Lecture 11

M. Siddikov

Marat.Siddikov@usm.cl

May 5, 2023
Outline
Today’s plan:
Potentials of Lienard-Wiechert in
medium. Cherenkov’s emission

Larmor’s formula (emission of nonrel-

ativistic particles)


Homework 2
 Available from Aula
 Deadline: Friday, 12/05/2023
Lienard-Wiechert potentials
Density of charge and current:

ρ (r , t) = e δ (r − w (t)) . R · u = cR − v · R =
⃗j (r , t) = e ẇ (t) δ (r − w (t)) = c |⃗r − w
⃗ (tr )| − (⃗r − w
⃗ (tr )) · ⃗v

Lienard-Wiechert’s potentials ⃗
dw d 2w
⃗
⃗v = , ⃗a =
1 dt dt 2
ϕ (r , t) = ×
4πε0 EM fields:
qc
|r − w (tr )| c − (r − w (tr )) · v ⃗ (⃗r , t) = q R ×
E (1)
4πε0
q 1
= c 2 − v 2 ⃗u + R × (⃗u × ⃗a )


4πε0 (R · u)
(R · u)3

⃗ (r , t) = µ0 qc v = v (tr ) ϕ
A
4π (R · u)3 c2 ⃗ (⃗r , t) = 1 R̂ × E
B ⃗
c
where tr = t − |r ′ − w (tr )|/c
−retarded distance/separation:
R = r − w (tr ) , u = c R̂ − v
R ≡ |R| = |⃗r − w
⃗ (tr )| = c (t − tr )
Special case: v = const
Electromagnetic fields:
1 q 1 − v 2 /c 2 R̂
ϕ (r , t) = × E =
4πε0 4πε0 1 − v 2 sin2 θ/c 2 3/2 R 2


qc
×q
2
(c t − r · v ) + (c 2 − v 2 ) (r 2 − c 2 t 2 )
2

q 1 1
=  B (r , t) = R̂ × E = 2 v × E
2
1/2 c c
R 1 − vc 2 sin2 θ
Poynting’s vector:
2
where θ is angle between R̂ and ⃗v

⃗ = 1 E
S ⃗ ×B ⃗ = 1 q
×
v µ0 µ0 c 2 R 4 4πε0
A (r , t) = ϕ (r , t)
2
c2 1 − v 2 /c 2 h i
×
3 R̂ × ⃗v × R̂
Recall that R = r − v t (instant, not 2
1 − v 2 sin θ/c 2
retarded separation).
... motion ⇒only additional factor If choose sphere around instant position o
... only for motion with constant v re- the charge, so dA = d|A|
˛ R̂
sult depends on instant R, in general dA · S = 0, dA · S = 0
dependence on retarded vectors
-charge does NOT emit EM waves
Lienard-Wiechert potentials
Effectively, this problem requires to
 2
 construct the spectral decomposition of
1 ∂ ρ
(1′′′ ) −∆ ϕ= the particular solution in case of motion
c 2 ∂t 2 ε0 of a charge with constant velocity.
1 ∂2
 
This might be useful for many applica-
(3′′′ ) − ∆ A = µ0 j
c 2 ∂t 2 tions (e.g. if we are interested only in
EM field in a given spectral range, e.g.
Evaluate the EM field of a point charge only in the form of the visible light)
moving with constant velocity ⃗v in the While mathematically we could just
Fourier space and find corresponding make Fourier of Lienard-Wiechert po-
potentials ϕ(⃗k), A(
⃗ ⃗k).
tentials, techncially it is difficult. For
(Landau, Section §64 and problem 1) this reason we’ll try to construct a so-
lution from scratch in Fourier space
ˆ
d 3k h
⃗
i
ρ (⃗r , t) = e δ (⃗r − ⃗
v t) = e exp i k · (⃗
r − ⃗
v t) .
(2π)3
ˆ
⃗j (⃗r , t) = e ⃗ d 3k h i
v δ (⃗r − ⃗
v t) = e⃗
v 3
exp i ⃗ k · (⃗r − ⃗v t)
(2π)
Lienard-Wiechert potentials
ˆ
1 ∂2 d 3k
 
′′′ ρ   h i
(1 ) −∆ ϕ= ϕ (⃗r , t) = φ ⃗ k, t exp i ⃗ k · ⃗r
c 2 ∂t 2 ε0 (2π) 3

1 ∂2
 
(3′′′ ) − ∆ ⃗ = µ0⃗j
A φ̈ 2
h
⃗
i
c 2 ∂t 2 + k φ = e exp −i k · ⃗
v t
c2
h i
Evaluate the EM field of a point charge   e exp −i ⃗ k ·⃗
vt
⇒φ ⃗ k, t = 2 ,
moving with constant velocity ⃗v in the

k2 − ⃗ v /c 2
k ·⃗
Fourier space and find corresponding
potentials ϕ(ω, ⃗k), A(ω,
⃗ ⃗k). h i
  e⃗v exp −i ⃗ k ·⃗vt
⇒A⃗ ⃗ k, t =  2
(Landau, Section §64 and problem 1) k2 − ⃗ v /c 2
k ·⃗
h i   
ρ (⃗r , t) = e δ (⃗r − ⃗ v t) −ie exp −i ⃗ k ·⃗
vt ⃗
k ·⃗ v ⃗v
ˆ ⃗ =
E 2 ⃗k− 
3 
c 2
d k h i
2
k − k ·⃗⃗ 2
=e 3
exp i ⃗ k · (⃗r − ⃗v t) v /c
(2π) h i
⃗j (⃗r , t) = e ⃗v δ (⃗r − ⃗v t) e⃗k ×⃗ v exp −i ⃗k ·⃗vt
ˆ ⃗ = i ⃗k × A
B ⃗ = 2
3

d k h
⃗
i
k2 − ⃗ v /c 2
k ·⃗
= e⃗ v exp i k · (⃗
r − ⃗
v t)
(2π)3
Puzzle ...
In our evaluations we assumed that v < c,
q 1 − v /c 2 2
R̂ which is always valid for massive particles
⃗ =
E
4πε0 1 − v 2 sin2 θ/c 2 3/2 R 2

in vacuum, according to Special Relativity.
−Assume we have dielectric with permittiv-
ity εr , permeability µr and refr. index
where θ is angle between R̂ and ⃗v √
n = εr µr > 1, so speed of light in it
c
⃗ (⃗r , t) = 1 R̂ × E
B ⃗ = 1 ⃗v × E ⃗ = cdiel =
n
c c2
∗
q 1 − v 2 /c 2 ⃗v × R̂ Lienard-Wiechert solution apparently re-
=
R2

3/2
4πε0 1 − v 2 sin θ/c 2
2 mains valid, but need to adjust ε0 → ε0 εr ,
µ0 → µ0 µr , c → cdiel
 2 −Assume a high energy charged particle with
⃗ = 1 E
S ⃗ ×B ⃗ = 1 q velocity
µ0 µ0 c 2 R 4 4πε0 c
<v <c

2 n
1 − v 2 /c 2 h i
×
3 R̂ × ⃗v × R̂ enters the dielectric. What happens with
2
1 − v 2 sin θ/c 2 Lienard-Wiechert solution?
h i  
R̂ × ⃗v × R̂ = ⃗v − ⃗v · R̂ R̂
Cherenkov radiation
¸
There is no EM waves in the re- The flux dA · S ̸= 0
gion where c > v > c/(n sin θ) ≡ −particle start emitting waves even if it has
c/(n cos θc ): constant velocity
Frank-Tamm formula
Energy loss per unit of length:
 
