OPTICAL OBSERVATIONS OF THE TYPE Ic SUPERNOVA 2007gr IN NGC 1058

Ground-based optical (UBVRI) photometry of SN 2007gr was obtained with the 0.76 m KAIT, the 1.0 m Nickel telescope at Lick Observatory, and the 0.8 m THU-NAOC Telescope (TNT) at NAOC Xinglong Observatory in China (Wang et al. 2008; Huang et al. 2012). The instrumental response functions of the former telescopes, obtained by multiplying the filter transmission functions by the quantum efficiency of the CCD detectors and the atmospheric transmission, can be seen in Figure 2 of Wang et al. (2009). The response curves of the TNT are not shown because the accurate transmission curves of the UBVRI filters are not available.

All of the KAIT and Nickel images were reduced using a mostly automated pipeline (Ganeshalingam et al. 2010). The TNT images were reduced with the Image Reduction and Analysis Facility (IRAF⁸) standard procedures, including correction for bias, flat-field division, and removal of cosmic rays. We applied host-galaxy subtraction to minimize contamination from the galaxy background and close field stars. Template images of the host galaxy were obtained when SN 2007gr had faded away with negligible flux. The KAIT, Nickel, and TNT templates of the host galaxy were obtained on 2010 December 2, 2008 August 30, and 2010 November 29, respectively. The point-spread function (PSF) fitting method was applied to obtain the instrumental magnitudes of SN 2007gr and 12 local standard stars as shown in Figure 1.

Instrumental magnitudes were then converted into the standard Johnson UBV (Johnson et al. 1966) and Kron–Cousins RI (Cousins 1981) systems, with the color terms and zeropoints determined by measuring a series of Landolt (1992) standards on the photometric nights. Table 1 lists the average color terms for the relevant telescopes and filters. The UBVRI magnitudes and uncertainties of the local standard stars are listed in Table 2. To alleviate systematic discrepancies in the photometric systems on different telescopes, we further applied S corrections (Stritzinger et al. 2002) to our photometry using our own spectra presented in this paper as well as those published by V08. After the S corrections, the KAIT and Nickel photometry agree to within 0.05 mag. The S corrections were not applied to the TNT data owing to the lack of accurate response curves of the corresponding filters. The final flux-calibrated and S-corrected UBVRI magnitudes are presented in Table 3, and the corresponding light curves are shown in Figure 2.

Zoom In Zoom Out Reset image size

Two late-time images of SN 2007gr were obtained with the HST/Wide Field and Planetary Camera 2 (WFPC2) on 2008 November 20 and 25 (GO-10877; PI: W. Li), corresponding to t ≈ 455 days after maximum light. The observations were made in the F450W, F555W, F675W, and F814W filters, for which the transmission curves are very similar to those of the Johnson–Cousin BVRI filters, as illustrated by the small coefficients required for the transformations between these two systems (Sirianni et al. 2005). The total exposure times were 800, 460, 360, and 700 s, respectively. Photometry was extracted using the DOLPHOT 2.0 WFPC2 module, which is an adaptation of HSTphot (Dolphin 2000). With the spectrum taken around 430 days after B-band maximum, we apply S corrections to convert the magnitudes of the HST system into those of the Johnson–Cousin system. The final magnitudes are listed in Table 3.

There are 13 low-resolution optical spectra of SN 2007gr obtained with the Kast double spectrograph (Miller & Stone 1993) on the 3 m Shane telescope at Lick Observatory and the Low Resolution Imaging Spectrometer (LRIS; Oke et al. 1995) on the 10 m Keck-I telescope, spanning from −7.8 to +430.2 days since B-band maximum. A journal of spectroscopic observations is given in Table 4.

