ASASSN-14ms: the Most Energetic Known Explosion of a Type Ibn Supernova and its Physical Origin

We commenced optical photometry of ASASSN-14ms with different instruments, including the 0.8 m Tsinghua University-NAOC Telescope (hereafter TNT) at Xinglong Observatory in China (Wang et al., 2008; Huang et al., 2012), the 2.4 m telescope of Yunnan Observatories at Lijiang Station (hereafter LJT; Wang et al. 2019) in China, and the 2.3 m Bok telescope (hereafter Bok; Williams et al. 2004) of NAOA in the USA. These observations spanned from about $-2$ days to +130 days from maximum brightness, and all of the data were reduced with standard IRAF¹¹1IRAF, the Image Reduction and Analysis Facility, is distributed by the National Optical Astronomy Observatory, which is operated by the Association of Universities for Research in Astronomy (AURA), Inc. under cooperative agreement with the National Science Foundation (NSF). routines. The SN instrumental magnitudes were converted to the Johnsons $UBV$ (Johnson et al., 1966) and Kron-Cousins $RI$ (Cousins, 1981) systems, using flux calibrations of six field stars from the Sloan Digital Sky Survey (SDSS) Data Release 9 catalog as listed in Table 1. Table 2 lists the final flux-calibrated $UBVRI$ photometry of ASASSN-14ms. A few reports for upper limits and detection magnitudes in the $V$ band, obtained at earlier phases from the ASAS-SN and amateur astronomers, are also included.

In addition, ultraviolet (UV) and optical photometry from the Ultra-Violet/Optical Telescope (UVOT; Roming et al., 2005) onboard the Neil Gehrels Swift Observatory ($Swift$; Gehrels et al., 2004) were available at two epochs around maximum light. The UVOT observations were obtained in the $uvw1$, $uvm2$, $uvw2$, $u$, $b$, and $v$ filters, with the images being reduced following that of the Swift Optical Ultraviolet Supernova Archive(SOUSA; Brown et al., 2014). An aperture of 3 arcsec is applied to measure the SN flux, with the corresponding aperture correction based on an average point-spread function. The final UVOT magnitudes are listed in Table 3.

Spectroscopic observations were collected using the Lijiang 2.4 m telescope and YFOSC system (Fan et al., 2015), the Xinglong 2.16 m telescope (+BFOSC), and the Keck 10 m telescope (+LRIS; Oke et al. 1995). All of the spectra were reduced using routine tasks within IRAF, and the flux was calibrated with spectrophotometric standard stars observed on the same nights. Telluric lines were removed from the spectra through comparison with the standard-star observations. A journal of observations is given in Table 4.

As redshift is not available for the galaxy SDSS J130408.52+521846.4 from the public catalogue, we collected the host-galaxy spectra from two telescopes. As the Keck spectrum was taken $\sim 50$ days after maximum brightness when the SN light was comparable to that of the host galaxy, it was also used to extract the host-galaxy spectrum to determine the redshift. Another spectrum of the host galaxy SDSS J130408.52+521846.4 was obtained on 2016 June 21.4 (t$\sim$ 6 months after the peak) with the 8.2 m Subaru telescope to cross-check the result from the Keck telescope. Figure 2 shows the host-galaxy spectra of ASASSN-14ms from Keck (blue) and Subaru (red). One can see that the narrow emission line at 6915 Å, with a measured full width at half-maximum intensity (FWHM) of $\sim 300$ km s^-1, is clearly present in both spectra. Assuming this feature to be H$\alpha$, a redshift of $z=0.0535$ can be inferred for SDSS J130408.52+521846.4. This value is consistent with $z=0.0539$ determined from another narrow emission feature (at 3928 Å) which can be attributed to [O ii] $\lambda$3727, as seen in the Keck spectrum (upper panel of Fig.2). We thus adopt a mean value of $z=0.0537\pm 0.0002$ for ASASSN-14ms. This redshift is consistent with that derived by Vallely et al. (2018) from a late-time LBT MODS spectrum of SDSS J130408.52+521846.4 (obtained at t$\sim$ 876 days after maximum light). Assuming H₀ = 73 km s^-1 Mpc^-1, $\Omega_{M}=0.27$, and $\Omega_{\Lambda}=0.73$ (Spergel et al., 2007), a distance modulus of $36.72\pm 0.14$ mag can be obtained for ASASSN-14ms, which is adopted in the following analysis.

