The Particle

Dirac Equation
Physics
Book “Modern Particle
Michaelmas TermPhysics”
2011 by
Prof. Mark Thomson ISBN-13: 978-1107034266

!+ !+
e- e+ e- e+
!- !-
!+ !+
e- e+ e- e+
!- !-

Handout 2 : The Dirac Equation

most of this material is beyond the scope of PHZ 7357 and recommended for further reading

Prof. M.A. Thomson Michaelmas 2011 45

Non-Relativistic QM (Revision)
• For particle physics need a relativistic formulation of quantum mechanics. But
first take a few moments to review the non-relativistic formulation QM
• Take as the starting point non-relativistic energy:

• In QM we identify the energy and momentum operators:

which gives the time dependent Schrödinger equation (take V=0 for simplicity)

(S1)

with plane wave solutions: where

•The SE is first order in the time derivatives and second order in spatial
derivatives – and is manifestly not Lorentz invariant.
•In what follows we will use probability density/current extensively. For
the non-relativistic case these are derived as follows

(S1)* (S2)
Prof. M.A. Thomson Michaelmas 2011 46
 •Which by comparison with the continuity equation

leads to the following expressions for probability density and current:

•For a plane wave

and

!The number of particles per unit volume is

! For particles per unit volume moving at velocity , have passing
through a unit area per unit time (particle flux). Therefore is a vector in the
particle’s direction with magnitude equal to the flux.

Prof. M.A. Thomson Michaelmas 2011 47

The Klein-Gordon Equation

•Applying to the relativistic equation for energy:
(KG1)
gives the Klein-Gordon equation:

(KG2)

•Using

KG can be expressed compactly as (KG3)

•For plane wave solutions, , the KG equation gives:

! Not surprisingly, the KG equation has negative energy solutions – this is

just what we started with in eq. KG1
" Historically the –ve energy solutions were viewed as problematic. But for the KG
there is also a problem with the probability density…
Prof. M.A. Thomson Michaelmas 2011 48
 •Proceeding as before to calculate the probability and current densities:

(KG2)* (KG4)

•Which, again, by comparison with the continuity equation allows us to identify

•For a plane wave

and
!Particle densities are proportional to E. We might have anticipated this from the
previous discussion of Lorentz invariant phase space (i.e. density of 1/V in the
particles rest frame will appear as E/V in a frame where the particle has energy E
due to length contraction).
Prof. M.A. Thomson Michaelmas 2011 49

The Dirac Equation

!Historically, it was thought that there were two main problems with the
Klein-Gordon equation:
" Negative energy solutions
" The negative particle densities associated with these solutions

!We now know that in Quantum Field Theory these problems are
overcome and the KG equation is used to describe spin-0 particles
(inherently single particle description ! multi-particle quantum
excitations of a scalar field).

Nevertheless:
!These problems motivated Dirac (1928) to search for a
different formulation of relativistic quantum mechanics
in which all particle densities are positive.

!The resulting wave equation had solutions which not only

solved this problem but also fully describe the
intrinsic spin and magnetic moment of the electron!
Prof. M.A. Thomson Michaelmas 2011 50
 The Dirac Equation :
•Schrödinger eqn: 1st order in
2nd order in
• Klein-Gordon eqn: 2nd order throughout
• Dirac looked for an alternative which was 1st order throughout:
(D1)
where is the Hamiltonian operator and, as usual,

•Writing (D1) in full:

“squaring” this equation gives

• Which can be expanded in gory details as…

Prof. M.A. Thomson Michaelmas 2011 51

• For this to be a reasonable formulation of relativistic QM, a free particle

must also obey , i.e. it must satisfy the Klein-Gordon equation:

• Hence for the Dirac Equation to be consistent with the KG equation require:
(D2)
(D3)
(D4)
!Immediately we see that the and cannot be numbers. Require 4
mutually anti-commuting matrices
!Must be (at least) 4x4 matrices (see Appendix I)
Prof. M.A. Thomson Michaelmas 2011 52
 •Consequently the wave-function must be a four-component Dirac Spinor
A consequence of introducing an equation
that is 1st order in time/space derivatives is that
the wave-function has new degrees of freedom !

• For the Hamiltonian to be Hermitian

requires
(D5)
i.e. the require four anti-commuting Hermitian 4x4 matrices.
• At this point it is convenient to introduce an explicit representation for .
It should be noted that physical results do not depend on the particular
representation – everything is in the commutation relations.
• A convenient choice is based on the Pauli spin matrices:

with

• The matrices are Hermitian and anti-commute with each other

Prof. M.A. Thomson Michaelmas 2011 53

Dirac Equation: Probability Density and Current

•Now consider probability density/current – this is where the perceived
problems with the Klein-Gordon equation arose.
•Start with the Dirac equation

(D6)
and its Hermitian conjugate

(D7)

•Consider remembering are Hermitian

•Now using the identity:

Prof. M.A. Thomson Michaelmas 2011 54

 gives the continuity equation (D8)

where
•The probability density and current can be identified as:

and

where

•Unlike the KG equation, the Dirac equation has probability densities which
are always positive.
• In addition, the solutions to the Dirac equation are the four component
Dirac Spinors. A great success of the Dirac equation is that these
components naturally give rise to the property of intrinsic spin.
• It can be shown that Dirac spinors represent spin-half particles (appendix II)
with an intrinsic magnetic moment of

(appendix III)

Prof. M.A. Thomson Michaelmas 2011 55

Covariant Notation: the Dirac " Matrices

•The Dirac equation can be written more elegantly by introducing the
four Dirac gamma matrices:

Premultiply the Dirac equation (D6) by

using this can be written compactly as:

(D9)

! NOTE: it is important to realise that the Dirac gamma matrices are not
four-vectors - they are constant matrices which remain invariant under a
Lorentz transformation. However it can be shown that the Dirac equation
is itself Lorentz covariant (see Appendix IV)
Prof. M.A. Thomson Michaelmas 2011 56
 Properties of the " matrices
•From the properties of the and matrices (D2)-(D4) immediately obtain:
and
•The full set of relations is

which can be expressed as:

(defines the algebra)

• Are the gamma matrices Hermitian?

" is Hermitian so is Hermitian.
" The matrices are also Hermitian, giving

" Hence are anti-Hermitian

Prof. M.A. Thomson Michaelmas 2011 57

Pauli-Dirac Representation
•From now on we will use the Pauli-Dirac representation of the gamma matrices:

which when written in full are

•Using the gamma matrices and can be written as:

where is the four-vector current.

(The proof that is indeed a four vector is given in Appendix V.)
•In terms of the four-vector current the continuity equation becomes

•Finally the expression for the four-vector current

can be simplified by introducing the adjoint spinor

Prof. M.A. Thomson Michaelmas 2011 58
 The Adjoint Spinor
• The adjoint spinor is defined as

i.e.

•In terms the adjoint spinor the four vector current can be written:

!We will use this expression in deriving the Feynman rules for the
Lorentz invariant matrix element for the fundamental interactions.

!That’s enough notation, start to investigate the free particle solutions

of the Dirac equation...
Prof. M.A. Thomson Michaelmas 2011 59

Dirac Equation: Free Particle at Rest

•Look for free particle solutions to the Dirac equation of form:

where , which is a constant four-component spinor which must satisfy

the Dirac equation

•Consider the derivatives of the free particle solution

substituting these into the Dirac equation gives:

which can be written: (D10)

•This is the Dirac equation in “momentum” – note it contains no derivatives.
•For a particle at rest

and
eq. (D10)
Prof. M.A. Thomson Michaelmas 2011 60
 (D11)

•This equation has four orthogonal solutions:

(D11) E=m (D11) E = -m

still have NEGATIVE ENERGY SOLUTIONS (Question 6)
• Including the time dependence from gives

Two spin states with E>0 Two spin states with E<0

!In QM mechanics can’t just discard the E<0 solutions as unphysical

as we require a complete set of states - i.e. 4 SOLUTIONS
Prof. M.A. Thomson Michaelmas 2011 61

Dirac Equation: Plane Wave Solutions

•Now aim to find general plane wave solutions:
•Start from Dirac equation (D10):
and use

Note

Note in the above equation the 4x4 matrix is

written in terms of four 2x2 sub-matrices
•Writing the four component
spinor as

Giving two coupled

simultaneous equations (D12)
for
Prof. M.A. Thomson Michaelmas 2011 62
 Expanding

•Therefore (D12)

gives

•Solutions can be obtained by making the arbitrary (but simplest) choices for
i.e. or

where N is the
giving and wave-function
normalisation

NOTE: For these correspond to the E>0 particle at rest solutions

!The choice of is arbitrary, but this isn’t an issue since we can express any
other choice as a linear combination. It is analogous to choosing a basis for
spin which could be eigenfunctions of Sx, Sy or Sz
Prof. M.A. Thomson Michaelmas 2011 63

Repeating for and gives the solutions and

! The four solutions are:

•If any of these solutions is put back into the Dirac equation, as expected, we obtain

which doesn’t in itself identify the negative energy solutions.

•One rather subtle point: One could ask the question whether we can interpret
all four solutions as positive energy solutions. The answer is no. If we take
all solutions to have the same value of E, i.e. E = +|E|, only two of the solutions
are found to be independent.
•There are only four independent solutions when the two are taken to have E<0.

!To identify which solutions have E<0 energy refer back to particle at rest (eq. D11 ).
• For : correspond to the E>0 particle at rest solutions
correspond to the E<0 particle at rest solutions
! So are the +ve energy solutions and are the -ve energy solutions
Prof. M.A. Thomson Michaelmas 2011 64
 Interpretation of –ve Energy Solutions
!The Dirac equation has negative energy solutions. Unlike the KG equation
these have positive probability densities. But how should –ve energy
solutions be interpreted? Why don’t all +ve energy electrons fall into
to the lower energy –ve energy states?
Dirac Interpretation: the vacuum corresponds to all –ve energy states
being full with the Pauli exclusion principle preventing electrons falling into
-ve energy states. Holes in the –ve energy states correspond to +ve energy
anti-particles with opposite charge. Provides a picture for pair-production
and annihilation.

.. .. ..
.. .. ..

mc2 mc2 mc2

" "
-mc2 -mc2 -mc2
.. .. ..
.. .. ..

Prof. M.A. Thomson Michaelmas 2011 65

Discovery of the Positron

!Cosmic ray track in cloud chamber: C.D.Anderson, Phys Rev 43 (1933) 491

B
e# 23 MeV 6 mm
Lead
Plate

e# 63 MeV

• e+ enters at bottom, slows down in the

lead plate – know direction
• Curvature in B-field shows that it is a
positive particle
• Can’t be a proton as would have stopped in the lead
Provided Verification of Predictions of Dirac Equation

!Anti-particle solutions exist ! But the picture of the vacuum corresponding to

the state where all –ve energy states are occupied is rather unsatisfactory, what
about bosons (no exclusion principle),….
Prof. M.A. Thomson Michaelmas 2011 66
 Feynman-Stückelberg Interpretation
!There are many problems with the Dirac interpretation of anti-particles
and it is best viewed as of historical interest – don’t take it too seriously.
Feynman-Stückelberg Interpretation:
!Interpret a negative energy solution as a negative energy particle which
propagates backwards in time or equivalently a positive energy anti-particle
which propagates forwards in time

e– (E<0) e+ (E>0)
" "
time

e+ e-

E>0 E<0
e– (E>0) e– (E>0)
NOTE: in the Feynman diagram the arrow on the
anti-particle remains in the backwards in time
direction to label it an anti-particle solution.
!At this point it become more convenient to work with anti-particle
wave-functions with motivated by this interpretation

Prof. M.A. Thomson Michaelmas 2011 67

Anti-Particle Spinors
•Want to redefine our –ve energy solutions such that:
i.e. the energy of the physical anti-particle.

We can look at this in two ways:

# Start from the negative energy solutions

Where E is understood to
be negative

•Can simply “define” anti-particle wave-function by flipping the sign

of and following the Feynman-Stückelburg interpretation:

where E is now understood to be positive,

Prof. M.A. Thomson Michaelmas 2011 68

 Anti-Particle Spinors
$ Find negative energy plane wave solutions to the Dirac equation of
the form: where
•Note that although these are still negative energy solutions
in the sense that

•Solving the Dirac equation

(D13)
! The Dirac equation in terms of momentum for ANTI-PARTICLES (c.f. D10)
•Proceeding as before: etc., …

•The same wave-functions that were written down on the previous page.
Prof. M.A. Thomson Michaelmas 2011 69

Particle and anti-particle Spinors

! Four solutions of form:

! Four solutions of form

! Since we have a four component spinor, only four are linearly independent
% Could choose to work with or or …
% Natural to use choose +ve energy solutions

Prof. M.A. Thomson Michaelmas 2011 70

 Wave-Function Normalisation
•From handout 1 want to normalise wave-functions
to particles per unit volume
•Consider
Probability density

which for the desired 2E particles per unit volume, requires that

•Obtain same value of N for

Prof. M.A. Thomson Michaelmas 2011 71

Charge Conjugation
• In the part II Relativity and Electrodynamics course it was shown that
the motion of a charged particle in an electromagnetic field
can be obtained by making the minimal substitution

with
this can be written

and the Dirac equation becomes:

•Taking the complex conjugate and pre-multiplying by

But and

(D14)
•Define the charge conjugation operator:

Prof. M.A. Thomson Michaelmas 2011 72

 D14 becomes:

•Comparing to the original equation

we see that the spinor describes a particle of the same mass but with
opposite charge, i.e. an anti-particle !
particle spinor " anti-particle spinor
•Now consider the action of on the free particle wave-function:

hence
similarly
!Under the charge conjugation operator the particle spinors and
transform to the anti-particle spinors and
Prof. M.A. Thomson Michaelmas 2011 73

Using the anti-particle solutions

•There is a subtle but important point about the anti-particle solutions written as

Applying normal QM operators for momentum and energy

gives and
!But have defined solutions to have E>0
!Hence the quantum mechanical operators giving the physical energy and
momenta of the anti-particle solutions are:
and
•Under the transformation :
Conservation of total angular momentum

!The physical spin of the anti-particle solutions is given by

0
A spin-up hole leaves the
In the hole picture: -mc2 negative energy sea in a spin
down state
.
Prof. M.A. Thomson Michaelmas 2011 74
 Summary of Solutions to the Dirac Equation
•The normalised free PARTICLE solutions to the Dirac equation:
satisfy
with

•The ANTI-PARTICLE solutions in terms of the physical energy and momentum:

satisfy

with

For these states the spin is given by

•For both particle and anti-particle solutions:
(Now try question 7 – mainly about 4 vector current )
Prof. M.A. Thomson Michaelmas 2011 75

Spin States
•In general the spinors are not Eigenstates of

(Appendix II)

•However particles/anti-particles travelling in the z-direction:

are Eigenstates of
Note the change of sign
of when dealing with
antiparticle spinors

z z
! Spinors are only eigenstates of for

Prof. M.A. Thomson Michaelmas 2011 76

 Pause for Breath…
•Have found solutions to the Dirac equation which are also eigenstates but
only for particles travelling along the z axis.
•Not a particularly useful basis
•More generally, want to label our states in terms of “good quantum numbers”,
i.e. a set of commuting observables.
•Can’t use z component of spin: (Appendix II)

•Introduce a new concept “HELICITY”

Helicity plays an important role in much that follows

Prof. M.A. Thomson Michaelmas 2011 77

Helicity
! The component of a particles spin along its direction of flight is a good quantum
number:

! Define the component of a particles spin along its direction of flight as HELICITY:

•If we make a measurement of the component of spin of a spin-half particle

along any axis it can take two values , consequently the eigenvalues
of the helicity operator for a spin-half particle are:

Often termed: “right-handed” “left-handed”

! NOTE: these are “RIGHT-HANDED” and LEFT-HANDED HELICITY eigenstates
! In handoutfrom
different 4 we will discuss RH and LH CHIRAL eigenstates. Only in the limit
are the HELICITY eigenstates the same as the CHIRAL eigenstates

Prof. M.A. Thomson Michaelmas 2011 78

 Helicity Eigenstates
!Wish to find solutions of Dirac equation which are also eigenstates of Helicity:

where and are right and left handed helicity states and here is
the unit vector in the direction of the particle.
•The eigenvalue equation:

gives the coupled equations:

(D15)

•Consider a particle propagating in direction

Prof. M.A. Thomson Michaelmas 2011 79

•Writing either or then (D15) gives the relation

(For helicity )

So for the components of BOTH and

•For the right-handed helicity state, i.e. helicity +1:

•Putting in the constants of proportionality gives:

Prof. M.A. Thomson Michaelmas 2011 80

 •From the Dirac Equation (D12) we also have

(D16)
Helicity

!(D15) determines the relative normalisation of and , i.e. here

•The negative helicity particle state is obtained in the same way.

•The anti-particle states can also be obtained in the same manner although
it must be remembered that
i.e.
Prof. M.A. Thomson Michaelmas 2011 81

! The particle and anti-particle helicity eigenstates states are:

particles anti-particles

! For all four states, normalising to 2E particles/Volume again gives

The helicity eigenstates will be used extensively in the calculations that follow.
Prof. M.A. Thomson Michaelmas 2011 82
 Intrinsic Parity of Dirac Particles
non-examinable
! Before leaving the Dirac equation, consider parity
! The parity operation is defined as spatial inversion through the origin:

•Consider a Dirac spinor, , which satisfies the Dirac equation

(D17)
•Under the parity transformation:
Try
so

(D17)
•Expressing derivatives in terms of the primed system:

Since anti-commutes with :

Prof. M.A. Thomson Michaelmas 2011 83

Pre-multiplying by

•Which is the Dirac equation in the new coordinates.

!There for under parity transformations the form of the Dirac equation is
unchanged provided Dirac spinors transform as

(note the above algebra doesn’t depend on the choice of )

•For a particle/anti-particle at rest the solutions to the Dirac Equation are:

with

etc.

!Hence an anti-particle at rest has opposite intrinsic parity to a particle at rest.

!Convention: particles are chosen to have +ve parity; corresponds to choosing

Prof. M.A. Thomson Michaelmas 2011 84

 Summary
!The formulation of relativistic quantum mechanics starting from the
linear Dirac equation

New degrees of freedom : found to describe Spin ½ particles

! In terms of 4x4 gamma matrices the Dirac Equation can be written:

! Introduces the 4-vector current and adjoint spinor:

! With the Dirac equation: forced to have two positive energy and two
negative energy solutions
!Feynman-Stückelberg interpretation: -ve energy particle solutions
propagating backwards in time correspond to physical +ve energy
anti-particles propagating forwards in time

Prof. M.A. Thomson Michaelmas 2011 85

! Most useful basis: particle and anti-particle helicity eigenstates

! In terms of 4-component spinors, the charge conjugation and parity

operations are:

! Now have all we need to know about a relativistic description of

particles… next discuss particle interactions and QED.

Prof. M.A. Thomson Michaelmas 2011 86

 Appendix I : Dimensions of the Dirac Matrices
non-examinable
Starting from
For to be Hermitian for all requires
To recover the KG equation:

Consider
with

Therefore
(using commutation relation)

similarly

Prof. M.A. Thomson Michaelmas 2011 87

We can now show that the matrices are of even dimension by considering
the eigenvalue equation, e.g.

Eigenvalues of a Hermitian matrix are real so

but
Since the are trace zero Hermitian matrices with eigenvalues of
they must be of even dimension

For N=2 the 3 Pauli spin matrices satisfy

But we require 4 anti-commuting matrices. Consequently the of the

Dirac equation must be of dimension 4, 6, 8,….. The simplest choice for
is to assume that the are of dimension 4.

Prof. M.A. Thomson Michaelmas 2011 88

 Appendix II : Spin
non-examinable
•For a Dirac spinor is orbital angular momentum a good quantum number?
i.e. does commute with the Hamiltonian?

Consider the x component of L:

The only non-zero contributions come from:

Therefore (A.1)
!Hence the angular momentum does not commute with the Hamiltonian
and is not a constant of motion
Prof. M.A. Thomson Michaelmas 2011 89

Introduce a new 4x4 operator:

where are the Pauli spin matrices: i.e.

Now consider the commutator

here

and hence
Consider the x comp:

Prof. M.A. Thomson Michaelmas 2011 90

 Taking each of the commutators in turn:

Hence

Prof. M.A. Thomson Michaelmas 2011 91

•Hence the observable corresponding to the operator is also not

a constant of motion. However, referring back to (A.1)

Therefore:

•Because

the commutation relationships for are the same as for the , e.g.
. Furthermore both S and Sz are diagonal
2

•Consequently and for a particle travelling along

the z direction
!S has all the properties of spin in quantum mechanics and therefore the
Dirac equation provides a natural account of the intrinsic angular
momentum of fermions
Prof. M.A. Thomson Michaelmas 2011 92
 Appendix III : Magnetic Moment
non-examinable
• In the part II Relativity and Electrodynamics course it was shown that
the motion of a charged particle in an electromagnetic field
can be obtained by making the minimal substitution

• Applying this to equations (D12)

(A.2)

Multiplying (A.2) by

(A.3)
where kinetic energy
•In the non-relativistic limit (A.3) becomes

(A.4)

Prof. M.A. Thomson Michaelmas 2011 93

•Now

which leads to
and
•The operator on the LHS of (A.4):

!Substituting back into (A.4) gives the Schrödinger-Pauli equation for

the motion of a non-relativisitic spin ½ particle in an EM field

Prof. M.A. Thomson Michaelmas 2011 94

 Since the energy of a magnetic moment in a field is we can
identify the intrinsic magnetic moment of a spin ½ particle to be:

In terms of the spin:

Classically, for a charged particle current loop

The intrinsic magnetic moment of a spin half Dirac particle is twice

that expected from classical physics. This is often expressed in terms
of the gyromagnetic ratio is g=2.

Prof. M.A. Thomson Michaelmas 2011 95

Appendix IV : Covariance of Dirac Equation

non-examinable
•For a Lorentz transformation we wish to demonstrate that the Dirac
Equation is covariant i.e.
(A.5)
transforms to (A.6)

where

and is the transformed spinor.

•The covariance of the Dirac equation will be established if the 4x4 matrix
S exists.
•Consider a Lorentz transformation with the primed frame moving with
velocity v along the x axis

where

Prof. M.A. Thomson Michaelmas 2011 96

 With this transformation equation (A.6)

which should be compared to the matrix S multiplying (A.5)

!Therefore the covariance of the Dirac equation will be demonstrated if

we can find a matrix S such that

(A.7)
•Considering each value of

where
and

Prof. M.A. Thomson Michaelmas 2011 97

•It is easy (although tedious) to demonstrate that the matrix:

with

satisfies the above simultaneous equations

NOTE: For a transformation along in the –x direction

!To summarise, under a Lorentz transformation a spinor transforms

to . This transformation preserves the mathematical
form of the Dirac equation

Prof. M.A. Thomson Michaelmas 2011 98

 Appendix V : Transformation of Dirac Current
non-examinable
!The Dirac current plays an important rôle in the description
of particle interactions. Here we consider its transformation properties.
•Under a Lorentz transformation we have
and for the adjoint spinor:
•First consider the transformation properties of

where
giving

hence
!The product is therefore a Lorentz invariant. More generally, the
product is Lorentz covariant
Prof. M.A. Thomson Michaelmas 2011 99

!Now consider

•To evaluate this wish to express in terms of

(A.7)

where we used
•Rearranging the labels and reordering gives:

!Hence the Dirac current, , transforms as a four-vector

Prof. M.A. Thomson Michaelmas 2011 100