DISCOVERY OF THE ULTRA-BRIGHT TYPE II-L SUPERNOVA 2008es

SN 2008es was discovered as an optical transient at α = 11^h56^m4913, δ = +54°27'257 (J2000.0) in unfiltered images (calibrated to the SDSS r band) taken by the 0.45 m ROTSE-IIIb telescope (Akerlof et al. 2003) at McDonald Obseratory on 2008 Apr 26 (ROTSE J115649.1+542726; Yuan et al. 2008). UT dates are used throughout this paper. ROTSE-IIIb continued to monitor the transient for ∼50 days. We began a monitoring program with the Palomar 60 inch (P60) telescope (Cenko et al. 2006) in the g, r, and i' bands on 2008 May 2, with continued observations for ∼100 days. Observations in the V band were obtained by the 1.3 m MDM telescope and the 4 K detector with 0.315 arcsec pixel⁻¹ spatial resolution, starting on 2008 May 24 and continuing for 45 days. Secondary standard stars were established through observations of stars from Landolt (1992), and included the determination of nightly extinction corrections. Observations on 2008 August 22–23 in the B, V, and i bands were obtained with the Palomar 200 inch telescope (P200) and the Large Format Camera.

The optical light curve measured from all four optical telescopes is shown in Figure 1 and Table 1. Magnitudes reported in this paper have been corrected for a Galactic extinction of E(B − V) = 0.012 mag (Schlegel et al. 1998). The ROTSE-IIIb and P60 data catch the rise of the source to its peak. We fit the rise of the unfiltered flux measured by ROTSE IIIb with a quadratic function (plotted with a black line in Figure 1), and estimate the time of explosion from the intercept with the time axis to be MJD 54574 ± 1. The error in the time of explosion reflects both the statistical error from the ROTSE-IIIb flux uncertainties and the systematic error from the shape of the function used for the fit. We measure the peak of the light curve from quadratic fits to the individual filter light curves from MJD 54580 to 54620, which yields a peak of MJD 54602 ±1, and corresponds to 23 ± 1 days after the time of explosion in the rest frame of the SN.

Zoom In Zoom Out Reset image size

Within days of the ROTSE discovery, we followed up the source with optical spectroscopy on 2008 May 1 and 8 with the 9.2 m Hobby–Eberly Telescope (HET) and the Marcario Low-Resolution Spectrograph and on 2008 May 2 with P200 and the Double Beam Spectrograph (Yuan et al. 2008). The transient was tentatively classified as a quasar because the spectra showed a featureless blue continuum and one broad emission feature that was associated with Mg ii λ2798 at z = 1.02. We obtained further follow-up spectra on 2008 June 1 and 16 with the HET and on 2008 August 1 with P200. The identification of the redshift and type of the SN was ambiguous (Miller et al. 2008a; Gezari et al. 2008b; Miller et al. 2008b) until later spectra obtained on 2008 August 1 (by our group with P200) and 2008 August 3 (by Chornock et al. 2008 with the 10 m Keck Low-Resolution Imager and Spectrograph) revealed broad Balmer emission lines, reported by Chornock et al. (2008) to be associated with a Type II SN at z = 0.21. Figure 2 shows the evolution of the spectra over time, and Table 2 gives the log of spectroscopic observations. In order to minimize uncertainties in the absolute flux calibration, the spectra are multiplied by a scaling factor so that their flux integrated over the r-band filter matches the P60 r-band photometry closest in time to the observation. We see no evidence of narrow absorption or emission lines superimposed on the SN spectrum indicating a contribution from the host galaxy.

