Chapter 7

Spin and SpinAddition

7.1

Stern-Gerlach Experiment Electron Spin

In 1922, at a time, the hydrogen atom was thought to be understood completely in terms
of Bohrs atom model, two assistants at the University of Frankfurt, Otto Stern and
Walther Gerlach, performed an experiment which showed that the electrons carry some
intrinsic angular momentum, the spin, which is quantized in two distinct levels. This was
one of the most important experiments done in the twentieth century, as its consequences
allowed for many interesting experimental and theoretical applications.

7.1.1

Electron in a Magnetic Field

To fully understand the concept of spin, we start by reviewing the some properties of a
classical charged particle rotating about its own symmetry axis. The angular momentum
~ , will create a magnetic dipole moment
due to this rotation, let us call it S
~ , proportional
to the angular momentum
~,
~ = g q S
(7.1)

~ = S
2mc
where is the gyromagnetic ratio and g is just called g-factor. For the electron we have
~ = ge

~ = e S

e ~
B ~
S = ge
S.
2 me c
~

(7.2)

The g-factor of the electron equals two, ge = 2, although for a classical angular momentum, it should be equal to 1. The fact, that the spin of the electron contributes twice
as strong to the magnetic moment as its orbital angular momentum, is called the anomalous magnetic moment of the electron. The constant B is known as Bohrs magneton 1 .
~
If such a magnetic dipole is subject to an external (homogeneous) magnetic field B
it starts to precess, due to the torque ~, known as the Larmor torque, exerted by the
magnetic field
~ =
~ = S
~ B
~,
~ B
(7.3)
1

Bohrs magneton is here given in Gaussian CGS units. In SI-units it looks the same, but the speed
e~
of light is removed, B = 2m
.
e

139

140

CHAPTER 7. SPIN AND SPINADDITION

The potential energy corresponding to this torque is given by

~.
H =
~B

(7.4)

~
Thus the Hamiltonian for a particle with spin in an exterior magnetic field of strength B
is of the form
~B
~.
H = S
(7.5)

7.1.2

Stern-Gerlach Experiment

In the Stern-Gerlach experiment silver atoms, carrying no orbital angular momentum but
with a single electron opening up a new s-orbital2 (l = 0), were sent through a special
magnet which generates an inhomogeneous magnetic field, see Fig. 7.1. The properties

Figure 7.1: Stern-Gerlach Experiment: The inhomogeneous magnetic field exerts a force
on the silver atoms, depending on the spin z-component. Classically a continuous distribution is expected, but the experiment reveals only two values of the spin z-component.
Figure from: http://en.wikipedia.org/wiki/Image:Stern-Gerlach experiment.PNG

of the silver atom in this state are such that the atom takes over the intrinsic angular
momentum, i.e. spin, of this outermost single electron.
The inhomogenity of the magnetic field causes a force F~ acting on the magnetic dipole
in addition to the torque
~ V =
~ (~ B)
~ ,
F~ =
(7.6)
2

This notation is often used in spectroscopy, where one labels the states of different angular momenta
by s (sharp, l = 0), p (principal, l = 1), d (diffuse, l = 2), f (fundamental, l = 3) and alphabetically from there on, i.e. g,h,i,...; Every azimuthal quantum number is degenerate in the sense that it
allows for 2 (2l + 1) bound electrons, which together are called an orbital.

7.1. STERN-GERLACH EXPERIMENT ELECTRON SPIN

141

where V = H if the particle is at rest. Since the force depends on the value of the spin, it
can be used to separate different spins. Classically, the prediction would be a continuous
distribution, bounded by two values, representing spins parallel and antiparallel to the
direction of the magnetic field. All the spins which are not perfectly (anti-)aligned with the
magnetic field would be expected to have components in that direction that lie in between
these maximal and minimal values. We can write the magnetic field as a homogeneous
and an inhomogeneous part, such that it is oriented parallel to the z-axis, i.e. Bx = By=0 ,
~ = Bz ~ez = (Bhom + z) ~ez .
B

(7.7)

~ which in the quantum

The force can then be expressed via the z-component of the spin S,
mechanical formalism will be an operator
F z = Sz .

(7.8)

The separation of the particles with different spin then reveals experimentally the eigenvalues of this operator.
The result of the experiment shows that the particles are equally distributed among
two possible values of the spin z-component, half of the particles end up at the upper spot
(spin up), the other half at the lower spot (spin down). Spin is a angular momentum
observable, where the degeneracy of a given eigenvalue l is (2l + 1). Since we observe two
possible eigenvalues for the spin z-component (or any other direction chosen), see Fig. 7.2,
we conclude the following value for s
2s + 1 = 2

s =

1
.
2

(7.9)

Figure 7.2: Spin 12 : The spin component in a given direction, usually the zdirection, of a spin 12 particle is always found in either the eigenstate with eigenvalue + 21 or with eigenvalue 21 . Figure from: http://en.wikipedia.org/wiki/
Image:Quantum projection of S onto z for spin half particles.PNG

142

CHAPTER 7. SPIN AND SPINADDITION

Result:
Two additional quantum numbers are needed to characterize the nature
of the electron, the spin quantum number s and the magnetic spin quantum number
ms = s, , +s . We conclude: spin is quantized and the eigenvalues of the corresponding observables are given by
Sz ~ ms =

~
2

~ 2 ~2 s (s + 1) = 3 ~2 .
S
4

(7.10)

The spin measurement is an example often used to describe a typical quantum mechanical measurement. Let us therefore elaborate this example in more detail. Consider a
source emitting spin 12 particles in an unknown spin state. The particles propagate along
the y-axis and pass through a spin measurement apparatus, realized by a Stern-Gerlach
magnet as described in Fig. 7.1, which is oriented along the z-axis, see Fig. 7.3.

Figure 7.3: Spin 21 measurement: Spin measurements change the state of the particles, if they are not in an eigenstate of the corresponding operator. Therefore subsequent measurements along perpendicular directions produce random results. Figure from:
http://en.wikipedia.org/wiki/Image:Sg-seq.svg

All particles leaving the Stern-Gerlach apparatus are then in an eigenstate of the Sz
operator, i.e., their spin is either up or down with respect to the z-direction. Lets
now concentrate on the spin up particles (in z-direction), that means we block up the
spin down in some way, and perform another spin measurement on this part of the
beam. If the second measurement is also aligned along the z-direction then all particles
will provide the result spin up, since they are all already in an eigenstate of Sz (see
the upper part of Fig. 7.3). The measurement of a particle being in an eigenstate of the
corresponding operator leaves the state unchanged, therefore no particle will flip its
spin.
If, however, we perform the spin measurement along a direction perpendicular to the
z-axis, lets choose the x-axis, then the results will be equally distributed among spin

7.2. MATHEMATICAL FORMULATION OF SPIN

143

up or spin down in x-direction (see the middle part of Fig. 7.3). Thus, even though
we knew the state of the particles beforehand, in this case the measurement resulted in a
random spin flip in either of the measurement directions. Mathematically, this property
is expressed by the nonvanishing of the commutator of the spin operators
[ Sz , Sx ] 6= 0 .

(7.11)

If we finally repeat the measurement along the z-direction the result will be random
again (see the lower part of Fig. 7.3). We do not obtain precise information about the
spin in different directions at the same time, due to the nonvanishing of the commutator
(7.11) there holds an uncertainty relation for the spin observables.
Of course, we could also choose an orientation of the Stern-Gerlach magnet along some
arbitrary direction. Let us assume we rotate the measurement apparatus by an angle
(in the z x plane), then the probability P+ to find the particle with spin up and P
to find the particle with spin down (along this new direction) is given by
P+ = cos2

7.2

and P = sin2

,
2

such that P+ + P = 1 .

(7.12)

Mathematical Formulation of Spin

Now we turn to the theoretical formulation of spin. We will describe spin by an operator,
more specifically by a 2 2 matrix, since it has two degrees of freedom and we choose
convenient matrices which are named after Wolfgang Pauli.

