Cosmic Dawn & First Light

The universe's first sunrise

For its first few hundred million years, the universe was a dark place. After recombination — when protons and electrons combined into neutral hydrogen at a redshift of about z ≈ 1100, roughly 380,000 years after the Big Bang — the cosmos went dark. There were no stars, no galaxies, no quasars; just a uniform fog of cold neutral hydrogen and helium, cooling and expanding, lit only by the fading afterglow of the cosmic microwave background as it slid from orange into the infrared. This is the cosmic dark ages.

Cosmic dawn is the end of that darkness. It is the era — somewhere around z ≈ 30 to 10, or about 100 to 250 million years after the Big Bang — when gravity finally pulled enough gas into the first dark-matter halos for it to collapse, heat up, and ignite the first stars. That first light flooded out into an otherwise pristine universe. The challenge is that we cannot easily see those stars directly: they are faint, far away, and buried in a universe full of neutral gas. Instead, the most powerful probe of cosmic dawn is indirect — the imprint that first light left on the hydrogen all around it, read off through the redshifted 21 cm line.

Why the universe went dark — and how it lit back up

The sequence of cosmic epochs after the Big Bang is tightly ordered, and cosmic dawn sits at a very specific link in the chain:

• Recombination (z ≈ 1100, ~380,000 yr). The plasma cools below ~3000 K, electrons bind to protons, and the universe becomes neutral and transparent. The cosmic microwave background is released.
• Dark ages (z ≈ 1100 → ~30). No luminous sources. Density perturbations grow under gravity but have not yet collapsed into stars. The only signal is the cooling CMB and the 21 cm line of the diffuse hydrogen.
• Cosmic dawn (z ≈ 30 → ~10). The first stars and the first accreting black holes switch on. Ultraviolet, X-ray, and Lyman-α radiation begin to alter the temperature and ionization of the surrounding gas.
• Epoch of reionization (z ≈ 10 → ~6). Accumulated ultraviolet photons ionize the bulk of the intergalactic hydrogen, ending with the universe almost fully ionized by about z ≈ 6 (roughly 1 billion years).

The trigger for cosmic dawn is structure formation. In the standard cold-dark-matter picture, gas falls into dark-matter "minihalos" of about 10⁵–10⁶ M☉. To form a star, that gas must cool and lose pressure support. With no carbon, oxygen, or other metals to radiate heat, the only efficient coolant in pristine gas is molecular hydrogen (H₂), which radiates weakly through rotational and vibrational lines. This sets the cooling threshold — and the minimum halo mass — that gates when the very first stars can form.

The first stars: Population III

The first generation of stars is called Population III (Pop III) — stars formed from gas of essentially primordial composition: about 75% hydrogen and 25% helium by mass, with a metallicity below roughly 10⁻⁴ of the solar value. Their lack of metals has dramatic consequences. Metal-poor gas cools inefficiently, so it cannot fragment into small clumps. Simulations suggest the first star-forming clouds collapsed into a small number of very massive objects — typically tens to a few hundred solar masses, dwarfing the ~0.3 M☉ median of stars forming today.

Such massive stars are hot, with surface temperatures of 10⁵ K and copious output in the ionizing ultraviolet. They burn through their fuel in only a few million years. The most massive of them end as pair-instability supernovae or collapse directly into black holes — potentially providing the seeds for the supermassive black holes seen, fully grown, in quasars less than a billion years after the Big Bang. Crucially, their supernovae dispersed the first heavy elements, so the very next generation of stars (Population II) already formed from enriched gas. Pop III star formation was therefore brief and self-terminating, which is one reason we have never directly observed a confirmed Pop III star.

The physics of the 21 cm signal

The probe of cosmic dawn is the hyperfine transition of neutral hydrogen. The proton and electron each have spin; the ground state of hydrogen is split into a slightly higher-energy "triplet" state (spins parallel) and a lower-energy "singlet" state (spins antiparallel). The transition between them has a rest wavelength of 21.106 cm and a rest frequency of

ν₀ = 1420.405751 MHz   (λ₀ = 21.106 cm)

The relative population of the two levels is described by the spin temperature T_s through a Boltzmann factor:

n₁ / n₀ = 3 · exp(−T★ / T_s),   with  k_B T★ = h ν₀  →  T★ = 0.068 K

Here the factor of 3 is the ratio of statistical weights. What we observe is not the line itself but the contrast of the hydrogen against the cosmic microwave background. The differential brightness temperature is

δT_b ≈ 27 x_HI (1+δ) [ 1 − T_γ / T_s ] · √[(1+z)/10] · ... mK

where x_HI is the neutral fraction, δ the local overdensity, and T_γ ≈ 2.725(1+z) K the CMB temperature. The sign of the bracket is everything: if T_s < T_γ, the gas is colder than the radiation behind it and the hydrogen appears in absorption (δT_b < 0); if T_s > T_γ, it appears in emission. Cosmic dawn is expected to produce a strong absorption trough — the universe's first stars print themselves as a dip, not a glow.

The Wouthuysen-Field coupling — how first light gets recorded

The reason cosmic dawn produces a sharp absorption feature is the Wouthuysen-Field effect, named for Siegfried Wouthuysen (1952) and George Field (1958). Three processes compete to set the spin temperature:

1. CMB photons drive T_s toward T_γ — washing out any contrast.
2. Collisions between hydrogen atoms drive T_s toward the kinetic gas temperature T_K. Collisions matter only in dense or early gas; in the diffuse dark-ages medium they fade as the universe expands.
3. Lyman-α photons from the first stars drive T_s toward T_K via the Wouthuysen-Field effect: a Lyman-α photon excites the atom to the 2p state, from which it can decay into a different hyperfine sublevel. Repeated scattering shuffles the hyperfine populations and locks T_s to the gas temperature.

Before the first stars, collisional coupling has faded, T_s ≈ T_γ, and the 21 cm signal vanishes. When cosmic dawn arrives, the first ultraviolet sources flood the universe with Lyman-α photons. These photons couple T_s to the gas temperature T_K — and at that moment the gas is much colder than the CMB, because it has been cooling adiabatically as (1+z)² while the radiation cools only as (1+z). So T_s plunges below T_γ and the signal drops into deep absorption. Later, X-rays from the first black holes and supernova remnants heat the gas above the CMB, the signal flips into emission, and finally reionization wipes out the neutral hydrogen and the signal disappears altogether. That full history — absorption trough, emission, fade — is the 21 cm "global signal."

Cosmic dawn by the numbers

Because the 21 cm line is redshifted, every epoch maps to a specific observing frequency. The relation is simply ν_obs = 1420.4 MHz / (1+z):

Epoch	Redshift z	Cosmic age	21 cm frequency	Expected signal
Recombination	~1100	~380,000 yr	1.29 MHz	CMB released; line decoupled
Dark ages	200 → 30	~3 → 100 Myr	7 → 46 MHz	Weak collisional absorption
Cosmic dawn (first stars)	~30 → 15	~100 → 270 Myr	46 → 89 MHz	Deep Lyman-α absorption trough
EDGES claimed trough	~17	~180 Myr	78 MHz	Absorption, ~0.5 K deep
X-ray heating	~15 → 10	~270 → 470 Myr	89 → 129 MHz	Absorption → emission
Reionization	~10 → 6	~470 → 940 Myr	129 → 203 MHz	Emission fading to zero

The scales involved are extreme. The expected global signal is tens to a couple hundred millikelvin against a Galactic synchrotron foreground that is thousands of kelvin at these frequencies — a contrast of one part in 10⁴ to 10⁵. The CMB temperature at z = 17 was about 49 K; the adiabatically cooled gas would have been near 6–7 K, giving a natural absorption depth around 0.2 K. The EDGES team reported roughly 0.5 K, about twice as deep as standard cosmology allows.

Hunting the signal — EDGES, SARAS, and the next arrays

Two complementary strategies are used to find cosmic dawn in the 21 cm line. Global signal experiments use a single well-calibrated dipole antenna to measure the sky-averaged spectrum, looking for a broad dip of a fraction of a kelvin. Interferometric power-spectrum experiments use arrays of dipoles to measure the spatial fluctuations of the signal across the sky.

• EDGES (Experiment to Detect the Global EoR Signature). A pair of broadband dipoles in Western Australia. In 2018 it reported a flat-bottomed absorption profile centred at 78 MHz (z ≈ 17), roughly 0.5 K deep — announced as the first fingerprint of cosmic dawn. The anomalous depth spawned a wave of exotic explanations, including baryon–dark-matter scattering that cools the gas, or an excess radio background above the CMB.
• SARAS 3. An Indian radiometer floated on a lake to suppress ground effects. In 2022 it reported a non-detection of the specific EDGES profile at 95.3% confidence, casting serious doubt on the claimed signal and underlining how hard foreground and instrumental systematics make this measurement.
• Interferometers — LOFAR, MWA, HERA, and the SKA. The Hydrogen Epoch of Reionization Array (HERA) in South Africa targets the power spectrum and has set increasingly tight upper limits, ruling out the coldest reionization models. The Square Kilometre Array (SKA-Low) aims ultimately to image the bubbles of ionized hydrogen growing during reionization and to map the cosmic-dawn fluctuations directly.

Because of the synchrotron foregrounds, the lowest dark-ages frequencies (a few MHz, below the ionospheric cutoff) are essentially impossible from the ground. This motivates proposed lunar far-side radio observatories — the Moon's farside is the most radio-quiet location in the inner Solar System — which would target the pristine dark-ages signal at z > 30.

Common misconceptions and edge cases

• Cosmic dawn is not the same as the CMB or "first light" in the radiation sense. Cosmologists sometimes call the release of the CMB at recombination the "surface of last scattering," but cosmic dawn refers specifically to the first stellar light, hundreds of millions of years later. The CMB is the universe's baby photo; cosmic dawn is its first sunrise.
• The 21 cm signal at cosmic dawn is absorption, not emission. A common intuition is that "turning on the first stars" should make the sky brighter. At cosmic dawn the opposite happens at 21 cm: starlight couples the spin temperature to the cold gas, pulling it below the CMB temperature and creating a dip. Emission only appears later, once X-rays heat the gas above the CMB temperature.
• Cosmic dawn ≠ reionization. Cosmic dawn is the onset of the first sources; reionization is the later, more energetic phase when those sources finish ionizing the bulk of intergalactic hydrogen. They overlap but are distinct epochs — cosmic dawn near z ≈ 15–30, reionization completing near z ≈ 6.
• EDGES is contested, not confirmed. It is tempting to treat the 78 MHz trough as the established discovery of cosmic dawn. It is not. The SARAS 3 non-detection means the field still awaits an independent confirmation before the signal can be called real.
• We have not seen a Pop III star. The James Webb Space Telescope has pushed galaxy detections back toward z ≈ 14, deep into the cosmic-dawn era, but a clean, individually confirmed Population III star or pure Pop III galaxy has not yet been identified. Candidates exist; confirmation does not.