HUBBLE SPACE TELESCOPE OBSERVATIONS OF THE AFTERGLOW, SUPERNOVA, AND HOST GALAXY ASSOCIATED WITH THE EXTREMELY BRIGHT GRB 130427A

We observed the location of GRB 130427A with HST on 2013 May 20, 23 days after the initial burst detection. A second epoch was obtained in 2014 April, almost a year after the initial burst. A log of these observations is shown in Table 1. For a more detailed study of the host galaxy we also utilize a longer (2228 s) WFC3/F606W observation obtained on 2014 May 15. The imaging data were reduced in the standard fashion; with on-the-fly processed data retrieved from the archive and subsequently redrizzled using astrodrizzle, for UVIS observations we separately corrected for pixel-based charge-transfer inefficiency (CTE; Anderson & ACS Team 2012; Anderson 2014). Photometry was performed in small apertures to minimize any contribution from underlying galaxy light, and maximize signal-to-noise, it was subsequently corrected using standard aperture corrections.¹⁶ We also use direct image subtraction to isolate the afterglow/SNe light at early epochs, this effectively removes the host contribution. These magnitudes may still contain some transient light, but since the second epoch magnitudes are a factor of ∼15–30 lower than observed at early times, this suggests that these epochs can be used for effective subtraction. The resulting photometry is shown in Table 1, while our HST images are shown in Figure 1.

Zoom In Zoom Out Reset image size

We also obtained grism spectroscopy centered at ∼8000 Å with the G800L grism on the Advanced Camera for Surveys (ACS), with a position angle chosen to minimize the contribution from the underlying host galaxy (see Figure 1). Again, we utilized the on-the-fly calibrated images, corrected for CTE and bias striping. We detected sources on a single F606W image and extracted these via aXe, subsequently drizzling each of the four exposures to create master spectra, which was flux calibrated using published sensitivity curves. We extracted the light from the GRB counterpart in a relatively small aperture (∼2 × FWHM). In principle, a given pixel in the grism image may be exposed to light of multiple different wavelengths from different spatial locations on the chip. For the second epoch of observations, we force an extraction of the same width at the position of the transient (as determined by a map between the first and second epoch of direct imaging). We then subtract this from the initial spectrum to obtain a host free spectrum. Since we have utilized a tight aperture around the SNe, we subsequently scale this subtracted spectrum to the host subtracted magnitudes of the afterglow/SNe.

The afterglow is offset (0.83 ± 0.03)'' from the centroid of the host galaxy light in F606W, and thus the latter is of little concern in our small (01) apertures. However, regions underlying GRBs are frequently among the most luminous parts of their hosts (Fruchter et al. 2006); therefore, some contamination may be expected. Our late time subtraction removes this from both our broadband photometry and grism spectroscopy. For the high-resolution observations reported here, this contamination is small (at most 6% in the UV and 3% in the optical) for our photometry. However, the contribution is somewhat larger in the grism observations. These observations disperse light not only from the region directly underlying the GRB, but also from other locations in the host (which represent contributions at different wavelengths, overlapping the transient light). In particular, the proximate, bright star-forming region considerably contaminates the SNe (>20% at the red end of our spectrum beyond 9000 Å). This region is also likely to be the dominant host contaminant in ground-based spectroscopy (since the host galaxy is resolved), and has quite different colors from the global host, implying potentially significant systematic errors when the broadband Sloan Digital Sky Survey (SDSS) colors of the host are used to attempt a subtraction (e.g., Xu et al. 2013; Melandri et al. 2014). Here, we can directly remove this contribution via the subtraction of the deep second epoch.

SNe Ib/c are generally weak UV emitters due to the strong metal line blanketing shortward of ∼3000 Å; therefore, to first order our F336W observations should be free of SN contribution. This is consistent with the UV colors of GRB 060218/SN 2006aj (from Šimon et al. 2010) and of XRF100316D/SN 2010bh (from Cano et al. 2011), which would predict a factor >10 decrease in flux between F606W and F336W. There are few UV spectra of SN Ic; however, if we graft the STIS UV observations of SN 2002ap onto the optical spectra of SN 1998bw following Levan et al. 2005, then we can obtain a first-order approximation of the likely UV spectral shape (see Figure 2).

Zoom In Zoom Out Reset image size

These colors and spectra suggest that it is reasonable to assume that the F336W light is dominated by the afterglow component. To confirm this, we utilize the UV to IR light curve from Perley et al. (2014). This exhibits a spectral slope of F_ν∝ν^β with β ≈ −0.92 and predicts U(AB) = 23.41 ± 0.10 at the time of the first epoch of HST F336W observations. The corresponding HST UV magnitude is F336W(AB) = 23.28 ± 0.02. Corrected for foreground absorption, this is consistent with the afterglow contributing 90% ± 10% of the measured F336W flux, confirming our assumptions above. This afterglow model predicts F160W(AB) = 21.84, somewhat fainter than the measured magnitude and suggesting that the SN makes up ∼20% of the light in this band, again in keeping with expectations of the few IR spectra of broad-lined SNe Ic obtained to-date (e.g., Bufano et al. 2012).