dW
=
dxparticle rad
ˆ
e2
 
1
= ω 1 − dω
4πε0 c 2 ϵ(ω)> 12 β 2 ϵ(ω)
β

2 ˆ
c2
 
e
= ω 1− dω
4πε0 c 2 c
v > n(ω) v 2 n2 (ω)
Clearly
dxparticle ≡ v dt
geometrically the field remains inside
the cone shown in the Figure Full derivation is lengthy and tedious, in
−Can use Lienard-Wiechert results essence-evaluation of flux through the sur-
face using spectral decomposition
Cherenkov radiation
− Assume the charge is moving with constant velocity v > c/n
−Now we are going to find spectral decomposition of the emitted waves (Frank-Tamm
formula)
−We’ll work in momentum space, assuming 4D Fourier transform over coordinates
and time,
ˆ
dω d 3k
φ(⃗x , t) = 1/2
e −iωt e ik·x Φ(ω, k),
(2π) (2π)3/2
ˆ
dω d 3k
A(⃗x , t) = 1/2
e −iωt e ik·x A(ω, k),
(2π) (2π)3/2

ω2
 
1
k 2 − 2 ϵ(ω) Φ(k, ω) = ρ(k, ω)
c ε0 ϵ(ω)
ω2
 
k 2 − 2 ϵ(ω) A(k, ω) = µ0 j (k, ω)
c

We assumed for the sake of generality that ϵ = ϵ(ω)

Cherenkov radiation

For motion with constant velocity ρ(r , t) = e δ (r − v t) , j = ρv

e
Fourier ⇒ ρ(⃗k, ω) = δ(ω − ⃗k · ⃗v )
2π
⃗j (⃗k, ω) = ⃗v ρ(⃗k, ω)
Solutions:
e δ(ω − ⃗k · ⃗v )
Φ(⃗k, ω) =
2πε0 ϵ(ω) k 2 − ω22 ϵ(ω)
c

⃗ ⃗k, ω) = ϵ(ω) ⃗v Φ(⃗k, ω)

A(
c2
 
⃗ (⃗k, ω) = i ωϵ(ω) ⃗v − ⃗k Φ(⃗k, ω)
E
c c
⃗ ⃗k, ω) = iϵ(ω)⃗k × ⃗v Φ(⃗k, ω)
B(
c
Cherenkov radiation
General evaluation is challenging, so let’s assume the patricle moves in x̂-
direction and evaluate the energy emitted through the surface of a cylinder of
radius b surrounding it, component-by-component
Dependence on impact parameter b (coordinate at the surface of the cylinder):
ˆ
1
⃗ (ω, b) =
E d 3k E
⃗ (k, ω)e ib·k
(2π)3/2
 We expect that fields are axially symmetric
... instead of working in polar coordinates can use Cartesian for simplicity:
 Consider e.g. that b = (0, 0, b, 0) (emission in ⊥ y -direction) and evaluate
normal component of the flux in that point (it will be the same for all points due
to axial symmetry).
evaluate explicitly components:
ˆ  
ie 3 ibk2 ωϵ(ω)v δ(ω − vk1 )
E1 (ω, b) = d ke − k1 2
2πε0 ϵ(ω)(2π)3/2 c2 k 2 − ωc 2 ϵ(ω)
´
Integral dk1 taken trivially, using δ-function (so we’ll get k1 = ω/v ).
ω2 ω2 ω2
Let’s define λ2 = 2 − 2 ϵ(ω) = 2 1 − β 2 ϵ(ω) and take remaining inte-


v c v
grals over dk2 dk3
Cherenkov radiation

 ˆ ∞ ˆ ∞
ieω 1 dk3
E1 (ω, b) = − − β2 dk2 e ibk2
2πε0 v 2 (2π)3/2 ϵ(ω) −∞ −∞ k22 + k32 + λ2
ˆ ∞
dk3 π
= 2
−∞ k22 + k32 + λ2 (λ + k22 )1/2
ˆ ∞
e ibk2

ieω 1
E1 (ω, b) = − √ − β2 dk2 2
2πε0 v 2 2π ϵ(ω) −∞ (λ + k22 )1/2
 1/2  
ieω 2 1
=− − β 2 K0 (λb)
2πε0 v 2 π ϵ(ω)
Similarly can get other components:
 1/2
e 2 λ
E2 (ω, b) = K1 (λb), E3 = 0
4πε0 v π ϵ(ω)

