Chapter 4

Spin-Orbit Coupling

4.1 Addition of Angular Momenta

4.1.1 Alternate set of angular momentum operators
We consider two independent systems that each carries angular momenta (each could be
either orbital angular momentum or spin). Before the two systems are coupled, each system
is described by the angular momentum operators Jˆ1 and Jˆ2 and their eigenstates |j1 , m1 i
and |j2 , m2 i satisfying the conditions

Jˆi2 |ji , mi i = ~2 ji (ji + 1)|ji , mi i,

Jˆiz |ji , mi i = ~m
pi |ji , mi i, (4.1)
ˆ
Ji± |ji , mi i = ~ ji (ji + 1) mi (mi ± 1)|ji , mi ± 1i,
for i = 1, 2. When we consider the combination of these two systems, the basis of the
combined system is created by

|j1 , j2 ; m1 , m2 i = |j1 , m1 i ⌦ |j2 , m2 i, (4.2)

where ⌦ indicates the tensor product of two states that describe the two independent systems.
When the total angular momentum of the two systems is considered, it is described by the
sum of the two angular momentum operators Jˆ = Jˆ1 + Jˆ2 . Because the two systems are
independent, the operators Jˆ1 and Jˆ2 commute with each other, and the components of each
system satisfies the commutation relations of the angular momentum operator. It is easy
to verify that the components of the total angular momentum operator Jˆ also satisfies the
commutation relationship
[Jˆi , Jˆj ] = i~"ijk Jˆk . (4.3)
We now look for four independent operators that commute with each other that includes
the operators Jˆ2 and Jˆz , to provide an alternative description of the combined system as
compared to the four commuting operators {Jˆ12 , Jˆ1z , Jˆ22 , Jˆ2z }.

31
32 CHAPTER 4. SPIN-ORBIT COUPLING

Since Jˆ1 and Jˆ2 each commute with Jˆ12 and Jˆ22 , so does Jˆ. More specifically, we note that
Jˆ2 and Jˆz both commute with the operators Jˆ12 and Jˆ22 ;

[Jˆ2 , Jˆ12 ] = [Jˆ2 , Jˆ22 ] = 0,

(4.4)
[Jˆz , Jˆ12 ] = [Jˆz , Jˆ22 ] = 0.

The operators Jˆ1z and Jˆ2z commute with the operator Jˆz = Jˆ1z + Jˆ2z . However, since

Jˆ2 = Jˆ12 + Jˆ22 + 2Jˆ1 · Jˆ2 = Jˆ12 + Jˆ22 + 2Jˆ1z Jˆ2z + Jˆ1+ Jˆ2 + Jˆ1 Jˆ2+ , (4.5)

the operators Jˆ1z and Jˆ2z do not commute with Jˆ2 . So, the four operators {Jˆ12 , Jˆ22 , Jˆ2 , Jˆz }
commute with each other, and forms a natural basis with which to describe the total angular
momentum of the combined system.

4.1.2 Addition rule for the angular momentum states

Classically, angular momentum is a vector, so when two angular momenta L ~ 1 and L~ 2 are
added together, the magnitude of the resulting angular momentum can be between |L1 L ~ ~ 2|
when the two vectors are aligned in opposite directions, and |L ~1 + L
~ 2 | when the two vectors
are aligned in the same direction. Quantum mechanically, the “magnitude” of the angular
momentum operator Jˆ is roughly represented by its eigenvalue j, so when the two angular
momentum operators Jˆ1 with eigenvalue j1 and Jˆ2 with eigenvalue j2 are added (without
loss of generality, we assume j1 j2 ), one can speculate that the resulting operator Jˆ will
have its eigenvalue j between the values j1 j2 and j1 + j2 . It turns out that this speculation
is exactly true, and the addition rule for angular momentum can be stated as follows:
• When two systems each with angular momentum of eigenvalues j1 and j2 are added,
the angular momentum eigenvalue of the added system runs from |j1 j2 | to j1 + j2
in increments of 1, and
• For each resulting j value, mj runs from j to j in increments of 1, i.e., mj =
j, j + 1, . . . , j 1, j.
Although a formal proof is quite complicated, we provide some examples in the next section
to illustrate the results. We note that the total number of states in the two systems is
(2j1 + 1)(2j2 + 1), since the degeneracy (total number of mj states) are 2j1 + 1 for the first
system and 2j2 + 1 for the second system, respectively. In the description of the combined
system, the total number of mj states is 2j + 1 for each j value, so the total number of states
is given by
Pj1 +j2 Pj1 +j2 P j1 j2 1
N = i=j1 j2 (2i + 1) = i=0 (2i) 0 (2i) + (j1 + j2 ) (j1 j2 ) + 1
= (j1 + j2 + 1)(j1 + j2 ) (j1 j2 )(j1 j2 1) + 2j2 + 1 (4.6)
= (j1 + j2 )2 (j1 j2 )2 + 2j1 + 2j2 + 1 = (2j1 + 1)(2j2 + 1),