All data were reduced with standard IRAF routines and IDL scripts described by Silverman et al. (2012). The spectra were flux calibrated using standard stars observed during the same night at airmasses similar to those of the SN. The slit was aligned along the parallactic angle for each of our spectral observations to minimize the chromatic effects of atmospheric dispersion (Filippenko 1982), but in any case the Keck/LRIS spectra were taken with an atmospheric dispersion compensator. Absolute fluxes of the spectra were not checked against the photometry, but the relative spectrophotometry is accurate: the colors synthesized from spectra agreed with the observed ones to within 0.05 mag (see also the discussion by Silverman et al. 2012). The spectral evolution of SN 2007gr is displayed in Figure 3. All spectra were corrected for the host-galaxy redshift of z = 0.001728 (Tifft & Cocke 1988).

Zoom In Zoom Out Reset image size

In an attempt to identify the ions responsible for the lines in the spectra of SN 2007gr and to constrain the corresponding parameters such as the photospheric velocity and temperature, we adopted the parameterized resonance scattering synthetic-spectrum code SYNOW (Fisher 2000; Branch et al. 2005). Figure 7 shows our best fit to the two spectra taken at t = −7.8 days and t = +12 days.

Zoom In Zoom Out Reset image size

To fit the spectrum taken at t = −7.8 days (Figure 7(a)), we assumed a photospheric velocity of v_phot = 10,000 km s⁻¹ and a blackbody temperature of T_bb = 9000 K. One can see that the SYNOW fit matches well the observed spectrum, with the main absorption features at ∼4300 Å (Mg ii λ4481), 4900 Å (Fe ii λλ4924, 5018, 5169), 6200 Å (Si ii λ6355), 7600 Å (O i λ7774), and 8200 Å (Ca ii λλ8542, 8662, 8498) being well reproduced. The small notches at ∼6400 Å and 7050 Å can be attributed to C ii λ6580 and C ii λ7235, respectively. Such identifications agree with those of V08 based on their spectrum. Note that the Fe ii λλ4924, 5018, 5170 features are clearly separated, which is very uncommon among SNe Ic.

There is a small notch at ∼5500 Å in the spectrum at t = −7.8 days that can be attributed to He i λ5876 at a velocity of 17,000 km s⁻¹. Another possible signature of helium is the weak absorption feature at ∼6700 Å, which might be produced by He i λ7065 at a velocity of 16,000 km s⁻¹. Owing to blending with C  ii λ6580, He i λ6678 is hard to identify. The SYNOW fit indicates that the helium layer has a velocity of about 19,000 km s⁻¹ and likely lies far above the photosphere. Possible helium features evolve rapidly and become nearly invisible in the spectrum at t = −2.8 days (see the inset in Figure 7(a)). The detection of helium features in very early-time spectra of SN 2007gr suggests that the progenitor had a thin helium envelope immediately before the explosion.

Figure 7(b) shows the SYNOW fit to the t = 12 day spectrum, with the adopted v_phot and T_bb being 7000 km s⁻¹ and 7400 K, respectively. It is apparent that the strengths of some lines of ionized species, such as Si ii and C ii, become weaker with decreasing photospheric temperature, while the corresponding neutral lines emerge at this phase. In particular, there are rather strong absorption features associated with C i lines in the range 8800–9400 Å. This indicates that SN 2007gr is a carbon-rich SN Ic (see a similar argument by V08). The multiple absorption features near 7200 Å could be identified as Si i λλ7020, 7289, 7420. Nevertheless, detailed modeling is needed to confirm these line identifications.

Figure 8 shows the spectral comparison of SN 2007gr with some notable SNe Ic such as SN 1994I (Filippenko et al. 1995), SN 2002ap (Foley et al. 2003), and SN 2004aw (Taubenberger et al. 2006) at several epochs (∼0 day, ∼2 weeks, ∼5 months, and ∼13 months since B-band maximum light). SN 2008D (Modjaz et al. 2009), a SN Ib, is also plotted for comparison. All of the spectra have been corrected for the redshift of the host galaxy.

