betaFeedback

WorksPhysics and Astronomy

A panchromatic photometric and spectroscopic modeling of SN1987A from 200 to 800d

Abstract

Type II supernovae (SNe) at 1-3yr after explosion can document not only the preSN mass loss history of massive stars but also the chemistry of the inner ejecta rich in metals and eventually molecules and dust. With the aim of modeling SN1987A, we present a grid of radiative-transfer models of Type II SNe at Zsun/3, covering an initial mass range between 11 and 21Msun, a fixed ejecta kinetic energy of 1.2e51erg, and ~0.08Msun of 56Ni. The influence of metallicity at such late times is found to be relatively weak. In this model set, the [OI]6330A flux at 1-2yr varies non-monotonically with Minit. It first rises with Minit due to the growing O abundance but drops for M>17-19Msun due to the higher ejecta density, with little change to the rest of the optical spectrum. Assuming CO molecules account for 90% of the cooling in the C/O shell leads to a 10-20% reduction in the [OI] flux, equivalent to an O(1Msun) reduction in Minit at M<19Msun. The fractional [OI] flux rises from 200 to ~600d but decreases beyond. We obtain a good agreement to SN1987A for all Minit apart from [OI], which appears in the optical as the only discriminant between models. The 13Msun model yields the best match, from UV to IR, but a higher Minit may result with allowance for strong CO and SiO cooling. Past 500d, a growing amount of dust below ~2000km/s is essential to account for the ~1.5mag optical excess fading relative to predictions from decay power. Adopting 0.1mic silicate grains, we find a dust mass of 1e-4Msun at 600d rising to a few 1e-3Msun at 700-800d. We also found that the 200-800d International Ultraviolet Explorer spectra of SN1987A are all contaminated by a background hot star, which also causes the relative UV brightening past 500d. Such contamination may complicate the interpretation of more distant SNe II at >~2yr once or if they fade to O(1e5Lsun) [abridged].

Introduction

Recent years have seen a renewed interest in the late-time evolution of Type II supernovae (SNe) past the nominal age of one year after explosion. Rather than continuously fading according to the declining decay power, a number of events have revealed a knee in their optical light curve, with a flattening occurring typically at about 600 d. This photometric change is associated with a dramatic change in spectral morphology from relatively narrow lines forming in the inner metal-rich ejecta to broad, boxy profiles forming in the outer ejecta (e.g., SN 2004et, [46]; SN 2013ej, [58]; SN 2017eaw, [85]). This transformation is thought to result from a shift in the dominant powering mechanism, that is from decay power early on to interaction power later on [21]. The two nearby SNe 2023ixf and 2024ggi have documented this phenomenon and its diversity further and better than ever before. SN 2023ixf exhibited signatures of interaction throughout the photospheric and nebular phase, both in the optical and in the ultraviolet (UV), which persist until today (e.g., [10]; [35]; [49]; [11]; [27]). SN 2024ggi showed signs of interaction with optically-thick circumstellar material early on [43]; [76]; [90] but little sign of interaction during the photospheric phase and the nebular phase prior to about 500 d [7]; [34]; [29]. Only later at ∼600\sim600 d were clear signatures of interaction detected in SN 2024ggi [42]. All such cases reveal the complicated mass-loss histories of Type II SN progenitors, with mass-loss rates well above expectations for red-supergiant stars (e.g., 10–−4 M⊙^{-4}\,M_{\odot} yr−1^{-1} for SN 2023ixf; [67]; [44]; [11]).

While this new window on late-stage massive star mass loss is of much interest, there is also a greater need to better understand the emission from the inner ejecta powered by radioactive decay alone. Such signatures can help constrain the progenitor mass and thus document the connection between mass loss and initial mass [37]; [45]; [25]. The optical spectrum at intermediate times of 300–600 d is also a mix of two emitting regions (i.e., inner and outer ejecta; [21]; [26]) and thus a proper modeling of each contribution is needed to understand the whole. Such modeling can also shed light on the formation, in unshocked metal-rich gas, of molecules [16]; [77]; [13]; [2]; [74]; [55] and dust [57]; [88]; [62]), as observed in SN 1987A. A final interest is to extend previous radiative-transfer simulations of nebular-phase SNe II to metallicities below the solar value – we start here with simulations at Z⊙/3Z_{\odot}/3.

SN 1987A remains undoubtedly the most studied SN to date and we will only briefly summarize the key relevant information for the work presented here on its nebular phase evolution (for a comprehensive review on SN 1987A, see, for example, Arnett et al. [6]). SN 1987A was monitored photometrically out to a few years (Hamuy et al. [39]; Hamuy & Suntzeff [38]; Bouchet et al. [14]; Walker & Suntzeff [84]; Bouchet & Danziger [12]), from which optical, infrared (IR), and tentative bolometric light curves could be constructed (Suntzeff et al. [82]; this led to the identification of 57Co^{57}\mathrm{Co} as an extra power source past ∼1000\sim1000 d). Spectroscopic monitoring was carried out at high cadence for several years and covering not just the optical, but also the near- and mid-IR (Catchpole et al. [15, 16]; Phillips et al. [70]; Spyromilio et al. [77]; Roche et al. [74]; Spyromilio et al. [79]; Meikle et al. [62]), as well as the UV with the International Ultraviolet Explorer (IUE; Pun et al. [72]). These data revealed the formation of CO and SiO molecules after about 200 d, but also of dust through the identification of emission profile asymmetry and optical fading after about 500 d (Lucy et al. [56]).

Most of the modeling of the SN radiation has focused on the light curve, from which constraints on the ejecta mass, 56Ni^{56}\mathrm{Ni} mass, as well as the 56Ni^{56}\mathrm{Ni} mixing have been made (see, for example, Woosley [89]; Blinnikov et al. [9]; Eastman & Kirshner [31]; Utrobin [83]). Early on, fluid instabilities developing after envelope shock passage have been recognized as a key ingredient for chemical mixing (Mueller et al. [66]; Ebisuzaki et al. [32]), augmented by the intrinsic asymmetries of the neutrino-driven explosion mechanism (see, for example, Wongwathanarat et al. 2015), in conformity, for example, with spatially resolved observations of SN 1987A taken many years after explosion (Abellán et al. [1]). The merger scenario is currently favored for SN 1987A (e.g., Menon & Heger [63]).

The modeling of line emission and gas properties in SN 1987A at nebular times allowed to set some constrains on the progenitor mass (i.e., 15–25 M⊙M_\odot), the physics of these ejecta at late times, as well as the inhomogenous, clumpy structure of these metal-rich inner ejecta (Fransson & Chevalier [36]; Li et al. 1993; Kozma & Fransson [47]). Modeling of the CO first overtone and fundamental constrained the amount of CO (see, e.g., Spyromilio et al. [77]; Liu et al. [55]). Modeling of the emission line profiles past 500 d constrained the location and amount of newly-formed dust in the inner ejecta (see, e.g., Lucy et al. [56]; Bevan & Barlow [8]; Dessart et al. [28]).

Here, we revisit previous work on the nebular-phase properties of SN 1987A by performing detailed radiative-transfer calculations, including dust at times, to model the UV to IR properties of SN 1987A between 200 and 800 d after explosion. In the next section, we present the numerical setup for our work, including the stellar evolution models performed at Z⊙/3Z_\odot/3, the treatment of the explosion, and the radiative transfer for the resulting ejecta. In Section 3, we present the ejecta, photometric, and spectral properties for our radiative-transfer models based on the explosion of stars born with a mass between 11 and 21 M⊙M_\odot. This includes a discussion of the nonmonotonicity of the [O I] λλ\lambda\lambda 6300.3, 6363.8 flux with MinitM_{\rm init} (§ 3.2), an assessment of the impact of CO cooling (§ 3.3), as well as the impact of metallicity (§ 3.4). We then compare in Section 4 our results to the observations of SN 1987A between 200 and 800 d after explosion, considering the “raw” models as well as counterparts in which some 0.1 μ\mum silicate dust is introduced in the inner ejecta below 2000 km s−1^{-1}. We conclude in Section 5. Additional illustrations are presented in Appendix § 6.

Numerical setup

PreSN evolution and explosion

The massive-star evolution models presented in this work were performed in a similar fashion to that described in Sukhbold et al. [80], with a few specific changes. All simulations were run with solar abundances scaled by a factor of 1/3, that is all elements from carbon and above have their abundance multiplied by 1/3. Rotation was ignored. The parameter αMLT\alpha_{\rm MLT} controlling convection was set to two. The wind mass loss rates were adopted from the recipes given in Nieuwenhuijzen & de Jager [68], with an additional metallicity scaling of (Z/Z⊙)0.5(Z/Z_\odot)^{0.5} or 0.577 for the present case.

The nuclear network included species up to bismuth (Z=83Z=83) and carried a maximum of 1703 isotopes in roughly 2000 zones. The detailed composition was evolved in a coproccessing mode. That is, the abundances were updated each zone cycle for the existing temperature, density, and convection, although the energy generation from the big nuclear network was not directly coupled to the stellar structure calculation. Energy generation for the stellar structure came from the 19 isotope network up until oxygen depletion and from the silicon quasi-equilibrium network after that. At the time of core collapse, all these models were red supergiant stars. While this would be unsuitable for the modeling of the photospheric phase properties of a Type II-peculiar SN like 1987A, it is of secondary importance for the nebular-phase modeling presented here (see, e.g., Dessart et al. [25]).

At the time of core collapse, the stars were exploded using a piston placed at the Lagrangian mass where the total entropy moving out first exceeded 4.0 kBk_{\rm B} per baryon. The piston was moved to impart the desired kinetic energy near 1.2×10511.2 \times10^{51} erg. Because these models produced a 56Ni^{56}\mathrm{Ni} mass of 0.1 M⊙M_\odot or more, the inner ejecta layers were excised such that the ejecta used for the radiative transfer calculations would all have 0.07–0.08 M⊙M_\odot as suitable for SN 1987A (see, e.g., Woosley [89]). The full grid contains explosion models from initial stellar masses between 11 and 21 M⊙M_\odot, ejecta masses between 9.1 and 17.1 M⊙M_\odot, a 56Ni^{56}\mathrm{Ni} mass of ∼0.075 M⊙\sim0.075\,M_\odot, and an O yield between 0.3 and 2.6 M⊙M_\odot – a summary of this model set is presented in Table 1.

ModelEkinE_{\rm kin} [foe]MejM_{\rm ej} [M⊙M_\odot]Hec_{\rm c} [M⊙M_\odot]H [M⊙M_\odot]He [M⊙M_\odot]12^{12}C [M⊙M_\odot]16^{16}O [M⊙M_\odot]Ne [M⊙M_\odot]24^{24}Mg [M⊙M_\odot]28^{28}Si [M⊙M_\odot]40^{40}Ca [M⊙M_\odot]52^{52}Fe [M⊙M_\odot]56^{56}Ni [M⊙M_\odot]57^{57}Ni [M⊙M_\odot]58^{58}Ni [M⊙M_\odot]
s15.20.8410.954.465.243.931.70(-1)1.002.20(-1)4.41(-2)8.65(-2)7.77(-3)6.95(-4)6.33(-2)2.79(-3)3.68(-3)
th151.1812.734.736.304.351.82(-1)1.182.59(-1)4.77(-2)1.48(-1)1.67(-2)1.20(-3)7.70(-2)2.33(-3)1.83(-3)
th111.179.133.275.213.229.63(-2)0.325.08(-2)2.49(-2)5.80(-2)4.32(-3)8.54(-4)7.84(-2)2.71(-3)1.48(-3)
th131.1710.824.045.673.751.51(-1)0.761.41(-1)3.67(-2)1.10(-1)9.24(-3)1.19(-3)7.79(-2)2.31(-3)2.61(-3)
th151.1812.734.736.304.351.82(-1)1.182.59(-1)4.77(-2)1.48(-1)1.67(-2)1.20(-3)7.70(-2)2.33(-3)1.83(-3)
th171.1714.235.546.714.891.92(-1)1.673.67(-1)6.96(-2)1.13(-1)1.02(-2)1.37(-3)7.61(-2)2.00(-3)1.53(-3)
th191.1915.306.297.165.463.02(-1)1.672.08(-1)4.34(-2)2.06(-1)1.09(-2)2.01(-3)7.67(-2)1.42(-3)1.16(-3)
th211.2117.107.147.355.962.51(-1)2.581.42(-1)4.83(-2)3.73(-1)1.53(-2)1.74(-3)7.86(-2)1.76(-3)1.02(-3)

Table 1. Summary of ejecta properties used as initial conditions for our radiative-transfer calculations. In addition to these models at Z⊙/3Z_\odot/3, we include the model s15.2 at Z⊙Z_\odot ([80, 25]) for comparison to th15 (see § 2 for discussion.)

Compared to the models of Sukhbold et al. [80], some differences arise (see th15 vs. s15.2 in the top rows of Table 1). Some of these are well understood. For example, the lower metallicity leads systematically to smaller mass loss rates and a larger preSN mass than the solar metallicity models computed in Sukhbold et al. [80]. This work also used updated pair neutrino losses (Sukhbold et al. [81]) that differ from those employed in Sukhbold et al. [80]. However, any small change, for example in zoning or time step, can send the star down a different evolutionary path (Farmer et al. [33]; Sukhbold et al. [81]), such that two stars with similar initial masses can have different preSN structures and fate. Hence one should not expect models s15.2 and th15 to have identical properties at core collapse.

Radiative transfer

Using the ejecta models described in the previous section as initial conditions, we performed 1D nonlocal thermodynamic equilibrium time-dependent radiative transfer from 200 to 800 d with CMFGEN (Hillier & Dessart 2012). In order to allow for macroscopic mixing in spherical symmetry (and without microscopic mixing) the original unmixed composition structure at 200 d was transformed into a shuffled-shell structure following the procedure described in Dessart & Hillier [22]. This complex struc-tion Notes: The columns give in turn the modelname (“s” stands for solar metallicity, “th” stands for third solar metallicity, followed by the model initial mass in M⊙M_\odot), the kinetic energy, the ejecta mass, the progenitor He-core mass, and the ejecta yields of important specific isotopes in order of increasing atomic mass (i.e., H, He, 12^{12}C, 16^{16}O, Ne, 24^{24}Mg, 40^{40}Ca, 48^{48}Cr, 52^{52}Fe, 56^{56}Ni, 57^{57}Ni, and 58^{58}Ni). For He, we give the sum of 3^{3}He and 4^{4}He, and for Ne we give the sum of 20^{20}Ne and 22^{22}Ne. Also, 1 foe ≡1051\equiv10^{51} erg.

temperature was sampled with a total of about 350 grid points to finely resolve all composition gradients as well as the optical-depth scale. The 56^{56}Ni-bubble effect (e.g., Woosley [89]), which leads to a strong expansion of 56^{56}Ni-rich regions but a modest compression of the surrounding material (typically a factor of two), was ignored (see discussion in § 5 of Dessart et al. [25]). Interaction with CSM was ignored such that the only power source at those nebular times was radioactive decay. Here, we accounted for the five two-step decay chains whose leading isotope are (in order of increasing atomic mass) 44^{44}Ti, 48^{48}Cr, 52^{52}Fe, 56^{56}Ni, and 57^{57}Ni. In order of importance, the 56^{56}Ni decay chain dominates at all times considered, but the 57^{57}Ni-decay chain is no longer negligible after about 500 d (and would dominate after about 1500 d; see Figure 1 of Dessart et al. [25]). Nonthermal processes were treated as by Li et al. (2012) and Dessart et al. [23]. The non-local energy deposition was calculated by solving the radiative-transfer equation with a gray absorption-only opacity to γ\gamma-rays set to 0.06 Ye cm2 g−10.06\,Y_{\rm e}\ {\rm cm}^{2}\ {\rm g}^{-1}. Ejecta clumping was ignored in this work.

Figure 1. Shuffled-shell composition profile versus Lagrangian mass for H, He, O, Si, Ca, and Fe in model th13 at 200 d and used as initial conditions for the radiative transfer calculations.

