© ESO 2022

1. Introduction

Core-collapse (CC) supernovae (SNe) are the final explosions of massive stars (≳8 M_⊙). Hydrogen-poor SNe represent CC in such stars that lost most – or even all – of their envelopes prior to explosion. This includes Type IIb SNe (some H left), SNe Ib (no H, some He), and SNe Ic (neither H nor He). We refer to these as stripped envelope (SE) SNe.

Even though SE SNe are less common than Type II SNe (e.g., Li et al. 2011; Graur et al. 2017), a fair number of well-observed objects exist. Presentations of such samples have highlighted how simple analytical models, such as the one initiated by Arnett (1982), provide reasonable matches with the observed light curves. In collecting sizeable samples, such approaches have revealed that the estimated ejecta masses are relatively low, which is often seen as an argument for binary interaction playing a major role in the stripping of the progenitor stars (Lyman et al. 2016; Taddia et al. 2015, 2018, 2019; Prentice et al. 2016, 2019; Drout et al. 2011; Barbarino et al. 2021).

The other main result from these studies is that the amount of ejected radioactive nickel is typically larger than for normal Type II SNe. The mean value from the recent sample of Type Ic SNe from the iPTF survey (Barbarino et al. 2021), for example, asserts that M_⁵⁶Ni = 0.19 ± 0.03 M_⊙.

A literature compilation by Anderson (2019) provides a calculated median of M_⁵⁶Ni = 0.032 M_⊙ for SNe II, and 0.163 and 0.155 M_⊙ for SNe Ib and Ic, respectively. That study was repeated and augmented by Meza & Anderson (2020, see also Sharon & Kushnir 2020) concluding that there is a real, intrinsic difference in the amount of radioactive nickel between SNe II and SE SNe, even if the exact numbers are sensitive to the specific methodology.

Our paper takes two modeling studies as the starting point. Exploiting state-of-the-art neutrino-driven explosion models for massive helium stars that have been evolved including mass loss, Ertl et al. (2020) note that for standard assumptions regarding the explosions and nucleosynthesis, their models predict light curves that are typically less luminous than many observed SNe Ib and Ic. Their upper limit on the peak luminosity is 10^42.6 erg s⁻¹. These authors remarked that many SNe Ibc appear to be too luminous to be made by their neutrino-driven models, proposing that magnetars could be a promising alternative to power these supernovae, rather than (or in addition to) radioactivity. Alternatively, they suggest that observers could pay more attention, for instance, to bolometric corrections, Malmquist bias, or evidence for circumstellar interaction that could overestimate the reported peak luminosities.

Following Ertl et al. (2020), Woosley et al. (2021) augmented that study by adding detailed radiation transport. Using the code SEDONA, they explored the same explosion models and could translate the limits on ejected nickel mass and bolometric luminosity to maximum light in common filter pass bands. They have no models brighter than M_r = −17.8 (or M_g > −17.5). The bottom line in Woosley et al. (2021) is that most SE SNe are best understood in the traditional sense of having “binary mass exchange, neutrino-powered explosions without rotation, and radioactivity-illuminated light curves”. Thus, these authors seem less keen to lean on the magnetar solution, even though they acknowledge that a sizeable fraction of the SE SNe might be out of reach (too bright) for their models.

Woosley et al. (2021) also occasionally discussed observational uncertainties, such as whether some specific SNe might have had their host extinction overestimated (see also Dessart et al. 2020) or whether some are really “normal” Type Ibc SNe. They also ask whether the bolometric light curves (LCs) have been improperly assembled or if overly simplistic modeling was used to derive the amount of radioactive nickel. They explicitly encourage observers to undertake new surveys and compare to their predicted pass-band LCs. Taking up that baton, our paper posits a simple goal in trying to address whether a reasonable number of well-observed normal SNe Ibc reach peak luminosities in excess of M_r = −17.8 even if carefully assessed for such parameters as distance and extinction. We also explore the caveats this type of investigation would have to consider.

We make use of the sample of SE SNe (Type Ib and Ic, collectively labeled SNe Ibc) provided by the Zwicky Transient Facility (ZTF, Graham et al. 2019; Bellm et al. 2019). In particular, Fremling et al. (2020) introduced the ZTF Bright Transient Survey (BTS), which provides a large and purely magnitude-limited sample of extragalactic transients in the northern sky, suitable for detailed statistical and demographic analysis. The early results of this survey were presented by Perley et al. (2020), while also introducing a web-based portal open to the public where specific sub samples can be constructed. We used this BTS sample explorer¹ to collect all Type Ibc SNe within the BTS. This is also an explicit purpose of this paper, namely: to advocate for the public BTS sample and to show how it can be used to address a specific scientific question.

