Calcium-rich Transient SN 2019ehk in a Star-forming Environment: Yet Another Candidate for a Precursor of a Double Neutron-star Binary

Figure 3 shows the spectral evolution of SN 2019ehk from the end of the first peak (t = − 7.7 days) to the tail phase (t = 208.3 days). Several absorption lines are identified in the spectra before the second maximum (t < 0 days): He i λ5876, He i λ 6678, Si ii λ 6355, O i λ 7774, and the Ca ii NIR triplet. Overall features indicate that SN 2019ehk is a member of SNe Ib, according to the standard criteria. The strong Na i D narrow absorption line at the redshift of the host galaxy M100 (z ∼ 0.002) indicates that the absorption within the host galaxy is substantial (see Section 3.2 for further details). After the second peak (t > 0 days), spectral features quickly evolve. Prominent Ca ii and [Ca ii] emission lines are developed already at t = 16.3 days, showing a rapid transition to the nebular phase. These features, a rapid evolution to the nebular phase and the prominent Ca emissions, are characteristics of a Ca-rich transient.

In Figure 4, we compare SN 2019ehk with SN Ib 2008D (Modjaz et al. 2009), SN IIb 1993J (Barbon et al. 1995), iPTF14gqr (De et al. 2018b), iPTF16hgs (De et al. 2018a), and Ca-rich transient PTF10iuv (Kasliwal et al. 2012) at t ∼ 0 days (i.e., around the maximum brightness). Overall, the spectrum of SN 2019ehk is similar to those of SN Ib 2008D and the Ca-rich transients (iPTF16hgs and PTF10iuv). Indeed, there are no significant differences between the Ca-rich transients and (some of the) SNe Ib at this stage. At a closer look, the line profiles of He i, O i, and Fe ii seen in SN 2019ehk show a better match to those of the Ca-rich transients than SN 2008D. The spectrum of iPTF14gqr in this early phase is distinct from the Ca-rich transients, confirming the argument by De et al. (2018b) that it should not be classified as a "canonical" Ca-rich transient.

Zoom In Zoom Out Reset image size

Emission lines of Ca ii and [Ca ii] become prominent for iPTF16hgs and PTF10iuv at t ∼ 20 days (Figure 5), which is the definition of the "Ca-rich" transient class (Kawabata et al. 2010; Perets et al. 2010). SN 2019ehk shows a strikingly similar spectrum. While the absorption lines of He i are still visible, the spectra of the Ca-rich transients are distinct from SNe Ib at this phase, and thus SN 2019ehk should be definitely classified as a Ca-rich transient, not as a canonical SN Ib. Interestingly, iPTF14gqr comes to resemble the other Ca-rich transients at this phase, showing strong [Ca ii] and Ca ii NIR triplet. Therefore, iPTF14gqr can also be classified as a Ca-rich transient according to the standard criterion. However, we should note that iPTF14gqr shows broader Ca emissions than the others, and some lines identified in the other Ca-rich transients are not visible in its spectrum. As such, iPTF14gqr could be classified as a peculiar Ca-rich transient, which might indicate that its origin is different from the canonical Ca-rich transients.

Zoom In Zoom Out Reset image size

The difference between (normal) SNe Ib and Ca-rich transients becomes more obvious in the later phases. All the Ca-rich transients show strong [Ca ii] and Ca ii emission lines, and weak or no [O i] at t ∼ 55 days (Figure 6). The velocity seen in SN 2019ehk, as measured from the full width at half maximum of a Gaussian function fitted to the [Ca ii] profile, is ∼5300 km s⁻¹ . This is again similar to other (canonical) Ca-rich transients (Kasliwal et al. 2012) (e.g., as compared to PTF10iuv). The [O i] seen in SN 2019ehk is very weak even among Ca-rich transients.

Zoom In Zoom Out Reset image size

Figure 7 shows the evolution of the [Ca ii] /[O i] ratio as compared with those measured from spectra of a sample of Ca-rich transients presented by Valenti et al. (2014). The [Ca ii] /[O i] ratio of SN 2019ehk at t ∼ 54 days is measured to be 60 ± 40, which is within the range generally seen in the Ca-rich transients, and is among the largest.

