Discovery of 90 Type Ia supernovae among 700,000 Sloan spectra: the Type-Ia supernova rate versus galaxy mass and star-formation rate at redshift ~0.1

Using a method to discover and classify supernovae (SNe) in galaxy spectra, we find 90 Type Ia SNe (SNe Ia) and 10 Type II SNe among the ~700,000 galaxy spectra in the Sloan Digital Sky Survey Data Release 7 that have VESPA-derived star-formation histories (SFHs). We use the SN Ia sample to measure SN Ia rates per unit stellar mass. We confirm, at the median redshift of the sample, z = 0.1, the inverse dependence on galaxy mass of the SN Ia rate per unit mass, previously reported by Li et al. (2011b) for a local sample. We further confirm, following Kistler et al. (2011), that this relation can be explained by the combination of galaxy “downsizing” and a power-law delay-time distribution (DTD; the distribution of times that elapse between a hypothetical burst of star formation and the subsequent SN Ia explosions) with an index of -1, inherent to the double-degenerate progenitor scenario. We use the method of Maoz et al. (2011) to recover the DTD by comparing the number of SNe Ia hosted by each galaxy in our sample with the VESPA-derived SFH of the stellar population within the spectral aperture. In this galaxy sample, which is dominated by old and massive galaxies, we recover a “delayed” component to the DTD of 4.5 +/- 0.6 (statistical) +0.3 -0.5 (systematic) X 10^-14 SNe Msun^-1 yr^-1 for delays in the range > 2.4 Gyr. The mass-normalized SN Ia rate, averaged over all masses and redshifts in our galaxy sample, is R(Ia,M,z=0.1) = 0.10 +/- 0.01 (statistical) +/- 0.01 (systematic) SNuM, and the volumetric rate is R(Ia,V,z=0.1) = 0.247 +0.029 -0.026 (statistical) +0.016 -0.031 (systematic) X 10^-4 SNe yr^-1 Mpc^-3. This rate is consistent with the rates and rate evolution from other recent SN Ia surveys, which together also indicate a ~t^-1 DTD.

💡 Research Summary

In this paper the authors present a novel spectroscopic search for supernovae (SNe) within the Sloan Digital Sky Survey Data Release 7 (SDSS DR7) galaxy sample, ultimately identifying 90 Type Ia supernovae (SNe Ia) and 10 Type II events among roughly 700,000 galaxy spectra. The key to the analysis is the use of VESPA (the VErsatile SPectral Analysis code) to reconstruct detailed star‑formation histories (SFHs) for each galaxy in three age bins: 0–0.42 Gyr, 0.42–2.4 Gyr, and >2.4 Gyr. By reversing the aperture correction applied by VESPA, the authors obtain the stellar mass actually probed by the 3‑arcsecond SDSS fiber, which is the region effectively monitored for transient light.

The detection pipeline first corrects each spectrum for Galactic extinction, masks bad pixels, and subtracts a galaxy model built from ten SDSS eigenspectra using singular‑value decomposition (SVD). Spectra with a reduced χ² > 1 are discarded as low‑S/N or lacking a transient. The residual is flattened, and a feature‑counting algorithm selects candidates that contain at least ten contiguous features each wider than 30 pixels. For each candidate the residual is refitted with a set of supernova templates drawn from the SNID library, but only those covering at least 2 900 Å are retained to avoid bias from narrow‑range templates. A merit function χ²_λ (χ² divided by the wavelength coverage of the template) is used to rank matches; Ia‑type events are required to have χ²_λ < 0.5, while core‑collapse types have analogous thresholds. Simulations with injected artificial SNe show an overall detection efficiency of ~70 % for SNe Ia and a purity exceeding 90 %.

The host galaxies are classified by stellar mass and specific star‑formation rate (sSFR) into passive (log sSFR < −12), star‑forming (−12 < log sSFR < −9.5), and highly star‑forming (log sSFR > −9.5) categories. The sample is dominated by massive, old (passive) galaxies (≈55 % of the total), with only 0.8 % of galaxies being highly star‑forming. Correspondingly, 47.8 % of the SNe Ia occur in passive hosts and 52.2 % in star‑forming hosts; virtually none are found in the highly star‑forming subset, reflecting the limited fiber coverage of the outer, younger regions.

Using the VESPA‑derived SFHs, the authors compute the SN Ia rate per unit stellar mass (SNuM = 10⁻¹² SNe yr⁻¹ M_⊙⁻¹). The mass‑normalized rate averaged over the whole sample is R_Ia,M(z = 0.1) = 0.10 ± 0.01(stat) ± 0.01(sys) SNuM. This reproduces the previously reported “rate‑size” relation: lower‑mass galaxies have a higher SN Ia rate per unit mass than massive galaxies. Converting to a volumetric rate yields R_Ia,V(z = 0.1) = 2.47 × 10⁻⁵ SNe yr⁻¹ Mpc⁻³, in agreement with other recent SN Ia surveys.

To probe the delay‑time distribution (DTD), the authors apply the method of Maoz et al. (2011), directly comparing the number of SNe Ia observed in each galaxy with its three‑bin SFH. They recover a “delayed” DTD component (t > 2.4 Gyr) of (4.5 ± 0.6 (stat) +0.3 − 0.5 (sys)) × 10⁻¹⁴ SNe M_⊙⁻¹ yr⁻¹. This normalization, together with the observed mass‑dependence of the SN Ia rate, is consistent with a power‑law DTD ∝ t⁻¹, the functional form expected for double‑degenerate (DD) progenitor scenarios where the merger time distribution of white‑dwarf binaries follows a similar power law. The “prompt” component (t < 420 Myr) is poorly constrained in this work because the sample contains few highly star‑forming galaxies where such short‑delay events would dominate.

Overall, the study demonstrates that spectroscopic surveys can serve as an efficient, independent avenue for discovering and characterizing supernovae, providing simultaneous access to the host galaxy’s stellar population via the same data. The authors’ results reinforce the view that the bulk of SNe Ia arise from progenitors with a t⁻¹ DTD, favoring the DD channel, and they illustrate how detailed SFH reconstructions enable direct DTD measurements. The methodology is readily applicable to upcoming massive spectroscopic campaigns (e.g., DESI, 4MOST), promising larger, higher‑redshift SN samples and tighter constraints on the progenitor physics of Type Ia supernovae.