The paper is organized as follows. In Sect. 2, we present the observations and explain the sample selection based on our optical photometry and spectroscopy. Section 3 presents a discussion of the different caveats in determining absolute magnitudes, including distances and extinction for this subsample. Finally, Sect. 4 presents our conclusions and a short discussion where we put our results in context.

2. Observations and sample

All photometric observations in this paper were conducted with the Palomar Schmidt 48-inch (P48) Samuel Oschin telescope as part of the ZTF survey, using the ZTF camera (Dekany et al. 2020). The light curves from the P48 come from the ZTF pipeline (Masci et al. 2019). All magnitudes are reported in the AB system.

The BTS SNe are regularly reported to the Transient Name Server (TNS²), and the LCs can be displayed using the above mentioned BTS sample explorer, which we used to construct our sample. We note again that the BTS is an untargeted sample of SNe that is virtually spectroscopically complete down to a magnitude of 18.5 (Perley et al. 2020).

The aim of the paper is to explore to what extent such normal³ Type Ibc SNe exist – namely, those that exceed the maximum brightness predicted by the models noted in the introduction. Our main aim is therefore not to construct a complete and non-biased sample. While such a sample would certainly be of interest in comparing the average properties of SNe Ibc with the models, it would require greater care in terms of completeness and corrections for Malmquist bias (see e.g., Ouchi et al. 2021) and extinction. We take a simpler approach, aiming at a reasonable sample (𝒪(10)) of normal SE SNe that is large enough not to be biased by statistical outliers. More explicit investigations of the sample luminosity-function, light-curve parameters, and extinction-correction properties are planned for a future work.

What is important for the selection process is the availability of enough data that are adequate for constructing the LCs, measuring the peak luminosity, and ensuring that the object is indeed a normal SN Ibc, both in terms of LC and spectra. In the first initial construction of the sample, we use the BTS explorer criteria provided in Table 1. The full BTS database included 4496 objects classified as SNe, where 3038 were SNe that passed these cuts⁴, including 218 SE SNe. The quality cuts in Table 1 ensure, for example, that our objects have accessible data from both before and after peak, and that the object was not detected too early in the survey, that is, when uncontaminated templates were not available.

From the initial list of 218 SE SNe, we meticulously excluded candidates that do not fulfill the next sets of selection criteria. Since the BTS explorer includes > 200 SNe Ibc, we can allow for rather strict cuts. These are based on data quality and are not supposed to bias the sample (any more than the requirement that the selected SNe are normal SNe Ibc). We note, in particular, that luminosity is not explicitly used in the sample cuts. We further requested that the classification Type was either Type Ib or a Type Ic. We thus removed all of the following types from the sample: Types Ic-BL, Ibn, Icn, IIb or Ib/c or Ib-pec, as well as anything labeled with a question mark. This cut excludes objects where other powering mechanism could be at play, such as shock cooling, circumstellar matter (CSM) interaction, or a central engine. The “Ib/c” class on BTS represents objects for which a separation into either Type Ib or Type Ic could not be made on the basis of the quality of the spectrum. For purity, we simply removed these from our sample as well. Finally, a few objects had different classifications on TNS as compared to our internal follow-up marshal, and we removed these as well⁵. Ultimately, this gave us 53 SNe Ib and 76 SNe Ic, or in total 129 Type Ibc SNe. The selection cuts are provided in Table 2.

This sample that fulfills our first set of BTS sample criteria is used to construct an initial luminosity function. The absolute peak luminosity function for these supernovae, with magnitudes as provided from the BTS, is presented in Fig. 1 in black full lines. These BTS absolute magnitudes are computed using the observed peak, given the observed redshift and Milky Way extinction, and applies a basic k-correction. This is already a significant result given the untargeted nature and the large size of the survey, and that the selection criteria used are mainly dependent on the data quality and cadence. The sample and the luminosity function⁶ is further refined throughout the rest of the paper.

Fig. 1. Luminosity function for Type Ibc SNe. The figure shows the number of objects per absolute magnitude bin (Mr) for different sample selections. The black distribution is for the 129 SNe Ibc initially selected from the BTS explorer and using the absolute magnitudes from that site. This distribution has an average and standard deviation of Mg/r = −17.61 ± 0.72 mag. The blue dashed distribution is for the 94 SNe Ibc that is kept after additional quality cuts have been implemented. These magnitudes are measured using GP on forced photometry data and yield Mr = −17.64 ± 0.54 mag. The red distribution of the final 14 normal SNe Ibc has an average and standard deviation of Mr = −17.90 ± 0.73 mag. The vertical black dashed line marks the upper limit of Mr = −17.8 from Woosley et al. (2021).