Since the ejecta gas conditions are cool and at most partially ionized, we limited the model atom to ionization states I–III for most species. In order of increasing atomic weight and then ionization, we included H I (26,36), He I (40,51), He II (13,30), C I (14,26), C II (14,26), N I (44,104), N II (23,41), O I (21,51), O II (54,123), Ne I (78,155), Ne II (22,91), Ne III (32,80), Na I (22,71), Mg I (39,122), Mg II (31,80), Si II (100,187), Si III (31,59), S I (106,322), S II (56,324), Ar I (56,110), Ar II (14,344), K I (25,44), Ca I (76,98), Ca II (21,77), Sc I (26,72), Sc II (38,85), Sc III (25,45), Ti II (37,152), Ti III (33,206), Cr II (28,196), Cr III (30,145), Fe I (413,1142), Fe II (228,2698), Fe III (96,1001), Co II (112,1005), Co III (88,1075), Ni I (56,301), Ni II (59,1000), Ni III (47,1000), Ba I (17,31), and Ba II (28,61). Here, the numbers in parentheses correspond to the number of super levels and full levels (for additional information on super levels, see Hillier & Miller [41]).

Molecules are ignored in this work. In § 3.3, we explore the potential impact of CO molecule formation on our results, in particular on the strength of the [O I] λλ\lambda\lambda 6300, 6363 Å line flux. Dust radiative transfer is however performed since it is a critical driver of the evolving radiative properties of SN 1987A after about 500 d ([56]). Practically, we used the same formalism as in Dessart et al. [28]. Dust is introduced as a post-processing step in a given converged CMFGEN calculation by specifying its spatial distribution, type, and temperature. Exploratory calculations were presented in Dessart et al. [28], which are now extended to more epochs, but using the same spatial distribution within the inner 2000 km s−1^{-1} of the ejecta, thus the metal-rich regions only. Theoretically, silicate grains are expected to form first and as small grains (Sarangi [75]) so we used 0.1 μ\mum silicates. The opacities for these dust grains are adopted from Sarangi [75], and originally from Draine & Li [30], Weingartner & Draine [86] and Zubko et al. [91]. In this work, we constrained the dust present in the ejecta primarily through its combined absorptive and scattering influence on the spectral energy distribution, thus considering the impact of dust on all line profiles between the UV and IR. Given the many degeneracies in this exercise, we did not explore with alternate dust types – we tried with a larger grain size of 1.0 μ\mum but this produced little difference.

Ejecta, photometric, and spectral properties for our model set th11 to th21

A detailed presentation of nebular-phase phase properties of Type II SN gas and radiation properties was provided in Dessart et al. [25] and Dessart (2025) – see also Fransson & Chevalier [37] and Jerkstrand et al. [45]. In this section, we focus only on a few specific items and present additional results alongside the comparison to SN 1987A in § 4. We used model th13 for most of the present discussion since it best matches SN 1987A.

Ejecta properties

Figure 2 shows the evolution of ejecta properties for model th13 from 200 to 800 d. Conditions are marginally nebular at 200 d (total ejecta electron-scattering optical depth τes\tau_{\mathrm{es}} of 1.28) and eventually fully nebular at 800 d (i.e., τes=0.011\tau_{\mathrm{es}} = 0.011). The temperature is initially ∼5000\sim5000 K, with the He-rich and H-rich regions above 6000 K and the Fe-rich regions below 4000 K. This complicated temperature profile arises from the fact that the conditions are optically thin and the coolants vary from shell to shell – more metal-rich regions typically have greater cooling power and thus have lower temperatures (Fig. A.1 in the appendix shows the dominant coolants in model th13 at 350 and 700 d). As time passes, the temperature exhibits qualitatively the same profile but drops everywhere as the heating source (i.e., radioactive decay) exponentially declines. By the end of the simulation at 800 d, the temperature minima are at 800–1000 K and located in the Fe-rich shells while the maxima are around 3500 K in the He-rich and H-rich regions. The O-rich material is at about 1500 K.

Figure 2. Evolution of ejecta properties for model th13 from 200 to 800 d. From top to bottom, we show the composition of the main species followed by the temperature and ionization of O and Fe. A reported ionization of zero corresponds to a neutral state, one to once ionized etc. We only show the inner ejecta where the bulk of the decay power is absorbed.

The bottom two panels of Figure 2 illustrate the ionization profile of O and Fe versus velocity. Higher ionizations for all species are encountered in the He/N shell. In the O-rich shell, the O ionization is initially about 0.4 but drops to below 0.1 at 800 d. Fe is once ionized throughout most of the ejecta, apart from the He/N shell (Fe2+^{2+}) and the Fe-rich regions (mostly neutral ion). These changes are relatively modest over such a long time span and will lead to a weak evolution of spectral properties with lines mostly from neutral and once ionized species.

Figure 3 illustrates the evolution of the decay power absorbed in the ejecta for our set of models th11 to th21 (i.e., MinitM_{\mathrm{init}} of 11 to 21 M⊙M_{\odot}). Although we included five two-step decay chains, the main contributor at those late times is 56^{56}Ni, and to a lesser extent 57^{57}Ni (e.g., ∼4\sim4% of the total at 800 d in model th13). Full trapping of γ\gamma-rays holds at 200 d but their escape starts at about that time such that only 20% (50%) of γ\gamma-rays are absorbed by the ejecta at 800 d in models th11 (th21). Since all ejecta models have the same EkinE_{\mathrm{kin}}, same 56^{56}Ni mass and distribution, these differences arise from the different MejM_{\mathrm{ej}} (i.e., 9.13 up to 17.10 M⊙M_{\odot} from th11 to th21). The larger is the γ\gamma-ray escape, the greater is the contribution from positrons, which varies from ∼12\sim12% in model th11 to ∼5\sim5% in model th21. Similarly, the growth of the γ\gamma-ray mean free path with time leads to reduction of the decay power absorbed in the O-rich gas (only present in the metal-rich inner ejecta) relative to the H-rich gas (present both in the inner, metal-rich ejecta and in the outer ejecta beyond 2000 km s−1^{-1}).

Figure 3. From top to bottom, evolution of the total decay power absorbed by the ejecta, the fraction of the total power emitted that is absorbed, the percentage fraction of the decay power absorbed by the ejecta that is kinetic (i.e., positrons), and the percentage fraction of the decay power absorbed in the O-rich and H-rich shells.

These variations will lead to a strengthening of cooling from the H-rich gas over the 200 to 800 d timespan, as well as a strengthening of emission from the Fe-rich gas where positrons are absorbed. Together with the changes in temperature and ionization, the spectral energy distribution arising from these models

els evolves significantly in time (Figure 4). In model th13, 62 % (38 %) of the emergent flux is channeled into the optical between 3500 Å and 1 µm (infrared, or > 1 µm) at 200 d, with essentially zero flux in the UV below 3500 Å (Figure 4). At 800 d, the fraction falling in the optical has dropped to 40 %, with about the same fraction in the IR (45 %) but now 15 % in the UV.

Figure 4. Cumulative flux versus wavelength and time in the nebular phase for the model th13. A grey shading indicates the ultraviolet and IR ranges on each side of the optical range.

The [O i] λλ 6300.3, 6363.8 doublet

Of particular interest is the evolution of the [O i] λλ 6300.3, 6363.8 line flux at nebular times since it is routinely used to evaluate the progenitor mass (e.g., [37, 45, 25]. In the context of SN 1987A, the analysis of the [O i] λλ 6300.3, 6363.8 line flux was also used to set constraints on the oxygen abundance, clumping, and optical depth effects (see, e.g., [78, 17] (Li & McCray 1992)).

In our simulations, following the increase in O abundance, or more specifically the size of the O-rich shell (i.e., the combination of the O/Si, O/Ne/Mg, and O/C shells), a greater fraction of the decay power is absorbed by the O-rich material for increasing MinitM_{\rm init} (Figure 3). If the ionization is relatively low in the O-rich shell, [O i] λλ 6300.3, 6363.8 may be the dominant coolant. If the conditions are identical in the O-rich shell in ejecta arising from the explosion of stars of different MinitM_{\rm init}, that [O i] flux will scale with MinitM_{\rm init}. However, complications may enter when the ionization is too high, in which case Fe ii may be an important coolant of the O-rich material, or too low, in which case Na i λλ 5896, 5890 or Mg i] λ 4571.1 may take over – this situation is however more relevant in stripped-envelope SNe ([24]). Another complication is related to the density. Forbidden lines thrive at low density, which is the reason behind the delayed emergence at about 200 d of [O i] λλ 6300.3, 6363.8 in SNe II, and even more so in low-energy Type II SNe because of their slower expanding, denser inner ejecta (see, e.g., SN 2005cs; [69]).