Zoom In Zoom Out Reset image size

Figure 8 shows the evolution of the He i line velocities of SN 2019ehk. The velocities are derived from a Gaussian fit to the He i λ5876 and He i λ6678 absorption line profiles. We do not see a clear evolution in the velocity of He i λ 5876, but it is likely contaminated by Na i D. The velocity measured for He i λ 6678 is around 6500 km s⁻¹ at t ∼ − 5 days, and declines to ∼3000 km s⁻¹ at t ∼ 54 days. The velocity and its evolution here roughly overlap with those of iPTF16hgs.

Zoom In Zoom Out Reset image size

The spectral features and their evolution show that SN 2019ehk clearly belongs to the Ca-rich transient class. It shows spectral properties very similar to those of iPTF16hgs, and these two objects also share striking similarities in their light curves (see Section 3.3). iPTF14gqr is distinct from SN 2019ehk, iPTF16hgs, and other Ca-rich transients in terms of spectral evolution, while its similarities with regard to late phase spectrum and light-curve properties (Section 3.3) to SN 2019ehk and iPTF16hgs suggest a link among these three objects.

The Galactic extinction is negligibly small for SN 2019ehk (E(B − V) = 0.026 mag; see Schlafly & Finkbeiner (2011)). However, the extinction within the host galaxy along the line of sight toward SN 2019ehk is highly uncertain. Our spectra show a very deep Na i D absorption line, indicating substantial extinction within the host galaxy (see Figure 3). We measure the equivalent width (EW) as 3.2 ± 0.3 Å; this is beyond the applicability of the extinction–EW relation suggested by Poznanski et al. (2010).

Given the similarities between SN 2019ehk and iPTF16hgs, we could match the spectra of these two objects in order to constrain the extinction. Figure 9 shows the spectral comparison at t ∼ 0 days, with different values of E(B − V) applied to SN 2019ehk (E(B − V) = 0.5 and 1.0 mag). With E(B − V) = 0.5 mag, the two spectra match quite well. The blackbody temperature of SN 2019ehk is ∼5400 K and ∼8100 K, assuming respective extinctions of E(B − V) = 0.5 and 1.0 mag. Since the spectral features of SN 2019ehk are similar to those of iPTF16hgs, the temperature should not be very different between the two objects; as a rough estimate, if the temperature were different by ≳ 50%, the differences in the spectral features would likely become significant (Nugent et al. 1995). This places a rough upper limit of E(B − V) ∼1.0 mag for SN 2019ehk.

Zoom In Zoom Out Reset image size

To compare the color evolution of SN 2019ehk and other Ca-rich transients, we convert the magnitudes of SN 2019ehk to the SDSS system using the relations given by Jordi et al. (2006). In the comparison using the SDSS system, we apply the AB magnitude. We also confirm that these SDSS (AB) magnitudes are consistent with those estimated from the spectra of SN 2019ehk. Figure 10 shows the g − r color evolution of SN 2019ehk as compared to iPTF16hgs, iPTF14gqr, and other Ca-rich transients. The color evolution of SN 2019ehk roughly matches those of iPTF16hgs and other Ca-rich transients, for E(B − V) between 0.5 and 1.0 mag. This range is consistent with that from the spectral matching to iPTF16hgs. Therefore, we estimate the host extinction to be between E(B − V) = 0.5 and 1.0 mag.

Zoom In Zoom Out Reset image size

One issue we face when trying to obtain the absolute magnitudes is determining the value of R_V . Usually, the Galactic value (R_V = 3.1) is adopted for CCSNe, while it is suggested to be lower (R_V ∼ 2) for SNe Ia (Wang et al. 2008). The host extinction is usually negligible for Ca-rich transients, and thus the typical value for Ca-rich transients is uncertain. In this paper, we adopt the Galactic value, i.e., R_V = 3.1, which is more likely the case than the smaller value, given our interpretation of SN 2019ehk, independent from R_V , as a member of CCSNe (see Section 4). While adopting R_V ∼ 2 will reduce the estimated absolute magnitude by ∼0.5–1.0 mag, this would not affect our main conclusions, given that we already consider a range of E(B − V) = 0.5–1.0 mag.

Figure 2 shows our multiband light curves. Note that these magnitudes are corrected only for the Galactic extinction (see Section 3.2). The light curve shows a clear rise to the first peak in the optical bands after its discovery. Similar initial emissions were also detected for iPTF14gqr and iPTF16hgs, while the rising part toward the first peak was missed for both of them. This early emission has not been seen in the other Ca-rich transients. The first peak is reached at t = − 10.8 days for SN 2019ehk.

