# How is the first detonation in Supernove type Ia triggered?

Ok, I read about the Supernova of type Ia and I found out that there are two detonations happening. First one is in helium shell around the white dwarf and second one seems to be triggered by the first one and it is the main part of whole supernova. I did not find how this helium shell explodes at all so that's what I am looking for. Thanks in advance!

Nobody really knows how type Ia supernovae detonate (or deflagrate) - there are a number of possibilities. The "vanilla" possibility is not what you state in your question, it is that the white dwarf accretes sufficient mass that it approaches the Chandrasekhar limit and becomes dense enough in its core to commence carbon burning.

However, the emerging diversity that is seen among type Ia supernovae, once thought to be a single population, suggests there may be other possibilities. There is some evidence that white dwarfs may explode at masses well below the Chandrasekhar limit. If a white dwarf in a binary accretes enough He-rich matter, this can become compressed enough to ignite He burning near the surface (this happens at a lower density threshold than Carbon burning). This then drives a shock wave into the white dwarf and the compression caused by this can ignite the carbon.

Why does the He "explode"? Well, the accreted He will form an electron-degenerate layer at the surface. A fundamental property of this degenerate gas is that the pressure is independent of the temperature. Thus, if the He ignites then at least initially, the temperature goes up but the pressure does not. Since the He fusion rate depends on something like $T^{40}$ this allows a runaway reaction to develop that could be characterised as an "explosion" in the surface layers.

## First Simulation of Type Ia Supernova

How can scientists use violent stellar explosions occurring outside the Milky Way Galaxy to determine the characteristics of the universe? Aren’t the many variations in these explosions so extreme as to make them useless for studying? After all, we can’t take these objects into the lab (or make them) in order to test the details of how they work.

While the case for using Type Ia supernovae to measure cosmic expansion may seem hopeless, astrophysicists have developed a suite of tools to address the above problems. For example, the conditions specific to each supernova cause them to vary in brightness. As described in a previous TNRTB, astronomers can develop models that match observations and allow them to correct for those variations.

Another approach is to build these supernovae in the lab—or at least in a powerful computer program. A team of American scientists recently demonstrated the first fully three-dimensional computer simulation of the stages preceding the supernova explosion. Type Ia supernovae occur under almost (but not quite) identical conditions. Astronomers can use these detailed computer simulations to test how various conditions impact a supernova’s brightness.

The results of this simulation confirm previous outcomes of earlier, less sophisticated attempts. Additionally, the new work demonstrates that the behavior inside the white dwarf just before detonation is more complex than expected.

While much work remains, this research represents a significant step in confirming the “standard candle” signature of Type Ia supernovae that make them useful probes of cosmic expansion. Continued improvements in these computer simulations will allow astronomers to better understand how different environments affect the brightness of Type Ia supernovae. This understanding will lead to better measurements of the cosmic expansion, and, RTB predicts, more evidence revealing God’s design of this universe.

## Finding Just the Right Type of Detonation

Type Ia supernovae are a cornerstone of extragalactic distance measurements, so it’s important that we understand them very well. Right now, we’re fairly certain that Type Ia supernovae are the result of white dwarfs exploding. But how they explode is still an open question.

#### To Explode a White Dwarf

A creation mechanism for Type Ia supernovae where a white dwarf accretes mass from a companion (upper panel) till it explodes as a supernova (lower panel). [NASA/CXC/M. Weiss]

So how do you make something like a white dwarf explode? Just add mass! If a white dwarf accretes enough matter from a nearby companion, it can approach the Chandrasekhar limit of 1.4 solar masses and explode. This process seems fairly straightforward, but it turns out there are several potential ways to explode a white dwarf.

One scenario involves “double detonation”, where the helium shell of a white dwarf detonates and causes the carbon core to detonate in turn. Another scenario considers white dwarfs in a binary, with one white dwarf accreting material and exploding to knock the other one away.

Interestingly, observations suggest that a combination of these two scenarios — double detonation in white dwarf binaries — may be the likely progenitor of many Type Ia supernovae. One important constraint in this model is that the exploding white dwarf’s mass remains just below the Chandrasekhar limit.

With this in mind, a group of researchers led by Ken Shen (University of California, Berkeley) considered sub-Chandrasekhar-mass explosion scenarios with a tricky but realistic assumption: that local thermodynamic equilibrium (LTE) does not hold.

Model and observed supernova light curves in different bands. The solid lines and the colored circles represent the model, while the unfilled shapes are observed supernovae. The model colors correspond to different white dwarf masses. The white dwarf’s carbon/oxygen ratio was assumed to be 50:50. [Shen et al. 2021]

#### Explosions Not in Equilibrium

When a system is in LTE, the energies and ionization levels of particles in the system are in some fixed relation with each other while temperature remains consistent across the system. There are astrophysical scenarios where LTE is a safe assumption, like in stars, but LTE certainly doesn’t hold in an event like a supernova.

To model explosions with non-LTE assumptions, Shen and collabors used two different modeling codes. A major difference between the two codes was computation time, and running the same explosion scenarios through both codes allowed Shen and collaborators to determine if the more time-efficient code could stand up to the other. The model outputs included spectra of the ensuing supernovae as well as their light curves across different filters.

#### Model Matches

A diagram showing the Phillips relation, with peak B-band brightness plotted against the decrease in B-band magnitude 15 days after the peak. Crosses correspond to observations of Type Ia supernovae. The colored shapes correspond to the models, with color representing white dwarf mass, shape representing the carbon/oxygen ratio, and shape outline representing the modeling code used. [Shen et al. 2021]

The model spectra are also good matches to observations, sometimes even out to 30 days after the peak. They are especially accurate near the peak, save for spectral features from “intermediate mass elements”, which generally include elements heavier than carbon up to calcium.

All in all, these initial non-LTE models of sub-Chandrasekhar detonations are an excellent match to a broad range of observed Type Ia supernovae near peak brightness! Future models will have to account for more conditions, but non-LTE seems to be the way to go.

#### Citation

“Non-local Thermodynamic Equilibrium Radiative Transfer Simulations of Sub-Chandrasekhar-mass White Dwarf Detonations,” Ken J. Shen et al 2021 ApJL 909 L18. doi:10.3847/2041-8213/abe69b

## How is the first detonation in Supernove type Ia triggered? - Astronomy

Ji-an Jiang 1 , 2 , Mamoru Doi 1 , 3 , 4 , Keiichi Maeda 5 , 3 , Toshikazu Shigeyama 4 , Ken’ichi Nomoto 3 , Naoki Yasuda 3 , Saurabh W. Jha 6 , Masaomi Tanaka 7 , 3 , Tomoki Morokuma 1 , 3 , Nozomu Tominaga 8 , 3 , Željko Ivezić 9 , Pilar Ruiz-Lapuente 10 , 11 , Maximilian D. Stritzinger 12 , Paolo A. Mazzali 13 , 14 , Christopher Ashall 13 , Jeremy Mould 15 , Dietrich Baade 16 , Nao Suzuki 3 , Andrew J. Connolly 9 , Ferdinando Patat 16 , Lifan Wang 17 , 18 , Peter Yoachim 9 , David Jones 19 , 20 , Hisanori Furusawa 7 , Satoshi Miyazaki 7 , 21

1 Institute of Astronomy, Graduate School of Science, The University of Tokyo, 2-21-1 Osawa, Mitaka, Tokyo 181-0015, Japan

2 Department of Astronomy, Graduate School of Science, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan

3 Kavli Institute for the Physics and Mathematics of the Universe (WPI), The University of Tokyo, 5-1-5 Kashiwanoha, Kashiwa, Chiba 277-8583, Japan

4 Research Center for the Early Universe, Graduate School of Science, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan

5 Department of Astronomy, Kyoto University, Kitashirakawa-Oiwake-cho, Sakyo-ku, Kyoto 606-8502, Japan

6 Department of Physics and Astronomy, Rutgers, The State University of New Jersey, 136 Frelinghuysen Road, Piscataway, NJ 08854, USA

7 National Astronomical Observatory of Japan, 2-21-1 Osawa, Mitaka, Tokyo 181-8588, Japan

8 Department of Physics, Faculty of Science and Engineering, Konan University, 8-9-1 Okamoto, Kobe, Hyogo 658-8501, Japan

9 Department of Astronomy, University of Washington, Box 351580, Seattle, WA 98195-1580, USA

10 Instituto de Física Fundamental, Consejo Superior de Investigaciones Científicas, c/. Serrano 121, E-28006, Madrid, Spain

11 Institut de Ciències del Cosmos (UB-IEEC), c/. Martí i Franqués 1, E-08028, v, Spain

12 Department of Physics and Astronomy, Aarhus University, Ny Munkegade 120, 8000 Aarhus C, Denmark

13 Astrophysics Research Institute, Liverpool John Moores University, IC2, Liverpool Science Park, 146 Brownlow Hill, Liverpool L3 5RF, UK

14 Max-Planck-Institut für Astrophysik, Karl-Schwarzschild-Str. 1, D-85748 Garching, Germany

15 Centre for Astrophysics and Supercomputing, Swinburne University of Technology, Hawthorn, Vic 3122, Australia

16 European Organisation for Astronomical Research in the Southern Hemisphere (ESO), Karl-Schwarzschild-Str. 2, 85748 Garching b. München, Germany

17 George P. and Cynthia Woods Mitchell Institute for Fundamental Physics and Astronomy, Department of Physics and Astronomy, Texas A&M University, 4242 TAMU, College Station, TX 77843, USA

18 Purple Mountain Observatory, Chinese Academy of Sciences, Nanjing 210008, China

19 Instituto de Astrofísica de Canarias, E-38205 La Laguna, Tenerife, Spain

20 Departamento de Astrofísica, Universidad de La Laguna, E-38206 La Laguna, Tenerife, Spain

21 SOKENDAI (The Graduate University for Advanced Studies), Mitaka, Tokyo, 181-8588, Japan

Type Ia supernovae (SNe Ia) arise from the thermonuclear explosion of carbon-oxygen white dwarfs 1 , 2 . Though the uniformity of their light curves makes them powerful cosmological distance indicators 3 , 4 , long-standing issues remain regarding their progenitors and explosion mechanisms 2 , 5 , 6 . Recent detection of the early ultraviolet pulse of a peculiar subluminous SN Ia has been claimed as new evidence for the companion-ejecta interaction through the single-degenerate channel 7 , 8 . Here, we report the discovery of a prominent but red optical flash at ∼ 0.5 days after the explosion of a SN Ia which shows hybrid features of different SN Ia sub-classes: a light curve typical of normal-brightness SNe Ia, but with strong titanium absorptions, commonly seen in the spectra of subluminous ones. We argue that the early flash of such a hybrid SN Ia is different from predictions of previously suggested scenarios such as the companion-ejecta interaction 8 − 10 . Instead it can be naturally explained by a SN explosion triggered by a detonation of a thin helium shell either on a near-Chandrasekhar-mass white dwarf ( ≳ 1.3 M ⊙ ) with low-yield 56 Ni or on a sub-Chandrasekhar-mass white dwarf ( ∼ 1.0 M ⊙ ) merging with a less massive white dwarf. This finding provides compelling evidence that one branch of the previously proposed explosion models, the helium-ignition scenario, does exist in nature, and such a scenario may account for explosions of white dwarfs in a wider mass range in contrast to what was previously supposed 11 − 14 .

A faint optical transient was discovered on UT April 4.345, 2016 through the newly established high-cadence deep-imaging survey which is optimized for finding Type Ia Supernovae (SNe Ia) within a few days after explosion with the Subaru/Hyper Suprime-Cam (HSC) 15 —“the MU lti-band S ubaru S urvey for E arly-phase S Ne Ia” ( MUSSES ). Close attention has been paid to one transient because its brightness increased by ∼ 6.3 times within one day of the first observation. We designated this fast-rising transient as MUSSES1604D (the official designation is SN 2016jhr)—the fourth early-phase SN candidate found in the April 2016 observing run of MUSSES.