Furthermore, optical depth effects remain important even at those nebular epochs. First, τes is still about unity in all models th11 to th21 at 200 d, so electron scattering alone can introduce some effect. The lines themselves may remain optically thick out to late times. In model th13 at 200 d, the line optical depth of [O i] λ 6300 is about 0.3 and thus barely optically thin whereas that of [O i] λ 6363 is about 0.1. The [O i] λλ 6300.3, 6363.8 doublet is also affected by overlapping Fe ii lines. As a result, the

Figure 6. Evolution of the fractional [O i] λλ 6300.3, 6363.8 line luminosity relative to total optical luminosity for our model set and given as a percentage. Filled dots indicate the epochs at which the models were computed.

evolution of [O i] λλ 6300.3, 6363.8 from 200 to 800 d is quite complicated and that it is only at 500–600 d that this emission appears strong, largely unaffected by Fe, and with the expected blue-to-red ratio of 3.1 (i.e., the ratio of the A values for these transitions; Figure 5). At times past 600 d, the reduction in decay power absorbed in the O-rich shell causes the Fe emission relative to [O i] λλ 6300.3, 6363.8 and alters its shape and flux significantly.

Figure 5. Evolution of the [O i] λλ 6300.3, 6363.8 line flux (grey shading) for model th13 from 200 to 800 d. The rest of the flux in this regions is due to Fe (primarily Fe ii), apart from the depression at ≃ 4200 Å (and weakening with time passing) corresponding to the He trough. Time is ordered from left to right and top to bottom.

Figure 6 shows the evolution of the fractional [O i] λλ 6300.3, 6363.8 luminosity (relative to the optical luminosity). It rises in all models until 500–600 d as the ionization and density decline but eventually drops beyond as other coolants kick in (e.g., the boost to the UV flux Figure 4; lines in the IR such as [Ne ii] 12.810 µm; see also (Dessart 2025)). Although the [O i] line flux increases with MinitM_{\rm init} in the mass range 11 to 17 M⊙M_\odot, further increase of MinitM_{\rm init} leads to its decrease. The [O i] line flux is in fact comparable for models th11 and th21 which stand 10 M⊙M_\odot apart on the main sequence, and correspond to a final O yield of 0.32 M⊙M_\odot (th11) compared to 2.58 M⊙M_\odot (th21). The origin of this non-monotonicity is that despite the growing O yield with MinitM_{\mathrm{init}}, the density of the O-rich material also grows and eventually quenches the formation of forbidden lines such [O i] λλ\lambda\lambda 6300.3, 6363.8. This highlights an important degeneracy of Type II SN nebular-phase spectra in their use as progenitor mass diagnostics.

Evaluating the influence of CO molecule formation and cooling

In this work as well as in most other works on nebular-phase modeling of SNe II, CO molecule formation and cooling are ignored in the global radiative transfer solution. Liljegren et al. [54] studied molecular formation (e.g., of CO and related molecules) in SNe but this calculation was not coupled to the full radiative transfer solution. Similarly, modeling of the CO first overtone and fundamental (see technique outlined in Mera et al. [65]) were performed by Medler et al. [60] for SN 2022acko and by Mera et al. [64] for SN 2024ggi using spectra from the James Webb Space Telescope. Exceptions to this are McLeod et al. [59] for Type II SN models and Liljegren et al. [53] for Type Ibc SN models. Furthermore, the impact of CO cooling is generally hard to gauge from SN observations since these typically capture the CO first overtone in the KK band (e.g., Rho et al. [73]) but generally lack the stronger fundamental at ≳4.3 μm\gtrsim4.3\,\mu\mathrm{m} (e.g., Spyromilio et al. [77]), and even more rarely document the emission from SiO at longer wavelengths (see, for example, Roche et al. [74]).

But one can test the maximum impact of molecular cooling on SN spectra without doing any molecular modeling. Indeed, in practice, CO cooling should “radiate away” a fraction of the decay power absorbed by the O-rich material and thus reduce the [O i] λλ\lambda\lambda 6300.3, 6363.8 line flux, thereby impacting the inferred progenitor mass. One can gauge the impact of CO cooling by assuming that CO is the main coolant of the O/C shell (the situation is more complicated since CO may be only a partial coolant, but of both the C/O shell and the O/Ne/Mg shell; Liljegren et al. [54]). In Dessart et al. [29], we estimated the reduction on the [O i] λλ\lambda\lambda 6300.3, 6363.8 line flux based on the ratio M(O/C-shell)/M(O-shell)M(\mathrm{O/C\text{-}shell})/M(\mathrm{O\text{-}shell}), which was at most 30 % for M>12 M⊙M > 12\,M_\odot in the massive star models of Sukhbold et al. [80].

Here, we directly evaluate the impact on the spectra by recalculating a sample of our models with a decay-power absorbed in the C/O shell scaled by a factor of 0.1 – this is equivalent to assuming that CO is responsible for 90 % of the cooling in the C/O shell – obviously this is not going to produce any CO emission in the IR but because it drains the power absorbed in the C/O shell it can mimic the impact CO would have if present. Figure 7 shows the corresponding optical spectra for model th15 from 200 to 700 d (left column) and for models th11 to th21 at 350 d. Surprisingly, we find that the impact is limited to the [O i] λλ\lambda\lambda 6300.3, 6363.8 emission (there is no noticeable effect in the ultraviolet or IR) and amounts to a reduction of order 10 % in that line flux. Thus, in the mass range between 11 and about 19 M⊙M_\odot, this would correspond to an underestimate of the inferred progenitor mass of order 1 M⊙M_\odot. This error is therefore small. It also suggests that when modeling SNe II at nebular epochs without CO cooling, one ought to select a model that overestimates the [O i] λλ\lambda\lambda 6300.3, 6363.8 line flux by about 10–20 % (an even greater overestimate should be allowed for if CO cooling affects the O/Ne/Mg shell or if SiO also forms). This is what we will do for the modeling of SN 1987A in § 4 by selecting model th13.

Figure 7. Exploration of the potential impact of CO cooling on the model th15 from 200 to 700 d (left) and on models th11 to th21 at 350 d (right). In each subfigure, the left panel shows the optical spectrum and the right panel zooms in on the [O I] λλ\lambda\lambda 6300.3, 6363.8 region. CO cooling is mimicked here by assuming it accounts for a maximum of 90 % of the cooling of the C/O shell.

The weak influence of metallicity

The new grid of models at one third the solar metallicity returns a paradoxical result in the negligible impact of that metallicity variation on nebular-phase spectra. Figure 8 illustrates this weak dependence for model th15 (this work) and the solar metallicity counterpart s15.2 (Sukhbold et al. [80]; Dessart et al. [25]) – this s15.2 model was recomputed with a better metal atom and more accurate composition in Dessart et al. [29]. The fluxes have been normalized by the decay-power absorbed in each model in order to cancel the slight offset in 56Ni{}^{56}\mathrm{Ni} mass between the two models (0.063 vs. 0.077 M⊙M_\odot; see Table 1).

Figure 8. Spectral comparison of the models th15 (at Z⊙/3Z_\odot/3) and s15p2 (at Z⊙Z_\odot) and at a post-explosion time of 350 d. The spectra are first normalized to their associated decay-power absorbed at that time, scaled by λ2\lambda^2 to yield a flat flux across optical and IR, and then divided by the maximum flux in model th15. The spectra were also rebinned to a resolution of 600 km s−1^{-1}. Line identifications are provided for most emission features such that this figure can be used for line identifications elsewhere in the document.

Evidently, the offset between the two models is extremely weak and limited to differences in lines from Fe, Co, and Ni in the IR range, which results from the differences in the yields from explosive nucleosynthesis (e.g., 58Ni{}^{58}\mathrm{Ni}), but also the differences in EkinE_{\mathrm{kin}} and MejM_{\mathrm{ej}}. Unlike the strong variations predicted during the photospheric phase (Dessart et al. [20]) and observed (see, e.g., Anderson et al. [5]; Anderson et al. [4]; Coulter et al. [18]), there is no variation from the forest of Fe lines contributing most of the flux below 5500 Å. It will be interesting to explore whether going to one solar would produce an effect. It may be that at nebular times, the spectrum would be indirectly affected by metallicity through changes in preSN evolution, for example with a prolonged and stronger influence of stellar rotation.

