Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Bursty or heavy? The surprise of bright Population III systems in the Reionization era

Alessandra Venditti, Julian B. Munoz, Volker Bromm, Seiji Fujimoto, Steven L. Finkelstein, John Chisholm

TL;DR

This study investigates the unexpected presence of bright Population III (Pop III) systems during the Epoch of Reionization as seen by JWST at z≈5.6–6.6. It develops a semi-analytic Pop III UV luminosity function framework, linking halo mass to star formation and UV output via a Gaussian conditional distribution and a flexible star-formation efficiency, duty cycle, and UV-boost parameters, then tests two viable late-Pop III scenarios: (i) Heavy, with formation in atomic-cooling halos up to ${M_ ext{up}^ ext{III}} \napprox 10^{10}~M_\odot$, and (ii) Bursty, with highly stochastic bursts in mini-halos, characterized by ${\sigma_ ext{UV}^ ext{III}} rightarrow 1.5$–2.0. The analysis finds that fitting the observed bright candidates requires either a high halo-mass cutoff or substantial burstiness, with both pathways implying a larger, hidden Pop III population and potential implications for reionization and high-energy backgrounds. Extrapolations to z≈8–12.5 suggest a persistent Pop III presence at faint magnitudes, and forthcoming surveys like JWST’s VENUS could uncover dozens of additional systems, enabling tighter discrimination between the Heavy and Bursty scenarios. Overall, the paper provides a robust framework to interpret JWST-era Pop III constraints and forecasts for ultra-deep surveys.

Abstract

The nature of the first, so-called Population III (Pop III) stars has for long remained largely unconstrained. However, the James Webb Space Telescope (JWST) finally opened new concrete prospects for their detection during the Epoch of Reionization (EoR), notably providing promising observational constraints on the Pop III ultra-violet luminosity function (UVLF) at $z \approx 5.6 - 6.6$. These preliminary data hint towards an unexpected population of UV-bright Pop III sources, which challenges the prevailing view that Pop III star formation is confined to molecular-cooling mini-halos. Here we show that there are two families of models that can explain these surprising observations, either by allowing for late-time Pop III formation within massive, atomic-cooling halos (with halo masses up to $M^\mathrm{III}_\mathrm{up} \gtrsim 10^{10} \, \mathrm{M_\odot}$) or by invoking a highly bursty Pop III star-formation activity (with a stochasticity parameter $σ^\mathrm{III}_\mathrm{UV} \gtrsim 1.5$). In these scenarios, Pop III systems would have to be either heavier or burstier than usually assumed, underscoring the need to reconsider common assumptions about Pop III star-formation sites, and the potential implications of JWST candidates for current and future observations.

Bursty or heavy? The surprise of bright Population III systems in the Reionization era

TL;DR

This study investigates the unexpected presence of bright Population III (Pop III) systems during the Epoch of Reionization as seen by JWST at z≈5.6–6.6. It develops a semi-analytic Pop III UV luminosity function framework, linking halo mass to star formation and UV output via a Gaussian conditional distribution and a flexible star-formation efficiency, duty cycle, and UV-boost parameters, then tests two viable late-Pop III scenarios: (i) Heavy, with formation in atomic-cooling halos up to , and (ii) Bursty, with highly stochastic bursts in mini-halos, characterized by –2.0. The analysis finds that fitting the observed bright candidates requires either a high halo-mass cutoff or substantial burstiness, with both pathways implying a larger, hidden Pop III population and potential implications for reionization and high-energy backgrounds. Extrapolations to z≈8–12.5 suggest a persistent Pop III presence at faint magnitudes, and forthcoming surveys like JWST’s VENUS could uncover dozens of additional systems, enabling tighter discrimination between the Heavy and Bursty scenarios. Overall, the paper provides a robust framework to interpret JWST-era Pop III constraints and forecasts for ultra-deep surveys.

Abstract

The nature of the first, so-called Population III (Pop III) stars has for long remained largely unconstrained. However, the James Webb Space Telescope (JWST) finally opened new concrete prospects for their detection during the Epoch of Reionization (EoR), notably providing promising observational constraints on the Pop III ultra-violet luminosity function (UVLF) at . These preliminary data hint towards an unexpected population of UV-bright Pop III sources, which challenges the prevailing view that Pop III star formation is confined to molecular-cooling mini-halos. Here we show that there are two families of models that can explain these surprising observations, either by allowing for late-time Pop III formation within massive, atomic-cooling halos (with halo masses up to ) or by invoking a highly bursty Pop III star-formation activity (with a stochasticity parameter ). In these scenarios, Pop III systems would have to be either heavier or burstier than usually assumed, underscoring the need to reconsider common assumptions about Pop III star-formation sites, and the potential implications of JWST candidates for current and future observations.

