Inside ITER: the machine and fusion gain

The parts of the machine

ITER is a tokamak. Its plasma, about $830\ \text{m}^3$ of deuterium and tritium at about $150$ million $^\circ\text{C}$, is held by magnetic fields inside a ring-shaped steel chamber.

• Magnets. $18$ D-shaped toroidal field coils make the field that runs around the ring (up to $11.8\ \text{T}$). Six ring-shaped poloidal field coils shape the plasma and hold it in place. The central solenoid, a $13\ \text{m}$ tall stack of six modules, works like the primary coil of a transformer: a changing current in it drives a current of up to $15\ \text{MA}$ in the plasma. All the magnets are superconducting and are cooled with helium to about $4\ \text{K}$.
• Vacuum vessel. A double-walled steel chamber, $19.4\ \text{m}$ wide, keeps the vacuum and is the first barrier that keeps tritium in.
• Blanket. $440$ modules line the inside of the vessel. They take the heat and the fast neutrons and shield the vessel and the magnets.
• Divertor. At the bottom, $54$ cassettes with tungsten targets catch the heat and the helium "ash" that flow out along the edge of the plasma. The heat flow there is up to about $10\text{–}20\ \text{MW}$ on each square metre, the highest in the machine.
• Heating. Beams of fast neutral atoms and two kinds of radio waves heat the plasma with a planned $50\ \text{MW}$.
• Cryostat. A vacuum chamber about $30\ \text{m}$ tall around everything insulates the cold magnets from the warm building.

Key idea

A fusion plasma loses energy all the time: hot particles escape across the magnetic field and the plasma radiates. To keep it hot, we heat it from outside. The fusion gain $Q$ compares the fusion power produced in the plasma with this heating power.

At $Q = 1$, called scientific breakeven, the plasma produces as much fusion power as it receives. No magnetic fusion experiment has reached it yet with real $\text{D–T}$ fuel: JET's record from 1997 was $16\ \text{MW}$ of fusion power from $24\ \text{MW}$ of heating, $Q \approx 0.67$. ITER's goal is $Q = 10$: $500\ \text{MW}$ of fusion power from $50\ \text{MW}$ of heating, in pulses of $400\text{–}600\ \text{s}$. In D–T fusion, $3.5\ \text{MeV}$ of the $17.6\ \text{MeV}$ goes to the helium nucleus, which is charged and stays in the plasma. At $Q = 10$ these helium nuclei heat the plasma about twice as much as the external heating does: a burning plasma.

Fusion gain

$$Q = \frac{P_\text{fusion}}{P_\text{heating}}$$

Gain is not electricity. $Q$ describes the plasma only. ITER has no turbines and will not produce any electricity. During a pulse the whole plant (heating systems, cooling water, cryogenics, magnets) draws far more electricity than the $50\ \text{MW}$ that reaches the plasma. According to ITER, it needs $110\ \text{MW}$ at the least and up to $620\ \text{MW}$ in peaks of $30\ \text{s}$; the cooling water and the cryogenic systems take about $80\ \%$ of it. A power plant would turn the heat into electricity with steam turbines, typically about one third of it. Even then, $500\ \text{MW}$ of heat would give only about $170\ \text{MW}$ of electricity: far less than ITER's peak demand of $620\ \text{MW}$.

Scientific gain is not commercial viability. A power plant must also produce more electricity than the whole plant uses, run for months instead of minutes, breed its own tritium, survive years of neutron damage and sell its electricity at a price people will pay. ITER is built to test the physics and the technology for that, not to show that fusion electricity is cheap.

Worked example

Fuel, blankets and radiation

Tritium fuel cycle. Only a small fraction of the fuel, about one per cent according to ITER, fuses on its way through the plasma. Cryopumps pump the exhaust out of the vessel, and the tritium plant separates the hydrogen isotopes and sends deuterium and tritium back to the fuelling system. Tritium is radioactive (half-life $12.3$ years, a weak beta emitter), so detritiation systems take it out of air and water.

Breeding. Tritium is very rare; today it comes mainly from heavy-water fission reactors. A power plant must make its own from lithium: $^{6}_{3}\text{Li} + \text{n} \to {}^{4}_{2}\text{He} + {}^{3}_{1}\text{H} + 4.8\ \text{MeV}$. In ITER, a few test blanket modules will try out ways to do this.

Radiation protection. The fast neutrons make the vessel and the blanket radioactive (activation), so parts inside the vessel are changed by remote-handling robots. A thick concrete bioshield surrounds the machine. The fuel leaves no long-lived high-level waste, and the reaction stops by itself if anything fails, but tritium, activated parts, the strong magnets and the heat are real hazards. ITER obtained the licence of a French nuclear installation in 2012.

Common misconception

"$Q = 10$ means that ITER gives ten times more electricity than it uses." $Q$ compares two powers in the plasma: the fusion power and the heating that reaches the plasma. ITER produces no electricity at all, and the whole plant draws far more from the grid than the $50\ \text{MW}$ that reaches the plasma.

A second mistake is "the plasma is $150$ million degrees, so the machine must be extremely hot everywhere". The plasma is hundreds of thousands of times less dense than air, and the magnetic field holds it away from the walls. A few metres away, the magnets are at about $4\ \text{K}$.

Summary

• ITER's superconducting magnets (toroidal field coils, poloidal field coils, central solenoid) hold and shape the plasma; the vacuum vessel, the blanket and the divertor surround it; beams and waves heat it.
• Fusion gain: $Q = P_\text{fusion} / P_\text{heating}$. $Q = 1$ is scientific breakeven; JET reached $0.67$ in 1997; ITER aims at $Q = 10$, $500\ \text{MW}$ from $50\ \text{MW}$.
• At $Q = 10$ the helium nuclei carry about $100\ \text{MW}$ and heat the plasma more than the external heating: a burning plasma.
• $Q$ is a plasma gain, not electricity. ITER produces no electricity, and a power plant needs much more than a high $Q$.
• Tritium is recycled in a closed loop and must later be bred from lithium; neutrons activate the structure, so ITER is a licensed nuclear installation.