The decline rate after the second peak (between t = − 0.6 and t = 13.3 days) is 0.09, 0.07, and 0.05 mag day⁻¹ in the V, R, and I bands, respectively. After t ∼ 13.3 days, the decline becomes slower. The decline rate between t = 26 and 38 days is 0.05, 0.05, and 0.03 mag day⁻¹in the V, R, and I bands, respectively. A bluer band shows a steeper decline.

We compare the R-band light curve of SN 2019ehk to those of canonical SNe IIb and Ib in Figure 11. The same comparison, but for the r band (see Section 3.2), is shown with iPTF14gqr, iPTF16hgs, and other Ca-rich transients in the top panel of Figure 12. In these figures, we also plot the magnitudes of SN 2019ehk reported by other sources (see Section 2). These early points are shown here only for demonstration purpose, because of the differences in the band passes even though the central wavelengths are close to the R or r band. For the sake of comparison, we shift the light curves vertically to match to the peak magnitude of SN 2019ehk, and horizontally to match to the peak date.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The middle and bottom panels of Figure 12 show the absolute magnitude light curves of SN 2019ehk in the r and g bands, respectively, as compared with those of iPTF14gqr, iPTF16hgs, and other (canonical) Ca-rich transients (see Section 3.2). To show the uncertainty associated with the extinction, the absolute magnitudes of SN 2019ehk are corrected for host excitation with E(B − V) = 0.5 mag (red filled circles) or E(B − V) = 1.0 mag (blue).

Figure 11 shows that the light-curve evolution is much faster for SN 2019ehk than for the well-studied SNe IIb and Ib. It is especially clear in terms of the rising time to the second peak since the (estimated) explosion date: 22.7, 20.0, 23.6, and 12.7 days for SNe 1993J, 2008D, 2008ax, and 2019ehk, respectively. The same tendency is also seen in the decline rate after the second maximum: 0.045, 0.05, 0.055, and 0.07 mag day⁻¹ in SNe 1993J, 2008D, 2008ax, and 2019ehk, respectively.

SN 2019ehk, iPTF16hgs, and other Ca-rich transients show similar rising times, as seen in Figure 12. For example, the rising time and the decline rate for iPTF16hgs in the r band are 12.7 days and 0.10 mag day⁻¹, respectively, which are both similar to those of SN 2019ehk. iPTF14gqr shows a faster rise and fall than these objects, suggesting a smaller amount of ejecta. In any case, the light-curve evolution of iPTF14gqr is qualitatively similar to SN 2019ehk and the Ca-rich transients, being much faster than canonical SNe Ib.

The decline rate of SN 2019ehk after t ∼ 10 days is within a range observed for the Ca-rich transients, although it is on the slower side. On the other hand, iPTF16hgs shows the fastest decline among the Ca-rich transients. If SN 2019ehk and iPTF16hgs belong to the same subpopulation within the Ca-rich transient class (along with iPTF14gqr, as argued in this paper), it might suggest that this subpopulation has much more diverse properties than the other (main) population within the Ca-rich transient class.

The r-band peak magnitude of SN 2019ehk at t ∼ 0 days is ∼1 mag brighter than those of PTF10iuv and iPTF16hgs in the case of E(B − V) = 0.5 mag. ¹⁸ Even taking the uncertainty in the host extinction into account, SN 2019ehk is much more luminous in its (second) peak than the bulk of the Ca-rich transients. It can be compared to the peak luminosity of iPTF14gqr, which also shows a distinctly brighter peak than the other comparison objects.

The first peak is reminiscent of early emission from some CCSNe, exemplified by SN IIb 1993J. The magnitude of this emission relative to the second-peak magnitude, as well as the declining rate (and the duration from the estimated explosion date), are indeed similar between SN 1993J and SN 2019ehk. The emission is well-understood as the "cooling envelope emission" for the case of SN 1993J, with its extended H-rich progenitor (Richmond et al. 1996). Similar early emission is also seen for iPTF14gqr and iPTF16hgs, while such a feature has not been detected for the other (canonical) Ca-rich transients. The existence of the first peak would suggest that SN 2019ehk, iPTF14gqr, and iPTF16hgs may form a subpopulation within the Ca-rich transient class (or a population of explosions that resemble the canonical Ca-rich transients, in terms of observational features). The similarities in the properties of the first peak may further suggest that they are related to the deaths of massive stars (Section 4.2 for further details).