matching the total number of states in the description of individual systems.

4.1. ADDITION OF ANGULAR MOMENTA 33

4.1.3 Addition of two spins

In this section, we consider addition of angular momentum for two spin 1/2 particles. In
the uncoupled representation, the four quantum numbers are {Ŝ12 , Ŝ1z , Ŝ22 , Ŝ2z }, and the four
possible states are
1 1 1 1
|⇠i = | , , ± , ± i. (4.7)
2 2 2 2
These corresponds to the four basis spin states {| ""i, | "#i, | #"i, | ##i}.
In the coupled representation, the states are simultaneous eigenstates of the four operators
{Ŝ 2 , Ŝz , Ŝ12 , Ŝ22 }, where
Ŝ 2 = (Ŝ1 + Ŝ2 )2 = Ŝ12 + Ŝ22 + 2Ŝ1 · Ŝ2 ,
(4.8)
Ŝz = Ŝ1z + Ŝ2z .

First, we note that the values of s, eigenvalue of the Ŝ 2 operator, greater than 1 are not
possible. This is because that would require ms (eigenvalue of the operator Ŝz ) of value
larger than 1, but ms = m1s + m2s and both m1s and m2s can only take values of ±1/2.
Next, we note that s = 1 is possible, since ms = 1 is possible. Since ms = 1 can only happen
when m1s = m2s = +1/2, this state is not degenerate. From this argument, we note that
|s = 1, ms = 1i = | ""i.
Applying the total spin lowering operator Ŝ = Ŝ1 + Ŝ2 yields, using Eq. 4.1,
p
Ŝ |s = 1, ms = 1i = ~ 2|s = 1, ms = 0i q
= (Ŝ1z + Ŝ2z )| ""i = ~ 13
22
+ 11
22
(| "#i + | #"i) = | "#i + | #"i.
p (4.9)
So, we conclude that |s = 1, ms = 0i = (| "#i + | #"i)/ 2.
Applying the total spin lowering operator once again on this state gives
p
Ŝ |s = 1, ms = 0i = ~ 2|s = 1, ms = 1i
= p12 (Ŝ1z + Ŝ2z )(| "#i + | #"i) (4.10)
q p
~
= p
2
13
22
+ 11
22
(| ##i + | ##i) = ~ 2| ##i,
and therefore |s = 1, ms = 1i = | ##i, as expected. These three states form the spin
triplet states
|s = 1, ms = 1i = | ""i p
|s = 1, ms = 0i = (| "#i + | #"i)/ 2 (4.11)
|s = 1, ms = 1i = | ##i
The next possible s value is s = 0, accompanied by ms = 0 as the only possible ms value.
This fourth state is formed by a linear superposition of the four basis spin states {| ""i, | "#
i, | #"i, | ##i}, and is orthogonal to all three triplet states in Eq. 4.11. We see that this fourth
state cannot contain | ""i nor | ##i states, and therefore has to be a linear superposition
of only | "#i and | #"i. In order for such a superposition state to be orthogonal to the
|s = 1, ms = 0i state, we derive the final state, known as the singlet state, as
p
|s = 0, ms = 0i = (| "#i | #"i)/ 2. (4.12)
34 CHAPTER 4. SPIN-ORBIT COUPLING