Figure 1 presents the observed g -, r -, i -band light curves of MUSSES1604D. The earliest photometry by Subaru/HSC indicates an apparent g -band magnitude of 25.14 ± 0.15 on April 4.345 (MJD 57482.345). One day later, MUSSES1604D brightened rapidly to ∼ 23.1 and 23.0 mag in the g and r bands, respectively. More surprisingly, the g -band observation on April 6 indicates that the transient “paused” brightening from April 5, showing a plateau-like evolution lasting for ∼ 1 day. At the same time, the transient also slowed down in its rate of brightening in the r band.

Follow-up observations indicated that MUSSES1604D is a SN Ia with a r -band peak absolute magnitude of ∼ -19.1 on April 26. Adopting a host galaxy redshift z of 0.11737, the rest-frame light curves in the B - and V -band absolute magnitudes from ∼ 4 days after the first observation are derived by applying a K-correction based on the best-fitting model with SALT2 16 . Because of the peculiar flash at early time, K-correction for the flash-phase light curves is performed by simplified spectral-energy distributions, estimated from the early color information of MUSSES1604D (see Methods). The rest-frame B -band light curve shows a peak absolute magnitude of about -18.8 and Δ m 15 ( B ) ≈ 1.0 mag, indicating a normal-brightness SN Ia 17 .

Color evolution within a few days after a SN explosion is crucial for identifying the early flash 8 , 10 . In contrast to another peculiar early-flash SN Ia iPTF14atg with B − V color evolution obtained only from ∼ 5 days after the discovery 7 , the specific survey strategy of the MUSSES project enables us to obtain the color information of MUSSES1604D from 1 day after the first observation (Figure 2), which shows a slightly red B − V color of about 0.2 mag at first, reddening further to about 0.5 mag in one day.

The interaction of SN ejecta with a non-degenerate companion star 8 , 18 , 19 (“companion-ejecta interaction”, CEI) or with dense circumstellar material 9 , 10 (“CSM-ejecta interaction”) are popular scenarios to explain the early optical flash. In order to produce a prominent optical flash comparable to that of MUSSES1604D, either a companion with a very extended envelope or a large-scale CSM distribution is required. In the CEI scenario, a prominent flash generated from the inner, hot region of ejecta can be observed through the hole that is carved out by a red-giant companion 8 , 19 . In the CSM-ejecta-interaction scenario, a more extended CSM distribution could generate a brighter flash but with longer diffusion time. Our best-fitting CEI model (Figures 2 & 3) and previous simulations of both two scenarios 8 , 10 , 19 all indicate that the particular blue color evolution is inevitable when producing the early flash as bright as that of MUSSES1604D (Extended Data Figure 1), which is incompatible with the red and rapid early color evolution observed for MUSSES1604D.

Peculiar spectral features have been discovered around the peak epoch (Figure 4). At first glance, the Si II λ 6355 line, the W-shaped S II feature, and the Ca II H & K absorptions are reminiscent of a normal SN Ia, while the weak Si II λ 5972 line suggests a higher photospheric temperature than those of SNe Ia with similar luminosities. On the other hand, prominent absorption features such as the Ti II trough around 4150 AA , usually attributed to low temperature, have been found at the same time, in contrast to the brightness indicated by the light curve. By inspecting near-maximum spectra of more than 800 non-subluminous SNe Ia, we found just three MUSSES1604D-like objects—SN 2006bt, SN 2007cq and SN 2012df (Extended Data Figures 2 & 3), indicating the rarity of such hybrid SNe Ia.

The peculiar spectral features and the early flash followed by a normal-brightness light curve observed for MUSSES1604D are incompatible with predictions of classical explosion mechanisms 20 , 21 through the hydrogen-accreting single degenerate channel, but suggested by a specific scenario in which the SN explosion is triggered by the He-shell detonation, so-called the double-detonation (DDet) scenario 12 , 13 , 22 , 23 . In principle, a He-shell detonation not only generates a shock wave propagating toward the center of the white dwarf (WD) and ignites carbon burning near the center, but also allocates 56 Ni and other radioactive isotopes such as 52 Fe and 48 Cr to the outermost layers where the optical depth is relatively low 12 , 23 . Therefore energy deposited by decaying radioactive isotopes diffuses out and consequently results in a prominent flash in the first few days after the explosion (see Methods). Observationally, the plateau-like light curve enhancement can be observed with the day-cadence observations. At the same time, a significant amount of not only iron group elements but also intermediate mass elements such as Ti and Ca will be produced in the outermost layers 12 , 13 , 23 . Vast numbers of absorption lines of these elements are very effective in blocking the flux in the blue part of the optical spectrum, thus leading to a relatively red B − V color evolution in general. Indeed, although a substantial amount of He is left after the detonation, the expected spectrum would not show a trace of He in the optical wavelength 24 . By assuming a progenitor star with a WD mass of 1.03 M ⊙ and a He-shell mass as low as ∼ 0.054 M ⊙ (as required to trigger the He detonation on the surface of a 1.03 M ⊙ WD 12 , 23 ), the prominent early flash, peculiar early color evolution and Ti II trough feature are reproduced simultaneously (Figures 2–4). Early-phase photometric behavior similar to that seen in our simulation has also been independently shown in a simulation of the sub-Chandrasekhar DDet model very recently 25 , validating our simulation and interpretation.

A potential issue in our simulation is the assumption of a sub-Chandrasekhar-mass WD with a thin He-shell. The amount of synthesized 56 Ni is sensitive to the mass of the exploding WD and determines the peak luminosity 12 , 23 . The DDet model requires a sub-Chandrasekhar-mass WD ( ∼ 1 M ⊙ ) for the peak luminosity of MUSSES1604D. However, DDet happening on such a WD would lead to a fast-evolving B -band light curve, which is inconsistent with a much slower-evolving light curve observed for MUSSES1604D. In addition, the early flash resulting from the corresponding He mass of 0.054 M ⊙ is much brighter than that of MUSSES1604D. We suggest two alternative scenarios that also involve He detonation to solve this issue. A He-ignited violent merger 14 can easily trigger a detonation in a thin He shell, and could produce the light curve of MUSSES1604D, but by fine-tuning the configuration of the binary system. Whether core detonation can be triggered by the thin-He-shell detonation through the WD-WD merger is also an open question 26 , 27 . Alternatively, the lower mass He can be detonated on the surface of a near-Chandrasekhar-mass WD, which provides a better and more straightforward account of the light curve and spectral features (Figures 2–4). Further investigation suggests that the best-fitting WD mass is in the range 1.28–1.38 M ⊙ but with a low-yield 56 Ni compared with the prediction by DDet (see Methods). This finding suggests that there could be a mechanism to reduce the mass of 56 Ni in the explosion triggered by the He detonation. For example, the shock wave generated by He detonation may trigger a deflagration rather than a detonation near the centre of the WD 28 , because the high degeneracy pressure of a near-Chandrasekhar-mass WD would inhibit the formation of a shock wave as strong as that seen in a sub-Chandrasekhar-mass WD. Although the observed peculiarities of MUSSES1604D could be naturally explained by this scenario, it is not yet clear how a thin He shell is formed on such a massive WD during binary evolution.

The discovery of MUSSES1604D indicates that the He-detonation-triggered scenario is also promising to explain early-flash SNe Ia in addition to other popular scenarios 8 − 10 . The prominent optical excess and peculiar color evolution in the earliest phase together with absorptions due to Ti II ions in around-maximum spectra can be used as indicators of this scenario. The slow-evolving B -band light curve makes the classical sub-Chandrasekhar DDet model previously supposed 11 , 12 unlikely. Recent work shows that the sub-Chandrasekhar DDet scenario could explain a part of normal SNe Ia if only a negligible amount of He exists at the time of the He-shell detonation 29 , 30 . Given that MUSSES1604D is best explained by a He shell that is thin but still more massive than required in the above scenario, it opens up a possibility that the He-detonation-triggered scenario would produce a range of observational counterparts, controlled by the masses of both the WD and the He shell. The discovery of MUSSES1604D thus provides the first observational calibration about the range and combination of these quantities realized in nature.

1. Filippenko, A. V. Optical Spectra of Supernovae. Ann. Rev. Astron. Astrophys. 35, 309–355 (1997).

2. Maoz, D., Mannucci, F. & Nelemans, G. Observational Clues to the Progenitors of Type Ia Supernovae. Ann. Rev. Astron. Astrophys. 52, 107–170 (2014).

3. Perlmutter, S., et al. Measurements of Ω and Λ from 42 High-Redshift Supernovae. Astrophys. J. 517, 565–586 (1999).

4. Riess, A. G., et al. Observational Evidence from Supernovae for an Accelerating Universe and a Cosmological Constant. Astron. J. 116, 1009–1038 (1998).

5. Hillebrandt, W. & Niemeyer, J. C. Type IA Supernova Explosion Models. Ann. Rev. Astron. Astrophys. 38, 191–230 (2000).

6. Whelan, J. & Iben, I., Jr. Binaries and Supernovae of Type I. Astrophys. J. 186, 1007–1014 (1973).

7. Cao, Y. et al. A strong ultraviolet pulse from a newborn type Ia supernova. Nature. 521, 328–331 (2015).

8. Kasen, D. Seeing the Collision of a Supernova with Its Companion Star. Astrophys. J. 708, 1025–1031 (2010).

9. Levanon, N., Soker, N. & García-Berro, E. Constraining the double-degenerate scenario for Type Ia supernovae from merger ejected matter. Mon. Not. R. Astron. Soc. 447, 2803–2809 (2015).

10. Piro, A. L. & Morozova, V. S. Exploring the Potential Diversity of Early Type Ia Supernova Light Curves. Astrophys. J. 826, 96 (2016).

11. Bildsten, L., Shen, K. J., Weinberg, N. N. & Nelemans, G. Faint Thermonuclear Supernovae from AM Canum Venaticorum Binaries. Astrophys. J. Lett. 662, L95–L98 (2007).

12. Fink, M., et al. Double-detonation sub-Chandrasekhar supernovae: can minimum helium shell masses detonate the core? Astron. & Astrophys. 514, 53 (2010).

13. Woosley, S. E. & Kasen, D. Sub-Chandrasekhar Mass Models for Supernovae. Astrophys. J. Lett. 747, 38 (2011).

14. Pakmor, R., Kromer, M., Taubenberger, S. & Springel, V. Helium-ignited Violent Mergers as a Unified Model for Normal and Rapidly Declining Type Ia Supernovae. Astrophys. J. Lett. 770, L8 (2013).

15. Miyazaki, S., et al. Hyper Suprime-Cam. Proc. SPIE Conf. Ser. 8446, 84460Z-1–84460Z-9 (2012).

16. Guy, J., et al. SALT2: using distant supernovae to improve the use of type Ia supernovae as distance indicators. Astron. & Astrophys. 466, 11–21 (2007).

17. Phillips, M. M. The absolute magnitudes of Type IA supernovae. Astrophys. J. Lett. 413, L105–L108 (1993).

18. Pan, K., Ricker, P. M. & Taam, R. E. Impact of Type Ia Supernova Ejecta on Binary Companions in the Single-degenerate Scenario. Astrophys. J. 750, 151 (2012).

19. Kutsuna, M. & Shigeyama, T. Revealing progenitors of type Ia supernovae from their light curves and spectra. Publ. Astron. Soc. Jap. 67, 54 (2015).

20. Nomoto, K., Thielemann, F. -K. & Yokoi, K. Accreting white dwarf models for type I supern. III. Carbon deflagration supernovae. Astrophys. J. 286, 644–658 (1984).

