2.06
398.68	10.03	3.42	1.09
427.51	18.81	1.79	0.58
472.64	26.33	1.29	0.41
514.01	15.04	2.06	0.69
778.53	249.47	0.26	0.060
14944.11	9184.50	0.0040	0.0055
677205.64	376223.75	<0.0016	...
898641.0a	17900	0.00032	0.00017

Notes. Using a count-to-luminosity conversion of 4.3 × 10⁴² erg s^-1 per count s⁻¹ based on the pure PL spectral fit, for a distance of D = 31 ± 2 Mpc. Swift data unless noted otherwise. ^aChanda data.

Download table as:  ASCIITypeset image

We restrict our spectral fitting to the time period 0–520 s after the onset of XRT 080109, where the contamination from X-ray field sources is negligible (Pooley & Soderberg 2008).

Different groups, performing analysis of the same Swift/XRT data, come to different conclusions: Soderberg et al. (2008b) exclude an absorbed single blackbody (BB) fit on statistical grounds (with χ²/ν = 26.0/17) and thus favor a simple power-law (PL) fit (with χ²/ν = 7.5/17); Li (2008) argues that the data can be fit by an absorbed two-component BB fit (with χ²/ν = 10.4/16) as well as by a pure PL fit, but excludes the two-component BB fit based on physical grounds arguing that the inferred BB radius (5 × 10⁹ cm) is much smaller than the possible progenitor radius. Mazzali et al. (2008) claim that for any combined BB and PL fit, the BB contribution has to be small, namely at most 14% (whose fit gives χ²/ν = 20.6/21), and that the implied small BB radius (∼10¹⁰ cm) is evidence for a GRB jet (see also Section 9). However, we find that the low S/N data of XRT 080109 alone do not well constrain the contributions from a soft thermal excess and a harder PL continuum. In contrast to Mazzali et al. (2008), we find excellent fits (χ²/ν = 12.57/23, see Figure 2.) for a combined BB–PL fit assuming a factor unity ratio of PL to bolometric BB flux (i.e., 50% bolometric contribution from a BB, see below) without increasing N_H relative to that in the pure PL fit (χ²/ν = 12.73/24, N_H = 5.2^+2.1_−1.8 × 10²¹ ${\rm cm^{\scriptscriptstyle -2}}$). This H-equivalent column density N_H, which is greater than that expected from the Galaxy (1.7 × 10²⁰ ${\rm cm^{\scriptscriptstyle -2}}$; Dickey & Lockman 1990), is consistent with the large reddening value inferred from the optical-NIR data (Soderberg et al. 2008b, Section 6), given standard conversions between N_H and reddening (Predehl & Schmitt 1995).

Zoom In Zoom Out Reset image size

For the combined BB–PL fit, the time-integrated PL photon index is Γ = 2.1^+0.3_−0.4 (where N(E) ∝ E^−Γ), the BB temperature is kT = 0.10 ± 0.01 keV and R^X_BB = 10¹¹ cm, i.e., a factor of 10 larger than computed by Mazzali et al. (2008). For comparison, the pure BB fit gives χ²/ν = 28.56/25 with R^X_BB = (1.0 ± 0.2) × 10⁹ cm and kT = 0.75 ± 0.07 keV (Table 2). Hence, the apparent low R^X_BB from the pure BB fit alone may be an artifact of incorrectly fitting a low signal-to-noise ratio (S/N) composite BB plus PL spectrum with just a pure BB model. For completeness, we list in Table 2 our pure BB and PL fits, whose parameters are broadly consistent with the analysis performed by the groups above; specifically, our pure PL fit gives a time-integrated photon index of Γ = 2.1^+0.3_−0.4.

Table 2. Model Parameters for X-ray Spectral Fits of XRT 080109

Parameter	Power-law (PL) Fit	Blackbody (BB) Fit	Combined PL+BB Fita
Quality of Fit (χ2/dof)	12.73/24	28.56/25	12.57/23
Implied NH [1021 ${\rm cm^{\scriptscriptstyle -2}}$]	5.2+2.1−1.8	<1.5	5.2+2.1−1.8
Time-integrated Γ, kT, and [Γ;kT] respectively	2.1+0.3−0.4	0.75 ± 0.07	2.1+0.3−0.4; 0.10 ± 0.01
LX,isob [ 1042 erg s−1]	12+5−2	4.6 ± 0.6	13 ± 2c
EX,isob [1045 erg]	5.8+2−1	2.4± 3	6.7 ± 4c

Notes. ^aAssuming the same (factor unity) ratio of PL to BB flux as in XRF 060218 (Butler et al. 2006). ^bUnabsorbed and time-averaged value over the range 0.3–10.0 keV (for PL fit) and bolometric (for BB fit), respectively, assuming D = 31 ± 2 Mpc. ^cListed only for the BB component. L_X,iso and E_X,iso are computed assuming R^X_BB = 10¹¹ cm as required from the combined PL and BB fits.

Download table as:  ASCIITypeset image

Although the exact fraction of soft BB emission in XRT 080109 is poorly constrained by the data, a case can be made for the possible presence of BB emission through comparison to XRF 060218. The high-S/N X-ray spectrum of XRF 060218 during the proposed shock breakout phase at t < 6 ks was predominantly nonthermal but also required at high confidence a soft (kT = 0.1–0.2 keV) BB (Campana et al. 2006; Butler et al. 2006) with a ratio of about unity between PL and BB flux (Butler et al. 2006). If we downsample the spectrum of XRF 060218 for t < 6 ks to contain 380 counts as in XRT 080109, we find that the XRF 060218 spectrum is fit in a strikingly similar fashion to the XRT 080109 spectrum: a BB fit has kT = 0.72^+0.9_−0.5 keV and local N_H < 3.6 × 10²¹ cm⁻² (χ²/ν = 35.22/23), while a PL fit has Γ = 1.7 ± 0.3 and local N_H = 2.2^+1.8_−1.5 × 10²¹ cm⁻² (χ²/ν = 9.25/23). From the low χ²/ν, we see that the PL model with N_H in excess of Galactic also effectively overfits the soft spectral excess in XRF 060218, which requires a PL plus BB given all the counts. We also note that the light curve of XRT 080109 is quite similar to that observed for XRF 060218/SN 2006aj, if we allow a time stretch by a factor of ∼0.1 (Figure 1). We discuss the implications and interpretation of the X-ray data in Section 9.

For photometric calibrations, the field of SN 2008D was observed in UBVRI during photometric nights on 2008 January 18, 19, and 20 with KAIT, in BVRI on 2008 January 12 with the Nickel 1 m telescope, and in BVr'i' on 2008 January 18 with the FLWO 1.2 m telescope. About a dozen Landolt (1992) standard star fields were observed at different airmasses throughout each photometric night. Photometric solutions to the Landolt standard stars yield a scatter of ∼0.02 mag for all the filters for the Nickel telescope and about 0.03 mag for KAIT. The SN 2008D field was also observed for several sets of UBVRI images with different depth in the photometric nights. The photometric solutions are used to calibrate a set of local standard stars in the SN 2008D field as listed in Table 3, and a finder chart is shown in Figure 3. The calibration based on the FLWO 1.2 m telescope observations ("FLWO calibration") is consistent within the uncertainties with the Lick calibration (from the KAIT and Nickel telescopes) in the filters that they have in common (BV).