Paper Structure

This paper contains 10 sections, 6 equations, 7 figures.

Table of Contents

  1. Introduction
  2. Modeling the Pop III UVLF
  3. The redshift $\approx 5.6 - 6.6$ Pop III UVLF
  4. Pop III star formation beyond mini-halos
  5. Role of stochasticity/burstiness
  6. A Pop III sea in the ultra-deep Universe?
  7. Summary and Conclusions
  8. Pop III SFE/UVLF model and simulations
  9. Alternative datasets
  10. Further modeling

Figures (7)

  • Figure 1: Left: SFR per unit halo mass ($\dot{M}_\star / {M_\mathrm{h}}$) as a function of halo mass (${M_\mathrm{h}}$) at $z = 6.1$ (with $\Delta z = 1)$, incorporating a UV enhancement of $\varepsilon_\mathrm{\star,UV} / \varepsilon_\star$. Our reference "Bursty"/"Heavy" models (with ${M_\mathrm{up}^\mathrm{III}} \sim 10^8/10^{10} ~{M_\odot}$ respectively, and ${\varepsilon_\mathrm{\star,UV}^\mathrm{III}} \sim 10^{-2.4}$) are shown as red/blue, thick, dashed lines. The model of Cruz_2024, and the corresponding Pop II model from Munoz_2023, are included as a reference. The blue crosses show the average $\dot{M}_\star / {M_\mathrm{h}}$ at the two extremes of the considered redshift range ($z \approx 5.6/6.6$, in lighter/darker shade respectively) for the dustyGadget simulations DiCesare_2023Venditti_2023 in bins of ${M_\mathrm{h}}$, assuming a rescaling of ${\eta_\mathrm{III}} = 0.3$ for the Pop III mass (see Appendix \ref{['app:model_details']} for more details). Right: UVLFs ($\Phi_\mathrm{UV}$) as a function of UV magnitude ($M_\mathrm{UV}$) resulting from the models in the left panel (with same color and linestyle, assuming ${\sigma_\mathrm{UV}^\mathrm{III}} = 1.5/0.7$ for the "Bursty"/"Heavy" models, respectively), compared with the data point and upper limits from Fujimoto_2025, including the AMORE6 candidate from Morishita_2025 in the faintest $M_\mathrm{UV}$ bin; filled/empty circles distinguish between $M_\mathrm{UV}$ bins containing spectroscopic/photometric candidates (see text and Appendix \ref{['app:alternative_datasets']} for more details). The grey, shaded area encompasses a range of UVLF fitting functions (from Bouwens_2021Finkelstein_Bagley_2022Munoz_2023) extrapolated at the faint end, included as a reference together with data points from Bouwens_2021 and Bouwens_2022 at $z = 5/6/7$ (green/golden/brown dots). While the model of Cruz_2024 assuming ${\sigma_\mathrm{UV}^\mathrm{III}} = 0.7$ (consistent with the value inferred from galaxy populations at the same redshift, as demonstrated by the agreement with observed data points of the Pop II black line when assuming ${\sigma_\mathrm{UV}^\mathrm{III}} \approx 0.7$) is unable to explain the data, models with enhanced stochasticity ("Bursty" model) or a larger high-mass cut-off ("Heavy" model) would be marginally consistent with GLIMPSE-16043 and AMORE6.
  • Figure 2: Joint posterior for the effective SFE normalization ${\varepsilon_\mathrm{\star,UV}^\mathrm{III}}$ (Equation \ref{['eq:epsilonstar_UV']}) and the high-mass cut-off ${M_\mathrm{up}^\mathrm{III}}$ (Equation \ref{['eq:fduty_III']}), resulting by fitting our Pop III UVLF model described in Section \ref{['sec:methodology']} against the observed data points from Fujimoto_2025 and Morishita_2025 at $z \approx 5.6 - 6.6$, with fixed ${\beta_\star^\mathrm{III}} = 0$ and ${\sigma_\mathrm{UV}^\mathrm{III}} = 1.5/0.7$ (red/blue curves). The dashed, thick lines show the values adopted for ${\varepsilon_\mathrm{\star,UV}^\mathrm{III}}$ and ${M_\mathrm{up}^\mathrm{III}}$ in our reference "Bursty" and "Heavy" models (refer to Figure \ref{['fig:PopIII_UVLF_ref']}): for