Zoom In Zoom Out Reset image size

Table 2. Log of Spectroscopic Observations

MJDa	Telescope/Instrument	Exposure Time (s)	Grismb	Slit	Resolving Powerb
54586.8	HET/LRS	2 × 600	300 l mm−1	20	300
54587.8	P200/DBSP	2 × 1200	600/4000 (b), 158/7500 (r)	10	1800 (b), 700 (r)
54593.8	HET/LRS	2 × 450	300 l mm−1	20	300
54618.2	HET/LRS	3 × 900	600 l mm−1	20	650
54632.2	HET/LRS	3 × 900	600 l mm−1	20	650
54678.7	P200/DBSP	2 × 1500	600/4000 (b), 158/7500 (r)	10	1800 (b), 700 (r)

Notes. ^aDays in JD−2400000.5. ^bValues for the blue and red side of P200/DBSP are labeled (b) and (r).

Download table as:  ASCIITypeset image

We measure the redshift of the SN to be z = 0.205 ± 0.001 from a Gaussian fit to the broad He ii λ4686 line in the 2008 May 1 and 8 spectra, with a full width at half-maximum (FWHM) of 5500 ±  500 km s⁻¹. Detection of the high ionization line He ii λ4686 at t = 14 to 20 days after explosion in the SN rest frame requires high temperatures and ionization. This is similar to the case of Type II-P SN 2006bp, for which He ii λ4686 was detected in the first few days after explosion, and was associated with outer layers of the SN ejecta that were flash ionized at shock breakout (Quimby et al. 2007b; Dessart et al. 2008).

Strong Balmer lines and Na i λ5892 did not appear until after more than 20 rest-frame days after the explosion. The broad Hα line in the 2008 August 1 spectrum can be fitted with a single Gaussian with an FWHM = (1.02 ± 0.06) × 10⁴ km s⁻¹. The spikes near the peak of Hα are residuals from a poor sky subtraction. The broad Hα flux is (3.7 ± 0.4) × 10⁻¹⁵ erg s⁻¹ cm⁻², corresponding to a luminosity of (3.7 ± 0.4) × 10⁴¹ erg s⁻¹. The Na i λ5892 and Hβ lines have P Cygni line profiles with a blue-shifted absorption component with a minimum at a velocity of (−7.4 ± 0.2) × 10³ km s⁻¹ and (−7 ± 1) × 10³ km s⁻¹, respectively, and a blue absorption wing that extends out to ∼1 × 10⁴ km s⁻¹. Figure 3 shows the broad emission lines, and their best-fitting Gaussian profiles. The lack of a strong P Cygni absorption component in the Hα line and P Cygni profiles in the Hβ and Na i lines is consistent with the bright Type II-L SNe 1979C and 1980K (Panagia et al. 1980; Branch et al. 1981; Uomoto & Kirshner 1986). The width of the Hα line is a measure of the expansion speed of the SN ejecta, while the absorption component of the P Cygni profile traces the outflow of the absorbing material in the line of sight (LOS). Both components indicate an SN blast wave velocity of ∼10000 km s⁻¹.

Zoom In Zoom Out Reset image size

The error in the redshift from the Gaussian fit to the He ii λ4686 line does not reflect systematic errors that may be introduced if the He ii line is blended or is not symmetric. The best-fitting SN II template in the Supernova Identification Code (SNID) database (Blondin & Tonry 2007) to the spectrum 105 days after explosion is SN II-L 1990 K on day +53 (Cappellaro et al. 1995) at z = 0.202 ± 0.006. A fit to all Type II-L SNe in the database gives the same best-fit result. If we restrict the fit to SN templates 1979C and 1980 K (as is done by Miller et al. 2008c), a worse fit is found, with z = 0.213 ±  0.011. The redshift measured from the best template fits is consistent with the redshift determined from the He ii line, and so we use this redshift but change the error bars to reflect the uncertainties in the template fit, z = 0.205^+.003_−.009.