Zoom In Zoom Out Reset image size

Table 3. Photometry of Comparison Stars in the Field of SN 2008D by KAIT and FLWO 1.2 m

ID	U (mag)	NU	B (mag)	NB	V (mag)	NV	R/r' (mag)	$N_{R/r^{\prime }}$	I/i' (mag)	$N_{I/i^{\prime }}$
1	16.771(029)	2	16.438(002)	3	15.563(006)	2	15.117(013)	3	14.640(016)	3
2	17.703(037)	2	17.732(003)	2	16.997(002)	4	16.590(005)	2	16.274(007)	6
3	13.611(005)	3	13.745(008)	3	13.234(010)	4	12.933(010)	4	12.639(015)	3
4	17.822(069)	2	17.865(009)	3	17.409(012)	4	17.061(010)	5	16.781(002)	6
5	⋅⋅⋅	⋅⋅⋅	17.994(011)	2	17.316(010)	4	16.841(013)	4	16.498(009)	5
6	⋅⋅⋅	⋅⋅⋅	18.132(014)	4	17.550(016)	3	17.115(005)	4	16.856(011)	6
7	12.466(005)	3	12.530(002)	3	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅	⋅⋅⋅
8	⋅⋅⋅	⋅⋅⋅	18.977(012)	4	17.924(011)	4	17.227(006)	5	16.789(003)	3
9	15.003(007)	3	14.969(003)	3	14.394(016)	3	14.041(010)	3	13.734(013)	3
1	⋅⋅⋅	⋅⋅⋅	16.415(049)	1	15.586(033)	1	15.326(045)	1	15.088(046)	1
3	⋅⋅⋅	⋅⋅⋅	13.657(031)	1	13.191(022)	1	13.089(031)	1	12.999(031)	1
10	⋅⋅⋅	⋅⋅⋅	16.498(049)	1	15.594(033)	1	15.298(045)	1	15.047(045)	1
11	⋅⋅⋅	⋅⋅⋅	16.908(051)	1	15.764(035)	1	15.344(047)	1	15.004(048)	1
12	⋅⋅⋅	⋅⋅⋅	18.068(089)	1	17.237(060)	1	17.038(079)	1	16.805(081)	1
13	⋅⋅⋅	⋅⋅⋅	15.659(040)	1	14.827(027)	1	14.585(037)	1	14.391(037)	1
14	⋅⋅⋅	⋅⋅⋅	14.364(033)	1	13.631(023)	1	13.446(031)	1	13.311(031)	1
15	⋅⋅⋅	⋅⋅⋅	17.584(086)	1	17.164(058)	1	17.087(076)	1	17.040(078)	1
16	⋅⋅⋅	⋅⋅⋅	16.054(045)	1	15.320(031)	1	15.105(041)	1	14.946(042)	1
17	⋅⋅⋅	⋅⋅⋅	17.309(055)	1	15.966(037)	1	15.486(049)	1	15.059(051)	1

Notes. Uncertainties (standard deviation of the mean) are indicated in parentheses. Stars (see Figure 3) in the top half of the table were observed with KAIT in UBVRI and the stars in the bottom half were observed with the FLWO 1.2 m telescope in BVr'i'.

Download table as:  ASCIITypeset image

KAIT followed SN 2008D in BVRI and unfiltered mode nightly (weather permitting) after its discovery. As SN 2008D occurred on a spiral arm of NGC 2770, we use image subtraction to remove the contamination of the host-galaxy emission. For template subtraction, we used pre-explosion images obtained with the FLWO 1.2 m telescope during the course of follow-up photometry for SN 2007uy, which also occurred in NGC 2770 (as shown in Figure 3).

For KAIT, the image subtraction and subsequent photometry reduction are performed with the KAIT photometry pipeline (M. Ganeshalingam et al. 2009, in preparation). Two independent image-subtraction routines were employed in the pipeline. One routine is based on the ISIS package (Alard & Lupton 1998) as modified by Brian Schmidt for the High-z Supernova Search Team (Riess et al. 1998), and the other is based on the IRAF²⁶ task PSFMATCH (Phillips & Davis 1995). PSF fitting photometry is performed on the subtracted images, and the results from the two routines are averaged whenever appropriate. Artificial stars are injected into the original KAIT images and extracted from the subtracted images to estimate the scatter of the measurements. To convert the KAIT instrumental magnitudes into the standard Johnson BV and Cousins RI system, the color terms for the KAIT filters as determined from many photometric calibrations are used in the conversion (as detailed by Modjaz et al. 2001 and Li et al. 2001). The final error bars for the magnitudes are the scatter in the artificial-star simulation and the calibration error added in quadrature.

We note that the FLWO 1.2 m telescope template images were taken with the Johnson BV and Sloan r'i' filters, while the KAIT observations of SN 2008D were taken with the Johnson BV and Cousins RI filters. The difference between the r'i' and RI filter transmission curves raises the possibility of a systematic uncertainty during the image-subtraction process. We investigated this issue with the artificial-star simulations and concluded that the systematic uncertainty is smaller than ∼0.03 mag (which is not reported in the final uncertainties).

SN 2008D was remotely observed with the Lick Observatory 1 m Nickel telescope in BVRI on 2008 January 11 and 12. These data were processed with proper bias and flat-field images, and the image subtraction and photometry were performed with the KAIT photometry pipeline (but modified to deal with the Nickel telescope images). The FLWO 1.2 m telescope template images were used to generate the subtracted images, and the Lick calibration and the proper color terms for the Nickel telescope filters are used to convert the instrumental magnitudes into the standard system. The final photometry of SN 2008D from the KAIT and Nickel observations is listed in Table 4.

The field of SN 2007uy in NGC 2770 was monitored by the FLWO 1.2 m telescope in the context of the CfA SN monitoring²⁷ efforts when SN 2008D was discovered, providing us with valuable pre-explosion images. We followed SN 2008D on a nightly basis (weather permitting) in BVr'i'. For performing photometry of FLWO 1.2 m telescope images, we adopted the image-analysis pipeline of the SuperMACHO and ESSENCE collaborations (see Rest et al. 2005, Garg et al. 2007, and Miknaitis et al. 2007 for details). The operations of this optical photometry pipeline are discussed in depth by Hicken et al. (2009). In brief, we employed differential photometry by measuring the brightness of the SN with respect to a set of stars (see Figure 3) in the SN field and employed the DoPHOT photometry package (Schechter et al. 1993) to measure the flux of the SN and its comparison stars. We performed image subtraction with the robust algorithm of Alard & Lupton (1998; see also Alard 2000), using as templates the 1.2 m images taken on 2008 January 8 of the field. The "FLWO calibration" and the color terms for Keplercam (Modjaz 2007) were used to convert the instrumental magnitudes into the standard system.

We present the 1.2 m FLWO BVr'i' photometry in Table 4. For multiple observations on the same night we report the weighted mean of those observations. Intranight observations did not reveal statistically significant variations for SN 2008D. We furthermore derive upper limits for SN 2008D based on 2008 January 9.44 images of the field of SN 2007uy, three hours before the onset of XRT 080109 (Section 2), and present them in Table 4. The upper limits were estimated by placing fake stars with a range of magnitudes at the position of SN 2008D on the template images and then determining when they are detected at varying thresholds above background.