each chosen value of ${\sigma_\mathrm{UV}^\mathrm{III}}$, these represent the best-fit values of Log${\varepsilon_\mathrm{\star,UV}^\mathrm{III}}$ corresponding roughly to the 2.3rd percentile of the Log$({M_\mathrm{up}^\mathrm{III}} / {M_\odot})$ marginalized posterior. The dotted, dark-red and solid, dark-blue lines further show the values of ${\varepsilon_\mathrm{\star,UV}^\mathrm{III}}$ and ${M_\mathrm{up}^\mathrm{III}}$ from the model of Cruz_2024 (as in Figure \ref{['fig:PopIII_UVLF_ref']}), and the maximum halo mass hosting Pop III stars at $z \approx 5.6 - 6.6$ from the dustyGadget simulation suite DiCesare_2023Venditti_2023 as a reference. Note that either large stochasticities (${\sigma_\mathrm{UV}^\mathrm{III}} \gtrsim 1.5$) or large high-mass cut-offs (${M_\mathrm{up}^\mathrm{III}} \gtrsim 10^{10} ~{M_\odot}$, still well below the most massive Pop III halo in the dustyGadget simulations) are required to fit the data within $2\sigma$.
  • Figure 3: UVLFs resulting from the "Bursty" and "Heavy" models from Fig. \ref{['fig:PopIII_UVLF_ref']} (red/blue, dashed lines), if extrapolated to $z = 8$ and $z = 12.5$ (with progressively thinner linestyle). For comparison, we show the observed binned UVLF and fitting functions from Bouwens_2021 and Donnan_2024, extrapolated towards fainter magnitudes. The bright end of the Pop III UVLF shows a modest increase up to $z \approx 12.5$, while a strong decay of the total (Pop II-dominated) UVLF is evident from observations. Lensing JWST surveys such as VENUS Fujimoto_2025_VENUS will allow deeper $M_\mathrm{UV}$ coverage (within the pink, shaded area, computed for $z = 10$ with $\Delta z = 1$).
  • Figure 4: Same as Figures \ref{['fig:PopIII_UVLF_ref']} and \ref{['fig:PopIII_UVLF_corner_ref']}, but also including the uncertain JOF-21739 Pop III candidate at bright $M_\mathrm{UV}^\mathrm{III} \approx -17.6$ in the fit. The "Bursty"/"Heavy" models in this case correspond to larger ${M_\mathrm{up}^\mathrm{III}} \sim 10^9 / 10^{12} ~{M_\odot}$, and ${\varepsilon_\mathrm{\star,UV}^\mathrm{III}} \sim 10^{-2.8} / 10^{-3.4}$ (with the usual reference ${\sigma_\mathrm{UV}^\mathrm{III}} = 1.5/0.7$). An additional ${\sigma_\mathrm{UV}^\mathrm{III}} = 2$ scenario is included in pink, which approximately corresponds to the minimum stochasticity allowing to fit the data within $2\sigma$ with a cut-off close to the atomic-cooling halo mass threshold; see e.g. the "Burstier" model (${M_\mathrm{up}^\mathrm{III}} \sim 10^8 ~{M_\odot}$, ${\varepsilon_\mathrm{\star,UV}^\mathrm{III}} \sim 10^{-3.1}$).
  • Figure 5: Same as Figures \ref{['fig:PopIII_UVLF_ref']} and \ref{['fig:PopIII_UVLF_corner_ref']}, but only including the AMORE6 Pop III candidate from Morishita_2025 and the $z = 6$ total UVLF constraints from Bouwens_2022 in the fit (as detailed in the text); shaded lines refer to the results of corresponding fits without the inclusion of the total UVLF constraints. While the latter allow ${M_\mathrm{up}^\mathrm{III}} \sim 10^8 ~{M_\odot}$ with ${\varepsilon_\mathrm{\star,UV}^\mathrm{III}} \sim 10^{-2.4}$ (for the "Bursty" model) and a very large ${\varepsilon_\mathrm{\star,UV}^\mathrm{III}} \sim 10^{-1.2}$ (for the "Heavy" model), values of ${M_\mathrm{up}^\mathrm{III}} \sim 10^8 / 10^{9.5} ~{M_\odot}$ and ${\varepsilon_\mathrm{\star,UV}^\mathrm{III}} \sim 10^{-2.5} / 10^{-2.1}$, closer to the case including GLIMPSE-16043 (Figures \ref{['fig:PopIII_UVLF_ref']} and \ref{['fig:PopIII_UVLF_corner_ref']}) are found for the "Bursty"/"Heavy" models when the total UVLF constraints are included.
  • ...and 2 more figures