21. Khokhlov, A. M. Delayed detonation model for type IA supernovae. Astron. & Astrophys. 245, 114–128 (1991).

22. Guillochon, J., Dan, M., Ramirez-Ruiz, E. & Rosswog, S. Surface Detonations in Double Degenerate Binary Systems Triggered by Accretion Stream Instabilities. Astrophys. J. Lett. 709, L64–L69 (2010).

23. Kromer, M., et al. Double-detonation Sub-Chandrasekhar Supernovae: Synthetic Observables for Minimum Helium Shell Mass Models. Astrophys. J. 719, 1067–1082 (2010).

24. Boyle, A., Sim, S. A., Hachinger, S. & Kerzendorf, W. Helium in Double-Detonation Models of Type Ia Supernovae. Astron. & Astrophys. 599, 46 (2017).

25. Noebauer, U. M., et al. Early light curves for Type Ia supernova explosion models. Preprint at (http://arxiv.org/abs/1706.03613) (2017).

26. Shen, K. J. & Bildsten, L. The Ignition of Carbon Detonations via Converging Shock Waves in White Dwarfs. Astrophys. J. 785, 61 (2014).

27. Tanikawa, A., et al. Hydrodynamical Evolution of Merging Carbon-Oxygen White Dwarfs: Their Pre-supernova Structure and Observational Counterparts. Astrophys. J. 807, 40 (2015).

28. Nomoto, K., Sugimoto, D. & Neo, S. Carbon Deflagration Supernova, an Alternative to Carbon Detonation. Astrophys Space Sci. 39, L37–L42 (1976).

29. Blondin, S., Dessart, L., Hillier, D. J. & Khokhlov. A. M. Evidence for sub-Chandrasekhar-mass progenitors of Type Ia supernovae at the faint end of the width-luminosity relation. Mon. Not. R. Astron. Soc. 470, 157–165 (2017).

30. Shen, K. J., Kasen, D., Miles, B. J. & Townsley, D. M. Sub-Chandrasekhar-mass white dwarf detonations revisited. Preprint at (http://arxiv.org/abs/1706.01898) (2017).

Author Contributions J.J. initiated the study, carried out analysis and wrote the manuscript as the principal investigator of the MUSSES project. M.D. contributed to the initiation of the MUSSES project, and assisted with manuscript preparation and analysis together with K.M. and T.S. K.M. and T.S. organized the efforts for theoretical interpretation with J.J. and M.D. K.M. investigated the He-detonation-triggered explosion models and conducted radiation-transfer calculations used to generate simulated light curves and spectra. T.S. developed and ran the radiation-transfer calculations used to generate simulated CEI-induced light curves. K.N. provided insights into the He-detonation-triggered explosion models and assisted with the analysis. N.Y., H.F. and S.M. are core software developers for HSC and are in charge of the HSC Subaru Strategic Program project. N.Y., N.T. and M.T. developed the HSC transient server for selecting real-time supernova candidates and contributed to Subaru/HSC observations and data reduction. T.M. contributed to the Subaru/HSC observation and to target-of-opportunity observations made with the 1.05-m Kiso Schmidt telescope. S.W.J. contributed SALT spectroscopy and data reduction. Ž.I., A.J.C., P.Y., P.R.-L., N.S., F.P., D.B., J.M., L.W., M.D.S., D.J., P.A.M. and C.A. are core collaborators of the MUSSES project who are in charge of follow-up observations (including proposal preparations) with the following telescopes: 3.5-m ARC, 10.4-m GTC, 8.1-m VLT, 2.5-m NOT, 2.5-m INT and 2-m LT. All of the authors contributed to discussions.

Author Information The authors declare no competing financial interests. Correspondence and requests for materials should be addressed to J.J. (email: ).

Acknowledgements The authors thank S. C. Leung and M. Kokubo for helpful discussions. We also thank the staff at the Southern African Large Telescope, the Gemini-North telescope, the Nordic Optical Telescope, the Isaac Newton Telescope, the Liverpool Telescope and the Kiso Schmidt telescope for observations and people who carried out follow-up observations which were unfruitful owing to the weather. Simulations for the He detonation models were carried out on a Cray XC30 at the Center for Computational Astrophysics, National Astronomical Observatory of Japan. The work is partly supported by the World Premier International Research Center Initiative (WPI Initiative), MEXT, Japan, Grants-in-Aid for Scientific Research of JSPS (16H01087 and 26287029 for M.D. and J.J. 26800100 and 17H02864 for K.M. 16H06341, 16K05287, and 15H02082 for T.S. 26400222, 16H02168, and 17K05382 for K.N. 15H02075, 16H02183, and 17H06363 for M.T. 15H05892 for S.M.) and the research grant program of the Toyota foundation (D11-R-0830). S.W.J. acknowledges support from the US National Science Foundation through award AST-1615455. M.D.S acknowledges generous support provided by the Danish Agency for Science and Technology and Innovation realized through a Sapere Aude Level 2 grant, the Instrument-center for Danish Astrophysics (IDA), and by a research grant (13261) from VILLUM FONDEN. The Hyper Suprime-Cam (HSC) collaboration includes the astronomical communities of Japan and Taiwan, and Princeton University. The HSC instrumentation and software were developed by the National Astronomical Observatory of Japan (NAOJ), the Kavli Institute for the Physics and Mathematics of the Universe (Kavli IPMU), the University of Tokyo, the High Energy Accelerator Research Organization (KEK), the Academia Sinica Institute for Astronomy and Astrophysics in Taiwan (ASIAA), and Princeton University. Funding was contributed by the FIRST program from Japanese Cabinet Office, the Ministry of Education, Culture, Sports, Science and Technology (MEXT), the Japan Society for the Promotion of Science (JSPS), Japan Science and Technology Agency (JST), the Toray Science Foundation, NAOJ, Kavli IPMU, KEK, ASIAA, and Princeton University. The Pan-STARRS1 Surveys (PS1) have been made possible through contributions of the Institute for Astronomy, the University of Hawaii, the Pan-STARRS Project Office, the Max-Planck Society and its participating institutes, the Max Planck Institute for Astronomy, Heidelberg and the Max Planck Institute for Extraterrestrial Physics, Garching, The Johns Hopkins University, Durham University, the University of Edinburgh, Queen’s University Belfast, the Harvard-Smithsonian Center for Astrophysics, the Las Cumbres Observatory Global Telescope Network Incorporated, the National Central University of Taiwan, the Space Telescope Science Institute, the National Aeronautics and Space Administration under Grant No. NNX08AR22G issued through the Planetary Science Division of the NASA Science Mission Directorate, the National Science Foundation under Grant No. AST-1238877, the University of Maryland, and Eotvos Lorand University (ELTE). This paper makes use of software developed for the Large Synoptic Survey Telescope. We thank the LSST Project for making their code available as free software at http://dm.lsst.org. This work is also based on observations obtained at the Gemini Observatory (program: GN-2016A-DD-7), which is operated by the Association of Universities for Research in Astronomy, Inc., under a cooperative agreement with the NSF on behalf of the Gemini partnership: the National Science Foundation (United States), the National Research Council (Canada), CONICYT (Chile), Ministerio de Ciencia, Tecnología e Innovación Productiva (Argentina), and Ministério da Ciência, Tecnologia e Inovação (Brazil).

Figure 1: The multi-band light curve of MUSSES1604D. Photometry in g , r and i bands (observer-frame) are in the AB system. Error bars denote 1- σ uncertainties. Dashed lines are best-fitting light curves derived from the non-early photometry (t ≳ 12 days) with SALT2 16 . The explosion epoch is estimated by adopting a classical t 2 fireball model for the early-flash phase (see Methods). The inset zooms in on the early-phase multi-band light curve by Subaru/HSC, which shows that the brightening in g -band “paused” after the second-night observation.

Figure 2: Comparative analysis of MUSSES1604D color evolution. The upper panel presents B − V color evolution of MUSSES1604D, iPTF14atg (early-flash), SN 2006bt (MUSSES1604D-like), SN 2012ht (transitional), SN 2015F and SN 2011fe (normal-brightness). The lower panel shows the color evolution predicted by CEI, He-detonation model for the sub-Chandrasekhar-mass WD and the newly proposed He-detonation models for the near-Chandrasekhar-mass WD under different He-shell mass assumptions. The B -band maximum occurred about 20 days after the explosion. As the bandpass difference between the rest-frame B / V band and the observer-frame g / r band is inconspicuous at z ∼ 0.1, the observed g − r color evolution is provided for reference. Error bars represent 1- σ uncertainties.
Figure 3: Rest-frame B - and V -band light curves of MUSSES1604D and simulations. K-corrections in the flash (open squares) and the post-flash phase (filled squares) are carried out with different methods. Each panel includes He-detonation models for sub-Chandrasekhar-mass WD (1.03 M ⊙ WD + 0.054 M ⊙ He-shell black dashed line) and massive WD (1.28 M ⊙ WD + 0.013 M ⊙ He-shell, orange long-dashed line 1.38 M ⊙ WD + 0.01 M ⊙ He-shell, red dotted line 1.38 M ⊙ WD + 0.03 M ⊙ He-shell, red solid line) conditions. The inset zooms in on the flash phase and also includes our best-fitting CEI model assuming a 1.05 M ⊙ red-giant companion (magenta dashed-dotted line). The first-night g -band data (blue circles) are included in panel a . The explosion epoch shown here is shifted (+0.3 days) from that estimated by the classical t 2 model (Figure 1) within the uncertainty from the simulations. Error bars denote 1- σ uncertainties.

Figure 4: An around-maximum spectral comparison of MUSSES1604D, other observed SNe Ia of different types, and models. In panel a , the spectrum of MUSSES1604D taken 2 days before the B -band maximum by Southern African Large Telescope (SALT) is compared with that of SN 2011fe (normal), SN 1999dq and SN 2000cx (shallow-silicon), SN 1999by (subluminous), iPTF14atg (early-flash) and SN 2012df (MUSSES1604D-like) at a similar epoch. Major absorption features are labeled on the spectrum of MUSSES1604D. In panel b , simulated spectra of the classical W7 deflagration model (top), the newly proposed He-detonation models with different assumed He-shell masses (middle three), and the classical double-detonation model for a sub-Chandrasekhar-mass WD (bottom) are compared with the MUSSES1604D spectrum (dark green) at the same epoch (two days before the B -band maximum).

### Methods

I. The Handbook for MUSSES1604D

The MUSSES project and discovery of MUSSES1604D The Subaru Hyper Suprime-Cam 15 (HSC) is a new-generation, wide-field camera which started to serve as a facility instrument of the 8.2-m Subaru telescope from 2014. With a total of 116 CCDs, a single HSC pointing covers 1.8 square degrees and reaches to a g -band limiting magnitude (5- σ ) of about 26.5 mag with exposure time of 300 s.

The MU lti-band S ubaru S urvey for E arly-phase S Ne Ia ( MUSSES ) is a newly established project which aims to systematically investigate the photometric and spectroscopic behavior of SNe Ia within a few days of their explosions (hereafter early-phase SNe Ia, ESNe Ia) with Subaru/HSC and other 1–10 m class telescopes around the world. In every semester, we plan to carry out 1–2 observing runs and each of them includes two stages: the Subaru/HSC survey (2–3 nights) and follow-up observations. In the survey stage, Subaru/HSC observes over 100 square degrees of sky with a g -band limiting magnitude of 26.0 (5- σ ) every night for finding ESNe Ia and obtaining their multi-band light curve information. Using the HSC transient pipeline and newly employed machine-learning classifiers, we are able to carry out real-time candidate selection during the survey and trigger photometric/spectroscopic follow-ups within one day after the Subaru observation. Because of the fast brightening of ESNe Ia, photometric follow-up observations can be conducted well with 1–4 m telescopes. The strategy of MUSSES gives a very large photometric dynamic range, enabling us to observe ESNe Ia even to redshift z ∼ 0.3.