The Swift/UVOT has followed SN 2008D intensively (Kong et al. 2008b; Immler et al. 2008a; Li et al. 2008; Immler et al. 2008b; Soderberg et al. 2008b). We retrieved the level-2 UVOT (Roming et al. 2005) data for SN 2008D from the Swift data archive in all available filters (i.e., in the UVW2, UVM2, UVW1, U, B, and V filters).

We used the UVOT observations of SN 2007uy on 2008 January 6 as the template images to reduce the data of SN 2008D (a total exposure time of 808 s in each of the filters). For the U, B, and V filters, we employed the photometric reduction and calibration as described by Li et al. (2006) to perform photometry of SN 2008D on the subtracted images, while for the Far-UV filters (UVW2, UVM2, UVW1), we used a 3'' aperture and employed the calibration of Poole et al. (2008). Many tests have shown that the two calibrations by Li et al. (2006) and Poole et al. (2008) are consistent with each other.

The UVOT photometry of SN 2008D is reported in Table 5, and is presented in Figure 4, along with the overall optical photometry, with respect to the start of XRT 080109 (Section 2). The UVOT BV values reported here are fully consistent with those in Soderberg et al. (2008b), while our U photometry is systematically fainter by 0.2 mag (consistent within systematic errors, see below), and our UVW1, UVW2, and UVM2 3σ upper limits are systematically fainter by 0.2–1 mag. Our analysis used larger binning periods than that of Soderberg et al., producing deeper images and enabling additional detections in UVW1. The S/N of the UV detections is poor, and thus precision photometry is difficult. Specifically, the choice of aperture size affects the final SN magnitude; for example, using a 2'' aperture for the UVW1 band gives magnitudes that are ∼0.23 mag brighter than for the 3'' aperture that we adopt (even after appropriate aperture corrections). Thus, we estimate the systematic uncertainty of the final UV photometry to be on the order of ∼0.25 mag.

Zoom In Zoom Out Reset image size

We obtained NIR (JHK_s) photometric measurements with the refurbished and fully automated 1.3 m Peters Automated Infrared Telescope (PAIRITEL)²⁸ located at FLWO. PAIRITEL is the world's largest NIR imaging telescope dedicated to time-domain astronomy (Bloom et al. 2006). The telescope and camera were part of the Two Micron All Sky Survey (2MASS; Skrutskie et al. 2006) project. The J-, H-, and K_s-band (1.2, 1.6, and 2.2 μm; Cohen et al. 2003) images were acquired simultaneously with the three NICMOS3 arrays with individual exposure times of 7.8 s.

The PAIRITEL reduction pipeline software (Bloom et al. 2006) is used to estimate the sky background from a star-masked median stack of the SN raw images. After sky subtraction, the pipeline is used to cross-correlate, stack, and subsample the processed images in order to produce the final image with an effective scale of 1''pixel⁻¹ and an effective field of view (FOV) of 10'×10'. The effective exposure times for the final "mosaic" images ranged from 15 to 20 minutes. Multi-epoch observations over the course of the night (totaling between 40 minutes and 4 hr) were obtained to search for intranight variations, but none were found at a statistically significant level.

For the NIR photometry, we used the image-analysis pipeline of the ESSENCE and SuperMACHO projects (Rest et al. 2005; Garg et al. 2007; Miknaitis et al. 2007). We performed difference photometry (following Wood-Vasey et al. 2008) using the image-subtraction algorithm of Alard & Lupton (1998), similar to our optical photometry. As a template image for the field of SN 2008D, we used our PAIRITEL images of SN 2007uy taken on 2008 January 9. Thus, we also derive NIR upper limits for SN 2008D based on images of the field of SN 2007uy, taken on 2008 January 9 08:26:19, 4.5 hr before t₀, the onset of XRT 080109 (see Section 2), and present them in Table 6. The upper limits were estimated in the same fashion as for the optical data. We performed the final calibration onto the 2MASS system by using as reference stars the field stars from the 2MASS catalog, of which there were 20–30 in the FOV. No color-term corrections were required since our natural system is already in the 2MASS system. We present our PAIRITEL photometry in Table 6.

We obtained one high-resolution spectrum with Keck I on 2008 January 12.5 (i.e., at Δt = 2.95 days), over the observed wavelength range 3440–6300 Å. We detect underlying H ii region emission lines (Balmer lines and [O ii] λ3727) as well as Na i and Ca ii H&K absorption lines in the host galaxy (see also Soderberg et al. 2008b; Malesani et al. 2009). The Na i and Ca ii H&K absorption lines have identical redshifts as the centroid of the [O ii] and Balmer emission lines, z = 0.00700 ± 0.00005. Since both emission and absorption lines show the same velocity, we consider the Na i and Ca ii H&K absorption lines to have interstellar origin (as opposed to circumstellar) and attribute the offset of ∼160 km s⁻¹ from the measured redshift of the nucleus of the host galaxy NGC 2770 (z_nuc = 0.006451 ± 0.000087; Falco et al. 1999) to the galaxy's rotation curve. Given that NGC 2770 is an inclined, edge-on, typical spiral (Sb/Sc) galaxy (Soderberg et al. 2008b; Thöne et al. 2009), rotation on the order of 160 km s⁻¹ at the distance of the SN from the nucleus (∼9 kpc) is plausible; the full rotation curve of NGC 2770 extends to ∼400 km s⁻¹ (Haynes et al. 1997).

Since the absorption lines are saturated, we can only place lower limits to the column densities of Na i and Ca ii H&K: $N_{\rm {Na\,{\scriptscriptstyle I}}} > 1.0 \times 10^{13}$ ${\rm cm^{\scriptscriptstyle -2}}$ and $N_{ {\rm Ca\,{\scriptscriptstyle II}}} > 1.4 \times 10^{13}$ ${\rm cm^{\scriptscriptstyle -2}}$. We measure an equivalent width (EW) for Na i D₂ of 0.67 ± 0.04 Å (at rest) and for Ca ii K of 0.55 ± 0.04 Å (at rest).

In Figure 5, we present the sequence of early-time optical spectra after subtracting the SN recession velocity, dereddened by E(B − V)_Host = 0.6 ± 0.1 mag (Section 6). Furthermore, the spectra are labeled according to age (Δt) with respect to date of shock breakout (t₀= 2008-01-09 13:32:49) and epoch (t_Vmax), defined as days relative to V-band maximum (2008 January 27.9 = JD 2454493.4, see Section 4). Knowing both reference dates is crucial for interpreting the spectra and their temporal evolution.

Zoom In Zoom Out Reset image size

These are amongst the earliest spectra presented of a normal SN Ib (starting 1.70 day after shock breakout, which means 17.0 days before V-band maximum) and provide insights into the different layers of the SN ejecta. As the photosphere recedes with time, it reveals lower-velocity material since the SN ejecta are in homologous expansion (where v ∝ R). Moreover, we obtained late-time spectra at t_Vmax = +91 and +132 days that show the advent of the nebular epoch, in which the spectrum is marked by strong forbidden emission lines of intermediate-mass elements [O i] and [Ca ii] (not shown in Figure 5). We discuss the late-time spectra in Section 8.

