ASASSN-15lh: A highly super-luminous supernova

See allHide authors and affiliations

Science  15 Jan 2016:
Vol. 351, Issue 6270, pp. 257-260
DOI: 10.1126/science.aac9613

The most luminous supernova to date

Supernovae are exploding stars at the end of their lives, providing an input of heavy elements and energy into galaxies. Some types have near-identical peak brightness, but in recent years a new class of superluminous supernovae has been found. Dong et al.y report the discovery of ASASSN-15lh (SN 2015L), the most luminous supernova yet found by some margin. It appears to originate in a large quiescent galaxy, in contrast to most super-luminous supernovae, which typically come from star-forming dwarf galaxies. The discovery will provide constraints on models of superluminous supernovae and how they affect their host galaxies.

Science, this issue p. 257


We report the discovery of ASASSN-15lh (SN 2015L), which we interpret as the most luminous supernova yet found. At redshift z = 0.2326, ASASSN-15lh reached an absolute magnitude of Mu,AB = –23.5 ± 0.1 and bolometric luminosity Lbol = (2.2 ± 0.2) × 1045 ergs s–1, which is more than twice as luminous as any previously known supernova. It has several major features characteristic of the hydrogen-poor super-luminous supernovae (SLSNe-I), whose energy sources and progenitors are currently poorly understood. In contrast to most previously known SLSNe-I that reside in star-forming dwarf galaxies, ASASSN-15lh appears to be hosted by a luminous galaxy (MK ≈ –25.5) with little star formation. In the 4 months since first detection, ASASSN-15lh radiated (1.1 ± 0.2) × 1052 ergs, challenging the magnetar model for its engine.

Only within the past two decades has the most luminous class of supernovae (super-luminous supernovae, SLSNe) been identified (1). Compared with the most commonly discovered SNe (Type Ia), SLSNe are more luminous by over two magnitudes at peak and rarer by at least 3 orders of magnitude (2). Like normal SNe, SLSNe are classified by their spectra as either SLSN-I (hydrogen-poor) or SLSN-II (hydrogen-rich). Yet, the physical characteristics of SLSNe may not be simple extensions from their low-luminosity counterparts (1). In particular, the power source for SLSNe-I is poorly understood (3). Adding to the puzzle, SLSNe tend to explode in low-luminosity, star-forming dwarf galaxies (46). The recent advent of wide-area, untargeted transient surveys has made the systematic discovery and investigation of the SLSNe population possible [(7, 8) and references therein].

The All-Sky Automated Survey for SuperNovae [ASAS-SN; (9)] scans the visible sky every two to three nights to depths of V ≈ 16.5 to 17.3 mag using a global network of 14-cm telescopes (9) in an untargeted search for new transients, particularly bright supernovae.

On 14 June 2015 (universal time dates are used throughout this paper), ASAS-SN triggered on a new source located at RA = 22h02m15s.45 Dec = –61°39′34″.6 (J2000), coinciding with a galaxy of then unknown redshift, APMUKS(BJ) B215839.70–615403.9 (10). Upon confirmation with our follow-up telescopes, we designated this new source ASASSN-15lh and published its coordinates (11).

By combining multiple epochs of ASAS-SN images, we extended the detections to fainter fluxes, finding prediscovery images of ASASSN-15lh from 8 May 2015 (V = 17.39 ± 0.23 mag), and the light curve through 19 September 2015 is shown in Fig. 1. The ASAS-SN light curve peaked at V = 16.9 ± 0.1 on approximately tpeak ~ JD2457179 (2015 June 05) based on a parabolic fit to the lightcurve (Fig. 1, dashed line). Follow-up images were taken with the Las Cumbres Observatory Global Telescope Network (LCOGT) 1-m telescopes, and the BV light-curves with the galaxy contribution subtracted are also shown.

Fig. 1 Multi-band light curve of ASASSN-15lh.

The V-band ASAS-SN light curve is shown as black solid dots, and upper limits are represented by black arrows. Swift and LCOGT 1-m data are shown as open circles.

We obtained an optical spectrum (3700 to 9200 Å) of ASASSN-15lh on 21 June 2015 with the du Pont 100-inch telescope. The steep spectral slope with relatively high blue flux motivated Swift UltraViolet and Optical Telescope (UVOT)/X-Ray Telescope (XRT) (12) target-of-opportunity observations starting on 24 June 2015. The six-band Swift light curve spanning from the ultraviolet (UV) to the optical (1928 to 5468 Å) is shown in Fig. 1. The Swift spectral energy distribution (SED), peaking in the UV, indicates that the source has a high temperature. We derive a 3σ x-ray flux limit of <1.6 × 10−14 ergs s–1 cm–2 (0.3 to 10 keV) from a total XRT exposure of 81 ks taken between 24 June and 18 September 2015.

The du Pont spectrum is mostly featureless (Fig. 2A, first from the top), except for a deep, broad absorption trough near ~5100 Å (observer frame). SNID (13), a commonly used SN classification software that has a spectral library of most types of supernovae except SLSN, failed to find a good SN match. However, we noticed a resemblance between the trough and a feature attributed to O II absorption near Embedded Image Å (rest frame) in the spectrum of PTF10cwr/SN 2010gx, a SLSN-I at z = 0.230 (3, 14, 15). Assuming that the ASASSN-15lh absorption trough (full width at half maximum of ~104 km s–1) was also due to the same feature indicated a similar redshift of z ~ 0.23. An optical spectrum (3250 to 6150 Å) obtained on the Southern African Large Telescope (SALT) revealed a clear Mg II absorption doublet (λλ2796, 2803) at z = 0.232, confirming the redshift expected from our tentative line identification. Subsequent Magellan/Clay (6 July) and SALT (7 July) spectra refined the redshift to z = 0.2326 (Fig. 2, C and D). The available rest frame spectra show continua with steep spectral slope, relatively high blue fluxes, and several broad absorption features also seen in PTF10cwr/SN 2010gx (Fig. 2A, features “a,” “b,” and “c”) and without hydrogen or helium features, which is consistent with the main spectral features of SLSNe-I (1, 3). The broad absorption feature near 4400 Å (Fig. 2, “d”) seen in PTF10cwr/SN 2010gx is not present in ASASSN-15lh. ASASSN-15lh thus has some distinct spectral characteristics in comparison with PTF10cwr/SN 2010gx and some other SLSNe-I (3).

Fig. 2 Rest-frame spectra of ASASSN-15lh (black) compared with SLSN-I PTF10cwr/SN 2010gx (red).

(A) The spectra are offset for clarity, labeled with phases and telescopes, and ranked by descending TBB (given on the right) from the top. The ASASSN-15lh spectra are blue and featureless, except for broad absorption features labeled “a,” “b,” and “c” (marked in blue), which match those of PTF10cwr/SN 2010gx at similar TBB. Absorption features “a” at ~4100 Å and “d” at ~4400 Å (marked in red) in PTF10cwr/SN 2010gx are commonly attributed to O II (3, 15). The ~4400 Å feature is not present in ASASSN-15lh. (B) Close-ups of the 4100 Å features, whose evolution in shape, depth, and velocity as a function of TBB is similar for both supernovae. (C and D) The ASASSN-15lh host redshift (z = 0.2326) is determined from the Mg II doublets seen in the SALT and Clay MagE spectra, with EW 0.55 ± 0.05 and 0.49 ± 0.05 Å in (C) and (D), respectively.