7.2.1

The PauliMatrices

~ is mathematically expressed by a vector whose components are

The spin observable S
matrices

~ = ~ ~ ,
S
2
where the vector ~ contains the


x
0
~ = y , x =
1
z

(7.13)

so-called Pauli matrices x , y , z :

1
0





0 i
1 0
, y =
, z =
.
i 0
0 1

(7.14)

~ (or the Pauli vector ~ ) can be interpreted as the generator of

Then the spin vector S
rotations (remember Theorem 6.1) in the sense that there is a unitary operator U ()
i ~~

U () = e ~ S = 1 cos

+ in
~ sin ,
2
2

(7.15)

~
~ of the state vectors in Hilbert space.
generating rotations around the -axis
by an angle ||
The scalar product ~ ~ is to be understood as a matrix
~ ~ = x x + y y + z z .

(7.16)

144

CHAPTER 7. SPIN AND SPINADDITION

Whats very interesting to note here is the fact that a spin 12 particle has to be rotated
by 2 2 = 4 (!) in order to become the same state, very much in contrast to our
classical expectation. It is due to the factor 12 in the exponent. This very interesting
quantum feature has been experimentally verified by the group of Helmut Rauch [16]
using neutron interferometry.

7.2.2

Spin Algebra

Since spin is some kind of angular momentum we just use again the Lie algebra 3 , which
~ by S
~
we found for the angular momentum observables, and replace the operator L

[ Si , Sj ] = i~ ijk Sk .

(7.17)

The spin observable squared also commutes with all the spin components, as in
Eq. (6.19)
h
i
2
~
S , Si = 0 .
(7.18)
Still in total analogy with Definition 6.1 we can construct ladder operators S
S := Sx i Sy ,

(7.19)

which satisfy the analogous commutation relations as before (see Eqs. (6.21) and (6.23))
[ Sz , S ] = ~ S
[ S+ , S ] = 2 ~ Sz .

(7.20)
(7.21)

The operators now act on the space of (2 component) spinor states, a twodimensional
Hilbert space which is equipped with a basis of eigenstates | s , ms i, labeled by their
~ 2 and Sz respectively
eigenvalues s and ms of S
1 1
1

 ,
 , 1 | i .

i
,
(7.22)
2 2
2
2
These two states, we call them up and down, are eigenstates of the z Pauli
matrix, which we can interpret as the spin observable in z-direction, with eigenvalues +1
and 1

z | i = + | i

(7.23)

z | i = | i .

(7.24)

Remark for experts: The Lie algebra we studied earlier was that of the threedimensional rotation
group SO(3), the group of orthogonal (hence the O) 3 3 matrices with determinant 1 (which is
indicated by S for special), while we here are studying the group SU (2), the group of unitary 2 2
matrices with determinant 1. The fact that these two Lie algebras look identical is not a mere coincidence,
but is due to the fact that SU (2) is the so called universal covering group of SO(3) relating those groups
in a very close way.

7.2. MATHEMATICAL FORMULATION OF SPIN

145

The eigenstates are orthogonal and normalized, i.e.

h |i = 0 ,

h |i = h |i = 1 .

(7.25)

Lets now gather the established facts to find a representation of the operators on the
aforesaid 2dimensional Hilbert space by noting that we are looking for a hermitian 2 2
matrix with eigenvalues 1, which is trivially satisfied by choosing


1 0
z =
.
(7.26)
0 1
The eigenstates take the form
 
1
|i =
0

 
0
|i =
.
1

(7.27)

We then construct ladder operators from the Pauli matrices, i.e.

= x i y ,

(7.28)

which satisfy the spin algebra (recall Eq. (7.17) or Eq. (7.20))
[ z , ] = 2 .

(7.29)

Since the following relations hold true for the ladder operators
| i = 2 | i

(7.30)

+ | i = 2 | i

(7.31)

| i = + | i = 0 ,

(7.32)

we can represent them as


+ = 2

0 1
0 0


,



0 0
= 2
.
1 0

We then easily get the representation of x and y from Eq. (7.28)





0 i
0 1
x =
, y =
.
1 0
i 0

(7.33)

(7.34)

Properties of the Pauli matrices:

The Pauli matrices satisfy a Lie algebra
[ i , j ] = 2 i ijk k ,

(7.35)

146

CHAPTER 7. SPIN AND SPINADDITION

their square is the identity (in two dimensions) and they are traceless
i2 = 1 ,

Tr i = 0 .

