Chapter 5

Quantization of the Spins

As pointed out already in chapter 3, the external degrees of freedom, position and momen-
tum, of an ensemble of identical atoms is described by the Scödinger field operator. As
for the quantization of the internal degrees of freedom of the same ensemble, we have an
appropriate mathematical framework, which is the quantization of collective angular mo-
mentum operators. A single two-level atom is often represented by a (fermionic) Pauli spin
operator, while an ensemble of two-level atoms is conveniently described by a (bosonic)
collective angular momentum operator. In this section, we will present a formal theory of
collective angular momentum algebra.

5.1 Quantization of the orbital angular momentum

5.1.1 Commutation relation
We do not need a new postulate to quantize an orbital angular momentum. A standard
quantization postulate, [q̂, p̂] = i~, automatically quantize an orbital angular momentum.
An orbital angular momentum operator, l = q × p, is decomposed into three cartesian
components:
ˆlx = q̂y p̂z − q̂z p̂y (5.1)
ˆly = q̂z p̂x − q̂x p̂z
ˆlz = q̂x p̂y − q̂y p̂z .

Using the commutation relations, [q̂i , p̂j ] = i~δij, we can easily derive the commutation
relation for ˆlx , ˆly and ˆlz :
h i
ˆlx , ˆly = [q̂y p̂z − q̂z p̂y , q̂z p̂x − q̂x p̂z ] (5.2)
= q̂y [p̂z , q̂z ] p̂x + q̂x [q̂z , p̂z ] p̂y
= i~ (q̂x p̂y − q̂y p̂x )
= i~ˆlz ,
h i
ˆly , ˆlz = i~ˆlx ,
h i
ˆlz , ˆlx = i~ˆly .

1
 If we define the total angular momentum operator by
ˆl2 = ˆl2 + ˆl2 + ˆl2 , (5.3)
x y z

we can show h i
ˆl2 , ˆli = 0 (i = x, y, z) . (5.4)

The above result indicates the total angular momentum ˆl2 and one of the three cartesian
components, for instance ˆlz , can be determined simultaneously without any quantum
uncertainty. A post-measurement state after such a simultaneous measurement of the two
observables, ˆl2 and ˆlz , is called an angular momentum eigenstate.

5.1.2 Angular momentum eigenstates

An angular momentum eigenstate is specified by the two eigenvalues |ν, mi corresponding
to the two observables ˆl2 and ˆlz :
ˆl2 |ν, mi = ~2 ν|ν, mi , (5.5)
ˆlz |ν, mi = ~m|ν, mi . (5.6)
In this case, z-direction is chosen as a quantization axis. We can alternatively choose x or
y direction as a quantization axis.
Now let us introduce a non-Hamiltonian operator defined by
ˆl± = ˆlx ± iˆly . (5.7)

These operators do not commute mutually and also with ˆlz :

h i
ˆl+ , ˆl− = 2~ˆlz , (5.8)
h i
ˆlz , ˆl± = ±~ˆl± . (5.9)

The total angular momentum is now expanded as

³ ´
ˆl2 = ˆl2 + 1 ˆl+ ˆl− + ˆl− ˆl+ . (5.10)
z
2
From (5.8) and (5.10), we have the following relations
ˆl+ ˆl− = ˆl2 − ˆl2 + ~ˆlz , (5.11)
z

ˆl− ˆl+ = ˆl2 − ˆl2 − ~ˆlz . (5.12)

z

If we create a new state by projecting ˆl+ or ˆl− on the angular momentum eigenstate
|ν, mi, the resulting new state still satisfies the original eigenvalue relation for ˆl2 :
³ ´ ³ ´
ˆl2 ˆl± |ν, mi = ˆl± ˆl2 |ν, mi = ~2 ν ˆl± |ν, mi , (5.13)

2
which suggests |ν, mi, ˆl+ |ν, mi and ˆl− |ν, mi are the eigenstates of ˆl2 with the identical
eigenvalue ν. On the other hand, those two new states satisfy slightly different eigenvalue
relations for ˆlz :
³ ´ ³ ´
ˆlz ˆl± |ν, mi = ˆl± ˆlz |ν, mi ± ~ˆl± |ν, mi = ~(m ± 1) ˆl± |ν, mi , (5.14)

which shows |ν, mi, ˆl+ |ν, mi and ˆl− |ν, mi are the eigenstates of ˆlz with the different eigen-
values ~m, ~(m + 1) and ~(m − 1), respectively. Because of this property of shifting an
eigenvalue by one, ˆl+ and ˆl− are called a rasing and lowering operator.
The norm of a state ˆl+ |ν, mi is calcurated by using (5.12)
¡ ¢
hν, m|ˆl− ˆl+ |ν, mi = ~2 ν − m2 − m , (5.15)

which must be non-negative so that the eigenvalue m is upper bounded by

m(m + 1) ≤ ν . (5.16)