In order to make the best use of the Subaru time, the MUSSES observing run in April 2016 adopted a specific survey mode which combines both HSC Subaru Strategic Program (HSC SSP 31 , 1-night g -band observation, from UT April 4.17 to UT April 4.67) and open-use observation (1.5-nights g - and r -band observation, from UT April 5.17 to 5.67 and April 6.43 to 6.67 respectively).

The supernova MUSSES1604D (official designation: SN 2016jhr) was discovered on UT April 4.345, 2016 at α (J2000) = 12h18m19s.85 and δ (J2000) = +00 ∘ 15’17.38” with a g -band magnitude of 25.14 mag upon discovery (Extended Data Figure 4). This was the fourth ESN candidate found in the April observing run. MUSSES1604D was located about 5.8” (to the southwest) from the host-galaxy center. The redshift of the host galaxy is 0.11737 ± 0.00001 according to the SDSS (Data Release 12) 32 . With cosmological parameters H 0 = 70 km s − 1 Mpc − 1 , Ω m = 0.30, Ω Λ = 0.70 and Ω ν = 0.00, we calculate a luminosity distance of 546.5 megaparsecs and a distance modulus of 38.69 mag for MUSSES1604D.

The host galaxy The red color with a visible H α emission feature suggests that the host galaxy of MUSSES1604D is a star-forming early-type galaxy 33 . Further analysis of the SDSS photometry and spectroscopy shows that the stellar mass is 3 − 7 × 10 10 M ⊙ , which is also consistent with an early-type galaxy, e.g. an S0 galaxy.

Follow-up observations Our scheduled early follow-up observations at La Palma island and Apache Point Observatory were lost owing to poor weather conditions. By HSC SSP r -band observation conducted at two days after our Subaru/HSC observations, we successfully took another r -band image of MUSSES1604D, which provides a crucial constraint on the timescale of the early-flash. Multi-band follow-up observations with the 8.0-m Gemini-North telescope, the 3.5-m Astrophysical Research Consortium (ARC) telescope, the 2.5-m Nordic Optical Telescope (NOT), the 2.5-m Isaac Newton Telescope (INT), the 2-m Liverpool Telescope (LT) and the 1.05-m Kiso Schmidt telescope have been conducted from about -8 days to +40 days after the B -band maximum. For the spectroscopic observations, we triggered the 9.2-m SALT and the 8.0-m Gemini-North telescope at specific epochs to get spectral evolution from about -2 days to one month after the B -band maximum (Extended Data Figure 2).

Data reduction and photometric calibration As MUSSES1604D resides at the edge of the host galaxy, contamination from the host is negligible except for the photometry of the earliest Subaru/HSC observation. The morphology of the host galaxy indicates a symmetric S0 galaxy. We thus built the host template with GALFIT 34 , 35 and performed the standard point spread function (PSF) photometry with the IRAF DAOPHOT package 36 on host-subtracted images. The photometry has been tested by subtracting the SN from the original image, using an artificial PSF star with the derived photometric magnitude. The average flux of the residual region is comparable with the surrounding region and well below the photometric error of the discovery image by Subaru/HSC (the 1- σ photometric error is 0.15 mag). PSF photometry is performed on host-subtracted images for all follow-up observations as well. The photometry is then calibrated to the standard SDSS photometric system by adopting a color term correction based on field stars 37 . For spectroscopic data reduction, all data were reduced with standard routines in IRAF.

II. Light curve fitting and K-correction

Considering the limited understanding of spectral features during the early optical flash phase of MUSSES1604D, we adopted different methods to derive the rest-frame light curves at flash and post-flash phases respectively. For the post-flash light curves, we firstly fit the observed light curves by applying the SALT2 model of SNe Ia spectrophotometric evolution which is built using a large data set including light curves and spectra of both nearby and distant SNe Ia 16 . After light curve fitting, K-correction is performed to get the rest-frame B - and V -band light curves according to the best-fitting spectral sequence model of MUSSES1604D with SNCosmo 38 . For the light curve in the flash phase (within 5 days after the explosion), we applied the color-based K-correction with a pseudo power-law spectral energy distribution (SED) function f ( ν ) = k ν α , where ν is the frequency of the light, k and α are parameters derived by the early color information of MUSSES1604D. Considering there is no indication of Na I D absorption lines in any of our spectra ( S / N ∼ 18 per resolution element near the wavelength of Na I D lines for the around-maximum spectrum) and the supernova was located far away from the center of an S0-type host, we only take into account the Galactic extinction given by E( B − V ) M W = 0.0263 mag (SFD, 1998 39 ). The rest-frame B - and V -band light curves are shown in Extended Data Figure 3.

The K-corrected rest-frame light curve of MUSSES1604D indicates a B -band peak absolute magnitude of -18.8, but with Δ m 15 ( B ) ≈ 1.0 mag, corresponding to a slow-evolving normal-brightness SN Ia according to the Phillips relation 17 . The V -band light curve of MUSSES1604D is consistent with typical normal-brightness SNe Ia, such as SN 2011fe. All photometric data in observer and rest frames are listed in Extended Data Table 1.

III. Explanations for the peculiarities of MUSSES1604D

The “peculiarities” of MUSSES1604D mainly include: 1) a prominent optical flash with peculiar color evolution at very early time 2) the red B − V color evolution in general 3) a normal-brightness SN Ia with prominent Ti II absorptions in the around-maximum spectrum 4) a slow-evolving B -band light curve. In this section, we compare different scenarios which may account for such peculiarities, and find the best solution.

The companion-ejecta interaction. We performed two-dimensional axisymmetric radiation hydrodynamic simulations of the explosions of a WD with a Chandrasekhar mass in binary systems to obtain light curves and spectra resulting from collisions between the ejecta and the companion star (Kutsuna and Shigeyama’s (K-S) CEI models 19 , 40 ). The ejecta are described by the W7 model 20 . The best-fitting light curves presented in Figure 3 are the outcomes expected from an explosion in a binary system with a separation of 2.5 × 10 13 cm when we observe this event from the companion side. The companion star is a red giant with a mass of 1.05 M ⊙ (the core mass is 0.45 M ⊙ ) and a radius of 8.9 × 10 12 cm, filling the Roche lobe. The initial mass of the companion was assumed to be 1.50 M ⊙ . Although the CEI-induced early flash could be prominent in this condition, we cannot reproduce the early light curves and B − V color evolution of MUSSES1604D because a strong but long-lasting flash will be produced after interacting with a red giant which has a more extended envelope 8 , 19 . For the spectral peculiarity (Figure 4), the prominent Ti II lines also contradict the predictions of typical explosion models through the hydrogen-accreting single degenerate channel 5 , 20 , 21 .

Further comparisons of early-phase light curves with both Kasen’s (K10) and K-S CEI models 8 , 19 are presented in panels a – c of Extended Data Figure 1. Note that K10 predicts a brighter early flash than K-S models because it assumes instantaneous thermalization in the shocked matter while K-S models approximately take into account thermalization processes between shocked matter and radiation (cooling of shocked matter by bremsstrahlung). As K10 noticed, the assumption of instantaneous thermalization tends to underestimate the energies of photons and also results in overestimating the emissivity from shocked matter. Therefore, K10 models produce a prominent flash even with a low-mass main-sequence companion while K-S models can only marginally produce a comparable early flash with a red-giant companion, and produce an even fainter early flash with a main-sequence companion. Despite the different assumptions in two CEI models, with an early flash as bright as that of MUSSES1604D, both K10 and K-S models predict blue color of B − V ≲ 0.1 in the first 4 days after the explosion, which is incompatible with the observations of MUSSES1604D.

The CSM-ejecta interaction. In the double-degenerate progenitor scenario where a SN Ia is generated from the merger of two WDs, a considerable amount of material from the disrupted secondary WD may get pushed out to a large radius 41 , 42 and possibly result in an early ultraviolet/optical flash due to the interaction with the ejecta 9 , 10 . The strong early light curve enhancement seen in MUSSES1604D requires a very extended CSM distribution 10 , 43 . Regardless of the physical possibility of reaching the CSM distribution under their assumptions, interactions with more extended CSM not only strengthen the early flash but also increase the diffusion time, resulting in a bluer and longer flash phase 10 . Panels d – f of Extended Data Figure 1 show the early light curves and color evolution predicted by CSM-ejecta interaction models. To produce a flash with a brightness comparable to that of MUSSES1604D, blue and slow color evolution is inevitable in these models, even after fine-tuning the CSM scale and the 56 Ni distribution of the inner ejecta. Therefore, the CSM-ejecta interaction cannot explain the prominent early flash and the rapid, red B − V color evolution observed for MUSSES1604D.

The He-detonation-triggered scenarios. Another scenario is the SN Ia explosion triggered by the detonation of the He layer. The He detonation generates radioactive materials as the nucleosynthesis ash. For example, He detonation on the surface of a Chandrasekhar-mass WD would leave 56 Ni as a main energy source in this layer with the mass fraction ( X 56 N i ) reaching to ∼ 20% (see below). The diffusion time 44 of optical photons through this He layer is estimated to be ∼ 2 days × ( κ / 0.2 c m 2 g − 1 ) 0.5 ( M H e / 0.02 M ⊙ ) 0. 5 ( V H e / 20 , 000 k m s − 1 ) − 0.5 . Here, a subscript “He” is used for quantities related to the He layer, and the He as a dominant element in the layer is assumed to be fully ionized. M H e and V H e are the mass and velocity of the He layer respectively, and κ is the opacity. The decay power at ∼ 2 days from the 56 Ni in the He layer is estimated to be ∼ 2.5 × 10 41 erg s − 1 ( X 56 N i / 0.2 ) × ( M H e / 0.02 M ⊙ ) . Therefore, the radioactivity in the He-detonation ash is predicted to produce a prompt flash lasting for a few days with the peak bolometric magnitude of ∼ -16, assuming the He mass M H e ∼ 0.03 M ⊙ . This scenario roughly explains the nature of the early flash found for MUSSES1604D. For the sub-Chandrasekhar WD, the abundance in the He ash is dominated by the other radioactive isotopes, 52 Fe and 48 Cr, and they power the early flash. Still, a similar argument as above applies.

The synthetic light curves and spectra expected from the He-shell detonation models are simulated as follows (Figures 2–4). We constructed a series of toy one-dimensional models which mimic the results of DDet hydrodynamic simulations 13 , 23 . The density structure is assumed to be exponential in velocity space, where the kinetic energy is specified by the energy generation for the assumed burned composition structure. A stratified structure in the composition and a uniform abundance pattern in each layer are assumed, where the distribution of the burning products is set to represent the DDet models 23 .

The model structures are shown in Extended Data Figures 5 and 6. Our sub-Chandrasekhar model and Chandrasekhar model are similar to a typical DDet model and the W7 model, respectively, in the mass coordinate. For the Chandrasekhar WD model, we replaced part of the 56 Ni-rich region with a Si-rich region, leading to a more centrally confined structure than in the W7 model. Note that we assume a stable Fe/Ni region in the core of the Chandrasekhar model, the mass of which is taken to be ∼ 0.2 M ⊙ similar to that in the W7 model. For each model, we run multi-frequency and time-dependent Monte-Carlo radiation transfer calculations 40 , which were updated to include radioactive energy input from the decay chains of 52 Fe/Mn/Cr and 48 Cr/V/Ti together with 56 Ni/Co/Fe. The code assumes LTE for the ionization, which is generally believed to be a good approximation in the early phase. For example, in the W7 model, LTE and NLTE simulations yield indistinguishable light curves (except for the U -band) until ∼ 25 days after the explosion, corresponding to ∼ 5 days after the B -maximum 45 , 46 . In addition, we do not include the non-thermal excitation of He generally He absorption lines in optical wavelength are invisible for the DDet models even with this effect 24 .