2.1. Photometry cuts

We proceed with those SNe that have good quality light curves. At this stage, we performed forced photometry (Masci et al. 2019; Yao et al. 2019) for the remaining SE SN subsample. For those resulting LCs, we furthermore require the following data quality cuts: i.) At least six epochs of photometry in either the g or r band; ii.) At least three epochs of g − r (sampled within ±3 days). We also require that iii.) photometry is available both before and after peak within ±3 days of estimated time of peak brightness, and iv.) that the photometry is accurate enough that we can determine the peak luminosity to better than 10% (0.1 mag).

These steps were accomplished using a Gaussian processing (GP) algorithm⁷ to interpolate the photometric data. The number of SNe that remains after each sample cut is presented in Table 2. We note again that selecting on the basis of cadence and data quality should not bias the sample in preferring some specific classes of SE SNe before others, or deselecting particular environments. There is, however, a Malmquist-like selection in that intrinsically very faint or fast transients will on average have fewer good-quality data-points. For the purpose of this study of the bright end of the luminosity function, where we want to find out if there exist luminous SNe, this is not a problem; but we note that there may exist a population of less luminous, nickel-poor SE SNe that are underrepresented or missing from this compilation. Fremling et al. (in prep.) are exploring ways to find such transients via their early shock-breakout cooling emission. The rationale for requiring two bands at this stage is that we also want to be able to construct bolometric LCs and to assess the host extinction; see Sects. 3.3 and 3.2. Only ten objects were removed in this step, mostly since we had already performed cuts on the data in the first selection (Table 1).

2.2. Distance cuts

Distances are often a major uncertainty in estimates of absolute luminosities and thus nickel masses. This is paradoxically often true for the most nearby, and therefore best observed, SNe in the literature, simply because in the local universe the peculiar motions of nearby galaxies make the relative distance uncertainties larger. To avoid SNe with large uncertainties from their distance estimates, we required for the SNe to be distant enough so as to be located in the Hubble flow (we used z > 0.015). None of the nearby hosts had a distance estimate from, for instance, Cepheids. Thus, seven rather well observed SNe⁸ were excluded from this study.

Redshifts were converted to distances using a flat cosmology with H₀ = 70 km s⁻¹ Mpc⁻¹ and Ω_m = 0.3. The rationale for this cut and the remaining uncertainties from these distance estimates to the absolute magnitudes are further discussed in Sect. 3.1.

2.3. Milky Way reddening

In our analysis, we correct all photometry for Galactic extinction, using the Milky Way (MW) color excess E(B − V)_MW toward the position of the SNe, as provided in Table 3. These were all obtained from Schlafly & Finkbeiner (2011). All reddening corrections are applied using the Cardelli et al. (1989) extinction law with R_V = 3.1. Supernovae experiencing significant amount of Galactic extinction (A_V > 1.0 mag) were already deselected in the BTS explorer search (Table 1). For this exercise, we furthermore remove SNe for which the MW extinction correction A_V > 0.5 mag, see Table 2. The argument is simply based on the fact that larger corrections imply larger uncertainties. The corrections for dust in the host galaxies is discussed below (Sect. 3.2). This process led to the removal of six objects (Table 2).

2.4. Light curve properties

To make sure we selected only normal SE SNe, since these are what we want to compare against, we made LC fits using a functional form that was also used for SNe in Taddia et al. (2015, see their Fig. 8). This was done using scipy.optimize.curve.fit and we require the fit to have χ² < 2 per degree of freedom. This step is made to avoid SNe with LC bumps, signs of CSM interaction, or, simply, poor photometry. This removed only a few SNe⁹.

In this exercise, we also used the LC fit with the analytical function to characterize the rise and decline parameters (again following the study of Taddia et al. 2015). In order to estimate the actual rise time with respect to an estimated time of explosion (first light), we followed the methodology employed by Miller et al. (2020) using both the pre-explosion upper limits and the rising part of the LC. Comparing the τ_fall vs. τ_rise distributions with those of the SDSS sample (Taddia et al. 2015) and, in particular, investigating the rise-time distribution, we decided to remove objects with τ_rise > 8 days. This effectively also removed all objects with t_rise > 35 days¹⁰. Again, the selection is carried out with the aim to focus this study on the normal population of SNe Ibc. Slow-rising SE SNe are by themselves also of large interest, in particular for understanding the population of single massive stars as progenitors, but for the scope of this investigation, such objects were de-selected.

Out of the initial 129 SE SNe, 94 remained after the above mentioned cuts. The absolute peak luminosity function for these supernovae is also presented in Fig. 1 (dashed blue lines). This is a significant contribution to the knowledge of the Type Ibc luminosity function, and the sample compares well with for example the recently published large iPTF sample of 44 SNe Ic by Barbarino et al. (2021), but with a higher degree of control on the selection functions. The results will be discussed further in the next sections, but for now, we proceed to a final culling of our sample.