First, we construct a quasi-bolometric light curve of SN 2019ehk by integrating the spectral energy distribution (SED) in the B, V, R, and I bands. The integration is performed by interpolating the SED with trapezium functions. The conversion from the observed magnitude into the flux is carried out using the filter function (Bessell 1990) and the zero-point flux in each band.

We show the bolometric light curve in Figure 13. The bolometric light curve is constructed assuming two different values of the host extinction: E(B − V) = 0.5 and 1.0 mag (see Section 3.2). The ratio of the optical flux (in the B, V, R, and I bands) to the total flux is assumed to be that seen in typical SNe Ib/c (60%; Drout et al. 2014), given the similarity in the maximum spectra; we note that this is largely consistent with the value adopted for iPTF16hgs (De et al. 2018a). The first peak of SN 2019ehk is also confirmed in the quasi-bolometric light curve. The second-peak luminosity, L_peak, is ∼6 × 10⁴¹ erg s⁻¹ in the case of low extinction (E(B − V) = 0.5 mag), which already exceeds the typical range found for the canonical Ca-rich transients (Kasliwal et al. 2012; Lunnan et al. 2017). It could even be as bright as L_peak ∼ 2 × 10⁴² erg s⁻¹ if E(B − V) = 1.0 mag, making it comparable to the peak luminosity of iPTF14gqr. The rise time to the second peak is estimated to be 14.4 days, which is similar to iPTF16hgs (12.6 days; De et al. 2018a) and other Ca-rich transients.

Zoom In Zoom Out Reset image size

We calculate the ⁵⁶Ni mass, M(⁵⁶Ni), as 0.02 − 0.07 M_⊙ from the second-peak luminosity (Arnett 1982), where the range reflects the extinction uncertainties. Even with E(B − V) = 0.5 mag, this is larger than the values found for most of the Ca-rich transients (Kasliwal et al. 2012) including iPTF16hgs (De et al. 2018a). The estimated ⁵⁶Ni mass is between iPTF16hgs and iPTF14gqr, or similar to iPTF14gqr, depending on the treatment of the host extinction toward SN 2019ehk. ¹⁹

We estimate the kinetic energy and the ejecta mass for SN 2019ehk from the analysis of the light-curve evolution. We fit the r-band light curve of SN 2019ehk to that of iPTF16hgs by stretching the light-curve width around the second maximum. We then obtain the ratio of the timescale of SN 2019ehk to iPTF16hgs as ∼1.2.

Figure 8 suggests that the expansion velocities are similar between SN 2019ehk and iPTF16hgs, with a ratio of ∼0.8 between two objects. We further test this via the overall spectral property. After bending the second-maximum (t ∼ 0 days) spectrum of SN 2019ehk to match to the continuum of iPTF16hgs (Figure 9), we introduce an artificial shift in the wavelength to the spectrum of SN 2019ehk, to find the best match in the positions of the absorption minima of different lines. We thereby find that the ratio of the line velocities of SN 2019ehk to iPTF16hgs is on average v_19ehk/v_16hgs ∼ 0.8, i.e., consistent with the estimate using the He velocity.

We then convert these ratios in the light-curve widths and the expansion velocities to the ejecta mass (M_ej) and the kinetic energy of the ejecta (E_k), using the following relations:

$$\begin{eqnarray}&&\tau \propto {M}_{\mathrm{ej}}^{3/4}\cdot {E}_{{\rm{k}}}^{-1/4},\end{eqnarray} \tag{ 1 }$$

$$\begin{eqnarray}&&{v}_{\mathrm{ej}}\propto {M}_{\mathrm{ej}}^{-1/2}\cdot {E}_{{\rm{k}}}^{1/2}.\end{eqnarray} \tag{ 2 }$$

For the normalization in this scaling method, we use the ejecta properties estimated for iPTF16hgs (M_ej = 0.38 M_⊙ and E_k = 2.3 × 10⁵⁰ erg; see De et al. (2018a)). We then obtain M_ej = 0.43 M_⊙ and E_k = 1.7 × 10⁵⁰ erg for SN 2019ehk.

