First Insights into the ISM at $z>8$ with JWST: Possible Physical Implications of a High [O III]$\mathbf{λ4363}$/[O III]$\mathbf{λ5007}$

TL;DR

This study uses JWST/NIRSpec rest-frame optical emission lines from three z>7.5 galaxies to probe ISM conditions and heating mechanisms in the early Universe. By comparing line ratios to local analogues and applying photoionization modelling, the authors find that two galaxies align with Green Peas/Blueberries while a third, S04590, exhibits an extreme RO3 that standard stellar photoionization and shocks cannot fully explain. They show that non-thermal heating from high-mass X-ray binaries or cosmic rays can plausibly reproduce the observed ratios, suggesting these processes may influence ISM heating and potentially LyC leakage during the Epoch of Reionization. The work highlights the need for larger JWST samples to determine how common such heating channels are in the early galaxies and to refine models of ISM physics at the highest redshifts.

Abstract

We present a detailed analysis of the rest-frame optical emission line ratios for three spectroscopically confirmed galaxies at $z>7.5$. The galaxies were identified in the James Webb Space Telescope (JWST) Early Release Observations field SMACS J0723.3$-$7327. By quantitatively comparing Balmer and oxygen line ratios of these galaxies with various low-redshift "analogue" populations (e.g. Green Peas, Blueberries, etc.), we show that no single analogue population captures the diversity of line ratios of all three galaxies observed at $z>7.5$. We find that S06355 at $z=7.67$ and S10612 at $z=7.66$ are similar to local Green Peas and Blueberries. In contrast, S04590 at $z=8.50$ appears to be significantly different from the other two galaxies, most resembling extremely low-metallicity systems in the local Universe. Perhaps the most striking spectral feature in S04590 is the curiously high [O III] $\lambda4363$/[O III] $\lambda5007$ ratio (RO3) of $0.048$ (or $0.055$ when dust-corrected), implying either extremely high electron temperatures, $\sim3\times10^4$ K, or gas densities $>10^4\ {\rm cm^{-3}}$. Observed line ratios indicate that this galaxy is unlikely to host an AGN. Using photoionization modelling, we show that the inclusion of high-mass X-ray binaries or a high cosmic ray background in addition to a young, low-metallicity stellar population can provide the additional heating necessary to explain the observed high RO3 while remaining consistent with other observed line ratios. Our models represent a first step at accurately characterising the dominant sources of photoionization and heating at very high redshifts, demonstrating that non-thermal processes may become important as we probe deeper into the Epoch of Reionization.

Figures (6)

• Figure 1: Observed frame NIRSpec spectra of S04590 at $z=8.498$ (top), S06355 at $z=7.666$ (middle) and S10612 at $z=7.663$ (bottom). The spectra have been normalised such that the maximum flux value is at 1. The locations of emission lines have been marked for each source and the best-fitting Gaussians (red) are shown above the full 1D spectra for each object with the same axis labels as the 1D spectra, where we also show the base of the Gaussian fit (blue) inferred from the local continuum around the line, broadly consistent with the global noise in the 1D spectra (gray). The missing data is due to the gap between the two NIRSpec detector arrays.
• Figure 2: ( Top Left) RO3 versus O32. The three high redshift galaxies are shown as coloured stars with error bars representing the 1$\sigma$ uncertainty on the flux ratios. Translucent data points have been corrected for dust. For comparison, we show 2D probability distribution functions (PDFs) of SDSS galaxies (blue) and SDSS QSOs (orange) for objects that have at least an 8$\sigma$ detection of the [O III] $\lambda 4363$ line. Low-redshift analogues of high-redshift galaxies including green peas Yang2017a, blueberries Yang2017b, extremely low metallicity galaxies Izotov2019, and extreme emission line galaxies Amorin2015 are shown at green triangles, blue crosses, red hexagons, and grey hexagons, respectively. Orange squares are a compilation of Seyfert 2 nuclei Armah2021. ( Top Right) RO3 versus [O III] $\lambda 5007$/H$\beta$. ( Bottom Left) RO3 versus [Ne III] $\lambda 3869$/[O II] $\lambda\lambda 3726, 3729$. ( Bottom Right) [O II] $\lambda\lambda 3726, 3729$/H$\delta$ versus [Ne III] $\lambda 3869$/H$\delta$. The thick black line is the AGN/star-forming galaxy demarcation as described in PZ2007. Unless otherwise specified, the symbols in all panels are the same.
• Figure 3: Contours of constant RO3 as a function of electron density and temperature. The colour of the line represents RO3 as given in the colour bar. The two thick magenta lines represent the observed (lower) and dust-corrected (upper) contours that best match S04590.
• Figure 4: Various strong line diagnostics (Left: RO3 versus O32, Centre: RO3 versus [Ne III] $\lambda 3869$/[O II] $\lambda 3726, 3729$, Right: RO3 versus [O III] $\lambda 5007$/H$\beta$) for the high redshift galaxies compared to CLOUDY models. The top row shows models that include only photoionization from stars, the middle row includes the impact of HMXBs, and the bottom row includes progressively stronger cosmic ray background intensities. The three high redshift galaxies are shown as coloured stars with error bars representing the 1$\sigma$ uncertainty on the flux ratios. Solid and translucent points represent intrinsic and dust-correct values, respectively for each galaxy. For reference, we show the location of SDSS galaxies and SDSS QSOs as the blue and orange background histograms. For all CLOUDY models, the colour of the line represents metallicity as given in the colour bar. Line style represents ionization parameter (solid, dashed, dotted, and dot-dashed correspond to $U$ values of $10^{-3.2}$, $10^{-2.6}$, $10^{-2.1}$, and $10^{-1.5}$, respectively, except for the cosmic ray panel where at a metallicity of $10^{-1}Z_{\odot}$ (blue), we have added an extra set of models with $U=10^{-0.5}$ shown as the solid line). Line thickness either represents stellar age (top), $L_X/{\rm SFR}$ (middle), or cosmic ray background intensity (bottom). In all panels, only a subset of CLOUDY models are shown for visual clarity.
• Figure 5: Various strong line diagnostics (Left: RO3 versus O32, Centre: RO3 versus [Ne III] $\lambda 3869$/[O II] $\lambda 3726, 3729$, Right: RO3 versus [O III] $\lambda 5007$/H$\beta$) for the high redshift galaxies compared to CLOUDY models assuming a BPASSv2.0 SED a single burst with an age of 5 Myr. Each row shows runs with different gas densities while the different colours represent variations in ionization parameter or metallicity, as shown in the colour bars.