Possible evidence for a pair-instability supernova nature of ultra-early JWST sources

TL;DR

The paper investigates Capotauro, a bright ultra-high-redshift source originally tagged as a $z\sim30$ galaxy but later shown to vary over ~800 days. It demonstrates that a metal-free pair-instability supernova (PISN) from a $\sim250$–$260\,M_\odot$ progenitor at $z\sim15$ can reproduce Capotauro’s brightness, time evolution, SED, and observed NIRSpec spectrum, while a cool Y0 brown dwarf can also mimic the spectrum but cannot explain the variability. By comparing PISN light curves, SEDs, and spectra to the data, the study identifies He130-like PISN models as favored and outlines observational tests—repeat NIRCam photometry, deep MIRI imaging, and high-resolution spectroscopy—to decisively distinguish between PISN and brown dwarf scenarios. The work assesses the expected PISN rate in the CEERS field, finding a non-negligible but uncertain chance of detecting such events, and argues that confirming Capotauro as a PISN would provide a rare window into Population III star evolution and the role of transients in ultra-high-redshift samples.

Abstract

Recent JWST observations have revealed a population of unexpectedly bright sources at ultra-high redshift ($z > 15$), challenging current models of early galaxy formation. One extreme example is 'Capotauro', an F356W-dropout identified in the CEERS survey and initially interpreted as a luminous galaxy at $z\sim30$, but subsequently found to be variable over an $\sim 800$ day baseline. Motivated by this variability, we explore the alternative hypothesis that Capotauro is a pair-instability supernova (PISN) originating from a massive ($250-260\,M_\odot$), metal-free star. Using state-of-the-art PISN light curves, spectral energy distributions, and synthetic spectra, we show that a PISN at $z\simeq 15$ can plausibly reproduce the observed brightness, temporal evolution, photometry, and NIRSpec spectrum. We compare this scenario with alternative interpretations, including a local Y0 brown dwarf, and discuss observational tests to discriminate among them. If confirmed, this event would provide a rare window onto Population III stars, and highlights the importance of transient contamination in ultra-high redshift galaxy samples.

Figures (4)

• Figure 1: Best-fit redshift light curves at $4.44\ \mu$m for different PISN models as a function of the observed time from the explosion. The data points with errors are the NIRCam (circle, 1st epoch) and NIRSpec (star, 2nd epoch) observations Gandolfi25.
• Figure 2: Spectral energy distributions of the five fiducial models fitting the light curve constraints as a function of the observed wavelength at the 1st observation epoch (Dec. 2022). The curves are color-coded as shown in the label. Also shown are the two measured NIRCam data points (black circles), and upper limits (gray triangles) Gandolfi25. For illustration, we show the transmissivity (cyan dashed line) due to the Ly$\alpha$ damping wing at the maximum redshift of the models ($z=15.32$); its effects on the SEDs are completely negligible.
• Figure 3: Top panel: Comparison between Capotauro spectrum (gray line; the black line shows the rebinned spectrum) and the He130 PISN model (orange) shown at its best-redshift ($z=15.32$), and 115 day after the explosion, i.e. at the 2nd epoch. The orange squares are the predicted photometric flux densities in the F444W (observed value shown as black star), and F777W filters. Also shown is the CEERS MIRI F777W 3$\sigma$ upper limit Gandolfi25. The spectrum of a prototypical cool Y0 brown dwarf (WISE 0359-54, Beiler23Beiler24), scaled to match the F444W photometric point, is shown as a purple line. Bottom: Fractional binned residuals of the PISN (orange filled points) and brown dwarf spectra (purple filled) after subtraction of the JWST spectrum. Open red points/light red bands indicate a poor fit, i.e. bins where the mean is $>3\sigma$.
• Figure 4: UV luminosity function at $z\simeq 14$ showing the putative location of Capotauro interpreted as a PISN at $z\simeq 15$. Data points are from Finkelstein24McLeod24Robertson24Donnan24Casey_2024Whitler25Asada25Naidu25; the lines show the fit provided by Harikane25 adopting a double-power law (solid) or a Schechter (dashed) function. The error on the PISN data point assumes a factor 2 uncertainty in the comoving volume.