These properties are similar to those obtained for iPTF16hgs (and the other Ca-rich transients in general), as expected from the similar properties seen in their light curves and spectra. The ejecta mass is smaller than that of iPTF14gqr, while the kinetic energy is similar (M_ej = 0.2 M_⊙ and E_k = 2 × 10⁵⁰ erg; see De et al. (2018b)). These properties are consistent with the slower evolution and lower line velocities in SN 2019ehk versus those in iPTF14gqr.

Here, we investigate the origin of the first peak in the light curves. Given the large uncertainty in the host extinction, we consider a range of the possible extinction between E(B − V) = 0.5 and 1.0 mag as roughly constrained by the color at the second peak (see Section 3.2). The possible energy source of the first peak is roughly divided into two categories: the radioactivity and the shock interaction. In Section 4.3, we argue against the former scenario, and thus in this section we focus on the shock-interaction power. This is further divided into two classes of the mechanism: the shock-cooling emission and the ongoing shock-powered emission. We discuss these scenarios one by one in this section.

The first peak observed for SN 2019ehk is reminiscent of the so-called cooling-envelope emission frequently observed for CCSNe with an extended envelope (e.g., see SN IIb 1993J in Figure 11), where the thermal energy deposited by the propagation of the shock wave is diffused out in the cooling timescale. The mechanism is indeed not limited to a "stellar envelope," but rather is applicable to a "confined" CSM as long as the CSM is already swept up by the ejecta and behaves like a cooling envelope.

For the analysis of this process, we follow the formalism presented by Piro (2015). In doing this, we use a semi-analytic code developed by Maeda et al. (2018) to simulate a similar emission process. We calibrated the code by comparing the output to the results of Piro (2015) for the same parameter set. Indeed, De et al. (2018a) used the same formalism to analyze the first peak seen in iPTF16hgs. We have also confirmed that we obtain a result consistent with that of De et al. (2018a), if we use the same input parameters; this guarantees a fair comparison between the natures of SN 2019ehk and iPTF16hgs.

As the input parameters, we adopt M_ej = 0.43 M_⊙ for the ejecta mass and E_k = 1.7 × 10⁵⁰ erg for the kinetic energy (see Section 4.1). The luminosity is sensitive to the radius of the envelope (or CSM) while the diffusion timescale is so to the envelope (CSM) mass, and there is no degeneracy between these two parameters to derive. We first try to obtain a rough match to the V-band light curve, and the same model is adopted to generate the B- and R-band light curves.

As shown in Figure 14, we can obtain a reasonable fit to the early V-band light curve. The envelope/CSM mass is ∼0.04–0.05 M_⊙ irrespective of the extinction assumption. The outer radius of the envelope/CSM is sensitive to the assumed extinction: 130 R_⊙ , 300 R_⊙ , and 1500 R_⊙ for E(B − V) = 0.5, 0.7, and 1.0 mag, respectively. While the mass is similar to that derived for iPTF16hgs (De et al. 2018a), the radius required for SN 2019ehk is substantially larger. This is a result of the brighter first peak luminosity (L_bump) in SN 2019ehk even for E(B − V) = 0.5 mag.

Zoom In Zoom Out Reset image size

We note that the expected model color is far too blue to be consistent with the observations of SN 2019ehk, as long as E(B − V) < 1.0 mag. The cooling-envelope emission predicts that the V-band peak is realized when the photospheric temperature decreases to ∼10,000–20,000 K, as an outcome of the shift of the blackbody peak from the UV to the optical bands. However, the observed color of the first peak is red (V − R > 0 mag) if E(B − V) < 1.0 mag. There is then no solution to explain the color via the cooling-envelope emission under the low-extinction assumption (Figure 14).

A physically reasonable solution with the cooling-envelope emission model can be found only if we assume a high value of the extinction (E(B − V) = 1.0 mag; see Figure 14). The derived radius can be as large as ∼1500 R_⊙ . This is comparable even to one of the largest SN IIP progenitors discovered through the pre-explosion image analysis (Huang et al. 2018). Our spectra do not exhibit any hydrogen absorption features, and such a large and He-rich "envelope" is physically not realized. On the other hand, the mass and radius here are very similar to those suggested for the "confined" CSM around CCSNe commonly inferred through the flash spectroscopy (Gal-Yam et al. 2014; Yaron et al. 2017) or early light-curve behaviors (Moriya et al. 2017b, 2018; Morozova et al. 2017, 2018; Förster et al. 2018). We suggest that this can be interpreted as the existence of the confined CSM around SN 2019ehk, which might further support its origin as a core collapse of a massive star. The corresponding mass-loss rate is ∼1.5 M_⊙ yr⁻¹, assuming a mass-loss velocity of 1000 km s⁻¹. This is high due to the large wind velocity assumed here, but it involves a large uncertainty—even more than an order of magnitude.

