Neutron Moderation

Why neutrons must be slowed

Every fission of U-235 releases two to three neutrons, and those neutrons are fast — they leave the splitting nucleus with a broad spectrum averaging about 2 MeV, moving near 10⁷ m/s. To keep a chain reaction going, each fission must, on average, trigger at least one more. The catch is that a fast neutron is a poor trigger: the U-235 fission cross section at 2 MeV is only around 1 barn, so a fast neutron usually leaks out of the core, scatters away, or is captured before it finds another fissile nucleus.

Slow the neutron down and the odds transform. At thermal energy (~0.025 eV) the U-235 fission cross section rises to roughly 580 barns — a factor of several hundred. The neutron lingers near each nucleus far longer (its de Broglie wavelength grows as it slows), giving the strong force time to act. Moderation is the deliberate process of harvesting that gain: surround the fuel with a moderator whose light nuclei absorb little but scatter a lot, and let the fast neutrons bounce their way down to thermal speeds.

Elastic scattering: billiard balls, not absorption

Moderation works almost entirely through elastic scattering — the neutron collides with a nucleus and bounces off, conserving kinetic energy and momentum, like one billiard ball striking another. No energy is absorbed by the nucleus; it is simply shared. The key physics is the mass mismatch.

For a head-on elastic collision between a neutron (mass 1) and a nucleus of mass number A, the neutron's energy after the collision can be anywhere between its full energy and a minimum fraction α of it, where

α = ( (A − 1) / (A + 1) )²

maximum fractional energy loss per collision = 1 − α = 4A / (A + 1)²

Against hydrogen (A = 1), α = 0, so a single head-on collision can stop the neutron dead — all its energy goes to the proton. Against carbon (A = 12), α ≈ 0.716, so even the best collision removes only ~28% of the energy, and the average removes much less. This is why light nuclei make the best moderators: a neutron and a proton have nearly equal mass, so the energy transfer per bounce is large.

Lethargy and the logarithmic energy decrement

Because each collision multiplies the energy by a random factor, neutron slowing-down is naturally described on a logarithmic scale. The average decrease in the natural log of energy per collision, ξ (the "average logarithmic energy decrement"), is a constant for a given nuclide and is independent of the neutron's energy:

ξ = 1 + ( α / (1 − α) ) · ln(α)        (exact)
ξ ≈ 2 / (A + 2/3)                       (good for A > ~10)

The number of collisions to slow a neutron from fission energy E₀ to thermal energy E_th is then simply

n = ln(E₀ / E_th) / ξ ≈ ln(2×10⁶ eV / 0.025 eV) / ξ ≈ 18.2 / ξ

Hydrogen, with ξ ≈ 1.0, thermalizes a neutron in about 18 collisions; graphite, with ξ ≈ 0.158, needs roughly 115. But collision count alone does not pick the winner — a moderator must also avoid eating the neutrons it is slowing.

Comparing moderators

A good moderator balances three demands: a large ξ (few collisions), a high macroscopic scattering cross section Σ_s (collisions happen quickly), and a tiny macroscopic absorption cross section Σ_a (neutrons survive the journey). The figure of merit that captures all three is the moderating ratio, ξΣ_s/Σ_a. The slowing-down power, ξΣ_s, tells you how fast a moderator slows neutrons; the moderating ratio tells you how cleanly it does so.

Moderator	A (scatterer)	ξ	Collisions to thermal	Moderating ratio ξΣ_s/Σ_a
Light water (H₂O)	1 (H)	1.000	~18	~62
Heavy water (D₂O)	2 (D)	0.725	~25	~6000
Helium	4	0.425	~43	~51
Beryllium	9	0.207	~88	~130
Graphite (C)	12	0.158	~115	~170

Light water slows neutrons in the fewest collisions of any practical moderator, but its hydrogen absorbs neutrons (capturing one to become deuterium), giving it the lowest moderating ratio of the bunch. That is why light-water reactors require enriched fuel (~3–5% U-235) to overcome the lost neutrons. Heavy water barely absorbs at all, so heavy-water reactors like CANDU can run on natural uranium (0.7% U-235). Graphite sits in between and was the moderator of the first reactor ever built, Fermi's Chicago Pile-1 in 1942.

Thermal neutrons and the Maxwell–Boltzmann floor

Slowing-down does not continue forever. Once a neutron's energy drops to the order of the thermal motion of the moderator atoms, it is as likely to gain energy in a collision as to lose it. The neutron reaches thermal equilibrium with the moderator and its speeds settle into a Maxwell–Boltzmann distribution at the moderator temperature. By convention the reference thermal neutron has the most probable speed at 20°C:

E_th = kT = (1.381×10⁻²³ J/K)(293.6 K) ≈ 0.0253 eV
v = √(2E/m_n) ≈ 2200 m/s

This is why the standard cross sections are quoted "at 2200 m/s." Raise the moderator temperature and the whole spectrum shifts upward — the neutrons get hotter, the average fission cross section changes, and the reactor's reactivity responds. This temperature feedback, together with Doppler broadening of the U-238 capture resonances, is one of the safety levers that lets a thermal reactor self-regulate.