At early times, our densely time-sampled series shows the dereddened spectra to consist of a blue, nearly featureless continuum (except around 4000 Å), with superimposed characteristic broad SN lines not attributed to hydrogen (Malesani et al. 2008; Blondin et al. 2008a; Valenti et al. 2008b; Soderberg et al. 2008b). In Section 5.3, we discuss the fleeting, double-absorption feature around 4000 Å, which is only seen in our spectrum at Δt= 1.84 days.

Figure 5 shows that the optical helium lines (blueshifted He i λλ4471, 5876, 6678, 7061) become apparent starting at Δt≈ 10–12 days (8–6 days before V-band maximum) and grow stronger over time, allowing this SN to be classified as an SN Ib (Modjaz et al. 2008a; Valenti et al. 2008a; see also Soderberg et al. 2008b; Mazzali et al. 2008; Malesani et al. 2009). This is in stark contrast to all SNe associated with GRBs that have been exclusively broad-lined SN Ic (e.g., Galama et al. 1998; Stanek et al. 2003; Hjorth et al. 2003; Malesani et al. 2004; Modjaz et al. 2006; Mirabal et al. 2006; Campana et al. 2006; Pian et al. 2006).

Perhaps the most intriguing aspect of the very early time spectra of SN 2008D is the transient, prominent, double-absorption feature around 4000 Å (Figure 6). It is only present in our second spectrum, the MMT+Blue Channel spectrum taken on January 11.41 (Δt = 1.84 day), and was first noted by Blondin et al. (2008a).

Zoom In Zoom Out Reset image size

Furthermore, Malesani et al. (2009) obtained spectra of SN 2008D at nearly the same epoch, January 11.32 (i.e., within two hours of our Gemini and MMT spectra, see Table 7) with wavelength range extending down to 3700 Å, and note that their spectra show the same double-absorption feature. The double-absorption feature is apparently absent from our very first spectrum, the Gemini+GMOS spectrum taken on January 11.26, but the spectrum is extremely noisy below ∼4500 Å due to the particular setup used. The spectrum of Soderberg et al. (2008b) on the same and subsequent nights apparently does not extend sufficiently far to the blue to include this feature at ∼4000 Å (their Figure 3). The double-absorption line is no longer visible in spectra taken ≳ 1 day later (i.e., Δt ≳ 2.99 days) and the two components appear to blend into a single, broad (FWHM ≈23,000 km s⁻¹) absorption feature (maximum absorption at rest frame ∼4100 Å), as clearly seen in the January 13.30 MMT+Blue Channel spectrum (Δt = 3.74 days) in Figure 6. Also, spectra taken by Malesani et al. (2009) do not show the feature in spectra taken after January 13.07, i.e., Δt ≳ 3.51 days.

The double-absorption feature appears to consist of two overlapping P-Cygni profiles, with rest-frame full width at half-maximum (FWHM) of ∼11,500 km s⁻¹ and ∼11,000 km s⁻¹ for the bluer (maximum absorption at rest-frame wavelength ∼3980 Å) and redder (maximum absorption at rest-frame wavelength ∼4240 Å) absorptions, respectively. Both absorption components are thus significantly narrower than other absorption features in the same spectrum, whose FWHM ranges between ∼15,000 km s⁻¹ and ∼30,000 km s⁻¹.

In the following section we attempt to address the line identification for this feature. Amongst the published supernova spectra, the closest match to the "W" feature that we can find is in the earliest spectrum of SN 2005ap (Quimby et al. 2007) from 2005 March 7 (∼1 week before maximum light), which shows a similar feature at slightly redder wavelengths (by ∼200 Å). Quimby et al. used the parameterized supernova spectrum synthesis code SYNOW (Fisher et al. 1999; Branch et al. 2002) to fit that spectrum with four ions. They ascribed the "W" feature to a combination of C iii, N iii, and O iii with a photospheric velocity of 21,000 km s⁻¹. We downloaded the spectrum of SN 2005ap from the SUSPECT database,³⁰ generated SYNOW models using only those three ions, and were able to satisfactorily reproduce the "W" feature in SN 2005ap using similar fitting parameters to those of Quimby et al. We then increased what SYNOW calls the "photospheric velocity" (v_phot) of the fit to match the observed wavelengths of the "W" feature in SN 2008D and found that v_phot≈ 30,000 km s⁻¹ produced a good match to the "W" as well as to some other weaker features, as shown in Figure 7 (using a continuum temperature of ∼13,000 K, consistent with the BB temperature as derived from our photometry data, Table 8). SYNOW did not produce any other strong features that are associated with these ions, matching the full extent of the observed MMT spectrum.

Zoom In Zoom Out Reset image size

Table 8. Blackbody Fits and Bolometric Light Curves of SN 2008D from UVW1 → K_s Data

Δt (days)	TBB (K)	RBB (1014 cm)	log(LBB) (${\rm erg\, s^{\scriptscriptstyle -1}}$)	log($L_{{\it UVW1}-K_s}$) (${\rm erg\, s^{\scriptscriptstyle -1}}$)
0.80	12000+3000−2000	2.75+0.36−0.37	42.04+0.27−0.21	41.92+0.18−0.17
1.80	10000+1600−1200	3.78+0.30−0.30	42.05+0.19−0.17	42.00+0.17−0.16
2.80	8700+1300−1000	4.64+0.47−0.46	41.93+0.17−0.15	41.89+0.15−0.14
3.80	7900+1000−830	5.19+0.47−0.47	41.86+0.15−0.13	41.83+0.14−0.13
4.80	7600+970−780	5.62+0.52−0.50	41.87+0.14−0.13	41.85+0.14−0.12
5.80	7400+910−740	6.12+0.56−0.54	41.90+0.14−0.12	41.89+0.13−0.12
6.80	7200+850−690	6.92+0.61−0.60	41.95+0.13−0.12	41.93+0.13−0.12
7.80	7000+800−660	7.64+0.65−0.64	42.00+0.13−0.11	41.98+0.13−0.12
8.80	6900+770−630	8.30+0.69−0.69	42.04+0.13−0.11	42.02+0.12−0.11
9.80	6800+740−610	8.89+0.74−0.73	42.08+0.12−0.11	42.05+0.12−0.11
10.80	6700+720−590	9.45+0.78−0.76	42.10+0.12−0.11	42.08+0.12−0.11
11.80	6600+700−580	9.95+0.79−0.81	42.13+0.12−0.11	42.11+0.12−0.11
12.80	6600+680−570	10.40+0.84−0.83	42.15+0.12−0.10	42.13+0.12−0.11
13.80	6500+660−550	10.80+0.92−0.84	42.17+0.11−0.10	42.15+0.12−0.11
14.80	6400+650−540	11.20+0.94−0.86	42.18+0.11−0.10	42.17+0.12−0.11
15.80	6400+630−530	11.60+0.90−0.93	42.19+0.11−0.10	42.18+0.12−0.11
16.80	6300+620−520	11.90+0.97−0.89	42.20+0.11−0.10	42.19+0.11−0.11
17.80	6200+600−510	12.30+0.98−0.96	42.21+0.11−0.10	42.20+0.11−0.10
18.80	6200+590−500	12.60+1.00−0.97	42.21+0.11−0.10	42.20+0.11−0.10
19.80	6100+570−480	12.90+1.00−0.98	42.21+0.11−0.10	42.20+0.11−0.10
20.80	6000+560−470	13.20+1.00−0.99	42.21+0.10−0.10	42.20+0.11−0.10
21.80	6000+540−460	13.40+1.10−1.00	42.20+0.10−0.09	42.20+0.11−0.10
22.80	5900+530−450	13.70+1.00−1.00	42.19+0.10−0.09	42.19+0.11−0.10
23.80	5800+520−440	13.90+1.10−1.00	42.18+0.10−0.09	42.18+0.10−0.10
24.80	5700+500−430	14.20+1.10−1.10	42.17+0.10−0.09	42.16+0.10−0.10
25.80	5600+480−410	14.50+1.10−1.10	42.15+0.09−0.09	42.15+0.10−0.09
26.80	5400+460−400	14.80+1.20−1.10	42.13+0.09−0.08	42.13+0.10−0.09
27.80	5300+440−380	15.20+1.20−1.20	42.11+0.09−0.08	42.11+0.10−0.09
28.80	5200+420−360	15.50+1.20−1.20	42.09+0.09−0.08	42.09+0.10−0.09
29.80	5100+400−350	15.70+1.20−1.20	42.08+0.09−0.08	42.08+0.10−0.09
30.80	5100+400−350	15.80+1.20−1.20	42.06+0.08−0.07	42.07+0.09−0.09