Zoom In Zoom Out Reset image size

Figure 8(a) shows the comparison around maximum light. Both SN 2007gr and the comparison SNe reveal features of intermediate-mass elements such as Ca ii, Si ii, Mg ii, and O i, as well as iron-group elements (Fe ii). Of our SN Ic sample, SN 2007gr exhibits the narrowest spectral lines, while SN 2002ap shows broader lines with less structure (especially in the region 4000–4500 Å). The expansion velocity of SN 2007gr is ∼7000 km s⁻¹ at maximum brightness as estimated using the Si ii λ6355 line, compared to a velocity of ∼ 18,000 km s⁻¹ at B maximum for SN 2002ap. This suggests that SN 2007gr is less energetic than SN 2002ap because the former has a smaller ejected mass (see discussion in Section 5).

Figure 8(b) shows the comparison at about two weeks after B-band maximum. As in the comparison SNe, the spectrum of SN 2007gr has evolved while maintaining most of its characteristics. The Ca ii NIR features gain strength, while the Si ii lines become weaker and are nearly invisible in the spectrum. One notable aspect of SN 2007gr is the appearance of three narrow absorption features at 4800–5000 Å, which may correspond to Fe ii λλ4924, 5018, 5169 lines. However, these features are less structured in the spectra of the comparison SNe, likely because of line blending. Moreover, SN 2007gr exhibits triple absorption features near 9000 Å, probably attributable to C i lines, which are also visible in the spectrum of SN 1994I. In contrast, only one broad feature is seen at 8800 Å in spectra of SN 2002ap and SN 2004aw. Such a difference in the absorption-line widths provides additional evidence that SN 2007gr had a lower energy per unit mass.

The comparison at t ≈ 5 months is shown in Figure 8(c). The spectra are now dominated primarily by the emission lines [O i] λλ6300, 6364, a blend of [Ca ii] λλ7291, 7324 and [O ii] λλ7300, 7330, and the Ca ii NIR triplet. These lines generally have asymmetric structures, with the profiles varying for different compositions and SNe. For example, the [O i] λλ6300, 6364 doublet displays a stronger red peak in SNe 1994I, 2002ap, and 2007gr, but seems to have a stronger blue peak in SN 2004aw. While the Mg i] line exhibits a prominent blue peak in SN 2007gr, it has a strong red component in SN 2002ap. Such diversity in the line profiles is likely related to the asymmetric distribution of the ejecta (Milisavljevic et al. 2010). The profile of the Ca ii NIR triplet is also different for the SNe Ib/c in our comparison sample. However, it is difficult to use the Ca ii NIR feature to probe the ejecta structure because it can form in different layers of the ejecta owing to a lower ionization energy, and it may also suffer possible contamination from the [C i] λ8727 line (Matheson et al. 2000).

A common feature of SN 2007gr and SN 2002ap is the strong Mg i] λ4571 emission, with the intensity ratio of Mg i] to [O i] being 0.177 ± 0.049 and 0.244 ± 0.045, respectively. The corresponding ratios are 0.103 ± 0.084 and 0.084 ± 0.095 in SN 1994I and SN 2004aw, respectively. Differences in the Mg i]/[O i] ratio may be interpreted as variations in the degree of stripping of the envelopes experienced by the progenitor star, with a higher ratio implying a greater degree of stripping as more of the inner layers are exposed to the observers (e.g., Foley et al. 2003). This is consistent with the trend that the observed Mg i]/[O i] ratio increases with time at later phases. However, a very high ratio of Mg i]/[O i] was also reported for the Type IIb SN 2001ig, whose outer layers of H and He were only partially stripped (Silverman et al. 2009), suggesting other possible explanations such as mixing. The higher Mg i]/[O i] ratio could be also explained if less of the C/O layer is observed owing to a lower transparency of the ejecta (H09). Moreover, the measured line ratios may be influenced by changes in the environment, as the [O i] and Mg i] lines are formed via different mechanisms (Kozma & Fransson 1998).