As clarified above, another emission mechanism is required if E(B − V) < 1.0 mag. An alternative mechanism is the situation in which the ejecta is still propagating within the optically thick CSM at the first peak of the light curve. This allows a larger photospheric radius as well as a lower temperature, and thus a redder color. If E(B − V) = 0.5 mag, the blackbody fit to the SED at the first peak suggests a photospheric temperature of ∼7000 K. The observed luminosity and temperature for the case of E(B − V) =0.5 mag indicate a blackbody radius of ∼11,000 R_⊙ (i.e., ∼8 × 10¹⁴ cm). Similarly, if we assume E(B − V) = 0.7 mag, then the photospheric temperature and the radius are ∼8500 K and ∼9400 R_⊙ (∼7 × 10¹⁴ cm), respectively. This is again within the range suggested for the confined CSM. We can derive the CSM density (and thus the mass-loss rate) under this ongoing shock interaction model, through the peak luminosity (interaction power) and the peak duration (diffusion timescale) (Moriya & Maeda 2014). ²⁰ The mass-loss rate thus derived is ∼4 × 10⁻³ M_⊙ yr⁻¹, assuming a constant wind velocity of ∼1000 km s⁻¹, which is the velocity seen in iPTF14gqr (De et al. 2018b); this velocity is also expected for a He star progenitor. Again, the required CSM density (the mass-loss rate) and the size are within the range estimated for the confined CSM around CCSNe.

The narrow emission lines of hydrogen in the earliest spectrum of SN 2019ehk were reported by Dimitriadis et al. (2019) and discussed in Jacobson-Galán et al. (2020a). This indicates that the ongoing interaction model may be more likely, while the shock-cooling scenario is not immediately rejected. The progenitor system of SN 2019ehk including these early emission lines in the spectrum is discussed in Section 4.3.

In summary, we suggest two possible interpretations: the cooling envelope/CSM emission (for E(B − V) ∼1.0 mag) or the ongoing CSM interaction (for E(B − V) < 1.0 mag) (see Section 4.3 for details regarding the difficulties presented by another mechanism where it is powered by radioactivity, e.g., by ⁵⁶Ni). The results of this analysis are summarized in Table 4, together with the properties derived from the second peak, where R_peak is the R-band peak magnitude of SN 2019ehk at t ∼ 0 days. While the two scenarios involve different mechanisms and result in different natures of the CSM, we come to the same conclusions in qualitative terms. In both cases, there is a dense CSM around SN 2019ehk. The progenitor might experience (unknown) unstable activity in the final stage of the stellar evolution of the massive star (Fuller & Ro 2018).

The ejecta properties of SN 2019ehk, iPTF14gqr, and iPTF16hgs are generally consistent with the expectations for a low-mass (<2M_⊙) He (or C+O) star explosion corresponding to a main-sequence mass of 8–12 M_⊙ , which defines a boundary between a CCSN explosion and a WD formation (Kawabata et al. 2010). In terms of the binary evolution, such a low-mass He or C + O star progenitor is a system of interest in the formation of double NS binaries; if a companion star is an NS, the low-mass SESN can leave a bound double NS binary, while the explosion of a more massive star should disrupt the progenitor binary system. The low-mass SESNe include the ultra-stripped envelope SN with an NS comparison (Tauris et al. 2013, 2015), which has been actively debated as a promising pathway to form double NS binaries. Indeed, De et al. (2018b) suggested iPTF14gqr as a robust candidate to be an ultra-stripped envelope SN. The scenario favored for iPTF16hgs (De et al. 2018a) is also the ultra-stripped envelope SN, while an eruptive event related to a WD is not rejected. In this paper, we suggest that SN 2019ehk is yet another candidate for a low-mass SESN. We note that a similar scenario has also been proposed by De et al. (2021).

It is possible that SN 2019ehk might have left a double NS binary. However, the orbital separation of the post explosion binary is likely wider than the genuine ultra-stripped envelope SNe. This is thus a candidate system for double NSs with a large separation, which will not merge within the Hubble time (Tauris et al. 2017), and such systems are seen in Galactic double NSs systems (Farrow et al. 2019). SN 2019ehk is thus an interesting object that could potentially provide insight regarding the conditions and fraction leading to the ultra-stripped envelope SNe among the low-mass SESN systems, which may then be linked to the formation scenario of double NSs.