We estimate the SN contribution by using the model above, with initial error bars accounting for the uncertainty in the F336W afterglow light discussed above and the intrinsic value of β, which we adopt to be 0.92 ± 0.1. For this range of models, we then subtract the afterglow spectrum from the measured grism data and neglect any host galaxy contamination. The extremum of this model is set by the assumption that both F336W and F160W are entirely dominated by afterglow, and by subtracting the resulting power-law index.

Our grism spectrum, both before and after subtraction of the afterglow and host light, is shown in Figure 2 (top and bottom). Broad features, consistent with those seen in other high velocity SN Ic associated with GRBs are clearly visible in the spectrum, even before the subtraction of the afterglow component. The absence of broad emission at Hα or He absorption rules out type II or Ib events, respectively. In the lower panel of Figure 2, we plot rest-frame wavelength versus luminosity comparisons of the afterglow subtracted and dereddened¹⁷ spectra of SN 2013cq with various GRB/SNe pairs.

The closest match for the overall spectral shape and luminosity is that of SN 1998bw (Galama et al. 1998; Iwamoto et al. 1998; Patat et al. 2001). The similarity in appearance of these SNe is primarily due to the overall spectral shape, with a drop in luminosity of a factor of ∼3 over the 5000–7000 Å range, substantially more than seen in other GRB/SNe pairs.

The broad colors of these SNe are similar, and if the kink at ∼6000 Å is interpreted as the Si ii (6355 Å) blend, then it would be indicative of a photospheric velocity similar to SN 1998bw at the same epoch (v_ph ∼ 15, 000 km s⁻¹), though we note that this feature is apparently stronger in SN 1998bw, where there is marked upturn in flux redward of it. This may suggest a somewhat higher velocity for SN 2013cq. Taken at face value, this would suggest that SN 2013cq is broadly similar in peak luminosity, ⁵⁶Ni production and kinetic energy to SN 1998bw.

While the similarity to SN 1998bw is very marked over the rest-frame spectral range from 5100–7000  Å it is much poorer around the central peak of an SN Ic at ∼5000 Å. In SN 2013cq this feature appears to be broader and blueshifted relative to SN 1998bw, as observed by Xu et al. (2013). Our comparison suggests that this peak may fit better with the slightly higher velocity SN 2003dh (Hjorth et al. 2003; Mazzali et al. 2003), the only one of the comparison SNe to arise from a luminous cosmological GRB. In this case, the SN 2003dh model shown in Figure 2 remains a factor of 50% less luminous than SN 2013cq, and is systematically redder (i.e., a smaller decrement between the peak and ∼6000 Å). However, the positions of the broadened lines do then appear generally similar.

Alternatively, this could be suggestive of a much higher velocity SN such as SN 2010bh (Bufano et al. 2012; Cano et al. 2011) as favored by Xu et al. (2013), who infer v_ph = 32, 000 km s⁻¹ from the Fe ii (5169 Å) at an epoch of 12.5 rest-frame days. This does provide a good match to the location and width of the feature; however, the overall spectral shape and luminosity of SN 2010bh are also very different from SN 2013cq. At 17 days, SN 2013cq appears to be much bluer and a factor >3 times more luminous than SN 2010bh, suggesting that it is not as good of an analog as SN 1998bw. To obtain a match in both luminosity and general spectral shape would require that the reddening in the direction of SN 2010bh had been significantly underestimated. Our plotted spectrum assumes E(B − V)_gal = 0.12 and E(B − V)_int = 0.14 (Bufano et al. 2012). To obtain a match would require dereddening the spectra with an additional E(B − V) = 0.4–0.5, well beyond the values allowed from the Na i D doublet observed in moderate resolution X-shooter spectroscopy (Bufano et al. 2012) and even the largest extinctions allowed by the afterglow spectrum (Olivares E. et al. 2012). Hence, we disfavor the suggestion that the underlying SNe is similar to SN 2010bh, due to the significant disparity in the luminosity. Instead, we favor an SNe similar in ⁵⁶Ni yield and kinetic energy to SN 1998bw and SN 2003dh, in which possible velocity structure within the ejecta gives rise to a range of features within the spectra that do not provide a unique match to any previous GRB/SNe pair. Indeed, the recent study of Melandri et al. (2014) also identifies a source that is much brighter than SN 2010bh, but spectrally intermediate between SN 1998bw and SN 2010bh.

