The density-bounded twilight of starbursts in the early Universe

TL;DR

This work shows that a substantial fraction of high-redshift galaxies exhibit Balmer line ratios inconsistent with Case B, beyond what dust can explain. By constructing density-bounded nebulae with CLOUDY using BPASS stellar SEDs across a range of metallicities, ionisation parameters, and densities, the authors reproduce the observed $H\alpha/H\beta$ and $H\gamma/H\beta$ trends, including a metallicity-driven turnover caused by Ly$\gamma$ absorption around higher metallicities. The results link ABEs to density-bounded phases during a transient fast breathing mode of star formation, typically lasting about $\sim$20 Myr, suggesting that leakage of ionising photons and altered Ly$\alpha$ escape are common in early galaxies. These findings have broad implications for ISM feedback, nebular diagnostics, and the interpretation of reionisation-era galaxy properties, including biases in $\xi_{\mathrm{ion}}$ and dust-correction methods when Case B is violated.

Abstract

The peculiar nebular emission displayed by galaxies in the early Universe presents a unique opportunity to gain insight into the regulation of star formation in extreme environments. We investigate 500 (109) galaxies with deep NIRSpec/PRISM observations from the JADES survey at $z>2$ ($z>5.3$), finding 52 (26) galaxies with Balmer line ratios more than $1σ$ inconsistent with Case B recombination. These anomalous Balmer emitters (ABEs) cannot be explained by dust attenuation, indicating a departure from Case B recombination. To address this discrepancy, we model density-bounded nebulae with the photoionisation code CLOUDY. Density-bounded nebulae show anomalous Balmer line ratios due to Lyman line pumping and a transition from the nebulae being optically thin to optically thick for Lyman lines with increasing cloud depth. The H$α$/H$β$ versus H$γ$/H$β$ trend of density-bounded models is robust to changes in stellar age of the ionising source, gas density, and ionisation parameter; however, increasing the stellar metallicity drives a turnover in the trend. This is due to stronger stellar absorption features around Ly$γ$ reducing H$β$ fluorescence, allowing density-bounded models to account for all observed Balmer line ratios. ABEs show higher [OIII]/[OII], have steeper ultra-violet slopes, are fainter, and are more preferentially Ly$α$ emitters than galaxies which are consistent with Case B and little dust. These findings suggest that ABEs are galaxies that have become density bounded during extreme quenching events, representing a transient phase of $\sim$20 Myr during a fast breathing mode of star formation.

Figures (11)

• Figure 1: Balmer line ratios for $z>2$ galaxies compared to Case B values in the 10 000 -- 20 000 K temperature range (blue shaded region). Top: H$\alpha$/H$\beta$ for galaxies with 5$\sigma$ detections of H$\alpha$ and H$\beta$, ordered by increasing line ratio. The 100 galaxies with the lowest H$\alpha$/H$\beta$ ratios are shown. 52 (14) galaxies with $z>2$ ($z>5.3$) show Balmer decrements which are more than $1\sigma$ below the Case B values, which cannot be explained by dust attenuation. Bottom: H$\gamma$/H$\beta$ values for galaxies which also have a 5$\sigma$ detection of H$\gamma$, ordered by decreasing line ratio. The 20 galaxies with the highest H$\gamma$/H$\beta$ ratios are shown. 12 galaxies with $z>5.3$ show values more than $1\sigma$ above the Case B range, which cannot be due to dust attenuation.
• Figure 2: H$\alpha$/H$\beta$ against H$\gamma$/H$\beta$ for galaxies with 5$\sigma$ detections of H$\alpha$, H$\beta$, and H$\gamma$. The arrow represents increasing dust attenuation for an SMC extinction curve ($\mathrm{R}_{\rm _V}=2.7$) with a length of $\mathrm{A}_{\rm _V}=1$ mag Gordon:2003aa. A Cardelli:1989aa dust curve with $\mathrm{R}_{\rm _V}=3.1$ (not shown) has a nearly identical direction. The blue box around H$\alpha$/H$\beta=2.8$ and H$\gamma$/H$\beta=0.47$ represents Case B values in the 10 000 -- 20 000 K temperature range. Many galaxies are not consistent with Case B line ratios and dust attenuation; galaxies which are at least 1$\sigma$ inconsistent are marked with red points. Following Yanagisawa:2024ac, the solid (dashed) red line represents Balmer absorption due to a shell of HI gas in the $n=2$ state, combined with dust $\mathrm{A}_{\rm _V}=0$ mag ($\mathrm{A}_{\rm _V}=2$ mag). The log column density of the excited neutral gas is labelled on the lines. This model of an excited hydrogen shell cannot explain galaxies with elevated H$\gamma$/H$\beta$.
• Figure 3: H$\alpha$/H$\beta$ against H$\gamma$/H$\beta$ for all density-bounded nebulae models in our parameter grid (Table \ref{['tab:cloudy_parameters']}). Each line represents a density-bounded model with increasing cloud depth until it becomes ionisation-bounded and converges to the Case B line ratios (blue box shaded region around H$\alpha$/H$\beta=2.8$ and H$\gamma$/H$\beta=0.47$). Our models bracket the bottom right of the observed distribution, and therefore can account for all observed values when combined with dust attenuation (arrow shows Gordon:2003aa SMC extinction curve with a length of $\mathrm{A}_{\rm _V}=1$ mag). The model-to-model variation is primarily driven by metallicity, and the trends are otherwise robust to changes in gas density, ionisation parameter, and stellar age, although the older 10 Myr stars tend to reach more extreme line ratios while still following the same trends set by the metallicity.
• Figure 4: The H$\alpha$/H$\beta$ and H$\gamma$/H$\beta$ ratios as a function of cloud depth for a set of density-bounded nebulae models. These represent gas with density $\mathrm{n_H}=10~\mathrm{cm^{-3}}$ being ionised by a 3 Myr binary instant starburst with $\log(\mathrm{U})=-1$. The $\log\mathrm{Z/\text{Z}_{\astrosun}}$ is varied from $-3$ to 0.2. All models show a varying H$\alpha$/H$\beta$ and H$\gamma$/H$\beta$ with increasing cloud depth, which is caused by the transition from the gas being optically thin to optically thick for Lyman lines. For low metallicity models, both H$\alpha$/H$\beta$ and H$\gamma$/H$\beta$ initially have low values, before converging to Case B as they become ionisation-bounded nebulae. However, high metallicity models begin with high H$\alpha$/H$\beta$ and H$\gamma$/H$\beta$, and converge down to the Case B values as they become ionisation bounded. This is driven by the change Lyman line absorption features with stellar metallicity, in particular the increasing absorption features around Ly$\gamma$ with increasing metallicity (Fig. \ref{['fig:bpass_plot']}).
• Figure 5: A comparison of the SEDs of 3 Myr SSP with metallicities of $\mathrm{Z}=0.001$ (red) and $\mathrm{Z}=0.014$ (blue) from the BPASS library. The Lyman line absorption features vary with metallicity. Notably, the higher metallicity stars have much stronger absorption around Ly$\gamma$, and therefore fewer continuum photons are available to pump electrons to the $\mathrm{n}=4$ state of hydrogen and cause H$\beta$ emission via subsequent radiative decay. The variation in absorption features is responsible for the turnover in the H$\alpha$/H$\beta$ versus H$\gamma$/H$\beta$ relation shown in Fig. \ref{['fig:DB_Ha_Hg_vs_Ha_Hb_allmod']}. These absorption features are not important in ionisation-bounded nebulae because the continuum photons which can pump Lyman lines are used up at smaller cloud depths than the ionising photons, meaning that most emission is from non-pumped gas.