Chapter 11 Radiation

Dipole Radiation
What is Radiation?
When charges accelerate, their fields can transport
energy irreversibly out to infinity—a process we call
radiation.

Assume the source is localized near the origen; we

would like to calculate the energy it is radiating at time t0.

Imagine a sphere, out at radius r. The power passing through its surface is the
1
integral of the Poynting vector: P (r , t )=∮ S⋅d a = μ ∮ E × B⋅d a
0
Because EM “news” travels at the speed of light, this energy actually left the
r r
source at the earlier time t 0 =t −
c
, so the power
P rad (t 0 )= lim P ( r , t 0 + )
radiated is r ∞ c
This is energy (per unit time) that is carried away and never comes back.

The area of the sphere is 4πr 2, so the Poynting vector must decrease (at large

1 1 1
r) no faster than . If it went 3
, then P would go , and Prad would be 0.
r
2
r r
 1
According to Coulomb’s law, electrostatic fields fall off like 2
(or faster, if the
r
1
charge is 0), and the Biot-Savart law says that magnetostatic fields go like 2 (or
r
1
faster), so S ∼ 4 , for static configurations. So static sources do not radiate.
r

Jefimenko’s equations indicate that time-dependent fields include terms

1
(involving ρ̇ and J̇ ) that go like ; these are the terms that are responsible for
r
EM radiation.

The study of radiation involves picking out the parts ofE & B that go like 1 at
r
1
large distances from the source, constructing from them the term in S,
2
r
integrating over a large spherical surface, and taking the limit as r.
Electric Dipole Radiation (Hertzian dipole)
2 tiny metal spheres are separated by a distance d and
connected by a fine wire; the charge on the upper sphere
is q(t), and the charge on the lower sphere is −q(t). Let that
we drive the charge back and forth through the wire, at an
angular frequency ω:
q (t )= q 0 cos ω t ⇐ p (t )= p0 cos ω t z^ ⇐ p 0 ≡ q 0 d
electric dipole
1 q 0 cos ω t r  q 0 cos ω t r – г±
⇒ V (r , t )= ( − ) ⇐ t r ± =t −
4 π ϵ0 г+ г– c
2
d
√ 2
where г ± = r ∓r d cos θ +
4
To make this physical dipole into a perfect dipole, we want the separation
distance to be extremely small:
d
approximation 1: d≪r: to 1 order in d: ⇒ г ± ≃ r ( 1∓ cos θ )
st
2r
г± r d cos θ
⇒ cos ω t r ± = cos ω ( t − ) ≃ cos ω ( t − ± )
c c 2c
ω (c t −r ) ω d cos θ ω (c t − r ) ω d cos θ
= cos cos ∓sin sin
c 2c c 2c
 In the perfect dipole limit we have, further,
c 2πc 1 1 d
Approximation 2: d ≪ ω ⇒ d ≪ λ ⇐ λ= ω , ≃ ( 1± cos θ )
г± r 2r
г± ω (c t − r ) ω d ω (c t − r )
⇒ cos ω ( t − ) ≃ cos ∓ cos θ sin
c c 2c c
p0 cos θ 1 ω (c t −r ) ω ω (c t −r )
⇒ V (r , θ , t )≃ ( cos − sin ) ⇐ p0 = q 0 d
4 π ϵ0 r r c c c
p 0 cos θ
ω0 ⇒ V≃ 2
potential of a stationary dipole
4 π ϵ0 r
We are interested in the fields that survive at large
distances from the source, in the so-called radiation zone:
c 1 ω
approximation 3: r ≫ ⇒ r ≫ λ ⇒ ≪
ω r c
p0 ω cos θ ω (c t − r )
⇒ V (r , θ , t )≃− sin
4 π ϵ0 c r c
The vector potential is determined by the current flowing
dq
in the wire: I (t )= z^ =− q 0 ω sin ω t z^
dt
 1 1 d
≃ ( 1± cos θ )

[ г± r

⇒ V (r , θ , t )=
2r

cos ω t r ± ≃ cos

1
ω (c t − r ) ω d

(
c
∓
2c
cos θ sin

−
ω (c t −r )
c
q 0 cos ω t r  q 0 cos ω t r –
)
⇐ cos ϑ ≃ 1 for ϑ ≪ 1
sin ϑ ≃ ϑ

4 π ϵ0 г+ г–
q0 d ω (c t −r ) ω d ω (c t −r )
( 1+ cos θ ) ( cos )
≃
4 π ϵ0 r [ 2r c
−
2c
cos θ sin
c
d ω (c t −r ) ω d ω (c t −r )
− ( 1− cos θ ) ( cos )
2r c
+
2c
cos θ sin
c ]
q0 d ω (c t − r ) ω d ω (c t − r )
= ( cos θ cos − cos θ sin )
4 π ϵ0 r r c c c
p0 cos θ 1 ω (c t −r ) ω ω (c t −r )
= ( cos − sin ) ⇐ p0 = q 0 d
4 π ϵ0 r r c c c
 + d /2
μ0 − q0 ω
ω (c t −г )
⇒ A (r , t )= ∫ z^ d z
sin
4 π −d /2 г c
μ0 p0 ω ω (c t −r )
⇒ A (r , θ , t )≈ − sin z^ ⇐ z^ = r^ cos θ − θ^ sin θ
4πr c
∂V 1 ∂V ^ p 0 ω cos θ ω (c t − r ) ω r ω (c t − r)
( ) r^
∇ V=
∂r
r^ +
r ∂θ
θ=
4 π ϵ0 c r 2
[
sin
c
+
c
cos
c
sin θ ω (c t −r ) ^ p 0 ω2 cos θ ω (c t − r)
+
r
2
sin
c ]
θ ≃
4 π ϵ0 c
2
r
cos
c
r^

∂A μ0 p0 ω2 cos θ sin θ ^ ω (c t − r )
=− ( r−
^ θ ) cos
∂t 4π r r c
∂A μ 0 p0 ω 2 sin θ ω (c t −r ) ^
⇒ E =− ∇ V − ≃− cos θ
∂t 4π r c
1 ∂ Ar
∇ ×A = ( )ϕ
∂
(r Aθ )− ^
r ∂r ∂θ
μ0 p0 ω ω ω (c t −r ) sin θ ω (c t −r ) ^
=− ( sin θ cos + sin )ϕ
4πr c c r c
μ0 p0 ω 2 sin θ ω (c t − r ) ^
⇒ B = ∇ × A ≃− cos ϕ
4πc r c
Electric dipole magnetic dipole
 Here E & B represent monochromatic waves of frequency ω traveling in the
radial direction at the speed of light. The fields are in phase, mutually
E0
perpendicular, transverse; the ratio of their amplitudes is = c, as expected.
B0
These are spherical waves, not plane waves, and their amplitude decreases like
1 as they progress. For large r, they are approximately plane over small region.
r
The energy radiated by an oscillating electric dipole is determined by the
Poynting vector: 2
2 2 4
E × B μ 0 p 0 ω sin θ μ 0 p 0 ω sin 2 θ
S (r , t )=
μ0
=
c 4π r [cos
ω (c t −r )
c
r^ ⇒ ⟨S ⟩ =
] 2
32 π c r 2
r^

There is no radiation along the axis of the dipole (here sin θ=0); the intensity
profile takes the form of a donut, with its maximum in the equatorial plane.

The total power radiated is found by integrating <S>

over a sphere of radius r:

⟨ P ⟩ =∮ ⟨ S ⟩⋅d a
2 4 2
μ0 p0 ω sin 2 θ μ0 p0
∫
2 4
= 2 2
r sin θ d θ d ϕ = ω
32 π c r 12 π c
Example 11.1: The strong frequency dependence of the power formula is what
accounts for the blueness of the sky. Sunlight passing through the atmosphere
stimulates atoms to oscillate as tiny dipoles.

The incident solar radiation covers a broad range of frequencies (white light),
but the energy absorbed and re-radiated by the atmospheric dipoles is stronger
at the higher frequencies because of the ω 4.

It is more intense in the blue than in the red. It is this re-radiated light that you
see when you look up in the sky (not directly at the sun).
 In the celestial arc  the sun’s rays, where the blueness is most pronounced,
the dipoles oscillating along the line of sight send no radiation to the observer;
light received at this angle is polarized  the sun’s rays.

The redness of sunset is the other side of the same coin: Sunlight coming in at
a tangent to the earth’s surface must pass through a much longer stretch of
atmosphere than sunlight coming from overhead. Accordingly, much of the blue
has been removed by scattering, and what’s left is red.
Magnetic Dipole Radiation
A wire loop of radius b with an alternating
current: I (t ) = I cos ω t
0

This is a model for an oscillating magnetic dipole

m (t )= π b2 I (t ) z^ = m 0 cos ω t z^ ⇐ m 0 = π b 2 I 0
The loop is uncharged, so the scalar potential=0
μ0 I0 ω (c t −г )
⇒ A (r , t )= ∫ cos d ℓ
4π г c
For a point r directly above the x axis, A must aim in the y direction, since the x
components from symmetrically placed points on either side of the x axis cancel.
μ0 I 0 b 1 ω (c t −г )
⇒ A (r , t )= y^ ∫ cos cos ϕ  d ϕ  ⇐ cos ϕ  for d ℓ y
4π г c
r =r sin θ x^ + r cos θ z^ , b = b cos ϕ x^ + b sin ϕ  y^
2 2 2 2
⇒ г =|r − b|= √ r + b − 2 r b cos ψ = √ r + b − 2 r b sin θ cos ϕ


For a “perfect” dipole, we want the loop to be extremely small:

1 1 b
approximation 1: b ≪ r ⇒ г ≃r − b sin θ cos ϕ ⇒ ≃ ( 1+ sin θ cos ϕ )
 

г r r
 ω (c t −г ) r b
≃ cos ω ( t − + sin θ cos ϕ )

