Maxwell's Equations

I Electrodynamics before Maxwell

~E
∇· ~ = ρ Gauss's law
0
~ ·B
∇ ~ =0 Magnetic monopoles do not exist

~ = − ∂B
~
~ ×E
∇ Faraday's law
∂t
∇ × B = µ0~J
~ ~ Ampere's law

I The above equations represented the state of e.m. theory over

a century ago when Maxwell began his work.

I They were not written in so compact a form in those days but

their physical content was familiar.
Maxwell's Equations

The integral forms of Maxwell's equations

~E
∇· ~ = ρ

ˆ ˆ0
~ · Ed
~ 3r = ρ 3
∇ d r
0
˛
q
E~ · d ~S = encl The integral form of Gauss's law (1)
s 0

~B
∇· ~ =0
˛
B~ · d ~S = 0 Net magnetic ux enclosed by a closed surface

s is zero since there are no magnetic monopoles

(2)
Maxwell's Equations
I

~ = − ∂B
~
~ ×E
∇
∂t
ˆ ˆ ~
~ · d ~S = − ∂ B · d ~S
~ ×E
∇
s ˛ s ∂t
dφ
E~ · d~l = − Integral form of Faraday's law (3)
dt

~ ×B
∇ ~ = µ0~J
ˆ ˆ
∇ × B · d S = µ0~J · d ~S
~ ~ ~
S ˛
B~ · d~l = µ0 Iencl Integral form of Ampere's law

(4)
Maxwell's Equations

I Now it so happens that there is a fatal inconsistency in these

formulas. Whereas

~ =−∂ ∇
~ ×E ~ ·B
   
~ · ∇
∇ ~ =0
∂t
I But

~ ×B ~ · ~J 6= 0
   
~ · ∇
∇ ~ = µ0 ∇ in general

I For steady currents ~ · ~J = 0

∇ but evidently when we go
beyond magnetostatics Ampere's law in its present form
cannot be right.
Maxwell's Equations
B
A

I Iencl is the total current passing through the loop or more

precisely the current piercing a surface that has the loop for its
I
boundary. In this case for surface A encl =I but for surface
B? No current passes through this surface and encl I = 0.
I Just as the surface chosen is important for Eq.4 similarly it
should be for Eq.3.
Maxwell's Equations
I The reason it does not matter for Eq.3
I ˛
dφ
E~ · d~l = −
s dt
where φ is the net ux across the surface.

S2 S1

I Let A be the loop and let S1 and S2 be the two surfaces.

S1 US2 = S
¸
is a closed surface. Now ~ ·B
∇ ~ =0 or B~ · d ~S = 0 or φs = 0.
I Hence we see net ux across a closed surface is zero.
Maxwell's Equations

φs1 + φs2 = 0
φs1 = −φs2

I The negative sign arises because of the direction in which the

vector normal to the surface is taken. So irrespective of which
surface is chosen the ux remains the same.

I We did not have the which surface problem in

magnetostatics beacuse the conict arises only when charge is
piling up somewhere.

I In Maxwell's time there was no experimental reason to doubt

that Ampere's law was of wider validity. The aw was a purely
theoretical one and Maxwell xed it by purely theoretical
arguments.
How Maxwell xed Ampere's Law :

∂E
!
~
~ · ~J = − ∂ρ = − ∂ 0 ∇
~ ·E
 
∇ ~ = −∇
~ · 0
∂t ∂t ∂t

E
!
∂ ~
~ · ~J + 0
∇ =0
∂t

So we can write ( ) as curl of something with µ0 multiplied.

~ = µ0~J + µ0 0 ∂ E
~
~ ×B
∇
∂t
~ ×B
 
~
Taking divergence of the above Eq. yeilds zero:∇ · ∇ ~ =
~E
~ · ~J + ∂ µ0 0 ∇·
 
µ0 ∇ ~ = −µ0 ∂ρ + µ0 0 1 ∂ρ = 0
∂t ∂t 0 ∂ t
So this xes the problem !!
Maxwell's Fourth Equation :
I The dierential form

~ = µ0~J + µ0 0 ∂ E
~
~ ×B
∇
∂t
I The integral form

ˆ  ˆ
∂E
!
~
~ ×B
~ · d ~S = µ0~J + µ0 0 · d ~S

∇
s ∂t
˛ ˆ
∂E
~
B~ · d~l = µ0 Iencl + µ0 0 · d ~S
∂t
I Such a modication changes nothing as far as magnetostatics
is concerned: when E~ is constant we still have ~ ×B
∇ ~ = µ0~J .
In fact Maxwell's term is hard to detect in ordinary e.m.
experiments where it must compete for recognition with ~J .
Maxwell's Fourth Equation :
I Apart from curing the defect in Ampere's law, Maxwell's term
has a certain aesthetic appeal. Just as changing magnetic eld
induces an electric eld so also a changing electric eld
induces a magnetic eld.

I Maxwell called his extra term the displacement current

J~d = 0 ∂∂E~t
I Resolution of the paradox:
I If the capacitor plates are close to each other then electric eld
between the plates
σ Q
E= =
0 0 A
∂E 1 dQ I
= =
∂t 0 A dt 0 A
Now ˛ ˆ
∂E
~
B~ · d~l = µ0 Iencl + µ 0 0 · d ~S
∂t
Maxwell's Fourth Equation :

I If we choose the at surface E~ = 0.

I I ´ = 0 but
For the baloon surface encl
´ ∂ E~
µ0 0 ∂ t · d S = µ0 0 0IA d ~S = µ0 I
~
A Bit of History :
I Maxwell axiomatised EM theory: realised that these four
equations (with a modication) along with the Lorentz force
law can explain all of EM theory (99% of the observable
world).
I Maxwell noted the which surface ambiguity in Ampere's law,
corrected it by introducing the displacement current (1865).
I The full set of Maxwell's equations predicted the exsitence of
E
EM waves ( ~ and B~ in free space obey the 3D wave equation)
The speed predicted by his equations (using the measured
values of electrical constants) was about 3.2 × 108 m/sec.
I A few years ago in 1849 H. Fizaue had determined the speed
of light to be 3.13 × 108 m/sec. Note that this was actually
less accurate than the astronomical method of the 17th
century Dutch astronomer Ole Romer.
I The close similarity of speeds prompted Maxwell to assume
that light is an e.m.wave.
I Calculation of properties of light, assuming that it is an e.m.
wave, closely matched experimental results.
A Bit of History:

I But do e.m waves really exist?

I In 1885 H. Hertz sloved Maxwell's equations with sources to

gure out what kind of sources would produce e.m. waves.

I based on these calculations Hertz built the rst receiver (1886)

and the rst em transmitter (1887). The waves produced were
just like light only with much larger λ(∼ 4m).
EM waves:
I Assume free space (ρ = 0,~j = 0)
I Maxwell's equations in free space:

~ ·E
∇ ~ =0 (5)

~ ·B
∇ ~ =0 (6)

~ = − ∂B
~
~ ×E
∇ (7)
∂t
~ = µ0 0 ∂ E
~
~ ×B
∇ (8)
∂t
Taking curl of Eq.7we have

~ =−∂ ∇
~ ×E ~ ×B
   
~ × ∇
∇ ~
∂t
E
!
∂ ∂ ~
~ ·E
~ −∇ E
 
~ ∇ 2 ~ =−
∇ µ0 0
∂t ∂t
∂2E
~
∇2 E
~ = µ0 0
∂t 2
EM waves :
I Similarly by taking curl of Eq.8 we get

∂2B
~
∇2 B
~ = µ0 0
∂t 2
I We see that in vacuum each cartesian component of E~ and
B~ satises the three dimensional wave equation
1 ∂2f
∇2 f =
v 2 ∂t 2
I So Maxwell's equations implt that empty space supports the
propagation of em waves travelling at a speed

1
v=√ = 3 × 108 m/sec
µ0 0

which happens to be precisely the velocity of light.

1D Wave equation :
I Let us now try to construct a simple equation whose solution
will give wave like behaviour.
I Superposition of disturbances travelling towards right and left
with velocity v.
u (x , t ) = f (x − vt ) + g (x + vt )
∂2u 1 ∂2u
=
∂x 2 v 2 ∂t 2
Using the method of separation of variables,
u (x , t ) = U (x ) T (t ) we have
T (t ) = exp (±ikvt )
and
U (x ) = exp (ikx )
where k can be positive or negative.

u (x , t ) = A exp (i (kx ± ωt ))
for waves travelling towards left and right. Here ω = kv
3D Wave equation :

I Let us consider a monochromatic em wave. Suppose moreover

that the waves are travelling in the x direction and have no y
or z dependence. These are called plane waves because the
elds are uniform over every plane perpendicular to the
direction of propagation.

∂2E
~ ∂2E
~
= µ 0 0
∂x 2 ∂x 2
E = E0 exp (i (kx − ωt ))
~ ~

At an instant t constant phase surface implies xed

x =⇒ plane normal to the X axis. These are called wavefronts.
Plane progressive wave in arbitrary direction :
I Let us consider a plane wave travelling in an arbitrary direction
X0
E~ =E
~ 0 exp i kx 0 − ωt

~ 0 exp i ~k · ~r − ω t
=E
  

I ~k · ~r in rotated frame=kx 0 .
I Wavefront ~k · ~r = const.= ω t + φ.
I At xed t this is the equation of a plane normal to ~k , whose
ω t +φ
distance from the origin = .
|~k |
ω∆t
I t → t + ∆t the plane shifts by
|~k |
I So we get a plane wavefront normal to ~k moving with a speed
ω
.
|~k |
Helmholtz Equation :

I Now in 3D the wave equation

1 ∂2u
∇2 u =
v 2 ∂t 2
Using the separation ansatz u (~r , t ) = U (~r ) T (t )
T (t ) = exp (−i ωt )
where ω = ±kv . For the spatial part we have

∇2 + k 2 U=0

I This is known as the Helmholtz equation.

Helmholtz Equation :
∇2 + k 2 U=0

I Separation of variables in Cartesian coordinates leads to the

plane wave solution.
I Spherically symmetric solution (use spherical polar coordinates:
assume that U depends only on r)
I Under these conditions
d d
 
1
2
∇ ≡ r2
No ϑ, φ dependence.
r dr
2 dr

d 2 dU
 
1
r + k 2U = 0
r 2 dr dr
d 2 U 2 dU
+ + k 2U = 0
dr 2 r dr
d 2U dU
r 2 + 2 + k 2 rU = 0
dr dr
Helmholtz Equation :

rU = A exp (ikr ) + B exp (−ikr )

A B
U (r ) = exp (ikr ) + exp (−ikr )
r r
A B
U (r , t ) = exp (i (kr − ωt )) + exp (−i (kr + ωt ))
r r
I At r = 0 the solution blows up hence the source should be at
the origin.
I At t = t0 say we make some changes to the source at the
origin. Therefore

kr − ωt = −ωt0
ω t0 = ω t − kr

which implies
t0 < t
Helmholtz Equation :

I Whereas for the second part of the solution

kr + ωt = ωt0
which implies
t < t0
which is an impossibility.
Poynting Vector, Poynting's Theorem :
I Energy needed to assemble a system of charges by bringing
them in from innity:

q1 q2 q1 q3 q1−n qn
 
1
W= + + ............ +
4π0 r12 r13 r1−n,n
n n
X X qi qj n
X qi qj
1 1 1
= =
4π0 r
i =1 j =i +1 ij
2 4π0 r
i 6=j =1 ij
n n
 
1 1 qj 1
qi  qi Vi
X X
= =
2 4π0 rij 2
i =1 j =1
I For continuum
ˆ
1
W = ρ (~r ) V (~r ) d 3 r
2

From Gauss's law we have

ˆ 
0
W ~E Vd 3 r

= ∇· ~
2
Poynting Vector, Poynting's Theorem :
ˆ n
0
W ~ · EV −E ~ V d 3r
  o
= ∇ ~ ~ ·∇
2
ˆ ˆ
0 0
= V E · dS +
~ ~ E 2d 3r
2 Σ 2 Ω

I In the above integral the rst term vanishes. [If we consider a

surface at innity S ∼ r 2 where E falls o at least as
1
r2 and

V falls o at least as
1
r so the surface integral vanishes when
the surface is taken at innity]. Therefore
ˆ
0
W = E 2d 3r
2 all space
I So we have ˆ
0
We = E 2d τ
2
Similarly ˆ
1
Wm = B 2d τ
2 µ0
Poynting Vector, Poynting's Theorem :
I Total energy stored in em elds
ˆ  
1 1
Uem = 0 E 2 + B 2
dτ
2 µ0

I Suppose we have some charge and current conguration which

at time t E~ and B~ . In time dt the work done on
produces elds
a charge q (say in time dt the charges move around a bit)

dW = F~ ·d~l = F
~ · ~v dt = q E~ + ~v × B
~ · ~v dt = q E
~ · ~v dt
 

Therefore
dW
= qE
~ · ~v
dt
for a point charge.
I For a charge distribution
ˆ ˆ ˆ
dW
= ρd τ E
~ · ~v = d τ E~ · (ρ~v ) = E~ · ~Jd τ
dt V V
Poynting Vector, Poynting's Theorem :
I So E
~ · ~J is the work done per unit time per unit volume or the
power delivered per unit volume. Now

~ = µ0~J + µ0 0 ∂ E
~
~ ×B
∇
∂t

~ · ∂E
1 ~
E~ · ~J = E~ · ~ − 0 E
~ ×B
 
∇
µ0 ∂t
Now
~ · E~ ×B
~ =B ~ ×E
~ −E ~ ×B
     
∇ ~ · ∇ ~ · ∇ ~

Hence

1 ∂E
~
E~ · ~J = B · ∇ × E − ∇ · E × B − 0 E~ ·

   
~ ~ ~ ~ ~ ~
µ0 ∂t
~ · ∂ B − 0 E
~ · ∂E − 1 ∇
1 ~ ~
=− B ~ · E~ ×B
 
~
µ0 ∂t ∂t µ0
Poynting Vector, Poynting's Theorem :
 
1 ∂ 1 1
E~ · ~J = − 0 E +
2
B 2 ~ · E~ ×B
 
− ∇ ~
2 ∂t µ0 µ0
I Therefore rate of work done on charge by em eld:

ˆ ˛ 
dW d
 
1 1 1
0 E 2 + B2 dτ − E~ × B~ · d ~S

=−
dt dt 2 µ0 µ0 S

where S is the surface bounding the volume V. This is

Poynting's Theorem -the work-energy theorem of
electrodynamics.

I The energy per unit time per unit area transported by the
elds is called the Poynting vector given by

1
~S = E~ × B~
 
µ0
Maxwell's Equations in free space (no sources, vacuum):
I In free space ρ = 0 and ~j = 0. Maxwell's eqns. take the form
~ ·E
∇ ~ =0
~ ·B
∇ ~ =0

~ = − ∂B
~
~ ×E
∇
∂t
~ = µ0 0 ∂ E = 1 ∂ E
~ ~
∇~ ×B
∂t c 2 ∂t
Now

E~ =E i ~k · ~r − ωt
  
~ 0 exp

B~ = B~ 0 exp i ~k · ~r − ωt
  

~ ·E
∇ ~ = i ~k · E
~ = 0 =⇒ ~k · E
~ =0
~ ·B
∇ ~ = i ~k · B
~ = 0 =⇒ ~k · B
~ =0

This tells us that E~ and B~ are both perpendicular to ~k .

Maxwell's Equations in free space :

~ ≡ i ~k
∇
and
∂
≡ −i ω
∂t
From

~ = − ∂B
~
~ ×E
∇
∂t

i ~k × E~= i ωB
~
~k × E
~
B~ =
ω
From

~ ×B
~ = 1 ∂E
~
∇
c 2 ∂t
Maxwell's Equations in free space :

we have

iω
i ~k × B~ = − 2 E~
c
ω 2
~k × ~k × E ~ =− E
 
~
c2

2
~k ~k · E
~ −E~ ~k · ~k = − ω E
   
~
2 c
ω2
k2 =
c2
ω = ±ck
Meaning of Complex EM elds :
I The Maxwell's equations are real and linear.

I If a set of complex elds satisfy them, then so do their real

and imaginary parts.

I By convention we usually take the real part.

E~ =E i ~k · ~r − ωt
  
I So when we say ~ 0 exp with

E~ 0 = E~ 0  e i φ
 
what we really mean is that

E~ = E k r t
   
~ 
0  cos ~ · ~ − ω + φ

I In fact as long as we are calculating linear properties we can

keep on using the complex elds throughout.

I Since dealing with exponentials is easier mathematically this is

what we prefer to do.

I Calculation of nonlinear quantities require special care like the

Poynting vector where we take a product of two elds.
Average of Poynting vector:
Let

ã = a0 e i (kx −ωt )
b̃ = b0 e i (kx −ωt )
Now

Real part of ã = a = |a0 | cos (kx − ωt + φa )

Real part of b̃ = b = |b0 | cos (kx − ω t + φb )

ˆ T
|a0 | |b0 |
habitime = 2 cos (kx − ω t + φa ) cos (kx − ω t + φb ) dt
T 02

|a0 | |b0 |
= cos (φa − φb ) T
2T
Re ∗ 
ã b̃

=
2

E~ ∗ × H~
!
Therefore
~S = Re
D E
2
Conservation of momentum in EM elds :
Trouble with Newton's Third Law:

I A moving charge produces both electric and magnetic elds.

I Because it is moving the electric eld is not given by

Coulomb's law but E~ still points radially outward from the
instantaneous position of charge.

I Moreover a moving point charge does not constitute a steady

current and B~ is not given by the Biot Savart law. But B~ still
circles around the axis.

Y
Fe I The electric forces are opposite to

q2⊗ Fm
each other.

I The magnetic forces are not!!

v2 Fm I Newton's third law does not hold.
q1 I What happens to momentum
v1  X conservation?
Fe
Conservation of momentum in EM elds :

I In electrostatics and magentostatics Newton's third law holds

but not in electrodynamics.

I Does this mean that conservation of momentum also does not

hold?

I Momentum conservation is rescued in electrodynamics by the

realization that the elds themselves carry momentum.

I In the case of the two point charges whatever momentum is

lost to the particles is gained by the elds.

I Only when the eld momentum is added to the mechanical

momentum of the charges is momentum conservation restored.
Maxwell's Stress Tensor :
I The total em force on the charges in volume V
ˆ  ˆ 
F~ E~ + ~v × B~ ρd τ = ρE
~ + ρ~v × B dτ
 
= ~
V V

I Hence the force per unit volume

~f = ρE
~ + ~j × B
~

~ − µ0 0 ∂ E
" #
1 ~
~ ·E~ E ~ ×B ×B
 
= 0 ∇ ~ + ∇ ~
µ0 ∂t

I Now

∂ ~S ∂ ~ ~  ∂ E~
~ × ∂B
~
µ0 = E ×B = ×B
~ +E
∂t ∂t ∂t ∂t
∂E ~ ~
~
×B −E × ∇~ ×E
 
= ~
∂t
Maxwell's Stress Tensor :
I Therefore

~f + µ0 0 ∂ S = 0 ∇
~
~ ·E~ E~ + 1 ∇ ~ ×B ~ ×B ~ − 0 E ~ ×E
     
~ × ∇ ~
∂t µ0
~ ·E~ E~ −E ~ ×E
h   i
= 0 ∇ ~ × ∇ ~
 
1 
~ ·B
~ B ~ −B ~ ×B
   
+  ∇ ~ × ∇ ~ 
µ0 | {z }

=0

I
E~ × ~ ×E = ijk Ej ∇~ ×E
h  i  
Now ~ ~
∇
i k
= ijk Ej klm ∂l Em
= kij Ej klm ∂l Em
= δil δjm − δim δjl Ej ∂l Em

= Ej ∂i Ej − Ej ∂j Ei
1
= ∂i E 2 − Ej ∂j Ei
2
Maxwell's Stress Tensor :

~S
" #  
~f + µ0 0 ∂ 1
= 0 (∂j Ej ) Ei + Ej ∂j Ei − ∂i E 2
∂t 2
i  
1 1
+ (∂j Bj ) Bi + Bj ∂j Bi − ∂i B 2
µ0 2
   
1 1 1
= 0 ∂j (Ei Ej ) − E 2 δij + ∂j (Bi Bj ) − B 2 δij
2 µ0 2

= ∂j Tij
Therefore

~ ·←
~f = ∇ → ∂ ~S
T − µ0 0
∂t
The total force on the charges (using the divergence theorem)Pg
353 (Griths)
‰ ˆ
←
→ d
F~ = T · d~a − µ0 0 ~Sd τ (9)
S dt V
Conservation of Momentum :
I We can write Eq. 9 in the form
ˆ ‰
d ~pmech d ~Sd τ + ←
→
= −µ0 o T · d~a (10)
dt dt V S
where ~pmech is the total mechanical momentum of the
particles contained in the volume V.
I This expression is similar in structure to Poynting's theorem.
I The rst integral represents momentum stored in the em elds
ˆ
~pem = µ0 0 ~Sd τ
V

I The second integral is the momentum per unit time owing in

through the surface.
I Any increase in the total momentum is equal to the
momentum owing in due to the elds.
I If V is all of space no momentum ows in or out and
~pmech + ~pem is constant.
Conservation of Momentum :

I Let ~pmech be the density of mechanical momentum and ~pem the

density of momentum in the elds

~pem = µ0 0 ~S

I We can then write Eq. 10

∂ ~ ·←
→
T


~pmech + ~pem = ∇
∂t

←
→
I Here −T is the momentum ux density playing the role of ~J
in the continuity equation or ~S in Poynting's theorem.
Angular Momentum :
I The energy carried by the em elds

 
1 1
uem = 0 E +
2
B 2
2 µ0

and momentum

~pem = µ0 0 ~S = 0 E~ ×B
 
~

I Hence angular momentum

~lem = ~r × ~pem = 0 ~r × E~ ×B
h  i
~

I Even perfectly static elds can harbor momentum and angular

momentum as long as E~ × B~ is non zero.
Electric elds in matter :
I Electric eld can distort the charge distribution of a dielectric
atom or molecule-stretching and rotating.

I The electric eld pulls the electrons and nucleus apart their
mutual attraction drawing them together -this when it reaches
a balance leaves the atom polarized.

I The atom now has a tiny induced dipole moment ~p in the

same direction as E~ .
~p = αE
~

where α is the atomic polarizability.

I In case of polar molecules the dipole in a uniform electric eld

E~ experiences a torque

N~ = ~p × E~
N~ is in such a direction as to line ~p parallel to E~ .
I The eect is the same in both cases-the material is polarized.
Electric elds in matter :

I Polarization P~ is the net dipole moment per unit volume.

I The eect of polarization is to produce bound charge

~ ·P
ρb = −∇ ~ within the dielectric and σb = P
~ · n̂ on the
surface.

I Field due to polarization of the medium is just the eld due to

bound charge plus the eld due to free charge

~ ·E
0 ∇ ~ = ρ = ρf + ρb
~ · P
~ + 0 E
 
∇ ~ = ρf

~ ·D
∇ ~ = ρf

where D~ = P~ + 0 E~ is the electric displacement vector.

Electric elds in matter :
I Magnetic polarization or magnetization M
~ results in a bound
current
~Jb = ∇
~ ×M
~

~ = µ0~J + µ0 0 ∂ E
~
~ ×B
∇
∂t

~J = ~Jb + ~Jp + ~Jf

I ~Jp = ∂ P~
Due to change in electric polarization we have
∂t and
~Jf is the current due to free charges.

Therefore

~ ×B ∂P~ ∂E
~
∇ × M + µ0 + µ0~Jf + µ0 0
 
∇ ~ = µ0 ~ ~
∂t ∂t
B~ ~
!
∂
−M = 0 E
~ +P~ + ~Jf
 
~ ×
∇
µ0 ∂t
Electric and Magnetic elds in matter :

~ = ∂ D + ~Jf
~
~ ×H
∇
∂t
where

D~ = 0 E~ + P~
B~ ~
H~ = − M
µ0
Boundary Conditions :
I The elds E
~, B
~, D
~ and H
~ will be discontinuous at a boundary
between two dierent media or at a surface that carries charge
density σ or surface current density K~ .
I The integral form of Maxwell's equations
˛
D~ · d~a = Qf encl
˛S
B~ · d~a = 0
˛S ˆ
d
E~ · d~l = − B~ · d~a
˛
dt S ˆ
d
H~ · d~l = If encl + D~ · d~a
dt s
I Considering a wafer thin pill box straddling the surface from
the rst equation we have

D~ 1 · ~a − D~ 2 · ~a = σf a
D1⊥ − D2⊥ = σf
Boundary Conditions :
I Similarly we have

B~ 1 · ~a − B~ 2 · ~a = 0
B1⊥ − B2⊥ = 0

I Hence we see that the normal component of B~ is continuous

and the normal component of D~ is continuous only in the
absence of surface charge density.
I For the tangential components an amperian loop running
perpendicular to the current yields
ˆ
d
E~ 1 · ~l − E~ 2 · ~l = − B~ · d~a = 0
dt S
since in the limit as the width of the loop goes to zero the
magnetic ux vanishes.
I Therefore
E1k − E2k = 0
Boundary Conditions :
I From the fourth equation in the limit as the width of the loop
goes to zero the ux vanishes and we have

H~ 1 · ~l − H~ 2 · ~l = If encl

l θ

~λ
K~

I = Kl cos ϑ
We choose a vector ~λ on the surface with length l and

l .

~
perpendicular to
Boundary Conditions :

K~ · ~λ = Kl cos ϑ = n̂ × ~l · K~
 

where n̂is a vector normal to the surface.

If encl =K n̂ × ~l = K~ f × n̂ · ~l
   
~f ·

Therefore
H1k − H2k = K~ f × n̂
 
Vector and Scalar Potentials :
Recap

I ~ ×E
∇
Electrostatics ~ = 0 =⇒ E ~V
~ = −∇
I ~ ·B
Magnetostatics ∇ ~ = 0 =⇒ B ~ ×A
~ =∇ ~
I On to electrodynamics:
~ ·B
∇ ~ =0
is still valid!
B~ = ∇
~ ×A
~
The new thing is A~ may also be time dependent.
~ ×E
I ∇ ~ 6= 0 In electrodynamics E
~ is no longer
conservative!

~ = − ∂B = − ∂ ∇ ∂A
!
~ ~
~ ×E ~ ×A
 
∇ ~ =∇~ × −
∂t ∂t ∂t

∂A
!
~
~ ×
∇ E~ + =0
∂t
Vector and Scalar Potentials :

~ + ∂A
I E ~ ~V
−∇
∂t is irrotational and can be written as

∂A
~
E~ ~V −
= −∇
∂t

I To summarize

~ V − ∂A
~
E~= −∇
∂t
B =∇×A
~ ~ ~
Gauge Transformations :

~ V − ∂A
~
E~= −∇
∂t
B =∇×A
~ ~ ~

I Replacing A~ by A~ + ∇χ where χ (~r , t ) is an arbitrary smooth

scalar, leaves B~ unchanged.
I A~ → A~ 0 = A~ + ∇χ
B~ → B~ 0 = ∇×~ A A~ + ∇χ ~ A =B
   
I ~ 0 = ∇×
~ ~ = ∇× ~ + ∇×
~ ~
∇χ ~

I This freedom of redening A~ was there in magnetostatics as

well but there we were conned to time independent χ (~r ).
Gauge transformations :
I How must V change?

~ V 0 − ∂ A = −∇
~0
E~ →E ~ V0 − ∂ A
 
~ 0 = −∇ ~ + ∇χ
∂t ∂t
∂A
~ ∂
= −∇V 0 − − ∇χ
∂t ∂t
∂A
~
 
∂χ
= −∇ V + 0
−
∂t ∂t

I To get E~ =E
~0 we must have
 
∂χ
∇ V 0
+ = ∇V
∂t
Therefore
∂χ
V0 = V −
∂t
upto a constant.
Gauge Transformations :

I The transformations

A~ → A~ 0 = A~ + ∇χ
∂χ
V → V0 = V −
∂t

leaves E~ and B~ unchanged.

Why use potentials ?

I Working with A~ and V has the following advantages:

I Four components (one for V and three for A ~ ) as opposed to
 forE and three for B )
six (three ~ ~
I Using A ~ , V automatically satises Faraday's law and Gauss
law formagnetism.
V ,A

I ~ is a 4-vector : makes is easier to go over to relativistic
c
electrodynamics.
I V , A~ are better suited for quantization.
Maxwell's Equations in terms of Potentials:
I The two source free Maxwell's equations (Gauss law for
magnetism and Faraday's law) are automatically satised if E~
and B~ are written in terms of potentials A~ and V .
I The other two Maxwell's equations

~ = ρ
~ ·E
∇
0

~ = µ0~j + µ0 0 ∂ E
~
~ ×B
∇
∂t

I can be expressed in terms of the potential (using B~ = ∇

~ ×A
~
∂A
and E~ ~V −
= −∇
~
∂ t ) as

~ · ∂A = ρ
~
−∇2 V − ∇ (11)
∂t 0
and

~ + µ0 0 ∂ A + µ0 0 ∂ ∇
2~
~ ·A
~ − ∇2 A ~ V = µ0~j
   
~ ∇
∇ (12)
∂t 2 ∂t
Coulomb Gauge :
I By exploiting Gauge freedom we can choose

~ ·A
∇ ~ =0

I THis particular choice of gauge is called Coulomb Gauge.

I In Eq. 11 putting ~ ·A
∇ ~ =0 we have

ρ
∇2 V = −
0
.

I This looks exactly like the Poisson equation in electrostatics.

I From Eq. 12 we have

∂2 ~ ∂V
 
1 1
∇ A − 2 2A
2~
= µ0~j − ~
∇
c ∂t c 2 ∂t
Coulomb Gauge:
I From
ρ
∇2 V = −
0
I For potentials which vanish innitely far away we get

ˆ
1 ρ (~r 0 , t ) 3 0
V (~r , t ) = d ~r
4π0 |~r − ~r 0 |
I V responds instantaneously to changes in ρ ! Doesn't this
violate causality?

I This violation is only apparent.

I V A~ are not observables - E~ and B~ are!

and

I Showing that the E~ and B

~ calculated from the potentials in the
Coulomb gauge are retarded (i.e. they obey causality) is
non-trivial but possible!

I It is easier to use other gauges where this retardation is obvious

in the potentials themselves-for example Lorentz Gauge.
Lorentz Gauge:
I For the Lorentz gauge we choose

∇ ~ + µ0 0 ∂ V = ∇
~ ·A ~ ·A
~+ 1 ∂V
=0 (13)
∂t c 2 ∂t
I Hence from Eq. 12 we have

∂2A
~
∇2 A
~ − µ0 0 = −µ0~j
∂t 2
I From Eq. 11 and Eq. 13 we have

∂2V ρ
∇2 V − µ0 0 =−
∂t 2 0

I We dene

∂2 1 ∂2
∇ 2 − µ0  0 = ∇2 − =
∂t 2 c 2 ∂t 2
I Therefore
A
~ = µ0~j
ρ
V = −
0