Potential formulation, moving charges

and radiation
1. Potentials and gauge
2. Retarded potential
3. Point dipole and half wave antenna
4. Moving point charge (Leinard Wiechart factor)
5. Uniformly moving point charge
6. Accelerated point charge
7. Brehmstralung and Synchrotron radiation
8. Čerenkov radiation
9. Radiation retardation
The time dependent potential formulation
∇.⃗ B = 0 ⇒ ⃗ B = ∇× ⃗ A always possible
∇× ⃗ E =−
∂
∂t
⃗
B
⇒ ∇× ⃗ E +
∂
∂t
⃗
A
(
=0
)
∂ ⃗
A
⃗ +
E =−∇ V The potential
∂t
⃗ ρ 2 ∂ ⃗ ρ
∇ . E =ϵ ⇒ ∇ V + ( ∇ . A ) = −ϵ
0 ∂t 0

The choice div. A = 0 leads to the Possion's equation. Poisson's

equation has no time dependence into it. This implies that if the
charge density changes, the potential must change instantaneously
at all points. This cannot be correct in dynamic situations.

Q: What condition will relate dependent J and A ?

The time dependent potential formulation
∂ ⃗
E
∇ ×⃗
B ⃗
= μ0 J +ϵ 0 μ 0
∂t

∇ ×∇ × ⃗
A
2⃗
= μ0 ⃗
J + ϵ0 μ 0 ∂
∂t
−
∂
(
∂t
⃗A
− ∇V )
2
⃗ 1
∇ A − 2
∂
c ∂t
A
2
⃗ ⃗
( 1 ∂V
= −μ0 J + ∇ ∇ . A + 2
c ∂t )

{
2
1 ∂ ⃗
A
The choice ⃗2
∇ A − 2 = −μ0 ⃗
J
2
⃗ 1 ∂V c ∂t
∇.A + 2 = 0 ⇒ 2
c ∂t 2 1 ∂ V ρ
∇ V − 2 2
= − ϵ0
called Lorentz gauge c ∂t
How to solve this in the Lorentz gauge
2
2 1∂ V ρ
∇ V − 2 2 = −ϵ
c ∂t 0

Use the Fourier transform method

}
∞
1 −i ω t
ρ( ⃗r , t) =
2π ∫ ρ(
̃ ⃗r , ω)e dω 2
∇ Ṽ + ω2 Ṽ
2 ρ̃
= −ϵ
-∞
∞
⇒ c 0
1
V ( ⃗r , t ) =
2π
∫ V (⃗r , ω)e d ω
̃ −i ω t
=− g̃
-∞

Now solve for the Green's function No theta or phi

2 dependence because we
∇ G( ⃗r , r⃗' )+ ω2 G( ⃗r , r⃗' )=−δ( ⃗r −r⃗' ) expect the solution to
2

c depend on the distance

spherical polar with R=| ⃗r −r⃗' | from the source only
1 ∂
[ ( ) 1 ∂ 1
]
2
2
∇= 2
R ∂R
R
2∂
+
∂ R sin θ ∂θ ( )
sin θ ∂ + 2 ∂ 2
∂θ sin θ ∂ϕ
How to solve this in the Lorentz gauge
2
d G 2 dG ω2
2
+ + 2 G = 0 (R≠0)
dR R dR c
2 To fix the constant A, need to
d ω 2
integrate both sides over a small
2
(GR)+ 2 GR = 0 sphere centered at R=0, with the
dR c delta fn in RHS
A ±i(ω/ c)R
G = e
R
As R → 0
A
( 1ω 2
)
2
G ≈ ω
1 ± i R − R ±.... A
2 G ≈
R c 2c R

( )
2
∫ d τ ∇ G + ω2 G ∫
2
=− d τ δ(⃗r −r⃗' )
small c small
sphere sphere

−4 π A 0 −1
The retarded/advanced potential

Ṽ (⃗r , ω) =∫ d τ '
ρ
̃
(
( ⃗' , ω)
r
ϵ0 )[ 1
4 π| ⃗r −r⃗' |
e±i(ω/ c)|⃗r −r⃗' |
]
V (⃗r ,t ) =∫ d τ '
d ω −i ωt
2π
e (ρ̃ ( r⃗' , ω)
ϵ0 )[ 1
4 π| ⃗r −r⃗' |
e ±i(ω/ c)|⃗r −r⃗' |
]
1 1 d ω −i ω t
̃ r ' , ω) ) [ e ]
r − r⃗' |
∫ e ( ρ( ⃗ ±i (ω /c)| ⃗
= d τ'
4 π ϵ0 | ⃗r −r⃗' | 2 π
1 1 d ω −i ω t ±i(ω/ c)| ⃗r −r⃗' | ⃗
=
4 π ϵ0
∫ d τ'
| ⃗r −r⃗' | 2 π
e e ρ̃ ( r ' , ω)

=
1
4 π ϵ0
∫ d τ'
1
| ⃗r −r⃗' | (
ρ r⃗' ,t±
| r
⃗ −
c
r⃗' |
)
The retarded time arises naturally in Lorenz gauge solutions. It is not
put in by some other considerations!
The retarded/advanced potential
The solution for A will have similar dependence on J, component by
component.

The solution appears to say that the ''information'' about a change of

charge at r' reaches the point r with speed c. This is an attractive
physical interpretation – but works only for the potentials.

By the same logic, one might try to ''retard'' the solution for E and B
and obtain the time dependent solution – IT DOESN'T WORK!

The actual E and B must be obtained by differentiating the potentials

and they look very different. They may no longer fall off as 1/r^2

The ''retarded'' integral for the potentials is often non-trivial to do.

However everything about ''radiation'' is contained in that retarded

potential term!
Radiation
The most important consequence of certain time varying charge and
current configurations is radiation. A part of the E and B fields fall off
as 1/r – a strikingly different behaviour.

These means that the Poynting vector integrated over a spherical

surface may give a constant value as the r dependence of E x B and
the surface area would cancel each other.

This outward energy flow is radiation from a source like a radio

antenna or something else, like an accelerated charge.

The complete E and B fields created by an antenna/accelerating

charge can be quite complicated. It is only one part that has the 1/r
dependence. However this is the term which we would need to
consider for calculating radiation.

The part of the field that falls off as 1/r is called the radiation field.
An oscillating (short) dipole
l q 0 cos (ω t ) Overall electrically neutral
z '=
2
l
↑ I =−q0 ω sin( ω t) small size l≪ ω
2πc

z ' =− −q cos (ω t) far field l≪|⃗ r |

2 0

[ ( ) ( ) ]
| ⃗r −k̂ l /2 | | ⃗r +k̂ l /2|
cos ω t− cos ω t−
q0 c c
V (⃗r , t) = −
4 π ϵ0 | ⃗r − k̂ l /2 | |⃗r +k̂ l/2 |
l/2
μ0 I (z ' , t−| ⃗r −k̂ z ' | /c)
Az (⃗r ,t ) = ∫
4 π −l /2 | ⃗r −k̂ z ' |
dz '

μ0 I l
≈
4π r
sin ω t−
( )
|⃗r |
c
where I =−q 0 ω

No moving or accelerating charges in this...

An oscillating (short) dipole
Question: Why did we not model the ''oscillating
dipole'' as two charged balls on a spring ? This
must give the same answer but will involve
calculating the retarded potentials and fields due to
moving/accelerating charges. We will do that later.

First we need to approximate the distances involved

̂
( l
| ⃗r ±k l /2| = r 1± cos θ
2r ) z

θ ⃗r
1
|⃗r ±k̂ l /2| ( l
= r 1∓ cos θ
2r ) y
Use these two to approximate V (⃗r , t)
x
The scalar potential with time variation

[ ( ) ( ) ]
| ⃗r −k̂ l /2 | | ⃗r +k̂ l / 2|
cos ω t − cos ω t−
q0 c c
V (⃗r ,t ) = −
4 π ϵ0 | ⃗r −k̂ l /2 | |⃗r +k̂ l / 2|
Use binomial and small angle approximation

≈
q 0 l cos θ 1
4 π ϵ0 r r [
cos ω t−
r
c ( ) − ω
c
sin ω t −
r
c ( )]
( )1
The first term ∼ 2 will reduce to electrostatic dipole as ω → 0
r
The second term (ω≠0) falls off slowly
This gives rise to the radiation term as r → ∞
Is the gauge condition satisfied ?
⃗ 1 ∂V
Lorenz gauge : ∇. A + 2 = 0
c ∂t