⇒ cos
c c c
ω (c t −r ) ω b sin θ cos ϕ  ω (c t −r ) ω b sin θ cos ϕ 
=cos cos −sin sin
c c c c
Assume the size of the dipole is small compared to the wavelength radiated:
c
approximation 2: b≪ ⇐ cos ϑ ≃ 1 , sin ϑ ≃ ϑ for ϑ ≪ 1
ω
ω (c t −г ) ω (c t −r ) ω b  ω (c t −r )
⇒ cos ≃ cos − sin θ cos ϕ sin
c c c c
2π
μ0 I 0 b ω (c t −r )
⇒ A (r , t )=
4πr
y∫
^
0
[
cos
c

+ b sin θ cos ϕ ×

1 ω (c t −r ) ω ω (c t −r )
( cos ) cos ϕ d ϕ
2π 2π
r c
− sin
c c
2π
]  

1 1
∫ cos ϕ d ϕ =0 , ∫ cos ϕ d ϕ = π vs ⟨cos ϕ ⟩ =
  2   2
∫ cos2 ϕ d ϕ  =
0 0 2π 0 2
μ 0 m 0 sin θ 1 ω (c t − r ) ω ω (c t −r ) ^
⇒ A (r , θ , t )= ( cos − sin )ϕ
4π r r c c c
μ 0 m 0 sin θ
static limit ω  0 ⇒ A (r , θ)= ϕ^ the potential of
2
4π r a magnetic dipole
 In the radiation zone,
ω μ 0 m 0 ω sin θ ω (c t − r ) ^
approximation 3: r ≫ ⇒ A (r , θ , t )=− sin ϕ
c 4πc r c
∂A μ 0 m 0 ω 2 sin θ ω (c t −r ) ^
E =− = cos ϕ
⇒
∂t 4πc r c
μ0 m 0 ω 2 sin θ ω (c t −r ) ^
B = ∇ × A =− cos θ
4πc
2
r c
These fields are in phase, mutually perpendicular, and transverse to the
E0
propagation direction ( r^ ), the ratio of their amplitudes is = c, as expected.
B0
They are similar in structure to the fields of an oscillating electric dipole, only
this time it is B/E that points in the θ
^ /ϕ
^ direction, whereas for electric dipoles
it’s the other way around.

The energy flux for magnetic dipole radiation is

2 2
E × B μ0 m 0 ω sin θ ω (c t − r )
S (r , t )= μ = ( cos ) r^
0 c 4πc r c
μ 0 m 20 ω 4 sin 2 θ μ 0 m 20 4
⇒ intensity ⟨ S⟩ = 2 3 2
r^ ⇒ total radiated power ⟨ P ⟩ = 3
ω
32 π c r 12 π c
 The intensity profile has the shape of a donut, the power radiated goes like ω 4.
There is one important difference between electric and magnetic dipole
radiation: For configurations with comparable dimensions, the power radiated
electrically is enormously greater.

P magnetic m 20 ω2 b 2 2
= 2 2= 2 ⇐ m0 = π b I 0 , p 0 = q 0 d , I 0 = q 0 ω , d = π b
P electric p0 c c
ωb
is the quantity we assumed was very small, not to mention being squared.
c
Ordinarily one should expect electric dipole radiation to dominate. Only when
the system is carefully contrived to exclude any electric contribution will the
magnetic dipole radiation reveal itself.

Example: Find the radiation resistance for the magnetic dipole as described.
2 2 μ 0 m 20 4 1 2 I (t )= I 0 cos ω t
P = I R ⇒ ⟨ P ⟩ =⟨ I ⟩ R md ⇒ 3
ω = I 0 R md ⇐
12 π c 2 m0 = π b 2 I 0
5 4 4
μ0 π 8π b b 2πc
( ) ( )
4 4 6
⇒ R md = b ω = μ 0 c ≃320 π ⇐ ω =
6c 3
3 λ λ λ
−6
For d =5 cm , λ =1 km ⇒ R md ≃2× 10 Ω∼10 R ed
−12
Linear Electric Quadrupole Radiation
μ 0 q 0 a2 ω 3 sin 2 θ ω (c t − r) ^
Er = sin θ
8πc r c
μ 0 q 0 a 2 ω 3 sin 2 θ ω (c t − r ) ^
Br = sin ϕ
8πc
2
r c
μ 0 q 20 a 4 ω 6 sin 2 2 θ
⇒ intensity ⟨ S⟩ = 2 3 2
r^
128 π c r
2 4
μ0 q0 a
⇒ total radiated power ⟨ P ⟩ = 3
ω6
60 π c
Radiation from an Arbitrary Source
A configuration of charge and current that is
entirely arbitrary, except that it is localized within
some finite volume near the origen.

The retarded scalar potential is

1 1 г
∫ ρ(r ,t− )dτ
 
V (r , t)=
4 π ϵ0 г c
2 2
where г = √ r + r − 2 r⋅r


Assume that r is far away, in comparison

to the dimensions of the source:
r r⋅r
 
1 1
approximation 1: r ≪ r ⇒ г ≃r ( 1− ) ⇒ ≃ ( 1+ )
 ^ ⋅r ^
r г r r
г r r⋅r
 
r
⇒ ρ ( r , t − )≃ ρ ( r , t 0 + )= ρ (r , t 0 )+ ρ̇ (r , t 0 )
^ ⋅r ^
   
+⋯ ⇐ t 0 =t −
c c c c
We can afford to drop them, provided
c c
approximation 2: r ≪ , ,⋯
|ρ̈ / ρ̇| |ρ⃛ / ρ̇|1 /2
c
For an oscillating system, each of these ratios is , and we recover the old
approximation 2.
ω
 As a procedural matter approximations 1 and 2 amount to keeping only the 1st-
order terms in r.
Discarding the higher-order terms
1 r^
V (r , t)≃ ( ∫ ρ (r , t 0 ) d τ + ⋅∫ r  ρ (r  , t 0 ) d τ 
 

4 π ϵ0 r r
r^ d 1 Q r⋅p ^ (t 0 ) r^ ⋅ṗ (t 0 )
+ ⋅ ∫ r ρ (r , t 0 ) d τ ) =
  
( + + )
c dt 4 π ϵ0 r r 2
rc
Because charge is conserved, Q is independent of time. The other 2 integrals
represent the electric dipole moment at time t0.

In the static case, the 1st 2 terms are the monopole and dipole contributions to
the multipole expansion for V ; the 3rd term would not be present.
k k k k i
J = ∑ J x^ k = ∑ J ∂k r = ∑ [∂ k ( J r )− r ∂ k J ]= ∑ ∇⋅( x J ) x^ i −( ∇⋅J ) r
k k k i

⇒ ∫ J (r , t 0 ) d τ = ∑ x^ i ∫ ∇ ⋅( x J ) d τ −∫ r ( ∇ ⋅J ) d τ
   i    

i
∂ρ d ∂ρ
= ∑ x^ i ∮ x J⋅d a + ∫ r
i 
dτ =
 
∫ ρ r d τ = ṗ ⇐
 
+ ∇ J =0
i ∂ t0 dt ∂t
[Problem 5.7]
 μ0 1 г μ0 μ 0 ṗ (t 0 )
⇒ A (r , t )= ∫ J ( r , t− ) d τ ≃
 
∫ J (r , t 0 ) d τ =
 

4π г c 4πr 4π r
It was unnecessary to carry the approximation of г beyond the 0th-order (г≅r): p
is already 1st order in r , any refinements are corrections of 2nd order (or higher).


1
In the radiation we keep only those terms that go like :
r
1
approximation 3: discard 2 terms in E and B
r
1 Q
For instance, the Coulomb field, E = 2
r
^ , from the 1 st
term in V(r, t),
does not contribute to the EM radiation.4 π ϵ 0 r

The radiation comes from the terms in which we differentiate the argument t0,
∇r r^
∇ t 0 =− =−
c c
1 r^ ⋅ṗ (t 0 ) 1 r⋅^ p̈ (t 0 ) 1 r⋅
^ p̈ (t 0 )
∇ V ≃∇ ( )≃ ∇ t 0 =− r^
4 π ϵ0 r c 4 π ϵ0 r c 4 π ϵ0 c 2
r
⇒ ∂ A ≃ μ0 p̈ (t 0 )
∂t 4π r
μ0 μ0 μ0
∇×A≃ ∇ × ṗ (t 0 )= ∇ t 0 × p̈ (t 0 )=− r^ × p̈ (t 0 )
4πr 4πr 4πrc
 μ0 μ0
E (r , t )≃ [( r^ ⋅p̈) r^ − p̈]= r^ ×( r^ × p̈ )
4πr 4πr r
⇒ ⇐ p̈ = p̈ (t 0 )= p̈ ( t − )
μ0 r^ c
B (r , t )≃− r^ × p̈ = × E (r , t )
4πrc c
Use spherical polar coordinates, with the z axis in the direction of p̈ (t 0 )= p̈ z^
μ 0 p̈ (t 0 ) sin θ
E (r , θ , t)≃ θ^
4π r
μ 0 p̈ (t 0 ) sin θ
B (r , θ , t)≃ ϕ^
4πc r
2
E × B μ 0 p̈ (t 0 ) sin 2 θ
⇒ Poynting vector S (r , t )≃ = r^
μ0 16 π c 2
r 2

2
μ 0 p̈ (t 0 )
⇒ P (r , t )=∮ S (r , t )⋅d a = = P rad (t 0 ) total radiated power
6πc
E and B are mutually perpendicular, transverse to the direction of propagation
E
( r^ ), and in the ratio = c , as always.
B
Example 11.2: (a) In the case of an oscillating electric dipole,
μ0 p0 ω2 sin θ ω (c t − r ) ^
p (t )= p0 cos ω t z^ E ≃− cos θ
⇒ 4π r c ⇐ same as
2
2 μ p ω sin θ ω (c t −r ) the earlier
p̈ (t )=− ω p 0 cos ω t z^ B ≃−
0 0
cos ϕ^
4πc r c
(b) For a single point charge q, the dipole moment is
μ 0 q2 a 2 2
p (t )= q d (t ) ⇒ p̈ (t )= q a (t) ⇒ Larmor formula P = ∝ a
6πc

In this section for a multipole expansion of the retarded potentials, use the
1
lowest order in r that is capable of producing EM radiation (fields being like

).
This turns out to be the electric dipole term. r

Because charge is conserved, an electric monopole does not radiate.