Worked calculations

// Average logarithmic energy decrement and collisions to thermalize
function xi(A) {
  if (A === 1) return 1.0;                 // exact limit for hydrogen
  const alpha = Math.pow((A - 1) / (A + 1), 2);
  return 1 + (alpha / (1 - alpha)) * Math.log(alpha);
}

function collisionsToThermal(A, E0 = 2e6, Eth = 0.025) {
  return Math.log(E0 / Eth) / xi(A);
}

console.log(`H  : xi = ${xi(1).toFixed(3)},  collisions = ${collisionsToThermal(1).toFixed(0)}`);  // ~18
console.log(`D  : xi = ${xi(2).toFixed(3)},  collisions = ${collisionsToThermal(2).toFixed(0)}`);  // ~25
console.log(`C  : xi = ${xi(12).toFixed(3)}, collisions = ${collisionsToThermal(12).toFixed(0)}`); // ~115
console.log(`U  : xi = ${xi(238).toFixed(4)}, collisions = ${collisionsToThermal(238).toFixed(0)}`); // ~2200 — useless

// Max fractional energy loss in one collision
function maxLoss(A) { return 4 * A / Math.pow(A + 1, 2); }
console.log(`Max loss H : ${(maxLoss(1) * 100).toFixed(0)}%`);   // 100%
console.log(`Max loss C : ${(maxLoss(12) * 100).toFixed(0)}%`);  // 28%
console.log(`Max loss U : ${(maxLoss(238) * 100).toFixed(2)}%`); // ~1.7%

// Thermal neutron speed from energy
function neutronSpeed(E_eV) {
  const E = E_eV * 1.602e-19;        // joules
  const m = 1.675e-27;               // neutron mass, kg
  return Math.sqrt(2 * E / m);       // m/s
}
console.log(`0.025 eV -> ${neutronSpeed(0.025).toFixed(0)} m/s`); // ~2188
console.log(`2 MeV    -> ${(neutronSpeed(2e6) / 1e7).toFixed(2)}e7 m/s`); // ~1.96e7

// Why thermal beats fast: fission probability scales with cross section
const sigma_fast = 1;    // barns at ~2 MeV
const sigma_th = 580;    // barns at 0.025 eV
console.log(`Thermal fission ${(sigma_th / sigma_fast).toFixed(0)}x more likely per nucleus encountered`);

Design trade-offs and the void coefficient

Choosing a moderator shapes the entire reactor. Light water is cheap, dense, and serves as both moderator and coolant — but its neutron absorption forces fuel enrichment, and because moderator and coolant are the same fluid, losing coolant also removes moderation. That gives light-water reactors a negative void coefficient: if the water boils or drains, moderation collapses and the chain reaction shuts itself down. This is an inherent safety feature.

Graphite-moderated, water-cooled designs decouple the two roles. The Soviet RBMK at Chernobyl used solid graphite for moderation and flowing water only for cooling. When that water boiled, it stopped absorbing neutrons while the graphite kept moderating just fine, so reactivity rose — a positive void coefficient. Combined with operator error and a flawed control-rod design, that feedback drove the 1986 runaway. Heavy-water reactors (CANDU) and gas-cooled graphite reactors (the UK Magnox/AGR fleet) make different bargains: superb neutron economy and natural-uranium fueling, at the cost of expensive heavy water or large graphite cores.

Where moderation shows up

• Power reactors. Nearly all commercial reactors are thermal reactors: PWRs and BWRs (light water), CANDU (heavy water), and AGR/RBMK (graphite). Moderation is what makes their fuel economics work.
• Research reactors and neutron sources. Beamlines for neutron scattering rely on cold and thermal moderators (liquid hydrogen, room-temperature water) to tune the neutron wavelength for materials science.
• Fast reactors deliberately omit it. Breeders and fast reactors use no moderator and a liquid-metal coolant (sodium) precisely to keep neutrons fast, enabling breeding of plutonium from U-238.
• Boron neutron capture therapy & medicine. Beam-shaping assemblies moderate accelerator neutrons down to epithermal energies for tumor treatment.
• Criticality safety. Engineers must account for accidental moderation — a fissile solution or wet powder can go critical with far less material than the dry metal, because water moderates the neutrons.

Common misconceptions

• "The moderator absorbs the neutrons." No — a good moderator scatters and barely absorbs. Absorption is the enemy; the whole point of a high moderating ratio is to slow neutrons without losing them.
• "Slower neutrons carry more energy to cause fission." Backwards. Slower neutrons carry less kinetic energy, but they are far more likely to be captured by a fissile nucleus — the cross section is what skyrockets, not the impact energy.
• "Heavier moderators are better because they're denser." The mass that matters is the mass number of the scattering nucleus, not bulk density. A lighter nucleus removes more energy per collision; heavy nuclei (like U-238) barely slow neutrons at all.
• "Moderator and coolant are always separate." In light-water reactors they're the same water, which is exactly what gives the safe negative void coefficient. In graphite reactors they're separate, which is what allowed Chernobyl's positive void coefficient.
• "Thermal neutrons are at absolute rest." They are in thermal equilibrium with the moderator, moving at ~2200 m/s at room temperature with a Maxwell–Boltzmann spread — fast by everyday standards, just slow compared to fission neutrons.