(7.36)

Furthermore, an interesting property arises if one considers not only the square of the
Pauli matrices, but the following object
~ (~ B)
~ = A
~B
~ + i ~ (A
~ B)
~ .
(~ A)

(7.37)

~=B
~ = P~ in Eq. (7.37) the cross product on the right side vanishes and
If we e.g. set A
we can express the kinetic energy as
p~ 2
(~ p~)2
=
.
2m
2m

(7.38)

It turns out to be actually the correct way to include spin into the kinetic part of the
Schrodinger equation.

7.2.3

Spin Measurements

Let us now verify on the theoretical side the conclusions about the spin measurements
which we presented in Fig. 7.3. So we assume to have already performed a measurement
in the z-direction and thus obtained a particle in the up state | i . If we again perform
a measurement in the same direction, which means we apply the operator z on the state
| i , we always get the result +1
h | z | i = h | i = 1 .
| {z }

(7.39)

+1 | i

It is easily verified that we will always get the result 1 if we perform the same measurement in the state | i .
Let us now consider a spin measurement in x-direction on that up state
h | x | i = h | i = 0 .
| {z }

(7.40)

|i

What we find is that, though individual results give 1, on average we find a zero result. The reason is that each individual result +1 or 1 occurs exactly to 50% of the
total outcome. That the individual measurements give 1 is obvious since these are the
eigenvalues of x . The corresponding eigenfunctions are given by
1
|+i =
2

 
1
1

1
|i =
2


1
.
1

(7.41)

7.3. TWO PARTICLES WITH SPIN

7.3

1
2

147

Two Particles with Spin

1
2

In this section we show how to describe a system of two particles with spin, how to
combine the two spins. The addition of the two spins of the constituent particles works in
the same way as the addition of any other angular momenta. To describe the states of the
combined spin we need the tensor product operation , which expresses a product
between elements of different (Hilbert) spaces, e.g. HA , HB . These tensor products form
an element of a larger (Hilbert) space HAB , where
HAB = HA HB .

(7.42)

To give this a little more substance, let us consider two spin 12 particles that each
can be in either of the states or . Then we conclude that there are 4 different ways
to combine these states: , , , and , which we construct by using the tensor
product operation (in the notation the tensor product is often omitted for brevity)
|i |i = |i |i = |i

(7.43)

|i |i = |i |i = |i

(7.44)

|i |i = |i |i = |i

(7.45)

|i |i = |i |i = |i .

(7.46)

Let us now try to define the spin operators for the composite system by looking
at the tensor product structure of the Hilbert space (see Eq. (7.42)). Since we know the
~ (A) and S
~ (B) acting in HA or HB respectively, we construct the
individual spin operators S
composite spin operator such that the individual operators are acting in their respective
subspace, i.e. we set
~ (AB) = S
~ (A) 1 (B) + 1 (A) S
~ (B) = S
~ (A) + S
~ (B) .
S

(7.47)

In total analogy we construct the operator for the spin component in z-direction as
Sz(AB) = Sz(A) 1 (B) + 1 (A) Sz(B) = Sz(A) + Sz(B) .

(7.48)

Next we calculate the spin z-component of the vector (Eq. (7.43)) by using the
relations discussed above, in particular Eqs. (7.23), (7.24) and Eq. (7.13)


Sz(A) + Sz(B) | i | i =
Sz(AB) | i =
= (Sz(A) | i) | i + | i (Sz(B) | i) =
~
~
| i) | i + | i ( | i) =
2
2
~
~
= ( + ) |i |i = ~ |i .
2
2
= (

(7.49)

Using the same methods, for the other combinations (Eqs. (7.44) - (7.46)), we find
Sz | i
Sz | i

=
=

~ |i
Sz | i = 0 ,

(7.50)
(7.51)

148

CHAPTER 7. SPIN AND SPINADDITION

where we dropped the label AB for ease of notation. If not indicated otherwise the operators now always act on the total space, analogously to Eqs. (7.47) and (7.48). Unluckily,
these naive combinations of two spins, although eigenstates of the Sz operator, are not all
~ 2 . The operator S
~ 2 we
of them simultaneous eigenstates of the squared spin operator S
now express in terms of tensor products

