Relativistic Quantum Mechanics.

An Introduction

by Walter Pfeifer

www.walterpfeifer.ch/

2004, revised 2008

 1

Contents
Contents 1

Preface 2

1 Elements of the theory of special relativity 3

1.1 Lorentz transformation for inertial systems in uniform relative motion 3
1.2 Partial derivatives and quantum mechanical operators 5
1.3 Electromagnetic quantities 7
2 The Dirac equation 11
2.1 The equation 11
2.2 Lorentz-covariance of the Dirac equation (form invariance) 15
2.3 Probability density, current density 19
2.4 Nonrelativistic limit of the Dirac equation with electromagnetic fields.
The Pauli term and the spin-orbit energy 21
2.5 The Dirac equation for particles with an anomalous magnetic moment 27
3 Wave functions of Dirac particles 30
3.1 The wave function of a free particle, helicity, the velocity operator 30
3.2 A Dirac particle in a homogeneous magnetic field 38
3.3 Dirac equation with central potential. Parity and total angular
momentum 43
3.4 Separation of the variables for the Dirac equation with central potential 46
3.5 Solution of the radial equations for a Dirac particle in a Coulomb field 53
3.6 Massless Dirac particles 59
3.7 Particles with arbitrary spin 61
3.8 Charge conjugation. The positron 66
4 Other relativistic quantum mechanical equations 69
4.1 The Klein-Gordon equation 69
4.2 The Klein-Gordon Schrödinger equation 70
References 71

Index 72
 2

Preface
Relativistic quantum mechanics are used to describe high-energy particles and
highly ionised atoms. They give a consistent formalism for spin-½ particles and
provide finer details of atomic and molecular spectra. In short, they are an
important tool of modern physics.
This book deals mainly with the Dirac equation, its properties, its applications and
its limiting cases. A formalism for particles with arbitrary spin and remarks on other
relativistic quantum mechanical equations are given.
This publication is an introduction and is directed towards students of physics and
interested physicists. The detailed developments and the numerous references to
preceding places make it easier to follow. However, knowledge of the elements of
quantum mechanics, relativistic mechanics and electrodynamics is a prerequisite.
In order to relieve the reader, we don't deal with rotations of the coordinate system
and not with Lorentz groups either. We have no renaming of matrices, no
Feynman daggers, no Einstein convention of summation over repeated indices, no
quantum field theory, no second quantisation and no natural units with = = c = 1 .
The system SI (MKSA system) is used without exception.
We use the following symbols: operators are written in bold letters, a three-
dimensional vector is marked with an arrow, two- or four-spinors or -vectors are
underlined and the symbols for matrices are doubly underlined.
 3

1 Elements of the theory of special relativity

1.1 Lorentz transformation for inertial systems in uniform relative motion

x Σ x' Σ'
(x,y,z,t)
(x',y',z',t')
v

z z'

The coordinate system Σ ′ is given parallel to a system Σ and has a constant

relative velocity v in the direction z ( z′ ) . At the time t = t ′ = 0 both origins
coincide.
For v  c (=velocity of light) the classical, nonrelativistic relations

t ′ = t , x ′ = x, y ′ = y , z′ = z − vt (1.1.1)

hold between the space-time coordinates of both systems for a given event. If the
velocity v is not negligible with respect to c , i.e. for the relativistic case, a Lorentz
transformation has to be performed as follows

ct − (v / c ) z z − vt
ct ′ = , x ′ = x, y ′ = y , z ′ = . (1.1.2)
1 − (v / c ) 1 − (v / c )
2 2

Since in relativistic transformations space- and time coordinates are interrelated

linearly, the following variables are used

x 0 = ct , x1 = x, x 2 = y , x 3 = z (1.1.3)

with corresponding expressions for x 0′, ", x 3′. Equations (1.1.2) read now

x 0 − (v / c ) x 3 x 3 − (v / c ) x 0
x 0′ = , x 1′ = x 1, x 2′ = x 2 , x 3′ = . (1.1.4)
1 − (v / c ) 1 − (v / c )
2 2

Introducing the matrix

 4

⎛ 1 −v / c ⎞
⎜ 0 0 ⎟
⎜ 1 − (v / c ) 1 − (v / c )
2 2
⎟
⎜ ⎟
0 1 0 0
a=⎜ ⎟ (1.1.5)
⎜ 0 0 1 0 ⎟
⎜ ⎟
⎜ −v / c 1 ⎟
0 0
⎜⎜ ⎟⎟
⎝ 1 − (v / c ) 1 − (v / c )
2 2
⎠

with elements a00 , a01, ⋅, a03 , a10 , a11, ", a33 , equations (1.1.4) can be written as
3
x i ′ = ∑ aik x k , i = 0,1,2,3, (1.1.6)
k =0

⎛ x 0′ ⎞
⎛ x0 ⎞ ⎜ ⎟
⎜ 1⎟ ⎜ x1′ ⎟
x
and by means of the column vectors x = ⎜ 2 ⎟ and x ′ = ⎜ ⎟ we have
⎜x ⎟ ⎜ x 2′ ⎟
⎜⎜ 3 ⎟⎟ ⎜ ⎟
⎝x ⎠ ⎜ x 3′ ⎟
⎝ ⎠

x′ = ax. (1.1.7)

Eq. (1.1.6) implies

∂x i ′
= aik . (1.1.8)
∂x k
Of course, motions in the x- or y-direction result in corresponding transformation
formulas. Four-vectors which transform like x are named contravariant. With the
help of (1.1.4) we calculate the gauge

( ) ( ) ( ) ( )
2 2 2 2
x 0′ − x1′ − x 2′ − x 3′

(x )0
2
(1− (v / c ) ) − ( x ) (1− (v / c ) ) −
2 3
2 2

(x ′) − (x ′)
2 2
= 1 2
(1.1.9)
1 − (v / c )
2

( ) − (x ) − (x ) − (x )
2 2 2 2
= x0 1 2 3
.

The first and the last line show that the form of this expression isn't changed by the
Lorentz transformation (1.1.4). The expression (1.1.9) is said to be form invariant.
There are other linear transformations which preserve the Lorentz metric (1.1.9): a
rotation of the coordinate system, the reflection of the space coordinates
( x ′ = − x, y ′ = − y , z′ = −z ) and the time reversal ( t ′ = −t ) . These transformations
and the translational transformation of the type (1.1.2) can be coupled individually
resulting in a transformation with the same Lorentz gauge. That is, all these
transformations form a group: the inhomogeneous Lorentz group.
 5

Transformations to a frame with parallel axes but moving in an arbitrary direction

are called boosts.
The general translation where the systems Σ and Σ′ are positioned arbitrarily
K
relative to the velocity v , can be constructed with a preceding instant rotation, a
translation along the z - axis (boost) and following rotations. In most cases it is not
necessary to deal with the mentioned (instant) rotations, i.e. the axes z and z′ can
K
be chosen in the direction of v . In this book we will restrict ourselves to this case,
and the transformation formulas (1.1.2) up to (1.1.7) will be used. Consequently
the Lorentz group will not be investigated.

1.2 Partial derivatives and quantum mechanical operators

The inversion of the transformation relations (1.1.4) reads

x 0′ + (v / c ) x 3′ x 3′ + (v / c ) x 0′
x =
0
, x = x ′, x 2 = x 2′, x 3 =
1 1
(1.2.1)
1 − (v / c ) 1 − (v / c )
2 2

3
x i = ∑ aik−1x k ′ or x = a x ′ with
−1
i.e.
k =0

⎛ 1 (v / c ) ⎞
⎜ 0 0 ⎟
⎜ 1 − (v / c ) 1 − (v / c ) ⎟
2 2

⎜ ⎟ (1.2.2)
−1 ⎜ 0 1 0 0 ⎟
a =⎜ ⎟,
0 0 1 0
⎜ ⎟
⎜ (v / c ) 1 ⎟
⎜ 0 0 ⎟
⎜ 1 − (v / c )2 1 − (v / c )
2
⎟
⎝ ⎠

which can easily be checked. We set up the partial derivatives

∂ ∂
∂ ∂x 3 ∂x 0 (
+ v /c) 3
∂ ∂ ∂x 0 ∂x
= 0 + 3 =
∂x 0′ ∂x ∂x ′ ∂x ∂x ′
0 0
1 − (v / c )
2

∂ ∂
=
∂x ′ 1 ∂x1
(1.2.3)
∂ ∂
= 2
∂x 2′ ∂x
∂ ∂
+ (v / c ) 0
∂
= ∂x ∂x .
3

3′
∂x 1 − (v / c )
2

In short,

∂ 3
∂ ∂x i 3
∂ −1
=∑ = ∑ a (1.2.4)
∂x k ′ i = 0 ∂x ∂x ′ i = 0 ∂x
i k i ik
 6

holds, where (1.2.2) has been used. The inverse relation reads

∂ 3
∂ ∂x i ′ 3
∂
∂x k
= ∑
i = 0 ∂x ′ ∂x
i k
= ∑
i = 0 ∂x ′
i
aik .

In electrodynamics the expression

1 ∂2 1 ∂2 ∂2 ∂2 ∂2 ∂2 3
∂2
c 2 ∂t 2
− ∇ 2
≡ − − − ≡ − ∑
c 2 ∂t 2 ∂x 2 ∂y 2 ∂z 2 (∂x 0 )2 i =1 (∂x i )2

appears. Applying eq. (1.2.3) repeatedly one obtains the Lorentz form invariance

∂2 3
∂2 ∂2 3
∂2
−∑ = −∑ (1.2.5)
(∂x 0′ )2 i =1 (∂x i ′ )2 (∂x ) i =1 ( ∂x )
0 2 i 2

in analogy with (1.1.9).

In quantum mechanics the energy operator reads

∂
E = i= . (1.2.6)
∂t
Therefore we write

∂ ∂ 1 1
= = E ≡ p0 ,
∂x 0
c∂t i=c i=

E K
where we have defined p 0 = . The x-component of the momentum operator p
c
reads

∂
px = −i= (1.2.7)
∂x
and we have

∂ −1 −1 1
= px ≡ p .
∂x 1
i= i=
∂ ∂
Inserting in (1.2.3) and defining p 2 ≡ −i= , p 3 ≡ −i= , we obtain the following
∂y ∂z
relation between operators

E′ p 0 − (v / c ) p 3 1 p 3 − (v / c ) p 0
≡ p 0′ = , p ′ = p1 , p 2′ = p 2 , p 3′ = . (1.2.8)
c 1 − (v / c ) 1 − (v / c )
2 2
 7

⎛ p0 ⎞
⎜ 1⎟
p
That is, the four-vector p = ⎜ 2 ⎟ transforms like x (cf.(1.1.4)). Consequently it is
⎜p ⎟
⎜⎜ 3 ⎟⎟
⎝p ⎠
contravariant. Due to the strong analogy with (1.1.9) the operator expression
(p ) − (p ) − (p ) − (p )
0 2 1 2 2 2 3 2
must be form invariant in a Lorentz transformation, i.e.

(p ′) − (p ′) − (p ′) − (p ′) = ( p 0 ) − ( p1 ) − ( p 2 ) − ( p 3 )
2 2 2 2 2 2 2 2
0 1 2 3

(1.2.9)
E ′2 E2
or 2
′ 2
′ 2
′
− px − py − pz = 2 − px 2 − py 2 − pz 2 .
2

c c
A similar expression comes from the relativistic relation between energy and
momentum (as quantities) of a free particle

E 2 = c 2 p 2 + m02c 4 . (1.2.10)
K
Following the rules of quantum mechanics we substitute E and p by the
K
corresponding operators E and p , (1.2.6) and (1.2.7), and obtain
K
E 2 − c 2 p 2 ≡ E 2 − c 2 ( px 2 + py 2 + pz 2 ) = m02c 4 (1.2.11)

Because the left hand side of (1.2.11) is form invariant, the rest mass m0 is the
same in every inertial system as we expect.

1.3 Electromagnetic quantities

K
In the theory of electromagnetism the electric field strength E and the magnetic
K
field strength B can be calculated starting from the scalar potential Φ and a
K
vector potential A like this
K
K ∂A K K K K
E = −grad Φ − , B = curl A ≡ ∇ × A,
∂t
(1.3.1)
∂Az ∂Ay
for example Bx = − .
∂y ∂z

If a charged particle of charge e moves in an electromagnetic field, the Lorentz

force
K K K K
F = e E +v ×B ( )
(1.3.2)

acts on the particle. Our formulas and quantities are written in the system SI. The
K
potentials Φ and A obey differential equations which contain the electric charge
K
density ρ and the electric current density J . The equations read
 8

1 ∂ 2Φ ρ
− ∇ 2Φ = , (1.3.3)
c ∂t
2 2
ε0
K
1 ∂2 A K K
− ∇ 2
A = μ J
0 . (1.3.4)
c 2 ∂t 2

The quantities μo and ε 0 are the permeability and the dielectricity respectively of
the vacuum. An additional constraint can be chosen. We take the “Lorenz gage”:
K 1 ∂Φ
div A + 2 = 0. (1.3.5)
c ∂t
In the framework of special relativity it is natural to introduce the contravariant four-
vectors

⎛Φ ⎞
A = ( A0 , A1 , A2 , A3 ) ≡ ⎜ , Ax , Ay , Az ⎟ (1.3.6)
⎝c ⎠

and J = ( J 0 , J 1, J 2 , J 3 ) ≡ (c ρ , J x , Jy , Jz ) (1.3.7)

whose components transform as

A0 − ( v / c ) A3 A3 − ( v / c ) A 0
A0′ = , A1′ = A1 , A2′ = A2 , A3′ = , (1.3.8)
1 − (v / c ) 1 − (v / c )
2 2

J 0 − (v / c ) J 3 J 3 − (v / c ) J 0
J 0′ = , J 1′ = J 1 , J 2′ = J 2 , J 3′ = . (1.3.9)
1 − (v / c ) 1 − (v / c )
2 2

when changing over from the inertial system Σ to Σ ′ (c.f. (1.1.4).

We now show that with these transformation rules the differential equations (1.3.3)
up to (1.3.5) are form invariant, i.e. in both systems they have the same form, that
is, the same electromagnetic laws hold.
Making use of the definitions (1.1.3), (1.3.6) and (1.3.7) we rewrite (1.3.3) in the
system Σ ′

∂ 2 A0′ 3
∂ 2 A0′ ρ′ J 0′
−∑ = = 2 = μ0 J 0′ , (1.3.10)
( ) ( ) cε 0 c ε 0
2 2
∂x ′0 i =1
∂x ′
i

where we have applied the basic relation

1
ε 0 μ0 = . (1.3.11)
c2
With the help of (1.2.5), (1.3.8) and (1.3.9) we obtain from (1.3.10)
 9

⎛ ⎞ 0
⎜ ∂ ∂ 2 ⎟ A − (v / c ) A J 0 − (v / c ) J 3
2 3 3

⎜ ( ∂x 0 )2 ∑
− = μ . (1.3.12)
i =1 ( ∂x ) ⎟
0

⎠ 1 − (v / c ) 1 − (v / c )
i 2 2 2
⎝

In the same way we deal with the z-component of eq. (1.3.4)

∂ 2 A3′ 3
∂ 2 A3′
−∑ = μ0 J 3′ (1.3.13)
( ) ( )
2 2
∂x ′
0 i =1
∂x ′i

⎛ ⎞ 3
⎜ ∂ ∂ 2 ⎟ A − (v / c ) A J 3 − (v / c ) J 0
2 3 0

⎜ ( ∂x 0 )2 ∑
− = μ0 . (1.3.14)
i =1 ( ∂x ) ⎟ ( ) ( )
i 2 2 2
⎝ ⎠ 1 − v / c 1 − v / c

We multiply (1.3.14) by v/c and add it to (1.3.12), which yields

⎛ ⎞
⎜ ∂ ∂2 ⎟ 0
2 3
−∑ A = μ0 J 0 , (1.3.15)
⎜ ( ∂x 0 ) 2
i =1 ( ∂x ) ⎟
i 2

⎝ ⎠

which is form invariant with respect to (1.3.10). Multiplying (1.3.12) by v/c and
adding it to (1.3.14) results in

⎛ ⎞
⎜ ∂ ∂2 ⎟ 3
2 3

⎜ ( ∂x 0 )2 ∑
− A = μ0 J 3 (1.3.16)
i =1 ( ∂x )
i 2 ⎟
⎝ ⎠

in form invariant accordance to (1.3.13). The x- or y-component of eq. (1.3.4) in

the system Σ ′ reads

⎛ ⎞
⎜ ∂2 3
∂2 ⎟ k′
⎜ −∑ 2 ⎟
A = μ0 J k ′ with k = 1, 2 . (1.3.17)
( ) ( )
2
⎜ ∂x 0′ i =1
∂x i′
⎟
⎝ ⎠

When handled in the same way as the preceding equations, it results immediately
in the form invariant form affiliated to the coordinate system Σ .
Finally we write eq. (1.3.5) with the help of (1.1.3) and (1.3.6) in the system Σ ′

∂A0′ ∂A1′ ∂A2′ ∂A3′

+ + + = 0. (1.3.18)
∂x 0′ ∂x 1′ ∂x 2′ ∂x 3′

With the help of (1.2.3) and (1.3.8) we obtain

⎛ ∂ ∂ ⎞ ⎛ ∂ ∂ ⎞
⎜ ∂x 0 + (v / c ) ∂x 3 ⎟ A0 − (v / c ) A3 ∂A1 ∂A2 ⎜ ∂x 3 + (v / c ) ∂x 0 ⎟ A3 − ( v / c ) A0
⎜ ⎟ + 1 + 2 +⎜ ⎟ = 0,
⎜⎜ 1 − (v / c )
2
⎟⎟ 1 − (v / c )
2 ∂x ∂x ⎜⎜ 1 − (v / c )
2
⎟⎟ 1 − (v / c )
2

⎝ ⎠ ⎝ ⎠
which results finally in
 10

∂A0 ∂A1 ∂A2 ∂A3

+ + + = 0, (1.3.19)
∂x 0 ∂x1 ∂x 2 ∂x 3
which has the same form as (1.3.18). Thus, we have proven that the field
equations (1.3.3), (1.3.4) relating the electromagnetic potential A with the sources
J and the Lorenz gauge condition (1.3.5) are form invariant in our Lorentz
transform.
 11

2 The Dirac equation

2.1 The equation

In non relativistic quantum mechanics the Schrödinger equation

∂ψ =2 2 K
i= = Hψ ≡ − ∇ ψ + V ( r , t )ψ (2.1.1)
∂t 2m
dominates. The complex wave function ψ depends on space and time. Planck's
constant h divided
K by 2π is marked by = . The potential V depends on the
position vector r of the particle (mass m) and t. In (2.1.1) the Hamiltonian H is
defined. This equation contains a differentiation with respect to the time and in ∇ 2
there are second order derivatives with respect to spatial coordinates. This is
inconsistent with the principles of special relativity, where time and spatial
coordinates have the same importance and are interrelated linearly in the
transformation formulas. That is why the Schrödinger equation is not form invariant
in a Lorentz transformation, which is not acceptable in situations where velocities
are not negligible compared with c.
The relativistic equation which Dirac was looking for, had to treat space and time
in a manifestly symmetric fashion. In addition the equation had to be form invariant
in a Lorentz transformation, and the Hamiltonian had to be Hermitian - a condition
which is well-known to guarantee real expectation values for the energy.

For a free particle with the rest mass m0 Dirac's equation reads
K
∂Ψ ( r , t ) ⎛ ∂ ∂ ∂ ⎞ K K
i= + i=c ⎜ α1 + α2 + α 3 ⎟Ψ ( r , t ) − β m0c 2Ψ ( r , t ) = 0 (2.1.2)
∂t ⎝ ∂x ∂y ∂z ⎠
K K
The second term can be written as a scalar product i=cα ⋅ ∇ or with the help of
K K
(1.2.7) as −cα ⋅ p , and using (1.2.6) we have

EΨ = c (α1 p1 + α 2 p 2 + α 3 p 3 + β m0c )Ψ , (2.1.3)

where p1 means px etc. The operator E acts on the left hand side, and on the
right hand side we have the operator c (α1 p1 + α 2 p 2 + α 3 p 3 + β m0c ) , which is
identical with E , because we demand that (2.1.3) holds for every Ψ . Applying
these operators once more to each side of (2.1.3) we obtain

⎛ 3 ⎞
E 2 Ψ = c 2 ⎜ ∑ α i 2 ( p i ) + ∑ (α iα k + α kα i ) p i p k + m0c ∑ (α i β + βα i ) p i + β 2m02c 2 ⎟Ψ .
3 3
2

⎝ i =1 i ≠ k =1 i =1 ⎠

(2.1.4)
We demand that Ψ satisfies also the relation (1.2.11) like this
 12

⎛ 3 ⎞
E 2Ψ = c 2 ⎜ ∑ ( p i ) + m02c 2 ⎟Ψ
2
(2.1.5)
⎝ i =1 ⎠

Comparing with (2.1.4) yields

α i2 = 1, α i β + βα i = 0 for i = 1, 2, 3,
α i α k + α kα i = 0 for i ≠ k = 1, 2, 3, (2.1.6)
β 2 = 1.

That is, the pairs α i , α k ( i ≠ k ) and α i , β anticommute. Therefore, they cannot

be numbers. In reality, it is possible to find matrices α i , β with the properties given
in (2.1.6). They are chosen to be square. Consequently, Ψ can not be a simple
function but it is a column vector of the same dimension containing functions ψ i
depending on space and time
K
⎛ψ 1 ( r , t ) ⎞
⎜ K ⎟
⎜ψ 2 (r ,t ) ⎟
K
Ψ (r ,t ) = ⎜ ⋅ ⎟ (2.1.7)
⎜ ⎟
⎜ ⋅ ⎟
⎜ψ ( rK , t ) ⎟
⎝ n ⎠

In this way, both sides of (2.1.3) are again column vectors with the same
dimension. The operator on the right hand side of (2.1.3) is the Hamiltonian Hf of
the free particle:

⎛ 3 ⎞
H f = c ⎜ ∑ α i p i + β m0c ⎟ (2.1.8)
⎝ i =1 ⎠

As mentioned above, H f has to be Hermitian. In quantum mechanics this property

of an operator guarantees that its expectation value is real. We show here that the
matrices in H f are also Hermitian i.e. matrix elements of α i or β in symmetric
positions with respect to the matrix diagonal are conjugate complex.

