Bursty star formation, chemical enrichment, and star cluster formation in numerical analogues of GN-z11

TL;DR

This study uses high-resolution cosmological zoom-in simulations with $\sim5$ pc spatial resolution to model a GN-z11–like galaxy at $z\sim10$, tracking 18 elements through stellar winds, CCSNe, Type Ia SNe, and AGB ejecta. It finds a rapid, centrally concentrated starburst in a $\sim3\times10^{10}\,M_\odot$ halo, where wind-driven pollution from fast-rotating massive stars raises log$(\mathrm{N/O})$ to about $-0.61$ for $\sim 10$ Myr before dilution from supernova ejecta, and a significant population of high-efficiency star clusters forms ($\epsilon_\mathrm{cluster}\sim20\%$). While the central N-enrichment and star-formation structure resemble GN-z11, the peak log$(\mathrm{N/O})$ remains below some observational lower limits, suggesting additional channels (e.g., supermassive stars) may be required. The results also reveal a dichotomy of star clusters (ex-situ and in-situ) with Na–O anti-correlations emerging in wind-polluted, metal-poor clusters, linking early galaxy assembly to globular-cluster–like chemistry. Overall, the work supports a picture where feedback-free, centrally concentrated starbursts in high-$z$ galaxies can drive rapid chemical enrichment and dense cluster formation, but it also highlights remaining gaps in matching the full GN-z11 abundance constraints and identifies avenues for refinement in yields and physics.

Abstract

The James Webb Space Telescope reveals anomalous nitrogen enrichment (high N/O ratios) in compact, star-forming galaxies, such as GN-z11 at $z\sim10$. The origin of this chemical signature provides an insight into the early star and galaxy formation processes, yet remains unclear. We performed high-resolution cosmological zoom-in simulations of massive galaxies at high redshift ($z\sim10$) in rare density peaks, incorporating various chemical evolution channels including stellar winds, core-collapse, Type Ia supernovae, and asymptotic giant branch stars. Our simulations reproduce several key features of high-redshift galaxies: (1) stars form with high efficiencies ($>0.1$) at the center of rare peak halos, creating very compact galaxies similar to GN-z11; (2) high N/O ratios emerge during the first 10-20 Myr of intense starburst, before being diluted by CCSNe; (3) multiple star clusters form in and around the galaxy with high efficiency ($\sim20\%$), some of which exhibit high N/O ratios and sodium-oxygen anti-correlations similar to those observed in local globular clusters. Although our simulations can reproduce the high log(N/O) values (up to -0.61, exceeding the solar value by 0.25 dex), they remain below the observational lower limits of GN-z11, indicating room for improvement through additional chemical evolution channels, such as supermassive stars.

Figures (22)

• Figure 1: Expected log(N/O) (top panel) and log(C/O) (bottom panel) as a function of metallicity, $\log({\rm O/H})+12$. The blue, red, gray, and green curves exhibit the IMF-weighted log(N/O) and log(C/O) of the ejecta for stellar wind (SW), CCSNe, AGBs (FRUITY yields), and AGBs (Monash yields), respectively. For AGBs, the ejecta components are diluted by the same mass of the gas component of the SSP particle. The orange curve represents the scaling values of log(N/O) and log(C/O) obtained from Nicholls+2017. The fiducial and conservative regions of GN-z11 from Cameron+2023 are shown with magenta solid and dotted lines. The magenta filled region is the observational values of GN-z11 Alvarez-Marquez+2025. The solar abundance values are shown with the solar symbols. Alt text: Two-panel line plot showing nitrogen-to-oxygen (top) and carbon-to-oxygen (bottom) ratios against oxygen abundance. Multiple curves represent theoretical yields from stellar winds, core-collapse supernovae, and two asymptotic giant branch star models. The curves demonstrate that stellar winds and asymptotic giant branch stars are possible origins of high nitrogen-to-oxygen ratios. Observational scaling relations and the measured values of GN-z11 are also shown for comparison.
• Figure 2: Distribution of star-forming halos as a function of time (redshift). Each dot represents the mass of the halo at the redshift. Color indicates the average metallicity of the stellar component in each halo. For visibility, we add a spread for time. Each group is sampled at the same redshift. Here we sample every $\sim 27.6~\mathrm{Myr}$. The blue curve exhibits the expected halo mass of the virial temperature at $10^4~\mathrm{K}$ obtained from BrommYoshida2011. The green curve is the evolution track of the primary halo we investigate in this study. The solid part has stars while the dashed part does not. Alt text: Scatter plot showing halo mass versus time and redshift. Dots represent individual halos with metallicity indicated for each halo. A reference curve shows expected halo mass for virial temperature of ten to the fourth Kelvin. An evolution track represents the primary halo evolution, with solid and dashed segments indicating presence or absence of stars.
• Figure 3: Left column: The projected gas surface density in the units of $M_{\odot}~\mathrm{pc}^{-2}$. Colors represent the surface density and the color bar is found in the top panels. Middle column: The projected stellar density in the units of $M_{\odot}~\mathrm{pc}^{-2}$. Right column: the surface SFR density in the units of $M_{\odot}~\mathrm{yr} ^{-1}~\mathrm{kpc}^{-2}$. Each panel covers the $20~\mathrm{kpc} \times 20~\mathrm{kpc}$ region. From top panels to bottom panels, we show the snapshots from $z=11.82$ to $z=9.8$. Alt text: Three-column panel showing projected densities across a twenty kiloparsec by twenty kiloparsec region. Panels arranged from top to bottom show snapshots at redshifts from eleven point eight two to nine point seven nine. Left column displays gas surface density in solar masses per square parsec. Middle column shows stellar density in solar masses per square parsec. Right column presents star formation rate surface density in solar masses per year per square kiloparsec. The central high-gas density region is the intersection of filamentary structures. Star-forming regions and stars are highly concentrated at the center.
• Figure 4: Same as figure \ref{['fig:20kpc']}, but for the $2~\mathrm{kpc} \times 2~\mathrm{kpc}$ regions. Alt text: Three-column panel showing projected densities across a two kiloparsec by two kiloparsec region. Panels arranged from top to bottom show snapshots at redshifts from eleven point eight two to nine point seven nine. Left column displays gas surface density in solar masses per square parsec. Middle column shows stellar density in solar masses per square parsec. Right column presents star formation rate surface density in solar masses per year per square kiloparsec. Gas accumulates at the center, forming a very compact and dense galaxy.
• Figure 5: Gas and stellar surface density maps at $z=10.46$. Left panels: Projected surface density maps of gas in units of $M_{\odot}~\mathrm{pc}^{-2}$. Right panels: Projected surface density maps of stars in units of $M_{\odot}~\mathrm{pc}^{-2}$. The orientations have been adjusted to present face-on and edge-on views in the top and bottom panels, respectively. Alt text: Two-column panel showing surface density maps across a four hundred parsec by four hundred parsec region. Left column shows gas surface density in solar masses per square parsec. Right column displays stellar surface density in solar masses per square parsec. The maps show formation of a compact, rotation-supported galaxy.