A comparison of the very late-time nebular spectra is shown in Figure 8(d). The Mg i] λ4571 emission becomes noticeably stronger in SN 2007gr and SN 2002ap, with the Mg i]/[O i] ratio continuously increasing. However, the Mg i] line can barely be observed in SN 2004aw, and the increasing Mg i]/[O i] trend does not hold for it. By t ≈ 400 days after maximum light, the Mg i] λ4571 emission line in SN 2007gr develops a nearly symmetric double-peaked profile, while the line profile in SN 2002ap still maintains the asymmetric structure seen earlier. A common feature for SN 2007gr and the other two comparison SNe is the weakening of the [Ca ii] λλ7291, 7324 plus [O ii] λλ7300, 7330 blend and the strengthening of the Na i lines with time. This can be explained with a gradual decrease in the density of the ejecta (Filippenko et al. 1990).

The expansion velocities of the ejecta were estimated by fitting a Gaussian profile to the minimum in the absorption lines. The spectra presented by V08 were also included in our analysis. The velocity evolution of different ions is shown in Figure 9. All velocities have been corrected for the redshifts of the host galaxy and the relativistic effect.

Zoom In Zoom Out Reset image size

The expansion velocity inferred from the Ca ii NIR triplet is apparently higher than for other elements. It evolved rapidly from ∼14,000 km s⁻¹ at t = −8 days to <11,000 km s⁻¹ at t = +10 days and maintained a constant velocity thereafter. This indicates that a considerable amount of calcium was distributed in the outermost layers of the ejecta. Such a distribution is perhaps caused by a primordial abundance of calcium that is mixed with the outer part of the envelope (e.g., Li & McCray 1993) and/or nucleosynthesis from an interaction of SN ejecta with circumstellar material (Gerardy et al. 2004). After t ≈ 20 days from maximum light, the velocity inferred from other elements such as Na, Mg, O, and Fe also remains nearly constant.

After the B-band maximum, Si ii λ6355 becomes weaker, and blending with Si i lines makes the measurement of its velocity quite uncertain. In the spectra taken at t ≈ 20 days the Si features can be hardly detected. The expansion velocity of Fe ii is estimated using the average measured from the three absorption lines at 4924, 5018, and 5169 Å; it follows a trend similar to that of O i λ7774, decreasing from ∼11,000 km s⁻¹ in the pre-maximum phase to <7000 km s⁻¹ at t > 20 days. It is worth pointing out that the measured velocity of Na i increased at t = 15 days, possibly because of recombination of Na ii to Na i (H09).

Pre-explosion images of SN 2007gr were obtained on 2001 July 3 with the WFPC2 onboard HST as part of program GO-9042 (PI: S. Smartt). The post-explosion WFPC2 images were obtained about 450 days after maximum light as mentioned in Section 2.2. In Figure 10, we show the pre- and post-explosion HST WFPC2 images of the site of SN 2007gr.

Zoom In Zoom Out Reset image size

To accurately determine the explosion site of SN 2007gr, we use a total of 14 stars in the F450W image and 12 stars in the F814W image to establish the geometric transformations from the post-explosion HST images to the pre-explosion ones. This yields a pixel position of [251, 119] for the SN position in the WF2 pre-explosion image (F450W), with an rms uncertainty of ∼002. A similar SN position is obtained in the pre-explosion F814W-band image, with an offset <0.5 pixel. As can be seen in Figure 10, no object was detected within the position uncertainty of the SN site in F450W or F814W images. Our result is consistent with that obtained by Crockett et al. (2008) using adaptive-optics observations on the 8.1 m Gemini North Telescope.