Using a luminosity distance of 1171 Mpc (standard Planck cosmology at z = 0.2326), Galactic extinction of E(BV) = 0.03 mag (16), assuming no host extinction (thus, the luminosity derived is likely a lower limit), and fitting the Swift and LCOGT flux measurements to a simple blackbody (BB) model, we obtain declining rest-frame temperatures of TBB from 2.1 × 104 to 1.3 × 104 K and bolometric luminosities of Lbol = 2.2 × 1045 to 0.4 × 1045 ergs s–1 at rest-frame phases relative to the peak of trest ~ 15 and ~50 days, respectively (Fig. 3). ASASSN-15lh’s bolometric magnitude declines at a best-fit linear rate of 0.048 mag day–1, which is practically identical to SLSN-I iPTF13ajg (17) at 0.049 mag day–1 during similar phases (~10 to ~50 days). Subsequently, the luminosity and temperature reach a “plateau” phase with slow changes, and a similar trend is also seen for iPTF13ajg though with sparser coverage. Overall, the temperature and luminosity time evolution resemble iPTF13ajg, but ASASSN-15lh has a systematically higher temperature at similar phases. The estimated BB radius of ~5 × 1015 cm near the peak is similar to those derived for other SLSNe-I (3, 17). These similarities in the evolution of key properties support the argument that ASASSN-15lh is a member of the SLSN-I class, but with extreme properties.

Fig. 3 Time evolution of blackbody temperatures, radii and bolometric luminosities for ASASSN-15lh (black) and SLSN-I iPTF13ajg (red).

Solid black dots show estimates derived from the full UV and optical bands, whereas the open circles show those from optical only. For trest < 10 days, only V-band is available, and the temperatures are estimated on the basis of linear extrapolation from MJD = 57191 – 57241.

The absolute magnitudes (AB) in the rest-frame u-band are shown in Fig. 4. Using either TBB or the spectra, there is little K-correction (18) in converting from B-band to rest-frame uAB with Embedded Image. The solid red points at trest ≳ 10 days include B-band data. Before ~10 days, we lack measurements in blue bands. To estimate Mu,AB at these earlier epochs, we assumed the BV = –0.3 mag color and K-corrections found for the later epochs with Swift photometry. We estimate an integrated bolometric luminosity radiated of ~(1.1 ± 0.2) × 1052 ergs over 108 days in the rest frame. Although our estimates at trest ≲ 10 days should be treated with caution, we can securely conclude that the peak Mu,AB is at or brighter than –23.5 ± 0.1, with a bolometric luminosity at or greater than (2.2 ± 0.2) × 1045 ergs s–1. Both values are without precedent for any supernova recorded in the literature. In Fig. 4, we compare ASASSN-15lh with a sample of SLSNe-I (3, 17). Although its spectra resemble the SLSNe-I subclass, ASASSN-15lh stands out from the luminosity distribution of known SLSNe-I, whose luminosities are narrowly distributed around M ~ –21.7 (2, 19). In table S1, we list the peak luminosities of the five most luminous SNe discovered to date, including both SLSN-I and SLSN-II. The spectral correspondence and similarities in temperature, luminosity, and radius evolutions between ASASSN-15lh and some SLSNe-I lead to the conclusion that ASASSN-15lh is the most luminous supernova yet discovered. Even though we find that SLSN-I is the most plausible classification of ASASSN-15lh, it is important to consider other interpretations given its distinct properties. We discuss alternative physical interpretations of ASASSN-15lh in the supplementary text, and given all the currently available data, we conclude that it is most likely a supernova, albeit an extreme one.

Fig. 4 Rest-frame absolute magnitude light curve of ASASSN-15lh near peak compared with other SLSNe-I.

Estimates of Mu,AB for ASASSN-15lh at trest ≳ 10 days are derived from B-band fluxes, which are subject to small K-corrections, whereas the less reliable Mu,AB estimates are based on V-band only for trest ≲ 10 days. The comparison sample (3, 17) includes the most luminous SLSNe-I previously known. At Mu,AB = –23.5, ASASSN-15lh stands out from the SLSNe-I luminosity distribution (2, 19). Its peak bolometric absolute magnitude is more than ~1 mag more luminous than any other SLSN-I.

The rate of events with similar luminosities to ASASSN-15lh is uncertain. On the basis of a simple model of transient light curves in ASAS-SN observations tuned to reproduce the magnitude distribution of ASAS-SN Type Ia supernovae (supplementary text), the discovery of one ASASSN-15lh–like event implies a rate of r ≃ 0.6 Gpc–3 yr–1 (90% confidence: 0.21 < r < 2.8). This is at least 2 times and can be as much as 100 times smaller than the overall rate of SLSNe-I, r ≃ 32 Gpc–3 yr–1 (90% confidence: 6 < r < 109) from (2), and suggests a steeply falling luminosity function for such supernovae.