2.5. Host galaxy extinction

This final cut was done to remove objects with different colors than the main population of SNe Ibc. The main rationale here being that we want to avoid large corrections for host-galaxy extinction. This is probably the greatest uncertainty that could be ingested from the observational side, over-correcting for extinction would make the SNe too luminous, which could be a reason for the apparent discrepancy between model predictions and observations.

A very red color for the MW-extinction corrected SN LC probably indicates significant host-galaxy extinction. There are a number of ways to compensate for this, as discussed in Sect. 3.2, but all of the methods come with a (fairly large) degree of uncertainty.

To exclude cases where extinction corrections would come with a large uncertainty, we simply deselected objects that are too red (g − r > [0.64 + 0.13] mag) at 10 days past peak, and also cut out objects that are significantly bluer (g − r < [0.64−0.13] mag) than the rest of the sample at this phase. These numbers and their uncertainties come from the investigation of Taddia et al. (2015) using Type Ibc SNe from SDSS. We furthermore rejected objects where the color information is simply not accurate enough to reliably perform these cuts, that is to say, we rejected any object for which we could not estimate (g − r) at 10 days past peak with an accuracy better than 0.2 mag. This is clearly one of the most severe cuts in the sample selection, removing 59+5+16 objects, leaving only 14 remaining (Table 2). The rationale for these cuts and the remaining uncertainties are further discussed in Sect. 3.2. The selection is illustrated in Fig. 2 where the gray area shows the typical colors of SE SNe at ten days past peak, g − r = 0.64 ± 0.13 mag (Taddia et al. 2015). The objects that survive this final cut are marked with black symbols. The red symbols constitute the majority of the objects, which have redder colors. The notion that they are also dimmed by extinction is supported by the fact that they are typically less luminous than the bluer SNe; there is a clear trend visible in this figure. Instead of attempting to correct for this dimming, for the final cut in this paper, we took the simple yet conservative step to remove all of these objects. We stress that this is very cautious with respect to the purpose of this study, as the red objects would only become more luminous with host extinction corrections (Sect. 3.5.2). The green symbols in Fig. 2 show the objects that were removed because the GP photometry at ∼10 days had overly large uncertainties on the color. Finally, we note a sub-population of bright and blue objects, marked with blue symbols in the upper left corner of the figure. Including these objects would also make our average SN Ibc magnitude brighter, and the required mass of radioactive material larger. Conservatively, we remove them on the basis that they do not have normal colors according to Taddia et al. (2015) and Stritzinger et al. (2018). The final selection leaves us with only 14 SNe. The properties of these SNe, with regards to the selection criteria detailed above, are provided in Table 3.

Fig. 2. Colors and color cuts for the sample selections of Type Ibc SNe. The figure shows the absolute peak magnitudes (Mr in the upper and Mg in the lower panel) for the 94 SNe selected versus their MW corrected colors in g − r at ∼10 days after peak, when these transients have the most uniform color distribution (Taddia et al. 2015; Stritzinger et al. 2018). The gray box includes those 14 SNe kept in the final sample as normal SNe Ibc where uncertainties in the extinction corrections are smaller. The color coding is explained in the main text. The horizontal dashed lines represent the maximum luminosities (Mg = −17.5, Mr = −17.8), according to Woosley et al. (2021). The data points also have uncertainties in magnitudes assigned, according to the error propagation in Sect. 3.4.

Most of the spectroscopy for the BTS is conducted with the robotic Palomar 60-inch telescope (P60; Cenko et al. 2006) equipped with the Spectral Energy Distribution Machine (SEDM; Blagorodnova et al. 2018). Further spectra were often obtained with other larger telescopes such as the Nordic Optical Telescope (NOT) using the Alhambra Faint Object Spectrograph (ALFOSC). The BTS provides all the classification spectra via the publicly available TNS. The SEDM spectra were reduced using the pipeline described by Rigault et al. (2019). We checked the spectra for these objects, and we confirm that they are all best fit with the reported subclasses of SNe.

3. Discussion

The properties of the final sample of SNe Ibc are listed in Table 3. The ZTF obtains regular photometry in g, r (and i) bands, and for these 14 SNe we have, on average, 13/14 r/g-band points between −20 and 20 days past peak, that is, a typical cadence of 3 days, although some are better sampled than 2 days. The SNe in this sample thus have relatively well-constrained explosion times, rise times and decline times, and we measure these parameters and list them in Table 4.

In Fig. 3, we show their LCs in absolute magnitudes. The magnitudes are in the AB system and have been corrected for distance modulus and MW extinction. They are plotted versus rest-frame days past estimated explosion epoch. This final absolute magnitude distribution is included in Fig. 1.