B1 = B2 = 0, B3 (ω, b) = ϵ(ω)βE2 (ω, b)

Cherenkov radiation
Evaluate the energy emitted through the surface of an infinite cylinder surround-
ing it
 Instead of energy emitted per unit of time it is more common to use energy
per unit of distance traveled;

dxparticle = v dt
dW 1 dW
=
dxparticle v dt
Flux through the surface:
ˆ
2πb ∞
 
dW 1
= Sa=⊥ = − B3 E1 dx =
dxparticle rad v µ0 v −∞
ˆ ∞ ˆ ∞
2πb 2πb
=− B3 (t)E1 (t)dt = − B3∗ (ω)E1 (ω)dt
µ0 −∞ µ0 −∞

where at the last step we jumped from configuration (coordinate) space to fre-
quency domain (Fourier)
Cherenkov radiation
Let’s assume that b is very large, so λb ≫ 1 and we can replace K0 , K1 with
their asymptotic expansions   −λb
ieω 1 e
E1 (ω, b) → 1− 2 √ ,
4πε0 c 2 β ϵ(ω) λb
r
e λ −λb
E2 (ω, b) → e , B3 (ω) = ϵ(ω)βE2 (ω)
4πε0 c 2 v ϵ(ω) b
  ˆ ∞ 2 r    
dW e λ∗ 1 −(λ+λ∗ )a
= Re −i ω 1− 2 e dω
dxparticle rad 0 c2 λ β ϵ(ω)
ω2 ω2 ω2
Earlier we defined λ2 = − 2 ϵ(ω) = 2 1 − β 2 ϵ(ω) .


v 2 c v∗
 If v < c/n, then λ is real, then λ = λ and thus exponent is suppressed for
large b → ∞
If v > c/n, then λ is imaginary, λ∗ = −λ and exponent gives one, yielding
ˆ
e2
   
dW 1
= ω 1 − dω =
dxparticle rad 4πε0 c 2 ϵ(ω)> 12 β 2 ϵ(ω)
β
ˆ
e2 c2
 
= ω 1− dω
4πε0 c 2 c
v > n(ω) v 2 n2 (ω)
This result is known as Frank-Tamm formula.
Cherenkov radiation
ˆ
e2
   
dW 1
= ω 1− 2 dω =
dxparticle rad 4πε0 c 2 ϵ(ω)> 12 β ϵ(ω)
β

2 ˆ
c2
 
e
= ω 1− dω
4πε0 c 2 c
v > n(ω) v 2 n2 (ω)
Note that for all materials
lim n(ω) = 1
ω→∞

(weak atomic forces cannot affect hard X -rays and γ-quants).

´
Since v < c, this implies that dω is cut from above by some frequency ωmax
which satisfies
n(ωmax ) = c/v ,
and thus integration over dω yields finite result (which depends on n(ω) for a
given material).
Emission of EM waves by isolated charges (summary)

We’ve seen so far that EM waves are NOT emitted if a charge moves with constant
velocity ⃗v (v < c)
 Flux of energy (Poynting vector):
 Locally, S = µ−1
0 E × B ̸= 0 ¸
Integration over closed surface: dA · S = 0
(i.e. only accelerated charges can emit EM waves)

In general emission of EM waves for rapidly moving accelerated particle gives

complicated expression.

For many applications we may assume that particle moves slowly (v ≪ c) and
significantly simplify our expressions.

We’ll see some examples in order to understand the general picture.

Emission of an accelerated charge

Evaluate the EM emission of the accelerated charge assuming that the typical velocity
of the charge in a chosen frame is always small, v ≪ c.
Larmor’s formula
Earlier we got Lineard-Wiechert solution for the single accelerated charge in general
case. Now we need to take limit of small velocities v /c ≪ 1 (equivalently c → ∞).
As we have seen earlier, the term without acceleration does not lead to emission
¸
( dA · S = 0), so we need to focus on the term ∼ a only.