For a redshift of z = 0.2326, the host galaxy of ASASSN-15lh has MK ≈ –25.5, which is much more luminous than the Milky Way. We estimate an effective radius for the galaxy of 2.4 ± 0.3 kpc and a stellar mass of M* ≈ 2 × 1011 M. This is in contrast to the host galaxies of other SLSNe, which tend to have much lower M* (46). However, given the currently available data, we cannot rule out the possibility that the host is a dwarf satellite galaxy seen in projection. The lack of narrow hydrogen and oxygen emission lines from the galaxy superimposed in the supernova spectra implies little star formation (SFR) < 0.3 M yr–1 by applying the conversions in (20). Las Cumbres Observatory Global Telescope (LCOGT) astrometry places ASASSN-15lh within 0″.2 (750 pc) of the center of the nominal host. A detailed discussion of the host properties is provided in the supplementary text.

The power source for ASASSN-15lh is unknown. Traditional mechanisms invoked for normal SNe likely cannot explain SLSNe-I (3). The lack of hydrogen or helium suggests that shock interactions with hydrogen-rich circumstellar material, invoked to interpret some SLSNe, cannot explain SLSNe-I or ASASSN-15lh. SLSN-I post-peak decline rates appear too fast to be explained by the radioactive decay of 56Ni (3)—the energy source for Type Ia supernovae. Both the decline rate of the late-time light curve and the integral method (21) will allow tests of whether ASASSN-15lh is powered by 56Ni, and we estimate that ≳30 M of 56Ni would be required to produce ASASSN-15lh’s peak luminosity. Another possibility is that the spindown of a rapidly rotating, highly magnetic neutron star (a magnetar) powers the extraordinary emission (2224). To match the peak Lbol and time scale of ASASSN-15lh, the light-curve models of (23) imply a magnetar spin period and magnetic field strength of P ≃ 1 ms and B ≃ 1014 G, respectively, assuming that all of the spindown power is thermalized in the stellar envelope. If efficient thermalization continues, this model predicts a Lbolt–2 power-law at late times. The total observed energy radiated so far (1.1 ± 0.2 × 1052 ergs) strains a magnetar interpretation because, for P ≲ 1 ms, gravitational wave radiation should limit the total rotational energy available to Embedded Image ergs (25) and the total radiated energy to a third of Embedded Image, which is ~1052 ergs (23).

The extreme luminosity of ASASSN-15lh opens up the possibility of observing such supernovae in the early universe. An event similar to ASASSN-15lh could be observed with the Hubble Space Telescope out to z ~ 6, and with the James Webb Space Telescope out to z ≳ 10 (19). A well-observed local counterpart will be critical in making sense of future observations of the transient high-redshift universe.

Supplementary Materials

Materials and Methods

Supplementary Text

Figures S1 to S5

Tables S1 to S6

References (2765)

References and Notes

Acknowledgments: We acknowledge B. Zhang, L. Ho, A. Gal-Yam, and B. Katz for comments; NSF AST-1515927, OSU CCAPP, Mt. Cuba Astronomical Foundation, TAP, SAO, CAS grant XDB09000000 (S.D.); NASA Hubble Fellowship (B.J.S.); FONDECYT grant 1151445, MAS project IC120009 (J.L.P.); NSF CAREER award AST-0847157 (S.W.J.); U.S. Department of Energy (DOE) DE-FG02-97ER25308 (T.W.-S.H.); NSF PHY-1404311 (J.F.B.); D. Victor for donating equipment (BN); FONDECYT postdoctoral fellowship 3140326 (F.O.E.), and Los Alamos National Laboratory Laboratory Directed Research and Development program (P.R.W). B.J.S. is a Hubble and Carnegie-Princeton Fellow. All data used in this paper are made public, including the photometric data (tables S1 to S6) and spectroscopic data at public repository WISeREP (26) (
View Abstract

Stay Connected to Science

Navigate This Article