Probing Infrared eXcess to Investigate Early-Universe Dust (PIXIEDust)

TL;DR

This study leverages about $2\times10^2$ hours of ALMA/NOEMA observations of ten spectroscopically confirmed $z>8$ galaxies to perform a comprehensive flux-, dust mass-, and dust-to-stellar-mass-ratio stacking analysis. No significant dust emission is detected; the derived 3σ limits are $M_{\rm dust} < 9.1\times10^{4}\,M_{\odot}$ and $M_{\rm dust}/M_{\ast} < 3.7\times10^{-4}$ (assuming $T_{\rm dust}=50$ K and $\beta_{\rm dust}=2$), with UV-based estimates generally predicting higher dust masses due to geometric uncertainties. The results, when compared with lower-redshift dust detections and theoretical models, support inefficient dust build-up in the $z>8$ Universe—likely due to limited SN dust production, slow ISM grain growth, or rapid dust removal via outflows—while providing public data and tools to enable future, broader probes of dust in the first 600 million years. The work highlights the importance of multi-wavelength constraints and motivates targeted, larger-area, and shorter-wavelength surveys to uncover the dusty early Universe with upcoming facilities and programs like CHAMPS and ALMA upgrades.

Abstract

Despite the implied presence of dust through reddened UV emission in high-redshift galaxies, no dust emission has been detected in the (sub)millimetre regime beyond $z > 8.3$. This study combines around two hundred hours of Atacama Large Millimeter/submillimeter Array (ALMA) and Northern Extended Millimeter Array (NOEMA) observations on ten $z > 8$ galaxies, revealing no significant dust emission down to a $1 σ$ depth of $2.0$, $2.0$, and $1.5 \,μ$Jy at rest-frame 158, 88 $μ$m, and across all the data, respectively. This constrains average dust masses to be below $< 10^{5}$ M$_{\odot}$ at $3 σ$ and dust-to-stellar mass ratios to be below $3.7 \times{} 10^{-4}$ (assuming $T_{\rm dust} = 50$ K and $β_{\rm dust} = 2.0$). Binning by redshift ($8 < z < 9.5$ and $9.5 < z < 15$), UV-continuum slope ($β_{\rm UV} \lessgtr -2$) and stellar mass ($\log_{10} M_{\ast}/{\rm M_{\odot}} \lessgtr 9$) yields similarly stringent constraints. Combined with other studies, these results are consistent with inefficient dust build-up in the $z > 8$ Universe, likely due to inefficient supernova production, limited interstellar grain growth and/or ejection by outflows. We provide data and tools online to facilitate community-wide high-redshift dust searches.

Figures (12)

• Figure 1: UV Magnitude of our stacking sample of spectroscopically confirmed z>8 galaxies (red squares) as a function of their redshift and age of the Universe. Galaxies with deep observations that are not stacked are shown in blue (Y1; Tamura2019) and grey squares (COS-z-0 and SPT-0615-JD), and compared to galaxy candidates with photometric redshift estimates from the GOODS-N and GOODS-S fields shown in blue pointsHainline2024.
• Figure 2: The left-hand side and middle panels show the stacks of the rest-frame 88 and 158 $\mu$m emissions of eight and five observations, respectively. The right-hand side panel shows the combined stacked emission across all fourteen observations. The continuum weighting is based on the standard deviation of the continuum maps, with dashed and solid contours indicating 2, 3, 4... $\sigma$ in 8 by 8 arcsecond poststamps. The $3 \sigma$ depths of the maps are detailed in Table \ref{['tab:stacking']}. The black indicators guide the eye to the central region where the rest-frame UV / optical emission of the sources is detected.
• Figure 3: The observations of $z > 8$ galaxies typically target the expected wavelengths of bright sub-mm spectral lines ([O iii] 52 $\mu$m, [N iii] 57 $\mu$m, [O iii] 88 $\mu{\rm m}$, and [C ii] 158 $\mu{\rm m}$), and subsequently probe different parts of the dust spectrum. The observational limits and detections of nineteen dust continuum observations across thirteen $z > 8$ sources are shown against the expected dust emission of a $z = 9$ galaxy with a dust mass of $10^6$ M$_{\odot}$ with a $\beta_{\rm dust} = 2$ at three different dust temperatures of $T_{z = 0} =$ 30, 50 and 70 K assuming a modified black-body. The single source with dust continuum detections at $z > 8$, Y1, is shown in blue. The stacks at rest-frame 88 and 158$\mu$m are shown as red arrows.
• Figure 4: The dust mass and dust-to-stellar mass ratio of the individual galaxies (black circles) and of the stack at two different stellar mass bins ($\log_{10} M_{\ast}/{\rm M_{\odot}} \lessgtr 9$; red upper limits) as a function of stellar mass in the left and right panel, respectively. The upper limits are drawn at $3 \sigma$ and the dust and stellar masses are corrected for lensing. These results are compared against individually-detected galaxies at $4.4 < z < 6$ indicated with grey circles Sommovigo2022 and $6 < z < 8$ galaxies indicated with grey squares Bakx2021Sommovigo2022REBELSWitstok2022Fudamoto2022DustTemperaturesAlgera2024Algera2024REBELS25Valentino2024. The sole source with a dust detection at $z = 8.3$ is shown as a blue star Tamura2019Tamura2023Bakx2020CII. Dust estimates from stacking experiments are shown as thin blue upper limits Ciesla2024, and scaling relations from semi-analytical and hydrodynamical models that account for dust production are shown as trend lines between $z = 7$ to $9.5$Popping2017DustproductionImara2018Vijayan2019DiCesare2023Esmerian2024Triani2021Dayal2022. For two dust production models Popping2017DustproductionVijayan2019, we show the maximum and average dust production scenarios in solid and dash-dotted lines respectively.
• Figure 5: The UV-derived dust mass against the infrared to UV-derived dust masses for our sample and stack. The difference between the UV- and infrared-derived dust masses indicate the limitations of the UV-based perspective on dust production and obscured star formation. The dark circles indicate the optically-derived upper limits based on $\beta_{\rm UV, intrinsic} = -2.63$, while the connected grey points indicate the $\beta_{\rm UV, intrinsic} = -2.23$ scenario as suggested for lower-redshift galaxies Meurer1999. Note that these are unavailable for sources with $\beta_{\rm UV} < -2.23$. Individual sources, as well as all the stacks (dark red for the complete stack, with orange colours indicating the binned stacks) lie on the side where the UV-derived dust masses are $\sim 1$ dex in excess of infrared-derived optically-thin dust masses with $T_{\rm dust} = 50$ K and $\beta_{\rm dust} = 2$. The UV-estimated dust mass for Y1 (in light blue) is also larger than the measured value Tamura2019.