The Type I Superluminous Supernova Catalogue II: Spectroscopic Evolution in the Photospheric Phase, Velocity Measurements, and Constraints on Diversity

The Big Picture

A superluminous supernova (SLSN) can outshine a normal supernova by a factor of 100. These aren’t just brighter versions of ordinary stellar explosions. They’re a distinct class of cosmic event, blazing for months across billions of light-years, and astronomers still can’t fully explain what powers them.

Normal supernovae get their light from radioactive nickel forged in the blast. SLSNe are so luminous they’d require several solar masses of nickel, yet their spectra show no trace of it. Something else must be at work: possibly a rapidly spinning, ultra-magnetized neutron star called a magnetar, or a violent collision between the explosion and material the dying star had previously shed. The debate has run for over a decade with no clear winner.

An international team led by Aysha Aamer at Queen’s University Belfast has now assembled the largest collection of SLSN spectra ever compiled. A spectrum is a fingerprint of light spread out by wavelength, revealing composition, temperature, and velocity. The team gathered 974 spectra from 234 SLSNe and applied statistical machine learning to search for hidden structure in the data.

Key Insight: By treating the spectra of 234 superluminous supernovae as a statistical population rather than individual curiosities, the team found that SLSNe are unexpectedly uniform beneath their apparent diversity, and that uniformity points toward a single class of internal engine powering them all.

How It Works

The team drew data from three sources: the ePESSTO+ survey, the FLEET transient search program at Harvard, and every published SLSN spectrum through December 2022. The resulting database is roughly three times larger than any previous SLSN spectral compilation.

SLSNe change rapidly, and different spectra were taken at different points in each explosion’s evolution. To compare them fairly, the researchers normalized timing not just relative to peak brightness but relative to the explosion itself, then scaled by each event’s characteristic decline time. Think of it as comparing aging across individuals by biological age rather than calendar date.

With this alignment in hand, the team constructed phase-binned average spectra: composite snapshots of what a typical SLSN looks like at each evolutionary stage.

• At peak brightness: SLSNe burn at 10,000–11,000 K with a steep blue continuum and weak absorption features. Most elements are fully ionized and can’t absorb light efficiently. A characteristic W-shaped feature from doubly ionized oxygen between 3,700 and 4,650 Å marks this phase.
• 40 days after peak: Temperatures fall to 5,000–6,000 K. Strong P-Cygni features appear, double-humped profiles where expanding gas absorbs light on one side and re-emits it on the other.
• A subtler signal: The average spectra suggest that some fraction of SLSNe carry helium into the explosion, a clue about their progenitor stars that hadn’t been established at population scale before.

To probe diversity more rigorously, the team turned to Principal Component Analysis (PCA), a technique that distills a complex dataset down to its most essential axes of variation. Think of it as finding the fewest “ingredients” needed to reconstruct any spectrum in the collection. When spectra were grouped using the explosion-scaled timeline rather than peak-brightness dates, PCA needed fewer components to capture the same fraction of variance. The population was more coherent than it first appeared.

Next came K-means clustering, an algorithm that sorts data into groups by similarity. Despite a handful of outliers with signs of interaction with surrounding material, the analysis turned up no statistically significant evidence for multiple distinct subclasses. SLSNe, for all their surface-level variety, form a single continuous population.

The final piece came from velocity measurements. The team tracked Fe II λ5169, an iron absorption line that traces the photosphere (the visible “surface” of the expanding explosion). Its velocity closely matched the radius inferred from blackbody fits, confirming the iron line forms right at the photosphere.

They also confirmed a correlation between a SLSN’s velocity and how quickly that velocity declines. All SLSNe undergo homologous expansion, where each layer of gas moves outward at a speed proportional to its distance from the center, like sections of a uniformly inflating balloon. The difference between events is one of scale: faster-expanding SLSNe slow down faster, while slower ones coast more gently. That behavior is exactly what you’d expect from an internal engine, a magnetar spinning down, pumping energy into the ejecta from the inside out.

Why It Matters

If magnetars really are the engines behind SLSNe, these explosions become rare laboratories for physics that can’t be recreated on Earth: magnetic fields a trillion times stronger than a hospital MRI, rotation rates of hundreds of cycles per second, energy output that dwarfs the Sun’s entire luminosity. The statistical uniformity reported here strengthens the case that one mechanism governs most or all of them.

The machine learning methods at work here (PCA, clustering, variance decomposition) aren’t exotic tricks bolted onto an astronomy problem. They’re the right tools for making sense of a dataset no human could assess by eye. With the Vera Rubin Observatory expected to discover thousands of new SLSNe over the coming decade, this kind of statistical framework will be essential. The 234 events analyzed here will look like a pilot study.

Bottom Line: The largest spectroscopic survey of superluminous supernovae to date finds they form a single, coherent population whose velocity structure and spectral evolution both point to internal powering, most likely a magnetar engine, rather than external interaction as the dominant energy source.