Notes. Adopted values: E(B − V)_Host = 0.6 ± 0.1 mag, D = 31 ± 2 Mpc, and t₀ = 2008-01-09 13:32:49 UT. The uncertainties include the uncertainty in reddening, but not the systematic uncertainty in distance.

Download table as:  ASCIITypeset image

While our SYNOW fit produces a good match to the data, we note that SYNOW fits do not produce unique line identifications and the suggested identification here needs to be confirmed by detailed spectral modeling (Mazzali & Lucy 1998; Baron et al. 1999; Dessart & Hillier 2008). Furthermore, while the "W" feature is no longer visible in spectra of SN 2008D taken ∼1 day later (see also Malesani et al. 2009), it seemed to linger for ∼1 week in the spectrum of SN 2005ap (at slightly lower velocities).

At face value, an identification of the "W" feature with C iii, N iii, and O iii at high velocities seems plausible. We expect the early supernova spectrum to have high velocities based on an extrapolation of the observed velocities at late times back to the earliest observations. In addition, what SYNOW designates as the "excitation temperature" is high (T_ex≈ 17,000 K) at these early times³¹ and many of the elements that will contribute to the later spectra are more highly ionized. The transient nature of the feature might then be explainable because the elements involved will recombine to lower ionization stages as the ejecta cool (at ∼15,000 K where most of doubly ionized elements recombine to singly ionized species).

Alternatively, the reason for the short-lived nature of this "W" feature could be asphericity in the SN ejecta. We have evidence from late-time spectra for asphericity in at least the inner SN core (see Section 8), which may be connected to some inhomogeneity in the SN ejecta density or chemical abundance profile at the outermost radii that the early-time spectra are probing. If indeed asphericity is involved, then existing spectral synthesis codes (such as SYNOW which assumes spherical symmetry) are challenged to accommodate that aspect.³²

The broadband data set (spanning 2200–24,100 Å) presented here lends itself uniquely to computing the bolometric output of SN 2008D. We determine it in two different ways: (a) by fitting a BB to the broadband photometry and deriving the equivalent BB luminosity (L_BB) by analytically computing the corresponding Planck function, and (b) by performing a direct integration of the UVW1BVRr'Ii'JHK_S broadband magnitudes ($L_{{\it UVW1} \rightarrow K_s}$), including the far-UV filters UVW2 and UVM2 in case of detections. In both cases, the Swift UV and our NIR data are crucial, as they provide a factor of at least two in wavelength coverage and leverage for the SEDs. While theoretical considerations might justify using a simple BB approximation for the SN emission shortly after shock breakout (Waxman et al. 2007; Chevalier & Fransson 2008), we do note that without detailed models of SN Ib/c atmospheres, it proves difficult to say how well BB fits approximate the true SN luminosity over the full light curve of SN 2008D. We address this question more in the remainder of the section, when comparing the bolometric luminosities computed via BB fits and those via direct integration.

We start by interpolating the light curves by Swift/UVOT (in UVW1, UBV), FLW0 1.2 m (in r'i'), and PAIRITEL (in JHK_s) onto a grid spaced regularly in time, as the observing times were different for different telescopes. We then sampled those light curves at intervals closest to the actual observing time of Swift/UVOT and the FLWO telescopes. After removing extinction (E(B − V)_Host = 0.6 ± 0.1 mag and assuming a Milky Way extinction law parameterization with R_V = 3.1; Fitzpatrick 1999), we converted our photometry into monochromatic fluxes at the effective wavelengths of the broadband filters, using zero points from Fukugita et al. (1995; for UBVr'i', since the UVOT UBV photometric system is close to the Johnson UBV with Vega defining the zero point; Poole et al. 2008) and Cohen et al. (2003) (for JHK_s). We included 0.05 mag of systematic error in the UVOT filters (Li et al. 2006) and 4% conversion error when using Fukugita et al. (1995).

For our first method, we perform a least-squares fit of a BB spectrum to the SED of each epoch and constrain the derived BB temperature (T_BB), BB radius (R_BB), and BB luminosity (L_BB) as a function of age (see Table 8 and Figure 8). The error bars include the uncertainties from the formal fits as well as uncertainties introduced by the uncertain reddening (±0.1 mag), which affect the derived BB fit parameters differently, depending on the underlying SED.

Zoom In Zoom Out Reset image size

We do not include the systematic uncertainty in distance (of order ∼20%). At later times (Δt≳ 8–10 days) the assumption of a BB continuum is expected to break down, based on the optical spectra where distinct absorption lines develop (see Figure 5). While the χ² values of the fits are strictly not acceptable except at early times (1 day <Δt< 5 days), the overall fits are reasonable even beyond this time with no obvious residual trends, and we consider the BB measured temperatures, radii, and luminosities still to be generally quite reliable. Thus, the UVOPTIR data are well fit by a BB, especially at early times (and Δt< 5 days), and we do not detect any NIR excess (e.g., due to potential dust formation). For clarification, we note that the optical/NIR data to which we fit a BB were obtained much later (Δt ≳ 10⁵ s) than when the main X-ray emission period of XRT 080109 occurred (Δt ≲ 10³ s, Section 2). For Δt = 1.8 days, we detect the SN in the Swift UVW2 and UVM2 band and include those data in the BB fit, though their inclusion does not have a significant effect on the outcome (the parameters change by less than 2%).

We present the resultant BB luminosity evolution in Figure 8 (right, the top panel) and include that of SN 1999ex for comparison (from Table 8 of Stritzinger et al. 2002, for their adopted values of E(B − V) = 0.28 mag, and explosion date of JD 2,451,480.5), but with a luminosity distance recomputed with H₀ = 73 ${\rm km\, s^{\scriptscriptstyle -1}\, Mpc^{\scriptscriptstyle -1}}$. Both SNe exhibit the initial dip from the cooling stellar envelope emission, with SN 2008D showing a much less pronounced contrast between dip and peak due to heating of ⁵⁶Ni than in SN 1999ex.