The peculiar early light curve and color evolution as well as the strong Ti II absorptions for MUSSES1604D can be naturally reproduced by the DDet scenario, as shown by the model for the 1.03 M ⊙ WD with 0.054 M ⊙ He (ash) layer in Figures 2–4. We note that the idea that the DDet model predicts an early flash by the radioactive decay of He ash was independently proposed by Noebauer et al. 25 in work posted on arXiv after we submitted this paper. Their model, which is qualitatively similar to our sub-Chandrasekhar model, leads to the early flash and the color evolution in the first few days after an explosion, powered by the decay of 52 Fe and 48 Cr, as is similar to our model prediction. However, their model lacks around/post-maximum light-curve and spectral information, therefore further comparison with our model is not possible.

Although the sub-Chandrasekhar DDet model can explain most of peculiarities of MUSSES-1604D, it has prominent defects in the resulting fast evolution of the simulated B -band light curve (see also refs 13 , 23 ). Note that the fast decline of B -band light curve predicted by this classical DDet scenario of the sub-Chandrasekhar-mass WD happens from ∼ 17 days after the explosion, and the magnitude becomes ∼ 1.4 mag fainter than the peak at t ∼ 25 days. The difference between the LTE and NLTE treatments in the first 25 days after the explosion is too small to account for such abnormal light curve evolution 45 , 46 , and thus it is unlikely that the LTE assumption accounts for this discrepancy. Another issue is that the quantity of radioactive isotopes in this model (0.054 M ⊙ of the He layer as a minimal He shell for the He detonation) produces a stronger flash than that of MUSSES1604D, suggesting that the observationally required mass of the He layer is lower. For further investigation of the classical DDet scenario, we ran a grid of models spanning WD masses from ∼ 0.9 to 1.4 M ⊙ , but all the models predicted very fast evolution in the B -band light curve and/or too-bright peak luminosity.

Indeed, this fast evolution in the B -band light curve has been recognized as one of the issues in the (sub-Chandrasekhar) DDet model 13 , because the Fe-peak and Ti/Cr in the He ash should start blocking the photons in the bluer bands once the temperature decreases after the maximum light, and this argument is not sensitive to the LTE or NLTE treatment. It has been shown that this problem could be remedied if the mass of the He layer is much smaller than that required by the classical DDet Model so as not to provide a large opacity 29 , 30 , 47 , partly based on an idea that such a small amount of He ( < 0.01 M ⊙ ) would lead to detonation when a substantial fraction of carbon is mixed in the He layer 48 . We have also confirmed from our model sequence that the light curves of sub-Chandrasekhar DDet models are indeed roughly consistent with normal (but relatively faint and fast-evolving) SNe Ia, once the He layer is removed. However, this scenario would not explain MUSSES1604D, as we do see prominent early flash and signatures of the He ash in the maximum spectrum.

To remedy the abnormal fast-evolution issue in the classical DDet scenario, we investigated additional models in which we allow that the relation between the WD mass and the final 56 Ni production expected in DDet is not necessarily fulfilled. By altering the WD mass, 56 Ni mass, and the He mass, the most straightforward choice we found is shown in Figures 2–4, where the models with 1.38 M ⊙ WD, 0.01–0.03 M ⊙ He-ash layer, and 0.43 M ⊙ of 56 Ni are presented. Additionally we investigated the model with 1.28 M ⊙ WD, 0.013 M ⊙ He-ash layer, and 0.44 M ⊙ of 56 Ni. While such a relatively less massive WD model can also give a slow-evolving light curve, the pre-maximum B − V color turns out to be too red. Therefore, we constrain the acceptable WD mass range between 1.28 and 1.38 M ⊙ . In addition, an even better consistency of the light curves and color evolution from ∼ 5 days after the B -band maximum could be expected for our preferred model (1.38 M ⊙ WD + 0.03 M ⊙ He-shell) once the NLTE effects were taken into account 45 , 46 .

From these analyses, we suggest two scenarios that involve He detonation. First is the He-ignited violent merger scenario 14 . In this case, the primary WD mass should still be ∼ 1 M ⊙ to produce the required peak luminosity. The accretion stream of He during the merging process may trigger a detonation even if the He mass is low 14 , 27 . If the secondary white dwarf is swept up by the ejecta, this would explain the slow-evolving light curve. However, there are two drawbacks to this scenario: (1) it is uncertain whether the core detonation can be triggered by the thin-He-shell detonation 26 , 27 and (2) it will involve fine-tuning of the merging configuration (e.g., the masses of the WDs) to reproduce the observational features of MUSSES1604D.

The second scenario is the He detonation on the surface of a nearly-Chandrasekhar-mass WD, as is motivated by our light curve and spectral models which reproduce the observational results quite well simply by assuming a standard Chandrasekhar-mass WD without fine-tuning. The amount of He mass is also consistent in this picture to trigger the detonation there. The evolutionary track of this binary system towards the He detonation on the surface of a WD more massive than 1.3 M ⊙ has been never discussed in the literature. Further investigations are needed to explore whether this scenario can be realized or not. Another drawback is that in the classical DDet scenario it will produce too much 56 Ni through core detonation, resulting in an over-luminous SN Ia. Still, the fact that this simple model explains all the main features of MUSSES1604D is striking, indicating that it is unlikely to be a mere coincidence. This would suggest that there could be a mechanism to reduce the mass of 56 Ni as compared to the classical DDet picture.

More realistic light curve and spectra might be realized if one takes into account the possible viewing-angle effect related to both the violent merger scenario and the He-ignited near-Chandrasekhar-WD scenario. Our one-dimensional models only address angle-averaged behavior. The Ti/Fe absorptions will be stronger than our one-dimensional prediction if the line of sight is to intersect a region of the He ash. The initial light-curve enhancement would also be dependent on the viewing angle, but this effect would be much less prominent than in the absorption.

Another issue is that the Si and S features of MUSSES1604D are not very well reproduced. In general, these features are qualitatively well explained but obtaining quantitatively good fits is an issue even with sophisticated NLTE modeling 49 . We find that these features are also sensitive to detailed composition structure even in one-dimensional simulations. Providing detailed fitting for these features is beyond the scope of this study, as these features are theoretically more uncertain than the features we have analyzed in this paper.

IV. The explosion epoch of MUSSES1604D

Extrapolating the explosion epoch of a SN Ia based on the 56 Ni -powered light curve is controversial because a considerable “dark phase” between the explosion and the radioactive decay from SN ejecta may exist for some SNe Ia 50 − 53 . For example, the best-observed SN Ia so far, SN 2011fe likely has a one-day dark phase though it was discovered at the brightness of ∼ 1/1000 of its peak brightness 52 , 54 . More stringent restrictions on the explosion epoch require not only deep-imaging observations but also specific radiation mechanisms at early times to light up the dark phase 55 . Thanks to the deep imaging capability of Subaru/HSC and the early flash of MUSSES1604D, the explosion time of MUSSES1604D can be pinpointed.

In He-detonation-triggered scenarios, the early optical flash is produced immediately from the radioactive decay at the surface of the SN ejecta, which is the earliest optical emission except for the almost non-detectable cooling emission from the shock-heated WD soon after the SN shock breakout 50 , 56 . Thus, MUSSES1604D was discovered at an earlier phase than any previously discovered SN Ia. Given an effectively negligible dark phase before the early flash, we adopt the classical t 2 fireball model (where t is the time since the explosion) for the rising phase of the early flash, assuming that neither the photospheric temperature nor the velocity changes significantly in estimating the explosion epoch of MUSSES1604D. The result indicates that the first observation of MUSSES1604D is at ∼ 0.51 + 0.08 − 0.06 days after the SN explosion. According to the best-fitting light curves derived from the post-flash multi-band photometry (dashed lines in Figure 1), the g -band magnitude reaches the same level of our first observation (25.14 mag) at t ∼ 3 days. This may imply that a non-negligible dark phase exists for non-early-flash SNe Ia.

V. MUSSES1604D-like SNe Ia and their rarity

The rate of occurrence of He-detonation-triggered SNe Ia can be constrained by estimating the fraction of MUSSES1604D-like SNe Ia. By inspecting over 1,000 SNe Ia from normal to various different subtypes which have at least one good spectrum from about -6 to +12 days after their B -band maximum through published resources and open SN databases 57 , 58 , three MUSSES1604D-like SNe Ia (without early-phase observations) have been found. The screening criteria and detailed properties of the MUSSES1604D-like SNe Ia are listed in Extended Data Table 2. In addition to three normal-brightness SNe Ia (-19.4 ≲ M B ≲ -18.7) we mentioned here, some subluminous SNe Ia also show good similarities to MUSSES1604D (e.g. 02es-like SNe Ia, PTF10ops and SN 2010lp, which also have slow-evolving light curve and similar spectral features to MUSSES1604D 59 , 60 ). However, due to insufficient information to classify these subluminous objects conclusively, the discussion here focuses on the best three MUSSES1604D-like SNe Ia, namely SN 2006bt, SN 2007cq, and SN 2012df 61 − 64 .

A normal-brightness SN Ia, SN 2006bt shows good similarity with MUSSES1604D in both light curve and spectral features except for the shallow Si II λ 5972 absorption seen in MUSSES1604D. Because there is no Na I D feature in the spectra of SN 2006bt and the SN is far away from the center of an S0/a host galaxy, the absolute magnitude is shown in Extended Data Table 2 without taking into account the host extinction. Well-organized follow-up observations for SN 2006bt indicate pre-maximum Ti II absorptions, a slow-evolving B -band light curve and similar B − V color evolution to MUSSES1604D.

SN 2007cq is classified as another MUSSES1604D-like SN Ia. In particular, the pre-maximum spectroscopy of SN 2007cq shows prominent Ti II absorptions from about 6 days before the B -band maximum, which is consistent with the prediction of the He-detonation models 12 , 23 . Note that SN 2007cq shows shallower intermediate element absorption features and bluer color than MUSSES1604D, which could be attributed to a larger amount of 56 Ni generated from the core explosion for SN 2007cq.

SN 2012df was located at the edge of an S0-like galaxy. The spectrum was taken near its brightness peak with an unfiltered absolute magnitude of ∼ -18.9 (without extinction correction). Despite the limited observational information for SN 2012df, high spectral similarity between two SNe Ia has been found at a similar epoch (Figure 4). Therefore we classify SN 2012df as a MUSSES1604D-like SN Ia. Comparisons of the spectral evolution and light curves of MUSSES1604D-like SNe Ia are presented in Extended Data Figures 2 and 3 respectively.

To obtain a conservative estimate of the event rate of MUSSES1604D-like SNe Ia, we eliminated all subluminous objects, even though some of them may have the same origin 59 , 60 . Statistically there are 4 MUSSES1604D-like objects (including MUSSES1604D) out of ∼ 800 SNe Ia with B -band peak absolute magnitude ≲ -18.7, corresponding to a fraction of MUSSES1604D-like SNe Ia of ∼ 0.5%.

The traditional SN Ia classification which is mainly based on SN brightness and spectral features will classify MUSSES1604D and iPTF14atg into two peculiar subtypes, even though both have strong early light curve enhancements, slow-evolving light curves, prominent Ti II absorptions, and similar color evolution and host environments 7 , 55 , implying that they might be intrinsically connected 65 . However, whether iPTF14atg is also triggered by the He-shell detonation is an open question because the Ti II absorptions and red color of subluminous SNe Ia in the post-flash phase can be attributed to the low temperature of ejecta, and the lack of early color information prevents us from further comparisons with MUSSES1604D at the flash phase. It is worth noting that the earliest B − V color of iPTF14atg at ∼ 5 days after the explosion is probably too red to be explained by CEI or CSM-ejecta interaction, but is in line with the predictions of He-detonation models (Figure 2). As a reference for the future work, in Extended Data Table 2, we list MUSSES1604D-like and iPTF14atg-like candidates selected from different SN Ia branches 59 , 60 , 66 − 68 . Similarities among these objects may suggest intrinsic connections between a number of SNe Ia of different subtypes.

