Cloudy with a chance of starshine: Possible photometric signatures of nebular-dominated emission in $1.5 < z < 8.5$ JADES galaxies

TL;DR

This study develops a photometric framework to identify nebular-dominated galaxies at $1.5 < z < 8.5$ by exploiting Balmer jumps and elevated ionising photon production efficiencies $ξ_ ext{ion, obs}$. Leveraging JWST/JADES medium-band imaging in GOODS-S/N, the authors define Balmer-jump candidates via rest-frame optical deficits and further filter for nebular dominance using $ξ_ ext{ion, obs} > 25.60$ and UV colour criteria, with high-redshift adjustments using $ ext{EW}([OIII]+Hβ)$ when Hα is unavailable. They report 2684 Balmer-jump candidates and 972 nebular-dominated candidates, finding that nebular-dominated systems often exhibit extreme line EWs and strong line luminosities, yet do not always show the largest Balmer jumps due to hotter HII regions and collisional effects. The analysis reveals redshift evolution in the incidence of Balmer jumps and nebular-dominated emission, suggesting the early Universe hosts a growing population of young, line-rich starbursts that may be powered by top-heavy star formation or AGN activity. However, definitive confirmation of the defining two-photon downturn requires continuum spectroscopy, underscoring the need for follow-up NIRSpec observations and nebular-dominated templates to robustly identify these signatures of primordial star formation.

Abstract

The discovery of high-redshift galaxies exhibiting a steep spectral UV downturn potentially indicative of two-photon continuum emission marks a turning point in our search for signatures of top-heavy star formation in the early Universe. We develop a photometric search method for identifying further nebular-dominated galaxy candidates, whose nebular continuum dominates over the starlight, due to the high ionising photon production efficiencies $ξ_\mathrm{ion}$ associated with massive star formation. We utilise the extensive medium-band imaging from JADES, which enables the identification of Balmer jumps across a wide range of redshifts ($1.5 < z < 8.5$), through the deficit in rest-frame optical continuum level. As Balmer jumps are a general recombination feature of young starbursts ($\lesssim 3$~Myr), we further demand a high observed $\log\, (ξ_\mathrm{ion, obs}/\mathrm{(Hz\ erg^{-1})}) > 25.60$ to power the strong nebular continuum, together with a relatively non-blue UV slope indicating a lack of stellar continuum emission. Our nebular-dominated candidates, constituting ${\sim}$10% of galaxies at $z \sim 6$ (decreasing to ${\sim}$3% at $z \sim 2$, not completeness-corrected) are faint in the rest-frame optical (median $M_\mathrm{opt} = -17.95$) with extreme line emission (median $\mathrm{EW}_\mathrm{Hα,rest} = 1567$ Å, $\mathrm{EW}_\mathrm{[O\ III] + Hβ,rest} = 2244$ Å). However, hot H II region temperatures, collisionally-enhanced two-photon continuum emission, and strong UV lines are expected to accompany top-heavy star formation. Thus nebular-dominated galaxies do not necessarily exhibit the biggest Balmer jumps, nor the largest $ξ_\mathrm{ion, obs}$ or reddest UV slopes. Hence continuum spectroscopy is ultimately required to establish the presence of a two-photon downturn in our candidates, thus advancing our understanding of primordial star formation and AGN.

Figures (17)