The maximum eigenvalue mmax = j is given by the relation, ν = j(j + 1). Similarly, the
norm of a state ˆl− |ν, j − ki is calculated by using (5.11)
£ ¤
hν, j − k|ˆl+ ˆl− |ν, j − ki = ~2 j(j + 1) − (j − k)2 + (j − k) (5.17)
= ~2 (k + 1)(2j − k) ,

which must be also non-negative so that k is upper bounded by

k ≤ 2j . (5.18)

Since the number of lowering operations k takes a positive integer, the maximum eigenvalue
j of the observable ˆlz takes only j = 1/2, 1, 3/2, · · ·. We use j instead of ν = j(j + 1) to
represent the eigenvalue of ˆl2 so that the angular momentum eigenstate is defined by
ˆlz |j, mi = ~m|j, mi , (5.19)
ˆl2 |j, mi = ~2 j(j + 1)|j, mi . (5.20)
Three lowest angular momentum eigenstates are
j = 1/2 | 12 , − 12 i | 21 , 21 i
j=1 |1, −1i |1, 0i |1, 1i
j = 3/2 |3/2, −3/2i |3/2, −1/2i |3/2, 1/2i |3/2, 3/2i

5.1.3 Recursion relation

Let us next establish the projection property of the rasing and lowering operators. We
write down a new state created by ˆl+ as
ˆl+ |j, mi = ~λjm |j, m + 1i , (5.21)

where λjm is a c-number constant we want to determine. Projecting hj, m + 1| from the
left of (5.21), we have
hj, m + 1|ˆl+ |j, mi = ~λjm . (5.22)

3
The adjoint of (5.22) is
hj, m|ˆl− |j, m + 1i = ~λ∗jm , (5.23)
which suggests
ˆl− |j, m + 1i = ~λ∗ |j, mi . (5.24)
jm

Combining (5.21) and (5.24) together with (5.12), we have

ˆl− ˆl+ |j, mi = ~2 |λjm |2 |j, mi (5.25)
£ ¤
= ~2 j(j + 1) − m2 − m |j, mi .

If we neglect an irrelevant phase factor, a constant λjm is given by

p
λjm = j(j + 1) − m(m + 1) . (5.26)

Using this result, we now establish the recursion relation for the raising and lowering
operators: p
ˆl+ |j, mi = ~ j(j + 1) − m(m + 1)|j, m + 1i , (5.27)
p
ˆl− |j, mi = ~ j(j + 1) − m(m − 1)|j, m − 1i . (5.28)

5.2 Connection of angular momentum algebra to an ensem-

ble of two-level atoms
5.2.1 Pauli spin operator
A two-level atom with an excited state |ei and ground state |gi is described as a “fictitious”
spin−1/2 particle with j = 1/2. An analogy we will use here is that an up-spin state
corresponds to an excited state, |ei = | 12 , 12 i, and a down-spin state corresponds to a
ground state, |gi = | 21 , − 12 i. We normally use a Pauli spin operator in place of an angular
momentum operator defined by
~
l̂ = σ̂ . (5.29)
2
The commutation relation (5.2) is now rewritten as

[σ̂i , σ̂j ] = 2iσ̂k (i, j, k = cyclic permutation of x,y,z) . (5.30)

The total angular momentum (5.3) and its commutator bracket with ˆli are also rewritten
as
σ̂ 2 = σ̂x2 + σ̂y2 + σ̂z2 , (5.31)
£ 2 ¤
σ̂ , σ̂i = 0 . (5.32)
In order to restrict the eigenvalues of σ̂z to ±1, which corresponds to the eigenvalues
of ˆlz = ±~/2, we introduce a new postulate:

[σ̂i , σ̂j ]+ = σ̂i σ̂j + σ̂j σ̂i = 2δij , (5.33)

which is called an anti-commutation relation. From (5.30) and (5.33), we have

σ̂i2 = 1 , (5.34)

4
 σ̂i σ̂j = iσ̂k (i 6= j) . (5.35)
The raising and lowering operators (5.7) can be re-defined as
1
σ̂± = (σ̂x ± iσ̂y ) , (5.36)
2
which satisfy the anti-commutator bracket:
[σ̂+ , σ̂− ]+ = 1 . (5.37)
From (5.33) and (5.34), σ̂+ and σ̂− also satisfy
2 2
σ̂+ = σ̂− =0 . (5.38)
The above result indicates that we cannot raise or lower the eigenvalue of σ̂z successively,
which should be the case for a spin-1/2 particle (or two-level atom).
The matrix representation of the spin operators and eigenstates of σ̂z are useful for
later use and now summarized below:
µ ¶ µ ¶ µ ¶
0 1 0 −i 1 0
σ̂x = , σ̂y = , σ̂z = . (5.39)
1 0 i 0 0 −1
µ ¶ µ ¶
0 1 0 0
σ̂+ = , σ̂− = . (5.40)
0 0 1 0
µ ¶ µ ¶
1 1 1 1 1 0
|ei = | , i = , |gi = | , − i = . (5.41)
2 2 0 2 2 1