Comparison to SN 1987A

We now compare the simulations from the above sections to the observations of SN 1987A from 200 to 800 d. We adopted an E(B−V)E(B-V) of 0.15 mag, a redshift of 0.00096, and a distance of 49.59 kpc (Pietrzyński et al. [71]). The photometric data that we used for the VV band is from Hamuy & Suntzeff [38]. We did not include other photometric comparisons because we compare our radiative transfer results to the full spectral energy distribution, covering from UV to the IR whenever possible, and thus our models are compared to data both in absolute and relative terms. The UV spectra from the IUE were taken from Pun et al. [72].1 The optical spectra are from Phillips et al. [70]. The infrared spectra were provided by Jason Spyromilio (priv. comm.) and originate from Meikle et al. [61, 62], Roche et al. [74], and Wooden et al. [88]. A slight correction (i.e., at the 10 % level) was applied to the optical spectra so that they matched the contemporaneous VV-band magnitude from Hamuy & Suntzeff [38].

Figure 9 shows a comparison between the optical spectrum of SN 1987A at 340 d and the model spectra for our set th11 to th21 at 350 d. Strikingly, all models match the observations throughout the optical with the exception of [O i] λλ\lambda\lambda 6300.3, 6363.8 in essentially all models. The match is good throughout for th11, but [O i] λλ\lambda\lambda 6300.3, 6363.8 is increasingly overestimated as we progress from th13 to th17, and that discrepancy is then reduced for the higher initial masses 19 and 21 M⊙M_\odot. This non-monotonicity was discussed earlier on in § 3.2. Since CO cooling is ignored in our simulations, we selected a model that overestimates the [O i] λλ\lambda\lambda 6300.3, 6363.8 flux by 10–20 %, that is model th13. Model th21 could be another option but its large MejM_{\mathrm{ej}} may be in tension with the light curve properties.

Figure 9. Spectral comparison between the observations of SN 1987A at 340 d and models th11 to th21 at 350 d. The observations have been corrected for reddening and redshift and the models have been scaled to the distance of SN 1987A. No other scaling was applied.

Figure 10 then shows how this model th13 compares with the VV-band light curve of SN 1987A. The offset with the data is less than 0.2 mag at times prior to about 550 d but then grows to reach 1.2 mag at 800 d. The knee in the light curve at 500 d is not associated with γ\gamma-ray escape since this varies smoothly in time (see Figure 3). The left panel of Figure 11 shows the spectral evolution of this same model th13 from 200 to 800 d. The model spectra are in agreement with observations until 500 d and subsequently deviate, with eventually a strong discrepancy at 800 d. The main offset is not a poor match of line profile shapes. It is instead primarily a strong overestimate of the observed flux by a factor two or more throughout the optical range.

Comparison of the observed V-band light curve of SN 1987A with models

Figure 10. Top panel: Comparison of the observed VV-band light curve of SN 1987A with the results for models th13 without (solid) and with (dashed) dust. The model photometry was corrected for extinction and distance. Bottom panel: Same as top but now showing the magnitude offset between observations and models.

Figure 11. Left: Spectral comparison between the observations of SN 1987A spanning from 198 until 806 d and model th13 at roughly contemporaneous epochs – dust is ignored for the models in that left column. The observations have been corrected for reddening and redshift and the models have been scaled to the distance of SN 1987A. No other scaling was applied. For the ordinate, the quantity FunitF_{\mathrm{unit}}, in units of erg cm−2^{-2} s−1^{-1} Å−1^{-1}, is 10−1010^{-10} at 200 and 350 d, 10−1110^{-11} at 500 d, 10−1210^{-12} at 600 d, and 10−1310^{-13} at 700 and 800 d. Right: Same as left, but first, an increasing amount of dust was added to the model th13 after 500 d. At all epochs, we added to the SN model spectrum a fixed contamination corresponding to 3 % of the flux from a blue supergiant star model (i.e., “SK-68D140” from Alkousa et al. [3]). [See § 4 for discussion.]

As indicated by Lucy et al. [56], newly-formed dust in the inner, metal-rich ejecta of SN 1987A is at the origin of this unexpected change in radiative properties of SN 1987A. Since small silicate dust grains are likely to form first at about 500 d in Type II SNe (Sarangi [75]), we recomputed the model th13 after distributing 0.1 μ\mum silicate dust grains uniformly below 2000 km s−1^{-1}. For each epoch from 500 to 800 d, we computed cases with a dust mass MdM_{\rm d} between 10−410^{-4} and 5×10−3 M⊙5 \times10^{-3}\,M_\odot. In Figure 10 and the right column of Figure 11, we compare these dusty th13 models with the VV-band light curve and the UV and optical spectra of SN 1987A. By invoking a dust mass of 10−410^{-4}, 4×10−34 \times10^{-3}, and 5×10−3 M⊙5 \times10^{-3}\,M_\odot, at 600, 700, and 800 d, the now dusty model th13 matches well the VV-band light curve with an offset from observations that is within 0.2 mag at all times. We also obtain a satisfactory match of the spectra throughout the optical range, in particular with the strong reduction in [O I] λλ\lambda\lambda 6300.3, 6363.8 and the presence of a relatively strong [Fe II] emission below 5500 Å. Evidently, the small blue-red profile asymmetries introduced by dust are not perfectly fitted (for a detailed modeling of the impact of dust on Hα\alpha, see, for example, Bevan & Barlow [8]) but our model yields a global consistency – the global reduction of the flux by dust is also much greater than the blue-red asymmetry on line profiles (which may also encode potential asymmetries of the ejecta density and composition). We also tested with an increased outer velocity bounding the dust at 3000 km s−1^{-1} but this caused a more global attenuation of the spectra (the entire ejecta emission suffered dust attenuation) whereas the observations support a differential influence of the dust, with a greater impact on the emission from the metal-rich ejecta and less so of the H-rich material.

Not discussed so far is the rising UV flux (i.e., relative to the optical) in the IUE data of SN 1987A at late nebular times (Pun et al. [72]). The UV flux is vanishingly small prior to 500 d apart from a weak Ly α\alpha emission. At 600 d, a flat UV flux is clearly present, and it continues to grow until the epoch shown here at 800 d. While the UV flux increases in the (dust-free) model th13 at those late times (see Figure 4), it does not rise by as much. To match the observed evolution in the IUE range, we introduced a small fraction of the flux (i.e., 3 %) from the blue supergiant spectrum model “SK-68D140” of Alkousa et al. [3] characterized by an effective temperature of 24 100 K, a log gg of 2.81, and a log L/L⊙L/L_{\odot} of 5.52. Pun et al. [72] suggested that the IUE spectra were contaminated by background stars after ∼ 900 d but our radiative transfer calculations for SN 1987A suggest that this contamination is clearly present at ≳600\gtrsim600 d – inspecting the spectra in log-space indicates that this contamination is present 200 d onwards (see below, the appendix, and in particular Fig. A.4–A.6).

This contamination is not surprising. At late times, the bolometric luminosity drops to ∼ 104010^{40} erg s−1^{-1} at 500 d and is as low as ∼ 103810^{38} erg s−1^{-1} at 800 d. The latter is comparable to the luminosity of any supergiant star and it is clear that the SN luminosity may be drowned by contaminating light. However, this contamination appears much earlier in the UV, when the SN is more luminous but UV faint, in particular if the contamination arises from a hot star like a blue supergiant. A hot (i.e., of order 20 kK) massive star can then outshine SN 1987A in the UV at nebular times. This contamination would be weak in the UV and stronger in the optical if the contaminating star was a red supergiant or a cluster of such cooler stars.

