Chapter 6 Maxwell Equations, Macroscopic

Electromagnetism, Conservation Laws

6.1 Maxwell's Displacement Current; Maxwell Equations
Coulomb's law D = 
Ampere's law J = 0  ×H= J
4 equations:
B
Faraday's law  ×E =0
t
Absence of free magnetic poles B= 0
All but Faraday's law were derived from steady-state observations. Thus they
are inconsistent.

The faulty equation is Ampere's law. It is from J =0    × H = J =0

is valid for steady-state problems.
 D
the continuity equation  J = 0
 J  =0  D = 
t t
D D D

 ×H= J   J  J  : displacement current
t t t
It means that a changing electric field causes a magnetic field, even without a
current. The term is of crucial importance for rapidly fluctuating fields.
 D
D=  ,  ×H= J 
The Maxwell equations:
equations
t
B
B= 0,  ×E =0
t
When combined with the Lorentz force eqn and Newton's 2nd law of motion,
these eqns provide a complete description of the classical dynamics of interacting
charged particles and EM fields.
6.2 Vector and Scalar Potentials
It is often convenient to introduce potentials, obtaining a smaller number of 2nd-
order equations, while satisfying some of the Maxwell equations identically.
A A
B= 0
B= ×A
× E =0
E= 
t t
2  
D= 0 E , D = 
  A =
t 0
2
D 2 1  A 1 
B=  0 H ,  × H = J 
 A 2 2
 A  2
= 0 J
t c t c t
Reduced 4 1st-order Maxwell equations to 2 2nd-order equations.

A  A '=A
B '=B 1  Lorenz


choose A =0 
 '= E '=E c2  t condition
t
2
2 1 2  2 1  A

 2 2
= ,  A 2 2
= 0 J
c t 0 c t
uncouple the 2nd-order equations
6.3 Gauge Transformations, Lorenz Gauge, Coulomb Gauge
The transformation is called agauge transformation, and the invariance of the
fields under such transformations is called gauge invariance.
1  
A  0
A '=A , '=
c2  t t
1  ' 1  2 1 2

A '  = 0= A  
c2  t c2  t c2  t 2
2
2 1  1  the new potentials А ' , ' will satisfy

 = A  
c2  t2 c2  t the Lorenz condition and the wave eqns

1   1  '
A  2
=0
A '=A , '=
A '  2
=0
c t t c t
2 1 2

 2 2
=0  restricted gauge transformation
c t
All potentials in this restricted class are said to belong to the Lorenz gauge.
The Lorenz gauge is commonly used because
(1) it leads to the wave eqs, which treat the potentials on equivalent footings;
(2) a concept indep. of the coordinate system and so fits into special relativity.
 
Coulomb, radiation, or transverse gauge A = 0
2 =  Poisson
0 eqn
1  x',t

x ,t =
4
 x 0 x '
d x'
3
instantaneous Coulomb potential 1

2
1  A 1  

2 A 2 2
= 0 J  2

 × =0  irrotational
c t c t t
J = J   J t   × J  = 0 longitudinal, irrotational   ×  × J =  J
2
 J
  J t = 0 transverse, solenoidal

 2 J  = J  
2 1
= 4 x x'
2
 Jt =  ×  × J x x '
1  'J x ' , t
J =
4
  x x '
d3 x ' 
 J = 0 1 

 t
2
 = 0 J 
1 J x' ,t c t
J t =
4
 × ×  x x '
d x'3
1

2
2 1  A

 A 2 2
= 0 J t
c t
the source for the wave equation for A can be expressed entirely in terms of the
transverse current.
 The radiation gauge stems from that transverse radiation fields are given by the
vector potential alone, the Coulomb potential contributing only to the near fields.

This gauge is useful in quantum electrodynamics. A quantum-mechanical

description of photons necessitates quantization of only the vector potential.

The Coulomb or transverse gauge is often used with no source

2
2 1  A A
=0, J = 0
= 0,  A 2 2
=0
E= , B= ×A
c t t
In the Coulomb gauge, the scalar potential "propagates" instantaneously, but
the vector potential propagates in finite speed of propagation с.

it is the fields, not the potentials, that matter. Another is that the transverse
current extends over all space, even if J is localized.
6.4 Green Functions for the Wave Equation

Form of the wave eqn:   2 1 2 

2 2
= 4 f x,t  f : source function
c t

1
Fourier
 x ,t =
2
 
 x , e it
d  x, = 

 x , t ei  t d t




transform
1 x ,  =
i t it
f x ,t = f x, e d f f x ,t e dt

2 

2 2
 k  x ,  = 4  f x ,   inhomogeneous Helmholtz wave equation
k = / c
The Helmholtz wave eqn is an elliptic partial differential eqn similar to the
Poisson eqn to which it reduces for k=0.

The Green function satisfies the inhomogeneous eqn

 2  k2 G k x , x ' = 4   x x'
If there are no boundary surfaces, the Green function depend only on R=x-x'
and must in fact be spherically symmetric.
1 d2 2 1 d2

2
R G k  k Gk = 4   R
2
R G k  k 2 G k =0 for R0
R dR R dR
ik R ikR

R Gk R = A e B e
 the delta function has influence only at R0. In that limit the equation reduces
to the Poisson equation, since kR1
1  ± e±i k R

lim G k R =
Gk R = A G k
R  B Gk R  G =
k
, A B= 1
k R0 R R
the 1st term represents a diverging spherical wave propagating from the origin,
while the 2nd represents a converging spherical wave. The choice of A and

depends on the boundary conditions in time.

To understand their different time behaviors, construct the corresponding time-

dependent Green functions
1 2
 2
x 2 2
G± x , t ; x ' , t ' = 4   x x'  t t'
c t

 2x  k 2 G± x ,  ; x ' , t ' = 4   x x ' ei  t '
1 
e±i k R

±
G R, =
2
  R
e i
d   t t'

1 R 
=   for k = nondispersive
R c c

±
G x ,t;x' ,t ' =
 t'
[ t
x
c
x '
]  G  : retarded Green function
x x ' G : advanced Green function
 the retarded Green function has a causal behavior: an effect observed at the
point x at time t is caused by the action of a source a distance R away at an
earlier or retarded time t'=t-R/c. Similar with the advanced Green function.

± x , t = G ± x , t ; x ' , t ' f x ' , t ' d 3 x ' d t  homogeneous solution

2 1 2
t 
 2 2
 in x , t = 0
c t
 x , t = in x , t   G
 3

x ,t ;x' ,t ' f x' ,t' d x' d t
1 2
t
2 2 2
 out x , t =0
c t

 x , t = out x , t  G x , t ; x ' , t ' f x ' , t ' d3 x ' d t
G+ guarantees at remotely early times there is no contribution from the integral.
G
assures that no signal from the source exists after the source shuts off.

[ f x ' , t ' ]ret evaluated at the retarded time

usually  in = 0
 x , t =  x x '
d3 x ' 
t ' =t
x x '
c
6.5 Retarded Solutions for the Fields: Jefimenko's Generalizations
of the Coulomb and Biot-Savart Laws; Heaviside-Feynman
Expressions for Fields of Point Charge
Use the retarded solution for the wave eqns
1 [ x ' , t ' ]ret 0 [ J x ' , t ' ]ret
x ,t =
4  0
 R
3
d x', A x,t =
4
 R
d3 x '

From the 2 eqns the E & B fields can be computed, but it is often useful to have
retarded integral solutions for the fields in terms of the sources.

[ ]
2
1  E 1 1 J 1 1 1 J
2 E
c2  t 2
=
0
 
c2  t
E x,t =
4  0
R '
c2  t '
3
d x'

ret
2
1  B 0 [  ' × J ]ret
2
 B 2
c t
2
= 0  × J
4
B x ,t =
R
 d3 x '

 ' [ f ]ret [  ' f ]ret : ' in the retarded bracket is a spatial gradient in x' with t'
fixed; ' outside the retarded bracket is a spatial gradient in x' with x and t fixed.

[ f x ' , t ' ]ret = f x ' , t R t ' / c = f x ' , t '  [t t' R t ' /c ] d t ' ,  = R/ R
R
 R  
R
[ ' ]ret =  ' []ret [ ] ' t =  ' []ret [ ]
t' c t ' c

ret ret

J R J 
R
[ ' × J ]ret =  ' ×[ J ]ret [ ] × ' t =  ' ×[ J ]ret [ ] ×
t' ret c t' ret c


E x,t =
4
 1
0
[ [  x ' , t ' ]ret
R

R
2

[  x' ,t '
t' ] ret cR

R
c2 R
1
[  J x' ,t '
t' ]]
ret
d3 x '

[ [ ] ]
0 
R  J x' ,t ' 
R
B x ,t = [ J x ' , t ' ]ret × 2
 × d3 x '
4 R t ' ret cR
Jefimenko's generalizations of the Coulomb and Biot-Savart laws
1 R 3
E x =
4  0
 x'
R

2
d x'

for  , J are time-independent
0 R 3
B x =
4

J x' × 2d x'
R

[ f x' ,t'
t' ] ret
=

t
[ f x ' , t ' ]ret  t ' = t
x
c
x '

 x ' , t ' =q  [x ' r0 t ' ] , J x ' , t ' =  v t ' fora point charge

[[ ] [ ]]
  Heaviside-Feynman expressions
q R 1  R v/c
E= 2
 vR
4  0 R c t R =1 retardation factor

ret ret
2  c
B=
0 q
4 [[ ]R

v× R
2
ret

1 
c t [ ]]

v× R
R ret
 [t t '
R t'
c
]=
 [t '  t ]

 there is a difference between []ret/t and [/t]ret because x'r0(t'). The
fields are functions of x and t, with t'=t-|x-r0(t')|/c.
dt 1 dt' 1 dR dR
=
= =1  v t' =
dt'  dt c dt dt'

2

E=
q
4  0 [[ ]
R

R
2
ret

[ R]ret 
c t R[ ]
R
2
ret

1 2
2
c t
2

[ R] ret
]  Feynman's expression

B=
0 q
4 [[ ]
2

v× R
 R 2
ret

1 
c [ R]ret  t [ ]] 
v× R
 ret
 Heaviside's expression

The sets are equivalent.

6.6 Derivation of the Equations of Macroscopic Electromagnetism
present a serious derivation of them from a microscopic starting point.

The derivation remains within a classical framework.

For dimensions large compared to 10-14m, the nuclei and electrons can be
treated as point systems,
b
b = 0,  ×e =0
t  microscopic Maxwell equations
1 e
e = ,  ×b 2
= 0 j
0 c t
no corresponding fields d and h because all the charges are included in and j.

The microscopic EM fields produced by these charges vary extremely rapidly.

The spatial variations occur over distances <10-10m, and the temporal fluctuations
occur from 10-13s for nuclear vibrations to 10-17s for electronic orbital motion.

All the microscopic fluctuations are averaged out, giving smooth and slowly
varying macroscopic quantities, to appear in the macroscopic Maxwell equations.

Only a spatial averaging is necessary because in a characteristic length there

are still many charges, but the associated characteristic time is about the range of
atomic & molecular motions, not appropriate to average out in time.
 The spatial average of a function with respect to a test function is defined as
! F x , t "= f x ' F x x ' , t d3 x '

[
3
f x = 3/ 4  R , r#R

2 examples
0, r$R
isotropic 2
r /R
2

f x =e  R2 3/ 2

A test function only needs to have

general continuity and smoothness
properties that permit a rapidly converging
expansion over distances of atomic dimensions.
 F F  F
! F x , t " = f x ' x 3
x ' , t d x ' =! ", ! F x , t " =! "
 xi  xi  xi t t
The space & time differentiation commute with the averaging operation.

E x , t = !e x , t " , B x , t =!b x , t "

! b "= 0  B = 0 0 E = ! x ,t "
 free
= % qj  x xj , bound
= % n
x,t  n
x , t =% q j  x xj
j free n molecues j n

x , t = % q j  [x x j t ]= free
 bound

! n
x , t " = f x ' n
x x ' , t d3 x ' = % q j  f x '  x x' xjn x n d3 x '
j n

=% qj f x xj n xn  x j n : order of
j n atom dim
=% qj [ f x xn x j n f x x n
j n

1 2 f
 % xjn & xjn ' x x n  ]
2 &'  x &  x'

molecular multipole moments :

molecular charge qn = % q j
j n

molecular dipole moment pn = % q j x j n

j n

molecular quadrupole moment

Q ' n & ' =3 % q j x j n & xjn '
j n
 1 2 f

! n
x , t " =q n f x xn pn f x xn  % Q 'n &'
x x n 
6 &'  x&  x '
1 2
=!q n  x xn " !p n  x xn "  % ! Q 'n &'
 x x n " 
6 &'  x&  x '
we can view the molecule as a collection of point multipoles.
The details of the molecular charge distribution is replaced in its effect by a
sum of multipoles for macroscopic phenomena.

Fourier transforms is an alternative approach to the spatial averaging

1
g x,t =
2
3  g( k , t ei kx d3 k and g( k , t =  g x ,t e i kx
d3 x

 * [ f )F ]= * [ f ] * [ F ]

convolution
faltung
f )F   
f y F x y dy

* [ f F]=
* [ f ])* [ F]
2
1

!F x , t "=
2
3  (f k F
( k , t ei kx d 3 k
* [! F x , t "]= (f k F
( k ,t

 f x d3 x = 1
(f 0 =1 =  f x e
i 0x
d3 x
3/2 r 2 / R2 k 2 R2 / 4
f= R
2
e Gaussian test function
(f k = * [ f x ]= e
 the Fourier transform of the averaged quantity contains only low wave numbers,
up to kmax=O(l/R), the inverse of the length scale of the averaging volume.

(f k  1  1
*[(F(x,t)] gives a true representation of the

long-wavelength aspects of (F(x,t).
i k x ' xn
* [! n
x , t "]= (f k ( n k , t  ( n k , t = n
x',t e d3 x '
the support for the product is confined to comparatively small wave numbers

( n k , t = ( n 0, t k k ( n 0, t =  n
x ' , t [1 i k x ' x n ] d3 x '

( n k , t + qn i kp n  quadrupole & higher
1

! n
x , t "=
2
3  (f k [ q
n
i kp n ] e
i k x xn
d3 k

=q n f x xn p n f x x n 
=! qn  x xn " ! p n  x x n " 

The Fourier transform has the advantage of giving a

complementary view of the averaging as a
cutoff in wave number space.
 2

! x , t "=  x , t P x , t  % Q ' & ' x , t 
&'  x&  x '
 x , t =! % qj  x xj  % qn  x x n "  macroscopic
j free n molecules charge density

where P x , t = ! % pn  x xn "  macroscopic

n molecules polarization
1
Q '& ' x , t = ! % Q 'n &'
 x xn "  macroscopic
6 n molecules quadrupole density

%
&

 x& [ 0 E &  P & %
'
 Q '& '
 x'
 = 
]  0 E = ! x ,t "

 Q '& '

D & = 0 E &  P & %   macroscopic displacement vector
'  x'

j x , t =% qj vj  x x j t = j free  j bound
j free = % qj vj  x xj

j free

j bound = % jn x , t  jn x , t =% q j v j  x xj
n molecues j n

! j n x , t " = % q j v j n  vn f x xn xj n
j n
The final result for the averaged microscopic current density is
 
! j& x , t " = J & x , t  [ D& x , t 0 E & x , t ]  % & ' , M, x , t
t ',  x'

% ! % [ pn & vn ' pn ' vn &]  x xn "
'  x' n molecules
2
1 
% ! % [ Q 'n &'
vn , Q 'n ,'
vn &]  x x n " 
6 ',  x'  x, n molecules

J x , t =! % qj v j  x xj  % qn vn  x x n "  macroscopic
j free n molecules current density

where M x , t = ! % mn  x x n "  macroscopic

n molecules magnetization
qj
where m n = % xj n×xj n  molecular magnetic moment
j n 2
D 1 E
 ×H =J ,  ×B 0 =! j "
t 0 t
1

B& H &= M&  ! % pn × v n &  x xn "
0 n molecules

1 
% & ' , ! % Q 'n -'
vn ,  x x n " 
6 ' ,-  x- n molecules
 Except the 1st term of RHS, the other terms are small since v's are small.

But if the medium as a whole 1

vn = v
B H = M  D 0 E × v
has a translational velocity v 0
For a medium in motion the electric polarization (and quadrupole density) enter
the effective magnetization.

Q& '   3 x & x ' r 2  & ' d3 x traceless molecular quadrupole moment
e rn=% qj x j n
2 2 2

Q 'n &'
= Qn &'
 e r n & ' 
j n

1

Q ' & ' = Q& '  ! % e r 2n  & '  x xn "  macroscopic quadrupole density
6 n molecule

1

   free  ! % qn  x xn "  2 ! % e r 2n  x xn " **
n molecule 6 n molecule

the traceless quadrupole density replaces the quadrupole density and the charge
density is augmented by an additional term.

The molecular charge & mean square radius terms represent the 1st two terms
in an expansion of the =0 molecular multipole as we go beyond the static limit.
sin k r 1
F k 2   x !ei kx " =0 part d 3 x =  x d3 x +   d 3 x  k2  r2  d3 x
kr 6

k. i
**
6.7 Poynting's Theorem and Conservation of Energy & Momentum
for a System of Charged Particles and Electromagnetic Fields
For a single charge the rate of doing work by external EM fields is qv/E. The
magnetic field does no work, since the magnetic force is perpendicular to v.

the total rate of doing work by the fields in a finite volume is V JE d 3 x
It represents a conversion of EM energy into mechanical or thermal energy. It
must be balanced by a corresponding rate of decrease of energy in the EM field.
D
 V
JE d x =
3
 V
[ E × H E
t
3
]d x  Ampere-Maxwell law

=  V [  E × H  E
D
t
 H
B
t ]
d 3 x   E × H = H × E
 × E B/ t =0
E × H

1 the medium is linear

 V
JE d x =
3
 V [ u
t
  E × H d x
] 3
 1
u  ED  BH
2
u

 S= JE Poynting theorem  S E × H Poynting vector
t
Since only its divergence appears in the conservation law, the Poynting vector
seems arbitrary to the extent that the curl of any vector field can be added to it.
Relativistic considerations show that it is unique.
 the time rate of change of EM energy in a volume, plus the energy flowing out
through the boundary surfaces of the volume per unit time, is equal to the
negative of the total work done by the fields on the sources within the volume.
d E mech 0
=V JE d x , E field =V u d x = V
3 3 2 2 2 3
E c B d x
dt 2
dE d E mech d E field

=  = 0S nS d a Poynting's theorem
dt dt dt
d P mech
EM force on a charged particle F= q E  v × B
= V  E  J ×B d 3 x
dt
1 E
=0 E , J= ×B 0
0 t
E

 E  J × B= 0 [ E E  B × c 2 B×  × B ]
t

= 0 [ E E c 2 B B E× ×E c 2 B×  × B E×B ]
t
d P mech d P field


dt

dt
[  P field 0 V E × B d 3 x = 0 0 V E × H d 3 x ]
= 0 V [E E  c B B
2 2 3
E× ×E c B×  × B ] d x
 1 1
the density of electromagnetic momentum g  2
E × H =  0 0 E × H = 2
S
c c
g is the electromagnetic momentum associated with the fields.
 E1  E2  E3  E2  E1  E1  E3
[E E E ×  × E ]1 = E 1   E2  E3
 x1  x2  x3  x1  x2  x3  x1
   1 
= E 21  E1 E2  E1 E3 E 21  E 22  E 23
 x1  x2  x3 2  x1
 1

[ E E E ×  × E ]& = % E& E ' EE & '
'  x' 2
1

T & ' 0 [ E & E '  c 2 B& B ' EE  c 2 BB  & ' ]  Maxwell stress tensor
2
conservation of
d 

Pmech  P field & = % V T & ' d x =0S % T & ' n ' d a  linear momentum
3

dt '  x' '

n : outward normal
1T&'n' is the &th component of the flow per unit area of momentum across the
surface into the volume, ie, it is the force per unit area transmitted across the
surface to act on the combined system of particles and fields inside the volume.

The conservation of angular momentum of the system of particles & fields can
be treated in the same way as we have handled energy & linear momentum.
6.8 Poynting's Theorem in Linear Dispersive Media with Losses
Poynting's theorem was derived with linear media with no dispersion or losses,
with  and  real and frequency independent. Actual materials exhibit dispersion
and losses.

E x ,t =  

E x , e it
d
with the linearity D x ,  =  E x , 
D x , t = D x, e i t
d [B x ,  =   H x ,  ]


reality of field
E x ,  = E * x ,  , D x ,  = D* x ,  ,   = * 
D D  ED

E
t
=  d  E *
 ' E  [ i    ] e
i  ' t
d'  E
t

t 2
1
=
2
 
d E *  ' E  [i  ' *  ' i  ]e i  ' t
d'   
'
'

assume E is dominated by frequency around  ' + 
 ' =      =  ' 
* d *

i'  ' i    =2  2 [  ] i   ' [   ]
d
D

E
t
+   d
*
E  ' E   2 [  ] e
i  ' t
d '

1  d

2 t
  d *
E  ' E 
d
[ *  ] e i  ' t
d' #
 B

H
t
+  
d
*
H  ' H   2 [  ] e
i  ' t
d '

1  d

2 t
  d H*  ' H 
d
[ *  ] e i  ' t
d' @

If  and  are real and frequency independent we go back the the last section.

The 1st terms in (#) & (@) evidently represent the conversion of electrical
energy into heat, while the 2nd terms must be an effective energy density.
( t cos 0 t  &
E=E ( 1
Let , and E t slowly varies relative to
( t cos 0 t '
H=H ( t
H 0
D B 2 2
 ueff

! E  H " = 0 {2 [  0 ] ! E x , t "  2 [ 0 ] !H x , t "}
t t t
1 d 2 1 d  2
where u eff = 3[ 0 ] ! E x , t "  3[ 0 ] ! H x , t "
2 d 2 d
Poynting's theorem in these circumstances reads
 u eff 2 2
 S = JE 0 {2 [ 0 ] ! E x , t "  2 [  0 ] ! H x , t "}
t
The 1st term in RHS describes the explicit ohmic losses, while the next terms
represent the absorptive dissipation in the medium, not counting conduction loss.
6.9 Poynting's Theorem for Harmonic Fields;
Field Definitions of Impedance and Admittance
assume that all fields and sources have a time dependence e-it
i t 1 it
E x , t = 3 [E x e ] [E x e  E* x ei  t ]
2
1 it * i t it * it

J x , t E x , t = [J x e J x e ][E x e E x e ]
4
1 2it
= 3 [ J * x E x  J x E x e ]
2
For time averages of products, the convention is to take 1/2 of the real part of
the product of one complex quantity with the complex conjugate of the other.
B= 0,  ×E i  B=0
For harmonic fields the Maxwell equations
E = ,  × H  i  D = J

the time-averaged rate of work done by the fields !

dW
dt
1
"= 3
2 V
J *E d3 x [ ]
 J *E d 3 x =
V

E[ × H* i  D * ] d 3 x =
V

[i  BH * ED*  E × H * ] d 3 x
V

1 1 1
S E × H* , w e = ED* electric , w m = BH* magnetic
2 4 energy density 4 energy density
1


2
 V
* 3
J E d x  2 i   V
we wm d x 
3
0 Sn d a = 0
S
Poynting theorem
 The real part of the eqn gives the conservation of energy for the time-averaged
quantities and the imaginary part relates to the reactive or stored energy and its
alternating flow.
1
If  V
we w m d 3 x is real

2
 V
3 [ J *E] d3 x = 0 3 [Sn] d a
S

with lossless dielectrics and perfect conductors, the steady-state, time-averaged

rate of doing work on the sources by the fields is equal to the average flow of
power into the volume through the boundary surface.
1
the complex power input
2
I *i V i = 0 Sn d a
Si
1 1


2
I *i V i =
2
 V
J *E d 3 x

2 i   V
we wm d3 x  0 S Si
Sn d a
If S-Si is taken to infinity, the surface integral is
real and represents escaping radiation.

At low frequencies the radiation is generally

negligible. No distinction need be
made between Si and S.

The input impedance

Z=R iX
V i= Z Ii
 assume the power flow out through S is real

R=
1
I i
2 { 
3[
V
J *E d 3 x ]2 0 S Si
Sn d a 4  2 [  V
wm we d 3 x ]
}
X=
1
I i2
{
4  3[  V
wm we d x ]
3
2[  V
*
J E d x ]
3
}
2
I
2 0
S Si
Sn d a is the "radiation resistance," important at high frequencies.
i
At low frequencies, ohmic losses are the only appreciable source of dissipation.
1 4

R4
I i
2  V
- E2 d 3 x , X4 2
I i
 V
wm we d3 x

Here - is the real conductivity, the energy densities are also real essentially.

The different frequency dependences of the low-frequency reactance for

inductances (X=L) and capacitances (X=-1/C) can be traced to the definition
ofL in terms of current and voltage (V=LdI/dt) and of C in terms of charge and
voltage (V=Q/C).
6.10 Transformation Properties of Electromagnetic Fields and
Sources Under Rotations, Spatial Reflections, and Time Reversal
A. Rotations
A rotation in 3d is a linear transformation of the coordinates of a point and the
sum of the coordinates' squares remains invariant  orthogonal transformation.
1 T
x ' & = % a& ' x ' a = a
vector
% a& ' a& , = ' ,  2
x ' =x 2


2
' & a = 1  a  det a
a =1 : proper rotation, obtainable from the original configuration

by a sequence of infinitesimal steps
a = 1 : improper rotation, a reflection plus a rotation
Scalars are invariant under rotations and so are
tensors of rank zero.

B ' & ' = % a & , a '  B,  are called 2nd-rank tensors or,
,,
commonly, tensors, like Maxwell stress tensor.

the active view of rotation—the coordinate axes are

considered fixed and the physical system is imagined
to undergo a rotation.

The electrostatic potential is a scalar under rotations

 If a physical quantity is a function of coordinates and when the physical system
is rotated, the quantity is unchanged, then it is a scalar function under rotations.
xx'
6 ' x ' =6 x
If a set of physical quantities transform under rotation of the system according
to V ' & x ' = % a& ' V ' x then they form the components of a vector.
'
the gradient of a scalar transforms as a vector, the divergence of a vector is a
scalar, and the Laplacian operator is a scalar.

A = B×C
A& = % & ' , B' C ,
A ' & = a % a& ' A '  a  det 7a & '7
', '

Under proper rotations, the cross product transforms as a vector.

B. Spatial Reflection or Inversion

Spatial reflection in a plane is to change the signs of the 5 components of the

vectors of all points and to leaving the components
to the plane unchanged.
For x - y plane: x = x , y , z  x ' = x , y , z
Space inversion corresponds to reflection of all 3 components of every vector
through the origin, x  x '= x
Spatial inversion or reflection is a discrete transformation that cannot in general
be accomplished by proper rotations.
a= 1
a& ' = & ' for inversion operation
 Polar vectors: V ' & x ' = % a& ' V ' x & x x'= x
V  V '= V
'

Axial or pseudovectors: A ' & = a % a& ' A' & x  x ' = x
A A'=A
'

speak of scalars or pseudoscalars, depending on whether the quantities do not or

do change sign under spatial inversion.

The triple scalar product a(b8c) is a pseudoscalar if a, b,
are polar vectors.

If a tensor of rank N transforms under spatial inversion with a factor (-1)N, it is a

tensor, while if the factor is (-1)N+1 it is a pseudotensor of rank N.

C. Time Reversal

The basic laws of physics are invariant to the sense of direction of time.

under the time reversal transformation, the related physical quantities

transform consistently so that the form of the equation is the same as before.
dp
t t'= t
= U x invariant  x  x '=x , p  p ' = p
dt
if an initial configuration of a system evolves into a final configuration, a
possible state of motion of the system is that the time-reversed final configuration
will evolve over the reversed path to the time-reversed initial configuration.
Physical Quantity Symbol Rotation Space Inversion Time Reversal
(rank of tensor) name
Coordinate x 1 Odd(vector) Even
Velocity v 1 Odd(vector) Odd
Momentum p 1 Odd(vector) Odd
Angular Momentum L=x×p 1 Even(pseudovector) Odd
Force F 1 Odd(vector) Even
Torque N=x×F 1 Even(pseudovector) Even
2
Kinetic energy p /2 m 0 Even(scalar) Even
Potential energy U x 0 Even(scalar) Even
Charge density  0 Even(scalar) Even
Current density J 1 Odd(vector) Odd
Electric field E
Polarization P 1 Odd(vector) Even
Displacement D
Magnetic induction B
Magnetization M 1 Even(pseudovector) Odd
Magnetic field H
Poynting vector S= E × H 1 Odd(vector) Odd
Maxwell stress tensor T& ' 2 Even(tensor) Even
D. Electromagnetic Quantities
the forms of the eqns governing EM phenomena are invariant under rotations,
space inversion, and time reversal.

electric charge is invariant under Galilean & Lorentz transformations and is a

scalar under rotations. Thus people assume that charge is also a scalar under
spatial inversion & time reversal.

The transformation properties of fields like E &  depend on the convention

chosen for the charge.

charge density is a scalar. And E is a polar vector, even under time reversal.
B
 × E : pseudovector, even under time reversal   ×E =0 Faraday's law
t

B : pseudovector, odd under time reversal
1 E
the Ampere-Maxwell eqn  × B 0 = J polar vector, odd under time
reversal on both sides. 0 t

consider the structure of a constitutive relation specifying the polarization P,

assuming 1st order in E, 2
E  E
B0 : E × B0, × B0 , 2
× B0,  odd time derivatives of E
t t
2 E zero or even time derivatives of E
B : B0B0 E ,
0 EB 0 B0 , B0B0 ,
t
 1 E
Up to 2 -order of B0

nd
P = :0 E  :1 × B0 :2 B0B0 E :3 EB0 B0 
0 t
odd time derivatives of E for the terms linear in B0
even time derivatives of E for the 0th and 2nd powers of B0.

At low frequencies, the response of all material systems is via electric force
1 E 2 E 2 E

P = :0 E  :1 × B0  : ' 2 B 0B0 2
 : '3 2
B0 B0  more realistic
0 t t t
At optical frequencies this equation permits an understanding of the gyrotropic
behavior of waves in an isotropic medium in a constant magnetic field.

In certain circumstances the constraints of space-time symmetries must be

relaxed in constitutive relations. The optical rotatory power of chiral molecules
B E
has the constitutive relations, P = 0 :0 E 9 and 0 M = : ' 0 B  9 '
t t
The terms involve pseudoscalar quantities 9 and 9' that reflect the underlying lack
of parity symmetry for chiral substances.
 6.11 On the Question of Magnetic Monopoles
No experimental evidence for the existence of magnetic charges or monopoles.

Dirac's argument is that the existence of magnetic monopole would offer an

explanation of the discrete nature of electric charge.

If there exist magnetic charge & current densities, then Maxwell eqns become
D  e
D = e ,  ×H=  Je  J e = 0
t  t
B  m
B =  m ,  ×E=  Jm  J m = 0
t t
transformation E =E ' cos 9  Z 0 H ' sin 9 , Z 0 D =Z 0 D ' cos 9  B ' sin 9
duality transformation:

;
impedance Z 0 = 0 / 0
Z 0 H = E ' sin 9  Z 0 H ' cos 9 , B = Z 0 D ' sin 9  B ' cos 9
For a real (pseudoscalar) angle 9, the transformation leaves quadratic forms like
E8H, (ED+H), and the components of the Maxwell stress tensor invariant.

Z 0 e =Z 0  ' e cos 9   ' m sin 9 , Z 0 J e =Z 0 J ' e cos 9  J ' m sin 9
Apply to the sources
m = Z 0  ' e sin 9   ' m cos 9 , J m = Z 0 J ' e sin 9  J ' m cos 9

the generalized Maxwell equations are invariant
D'  B'
D ' =  ' e ,  × H ' =  J ' e , B ' =  ' m , ×E ' =  J 'm
t t
 The invariance of the EM eqns under duality transformations shows that it is
only a convention to think of electric charge instead of magnetic charge.

Ifall particles have the same ratio of magnetic to electric charge, one can make
a duality transformation so that m=0, Jm=0, so we have the usual Maxwell eqns.

If we choose the electric and magnetic charges of an electron qe=-e, qm=0, it is

known that for a proton, qe=+e and |qm(nucleon)|<2810-24Z0e.

The conclusion is that the particles of ordinary matter possess only electric
charge or they all have the same ratio of magnetic to electric charge.

m is a pseudoscalar density, odd under time reversal, and

Jm is a pseudovector density, even under time reversal.

Since the symmetries of m under spatial inversion & time reversal are opposite
to those of e, then a particle with both electric & magnetic charges that space
inversion and time reversal are no longer valid symmetries of the laws of physics.

the present evidence is that the symmetry violation is extremely small and
associated with the weak interactions.
 Considering the quantum mechanics of an electron near a magnetic monopole,
Dirac showed that consistency required the quantization condition,
g : magnetic charge
eg &g n
= = n = 0, ± 1, ± 2,   e2 1
4< Z0 e 2 &= + fine structure const
4  0 < c 137
The discrete nature of electric charge thus follows from the existence of a
monopole. The magnitude of e is determined in terms of the magnetic charge g.
With the known value of the fine structure constant, we can also infer the
existence of magnetic monopoles with the magnetic "fine structure" constant
g2 n2 4  0 < c 137
= 2
= n2  Dirac monopoles
4  0 < c 4 e 4
The coupling strength is enormous, making their extraction from matter with dc
magnetic fields and their subsequent detection very simple in principle.
6.12 Discussion of the Dirac Quantization Condition
We consider the deflection at large impact parameters of a particle by the
field of a stationary magnetic monopole.

For b=0, the change in the state of motion of the charged

particle can be determined by the impulse of the force,
assuming the particle is undeflected.
Fx+ 0
g r
B= 3

eg vb
4 r Fy= e v Bx=
4  b2  v 2 t 2 3 / 2

egvb dt eg

> py =
4
  2 2 2 3/2
b v t
=
2b

the particle is deflected out of the plane, ie,

in the azimuthal direction. So the particle's
angular momentum is changed
eg
> Lz = b > py =
2
The change in the z component of angular momentum of the particle depends
only on the product eg and is a universal value for a charged particle passing a
stationary monopole, no matter how far away.
 If we assume that any change of angular momentum
must occur in integral multiples of <, we are led to the
Dirac quantization condition. e g eg n
=n <
=
2 4< 2
The universal character of the change in the angular
momentum of a charged particle in passing a magnetic
monopole is from the angular momentum contained in
the fields of an electric charge in the presence of a
magnetic monopole.
g 1 g n' 1
H=
4  0

r'
=
4  0 r ' 2
Pem 
c2
 3
E×H d x

1
E=
e
4  0

1
r
=
e
4  0 r
n
2
Lem 
c2
 x × E × H d3 x

P em =0  by symmetry
L em is independent of choice of origin

4 1 1


0
L em = e r  r [n nH H] d x
n × n × H d3 x = e 3

4  Lem = e  B n d x  B=  H , r H n = H n nH

3
0

= e  n B d x e 0 n Bn d a  integration by parts

3
S
S
 g
0 S
n Bn S d a =
4
0 n d ? = 0  n : radially directed and
no angular average
L em =
eg R
4 R
B = g  x R
In the collision process, the total angular momentum of the system (the particle
& the EM field) is conserved, ie, the change in the angular momentum of the
particle is balanced by the change in the EM angular momentum caused by the
reversal of the direction R.

Dirac's original argument: the

standard Hamiltonian interaction
e e2
H int = e pA  AA
m 2m
The magnetic charge is imagined to
be the end of a line of dipoles or a
tightly wound solenoid that stretches
off to infinity.
0 1
dA x = d m×
4 x x '
g 1

AL =
4

L
d ×
x x '
 the vector potential has a curl that is directed radially outward from the end of
the string, varies inversely with distance squared, with total outward flux g.

On the string the vector potential is singular, equivalent to a field B' in the
solenoid and bringing a return contribution of flux -g in along the string.

Bmonopole =  × A B'  B ' : exists only on the string (inside the solenoid)

to describe the interaction of the electron with a magnetic monopole, rather

than with a solenoid, Dirac asked that the electron never "see" the singular
field B', so the electronic wave function to vanish along the string. [Not Good]

a choice of different string positions is equivalent to different choices

of gauge for the vector potential.

The difference of the 2 vector potentials

g 1
> A=
4
 C
d  ×
x x '
 C=L ' L

g

AL ' x = AL x   ?C x  Problem 5.1
4
A  A ' = A  : ?C
1 :
:= g
 '= 4
c t
 a change in the gauge of the EM potentials leaves the form of the Schrodinger
en invariant, provided the wave function is transformed according to
e g ?C

[ ]
i
> ?C = 4  as an electron
i e :/ < 4<
[@  @ ' = @ e
@  @ ' =@ e ] 
crosses the surface S
eg

=2  n n = 0, ± 1, ± 2,   the wave function is single-valued
<
The arguemnt follows from the general requirements of gauge invariance and
single-valuedness of the wave function, independent of the location of the
monopole's string.
6.13 Polarization Potentials (Hertz Vectors)
use potentials as auxiliary fields to determine the fundamental EM fields.
D=  E  P ext
 P ext , M ext : external polarization densities
B =  H  0 M ext
2 A 2
 P ext
  A =  0  × M ext
t
2
t 

 A    =0

2
2 1 t
 2
 = P ext
t 
 Ae
A =:   0  × A m

t  A e , A m : vector polarization potentials
1 Hertz vectors
=: A e


[
  2 A e 
2 A e
t
2 ]
 P ext = 0



t [ 2 Ae 
2 A e
t
2 ] [
 P ext  0  ×  2 A m 
2 A m
t
2 ]
 M ext = 0
 2 A e 2 0
 2
 A e = Pext  ×V

t 
2
 Am 2 V 9
 2
 A m = M ext  
t t t
2 A e
  2 A e = Pext

 t2  the arbitrary functions V & 9 may be
2 A m 2
removed by a gauge transformation [Problem 6.23]
 2
 A m = M ext
t
2
1  Ae  Am  Am 1 1
E=  A e  2
0  × = 0  ×   ×  × Ae P ext

 t t t  
 Ae
B=  ×  0  ×  × A m
t
For a simple case
M ext = m t  x x0
Am x , t =
m t ;  R  R =x x 0
4R
Selected problems: 6.1, 6.3, 6.10, 6.14, 6.18, 6.21