V ( ⃗r , t ) =
4 π ϵ0 r r [
q 0 l cos θ 1
( )
cos ω t−
r
c ( )]
− ω
c
sin ω t −
r
c

[
q 0 l cos θ 1
( )r
( )] r
2
∂V ω
= − ω sin ω t− − cos ω t−
∂t 4 π ϵ0 r r c c c
μ0 I l
Az ( ⃗r ,t ) =
4π r
sin ω t−
r
c ( ) where I =−q0 ω

∂ Az
∂z
=
(
1 −q 0 ω l
c 2 4 π ϵ0 )[
1 z
r r ( ) r
c( )( )]
ω
− 2 sin ω t− + cos ω t −
r
r
c
−
z
r
z
Since =cos θ , the two expressions are identical
r
We could have used this to calculate V ( ⃗r , t ) from A( ⃗r , t )
 ∣ ∣
Calculating ⃗
E and ⃗
B from V and ⃗
A
ϵ̂r r ϵ̂θ r sin θ ϵ̂ϕ
⃗ ⃗ 1 ∂ ∂ ∂
B = ∇×A = 2 ∂θ ∂ϕ
r sin θ ∂r
A z cos θ −r A z sin θ 0
Br = 0
Bθ = 0

Bϕ =
μ0 I l
4π r [
sin θ ω
c ( )
cos ω t−
r
c
1
( )]
+ sin ω t−
r
r
c
1
There is one term which falls off as ∼
r

Bϕ =
μ0 I l
4π r [
sin θ ω
c ( )]
cos ω t−
r
c
Calculating ⃗
E and ⃗
B from V and ⃗
A
∂ ⃗A
⃗ = −∇ V −
E
∂t

Er =
q l cos θ 1
4 π ϵ0 r r
2 [ ( )
cos ω t− ( )]
r
c
− ω
c
sin ω t −
r
c

Eθ =
q l sin θ
4 π ϵ0 r 2
[( ) ( ) ( )]
1 ω2 r ω
− 2 r cos ω t− − sin ω t−
r c c c
r
c
Eϕ = 0
1
Only component that falls off as ∼ :
r
Eθ = −
ql ω2 sin θ
2
4 π ϵ0 c r ( )
cos ω t−
r
c
Power radiated by the dipole
⃗ 1 ⃗ ⃗
Consider S =μ E × B over a sphere with R → ∞
0
We only need to consider Both expressions
are equivalent.
ql ω2 sin θ r
( )

I 0=−q ω
Eθ = − 2
cos ω t−
4 π ϵ0 c r c They show two
μ 0 ql ω 2 sin θ different ways of
Bϕ = −
4π c r
cos ω t −
r
c ( ) viewing the source
of radiation.
2
R
∯ ⃗S . d ⃗a = μ
0
∫ E θ B ϕ 2 π sin θ d θ Either as a dipole
R or as a ''current
q2 l 2
=
6 π ϵ0 c 3
4 2
ω cos ω t−
R
c ( ) element'' of an
antenna.
2 4 2 2
I0
〈 P radiated 〉 =
1 (ql ) ω
4 π ϵ0 3c3
= √ ()
2 π μ0 l
3 ϵ0 λ 2
Radiation pattern and antenna impedance
No intensity along the axis.

Maximum intensity on the

equitorial plane.

In the polar plot the radial

distance is the magnitude of
the quantity at a certain
angle.

CC BY-SA 3.0, https://commons.wikimedia.org/w/index.php?curid=1085864

The far field pattern tells us how much the dipole is radiating.

The near field pattern will be necessary if we want to calculate the

effect of one dipole on another nearby dipole. (i.e. How would two
antennas interfere, etc. ?)
Radiation pattern and antenna impedance
2 4 2
2π μ0 l I 0
1 (ql ) ω
〈 P radiated 〉 = 4 π ϵ
0 3c 3
=
3 ϵ 0 λ 2√ ()
Radiated power = Real part of Impedance × r.m.s. current
μ0
√
The quantity ϵ ≈377 Ω sets the impedance scale
0
This is called the Radiation resistance of an antenna
This does NOT tell us the reactive part of the impedenance.
Also the result is correct only for l ≪λ
2
2D
←−−−−−−−−−−−− −−−−−−−−−−−−→
Antenna dimension λ
D Far field
Reactive near field Radiative near field
Wavefront shape (Fraunhoffer)
√
3
D
≈0.62 keeps changing
λ
The half wave dipole
The short dipole result will not hold unless l ≪λ .
l=λ/2 is a common configuration called a half wave antenna.
l 1
But setting = in the earlier formula won't work!
λ 2
Also : If l=λ IT WILL NOT RADIATE AT ALL! Why?
For l>λ /2 some parts will start having oppposite currents....