We investigate the operator H f , which, in fact, is a matrix. With the states

Ψ a and Ψ b we form the integral

Ψ a H fΨ b = ∫Ψ †a H fΨ bdτ (
with Ψ a = ψ a1 *,ψ a2 *," ,ψ an * .
†
) (2.1.9)

On the other hand

( )
†
H fΨ a Ψ b = ∫ H fΨ a Ψ bdτ = ∫Ψ a H f Ψ bdτ ,
† †
(2.1.10)

( HΨ )
†
=Ψ a H
† †
where the well-known mathematical relation a has been used.
Hermiticity of the operator H f means that the expression (2.1.9) equals (2.1.10).
 13

Therefore, we have H f = H f
†
i.e. the matrix H f is Hermitian. Because the
coefficients in H f are arbitrary, the matrices α i and β are also Hermitian and
their diagonal matrix- elements are real. Their traces tr (i.e. the sums of the
diagonal elements) vanish for the following reason: According to (2.1.6) one has
α i = −α k α i α k ( i ≠ k ) . The well-known relation concerning the trace of the
product of two matrices, say A and B ,

tr ( AB ) = tr ( BA ) ,

yields tr α i = −tr α k α i α k = −tr α k α i = −tr α i , for which reason

2

tr α i = 0. (2.1.11)

The same relation is true for β . As will be shown at the end of this section the

{ }
choice of the matrix set α i , β is not unique. We choose β to be diagonal. It's
usual to make up the matrix set by the Pauli spin matrices ( 2 × 2 matrices)

⎛ 0 1⎞ ⎛ 0 −i ⎞ ⎛1 0 ⎞
σ1 = ⎜ ⎟, σ 2 = ⎜ ⎟, σ3 = ⎜ ⎟, (2.1.12)
⎝ 1 0⎠ ⎝ i 0⎠ ⎝ 0 −1⎠

which meet σ 2i = 1 ( unity 2 × 2 matrix, i = 1, 2, 3 ) (2.1.13)

and σ 1σ 2 + σ 2 σ 1 = 0 , σ 1σ 2 = iσ 3 with cyclic permutations. (2.1.14)

{ }
Mostly, the matrix set α i , β is represented by the following 4 × 4 matrices

⎛0 σi ⎞ ⎛1 0 ⎞
αi = ⎜ ⎟ for i = 1, 2, 3, β =⎜ ⎟, (2.1.15)
⎜σ 0 ⎟⎠ ⎝ 0 −1⎠
⎝ i

which is named the Pauli realization. Explicitly the set reads

⎛0 0 0 1⎞ ⎛ 0 0 0 -i ⎞
⎜ ⎟ ⎜ ⎟
0 0 1 0⎟ 0 0 i 0⎟
α1 = ⎜ , α2 = ⎜ ,
⎜0 1 0 0⎟ ⎜ 0 -i 0 0 ⎟
⎜⎜ ⎟ ⎜⎜ ⎟⎟
⎝1 0 0 0 ⎟⎠ ⎝ i 0 0 0⎠
(2.1.16)
⎛0 0 1 0⎞ ⎛1 0 0 0 ⎞
⎜ ⎟ ⎜ ⎟
0 0 0 −1⎟ 0 1 0 0⎟
α3 = ⎜ , β= ⎜ .
⎜1 0 0 0⎟ ⎜ 0 0 −1 0 ⎟
⎜⎜ 0 −1 0 0 ⎟⎠
⎟ ⎜⎜ 0 0 0 −1⎟⎟
⎝ ⎝ ⎠
 14

The matrices are evidently Hermitian and meet the relations (2.1.6) and (2.1.11).
{ }
The matrix set α , β can be turned into an equivalent set α ′ , β ′ by unitary
i { } i

transformation using a unitary matrix U ( with U −1

=U
†
) as follows
α i ′ = Uα i U −1 , β ′ = U β U −1. (2.1.17)

{ }
Again, it can easily be shown that the new matrices α i ′ , β ′ are Hermitian and
meet the relations (2.1.6) and (2.1.11). Therefore Dirac's equation (2.1.3) must

{ }
also hold with the matrix set α i ′ , β ′ . We name the corresponding solution Ψ ′ ,
which satisfies

⎛ 3 ⎞
EΨ ′ = c ⎜ ∑ α i ′ p i + β ′m0c ⎟Ψ ′.
⎝ i =1 ⎠

{ }
If we substitute here α i ′ , β ′ by the expressions of (2.1.17) and multiply the
−1
equation by U from the left, we obtain

⎛ 3 ⎞ −1
( )
E U Ψ ′ = c ⎜ ∑ α i p i + β m0c ⎟ U Ψ ′ .
−1

⎝ i =1 ⎠
( )
That is

Ψ = U −1Ψ ′ or UΨ = Ψ ′ . (2.1.18)

Apparently, the solution Ψ ′ { }

of the Dirac equation containing α i ′ , β ′ is simply
interrelated with Ψ due to (2.1.18).

In section 2.3 we will learn that the probability density ρ of the particle is
calculated analogously to classical quantum mechanics, namely

⎛ψ 1 ⎞
⎜ ⎟
ψ
ρ = Ψ Ψ ≡ (ψ 1*,ψ 2 *,ψ 3 *,ψ 4 * ) ⎜ 2 ⎟ .
†

⎜ψ 3 ⎟
⎜⎜ ⎟⎟
⎝ψ 4 ⎠

With Ψ ′ the density reads

ρ ′ = Ψ ′†Ψ ′ = (UΨ ) (UΨ ) = Ψ

† −1
U UΨ = Ψ U UΨ = Ψ Ψ = ρ .
† † † †
(2.1.19)
 15

{ }
That is, for both sets α i ′ , β ′ and {α } , β
i
the density function is identical. One
can say that unitary transformations do not change the physics. We will use this
fact in the next section.
For a particle which is not free, unlike the situation in (2.1.3), but influenced by
K
electromagnetic forces, the electromagnetic potentials Φ and A , (1.3.1), have to
be built in the Dirac equation in the same way as in the Schrödinger equation. The
principle of minimal coupling of the quantum mechanics - or minimal substitution -
orders that the potential energy eΦ (for a charge e) has to be subtracted from the
energy operator E , and to each p i a component −eAi is added. That is, instead
of (2.1.3) we have

⎛ 3 ⎞
( − Φ )Ψ = ⎜ ∑ α i ( p − eA ) + β m0c ⎟Ψ
i i
E e c (2.1.20)
⎝ i =1 ⎠

In section 2.4 it will be shown that the asymptotic form of this equation for v  c is
the classical Pauli equation with two-component spinors.

2.2 Lorentz-covariance of the Dirac equation (form invariance)

Before we go on, we have to show that the Dirac equation is really relativistic i.e.
that it keeps its form in a transformation from one inertial system Σ to another
system Σ′ . We deal first with a free particle. We know, how the operators p i in
the Dirac equation (2.1.3) have to be transformed (see (1.2.8)), but the
K
{ }
transformation rules of the set α i , β and of the four-vector Ψ ( r , t ) have to be
discussed. As will be mentioned in the context with (2.2.14), there will be a certain
arbitrariness for choosing these rules.
We demand that (2.1.3) holds also in the inertial system Σ′ like this

(1 2 3 0
⎦⎥)
⎡E ′− c α ′ p1′ + α ′ p 2′ + α ′ p 3′ − β ′m c 2 ⎤Ψ ′ ( rK′, t ′ ) = 0
⎣⎢
(2.2.1)

Since the coordinate systems are equivalent, the relations (2.1.6) hold also in Σ′
i.e.

β ′2 = α ′2i = 1, α i ′ β ′ +β ′α i ′ =0 for i = 1, 2, 3,
. (2.2.2)
α i ′α k ′ + α k ′α i ′ = 0 for i ≠ k = 1, 2, 3.

The matrices must also be Hermitian i.e.

α ′†i = α i ′ , β ′† = β ′. (2.2.3)

It can be shown that every set {α ′} , β ′

i
which meets (2.2.2) and (2.2.3) is

{ }
interrelated with the corresponding set α i , β by a unitary transformation this way
 16

α ′i = U † α i U , β ′ = U† βU for i = 1, 2, 3 with U = U
† −1
(2.2.4)

This is Pauli's theorem, which we give here without proof. Due to (2.1.18) and
(2.1.19) the Dirac equation generates the same density ρ (and the same physics)

{ } { }
for both sets α i ′ , β ′ and α i , β . Therefore, without loss of generality we can

{ }
use the same set α i , β in the Lorentz system Σ′ and in Σ . We shall no longer
differentiate between both sets of matrices:

β ′ = β , α i ′ = α i , i = 1, 2, 3 . (2.2.5)

K
( )
We now turn to the wave function Ψ ′ ( r ′, t ′ ) ≡ Ψ ′ x′ , (2.2.1), where x ′ is given in

(1.1.4) up to (1.1.7). Because Ψ ( x ) is a four-vector we cannot expect to obtain

( ) ( )
Ψ ′ x′ merely by inserting x x′ in Ψ (see (1.2.1)). However, a combination of
the components ψ 1 ( x ) ,ψ 2 ( x ) ,ψ 3 ( x ) ,ψ 4 ( x ) may occur. Since both the Dirac
equation (2.1.2) as well as the Lorentz transformation (1.1.4) are linear in space-
time coordinates, we demand that the transformation of Ψ ( x ) is linear in the
components of Ψ and make the ansatz

( )
Ψ ′ x′ = S ( a )Ψ ( x ) . (2.2.6)

Our Lorentz transformation matrix a is defined in (1.1.5) and S is a 4 × 4 matrix

which is a function of v (contained in a ) and operates on the four-component
vector Ψ ( x ) . On the right hand side of (2.2.6), following (1.2.2), x is written as
−1
x = a x ′ .The principle of relativity does not differentiate the inertial system Σ from
Σ′ . Therefore the inverse of S , i.e. S , must exist and Ψ ( x ) is constructed in
−1

( )
terms of Ψ ′ x′ like this

( )
Ψ ( x ) = S −1 ( a )Ψ ′ x ′ . (2.2.7)

We investigate the structure of S ( a ) . It has to insure that the Dirac equation is

form invariant in a transformation from the inertial system Σ to Σ′ . We multiply eq.
(2.1.2) from the left by β and introduce the unit matrix

⎛1 0 0 0⎞
⎜ ⎟
0 1 0 0⎟
α0 ≡ 1≡ ⎜ (2.2.8)
⎜0 0 1 0⎟
⎜⎜ ⎟
⎝0 0 0 1 ⎟⎠
 17

in the first term and obtain

⎡ ⎛ ∂ ∂ ∂ ∂ ⎞ ⎤ K
c ⎢i= ⎜ βα 0 + βα 1 + βα 2 + βα 3 ⎟ − β m0c ⎥Ψ ( r , t ) = 0 .
2
(2.2.9)
⎣ ⎝ c ∂t ∂x ∂y ∂z ⎠ ⎦

Using (1.1.3) and 2.1.6) we have

⎛ 3 ∂ ⎞
⎜ i= ∑ βα μ μ − m0c ⎟Ψ ( x ) = 0 . (2.2.10)
⎝ μ =0 ∂x ⎠

Expressing Ψ ( x ) by means of (2.2.7) yields

⎛ 3
⎝ μ =0
−1 ∂
∂x
−1 ⎞
⎜ i= ∑ βα μ S ( a ) μ − m0cS ( a ) ⎟Ψ ′ x ′ = 0 .
⎠
( )
Multiplying with S ( a ) from the left and using S ( a ) S (a ) = 1
−1
(unity matrix) we
obtain

⎛ 3
⎝ μ =0
−1 ∂
∂x
⎞
⎜ i= ∑ S ( a ) βα μ S ( a ) μ − m0c ⎟Ψ ′ x ′ = 0 .
⎠
( ) (2.2.11)

Equation (1.2.4) conveys

∂ 3
∂
∂x μ
= ∑ a ,
ν νμ
ν =0 ∂x ′
(2.2.12)

which we insert in eq. (2.2.11) and obtain

⎛ 3 ∂ ⎞
⎜ i= ∑ S ( a ) βα μ S ( a ) ∑ aνμ ν − m0c ⎟Ψ ′ x ′ = 0 . ( )
3
−1

⎝ μ =0 ν =0 ∂x ′ ⎠

We interchange the indices μ and ν like this

⎡ 3 ⎛ 3
⎣ μ =0 ⎝ ν =0
−1 ⎞ ∂
⎠ ∂x ′
⎤ ′ ′
⎢i= ∑ ⎜ ∑ S ( a ) βα ν S ( a ) aμν ⎟ μ − m0c ⎥Ψ x = 0 .
⎦
( ) (2.2.13)

In order to achieve form invariance between (2.2.10) and (2.2.13) the following
relations must hold

∑ S ( a )βα ν S ( a ) aμν
3
−1
= βα μ . (2.2.14)
ν=0

−1
On both sides we multiply by aλμ (cf. (1.2.2)) and sum over μ . By means of
3

∑
μ
−1
aλμ aμν
=0
= δνλ we obtain

S ( a ) βα λ S ( a ) = ∑ aλμ−1 βα μ ,
3
−1
(2.2.15)
μ =0
 18

which is the conditional equation for the matrix S ( a ) . It holds also for general
Lorentz transformations including instant rotations of the coordinate system and
even for reflections. In fact, this equation forces the form invariance of the Dirac
equation. We make clear that in reality here one defines a four-component wave
function to be a Lorentz function, if it transforms according to (2.2.7) by means of
the relation (2.2.14).

Here we do not construct the matrix S ( a ) step by step (as does Greiner, 1990,
p.110), but we give S ( a ) directly and check it by means of the system of
equations (2.2.15). For our Lorentz transformation, i.e. the translation, we assert

⎛ γ +1 0 − γ −1 0 ⎞
⎜ ⎟
1 ⎜ 0 γ +1 0 γ − 1⎟ 1
S (a) = ⎜ ⎟ with γ = . (2.2.16)
2 ⎜− γ −1 γ +1 1 − (v / c )
2
0 0 ⎟
⎜ ⎟
⎝ 0 γ −1 0 γ + 1⎠

Obviously S is Hermitian. The inverse matrix reads

⎛ γ +1 0 γ −1 ⎞
0
⎜ ⎟
1 ⎜ 0 γ +1 0 − γ − 1⎟
S (a ) =
−1
⎜ ⎟. (2.2.17)
2 ⎜ γ −1 0 γ +1 0 ⎟
⎜ 0 − γ −1 γ + 1 ⎟⎠
⎝ 0

−1
The condition S S = 1 is checked easily. By matrix multiplication the relation

−1 −1
= β S β = β Sβ or β S β = S
† †
S (2.2.18)

can be shown directly. In order to test S ( a ) we insert equations (2.2.16), (2.2.17),

(2.1.16), (2.2.8) and (1.2.2) in (2.2.15). For λ = 0 this equation reads

S ( a ) βα 0 S (a ) = a
−1 −1
00 βα 0 + a03
−1
βα 3

⎛ γ 0 γv / c 0 ⎞
⎜ ⎟
0 γ 0 −γ v / c ⎟
and for both sides we obtain ⎜ .
⎜ −γ v / c 0 −γ 0 ⎟
⎜⎜ ⎟
⎝ 0 γv / c 0 −γ ⎟⎠

For λ = 1 we have S ( a ) βα1S (a ) = a

−1 −1
11 βα 1 and on both sides the result
⎛0 0 0 1⎞
⎜ ⎟
⎜0 0 1 0⎟
.
⎜0 −1 0 0⎟
⎜⎜ −1 ⎟
0 ⎟⎠
⎝ 0 0
 19

The cases λ = 2 and λ = 3 are checked analogously.

Thus, we have found that the Dirac equation of a free particle remains form
invariant in a Lorentz transformation, and therefore it is relativistic.
We show now that the equation (2.1.20), which contains electromagnetic fields, is
also relativistic. With the aid of (1.2.6) we write (2.1.3) like this

⎛ 0 3 2⎞
⎜ c p − ∑ α i c p − β m0c ⎟Ψ = 0 .
i
(2.2.19)
⎝ i =1 ⎠

Using the definition (1.3.6) the equation (2.1.20) turns into

⎛ ⎞
( ) α i c ( p i − eAi ) − β mo c 2 ⎟Ψ = 0
3

⎜
⎝
c p 0
− ceA 0
− ∑
i =1 ⎠
(2.2.20)

Because the four-vectors (p 0

, p1 , p 2 , p 3 ) and (A ,A ,A ,A )
0 1 2 3
transform in an
identical way (cf. (1.2.8) and (1.3.8)) and because equations (2.2.19) and (2.2.20)
have an identical structure, the equation (2.2.20) is also form invariant in a Lorentz
transformation in the same way as (2.2.19).

2.3 Probability density, current density

In order to obtain a relation between the probability density and the current density
of a particle we multiply equation (2.1.2) from the left by Ψ ≡ (ψ 1*,ψ 2 *,ψ 3 *,ψ 4 * )
†

and obtain

∂Ψ 3
∂Ψ
i=Ψ = −i=c ∑Ψ α i i + m0 c 2Ψ βΨ ,
† † †
(2.3.1)
∂t i =1 ∂x

where substitutions from (1.1.3) have been made. The adjoint of eq. (2.1.2) reads

∂Ψ ∂Ψ
† 3 †
−i= = i=c ∑ α †i + m0 c 2Ψ † β †
∂t i =1 ∂x
i

∂Ψ
†
and α i as mentioned in the context of eq. (2.1.10).
†
Notice the sequence of
∂x i

We multiply the equation from the right by Ψ , take into consideration the
Hermiticity of the Dirac matrices α i = α i , β = β ( † †
) (see the context of (2.1.10))
and obtain

∂Ψ ∂Ψ
† 3 †
-i= Ψ = i=c ∑ i α iΨ + m0c 2Ψ † βΨ (2.3.2)
∂t i =1 ∂x

Then, subtraction of (2.3.2) from (2.3.1) yields

∂ ∂
( ) ( )
3
i= Ψ †Ψ = −i=c ∑ i Ψ † α iΨ . (2.3.3)
∂t i =1 ∂x
 20

We name
4
Ψ †Ψ ≡ ∑ψ *kψ k = ρ (2.3.4)
k =1

Clearly, ρ is positive definite, i.e. real and positive at every position and time.
Furthermore we mark the spatial component

cΨ α iΨ ≡ j i , i = 1, 2, 3 .
†
(2.3.5)

With (2.3.4) and (2.3.5) equation (2.3.3) reads

∂ K K
ρ + div j = 0 with j = ( j 1 , j 2 , j 3 ) . (2.3.6)
∂t
This differential equation has the familiar form associated with the conservation of
K
flow of a fluid of density ρ and current density j , in which there are neither
K
sources nor sinks. It is thus reasonable to interpret ρ ( r , t ) as the probability
K K
density and j ( r , t ) as the probability density current of the particle. The density
K K K
ρ ( r , t ) is directly measurable but j ( r , t ) is not, because the uncertainty relation
prohibits to measure simultaneously position and velocity exactly.
We expect that the flow conservation equation (2.3.6) holds in every inertial
system. First we show that ( c ρ , j 1 , j 2 , j 3 ) is a contravariant four-vector (cf.
(1.1.6/7)). Using (2.3.4) and (2.2.8) we write

c ρ = cΨ α 0Ψ ,
†
(2.3.7)

which we name j 0 . Taking into account eq. (2.3.5), we have

j μ ( x ) = cΨ ( x ) α μΨ ( x ) for μ = 0,1, 2, 3 .
†
(2.3.8)

In the system Σ′ a component reads

( )
j μ ′ x′ = cΨ ′ x′ α μΨ ′ x′
†
( ) ( )
where we insert (2.2.6) and obtain

( ) (
j μ ′ x ′ = c S ( a )Ψ ( x ) α μ S ( a )Ψ ( x ) )
†

(2.3.9)
= cΨ
†
( x ) S † ( a ) α μ S ( a )Ψ ( x ) .
With the help of (2.2.18) we write

( )
j μ ′ x′ = cΨ
†
( x ) β S −1 ( a ) βα μ S ( a )Ψ ( x ) . (2.3.10)

Multiplying equation (2.2.14) by S

−1
( a ) and by S ( a ) , we obtain
 21

( a ) βα S ( a ) ,
3

∑ βα ν aμν
−1
=S μ
(2.3.11)
ν =0

which we insert in (2.3.10) like this

( )
3
j μ ′ x ′ = cΨ
†
( x ) β ∑ βα ν aμνΨ ( x )
ν =0
3 3
(2.3.12)
= ∑ aμν cΨ ( x ) α νΨ ( x ) = ∑ aμν j ( x ) .
ν
†

ν =0 ν =0

Obviously, the four-vector j = ( j 0 , j 1 , j 2 , j 3 ) transforms like x . Consequently, the

K
triple j = ( j 1 , j 2 , j 3 ) is a spatial vector, i.e. its components transform under spatial
rotation as the components of a three-vector, for example as ( x1 , x 2 , x 3 ) .

With the aid of (2.3.7) we write the flow conservation equation (2.3.6) in the
system Σ′ like this

0=
i′
∂c ρ ′ ( x ′ ) 3 ∂j x
+∑
′
≡∑
( )
3 ∂j
μ′
x′
.
( ) (2.3.13)
c ∂t ′ i =1 ∂x i ′ μ =0 ∂x μ ′
We insert (2.3.12) and (1.2.4) this way
3 3
∂ −1 3 3
∂ 3 ⎛ 3 −1 ⎞ν
0 = ∑∑ λ λμ ∑ μν
a a j ν
( x ) = ∑ λ ∑ ⎜ ∑ λμ μν
a a ⎟j ( x )
μ =0 λ =0 ∂x ν =0 λ = 0 ∂x ν = 0 ⎝ μ = 0 ⎠ (2.3.14)
3
∂ 3 3
∂j ( x ) λ
= ∑ λ ∑ δ λν j ν ( x ) = ∑ λ
.
λ =0 ∂x ν =0 λ = 0 ∂x

Thus, the flow conservation equation is form invariant in a Lorentz transformation.

This is a further success of the Dirac theory.

