39
Part One · Quantum mechanics

Angular momentum

Turn a neutron through a full circle, 360 degrees, and it does not come back to itself. Its wavefunction has changed sign. Turn it a second full circle and it is back. This was measured in 1975, by splitting neutron beams in a crystal interferometer and turning one half in a magnetic field. The possibility follows from three commutators and nothing else, and this chapter derives it.

On the plate: an arrow turning, and beside it, the sign of a spin-½ state turned with it.

Everything from three commutators

Forget orbits and spherical harmonics. Define angular momentum as any three Hermitian operators obeying [Jx,Jy]=iℏJz[J_x, J_y] = i\hbar J_z and its cyclic partners (Chapter 29). Then J2J^2 commutes with each JiJ_i, so label states by J2J^2 and JzJ_z: ∣j,m⟩|j, m\rangle. Build the ladder operators J±=Jx±iJyJ_\pm = J_x \pm iJ_y. The commutators give [Jz,J±]=±ℏJ±[J_z, J_\pm] = \pm\hbar J_\pm, so J±J_\pm raise or lower mm by one, just like a†a^\dagger and aa in Chapter 36.

But Jz2≤J2J_z^2 \le J^2, so the ladder has a top m=jm = j and a bottom m=−jm = -j, and you get from one to the other in whole steps. So 2j2j is a whole number. That allows j=0,12,1,32,…j = 0, \tfrac12, 1, \tfrac32, \dots: half-integers are not forbidden. Choose j=j = 1.5 to see the 2j+12j + 1 rungs. Established

DeeperJ−J+=J2−Jz2−ℏJzJ_-J_+ = J^2 - J_z^2 - \hbar J_z. Applied to the top state, where J+∣j,mmax⁡⟩=0J_+|j,m_{\max}\rangle = 0, this gives the eigenvalue ℏ2mmax⁡(mmax⁡+1)\hbar^2 m_{\max}(m_{\max}+1). Calling mmax⁡=jm_{\max} = j gives J2=ℏ2j(j+1)J^2 = \hbar^2 j(j+1); the bottom gives mmin⁡=−jm_{\min} = -j.

How hard each step is

The size of each step is fixed by the algebra:

J±∣j,m⟩=ℏj(j+1)−m(m±1)  ∣j,m±1⟩J_\pm|j, m\rangle = \hbar\sqrt{j(j+1) - m(m\pm1)}\;|j, m\pm1\rangle

The plate labels each rung with the factor for stepping up. Steps are largest in the middle of the ladder and vanish off the ends. From these numbers you can write down the matrices of JxJ_x and JyJ_y for any jj. For j=12j = \tfrac12 they are the Pauli matrices of Chapter 8, times ℏ/2\hbar/2.

The vector model

A state ∣j,m⟩|j, m\rangle has a definite length ℏj(j+1)\hbar\sqrt{j(j+1)} and a definite zz-component mℏm\hbar, but its xx and yy components are uncertain. Picture it as a vector somewhere on a cone around the zz-axis. Choose m=m = 1.5 for your jj.

The vector is longer than its largest projection, since j(j+1)>j\sqrt{j(j+1)} > j, so it can never point straight along zz. It cannot, because that would make JxJ_x and JyJ_y both exactly zero, and they do not commute. The cone is a picture, not a mechanism; there is no hidden vector going round.

Rotations and the minus sign

Angular momentum generates rotations (Chapter 29): turning a state by angle θ\theta about zz is R(θ)=e−iθJz/ℏR(\theta) = e^{-i\theta J_z/\hbar}, which multiplies ∣j,m⟩|j, m\rangle by e−imθe^{-im\theta}. For whole-number mm, a turn of 2π2\pi gives 1. For half-integer mm, it gives e∓iπ=−1e^{\mp i\pi} = -1.

Turn through 360°. The arrow is back after 360°. The spin-½ amplitude, drawn as a pointer turning at half the speed, is reversed, and needs 720° to return. A sign on its own is unobservable, but a sign relative to an unrotated partner is not.

Seeing the sign

In 1975 two groups, Samuel Werner’s in Missouri and Helmut Rauch’s in Vienna, split a neutron beam in two inside a single silicon crystal, passed one half through a magnetic field that turned its spin by a controllable angle, and recombined them. The interference pattern repeats only every 720° of turning, not every 360°; the measured periods agreed with 720° to within a few percent. The plate shows the expected intensity, I∝1+cos⁡(θ/2)I \propto 1 + \cos(\theta/2).

So the minus sign is physical. It is how fermions, all particles with half-integer spin, differ from bosons, and Part Five will show it is linked to the Pauli exclusion principle. Established

Two descriptions of turning

Ordinary rotations in space form the group SO(3) (Chapter 20). The 2×22\times2 unitary matrices with determinant one form SU(2). Every rotation corresponds to two SU(2) matrices, UU and −U-U. Integer spins see only SO(3). Half-integer spins see all of SU(2), the minus sign included. The quantum world turned out to use the larger group.

DeeperSU(2) is topologically a three-sphere, which is simply connected; SO(3) is the three-sphere with opposite points identified, so a loop of rotations through 2π2\pi cannot be shrunk to nothing, but one through 4π4\pi can. Dirac’s belt trick shows the same fact with a ribbon.

What you now have

The full spectrum of angular momentum from its commutators, half-integers included, ladder factors, the vector model, rotation operators, the 720° spinor and the neutron experiments that saw it, and SU(2) as the double cover of rotations.

· · ·

Three questions to carry with you.

Why must 2j be a whole number?tap
The ladder runs from m = −j to m = +j in whole steps.
Why can’t the angular momentum point exactly along z?tap
That would make J_x and J_y both zero, impossible for non-commuting operators; and √(j(j+1)) exceeds j.
What did the 1975 neutron experiments show?tap
That a spin-½ state changes sign under a 360° rotation and needs 720° to return.

And one to work out. For j=32j = \tfrac32, J+∣32,12⟩=c ℏ ∣32,32⟩J_+|{\tfrac32}, \tfrac12\rangle = c\,\hbar\,|{\tfrac32}, \tfrac32\rangle. What is cc?

c =

Sources: J. J. Sakurai and J. Napolitano, Modern Quantum Mechanics, ch. 3. S. A. Werner, R. Colella, A. W. Overhauser and C. F. Eagen, Phys. Rev. Lett. 35, 1053 (1975). H. Rauch et al., Phys. Lett. A 54, 425 (1975). Checked by code: [Jx, Jy] = iJz for j = ½ to 2; J² = j(j+1); R(2π) = −1 for half-integer j, +1 for integer.

Next · Chapter 40
Spin-1/2 and Stern–Gerlach

Silver atoms through a magnet split into exactly two beams. Not a smear. Two.

Turn the page