No host galaxy is detected in the pre-SN Sloan Digital Sky Survey (SDSS) Data Release 6 (DR6) images (Adelman-McCarthy et al. 2008), down to a limiting magnitude of r ∼ 23. This constrains the host galaxy to be a dwarf galaxy with M_r ≳ −17 at z = 0.205. The closest galaxy detected by SDSS is located 83 northeast of the SN, with r = 23.1 ± 0.4 and g − r = −0.1 ± 0.5 mag. At a redshift of z = 0.205, this angular separation corresponds to a projected separation of ∼40 kpc, which is much larger than that would be expected if the galaxy were the SN host. During our 2008 August 1 spectrum, we placed the slit at a position angle so that this galaxy would fall in the slit, but we did not detect any emission lines from the galaxy that could be used to determine its redshift. The accuracy of the SN redshift could be greatly improved with future spectroscopic observations at the position of the SN, after the SN has faded significantly, when emission lines from the host galaxy that are now outshined by the SN may be detectable.

We requested Target of Opportunity (ToO) observations with the Ultraviolet/Optical Telescope (UVOT; Roming et al. 2005) and X-Ray Telescope (XRT; Burrows et al. 2005) on board the Swift observatory (Gehrels et al. 2004) on 2008 May 14, with continued monitoring through 2008 September 2. The spectral energy distribution (SED) of the source on 2008 May 14 was well described by a blackbody with T ∼ 10⁴ K for z ∼ 1. The corresponding bolometric luminosity of ∼10⁴⁶ erg s⁻¹, in combination with the lack of X-ray emission indicative of persistent quasar activity, suggested that the source may have been a flare from the accretion of a tidally disrupted star around a central black hole of 10⁸M_☉ (Gezari & Halpern 2008). This interpretation was soon ruled out, however, by the rapidly cooling temperature of the emission seen in the later Swift observations, and the subsequent appearance of SN-like broad features as the spectra evolved (Miller et al. 2008a; Gezari et al. 2008b).

No X-ray source was detected in any of the individual XRT epochs. The count rate in a 23'' extraction radius for the accumulated exposure time of 50.27 ks is 4.8 × 10⁻⁴ counts s⁻¹, and places a 3σ upper limit to the 0.3–10 keV flux, using the standard energy conversion factor of 2.4 × 10⁻¹⁴ erg s⁻¹ cm⁻² s⁻¹, which corresponds to a luminosity at z = 0.205 of L_X < 2.4 × 10⁴² erg s⁻¹ using a luminosity distance for z = 0.205 of d_L = 1008 Mpc (H₀ = 70 km s⁻¹ Mpc⁻¹, Ω_m = 0.30 and Ω_Λ = 0.70).

The UVOT magnitudes were measured with a 3'' aperture and an aperture correction based on an average point-spread function (PSF). We convert the magnitudes into flux densities using the count rate to flux conversions determined for the Pickle standard stars (Poole et al. 2008). The UVOT uvw2 (λ_eff = 203 nm), uvm2 (λ_eff = 223 nm), uvw1 (λ_eff = 263 nm), u (λ_eff = 350 nm), b (λ_eff = 433 nm), and v (λ_eff = 540 nm) band magnitudes are plotted in Figure 1 and listed in Table 3. In Figure 4, we show the SED of the SN over time from 200 to 800 nm, measured by UVOT in combination with optical data points from the P60 data closest in time to the UVOT observations. The UVOT b and v bands are interchangeable with the Johnson B and V system, but the u band has a bluer response than the Johnson U band, and we make this distinction throughout the paper.

Zoom In Zoom Out Reset image size

We fit the SEDs over time with a blackbody at z = 0.205, and fit temperatures starting t = 26 days after explosion with 14000 K, cooling down to 6400 K at t = 91 days, with the corresponding blackbody radii expanding from 3 × 10¹⁵ cm to 5 × 10¹⁵ cm. These radii are ≳50 times larger than a standard red supergiant photosphere, and mostly likely correspond to the radius of the expanding SN blast wave. We see evidence of an excess above that of a pure blackbody in the uvw2 band at early times (26–27 observed days after explosion) that may be the effect of strong Fe iv and Fe iii features in that region, seen in the spectra of SN II-P 2006bp at similar photospheric temperatures (Dessart et al. 2008).

