678 words
3 minutes
Spin-Orbit Interaction as a 2×2 Spinor Operator

Spin-orbit coupling is one of the contributions to the fine structure of hydrogen, and it keeps showing up after atomic physics too, in condensed matter for instance (the Rashba effect, topological insulators). The interaction Hamiltonian is

WS−O=e28πϵ0me2c2R3 L⃗⋅S⃗W_{S-O} = \frac{e^2}{8\pi\epsilon_0 m_e^2 c^2 R^3}\, \vec{L} \cdot \vec{S}

where RR is the (operator) magnitude of the electron’s position, L⃗\vec{L} is the orbital angular momentum and S⃗\vec{S} is the spin. In this post we write it out as an explicit 2×22 \times 2 matrix operator acting on spinors, in the basis of SzS_z eigenstates ∣+⟩,∣−⟩|{+}\rangle, |{-}\rangle. Once it is in that form, you can more or less read off what the interaction does.


Decomposing L⃗⋅S⃗\vec{L} \cdot \vec{S} with ladder operators#

We use the usual raising and lowering operators,

L±≡Lx±iLy,S±≡Sx±iSyL_\pm \equiv L_x \pm i L_y, \qquad S_\pm \equiv S_x \pm i S_y

and the claim is that

L⃗⋅S⃗=12(S+L−+S−L+)+LzSz\vec{L} \cdot \vec{S} = \frac{1}{2}(S_+ L_- + S_- L_+) + L_z S_z

To prove it we start from the right-hand side and work back to L⃗⋅S⃗\vec L\cdot\vec S. Writing the ladder operators in terms of their components,

12(S+L−+S−L+)+LzSz\frac{1}{2}(S_+ L_- + S_- L_+) + L_z S_z=12((Sx+iSy)(Lx−iLy)+(Sx−iSy)(Lx+iLy))+LzSz= \frac{1}{2}\Big((S_x + iS_y)(L_x - iL_y) + (S_x - iS_y)(L_x + iL_y)\Big) + L_z S_z

and multiplying out the brackets,

=12(SxLx−iSxLy+iSyLx+SyLy+SxLx+iSxLy−iSyLx+SyLy)+LzSz= \frac{1}{2}\Big(S_x L_x - iS_x L_y + iS_y L_x + S_y L_y + S_x L_x + iS_x L_y - iS_y L_x + S_y L_y\Big) + L_z S_z

The four imaginary cross terms cancel in pairs, and the real terms each appear twice,

=12(2SxLx+2SyLy)+LzSz=SxLx+SyLy+SzLz=S⃗⋅L⃗■= \frac{1}{2}\Big(2S_x L_x + 2S_y L_y\Big) + L_z S_z = S_x L_x + S_y L_y + S_z L_z = \vec{S} \cdot \vec{L} \quad \blacksquare

The reason to bother with this form is that S±S_\pm act very simply on spin-12\frac{1}{2} states, while L±L_\pm act only on the orbital part. Spin and orbital operators act on different spaces, so they commute, and we can deal with the two pieces separately.


The explicit 2×22 \times 2 matrix#

For a spin-12\frac{1}{2} electron, Sx=ℏ2σxS_x = \frac{\hbar}{2}\sigma_x and Sy=ℏ2σyS_y = \frac{\hbar}{2}\sigma_y. The ladder operators are then

S+=Sx+iSy=ℏ2(0110)+i⋅ℏ2(0−ii0)=ℏ2(0200)=ℏ(0100)S_+ = S_x + iS_y = \frac{\hbar}{2}\begin{pmatrix}0 & 1\\1 & 0\end{pmatrix} + i\cdot\frac{\hbar}{2}\begin{pmatrix}0 & -i\\i & 0\end{pmatrix} = \frac{\hbar}{2}\begin{pmatrix}0 & 2\\0 & 0\end{pmatrix} = \hbar\begin{pmatrix}0 & 1\\0 & 0\end{pmatrix}S−=Sx−iSy=ℏ(0010)S_- = S_x - iS_y = \hbar\begin{pmatrix}0 & 0\\1 & 0\end{pmatrix}

and Sz=ℏ2(100−1)S_z = \frac{\hbar}{2}\begin{pmatrix}1 & 0\\0 & -1\end{pmatrix}. So S+S_+ only has an upper-right entry and S−S_- only a lower-left one, which is what raising and lowering should look like in this basis.