Code availability. The post-flash light curve fitting and K-correction are carried out with the SALT2 model and SNCosmo, which are available at http://supernovae.in2p3.fr/salt/doku.php & https://sncosmo.readthedocs.io/en/v1.5.x/ respectively. We have not made publicly available the code for the companion-ejecta interaction (CEI) models or the radiation-transfer code used for He-detonation simulations, because they are not prepared for the open-use. Instead, the simulated light curves and spectra for the He-detonation models shown in this paper are available upon request.

Data availability. The Source Data for Figures 1, 3 and 4 are available in the online version of the paper. Photometric and spectroscopic data will also be made publicly available on WISeREP3 (http://wiserep.weizmann.ac.il/).

31. Miyazaki, S., et al. Wide-field imaging with Hyper Suprime-Cam: Cosmology and Galaxy Evolution, A Strategic Survey Proposal for the Subaru Telescope. http://hsc.mtk.nao.ac.jp/ssp/wp-content/uploads/2016/05/hsc_ssp_rv_jan13.pdf (2014).

32. Bolton, A. S., et al. Spectral Classification and Redshift Measurement for the SDSS-III Baryon Oscillation Spectroscopic Survey. Astron. J. 144, 144 (2012).

33. Shimasaku, K., et al. Statistical Properties of Bright Galaxies in the Sloan Digital Sky Survey Photometric System. Astron. J. 122, 1238–1250 (2001).

34. Peng, C. Y., Ho, L. C., Impey, C. D. & Rix, H. -W. Detailed Decomposition of Galaxy Images. II. Beyond Axisymmetric Models. Astron. J. 139, 2097–2129 (2010).

36. Stetson, P. B. DAOPHOT: A Computer Program for Crowded-Field Stellar Photometry. Publ. Astron. Soc. Pac. 99, 191–222 (1987).

37. Doi, M., et al. Photometric Response Functions of the Sloan Digital Sky Survey Imager. Astron. J. 139, 1628–1648 (2010).

39. Schlegel, D. J., Finkbeiner, D. P. & Davis, M. Maps of Dust Infrared Emission for Use in Estimation of Reddening and Cosmic Microwave Background Radiation Foregrounds. Astrophys. J. 500, 525–553 (1998).

40. Maeda, K., Kutsuna, M. & Shigeyama, T. Signatures of a Companion Star in Type Ia Supernovae. Astrophys. J. 794, 37 (2014).

41. Fryer, C. L., et al. Spectra of Type Ia Supernovae from Double Degenerate Mergers. Astrophys. J. 725, 296–308 (2010).

42. Shen, K. J., Bildsten, L., Kasen, D. & Quataert, E. The Long-term Evolution of Double White Dwarf Mergers. Astrophys. J. 748, 35 (2012).

43. Levanon, N. & Soker, N. Early UV emission from disk-originated matter (DOM) in type Ia supernovae in the double degenerate scenario. Mon. Not. R. Astron. Soc. 470, 2510–2516 (2017).

44. Arnett, D. Supernovae and Nucleosynthesis: An Investigation of the History of Matter from the Big Bang to the Present. Princeton University Press (1996).

45. Kasen, D., Thomas, R. C. & Nugent, P. Time-dependent Monte Carlo Radiative Transfer Calculations for Three-dimensional Supernova Spectra, Light Curves, and Polarization. Astrophys. J. 651, 366–380 (2006).

46. Kromer, M. & Sim, S. A. Time-dependent three-dimensional spectrum synthesis for Type Ia supernovae. Mon. Not. R. Astron. Soc. 398, 1809–1826 (2009).

47. Sim, S. A.,et al. Detonations in Sub-Chandrasekhar-mass C+O White Dwarfs. Astrophys. J. Lett. 714, L52–L57 (2010).

48. Shen, K. J., & Moore, K. The Initiation and Propagation of Helium Detonations in White Dwarf Envelopes. Astrophys. J. 797, 46 (2014).

49. Nugent, P., Phillips, M., Baron, E., Branch, D. & Hauschildt, P. Evidence for a Spectroscopic Sequence among Type 1a Supernovae. Astrophys. J. Lett. 455, L147–L150 (1995).

50. Piro, A. L., & Nakar, E. What can we Learn from the Rising Light Curves of Radioactively Powered Supernovae? Astrophys. J. 769, 67 (2013).

51. Piro, A. L., & Nakar, E. Constraints on Shallow 56 Ni from the Early Light Curves of Type Ia Supernovae. Astrophys. J. 784, 85 (2014).

52. Mazzali, P. A., et al. Hubble Space Telescope spectra of the Type Ia supernova SN 2011fe: a tail of low-density, high-velocity material with Z < Z ⊙ . Mon. Not. R. Astron. Soc. 439, 1959–1979 (2014).

53. Zheng, W., et al. Estimating the First-light Time of the Type Ia Supernova 2014J in M82. Astrophys. J. Lett. 783, L24 (2014).

54. Nugent, P. E., et al. Supernova SN 2011fe from an exploding carbon-oxygen white dwarf star. Nature. 480, 344–347 (2011).

55. Cao, Y., et al. SN2002es-like Supernovae from Different Viewing Angles. Astrophys. J. 832, 86 (2016).

56. Piro, A. L., Chang, P. & Weinberg, N. N. Shock Breakout from Type Ia Supernova. Astrophys. J. 708, 598–604 (2010).

57. Yaron, O. & Gal-Yam, A. WISeREP–An Interactive Supernova Data Repository. Publ. Astron. Soc. Pac. 124, 668–681 (2012).

58. Guillochon, J., Parrent, J., Kelley, L. Z. & Margutti, R. An Open Catalog for Supernova Data. Astrophys. J. 835, 64 (2017).

59. Maguire, K., et al. PTF10ops – a subluminous, normal-width light curve Type Ia supernova in the middle of nowhere. Mon. Not. R. Astron. Soc. 418, 747–758 (2011).

60. Kromer, M., et al. SN 2010lp–a Type Ia Supernova from a Violent Merger of Two Carbon-Oxygen White Dwarfs. Astrophys. J. Lett. 778, L18 (2013).

61. Foley, R. J., et al. SN 2006bt: A Perplexing, Troublesome, and Possibly Misleading Type Ia Supernova. Astrophys. J. 708, 1748–1759 (2010).

62. Ganeshalingam, M., et al. Results of the Lick Observatory Supernova Search Follow-up Photometry Program: BVRI Light Curves of 165 Type Ia Supernovae. Astrophys. J. Suppl. 190, 418–448 (2010).

63. Scalzo, R., et al. Type Ia supernova bolometric light curves and ejected mass estimates from the Nearby Supernova Factory. Mon. Not. R. Astron. Soc. 440, 1498–1518 (2014).

64. Ciabattari, F., et al. Supernova 2012df = Psn J17481875+5218023. Central Bureau Electronic Telegrams. No. 3161 (2012).

65. Kromer, M., et al. The peculiar Type Ia supernova iPTF14atg: Chandrasekhar-mass explosion or violent merger? Mon. Not. R. Astron. Soc. 459, 4428–4439 (2016).

66. Ganeshalingam, M., et al. The Low-velocity, Rapidly Fading Type Ia Supernova 2002es. Astrophys. J. 751, 142 (2012).

67. Li, W., et al. SN 2002cx: The Most Peculiar Known Type Ia Supernova. Publ. Astron. Soc. Pac. 115, 453–473 (2003).

68. Foley, R. J., et al. Type Iax Supernovae: A New Class of Stellar Explosion. Astrophys. J. 767, 57 (2013).

Extended Data Figure 1: Comparison of MUSSES1604D observations and different model simulations at flash phase. Symbols for MUSSES1604D data are the same as those in Figures 1–3 and the results from our best-fitting He-detonation model (1.38 M ⊙ WD + 0.03 M ⊙ He-shell, red solid lines) are shown in each panel. Panels a-c present early B -band ( a ) and V -band ( b ) light curves and B − V color evolution ( c ) generated by different CEI simulations observed from the companion side. Dashed lines correspond to the K10 models with different binary-system compositions (MS, main-sequence star RG, red-giant star) 8 . The magenta dashed-dotted line denotes our best-fitting K-S CEI model 19 . Although an early flash as bright as that of MUSSES1604D could be produced with specific CEI models, the predicted color is very blue at the CEI-flash phase. Panels d and e are V -band light curves simulated by the CSM-ejecta interaction with deep ( d ) and shallow ( e ) 56 Ni distribution for the inner ejecta (Piro & Morozova, P16 10 ). Dotted lines correspond to an external mass of M e = 0.3 M ⊙ with different outer radii R e . Panel f is the color evolution under the same assumptions as in e . Similar to CEI models, combinations of early light curves and color evolution predicted by the CSM-ejecta interaction are different from the observed features of MUSSES1604D.

Extended Data Figure 2: Spectral evolution of MUSSES1604D and analogues. Spectra for MUSSES1604D (dark green) are compared with those of the analogous SNe Ia SN 2006bt, SN 2007cq and SN 2012df at similar epochs. Late-phase spectra of SN 2011fe are included for reference. SALT/RSS follow-up observations were carried out -2 and 12 days after the B -band maximum and the other two spectra were taken by Gemini-N/GMOS 3 and 26 days after the B -band maximum.

Extended Data Figure 3: Rest-frame B - and V -band light curves for MUSSES1604D and other SNe Ia. K-corrections in flash (open squares) and post-flash phase (filled squares with dashed lines) of MUSSES1604D were carried out with different methods (see Methods). An excellent light curve match is shown for MUSSES1604D, SN 2006bt and SN 2007cq. Another peculiar early-flash SN Ia iPTF14atg also shows similar light curves though its brightness is ∼ 1 magnitude fainter than MUSSES1604D. Light curves of a normal SN Ia, SN 2011fe (black dotted lines) are provided for reference. Magnitudes shown here are in the Vega system and the error bars denote 1- σ uncertainties.

Extended Data Figure 4: Early Subaru/HSC g -band images for MUSSES1604D. The left panel shows the earliest Subaru/HSC image of MUSSES1604D ( α (J2000) = 12h18m19s.85, δ (J2000) = +00 ∘ 15’17.38”) taken on UT April 4.345, 2016, when the g -band magnitude of MUSSES1604D was 25.14 ± 0.15. The supernova then brightened rapidly to ∼ 23.1 mag in one day (right panel).

Extended Data Figure 5: Composition structures of models used for radiation-transfer simulations. The composition structures shown here are He-detonation models for the sub-Chandrasekhar-mass WD (1.03 M ⊙ WD + 0.054 M ⊙ He-shell panels a & c ) and the Chandrasekhar-mass WD (1.38 M ⊙ WD + 0.03 M ⊙ He-shell panels b & d ). The mass fractions of selected elements are shown as a function of velocity ( a , b ) or mass coordinate ( c , d ). Colors used for selected elements are same for all panels.

Extended Data Figure 6: Density structures of the models used for radiation-transfer simulations. The density structures (as a function of velocity) shown here are He-detonation models for the sub-Chandrasekhar-mass WD (1.03 M ⊙ WD + 0.054 M ⊙ He-shell black dashed line) and the Chandrasekhar-mass WD (1.38 M ⊙ WD + 0.03 M ⊙ He-shell red solid line).

Extended Data Table 1: Imaging observations of MUSSES1604D

Notes. The magnitudes in g , r and i bands (observer-frame AB system) have been transferred to the standard SDSS photometric system by adopting a color term correction based on field stars. Rest-frame B - and V -band magnitudes are in the Vega system. Numbers in parenthesis correspond to 1- σ statistical uncertainties in units of 1/100 mag.