2.4 Nonrelativistic limit of the Dirac equation with electromagnetic fields.

The Pauli term and the spin-orbit energy
We start with the Dirac equation (2.1.20), which contains electromagnetic fields

K ⎛ 3 K ⎞ K
EΨ ( r , t ) = ⎜ c ∑ α i ( p i − eAi ) + β m0c 2 + eΦ ( r , t ) ⎟Ψ ( r , t ) . (2.4.1)
⎝ i =1 ⎠
K K
Clearly, the potential Φ ( r , t ) is connected with the 4 × 4 unit matrix. Since Ψ ( r , t )
is an eigenfunction for the energy operator we have
K K K
EΨ ( r , t ) = EΨ ( r , t ) ≡ ( E ′ + m0c 2 )Ψ ( r , t ) . (2.4.2)

The nonrelativistic limit is characterized by

E ′  mo c 2 . (2.4.3)
 22

K
We split the four-component vector Ψ ( r , t ) in two two-component spinors
ϕ and χ like this
K
K ⎛ ϕ (r ,t ) ⎞
Ψ ( r , t ) = ⎜⎜ K ⎟⎟ . (2.4.4)
⎝ χ (r ,t ) ⎠

Using (2.1.15), (2.4.2) and (2.4.4) we write eq. (2.4.1) this way

⎛ϕ ⎞
(E′ + m c ) ⎜ χ ⎟
0
2

⎝ ⎠
⎡ 3 ⎛0 σi ⎞
⎛1 0 ⎞ ⎛ 1 0 ⎞⎤ ⎛ ϕ ⎞
= ⎢c ∑ ( p i − eAi ) ⎜ ⎟ + m0c ⎜
2
⎟ + eΦ ⎜ ⎟⎥ ⎜ ⎟ (2.4.5)
⎜ 0 ⎠⎟
⎢⎣ i =1 ⎝σ i ⎝ 0 −1⎠ ⎝ 0 1 ⎠ ⎥⎦ ⎝ χ ⎠
⎛ϕ ⎞ ⎛σ χ ⎞ ⎛0⎞ ⎛ϕ ⎞
or E ′ ⎜ ⎟ = c ∑ ( p i − eAi ) ⎜ i ⎟ − 2m0c 2 ⎜ ⎟ + eΦ ⎜ ⎟ .
3

⎜σ ϕ ⎟
⎝χ⎠ i =1 ⎝ i ⎠ ⎝χ⎠ ⎝χ⎠

This relation represents a pair of equations with two-component spinors. We write

them explicitly

( E ′ − eΦ ) ϕ = c ∑ ( p i − eAi )σ i χ
3

i =1
(2.4.6)
( E ′ − eΦ + 2m c ) χ = c ∑ ( p − eA )σ ϕ .
3
2 i i
0 i
i =1

We insert the second equation, χ = ( E ′ − eΦ + 2m0c 2 ) c ∑ ( p i − eAi )σ i ϕ , in the

3
−1

i =1
first equation. The resulting relation contains the nonrelativistic spinor wave
function ϕ .

E ′ϕ

= c ∑ ( p i − eAi )σ i c ∑ ( p j − eA j ) σ j ϕ + eΦϕ
3 3
1
(2.4.7)
i =1 ⎛ E ′ − eΦ ⎞ j =1
2m0c ⎜ 1 +
2
⎟
⎝ 2m0c 2 ⎠

In addition to (2.4.3) we demand that the potential energy eΦ is also small, i.e.

E ′ − eΦ  m0c 2 (2.4.8)

and therefore
−1
⎛ E ′ − eΦ ⎞ E ′ − eΦ
⎜1+ 2 ⎟
≅ 1− .
⎝ 2m0 c ⎠ 2m0 c 2

So, equation (2.4.7) reads

 23

1 ⎛ E′ ⎞ 3
( ) ( p j − eA j )σ j ϕ
3
E ′ϕ ≅ ⎜1− 2 ⎟ ∑
2m0 ⎝ 2m0c ⎠ i =1
p i
− eA i
σ i ∑
j =1
. (2.4.9)
( pi − eAi )σ i 2meΦc 2 ∑ ( p − eA )σ
3 3
1
+
2m0
∑
i =1 j =1
j j
j
ϕ + eΦϕ
0

We deal with the first expression on the right hand side of (2.4.9)

∑ ( p − eA )( p − eA ) σ σ
3
i i j j
i j
=
i , j =1

∑ ( p p + e A A ) (σ σ )
2

∑(p )σ
3 3
− eA + +σ jσ i
k k 2 i i 2 i j
k i j
(2.4.10)
k =1 i < j =1

{
−e ⎡⎣( p1 A2 + A1 p 2 ) σ 1σ 2 + ( p 2 A1 + A2 p1 ) σ 2 σ 1 ⎤⎦ + cyclic terms . }
Using the relations (2.1.13), (2.1.14) and

∂ K
p i ( A j ") = −i= i (
A j ( r ) ")
∂x
K (2.4.11)
⎛ ∂A j ( r ) ⎞ j K ⎛ ∂ ⎞ j K j K
= −i= ⎜ ⎟" − A ( r ) i= ⎜ i " ⎟ = ( p A ( r ) )" + A ( r ) ( p ")
i i

⎝ ∂x ⎠ ⎝ ∂x ⎠
i

K
, where three points represent a function of r , we obtain

∑ ( p − eA )( p − eA ) σ σ
3
i i j i
i j
i , j =1
2

= ∑ ( p − eA ) 1+ 0
3
k k

k =1

{
−e ⎡⎣ p1 A2 − A2 p1 − ( p 2 A1 − A1 p 2 ) ⎤⎦ σ 1σ 2 + cyclic terms }
2

= ∑ ( p − eA ) 1 − e {⎡⎣ p }
3
k k 1
A2 − A2 p1 ⎤⎦ iσ 3 + cyclic terms (2.4.12)
k =1
2
K
= ∑ ( p k − eAk ) 1 − e [ p× A]3 iσ 3 + cyclic terms { }
3

k =1
2
K K
= ∑ ( p − eA ) 1 − e=∑ ⎡⎣∇ × A⎤⎦ σ
3 3
k k
l l
k =1 l =1
2

= ∑ ( p − eA ) 1 − e= ∑ B σ
3 3
k k
l l
.
k =1 l =1

In the last line equation (1.3.1) has been inserted. For later applications we make
up the following modification of (2.4.12) setting Ak equal to zero

K
∑ σ i pi σ j p j = ∑ ( pk ) ≡ p2 .
3 3
2
(2.4.13)
i , j =1 k =1
 24

In order to show clearly the so-called Pauli term, we choose a special case of eq.
(2.4.9) with
K
E′ eΦ ( r )
≅ 0, ≅0. (2.4.14)
2m0c 2 2m0c 2

Insertion of (2.4.14) and (2.4.12) in (2.4.9) results in

K ⎡ 1 e= 3 ⎤ K
∑(p − eAk ) + eΦ −
3
E ′ϕ ( r , t ) ≅ ⎢ ∑ Bl σ l ⎥ ϕ ( r , t ) .
k 2
(2.4.15)
⎣ 2m0 k =1 2m0 l =1 ⎦

This is the well-known Pauli equation, which was postulated by W. Pauli in the
framework of classical quantum mechanics, where the last term in (2.4.15) was
derived independently. The expression
= K K K
σ =S ≡S (2.4.16)
2
is the well-known spin operator for two-spinors. With its help the last expression in
(2.4.15) can be written as
e K K
− B ⋅S . (2.4.17)
m0

The two components of ϕ , therefore, describe the spin degrees of freedom in the
nonrelativistic equation for spin-½ fermions. Since spin exists both at low as well
as at high velocities, this implies that the Dirac equation describes fermions with
spin ½. This is another remarkable consequence of this equation.
The last term of (2.4.15) contains the correct value of the magnetic moment

e= J
μ= = 9.2740154 ⋅ 10 −24 , (2.4.18)
2m0 T

which is positive for a positive charge e .The observed magnetic moment of the
electron agrees to about 1%o.
In order to show explicitly the gyromagnetic ratio g = 2 of the electron, we take a
K K
homogeneous magnetic field B . Its vector potential A reads
K 1K K
A = B × r or A1 = ( B2 x 3 − B3 x 2 ) with cyclic permutations.
1
2 2
K K
The relation B = curl A (cf. (1.3.1)) can easily be shown. Supposed the magnetic
K
field (and A ) is feeble, i.e. ( Ak ) and ( p k Ak ) are negligible, the first expression
2

on the right hand side of (2.4.15) reads with the aid of (1.2.7), (2.4.11) and the
K K K K K K
( )
geometric relation B × r ⋅ p = B ⋅ ( r × p )
 25

2
1 K2 e
∑(p )
3 3
1
2m0 k =1
k
− eA k
≅
2m0
p −
m0
∑A p
k =1
k k

K e 1 K K K 1 K2 e K K K
=
1
2m0
p2 −
m0 2
(
B×r ⋅p =
2m0
p −
2m0
)B ⋅ (r × p) (2.4.19)

1 K e K K
= p2 − B ⋅L .
2m0 2m0
K K K
Here the well-known angular momentum L = r × p has been used. Inserting
(2.4.19) and (2.4.16) in (2.4.15) we obtain

K ⎡ 1 K2 e K K K⎤ K
E ′ϕ ( r , t ) ≅ ⎢ p − (
L + 2S ⋅ B ⎥ ϕ ( r , t ) )
(2.4.20)
⎣ 2m0 2m0 ⎦
K
The coefficient of S is the gyromagnetic ratio, and it has the correct value 2.

In the second special case we exclude permanent magnetic fields and set

Ai = 0 , (2.4.21)

for which reason equation (2.4.9) results in

K 1 ⎛ E′ ⎞ 3 3
K
2 ⎟∑ i ∑
E ′ϕ ( r , t ) ≅ ⎜ 1 − σ p i
σ j p j ϕ (r ,t )
2m0 ⎝ 2m0c ⎠ i =1 j =1
K (2.4.22)
1 3 i eΦ ( r , t ) K K K
3
+
2m0 i =1
∑ σ i
p 2 ∑
2m0c j =1
σ j p j ϕ ( r , t ) + eΦ ( r , t ) ϕ ( r , t ) .

Analogously to (2.4.11) we have

K
K ⎛ ∂Φ ( r , t ) ⎞ K
p (Φ ( r , t )") = −i= ⎜ ⎟ " + Φ ( r , t ) ( p ") ,
i i
(2.4.23)
⎝ ∂x
i
⎠

which we insert in (2.4.22) making use of (2.4.13)

K ⎡ 1 ⎛ E′ ⎞ 3 eΦ 3 2⎤ K
E ′ϕ ( r , t ) = ⎢ ⎜ 1 − 2 ⎟∑ ( p )
k 2
+ 2 2 ∑(
pk ) ⎥ ϕ ( r , t )
⎣ 2m0 ⎝ 2m0c ⎠ k =1 4m0 c k =1 ⎦
K (2.4.24)
i=e 3 ⎛ ∂Φ ( r , t ) ⎞ 3 K K K
2 2 ∑ i ⎜ ⎟ ∑ σ j p ϕ ( r , t ) + eΦ ( r , t ) ϕ ( r , t ) .
− σ j

4m0 c i =1 ⎝ ∂x i
⎠ j =1

We establish the following relation

 26

K
⎛ ∂Φ ( r , t ) ⎞ 3
⎟ ∑ σ j ( p ")
3

∑ σi ⎜ i

⎝ ∂x
i
i =1 ⎠ j =1
K
2 ∂Φ ( r , t )
( p j ")
3
= ∑σ j
j =1 ∂x j

K K
⎪⎧ ⎛ ∂Φ ( r , t ) 2 ∂Φ ( r , t ) 1 ⎞ ⎪⎫
+ ⎨σ 1σ 2 ⎜ p − p ⎟ + cyclic terms ⎬"
⎝ ∂x ∂x
1 2 (2.4.25)
⎩⎪ ⎠ ⎭⎪
K
∂Φ ( r , t ) j
( { K K
p ") + i σ 3 ⎡⎣grad Φ ( r , t ) × p ⎤⎦ 3 + cyclic terms " }
3
=∑
j =1 ∂x j

K
∂Φ ( r , t ) j K K
( )
3 3
=∑ p " + i ∑ σ k ⎡⎣grad Φ ( r , t ) × p ⎤⎦ k " ,
j =1 ∂x j
k =1

where (2.1.13) and (2.1.14) have been applied. We insert (2.4.25) in (2.4.24)
obtaining
K K
K ⎡⎛ E ′ − eΦ ( r , t ) ⎞ p 2 K ⎤ K
E ′ϕ ( r , t ) = ⎢⎜ 1 − 2 ⎟ + eΦ ( r , t ) ⎥ ϕ ( r , t )
⎢⎣⎝ 2m0c ⎠ 2m0 ⎥⎦
K
= 2e 3 ∂Φ ( r , t ) ∂ K
2 2 ∑
− ϕ (r ,t ) (2.4.26)
4m0 c i =1 ∂x i
∂x i

=e 3
K K K
+
4m02c 2
∑σ
k =1
k
⎡⎣grad Φ ( r , t ) × p ⎤⎦ k ϕ ( r , t ) .

K
Further simplifications can be made, if Φ ( r , t ) is spherically symmetric, i.e.
K
Φ ( r , t ) = Φ ( r , t ) , which results in
K
3
∂Φ ∂ ∂Φ ∂ ∂Φ r
∑
i =1 ∂x ∂x
i i
=
∂r ∂r
, grad Φ =
∂r r
. (2.4.27)

K K K K
The angular momentum operator L = r × p , the spin operator S , (2.4.16), and the
relations (2.4.27) are inserted in (2.4.26) like this
K K
K ⎡ p2 E ′ − eΦ ( r , t ) p 2 ⎤ K
E ′ϕ ( r , t ) = ⎢ − 2
+ eΦ ( r , t ) ⎥ ϕ ( r , t )
⎣ 2m0 2m0c 2m0 ⎦
. (2.4.28)
⎡ e= 2 ∂Φ ( r , t ) ∂ e 1 ∂Φ ( r , t ) K K ⎤ K
+ ⎢− + S ⋅ L⎥ϕ (r ,t )
⎣ 4m0
2
∂r ∂r 2m02c 2 r ∂r ⎦

The first and the third term on the right hand side of (2.4.28) give the nonrelativistic
Schrödinger equation. The last term is the spin-orbit energy, which is well-known
in quantum mechanics, where it was introduced artificially in order to explain the
energy fine structure of atoms. However, using the Dirac equation this term
emerges automatically. It is also very important in nuclear physics (together with
the last but one expression) for the classification of single-particle states of nuclei,
but there the potential is not only of electromagnetic origin.
 27

2.5 The Dirac equation for particles with an anomalous magnetic moment
In the nonrelativistic equation (2.4.15),

K ⎡ 1 e= ⎤ K
∑(p − eAk ) + eΦ −
3 3
E ′ϕ ( r , t ) = ⎢ ∑B σ ⎥ϕ (r ,t ) ,
k 2
l l
(2.5.1)
⎣ 2m0 k =1 2m0 l =1 ⎦

the last term represents the interaction energy between the normal magnetic
moment

e=
μ= (2.5.2)
2m0

of the Dirac particle and the magnetic field. On the other hand, experimentally, the
magnetic moment of the proton (rest mass M p ) has been determined to

ep =
μ p = (1 + K p ) with K p = 1.79 and ep = e , (2.5.3)
2M p

(where e is the elementary charge), and for the neutron has been found

en =
μn = K n with K n = −1.91 and en = e . (2.5.4)
2Mn

Therefore, the Dirac equation of these particles has to be extended in order to

obtain the following additional terms in the nonrelativistic version (2.5.1):
ep = 3
for protons: − Kp
2M p
∑B σ
l =1
l l
and
(2.5.5)
e= 3
for neutrons: − K n n
2Mn
∑B σ
l =1
l l
.

We maintain that the extended, relativistic Dirac equation for the proton reads
(cf.(2.1.20))

K ⎡ 3 ⎤ K
EΨ ( r , t ) = ⎢c ∑ α i ( p i − eAi ) + β M pc 2 + eΦ ⎥Ψ ( r , t )
⎣ i =1 ⎦
(2.5.6)
e= 3
⎛ Ei ⎞ K
− βK p ∑ ⎜ Σ i Bi − iα i
2M p i =1 ⎝ c ⎠
⎟Ψ ( r , t ) ,

⎛σ 0⎞
where Σ i = ⎜ i ⎟ , and Bi and Ei are the components of the strengths of the
⎜0 σ ⎟
⎝ i ⎠

magnetic and electric external fields respectively. For the neutron an analogous
equation holds, which differs by the absence of the terms eΦ and eAi .

In order to show that (2.5.6) achieves the right nonrelativistic form, we write this
equation explicitly with spinors from (2.4.4)
 28

K K
⎛ ϕ (r ,t ) ⎞ 3 ⎛ 0 σi ⎞ ⎛ ϕ (r ,t ) ⎞
E ⎜⎜ K ⎟⎟ = c ∑ ⎜ ⎟ ( p − ep A ) ⎜⎜ K ⎟⎟
i i

⎝ χ (r ,t ) ⎠ ⎝ χ (r ,t ) ⎠
⎜ 0 ⎟⎠
i =1 ⎝ σ i
K
⎡⎛ 1 0 ⎞ ⎛ 1 0⎞ ⎤ ⎛ ϕ (r ,t ) ⎞
+ ⎢⎜ ⎟ M pc + ⎜ ⎟ epΦ ⎥ ⎜⎜ K ⎟⎟
2
(2.5.7)
⎢⎣⎝ 0 −1⎠ ⎝ 0 1⎠ ⎥⎦ ⎝ χ ( r , t ) ⎠
⎛1 0 ⎞ ep = 3 ⎡⎛ σ i 0⎞ ⎛0 σ i ⎞ E ⎤ ⎛ ϕ ( rK , t ) ⎞
−⎜ ⎟ Kp ∑ ⎢⎜
⎜
⎟ Bi − i ⎜
σ i ⎟⎠ ⎜σ
⎟ i ⎥ ⎜⎜ K ⎟⎟
0 ⎟⎠ c ⎥ ⎝ χ ( r , t ) ⎠
⎝ 0 −1⎠ 2M p i =1 ⎢⎝ 0
⎣ ⎝ i ⎦

or using (2.4.2)

⎛ ϕ (r ,t ) ⎞ ⎛σ i χ ⎞ ep = 3 ⎛σ χ ⎞
⎟⎟ ≅ c ∑ ( p − eA ) ⎜⎜
3
E ′ ⎜⎜ i i
⎟ + K pi ∑ Ei ⎜ i ⎟
⎝ χ (r ,t ) ⎠
⎟ 2M pc i =1 ⎜⎝ −σ i ϕ ⎟⎠
i =1 ⎝σ iϕ ⎠
(2.5.8)
⎛0⎞ ⎛ϕ ⎞ e= 3 ⎛ σ iϕ ⎞
− 2M pc 2 ⎜ ⎟ + eΦ ⎜ ⎟ − K p p ∑ Bi ⎜ ⎟.
⎜ −σ χ ⎟
⎝χ⎠ ⎝χ⎠ 2M p i =1 ⎝ i ⎠

Due to (2.4.8) in the nonrelativistic limit the quantity Φ is relatively small. Because
∂Φ
of Ei = − i (cf. (1.3.1)), neglecting time dependence) the electrostatic energy
∂x
=ep =e ∂Φ =
Ei = − p is extremely small, since  2 ⋅ 10 −14 m is a very small
M pc M pc ∂x i
M pc
=ep
length. Therefore Ei can be neglected relative to the energy cp . We obtain
M pc
the following couple of equations

⎛ ep = ⎞
( p i − eAi )σ i χ
3 3
⎜⎜ E ′ − epΦ + K p
2M p
∑B σ i ⎟
i ⎟
ϕ ≅ c ∑ (2.5.9)
⎝ i =1 ⎠ i =1

⎛ e= ⎞
( pi − eAi )σ i ϕ . (2.5.10)
3 3
and ⎜ E ′ − epΦ + 2M pc 2 − K p p
⎜ 2M p
∑ Bi σ ⎟
i ⎟
χ ≅ c ∑
⎝ i =1 ⎠ i =1

ep =
In (2.5.10) the expression K p Bi is much smaller than the energy M pc 2 and
2M p
can be neglected. We insert (2.5.10) in (2.5.9). With the help of (2.4.8) we obtain

∑ ( p − eA )σ ∑ ( p − eA )σ ϕ
3 3
1
E ′ϕ ≅ i i
i
j j
j
2M p i =1 j =1
(2.5.11)
ep = 3
+ epΦϕ − K p
2M p
∑B σ ϕ .
i =1
i i

With the aid of (2.4.12) equation (2.5.11) results in

 29

K ⎡ 1 ⎤ K
∑(p − ep Ak ) + epΦ ⎥ ϕ ( r , t )
3
E ′ϕ ( r , t ) ≅ ⎢ k 2

⎣⎢ 2M p k =1 ⎦⎥ (2.5.12)
ep = K
− (1 + K p )
3

2M p
∑ Bl σ l ϕ ( r , t ) .
l =1

This equation contains the anomalous magnetic moment of (2.5.3) and is in

agreement with measurements. For neutrons an analogous equation results with
e=
μn = K n n , (2.5.4), where the terms eΦ and eAk are absent.
2Mn

Thus, we have shown that the extension in the Dirac equation (2.5.6) contributes
to the correct magnetic moment in the nonrelativistic limit. From the considerations
in context with (2.1.9) we know that the additional term in (2.5.6) has to be
Hermitian. The summands of the extension in (2.5.6),

ep = ep =
Kp Bi β Σ i and − Kp Ei βα i , (2.5.13)
2M p 2M pc

contain real coefficients and the matrix products β Σ i and iβα i respectively. We
show that these matrices are really Hermitian like this

⎛σ 0 ⎞⎛ 1 0 ⎞
( )
†
βΣ i =Σi β =Σiβ =⎜ i
† †
⎟⎜ ⎟ = βΣ i ,
⎜0 σ i ⎟⎠ ⎝ 0 −1⎠
⎝
⎛0 σ i ⎞⎛ 1
( )
† 0⎞
iβα i = −iα i β = −iα i β = −i ⎜ ⎟⎜ ⎟ = iβα i .
† †
⎜σ 0 ⎟⎠ ⎝ 0 −1⎠
⎝ i

Thus, the whole Hamiltonian in (2.5.6) is Hermitian.