4.1.4 Clebsch-Gordan coefficients

The basis state |j, mj i of the coupled system can be expressed in a linear superposition of the
uncoupled basis states |j1 , j2 ; m1 , m2 i, as by definition, basis states span the entire Hilbert
space for the quantum system. By noting that the identity operator is given by the sum of
the outer product of all basis states
j1 j2
X X
Iˆ = |j1 , j2 ; m1 , m2 ihj1 , j2 ; m1 , m2 |, (4.13)
m1 = j1 m2 = j2

we can express the basis states of the coupled system as

j1 j2
X X
|j, mj i = |j1 , j2 ; m1 , m2 ihj1 , j2 ; m1 , m2 |j, mj i. (4.14)
m1 = j1 m2 = j2

Inner product of two quantum states is a complex number, and the complex numbers
hj1 , j2 ; m1 , m2 |j, mj i are called Clebsch-Gordan coefficients.

4.2 Spin-Orbit Coupling

4.2.1 Orbiting spins and the term notation
For the case of a hydrogen atom, an electron with spin 1/2 orbits around the nucleus (proton),
itself with spin 1/2. For now, we ignore the nuclear spin, and focus on the interaction between
the electron spin and its orbit around the nucleus. This leads to fine structures in the atomic
spectrum. The interaction between the resulting energy levels with the nuclear spin is called
the hyperfine interaction: this interaction is typically several orders of magnitude smaller,
and we will consider that in later chapters.
The total angular momentum Jˆ under these conditions arise from two sources of angular
momenta, the spin Ŝ and the orbital angular momentum L̂: Jˆ = L̂ + Ŝ. The operators L̂
and Ŝ commute, as they are completely independent degrees of freedom. The eigenstates we
consider are simultaneous eigenstates of the operators {Jˆ2 , Jˆz , L̂2 , Ŝ 2 }, which all commute
with each other. The eigenvalue equations are given by
Jˆ2 |j, mj , l, si = ~2 j(j + 1)|j, mj , l, si
Jˆz |j, mj , l, si = ~mj |j, mj , l, si
(4.15)
L̂2 |j, mj , l, si = ~2 l(l + 1)|j, mj , l, si
Ŝ 2 |j, mj , l, si = ~2 s(s + 1)|j, mj , l, si,
where j runs from |l s| to l + s, and for each j, mj runs from j to j. We note that
1 ˆ2
L̂ · Ŝ|j, mj , l, si = 2
(J L̂2 Ŝ 2 )|j, mj , l, si
~2 (4.16)
= 2
[j(j + 1) l(l + 1) s(s + 1)] |j, mj , l, si.

For s < l, there are (2s + 1) possible values of j: the total number of possible j values is
called multiplicity. For one-electron atom (such as the hydrogen atom) with s = 1/2, there
4.2. SPIN-ORBIT COUPLING 35

are two possible values of j for l 1, j = l 12 and j = l + 12 , and the multiplicity is

2 · 12 + 1 = 2. The spin-orbit coupling of the atomic levels are typically denoted using the
term notation (or term symbol)
2s+1
Lj , (4.17)
where L = S, P, D, F, . . . for l = 0, 1, 2, 3, . . . For example, the doublet P states cor-
responding to l = 1 are 2 P1/2 and 2 P3/2 states, and the doublet D states corresponding to
l = 2 are 2 D3/2 and 2 D5/2 states.