The second method for computing the bolometric light output is by direct integration of the observed broadband photometry. Here we are in the favorable position of having NIR data, as the NIR contribution to the bolometric output is usually unknown for SNe Ib/c and can be quite large (30%–50%) for certain SNe at late time (>20 days; Tominaga et al. 2005; Tomita et al. 2006; Modjaz 2007), and even larger for dust-producing SNe (SN 2006jc, e.g., Smith et al. 2008; Di Carlo et al. 2008; Modjaz 2007).

We integrate the total optical and NIR monochromatic fluxes in the region between the effective wavelengths of the UVW1 and K_s filters via the trapezoid approximation. For this purpose, we extrapolate the SED to zero at the blue UVW1 (2220 Å) and red K_s (24,210 Å) edges of the total set of filters. We plot the resultant bolometric light curve ($L_{{\it UVW1} \rightarrow K_s}$) in Figure 8 (right, the bottom panel) and, for comparison, also that of SN 1999ex (based on U → z integration; Stritzinger et al. 2002).

For SN 2008D, we find that the bolometric luminosity based on direct UVW1 → K_s integration is very similar to that based on the BB fits (within ≲ 0.04 dex or ≲8%), except for the first three data points at Δt = 0.8, 1.8, and 2.8 days (different by a factor 1.32, 1.13, and 1.09, respectively). SN 2008D reaches bolometric peak at Δt = 19.2 ± 0.3 days with a bolometric peak luminosity of logL_bol = 42.2 ± 0.1 ${\rm erg\, s^{\scriptscriptstyle -1}}$, i.e., M_bol = −16.8 ± 0.1 mag.

At Δt = 0.84 day, the peak of the SED is at wavelengths bluer than the observed wavelength coverage and explains the missing flux in the direct integration, which is the absolute lower limit to the total light output. We note that for SN 2008D, the agreement between bolometric values derived from BB fits and those based on direct integration is much better than for SN 1999ex; if the emission from SN 1999ex can be as well described by a BB as that of SN 2008D, then the true bolometric peak luminosity of SN 1999ex is not −17.0 mag (Figure 8), but rather −17.7 mag; i.e., the SN is more luminous than previously thought, because it was not observed in the UV nor in the NIR whose data could have been used to construct a full bolometric output.

In Figure 9, we present the contribution of different passbands to the full bolometric luminosity output $L_{{\it UVW1} \rightarrow K_s}$ as a function of SN age. At early times, the UV is 40% of the bolometric luminosity. At later times (30 days after outburst), the NIR is 20%. Most supernova observations lack these measurements, so their derived bolometric flux has significant uncertainties.

Zoom In Zoom Out Reset image size

Now we compare our data to the fits of Soderberg et al. (2008b) via the cooling BB envelope model of Waxman et al. (2007), and to the predictions of Chevalier & Fransson (2008) as the source for the early-time emission, where the stellar envelope expands and cools after the passage and breakout of the shock wave.

In both models, the time evolution of the radius and the temperature of the cooling envelope are weak functions of the ejecta mass (M_ej) and ejecta kinetic energy (E_K), and a stronger function of the radius of the progenitor before explosion (R_⋆) to the power of 1/4. Both models assume that the photophere is in the outer shock-accelerated part of the SN ejecta density profile, but their methods differ. In the following, we use our early-time light-curve data to provide constraints on R_⋆ if the ratio E_K/M_ej can be measured independently from the SN light curves and spectra (Soderberg et al. 2008b; Mazzali et al. 2008; Tanaka et al. 2009).

Analytically, we can rewrite the equations for the radius and the temperature of the cooling envelope as given by W07 and CF08 (Equations (18) and (19), and Equations (2) and (5), respectively), and generalize such that

$$\begin{equation} \frac{R_{\star }}{ 10^{11}\, {\rm cm}} = C T_{ {\rm BB,ev}}^4 R_{\rm {BB,14}}^{0.4} \Delta t_{\rm {d}}^{\alpha } \left(\frac{E_{\rm {K,51}}}{M_{\rm {ej,\odot }} }\right)^{\beta }, \end{equation} \tag{ 1 }$$

with

$$\begin{eqnarray} C & = 0.286\;\;\, \rm { for \; W07 \;\, or }\quad & 3.028 \;\;\;\, \rm { for \; CF08}, \nonumber \\ \alpha & = 1.69\;\;\;\; \rm { for \; W07 \;\; or }\quad & 1.61 \;\;\;\;\;\, \rm { for \; CF08}, \nonumber \\ \beta & = -0.24\;\, \rm { for \; W07 \;\, or }\quad & -0.27 \; \rm { for \; CF08}, \nonumber \end{eqnarray} \tag{ 2 }$$

where T_BB,ev is the BB temperature of the photosphere in eV, R_BB = 10¹⁴R_BB,14 cm, Δt_d time in days after shock breakout, E_K = 10⁵¹E_K,51 erg, and M_ej = M_ej,⊙ M_☉. Thus, the derived BB fits to the early-time data yield independent measurements of R_⋆ for each epoch, for a given E_K/M_ej, whose uncertainty does not influence the value of R_⋆ significantly, given the very weak dependence of R_⋆ on E_K/M_ej. While the power-law scalings of the parameters are fairly similar for the two models, their largest discrepancy lies in a factor of ∼10 difference in the constant C, which is primarily caused by the factor of ∼2 difference in the temperature equations, as noted by CF08, and by the high sensitivity of R_⋆ on T_BB.

Here we explicitly calculate R_⋆ for the two models with our own data combined with published values of E_K/M_ej and include all relevant sources of uncertainties; using the slightly different values for E_K/M_ej for SN 2008D from the literature (Soderberg et al. 2008b; Mazzali et al. 2008; Tanaka et al. 2009), and our estimates for T_BB and R_BB (Table 8), we compute R_⋆ for each of the first four epochs (0.8 <Δt< 3.8 days), and for the range of adopted extinction values (E(B − V)_Host = 0.6 ± 0.1 mag). Taking an average over the epochs and using the different reddening values as bounds, we obtain for the W07 model R^W07_⋆ = (8.6 ± σ_Ex + σ_E/M) × 10¹⁰ cm, where σ_Ex and σ_E/M refer to the systematic error due to uncertain extinction and uncertain E_K/M_ej, respectively, with σ_Ex = 4.8 and σ_E/M = 0.4. In other words, R^W07_⋆ = 1.2 ± 0.7 ± 0.06 R_☉. Likewise for the CF08 model, R^CF08_⋆ = (86 ± 45 ± 4) × 10¹⁰ cm =12 ± 7 ± 0.6 R_☉. Thus, R_⋆ as derived from the cooling stellar envelope emission is 10 times larger according to the CF08 model than according to the W07 model. CF08 had noted this fact, and furthermore, that their larger progenitor size is consistent with R^X-ray_⋆≈ 9 R_☉ based on their simple interpretation of the published X-ray spectrum of XRT 080109 as a purely thermal X-ray breakout spectrum.

In addition to the literal application of both models, it is important to keep in mind the caveats and simplifications that enter them, as both models are consistent with the basic idea of emission from a post-breakout cooling stellar envelope. Any simplifications that lead to a slight misestimation of T_BB,ev will lead to a large effect onto R_⋆ due to its large temperature sensitivity. Caveats here include equating effective temperature (as given by the models) with observed color temperature, and the fact that for temperatures during the time of observations (10⁴ K), the opacity is probably not set by simple Thomson scattering as assumed in the models, but is composition dependent (E. Waxman, 2009, private communication). These aspects mark areas of improvement that need to be addressed if measurements like the ones for SN 2008D are to be used to routinely and precisely measure SN progenitor sizes in the future.