The relativistic form invariance of the extended equation (2.5.6) in a Lorentz
transformation has been shown by W. Greiner (Greiner, 1990, p. 197).
 30

3 Wave functions of Dirac particles

3.1 The wave function of a free particle, helicity, the velocity operator
We start from eq. (2.1.3), which does not contain a potential (i.e. the particle is
free) using (1.2.6)
K
∂Ψ ( r , t ) K ⎛ 3 ⎞ K K
i= ≡ EΨ ( r , t ) = ⎜ c ∑ α i p i + β m0c 2 ⎟Ψ ( r , t ) ≡ Hf Ψ ( r , t ) (3.1.1)
∂t ⎝ i =1 ⎠

In order to split off the time dependence of Ψ we make the ansatz

K K
Ψ ( r , t ) = ψ ( r ) e −iEt / = , (3.1.2)

which is an eigenfunction of the energy operator like this

K K
E ψ ( r ) e −iEt / = = Eψ ( r ) e −iEt / = . (3.1.3)

Consequently, E is the energy of the particle.

∂
It is easy to show that every component of the momentum operator p k = −i= ,
∂x k
(1.2.7), commutes with the Hamilton operator of the free particle,
3
Hf = c ∑ α i p i + β m0c 2 , (3.1.1). Consequently, there exist eigenfunctions with
i =1
defined energy and momentum, as we know from the methods K of quantum
mechanics. The following ansatz is such an eigenfunction of p

K KK ⎛ 3
⎞
ψ ( r ) = ψ 0eip⋅r / = ≡ ψ 0 exp ⎜ i∑ p i x i / = ⎟ . (3.1.4)
⎝ i=1 ⎠

The factor ψ o is a complex four-vector. When the component px of the

K
momentum operator acts on ψ ( r ) , we obtain

K ∂ ⎛ 3 ⎞
p1ψ ( r ) = −i= 1 ψ 0 exp ⎜ i∑ p i x i / = ⎟
∂x ⎝ i =1 ⎠
(3.1.5)
⎛ 3
⎞
= p1ψ 0 exp ⎜ i∑ p i x i / = ⎟
⎝ i =1 ⎠

as we demand. We can write also

K K K K
pψ ( r ) = pψ ( r ) . (3.1.6)
K
The function Ψ ( r ,t ) now has the familiar form of a plane wave function moving in
K
p -direction (if E is chosen positive) like this
 31

K KK
Ψ ( r ,t ) = ψ 0ei( p⋅r -Et ) / = . (3.1.7)

Since the matrices α i are composed of the 2 × 2 -matrices σ i and 1 (cf.

(2.1.15)), it is natural to split the four components of ψ 0 in two groups (cf. (2.4.4))
this way

⎛ϕ0 ⎞
ψ0 =⎜ ⎟ (3.1.8)
⎜χ ⎟
⎝ 0⎠

with the two-component spinors

⎛ ϕ01 ⎞ ⎛ χ 01 ⎞
ϕ0 = ⎜ 2 ⎟ and χ 0 = ⎜ 2 ⎟ . (3.1.9)
⎝ ϕ0 ⎠ ⎝ χ0 ⎠

The expressions ϕ0i and χ 0i , ( i = 1, 2 ) are complex. Arbitrarily we normalize

ϕ †0 ϕ 0 = ϕ01 * ϕ01 + ϕ02 * ϕ02 = 1 . (3.1.10)

K
Anyway, we will have to put a normalization factor, say N / V , in Ψ ( r ,t ) .

We insert (3.1.7), (3.1.3) and (3.1.6) in eq. (3.1.1) and obtain

⎛ϕ0 ⎞ K K ⎛ 3 ⎞⎛ϕ0 ⎞ K K
E ⎜ ⎟ ei( p⋅r −Et ) / = = ⎜ c ∑ α i p i + β m0c 2 ⎟ ⎜ ⎟ ei( p⋅r −Et ) / =
⎜χ ⎟ ⎝ i =1 ⎠ ⎜⎝ χ 0 ⎟⎠
⎝ 0⎠

⎛ϕ0 ⎞ 3 ⎛ϕ0 ⎞ ⎛ϕ0 ⎞

or E ⎜ ⎟ = c ∑ p i α i ⎜ ⎟ + m0c 2 β ⎜ ⎟
⎜χ ⎟ ⎜χ ⎟ ⎜χ ⎟
⎝ 0⎠ i =1 ⎝ 0⎠ ⎝ 0⎠
(3.1.11)
3 ⎛ 0 σ i ⎞⎛ϕ0 ⎞ ⎛ 1 0 ⎞⎛ϕ0 ⎞
= c ∑ pi ⎜ ⎟ ⎜ ⎟ + m0c ⎜
2
⎟⎜ ⎟ .
⎜σ 0 ⎟⎜ χ ⎟ 0 −1⎠ ⎜ χ ⎟
i =1 ⎝ i ⎠⎝ 0 ⎠ ⎝ ⎝ 0⎠

where we have used the expressions (2.1.15). Equation (3.1.11) represents the
following system
3
Eϕ 0 = c ∑ p i σ i χ 0 + m0c 2 ϕ 0
i =1
3
E χ 0 = c ∑ p i σ i ϕ 0 − m0c 2 χ 0
i =1

(E − m c )ϕ
3
or 0
2
0
− c ∑ pi σ i χ 0 = 0
i =1
(3.1.12)
(E + m c ) χ
3

0
2
0
− c∑ p σ i ϕ0 = 0 .
i

i =1

This linear, homogeneous equation for the variables (spinors) ϕ 0 and χ 0 is

solvable, if it satisfies the secular equation
 32

( E − m0c 2 )1
3
−c ∑ p i σ i
i =1
=0
( E + m c )1
3
−c ∑ p σ i i
0
2
(3.1.13)
i =1

(E − m02c 4 ) 1 − c 2 ∑ p i σ i p j σ j = 0 .
3
2
or
i , j =1

The double sum can be written by means of (2.4.13)

K
∑ ( p k ) 1 ≡ p 2 1.
3 3

∑
2
p i j
p σ i
σ j
= (3.1.14)
i , j =1 k =1

From (3.1.13) and (3.1.14) we obtain

K
E 2 = m02c 4 + c 2 p 2 , (3.1.15)

which is the well-known relation of relativistic mechanics (1.2.10). Equation

(3.1.15) is the direct consequence of the energy condition (2.1.5) included in the
Dirac equation.
Equation (3.1.15) has two solutions for the energy:

+ m02c 4 + c 2 p 2 and − m02c 4 + c 2 p 2 .

As soon as interaction of the particle with (say) electromagnetic radiation exists,

the positive energy state Ψ + E must decay immediately in a negative energy state
Ψ − E plus photons, and consequently, it would be very instable in contrast to the
reality. Dirac circumvented this difficulty by the hypothesis that all the states with
negative energy are occupied (Dirac sea). Since Dirac particles are fermions (cf.
section 2.4) the Pauli principle prevents transitions into this sea. In spite of its
infinite charge and mass, this sea, being homogeneous, exerts no direct influence
on us. Although Dirac could interpret antiparticles (say positrons) as holes in this
sea, it is "incredible and hard to swallow" (Yndurain, 1996, p. 58). It is possible to
reformulate the theory without the hypothesis of the sea.
In the following we will deal with positive energies:

E = + m02c 4 + c 2 p 2 (3.1.16)

with the exception of section 3.8. From (3.1.12) we have

3 3
c ∑ pi σ i c ∑ pi σ i
χ0 = i =1
ϕ0 or ϕ0 = i =1
χ0 . (3.1.17)
E + m0c 2
E − m0c 2

We insert (3.1.8) and (3.1.17) in (3.1.7) and obtain

 33

⎛ ϕ0 ⎞
⎜ 3 ⎟ KK
K N ⎜ ⎟ ei( p⋅r −Et ) / = ,
Ψ ( r ,t ) = c∑ p σ i
i
(3.1.18)
V ⎜ ⎟
⎜⎜
i =1
ϕ ⎟
0⎟
⎝ E + m0c
2
⎠

where we have introduced a representative volume V . We demand the following

normalization condition
K K
∫Ψ ( r ,t )Ψ ( r ,t ) dV = 1 ,
†
(3.1.19)
V

with which we determine the constant N . The adjoint free particle wave function
reads

⎛ 3
⎞
K ⎜
N * † † i =1
c ∑ σ i pi ⎟ K K
Ψ † ( r ,t ) = ⎜ ϕ 0 ,ϕ 0 ⎟ e −i( p⋅r −Et ) / = . (3.1.20)
V⎜ E + m0c ⎟2

⎜ ⎟
⎝ ⎠

We write eq. (3.1.19) like this

⎛ 2⎛
3
⎞
2
⎞
⎜ c ⎜∑p σ i ⎟i
⎟
2⎜ †
N ϕ 0ϕ 0 + ϕ 0
† ⎝ i =1 ⎠ ϕ 0 ⎟ = 1. (3.1.21)
⎜
⎜ ( E + m0c ) ⎟⎟
2 2

⎜ ⎟
⎝ ⎠

With the help of (3.1.10) and (3.1.14) we obtain

⎛ K ⎞
⎜ c 2 p2
2
⎟ = 1.
N 1+
⎜ ( E + m c 2 )2 ⎟
⎝ 0 ⎠

Making use of (3.1.15) we have

1+
K
c 2 p2
=
(E + m c ) + E − m c
0
2 2 2 2
0
4

=
2E
(E + m c
0 )
2 2
(E + m c ) 0
2 2 E + m0c 2

E + m0c 2
and N = , (3.1.22)
2E

where the positive root has been taken.

In the nonrelativistic limit with v / c → 0 and p → 0 the spinor χ 0 becomes

negligible (see (3.1.17)) and the four-component vector ψ 0 , (3.1.8/9), effectively
contains only two components:
 34

⎛ ϕ01 ⎞
⎜ 2⎟
ϕ
ψ0 ≅ ⎜ 0 ⎟.
⎜0⎟
⎜⎜ ⎟⎟
⎝0⎠
K
Then, the free particle function Ψ ( r ,t ) in essence is a two-spinor, which we
expect, because Dirac particles are spin ½ particles (section 2.4).
K
We continue to specify the wave function Ψ ( r ,t ) , (3.1.18), i.e. to determine the
spinor ϕ 0 . We introduce the helicity operator

⎛ 3 ⎞
K K
= ⎛σ ⋅ p 0 ⎞ ⎜ ∑
= ⎜ i =1
σ i ⋅ pi 0 ⎟
Sp = ⎜ K K⎟ ≡ ⎟. (3.1.23)
2p⎝ 0 σ ⋅ p⎠ 2 p ⎜ 3
i ⎟
⎜
⎝
0 ∑
i =1
σi ⋅p ⎟
⎠

We show that this operator commutes with the Hamilton operator like this

⎡ 3 ⎤
⎡⎣Hf ,Sp ⎦⎤ = ⎢c ∑ α i p i + β m0c 2 ,Sp ⎥
⎣ i =1 ⎦
K K K K
⎡ ⎛ 0 σ ⋅ p⎞ ⎛1 0 ⎞ 2 ⎛
σ ⋅p 0 ⎞ = ⎤
= ⎢c ⎜ K K ⎟+⎜ ⎟ m0c , ⎜ K K⎟ ⎥
⎣⎢ ⎝ σ ⋅ p 0 ⎠ ⎝ 0 −1⎠ ⎝ 0 σ ⋅ p ⎠ 2 p ⎦⎥
K K K K
⎡⎛ 0 σ ⋅ p ⎞ ⎛σ ⋅ p 0 ⎞⎤ = (3.1.24)
= c ⎢⎜ K K ⎟ ,⎜ K K ⎟⎥
⎣⎢⎝ σ ⋅ p 0 ⎠ ⎝ 0 σ ⋅ p ⎠ ⎦⎥ 2 p
K 2 K 2
c= ⎜ 0
⎛ (σK ⋅ p ) ⎞⎟ c = ⎛⎜ 0 (σK ⋅ p ) ⎟⎞
= − = 0,
2 p ⎜⎜ σK ⋅ pK 2 ⎟⎟ 2 p ⎜⎜ K K 2 ⎟⎟
⎝( ) 0 ⎠ ⎝( σ ⋅ p ) 0 ⎠

whereK (2.1.15) has been used. Apparently, the helicity operator commutes also
with p , for which reason a wave function can be found, which is eigenfunction of
these operators and of Hf .
K K
We construct this function and put the z-axis in the direction of p i.e. p = ( 0, 0,p ) .
K
Then the function Ψ ( r , t ) ,( 3.1.18), reads

⎛ ϕ0 ⎞
K N ⎜ ⎟ i( pz −Et ) / =
Ψ (z) ( r , t ) = ⎜ cpσ z ⎟e .
V⎜ ϕ 0⎟
⎝ E + m0c
2
⎠

Due to (3.1.23) we have

 35

Sp = Sp , x + Sp , y + Sp , z

= ⎛σ x 0 ⎞ = ⎛σ y 0 ⎞ = ⎛σ z 0 ⎞
= ⎜ ⎟ px + ⎜ ⎟ py + ⎜ ⎟p .
2p ⎜ 0 ⎟
σx⎠ 2p ⎜ 0 σy ⎟ 2p ⎜ 0 σ ⎟ z
⎝ ⎝ ⎠ ⎝ z⎠

K
If Sp acts on Ψ ( z ) ( r , t ) , only the term Sp ,z contributes. With the given choice of
the coordinate system and with the eigenvalue of pz this operator reduces to

⎛1 0 0 0⎞
= p ⎛σ 3 0 ⎞ = ⎜ 0 −1 0 0⎟
⎟
Sp,z = ⎜ ⎟= ⎜ . (3.1.25)
2 p ⎜⎝ 0 σ 3 ⎟⎠ 2 ⎜ 0 0 1 0⎟
⎜⎜ 0 0 ⎟
0 −1⎟⎠
⎝

Obviously, when we set

⎛ 1⎞ ⎛0⎞
ϕ 0 = ϕ 0+ ≡ ⎜ ⎟ or ϕ 0 = ϕ 0− ≡ ⎜ ⎟ , (3.1.26)
⎝0⎠ ⎝ 1⎠
K
see (3.1.9), the function Ψ ( r ,t ) is an eigenfunction of Sp,z . Namely, we have for
instance

⎛ ϕ 0 ,+ ⎞
⎜ ⎟ N i( pz −Et ) / =
Sp,zΨ E ,p,+ ≡ Sp,z ⎜ cpσ ⎟ e
z
⎜ E + m c 2 0 ,+ ⎟⋅ ϕ V
⎝ 0 ⎠
⎛ 1 ⎞
⎛ 1 0 0 0 ⎞⎜ ⎟
⎜ ⎟⎜ 0 ⎟ N
= 0 −1 0 0 ⎟
= ⎜ e(
i pz − Et ) / =
⎜ cp ⎟
2 ⎜0 0 1 0 ⎟⎜ 2 ⎟ V
⎜⎜ ⎟⎟ ⎜ E + m0c ⎟
⎝ 0 0 0 −1⎠ ⎜ ⎟
⎝ 0 ⎠
⎛ 1 ⎞
⎜ ⎟
⎜ 0 ⎟ N
= =
e(
i pz −Et ) / =
= ⎜ cp ⎟ = Ψ E ,p,+
2⎜ 2 ⎟ V 2
⎜ E + m 0 c ⎟
⎜ ⎟
⎝ 0 ⎠ (3.1.27)

=
with the eigenvalue + . Analogously, the function
2

⎛ 0 ⎞
⎜ ⎟
⎜ 1 ⎟ i pz −Et / =
K N
Ψ E ,p,− ( r ,t ) = ⎜ 0 ⎟e ( )
, (3.1.28)
V⎜ ⎟
⎜ −cp ⎟
⎜ E + m c2 ⎟
⎝ 0 ⎠
 36

=
which contains ϕ 0,− , generates the eigenvalue − , if Sp,z acts on it. The
2
normalization condition (3.1.10) is met by the choices (3.1.26) for ϕ 0 .

We interpret the result (3.1.27) and the corresponding one for (3.1.28) as follows.
In (2.4.16) the spin operator for two-spinors is given. We generalize it for four-
spinors like this

⎛1 0 0 0⎞
K ⎜ ⎟
K = ⎛σ 0 ⎞ = ⎜ 0 −1 0 0⎟
S= ⎜ K ⎟, for example Sz = (3.1.29)
2⎝0 σ ⎠ 2 ⎜0 0 1 0⎟
⎜⎜ ⎟
⎝0 0 0 −1⎟⎠

= K
Using (2.5.6) we formulate Sz = Σ z . We write the quadrate of S with the help of
2
(2.1.13) this way

⎛ 3 2 ⎞
K 2 =2 ⎛ σ
K 2
0 ⎞ ⎜ ∑
= ⎜ i =1
2
σi 0 ⎟
=2 3
S = ⎜ ⎟ = ⎟ = 1 (3.1.30)
4 ⎜⎝ 0 σK 2 ⎟⎠ 4 ⎜ 3
2⎟ 4
⎜ 0
⎝
∑
i =1
σi ⎟
⎠
K
or S 2 = =2 ⋅ s ( s + 1) ⋅ 1 with s = 1 / 2 . (3.1.31)

This means in analogy with quantum mechanics that the Dirac particles have spin
=s = = / 2 , which we have found already in section 2.4. Because Sp,z , (3.1.25),
agrees with Sz , (3.1.29), we see that in (3.1.27) the spin = / 2 of the particle is
K
directed along the z-axis or along p . In the state (3.1.28) it looks backward. In
section 3.6 we will see that the (massless) neutrinos show a special behaviour of
the helicity.
We now deal with the velocity operator and its expectation value. In
nonrelativistic quantum mechanics there is a correspondence between the time
derivative of an operator O (which does not depend explicitly on time) and its
commutator with the Hamilton operator like this
dO 1
= [O , H ] . (3.1.32)
dt i=
We extend this theorem (Heisenberg equation) over the relativistic theory. We
formulate it with the position vector as operator
K
dr 1 K
= [ r , Hf ] , (3.1.33)
dt i=
which represents the velocity operator. We calculate the right hand side of (3.1.33)
using (2.1.8)
 37

K
dr c ⎡ K 3 = ∂ ⎤
= ⎢r , ∑ α i + β m0c 2 ⎥
dt i= ⎣ i =1 i ∂x i
⎦
.
⎡K 3 ∂ ⎤
= − ⎢r , ∑ α i i ⎥
⎣ i =1 ∂x ⎦
K
The first component of r yields

dx 1 ⎡ ∂ ⎤ ⎛ ∂ ∂ ⎞
= −c ⎢ x 1 ,α 1 1 ⎥ = −cα 1 ⎜ x 1 1 − 1 − x 1 1 ⎟ = cα 1
dt ⎣ ∂x ⎦ ⎝ ∂x ∂x ⎠
K
i.e.
dr
dt
K K
(
= cα with α = α 1 ,α 2 ,α 3 . ) (3.1.34)

This form of the velocity operator is not so strange as it looks. To show this, we
calculate the average or expectation value of the velocity. In analogy with
nonrelativistic quantum mechanics we write it like this
K
K dr † K K K
v ≡ = ∫Ψ ( r ,t ) cαΨ ( r ,t ) dV , (3.1.35)
dt V
K
which is, in fact, the integral over the current density j (cf. (2.3.5)). We restrict
ourselves on particles moving in the z-direction. Due to (3.1.27) and (3.1.22) the
four-spinor wave function reads
K
Ψ E ,p,+ ( r ,t )
⎛ 1 ⎞ ⎛ 1 ⎞
⎜ ⎟ ⎜ ⎟
⎜ 0 ⎟ ⎜ 0 ⎟
N i( pz − Et ) / = E + m0c 2 . (3.1.36)
= ⎜ cp ⎟e = ⎜ cp ⎟ ei( pz −Et ) / =
V⎜ 2 ⎟ 2EV ⎜ 2 ⎟
⎜ E + m0c ⎟ ⎜ E + m0c ⎟
⎜ ⎟ ⎜ ⎟
⎝ 0 ⎠ ⎝ 0 ⎠
K
The z-component of v in (3.1.35) reads

⎛ 1 ⎞
⎛0 0 1 0 ⎞⎜ ⎟
⎜ ⎟ 0
E + m0c ⎛
2
cp ⎞ ⎜0 0 0 −1⎟ ⎜ ⎟
vz = ⎜ 1, 0 , , 0 ⎟ c ⎜ cp ⎟ ∫ dV
2EV ⎝ E + m0c 2 ⎠ ⎜ 1 0 0 0 ⎟⎜ 2 ⎟V
⎜⎜ ⎟ E + m0c ⎟ (3.1.37)
⎝ 0 −1 0 0 ⎟⎠ ⎜⎜ ⎟
⎝ 0 ⎠
E + m0c 2 cp c2p
= 2 c = .
2E E + m0c 2 E

The other components vanish. Using the expressions

m0v m0c 2
p= , E= (3.1.38)
1 − (v / c ) 1 − (v / c )
2 2
 38

of the relativistic mechanics we obtain from (3.1.37)

K
vz = v , (3.1.39)

as we expect.

3.2 A Dirac particle in a homogeneous magnetic field

In cyclotrons or in similar charged particle accelerators, particles circle in a
magnetic field, which we assume to be homogeneous. In this section we deal with
the wave function of a Dirac particle in such a field.
We give an electromagnetic vector field with the following components

A1 = A3 = 0, A2 = Bx and Φ = 0 . (3.2.1)

Due to (1.3.1) the components of the magnetic field read

∂A2 ∂A1
Bx = By = 0. Bz = − =B. (3.2.2)
∂x ∂y

This is a homogeneous magnetic field. We start from (2.1.20)

K ⎡ 3 ⎤ K
EΨ ( r ,t ) = ⎢c ∑ α i ( p i − eAi ) + β m0c 2 ⎥Ψ ( r ,t ) . (3.2.3)
⎣ i =1 ⎦

As in (2.4.4) or (3.1.8) we split up

K
⎛ ϕ ( r ,t ) ⎞
Ψ = ⎜⎜ K ⎟⎟ . (3.2.4)
⎝ χ ( r ,t ) ⎠

In analogy with (3.1.12) we can write (3.2.3) as a pair of equations containing two-
spinors like this

(E − m c )ϕ = c ∑σ ( p − eAi ) χ
3
2 i
0 i
i =1
(3.2.5)
(E + m c ) χ = c ∑σ ( p − eA ) ϕ .
3
2 j j
0 j
j =1

Elimination of χ results in an equation of second order in p i for ϕ this way

( E 2 − m02c 4 )ϕ = c 2 ∑ ( pi − eAi )( p j − eA j ) σ i σ j ϕ
3

i , j =1
(3.2.6)
⎡ 3 ⎤
= c ⎢ ∑ ( p k − eAk ) 1 − e=∑ Bl σ l ⎥ ϕ ,
3
2 2

⎣ k =1 l =1 ⎦

where we have made use of (2.4.12). Insertion of (3.2.2) yields

 39

( ) 2 ⎡
⎢ ∑ ( p ) − ep A − eA p + e A ( ) ⎤
3 2
k 2 ( 2) ( 2) ( 2) ( 2) ( 2)
E 2
− m0
2 4
c ϕ = c 2
− e=Bσ z ⎥ ϕ .
⎣ k =1 ⎦

From (2.4.11) and (3.2.1) we take

( ) ( ) ( )
p ( 2) A( 2) " = p ( 2) A( 2) " + A( 2) p ( 2) " = 0 + Bx p ( 2) " and obtain ( )
(E 2
− m02c 4 ) ϕ = c 2 ⎡⎣ px 2 + py 2 + pz 2 − e2Bxpy + e 2B 2 x 2 − e=Bσ z ⎤⎦ ϕ . (3.2.7)

For ϕ we make the ansatz

ϕ = f (x)e ( )
i ky y + kz z
(3.2.8)

with the numbers k y and k z . Equation (3.2.7) results in

⎡ d2 ⎤
(E 2
− m02c 4 ) f ( x ) = c 2 ⎢ − =2 2 + = 2k y2 + = 2k z2 − e2Bx =k y + e 2B 2 x 2 − e=Bσ z ⎥ f ( x ) or
⎣ dx ⎦

E2 ⎡ ⎤
− m02c 2 − = 2k z2 ⎢ 2
⎛ ⎞
2
⎛ = k ⎞
2
⎥
2 1 d e
c f ( x ) = ⎢− + ⎜⎜ B ⎟⎟ ⎜ x − y
⎟ − σ z ⎥ f ( x ) . (3.2.9)
e =B ⎢
e dx
B
2
⎝ = ⎠ ⎝ eB ⎠ ⎥
⎣ = ⎦

With the new variable

e ⎛ =k ⎞ ⎡ d2 eB d 2 ⎤
ξ= B⎜x − y ⎟, ⎢i.e. = ⎥ (3.2.10)
= ⎝ eB ⎠ ⎣ dx 2 = dξ 2 ⎦

and the constant

⎛ E2 ⎞
a = ⎜ 2 − m02c 2 − =2k z2 ⎟ / ( e=B ) (3.2.11)
⎝c ⎠

we obtain from (3.2.9)

⎛ d2 ⎞
a f (ξ ) = ⎜ − 2 + ξ 2 − σ z ⎟ f (ξ ) (3.2.12)
⎝ dξ ⎠

with f (ξ ) ≡ f ( x (ξ ) ) . We choose f (ξ ) so, that it is an eigenfunction of σ z , i.e.

⎛1 0 ⎞
σ z f (ξ ) ≡ ⎜ ⎟ f (ξ ) = μ f (ξ ) . (3.2.13)
⎝ 0 −1⎠

This eigenvalue equation is satisfied by

 40

⎛ f (ξ ) ⎞
f (ξ ) = ⎜ +1 ⎟ with μ = +1 and
⎝ 0 ⎠
(3.2.14)
⎛ 0 ⎞
f (ξ ) = ⎜ ⎟ with μ = −1 .
⎝ f−1 (ξ ) ⎠
Taking into account both solutions (3.2.14) the equation (3.2.12) is summarised
like this

⎛ d2 2⎞
⎜ 2 − ξ ⎟ fμ ( ξ ) = − ( a + μ ) f μ ( ξ ) , (3.2.15)
⎝ dξ ⎠

which has the following solution

fμ ,n (ξ ) = Ce −ξ Hn (ξ ) .
2
/2
(3.2.16)

The expression Hn (ξ ) is a Hermite polynomial, C is a normalising constant and

a + μ −1
n= = 0,1, 2," . (3.2.17)
2
From this formula and from (3.2.11) we obtain

E 2 = m02c 4 + c 2 = 2k z2 + c 2 =eB ( 2n − μ + 1) . (3.2.18)

Hence, the energy E is quantized! In order to see the order of magnitude of the
quanta, we calculate the basis quantum of E 2 − m02c 4 for an electron in the field
of 2 Tesla. Due to (3.2.18) we get for μ = n = 1, k z = 0 :

E 2 − m02c 4 = 2c 2 =eB
= 2 ⋅ ( 3 ⋅ 108 ) 1.05 ⋅ 10−341.6 ⋅ 10−19 2 J 2
2
(3.2.19)

= (15.4 eV ) ,
2

which is tiny in view of technical applications. On the other hand, we calculate the
order n for an energy 10 MeV = E 2 − m02c 4 by means of (3.2.19):

E 2 − m02c 4
n= = 4.22 ⋅ 1011
2c =eB
2

i.e. the order of the function (3.2.16) is very high, and the energy steps are
relatively small so that practically every similar energy value can be realized.

With the aid of (3.2.5), (3.2.1), (3.2.8), (3.2.13) and (3.2.10) the spinor χ is
expressed by ϕ like this
 41

χ=
c
E + m0c 2 1
(
σ p x + σ 2 ( p y − eBx ) + σ 3 p z ϕ )
⎛ ⎞
+ σ 2 ( =k y − eBx ) + =k z μ ⎟ ϕ
c d
= 2 ⎜
−i=σ 1 (3.2.20)
E + m0c ⎝ dx ⎠
c ⎛ ⎛ d ⎞ ⎞
= ⎜ − =eB ⎜ iσ + σ ξ ⎟ + =k μ ⎟ϕ .
E + m0c 2 ⎝ ⎝ dξ
z
⎠
1 2
⎠

We will need the mathematical relations

d
dξ
(2 2

)
e −ξ / 2H n (ξ ) = e −ξ / 2 ( −ξ H n (ξ ) + 2nH n −1 (ξ ) )
(3.2.21)
=e −ξ 2 / 2
(ξ H (ξ ) − H (ξ ) ) .
n n +1

We intend to write the vectors ϕ , χ and Ψ explicitly. From (3.2.8), (3.2.16),

(3.2.17), and for μ = +1 we have

a i( k y + k z ) ⎛ H (ξ ) ⎞
, ϕ μ =+1 = Ce y z e −ξ / 2 ⎜ n
2
n= ⎟ (3.2.22)
2 ⎝ 0 ⎠

and with the help of (3.2.20), (3.2.21) and (2.1.12) we obtain

(
i k y +k z )
Cce y z ⎛0⎞
χ μ =+1 = =eBie −ξ ⎡⎣ξ H n (ξ ) − 2nH n −1 (ξ ) − ξ H n (ξ ) ⎤⎦ ⎜ ⎟
2
/2

E + mo c 2 ⎝ 1⎠
i( k y + k z )
Cce y z ⎛ 1⎞
+ = k μ e −ξ 2 / 2
H ( ξ ) ⎜ ⎟ (3.2.23)
E + mo c 2
z n
⎝0⎠
i( k y + k z )
Cce y z −ξ 2 / 2 ⎛ =k z ⋅ 1 ⋅ H n ( ξ ) ⎞
= e ⎜ ⎟.
E + mo c 2 ⎜ −i =eB ⋅ 2n ⋅ H (ξ ) ⎟
⎝ n −1 ⎠

a
For μ = −1, n = − 1 one obtains
2

⎛ 0 ⎞
ϕ −1 = Ce ( )
i ky y + kz z
e −ξ
2
/2
⎜ ⎟,
⎝ H n ( ξ ) ⎠
i( k y + k z )
(3.2.24)
Cce y z −ξ 2 / 2 ⎛ i =eBH n +1 (ξ ) ⎞
χ −1 = e ⎜⎜ ⎟⎟ .
E + m0c 2 ⎝ − =k z ⋅ 1 ⋅ H n ( ξ ) ⎠

From classical and relativistic mechanics we know that, if the momentum of the
electron is perpendicular to the magnetic field ( pz = =k z = 0 ) , the particle circles in
a plane perpendicular to the field. We set the position vector in the x-direction ,
through which the momentum is directed parallel to the y-axis.
 42

We calculate the current density j 2 depending on x by means of (2.3.5) and

(2.1.15):

j 2 (ξ ( x ) ) = cΨ (ξ ) α 2Ψ (ξ )
†

⎛ 0 σ 2 ⎞⎛ϕ ⎞ (3.2.25)
= c (ϕ*, χ * ) ⎜ ⎟ ⎜ ⎟ = c ⎡⎣ϕ * σ 2 χ + χ * σ 2 ϕ ⎤⎦ .
⎜σ 0 ⎟ χ
⎝ 2 ⎠⎝ ⎠

Using (3.2.22) and (3.2.23) we can write for μ = 1 and k z = 0

Cc ⎛ − =eB 2nH n −1 ⎞
e −ξ ( H n , 0 ) ⎜
2
j2 = ⎟
E + m0c 2
⎝ 0 ⎠

) ⎛⎜⎝ iH
0 ⎞
+
Cc
E + m0c 2 (
e −ξ 0,i =eB 2nH n −1
2

⎟
n ⎠
(3.2.26)

Cc
=eB ⋅ 4n ⋅ e −ξ H n (ξ ) H n −1 (ξ ) .
2
=−
E + m0c 2

The lowest order Hermite polynomials read

n Hn (ξ )
0 1
1 2ξ (3.2.27)
2 4ξ 2 − 2
3 8ξ 3 − 12ξ .

For n = 1 we obtain
Cc
j 2 ,n =1 = =eB ⋅ 4 ⋅ e −ξ 2
2ξ ,
E + m0c 2

which has an extremum at ξ = 1 / 2 i.e. (see (3.2.10)) at

1 = p
x= + y . (3.2.28)
2 eB eB

Due to (1.2.10) and (3.2.19) we have for n = μ = 1:

E 2 − m02c 4
py = = 2= eB . (3.2.29)
c
Inserting (3.2.29) in (3.2.28) we obtain

3 2= 3 py
x= = . (3.2.30)
2 eB 2 eB
 43

This is a characteristic value of the radial distribution of j 2,n =1 . Classical and

relativistic mechanics give the following expression for the radius of the circular
path of the electron

p
r = , (3.2.31)
eB
which is near to the characteristic value (3.2.30).

3.3 Dirac equation with central potential. Parity and total angular
momentum
We introduce a central potential V ( r ) into the Hamilton form of the Dirac equation
(2.1.8) like this
3
H = c ∑ α i p i + β m0c 2 + V ( r ) 1 (3.3.1)
i =1

In our applications the potential V ( r ) will be electrostatic. Therefore the equation

will be form invariant in a Lorentz transformation as shown in context with (2.2.20).
K K
The following two operators characterize the solution of EΨ ( r,t ) = HΨ ( r,t ) ,
because they commute with H .
a) We define the parity operator P this way

P = β P0 , (3.3.2)

where the internal parity operator P0 replaces the spatial coordinates ( x1 ,x 2 ,x 3 )

by ( − x1 , − x 2 , − x 3 ) . Some authors add a phase factor to P0 . Due to (2.1.6) the
quadrate of P meets

P 2 = β P0 2 = 1.
2

Consequently, eigenfunctions of P have the eigenvalues +1 or -1, which mark

states of positive or negative parity. We investigate the commutator

∂
[H,P ] = −c ∑ ⎡⎢α i ⋅ i= ⎤
3
,P ⎥
i =1 ⎣ ∂x i
⎦
3
⎛ ∂ ∂ ⎞
= −ci=∑ ⎜ α i β i P0 − βα i P0 i ⎟ (3.3.3)
i =1 ⎝ ∂x ∂x ⎠

( ∂
)
3
= −ci=∑ α i β + βα i P0 = 0,
i =1 ∂x i

∂ ∂
where we have used (2.1.6) and P0 = − i P0 . Due to the vanishing
∂x i
∂x
commutator (3.3.3) there are eigenfunctions of the Hamiltonian which are also
eigenfunctions of P with a definite parity.
 44

K
b) In quantum mechanics the operator J of the total angular momentum appears
as follows
K K K
J = L+S. (3.3.4)
K
The spin operator S is given in (3.1.29), and from section 2.4 we know that the
K
angular momentum operator L reads
K K K
L = r ×p. (3.3.5)

We now calculate
K K K
⎡⎣H , J ⎤⎦ = ⎡⎣H , L ⎤⎦ + ⎡H ,S ⎤ . (3.3.6)
⎣ ⎦
K
First we take the x-term of ⎡⎣H , L ⎤⎦ , i.e.

⎡ 3
⎤
[H ,Lx ] = ⎡⎣H ,ypz − zpy ⎤⎦ = ⎢c ∑ α i p i + β m0c 2 + V ( r ),ypz − zpy ⎥ (3.3.7)
⎣ i =1 ⎦

Obviously, the term β m0c 2 does not contribute. The spherical potential yields in
(3.3.7)

⎡⎣V ( r ) , Lx ⎤⎦Ψ = ⎡⎣V ( r ) ,ypz − zpy ⎤⎦Ψ

⎛ ∂ ∂ ⎞ ⎛ ∂ ∂ ⎞
= −i=V ( r ) ⎜ y − z ⎟Ψ + i= ⎜ y − z ⎟V ( r )Ψ .
⎝ ∂z ∂y ⎠ ⎝ ∂z ∂y ⎠

Taking into account r = x 2 + y 2 + z 2 and

⎛ ∂ ∂ ⎞ ⎛ ∂V ( r ) ∂r ∂V ( r ) ∂r ⎞ ⎛ ∂ ∂ ⎞
⎜y − z ⎟V ( r )Ψ = ⎜ y −z ⎟Ψ + V ( r ) ⎜ y − z ⎟Ψ
⎝ ∂z ∂y ⎠ ⎝ ∂r ∂z ∂r ∂y ⎠ ⎝ ∂z ∂y ⎠
∂V ( r ) ⎛ 2z 2y ⎞ ⎛ ∂ ∂ ⎞
= ⎜y −z ⎟Ψ + V ( r ) ⎜ y − z ⎟Ψ
∂r ⎝ r r ⎠ ⎝ ∂z ∂y ⎠

we obtain

⎣⎡V ( r ) ,ypz − zpy ⎦⎤ = 0 . (3.3.8)

i.e. the potential term in (3.3.7) does not contribute either.

We investigate the contribution of the first term of the Hamiltonian in (3.3.7). Since
Lx commutes with px ≡ p1 it follows
 45

⎡ 3 ⎤
⎢c ∑ α i p , Lx ⎥Ψ = c ⎡⎣α 2 p + α 3 p , Lx ⎤⎦Ψ
i ( 2) (3 )
⎣ i =1 ⎦
⎡ ∂ ∂ ∂ ∂ ⎤
= −c = 2 ⎢α 2 +α3 ,y − z ⎥Ψ
⎣ ∂y ∂z ∂z ∂y ⎦
(3.3.9)
⎡∂ ∂⎤ ⎡∂ ∂ ⎤
= −c = 2 α 2 ⎢ ,y ⎥Ψ + c =2 α 3 ⎢ ,z ⎥Ψ
⎣ ∂y ∂z ⎦ ⎣ ∂z ∂y ⎦
⎛ ∂ ∂ ⎞
= −c = 2 ⎜ α 2
⎝ ∂z
−α3 ( 2)
( (3)
⎟Ψ = ci= α 3 p − α 2 p Ψ .
∂y ⎠
)
This expression represents [H , Lx ] . Cyclic permutations in (3.3.9) generate the
analogous expressions for ⎡⎣H , Ly ⎤⎦ and [H , Lz ] .

Now, we deal with the last commutator in (3.3.6)

K ⎡ 3 K⎤ K
⎡H ,S ⎤ = ⎢c ∑ α p i ,S ⎥ + ⎡ m0c 2 β ,S ⎤ (3.3.10)
⎣ ⎦
⎣ i =1 ⎦ ⎣ ⎦
i

Using (2.1.15) and (3.1.29) we write the commutator

K K
K = ⎛ 1 0 ⎞⎛ σ 0 ⎞ = ⎛ σ 0 ⎞ ⎛ 1 0 ⎞
⎡ β ,S ⎤ = ⎜ − =0,
⎦ 2 0 −1⎟⎜ 0 σK ⎟ 2 ⎜ 0 σK ⎟ ⎜ 0 −1⎟
(3.3.11)
⎣
⎝ ⎠⎝ ⎠ ⎝ ⎠⎝ ⎠

i.e. the last commutator in (3.3.10) vanishes. We write the preceding commutator
K
with the x-component of S :

⎡3 ⎤
( ) (
c ⎢ ∑ α i p i , S x ⎥ = cp ( ) α 2 S x − S x α 2 + cp ( ) α 3 S x − S x α 3
⎣ i =1 ⎦
2 3
)
= ( 2) ⎛⎜ ⎛ 0 σ y ⎞ ⎛σ 0 ⎞ ⎛σ x 0 ⎞⎛ 0 σ y ⎞⎞
=c p ⎜ ⎟⎜ x ⎟−⎜ ⎟⎜ ⎟⎟
2 ⎜ ⎜σ 0 ⎟ ⎝⎜ 0 σ x ⎠⎟ ⎝⎜ 0 σ x ⎠⎟ ⎜ σ y 0 ⎟⎟
⎝⎝ y ⎠ ⎝ ⎠⎠
= (3) ⎛ ⎛ 0 σ z ⎞ ⎛ σ x 0 ⎞ ⎛σ x 0 ⎞⎛ 0 σ z ⎞⎞
+c p ⎜⎜ ⎟⎜ ⎟−⎜ ⎟⎜ ⎟⎟
2 ⎜ ⎜σ 0 ⎟⎠ ⎜⎝ 0 σ x ⎟⎠ ⎜⎝ 0 σ x ⎟⎠ ⎜⎝ σ z 0 ⎟⎠ ⎟⎠
⎝⎝ z
=⎛ 2 ⎛ 0 σ yσ x − σ xσ y ⎞ ⎛ 0 σ zσ x − σ xσ z ⎞ ⎞
= c ⎜ p( ) ⎜ ⎟ + p(3) ⎜ ⎟⎟
2⎜ ⎜σ σ − σ σ ⎟ ⎜σ σ − σ σ ⎟
⎝ ⎝ y x x y
0
⎠ ⎝ z x x z
0 ⎠ ⎟⎠

=⎛ ⎛ 0 2iσ z ⎞ ⎛ 0 2iσ y ⎞ ⎞
= c ⎜ − p( ) ⎜ (3)
⎟+ p ⎜ ⎟ ⎟ i.e.
2
⎜ 0 ⎠⎟ ⎜ 2iσ 0 ⎟⎟
2⎜
⎝ ⎝ 2iσ z ⎝ y ⎠⎠

⎡3 ⎤
(
c ⎢ ∑ α i p i ,Sx ⎥ = −ci= p ( 2) α 3 − p ( 3) α 2 ,
⎣ i =1 ⎦
) (3.3.12)

where (2.1.15) and (2.1.14) have been used. Inserting (3.3.8), (3.3.9), (3.3.11) and
(3.3.12) in (3.3.6) yields
 46

[H , J x ] = 0 (3.3.13)

and by means of cyclic permutations

K
⎡⎣H , J ⎤⎦ = 0 . (3.3.14)

Finally we investigate the commutator of the operators introduced in a) and b),

K K
⎡⎣P , J ⎤⎦ . We again take the x-component of J :

[P , J x ] = ⎡⎣ β P0 ,Lx ⎤⎦ + ⎡⎣ β P0 ,Sx ⎤⎦ . (3.3.15)

⎛ ∂ ∂ ⎞ ⎛ ∂ ∂ ⎞
Since P0 ⎜ y − z ⎟ = ⎜y − z ⎟ P0 , the first commutator on the right hand
⎝ ∂z ∂y ⎠ ⎝ ∂z ∂y ⎠
= ⎛ 1 0 ⎞ ⎛σ x 0 ⎞
side of (3.3.15) vanishes. Because β Sx = ⎜ ⎟⎜ ⎟ = Sx β , the last term
2 ⎝ 0 −1⎠ ⎜⎝ 0 σ x ⎟⎠
in (3.3.15) disappears also and we have
K
[P , J x ] = 0 and ⎣⎡P , J ⎦⎤ = 0 . (3.3.16)
K
Due to (3.3.3), (3.3.14) and (3.3.16) the operators H , P and J commute mutually.
Therefore, there are energy eigenfunctions, which are simultaneously
eigenfunctions of the parity operator and of the total angular momentum operator.
This property will be used in the next section.

3.4 Separation of the variables for the Dirac equation with central potential
We are looking for the solution of the Dirac equation with the Hamiltonian (3.3.1)

⎛ 3 ⎞ K K
⎜ c ∑ α i p + β m0c + V ( r ) ⎟Ψ ( r ) = EΨ ( r ) .
i 2
(3.4.1)
⎝ i =1 ⎠
K
As in (3.2.4) or in (2.4.4) the four-spinor Ψ ( r ) is split up in two two-spinors like
this
K
K ⎛ ϕ (r ) ⎞
Ψ ( r ) = ⎜⎜ K ⎟⎟ . (3.4.2)
⎝ χ (r ) ⎠
K
In the preceding section is shown that there are energy eigenfunction Ψ ( r ) which
are also eigenfunctions of the parity operator P , i. e.
K K K K
K ⎛ ϕ ( r ) ⎞ ⎛ 1 0 ⎞ ⎛ ϕ ( −r ) ⎞ ⎛ ϕ ( −r ) ⎞ ⎛ ϕ (r ) ⎞
PΨ ( r ) = β P0 ⎜⎜ K ⎟⎟ = ⎜ ⎟ ⎜⎜ K ⎟⎟ = ⎜⎜ K ⎟⎟ = λ ⎜⎜ K ⎟⎟ (3.4.3)
⎝ χ ( r ) ⎠ ⎝ 0 −1⎠ ⎝ χ ( − r ) ⎠ ⎝ − χ ( − r ) ⎠ ⎝ χ (r ) ⎠
The eigenvalue λ is +1 or -1. From (3.4.3) follows that
 47

K K K K
for λ = +1 : ϕ ( −r ) = ϕ ( r ) , − χ ( − r ) = χ ( r )
K K K K (3.4.4)
and for λ = −1 : ϕ ( −r ) = −ϕ ( r ) , − χ ( −r ) = − χ ( r )
K
i.e. the internal parity of ϕ ( r ) corresponds to λ (it is positive if λ = +1 etc.). The
K K
spinors ϕ ( r ) and χ ( r ) have opposite internal parities.
K
On the other hand, (3.3.14) reveals that Ψ ( r ) is an eigenfunction of J z and J 2 . As
in quantum mechanics the corresponding eigenvalues read =m and = 2 j ( j + 1) .
K
This is true for each of the four components of Ψ ( r ) .

In atomic and nuclear physics such functions are formed by coupling spherical
K
harmonics Yl ,ml ( r / r ) with spin eigenfunctions xs ,ms . As we have seen in section
3.1, Dirac particles have spin s = ½. The coupling is performed by means of
Clebsch Gordan coefficients ( l ml s ms | jm ) resulting in the function
K K
Ω jml ( r /r ) = ∑ ( l m ,s = 1/ 2,m
ml ms
l s | j m )Yl ml ( r /r ) x1/ 2 ms . (3.4.5)

The spherical harmonics are eigenfunctions of the angular momentum operators

Lz and L2 ,cf. (3.3.5), with the eigenvalues =ml and = 2l ( l + 1) respectively and
with ml ≤ l . Analogously the spin operators Sz and S 2 ,cf. (2.4.16), act on the
spin functions x1 / 2,ms and meet Sz x1 / 2,ms = =ms x1 / 2 ,ms and
1⎛ 1 ⎞ 2
S 2 x1 / 2 ,ms =
⎜ + 1⎟ = x1 / 2,ms . Because ms can only be ½ or -½ , there are just two
2⎝2 ⎠
spin functions, which can be written as spinors

⎛ 1⎞ ⎛0⎞
x 1 / 2,1 / 2 = ⎜ ⎟ , x 1 / 2 ,−1 / 2 = ⎜ ⎟ . (3.4.6)
⎝0⎠ ⎝ 1⎠

The Clebsch Gordan coefficients are real, fulfil l − s ≤ j ≤ l + s, ml + ms = m , and

read

j +m
(l = j − 1 / 2,ml = m − 1 / 2,1 / 2,ms = 1 / 2 | jm ) =
2j
j −m
(l = j − 1 / 2,ml = m + 1 / 2,1 / 2,ms = −1 / 2 | jm ) =
2j
(3.4.7)
j − m +1
( l = j + 1 / 2,ml = m − 1 / 2,1 / 2,ms = 1 / 2 | jm ) = −
2j + 2
j + m +1
(l = j + 1 / 2,ml = m + 1 / 2,1 / 2,ms = −1 / 2 | jm ) = .
2j + 2

Inserting (3.4.7) and (3.4.6) into (3.4.5) yields the two-spinors

 48

K
K 1 ⎛⎜ j + m Yl = j −1 / 2,ml =m −1 / 2 ( r / r ) ⎞⎟
Ω jm,l = j −1 / 2 ( r / r ) = K , (3.4.8)
2 j ⎜ j − m Yl = j −1 / 2 ,m =m +1 / 2 ( r / r ) ⎟
⎝ l ⎠
K
K 1 ⎛⎜ − j − m + 1Yl = j +1 / 2,ml =m −1 / 2 ( r / r ) ⎞⎟
Ω jm,l = j +1 / 2 ( r / r ) = K . (3.4.9)
2 j + 2 ⎜ j + m + 1Yl = j +1 / 2 ,m =m +1 / 2 ( r / r ) ⎟
⎝ l ⎠
K
The internal parity of Ylml ( r / r ) depends on l like this
K K K
P0 Ylml ( r / r ) = Ylml ( −r / r ) = ( −1) Ylml ( r / r ) .
l
(3.4.10)

Since the l-values of the expressions in (3.4.8) and (3.4.9) differ by 1, these
functions have opposite internal parity. Therefore they are able to represent the
K K K
r / r -dependence of ϕ ( r ) and χ ( r ) set out in (3.4.4). Consequently, we assign
K
the ordinal numbers j ,m,l = j − 1 / 2 to ϕ ( r ) and the numbers j ,m,l ′ = j + 1 / 2 to
K
χ ( r ) . An inverse assignment is also possible, but it does not bring new results as
we shall see at the end of this section.
We make the following ansatz
K K
ϕ ( r ) = ig ( r ) Ω jm,l = j −1 / 2 ( r / r )
K K . (3.4.11)
χ ( r ) = −f ( r ) Ω jm,l ′= j +1 / 2 ( r / r )

In this way, the two-spinors are separated in a function dependingK on the

magnitude r and in a two-spinor depending on the direction of r . This one shows
K
the expected behaviour with respect to P and J (given in (3.4.3/4) and derived in
section 3.3) and is an eigenfunction of S 2 with the eigenvalue 1 / 2 (1 / 2 + 1) = 2 .
K
⎛ ϕ (r ) ⎞
We now look for the functions g ( r ) and f ( r ) so that ⎜⎜ K ⎟⎟ is a solution of the
⎝ χ (r ) ⎠
energy equation (3.4.1). Using (2.1.15) we write it this way

3 ⎛ 0 σ i pi ⎞ ⎛ ϕ ⎞ ⎛ 1 0 ⎞ ⎛ϕ ⎞ ⎛ϕ ⎞ ⎛ϕ ⎞
c∑ ⎜ ⎟⎜ ⎟ + ⎜ ⎟ m0c 2 ⎜ ⎟ + V ( r ) ⎜ ⎟ = E ⎜ ⎟ , (3.4.12)
⎜
i =1 σ p
⎝ i
i
0 ⎠⎟ ⎝ χ ⎠ ⎝ 0 −1⎠ ⎝χ⎠ ⎝χ⎠ ⎝χ⎠

which represents the following pair of equations

3
c ∑ σ i p i χ + m0c 2 ϕ + V ϕ = Eϕ , (3.4.13)
i =1

3
c ∑ σ i p i ϕ − m0c 2 χ + V χ = E χ . (3.4.14)
i =1

Bringing (3.4.14) in the form

 49

K K
c ∑ σ i p i ϕ ( r ) = ( E + m0c 2 − V ( r ) ) χ ( r )
3
(3.4.15)
i =1

makes clear that the expression on the left hand side of (3.4.15) is eigenfunction of
K
J 2 and J z ( J commutes with V ( r ) as shown in section 3.3) with the same value
j as χ , furthermore it has the same parity and s-value as χ . Because these
K
properties concern the angle- and spin dependent part of χ ( r ) , i.e.
K
Ω jm,l ′= j +1 / 2 ( r / r ) , the following proportionality must hold
3
K
The angle- and spin dependent part of c ∑ σ i p i ϕ ( r )
i =1 (3.4.16)
K
is proportional to Ω jm,l ′= j +1 / 2 ( r / r ) .

In order to strengthen this statement we transform

3
K 3
∂ K
i ( ( )
∑ σ i pi ϕ ( r ) = ∑ =σ i g r Ω jm,l = j −1 / 2 ( r / r ) )
i =1 i =1 ∂x
3
⎡⎛ ∂ ⎞ K ∂ K ⎤
= =∑ σ i ⎢⎜ i g ( r ) ⎟ Ω jm,l = j −1 / 2 ( r / r ) + g ( r ) i Ω jm,l = j −1 / 2 ( r / r ) ⎥ .
i= ⎣⎝ ∂x ⎠ ∂x ⎦

With the detail

∂ dg ( r ) ∂r dg ( r ) ∂ (x ) + (x ) + (x )
1 2 2 2 3 2
dg ( r ) x i
g ( r ) = = =
∂x i dr ∂x i dr ∂x i dr r
we obtain
3
K
∑σ p ϕ (r )
i =1
i
i

(3.4.17)
dg ( r ) 3 xi K 3
δ K
==
dr i =1
∑ σ i Ω jm,l = j −1 / 2 ( r / r ) + =g ( r ) ∑ σ i i Ω jm,l = j −1 / 2 ( r / r ) .
r i =1 δx

The first sum on the right hand side of (3.4.17) can be written like this with the help
of (2.1.12) and (3.4.8)

⎛ j +m K ⎞
⎜ Yl = j −1 / 2 ,ml =m −1 / 2 ( r / r ) ⎟
3
xi K 1 ⎛ x3 x1 -ix 2 ⎞ ⎜ 2j ⎟
∑σ i Ω jm,l = j −1 / 2 ( r / r ) = ⎜
r ⎝ x1 + ix 2 −x3 ⎠ ⎜
⎟ ⎟
i =1 r j −m K
⎜
⎜ Yl = j −1 / 2,ml =m +1 / 2 ( r / r ) ⎟⎟
⎝ 2j ⎠
⎛ x3 j + m K ⎛ x1 x 2 ⎞ j − m K ⎞
⎜ Yl = j −1 / 2 ,ml = m −1 / 2 ( r / r ) + ⎜ −i ⎟ Yl = j −1 / 2 ,ml =m +1 / 2 ( r / r ) ⎟
r 2j ⎝ r r ⎠ 2j
= ⎜⎜ ⎟.
⎟
⎜ ⎛⎜ x +i x ⎞⎟ j + mYl = j −1 / 2,m =m −1 / 2 ( rK / r ) − x j −m K
1 2 3

⎜ r Yl = j −1 / 2,ml = m +1 / 2 ( r / r ) ⎟⎟
⎝⎝ r ⎠ 2 j r 2 j ⎠
l

(3.4.18)
 50

The theory of spherical harmonics says that this spinor equals

⎛ j − m +1 K ⎞
⎜− Yl = j +1 / 2,ml =m −1 / 2 ( r / r ) ⎟
2j + 2
− ⎜⎜ ⎟.
⎟ (3.4.19)
⎜ j + m + 1 Yl = j +1 / 2,m =m +1 / 2 ( r / r ) ⎟
K
⎜ 2j + 2 ⎟
⎝ ⎠
l

Due to (3.4.9) we have therefore

3
xi K K
∑σ i
i =1 r
Ω jm,l = j −1 / 2 ( r / r ) = −Ω jm,l = j +1 / 2 ( r / r ) . (3.4.20)

We now investigate the last sum in (3.4.17). First, we will show that (3.4.20) holds
3
xk
also with exchanged l-values. For that, we apply the operator ∑ σ k on both
k =1 r
sides of (3.4.20) and we obtain
3
xk 3 xi K 3
xk K
∑σ
k =1
k ∑
r i =1
σ i Ω jm,l = j −1 / 2 ( r / r ) = −∑ σ k Ω jm,l = j +1 / 2 ( r / r )
r k =1 r
(3.4.21)

In analogy with (2.4.13) we have

3
xi xk 3
( x ) =1 k 2

∑
i ,k =1
σ i
σ
r k r
= 1∑k =1 r2
(3.4.22)

and consequently

K 3
xk K
Ω jm,l = j −1 / 2 ( r / r ) = −∑ σ k Ω jm,l ′= j +1 / 2 ( r / r ) . (3.4.23)
k =1 r

We set up an expression for the last sum in (3.4.17) by applying the

3
operator ∑ σ i p i on both sides of (3.4.23).
i =1

3
K 3 3
xk K
∑
i =1
σ i p Ω jm,l = j −1 / 2 ( r / r ) = −∑ σ i p ∑ σ k Ω jm,l ′= j +1 / 2 ( r / r ) .
i

i =1
i

k =1 r
(3.4.24)

Making use of (2.1.13) and (2.1.14) we deal with the general double sum which
K K
contains arbitrary vectors A and B :
 51

∑σ
i ,k =1
i
Ai σ k Bk

( )
3
= ∑ σ i Ai Bi + σ 1σ 2 A1B2 + σ 2 σ 1A2B1 + cyclic permutations
2

i =1
3
(3.4.25)
= ∑ Ai Bi 1 + iσ 3 ( A1B2 − A2B1 ) + cyclic permutations
i =1
K K
( )
3 3
= ∑ Ai Bi 1 + i∑ σ j A × B
j
i =1 j=1

in analogy with (2.4.12). Using (3.4.25) and (1.2.7) we transform the expression
(3.4.24):
3
K
∑σ p Ω
i =1
i
i
jm,l = j −1 / 2 (r / r)
K
⎛ 3 i xi 3
⎛K r⎞ ⎞ K
= −⎜∑ p + i∑ σ j ⎜ p × ⎟ ⎟ Ω jm,l ′= j +1 / 2 ( r / r ) (3.4.26)
⎜ i =1 r ⎝ ⎟
r ⎠j ⎠
⎝ j=1

3
δ xi K 1 3 K
= i=∑ i Ω jm,l ′= j +1 / 2 ( r / r ) + i ∑ σ i Li Ω jm,l ′= j +1 / 2 ( r / r ) ,
i =1 δ x r r i =1
K K K K K
where L = r × p = − p × r , as per (3.3.5), has been inserted. We continue
transforming (3.4.26):
3
K i 3
K
∑ σ i pi Ω jm,l = j −1 / 2 ( r / r ) =
i =1 r
∑σ L Ω
i =1
i i jm,l ′= j +1 / 2 (r / r)

3 K 3
x i 2x i K 3
xi δ K
+i= Ω jm,l ′= j +1 / 2 ( r / r ) + i=∑ Ω jm,l ′= j +1 / 2 ( r / r ) + i= ∑ Ω jm,l ′= j +1 / 2 ( r / r )
i =1 −2r i =1 r δ x
3 i
r

xi ∂ 3
The last term vanishes by the following reason: the sum ∑
i =1 r ∂x
i
is the scalar
K K
product of the unity position vector r / r and the gradientK ∇ . Therefore, it is the
projection of the gradient vector onto the direction of r and it results in the
δ
component of the gradient. Because Ω jml ′ ( r ′ / r ) depends only on directions
δr
∂ K
and not on r, there results Ω jml ′ ( r / r ) = 0 . Finally we have
∂r
3
K i⎡ 3
⎤ K
∑ σ i pi Ω jm,l = j −1 / 2 ( r / r ) = r ⎢2= + ∑ σ i Li ⎥ Ω jm,l ′= j +1/ 2 ( r / r ) .
i =1 ⎣ i =1 ⎦
(3.4.27)

In order to transform the last summand in (3.4.27) we use (3.3.4) with the two-
spinor operator (2.4.16) and form

K 2 ⎛ K = K ⎞2 K2 ⎛ = K ⎞2 K K
J = ⎜ L + σ ⎟ = L + ⎜ σ ⎟ + =σ ⋅ L . (3.4.28)
⎝ 2 ⎠ ⎝2 ⎠

We obtain
 52

K K 3 ⎛ K 2 K2 ⎛ = K ⎞2 ⎞
=σ ⋅ LΩ jml ′ ≡ =∑ σ i Li Ω jml ′ = ⎜ J − L − ⎜ σ ⎟ ⎟ Ω jml ′
⎜ ⎝ 2 ⎠ ⎟⎠
i =1 ⎝ (3.4.29)
⎡ 3⎤ ⎛ 3⎞
= ⎢ j ( j + 1) − l ′ ( l ′ + 1) − ⎥ =2 Ω jml ′ = ⎜ − j − ⎟ =2 Ω jml ′ ,
⎣ 4⎦ ⎝ 2⎠

where we have set the eigenvalues mentioned in context with (3.4.5) and the
relation l ′ = j + 1 / 2 (see (3.4.11)). Inserting (3.4.29) in (3.4.27) yields
3
i= ⎛ 3⎞ i= ⎛ 1⎞
∑σ p Ω
i =1
i
i
jm,l = j −1 / 2 = ⎜ 2 − j − ⎟ Ω jm,l ′= j +1 / 2 = − ⎜ j − ⎟ Ω jm,l ′= j +1 / 2 .
r ⎝ 2⎠ r ⎝ 2⎠
(3.4.30)

We insert (3.4.20) and (3.4.30) into (3.4.17)

3
K
∑σ p ϕ (r )
i =1
i
i

(3.4.31)
dg ( r ) K g (r ) K
= −= Ω jm,l ′=+1 / 2 ( r / r ) + = ( j − 1 / 2 ) Ω jm,l ′=+1 / 2 ( r / r ) ,
dr r
which confirms the hypothesis (3.4.16). Analogously one derives
3
K
∑σ p χ (r )
I =1
i
i

(3.4.32)
df ( r ) K f (r ) ⎛ 3⎞ K
= −i= Ω jm,l = j −1 / 2 ( r / r ) + i= ⎜ − j − ⎟ Ω jm,l = j −1 / 2 ( r / r ) .
dr r ⎝ 2⎠

Now, the expressions (3.4.11) and (3.4.32) are inserted in (3.4.13) and (3.4.31)
into (3.4.14). The angular spinors Ω can be omitted. We obtain the following
differential equations for the radial functions f and g

df ( r ) f (r ) ⎛ 3⎞
−i=c − i=c ⎜ j + ⎟ = i ( E − m0c − V ( r ) ) g ( r )
2

dr r ⎝ 2⎠
(3.4.33)
dg ( r ) g (r ) ⎛ 1⎞
− =c +=c ⎜ j − ⎟ = − ( E + m0c − V ( r ) ) f ( r ) .
2

dr r ⎝ 2⎠

With the substitution

G ( r ) = rg ( r ) and F ( r ) = rf ( r ) i.e. with

dG ( r ) dg ( r ) dF ( r ) df ( r ) (3.4.34)
= g (r ) + r and = f (r ) + r ,
dr dr dr dr

one finally gets the radial equations

dF ( r ) F (r )
=c − =cκ + ( E − m0c 2 − V ( r ) ) G ( r ) = 0
dr r
(3.4.35)
dG ( r ) G (r )
=c + =cκ − ( E + m0c 2 − V ( r ) ) F ( r ) = 0 ,
dr r
 53

where we have written κ instead of − ( j + 1 / 2 ) as usual in the literature.

If one makes an inverse choice for l and l ′ (see (3.4.11) and the preceding
paragraph) i.e. if one takes l = j + 1 / 2 and l ′ = j − 1 / 2 , as one can show, in
(3.4.35) only κ changes sign. In the next section we will see that the energy of the
bound states depends on κ and therefore it is independent of the relative choice
of l and l ′ . We stress

κ = j + 1/ 2 . (3.4.36)

3.5 Solution of the radial equations for a Dirac particle in a Coulomb field
We solve the coupled radial equations for a Dirac particle in a Coulomb potential
and determine the energy values for the bound states. The potential is given by a
point nucleus of charge Ze and a particle of charge −e and reads

Ze 2 Ze 2 As
V (r ) = − ≡− with ε 0 = 8.8542 ⋅ 10 −12
4πε 0 r r Vm
. (3.5.1)
 e2
and e ≡ 2

4πε 0

We insert it in (3.4.35) this way

dG ( r ) G ( r ) ⎛ m0c 2 + E Ze 2 ⎞
= −κ +⎜ + ⎟ F (r )
dr r ⎝ =c =cr ⎠
(3.5.2)
dF ( r ) F ( r ) ⎛ m0c 2 − E Ze 2 ⎞
=κ +⎜ − ⎟G (r ) .
dr r ⎝ =c =cr ⎠

We use the substitutions

α1 = ( m0c 2 + E ) / =c, α 2 = ( m0c 2 − E ) / =c, α = e 2 / =c, (3.5.3)

where α ≅ 1 / 137 is the fine structure constant, and obtain

dG ( r ) G (r ) ⎛ Zα ⎞
= −κ + ⎜ α1 + ⎟ F (r )
dr r ⎝ r ⎠
(3.5.4)
dF ( r ) F (r ) ⎛ Zα ⎞
=κ + ⎜α2 − ⎟G (r ) .
dr r ⎝ r ⎠

We introduce the variable

d d
ρ = α1α 2 r with = α1α 2 , (3.5.5)
dr dρ

and define G ( ρ ) ≡ G ( r ( ρ ) ) , F ( ρ ) ≡ F ( r ( ρ ) ) ,
 54

dG ( ρ ) G(ρ) ⎛ Zα ⎞
α1α 2 = −κ α1α 2 + ⎜ α1 + α1α 2 F (ρ )
ρ ⎟⎠
i.e.
dρ ρ ⎝
dF ( ρ ) F (ρ) ⎛ Zα ⎞
α1α 2 = κ α1α 2 + ⎜ α 2 − α1α 2 G(ρ )
dρ ρ ⎝ ρ ⎟⎠

dG ( ρ ) G(ρ) ⎛ α1 Zα ⎞
or = −κ + ⎜⎜ + ⎟F (ρ )
dρ ρ ⎝ α2 ρ ⎟⎠
(3.5.6)
dF ( ρ ) F (ρ) ⎛ α 2 Zα ⎞
=κ + ⎜⎜ − ⎟G ( ρ ) .
dρ ρ ⎝ α1 ρ ⎟⎠

With the transformation

F ( ρ ) = F ( ρ ) e − ρ and G ( ρ ) = G ( ρ ) e − ρ (3.5.7)

we have

dG ( ρ )  G ( ρ ) ⎛ α1 Zα ⎞ 
= G(ρ) −κ + ⎜⎜ + ⎟F (ρ ) , (3.5.8)
dρ ρ ⎝ α2 ρ ⎟⎠

dF ( ρ )  F ( ρ ) ⎛ α 2 Zα ⎞ 
= F (ρ) + κ + ⎜⎜ − ⎟G ( ρ ) . (3.5.9)
dρ ρ ⎝ α1 ρ ⎟⎠

The following series are helpful

∞ ∞
F ( ρ ) = ρ s ∑ am ρ m , G ( ρ ) = ρ s ∑ bm ρ m with a0 ≠ 0, b0 ≠ 0 . (3.5.10)
m =0 m =0

Inserting them in (3.5.8) and writing only the summands with the factor ρ s + μ −1 ,
where μ is an arbitrarily chosen m-value, we obtain

" + ( s + μ ) bμ ρ s + μ −1 + "
α1
= " + bμ −1ρ s + μ −1 " + " − κ bμ ρ s + μ −1 + " + a ρ s + μ −1 + " + Zα aμ ρ s + μ −1
α 2 μ −1

i.e., the coefficients of ρ s + μ −1 meet the following equation

α1
−Zα aμ − a + ( s + μ + κ ) bμ − bμ −1 = 0 . (3.5.11)
α 2 μ −1

Analogously, equation (3.5.9) results in

α2
( s + μ − κ ) aμ − aμ −1 + Zα bμ − b = 0. (3.5.12)
α1 μ −1

Since we presuppose a−1 = b−1 = 0 , for μ = 0 we have

 55

−Zα a0 + ( s + κ ) b0 = 0,
( s − κ ) a0 + Zα b0 = 0.
To obtain a non-trivial solution the determinant ( Zα ) + s 2 − κ 2 must vanish, i.e.
2

s = ± κ 2 − ( Zα ) .
2
(3.5.13)
K
⎛ ϕ (r ) ⎞
We investigate the sign of s. The wave function ⎜ K ⎟ must be normalizable i.e.
⎝ χ (r ) ⎠

(∫ ϕ ( rK )
V
2 K 2
)
+ χ ( r ) dV = finite, even in a volume near r = 0,

that is, due to (3.4.11) and (3.4.34)

a ⎛ ⎛ G ( r ) ⎞2 ⎛ F ( r ) ⎞2 ⎞
∫0 ⎜⎜ − ⎜⎝ r ⎟⎠ + ⎜⎝ r ⎟⎠ ⎟⎟ r dr = finite, or
2

⎝ ⎠
a α1α 2

∫ ( −b + a02 ) ρ 2s d ρ = ( −b02 + a02 )

1 α1α 2
2
ρ 2s +1 |0a = finite
2s + 1
0
0

1
i.e. 2s + 1 ≥ 0 or s ≥ − .
2

1
It can be shown (Greiner, 1990, p.179) that the range of values 0 ≥ s ≥ − is also
2
excluded. Therefore,

s = + κ 2 − ( Zα )
2
(3.5.14)

holds. Now, we multiply (3.5.11) by α 2 and (3.5.12) by α1 and subtract this

equation. We obtain

⎡( s + μ + κ ) α 2 − Zα α1 ⎤ bμ = ⎡( s + μ − κ ) α1 + Zα α 2 ⎤ aμ . (3.5.15)
⎣ ⎦ ⎣ ⎦

When μ is very large, we have

α 2 bμ ≅ α1aμ . (3.5.16)

Inserting eq. (3.5.16) in ( 3.5.11) and taking into account μ ≥ s + κ results in

α1 α1 α1
μ aμ − Zα aμ ≅ aμ −1 + a ,
α2 α2 α 2 μ −1

α1
which yields for μ  Zα the relations
α2
 56

2 2
aμ ≅ aμ −1 and b μ ≅ bμ −1 . (3.5.17)
μ μ

Due to (3.5.10) and (3.5.17) the function F ( ρ ) can be written for large m's like this

⎛ 2 4 ⎞
F ( ρ ) ≅ " + ρ s ⎜⎜ am ρ m + am ρ m +1 + am ρ m + 2 + " ⎟⎟ ,
⎝ m +1 ( m + 1)( m + 2 ) ⎠

which behaves asymptotically like

2ρ
⎛ 2 ρ ( 2 ρ )2 ( 2ρ )
n
⎞
ρ a0e
s
= ρ a0 ⎜ 1 +
s
+ +"+ + "⎟ .
⎜ 1 1⋅ 2 n! ⎟
⎝ ⎠

Consequently, the asymptotic behaviour of

F (r ) F ( ρ ) F ( ρ ) − ρ
f (r ) = = α1α 2 = e α1α 2 (3.5.18)
r ρ ρ

is like ρ s −1e ρ , which is not normalizable. Therefore, the series (3.5.10) must stop
for a certain value m = ν , i.e.

aν +1 = bν +1 = 0 . (3.5.19)

(Now, f ( r ) is normalizable because of the factor e − ρ in (3.5.18)). We write

equation (3.5.11) with μ = ν + 1, i.e.

α1
−Zα aν +1 − a + ( s + ν + 1 + κ ) bν +1 − bν = 0 .
α2 ν

Due to (3.5.19) this equation becomes

α1
− a = bν for ν = 0,1, 2," , (3.5.20)
α2 ν

which we insert in (3.5.15) with μ = ν like this

α1
− ⎡⎣( s + ν + κ ) α 2 − Zα α1 ⎤⎦ a = ⎡( s + ν − κ ) α1 + Zα α 2 ⎤⎦ aν , (3.5.21)
α2 ν ⎣

2E
i.e. 2 α1α 2 ( s + ν ) = Zα (α1 − α 2 ) = Zα , (3.5.22)
=c
where (3.5.3) is applied. The equation (3.5.22) is squared and reads

(m c 2
0
4
− E 2 ) ( s + ν ) = E 2Z 2α 2 or
2

1 1 ⎛ Z 2α 2 ⎞ (3.5.23)
= ⎜ 1 + ⎟.
E2 m02c 4 ⎜⎝ ( s + ν )2 ⎟⎠
 57

We define the principal quantum number n using (3.4.36)

n ≡ ν + κ ≡ ν + j + 1 / 2, n = 1, 2," . (3.5.24)

By means of (3.5.24) and (3.5.14) equation (3.5.23) reads

1
−
⎡ ⎤ 2

⎢ ( Zα ) ⎥
2

E = m0c ⎢1 +
2
2⎥
. (3.5.25)
⎢
⎢⎣ (
n − κ + κ 2 − ( Zα )
2
) ⎥
⎥⎦

e 2 1
Because of α ≡ ≅ (cf.(3.5.3)) for small Z's we have
=c 137

( Zα )
2
1 (3.5.26)

and when we neglect this quantity relative to κ 2 , equation (3.5.25) becomes

−1 / 2
⎡ ( Zα )2 ⎤
E ≅ m0c ⎢1 +
2
⎥
⎢⎣ n 2 ⎥⎦

and is nearly independent of j = κ − 1 / 2 for a given n. Therefore, in this case j

has only a little influence on E and produces only a fine structure.
However, this fine structure is worth investigating. With the help of the
approximations for a small quantity x

x 3x 2 x
(1 + x ) , (1 − x ) ≅ 1 + 2 x and (1 − x )
−1 / 2 −2 1/ 2
≅ 1− + ≅ 1−
2 8 2 (3.5.27)
for x  1
we write (3.5.25) this way
 58

1
−
⎡ ⎤ 2

⎢ ⎥
⎢ ( Zα )
2 ⎥
E ≅ m0c 2 ⎢1 + ⎥
⎢ ⎛ Zα ⎞ ⎥
2
⎞
⎢ ⎜ n − κ + κ ⎜1− ( ) ⎟ ⎟ ⎥
⎛ 2

⎢ ⎜ ⎜ 2κ 2 ⎟⎠ ⎟ ⎥
⎣ ⎝ ⎝ ⎠ ⎦
⎡ ⎤
⎢ ⎥
⎢ ( α ) ⎥
2
α
4
Z 3 ⎛ Z ⎞
≅ m0c 2 ⎢1 − + ⎜ ⎟ ⎥
κ ⎛ Zα ⎞2 ⎞ 8 ⎝ n ⎠ ⎥
2
⎢ ⎛
⎢ 2 ⎜⎜ n − 2 ⎜ κ ⎟ ⎟⎟ ⎥
⎣⎢ ⎝ ⎝ ⎠ ⎠ ⎦⎥
⎡ ( Zα )2 ⎛ ( Zα )2 ⎞ 3 ⎛ Zα ⎞ 4 ⎤
≅ m0c ⎢1 − 2
⎜1+ ⎟+ ⎥
⎢⎣ 2n 2 ⎜⎝ κ n ⎟⎠ 8 ⎜⎝ n ⎟⎠ ⎥
⎦
⎡ ( Zα )2 ( Zα )4 ⎛ n 3 ⎞⎤
≅ m0c ⎢1 − 2
− ⎜ − ⎟⎥ ,
⎢⎣ 2n 2 2n 4 ⎝ j + 1 / 2 4 ⎠ ⎥⎦ (3.5.28)

where we have inserted κ = j + 1 / 2 from (3.4.36). The last expression is the

famous Sommerfeld structure formula. The ionisation energy reads

⎡ ( Zα )2 ( Zα )4 ⎛ n 3 ⎞⎤
Eion. = m0c − E ≅ m0c ⎢
2
+ 2
⎜ − ⎟⎥ . (3.5.29)
⎢⎣ 2n
2
2n 4 ⎝ j + 1 / 2 4 ⎠ ⎥⎦

The first term on the right hand side of (3.5.29) is

( Zα )
2
Z 2e 4 Z 2e 4
m0c 2
= m0 = m
2=2n 2 ( 4πε 0 )
0
2n 2 2= 2 n 2 2
(3.5.30)
(using (3.5.3) and (3.5.1))

and agrees with the Schrödinger- or Bohr-formula.

Table 3.5.1. Ionisation energies of the hydrogen atom (Z = 1). The energies were
e 2
evaluated with m0c 2 = 511.0041 keV and α = = 1/137.036 by means of eqs.
=c
(3.5.30) and (3.5.25).
 59

main total nonrel. relativistic

quantum angular ionisation ionisation
number momentum energy/eV energy/eV
n j eq. (3.5.30) eq. (3.5.25)

1 1/ 2 13.60583 13.60601
2 1/ 2 3.40146 3.40151
2 3/ 2 " 3.40147
3 1/ 2 1.51176 1.551178
3 3/ 2 " 1.551177
3 5/ 2 " 1.551176

The total energy splitting of fine structure levels, ΔEfine , for a given n is determined
by the extreme values of j +1/2. The minimal value is ½ + ½ = 1 and the maximum
is given by n = ν + j + 1 / 2 (see (3.5.24)). I.e. n = j + 1 / 2 holds for ν = 0 .
Therefore, we have

( Zα )
4

ΔEfine = E j +1 / 2=n − E j +1 / 2=1 = m0c 2

( n − 1) (3.5.31)
2n 4
in satisfactory agreement with experimental data of the first half of the twentieth
century. This was a further success of the Dirac theory. To show the order of
magnitude of the fine structure splitting we calculate

ΔEfine ( Zα )
2

≅ ( n − 1) .
Eion. n2

ΔEfine
With n = 2 and Z = 1 we have ≅ 1.33 ⋅ 10−5 in agreement with table 3.5.1.
Eion.

In 1947 precise measurements revealed further splittings of the hydrogen levels,

i.e. the hyperfine structure, and small systematic shifts (Lamb shift). If one takes
into account the coupling between the proton and the magnetic moment of the
electron (hyperfine structure) and the interaction of the electron with the vacuum
fluctuations of the electromagnetic field (Lamb shift), the correspondence between
the calculated and the measured data is excellent.

3.6 Massless Dirac particles

Free, massless Dirac particles are described by equation (3.1.1) by putting
m0 = 0 :
K
∂Ψ ( r ,t ) 3
K K
i= = c ∑ α i p iΨ ( r ,t ) ≡ Hf ,m0 =0Ψ ( r ,t ) . (3.6.1)
∂t i =1

Due to (3.1.18) and (3.1.16) the solution reads

 60

⎛ ϕ0 ⎞
⎜ 3 ⎟ KK
K N ⎜ ⎟ ei( pr −Et ) / = .
Ψ ( r ,t ) = ∑ σip i
(3.6.2)
⎜
V i =1 ⎟
⎜⎜ ϕ 0 ⎟⎟
⎝ p ⎠

According to (3.1.24), the Hamilton operator Hf commutes with the helicity

operator Sp . Therefore, solutions exist which are simultaneous eigenfunctions of
Hf and Sp , which we form as follows. As in (3.1.27) and (3.1.28) we put the z-axis
K
in the direction of p , use (3.1.16) and obtain the eigenfunctions

⎛ 1⎞
⎜ ⎟
Ψ E ,p,m =
N ⎜ 0 ⎟ ei( pz −Et ) / = ,
0 ,+
V ⎜ 1⎟
⎜⎜ ⎟⎟
⎝0⎠
(3.6.3)
⎛0⎞
⎜ ⎟
Ψ E ,p,m =
N ⎜ 1 ⎟ ei( pz −Et ) / =
0 ,−
V ⎜0⎟
⎜⎜ ⎟⎟
⎝ −1⎠

with the helicity eigenvalues + = / 2 and − = / 2 respectively. Furthermore, we

have found in (3.3.3) that the parity operator P , (3.3.2), commutes also with H .
However, it does not commute with Sp , which is proven this way

⎡ ⎛ 3 ⎞ ⎤
⎜ ∑σ i p
i
⎢ 0 ⎟ ⎥
=
⎡⎣Sp , P ⎤⎦ = ⎢ ⎜ i =1 ⎟ , β P0 ⎥
⎢2 p ⎜ 3
⎟ ⎥
⎢ ⎜ 0
⎝
∑ σ i pi ⎟
⎠
⎥
⎣ i =1 ⎦
= ⎛ 3 ⎛σ i 0 ⎞⎛ 1 0 ⎞ i 3 ⎛1 0 ⎞ ⎛σ i 0⎞ ⎞
= ⎜ ∑⎜ ⎟⎜ ⎟ p P0 − ∑ ⎜ ⎟⎜ ⎟ P0 p ⎟
i
(3.6.4)
2 p ⎜⎝ i =1 ⎜⎝ 0 σ i ⎟⎠ ⎝ 0 −1⎠ i =1 ⎝ 0 −1⎠ ⎜ 0
⎝ σ i ⎟⎠ ⎟
⎠
= ⎛σ i 0 ⎞ i = ⎛σ i 0 ⎞ i
⎟ ( p P0 − P0 p ) =
3 3
=
2p
∑ ⎜
⎜
i =1 ⎝ 0
⎟
−σ i ⎠
i

2p
∑ ⎜⎜ 0 ⎟ p P ⋅2 ≠ 0
−σ i ⎟⎠
i =1 ⎝

Therefore, the helicity eigenfunctions (3.6.3) are not eigenfunctions of P , which

can be checked directly using (3.3.2). Consequently, these states have not a
definite parity.
In reality neutrinos show these properties. The measured mass of the neutrino is
much less than a thousandth of the electron rest mass. Therefore, it's reasonable
to treat them as massless quantas. Because neutrinos have spin ½, the Dirac
equation for m0 = 0 should be the fundamental equation for the neutrinos.

The antineutrino ν is emitted during the β − decay in the reaction

 61

n → p + e− + ν ,

and the neutrino ν occurs in β + decays.

Detailed experimental analyses have shown that the spin of the neutrino is
antiparallel to the momentum direction, i.e. it looks backward. Therefore, we can
describe it by means of Ψ E ,p,m0 ,− , (3.6.3), which is a helicity eigenfunction with
eigenvalue − = / 2 . In accordance with our predictions the neutrino has no definite
parity. Namely, measurements reveal that in weak interactions parity violation
occurs.
Analogously, experimentally, the spin of the antineutrio has been found to be
parallel to the momentum, i.e. the helicity is well defined, and one has detected
also, that the parity is not definite, as we expect.
The ability to describe the relation between definite helicity and undefined parity of
neutrinos is an additional highlight of the Dirac theory.

3.7 Particles with arbitrary spin

Here we construct wave functions which describe particles with spin 1, 3/2, 2, ...
out of the Dirac equation, and we study, by what kind of wave equation they are
generated.
We start with wave functions for free Dirac particles used
K in (3.1.27) and (3.1.28),
where the z-axis has the direction of the momentum p :

⎛ 1 ⎞
⎜ ⎟
⎜ 0 ⎟
K N
Ψ + ( r ,t,p ) = ⎜ cp ⎟ ei( pz −Et ) / =
V ⎜ ⎟
⎜ m0c + E ⎟
2

⎜ ⎟
⎝ 0 ⎠
(3.7.1)
⎛ 0 ⎞
⎜ ⎟
⎜ 1 ⎟
K N
Ψ − ( r ,t,p ) = ⎜ 0 ⎟ ei( pz −Et ) / = .
V ⎜ ⎟
⎜ −cp ⎟
⎜ m c2 + E ⎟
⎝ 0 ⎠

As shown at the end of section 3.1, they are eigenfunctions of the spin operators
Sz and S 2 with the eigenvalues ±=s and = 2s ( s + 1) respectively with s = ½. We
concentrate on the four-component spinors

⎛ 1 ⎞ ⎛ 0 ⎞
⎜ ⎟ ⎜ ⎟
⎜ 0 ⎟ ⎜ 1 ⎟
ω + ( p ) = ⎜ cp ⎟ , ω − ( p ) = ⎜ 0 ⎟ (3.7.2)
⎜ ⎟ ⎜ ⎟
⎜ m0c + E ⎟ ⎜ −cp ⎟
2

⎜ ⎟ ⎜ m c2 + E ⎟
⎝ 0 ⎠ ⎝ 0 ⎠
 62

from (3.7.1). In order to represent a spin-s particle, we combine 2s spinors of this

kind which we mark with α , β ," ,τ . For s = 3/2, for instance, the following
combination (multispinor) is possible

⎛ 1 ⎞ ⎛ 0 ⎞ ⎛ 1 ⎞
⎜ ⎟ ⎜ ⎟ ⎜ ⎟
⎜ 0 ⎟ ⎜ 1 ⎟ ⎜ 0 ⎟
ω + ,α ( p ) ω − ,β ( p ) ω + ,γ ( p ) = ⎜ cp ⎟ ⎜ 0 ⎟ ⎜ cp ⎟ . (3.7.3)
⎜ ⎟ ⎜ ⎟ ⎜ ⎟
⎜ m0c + E ⎟ ⎜ −cp ⎟ ⎜ m0c + E ⎟
2 2

⎜ ⎟ ⎜ m c2 + E ⎟ ⎜ ⎟
⎝ 0 ⎠α ⎝ 0 ⎠β ⎝ 0 ⎠γ

We achieve symmetry with respect to α , β ,γ in sums of multispinors of this kind:

ω αβγ ( p,i = 0 ) = ω + ,α ( p ) ω + ,β ( p ) ω + ,γ ( p ) ,
ω αβγ ( p,i = 1) = ω + ,α ( p ) ω + ,β ( p ) ω − ,γ ( p ) + ω + ,α ( p ) ω − ,β ( p ) ω + ,γ ( p )
+ ω − ,α ( p ) ω + ,β ( p ) ω + ,γ ( p ) ,
(3.7.4)
ω αβγ ( p,i = 2 ) = ω + ,α ( p ) ω − ,β ( p ) ω − ,γ ( p ) + ω − ,α ( p ) ω + ,β ( p ) ω − ,γ ( p )
+ ω − ,α ( p ) ω − ,β ( p ) ω + ,γ ( p ) ,
ω αβγ ( p,i = 3 ) = ω − ,α ( p ) ω − ,β ( p ) ω − ,γ ( p ) .

Obviously, the index i gives the number of ω − ( p ) - spinors in every triple. For an
arbitrary spin s we have 2s + 1 expressions ω ( p,i ) in analogy with (3.7.4)
containing "products" with 2s four-spinors.

Returning to s = 3/2, we define the z-component of the spin operator, Sz,α ,β ,γ , as a

generalization of (3.1.29) like this

Sz,α ,β ,γ = Sz,α 1β 1γ + 1α Sz,β 1γ + 1α 1β Sz,γ . (3.7.5)

An α -type matrix ( Sz,α or 1α ) is allowed to act only on an α -spinor, a β -type

matrix on an β -spinor and so on. For an arbitrary spin s the operator Sz,α ,β ,",τ
contains 2s summands, which are composed of 2s - 1 unity matrices and of one
Sz matrix. The operator Sz,α ,β ,γ of (3.7.5) acts on the expressions (3.7.4) this way
 63

= ⎛= ⎞
Sz,α ,β ,γ ω αβγ ( i = 0 ) = ω + ,α ω + ,β ω + ,γ + ω + ,α ⎜ ω + ,β ⎟ ω + ,γ
2 ⎝2 ⎠
⎛= ⎞ 3
+ ω + ,α ω + ,β ⎜ ω + ,γ ⎟ = =ω αβγ ( i = 0 ) ,
⎝2 ⎠ 2
=
Sz,α ,β ,γ ω αβγ ( i = 1) = (ω + ,α ω + ,β ω − ,γ + ω + ,α ω + ,β ω − ,γ − ω + ,α ω + ,β ω − ,γ )
2
=
+ (ω + ,α ω − ,β ω + ,γ − ω + ,α ω − ,β ω + ,γ + ω + ,α ω − ,β ω + ,γ ) (3.7.6)
2
=
+ ( −ω − ,α ω + ,β ω + ,γ + ω − ,α ω + ,β ω + ,γ + ω − ,α ω + ,β ω + ,γ )
2
=
= ω αβγ ( i = 1) ,
2
=
Sz,α ,β ,γ ω αβγ ( i = 2 ) = − ω αβγ ( i = 2 ) ,
2
3
Sz,αβγ ω αβγ ( i = 3 ) = − =ω αβγ ( i = 3 ) ,
2
where the parameter p was omitted. Generally (with arbitrary spin s) one obtains,
instead of (3.7.6),

Sz,α ,β ,",τ ω α ,β ,",τ ( p,i ) = ( s − i ) =ω α ,β ,",τ ( p,i ) , (3.7.7)

which is fulfilled by (3.7.6). Relation (3.7.7) comprises 2s + 1 eigenvalue equations

with eigenvalues s =, ( s − 1) =, " , ( −s + 1) =, − s= .

Now, we will show that the expressions ω α ,β ,",τ ( p,i ) are also eigenfunctions of S 2
with the well-known eigenvalues s ( s + 1) = 2 . We again restrict ourselves on s =3/2
and write explicitly

Sα2 ,β ,γ = Sx2,α ,β ,γ + Sy2,α ,β ,γ + Sz,2 α ,β ,γ . (3.7.8)

Omitting the unity matrices, we can write

Sx2,α ,β ,γ = ( Sx ,α + Sx ,β + Sx ,γ )( Sx ,α + Sx ,β + Sx ,γ )
(3.7.9)
= Sx2,α + Sx2,β + Sx2,γ + 2Sx ,α Sx ,β + 2Sx ,α Sx ,γ + 2Sx ,β Sx ,γ .

Due to (3.1.29) and (2.1.13) we have

⎛0 1 0 0⎞
⎜ ⎟
= ⎜1 0 0 0⎟ =2
Sx = and Sx =
2
⋅ 1 = Sy 2 = Sz 2 . (3.7.10)
2 ⎜0 0 0 1⎟ 4
⎜⎜ 0 ⎟
⎟
⎝ 0 1 0⎠

Provisionally we concentrate on the multispinor ω α ,β ,γ ( p,i = 0 ) and apply S-

operators of (3.7.9) on it. For example:
 64

Sx ,α Sx ,β ω α ,β ,γ ( p,i = 0 )
⎛ 1 ⎞ ⎛ 1 ⎞ ⎛ 1 ⎞
⎛0 1 0 0⎞ ⎜ ⎟ ⎛0 1 0 0⎞ ⎜ ⎟ ⎜ ⎟
2 ⎜ ⎟ ⎜ ⎟ ⎜1 ⎟
0 0 0
= 1 0 0 0⎟ 0 0 0⎟ ⎜ ⎟ ⎜ ⎟
= ⎜ ⎜ cp ⎟ ⎜ ⎜ cp ⎟ ⎜ cp ⎟
4 ⎜0 0 0 1⎟ ⎜ ⎟ ⎜0 0 0 1⎟ ⎜ ⎟ ⎜ ⎟
⎟ ⎜ m0c + E ⎟ ⎜⎜ ⎟ ⎜ m0c + E ⎟ ⎜ m0c + E ⎟
2 2 2
⎜⎜ ⎟ ⎟
⎝0 0 1 0 ⎠α ⎜ ⎟ ⎝0 0 1 0 ⎠β ⎜ ⎟ ⎜ ⎟
⎝ 0 ⎠α ⎝ 0 ⎠β ⎝ 0 ⎠γ
⎛ 0 ⎞ ⎛ 0 ⎞ ⎛ 1 ⎞
⎜ ⎟ ⎜ ⎟ ⎜ ⎟
2 ⎜
1 ⎟ ⎜ 1 ⎟ ⎜ 0 ⎟
=
= ⎜ 0 ⎟ ⎜ 0 ⎟ ⎜ cp ⎟ ,
4⎜ ⎟ ⎜ ⎟ ⎜ ⎟
⎟ ⎜ m0c + E ⎟
2
⎜ cp ⎟ ⎜ cp
⎜ m c2 + E ⎟ ⎜ m c2 + E ⎟ ⎜ ⎟
⎝ 0 ⎠α ⎝ 0 ⎠β ⎝ 0 ⎠γ

Sx2,α ,β ,γ ω α ,β ,γ ( p,i = 0 )
⎛ 0 ⎞ ⎛ 0 ⎞ ⎛ 1 ⎞
⎜ ⎟ ⎜ ⎟ ⎜ ⎟
⎜ 1 ⎟ ⎜ 1 ⎟ ⎜ 0 ⎟
3 2 2
= = ω α ,β ,γ ( p,i = 0 ) + =2 ⎜ 0 ⎟ ⎜ 0 ⎟ ⎜ cp ⎟
4 4 ⎜ ⎟ ⎜ ⎟ ⎜ ⎟
⎟ ⎜ m0c + E ⎟
2
⎜ cp ⎟ ⎜ cp
⎜ m c2 + E ⎟ ⎜ m c2 + E ⎟ ⎜ ⎟
⎝ 0 ⎠α ⎝ 0 ⎠β ⎝ 0 ⎠γ
⎛ 0 ⎞ ⎛ 1 ⎞ ⎛ 0 ⎞
⎜ ⎟ ⎜ ⎟ ⎜ ⎟
⎜ 1 ⎟ ⎜ 0 ⎟ ⎜ 1 ⎟
2
+ =2 ⎜ 0 ⎟ ⎜ cp ⎟ ⎜ 0 ⎟
4 ⎜ ⎟ ⎜ ⎟ ⎜ ⎟
⎟ ⎜ m0c + E ⎟ ⎜
2
⎜ cp cp ⎟
⎜ m c2 + E ⎟ ⎜ ⎟ ⎜ ⎟
⎠ β ⎝ m0c + E ⎠γ
2
⎝ 0 ⎠α ⎝ 0
⎛ 1 ⎞ ⎛ 0 ⎞ ⎛ 0 ⎞
⎜ ⎟ ⎜ ⎟ ⎜ ⎟
⎜ 0 ⎟ ⎜ 1 ⎟ ⎜ 1 ⎟
2
+ =2 ⎜ cp ⎟ ⎜ 0 ⎟ ⎜ 0 ⎟
4 ⎜ ⎟ ⎜ ⎟ ⎜ ⎟
⎜ m0c + E ⎟ ⎜
2
cp ⎟ ⎜ cp ⎟
⎜ ⎟ ⎜ m c2 + E ⎟ ⎜ m c2 + E ⎟
⎝ 0 ⎠α ⎝ 0 ⎠β ⎝ 0 ⎠γ

and analogously
 65

Sy2,α ,β ,γ ω α ,β ,γ ( p,i = 0 )
⎛ 0 ⎞ ⎛ 0 ⎞ ⎛ 1 ⎞
⎜ ⎟ ⎜ ⎟ ⎜ ⎟
⎜ 1 ⎟ ⎜ 1 ⎟ ⎜ 0 ⎟
3 2 2
= = ω α ,β ,γ ( p,i = 0 ) − =2 ⎜ 0 ⎟ ⎜ 0 ⎟ ⎜ cp ⎟
4 4 ⎜ ⎟ ⎜ ⎟ ⎜ ⎟
⎟ ⎜ m0c + E ⎟
2
⎜ cp ⎟ ⎜ cp
⎜ m c2 + E ⎟ ⎜ m c2 + E ⎟ ⎜ ⎟
⎝ 0 ⎠α ⎝ 0 ⎠β ⎝ 0 ⎠γ
⎛ 0 ⎞ ⎛ 1 ⎞ ⎛ 0 ⎞
⎜ ⎟ ⎜ ⎟ ⎜ ⎟
⎜ 1 ⎟ ⎜ 0 ⎟ ⎜ 1 ⎟
2
− =2 ⎜ 0 ⎟ ⎜ cp ⎟ ⎜ 0 ⎟
4 ⎜ ⎟ ⎜ ⎟ ⎜ ⎟
⎟ ⎜ m0c + E ⎟ ⎜
2
⎜ cp cp ⎟
⎜ m c2 + E ⎟ ⎜ ⎟ ⎜ ⎟
⎠ β ⎝ m0c + E ⎠γ
2
⎝ 0 ⎠α ⎝ 0
⎛ 1 ⎞ ⎛ 0 ⎞ ⎛ 0 ⎞
⎜ ⎟ ⎜ ⎟ ⎜ ⎟
⎜ 0 ⎟ ⎜ 1 ⎟ ⎜ 1 ⎟
2
− =2 ⎜ cp ⎟ ⎜ 0 ⎟ ⎜ 0 ⎟
4 ⎜ ⎟ ⎜ ⎟ ⎜ ⎟
⎜ m0c + E ⎟ ⎜
2
cp ⎟ ⎜ cp ⎟
⎜ ⎟ ⎜ m c2 + E ⎟ ⎜ m c2 + E ⎟
⎝ 0 ⎠α ⎝ 0 ⎠β ⎝ 0 ⎠γ

and

Sz,2 α ,β ,γ ω α ,β ,γ ( p,i = 0 )
⎛ 1 ⎞ ⎛ 1 ⎞ ⎛ 1 ⎞
⎜ ⎟ ⎜ ⎟ ⎜ ⎟
⎜ 0 ⎟ ⎜ 0 ⎟ ⎜ 0 ⎟
3 2 2 .
= = ω α ,β ,γ ( p,i = 0 ) + 3 =2 ⎜ cp ⎟ ⎜ cp ⎟ ⎜ cp ⎟
4 4 ⎜ ⎟ ⎜ ⎟ ⎜ ⎟
⎜ m0c + E ⎟ ⎜ m0c + E ⎟ ⎜ m0c + E ⎟
2 2 2

⎜ ⎟ ⎜ ⎟ ⎜ ⎟
⎝ 0 ⎠α ⎝ 0 ⎠β ⎝ 0 ⎠γ

Finally, we have

S 2α ,β ,γ ω α ,β ,γ ( p,i = 0 )
⎛3 2⎞ 3 5 (3.7.11)
= 3 ⎜ + ⎟ = 2 ω α ,β ,γ ( p,i = 0 ) = ⋅ =2 ω α ,β ,γ ( p,i = 0 ) .
⎝4 4⎠ 2 2

Generally, for arbitrary spin s, one finds

Sα2 ,β ,",τ ω α ,β ,",τ ( p,i = 0 ) = s ( s + 1) =2 ω α ,β ,",τ ( p,i = 0 ) ,

which is fulfilled by (3.7.11).

A similar but more involved calculation shows that an analogous relation holds for
all states ω α ,β ,",τ ( p,i ) of a given s:

Sα2 ,β ,",τ ω α ,β ,",τ ( p,i ) = s ( s + 1) =2 ω α ,β ,",τ ( p,i ) . (3.7.12)

 66

Equations (3.7.7) and (3.7.12) show that the symmetric multispinors ω α ,β ,",τ ( p,i )
belonging to a given spin s may be interpreted as spinor part of the wave functions
of a particle with spin s, where the z-component can assume 2s + 1 different
values.
The whole wave function of such a particle reads in analogy with (3.7.1)
K N
Ψ α ,β ,",τ ( r ,t,p,i ) = ω α ,β ,",τ ( p,i ) ei( pz-Et ) / = . (3.7.13)
V

We define a Dirac equation which can only act on the α -part of (3.7.13) according
to (3.1.1)

⎛ 3 ⎞ K
⎜ c ∑ α i p + β m0c − E ⎟ Ψ α ( r ,t,p ) = 0 ,
i 2
(3.7.14)
⎝ i =1 ⎠α

where Ψ α has the form of the function (3.7.1). No matter if Ψ α is of the type
Ψ + or Ψ − (cf. (3.7.1)) it is a solution of the Dirac equation (3.7.14). Therefore the
multispinor function (3.7.13) is also a solution of (3.7.14), i.e.

⎛ 3 ⎞ K
⎜ c ∑ α j p + β m0 c − E ⎟ Ψ α ,β ,",τ ( r ,t,p,i ) = 0 .
j 2

⎝ j =1 ⎠α

Generally, we realize the relations

⎛ 3 ⎞ K
⎜ c ∑ α j p + β m0 c − E ⎟ Ψ α ,β ,",τ ( r ,t,p,i ) = 0 with ς = α , β ," ,τ . (3.7.15)
j 2

⎝ j =1 ⎠ς

These are named the Bargmann-Wigner equations. Of course, they hold also with
K
linear combinations of functions Ψ α ,β ,",τ ( r ,t,p,i ) with various i-values.

There exist numerous types of mesons with spin 1, 3/2 and 2.

3.8 Charge conjugation. The positron

In this section we deal with an interesting relation between the solution of the Dirac
equation for a particle with charge q (in fact, for a state with negative energy) and
a solution for a particle with the same mass but charge -q. We restrict ourselves to
the pair: electron (charge - e ) and positron (charge + e ). The symbol e stands for
the elementary charge.
We start from the Dirac equation (2.4.1) containing electromagnetic fields acting
on an electron with charge - e

⎡ ∂ 3
⎛ ∂ i ⎞ 2⎤ K
⎢i= ∂t + e Φ + c ∑ ⎜ i=α i ∂x i − e A ⎟ − β m0c ⎥Ψ el ( r ,t ) . (3.8.1)
⎣ i =1 ⎝ ⎠ ⎦
K
The energy dependence of Ψ el ( r ,t ) is given by (3.1.2):
 67

K K
Ψ el ( r ,t ) = ψ ( r ) e −iEt / = . (3.8.2)

By way of exception we choose the negative solution for the energy in (3.1.15)

E = − m0 2c 4 + c 2 p 2 = − E . (3.8.3)

According to (3.8.2) the corresponding negative-energy solution reads

K K
Ψ el ,− ( r ,t ) = ψ el ( r ) ei E t / = . (3.8.4)

Now, we take complex conjugates in (3.8.1) and left multiply by β

⎡ ⎛ ∂ ⎞ 3
⎛ ∂ i ⎞ 2⎤ K
β
⎢ ⎜ −i= + e Φ ⎟ ∑ ⎜ −i= βα i * i − β e A ⎟ − m0c ⎥Ψ * el ,− ( r ,t ) = 0 (3.8.5)
+ c
⎣ ⎝ ∂t ⎠ i =1 ⎝ ∂x ⎠ ⎦

taking into account that Φ and Ai are real. According to (3.8.4) and (3.8.2) the
four-spinor Ψ * el ,− has the exponential energy dependence e −i E t / = . We multiply
−1
(3.8.5) from the left by a non-singular 4 × 4 -matrix C and insert C C = 1 in front of
Ψ * el ,− , which yields

⎡ −1 ⎛ ∂ ⎞ 3
⎛ −1 ∂ ⎞ ⎤
⎟ ∑ ⎜ −i=C βα i * C
−1
⎢C β C ⎜ − i= + e Φ + c − C β C e Ai ⎟ − m0c 2 ⎥
⎣ ⎝ δ∂t ⎠ i =1 ⎝ ∂x i
⎠ ⎦
K
⋅CΨ * el ,− ( r ,t ) = 0 .
(3.8.6)
On the other hand, analogously to (3.8.1), we write the Dirac equation for a
positron, which we multiply also with β from the left

⎡ ⎛ ∂ ⎞ 3
⎛ ∂ i ⎞ 2⎤ K
⎢ β ⎜ i= ∂t − e Φ ⎟ + c ∑ ⎜ i= βα i ∂x i + β e A ⎟ − m0c ⎥Ψ pos ,+ ( r ,t ) = 0 . (3.8.7)
⎣ ⎝ ⎠ i =1 ⎝ ⎠ ⎦

The wave function Ψ pos ,+ of the positron is taken for positive energy.

The expressions in square brackets in (3.8.6) and (3.8.7) agree, if

−1 −1
CβC = − β , C βα i * C = − βα i . (3.8.8)

The solution of the condition (3.8.8) reads

−1
C = iβα 2 , C = −iα 2 β . (3.8.9)

−1
Obviously C C = 1 is fulfilled. We check the relations in (3.8.8)
 68

= i βα 2 β ( −i ) α 2 β = − β βα 2 α 2 β = − β ,
−1
Cβ C

= i βα 2 βα1 ( −i ) α 2 β = − βα 2 βα 2 α 1 β = α 1 β = − βα 1 ,
−1
C βα 1 * C

C βα 2 * C
−1
( )
= i βα 2 β −α 2 ( −i ) α 2 β = − βα 2 .

In this way the square brackets of (3.8.6) and (3.8.7) are identical and the
solutions must agree:
K K K
Ψ pos ,+ ( r ,t ) = CΨ * el ,− ( r ,t ) ≡ iβ α 2Ψ * el ,− ( r ,t ) . (3.8.10)

Hence, by means of the negative energy wave function of a particle the positive
energy function of the corresponding antiparticle can be formed in a simple way.
The transformed function iβ α 2Ψ * el ,− is named charge-conjugate function of the
electron.
 69

4 Other relativistic quantum mechanical equations

In order to describe spin zero particles, two equations have been set up, which we
outline here shortly.

4.1 The Klein-Gordon equation

This equation is given by the relativistic operator relation (1.2.11) for a free
particle,
2

∑(p )
3
E −c p ≡ E −c
2 2 2 2 2 i
= m02c 4 , (4.1.1)
i =1

which is form invariant in Lorentz transformations. By means of (1.2.6) and (1.2.7)

it generates the following equation

⎛ ∂2 ∂2 ∂2 ∂ 2 m0 2c 2 ⎞ K
⎜ 2 2 − − − + ⎟Ψ ( r ,t ) = 0 . (4.1.2)
⎝ c ∂t ∂x ∂y ∂z = ⎠
2 2 2 2

This wave equation has the solution

K K K K
Ψ ( r ,t ) = exp ⎡⎣i ( p ⋅ r − Et ) / = ⎤⎦ , r = ( x,y ,z ) . (4.1.3)

Inserting (4.1.3) in (4.1.2) yields the relativistic relation (1.2.10) or (3.1.15), i.e.

E2
2
= p 2 + m02c 2 .
c
Treating the equation (4.1.2) in the same way as in the Dirac equation in (2.3.1) up
to (2.3.3) one obtains by subtracting

1 ⎛ ∂2 ∂2 ⎞ 1 ∂⎛ ∂ ∂ ⎞
⎜Ψ * Ψ −Ψ Ψ * ⎟ = 2 ⎜Ψ * Ψ −Ψ Ψ * ⎟ . (4.1.4)
c2 ∂t ∂t ⎠ c ∂t ⎝ ∂t ∂t ⎠
2 2
⎝

With arguments used in section 2.3 one expects that the expression

∂Ψ ∂Ψ *
Ψ* −Ψ (4.1.5)
∂t ∂t
is proportional to the probability density ρ . However, this interpretation is doubtful
because the expression (4.1.5) can become negative, which is not permissible for
a probability. This is one of the shortcomings of the Klein-Gordon equation.
On the other hand, if ρ is defined like in (2.3.4) as
K
ρ = Ψ ( r ,t ) ,
2

it can be shown, that it isn't relativisticaly invariant (Yndurain, 1996, p. 24), which is
also unsatisfactory.
 70

Yndurain also criticizes that, being of second order in the time derivative, the
K K
values of the wave function Ψ ( r ,t0 ) at time t0 do not uniquely determine Ψ ( r ,t ) :
K
knowledge of ∂Ψ ( r ,t ) / ∂t is also necessary.

There arise problems, when electrostatic fields, e.g. the field of a point charge,
have to be built in the Klein-Gordon equation. The often used form

=cα ⎞
2
⎛ K K
⎜E + ⎟ Ψ ( r ,t ) = ( m0 c + c p )Ψ ( r ,t ) ,
2 4 2 2

⎝ r ⎠

unfortunately, is not acceptable: it is not even compatible with the superposition

principle.

4.2 The Klein-Gordon Schrödinger equation

In order to avoid the quadrate of the energy operator, the equation (4.1.1) is
rearranged and the (positive) square root of the operators is taken:
K K
EΨ ( r ,t ) = m02c 4 + p 2c 2 Ψ ( r ,t )
⎛ p2 p4 ⎞ K (4.2.1)
= m0c 2 ⎜ 1 + 2 2
− 4 4
+ " ⎟Ψ ( r ,t )
⎝ 2m 0 c 8 m0 c ⎠

However, F. Gross (Gross, 1993, p.92) stresses that (4.2.1) is not manifestly
covariant. Dealing with this equation many problems are encountered. Anyhow, it
is an active area of current research.
 71

References
Bjorken, J.D., Drell, S.D. (1964), Relativistic Quantum Mechanics, McGraw-Hill,
Inc.
Greiner, W. (1990), Relativistic Quantum mechanics, Springer-Verlag, Berlin.
Gross, F. (1993), Relativistic Quantum Mechanics and Field Theory, John Wiley &
Sons, Inc, New York.
Pfeifer, W. (1998), An Introduction to the Interacting Boson Model of the Atomic
Nucleus, www.walterpfeifer.ch.
Pfeifer, W. (2003), The Lie Algebras su(N), An Introduction. Birkhäuser Verlag,
Basel.
Yndurain, F.J. (1996), Relativistic Quantum Mechanics and Introduction to Field
Theory, Springer-Verlag, Berlin.
 72

Index
angular momentum 25 Klein-Gordon Schrödinger equation
angular momentum operator 26, 44 70
antineutrino 60 Lamb shift 59
antiparticle 32, 68 Lorentz convention 8
arbitrary spin 61 Lorentz force 7
Bargman-Wigner equations 66 Lorentz group 4
Bohr-formula 58 Lorentz transformation 3, 18
boost 5 Lorentz-covariance 15
bound states 53 magnetic field strength 7
central potential 43, 46 magnetic moment 24
charge conjugation 66 anomalous 27
charge density 7 normal 27
circular path of the electron 43 massless Dirac particles 59
Clebsch Gordan coefficients 47 minimal substitution 15
conservation of flow of a fluid 20 momentum operator 6, 30
contravariant 4 multispinor 62
Coulomb potential 53 negative energy 66
covariant 70 negative energy state 32
current density 19 neutrino 60
cyclotron 38 nonrelativistic limit of the Dirac
density 16 equation 21
dielectricity 8 normalization factor 31
Dirac particles 34 parity operator 43, 60
Dirac sea 32 internal 43
electric current density 8 parity violation 61
electric field strength 7 Pauli equation 24
electromagnetic field 19, 66 Pauli realization 13
electromagnetic potentials 15 Pauli spin matrices 13
energy operator 6, 30 Pauli term 21, 24
energy splitting 59 Pauli's theorem 16
fine structure 57 permeability 8
flow conservation equation 20 plane wave 30
form invariant 19 point nucleus 53
free particle 7, 30 positron 66
Gauss system 7 principal quantum number 57
Greiner 29, 55 principle of minimal coupling 15
Gross 70 probability density 14, 19, 69
gyromagnetic ratio 24 probability density current 20
Hamiltonian 12 rest mass 7
helicity 30, 60 rotation of the coordinate 4
helicity operator 34 scalar potential 7
Hermite polynomial 40 Schrödinger equation 11, 26
Hermitian 11, 12, 29 Schrödinger-formula 58
homogeneous magnetic field 24, 38 secular equation 31
hyperfine structure 59 Sommerfeld structure formula 58
inertial system 15 special relativity 11
internal parity 47, 48 spherical harmonics 47, 50
Klein-Gordon equation 69 spherical potential 44
 1

spin eigenfunctions 47 total angular momentum 43

spin operator 24, 26, 61 trace 13
spin-½ fermions 24 unit matrix 16
spinor 31 unitary transformation 14
spin-orbit energy 21, 26 vector potential 7
superposition principle 70 velocity operator 30, 36
technical applications 40 Yndurain 32, 69
theorem of Ehrenfest 36