If charge were not conserved, 1 Q (t 0 ) 1 Q̇ (t 0 )

V mono = ⇒ E mono = r^
the 1st term in V would read 4 π ϵ0 r 4 π ϵ0 c r
You might think that a charged sphere whose radius oscillates in and out would
radiate, but it doesn’t—the field outside, according to Gauss’s law, is exactly
Q
2
r^ , regardless of the fluctuations in size.
4 π ϵ0 r
 In the acoustical analog, monopoles do radiate: transverse vs longitudinal.

If the electric dipole moment or its 2nd time derivative should happen to vanish,
then there is no electric dipole radiation, and one must look to the next term: the
one of 2nd order in r .


This term can be separated into 2 parts, one of which is related to the magnetic
dipole moment of the source, the other to its electric quadrupole moment.

If the magnetic dipole and electric quadrupole contributions vanish, the r 3

term must be considered, etc.

Selected problems: 4, 10, 18, 22, 26

Point Charges
Power Radiated by a Point Charge
The fields of a point charge q in arbitrary motion

q c (1− β 2 ) ( �^ − β) + ⃗� ×[( �^ − β)× β̇]

E (r , t)= 2 3
($)
4 π ϵ0 � c (1− �^⋅β)
�^
B (r , t )= × E (r , t )
c
The 1st term in ($) is the velocity field, and the 2nd one is the acceleration field.
2
E × B E ×( �^ × E ) E �^ −( �^⋅E ) E
Poynting vector S (r , t)≃ = =
μ0 μ0 c μ0 c
Not all of this energy flux constitutes radiation; some of it is just field energy
carried along by the particle as it moves.

The radiated energy is the stuff that, in effect, detaches itself from the charge
and propagates off to infinity.

To calculate the total power radiated by the particle at time tr, consider a huge
sphere of radius , centered at the position of the particle (at tr), wait the
appropriate interval c(t−tr)=� for the radiation to reach the sphere, and at that
moment integrate the Poynting vector over the surface.
 Tthe area of the sphere ∝ � 2, so any term in S that goes
⃗�
1 1
like 2 will yield a finite answer, but terms like 3 or
1 � �
4 will contribute nothing in the limit r ∞.
�
So only the acceleration fields represent true radiation:
q �^ ×[( �^ − β)× β̇]
Radiation fields E rad =
4 π ϵ 0 � c (1− �^⋅β)3
The velocity fields carry energy, as the charge moves this energy is dragged
along—but it’s not radiation.
E 2rad �^ −( �^⋅E rad ) E rad E 2rad ⃗�
E rad ⊥ �^ ⇒ S rad ≃ = �^ Θ
μ0 c μ0 c
Let v (t r )= 0 at rest ⇒ β (t r )=0
q �^ ×( �^ × β̇) μ 0 q c
⇒ E rad = = [( �^⋅β̇) �^ − β̇]
4 π ϵ0 c � 4π�
2
1 μ0 q c μ 0 c q 2 β̇ 2
( ) [ β̇ −( �^⋅β̇) ] �^ = ^
2 2 2
⇒ S rad ≃ 2 2
sin Θ �
^ ( #) ⇐ cos Θ = �
^ ⋅β̇
μ0 c 4π� 16 π �
No power is radiated in the forward or backward direction—rather, it is emitted
in a donut about the direction of instantaneous acceleration.
 a a
 β̇ = , β̇ =
The total power radiated is c c
a Larmor
2 2 2 2 2
μ 0 c q β̇ sin θ 2 μ q
P = ∮ S rad⋅d a = ∫ � sin θ d θ d ϕ ⇒ P = 0
formula
16 π 2
� 2
6 π c
(again)
Although we derived them on the assumption that v=0, the result actually holds
to good approximation as long as v≪c.

Suppose someone is firing a stream of bullets out the window of a moving car.
The rate Nt at which the bullets strike a stationary target is not the same as the
rate Ng at which they left the gun, because of the motion of the car.

We can check that Ng=(1−β)Nt, if the car is moving towards the

v
target, and N g =(1 − �^⋅β) N t for arbitrary directions, where β ≡ , β =|β|.
c
If d W is the rate at which energy passes through the sphere at radius , the
dt d W d W /d t dW ∂ tr 1
rate of energy left the charge = =(1 − �^⋅β) ⇐ =
d tr ∂ tr / ∂ t dt ∂ t 1− �^⋅β
Source Moving, Detector Stationary

Let source S move toward D at speed vS

v v v v
λ = v T − vS T ⇒ f =  =
 
= =f ⩾f
λ v T − v S T v / f − vS / f v − vS
c c fg 1
λt = c T g − v T g ⇒ f t = = = ⇐ f=
λ t c T g − v T g 1− β T
⇒ N g =(1 − β) N t
 So the ratio of the energy rate is precisely the ratio of Ng to Nt; it’s a purely
geometrical factor.