4.2.2 Spin-orbit coupling: induced magnetic field

In a one-electron atom where the there is one electron in the outer-most orbit, the spin of
this valence electron interacts with the shielded Coulomb field due to the nucleus and the
remaining (core) electrons forming a closed shell. Examples of such atoms are alkali atoms
in the first column of the periodic table, or singly ionized atomic ions of atoms that have
filled s states (such as the alkali-earth atoms in the second column of the periodic table). In
these atoms, the spin-orbit coupling modifies the basic structure of the hydrogen-like atom
in two ways that have roughly the same order-of-magnitude correction: one by the induced
magnetic field by the spin, and the other by the relativistic correction.
~ the observer experiences
If an observer is moving with a velocity ~v in an electric field of E,
magnetic field given by
~ = ~v ~
B ⇥ E, (4.18)
c
where c is the speed of light in vacuum, 2 = 1 2
, and = |~v /c|. To first order in , we
get
~ ' 1 ~v ⇥ E
B ~ = p~ ~
⇥ E. (4.19)
c me c
If the observer is an electron, its magnetic moment interacts with this “induced” magnetic
field via the interaction Hamiltonian
✓ ◆
0 1 1
~ = µ ~
v ~ = 1 µ ~ ~
Ĥ = ~ ·B
µ ~· ⇥B (E ⇥ p~). (4.20)
2 2 c 2 me c
Here, the factor of 1/2 in front is called the Thomas factor. Since for electric field created
by the nucleus arises from the central potential and only has a raidal component, we can
~ =E
express it as E ~ r = d (r) r̂, that leads to
dr

0 1 µ~ 1 d
Ĥ = (r) (~r ⇥ p~) · µ
~ ⌘ f (r)L̂ · Ŝ, (4.21)
2 me c r dr
e e d (r)
where L̂ = ~r ⇥ p~, µ
~ = me c
Ŝ, and f (r) = 2m2e c2 r dr
. We can see that the interaction
Hamiltonian takes the form of a spin-orbit interaction L̂ · Ŝ.
For the one-electron atom without the e↵ect of induced magnetic field, we have a hydrogen-
like Hamiltonian
p̂2 p̂2 L̂2
Ĥ0 = + V (r) = r + + V (r), (4.22)
2me 2me 2me r2
36 CHAPTER 4. SPIN-ORBIT COUPLING

with solutions |nli, when expressed in spatial coordinates, gives the wavefunction hr|nli =
'(~r) = Rnl (r)Ylm (✓, '). When the e↵ect of the induced magnetic field is considered as a
perturbation to this Hamiltonian,

p̂2r L̂2 1
Ĥ = Ĥ0 + Ĥ 0 = + 2
+ V (r) = f (r)(Jˆ2 L̂2 Ŝ 2 ), (4.23)
2me 2me r 2
where we used the relationship given by Eq. 4.16.
Using the solutions of the hydrogen (-like) atom solution, we get the equation
✓ ◆
~2
Ĥ|nli = En + f (r)[j(j + 1) l(l + 1) s(s + 1)] |nli = E|nli. (4.24)
2

It is important to note that |nli are not eigenstates of the full Hamiltonian Ĥ, since f (r)
still depends on the radial coordinate r. However, we can apply the concept of expansion
(perturbation theory) if the expectation value of hf (r)i of f (r) is very small. We consider
the expectation value of the energy for a state characterized by the quantum numbers |nlsi,
or equivalently, |nlji, given by

~2
Enlj = hnlj|Ĥ|nlji = En + [j(j + 1) l(l + 1) s(s + 1)]hf (r)inl , (4.25)
2
1 1
since f (r) does not depend on the spin degree of freedom s. Given j is either l + 2
or l 2
,
we find that
1 1 3 1 1
j(j + 1) l(l + 1) s(s + 1) = (l ± )(l + 1 ± ) l(l + 1) = ±(l + ) . (4.26)
2 2 4 2 2
So, to first order in hf (r)inl , the two energy levels become

+ ~2
Enlj ⌘ Ej=l+1/2 = En + lhf (r)inl (4.27)
2
~2
Enlj ⌘ Ej=l 1/2 = En (l + 1)hf (r)inl . (4.28)
2
To estimate the value of hf (r)inl , we realize that the central potential is given by V (r) =
Ze2 /r, and that ✓ ◆
1 d Ze2 Ze2 1
f (r) = = . (4.29)
2m2e c2 r dr r 2m2e c2 r3
The expectation value of this function is given by
Z 1
Ze2 |Rnl (r)|2 2 Ze2 2
hf (r)i = 2 2 3
r dr = 2 2 3 3
. (4.30)
2me c 0 r 2me c a0 n l(l + 1)(2l + 1)