Despite these differences, it is clear that the features observed in the spectrum of SN 2013cq are broadly compatible to the range observed in SN/GRB pairs, though a full model description of the spectrum is beyond the scope of this paper. To obtain approximate bounds on the luminosity of the SN, we integrate our afterglow and host subtracted spectra (allowing for the errors in the afterglow parameters) through a rest-frame V-band filter to obtain a luminosity relative to SN 1998bw. This suggests that SN 2013cq has a luminosity factor k = L_V(2013cq)/L_V(1998bw) of 0.77 ± 0.10. Given the suggestions that SN 2013cq is more rapidly evolving than SN 1998bw (Xu et al. 2013), these observations (at the V-band peak of SN 1998bw) may underestimate by true peak luminosity, which may be even closer to that of SN 1998bw.

GRB 130427A is unusual as a luminous GRB at a redshift at which the SNe are open to detailed study. Indeed, it is the highest luminosity burst for which there is spectroscopic evidence for an SN. In this regard, the similarity of the SN to that seen in a burst (GRB 980425) that was six orders of magnitude less energetic is striking. As we show in Figure 3, there is no correlation between the GRB energetics and the inferred peak magnitudes of their SNe. These similarities in SNe peak luminosity, and in their spectra, suggest similarities in the ejected ⁵⁶Ni masses and kinetic energies. Recent modeling (Mazzali et al. 2013) has used these diagnostics to suggest that most broad-lined SN Ic associated with low-luminosity GRBs in turn arise from a relatively small range of ZAMS progenitor masses, perhaps between 30 and 50 M_☉. The detection of a similar SN in a highly luminous GRB extends this across a broader range of energy.

Zoom In Zoom Out Reset image size

Our HST observations clearly resolve the afterglow from the host galaxy, showing the GRB to lie at a spatial offset of 083 (4.0 kpc at z = 0.34) from the center of its host. This is a relatively large offset for a GRB from its host galaxy (>75% of those in Bloom et al. 2002) though by no means exceptional (see Figure 3). In the F606W observations, the galaxy exhibits a bar-like structure with a face-on disk visible beyond this, a weak spiral structure is also apparent, making the host one of few to be classified as a spiral (Svensson et al. 2010). The F336W imaging shows weak star formation close to the center of the galaxy, but the most striking feature is a strong star-forming region at a similar radial offset to the GRB, but offset from the GRB position by 03 (1.5 kpc). This region is the strongest region of star formation in the host galaxy. While the magnitude of star formation underlying the GRB position remains uncertain because of contamination with the late-time afterglow, the offset region is at least a factor of ∼2 brighter. It also shows readily detectable O iii emission in the late-time grism spectrum. The weak spiral arm in the direction of the GRB also appears distorted, and so it may be that star formation has been triggered via a tidal interaction. In the larger field around GRB 130427A, we note several galaxies of similar magnitude (see Figure 1). In our grism spectroscopy, these galaxies do not exhibit strong spectral lines, but they may well imply that the host of GRB 130427A lies within an association.

The magnitude of the galaxy in a large aperture at late times after the afterglow/SNe has significantly faded is F336W(AB) = 22.84 ± 0.07 at a rest-frame wavelength of approximately 2500 Å. This corresponds to a UV star-formation rate of ∼1.1 ± 0.1 M_☉ yr⁻¹ (error statistical only), broadly in agreement with that inferred from the SDSS observed spectral energy distribution of the galaxy of $2^{+5}_{-1}$ M_☉ yr⁻¹ (Xu et al. 2013) and suggesting a relatively (but not extremely) low specific star formation rate by the standards of GRB hosts (Svensson et al. 2010). The compact star-forming region close to the GRB contains ∼10% of this star formation, making a highly luminous star forming region comparable to that seen in the host of GRB 980425/SN 1998bw (Hammer et al. 2006; Christensen et al. 2008) or in a handful of other local galaxies (Beck et al. 1996). At late times, UV emission is visible close to the GRB position, indicating that lower level star formation is likely arising proximate to the GRB. This may still contain some afterglow light, but this region is at least a factor >1.5 fainter than the brightest star-forming region in the host.

Some previously extremely energetic bursts (e.g., GRB 080319B; Tanvir et al. 2010) have shown extremely faint and small host galaxies, while the hosts of GRBs 980425, 030329, and 060218 are also sub-luminous and LMC-like. However, as show in Figure 3 the overall population of GRB–SN host luminosities shows no discernible correlation with the energy of the GRB (see also Levesque et al. 2010), and the host of GRB 130427A is in keeping with these expectations. It is intriguing to note that the GRB host galaxy with the closest properties to that of GRB 130427A/SN 2013cq is in fact the host of SN 1998bw. It is also a rare example of a spiral galaxy, in which the GRB occurs at a moderately large offset from the nucleus and from the strongest region of star formation within the host galaxy. This is shown graphically in Figure 4. Given the similarities in SN and environment, GRB 130427A would seem like a close analog of GRB 980425A if it were not for the factor of 10⁶ difference in their γ-ray energy releases.

Zoom In Zoom Out Reset image size