For illustration, we overplot in Figure 8 the predicted values for the Waxman model using the parameters put forth by Soderberg et al. (2008b) for SN 2008D, namely R_⋆ = 10¹¹ cm, M_ej = 5 M_☉, and E_K = 2 × 10⁵¹ erg (see their caption to Figure 3). Our values are consistent with their fits, within the uncertainties. For comparison, we also plot the predicted values for the model of Chevalier & Fransson (2008) with R_⋆ = 9 R_☉.

We can now directly compare our estimated R_⋆ of XRT 080109/SN 2008D with stellar radii of Wolf-Rayet (WR) stars, which possess dense winds (see Abbott & Conti 1987; Crowther 2007 for reviews), and are the most likely progenitors of stripped-envelope SNe (e.g., Filippenko 1991), including that of XRT 080109/SN 2008D (Soderberg et al. 2008b). Specifically, bona fide WN stars—the nitrogen-sequence spectral subtype of WR stars with their helium layer intact but without H (see Smith & Conti 2008 for the distinction between WN and WNH, i.e., H-rich WN stars)—are the likely progenitors of some SNe Ib. Their radii are in the range 1–20 R_☉  with typical sizes between 5 and 10 R_☉ (Herald et al. 2001; Hamann et al. 2006; Crowther 2007). Provided that those models and their data are accurate, R^W07_⋆ ≈ 0.5−1.9 R_☉ as suggested by W07 is smaller by a factor of 5–10 than typical values for such WN radii, while the size of R^CF08_⋆ = 5–20 R_☉ as suggested by CF08 is more in line with data of typical WN stars.

Independent of the considerations above, Soderberg et al. (2008b) infer from their extensive radio observations that the progenitor had a steady wind with a mass-loss rate consistent with those of WR stars.

Figure 10 compares the optical light curves of SN 2008D with those of other SNe with either observed shock breakout or subsequent phase of cooling stellar envelope emission, where the SNe are offset to match the peak magnitude of SN 2008D in the respective bands. Careful comparison with other stripped-envelope SNe caught shortly (∼1 day) after shock breakout (SN 1993J, SN 1999ex) reveals that while the light curves (and spectra) of SN 2008D and SN 1999ex are similar during the ⁵⁶Ni-decay phase, the light curves related to the cooling stellar envelope are different among these three SNe. Furthermore, we note that the B−V color evolution of SN 2008D is very similar to that of SN 1999ex during the time of ⁵⁶Ni decay, but very different during the shock-breakout phase (Figure 11). The color curve of SN 1999ex has a steep rise to and decline from the red during 1 day<Δt< 10 days, peaking at B − V = 1 mag, while that of SN 2008D evolves gradually during this time and reaches B − V = 0.5 mag at most.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

We leave it as a future topic to model the diversity of the very early-time light curves of SNe 1993J, 1999ex, 2006aj, and 2008D in terms of progenitor radius, E_K, and M_ej, and to compare these values with those derived from other methods (e.g., light curve and nebular models). We note, however, that the temporal evolution of the inferred BB temperature and radius of SN 2008D are quite different from those derived for SN 1999ex (Figure 8, the left panel; Stritzinger et al. 2002).

Comparing the light-curve shapes of the different SNe with respect to date of V-band maximum, we find that SN 2008D declines more slowly than any of the other SNe and has a wider peak. One route of parameterizing the SN light-curve shape is using the Δm₁₅(X) parameter, a formalism initially developed for SNe Ia (Phillips 1993), which is defined as the decline (in magnitudes) during the first 15 days after maximum brightness in the passband X. For SN 2008D, we measure Δm₁₅(U) = 1.2 mag, Δm₁₅(B) = 0.8 mag, and Δm₁₅(V) = 0.6 mag.

The very early spectra (Δt≈ 3–5 days) actually resemble those of SN 2006aj (see the top two spectra in Figure 12; see also Mazzali et al. 2008; Malesani et al. 2009), the SN connected with GRB/XRF 060218, which initially led to the identification of this SN as a broad-lined SN (Blondin et al. 2008b; Kong et al. 2008a, but see Malesani et al. 2008). This spectral resemblance suggests that in both systems, the material constituting the photosphere at early times is at high velocity (Mazzali et al. 2008), as expected from homologous expansion. Indeed, the velocity we determined via SYNOW, v_phot≈ 30,000 km s⁻¹ at Δt = 1.8 days in the spectrum with the double-absorption feature (see Section 5.3), is similar to those of broad-lined SNe Ic at the same epoch measured in a slightly different manner (Mazzali et al. 2008).

Zoom In Zoom Out Reset image size

However, while SN 2006aj and other broad-lined SNe Ic continue to have a significant amount of mass at high velocity (at least 0.1 M_☉ at 20,000–30,000 km s⁻¹ of the total of 2 M_☉ ejected mass as derived from models in Mazzali et al. 2006), at Δt ≈ 12 days, the receding photosphere in SN 2008D reveals the lower-velocity layers and the development of He i lines which SN 2006aj and other broad-lined SNe Ic do not show (Figure 12). Spectral synthesis modeling yields an estimate of ∼0.03 M_☉ at v>0.1c for SN 2008D (Mazzali et al. 2008; Tanaka et al. 2009), a factor of 10–50 less than for GRB-SNe.

As Figure 12 shows, the later spectral evolution of SN 2008D resembles that of a normal SN Ib, where the He i lines are as pronounced as in SN 2005hg and more so than in SN 1999ex.

We now investigate the temporal behavior of the He i lines, which are the spectral hallmarks of SNe Ib. In SN 2008D and in a larger sample of SNe Ib, we determine the blueshifts in the maximum absorption in the He i lines via the robust technique of measuring SN line profiles as presented by Blondin et al. (2006).

We refer to our velocity measurements as "He i absorption-line velocities" in the rest of this work. In Figure 13 (see also Table 9), we present the measured values for He i λ 5876, the strongest optical He i line in SN 2008D, along with absorption-line velocities of a sample of other SNe Ib (see Modjaz 2007), measured in the same fashion.

Zoom In Zoom Out Reset image size

In SN 2008D, the He i lines become apparent as early as Δt≈ 6 days (i.e., t_Vmax ≈ −13 days) and are fully identified at Δt≈ 11.7 days (t_Vmax ≈ −6.9 days). Figure 13 reveals that the velocities span a large range for normal SNe Ib; e.g., at maximum light, they range from −14,000 km s⁻¹ (for SN 2004gq) to ∼−10,000 km s⁻¹ (for SN 2005hg), and seem to follow a power-law decline. For comparison, Branch et al. (2002) showed that the He i velocities in the SNe Ib of their limited sample followed a standard pattern and traced fairly closely (within 2000 km s⁻¹ for the same epoch) the photospheric velocities derived from synthetic fits to Fe ii lines, which we overplot in Figure 13 (dashed line, using their power-law fit v_phot ∝ t^−2/(n−1) with n = 3.6). However, the SNe Ib in our sample show a larger departure (up to 6000 km s⁻¹) from the corresponding standard photospheric velocity, larger than those of SN 1998dt, which Branch et al. (2002) had declared as the exception in their sample. In addition, both SNe 2008D and 2005hg show He i line velocities that are lower than the corresponding photospheric velocity for dates before maximum light. This might mean that the line-emitting material is located at radii smaller than the bulk of the material that makes up the photosphere, or that there are optical-depth effects at play.

It has been long suggested that the He i lines are due to nonthermal excitation and ionization of gamma-ray photons produced during the radioactive decay of ⁵⁶Ni and ⁵⁶Co (Harkness et al. 1987; Lucy 1991; Mazzali & Lucy 1998). The fact that the gamma-ray photons need time to diffuse through the ejecta to reach the He layer has been invoked to explain the late emergence (after maximum light) of He i in some SNe Ib (e.g., Filippenko 1997, and references therein). A challenge for those who study the complicated nonthermal excitation effects in SN line formation (Baron et al. 1999; Kasen et al. 2006; Dessart & Hillier 2008) would be to self-consistently explain for SN 2008D its early emergence of He i lines (∼5–6 days after explosion, and ∼12–13 days before maximum light) by excitation of the relatively small amount of synthesized ⁵⁶Ni (0.05–0.1 M_☉, Soderberg et al. 2008b; Mazzali et al. 2008; Tanaka et al. 2009).

Figure 14 shows the four NIR spectra of SN 2008D (taken with the IRTF) in which we detect five prominent blueshifted He i lines. We tabulate their velocities as inferred from maximum absorption in Table 9.

Zoom In Zoom Out Reset image size

Comparison of the NIR spectra of proper SNe Ib 1999ex, 2001B, and 2008D with those of SNe Ic (see Taubenberger et al. 2006 and references therein) shows that all SNe display a strong feature at ∼1.04 μm but only SNe Ib show additional absorption at 1.95–2.0 μm due to blueshifted He i 2.058 μm, while none of the SNe Ic do. For the few SNe Ib in general (Gerardy et al. 2004) and for SN 2008D in particular, the absorption velocity of the strong blueshifted He i 2.058 μm is very similar to the strong blueshifted He i 1.083 μm if we identify the redder absorption trough of the multicomponent feature at 1 μm as He i 1.083 μm. As Figure 15 shows, the blueshifted He i 2.058 μm line is narrower and more well defined than the broad blend at 1 μm.

Zoom In Zoom Out Reset image size

In conclusion, SN 2008D gives supporting evidence to the argument that the broad absorption at ∼1.04 μm is a poor diagnostic for helium, since it is most likely a blend of Fe ii, C i, Ca ii, and S i at the same wavelengths (Millard et al. 1999; Gerardy et al. 2004; Sauer et al. 2006) and He i in SNe Ib. However, claims of significant He in optically classified SN Ic rely on identifying the same strong absorption feature at ∼1.04 μm exclusively with He i 1.083 μm (Filippenko et al. 1995; Clocchiatti et al. 1996; Patat et al. 2001). Our study of SN 2008D implies that determining unambiguously if indeed helium is present in SNe normally classified as Type Ic requires a K-band spectrum to test for the presence of the clean and unblended He i 2.058 μm feature.

Here we compare three properties of XRT 080109 with those observed in a sample of classical XRFs: duration (as parameterized by T₉₀), the shape of the X-ray spectrum, and X-ray-to-optical flux ratio. In addition, we discuss its energy output.

While in each case XRT 080109 is not necessarily an obvious outlier in the observed distribution, together all three considerations paint a strong picture that XRT 080109, along with XRF 060218/SN 2006aj, might be different phenomenologically from the rest of classical XRFs. However, future studies trying to uncover the inherent distribution will need to carefully consider detector effects and rest-frame properties.

• 1.  
  Duration: XRFs observed by HETE-2 and Swift are seen to have similar temporal profiles and durations compared to long-duration GRBs, exhibiting a FRED (i.e., fast rise exponential decline) profile. XRT 080109 can be adequately fit by such a FRED profile (Section 9.1), but the transient's duration is an order of magnitude longer than the median T₉₀ for other XRFs and X-ray-rich GRBs (XRRs), which is roughly 30 s (Sakamoto et al. 2005, 2008). Both XRT 080109 and XRF 060218/SN 2006aj have the longest durations (T₉₀ = 470 ± 30 s and 2100 ±100 s, respectively). We note the caveat that satellites are typically inhibited from triggering on long timescales due to background modulations, and XRT 080109 could not have been detected with any other survey X-ray telescope before Swift.
• 2.  
  X-ray spectrum: XRF 060218/SN 2006aj has a significant X-ray component that (a) is thermal and (b) appears to soften over time, with very high S/N (Campana et al. 2006; Butler 2007a). The X-ray spectrum of XRT 080109 is of much lower S/N, and is well fit by a combined PL+BB (Section 2). While the X-ray emission of XRF 060218 fades at a rate comparable to that often observed for GRB X-ray afterglows, its spectrum is markedly softer (e.g., Soderberg et al. 2006) and also softens strongly in time (Butler 2007a). This is in contrast to GRB X-ray afterglows, which exhibit little spectral evolution at the same epoch: GRB and XRF X-ray afterglows have photon indices of Γ≈ 2, while that of XRF 060218/SN 2006aj was Γ = 3.41 ± 0.13 (Butler & Kocevski 2007b). There is evidence that the XRT 080109 late-time X-ray spectrum (from the Chandra data) was also softer with Γ = 3.6^+1.7_−1.4 (Pooley & Soderberg 2008; Soderberg et al. 2008b) than the early-time spectrum.
• 3.  
  We also compare the radiation output in XRT 080109 for different wavelengths with that of XRFs and their afterglows.

At the dates of our optical and NIR photometry (at t − t_o = Δt = 0.84 and 0.71 day, i.e., ∼70 and 60 ks), the X-ray luminosity is in the range ∼2 × 10³⁸⁻³⁹ ${\rm erg\, s^{\scriptscriptstyle -1}}$. Thus, the X-ray luminosity (for 0.3–10.0 keV) is 3–4 orders of magnitude lower than the luminosity inferred from the optical observations (see Section 6.1); thus, L_X/L_opt ≈ 10⁻³to10⁻⁴ at 70 ks for XRT 080109/SN 2008D. This corresponds to a ratio of fluxes f_O/X = 4 × (10²–10³), while the observed ratio is f_O/X < 70 for classical XRF afterglows (D'Alessio et al. 2006) and GRB afterglows (De Pasquale et al. 2003) at comparable times (using the same definition and units of f_O/X as in D'Alessio et al. 2006). In other words, the optical to X-ray ratio for XRT 080109/SN 2008D is a factor of ∼10³ larger than for XRF afterglows. For XRF 060218/SN 2006aj, the ratio cannot be precisely determined since it was not observed in the desired bandpass at such early time.

The two main features commonly observed in XRFs are not measured for XRT 080109, namely detection of gamma-ray emission and peak of the energy distribution (E_peak).

In summary, XRT 080109, along with XRF 060218/SN 2006aj, appears qualitatively different from the observed distribution of classical XRFs, when considering all three discussed properties together (duration, spectral evolution, and optical to X-ray ratio). Finally, the prompt high-energy emission released in XRT 080109 (E_X,iso ≈ 6 × 10⁴⁵ erg) is 3–4 orders of magnitude lower than in XRFs (Soderberg et al. 2008b).