The high signal-to-noise ratio (S/N) MDM 1.3 m V-band observations, with Δ mag ∼0.01, measure a linear decay from 9 to 53 days after the peak of 2.95 ± 0.02 mag (100 days)⁻¹, or 3.55 mag (100 days)⁻¹ in the SN rest frame. This decay rate is just above the transition of the average decay rates in the first 100 days after maximum from Type II-plateau (II-P) to Type II-linear (II-L) classes of SNe defined by Patat et al. (1994) to be 3.5 mag (100 days)⁻¹ in the B band (at z ∼ 0.2 the V band is close to the rest-frame B band). This classification is nonexclusive, however, since Patat et al. (1994) point out that there is a continuum in rates between the plateau and linear subclasses. Type II-L SNe are on average brighter than Type II-P SNe, and II-L SNe appear to have a bimodal distribution of absolute peak magnitudes, with a subclass of bright II-Ls with 〈M_B〉 = −19.27 (Richardson et al. 2002). Yet even the bright II-L SNe are still 3 mag fainter than the peak magnitude of SNe 2008es.

The Swift and MDM V-band observations show evidence of a flattening of the optical light curve ∼65 rest-frame days after maximum, with a slope that is consistent with the ⁵⁶Co decay rate of 0.98 mag (100 days)⁻¹ (shown with a dotted line in Figure 1, stretched in time to match the time dilation of SN 2008es). The estimated luminosity at the time of the MDM V-band data point at 104 days after explosion in the SN rest frame is ≈1 × 10⁴² erg s⁻¹ (a factor of ∼3 below the luminosity on day 75 in the SN rest frame), and using the expression $L = 1.42 \times 10^{43} e^{-t_{d}/111} M_{Ni}/$ M_☉ erg s⁻¹ (Sutherland & Wheeler 1984) implies an upper limit to the initial ⁵⁶Ni mass of ≲0.2 M_☉, which is close to the upper end of the range of masses measured for Type II SNe (e.g., 0.3 M_☉ for SN 1992am; Schmidt et al. 1994).

In Figure 5, we compare the u − B, B − V, and g − r colors of SN 2008es over time to the evolution of the U − B and B − V colors of SN II-L 1979C which have a linear slope of ∼0.85 mag (100 days)⁻¹ (Panagia et al. 1980). The K-correction at the redshift of SN 2008es makes the g − r color close to rest-frame B − V. We also show the u − B and B − V light curves of SN II-P 2006bp measured by Swift (Immler et al. 2007). All curves are shown in days since explosion in the SN rest frame. The light curve of SN 2008es has a much slower evolution of the B − V color than seen in prototypical Type II-P SN 2006bp, and is in closer agreement with the colors and linear slopes of prototypical bright Type II-L SN 1979C.

Zoom In Zoom Out Reset image size

We estimate the bolometric luminosity of SN 2008es over time (as shown in Figure 6) using the blackbody curve fits shown in Figure 4. The luminosity during the optical peak, 23 days after the explosion in the SN rest frame, is 2.8 × 10⁴⁴ erg s⁻¹, and decays exponentially as L_bol(t) = L_bol(t = 0) × e^−αt erg s⁻¹, where L_bol(t = 0) = 6.5 × 10⁴⁴ erg s⁻¹, α = 0.040 ± 0.004, and t is the rest-frame days since explosion. We estimate the total energy radiated by integrating under the curve from day 21 to 75 in the SN rest frame to get E_tot = 5.6 × 10⁵⁰ erg, a value comparable to the radiated energy of the extreme SN 2006gy (Ofek et al. 2007; Smith et al. 2007) which emitted ∼1 × 10⁵¹ erg s⁻¹ in the first two months. These radiated energies, so close to the total kinetic energy produced in a core-collapse explosion (typically in the range of 0.5–2.0 × 10⁵¹ erg), require an efficient conversion of the kinetic energy of the SN ejecta into radiation.