In the appendix, we show additional illustrations of the model comparison to the UV, optical, and IR observations of SN 1987A, together with variation of the dust temperatures at 200, 250, 300, 350, and 400 K (Figs. A.4–A.6). Higher dust temperatures are necessary at earlier times (i.e., ∼ 400 K at 600 d compared to 250 K at 700 and 800 d), although there might be a SiO contribution to the IR emission, thus unrelated to thermal dust emission ([74, 88]). In our simulations of dust in the model th13, we find that the dust is optically thick throughout the spectrum from the ultraviolet to the infrared. Figure 12 illustrates the evolution of the total ejecta dust optical depth in model th13 at 800 d for 5×10−3 M⊙5 \times10^{-3}\,M_{\odot} of 0.1 μm silicates distributed uniformly in the inner ejecta below 2000 km s−1^{-1}. This leads to a more modest IR emission compared to optically-thin conditions for the same dust mass, but also means that blue-red profile asymmetries as well as flux attenuation affects all lines in the UV and IR. As shown in Figs. A.4–A.6, with our choice of dust, we tend to overestimate the blue-red asymmetry of NIR line profiles. Reducing the dust mass may improve the match in the NIR but would deteriorate the match in the optical at 600–800 d. An alternative could be to reduce the grain size to 0.01 μm, as this would boost the opacity in the UV relative to the optical and IR (our models do tend to overpredict the flux around 3000 Å), but this would introduce a mismatch in the optical where the dust opacity would then be smaller. Given all these complications and the relatively good agreement of our models with the observations of SN 1987A, we leave it at that. The comparison of the dusty-model th13 with SN 1987A also shows a rough agreement to the main lines present in the IR (Figs. A.4–A.6), but it is difficult to be more specific given the very low resolution and poor signal of the observations in that wavelength range.

Figure 12. Variation of the total ejecta dust optical depth associated with scattering and absorption in model th13 at 800 d for 5×10−3 M⊙5 \times10^{-3}\,M_{\odot} of 0.1 μm silicates distributed uniformly in the inner ejecta below 2000 km s−1^{-1}.

Conclusions

We have presented nonlocal thermodynamic equilibrium time-dependent radiative-transfer models for ejecta resulting from the explosion of progenitor stars initially between 11 and 21 M⊙M_{\odot} and evolved at Z⊙/3Z_{\odot}/3. Since the original aim was to confront observations of SN 1987A, we adopted an EkinE_{\rm kin} of 1.2×10511.2 \times10^{51} erg and a 56^{56}Ni mass of 0.07–0.08 M⊙M_{\odot}. Macroscopic mixing in the ejecta was enforced by adopting a shuffled-shell composition structure. The times considered covered from 200 to 800 d. Five two-step decay chains were included, including those led by 56^{56}Ni and 57^{57}Ni. Molecular formation and cooling were ignored but dust in the form of 0.1 μm silicate grains was introduced as a post-treatment stage. In contrast to previous similar work, the possibility of interaction with circumstellar material was ignored in order to focus on the evolution of the emission from the inner, metal-rich and decay-powered ejecta – no other power source was included. Previous work like Dessart (2025) studied similar configurations but were limited to earlier times (i.e., ≲500\lesssim500 d), to solar-metallicity progenitors, ignored dust, and did not include any detailed comparison to observations. However, we found little difference in results with the solar-metallicity counterparts so that aspect is little studied in this work – going to even lower metallicity (Z⊙/10Z_{\odot}/10?) or working with rotating stars may lead to visible variations but this is left to future work.

Our simulations follow a considerable drop in bolometric luminosity from ∼1041\sim10^{41} erg s−1^{-1} at 200 d down to ∼1038\sim10^{38} erg s−1^{-1} at 800 d, and thus down to luminosities typical not of one SN but of standard massive stars. Over that time span, γ-ray escape increases from essentially zero to 50–80 %, the more so for lower MinitM_{\mathrm{init}} (all models have the same EkinE_{\mathrm{kin}} but differ in MejM_{\mathrm{ej}}), with a growing fraction of positrons from 3 % to 6–13 %, and an increasing (decreasing) fraction of the decay power absorbed by the H-rich (O-rich) material. While the optical range contains the bulk of the flux at all times, both the UV and the IR contain a growing fraction of the total. In some sense, the spectral energy distribution increasingly spreads out in wavelength from UV to IR as time passes.

The [O i] λλ\lambda\lambda 6300.3, 6363.8 line flux shows a complicated trend in our simulations. It shows a rise from 200 to 500–600 d followed by a decline, probably as a result of the decrease of the decay power absorbed in the O-rich material. The fractional line flux in [O i] λλ\lambda\lambda 6300.3, 6363.8 grows for increasing MinitM_{\mathrm{init}} up to 17–19 M⊙M_{\odot}, but it decreases for larger masses as a result of the greater density in the inner ejecta of more massive progenitors. This non-monotonicity compromises the use of that line flux for progenitor mass determination, in particular in its ability to exclude higher-mass progenitors as the source of SNe II. These results reveal an important degeneracy of nebular-phase spectra of SNe II.

We also evaluated the potential impact that CO cooling would have on nebular-phase spectra. Using the simplistic assumption that CO molecules would cause 90 % of the cooling of the C/O shell, we found that the impact was essentially limited to the [O i] λλ\lambda\lambda 6300.3, 6363.8 line flux, with a small reduction of 10–20 %. This amounts to a potential underestimate of the inferred MinitM_{\mathrm{init}} by about 1 M⊙M_{\odot} for models below 17 M⊙M_{\odot} and that ignore CO cooling. CO cooling may also affect the O/Ne/Mg shell and would in that case quench further the [O i] line flux as well as IR lines like [Ne ii] 12.810 μ\mum. And such effects would be exacerbated with the formation of SiO and the associated cooling. All such coolants would tend to reduce the [O i] line flux and lead to an underestimate of MinitM_{\mathrm{init}} – this would allow a higher mass model (i.e., th15 or perhaps th17) as a better match for SN 1987A than our current better-match model th13.

The comparison to SN 1987A led to some surprises. As has been advocated by single-line studies, the blue-red asymmetry of line profiles like Hα\alpha after 500 d requires the presence of dust. In our best match model th13, we computed dust radiative transfer models from UV to IR and found that dust is indeed essential to match not just profile asymmetries but the global attenuation of the entire spectrum, by about 1.5 mag in the VV-band. The observed fading in SN 1987A at 500 d cannot arise from γ\gamma-ray escape, which has a more gradual evolution. Invoking an increase in mass of 0.1 μ\mum silicates from about 10−410^{-4} M⊙M_{\odot} at 600 d to a few 10−310^{-3} M⊙M_{\odot} at 800 d reconciles the light curve and spectra of our model th13 with SN 1987A. The greater attenuation of lines like [O i] λλ\lambda\lambda 6300.3, 6363.8 compared to Hα\alpha or Fe emission from the H-rich material requires the dust to be relatively deep in the ejecta, typically below 2000 km s−1^{-1} – placing it below 3000 km s−1^{-1} deteriorates the model agreement with SN 1987A. Some discrepancies remain for some lines, for example around 3000 Å, which arises from emission by Fe lines at the base of the outer H-rich ejecta.

Given all these uncertainties, we thus adopted our (dust-forming) 13 M⊙M_{\odot} Z⊙Z_{\odot}/3 model as best matching the nebular phase properties of SN 1987A. Its O yield is 0.79 M⊙M_{\odot} and thus smaller than inferred by early studies of SN 1987A ([78, 17] (Li & McCray 1992)), but the level of physical consistency of the present work is also greater. Furthermore, we neglect any clumping of the ejecta material, which might also seem to contradict those studies. In reality, this arises in part from a conflict in language. Indeed, even though we neglect clumping, that is we do not invoke any process that could compress any material (e.g., we neglect the 56^{56}Ni bubble effect), there is strong segregation of material in the adopted shuffled-shell composition of our models. For example, in model th13 the volume bounded by a velocity of 2500 km s−1^{-1} is occupied at a 37 % level by H-rich material, at a 29 % level by He-rich material, at a 30 % level by O-rich material, and at a 4 % level by Fe-rich material. That is, the volume filling factor of O-rich material is 30 %, and thus no dissimilar to the 10 % volume filling factor inferred by Spyromilio & Pinto [78]. However, this is not clumping since the term generally implies some compression of the gas. This configuration instead reflects the inhomogeneous nature of the composition within the inner, metal-rich regions of the ejecta (Figure 1). In fact, the original progenitor composition at core collapse is similarly characterized by distinct shells (i.e., H-rich, He-rich etc) in mass space so that only a fraction is occupied by each shell within some given Lagrangian mass. Whether this material is organized in spherical shells as in our 1D setup or in individual, uncompressed blobs in the ejecta is simply a question of spatial arrangement.

