When did the Universe turn dark?

For the first half-billion years after the Big Bang, we do not know whether the Universe was already hiding its newborn stars behind dust, or was still almost perfectly transparent. The two possibilities differ by an order of magnitude: a million dust-obscured galaxies, or essentially none.

The cosmic star-formation history and its missing dust-obscured half Top panel: the total star-formation rate (teal), measured from starlight and corrected for dust, is traced out to redshift 14, but the part hidden by dust (red) is only measured up to redshift 6, and its shaded uncertainty widens rapidly beyond redshift 3 until it spans more than two orders of magnitude and runs off the bottom of the plot. A gold diagonal arrow marks my benchmark: dust directly detected out to redshift 8.3. Bottom panel: the fraction of star formation hidden by dust — about 55 percent today, peaking near 80 percent at cosmic noon, then falling through half around redshift 4 to 5 and on to roughly a quarter by redshift 6, with an uncertainty that by redshift 8 stretches from a few percent to almost everything. Gold markers show my anchors: one galaxy 93 percent hidden at redshift 8.3, while my stack of ten galaxies beyond redshift 8 is nearly transparent. 0 2 4 6 8 10 12 14 10⁻¹ 10⁻² 10⁻³ 10⁻⁴ 13.8 3 1 0.5 0.3 0% 50% 100% Redshift — further right is deeper into the past Age of the Universe (billions of years) Stars formed per year per volume a million hidden galaxies… …or almost none? ? Total star formation at z ≈ 0.0 Dust-obscured star formation at z ≈ 0.1 Total star formation at z ≈ 0.3 Total star formation at z ≈ 0.5 Total star formation at z ≈ 0.7 Dust-obscured star formation at z ≈ 0.7 Total star formation at z ≈ 0.8 Dust-obscured star formation at z ≈ 0.9 Dust-obscured star formation at z ≈ 1.0 Total star formation at z ≈ 1.1 Total star formation at z ≈ 1.1 Dust-obscured star formation at z ≈ 1.5 Total star formation at z ≈ 1.5 Dust-obscured star formation at z ≈ 1.7 Total star formation at z ≈ 2.0 Dust-obscured star formation at z ≈ 2.2 Total star formation at z ≈ 2.3 Total star formation at z ≈ 2.7 Dust-obscured star formation at z ≈ 2.7 Dust-obscured star formation at z ≈ 2.7 Total star formation at z ≈ 3.0 Dust-obscured star formation at z ≈ 3.0 Dust-obscured star formation at z ≈ 4.0 Dust-obscured star formation at z ≈ 4.5 Dust-obscured star formation at z ≈ 4.5 Total star formation at z ≈ 4.9 Dust-obscured star formation at z ≈ 5.2 Dust-obscured star formation at z ≈ 5.5 Total star formation at z ≈ 6.0 Dust-obscured star formation at z ≈ 7.0 Dust-obscured star formation at z ≈ 7.2 Total star formation at z ≈ 7.9 Total star formation at z ≈ 9.0 Total star formation at z ≈ 9.0 Total star formation at z ≈ 9.8 Total star formation at z ≈ 10.0 Total star formation at z ≈ 10.2 Total star formation at z ≈ 11.0 Total star formation at z ≈ 11.0 Total star formation at z ≈ 12.2 Total star formation at z ≈ 12.7 Total star formation at z ≈ 13.5 Total star formation at z ≈ 14.5 all star formation hidden by dust total star formation (HST / JWST, dust-corrected) dust-obscured only (ALMA / submm) Bakx et al. 2021: benchmark — dust directly detected out to z = 8.31, the most distant to date my record: dust seen at z = 8.3 fraction of star formation hidden by dust Bakx et al. 2025: one galaxy at z = 8.31 forms 93% of its stars behind dust 93% hidden (z = 8.3) PIXIEDust (Bakx et al. 2025): my stack of ten z > 8 galaxies is nearly transparent …yet my z > 8 stack: nearly transparent

Each point of light is a measurement of how quickly the Universe formed stars at that moment in cosmic history. Teal points are the total rate, measured from starlight and corrected for the dust we know about; red points are only the part caught directly through its warm dust glow. (Even the backdrop is part of the story: the dark lanes crossing the Milky Way behind this figure are nearby cosmic dust, blotting out starlight.) Dust hides most star formation for most of cosmic time — but beyond a redshift of six its contribution has never been measured, except in one galaxy (↗) where I detected dust just 600 million years after the Big Bang. The lower panel distils it into one number — the fraction of all star formation hidden by dust: about 55% today, peaking near 80% at cosmic noon and still dominant back to redshift four or five (Zavala et al. 2021). Beyond that the measurements stop agreeing: the shaded range opens from a few percent to almost everything, and in the first billion years my own two anchors pull in opposite directions. Pinning down when dust-obscured galaxies came to dominate — the day the Universe turned dark — is the key question driving my research.

Hi — I'm Tom

I'm an astronomer at Chalmers University of Technology in Gothenburg, Sweden, hunting the hidden half of cosmic star formation with the world's largest infrared and radio telescopes — above all ALMA. The research page walks through how each of my projects feeds into the question above; my CV has the short version.

Nanoscale technology for extragalactic science

New instruments enable larger and deeper observations than before.
Here's me testing one such prototype in Chile!

During these studies, I get to use big sub-mm telescopes

This colour of light can probe the insides of these extremely active and dust-obscured galaxies. Light from newly-created stars is absorbed by cold interstellar dust, which re-emits it in far-infrared colours. To get an idea what I've managed to see with them, check out my publications.

Something unclear, send me an email!

Click this beautiful red button to do so!