Zoom In Zoom Out Reset image size

Figure 6 shows the evolution of the bolometric luminosity of SN 2008es in comparison to two Type II-P SNe with detailed non-LTE modeling of their photometry and spectra by Dessart et al. (2008): SNe 2005cs and 2006bp. The luminosity curves have been scaled by a factor of 50 for comparison with SN 2008es. The behavior of SN 2008es is markedly different than the SNe II-P, with a factor of 50 greater luminosity at 23 days after explosion compared to the peak luminosity of SN 2006bp that occurs a couple of days after explosion, and with a much slower decline in luminosity over 75 days in comparison to ∼15 days in SNe 2005cs and 2006bp. Figure 6 also shows T_bb and R_bb for SN 2008es in comparison to the Type II-P SNe 2005cs and 2006bp, with the radii of the Type II-P SNe scaled up by a factor of 5 for comparison. In contrast to SN 2008es which has T = 14000 K at 23 days after explosion, the photospheric temperatures of the Type II-P SNe have already cooled to 6000 K by day 20, and their photospheric radii are ∼5 times smaller. It is clear that there is a different mechanism powering the long-lived, and extremely luminous, emission from SN 2008es.

Given the estimate of the expansion speed, v_s, from the broad Hα line of 10000 km s⁻¹, the radius of the ejecta at the time of the peak of the optical light curve, t_peak = 23 days after explosion in the SN rest frame, is expected to be R_ph = v_st_peak = 2 × 10¹⁵ cm, which is in good agreement with the radius of the blackbody fit at that time (R_bb ∼ 3 × 10¹⁵ cm). If the luminosity during the optical maximum is powered by interaction with the CSM, this requires a density of order n ∼ L/(2πR²v³_sm) ∼ 2 × 10⁹ cm⁻³, where m is the mean mass per particle (2.1 × 10⁻²⁴ g for H+He) and we use R_ph = 3 × 10¹⁵ cm. This implies an enclosed mass within R_ph of ∼0.2 M_☉, and corresponds to a mass-loss rate of $\dot{M} \sim M/(tv_{s}/v_{w}) \sim 2.5 \times 10^{-3} (v_{w}$/10 km s⁻¹) M_☉ yr⁻¹, where v_w is the velocity of the stellar wind (∼10 km s⁻¹ for a red supergiant progenitor).

We can place an upper limit on the progenitor's mass-loss rate if we assume that X-ray emission is produced by the interaction of the SN shock with the ambient CSM. In the forward shock, a H+He plasma will be heated to T = 1.36 × 10⁹[(n − 3)/(n − 2)]²v²_s4 K (Chevalier & Fransson 2003), where n is the ejecta density parameter (in the range of 7–12) and v_s4 is the shock velocity in 10⁴ km s⁻¹, and will produce an X-ray luminosity of $L_{X}=3.3\times 10^{60}\Lambda (T)({{\dot M}_{-6}}/v_{w1})^{2}(v_{s4}t_{d})^{-1}$ erg s⁻¹ (Immler et al. 2002), where v_w1 is the stellar wind speed in units of 10 km s⁻¹, t_d is the time since explosion in days, $\dot{M}_{-6}$ is the progenitor's mass-loss rate in units of 10⁻⁶ M_☉ yr⁻¹, and Λ(T) = 2.4 × 10⁻²³g_ffT^0.5₈ is the cooling function for an optically thin heated plasma, where g_ff is the free–free Gaunt factor, and T₈ is the temperature in units of 10⁸ K. Using the width of the broad Hα line to estimate v_s4 ∼ 1, and using t_d = 23, the upper limit on L_X(0.3 − 10) keV implies that $\dot{M} \lesssim 5 \times 10^{-4} v_{w1}$ M_☉ yr⁻¹. This upper limit to the mass-loss rate is still a factor of 5 below what is required to power the observed peak bolometric luminosity purely by interaction with the CSM.