The power radiated by the particle into a patch of area � 2 sin θ d θ d ϕ= � 2 d Ω,

Ω=sin θ d θ d ϕ is the solid angle, on the sphere is given by
where d
2 2 2
dP E rad � q 2 �
|^ ×[( �
^ − β)× β̇]| a 1
=(1− �^⋅β) = 2 5
⇐ β̇ = ⇒ γ ≡
dΩ μ0 c 16 π ϵ 0 c (1− �^⋅β) c √ 1− β 2
2 6 2 6
μ0 c q γ μ0 q γ
(a −|β × a| ) Liénard’s generalization
2 2 2 2
⇒ P= ( β̇ −|β × β̇| )=
6π 6πc of the Larmor formula
The factor γ 6 means that the radiated power increases enormously as the
particle velocity approaches the speed of light.

Example 11.3: Let v & a are instantaneously collinear (at tr) as in straight-line
motion. Find the angular distribution of the radiation & the total power emitted.
dP q 2 |^� ×( �^ × β̇)|2
v∥a∥ z^ ⇒ β∥ β̇∥ z^ ⇒ ( �^ − β)× β̇ = �^ × β̇ ⇒ =
d Ω 16 π 2 ϵ 0 c (1− �^⋅β)5
2 2 2
�^ ×( �^ × β̇)=( �^⋅β̇) �^ − β̇ ⇒ |�^ ×( �^ × β̇)| = β̇ −( �^⋅β̇)
dP μ0 q2β̇ sin θ
2 2
v
⇒ = 2 5
⇐ β ≡ , β = β z^ , β̇ = β̇ z^
d Ω 16 π ϵ 0 c (1− β cos θ ) c
 dP q 2 |^� ×[( �^ − β)× β̇]|2
P =∫ d Ω= 2
∫ 5
sin θ d θ d ϕ
dΩ 16 π ϵ0 c (1− �^⋅β)
�^ ⋅β^ = cos θ , β⋅^ β̇^ =cos ψ
Choose β^ = z^ , β̇^ =sin ψ x^ + cos ψ z^ , ψ fixed ⇒
�^⋅β̇^ = cos Θ = cos ψ cos θ
�^ = sin θ (cos ϕ x^ + sin ϕ y^ )+ cos θ z^
+ sin ψ sin θ cos ϕ
^ 2 = cos2 Θ = cos2 ψ cos 2 θ + sin 2 ψ cos θ sin θ cos ϕ + sin 2 ψ sin 2 θ cos 2 ϕ
⇒ ( �^⋅β̇)
⇒ |^� ×[( �^ − β)× β̇]|2 = β̇ 2 [(β 2 − 1) cos2 Θ + 2 β cos ψ cos Θ (1− �^⋅β)+ (1− �^⋅β)2 ]
2
cos Θ 4π 4 2
β̇ (β −1) ∫

[
2 2 6 2
5
sin θ d θ d ϕ =− γ β̇ − 8 π γ (β⋅β̇)
(1− β cos θ) 3
cos Θ 32 π 6
⇒ 2 β β̇ cos ψ ∫
2
sin θ d θ d ϕ = γ (β⋅ β̇)2

(1 − β cos θ)4 3
1
β̇ ∫
2
3
sin θ d θ d ϕ = 4 π γ 4 2
β̇
(1− β cos θ )
q24π 4 2 32 π 6
⇒ P= 2
16 π ϵ 0 c
−
3
γ
[β̇ −8 π γ
6
(β⋅β̇)
2
+
3
γ (β⋅β̇)
2
+ 4 π γ
4
β̇
2
]
q2 8π 4 2 8π 6 μ0 c q2 γ 6
= 2 [
16 π ϵ 0 c 3
γ β̇ +
3
γ (β⋅β̇)
2
=
] 6π
2 2
( β̇ −|β × β̇| )
 dP 2
=� ×( #) for v = 0 (β = 0)
dΩ
⇒
dP 1
∝ 5
for v  c (β  1)
d Ω (1− β cos θ )
Although there is still no radiation in precisely the
forward direction, most of it is concentrated within an
increasingly narrow cone about the forward direction.
dP μ0 q2 a2 2
sin θ
P=∫ d Ω= 2
∫ 5
sin θ d θ d ϕ
dΩ 16 π c (1− β cos θ )
2 2 1 2 2 2 2 2 6
μ0 q a (1− x ) μ q a 4 μ q a γ
= ∫ d x= 0
= 0

8 π c −1 (1 − β x ) 5
8 π c 3 (1− β )
2 3
6πc

consistent with the Liénard formula, for the case of collinear v and a.

The angular distribution of the radiation is the same whether the particle is
accelerating or decelerating; it only depends on a2, and is concentrated in the
forward direction (with respect to the velocity) in either case.

