A radio-detected type Ia supernova with helium-rich circumstellar material

Abstract

Type Ia supernovae (SNe Ia) are thermonuclear explosions of degenerate white dwarf stars destabilized by mass accretion from a companion star¹, but the nature of their progenitors remains poorly understood. A way to discriminate between progenitor systems is through radio observations; a non-degenerate companion star is expected to lose material through winds² or binary interaction³ before explosion, and the supernova ejecta crashing into this nearby circumstellar material should result in radio synchrotron emission. However, despite extensive efforts, no type Ia supernova (SN Ia) has ever been detected at radio wavelengths, which suggests a clean environment and a companion star that is itself a degenerate white dwarf star^4,5. Here we report on the study of SN 2020eyj, a SN Ia showing helium-rich circumstellar material, as demonstrated by its spectral features, infrared emission and, for the first time in a SN Ia to our knowledge, a radio counterpart. On the basis of our modelling, we conclude that the circumstellar material probably originates from a single-degenerate binary system in which a white dwarf accretes material from a helium donor star, an often proposed formation channel for SNe Ia (refs. ^6,7). We describe how comprehensive radio follow-up of SN 2020eyj-like SNe Ia can improve the constraints on their progenitor systems.

Fig. 1. The first spectrum of SN 2020eyj is consistent with a type Ia(–CSM).

The SEDM classification spectrum of SN 2020eyj, obtained about 12 days after peak and shown in black, is compared with type Ia-91T SN 2001V, type Ia–CSM PTF11kx, type Ia SN 2004eo and type Ic SN 1994I. Phases are relative to peak, which—in the case of SN 2020eyj—has an uncertainty of a couple of days. Several important absorption features are indicated at the expected wavelengths. Notably, the spectrum of SN 2020eyj lacks any sign of O i 7,774 Å absorption. Spectra have been corrected for MW reddening. Telluric features are indicated by crossed circles.

Fig. 2. The spectra of SN 2020eyj in the tail phase are dominated by CSM interaction.

The spectra of SN 2020eyj at late phases (in black) are compared with the prototypical type Ibn SN 2006jc and the type Ia–CSM SN PTF11kx. The spectra show features common to SNe Ia–CSM, such as the quasi-continuum blueward of 5,700 Å and broad Ca ii emission. The main SN emission features are identified in the top spectrum. The emission lines in SN 2020eyj show strong asymmetry, with attenuated red wings (Extended Data Fig. 3). The bottom spectrum is of the host of SN 2020eyj, obtained at 679 days, some 300 days after the SN had faded below the detection limit of the ZTF. Some unresolved galaxy lines are marked. Phases are relative to first detection, which—in the case of SN 2006jc—was at or after the peak. Spectra have been corrected for MW reddening. Telluric features are indicated by crossed circles.

Fig. 3. The multiband light curve of SN 2020eyj can be divided into a diffusion-peak phase and a long-lived interaction-powered tail phase.

The light curves of SN 2020eyj are shown with overplotted SN Ia template fits to the initial peak (see ‘Light-curve analysis’ section in Methods). The most recent mid-infrared epoch (W1 and W2) is outside the date range plotted here and is shown in Extended Data Fig. 2 . Open circles indicate synthetic photometry derived from the spectra. Phase is in rest-frame days since first detection. Apparent magnitudes on the left y axis, absolute magnitudes on the right y axis, in which μ is the distance modulus. Non-detections with 5σ upper limits are indicated by triangles. The photometry has been binned into one-night bins and has been corrected for MW reddening. The diamond markers at the top indicate the epochs of spectroscopy. The bottom panel shows the g − r colour for the nights in which both g and r photometry was obtained, overplotted with the colour evolution of a typical SN Ia. The error bars represent 1σ uncertainties.

Fig. 4. The radio detections of SN 2020eyj at 5.1 GHz can be reconciled with CSM interaction.

For the wind model, in which the CSM follows a density profile of ρ ∝ r⁻², we assume a pre-SN wind velocity of 1,000 km s⁻¹ and adopt a mass-transfer rate as inferred from fitting the bolometric light curve of SN 2020eyj. Depending on the level of line-of-sight extinction affecting the bolometric light curve (see ‘Bolometric light curve’ section in Methods), the wind model fits the observations (in black, with 1σ uncertainties) well for the microphysics parameter ϵ_B = 1.7 × 10⁻³ (1.5 × 10⁻⁵) and a CSM mass of M_csm = 0.3 M_⊙ (1 M_⊙) within 10¹⁷ cm (see ‘Optically thick wind’ section in Methods) when E(B − V) = 0 mag (0.5 mag). For the shell model, in which the CSM is concentrated in a constant-density CSM shell, we assume ϵ_B = 0.1 and obtain a best estimate for the CSM mass of M_csm = 0.36 M_⊙ and a CSM interaction end time of t_end = 665 days (a width of 8.6 × 10¹⁶ cm; see ‘CSM shells’ section in Methods). In both the wind and shell model fits, ϵ_e = 0.1 is assumed. We also show radio light curves from a model involving a DD SN Ia interacting with the ISM (see ‘ISM’ section in Methods). To fit the individual radio detections, this model requires unusually high ISM densities and neither fit reproduces the observed decline in flux, ruling out the DD scenario.

Extended Data Fig. 1. The light curves of SN 2020eyj are consistent with a SN Ia and its H-rich analogue SN Ia-CSM PTF11kx.

We simultaneously fit the g, r and i light curves of the initial peak phases of both SN 2020eyj and PTF11kx with the SN Ia light curve fitter SNooPy. SN 2020eyj is well fit with stretch factor 1.2 and E(B − V) = 0.5 ± 0.1 mag. Similarly, PTF11kx is well fit with stretch factor 1.2 and E(B − V) = 0.25 ± 0.02 mag. Panels show the absolute-magnitude light curves of SN 2020eyj and PTF11kx, after correcting for the host extinction derived from the fit. a, g band. b, r band for SN 2020eyj and r/R band for PTF11kx. c, i band. Open circles indicate synthetic photometry derived from the spectra. The error bars represent 1σ uncertainties.

Extended Data Fig. 2. SN 2020eyj was accompanied by a bright mid-infrared counterpart.

a, A co-added image of the last NEOWISE-R epoch before the SN explosion, without any sign of the SN host. b, The co-added image in the W1 filter of the November 2020 NEOWISE-R epoch, 261 days after first detection, with SN 2020eyj clearly visible. c, A mid-infrared light-curve comparison of SN 2020eyj in the W2 filter (4.6 μm) to a sample of SNe observed with Spitzer at 4.5 μm, adapted from ref. , including Type IIn SNe (in black), Type Ia-CSM SNe (in red) and Type Ibn SN 2006jc (in lilac). Furthermore, the light curves of a sample of SNe Ia from ref.  is plotted in green. SN 2020eyj (large pink circles) is among the brightest SNe observed in the mid-infrared and is 6–10 magnitudes brighter than normal SNe Ia, depending on the phase. The error bars represent 1σ uncertainties.

Extended Data Fig. 3. The He and Hα emission-line profiles in the late spectra of SN 2020eyj show notable asymmetry.

The He i emission lines at 5,876 Å, 6,678 Å and 7,065 Å all show strong attenuation in the red wings and an apparent blue shift over time between the 131 and 251 days epochs. Such line asymmetry is commonly observed in SNe Ia-CSM (ref. ) and is interpreted as resulting from the condensation of dust in the ejecta or shocked CSM, obscuring the red wing (see ‘Dust properties’ section in Methods), but may also be a result of optical depth effects. The Hα emission line at 131 days also shows asymmetry and there is a (minor) decline in flux between the two epochs shown here. By 329 days, the Hα luminosity has dropped to the level of the line emission in the host spectrum (see ‘Optical spectroscopy’ section in Methods).

Extended Data Fig. 4. The position of the radio detection is consistent with the position of SN 2020eyj in the optical.

a, The average position of the e-MERLIN detections (black circle, 0.01″ uncertainty), the position reported in GaiaAlerts (G band, green circle, 0.06″ uncertainty) and the position of SN 2020eyj in the ALFOSC epoch at 382 days (r band, red circle, 0.1″ uncertainty), overlaid on a 4″ × 4″ Pan-STARRS1 i-band dataset of the host. b, A 3′ × 3′ colour composite image, obtained with NOT/ALFOSC, of the compact star-forming host galaxy of SN 2020eyj and its environment.

Extended Data Fig. 5. The spectra of SN 2020eyj do not match Type Ic or Ibn SNe at early epochs.

The SEDM classification spectrum of SN 2020eyj compared with Type Ibn SN 2006jc and Type Ic SN1994I at similar epochs, about 12 days after peak. At 15 days post-peak, SN 2006jc already showed strong He emission lines and developed the quasi-continuum typical for CSM-interaction-dominated spectra. These features are not observed in SN 2020eyj at 12 days post-peak but do become prominent at late phases (Fig. 2). At 13 days post-peak, SN 1994I has grown redder compared with its peak spectrum shown in Fig. 1 and the O i 7,774 Å absorption feature has become more prominent, whereas in SN 2020eyj, this feature is not present.

Extended Data Fig. 6. The bolometric light curve of SN 2020eyj can be described with a radioactive decay model for the peak phase and an optically thick wind for the tail phase.

For the initial SN Ia peak of SN 2020eyj, we adopt the bolometric light curve (solid lines) accompanying the SN Ia template fit to the gri photometry (see ‘Light-curve fits’ section in Methods), assuming no line-of-sight extinction (in green) and an extinction of E(B − V) = 0.5 mag (in orange). Overplotted are the associated bolometrically corrected luminosities up to 40 days. From epoch 46 days onward, the SN Ia template fit no longer accurately describes the observed (g-band) photometry (Extended Data Fig. 1). The dotted lines show the continuation of the bolometric light curve of the underlying SN Ia. The three measurements in the tail phase are based on the integration of the two Keck spectra, extrapolated to the UV, and a bolometrically corrected photometric ALFOSC epoch. The dashed lines represent the fits to the tail-phase measurements using the analytical model from ref. , following the same colour scheme for the level of extinction. In the transition region from the diffusion peak to the CSM-interaction-powered tail, between 50 and 100 days, the sum of the models would overestimate the luminosity, suggesting that the CSM configuration is more complicated than a simple wind-like density profile. The red diamonds show the infrared luminosity of SN 2020eyj (Extended Data Table 3), which are not included in the model fits.