Days (rest-frame) relative to the estimated date of the B -band maximum, UT April 26.27, 2016.

K-correction for the flash-phase (April 4–8) observations is carried out by using the power-law spectral energy distribution models derived from the color of the early flash. For post-flash observations, K-correction is performed according to the best-fitting spectral sequence model of MUSSES1604D. The Galactic extinction (E( B − V ) M W = 0.0263 mag) has been corrected.

Extended Data Table 2: Properties of MUSSES1604D- and iPTF14atg-like SNe Ia

Notes. For each property, we use “+”, “ | ” and “-” footnotes as “support”, “neutral” and “opposite” respectively to show the similarity between candidates and MUSSES1604D/iPTF14atg. For all three MUSSES1604D-like SNe Ia, the host extinction are neglected because of the relatively distant location of SNe to the center of their S0/a host galaxies and the non-detection of Na I D lines in their spectra. Galactic extinction has been applied with E( B − V ) M W of 0.1096 mag and 0.050 mag for SN 2007cq and SN 2006bt respectively.

The absolute magnitude for iPTF14atg, 02es-like (SN 2002es, SN 2010lp, PTF10ops) and all normal-brightness SNe Ia was calculated by using cosmological parameters H 0 = 70.0 km s − 1 Mpc − 1 , Ω m = 0.30, Ω Λ = 0.70 and Ω ν = 0.00. For 02cx-like SNe Ia, we adopt the value from the related paper 68 .

The B − V color information at around the B -band maximum. Here, we define B − V ≥ 0.4 mag, 0.4 mag > B − V ≥ 0.2 mag, 0.2 mag > B − V ≥ 0.1 mag, 0.1 mag > B − V ≥ -0.1 mag and -0.1 mag > B − V as “ultra-red”, “red”, “marginal-red”, “normal” and “blue”, respectively.

Spectral features at around the B -band maximum. For normal-brightness and subluminous SNe Ia, we used spectra taken on the closest epoch to t = -2 and t = 0, respectively (relative to the B -band maximum) for the similarity comparisons.

The relative strength of Ti II absorptions near the B -band maximum. The strength is relative to normal-type SNe Ia, e.g. SN 2011fe.

We define the equivalent width (EW) of Si II λ 5972 line as: EW (Si I I λ 5972 ) ≤ 10 AA , 10 AA < EW (Si I I λ 5972 ) ≤ 30 AA , 30 AA < EW (Si I I λ 5972 ) as “Shallow”, “Intermediate” and “Deep”, respectively.

The relative evolution speed of Ti II absorptions in the first 10 ± 2 days after the B -band maximum. The evolution speed is relative to SN 2011fe and iPTF14atg for normal-brightness and subluminous SNe Ia respectively.

Unfiltered photometry without considering the Galactic extinction E( B − V ) M W = 0.0393 mag.

If you find a rendering bug, file an issue on GitHub. Or, have a go at fixing it yourself – the renderer is open source!

## First discovery of a binary companion for a Type Ia supernova

The blue-white dot at the centre of this image is supernova 2012cg, seen by the 1.2-metre telescope at Fred Lawrence Whipple Observatory. At 50 million light-years away, this supernova is so distant that its host galaxy, the edge-on spiral NGC 4424 in the constellation Virgo, appears here as only an extended smear of purple light. Image credit: Peter Challis/Harvard-Smithsonian CfA. A team of astronomers including Harvard’s Robert Kirshner and Peter Challis has detected a flash of light from the companion to an exploding star. This is the first time astronomers have witnessed the impact of an exploding star on its neighbour. It provides the best evidence on the type of binary star system that leads to Type Ia supernovae. This study reveals the circumstances for the violent death of some white dwarf stars and provides deeper understanding for their use as tools to trace the history of the expansion of the universe. These types of stellar explosions enabled the discovery of dark energy, the universe’s accelerating expansion that is one of the top problems in science today.

The subject of how Type Ia supernovae arise has long been a topic of debate among astronomers.

“We think that Type Ia supernovae come from exploding white dwarfs with a binary companion,” said Howie Marion of The University of Texas at Austin (UT Austin), the study’s lead author. “The theory goes back 50 years or so, but there hasn’t been any concrete evidence for a companion star before now.”

Astronomers have battled over competing ideas, debating whether the companion was a normal star or another white dwarf.

“This is the first time a normal Type Ia has been associated with a binary companion star,” said team member and professor of astronomy J. Craig Wheeler (UT Austin). “This is a big deal.”

The binary star progenitor theory for Type Ia supernovae starts with a burnt-out star called a white dwarf. Mass must be added to that white dwarf to trigger its explosion &mdash mass that the dwarf pulls off of a companion star. When the influx of mass reaches the point that the dwarf is hot enough and dense enough to ignite the carbon and oxygen in its interior, a thermonuclear reaction starts that causes the dwarf to explode as a Type Ia supernova.

For a long time, the leading theory was that the companion was an old red giant star that swelled up and lost matter to the dwarf, but recent observations have virtually ruled out that notion. No red giant is seen. The new work presents evidence that the star providing the mass is still burning hydrogen at its centre, that is, that this companion star is still in the prime of life.

According to team member Robert P. Kirshner of the Harvard-Smithsonian Center for Astrophysics, “If a white dwarf explodes next to an ordinary star, you ought to see a pulse of blue light that results from heating that companion. That’s what theorists predicted and that’s what we saw.

“Supernova 2012cg is the smoking &mdash actually glowing &mdash gun: some Type Ia supernovae come from white dwarfs doing a do-si-do with ordinary stars.”

Located 50 million light-years away in the constellation Virgo, Supernova 2012cg was discovered on 17 May 2012 by the Lick Observatory Supernova Search. Marion’s team began studying it the next day with the telescopes of the Harvard-Smithsonian Center for Astrophysics.

“It’s important to get very early observations,” Marion said, “because the interaction with the companion occurs very soon after the explosion.”

The team continued to observe the supernova’s brightening for several weeks using many different telescopes, including the 1.2-metre telescope at Fred Lawrence Whipple Observatory and its KeplerCam instrument, the Swift gamma-ray space telescope, the Hobby-Eberly Telescope at McDonald Observatory, and about half a dozen others.

“This is a global enterprise,” Wheeler said. Team members hail from about a dozen U.S. universities, as well as institutions in Chile, Hungary, Denmark, and Japan.

What the team found was evidence in the characteristics of the light from the supernova that indicated it could be caused by a binary companion. Specifically, they found an excess of blue light coming from the explosion. This excess matches with the widely accepted models created by U.C. Berkeley astronomer Dan Kasen for what astronomers expect to see when a star explodes in a binary system.

“The supernova is blowing up next to a companion star, and the explosion impacts the companion star,” Wheeler explained. “The side of that companion star that’s hit gets hot and bright. The excess blue light is coming from the side of the companion star that gets heated up.”

Combined with the models, the observations indicate that the binary companion star has a minimum mass of six suns.

“This is an interpretation that is consistent with the data,” said team member Jeffrey Silverman, stressing that it is not concrete proof of the exact size of the companion, like would come from a photograph of the binary star system. Silverman is a postdoctoral researcher at UT Austin.

Only a few other Type Ia supernovae have been observed as early as this one, Marion said, but they have not shown an excess of blue light. More examples are needed.

“We need to study a hundred events like this and then we’ll be able to know what the statistics are,” Wheeler said.

The work has just been published in The Astrophysical Journal.

This press release is being issued jointly with The University of Texas at Austin.

## How is the first detonation in Supernove type Ia triggered? - Astronomy

The model of a presupernova's carbon-oxygen (C-O) core with an initial mass of 1.33 M_solar, an initial carbon abundance X_C^ <(0)>= 0.27, and a mean rate of increase in mass of 5 x 10^ <-7>M_solar/yr through accretion in a binary system evolved from the central density and temperature rho_c = 10^9 g/cm^3 and T_c = 2.05 x 10^8 K, respectively, by forming a convective core and its subsequent expansion to an explosive fuel ignition at the center. The evolution and explosion equations included only the carbon burning reaction C-12 + C-12 with energy release corresponding to the complete conversion of carbon and oxygen (at the same rate as that of carbon) into Ni-56. The ratio of mixing length to convection-zone size alpha_c was chosen as the parameter. Although the model assumptions were crude, we obtained an acceptable (for the theory of supernovae) pattern of explosion with a strong dependence of its duration on alpha_c. In our calculations with sufficiently large values of this parameter, alpha_c = 4.0 x 10^ <-3>and 3.0 x 10^<-3>, fuel burned in the regime of prompt detonation. In the range 2.0 x 10^ <-3>> alpha_c > 3.0 x 10^<-4>, there was initially a deflagration with the generation of model pulsations whose amplitude gradually increased. Eventually, the detonation regime of burning arose, which was triggered from the model surface layers (with m = 1.33 M_solar) and propagated deep into the model up to the deflagration front. The generation of model pulsations and the formation of a detonation front are described in detail for alpha_c = 1.0 x 10^<-3>.

## The origin of Type Ia supernovae revealed by manganese abundances

(a) Near-Chandrasekhar mass explosions: In a binary system of one white dwarf that is made of carbon and oxygen, mass accretion from the companion star (a main se-quence star or red giant) causes winds of material from the white dwarf, which regu-lates the mass accretion onto the white dwarf, and increases the white dwarf mass. Subsonic waves from the explosion at the center of near-Chandrasekhar mass white dwarf trigger a detonation in the outskirts. This explosion can produce a lot of manga-nese (Mn) and nickel (Ni) as well as iron (Fe). (b) An example of sub-Chandrasekhar mass explosions: In a binary system of two white dwarfs (at least one white dwarf consists of carbon and oxygen), the smaller one is dis-rupted by tidal forces and merges with the larger one. A detonation in a thin helium enve-lope around the white dwarf triggers a carbon detonation at the center. This explosion can produce more silicon (Si) and sulfur (S), as well as iron (Fe), and unburnt carbon and oxygen. Credit: The Astrophysical Journal

A research team at the Kavli Institute for the Physics and Mathematics of the Universe (Kavli IPMU) consisting of Visiting Scientist Chiaki Kobayashi, Project Researcher at the time Shing-Chi Leung (currently at the California Institute of Technology), and Senior Scientist Ken'ichi Nomoto have used computer simulations to follow the explosion, nuclear reaction, production of elements, and evolution of elemental abundances in galaxies. As a result, they placed stringent constraints on the origin of Type Ia supernovae.

A Type Ia supernova is a type of supernova that is not related to the death of a massive star. Instead, a Type Ia supernova is a luminous explosion of a star that occurs in a binary system, where two relatively low-mass stars are evolving together. Because of their relatively constant luminosity, Type Ia supernovae have been used as a standard "candle" to measure the expansion of the universe, a result for which the 2011 Nobel Prize in Physics was awarded. However, the progenitor star of a Type Ia supernova is unknown, and has been the topic of debate for around a half century.

"As usual for normal supernovae, Type Ia supernovae produce "metals"—or, in astronomical terms, chemical elements heavier than hydrogen and helium, the latter pair tracing their origin to the Big Bang—but Type Ia supernovae produce different elements, such as manganese (Mn), nickel (Ni), and iron (Fe). These elemental abundances can be measured in spectral features of nearby stars, which keep a "record" of supernovae from the past, like fossils do in archaeology,'' Kobayashi, who is also an associate professor at the University of Hertfordshire in the United Kingdom, said. Therefore, the evolution of elemental abundances in galaxies can provide a stringent constraint on the true origin of Type Ia supernovae.

