An emerging baryon cycle in a galaxy 500 million years after the Big Bang

This paper presents a detailed study of Gz9p3, a merging galaxy at redshift z = 9.311, observed only 500 million years after the Big Bang. The authors use ultra-deep JWST spectroscopy to investigate the transition from a feedback-free starburst (FFB) phase to a feedback-regulated baryon cycle. The study reveals that Gz9p3 is caught in this transition, showing evidence of intense recent star formation followed by the emergence of efficient stellar feedback.

The analysis of the rest-frame UV spectrum reveals a multiphase galactic outflow, with low-ionization state (LIS) and high-ionization state (HIS) absorption lines coherently blueshifted by 150–200 km s−1. The covering fractions are measured to be Cf,LIS > 0.6 and Cf,HIS > 0.8, which, combined with the similar kinematics, points to a multiphase outflow structure. A key result is the first direct measurement of the electron density of the cool outflowing gas at high redshift, derived from the relative populations of ground and fine-structure levels of C II and Si II. The joint modeling yields log(ne/cm−3) = 1.22+0.44−0.71.

This direct density measurement allows the authors to calculate a mass-loading factor (η = Ṁout/SFR) of η = 12.3+21.7−9.9, which is among the highest yet measured for galaxies of comparable stellar mass and lies above predictions from simulations like TNG and FIRE. The paper notes that conventional emission-line prescriptions, which rely on assumed densities, would severely underestimate this value (e.g., η[O III] = 0.13+0.08−0.06 with an assumed ne = 380 cm−3). The extreme mass loading is explained by the galaxy's rapidly evolving star-formation history, which shows a short, intense burst approximately 10–20 Myr before observation. The outflow is normalized by the lower current SFR, producing an enhanced apparent η. This temporal sequence is consistent with the delayed onset of feedback expected in FFB models, and the inferred star-formation efficiency (ε ≃ 0.17–0.28) aligns with FFB predictions.

The paper also investigates whether this feedback permanently removes baryons or redistributes them. The observed outflow velocity (160 km s−1) is below the estimated escape velocity of the host halo (vesc ≃ 350–650 km s−1 at 2 kpc), suggesting the cool outflowing gas is unlikely to escape. Spatially resolved damped Lyα absorption (DLA) fitting across four sightlines reveals a substantial neutral-gas reservoir (log(NHI/cm−2) ≃ 21.6–22.1) that is spatially coincident with the galaxy's extended stellar tail and in-situ star formation, indicating merger-driven gas redistribution. The authors conclude that Gz9p3 captures an early stage in the establishment of a self-regulated baryon cycle, where merger-driven gas redistribution and stellar feedback work together, with the expelled material remaining available for future recycling through the circumgalactic medium.

Improvements to AI Systems:

1. Multiphase Outflow Modeling with Density Inference

• Implement a Bayesian joint-modeling framework that directly infers electron density from fine-structure line ratios (C II, Si II) without assuming fixed densities, enabling accurate mass-loading factor (η) calculations.

• The improved system can predict outflow properties (velocity, covering fraction, density) from UV spectra of high-redshift galaxies, avoiding the severe underestimation seen with emission-line prescriptions (e.g., η[O III] vs. η from direct density).

2. Temporal Star-Formation History Deconvolution

• Integrate a time-resolved SFR reconstruction module that separates recent bursts (10–20 Myr) from current SFR, allowing the AI to normalize outflow rates correctly and distinguish intrinsic feedback efficiency from apparent enhancement due to declining SFR.

• The system can then forecast whether a galaxy is in a feedback-free starburst (FFB) phase or transitioning to feedback-regulated cycling, based on the ratio of burst age to dynamical timescale.

3. Gravitational Binding and Baryon Retention Prediction

• Add a halo escape-velocity estimator (using halo mass and radius) to compare with observed outflow velocities, enabling the AI to classify outflows as recyclable (bound) vs. escaping (unbound).

• The improved system can predict whether expelled gas will re-accrete via the circumgalactic medium, and quantify the net baryon retention fraction over cosmic time.

4. Spatially Resolved DLA Mapping and Merger-Driven Redistribution

• Develop a multi-sightline neutral-hydrogen column density (NHI) fitting algorithm that maps gas reservoirs across galaxy merger features (e.g., stellar tails), linking gas distribution to merger geometry.

• The AI can then simulate merger-driven gas redistribution and its coupling to star formation, providing predictions for future JWST observations of merging galaxies at z > 9.

5. Simulation-Calibrated Feedback Efficiency Tuning

• Use the measured η = 12.3 and star-formation efficiency ε = 0.17–0.28 to recalibrate sub-grid feedback parameters in cosmological simulations (e.g., TNG, FIRE).

• The improved system can generate more accurate synthetic spectra and outflow statistics for early galaxies, reducing discrepancies between observed and simulated mass-loading factors.

What the Improved AI System Can Do:

• Automatically analyze JWST/NIRSpec spectra of z > 8 galaxies to produce joint constraints on outflow density, mass loading, and baryon recycling potential.

• Distinguish between genuine feedback-driven quenching and temporary SFR dips, preventing misclassification of high-z galaxies.

• Predict the fate of outflowing gas (escape vs. recycle) and the resulting impact on subsequent star formation episodes.

• Provide real-time recalibration of galaxy formation models using direct observational inputs from the epoch of reionization.