Nebular-phase spectra of superluminous supernovae: physical insights from observational and statistical properties

Abstract

We study the spectroscopic evolution of superluminous supernovae (SLSNe) later than 100 days after maximum light. We present new data for Gaia16apd and SN2017egm, and analyse these with a larger sample comprising 41 spectra of 12 events. The spectra become nebular within 2-4 $e$-folding times after light curve peak, with the rate of spectroscopic evolution correlated to the light curve timescale. Emission lines are identified with well-known transitions of oxygen, calcium, magnesium, sodium and iron. SLSNe are differentiated from other Type Ic SNe by a prominent O I $\lambda$7774 line and higher-ionisation states of oxygen. The iron-dominated region around 5000 \AA\ is more similar to broad-lined SNe Ic than to normal SNe Ic. Principal Component Analysis shows that 5 `eigenspectra' capture 75% of the variance, while a clustering analysis shows no clear evidence for multiple SLSN sub-classes. Line velocities are 5000--8000 km/s, and show stratification of the ejecta. O I $\lambda$7774 likely arises in a dense inner region that also produces calcium emission, while [O I] $\lambda$6300 comes from further out until 300--400 days. The luminosities of O I $\lambda$7774 and Ca II suggest significant clumping, in agreement with previous studies. Ratios of [Ca II]$\lambda$7300/[O I]$\lambda$6300 favour progenitors with relatively massive helium cores, likely $\gtrsim 6$ M$_\odot$, though more modelling is required here. SLSNe with broad light curves show the strongest [O I] $\lambda$6300, suggesting larger ejecta masses. We show how the inferred velocity, density and ionisation structure point to a central power source.