2
~ 2 = (S
~ (AB) )2 = S
~ (A) 1 (B) + 1 (A) S
~ (B) =
S
~ (A) )2 1 (B) + 2 S
~ (A) S
~ (B) + 1 (A) (S
~ (B) )2 .
= (S

(7.52)

~ (A) )2 by using a property of the Pauli matrices Eq. (7.36)

We calculate (S
~ (A) )2 =
(S

~2
3
~2
x2 + y2 + z2 =
3 1 = ~2 .
4
4
4

(7.53)

Using this result we rewrite Eq. (7.52)

~ 2 = 3 ~2 1 (A) 1 (B) + 2 S
~ (A) S
~ (B) + 3 ~2 1 (A) 1 (B) =
S
4
4
2
~
[ 6 1 1 + 2 (x x + y y + z z )] .
(7.54)
=
4
~ 2 on our spin states (Eqs. (7.43) - (7.46)) we find that,
Computing the action of S
~ 2 corresponding to a quantum number
though and are indeed also eigenstates of S
s=1
2
~ 2 | i = ~ [ 6 1 1 + 2 (x x + y y + z z )] | i | i =
S
4

~2 
=
6 | i + 2 | i + 2 (i)2 | i + 2 | i =
4
~2
8 | i = 2 ~2 | i = ~2 1 (1 + 1) | i ,
(7.55)
=
4

~ 2 | i = 2 ~2 | i ,
S

(7.56)

the states and are no eigenstates of the squared spin operator

~2 |i = S
~ 2 | i = ~2 (| i + | i) .
S

(7.57)

However, we can form linear combinations of the states and , where we choose
the appropriate weights5 12 for normalization, which are eigenstates to the quantum
numbers s = 1 and s = 0
~2
S
4

1
2

(| i + | i) = 2 ~2

1
2

(| i + | i)

(7.58)

Generally, naively forming tensor products of simultaneous eigenstates of squared angular momentum
and angular momentum z-component operators does not give an eigenstate of the corresponding operators
on the tensor product space. There is, however, a simple way of finding these eigenstates, by calculating
the so called Clebsch-Gordon coefficients, which though mathematically simple can be a quite tiresome
procedure.
5
They are the Clebsch-Gordon coefficients for the addition of two spin 12 s.

7.3. TWO PARTICLES WITH SPIN

~2
S

1
2

1
2

149

(| i | i) = 0 .

(7.59)

So we find a triplet of states | s, ms i, belonging to the spin quantum number s = 1,

with magnetic spin quantum numbers ms = 1, 0, 1 ,
| 1, 1 i = | i
| 1, 0 i =

1
2

(| i + | i)

| 1, +1 i = | i

(7.60)
(7.61)
(7.62)

and a singlet with quantum numbers s = 0 and ms = 0

| 0, 0 i =

1
2

(| i | i) .

(7.63)

Remark I:
Of course, we could have also calculated the state (7.61) by applying the
lowering operator S (see Eq. (7.19)) to the | i state


S | i = SA + SB | i = ~ (| i + | i) .
(7.64)
Remark II: Statistics.
We divide particles into two groups, those of integer spin
values, called bosons, and those of half-integer spin values, called fermions. One can
then conclude, that wave functions describing bosons must always be even/symmetric
functions, while the ones, describing fermions must be odd/antisymmetric functions. This
causes for both types of particles to be subjected to different statistical behavior, fermions
are governed by the FermiDirac statistics while bosons follow the BoseEinstein
statistics.
One very important conclusion from this spinstatistic relation is the so called Pauli
exclusion principle.

Proposition 7.3.1 (Pauli exclusion principle) No two or more fermions

may occupy the same quantum state simultaneously.

Therefore, if we describe two (or more) electrons in the same spin state (symmetric
spin states), they can not be located at the same position, since this would also be a
symmetric state and thus the total wave function would be symmetric, which is impossible
for fermions.

150

CHAPTER 7. SPIN AND SPINADDITION