CNO Cycle

What the CNO cycle is

The CNO cycle — sometimes called the carbon cycle or the Bethe-Weizsäcker cycle — is the second of the two main hydrogen-burning mechanisms in stars. Like the proton-proton chain, its net result is the conversion of four hydrogen nuclei into one helium-4 nucleus, releasing 26.73 MeV of energy. Unlike the pp chain, the CNO cycle uses heavy nuclei — carbon, nitrogen, and oxygen — as catalysts. The protons are added one at a time to a CNO nucleus, which transmutes through a sequence of intermediate isotopes and emits an alpha particle (a ⁴He nucleus) at the end, restoring the original catalyst.

The cycle was proposed independently by Hans Bethe and Carl Friedrich von Weizsäcker in 1938. Bethe published the more complete analysis, including all the alternative branches and a full energy budget, and won the 1967 Nobel Prize for the work. The CNO cycle and the proton-proton chain together solve the problem that had defeated physics for half a century: where does the Sun (and other stars) get its energy?

The catalytic structure is what gives CNO its distinctive properties. Because the fusion happens via proton capture on a heavy nucleus rather than between two light protons, the energetics depend on the Coulomb barrier of carbon (Z=6), nitrogen (Z=7), or oxygen (Z=8) rather than the much smaller barrier of hydrogen (Z=1). The resulting reaction rates are much more sensitive to temperature, much smaller in magnitude per particle pair, but capable of being driven to very high luminosities at high enough temperatures.

The cycle's six reactions

The main CNO cycle (sometimes labeled CNO-I) consists of these six reactions:

(1) ¹²C + p → ¹³N + γ                  (S₁₇ ≈ 1.4 keV·b at solar)
(2) ¹³N → ¹³C + e⁺ + ν_e               (β⁺, t½ = 9.97 min)
(3) ¹³C + p → ¹⁴N + γ                  (S₁₇ ≈ 7.6 keV·b)
(4) ¹⁴N + p → ¹⁵O + γ                  (S₁₇ ≈ 1.7 keV·b — the bottleneck)
(5) ¹⁵O → ¹⁵N + e⁺ + ν_e               (β⁺, t½ = 122.2 s)
(6) ¹⁵N + p → ¹²C + ⁴He                (S₁₇ ≈ 67 MeV·b — explosive)

Adding the six steps cancels the catalysts: ¹²C appears on the left of (1) and the right of (6), so it returns; ¹³N, ¹³C, ¹⁴N, ¹⁵O, and ¹⁵N are all created and destroyed in equal numbers within the cycle. The four protons consumed in (1), (3), (4), and (6) become one ⁴He emitted in (6) and two positrons + two neutrinos from (2) and (5).

The S-factors give a sense of the cross-sections at the energies relevant to stellar interiors. The ¹⁵N(p,α)¹²C reaction (step 6) is by far the fastest because it is exothermic by 4.97 MeV and proceeds through a broad resonance. The ¹⁴N(p,γ)¹⁵O reaction (step 4) is the slowest because the resulting compound nucleus has no convenient resonance at the Gamow peak energy. As a result, ¹⁴N accumulates: nearly every catalyst nucleus in the burning region eventually finds itself stuck as ¹⁴N waiting for the next proton capture.

The ¹⁴N bottleneck and equilibrium abundances

The slowness of step 4 has two important consequences. First, it sets the rate of the entire cycle: at any temperature relevant to stellar burning, the bottleneck step is rate-limiting, and the energy production rate scales with the ¹⁴N(p,γ)¹⁵O cross-section. The cycle equilibrates so that the fluxes through every step are equal, which means slower steps must hold higher abundances of their reactants.

Second, the equilibrium isotopic abundances are dramatically skewed. Detailed balance gives

X(¹⁴N) / X(¹²C) ≈ ⟨σv⟩(¹²C+p) / ⟨σv⟩(¹⁴N+p) ≈ 30–100

at typical CNO-burning temperatures. So ¹⁴N becomes the most abundant CNO isotope by mass, much more so than its primordial abundance suggests. The carbon abundance drops, the oxygen abundance drops slightly, and the total CNO mass is conserved (since the cycle is catalytic) but redistributed almost entirely into ¹⁴N.

This abundance signature is observed unambiguously in the surface layers of stars that have undergone first dredge-up after the main sequence. Convective penetration into the former core brings CNO-cycled material to the surface, and spectroscopy reveals the predicted N-enhancement and C-depletion. Globular cluster red giants, asymptotic-giant-branch stars, and post-mass-transfer binaries all show this fingerprint.

Why ε ∝ T¹⁵ to T²⁰

The energy generation rate of any thermonuclear reaction depends on the temperature through two factors: the thermal kinetic-energy distribution of the reactants (a Maxwell-Boltzmann factor) and the tunneling probability through the Coulomb barrier (the Gamow factor). The product of these two has a sharp peak — the Gamow peak — at an energy

E_G = (π² Z₁² Z₂² e⁴ μ / 2 ħ²)^(1/3) (k_B T)^(2/3)

where μ is the reduced mass and Z₁ Z₂ is the product of the reactant charges. Higher Coulomb barriers mean higher Gamow peak energies, which means a steeper temperature dependence. For pp-chain reactions (Z₁ Z₂ = 1), the Gamow peak at solar temperatures sits around 6 keV. For ¹⁴N(p,γ)¹⁵O (Z₁ Z₂ = 7), it sits around 27 keV.

The reaction rate scales with temperature as approximately

⟨σv⟩ ∝ T⁻²/³ exp[-3(τ/2)^(1/3)]
where τ = (3 E_G / k_B T)

For pp at T = 1.5 × 10⁷ K this gives a local power-law slope d ln ε / d ln T ≈ 4. For CNO at the same temperature, the slope is ≈ 18. The bottom line: in the range 1.5–3 × 10⁷ K relevant to most CNO-burning stellar cores, the rate scales between T¹⁵ and T²⁰. Higher temperatures push toward the larger exponent; lower temperatures soften it slightly.

This extreme sensitivity creates a positive feedback loop. A small increase in core temperature gives a much larger increase in luminosity, which the star must shed by expanding (to lower the temperature) or by transporting energy faster outward. CNO-burning cores cannot transport their luminosity by radiation alone — the radiative gradient becomes super-adiabatic — so they convect. Convective cores are therefore the universal signature of CNO-dominant burning and a key structural difference between massive stars (M > 1.3 M☉) and low-mass stars.

Where pp gives way to CNO

The two cycles operate simultaneously in any hydrogen-burning star, but only one dominates. The crossover temperature, where ε_pp = ε_CNO, depends on metallicity (because CNO requires existing catalysts) and slightly on density:

Metallicity	Crossover T (K)	Crossover mass (M☉)	CNO catalyst	Sun-like-luminosity equivalent
Z = 0.02 (solar)	1.8 × 10⁷	~1.3	0.8% by mass	~3 L☉
Z = 0.014 (Asplund 2009)	1.85 × 10⁷	~1.35	0.6% by mass	~3.5 L☉
Z = 0.001 (halo stars)	2.4 × 10⁷	~1.6	0.04% by mass	~10 L☉
Z = 10⁻⁴ (very low Z)	2.8 × 10⁷	~2.0	0.004% by mass	~30 L☉
Z = 10⁻⁶ (extreme halo)	3.5 × 10⁷	~3.5	0.00004% by mass	~150 L☉
Z = 0 (Population III)	see notes	see notes	none initially	see notes

The Population III row is distinctive: with no CNO catalysts to start, primordial stars must run on pp until contraction-driven heating produces enough triple-alpha-cycle ¹²C to bootstrap CNO. This typically requires central temperatures above 10⁸ K for a brief moment of triple-alpha burning, after which CNO ignites at ~3 × 10⁷ K once even trace carbon is present. The transition is rapid (<10⁴ years) and means that no Population III star spends a substantial fraction of its life on pure pp burning at high luminosity.

For solar-metallicity stars, the crossover at ~1.3 M☉ corresponds neatly to a structural transition: stars below 1.2 M☉ have radiative cores and convective envelopes; stars above 1.3 M☉ have convective cores and radiative envelopes. The Sun, at 1 M☉, falls cleanly into the radiative-core / pp-dominant regime, with CNO at the 0.8% level acting essentially as a perturbation.

Worked numerical example: CNO rate at 1.5 × 10⁷ K vs 2 × 10⁷ K

The CNO rate exponent at T ≈ 1.5–2 × 10⁷ K is approximately

n_CNO = d ln ε / d ln T ≈ 18

If we increase the temperature from T₁ = 1.5 × 10⁷ K to T₂ = 2.0 × 10⁷ K — a factor of T₂/T₁ = 4/3 — then

ε_CNO(T₂) / ε_CNO(T₁) = (T₂/T₁)^n = (4/3)^18 ≈ 178

So a 33% temperature increase boosts CNO output by a factor of nearly 200. Compare with the pp chain at the same temperatures: with n_pp ≈ 4,

ε_pp(T₂) / ε_pp(T₁) = (4/3)^4 ≈ 3.16

A 33% temperature increase gives only a factor of 3 in pp rate. So between 1.5 × 10⁷ K (Sun's centre) and 2 × 10⁷ K (a ~1.5 M☉ star's centre), CNO burning grows by 178/3.16 ≈ 56× relative to pp. That ratio explains why the crossover is so sharp.

Concrete check on the Sun: the Sun's central temperature is ~1.567 × 10⁷ K. At this temperature, the CNO contribution to total energy generation is computed in standard solar models at about 0.8% of the total — meaning ~3.1 × 10³¹ erg/s from CNO out of the total 3.85 × 10³³ erg/s. Borexino's measurement of CNO neutrinos in 2020 gave a flux of (7.0 +3.0/−2.0) × 10⁸ /cm²/s at Earth, consistent with that 0.8% figure to within ~30%. The detection ended a 70-year theoretical wait for direct experimental confirmation that the Sun runs the CNO cycle.

Convective cores in CNO-burning stars

The Schwarzschild criterion for convective stability is

(d ln T / d ln P)_radiative < (d ln T / d ln P)_adiabatic = 1 - 1/γ ≈ 0.4

The radiative gradient depends on luminosity and opacity. In CNO-burning cores, the local luminosity is enormous because the energy generation is so concentrated, so the radiative gradient is steep and the inequality is reversed: the gradient must be set adiabatically by convection.

The size of the convective core depends on stellar mass. For 1.5 M☉ it is ~10% of the stellar mass; for 5 M☉ it is ~30%; for 20 M☉ it is ~50%. The convective core is well mixed, so its hydrogen and helium abundances are uniform, and the entire convective region exhausts hydrogen at the same time. Once core hydrogen runs out, the convective core shuts off and the star transitions to shell hydrogen burning, with rapid implications for stellar structure (see the Hertzsprung-gap article).

Convective overshooting at the core boundary mixes additional hydrogen-rich material into the burning region from above. This extends the main-sequence lifetime by 10–25% and fundamentally affects predictions for cluster ages, supernova rates, and chemical yields. Stellar evolution codes parametrize overshooting with a single parameter (typically α_ov ≈ 0.1–0.3 in pressure-scale-height units), and the value is tuned against eclipsing binary observations and asteroseismic data.

Variants and extensions

• CNO-II, III, IV. Side branches that involve alternative paths through ¹⁵N, ¹⁶O, ¹⁷O, ¹⁷F, and ¹⁸F. CNO-II opens up at slightly higher temperatures via ¹⁵N(p,γ)¹⁶O instead of ¹⁵N(p,α)¹²C; CNO-III and IV operate via ¹⁷O(p,γ)¹⁸F and similar reactions. Together they represent <1% of cycle flux at solar core conditions but become important in massive-star envelopes.
• Hot CNO. At T > 5 × 10⁷ K (nova explosions, X-ray bursts), proton capture rates exceed beta-decay rates of the unstable intermediates ¹³N and ¹⁵O. The cycle then proceeds via ¹³N(p,γ)¹⁴O instead of waiting for ¹³N to decay. The bottleneck shifts to the beta-decay timescales of ¹⁴O (70.6 s) and ¹⁵O (122 s), making the cycle's energy generation rate temperature-independent in the hot CNO regime.
• Rapid proton (rp) process. At still higher temperatures and densities (T > 10⁸ K, X-ray bursts on accreting neutron stars), proton capture is so fast that the reaction sequence breaks out of CNO entirely and proceeds along the proton drip line, building elements up to A ~ 100 in seconds.
• Ne-Na cycle. A higher-mass extension running on neon and sodium catalysts (²⁰Ne, ²¹Ne, ²²Na, ²²Ne, ²³Na), operating in stars heavier than ~5 M☉ at temperatures above ~3 × 10⁷ K. The bottleneck step ²²Ne(p,γ)²³Na produces a sodium-22 isotope whose decay creates a characteristic gamma-ray line.
• Mg-Al cycle. Even higher-mass extension on Mg and Al isotopes, operating at T > 5 × 10⁷ K in massive stars. Produces the long-lived radioisotope ²⁶Al, whose 1.809 MeV decay line is detected by gamma-ray telescopes such as INTEGRAL across the Galactic plane.

Where the CNO cycle shows up

• Massive-star main sequences. Every star above ~1.3 M☉ is CNO-dominated. This includes A-type stars like Vega (2.1 M☉) and Sirius (2.06 M☉), B-type stars like Spica (~10 M☉), and O-type stars like the most massive observed binaries (R136a1 at ~250 M☉). All have convective cores and CNO-driven energy generation.
• Globular cluster turnoff stars. Although globular clusters have low-mass turnoffs (~0.85 M☉) where pp would normally dominate, the CNO contribution is still measurable because of slightly elevated central temperatures from low metallicity. Spectroscopic anti-correlations of N vs. C, Na vs. O, Al vs. Mg in stars within a single cluster reveal CNO-cycled material that has been processed and mixed.
• Nova explosions. A classical nova is a thermonuclear runaway on the surface of a white dwarf accreting hydrogen-rich material from a companion. The hot CNO cycle drives the runaway. Observed nova ejecta show enormous nitrogen and oxygen enhancements, in line with CNO bottleneck predictions. The 1670 outburst of CK Vul, the 1934 outburst of DQ Her, and modern recurrent novae like RS Oph all show this signature.
• Borexino CNO neutrino detection (2020). Borexino at Gran Sasso reported the first direct detection of solar CNO neutrinos in November 2020 with statistical significance >5σ. The measured flux was (7.0 +3.0/−2.0) × 10⁸ /cm²/s, consistent with the standard solar model prediction. This was the last unobserved component of solar nuclear burning to be detected and provides an independent constraint on the Sun's central metallicity (the "solar abundance problem").
• Galactic ²⁶Al gamma-ray emission. The 1.809 MeV gamma-ray line from ²⁶Al decay is observed across the Milky Way's plane by INTEGRAL/SPI. Total Galactic ²⁶Al mass is estimated at ~2 M☉, produced predominantly in the Mg-Al cycle of massive star cores and Wolf-Rayet stars. The line traces ongoing massive-star nucleosynthesis in the Galaxy.

Common pitfalls

• Confusing catalyst with reactant. The CNO nuclei are catalysts, not consumed. Their total mass in the burning region is conserved over a cycle. Hydrogen is the only true reactant; helium is the only true product.
• Forgetting the metallicity dependence. In Population II and III environments, the CNO cycle starts off subdominant simply because there is no carbon to catalyse it. Stars must first run pp until self-produced carbon enables CNO. Treating CNO crossover at 1.3 M☉ as universal is wrong for low-Z populations.
• Underestimating the temperature exponent. Quoting T¹⁰ or even T¹² is too soft. At realistic CNO core temperatures the exponent is 15–20. The structural consequence — convective cores — only follows from the steeper value.
• Neglecting ¹⁴N accumulation. Surface composition of CNO-processed material is ¹⁴N-rich and ¹²C-poor. Misinterpreting this signature as elemental abundance evidence for primordial nitrogen production has led to historical confusion. Always check whether the spectroscopic target has undergone dredge-up or rotational mixing.
• Treating CNO and pp as mutually exclusive. Both run simultaneously in any hydrogen-burning star; the question is which dominates. The Sun runs both, with CNO at the 0.8% level. Stating "the Sun uses pp" without qualification ignores CNO-cycle physics that shows up in solar neutrino measurements.

Summary

The CNO cycle is the dominant hydrogen-burning mechanism in stars more massive than ~1.3 M☉ at solar metallicity. It uses carbon, nitrogen, and oxygen as catalysts and proceeds through six reactions whose slowest step — ¹⁴N(p,γ)¹⁵O — sets the rate of the entire cycle and causes nitrogen-14 to accumulate to abundances 30–100× initial. The energy generation rate scales as T¹⁵–²⁰, making the cycle far more thermally sensitive than the pp chain and forcing CNO-burning cores to be convective. The Sun runs CNO at the 0.8% level — confirmed directly by Borexino's 2020 detection of CNO neutrinos — but in a 5 M☉ B-type star CNO produces essentially 100% of the luminosity. The cycle's variants (hot CNO, Ne-Na, Mg-Al, rp-process) extend its physics into novae, X-ray bursts, and the most massive stellar interiors, leaving fingerprints on cosmic chemical evolution that span the full mass range of the stellar population.