Blue Monsters and Dusty Descendants: Reconciling UV and IR Emission from Galaxies from z=7, up to z= 14

Abstract

Recent JWST observations reveal massive, UV-bright galaxies at $z > 10$ with little apparent dust attenuation, whereas ALMA detections at $z \simeq 7$ show similarly massive systems that are already dust-rich and IR-luminous. This raises a fundamental question: can a single physical model of star formation and dust production explain both populations across cosmic time? We address this using a minimal framework with only two free parameters--the instantaneous star formation efficiency ($ε_\star$) and the dust yield per Type II supernova ($y_d$)--and predict the rest-frame UV and IR luminosity functions (LFs) from $z \simeq 14$ to 7. For a uniform ISM, we find a UV-IR tension at the bright end of the LFs at $z \ge 7$. The UV LF requires low dust yields ($y_d \lesssim 0.01\,M_\odot$), whereas the $z=7$ IR LF requires higher yields ($y_d \sim 0.1\,M_\odot$) unless the star formation efficiency is boosted above $ε_\star \approx 5$-10%. We show that incorporating a porous, turbulent ISM largely resolves this tension: turbulence opens low-column-density sightlines that enhance the UV escape fraction while leaving the total absorbed energy--and thus the IR luminosity--nearly unchanged once radiative-transfer--induced flattening of the attenuation curve is included. Large-grain dust distributions, while reducing UV opacity, play a secondary role once ISM porosity and radiative transfer are taken into account. At $z > 10$, however, even strong turbulence cannot reproduce the bright end of the UV LF at high dust yield. This could be resolved either by efficient dust removal in early massive systems or by substantial ISM dust growth by $z \simeq 7$. Our results highlight dust physics as a key lever for interpreting the rapidly growing UV and IR constraints within the broader context of early galaxy formation.

Figures (12)

• Figure 1: From top to bottom: Logarithm of the SFR, stellar (dashed), dust mass (solid), and dust to stellar mass ratio build up as a function of galaxy age for halos in the mass range $10^7-10^{12}\ \mathrm{M_{\odot}}$ at $z=10$ (see colorbar) according to the minimal model described in eq. \ref{['som22']} - \ref{['md']}. Here we assume $\epsilon_{\star}=0.1$ and $y_d=0.1\ \mathrm{M_{\odot}}$ per SN event. The grey shaded band in the bottom panel shows the dust-to-stellar mass ratio compared to that inferred for REBELS galaxies at $z=7$ in Sommovigo22_z7. The vertical line across the second to last panel marks 3.5 Myr.
• Figure 2: Intrinsic UV magnitude $\rm M_{\rm UV}^{\rm int}$ vs. stellar mass ($\rm M_\star$) at $z=10$, assuming $\epsilon_{\star} = 0.1$. Colored points represent spectroscopically confirmed $z>10$ galaxies (see text for detailed references), color-coded by inferred $\tau_V$ (increasing from blue to red; values labeled).
• Figure 3: Left: Dust opacity $\kappa_\lambda$ for Milky Way (Draine03; pink) and stellar dust (Hirashita19; teal). Solid lines show total extinction and dotted lines absorption only; differences arise solely from the adopted grain–size distributions (inset). Diamonds mark the UV opacity at $1500$ Å used in our RT calculations. Right: Time evolution of the V–band optical depth $\tau_V$ for the same halos shown in Fig. \ref{['Fig1']}, assuming Milky Way dust, $\epsilon_{\star} = 0.1$, and a dust yield of $y_d = 0.1\,{\rm M_\odot}$ per SN. The horizontal bar illustrates the uncertainty in the time at which the galaxy becomes optically thick for a halo of mass $M_h = 10^{10.86}\,{\rm M_\odot}$, corresponding to the $16$–$84$th percentile range in $\tau_V$ arising from galaxy-to-galaxy variations in size. Horizontal dashed and dotted grey lines mark $\tau_V = 1$ and $0.1$.
• Figure 4: Attenuation curves obtained for Milky Way dust (left) and stellar–dust grain–size distributions (right). Different shades correspond to different prescriptions for the effective transmission, from darker to lighter: point–source geometry (Eq. \ref{['T1500_sphere']}); stars and dust mixed in a spherical configuration (Eq. \ref{['T1500_mix']}); and the best–fit median attenuation curves (for given $A_V$) from Sommovigo25_TNG, summarised in their Eqs. 7–9 by post–processing $\sim 6500$ TNG50/TNG100 local galaxies with SKIRT over 51 lines of sight per source. Solid (dashed) lines correspond to the optically thick (dashed) cases with $A_V = 4$ ($A_V = 0.1$), respectively. These values bracket the $\tau_V$ values that we predict for our massive halos in Fig. \ref{['Fig_opacity_tau']}.
• Figure 5: Attenuated UV magnitude ($M_{\rm UV}$) versus stellar mass ($M_\star$) at $z=10$, assuming $\epsilon_{\star}=0.1$, MW dust, and varying SN dust yields $y_d = 10^{-3},\,10^{-2},\,0.1\,{\rm M_\odot}$ (dash–dotted, dashed, and dotted curves). The magenta dotted line shows the point–source case $T_{1500,{\rm ps}}$ (Eq. \ref{['T1500_sphere']}), which corresponds to a steeper attenuation curve (see Fig. \ref{['Fig_att_curve']}). The solid line indicates the intrinsic (unattenuated) $M_{\rm UV}$–$M_\star$ relation. Colored symbols denote spectroscopically confirmed $z>10$ galaxies (see text for references), color–coded by the inferred $\tau_V$ (increasing from blue to red; values labeled). The shaded region around the dotted curve reflects the 16th–84th percentile scatter in attenuated $M_{\rm UV}$ due to galaxy–to–galaxy variations in size.