---
title: Deuterium Bottleneck in Primordial Nucleosynthesis
url: https://www.emergentmind.com/topics/deuterium-bottleneck
type: topic
---

# Deuterium Bottleneck in Primordial Nucleosynthesis

Searching arXiv for the cited papers to ground the article in current paper metadata.
The **deuterium bottleneck** is the delay in primordial nucleosynthesis caused by the fact that deuterium is weakly bound and therefore easily photodissociated in the high-entropy radiation bath of the early universe. Even once the reaction
\[
p + n \rightarrow D + \gamma
\]
can occur, the reverse destruction by energetic photons keeps the equilibrium deuterium abundance tiny until the temperature falls enough that photodissociation becomes ineffective. Only after deuterium survives in appreciable numbers can the rest of the light-element reaction network proceed efficiently, so deuterium functions as the gateway nucleus for synthesis beyond free nucleons [1511.03843].

## 1. Standard cosmological meaning

In standard Big Bang nucleosynthesis, the bottleneck is fundamentally set by the competition between deuterium formation and photodissociation in a high-\(\gamma/b\) radiation bath. Although the deuteron is bound by \(B_D \simeq 2.22\) MeV, it cannot survive at very early times because the Universe contains an enormous excess of photons over baryons, \(\eta^{-1}\sim 10^9\). Even when the average photon energy has dropped below \(B_D\), the high-energy tail of the blackbody still contains enough photons above \(2.22\) MeV to photodissociate newly formed deuterons through \(\mathrm{D}(\gamma,p)n\). Thus \(p(n,\gamma)\mathrm{D}\) can occur, but its product is immediately destroyed [1511.03843].

Once the bottleneck lifts, reactions such as
\[
D + p \rightarrow {}^3\mathrm{He} + \gamma,
\]
\[
D + D \rightarrow {}^3\mathrm{He} + n,
\]
\[
D + D \rightarrow T + p
\]
follow, and then \(A=3\) nuclei feed production of \(^{4}\)He and, through later channels, \(^{7}\)Be/\(^{7}\)Li. In standard BBN language, almost all synthesis beyond free nucleons must pass through surviving deuterium [1203.5701].

This suggests that the deuterium bottleneck is not merely one reaction threshold among many. It is the phase boundary between a nearly nucleon-only plasma and a network in which rapid flow toward \(A=3\), \(A=4\), and mass-7 nuclei becomes possible. The paper on improved deuterium rates does not alter that standard explanation; it treats the bottleneck itself as background BBN physics and addresses instead the next stage, namely how fast deuterium is burned once it can first survive [1511.03843].

## 2. Release of the bottleneck and post-bottleneck processing

The release of the bottleneck is the moment at which newly made deuterium can survive long enough to act as the stepping stone to heavier light nuclei. Physically, that release occurs when photodissociation becomes ineffective relative to \(n+p\to \mathrm{D}+\gamma\). The standard bottleneck-lifting condition is described as the usual BBN epoch at \(T \sim 0.07\)–\(0.1\) MeV, corresponding to times of order \(10^2\) s, precisely the era relevant to late-decay scenarios with lifetimes \(\sim 100\)–500 s [1203.5701].

After release, the surviving deuterium abundance is set by how rapidly deuterium is processed into \(^3\)He and \(^3\)H. The reactions
\[
\mathrm{D}(p,\gamma)^3\mathrm{He},\qquad \mathrm{D}(d,n)^3\mathrm{He},\qquad \mathrm{D}(d,p)^3\mathrm{H}
\]
govern deuterium destruction after that release, and hence control how much deuterium remains frozen out as the primordial residual. In that sense they are post-bottleneck processing reactions rather than the reactions that determine the existence of the bottleneck in the first place [1511.03843].

The sensitivity coefficients quoted for the final deuterium abundance make that role explicit. For \(\mathrm{D}(p,\gamma)^3\mathrm{He}\),
\[
\frac{\Delta{\rm (D/H)}}{{\rm D/H}} =-0.32\frac{\Delta\langle\sigma v\rangle_{\mathrm{d(p,\gamma)^3He}}}{\langle\sigma v\rangle_{\mathrm{d(p,\gamma)^3He}}},
\]
and for the two deuteron-deuteron channels,
\[
\frac{\Delta{\rm (D/H)}}{{\rm D/H}} =-0.54\frac{\Delta\langle\sigma v\rangle_{\mathrm{d(d,n)^3He}}}{\langle\sigma v\rangle_{\mathrm{d(d,n)^3He}}}
-0.46\frac{\Delta\langle\sigma v\rangle_{\mathrm{d(d,p)^3H}}}{\langle\sigma v\rangle_{\mathrm{d(d,p)^3H}}}.
\]
These coefficients summarize the post-release regime: once deuterium exists in appreciable amount, faster destruction channels reduce final \(\mathrm{D/H}\), with the two \(d+d\) reactions having even larger leverage on the final abundance than \(d(p,\gamma)\) [1511.03843].

A direct numerical consequence appears in the updated abundance calculation. Starting from a nominal calculation with updated cosmological parameters but older rates, the quoted value is
\[
{\rm D/H}=2.635\times10^{-5}.
\]
Updating only the two \(d+d\) rates changes this to
\[
{\rm D/H}=2.526\times10^{-5},
\]
and updating also \(d(p,\gamma)\) gives
\[
{\rm D/H}=2.452\times10^{-5}.
\]
The full Monte Carlo result is
\[
\mathrm{D/H} = (2.45\pm0.10)\times10^{-5}\qquad (2\sigma),
\]
equivalently
\[
(2.45\pm0.05)\times10^{-5}\qquad (1\sigma).
\]
This is exactly the outcome expected when post-bottleneck destruction channels are increased: after deuterium begins to survive, it is burned away more efficiently, leaving a smaller residual frozen abundance [1511.03843].

## 3. Observational deuterium and the lithium connection

At the baryon density determined by the microwave anisotropy spectrum, standard BBN predicts deuterium fairly well, but the comparison is not exact. Using the WMAP baryon-to-photon ratio
\[
\eta = 6.16 \times 10^{-10},
\]
the quoted standard BBN predictions are
\[
\mathrm{D/H} = (2.54 \pm 0.17)\times 10^{-5}
\]
from Cyburt, Fields, Olive (2008), and
\[
\mathrm{D/H} = 2.59 \times 10^{-5}
\]
with estimated error \(\pm 0.15\times 10^{-5}\) from Coc et al. (2012). These are compared with the observationally inferred weighted mean from nine quasar absorption systems,
\[
\mathrm{D/H} = (3.05 \pm 0.22)\times 10^{-5},
\]
with scale factor
\[
S = 2.0 = \sqrt{\chi^2/8}
\]
applied to inflate the uncertainty [1203.5701].

The same work emphasizes the large dispersion in the individual measurements. It quotes a sample variance corresponding to
\[
0.62 \times 10^{-5},
\]
much larger than expected from the formal individual errors, and argues that systems with
\[
\mathrm{D/H} \simeq 4 \times 10^{-5}
\]
may better represent the primordial or “post-BBN” abundance, while systems with lower D/H may have undergone local deuterium destruction. Alternative subsets are also quoted:
\[
\mathrm{D/H} = (2.80 \pm 0.20)\times 10^{-5}
\]
excluding Levshakov et al., with sample variance \(0.52 \times 10^{-5}\), and
\[
\mathrm{D/H} = (3.11 \pm 0.21)\times 10^{-5}
\]
excluding Pettini & Bowen, with sample variance \(0.55 \times 10^{-5}\) [1203.5701].

The deuterium bottleneck becomes especially consequential because of the cosmological \(^{7}\)Li problem. Standard BBN predicts
\[
{}^{7}\mathrm{Li/H} = (5.07^{+0.71}_{-0.62}) \times 10^{-10}
\]
or
\[
{}^{7}\mathrm{Li/H} = 5.24 \times 10^{-10},
\]
whereas observations of metal-poor stars give values around
\[
{}^{7}\mathrm{Li/H} = (1.23^{+0.34}_{-0.16}) \times 10^{-10}
\]
and
\[
{}^{7}\mathrm{Li/H} = (1.58 \pm 0.31) \times 10^{-10}.
\]
The excess is therefore by a factor of roughly \(4\)–\(5\) [1203.5701].

Many proposed solutions to this lithium problem alter nuclear processing in ways that affect deuterium. The classes of mechanisms discussed include altered or resonant nuclear reaction rates, decay of massive particles during or after BBN, photon cooling by an axion condensate, and variation of fundamental constants. The paper repeatedly states that a “tight correlation” exists between the post-BBN D/H and \(^{7}\)Li/H abundances, and its focused assessment summarizes the effect as a quantitative D–Li anticorrelation: lower \(^{7}\)Li/H \(\Rightarrow\) higher D/H [1203.5701].

## 4. Nonstandard nucleosynthesis and deuterium as a probe of new physics

A central result of the beyond-Standard-Model discussion is that deuterium is not only the first stable stepping stone of standard synthesis but also one of the main byproducts of nonthermal helium breakup. Any post-BBN mechanism that injects hadrons or photons therefore tends naturally to overproduce D unless carefully constrained [1203.5701].

The case treated most explicitly is late decay of massive gravitinos in supersymmetric scenarios. Gravitino decay at lifetimes of order \(100\)–500 s injects hadronic and electromagnetic energy, leading to photo-erosion and spallation of \(^{4}\)He, production of extra deuterium, production of free neutrons, and destruction of freshly synthesized \(^{7}\)Be, which ordinarily later decays to \(^{7}\)Li. For shorter lifetimes, neutrons trigger
\[
{}^{7}\mathrm{Be}(n,p){}^{7}\mathrm{Li},
\]
followed by
\[
{}^{7}\mathrm{Li}(p,\alpha){}^{4}\mathrm{He},
\]
thereby reducing final \(^{7}\)Li. For longer lifetimes (\(\gtrsim 10^4\) s), the \(A=3\) products of spallation instead feed
\[
{}^{3}\mathrm{He}(\alpha,\gamma){}^{7}\mathrm{Be}
\]
and
\[
T(\alpha,\gamma){}^{7}\mathrm{Li},
\]
increasing lithium [1203.5701].

The favorable window for solving the lithium problem is therefore one in which neutron injection destroys \(^{7}\)Be/\(^{7}\)Li, but the unavoidable collateral effect is increased deuterium from helium breakup. This tradeoff is quantified in the quoted best-fit results from Cyburt et al. (2010): gravitino masses
\[
m_{3/2} \sim 4 - 5\ \mathrm{TeV}
\]
and abundance parameter
\[
\zeta_{3/2} \equiv m_{3/2} n_{3/2}/n_\gamma \sim 5 \times 10^{-11} - 5 \times 10^{-10}
\]
give typical post-BBN abundances
\[
\mathrm{D/H} \approx 3.2 \times 10^{-5},
\qquad
{}^{7}\mathrm{Li/H} \approx 2.4 \times 10^{-10}.
\]
The corresponding fit improvement in \(\chi^2\) is from
\[
\chi^2 = 31.7
\]
for standard BBN to
\[
\chi^2 \simeq 5.4 - 7.0
\]
for benchmark points [1203.5701].

The same study gives representative lithium targets and corresponding D/H values:
\[
{}^{7}\mathrm{Li/H} = 1.23 \times 10^{-10}
\quad \leftrightarrow \quad
\mathrm{D/H} = 4.4 \times 10^{-5},
\]
\[
{}^{7}\mathrm{Li/H} = 1.58 \times 10^{-10}
\quad \leftrightarrow \quad
\mathrm{D/H} = 3.9 \times 10^{-5},
\]
\[
{}^{7}\mathrm{Li/H} = 2.34 \times 10^{-10}
\quad \leftrightarrow \quad
\mathrm{D/H} = 3.3 \times 10^{-5}.
\]
These numbers make deuterium a decisive constraint on nonstandard processing: reducing \(^{7}\)Li generally enhances D/H [1203.5701].

The improved reaction-rate analysis reinforces the same point from the opposite direction. It states that many attempts to reconcile Li observations with models lead to an increased D prediction, and derives a qualitative lithium–deuterium anti-correlation for late neutron injection. In that treatment, extra neutrons destroy \(^7\)Be via
\[
\frac{dY_{^7{\rm Be}}}{dt} = -Y_{^7{\rm Be}}\,\rho\,N_A\langle\sigma v\rangle_{\rm be7np}\,\delta Y_n,
\]
while simultaneously overproducing deuterium through
\[
\frac{dY_{\rm D}}{dt} = +Y_{\rm H}\,\rho\,N_A\langle\sigma v\rangle_{\rm pn\gamma}\,\delta Y_n.
\]
A plausible implication is that precision D/H acts not only as a consistency check on standard BBN but also as a veto on many attempted lithium fixes [1511.03843].

## 5. Fragility of deuterium and post-BBN chemical evolution

A major interpretive issue is that primordial or post-BBN deuterium abundance is not necessarily identical to the abundance later measured in individual astrophysical systems. The relevant physical reason is the extreme fragility of deuterium. The paper states explicitly that deuterium is destroyed in stars at
\[
T \sim 10^5\ \mathrm{K},
\]
and that D can only decrease in chemical evolution [1203.5701].

This one-way behavior motivates the argument that high-redshift absorbers with low D/H need not define the primordial value. Instead, those systems may have experienced local astration. The same study therefore proposes that the highest observed D/H systems are likely closer to the primordial or post-BBN abundance, while lower D/H systems reflect local destruction [1203.5701].

The chemical-evolution framework invoked is a hierarchical structure formation model with multiple star-formation modes, including an early intermediate-mass population designed to allow significant deuterium destruction without strong heavy-element production. In that model, global D destruction begins around redshift \(z \sim 3\), corresponding to the peak cosmic star formation rate; average D/H can be reduced modestly while leaving \(^{7}\)Li nearly unchanged at low metallicity; and local systems can show much stronger depletion. Earlier work is cited indicating that D/H destruction factors can range from
\[
2 \text{ to } 10,
\]
depending on gas fraction and star formation history [1203.5701].

The improved-rate paper also extends deuterium evolution beyond primordial freeze-out. After cosmic astration, its model gives
\[
{\rm D/H}(z\approx 3)\approx (2.42\pm0.05)\times10^{-5},
\]
still compatible at \(2\sigma\) with the observed DLA values [1511.03843]. This suggests that the deuterium bottleneck determines primordial formation and early survival, whereas subsequent astrophysical processing can only move D/H downward.

## 6. Limiting cases and broader uses of the concept

The bottleneck can be illuminated by considering the limiting case in which deuterium is not stable. In universes where the strong interaction is slightly weaker so that deuterium has no bound state, while \(A=3\) nuclei and \(^{4}\)He remain bound, the standard BBN flow into \(^{3}\)He and \(^{4}\)He is suppressed and low-mass stellar hydrogen burning via the usual \(pp\)-chain is likewise disabled [1612.04741].

That study introduces
\[
\Delta_d \equiv m_d - (2m_p + m_e),
\]
with \(\Delta_d>0\), and examines \(\Delta_d = \mathcal{O}(1\,\mathrm{MeV})\). Assuming kinetic and chemical equilibrium, the deuterium abundance becomes
\[
n_d = n_p^3 \left(\frac{g_d}{g_e g_p^2}\right) 2^{3/2} \frac{(2 \pi)^3}{(m_e m_p T)^{3/2} T^3} \exp\!\left[-\frac{\Delta_d}{T}\right],
\]
or numerically
\[
\chi_d \equiv \frac{n_d}{n_p} = 2.3\times 10^{-3}\, n_{30}^2 T_9^{-3} \exp\!\left[-11.63\frac{\Delta_{\rm mev}}{T_9}\right].
\]
For \(\Delta_{\rm mev}=1\), \(T_9=1\), and \(n_{30}=1\), this gives \(\chi_d \sim 2\times10^{-8}\), but the paper emphasizes that stellar interiors and BBN generally do not reach that NSE abundance because the forward weak production of deuterium is too slow compared to deuterium decay [1612.04741].

Using
\[
\langle \sigma v \rangle_{pp} = 6.34 \times 10^{-39}\, {\rm cm}^3\,{\rm s}^{-1}\, T_9^{-2/3} f(T_9) \exp\!\left[-3.380 T_9^{-1/3}\right],
\]
with
\[
f(T_9)=1+0.123T_9^{1/3}+1.09T_9^{2/3}+0.938T_9,
\]
the associated \(pp\) timescale is
\[
\tau_{pp}\approx 1.6\,{\rm yr}\, n_{30}^{-1} \exp\!\left[3.380 T_9^{-1/3}\right].
\]
At \(T_9=1\), \(n_{30}=1\), this gives \(\tau_{pp}\approx 47\) yr, while an assumed deuterium lifetime is \(\lambda_d^{-1}\sim10^{-16}\,\mathrm{s}\). Steady state then gives
\[
n_d = \frac{1}{2}n_p^2 \langle \sigma v \rangle_{pp}\lambda_d^{-1},
\]
and for \(T_9=1\), \(n_{30}=1\),
\[
\chi_d \approx 3\times10^{-26}.
\]
The paper identifies this as a transformation of the standard deuterium bottleneck from a thermodynamic bottleneck into a kinetic one [1612.04741].

In modified BBN without stable deuterium, the replacement reaction is effectively
\[
n+p+p \rightarrow {^3\mathrm{He}}+\gamma,
\]
with production rate
\[
\frac{dY_3}{dt}\biggr|_{+} = Y_n Y_p^2\,\frac{\rho_b^2 N_A^2}{2\Gamma(d)} \,\langle np\rangle \langle dp\rangle .
\]
Because this is a three-body effective process proportional to \(\rho_b^2\), BBN densities are too low for it to compete with Hubble expansion once NSE fails. The resulting helium abundance is only
\[
X_4 \sim 10^{-14},
\]
with even smaller abundances of lithium and beryllium [1612.04741].

The same paper argues, however, that the bottleneck is severe rather than absolutely fatal. Gravitational contraction can power stars, stars above the pure-hydrogen Chandrasekhar mass
\[
M_{\rm Ch}\approx 5.6\,M_\odot
\]
can undergo collapse and explosive nucleosynthesis, a triple-nucleon process can bridge the \(A=2\) gap in hot dense stellar cores, and once trace carbon exists, the CNO cycle can operate without any deuterium in the catalytic loop. A plausible implication is that the deuterium bottleneck is a central organizing principle of ordinary nucleosynthesis, but its removal does not by itself constitute an absolute anthropic prohibition [1612.04741].

Source: https://www.emergentmind.com/topics/deuterium-bottleneck