Fig. 3. Light curves of the final sample of 14 Type Ibc SNe plotted in separate panels. We plot g- (green squares) and r-band (red circles) photometry in absolute AB magnitudes. These are corrected for distance and MW extinction. The x-axis gives rest-frame days since r-band peak, where the redshifts and explosion dates are provided in Tables 3 and 4. The dashed lines are the GP interpolations with error regions that were used to estimate peak explosion magnitudes and their uncertainties. Black arrows on top indicate the dates when we obtained spectra.

Next, we briefly discuss some of the selection cuts and the corrections and their uncertainties, given the main aim of this investigation. We make an effort to quantify the uncertainties involved in the different steps to be able to propagate these to the final luminosity function.

3.1. Distance estimates

An important uncertainty when estimating absolute luminosities (and nickel masses) for SNe is the uncertainties in the distance estimates. Such uncertainties are often underappreciated in the SN literature. In particular, many studies focus on nearby objects where good data quality is easier to acquire, but where the relative uncertainties due to peculiar motions of the host galaxies can be considerable.

As an example, we mention SN 2020oi (ZTF20aaelulu), a nearby Type Ic SN that was part of our initial BTS sample of SNe Ibc. For SN 2020oi in the host galaxy M100, Horesh et al. (2020) adopted a distance of 14 Mpc, corresponding to a distance modulus of 30.72 ± 0.06 mag. For an Arnett type of model, the nickel mass basically scales linearly with peak luminosity and a distance modulus uncertainty of 0.06 mag translates to a relative uncertainty on the ejected nickel mass of 5.5%.

However, the NASA Extragalactic Database (NED¹¹) includes multiple different distance estimates for this nearby host (and many others). Following the approach of Steer (2020), an uncertainty from median combining many of those estimates would be 16.4 ± 2.35 Mpc, which would correspond to an uncertainty in the nickel mass of 29%. To illustrate this, we note that the second published paper on SN 2020oi use a distance of 16.22 Mpc (Rho et al. 2021). They quote a nickel mass with an uncertainty of 15%, but we note that the difference in distance as adopted by these two studies of the same SN amounts to a 33% difference in flux. Somewhat ironically, the studies reach similar conclusions since they also adopt different amounts of host extinction, which in this case, happens to work in the direction of decreasing the differences. We note that SN 2020oi would also have been deselected from our sample due to the large host extinction, which makes it difficult to accurately determine the intrinsic luminosity.

We note that ZTF is an untargeted survey. Therefore, in contrast to most previous samples of SE SNe, we are not biased towards nearby and massive galaxies. The redshift distribution of our (94 object) sample has a mean value and rms of z = 0.036 ± 0.003, which means that peculiar velocities for the host galaxies are of less importance. Estimating a typical peculiar velocity of 300 km s⁻¹ (Davis et al. 2011) means that for our cut-off value z = 0.015, we have an uncertainty on cz of < 7%, whereas for the mean redshift of the sample ($\overline{z}=0.036$ within the errors for the three samples), this gives a typical flux error of 3%. For our distance estimate uncertainties for the individual SNe in the final sample, we used an individual uncertainty from peculiar velocities of 150 km s⁻¹ and for the cosmology, we included a systematic uncertainty of ±3 km s⁻¹ Mpc⁻¹ on the Hubble constant (Sect. 3.4).

3.2. Host extinction

Correcting for host extinction is probably the most difficult part in determining the luminosity function for any type of SN. Barbarino et al. (2021) used two different approaches for their SN Ibc sample, both from narrow absorption lines of NaI D in the spectra and using the SN colors to correct for reddening. There are pros and cons with both of these methods, and they are certainly both affected by uncertainties. Overall, on a sample level, the main results of Barbarino et al. (2021) were not much affected by the choice of method, but for the individual SN the actual correction can vary substantially. It is widely accepted that there is some relation between deep host-galaxy sodium absorption lines and the amount of extinction, but the scatter is large and the implementations differ (e.g., Turatto et al. 2003; Poznanski et al. 2012; Blondin et al. 2009; Phillips et al. 2013). For the ZTF SNe, we have often rather low-quality spectra, and we have not adopted these methods here.

The other available methodology is based on the fact that SNe Ibc often have similar colors at some phase after peak. This was first noted by Drout et al. (2011) and implemented by Taddia et al. (2015), then further developed by Stritzinger et al. (2018). The basic assumption here is the uniformity of these events, and interpreting redder events as being affected by host galaxy extinction. The investigations of Stritzinger et al. (2018) and Taddia et al. (2015) define a range of colors for normal, un-reddened SNe Ibc, and we adopted the cuts from the latter study¹² on the uncertainties and actual colors at 10 days past peak (Fig. 2, Table 2).