Now we substitute these into the ladder decomposition. Each of the three terms becomes a 2×22\times 2 matrix whose entries are orbital operators, i.e. they still act on the spatial wavefunction,

12S+L−=ℏ2(0L−00),12S−L+=ℏ2(00L+0),LzSz=ℏ2(Lz00−Lz)\frac{1}{2}S_+ L_- = \frac{\hbar}{2}\begin{pmatrix}0 & L_-\\0 & 0\end{pmatrix}, \qquad \frac{1}{2}S_- L_+ = \frac{\hbar}{2}\begin{pmatrix}0 & 0\\L_+ & 0\end{pmatrix}, \qquad L_z S_z = \frac{\hbar}{2}\begin{pmatrix}L_z & 0\\0 & -L_z\end{pmatrix}

Adding them up,

L⃗⋅S⃗=ℏ2(LzL−L+−Lz)\vec{L} \cdot \vec{S} = \frac{\hbar}{2}\begin{pmatrix}L_z & L_-\\L_+ & -L_z\end{pmatrix}

and putting this back into WS−OW_{S-O} gives

WS−O=e28πϵ0me2c2 ℏ2(Lz/R3L−/R3L+/R3−Lz/R3)\boxed{W_{S-O} = \frac{e^2}{8\pi\epsilon_0 m_e^2 c^2}\,\frac{\hbar}{2} \begin{pmatrix}L_z/R^3 & L_-/R^3\\L_+/R^3 & -L_z/R^3\end{pmatrix}}

Reading the matrix#

I think the matrix form is worth looking at for a minute, since it shows directly what spin-orbit coupling does.

  • Diagonal entries (Lz/R3L_z/R^3 and −Lz/R3-L_z/R^3). These connect spin-up to spin-up and spin-down to spin-down. They give an energy contribution proportional to mlmsm_l m_s, i.e. the projection of L⃗\vec{L} along zz, with a sign depending on whether the spin is up or down. On their own they are not the physical splitting. The eigenvalues of the full matrix (diagonal and off-diagonal together) depend only on jj and not on mjm_j, as rotational invariance requires. What spin-orbit coupling actually splits is the j=l+12j = l + \tfrac{1}{2} multiplet from the j=l−12j = l - \tfrac{1}{2} multiplet, through L⃗⋅S⃗=ℏ22[j(j+1)−l(l+1)−34]\vec{L}\cdot\vec{S} = \tfrac{\hbar^2}{2}\left[j(j+1) - l(l+1) - \tfrac{3}{4}\right].

  • Off-diagonal entries (L±/R3L_\pm/R^3). These flip the spin and at the same time change mlm_l by ∓1\mp 1, so that mj=ml+msm_j = m_l + m_s is conserved. They mix the spatial and spin degrees of freedom, and they are the reason mlm_l and msm_s separately are not good quantum numbers once spin-orbit coupling is switched on. Only jj and mjm_j are.

This is also why, for the fine structure of hydrogen, one works in the ∣n,l,j,mj⟩|n, l, j, m_j\rangle basis and not in ∣n,l,ml,ms⟩|n, l, m_l, m_s\rangle.

Spin-Orbit Interaction as a 2×2 Spinor Operator
https://rohankulkarni.me/posts/notes/spin-orbit-interaction/
Author
Rohan Kulkarni
Published at
2023-07-20
License
CC BY-NC-SA 4.0
Share Post
Comments