However, there is a bright, point-like source located at a distance of ∼013 from the SN site, which corresponds to ∼6.6 pc if a distance of 10.60 Mpc is assumed for NGC 1058 (Schmidt et al. 1994). The PSF profile of this object is found to be similar to that of SN 2007gr, implying that it may be a single star or a compact star cluster. DOLPHOT photometry of this source in the pre-explosion WFPC2 images gives F450W = 21.37 ± 0.03 mag and F814W =20.82 ± 0.03 mag, slightly brighter than the values obtained by Crockett et al. (2008) using HSTphot. We further examined the post-explosion images, yielding 21.47 ± 0.03 mag for F450W, 21.29 ± 0.02 mag for F555W, 21.12 ± 0.03 mag for F675W, and 20.85 ± 0.02 mag for F814W. Assuming the photometric result from DOLPHOT is more accurate than that from HSTphot, the star cluster close to SN 2007gr became slightly dimmer in the post-explosion images (especially in F450W), thus favoring the hypothesis that the progenitor of SN 2007gr was a member of the closest star cluster. However, this possible change in magnitude might suffer from light contamination from the supernova, and it should be checked with future HST images. Adopting the extinction correction and distance to NGC 1058 obtained in Section 4, we find absolute magnitudes of M_F450W = −9.01 ± 0.35 mag, M_F555W = −9.12 ± 0.35 mag, M_F675W = −9.22 ± 0.35 mag, and M_F814W = −9.43 ± 0.35 mag for the post-explosion images.

To probe the nature of this object, we fit the observed spectral energy distribution (SED) with distinct theoretical models of a single star and a star cluster. The star spectra are taken from the stellar spectral flux library by Pickles (1998), while the star-cluster spectra are computed with the starburst99 code (Leitherer et al. 1999). Note that the SEDs of clusters or single stars are significantly affected by the adopted metallicity. For SN 2007gr, the relative oxygen abundance log(O/H) + 12 is estimated to be 8.50 ± 13 dex at the explosion site according to Modjaz et al. (2011), which is about 37% of the solar value. We thus adopt a subsolar metallicity of Z = 0.008 for the SED model in the analysis. A χ² test is used to examine which members of the SED families are most compatible with the observed SED.

Figure 11 shows the HST WFPC2 integrated photometry of the object closest to the SN and the best-fit SED models. Fitting the HST photometry for the stellar model yields the solution of a yellow (F0-type) supergiant, with a χ² of 8.7 for four data points. Besides this source, there are also several others around the SN site with similarly high luminosity (see also Figure 2 of Crockett et al. 2008). This suggests that these brighter sources (including the one closest to the SN) are less likely to be single stars because of their short lifetimes and scarcity in spiral galaxies. On the other hand, the star-cluster model seems to yield a better fit than single stars, with a χ² value of 4.8 for the post-explosion images and 0.9 for the pre-explosion ones.

Zoom In Zoom Out Reset image size

Both of the pre- and post-explosion SEDs favor a star cluster at an age of 30–38 Myr, with a turn-off mass of ∼8–9 M_☉ (Girardi et al. 2000; Marigo et al. 2008). Such a low mass would not produce a Wolf–Rayet star via single-star evolution, and it is therefore expected that the progenitor of SN 2007gr was a low-mass Wolf–Rayet star resulting from an interacting binary. We note that Crockett et al. (2008) found two solutions for the age of this assumed cluster: 7.0 Myr and 20–30 Myr. This difference is caused primarily by the solar metallicity used in their fitting (versus Z = 0.008 adopted in our analysis). We caution, however, that SN 2007gr does not really coincide with the assumed star cluster, and it is possible that the supernova is not a member of this star cluster. The site should be scrutinized with HST in several years, when the supernova becomes significantly dimmer.

The emission-line profiles in the nebular spectra of SNe are powerful probes of the geometric structure of the ejecta in the explosion. The forbidden lines of oxygen, magnesium, and calcium are the strongest lines in the nebular phase, and have been usually used for studies of this kind. For example, the double-peaked [O i] λ6300 profile seen in some SNe Ib/c has been attributed to a geometric effect of a torus-like structure of O-rich ejecta (Maeda et al. 2008; Modjaz et al. 2008). Taubenberger et al. (2009) compared the line profiles of Mg i] λ4571 and [O i] λ6300 with a sample of 39 CC SNe, suggesting that O and Mg are similarly distributed within the ejecta. Milisavljevic et al. (2010) recently questioned the torus-like geometry of the ejecta as an explanation of the observed double-peaked [O i] λλ6300, 6364 emission through a reexamination of the line profiles. In our analysis, we compare the profiles of Mg i] and [O i] in the six spectra of SN 2007gr obtained from t = +78 days through +430 days (Figure 10), with an attempt to put some constraints on the ejecta geometry of SN 2007gr.

