The AURORA Survey: Ionizing Photon Production Efficiency with Minimal Nebular Dust Attenuation Systematics

TL;DR

This work tackles the uncertainty in the ionizing photon production efficiency, $\xi_{\rm ion}$, by exploiting JWST/AURORA data to apply precise, object-specific nebular dust corrections for 63 star-forming galaxies across $z=1.5-6.9$. Through emission-line fluxes, SED fitting, and both Balmer/Paschen line diagnostics, the authors derive intrinsic H$\alpha$ luminosities and UV luminosities, enabling $\xi_{\rm ion,0}$ estimates with reduced dust-systematics; a subset of 23 galaxies also provides individualized nebular attenuation curves. They report significant positive correlations of $\xi_{\rm ion,0}$ with redshift, O32, and $W_{\lambda}([\mathrm{O\,III}])$, and negative correlations with $A_V$ and direct-method metallicity, while addressing how different dust prescriptions can alter these trends, particularly flattening the $\xi_{\rm ion,0}-L_{\rm UV}$ relation. The results emphasize that while nebular dust curves substantially improve $\xi_{\rm ion}$ precision, uncertainties in stellar UV attenuation remain a major obstacle for robust inclusion in reionization models, underscoring the need for UV-to-IR, energy-balance constraints to fully close the dust-correction loop.

Abstract

We present ionizing photon production efficiencies ($ξ_{\rm ion}$) for 63 z=1.5-6.9 star-forming galaxies using precise nebular dust attenuation corrections from the JWST/AURORA survey. A subset of objects within AURORA have individually-determined nebular dust attenuation curves, which vary significantly in shape and normalization, resulting in reduced systematic uncertainty when constraining the total attenuation of H$α$ luminosity, and thus the intrinsic ionizing output within our sample. We find evidence for positive correlations between $ξ_{\rm ion}$ and redshift, equivalent width of [OIII]$λ$5007, and O32=[OIII]$λ$5007/[OII]$λ$3726,3729, and negative correlations between $ξ_{\rm ion}$ and stellar attenuation, UV luminosity (L$_{\rm UV}$), stellar mass, and direct-method metallicity. We test alternate dust prescriptions within this sample, and find that the total attenuation is lower when using the commonly-assumed Galactic extinction curve or when assuming that stellar attenuation is equal to nebular attenuation. We also find that assuming either of these alternate dust prescriptions can change the slope of relationships between $ξ_{\rm ion}$ and galaxy property, notably inducing a flat trend between $ξ_{\rm ion}$ and L$_{\rm UV}$ within AURORA. While the novel nebular dust curves derived from AURORA spectroscopy reveal obscured ionizing photon production within star-forming galaxies at these redshifts, a more complete understanding of stellar attenuation is required to fully reduce dust systematics on $ξ_{\rm ion}$ for inclusion in reionization models.

Figures (7)

• Figure 1: Estimation of reddening for GOODSN-22384 ($z=2.993$). The logarithm of observed line flux ratios $R$ relative to H$\alpha$, normalized by the intrinsic line ratio, are shown as a function of the wavelength of a significantly detected line. Solid, black points represent $R$ for a suite of Balmer and Paschen recombination lines for this object. These $R$ values are fit by assuming the nebular attenuation curve derived for this object in R25, shown as the grey curve. The red curve shows an alternate fit assuming the average AURORA nebular attenuation curve. In blue, we display the cardelliRelationshipInfraredOptical1989 curve with $E(B-V)_{\rm neb}$ determined using only the H$\alpha$/H$\beta$ ratio. Total attenuation in magnitudes on H$\alpha$ is reported as $A_{H\alpha}$, which differs depending on the method for constraining reddening. We also list attenuation curve normalizations $R_V$ and fitted $E(B-V)_{\rm neb}$ values for each method.
• Figure 2: Ionizing photon production efficiency as a function of redshift and $\rm M_{\rm UV}$ within AURORA. Datapoints highlighted with black, concentric circles are within the AURORA$_{\rm indv}$ subsample, which each have individual nebular attenuation curves. The rest of the analysis sample uses the AURORA average curve for determining $E(B-V)_{\rm neb}$. Spearman correlation coefficients are shown in orange for the full sample and black for the AURORA$_{\rm indv}$ subsample. Linear fits and 1$\sigma$ confidence intervals to the full analysis sample are shown as dark orange curves with light orange shaded regions, while fits to AURORA$_{\rm indv}$ are shown as black curves with grey shaded regions. Spearman correlation tests and linear regression reveals elevated $\xi_{\rm ion,0}$ at high redshift, and mild, positive evolution with $L_{\rm UV}$.
• Figure 3: Relationships between ionizing photon production efficiency, O32, and oxygen abundance within AURORA. Datapoints, trends, and Spearman correlations are colored as in Figure \ref{['fig:xiion_best']}. Oxygen abundances are determined via the direct method, utilizing detections of temperature-sensitive auroral emission lines. The Spearman correlation test reveals no strong trend between $\xi_{\rm ion,0}$ and $12+\textrm{log(O/H)}$, despite significant correlations between $\xi_{\rm ion,0}$ and O32 and anti-correlations between $12+\textrm{log(O/H)}$ and O32. Linear regression reveals a marginally significant ($1\sigma$), negative trend between $\xi_{\rm ion,0}$ and $12+\textrm{log(O/H)}$. Expanding the sample to include additional lower-metallicity objects is required to fully probe this trend.
• Figure 4: Relationships between $\xi_{\rm ion,0}$ and stellar mass $\rm M_{\rm *}$, stellar $A_V$, and [O iii]$\lambda5007$ equivalent width within AURORA. Datapoints, trends, and Spearman correlations are colored as in Figure \ref{['fig:xiion_best']}. We find strong trends between $\xi_{\rm ion,0}$ and $\textrm{A}_{\rm V}$ and $\xi_{\rm ion,0}$ and $W_{\lambda}$( [O iii]), with the Spearman correlation test showing no correlation between $\xi_{\rm ion,0}$ and $\rm M_{\rm *}$. $W_{\lambda}$( [O iii]) is an effective indirect tracer of $\xi_{\rm ion,0}$, and the calibration derived here from AURORA galaxies has minimal nebular dust systematics.
• Figure 5: The distribution of the total attenuation in magnitudes on H$\alpha$, $\textrm{A}_{\mathrm{H}\alpha}$, for various dust prescriptions. In orange, the fiducial $\textrm{A}_{\mathrm{H}\alpha}$ values are displayed, determined using individual dust curves for objects in AURORA$_{\rm indv}$ and the average AURORA curve for all other objects, using simultaneous fits to significantly-detected H i recombination line ratios. In pink, the AURORA average curve was used for all objects, with $E(B-V)_{\rm neb}$ determined from the Balmer decrement. In blue, the Milky Way cardelliRelationshipInfraredOptical1989 extinction curve was used with a Balmer decrement to determine $E(B-V)_{\rm neb}$. In green, $\textrm{A}_{\mathrm{H}\alpha}$ values were determined directly from fits to stellar population synthesis models, with the calzettiDustContentOpacity2000 stellar attenuation curve assumed. Medians and 16th and 84th percentiles of the distribution are shown as solid points with errorbars. Within AURORA, the average AURORA curve and Balmer decrement result in a similar $\textrm{A}_{\mathrm{H}\alpha}$ distribution as our fiducial analysis, while using the Milky Way curve or $\textrm{A}_{\mathrm{H}\alpha}$ derived from the SED result in lower median $\textrm{A}_{\mathrm{H}\alpha}$ values (resulting in biased $\xi_{\rm ion,0}$).