The progenitor stars of Type Ia supernovae are a type of white dwarf that are made of carbon and oxygen. White dwarfs form after the deaths of intermediate-mass stars, where electron degeneracy pressure supports the star against collapsing under its own gravity. However, if a white dwarf exceeds its upper mass limit—also called the Chandrasekhar mass limit (named after physicist Subrahmanyan Chandrasekhar)—this leads to nuclear reactions that cause it to explode.

Therefore, in a binary system containing a near-Chandrasekhar-mass white dwarf, mass accretion from a companion star can cause an explosion, which is one of the two proposed scenarios (the "single degenerate scenario") for Type Ia supernovae. In the other scenario, two white dwarfs are formed in a binary system (the "double degenerate scenario"), which merge together to cause an explosion—namely, a sub-Chandrasekhar-mass explosion.

Evolution of oxygen (left) and manganese (right) in the solar neighborhood of the Milky Way Galaxy. The x-axis shows the metallicity (iron abundance relative to hydrogen), which is a proxy of time increasing from the left to right. The y-axis shows the oxygen and manganese abundances, relative to iron. The points are for the elemental abundances observed in nearby stars with high-resolution spectroscopy. From the comparison, it is found that at least 75 percent of Type Ia supernovae are near-Chandrasekhar mass ex-plosions. Credit: The Astrophysical Journal

To investigate both cases, the research team run detailed calculations (2-dimensional hydrodynamical simulations and nucleosynthesis) of both near-Chandrasekhar-mass and sub-Chandrasekhar-mass explosions, and calculated the evolution of the Milky Way Galaxy, something that had not been done in previous research.

"Between these two cases, we find a critical difference in the evolution of elemental abundances, in particular for the element manganese,'' Kobayashi explained. In the first simulation, the explosion provided high-temperature and high-density matter where a lot of manganese was produced, while in the second simulation, there was no such matter and hence not enough manganese was produced.

The research team then incorporated the production amount of each chemical element into their galaxy model to predict the evolution of elements in the Milky Way. Compared to observational data, namely, elemental abundances measured in nearby stars with high-resolution spectroscopy, they found that at least 75 percent of Type Ia supernovae are near-Chandrasekhar mass explosions. In both cases, the research found, the produced iron mass is roughly the same—that is, 60 percent of the mass of the Sun—which is about 10 times larger than in normal supernovae from massive stars.

"The chemical evolution of galaxies is powerful for solving long-standing problems in nuclear astrophysics. Not only manganese but also nickel abundances are updated in our calculations with the latest nuclear reactions. Nickel was overproduced in previous calculations, but now the predicted abundance is consistent with observations,'' Kobayashi added. As a result of their findings, the nickel overproduction problem is finally solved, after two decades of studies.

More interestingly, the research team also showed that a larger contribution from sub-Chandrasekhar-mass explosions is preferred to near-Chandrasekhar-mass explosions from the available observations in different galaxies—dwarf spheroidal galaxies around the Milky Way, for example.

Kobayashi and her team noted that the elemental abundances of millions of stars will be obtained with ongoing and future international projects, such as APOGEE (Apache Point Observatory Galactic Evolution Experiment), HERMES-GALAH (GALactic Archeology with HERMES), WEAVE (WHT Enhanced Area Velocity Explorer), 4MOST (4-meter Multi-Object Spectroscopic Telescope), MSE (The Maunakea Spectroscopic Explorer), in the new research area of "Galactic Archeology," or the study of the history of the Milky Way Galaxy, and their findings will be tested further in future research.

## A Rare Double-Detonation Supernova Caught in the Act

There’s more than just one way for a star to explode. Supernovae — perhaps the most dramatic form of star death — come in many flavors, and astronomers are still learning about the vast diversity of these stellar explosions.

#### When Stars Steal Mass

This artist’s rendering depicts one kind of Type Ia supernova mechanism: the singly degenerate model, in which a white dwarf siphons mass from its companion, exceeds the Chandrasekhar mass, and explodes. [NASA/CXC/M. Weiss]

However, this isn’t the only way a Type Ia supernova can happen. In the double-detonation model, the explosion of the white dwarf is triggered by the ignition of an accreted helium shell. In this case, the white dwarf can be far less massive than the Chandrasekhar limit, leading to unexpectedly dim explosions.

Past studies have explored the minimum helium shell mass necessary (

0.01 solar mass) for this process and found that helium-shell detonations can efficiently cause core detonations, but there’s still plenty we don’t know about these events. The best way to learn about supernovae — double-detonation or otherwise — is to spot them soon after they happen.

A comparison of ZTF 18aaqeasu’s optical light curve (red circles) to normal (orange hexagons) and sub-luminous Type Ia supernovae. [Adapted from De et al. 2019]

#### A Survey Spies a Supernova

In May 2018, an unusual supernova was detected by the Zwicky Transient Facility, an optical survey that hunts for fleeting events like stellar flares, fast-rotating asteroids, and the visible-light counterparts of gravitational-wave events. Within days of its detection, a team led by Kishalay De (Caltech) began to collect photometric observations and spectra of the object.

The photometry revealed that the object, ZTF 18aaqeasu, was unusually red and less luminous than a typical Type Ia supernova, making it a good candidate for the double-detonation scenario.

Its spectra were unusual even for a sub-luminous supernova, taking much longer to develop the silicon absorption feature typically seen in this type of event. Even stranger, the spectra exhibited a never-before-seen cutoff in the flux at short wavelengths, likely due to the presence of metals like iron and titanium.

Comparison of observed spectra (black) to helium-shell double-detonation models (green and orange). [Adapted from De et al. 2019]

#### An Unusual Event

In order to derive the properties of ZTF 18aaqeasu, De and collaborators compared their photometric and spectroscopic data to models, finding that the event was likely caused by the ignition of a 0.15 solar mass helium shell, which led to the explosion of a 0.76 solar mass white dwarf.

The combination of a massive helium shell with a low-mass white dwarf makes ZTF 18aaqeasu unique among Type Ia supernovae SN 2016jhr (one of the only supernovae previously linked to a helium-shell detonation event) featured a much more massive white dwarf with a less massive helium shell.

Can we expect to find more supernovae like ZTF 18aaqeasu? Similarly luminous supernovae should be detectable out to about 1.3 billion light-years, but so far there have been none reported with similar spectral features and unusually red color. This may indicate that double-detonation events featuring massive helium shells might be rare — adding an elusive new member to the Type Ia supernova family.

#### Citation

“ ZTF 18aaqeasu (SN2018byg): A Massive Helium-shell Double Detonation on a Sub-Chandrasekhar-mass White Dwarf,” Kishalay De et al 2019 ApJL 873 L18. doi:10.3847/2041-8213/ab0aec

## Unusual type Ia supernova reported in Nature this week

The large japanese Subaru telescope in Hawaii is equipped with a superfast supernova hunting instrument MUSSES - MUlti-band Subaru Survey for Early-phase SNe Ia”. In this new Nature-paper the main author Ji-an Jiang of University of Tokyo with a long list of international cooperators describe the discovery of an unususal variety of the elsewise rather ususal supernova type Ia.

It is known that a type Ia supernova explosion occurs in a binary star system, but the excact process has been debated for years. The two main theories concern either a merger of two white dwarves late in their life or the influx of stellar gases from a larger companion star to a close white dwarf until it reaches the critical chandrasekar limit, causing it to become unstable and explode. Possibly both types of explosions are possible. The discovery with MUSSES of a type Ia supernova only one day after it occurred has shed new light on the controversy. Because of the very early discovery of the explosion, it has been possible to follow the development of the process in fine details, and it seems that a third explanation may be possible.

The energy output from the supernova shows an unexpected sudden high rise a few days after the explosion. In the paper the authors suggest that this can be caused by a sudden influx of helium onto the surface of a white dwarf, causing a detonation on the surface. The pressure waves from this nuclear explosion then propagates to the carbon in the core of the star, causing it to detonate as a supernova.

## What Triggers a Type Ia Supernova? Chandra Finds New Evidence

What makes a star go boom? A new look at Tycho’s supernova remnant by the Chandra X-ray telescope has supplied astronomers with previously unseen evidence for what could trigger specific type of supernova, a Type Ia supernova explosion. Astronomers have spotted what appears to be material that was blasted off a companion star to a white dwarf when it exploded, creating the supernova seen by Danish astronomer Tycho Brahe in 1572. There is also evidence that this material blocked the explosion debris, creating an “arc” and a “shadow” in the supernova remnant.

There are two main types of supernovae. One is where a massive star – much bigger than our sun — burns all its nuclear fuel and collapses in on itself, which ignites a supernova explosion. Type Ia supernovae, however, are different. Smaller stars eventually turn into white dwarfs at the end of their lives, becoming an ultra-dense ball of carbon and oxygen about the size of the Earth, with the mass of our Sun. In some instances, though, a white dwarf somehow ignites, creating an explosion so bright that it can be seen billions of light years away, across much of the Universe. But astronomers really haven’t understood what causes these explosions to start.

There are a couple of popular theories: one scenario for Type Ia supernovas involves the merger of two white dwarfs. In this case, no companion star or evidence for material blasted off a companion should exist. In the other theory, a white dwarf pulls material from a “normal,” or Sun-like, companion star until a thermonuclear explosion occurs.

Both scenarios may actually occur under different conditions, but the latest Chandra result from Tycho supports the latter one.

This is an artist's impression showing an explanation from scientists for the origin of an X-ray arc in Tycho's supernova remnant. Credit: NASA/CXC/M.Weiss

The new Chandra images show the famous leftovers of Tycho’s supernova, and reveal for the first time an arc of X-ray emission within the supernova remnant. The shape of the arc is different from any other feature seen in the remnant. This supports the conclusion that a shock wave created the arc when a white dwarf exploded and blew material off the surface of a nearby companion star.

In addition, this new study seems to show how resilient some stars can be, as the supernova explosion appears to have blasted very little material off the companion star. Previously, studies with optical telescopes have revealed a star within the remnant that is moving much more quickly than its neighbors, hinting that it could be the missing companion.

“It looks like this companion star was right next to an extremely powerful explosion and it survived relatively unscathed,” said Q. Daniel Wang of the University of Massachusetts in Amherst, a member of the research team whose paper will appear in the May 1st issue of The Astrophysical Journal. “Presumably it was also given a kick when the explosion occurred. Together with the orbital velocity, this kick makes the companion now travel rapidly across space.”

This image shows iron debris in Tycho's supernova remnant. The site of the supernova explosion is shown, as inferred from the motion of the possible companion to the exploded white dwarf. The position of material stripped off the companion star by the explosion, and forming an X-ray arc, is shown by the white dotted line. This structure is most easily seen in an image showing X-rays from the arc's shock wave. Finally, the arc has blocked debris from the explosion creating a "shadow" in the debris between the red dotted lines, extending from the arc to the edge of the remnant. Credit: NASA/CXC/Chinese Academy of Sciences/F. Lu et al.

Using the properties of the X-ray arc and the candidate stellar companion, the team determined the orbital period and separation between the two stars in the binary system before the explosion. The period was estimated to be about 5 days, and the separation was only about a millionth of a light-year, or less than a tenth the distance between the Sun and the Earth. In comparison, the remnant itself is about 20 light-years across.

Other details of the arc support the idea that it was blasted away from the companion star. For example, the X-ray emission of the remnant shows an apparent “shadow” next to the arc, consistent with the blocking of debris from the explosion by the expanding cone of material stripped from the companion.

“This stripped stellar material was the missing piece of the puzzle for arguing that Tycho’s supernova was triggered in a binary with a normal stellar companion,” said Fangjun Lu of the Institute of High Energy Physics, Chinese Academy of Sciences in Beijing. “We now seem to have found this piece.”

Because Type Ia supernova are all of similar brightness, they are used as a standard candle to measure the expansion of the Universe, and this new observation by Chandra has helped to answer at least part of the long-standing – and critical — question of what triggers these bright explosions.