The B-band light curve and B − V colors of bright SN II-L 1979C were successfully modeled by Blinnikov & Bartunov (1993) with a red supergiant progenitor with an extended structure (R ∼ 6000 R_☉), a small hydrogen envelope mass (∼1–2 M_☉), and a dense superwind with a mass-loss rate of $\dot{M} \approx 10^{-4}$M_☉ yr⁻¹. In this model, the large peak luminosity of SN 1979C (M_B = −19.6) is produced by the dense stellar wind that reprocesses the UV photons produced at shock breakout, and can be boosted up to M_B ∼ −22 by increasing the mass-loss rate to $\dot{M} \approx 10^{-3}$ M_☉ yr⁻¹. Their model predicts a steep drop in luminosity 50 days after explosion, when the recombination (in the diffusion regime) reaches the silicon core. After this point, the light curve has an extended tail powered by radioactive decay and/or emission from the shock propagating into the CSM. These predictions are consistent with the sharp drop and subsequent ⁵⁶Co decay slope tentatively detected in the V-band light curve of SN 2008es. The Blinnikov & Bartunov (1993) model also explains the absence of the absorption component in Hα, which is due to the fact that at late stages the temperature inside the dense envelope is lower than the shock wave temperature.

Young et al. (2005) find these values for the radius and superwind unrealistic, and prefer a model in which the bright luminosity at the early phase of the SN light curve is the result of an optical afterglow from a GRB jet interacting with the progenitor's hydrogen envelope. The main problem with this model is that the optical afterglow emission produced from the external shocks created by the interaction of the relativistic jet with the H envelope will emit nonthermal synchrotron radiation, which is in direct contradiction with the cooling blackbody SED of SN 2008es.

Another possibility, favored by Miller et al. (2008c), is that the luminosity is powered by the interaction of the SN blast wave with a massive circumstellar shell that was ejected in an episodic eruption (Smith & McCray 2007). Similarly, the luminosity could be powered by an interaction between such shells (Woosley et al. 2007). Both of these models produce their luminosity via the dense shell that decelerates the blast wave, and converts the bulk kinetic energy into radiation through shocks. The intermediate-width (a few thousand km s⁻¹) P Cygni lines produced from the dense postshock gas, and the slow escape of radiation due to photon diffusion of the thermal energy through the opaque shell in these models with massive shells are not consistent with the broad, symmetric Hα line emission and relatively fast rise time of SN 2008es.

If the bolometric luminosity of SN 2008es was powered solely by thermalization of gamma rays from radioactive cobalt decay, then the luminosity at t_d = 23 would require 25 M_☉ of ⁵⁶Ni to be synthesized in the explosion. This is 3 orders of magnitude larger than that typically produced in Type II SNe, and much larger than the ejecta mass which is expected to be less than 5 M_☉ to avoid a plateau from H recombination in the optical light curve. Although it is possible for such a large amount of ⁵⁶Ni to be produced by an extremely massive progenitor star (>100 M_☉) that underwent a pair-instability supernova (Ober et al. 1983), we already find evidence for a transition to a radioactive heating-dominated light curve at late times, which suggests that the light curve was not dominated by radioactive decay near the peak. Furthermore, the calculations of Scannapieco et al. (2005) predict a long, slow rise to maximum, and slow expansion speeds of ∼5000 km s⁻¹, which is neither consistent with the 23 days rise to maximum nor the 10000 km s⁻¹ expansion speed seen in SN 2008es.

Given the contradictions with the more exotic scenarios discussed above, the extreme luminosity of Type II-L SN 2008es seems most likely to be on the bright end of a range of luminosities and light curves possible with variations of progenitor star H envelope and stellar wind parameters presented in the models by Blinnikov & Bartunov (1993). However, even the largest red supergiant stars have radii ≲1500 R_☉ (Levesque et al. 2005), and thus a progenitor star with the extreme extended structure assumed by Blinnikov & Bartunov (1993) is problematic. Type II-L SN light curves with peak magnitudes of M_B ∼ −19 were reproduced by Swartz et al. (1991) for core-collapse models with more realistic envelope radii (1200 R_☉). More detailed modeling is needed to determine if the required parameters for SN 2008es are consistent with stellar evolution.