2025ApJ...981..134P

We have used a combined sample of JADES and CEERS objects in order to constrain ionizing photon production efficiency (ξ_ion) from JWST/NIRSpec and JWST/NIRCam data. We examine 163 objects at 1.06 < z < 6.71 with significant (3σ) spectroscopic detections of Hα and Hβ in order to constrain intrinsic Hα luminosities corrected from nebular dust attenuation via Balmer decrements. We constrain dust-corrected UV luminosities from best-fit spectral energy distribution modeling. We find a sample median ${{\rm{log}}}_{10}({\xi }_{{\rm{ion,0}}}/{{\rm{erg}}\,{\rm{Hz}}}^{-1})=25.2{9}_{-0.37}^{+0.29}$, assuming f_esc = 0 for the escape fraction of Lyman continuum (LyC) emission. We find significant (Spearman r_s = 0.19, p = 0.01) correlation between ξ_ion,0 and z, with objects at z > 4.64 having median ${{\rm{log}}}_{10}({\xi }_{{\rm{ion,0}}}/{{\rm{erg}}\,{\rm{Hz}}}^{-1})=25.3{8}_{-0.38}^{+0.38}$, with those below having ${{\rm{log}}}_{10}({\xi }_{{\rm{ion,0}}}/{{\rm{erg}}\,{\rm{Hz}}}^{-1})=25.2{4}_{-0.33}^{+0.30}$. We also find significant, positive correlations between ξ_ion,0 and L_UV; W_λ([O III]); [O III]λ5007/[O II]λλ3726, 3729; and inverse correlations with metallicity. In contrast with some previous results, we find no trends between ξ_ion,0 and stellar mass, stellar dust attenuation, or UV slope. Applying a multivariate fit to ξ_ion,0, z, and M_UV to an empirically motivated model of reionization, and folding in f_esc estimates from direct observations of the LyC at z ∼ 3 from the Keck Lyman Continuum Spectroscopic Survey, we find that the number of ionizing photons entering the IGM causes reionization to end at z ∼ 5-7.