However, in this paper we remain cautious on the actual and quantitative host-reddening correction. Our conservative approach is to not apply any correction for host galaxy reddening, and we simply removed the objects for which such a correction would have been needed. This culls a large fraction of our sample, but also alleviates the main problem. Figure 2 illustrates the situation, where absolute magnitudes in the r band (M_r, top) and g band (M_g, bottom) are plotted versus MW corrected colors at 10 days past peak. The black vertical line shows g − r = 0.64 mag, which is the normal unreddened color for SNe Ibc, and the gray region shows the 1σ deviation on this number (±0.13 mag) from Taddia et al. (2015). The red symbols show the large fraction of SNe that have redder colors and are therefore suspected to be affected by host galaxy reddening. These are excluded from the final sample. On the left-hand side of the gray region, there are also a number of SNe (5) that have bluer colors than the typical SN Ibc. These are marked with blue symbols and are also de-selected (Sect. 3.2, Table 2).

We note that there is indeed a correlation between absolute magnitude in these two bands and color at 10 days. The slope of the correlation is also larger in the g band, as expected if this is, in fact, primarily due to extinction by dust.

3.3. Luminosities and bolometric corrections

As a final exercise, we attempted to construct bolometric LCs for our final sample and use analytic expressions to estimate the amount of radioactive ⁵⁶Ni needed to power the peaks of these LCs. We follow the procedure outlined by Lyman et al. (2014) in order to construct the bolometric LCs from the g- and r-band LCs. This is a well-established procedure for normal Type Ib and Type Ic SNe, and we have secured that our final objects constitute such a sample. We thereafter estimated the nickel-mass following a simple Arnett model (Arnett 1982; Tartaglia et al. 2021). This provides final bolometric luminosities with corresponding nickel masses of M_Ni = 0.25 ± 0.05 M_⊙ for the sample of 14 SNe Ibc. This compares well with the values from the investigation of Barbarino et al. (2021), with M_Ni = 0.19 ± 0.03 M_⊙ for 41 SNe Ic, which used a similar approach. The larger sample of 94 SNe Ibc has a mean value of M_Ni = 0.16 M_⊙, but we note that no host-extinction corrections were applied to that sample. There is an ongoing discussion regarding the extent to which the simple models used here infer a realistic nickel mass; and alternatives have been suggested in the literature (Dessart et al. 2016; Khatami & Kasen 2019; Afsariardchi et al. 2021). This is the reason we mainly stick to the pass-band magnitudes in this observational paper, in order to make direct comparisons with the predictions from the radiation transport of Woosley et al. (2021). Work on estimating nickel masses for this sample using multiple methods is ongoing.

3.4. Error propagation

Apart from presenting mean values and rms uncertainties on the absolute magnitudes for the sample populations, we have also propagated the uncertainties for the individual objects through the different steps as outlined above. For each individual supernova, we include the photometric uncertainty on the peak magnitude as estimated from our GP analysis, a 15% uncertainty in the correction for MW extinction, a 150 km s⁻¹ uncertainty included in the peculiar velocity correction. and a systematic ±3 km s⁻¹ Mpc⁻¹ error on the adopted Hubble constant. These uncertainties are then provided as error bars on the y-axis for the black symbols in Fig. 2. These magnitude errors are mostly < 15%.

3.5. Final sample

The sample criteria so far have been strict and objective, without dwelling on any individual SN. The three sample distributions in Fig. 1 actually all have the same mean values within the errors, but the final sample is somewhat limited by statistics and four objects at M_r ∼ −19 are significantly more luminous than the model limit whereas there are no objects left in the bin between −18 and −19. Here, we briefly review first the four final objects that are substantially brighter than the model limits. Thereafter, we also take an individual look at four objects with M_r ∼ −18.5, which were de-selected due to their red colors, and discuss if these are also robustly brighter than the investigated limit magnitude.

3.5.1. The bright ones

Four objects in the final sample have absolute magnitudes brighter than −19. This is substantially more luminous than the predictions from the explosion models, and also on the bright end of the entire luminosity distribution. Although there are also several SNe robustly around −17.8, which are challenging the predictions, we first individually look at the top four. Treating samples on an overall statistical level is certainly more objective, whereas scrutinizing individual objects can illuminate some of the sample caveats. For these objects, we also display the classification spectra in Fig. 4. For most objects, we have multiple spectra to support the classification (as indicated on top in Fig. 3).

Fig. 4. Spectra for the bright and the red SNe discussed in Sects. 3.5.1 and 3.5.2. We display the classification spectrum from TNS together with a named template spectrum to illustrate the quality of our spectra. For most objects we also have additional spectra (see Fig. 3), and we provide some of these here as well. All these spectra are made available on Wiserep. Our spectra are shown in red, where the black line is a smoothed version. The name of the used telescope is also provided. The classification spectrum for SN 2019pfp is from Terreran (2019) and the one from SN 2020abqx from Burke et al. (2021).