The diversities seen in these objects are in line with the expectation from a core-collapse event of a massive star, as there can be diversities in the progenitor stars, e.g., in the progenitor mass and in the properties of the envelope at the time of the explosion (Tauris et al. 2015). Indeed, given the low-mass ejecta, a small difference in the progenitor mass in the low-mass SESN scenario can easily lead to a large diversity in the observational properties (Moriya et al. 2017a; Suwa et al.2015; Yoshida et al. 2017; Müller et al. 2018). For example, Tauris et al. (2015) predict that the ejecta mass ranges from ∼0.01 M_⊙ to ∼0.5 M_⊙ for the progenitors leading to ultra-stripped envelope SNe. The range covers the properties derived for SN 2019ehk, iPTF14gqr, and iPTF16hgs.

SN 2019ehk, iPTF14gqr, and iPTF16hgs share a unique observational property; they show a "first" peak with a duration of a few days observed in the optical wavelength. This feature can be naturally interpreted as the existence of a substantial amount of CSM, as has been interpreted for many CCSNe. The models based on an old system, e.g., a WD, have difficulties creating such a short-duration early emission. The so-called .Ia scenario would create a double-peaked light curve in the "bolometric" luminosity due to existence of powering radioactive species at different layers. However, in this model, the first peak is generally in the UV and the second peak is in the optical; the optical light curve is indeed predicted not to show the double peaks (Shen et al. 2010). An exception is the double-detonation model of a sub-Chandrasekhar mass WD, which could create the double peaks in the optical (Jiang et al. 2017; Maeda et al. 2018). However, given the relatively low luminosity of SN 2019ehk and the other objects (as compared to SNe Ia), this will result in substantial absorption by various Fe-peak lines at the second peak (Maeda et al. 2018), and the resulting spectrum would never resemble that of an SN Ib.

Through "flash spectroscopy," De et al. (2018b) suggested the existence of a confined CSM around iPTF14gqr. For iPTF16hgs, the properties of the first peak are explained by the cooling-envelope emission from a He-rich envelope, without the need for a dense CSM. The first peak observed for SN 2019ehk can be explained either by a very dense and confined CSM (if the extinction is high) or a moderately dense CSM (if the extinction is low; see Section 4.2). Either way, irrespective of an association with a He-rich envelope or a confined CSM, the double-peak light curve suggests a link to the CCSN, especially to the low-mass SESN. Given the classification of SN 2019ehk and iPTF16hgs as Ca-rich transients, as well as the similarities of iPTF14gqr to these objects, the existence of multiple populations is indicated within the Ca-rich transients (see Section 4.4).

Indeed, the difference in the properties of the first peak may indicate further subclassification among the low-mass SESNe. As mentioned in Section 4.2, the narrow emission lines of hydrogen in the earliest spectrum of SN 2019ehk were reported by Dimitriadis et al. (2019) and discussed in Jacobson-Galán et al. (2020a). Importantly, this indicates that the He star progenitor did not experience a strong Roche lobe mass transfer toward the SN explosion for the case of SN 2019ehk. On the other hand, iPTF14gqr showed noticeable He lines but not hydrogen lines in its flash spectrum. This variation may be expected in a low-mass SESN with a NS companion star; in such a scenario, the low-mass He star is first produced via common envelope interaction with an NS companion (Tauris et al. 2013). If the reduction in the orbital separation in the common envelope phase is substantial, the progenitor star will further fill the Roche lobe, and the entire H envelope (and perhaps further, the He layer as well) may be stripped away to become an ultra-stripped envelope SN. On the other hand, if the binary separation after the common envelope phase is modest, a small amount of H envelope will be left, and this could be the situation for SN 2019ehk.

Most Ca-rich transients show a large offset from a host galaxy (Lunnan et al. 2017). This indicates old stellar populations (e.g., an explosion or an eruptive event of a WD) as their origin. Kasliwal et al. (2012) and Lunnan et al. (2017) showed that the (canonical) Ca-rich transients have homogeneous luminosity distribution, indicating similar ⁵⁶Ni masses. The evolutions in the spectra and light curve also show homogeneous natures (Section 3; see also De et al. (2018a)).