From the H_α luminosity of flux-calibrated Subaru spectrum, we estimated the star formation rate using the conversion relation from Kennicutt (1998), which is half of the value derived by Vallely et al. (2018), i.e.,$\sim$0.005 M_⊙ yr^-1. This difference is likely due to that the Subaru spectrum has relatively low signal-to-noise ratio. Modeling of the photometric spectral energy distribution for galaxy SDSS J130408.52+521846.4 reveals that this galaxy has a stellar mass $M_{*}$=2.6^+0.8_-1.5$\times$10⁸ M_⊙ (Vallely et al., 2018). This indicates a specific star formation rate of log(sSFR)$\sim-$10.11 M_⊙ yr^-1 for the host galaxy of ASASSN14ms well within the range of core collapse SNe (see Fig.9 of Xiang et al. 2021).

Figure 3 shows the optical light curves of ASASSN-14ms, with two UV data points from Swift/UVOT overplotted. The earliest prediscovery image was taken on 2014 December 15.51, with a nondetection limit of $\sim 17.1$ mag in the $V$ band. The absence of early-phase data does not allow us to place stringent constraints on the explosion date and hence the rise time. The $V-$band light curve is better sampled around maximum light compared to other bands. Applying a polynomial fit to the $V$ light curve yields $V_{\rm max}=16.42\pm 0.04$ mag on JD 2,457,023.85. Likewise, the $B$-band light curve reaches a peak of $16.30\pm 0.05$ mag at a similar phase, while the $U$-band light curve seems to have a slightly earlier peak with $U_{\rm max}\approx 14.9$ mag. Maximum brightness is not clear in the $R$ and $I$ bands because of sparse sampling. On the other hand, we notice that all of the light curves seem to experience three-stage post-peak evolution, with breaks occurring at $t\approx 20$ days and $t\approx 40$ days after maximum. During the period from $t\approx 20$ days to $t\approx 40$ days after maximum, the light curves declined at a relatively faster pace.

Taking the distance modulus $\mu=36.72\pm 0.14$ mag and a Galactic reddening of $E(B-V)=0.01$ mag, we derive an absolute $V$-band peak magnitude of $-20.33\pm 0.15$ mag for ASASSN-14ms. The corresponding values in $B$ and $U$ are $-20.46\pm 0.15$ mag and $\gtrsim-21.9$ mag, respectively. With the available Swift $UV$ points, one can find that this object is also very luminous in the UV, with the luminosity in the $uvw1$, $uvm2$, and $uvw2$ bands being brighter than $-22.1$ mag. Detailed photometric parameters are listed in Table 3. These estimates indicate that ASASSN-14ms is one of the most luminous SNe Ibn ever recorded (e.g., Pastorello et al., 2016; Hosseinzadeh et al., 2017), even without considering possible extinction due to the host galaxy.

Figure 4 shows the $V$-band light curve of ASASSN-14ms compared with those of well-observed SNe Ibn such as SN 2006jc (Pastorello et al., 2007), SN 1991D (Benetti et al., 2002), SN 2010al (Pastorello et al., 2015a), SN 2014av (Pastorello et al., 2016), 2015U(Shivvers et al., 2016), ASASSN-15ed (Pastorello et al., 2015d), and 2019uo (Gangopadhyay et al., 2020). The light curves of other subclasses are also plotted for comparison, including broad-lined Type Ic SN 1998bw (Galama et al., 1998), Type Ia SN 2005cf (Wang et al., 2009), Type Ic SN 2007gr (Chen et al., 2014), and Type Ib SN 2009jf (Sahu et al., 2011). It is readily seen that ASASSN-14ms and other comparison SNe Ibn are characterized by fast post-peak declines. For example, the decline within the first 40 days is about 3.6 mag for ASASSN-14ms, 4.1 mag for ASASSN-15ed, 4.2 mag for SN 2010al, and 3.4 mag for SN 2014av.²²2The flattening behavior seen in SN 2014av after 20 days from maximum light is likely due to the interaction of the SN ejecta with the CSM from the progenitor. In contrast, the corresponding magnitude decline is about 2.0, 2.1, 2.0, and 1.6 for SN 1998bw, SN 2005cf, SN 2007gr, and SN 2009jf (respectively). This large discrepancy in the post-maximum evolution indicates that SNe Ibn are not primarily powered by radioactive decay in a relatively long phase after peak brightness (see discussions in §4) and their high luminosities are likely caused by other mechanisms.

Owing to the high luminosity of ASASSN-14ms, we further explore its possible connection with superluminous supernovae (SLSNe). To better quantify the discrepancy in light-curve evolution between ASASSN-14ms, other well-observed SNe Ibn, and Type I/II SLSNe, we examine the absolute peak magnitude in $V$ ($M_{V,\mathrm{p}}$) and the magnitude decline measured within the first 30 days after the peak ($\Delta m_{30}(V)$). As shown in Figure 5, the absolute $V$ magnitude of most SNe Ibn is in the approximate range $-17$ to $-20$ around maximum light. Among them, ASASSN-14ms is at the most luminous end with $M_{V,\mathrm{p}}\approx-20.5$ mag, close to the faint end of SLSNe. For SNe Ibn and SLSNe, the decline rate measured within 30 days after the peak tends to show a correlation with the corresponding absolute peak magnitudes, with brighter objects having slower decline rates. Applying a linear fit to the observed data for SNe Ibn and SLSNe yields $M_{V,\mathrm{p}}=-22.75^{+0.06}_{-0.02}+1.08^{+0.01}_{-0.02}\times\Delta m_{30}$. This trend suggests that these two subclasses of SNe may have a similar energy source. If their early-time radiation is dominated by ejecta-CSM interaction in both cases (see discussions in §4), the difference in decline rate might suggest that SNe Ibn tend to have less ejecta mass and/or less extended CSM than the SLSNe. However, the establishment of this anti-correlation among SLSNe and SNe Ibn requires more statistically significant evidence.

In Figure 6, we compare the color evolution of ASASSN-14ms with that of the comparison SNe Ibn. In the first 20 days, ASASSN-14ms seems to exhibit $B-V$ color evolution that is quite similar to that of SN 2010al and SN 2014av. After $t\approx 20$ days, both SN 2010al and SN 2014av tend to maintain a constant $B-V$ color of $\sim 0.6$ mag, whereas ASASSN-14ms evolves progressively redward until $t\approx 30$ days, subsequently evolving blueward and reaching a similar color as the former two events at $t\approx 50$ days. ASASSN-15ed may have a similar evolutionary trend as ASASSN-14ms, except that it is systematically redder by $\sim 0.2$ mag after $t\approx 10$ days. Among the comparison sample, ASASSN-15ed and SN 2006jc exhibit the fastest and slowest color evolution, respectively. One can see that SN 2010al and SN 2014av also show similar $R-I$ color evolution, with a reddening rate of $\lesssim 0.01$ mag d${}^{-}1$ during the first 40 days. Slow $R-I$ color evolution is also found in ASASSN-14ms, but overall it is significantly bluer than SN 2010al, SN 2014av, and SN 2006jc. The $R-I$ evolution of SN 2015U might be consistent with that of ASASSN-14ms, but the comparison is limited by the large uncertainty in the former’s color at $t\approx 10$–20 days.

Twelve optical spectra were obtained of ASASSN-14ms, covering the approximate phases $-2.5$ days to +50.0 days from maximum light; they are displayed in Figure 7. The earlier spectra are characterized by a very blue continuum, suggesting a high temperature for the emitting region at early phases and consistent with the strong UV emission revealed by Swift photometry obtained around the time of maximum light. Later, some narrow lines with P Cygni profiles emerge in the spectra. The feature at $\sim 5900$ Å, persisting in almost all of the spectra, is likely due to He i $\lambda$5876. Other helium lines can be also detected at 4470 Å, 6700 Å, and 7060 Å(see dashed lines in Fig. LABEL:fig-specomp). Note that these line profiles of helium (both absorption and emission components) get only slightly stronger from the near-maximum-light phase to $t\approx 50$ days after maximum brightness, indicating that they are primarily formed in a slowly-moving helium shell instead of in SN ejecta.

To further examine the spectral features of ASASSN-14ms, we compare it with different subtypes of core-collapse SNe in which the hydrogen envelope was stripped away before the explosion, as shown in Figure 8. The comparison objects include SNe Ibn such as SN 2010al (Pastorello et al., 2015a) and ASASSN-15ed (Pastorello et al., 2015d), the peculiar Type Ib SN 1991D (Benetti et al., 2002), and the normal Type Ib SN 2009jf (Sahu et al., 2011). At one week after maximum light, ASASSN-14ms shows narrow P Cygni profiles of helium lines at $\sim 4470$ Å, 5860 Å, and 6700 Å  similar to those seen in SN 2010al and ASASSN-15ed. In particular, all are found to show the “M”-shaped feature at 6600–6700 Å, which is likely formed by He i or Ne i (see the SYNOW fit shown in Figure 9). Note that the weak emission feature on the left side of He i $\lambda$6678 could be H$\alpha$, as similarly seen in SN 2010al and SN 2009jf, and it is not unexpected that some hydrogen mixed into the helium shell. At about four weeks after peak brightness, the line profiles of the two comparison SNe Ibn (SN 2010al and ASASSN-15ed) and the peculiar SN Ib (SN 1991D) become apparently stronger (see the lower panel of Fig. 8), with the absorption component of He i $\lambda$5876 having a velocity of 5000–6500 km s^-1. In contrast, the corresponding blueshifted velocity is only $\sim 2000$ km s^-1 in ASASSN-14ms, suggesting that its He i line is still primarily formed from the CSM.

Figure 10 shows the temporal evolution of the line profiles of He i $\lambda$5876, He i $\lambda$6678, and O i $\lambda$7774. One can see that the absorption components of these lines tend to have blueshifted velocities increasing from $\sim 800$ km s^-1 at around maximum light to $\sim 2000$ km s^-1 about 1 month later. Such an evolution in the above line profiles is similarly seen in a few spectra from Vallely et al. (2018) (see gray spectra in Fig. 10). This indicates that the CSM was abundant with helium and oxygen, and that the CSM surrounding ASASSN-14ms was only slightly accelerated by the SN ejecta, suggesting that the CSM was either more distant or the mass was relatively larger.

To further constrain the elements in the ejecta of ASASSN-14ms, we use SYNAPPS and the companion code SYN++ (Thomas et al. 2011) to fit the $t\approx+24.5$ d spectrum. We first fit the spectrum automatically with the SYNAPPS through the reduced chi-square technique, which involves more than a dozen of ions. We then ignored some unimportant ions and used 6 main ions (including He I, C I, O I, Ne I, Ca II and Fe II) to fit the main features in this spectrum, with the best-fit result (judged by eye) shown in Figure 9 and the parameters listed in Table 5. From the best-fit result, one can see that the intermediate-mass elements such as He i, O i, Ne i, and Ca ii (and perhaps C i) can be identified in the spectrum of ASASSN-14ms. Note that the Ne I features around 6000 Å can also be seen in Type Ib supernova SN 1991D (Bennett et al. 2002). Abundant Fe ii elements are also needed to reproduce the prominent absorptions around 4500 Å and 5100 Å. However, it is not clear which elements lead to the broad trough near 3600 Å.

Around maximum light, the expansion velocity inferred from the absorption minima of He i lines for ASASSN-14ms is measured to be $\sim 1000$ km s^-1. As shown in Figure 11, other SNe Ibn show similarly low velocities, as measured at comparable phases. In comparison, normal SNe Ib have expansion velocities ranging from $\sim 7000$ km s^-1 to $\sim 12,000$ km s^-1, much larger than the typical velocity derived for SNe Ibn. Moreover, these two subclasses with prominent He i features have distinct peak luminosities, with SNe Ibn being on average much more luminous than SNe Ib. We note that SN 1991D shows the P Cygni line profiles lying between those of normal SNe Ib and SNe Ibn, with a blueshifted velocity of $\sim 5000$ km s^-1; it may be regarded as a transitional object linking these two subclasses. This suggests that SNe Ib and SNe Ibn may not be distinctly different, and their observed differences are likely caused by the surrounding CSM.

A classic theoretical model is presented by Arnett (1982), where the radioactive decay of ⁵⁶Ni powers the ejecta in a homologous expansion with a uniform density distribution. In fitting the observations, we consider the radioactive decay of ⁵⁶Ni via ⁵⁶Co to ⁵⁶Fe as the only energy source (e.g., Valenti et al., 2008) and employ the temperature calculation method described by Nicholl et al. (2017). The gamma-ray leakage from the ⁵⁶Ni decay is also considered in our model, which modifies the luminosity input by a factor of $1-\exp(-3\kappa_{\gamma}M_{\mathrm{ej}}/4\pi v_{\mathrm{ej}}^{2}t^{2})$ (e.g., Wang et al., 2015).

The free parameters for the model fit include the explosion time ($T_{0}$), the ⁵⁶Ni mass ($M_{\mathrm{Ni}}$), and the ejecta mass and velocity ($M_{\mathrm{ej}}$, $v_{\mathrm{ej}}$). We set $\kappa_{\gamma}=0.027$ cm² g^-1 as the opacity for the gamma-ray photons from ⁵⁶Ni and ⁵⁶Co decay (e.g., Cappellaro et al., 1997; Mazzali et al., 2000; Maeda et al., 2003; Dai et al., 2016). Based on the MCMC fitting, the best-fit parameters are shown in Table LABEL:Tab:_Fit_models. This model cannot fit the data well (Fig. 12), with $\chi^{2}/\mathrm{d.o.f}=15$. Moreover, it is physically irrational to have newly-synthesized ⁵⁶Ni heavier than the ejecta themselves. Thus, ASASSN-14ms cannot be explained by the radioactive decay model alone, and an additional energy input is needed. A similar conclusion was also drawn by Vallely et al. (2018) by applying the radioactive-decay model fit to the bolometric light curves of ASASSN-14ms.

Actually, for a core-collapse SN there is an expected upper limit of 0.2 to the mass ratio between the ⁵⁶Ni and ejecta (Umeda & Nomoto, 2008). Thus, we take this as one of the criteria for selecting the optimum parameters in the following fitting process with models involving radioactive decay.

A supernova explosion is likely to result in a millisecond magnetar, whose rotation energy can be as high as $\sim 10^{52}$ erg. Thus, the spindown of a nascent magnetar is widely accepted as an alternative power source for luminous SNe (e.g., Kasen & Bildsten, 2010; Woosley, 2010; Inserra et al., 2013; Nicholl et al., 2014; Wang et al., 2015, 2016).

We adopt the magnetar model developed by Nicholl et al. (2017), where the recession of the photosphere is taken into account. In addition to the free parameters related to radioactive decay input, this model introduces the dipole magnetic field strength ($B$) and the initial spin period ($P_{0}$) of a magnetar. In the process of fitting the model, we fix $M_{\mathrm{Ni}}=0$ owing to the insignificant contribution of ⁵⁶Ni decay to a good fit. The best-fit parameters are reported in Table 6. In this scenario, a rapid rotation in the magnetar can provide a strong wind to power the high luminosity of ASASSN-14ms. Our model requires a magnetar with an initial spin period of 5.6 ms, which is longer than $P_{0}=1$ ms derived by Vallely et al. (2018) based on their data, though both values are within the breakup limit of a neutron star (i.e., $\sim 0.8$ ms; Haensel et al., 1995). The magnetic field strength of the magnetar is inferred to be $2.87\times 10^{14}$ G, which is lower than $B\approx 10^{15}$ G obtained by Vallely et al. (2018). As shown in Figure LABEL:fig:_Compare_LT, the magnetar model can reproduce the post-peak observations except for a slight deviation at t$\sim$20-40 days after V-band maximum light, with a reduced chi-square $\chi^{2}/\mathrm{d.o.f.}=3.5$. Moreover, the magnetar model seems to predict a slow rise time for the light curve, which is inconsistent with the fast-rise trend seen in some SNe Ibn. Therefore, the spindown of a newly-born magnetar might not be the main power source for the emission of ASASSN-14ms.

Inspired by the discoveries of double-peaked light curves in some core-collapse SNe, some studies suggest that cooling emission of shocked extended material may contribute to the first peak preceding the primary one (Nakar & Piro, 2014; Piro, 2015; Piro et al., 2020). In this scenario, the extended envelope, with a mass of $M_{\mathrm{e}}$ and a radius of $R_{\mathrm{e}}$, is shocked by a blast wave (with an energy of $E_{\mathrm{e}}$), and then it radiates by cooling. In addition, part of the core (with a mass of $M_{\mathrm{c}}$) is ejected by the explosion, which sweeps up the extended envelope and they expand outward together. Thus, the total ejecta mass can be defined as $M_{\mathrm{ej}}=M_{\mathrm{c}}+M_{\mathrm{e}}$ (e.g., Vreeswijk et al., 2017).

Here we fit the observations of ASASSN-14ms with a hybrid model involving shock cooling (SC) (Piro, 2015) and radioactive decay (RD) as considered in §4.1. The best-fit results are shown in Figure LABEL:fig:_Compare_LT and the corresponding parameters are presented in Table LABEL:Tab:_Fit_models ($\chi^{2}/\mathrm{dof}=6.9$). Although this model reproduces the late-time luminosity and temperature, it fails to attain a luminosity as high as $10^{44}$ erg s^-1 at early times. Moreover, this fit requires an unusually large mass ($M_{\mathrm{e}}=3.32~{}M_{\odot}$) and radius ($R_{\mathrm{e}}=9.8\times 10^{13}$ cm) for the envelope of the progenitor star, which is inconsistent with the properties of WR stars as implied by other SNe Ibc. It is worth noting that the inferred radius is close to the upper limit that we assumed in the fitting.

Another hybrid model (SC+MG), involving shock cooling and spindown of a magnetar, provides a better fit to the observations of ASASSN-14ms ($\chi^{2}/\mathrm{d.o.f.}=2.6$). In this fit, the early-time emission is primarily due to shock cooling, while a magnetar with $P\approx 5$ ms and $B\approx 6\times 10^{14}$ G is invoked to explain the late-time light curves. The inferred mass of the stellar envelope ($M_{\mathrm{e}}=0.9~{}M_{\odot}$) is more reasonable than that required by the SC+RD model. However, similar to the SC+RD model, this model requires an unusually large radius ($R_{\mathrm{e}}=9\times 10^{13}$ cm) for the envelope of the progenitor star.

An exploding star could possibly be surrounded by dense CSM formed through stellar winds, giant eruptions, binary interaction, or a pulsational pair instability (Smith, 2014; Woosley, 2017, and references therein). The interaction between the CSM and outward ejecta would produce a large amount of radiation as well as narrow emission lines as seen in SNe Ibn, SNe IIn, and some SLSNe with spectra similar to those of SNe IIn (for reviews, see Gal-Yam, 2017, 2019).

In this scenario, we explore the possibility that ASASSN-14ms was powered by not only radioactive decay (see §LABEL:subsec:_rd), but also the forward and reverse shocks produced by the interaction of ejecta with a steady-state wind ($\rho\propto r^{-2}$; e.g., Chatzopoulos et al., 2012, 2013; Vallely et al., 2018; Wang et al., 2019). The steady-state wind could be characterized by mass $M_{\mathrm{CSM}}$, inner radius $r_{1}$, and density at inner radius $\rho_{\mathrm{CSM,1}}$. Here we consider ejecta divided at a dimensionless radius of $x_{0}=0.7$, and adopt $\delta=2$ and $n=7$, which are power-law exponents of the inner and outer density profiles, respectively. In addition, $\epsilon$ is introduced to represent the efficiency of converting the kinetic energy into radiation during the interaction process.

As one can see, the CSM+RD model provides a better fit for the evolution of $L_{\mathrm{bol}}$ and $T_{\mathrm{eff}}$ (Fig. LABEL:fig:_Compare_LT) in comparison with other models, with a relatively small $\chi^{2}/\mathrm{d.o.f.}=2.2$; the best-fit model parameters are listed in Table LABEL:Tab:_Fit_models. We note that this model is also favored for ASASSN-14ms according to the fit by Vallely et al. (2018). However, they focused on the fits to the bolometric light curve while our model consider fits to both bolometric luminosity and effective temperature, which results in different choice about the best-fit parameters. For example, they derive the masses of the ejecta and CSM as $M_{\mathrm{ej}}=4.28M_{\odot}$ and $M_{\mathrm{CSM}}=0.51M_{\odot}$, respectively, which are almost half of those inferred from our model ($M_{\mathrm{ej}}\approx 9.0M_{\odot}$ and $M_{\mathrm{CSM}}=0.9M_{\odot}$). In addition, they require an obviously larger ⁵⁶Ni ($M_{\mathrm{Ni}}=0.23M_{\odot}$) than our result ($M_{\mathrm{Ni}}=0.04M_{\odot}$). Our fitting results based on the CSM+RD model suggest that ASASSN-14ms originated from a progenitor star with a high mass-loss rate, $6.7(v_{\mathrm{w}}/1000$ km s${}^{-1})~{}M_{\odot}$ yr^-1. In principle, it is difficult for a single red supergiant star to drive such a powerful wind except that it suffered intense pulsational ejections before exploding (i.e., Yoon et al., 2010). Although binary interaction could be responsible for such a strong outflow, the velocity expected for this outflow (i.e., 10–100 km s^-1) is quite slow compared to that of the He i lines of ASASSN-14ms ($\sim 1000$ km s^-1; see § LABEL:subsec:_spec). Another competitive progenitor candidate is a luminous blue variable (LBV), which could generate substantial mass loss during its eruptive phases, with a mass-loss rate of $\lesssim 10~{}M_{\odot}$ s^-1 (Smith, 2006). The typical velocity of the wind driven by WR/WN stars is $v_{\mathrm{w}}\sim$100–1000 km s^-1 (Bhirombhakdi et al., 2019, and references therein), in accordance with the He i velocity observed in ASASSN-14ms.

Among the well-studied sample of SNe Ibn, SN 2006jc was observed to undergo a giant LBV-like eruption two years before the SN explosion, consistent with a WR progenitor star evolved from an LBV (Foley et al., 2007; Pastorello et al., 2007). Note that the outbursts detected shortly before some SNe IIn resemble the activity of LBVs, which favors the origin of some SNe IIn from LBVs (e.g., Gal-Yam & Leonard, 2009; Foley et al., 2011; Smith et al., 2011; Thöne et al., 2017). A recent study of SN 2017hcc (SN IIn) indicates that it may have had an LBV progenitor with a mass-loss rate of $0.12(v_{\mathrm{w}}/20$ km s${}^{-1})~{}M_{\odot}$ yr^-1 before the explosion (Kumar et al., 2019). SN 2011hw may represent a transitional event between an SN IIn and an SN Ibn, characterizing by narrow emission features of both helium and hydrogen in the spectra. The presence of some residual H in the spectra of SN 2011hw suggests that its precursor was likely a post-LBV star — a WN-type star (Smith et al., 2012; Pastorello et al., 2015c). The similarity between SN 2006jc and SN 2011hw suggests a possibility that their progenitors were both in the transition stage from an LBV to a WR star but exploded at different phases (Smith et al., 2012).

Therefore, we propose that ASASSN-14ms likely originated from a WR star transited from an LBV, and the surrounding CSM was ejected during an eruption of a WN-type star before the explosion. This scenario could be true, as there have already been detections of LBV-like eruptions from WR stars, such as MCA-1B observed in M33 (Smith et al., 2020, and references therein). A WR+LBV binary is an alternative progenitor system: ejecta driven by the explosion of a WR star expand and finally catch up with CSM ejected by an LBV (e.g., Pastorello et al., 2007), which might lead to a bright transient analogous to ASASSN-14ms. The MCMC fitting results favor this hybrid scenario involving two power sources, suggesting that ASASSN-14ms could result from an explosion of a post-LBV WN-type star (i.e., type WN9-11 or Opfe). It is worth noting that there is inevitably model parameter degeneracy that is hard to break owing to so many parameters involved in this model. Thus, the best-fit results presented here are representative but possibly not the only set of parameters that can reproduce the observations.