The Deuterium Bottleneck: The Delay That Set Big Bang Nucleosynthesis

What the bottleneck is: the mandatory first rung

Building any nucleus heavier than a single proton requires passing through deuterium (²H, or D), a bound proton–neutron pair. There is no way around it: helium-4 cannot be assembled by slamming four nucleons together simultaneously — the universe is far too dilute for four-body collisions. Instead, nucleosynthesis proceeds in two-body steps, and every path begins with radiative capture:

• p + n → D + γ (the gateway reaction)
• D + D → ³He + n, D + p → ³He + γ
• ³He + D → ⁴He + p, and so on up to ⁷Li

The catch is that deuterium is fragile. Its binding energy is only 2.22 MeV — the weakest of all the light nuclei (compare ⁴He at 28.3 MeV, or 7.07 MeV per nucleon). Because the whole reaction network is gated by this one weakly-bound isotope, the rate at which D can survive sets the pace for everything downstream. The bottleneck is therefore not about a lack of neutrons or protons; both are plentiful. It is about keeping the first product intact long enough for the next collision to happen.

The mechanism: too many photons, not too much heat

The counterintuitive heart of the bottleneck is that it persists to temperatures far below deuterium's binding energy. Naively, once the mean thermal energy kT drops below 2.22 MeV — around T ≈ 0.8 MeV — deuterons should survive. They don't, and the reason is the enormous photon-to-baryon ratio.

The baryon-to-photon ratio is η ≈ 6×10⁻¹⁰, meaning there are roughly 10⁹ photons for every baryon. The photon energies follow a blackbody (Planck) distribution, and the number of photons with energy above the 2.22 MeV dissociation threshold falls off exponentially, roughly as exp(−B_D / kT). Deuterium can only accumulate once the fraction of photons above threshold drops below ~η, i.e. when:

• exp(−B_D / kT) ≲ η ≈ 10⁻⁹
• ⇒ B_D / kT ≳ ln(10⁹) ≈ 21
• ⇒ kT ≲ 2.22 MeV / 21 ≈ 0.1 MeV

So the bottleneck breaks not at 2.2 MeV but near 0.07–0.08 MeV — about 30 times cooler. It is a competition between the runaway high-energy tail of a huge photon bath and a tiny population of baryons: the photons win until the tail finally thins out.

Characteristic numbers and a worked estimate

Plug in the numbers and the timeline falls out. Using kT ≈ 0.07 MeV as the release temperature and the radiation-dominated relation T ∝ t^(−1/2) (specifically T ≈ 1.5×10¹⁰ K × t^(−1/2) with t in seconds), a temperature of ~8×10⁸ K corresponds to:

• t ≈ 180 s ≈ 3 minutes after the Big Bang

Meanwhile, neutrons are decaying with a mean lifetime of τ_n ≈ 880 s (half-life ~611 s). The neutron-to-proton ratio froze near n/p ≈ 1/6 at weak freeze-out (T ≈ 0.8 MeV, t ≈ 1 s); during the 3-minute wait it decays to about n/p ≈ 1/7. This is the crucial cost of the delay — every neutron lost to decay is one fewer available for helium.

Once the bottleneck breaks, nearly all surviving neutrons are locked into ⁴He almost instantly. The resulting helium mass fraction is:

• Y_p = 2(n/p) / (1 + n/p) ≈ 2(1/7) / (8/7) ≈ 0.25

Observations pin Y_p ≈ 0.247. Leftover, unburned deuterium freezes out at D/H ≈ 2.5×10⁻⁵, a number exquisitely sensitive to η.

How we observe it: deuterium as a baryometer

We cannot watch the bottleneck directly, but its fingerprint survives in the primordial deuterium abundance. Deuterium is a uniquely clean cosmological probe because stars only destroy it (it is 'astrated' — burned in stellar interiors) and never create net amounts. So any deuterium we find sets a firm floor on the primordial value.

• Where it's measured: in the absorption spectra of distant quasars whose light passes through nearly pristine, metal-poor gas clouds at redshifts z ≈ 2–3. Damped Lyman-α systems show the deuterium Lyman line slightly blueshifted from hydrogen's by the isotope shift.
• The result: D/H = (2.5 ± 0.03)×10⁻⁵, one of the most precise numbers in cosmology.

Because deuterium survival depends steeply on η (higher baryon density → faster D-burning → less leftover D), measuring D/H yields the cosmic baryon density: Ω_b h² ≈ 0.0224. Remarkably, this agrees to within a percent with the completely independent value from the cosmic microwave background (Planck), a triumph that validates the entire bottleneck picture. The abundance also tightly constrains the number of relativistic neutrino species, N_eff ≈ 3.

How it differs from stellar and other nucleosynthesis

The deuterium bottleneck is specific to the early universe, and contrasting it with stellar fusion sharpens why it matters.

• Versus stellar hydrogen burning: In the Sun's core, the p–p chain also begins with a weak-interaction step (p + p → D + e⁺ + ν), but there is no photodissociation problem — the photon-to-baryon ratio inside a star is minuscule (~10⁻¹⁰ times the cosmic value at the relevant temperatures), so deuterium is consumed the instant it forms. Stars have all the time in the world; the bottleneck there is the slow weak first step, not the radiation field.
• Versus the neutron-capture (r- and s-) processes: Heavy elements past iron form by neutron capture in supernovae and neutron-star mergers, a completely different physics regime, and are irrelevant to BBN.
• Versus the 'gap' problem: BBN also stalls at mass numbers 5 and 8 because no stable nucleus exists there (no ⁵He or ⁸Be). That is a nuclear-structure gap, distinct from the deuterium bottleneck, which is a photodissociation kinetics problem. Together, the two prevent BBN from producing much beyond lithium.

Significance and open questions

The bottleneck is not a footnote — it is the pacemaker of primordial chemistry. Its resolution at ~3 minutes, set against the ~880 s neutron lifetime, is precisely what fixes the helium fraction near a quarter of all ordinary matter by mass, a prediction confirmed by observation and by three foundational figures of BBN theory: George Gamow, Ralph Alpher, and Robert Herman in the 1940s, later refined by Hoyle, Wagoner, Fowler, and the Peebles group in the 1960s.

The bottleneck also acts as a sensitive dial on new physics. Shift η, the expansion rate, N_eff, or the neutron lifetime, and you shift when the bottleneck breaks and therefore every abundance.

• The lithium problem: BBN predicts ⁷Li/H about 3× higher than the value seen in old, metal-poor halo stars — a persistent, unresolved discrepancy that may point to stellar depletion, uncertain nuclear rates, or physics beyond the Standard Model.
• Neutron lifetime tension: 'beam' and 'bottle' experiments disagree on τ_n by ~9 seconds, which feeds directly into the predicted helium and deuterium yields and remains an active experimental controversy.

Landmark temperatures and times during Big Bang Nucleosynthesis, showing where the deuterium bottleneck sits

Epoch / event	Temperature (MeV / K)	Time after Big Bang	What happens
Weak freeze-out (n/p ratio locked)	~0.8 MeV / 9×10⁹ K	~1 s	Neutron-to-proton ratio freezes near 1:6, then drifts to ~1:7 by decay
Naive expectation for D formation	~2.2 MeV / 2.6×10¹⁰ K	~0.2 s	Where D would survive if photons did not outnumber baryons 10⁹:1
Deuterium bottleneck (actual onset)	~0.07–0.08 MeV / 8–9×10⁸ K	~3 min (180 s)	Photodissociation finally loses; D accumulates and fusion ignites
Helium-4 assembly	~0.07 MeV / 8×10⁸ K	~3–5 min	Nearly all neutrons swept into ⁴He; mass fraction Y_p ≈ 0.247
BBN freeze-out (reactions cease)	~0.03 MeV / 3×10⁸ K	~20 min	Density/temperature too low; abundances frozen for cosmic time