The ratio of the correction term ~2 hf (r)inl to the atomic energy is given by

~2 hf (r)i (Z↵)2 1
= , (4.31)
|E| n l(l + 1/2)(l + 1)
4.2. SPIN-ORBIT COUPLING 37

where ↵ ⌘ e2 /(4⇡"0 ~c) = 1/(137.037) = 7.297 ⇥ 10 3 is called the fine structure constant.
For a singly-charged atom (Z = 1), we confirm that this correction is indeed very small,
on the order of ⇠ ↵2 . The atomic energy levels corresponding to l 1 are split into two
due to spin-orbit coupling, where the energy level corresponding to j = l + 12 increases by
1 (Z↵)2
a fraction (2L+1)(l+1) n
, and the other level corresponding to j = l 21 decreases by a
1 (Z↵)2
fraction l(2L+1) n
.

4.2.3 Spin-orbit coupling: relativistic correction

When relativistic e↵ects are considered, the kinetic portion of the Hamiltonian should be
represented as
1/2 p̂2 p̂4
Ĥ = p̂2 c2 + m2e c4 me c2 + V (r) ' + · · · + V (r), (4.32)
2me 8m3e c2
for |p| ⌧ mc. When the second term in the expansion is treated as a small perturbation
Ĥ 0 = p̂4 /(8m3e c2 ), the correction term in the energy can be estimated using the eigenstates
of the bare hydrogen atom solutions as
1
hnl|Ĥ 0 |nli = 3 2
hp̂4 inl . (4.33)
8me c
Noting that
p̂2 |nli = 2me (En V )|nli, (4.34)
we can evaluate
hnl|p̂4 |nli = 4m2e hnl(En V )2 |nli. (4.35)
Noting that h1/ri = 1/(a0 n2 ) and h1/r2 i = 2/[(2l + 1)a20 n3 ], and therefore hnl|V |nli =
Ze2 /(a0 n2 ) and hnl|V 2 |nli = 2(Ze2 )2 /[(2l + 1)a20 n3 ], we can evaluate the correction term
using the condition En = m(Ze2 )2 /(2~2 n2 ) as

0 (Z↵)2 8n
hnl|Ĥ |nli = |En | 2
3 (4.36)
4n 2l + 1

Therefore,
⇥ the⇤ fractional correction
⇥ ⇤ energy for the two fine-structure split energy levels are
Z↵ 2 4n Z↵ 2 4n
( 2n ) l+1 3 and ( 2n ) l 3 , respectively.

4.2.4 Spin-orbit coupling: Darwin term

In the Schrödinger equation, we have considered the electron to be a point particle such that
the interaction between the electron and the Coulomb field of the nucleus is local. The first
order correction to this assumption is that the electron’s interaction with the Coulomb field
has a non-local nature over a finite length scale, corresponding to the Compton wavelength
of the electron, given by ~/me c. This correction term is expressed in the Hamiltonian as

~2
ĤD = r2 V (R). (4.37)
8m2e c2
38 CHAPTER 4. SPIN-ORBIT COUPLING

Noting that r2 (1/r) = 4⇡ (~r), we find that

⇡e2 ~2
ĤD = (~r). (4.38)
2m2e c2

When we take the average of this potential over the electron wavefunction, we get

⇡e2 ~2 ~ 2
hĤD i = | (0)| , (4.39)
2m2e c2

where (~0) is the value of the wavefunction at the origin. Therefore the Darwin term only
a↵ects the s electrons with non-zero wavefunction value at the origin. Given that the s
electrons are spread out over a length scale of Bohr radius a0 , we find that

1 m3 e 6
| (~0)|2 ' 3 = e6 . (4.40)
a0 ~

This leads to the estimate of the Darwin term as

e8
hĤD i ' me c2 = me c2 ↵ 4 , (4.41)
~4 c 4
and
me c2 ↵ 4
hĤD i / hĤ0 i ' = ↵2 . (4.42)
me c2 ↵ 2