5.2.2 Collective spin operators

An internal state of the ensemble of identical two-level atom is expanded by a set of 2N
orthogonal states such as |ei1 |gi2 · · · |eiN , where N is a total number of atoms. We can
introduce collective spin operators to represent the internal state of the N atoms in a
corresponding Hilbert space of 2N dimensions:
N
1X
Jˆi = σni (i = x, y, z) , (5.42)
2
n=1

Jˆ2 = Jˆx2 + Jˆy2 + Jˆz2 , (5.43)

h i
Jˆi , Jˆj = iJˆk (i, j, k = cyclic permutation of x,y,z) , (5.44)

Jˆ± = Jˆx ± iJˆy , (5.45)

h i
Jˆ+ , Jˆ− = 2Jˆz . (5.46)
While the Pauli spin operator σ̂± satisfy the anti-commutation relation (5.37) and a
continuous excitation or de-exciation is inhibited as demonstrated by (5.38), the collective
spin operator Jˆ± does not have such a constraint. Because of this difference, the Pauli
spin operator is said “fermionic” and the collective spin operator is called “bosonic”. Of
course, such a terminology is a simple analogy to fermionic and bosonic algebra and does
not mean a particular atom of interest is either a fermion or boson.

5
5.3 Various quantum states of an ensemble of atoms
5.3.1 Angular momentum eigenstate
A simultaneous eigenstate of Jˆ2 and Jˆz is called an angular momentum eigenstate or Dicke
state [1] and defined by
Jˆ2 |J, M i = J(J + 1)|J, M i , (5.47)
Jˆz |J, M i = M |J, M i , (5.48)
where J = N/2, M = −J, −J + 1, · · · J − 1, J and N is a total number of atoms.
Using the recursion relation (5.27), the first, second and third excited Dicke states can
be constructed from the ground state |J, −Ji:
√
Jˆ+ |J, −Ji = 2J|J, −J + 1i , (5.49)
p
Jˆ+ |J, −J + 1i = 2(2J − 1)|J, −J + 2i , (5.50)
p
Jˆ+ |J, −J + 2i = 3(2J − 2)|J, −J + 3i , (5.51)
..
.
p
Jˆ+ |J, M − 1i = (J + M )(J − M + 1)|J, M i . (5.52)

From (5.49) to (5.52), we have the mathematical construction of an arbitrary angular

momentum eigenstate (Dicke state), which is analogous to the mathematical construction
of a photon number eigenstate (Fock state):
µ ¶1/2
1 2J (M +J)
|J, M i = Jˆ+ |J, −Ji , (5.53)
(J + M )! J +M

l
1 ¡ + ¢n
|ni = √ â |0i . (5.54)
n!
The highest excited state |J, Ji and the ground state |J, −Ji are defined by

Jˆ+ |J, Ji = 0 , (5.55)

Jˆ− |J, −Ji = 0 , (5.56)

As shown in Fig. 5.1, there are (N + 1) mutually orthogonal states in this angular mo-
mentum eigenstates, M = −J, −J + 1, · · · , J, while the dimension of the Hilbert space for
N two-level atoms is 2N . Where are the missing states of a total number of 2N − (N + 1)?
We will come back to this question in the next section 5.4.

6
 Figure 5.1: Angular momentum eigenstates (or Dicke states).

5.3.2 Spin coherent state

A state that can be created by the arbitrary rotation of the ground state |J, −Ji in an
extended Bloch sphere is called a spin coherent state or Bloch state [2]. We start with the
introduction of the rotation operator (Fig. 5.2). If we rotate (x,y,z) axes by an angle ϕ
around z-axis, the corresponding collective spin operators are transformed to

Jˆz = Jˆz (5.57)

Jˆn = Jˆx sin ϕ − Jˆy cos ϕ
Jˆk = Jˆx cos ϕ + Jˆy sin ϕ .

Figure 5.2: Definition of a rotation operator.

The rotation by an angle θ around a new axis n is thus given by

ˆ
R̂θϕ = e−iθJn (5.58)
(Jˆx sin ϕ−Jˆy cos ϕ)
= e−iθ³ ´
= exp ζ Jˆ+ − ζ ∗ Jˆ− ,
³ ´ ³ ´
where ζ = 2θ e−iϕ . We used Jˆx = 21 Jˆ+ + Jˆ− and Jˆy = 2i 1
Jˆ+ − Jˆ− to derive the
third equality. A spin coherent state is mathematically constructed by the rotation of the
ground state |J, −Ji (see Fig 5.3), which is analogous to the mathematical construction
of a coherent state via displacement of the vacuum state:

7
 Figure 5.3: A spin coherent state (or Bloch state).

|θ, ϕi = R̂θϕ |J, −Ji , (5.59)

l
¡ + ¢
|αi = exp αâ − α∗ â |0i . (5.60)

Next let us derive the eigenvalue equation for spin coherent states. Collective angular
momentum operators Jˆk and Jˆn are expressed in terms of the rasing and lowering operators
using (5.45) and (5.57),
1 ³ ˆ −iϕ ´
Jˆk = J+ e + Jˆ− eiϕ , (5.61)
2
i ³ ˆ −iϕ ´
Jˆn = J+ e − Jˆ− eiϕ . (5.62)
2
Rotation around n-axis by an angle θ translates these operators into

R̂θϕ Jˆn R̂θϕ

−1
= Jˆn , (5.63)

R̂θϕ Jˆk R̂θϕ

−1
= Jˆk cos θ + Jˆz sin θ . (5.64)
Thus, we can obtain the following operator expression:
h³ ´ i
R̂θϕ Jˆ− R̂θϕ
−1
= R̂θϕ Jˆk + iJˆn e−iϕ R̂θϕ−1
(5.65)
½ µ ¶ µ ¶ ¾
ˆ 2 θ ˆ 2 θ
= iϕ
Jz sin θ + e cos J− − e −iϕ
sin Jˆ+ e−iϕ .
2 2

If we project the rotation operator R̂θϕ from the left of (5.56) and use Iˆ = R̂θϕ
−1
R̂θϕ , we
obtain ³ ´
R̂θϕ Jˆ− R̂θϕ
−1
R̂θϕ |J, −Ji = R̂θϕ Jˆ− R̂θϕ
−1
|θ, ϕi = 0 . (5.66)

By substituting (5.65) into (5.66), we have the eigenvalue equation for spin coherent states,
which is analogous to the eigenvalue equation for coherent states:
½ µ ¶ µ ¶ ¾
θ θ ˆ
Jˆz sin θ + e cos
iϕ 2
Jˆ− − e −iϕ
sin2
J+ |θ, ϕi = 0 , (5.67)
2 2

8
 l

â|αi = α|αi . (5.68)

Since a spin coherent state |θ, ϕi is obtained by simple rotation of the ground state
|J, −Ji, it is an eigenstate of the total angular momentum with the identical eigenvalue,

Jˆ2 |θ, ϕi = J(J + 1)|θ, ϕi . (5.69)

Therefore, it is possible to expand a spin coherent state |θ, ϕi in terms of angular momen-
tum eigenstates |J, M i(M = −J, −J + 1, · · · , J). In order to derive this expansion, we
rewrite (5.58) using Baker-Hausdorf relation [3],
ˆ 2 ˆ ∗ ˆ
R̂θϕ = eτ J+ eln(1+|τ | )Jz e−τ J− , (5.70)

where τ = e−iϕ tan 2θ . Using (5.70) in (5.59), we have

1 ˆ
|θ, ϕi = eτ J+ |J, −Ji (5.71)
(1 + |τ |2 )J
J
X τ M +J ˆM +J
1
= J |J, −Ji
2 J (M + J)! +
M =−J (1 + |τ | )
XJ µ ¶1/2
1 2J
= J
τ M +J |J, M i ,
2 M + J
M =−J (1 + |τ | )

l
2
X∞ − |α|
e 2 αn
|αi = √ |ni . (5.72)
n=0
n!

The inner product of two spin coherent states has an analogous non-orthogonality
relation with two coherent states:

hθ, ϕ|θ0 , ϕ0 i = cos4J (Φ/2) , (5.73)

where
cos Φ = cos θ cos θ0 + sin θ sin θ0 cos(ϕ − ϕ0 ) . (5.74)
Since Φ is an angle between two vectors (θ, ϕ) and (θ0 , ϕ0 ) in the extended Bloch sphere,
a spin coherent state is only pair-wise orthogonal when Φ = π. A set of spin coherent
states, however, forms a complete set [2],
Z J
X
dΩ
(2J + 1) |θ, ϕihθ, ϕ| = |J, M ihJ, M | (5.75)
4π
M =−J

= Iˆ ,

where dΩ is a differential solid angle. Therefore, spin coherent states from an overcomplete
set just as coherent states.