In this paper, we present the first study of stripped-envelope core-collapse supernovae (SN Ibc) host environments based completely on SNe discovered by untargeted optical transient searches. Previous studies of this type have drawn primarily from galaxy-targeted SN searches, which biases the host environment population towards high-luminosity, high-metallicity galaxies. We present new spectroscopy of ~60 host environments in total, including metallicity measurements and age estimates for young stellar populations. We find that the metallicity distribution of SN Ib and Ic are not statistically different, but SN Ic-BL come from significantly lower-metallicity environments. Synthesizing all past studies, we show that differences in these distributions are obscured among SNe discovered by galaxy-targeted surveys, which offer a ~40% smaller baseline for measuring metallicity differences. These observations question the role of metal-line driven winds in stripping the progenitor stars of Type Ibc SNe and underscore the necessity of using untargeted transient searches to uncover the properties of their host environment populations.
We present new spectroscopy of ~60 SN Ibc host environments and measure metallicities at the explosion site (~75%) or, if necessary, at the host galaxy nucelus (~25%). Analyzing the full metallicity distribution, we find that the SN Ib and Ic distributions are not significantly different, with median metallicities the same within ~0.1 dex, nor are the SN Ib and IIb distributions statistically different. However, the SN Ic-BL are found at significantly lower metallicities.
Combining all previous spectroscopic studies of SN Ibc host environments, we present the metallicity distribution for ~170 SN Ibc in total and look for differences by SN type. The combined sample does not reveal a signficant difference between SN Ib and Ic or IIb, but reinforces the finding of lower metallicities for SN Ic-BL. The figures below illustrate the metallicity distributions from the combined sample in continuous and cumulative form, with various cuts on the sample.
A condensed, 4-panel version of this cumulative distribution figure appears in the paper:
We discuss the implications for these observations for SN progenitor models. The non-detection of a statistically significant difference between SN Ib and Ic places a strong limit on the role of metal-line driven winds in the stripping of their progenitor stars, suggesting (possibly binary evolution) may play more important roles. The lower metallicities for SN Ic-BL, as uncovered by untargeted SN searches, yields closer agreement with the progenitors of GRB-SNe and constrains the role of metallicity in the production of GRB central engines.
We use the equivalent with of Hβ to estimate the age of the young stellar population of the host environment of the SNe. The sample size for these age measurements is not sufficient to identify statistically significant differences, but the results are suggestive of lower median ages for SN Ic-BL and IIb relative to SN Ib and Ic, which would be consistent with higher mass progenitors for those explosion classes.
We perform a detailed analysis of the systematics that may influence studies of SN Ibc host environments, identifying 5 major classes of effects: 1) the effect of galaxy-targeted SN searches, 2) the ability to isolate the SN explosion site, 3) uncertainties in SN classification, 4) selection effects in spectroscopic follow-up of optical transients, 5) depth limits for host environment spectorscopy.
We find that the dominant systematic effect is introduced by galaxy-targeted SN searches, which substantially under-represent low-metallicity host galaxies and therefore provide a ~40% smaller baseline for recovering differences in metallicity distributions. Moreover, we show that mixing unequal samples of targeted and untargeted SNe of different types can bias observational results.
The ability to isolate the SN explosion site is controlled primarily by two factors: redshift, which sets the physical resolution, and spectroscopic slit placement (at the galaxy nucleus or explosion site). The figure below illustrates that there is no evidence in the combined dataset that physical resolution is the limiting factor obscuring a large (e.g. the 0.2 dex model shown) difference between the metallicities of SN Ib and Ic. A separate analysis of the role of slit placement suggests that nuclear spectroscopy introduces only a small bias that should have a minimal effect on the ability to distringuish SN Ib and Ic. However, the additional scatter introduced by photometric estimates of galaxy metallicity requires a significantly larger sample size to detect a small difference in metallicity distributions.
Simulating observational surveys of SN Ibc host environments, we find that a sample at least twice as large as the combination of all host environment spectroscopy to date would be required to unambiguously confirm a small difference (~0.1 dex in the median) between the metallicity distributions of SN Ib and Ic. The range of KS test p-values recovered by surveys of different sizes given a certain difference in the intrinsic metallicity distributions is illustrated in this figure: