The Gamow Peak

The puzzle: stars are too cold to burn

Two protons repel each other electrostatically. To bring them close enough for the strong nuclear force to take over — within a few femtometres — they have to climb a Coulomb potential whose summit is about 550 thousand electron-volts. The centre of the Sun, where fusion runs, is hot, but not that hot: at 15.7 million kelvin the typical thermal energy is only kT ≈ 1.35 keV. The average proton carries about four hundred times too little energy to surmount the barrier. Heat the gas to billions of kelvin and you could do it classically — but the Sun's core sits at fifteen million, and it has been fusing happily for 4.6 billion years.

Arthur Eddington insisted in the 1920s that the Sun must be powered by the subatomic conversion of hydrogen to helium, and when a colleague objected that the core was too cold, he famously retorted that the critic should "go and find a hotter place." The hotter place turned out not to be a higher temperature but a deeper layer of physics. Protons do not climb the barrier — they tunnel through it. And the precise interplay between how many protons are fast enough and how easily each one tunnels produces a sharp spike in energy space. That spike is the Gamow peak: the narrow band of energies where essentially all of a star's fusion is transacted.

Two opposing exponentials

The reaction rate per pair of nuclei is an integral over energy of three things multiplied together: the number of particles at each energy, the probability that such a pair tunnels through the barrier, and the intrinsic nuclear cross-section. The first two are both exponential, and they pull in opposite directions.

The Maxwell-Boltzmann tail. In a thermal gas, the fraction of particle pairs with relative energy E follows

f(E) ∝ √E · exp(−E / kT)

High-energy particles exist, but they are exponentially rare. The number falls off steeply as E climbs above kT.

The tunnelling (Gamow) factor. The quantum probability of penetrating the Coulomb barrier — Gamow's 1928 result — is

P_tunnel(E) ∝ exp(−2πη) = exp(−√(E_G / E))

where η is the Sommerfeld parameter and E_G is the Gamow energy (defined below). This rises steeply with E — faster particles tunnel far more easily, because the barrier they must traverse is thinner.

One factor crashes with energy; the other climbs. Their product, exp(−E/kT − √(E_G/E)), has a maximum at an intermediate energy E₀ and falls away on both sides. That product is the Gamow peak. The remarkable consequence is that the fusion is done neither by the typical thermal particles (too slow to tunnel) nor by the rare super-fast ones (too few to matter), but by a compromise population in a thin energy sliver around E₀.

The governing equations

The Sommerfeld parameter for two nuclei of charges Z₁ and Z₂, reduced mass μ, and relative velocity v is

η = Z₁ Z₂ e² / (ħ v · 4πε₀)   (SI)
2πη = (2π Z₁ Z₂ e²) / (ħ v · 4πε₀) ∝ Z₁ Z₂ / √E

It is convenient to package the energy dependence into the Gamow energy E_G, defined so that 2πη = √(E_G / E):

E_G = 2 μ c² (π α Z₁ Z₂)²

where α ≈ 1/137 is the fine-structure constant and μ is the reduced mass. For two protons (Z₁ = Z₂ = 1, μ = m_p/2), E_G ≈ 493 keV in convenient units (often quoted as the related b² with b ≈ 22.2 keV^1/2, so √E_G ≈ 22.2 keV^1/2). The cross-section is written by factoring out the two strong energy dependences:

σ(E) = [ S(E) / E ] · exp(−√(E_G / E))

Here S(E) is the astrophysical S-factor — what is left after removing the 1/E geometric piece and the tunnelling exponential. Because S(E) is smooth and slowly varying, it is the quantity experimenters actually measure and extrapolate. The thermally averaged reaction rate is then the integral over the Maxwell-Boltzmann distribution:

⟨σv⟩ ∝ ∫₀^∞ S(E) · exp(−E/kT − √(E_G/E)) dE

Treating S(E) as constant, the integrand exp(−E/kT − √(E_G/E)) is sharply peaked. Setting its derivative to zero gives the Gamow peak energy

E₀ = ( √(E_G) · kT / 2 )^(2/3) = ( b kT / 2 )^(2/3),  with b = √(E_G)

and, expanding the integrand as a Gaussian around E₀, an effective 1/e full width

ΔE = (4 / √3) · √(E₀ kT) ≈ 2.31 · √(E₀ kT)

So the peak's location and width are fixed by just two inputs: the Gamow energy (set by the charges of the colliding nuclei) and the temperature (set by where in the star you are).

The key numbers

For the proton-proton reaction at the Sun's central temperature T = 15.7 × 10⁶ K, so kT = 1.354 keV:

Quantity	Symbol	Value (solar p-p)
Central temperature	T_c	1.57 × 10⁷ K
Thermal energy	kT	1.35 keV
Coulomb barrier height	E_C	≈ 550 keV
Gamow peak energy	E₀	≈ 5.9 keV
Ratio E₀ / kT	—	≈ 4.4
Peak 1/e full width	ΔE	≈ 6.4 keV
Barrier / peak ratio	E_C / E₀	≈ 93
Penetration probability at E₀	P	≈ 10⁻⁹

Read those rows together and the picture is stark. The reactions occur at about 5.9 keV — roughly four-and-a-half times the mean thermal energy, yet a factor of ninety below the classical barrier. The particles doing the fusing live on the exponential far tail of the velocity distribution, and even then each pair has only about a one-in-a-billion chance of tunnelling per encounter. The combination of "vanishingly few fast particles" and "tiny tunnelling odds" is precisely why the proton-proton reaction is so slow that a given proton in the Sun's core waits, on average, about 9 billion years to fuse — and that glacial rate is exactly what gives the Sun a multi-billion-year lifetime.

Worked example: locating the solar p-p peak

Let's reproduce E₀ from scratch. For the p-p reaction, Z₁ = Z₂ = 1 and the reduced mass is μ = m_p/2. The Gamow energy parameter is conventionally given as √(E_G) = b where

b = π α Z₁ Z₂ √(2 μ c²)
  = π (1/137.036)(1)(1) · √(2 · (938.27/2) MeV)
  = 0.022925 · √(938.27 MeV)
  = 0.022925 · 968.6 keV^(1/2) · (unit handling)
  ≈ 22.2 keV^(1/2)

(equivalently E_G = b² ≈ 493 keV). Now plug into the peak formula with kT = 1.354 keV:

E₀ = ( b · kT / 2 )^(2/3)
   = ( 22.2 · 1.354 / 2 )^(2/3) keV
   = ( 15.03 )^(2/3) keV
   = 6.09 keV   ≈ 5.9 keV

And the width:

ΔE = (4/√3) √(E₀ kT)
   = 2.309 · √(6.09 · 1.354) keV
   = 2.309 · √(8.25) keV
   = 2.309 · 2.87 keV
   = 6.6 keV   ≈ 6.4 keV

The small residual differences from the table values come from using slightly different reference temperatures (15.7 vs 15.6 × 10⁶ K) and from the screening and S-factor corrections folded into precise solar-model numbers. But the back-of-envelope calculation lands within a few percent — the Gamow peak is genuinely a two-line estimate once you know the charges and the temperature.

Why reaction rates are so temperature-sensitive

Because E₀ ∝ T^(2/3) and the height of the integrand depends exponentially on the combination, the reaction rate near a reference temperature behaves like a steep power law:

rate ∝ T^n,    n = (E₀ / kT − 2) / 3 · 2 ≈ (τ − 2)/3,   τ = 3 E₀ / kT

The local logarithmic slope n grows with the charges of the nuclei, because heavier charges push E₀ higher relative to kT. The result is the single most consequential fact in stellar structure: small temperature changes drive enormous rate changes.

Reaction	Z₁·Z₂	Typical T	E₀ at that T	Rate exponent n
p + p (p-p chain)	1	1.5 × 10⁷ K	≈ 5.9 keV	≈ 4
p + ¹⁴N (CNO cycle)	7	1.5 × 10⁷ K	≈ 26 keV	≈ 18
3α (triple-alpha)	4 (per pair)	1.0 × 10⁸ K	≈ 83 keV	≈ 40
¹²C + ¹²C (carbon burning)	36	8 × 10⁸ K	≈ 2.4 MeV	≈ 30
¹⁶O + ¹⁶O (oxygen burning)	64	2 × 10⁹ K	≈ 6 MeV	≈ 35

The CNO cycle's n ≈ 18 means the rate doubles for a temperature rise of only ~4 percent. That is why CNO burning, with its high-charge p + ¹⁴N bottleneck, switches on sharply above roughly 1.3–1.7 solar masses and produces the steep mass-luminosity relation of upper-main-sequence stars, while the gentle p-p chain (n ≈ 4) governs the Sun. The triple-alpha process is even more violent: n ≈ 40 near 10⁸ K, which is exactly why the helium flash in a degenerate red-giant core is so explosive — a tiny temperature rise unleashes a runaway before the gas can expand and cool.

How it is pinned down — the S-factor and underground labs

You cannot measure σ(E) directly at the few-keV stellar energies: the cross-section is so suppressed by the Gamow factor (≈ 10⁻⁹ for solar p-p) that no terrestrial beam is intense enough to register a count rate. The strategy instead is to measure the smooth S-factor at the highest energies the lab can reach — tens to hundreds of keV — and extrapolate S(E) down to E₀, where it is nearly flat.

The flagship effort is LUNA (Laboratory for Underground Nuclear Astrophysics), an accelerator buried under 1,400 m of rock at the Gran Sasso laboratory in Italy. The rock shields out cosmic rays, cutting background by roughly six orders of magnitude, so reactions like ³He(³He,2p)⁴He and ¹⁴N(p,γ)¹⁵O can be measured at energies that actually overlap the solar Gamow window. LUNA's 2004 measurement of the ¹⁴N(p,γ)¹⁵O reaction — the slowest, rate-limiting step of the CNO cycle — roughly halved the previously accepted S-factor, which in turn shifted predictions for the CNO neutrino flux and the inferred age of globular clusters.

Those CNO neutrinos were finally detected by the Borexino experiment, also at Gran Sasso, which announced the first direct observation of solar CNO neutrinos in 2020 and a refined measurement in 2023 — confirming that about 1 percent of the Sun's energy comes from CNO burning, exactly as the Gamow-peak-weighted rates predict. Meanwhile the proton-proton, beryllium-7, and boron-8 neutrino fluxes measured by Borexino, SNO, and Super-Kamiokande all trace different reactions whose rates are set by their respective Gamow peaks, giving an independent, real-time confirmation of the temperature structure of the solar core.

History: Gamow, Atkinson, Houtermans, Bethe

• 1928 — George Gamow. Working in Göttingen, Gamow solved the quantum problem of a particle escaping a Coulomb barrier and applied it to alpha decay, explaining the Geiger-Nuttall law. The exponential penetration factor exp(−2πη) is "the Gamow factor."
• 1929 — Atkinson & Houtermans. Robert Atkinson and Fritz Houtermans turned the tunnelling result around: if particles can tunnel out of a nucleus, light nuclei can tunnel in to fuse. Their paper showed thermonuclear fusion is possible at stellar temperatures and is the first quantitative theory of stellar energy generation.
• 1938 — Carl von Weizsäcker / 1938-39 — Hans Bethe & Charles Critchfield. Bethe and Critchfield worked out the proton-proton chain (1938); Bethe and von Weizsäcker independently identified the CNO (carbon-nitrogen-oxygen) cycle. Bethe's 1939 paper "Energy Production in Stars" tied the networks together and won him the 1967 Nobel Prize in Physics.
• 1957 — B²FH. Burbidge, Burbidge, Fowler & Hoyle's monumental review systematised stellar nucleosynthesis; William Fowler's experimental program to measure the relevant S-factors won him a share of the 1983 Nobel Prize.
• 1990s–present — LUNA & Borexino. Underground accelerators and detectors push direct measurements into the Gamow window and detect the neutrinos that the Gamow-peak rates produce, closing the loop between the theory and the Sun itself.

Subtleties and common misconceptions

• The peak is not at kT, and not at the barrier. A frequent error is to assume fusion happens at the mean thermal energy (it is too slow there) or just below the barrier (there are essentially no particles there). It happens at E₀ ≈ 4–5 kT for solar p-p — a genuine compromise the two exponentials negotiate.
• "Tunnelling means the barrier doesn't matter" is wrong. The barrier matters enormously: the tunnelling probability at E₀ is only ~10⁻⁹, and it is this suppression that makes the Sun burn slowly enough to live for billions of years rather than detonating.
• Electron screening shifts the rate. In the dense solar plasma, electrons partly shield the nuclear charges, lowering the effective barrier and enhancing the rate by a few percent (the Salpeter weak-screening correction). Precise solar models must include it.
• The Gaussian approximation is only approximate. Replacing the true integrand by a Gaussian around E₀ is excellent for non-resonant reactions but fails when a nuclear resonance sits inside or near the Gamow window — then the rate is dominated by the resonance and the simple E₀, ΔE formulas don't apply. The triple-alpha process, which proceeds through the Hoyle resonance in ¹²C, is the classic example.
• S(E) is "slowly varying," not constant. The whole point of factoring out the Gamow factor is that S(E) is smooth — but it still has a mild slope and occasional resonant structure, which is why measuring its energy dependence near the window (not just one point) matters for the extrapolation.
• Different reactions have different peaks at the same temperature. Because E₀ ∝ (Z₁Z₂)^(2/3), every reaction in a star sits at its own Gamow energy. At the Sun's core, p+p peaks near 6 keV while p+¹⁴N peaks near 26 keV — the high-charge reactions are pushed to higher energies and are therefore far more temperature-sensitive.