Hydrogen-Poor Superluminous Supernovae in the Nebular Phase: Spectral Diversity Due to Ejecta Ionization as a Probe of the Power Source

TL;DR

This work provides the largest nebular-phase spectral census for hydrogen-poor SLSNe to date, linking late-time ionization signatures to the magnetar engine and progenitor mass loss. By measuring lines near 6300, 7300, and 7774 Å and their ratios, the study reveals a continuum of ejecta ionization, from highly ionized cases with [O II] and [O III] to mostly neutral inner regions, and shows that instantaneous magnetar power strongly modulates this ionization, especially for lower ejecta masses. A robust early-time correlation emerges between the spectral ionization proxy $L_{7300}/L_{6300}$ and light-curve timescales, establishing a link between early energy deposition and late-time spectra. The findings support magnetar-driven ionization as a dominant driver of nebular-phase properties while highlighting clumping as a natural consequence of engine interaction, with rare outliers likely powered by late-time CSM interaction; no late-time hydrogen emission is detected in the sample, constraining hydrogen-rich shell scenarios.

Abstract

We present a large sample of 39 nebular-phase optical spectra of 25 hydrogen-poor superluminous supernovae (SLSNe-I) and jointly analyze them with previously published spectra of 12 events. We measure the properties of key emission features, namely those at 6300, 7300, and 7774 angstroms (associated with [O I], [Ca II]/[O II], and O I, respectively), and find that SLSNe exhibit much wider spectral diversity than normal SNe Ic, primarily in the line ratio $L_{7300}/L_{6300}$, which is highly sensitive to ejecta ionization. Some events exhibit weak [O I] and a clear [O II] contribution to the 7300 angstrom feature, enhancing the ratio, along with [O III] lines at 4363 and 5007 angstroms. Other SLSNe show weak or no lines of ionized oxygen. Moreover, we find that the population exhibits decreasing $L_{7300}/L_{6300}$ over time, while a few outliers instead display sustained high or increasing ratios for extended periods. The ratio $L_{7300}/L_{6300}$ is also correlated with the rise and decline times of the light curves, with slower events exhibiting higher ionization, the first robust connection between early light curve and late-time spectral properties, likely due to the magnetar's impact: slower-evolving SLSNe are generally powered by engines with longer spin-down timescales, which deposit more energy at later phases. Among the events with decreasing $L_{7300}/L_{6300}$, SLSNe with high ionization are on average powered by magnetars with higher thermalized spin-down power, a correlation that is most significant for events with $M_{\rm ej}\lesssim12$ M$_{\odot}$. The ionization in the outliers with increasing $L_{7300}/L_{6300}$ may be due to late CSM interaction. $L_{7300}/L_{6300}$ and its evolution are therefore key diagnostics of SLSN engines and progenitor mass loss.

Figures (17)

• Figure 1: Comparison of our nebular sample (blue) magnetar parameters (left) and light curve rise and decline timescales (right) with those for the full SLSN light curve sample (gray; excluding bronze events) from Gomez2024. The nebular sample is representative of nearly the full range of properties, though with a slight over-representation of slowly evolving SLSNe.
• Figure 2: Spectra of our SLSN sample in order of normalized phase (rest-frame days since peak normalized by the characteristic decline time $t_{\rm d}$). Original and smoothed spectra are shown. Temporal differences are clearly apparent, with spectra at $t/t_{\rm d}\lesssim3$ exhibiting broad, underdeveloped lines at 6300 Å (the location of [O1]) with often strong emission at 7300 Å (the location of [Ca2] and [O2]). A subset of these show obvious [O3] lines at 4363 and 5007 Å (SN 2020abjc, SN 2022le). At later phases, the 6300 Å emission is narrower and comparable to or stronger than the emission at 7300 Å. In addition, emission near the O1 recombination line at 7774 Å is present in events with sufficient coverage and S/N; this emission exhibits a red shoulder in several cases.
• Figure 3: Spectra of our sample zoomed-in on the 6300 Å feature with the single (orange) or multi (green) component Gaussian fits to the line profiles shown. The location of the [O1] $\lambda6300$ doublet is marked with vertical dashed lines. The phases correspond to rest-frame days since peak.
• Figure 4: Line width (FWHM) versus central wavelength for the fits shown in Figure \ref{['fig:OI']} with the location of the [O1] $\lambda6300$ doublet marked (blue dashed lines). The green markers are profiles for which the best fit consists of two or three components (the values for each component are connected by a dashed line). The transparency of the points is proportional to the normalized phase of the spectra, with lighter shades corresponding to $t/t_{\rm d}<4$ and solid points to $t/t_{\rm d}>4$. A temporal trend is apparent where spectra taken at earlier phases have broad profiles, corresponding to high velocities, that are systematically centered redward of [O1]. Several events with lower velocities have components with an appreciable blueshift.
• Figure 5: Same as Figure \ref{['fig:OI']} but for the 7300 Å feature. The location of the [Ca2] $\lambda7300$ doublet is marked with vertical dashed lines. The profiles that are best fit with two components (green) mostly show excess emission redward of [Ca2] with the exception of SN 2020wnt which has a prominent component blueward.