Variants of the dipole antenna.

The half wave dipole

The ''short'' dipole that we analyzed can be used to build up a

solution, if we know the current at each point of the dipole.

However the current at each point must be consistent with the ''near
field'' produced by the other parts.

This makes the ''exact'' solution a difficult self-consistent problem.

We generally assume a reasonable current pattern that goes to zero

at the ends and is maximum at the feed-point.

It so happens that the ''numerically exact'' solution agree quite closely

with the result from the profile shown.
The half wave dipole
small dipole result derived earlier
⃗
R
⃗r Eθ = −
ql ω2 sin θ
2
4 π ϵ0 c r ( )
cos ω t−
r
c

(−λ <z ' < λ

)
z'→

R = r−z ' cos θ

θ 4 4
l → dz ' I 0 =−q ω

(
I ( z ' , t) = I 0 sin ω t cos
2π z'
λ )
Far field due to a segment between z ' to z ' +dz '

dE θ =
( I0
)
sin θ ω
4 π ϵ0 c R c
R
( ) ( )
cos ω t − cos
c
2π z'
λ
dz '

dBϕ =
( ) μ 0 I 0 sin θ ω
4π R c ( ) ( )
R
cos ω t− cos
c
2π z'
λ
dz '
The half wave dipole
2π z '
u = R=r−z ' cos θ (change variables)
λ
π /2
K =
1
( ) R
∫ R cos ω t −c cos u du
−π/ 2

[( ) ]
π/2
1 r
≈ ∫
r −π/ 2
cos ω t− +u cos θ cos u du
c
( r ≫z ' )
π/ 2
=
1
r ( )∫
cos ω t−
r
c −π /2
cos (u cos θ)cos u du +
π/ 2
1
r ( )∫
sin ω t−
r
c −π/ 2
sin(u cos θ) cos u du
=0

K =
2
r ( )
cos ω t−
r cos(π/ 2 cos θ)
c sin 2 θ
The half wave dipole

Eθ = ( )
I0
4 π ϵc
K = ( ) ( )
I0
2 π ϵc r
cos ω t −
r cos (π/ 2 cos θ)
c sin θ

Bϕ =
( )
μ0 I 0
4π
K =
( ) ( )
μ0 I 0
2πr
cos ω t−
r cos(π / 2 cos θ)
c sin θ

Integrating the Poynting vector over a large sphere

2

( )
π
1 μ 0 2 cos (π/ 2cos θ)
√
〈 P radiated 〉 = 4 π ϵ0 I 0 ∫
0 sin θ
sin θ d θ

( )
2
I0
= 73 (ohms)×
2
This approximately 75 Ohms impedance is often encountered
in dealing with cables connecting antennas to amplifiers etc.
What is the reason ? [Discuss later]
Magnetic dipole radiation
We saw that an oscillating electric dipole radiates. A natural
question is what does an oscillating magnetic dipole do ?
z Such a current will
I =I 0 cos ω t produce no electric
potential. Why ?
y
The loop area → 0
m = I x area = constant
x Q: Does it satisfy the Lorenz gauge?
Does it give the correct static limit ?
The solution No Radiation Radiation
V (⃗r , t ) = 0
⃗A(⃗r , t ) =
4π ( )[
μ 0 m sin θ
r
1
r ( )
r ω
cos ω t− − sin ω t−
c c
r
c ( )] ϵ̂ϕ
Magnetic dipole radiation
The far field
2
⃗ μ 0 mω
⃗
E =
−∂
∂t
A
=
4πc ( ) ( )
2
sin θ
r
r
cos ω t − ϵ̂ϕ
c
−μ 0 mω
⃗ =
B ∇ × ⃗A =
4πc
2 ( ) ( )
sin θ
r
r
cos ω t− ϵ̂θ
c
2 4
1 m ω
The radiated power 〈 P radiated 〉 = 4 π ϵ 5
0 3c

This power is small in comparison to an electric dipole of similar size

with I 0 → q ω and πa →d
where a is the radius of the current loop and d the dipole length