SN 2019ieh/ZTF19abauylg. This SN has a well-monitored LC and the redshift is secure from host galaxy emission lines in the SN spectrum. The classification spectrum is from P60 and is shown in Fig. 4. A better spectrum was obtained with NOT and is best fit by a Type Ic template using SNID (Blondin et al. 2009). The same is true for an earlier Lick spectrum, although templates with SNe Ic-BL are also viable fits at that phase.

SN 2019lfj/ZTF19abfiqjg. The redshift is secure from the host galaxy spectrum (SDSS). The LC is well sampled in the r band and peaks above −19; it is poorly matched with SN Ia LCs using SALT2 (Guy et al. 2007). The classification is based on a single host-contaminated P60 spectrum. The best SNID fit template is the Type Ic SN 2004aw.

SN 2019qvt/ZTF19abztknu. The redshift in this case also comes from host lines in late SN spectra. The LC is well sampled and the Type Ib classification is secure from He lines in later spectra. In Fig. 4 we display both the classification spectrum obtained with the Liverpool Telescope, as well as a later NOT spectrum, together with matched Type Ib template spectra.

SN 2020abqx/ZTF20acvebcu. For this SN Ib, the classification spectrum from Burke et al. (2021) includes also galaxy lines, securing the redshift (host z also known from SDSS). Also SNID finds good matches with a SN Ib at this redshift (Fig. 4). The classifiers note that the He Iλ5876 is particularly strong. We do not have a detailed spectroscopic sequence to establish the classification further. The LC is not well fit with a Type Ia SN LC with a secondary r-band bump.

These are all thus clearly luminous supernovae with secure peak photometry and redshifts. Some of the objects have classifications based on low-resolution and mediocre signal-to-noise spectra from robotic telescopes, where the potential confusion could come from peculiar SNe Ia or Type Ic-BL SNe.

3.5.2. The red ones

We also discuss four objects with brightness significantly above the theoretical limit, but which were excluded from the final sample because they were slightly too red, as shown in Fig. 2¹³. Looking also at these individual objects, we conclude that they are all positioned in (projected on) large star-forming galaxies, which is consistent with them suffering from some extinction. For three out of four, there were no previous host galaxy redshifts, but our later spectra secure these from host galaxy lines. Again, the light curves are well sampled and there are no doubts with regard to their redshifts or photometry. In all cases, early robotic spectra are complemented with later spectroscopy from larger facilities, and also here we have no reason to reclassify any of the objects. Some of these spectra are also displayed in Fig. 4, where we show the classification spectra from TNS and the closest template according to SNID. In some of these cases, there are also earlier P60 spectra, but these were not of high-enough quality to warrant a secure classification. These are thus SE SNe brighter than the investigated limits, and any corrections for host galaxy extinction would only make them even brighter.

The main conclusion based on our investigation of these individual SNe is that for some of the objects, the exact sub-classifications might be questioned, but that, overall, we often have multiple spectra and supporting observations also from larger facilities. Redshifts derived from the supernova features may come with larger uncertainties, but for the objects investigated in Sects. 3.5.1 and 3.5.2, all redshifts were well established from host emission lines. There are thus clearly normal SE SNe that reach above the brightness limits investigated in this study. It is also clear that many of the red objects are normal SNe Ibc that are already brighter than the limit. For the sample of 94 objects, there were 29 such SNe (31%). Correcting for host extinction only makes that large sample brighter.

4. Summary and conclusions

In this paper, we present a selection of SE SNe from the BTS sample. Starting with 129 selected Type Ib and Type Ic SNe from the BTS, we present an initial luminosity distribution for these objects, as shown in Fig. 1. The mean absolute magnitude and the rms for this distribution is M_g/r = −17.61 ± 0.72, and 36% of the SNe appear brighter than the limit of −17.8 that Woosley et al. (2021) suggested as the upper limit on the brightness from their radiation transport calculations (based on state-of-the-art explosion models). This already supports previous studies reporting large luminosities and nickel masses for Type Ibc SNe.

A main driver in this paper has been to use the well-characterized BTS sample together with strict selection cuts to weed out the normal SNe Ibc. One of the largest cuts in the selection of the final sample was based on the colors of the SNe. This is discussed in Sect. 3.2 and illustrated in Fig. 2. Correcting for extinction would make the red objects to the right even more luminous, further amplifying the discrepancy between the model predictions and the observed luminosity function. Several of these bright and red objects are clearly SE SNe more luminous than the theoretical cut (Sect. 3.5.2). We also note the objects marked in blue that we also de-selected from the sample. The rationale for omitting these objects was not that they are affected by dust, but merely that they are outside the region of normal SN Ibc colors. It is noteworthy that they are all more luminous than M_r = −17.8, and including some of these objects would clearly push the luminosity function to even brighter magnitudes. Similarly, declaring some of them as normal, un-distinguished SNe would effectively push the black vertical line to the left, and also make the final sample more luminous.