However, there are some Ca-rich transients, including SN 2019ehk, that show star-forming activity in their explosion sites. SN 2019ehk is on a dust lane along a spiral arm of M100. Another Ca-rich transient, iPTF16hgs, is located at ∼6 kpc away from the core of a star-forming galaxy (De et al. 2018a), which is atypical as a site of the Ca-rich transient. As yet another example, a peculiar Ca-rich transient, iPTF15eqv, is located on an arm of a spiral galaxy. Early phase data for iPTF15eqv are missing, as it was discovered after the maximum, but it is definitely much brighter than the canonical Ca-rich transients with an estimated ⁵⁶Ni mass of M(⁵⁶Ni) = 0.07 M_⊙ for iPTF15eqv (Milisavljevic et al. 2017). Among the low-mass SESN candidates discussed in this paper, iPTF14gqr does not clearly show star-forming activity in its explosion site; it is offset by ∼30 kpc from the core of a star-forming galaxy (or ∼15 kpc from the nearest spiral arm), which is consistent with the old environment generally derived for the Ca-rich transients, while the host galaxy shows a tidally interacting environment (De et al. 2018b).

SN 2019ehk, iPTF16hgs, and iPTF14gqr share a distinct feature in their light curves; a first peak with a duration of a few days to a week. ²¹ This is not seen in the other canonical Ca-rich transients. Analysis of the first peaks suggests that they originate in the CCSN events. At the same time, they also show some diversities in their observational properties. iPTF16hgs is nearly indistinguishable from the canonical Ca-rich transients, except for the first peak and its potentially young environment. iPTF14gqr shows characteristics quite different from those of the Ca-rich transients, e.g., its early-phase spectra and the double-peaked light curve, while it evolves to resemble the canonical Ca-rich transients in a late phase. The almost featureless spectra of iPTF14gqr around the peak are likely due to the high luminosity and temperature as well as the high expansion velocities. The properties of SN 2019ehk are in between those of iPTF16hgs and iPTF14gqr; spectroscopically, it is indistinguishable from the canonical Ca-rich transients, but it shows a high luminosity, a clear first peak, and a star-forming environment.

As a consequence of the diversity in the observational properties, the inferred properties of the ejecta and circumstellar environment are also diverse. The ejected ⁵⁶Ni masses span from M(⁵⁶Ni) = 0.01 M_⊙ for iPTF16hgs to M(⁵⁶Ni) = 0.05 M_⊙ for iPTF14gqr. SN 2019ehk is in between. The ejecta mass may range from ∼0.2M_⊙ (iPTF14gqr; see De et al. (2018b)) to ∼0.4 M_⊙ (iPTF16hgs and SN 2019ehk; see De et al. (2018a) and this paper). There also seems to be some diversity in the nature of the circumstellar environment (Section 4.2).

In summary, we suggest that SN2019ehk, iPTF16hgs, and iPTF14gqr (and potentially iPTF15eqv) form a subpopulation within the Ca-rich transient class. They not only explode within the young environment (with the possible exception of iPTF14gqr) but also have observational properties beyond the diversity seen in the canonical Ca-rich transients. Their properties are summarized in Table 5, together with those of PTF10iuv as a representative of the canonical Ca-rich transients. We identify four such objects (SN 2019ehk, iPTF14gqr, iPTF15eqv, and iPTF16hgs), i.e., ∼10% of the whole sample of Ca-rich transients. Note that the fraction here might be an overestimate, given the systematically higher peak immensities for these events versus those of the canonical Ca-rich transients. The rate of Ca-rich transient is estimated as ∼33–95% of SNe Ia in the local universe, which corresponds to ∼10–30% of CCSNe. Therefore, the rate of this subpopulation (including SN 2019ehk) is roughly ∼1%–3% of CCSNe, or even larger. This rate is marginally consistent with the expectation for the ultra-stripped envelope SN scenario to be a main evolutionary pathway toward double NS binaries (Tauris et al. 2013, 2015), i.e., 0.1%–1%. Indeed, while the estimate here is very rough, the relatively high rate of the subpopulation under consideration might indicate that it would contain not only the extreme case of a ultra-stripped envelope SNe (i.e., a close binary with a NS), but those of a low-mass helium star explosion in a binary system with a massive main-sequence star, or in a binary with an NS but with a large binary separation, which may indeed be the case for SN 2019ehk (Section 4.3).