A 13 M⊙M_{\odot} progenitor lies at the lower end of the proposed range in the literature. The progenitor luminosity came chiefly from the helium shell and supported the He core mass to be 5 to 7 M⊙M_{\odot} from the inferred luminosity of 3–6 ×1038\times10^{38} erg s−1^{-1} ([6]). The model th13 has a He-core mass of 4.04 M⊙M_{\odot}, and a luminosity of 3.21 ×1038\times10^{38} erg s−1^{-1} at core collapse, so low but within the bounds ([89]). We conducted tests with clumping (i.e., compression) of the O-rich material and found that it can contribute to reducing the [O i] λλ\lambda\lambda 6300.3, 6363.8 line flux, but at the expense of Na i λ\lambda 5890, 5896 and Mg i] λ\lambda 4571, which would then be overpredicted relative to SN 1987A. Revisions upward of this progenitor mass might arise from a proper treatment of molecular cooling, since both CO and SiO would drain some of the [O i] λλ\lambda\lambda 6300.3, 6363.8 flux. Fully coupled molecule formation and cooling within the radiative transfer is however challenging in Type II SN ejecta ([59]).

Importantly, we found that the IUE spectra of SN 1987A are dominated by a contamination from a background hot star at all times in the 200–800 d interval considered here. Indeed, although our models brighten in the UV relative to the optical as time passes, the observed UV brightening (in a relative sense) of SN 1987A is much greater. By adding a constant flux from a B 0.7 Ia supergiant model (scaled by 0.04), we reconcile models and UV observations of SN 1987A. Such contamination is typically ignored in the interpretation of observations (although it is clearly a recognized challenge) but could be a major limitation for the understanding of late time observations of Type II SN at larger distances than SN 1987A. Once or if these have reached down to powers of 103810^{38} erg s−1^{-1}, the potential contamination in the UV from nearby blue supergiants or in the optical from red supergiants becomes a concern.

Acknowledgements. LD thanks Stan Woosley for fruitful discussions and for computing the stellar evolution and explosion models used as initial conditions for the radiative transfer calculations presented in this work. LD thanks Jason Spyromilio for providing the IR data of SN 1987A. LD acknowledges support from the ESO Scientific Visitor Program for a visit to ESO-Garching during the summer of 2025 and during which this work was initiated. This work was granted access to the HPC resources of TGCC under the allocations 2024 – A0170410554 and 2025 – A0190416871 on Irene-Rome made by GENCI, France. This research has made use of NASA’s Astrophysics Data System Bibliographic Services.

Additional illustrations

In this appendix, we provide additional information to complement what was discussed in the main text. Figure A.1 present the main coolants in different regions of the ejecta, primarily controlled by the large variations in composition in the adopted shuffled-shell structure of our models. Figures A.2 and A.3 present a break down of the main contributors to the (dust-free) th13 spectrum from UV to IR at 350 and 800 d. While elements of low atomic weight tend to contribute only a few lines, heavier elements and in particular those from the iron group contribute forests of lines. Figures A.4–A.6 present a UV-to-IR comparison of the best-matching th13 model, augmented by a contaminating blue supergiant star model and at times by dust, and at post-explosion epochs of 200, 350, 500, 600, 700, and 800 d. For better visibility, the spectra may be shown in the log, as FλF_{\lambda}, or as FνF_{\nu}.

Figure A.1. Top: Illustration of ejecta and radiative properties for model th13 at 350 d. The top row shows the mass fraction of H, He, C, O, Ne, Si, and Fe as a function of velocity, which reveals the location and extend of the main shells. Also shown is the normalized cumulative absorbed decay power integrated outwards from the innermost ejecta layer (light grey line). The bottom row shows the variation of the local cooling power, as a percentage of the local total heating power, for the top 14 coolants ejecta-wide. The term “COL” refers to collisional excitation, “FF” to free-free emission, and “NT” to nonthermal excitation. The term “AC.R(V).” refers to cooling from expansion. Bottom: Same as top but now at 700 d.

Figure A.2. Spectral decomposition of model th13 at 350 d showing the main contributors to the spectrum. The spectra were scaled to the distance of SN 1987A. The shading corresponds to the flux arising from the corresponding species (i.e., all ions from the species indicated at top right of each panel) relative to the continuum flux. Over wide spectral regions, this species-dependent flux deviates from the local flux because of overlap with other species (e.g., absorption or emission from Fe lines etc).

Figure A.3. Same as Fig. A.2 but now for model th13 at 800 d.

Figure A.4. Spectral comparison between SN 1987A and model th13 at about 200 d (top) and 350 d (bottom). The observations were corrected for redshift and reddening and the models were scaled to the distance to SN 1987A as well as rebinned to a resolution of 500 km s−1500\ \mathrm{km}\ \mathrm{s}^{-1}. Dust is ignored at both epochs but the model spectra are augmented with a fixed (non-time-evolving) contribution amounting to 3 % of the flux from the B 0.7 Ia model named “SK-68D140” [3 shown as a dashed grey line].

Figure A.5. Same as Fig. A.4 but now at 500 d (top) and 600 d (bottom). The 600 d model includes dust from 10−4 M⊙10^{-4}\,M_\odot 0.1 μm0.1\,\mu\mathrm{m} silicates at 400 K and distributed uniformly below 2000 km s−1^{-1}.

Figure A.6. Same as Fig. A.4 but now at 700 d (top) and 800 d (bottom). Both models include dust from 4 and 5 ×10−3 M⊙\times10^{-3}\,M_{\odot} 0.1 μm\mu\mathrm{m} silicates at 250 K and distributed uniformly below 2000 km s−1\mathrm{s}^{-1}.