When a high speed electron hits a metal target it rapidly decelerates, giving off
what is called bremsstrahlung, or “braking radiation.”
Problem 11.15: Find θmax at which the maximum radiation is emitted in Ex. 11.3.
d dP d sin θ 2 sin θ (2 cos θ − 2 β cos 2 θ −5 β sin 2 θ )
=0 ⇒ 0= =
dθ dΩ d θ (1− β cos θ)5
(1 − β cos θ)6
2 ± √ 15 β 2 + 1− 1 choose  sign
⇒ 3 β cos θ + 2 cos θ − 5 β =0 ⇒ cos θ =
3β to fit β  0
2
−1 √ 15 β + 1−1 dP
⇒ θ max = cos and θ min =0, π for sin θ min =0 ⇒ =0
3β d Ω |min
For ultra-rela- ⇒ β  1 ⇒ β =1 −δ , δ ≪ 1 ⇒ γ = 1 1
≃
tivistic speeds √1 − β2 √ 2 δ
⇒
√ 15 β 2 + 1−1 = √ 15 (1 − δ )2 + 1− 1 ≃ √ 16 − 30 δ −1 ≃ 1+ δ ( 3− 15 δ )
3β 3 (1− δ ) 3 (1− δ ) 3 4
2
θ 1−β 1
≃1−

dP
δ
4
⇒ cos θ max ≃ 1−

μ 0 q2
max

2
≃ 1−

β̇ 2 sin θ 2max
δ
4
⇒ θ max ≃

μ 0 q2 β̇ 2
√ √
δ
2
≃
2
θ 2max
≃
2γ

⇒ = 2 5
≃ 2 5
d Ω |max 16 π ϵ 0 c (1− β cos θ max ) 16 π ϵ 0 c [1−(1− δ ) (1− δ / 4)]
2 2 2 2 2 2
μ 0 q β̇ δ /2 μ 0 q β̇δ /2 μ0 q β̇4
5
( )
8
≃ 2 5
≃ 2 5
≃ 2
γ
16 π ϵ 0 c [1 −(1 −δ ) (1 − δ / 4)] 16 π ϵ 0 c (5 δ / 4) 2 π ϵ0 c 5
Radiation Reaction
An accelerating charge radiates. This radiation carries off energy, which comes
at the expense of the particle’s kinetic energy.

Under the influence of a given force, a charged particle accelerates less than a
neutral one of the same mass.

The radiation exerts a force (Frad) back on the charge—a recoil force, like that
of a bullet on a gun―the radiation reaction force (from conservation of energy).

For a nonrelativistic particle (v≪c), the total power radiated is given by the
μ0 q2 a2
Larmor formula: P =
6πc
Conservation of energy asks that this is also the rate at which the particle loses
(only correct averagely) μ0 q 2 a 2
energy, under the influence of the radiation reaction force: F rad⋅v =−
6πc
In the calculation of the radiated power the velocity fields played no part, since
they fall off too rapidly as a function of � to make any contribution.

The velocity fields do carry energy—they just don’t transport it out to infinity.
As the particle accelerates/decelerates, energy is exchanged between it and the
velocity fields, as energy is also radiated away by the acceleration fields.
 The earlier consideration accounts only for the latter. If we want to know the
recoil force by the fields on the charge, we should consider the total power lost at
any instant, not just the portion that eventually escapes in the form of radiation.

The energy lost by the particle in any given time interval must equal the energy
carried away by the radiation plus the energy pumped into the velocity fields.

If we consider only intervals over which the system returns to its initial state,
eg, the periodic motion, then the energy in the velocity fields is the same at both
ends, and the only net loss is in the form of radiation.
t2
μ0 q2 t2

∫ ∫
2
F rad⋅v d t =− a d t ⇐ v (t 1 )= v (t 2 ) , a (t 1 )= a (t 2 )
t1 6πc t1
t2 t2 t2 t2 2
dv dv dv d v
∫
t1
a2 d t = ∫
t1
⋅
dt dt
d t = v⋅
dt t
−∫ | 1 t1 dt 2
⋅v d t

t2
μ 0 q2 μ0 q2
⇒ ∫ ( F rad − ȧ )⋅v d t = 0 ⇒ F rad = ȧ Abraham-Lorentz formula
t1 6πc 6πc
The expression tells nothing about the component of Frad  v, it only tells the
time average of the parallel component—the average over special time intervals.

However, it represents the simplest form the radiation reaction force could take,
consistent with conservation of energy.
 The Abraham-Lorentz formula has disturbing implications. Suppose a particle is
subject to no external forces; then Newton’s 2nd law says
μ0 q 2 t
μ0 q2
⇒ τ electron =6 × 10
−24
F rad = ȧ = m a ⇒ a (t)= a0 e τ
⇐ τ≡ s
6πc 6πmc
The acceleration increases exponentially with time unless a0=0!

The systematic exclusion of such runaway solutions has a more unpleasant

consequence: If you do apply an external force, the particle starts to respond
before the force acts! [Problem 11.19]
This acausal preacceleration jumps the gun by only a short time τ; however,
it is unacceptable that the theory should countenance it at all.

Example 11.4: Calculate the radiation damping of a charged particle attached

to a spring of natural frequency ω0, driven at frequency ω.

The equation of motion is

m ẍ = F spring + F rad + F drive =− m ω 20 x + m τ x⃛ + F drive ( ω) ⇒ x (t )= x 0 cos (ω t + δ )
2 2 2
⇒ ⃛x =− ω ẋ ⇒ m ẍ + m γ ẋ + m ω 0 x = F drive ⇐ damping factor γ = ω τ
2
⇒ F damping =− γ m ẋ =− m ω τ ẋ
The Mechanism Responsible for the Radiation Reaction
The fields of a point charge blow up right at the particle, so it’s hard to see how
one can calculate the radiation reaction force they exert.

Avoid this problem by considering an extended charge distribution with the

field is finite everywhere; then take the limit as the size of the charge goes to 0.

In general, the EM force of one part (A) on another part (B) is

not equal and opposite to the force of B on A.
If the distribution is divided up into infinitesimal chunks, and
the imbalances are added up for all such pairs, the result is a
net force of the charge on itself.
It is this self-force, resulting from the breakdown
of Newton’s 3rd law within the structure of the particle,
that accounts for the radiation reaction.

Consider a “dumbbell” in which the total charge q is

divided into 2 halves separated by a fixed distance d.

Although it’s an unlikely model for an elementary

particle: in the point limit (d0) any model must
yield the Abraham-Lorentz formula, to the extent
that conservation of energy dictates that answer.
 Assume the dumbbell moves in the x direction, and is instantaneously at rest
[β(tr=0)=0] at the retarded time. The electric field at 1 due to 2 is
q /2 c (1− β 2 ) ( �^ − β) + ⃗� ×[( �^ − β)× β̇] q (c + ⃗� ⋅β̇) ⃗� −� 2 β̇
E1 = 2 3
= 3
4 π ϵ0 � c (1− �^⋅β) 8 π ϵ0 c�
2 2
⃗� = ℓ x^ + d y^ ⇒ � = √ ℓ + d , ⃗� ⋅β̇ = ℓ β̇
Only interested in the x component of E1, since the y components will cancel
when we add the forces on the 2 ends.
q ℓ c − β̇ d 2
E1 x = 2 2 3 /2
⇒ E 2 x = E 1 x by symmetry
8 π ϵ0 c ( ℓ + d )
q q2 ℓ c2 − a d 2 q2 a 2
⇒ F self = (E 1 + E 2 )= 2 2 2 3/2
x^ ≃ 3
( ℓ − 2
d ) x^ for ℓ ≪ d
2 8 π ϵ 0 c (ℓ + d ) 8 π ϵ0 d c
1 1
x (t )= x (t r )+ ẋ (t r ) (t − t r ) + ẍ (t r ) ( x − t r )2 + x⃛ (t r ) (t − t r )3 +⋯
2 3!
1 1
⇒ ℓ = x (t )− x (t r )= a T 2 + ȧ T 3 +⋯ ⇐ T ≡ t − t r , ẋ (t r )=0
2 6
2 2 2 3
a T ȧ T a T
(c T )2 = ℓ2 + d 2 2 2 2
⇒ d = √ c T − ℓ = c T 1−( +
2c 6c √ +⋯ ) =c T −
8c
+⋯
 Need to solve the equation for T as a function of d. There is a systematic
procedure for doing this, known as reversion of series,
d
0th -order: d ≃ c T ⇒ T≃
c d a2 3 4
2 3 2 3
⇒ T = + d + O (d )
st a d d a d c 8c 5
1 -order: d ≃c T − ⇒ T≃ +
8c c 3
c 8 c5
2
a 2 ȧ 3 q a (t r ) ȧ (t r )
(− + O (d ) ) x^
4
⇒ ℓ= 2 d + d + O ( d ) ⇒ F self ≃ +
2c 6c 3
4 π ϵ0 2
4 c d 12 c 3

d
⇒ a (t r )= a (t) + ȧ (t ) (t r − t ) +⋯= a (t )− ȧ (t) T +⋯= a (t )− ȧ (t ) +⋯
c
q2 a (t ) ȧ (t )
⇒ F self = (− + + O (d ) ) x^
4 π ϵ0 2
4 c d 3c
3

The 1st term ∝ a(t); if we pull it over to the other side of Newton’s 2nd law, it
simply adds to the dumbbell’s mass.
2
1 q
In effect, the total inertia of the charged dumbbell m =2 m 0 +
4 π ϵ0 4 d c2
In special relativity, it is not surprising that the electrical repulsion of the
charges should enhance the mass of the dumbbell.
 2
1 (q /2)
For the potential energy of this configuration is and according
4 π ϵ0 d
to Einstein’s formula E = m c 2, this energy contributes to the inertia of the object.
2
int μ 0 q ȧ
The 2 term is the radiation reaction: F rad =
nd

12 π c
The term survives in the “point dumbbell” limit d0. But it differs from the
Abraham-Lorentz formula by a factor of 2.

But this is only the self-force associated with the interaction between 1 and 2.
There remains the force of each end on itself. When the latter is included, the
μ 0 q 2 ȧ
result is F rad = , reproducing the Abraham-Lorentz formula exactly.
6πc
q 1
F rad (q)= F (q)+ 2 F rad ( ) = F rad (q) + 2⋅ 2 ⋅F rad (q) ⇐ F rad (q) ∝ q
int int 2
rad
2 2
2
int μ 0 q ȧ
⇒ F rad (q)=2 F rad (q)=
6πc
Conclusion: The radiation reaction is due to the force of the charge on itself—or
the net force exerted by the fields generated by different parts of the charge
distribution acting on one another.