We used the ZTF BTS sample and a series of selection criteria to investigate whether normal SE SNe can be more luminous than M_r = −17.8. We find that they certainly can. This puts the ball back in the court of the theoretical models, implying modifications to either: fundamental core-collapse explosion models, alternative powering mechanisms (such as magnetars), more sophisticated radiative transport schemes to translate bolometric luminosities to pass-band limits, or probably a combination of these.

Acknowledgments

We thank Schuyler van Dyk for comments. Thanks also to the referee for encouraging comments. Based on observations obtained with the Samuel Oschin Telescope 48-inch and the 60-inch Telescope at the Palomar Observatory as part of the Zwicky Transient Facility project. ZTF is supported by the National Science Foundation under Grant No. AST-2034437 and a collaboration including Caltech, IPAC, the Weizmann Institute for Science, the Oskar Klein Center at Stockholm University, the University of Maryland, Deutsches Elektronen-Synchrotron and Humboldt University, the TANGO Consortium of Taiwan, the University of Wisconsin at Milwaukee, Trinity College Dublin, Lawrence Livermore National Laboratories, and IN2P3, France. Operations are conducted by COO, IPAC, and UW. SED Machine is based upon work supported by the National Science Foundation under Grant No. 1106171. The ZTF forced-photometry service was funded under the Heising-Simons Foundation grant 12540303 (PI: Graham). This work was supported by the GROWTH project (Kasliwal et al. 2019) funded by the National Science Foundation under PIRE Grant No 1545949. The Oskar Klein Centre was funded by the Swedish Research Council. Gravitational Radiation and Electromagnetic Astrophysical Transients (GREAT) is funded by the Swedish Research council (VR) under Dnr 2016-06012. Partially based on observations made with the Nordic Optical Telescope, owned in collaboration by the University of Turku and Aarhus University, and operated jointly by Aarhus University, the University of Turku and the University of Oslo, representing Denmark, Finland and Norway, the University of Iceland and Stockholm University at the Observatorio del Roque de los Muchachos, La Palma, Spain, of the Instituto de Astrofisica de Canarias. Some of the data presented here were obtained with ALFOSC. M. M. K. acknowledges generous support from the David and Lucille Packard Foundation.

References

All Tables

All Figures

Fig. 1. Luminosity function for Type Ibc SNe. The figure shows the number of objects per absolute magnitude bin (Mr) for different sample selections. The black distribution is for the 129 SNe Ibc initially selected from the BTS explorer and using the absolute magnitudes from that site. This distribution has an average and standard deviation of Mg/r = −17.61 ± 0.72 mag. The blue dashed distribution is for the 94 SNe Ibc that is kept after additional quality cuts have been implemented. These magnitudes are measured using GP on forced photometry data and yield Mr = −17.64 ± 0.54 mag. The red distribution of the final 14 normal SNe Ibc has an average and standard deviation of Mr = −17.90 ± 0.73 mag. The vertical black dashed line marks the upper limit of Mr = −17.8 from Woosley et al. (2021).

In the text

Fig. 2. Colors and color cuts for the sample selections of Type Ibc SNe. The figure shows the absolute peak magnitudes (Mr in the upper and Mg in the lower panel) for the 94 SNe selected versus their MW corrected colors in g − r at ∼10 days after peak, when these transients have the most uniform color distribution (Taddia et al. 2015; Stritzinger et al. 2018). The gray box includes those 14 SNe kept in the final sample as normal SNe Ibc where uncertainties in the extinction corrections are smaller. The color coding is explained in the main text. The horizontal dashed lines represent the maximum luminosities (Mg = −17.5, Mr = −17.8), according to Woosley et al. (2021). The data points also have uncertainties in magnitudes assigned, according to the error propagation in Sect. 3.4.

In the text

Fig. 3. Light curves of the final sample of 14 Type Ibc SNe plotted in separate panels. We plot g- (green squares) and r-band (red circles) photometry in absolute AB magnitudes. These are corrected for distance and MW extinction. The x-axis gives rest-frame days since r-band peak, where the redshifts and explosion dates are provided in Tables 3 and 4. The dashed lines are the GP interpolations with error regions that were used to estimate peak explosion magnitudes and their uncertainties. Black arrows on top indicate the dates when we obtained spectra.

In the text

Fig. 4. Spectra for the bright and the red SNe discussed in Sects. 3.5.1 and 3.5.2. We display the classification spectrum from TNS together with a named template spectrum to illustrate the quality of our spectra. For most objects we also have additional spectra (see Fig. 3), and we provide some of these here as well. All these spectra are made available on Wiserep. Our spectra are shown in red, where the black line is a smoothed version. The name of the used telescope is also provided. The classification spectrum for SN 2019pfp is from Terreran (2019) and the one from SN 2020abqx from Burke et al. (2021).

In the text