References

  1. [1]Abellán, F. J., Indebetouw, R., Marcaide, J. M., et al. 2017, ApJL, 842, L24
  2. [2]Aitken, D. K., Smith, C. H., James, S. D., et al. 1988, MNRAS, 231, 7P
  3. [3]Alkousa, T., Crowther, P. A., Bestenlehner, J. M., et al. 2025, A&A, 699, A314
  4. [4]Anderson, J. P., Dessart, L., Gutiérrez, C. P., et al. 2018, Nature Astronomy, 2, 574
  5. [5]Anderson, J. P., Gutiérrez, C. P., Dessart, L., et al. 2016, A&A, 589, A110
  6. [6]Arnett, W. D., Bahcall, J. N., Kirshner, R. P., & Woosley, S. E. 1989, ARA&A, 27, 629
  7. [7]Baron, E., Ashall, C., DerKacy, J. M., et al. 2025, ApJ, 994, 249
  8. [8]Bevan, A. & Barlow, M. J. 2016, MNRAS, 456, 1269
  9. [9]Blinnikov, S., Lundqvist, P., Bartunov, O., Nomoto, K., & Iwamoto, K. 2000, ApJ, 532, 1132
  10. [10]Bostroem, K. A., Sand, D. J., Dessart, L., et al. 2024, ApJL, 973, L47
  11. [11]Bostroem, K. A., Valenti, S., Sand, D. J., et al. 2026, ApJ, 1004, 23
  12. [12]Bouchet, P. & Danziger, I. J. 1993, A&A, 273, 451
  13. [13]Bouchet, P., Moneti, A., Slezak, E., Le Bertre, T., & Manfroid, J. 1989, A&AS, 80, 379
  14. [14]Bouchet, P., Phillips, M. M., Suntzeff, N. B., et al. 1991, A&A, 245, 490
  15. [15]Catchpole, R. M., Menzies, J. W., Monk, A. S., et al. 1987, MNRAS, 229, 15P
  16. [16]Catchpole, R. M., Whitelock, P. A., Feast, M. W., et al. 1988, MNRAS, 231, 75P
  17. [17]Chugai, N. N. 1994, ApJL, 428, L17
  18. [18]Coulter, D. A., Larison, C., Pierel, J. D. R., et al. 2026, submitted to Science, arXiv:2601.04156arxiv.org/abs/2601.04156
  19. [20]Dessart, L., Gutiérrez, C. P., Hamuy, M., et al. 2014, MNRAS, 440, 1856
  20. [21]Dessart, L., Gutiérrez, C. P., Kuncarayakti, H., Fox, O. D., & Filippenko, A. V. 2023, A&A, 675, A33
  21. [22]Dessart, L. & Hillier, D. J. 2020, A&A, 643, L13
  22. [23]Dessart, L., Hillier, D. J., Li, C., & Woosley, S. 2012, MNRAS, 424, 2139
  23. [24]Dessart, L., Hillier, D. J., Sukhold, T., Woosley, S. E., & Janka, H. T. 2021a, A&A, 656, A61
  24. [25]Dessart, L., Hillier, D. J., Sukhold, T., Woosley, S. E., & Janka, H. T. 2021b, A&A, 652, A64
  25. [26]Dessart, L., Jacobson-Galán, W. V., Bostroem, K. A., et al. 2026a, A&A, in press, arXiv:2605.14081arxiv.org/abs/2605.14081
  26. [27]Dessart, L., Jacobson-Galán, W. V., Bostroem, K. A., et al. 2026b, arXiv:2604.05145arxiv.org/abs/2604.05145
  27. [28]Dessart, L., John Hillier, D., & Sarangi, A. 2025a, A&A, 698, A293
  28. [29]Dessart, L., Kotak, R., Jacobson-Galán, W., et al. 2025b, A&A, 704, L6
  29. [30]Draine, B. T. & Li, A. 2007, ApJ, 657, 810
  30. [31]Eastman, R. G. & Kirshner, R. P. 1989, ApJ, 347, 771
  31. [32]Ebisuzaki, T., Shigeyama, T., & Nomoto, K. 1989, ApJL, 344, L65
  32. [33]Farmer, R., Fields, C. E., Petermann, I., et al. 2016, ApJS, 227, 22
  33. [34]Ferrari, L., Folatelli, G., Ertini, K., et al. 2025, A&A, 703, A12
  34. [35]Folatelli, G., Ferrari, L., Ertini, K., Kuncarayakti, H., & Maeda, K. 2025, A&A, 698, A213
  35. [36]Fransson, C. & Chevalier, R. A. 1987, ApJL, 322, L15
  36. [37]Fransson, C. & Chevalier, R. A. 1989, ApJ, 343, 323
  37. [38]Hamuy, M. & Suntzeff, N. B. 1990, AJ, 99, 1146
  38. [39]Hamuy, M., Suntzeff, N. B., Gonzalez, R., & Martin, G. 1988, AJ, 95, 63
  39. [41]Hillier, D. J. & Miller, D. L. 1998, ApJ, 496, 407
  40. [42]Hueichapán, E., Cartier, R., Prieto, J. L., et al. 2026, ApJ, 1000, 80
  41. [43]Jacobson-Galán, W. V., Davis, K. W., Kilpatrick, C. D., et al. 2024, ApJL, 972, 177
  42. [44]Jacobson-Galán, W. V., Dessart, L., Kilpatrick, C. D., et al. 2025, ApJL, 994, L14
  43. [45]Jerkstrand, A., Fransson, C., Maguire, K., et al. 2012, A&A, 546, A28
  44. [46]Kotak, R., Meikle, W. P. S., Farrah, D., et al. 2009, ApJ, 704, 306
  45. [47]Kozma, C. & Fransson, C. 1998a, ApJ, 496, 946
  46. [49]Kumar, A., Dastidar, R., Maund, J. R., Singleton, A. J., & Sun, N.-C. 2025, MNRAS, 538, 659
  47. [53]Liljegren, S., Jerkstrand, A., Barklem, P. S., et al. 2023, A&A, 674, A184
  48. [54]Liljegren, S., Jerkstrand, A., & Grumer, J. 2020, A&A, 642, A135
  49. [55]Liu, W., Delgarn, A., & Lepp, S. 1992, ApJ, 396, 679
  50. [56]Lucy, L. B., Danziger, I. J., Gouiffes, C., & Bouchet, P. 1989, in IAU Colloq. 120: Structure and Dynamics of the Interstellar Medium, ed. G. Tenorio-Tagle, M. Moles, & J. Melnick, Vol. 350, 164
  51. [57]Lucy, L. B., Danziger, I. J., Gouiffes, C., & Bouchet, P. 1991, in Supernovae, ed. S. E. Woosley, 82
  52. [58]Mauerhan, J. C., Van Dyk, S. D., Johansson, J., et al. 2017, ApJ, 834, 118
  53. [59]McLeod, C., Hillier, D. J., & Dessart, L. 2024, MNRAS, 532, 549
  54. [60]Medler, K., Mera, T., Ashall, C., et al. 2026, arXiv e-prints, arXiv:2606.31176arxiv.org/abs/2606.31176
  55. [61]Meikle, W. P. S., Allen, D. A., Spyromilio, J., & Varani, G. F. 1989, MNRAS, 238, 193
  56. [62]Meikle, W. P. S., Spyromilio, J., Allen, D. A., Varani, G. F., & Cumming, R. J. 1993, MNRAS, 261, 535
  57. [63]Menon, A. & Heger, A. 2017, MNRAS, 469, 4649
  58. [64]Mera, T., Ashall, C., Hoeflich, P., et al. 2026a, ApJ, 997, 330
  59. [65]Mera, T., Hoeflich, P., Burns, C. R., et al. 2026b, ApJ, 1003, 135
  60. [66]Mueller, E., Fryxell, B., & Arnett, D. 1991, A&A, 251, 505
  61. [67]Nayana, A. J., Margutti, R., Wiston, E., et al. 2025, ApJ, 985, 51
  62. [68]Nieuwenhuijzen, H. & de Jager, C. 1990, A&A, 231, 134
  63. [69]Pastorello, A., Valenti, S., Zampieri, L., et al. 2009, MNRAS, 394, 2266
  64. [70]Phillips, M. M., Hamuy, M., Heathcote, S. R., Suntzeff, N. B., & Kirhakos, S. 1990, AJ, 99, 1133
  65. [71]Pietrzyński, G., Graczyk, D., Gallenne, A., et al. 2019, Nature, 567, 200
  66. [72]Pun, C. S. J., Kirshner, R. P., Sonneborn, G., et al. 1995, ApJS, 99, 223
  67. [73]Rho, J., Geballe, T. R., Banerjee, D. P. K., et al. 2018, ApJL, 864, L20
  68. [74]Roche, P. F., Aitken, D. K., & Smith, C. H. 1991, MNRAS, 252, 39P
  69. [75]Sarangi, A. 2022, A&A, 668, A57
  70. [76]Shrestha, M., Bostroem, K. A., Sand, D. J., et al. 2024, ApJL, 972, L15
  71. [77]Spyromilio, J., Meikle, W. P. S., Learner, R. C. M., & Allen, D. A. 1988, Nature, 334, 327
  72. [78]Spyromilio, J. & Pinto, P. A. 1991, in ESO Conference and Workshop Procs, ed. I. J. Danziger & K. Kjær, Vol. 37, 423
  73. [79]Spyromilio, J., Stathakis, R. A., Cannon, R. D., et al. 1991, MNRAS, 248, 465
  74. [80]Sukhbold, T., Ertl, T., Woosley, S. E., Brown, J. M., & Janka, H.-T. 2016, ApJ, 821, 38
  75. [81]Sukhbold, T., Woosley, S. E., & Heger, A. 2018, ApJ, 860, 93
  76. [82]Suntzeff, N. B., Phillips, M. M., Elias, J. H., Depoy, D. L., & Walker, A. R. 1992, ApJL, 384, L33
  77. [83]Utrobin, V. 1993, A&A, 270, 249
  78. [84]Walker, A. R. & Suntzeff, N. B. 1991, PASP, 103, 958
  79. [85]Weil, K. E., Fesen, R. A., Patnaude, D. J., & Milisavljevic, D. 2020, ApJ, 900, 11
  80. [86]Weingartner, J. C. & Draine, B. T. 2001, ApJ, 563, 842
  81. [88]Wooden, D. H., Rank, D. M., Bregman, J. D., et al. 1993, ApJS, 88, 477
  82. [89]Woosley, S. E. 1988, ApJ, 330, 218
  83. [90]Zhang, J., Dessart, L., Wang, X., et al. 2024, ApJL, 970, L18
  84. [91]Zubko, V., Dwek, E., & Arendt, R. G. 2004, ApJS, 152, 211

Paper details

vlnc:W_vubcc58t
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 10

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents