Magnetic fields and the Zeeman effect
Wind a long, thin coil of wire and run a current through it. The magnetic field is all inside; outside, there is none at all. Send electrons around the outside, never through, and recombine them. Their interference pattern shifts as you change the current. The electrons respond to a magnetic field they never touch. Yakir Aharonov and David Bohm predicted this in 1959, and Akira Tonomura’s team confirmed it beyond doubt in 1986.
On the plate: two electron paths passing either side of a sealed magnetic region, and the fringes where they meet.
How a charge feels a magnetic field
In Chapter 15 the magnetic field was written as , with the vector potential. Quantum mechanics brings in the field through , not , by the rule called minimal coupling:
is not unique: adding the gradient of any function leaves unchanged. The Schrödinger equation stays the same if the wavefunction changes phase at the same time, . This gauge invariance looks like bookkeeping here. In Part Five it becomes the principle from which the forces of the Standard Model are built. Established
Landau levels
A classical electron in a uniform field circles at the cyclotron frequency . Quantum mechanically, the motion in the plane is a harmonic oscillator (Chapter 35), so the energies are
These are Landau levels. Each is enormously degenerate: one state for every flux quantum passing through the sample, about per square metre per tesla. Set the field to 10 T. The spacing grows with and each level holds more states. When a two-dimensional sheet of electrons fills a whole number of these levels, its Hall resistance locks to with astonishing precision: the quantum Hall effect, which Part Four takes up.
Splitting spectral lines
In 1896 Pieter Zeeman put a sodium flame between the poles of a magnet and saw its yellow lines broaden. A magnetic moment in a field has energy , so each level shifts by , with the Bohr magneton eV/T and the Landé factors of Chapter 41.
Raise the field to 1.0 T. Because the upper and lower levels have different , the D₁ line splits into four components and D₂ into six. This pattern was called anomalous, because the simple three-line splitting that classical theory predicted only appears for atoms with no net spin. Its explanation needed spin and . Established
When the field wins
The Zeeman formula assumes the field is weak compared with spin–orbit coupling, so and stay locked into . In a strong field, they decouple and each precesses about separately, and the energies become : the Paschen–Back regime.
The plate solves the sodium 3p levels exactly for any field. Stretch the range to 60 T. At low field, two groups fan out with slopes . Around 25 T, where matches the spin–orbit energy, the levels reorganize into groups set by .
A field it never touches
Back to the coil. Outside it, but is not zero; it circles the coil. An electron going round one side picks up a phase , and the difference between the two sides is , set by the flux inside, which neither path enters.
Change the enclosed flux to 0.25 flux quanta . The fringes slide by that fraction of a fringe. Tonomura’s group shielded a tiny ring magnet inside a superconductor, so no field could leak out and the electrons could not reach it, and still saw the shift. The vector potential, or more precisely the phase it produces around a loop, is physical. Established
What you now have
Minimal coupling and gauge invariance, Landau levels and their degeneracy, the Zeeman effect and the anomalous pattern, the Paschen–Back regime, and the Aharonov–Bohm effect.
· · ·
Three questions to carry with you.
And one to work out. An electron spin () flips when . With GHz/T, what frequency flips it at 0.35 T, in GHz? (This is how electron spin resonance spectrometers work.)
Sources: P. Zeeman, Phil. Mag. 43, 226 (1897). Y. Aharonov and D. Bohm, Phys. Rev. 115, 485 (1959). A. Tonomura et al., Phys. Rev. Lett. 56, 792 (1986). L. D. Landau, Z. Phys. 64, 629 (1930). CODATA 2022 μ_B. Checked by code: μ_B = 5.788×10⁻⁵ eV/T = 13.996 GHz/T; ħω_c = 0.1158 meV/T; eB/h = 2.42×10¹⁴ m⁻² T⁻¹; Na 3p spin–orbit ζ = 1.42 meV; 6×6 Hamiltonian agrees with the 2×2 block solution.