At t = +78 days, [O i] displayed a nearly symmetric double-peaked structure, with the peaks centered at 6260 Å and 6300 Å. The feature became asymmetric in the t = +129 day spectrum and thereafter, but the gradually decreasing intensity ratio of the blue peak to the red peak is inconsistent with the expected evolution of the intensity ratio of [O i] λ6300 and λ6364. Moreover, we notice that the wavelength difference between these two peaks is 40 Å, smaller than the true separation of the [O i] λλ6300, 6364 doublet. These results indicate that the emission peaks seen at 6300 Å and 6260 Å are unlikely to be from the two distinct components of the [O i] λλ6300, 6364 doublet. Attributing the blue peak seen at 6260 Å to blueshifted [O i] λ6300 yields a velocity shift of about 1800 km s⁻¹. As can be seen in Figure 12(a), the multiple emission peaks do not show any noticeable shift with time in velocity space. Moreover, the relative strength of the blueshifted peak (marked by the dotted lines) decreases gradually relative to the emission peak near zero velocity (marked by the dashed lines).

Zoom In Zoom Out Reset image size

Close inspection of Figure 12(a) reveals that there are two other minor emission features with central wavelengths at 6320 Å and 6360 Å. We suggest that these are likely produced by blueshifted and rest-frame [O i] λ6364, as the velocity shift inferred from the blue peak is ∼1800 km s⁻¹, consistent with the value derived from the [O i] λ6300 line. To confirm our identifications, we examine the velocity structure of the Mg i] λ4571 line in Figure 12(b). As can be seen, the Mg i] λ4571 line also displays a double-peaked feature in SN 2007gr, with the blue-side peak being shifted by about ∼1800 km s⁻¹ with respect to the rest-frame (or the red-side) component. Figure 12(c) shows the comparison of the [O i] λλ6300, 6364 doublet with the scaled Mg i] λ4571 line profile in the t = +430 day spectrum.

The Mg i] line shows a stronger blueshifted peak in the early nebular phase, and the rest-frame peak appears in the later spectra (after t = 144 days). In comparison, the [O i] λ6300 line exhibits both the rest-frame and blueshifted components during the nebular phase, and the blueshifted peak gradually weakens with time. This feature can be understood in light of torus-like ejecta produced by an aspherical explosion. This asphericity was also suggested by the spectropolarimetric observation of SN 2007gr (Tanaka et al. 2008). The blueshifted peak, which arises in the ejecta closer to the observer, is present in all of the nebular spectra. The emission from the rear side of the SN is partially scattered and/or absorbed by the ejecta, and it gains strength at late phases when the ejecta become optically thin.

We note that the flux ratio of the blueshifted to the rest-frame peaks is different for the Mg i] and [O i] lines. The strong blueshifted peak in the Mg i] line leads to the mismatch between the [O i] and Mg i] profiles in the early nebular spectra (see also the mismatch shown by H09). Even in the t = +430 day spectrum, the blueshifted peak of the Mg i] line is slightly stronger than that of the [O i] line (see Figure 12(c)). This difference suggests that oxygen and magnesium have different distributions within the ejecta. More magnesium is located in the outer part, while most of the oxygen is located on the rear side of the ejecta.

At t = +430 days, the flux ratio of [O i] λ6300 to [O i] λ6364 Å is estimated to be 3:1. According to theoretical calculations (Spyromilio 1991; Williams 1994), this ratio is about 1:1 when the ejecta are optically thick at early times. Examining the spectra at t = 129 days, we find that the flux ratio of the [O i] doublet is also close to 3:1. This suggests that SN 2007gr already became optically thin in the early nebular phase, and hence the explosion ejected a small mass of material. The small ejecta mass, together with the low observed expansion velocities (hence low kinetic energy per unit mass), imply a low-energy explosion for SN 2007gr.