Magnetic Confinement
Source lecture(s): PC368 Lec 1
Intuition
A charged particle cannot cross a magnetic field line easily — it spirals around it with a Larmor radius of a millimetre or so in a fusion field. Perpendicular confinement is therefore free. Parallel confinement is the entire problem: along the field line the particle is unimpeded, and it leaves at thermal speed.
The history of magnetic confinement is a sequence of increasingly clever answers to "where does the field line go?"
Attempt 1: the straight solenoid
Uniform \(B\hat{z}\). Particles spiral tightly, and stream straight out the ends at \(\sim10^6\) m/s. Confinement time: microseconds. Fails.
Attempt 2: the magnetic mirror
Make the field stronger at both ends. Conservation of the magnetic moment \(\mu = mv_\perp^2/2B\) converts parallel motion into perpendicular as a particle enters the strong-field region, and if \(v_\perp\) uses up all the energy the particle turns around. This genuinely works — for some particles. Those with too little \(v_\perp\) escape through the loss cone, and because collisions continuously refill the loss cone, the plasma drains. Fails, but instructively — the same physics traps the Van Allen belts.
Attempt 3: bend it into a torus
No ends, no end losses. Except a purely toroidal field is non-uniform by geometry — \(B \propto 1/R\) — so it has both a gradient and curvature, and both drive charge-dependent drifts:
Ions drift up, electrons drift down. The resulting vertical electric field drives an \(\mathbf{E}\times\mathbf{B}\) drift that is outward for both species, and the entire plasma walks into the outer wall in milliseconds. Fails — and you can watch it fail in the drift orbit lab, curvature-drift mode.
Attempt 4: twist the field — tokamaks and stellarators
The cure is to make each field line spiral, so it spends half its time on top and half on the bottom. A particle following it samples both signs of the vertical drift, which then averages away. The twist is measured by the safety factor
— the number of toroidal turns per poloidal turn. Two ways to get the poloidal field:
- Tokamak. Drive a large toroidal current through the plasma itself; that current makes \(B_\theta\). Simple, axisymmetric, excellent confinement — but the current is a free energy source for kink instabilities, needs \(q > 1\) at the edge (Kruskal–Shafranov), and is fundamentally pulsed unless driven non-inductively. Disruptions — sudden loss of the whole current — are the defining engineering hazard.
- Stellarator. Build the twist into external coils instead. No plasma current, so no disruptions and inherently steady-state; the price is a fiendish 3-D coil geometry and historically worse confinement. Wendelstein 7-X showed that modern optimisation largely closes the gap.
The two devices worth knowing
ITER (Saint-Paul-lès-Durance, France) — the largest tokamak ever built, designed for \(Q \ge 10\): 500 MW of fusion power from 50 MW of plasma heating, the first burning plasma where alpha self-heating dominates. Originally budgeted near $4B, now above $25B; a seven-member international collaboration with additional participants including Thailand.
TT-1 (Thailand Tokamak 1) — the first tokamak in Thailand and in ASEAN, a refurbished HT-6M donated by ASIPP, China, installed at TINT in Ongkharak, Nakhon Nayok.
| Milestone | Date |
|---|---|
| Commissioning at ASIPP, China | May 2022 |
| Disassembly and transport | August 2022 |
| Assembly at TINT | November 2022 |
| Commissioning at TINT | January 2023 |
| Ownership transfer | April 2023 |
Its stated development path is three steps: a foundation phase reaching \(10^6\) K and building domestic engineering capability; a superconducting device reaching fusion-relevant \(10^8\) K within ten years; and, at thirty years, a Thai-designed power plant.
Common mistakes
- "The magnetic field holds the plasma like a container." It restricts motion across field lines only. Everything hard about confinement is about the parallel direction and about drifts that carry particles across lines anyway.
- Forgetting that the toroidal field is non-uniform by necessity. \(B \propto 1/R\) is forced by Ampère's law around the central column — you cannot design it away, only compensate for it.
- Thinking \(q\) is about safety margins in the engineering sense. It is a geometric winding number; the name is historical, though staying above \(q = 2\) really does keep you safe from the most violent kinks.
Related concepts
- Drift motions — why the naive torus fails
- Magnetic mirror · Loss cone — the open-ended branch
- MHD equilibrium · Grad–Shafranov (eq.)
- Kink instability — the price of driving current
- Lawson criterion — what confinement has to achieve
Knowledge graph position
Prerequisites: Drift motions, Larmor radius, adiabatic invariants. Leads to: MHD equilibrium, MHD stability, pinch equilibria.
Quiz
Q1 (conceptual). Why does bending a solenoid into a torus not, by itself, confine a plasma?
Answer
It removes end losses but introduces \(B \propto 1/R\), hence ∇B and curvature drifts. These are charge-dependent, so ions and electrons separate vertically; the resulting \(\mathbf{E}\) drives an \(\mathbf{E}\times\mathbf{B}\) drift that is radially outward for both species. The plasma is lost in milliseconds.
Q2 (computational). A tokamak has \(R = 6\) m, \(a = 2\) m, \(B_\phi = 5\) T and a plasma current giving \(B_\theta = 0.6\) T at the edge. What is the edge safety factor?
Answer
\(q = rB_\phi/(RB_\theta) = (2)(5)/((6)(0.6)) = 2.8\) — comfortably above the Kruskal–Shafranov limit of \(q = 1\), and above the \(q = 2\) working threshold.
Q3 (MCQ). The essential difference between a tokamak and a stellarator is:
- (a) the tokamak is toroidal and the stellarator is linear
- (b) the poloidal field comes from plasma current in one, external coils in the other
- (c) only the stellarator confines particles magnetically
- (d) the stellarator does not need a toroidal field
Answer
(b). Both are toroidal and both need helical twist; they differ only in where the poloidal component comes from — and every other difference (disruptions, steady-state operation, coil complexity) follows from that one choice.