• Figure 1: The NIRSpec PRISM spectrum (dark green), HST+NIRCam photometry (black) and $2\times 2$ arcsec RGB (F444W, F200W, F115W) cutout for GS-9422, the nebular-dominated galaxy candidate reported by Cameron2024. Vertical dashed lines indicate the location of the Balmer limit at 3646 Å (green), H$\beta$ (yellow), [OIII] $\lambda 5007$ (orange) and H$\alpha$ (red). The SNR in 0.3 arcsec diameter circular apertures is denoted at the bottom for each filter. Non-detections are depicted as downward arrows located at the 3$\sigma$ upper limit. The spectrum of GS-9422 exhibits a prominent downturn at ${\sim}1.1$ µ m, possibly due to two-photon continuum emission (or perhaps damped Ly$\alpha$ absorption), suggesting nebular-dominated emission Cameron2024Katz2025. GS-9422 also exhibits a discontinuity (at 3646 Å rest-frame, ${\sim}2.5$µ m observed-frame) in its continuum levels between the rest-frame UV and optical: the Balmer jump, a general recombination feature of all young ($\lesssim 3$ Myr) starbursts. Medium-band photometry, narrow enough to fit between the prominent rest-frame optical emission lines ([OIII] and H$\alpha$), reveals the low-lying rest-frame optical continuum level (measured with F410M, red), being $\Delta m_\mathrm{jump} = -0.72$ mag fainter than the rest-frame UV (measured with F200W, dark blue). The photometric boost caused by the strong H$\alpha$ emission in F444W, together with the 1500 Å continuum measurement (via F115W), establish the high ionising photon production efficiency $\log\, (\xi_\mathrm{ion, obs} /\mathrm{(Hz\ erg^{-1})})= 25.70$ in this source, supporting the nebular-dominated scenario.
• Figure 2: Top panel: Various components of the nebular continuum emission, assuming pure hydrogen gas at $T=20000~\mathrm{K}$, generated using PyNeb Luridiana2015. Shown are the two-photon continuum (light green), free--bound emission (blue) which is responsible for the Balmer jump at 3646 Å and Paschen jump at 8206 Å, free--free emission (purple), and the total nebular continuum (ff + fb + $2\gamma$, red). All components are normalised by the peak value of the two-photon continuum at 1615 Å. Middle panel: The temperature dependence of the total nebular continuum, with ff + fb + $2\gamma$ at $T = 10000~\mathrm{K}$ (dashed), $T = 20000~\mathrm{K}$ (solid) and $T = 30000~\mathrm{K}$ (dot-dashed). The Balmer jump decreases in $\left | \Delta m\right |$ with increasing temperature, so top-heavy, metal-poor star formation (resulting in higher HII region temperatures) is not necessarily powering the largest Balmer jumps. Bottom panel: The total nebular continuum with varying amounts of collisional enhancement (caused by high HII region temperatures and non-negligible neutral HI fractions) of two-photon continuum emission. Shown are regular two-photon continuum emission ($2\gamma \times 1$, red), collisionally-enhanced two-photon emission with $2\gamma \times [2, 3, 4]$ (orange, green, blue, respectively) and collisionally-suppressed two-photon emission with $2\gamma \times 0.1$ due to high density $n_\mathrm{e} \sim 10^{5}~\mathrm{cm}^{-3}$ gas (dark red).
• Figure 3: Left panel: For normal starbursts Salpeter1955Kroupa2001Chabrier2003 with $\log\, (\xi_\mathrm{ion, *} /\mathrm{(Hz\ erg^{-1})}) \approx 25.85$ such as the 1 Myr old starburst shown generated with Bagpipes Carnall2018, the normalisation of the nebular continuum (green) is too low for the steep decline of the two-photon continuum emission to be seen over the rise of the stellar continuum (blue) in the total spectrum (red). Right panel: It is only with top-heavy star formation Zackrisson2011, where the ionising photon production efficiency is much higher (here $\log\, (\xi_\mathrm{ion, *} /\mathrm{(Hz\ erg^{-1})}) \approx 26.70$) , so that the normalisation of the nebular continuum becomes substantial relative to the ionising starlight that is powering it, that the two-photon turnover is very prominent in the total spectrum: the galaxy is nebular-dominated Cameron2024Katz2025.
• Figure 4: Left panel: The Balmer jump, as traced by the magnitude difference $m_{3640}-m_{3650}$ between the continuum levels immediately blueward and redward of the Balmer limit at 3646 Å, shown as a function of HII region temperature for different levels of collisionally-enhanced two-photon continuum emission (various colours). We assume only the nebular continuum, including the H and HeI contribution Luridiana2015, but excluding any stellar continuum component (which would reduce the magnitude shift, arrow). Right panel: Low-resolution PRISM spectroscopy and photometry demand longer wavelength baseline measurements $m_\mathrm{3000}-m_\mathrm{5750}$, to avoid blending of emission lines and continuum. Here the continuum level in the rest-frame optical $m_\mathrm{5750}$ is measured using a medium-band filter that lies between [OIII] $\lambda 5007$ and H$\alpha$. HeI $\lambda 5876$ emission (maximally ${\sim}100~$Å EW), can cause a photometric boost in the medium band, causing the Balmer jump to be underestimated by up to 0.16--0.32 mag (depending on filter spectral resolution $R$, arrows). Photometric (pentagons) and spectroscopic (squares) measurements of the Balmer jump in the spectroscopically-identified nebular-dominated galaxy candidates reported by Katz2025: GS-9422 (green), 2198_7807 ($z=5.347$, cyan), 1210_5217 ($z=4.888$, blue), 2561_17467 ($z=3.99$, light purple), 2756_301 ($z=3.99$, dark purple). Motivated by the smallest Balmer jumps seen in current observations of nebular-dominated galaxy candidates Cameron2024Katz2025, as well as Pop III models Zackrisson2011Nakajima2022, we set our minimal Balmer jump threshold (dashed horizontal line) as $\Delta m_\mathrm{jump} < -0.15$.
• Figure 5: SEDs of Balmer-jump galaxy candidates, selected as having UV--optical colours $\Delta m_\mathrm{jump} < -0.15$. Bagpipes fits to the full photometry are shown in light blue. As in Fig. \ref{['fig:GS-9422']}, the RGB (F444W, F200W, F115W) cutout spans $2\times 2$ arcsec. Left panels: Candidates where the UV--optical flux density gap is assessed to mostly be due to the Balmer jump (through visual inspection of the Bagpipes fit). Right panels: Candidates where the UV--optical flux density gap is due to a combination of a Balmer jump and blue slope (top and middle), and a blue slope (bottom).