First need to find retarded time tr as a function of r , t:

c (t − tr ) = |r − w (tr )|

c 2 (t − tr )2 = r 2 + w 2 (tr ) − 2r · w (tr )
In general we expect that t − tr = |r − w (tr )| /c = O(1/c), so we may expand:
δ1 δ2
tr ≈ t + + 2 + ...
c c
We’ll assume for simplicity that locally v (t) = 0 to simplify algebra
 2
δ1 a (t) δ1
w (tr ) ≈ w (t) + v (t) + + ...
c 2 c

|r − w (t)|
c (t − tr ) = |r − w (tr )| ≈ |r − w (t)| ⇒ tr ≈ t −
c
Larmor’s formula

(General) Lienard-Wiechert solution:

2 2


⃗ (⃗r , t) = q R c − v u + R × (⃗u × a )
⃗ ⃗
E
4πε0 (R · u)3
⃗ (⃗r , t) = 1 R̂ × E
B ⃗
c

⃗
dw d 2w
⃗
⃗v = , ⃗a =
dt dt 2
⃗ = ⃗r − w
R ⃗ (tr ) , ⃗ − ⃗v
⃗u = c R

R · u = cR − v · R = c |⃗r − w
⃗ (tr )| − (⃗r − w
⃗ (tr )) · ⃗v

Need to simplify assuming v ≪ c

Larmor’s formula Recall A × [B × C ] with A ≡ B ≡ n, C =
⃗ = ⃗r − w
R ⃗ (tr ) , n ≡ R̂ n × a:
⃗ v ≈ c Rn
⃗u = c R−⃗  

R̂ · u ≈ R A×[B×C ] = n n · [n × a]−n×a (n · n)
 

⃗ (⃗r , t) = q R ×
| {z } | {z }
E =0 =1
4πε0
2 2


c − v ⃗u + R × (⃗u × ⃗a )
 3 ⃗ (⃗r , t) = − q [n × a]
⇒B
R̂ · u 4πε0 c 3 R
q n × [n × a]
≈ Both E and B are transverse to direction
4πε0 c 3 R
of propagation ⃗n

⃗ ≈ 1 E
S ⃗ ×B ⃗ = 1 |B| ⃗ 2 ⃗n
µ0 µ0
⃗ (⃗r , t) = 1 n × E = qn × [ n × [n × a]]
B
c 4πε0 c 3 R 
q
2
1
⃗ 2=
|B| |a (t0 )|2 sin2 θ
4πε0 c 3 r2
| {z }
const

where θ is angle between n

⃗ and a (t0 )
Larmor’s formula Total emission (power per unit of time):
ˆ
I = r 2 dΩ S · n =
Evaluate the EM emission of the
accelerated charge assuming that the  2 
q µ0
typical velocity of the charge in a = |a (t0 )|2
12πc
chosen frame is always small, v ≪ c. | {z }
const

⃗ (⃗r , t) ≈ q n × [n × a]
E Note that I does not depend on radius r
4πε0 c 3 R
of the surface used for estimates

Spacial distribution of flux density (as-

⃗ (⃗r , t) = − q [n × a]
⇒B
sume axis ẑ points in direction of a, so
4πε0 c 3 R a ∼ ẑ)

⃗ ≈ 1 E
S ⃗ ×B ⃗ = 1 |B| ⃗ 2 ⃗n
µ0 µ0
 2
⃗ 2= q 1
|B| |a (t0 )|2 sin2 θ
4πε0 c 3 r2
| {z }
const
Case of harmonic oscillations: I ∼ ω 4
Some observations:
We could expect that energy loss leads to appearence of reaction force which
decelerates particle (to keep total energy constant).

However, Larmor’s (approximate) result suggests that there is NO mechanical force

acting on a particle which moves with constant acceleration a:

−Force acting on particle is related to it momentum pas

dp dp
F = = − em , p + p em = const
dt dt
where p em is the momentum carried by EM field.

1
´
− Lecture 6: In general p em = c2
dV S

−For Larmor’s (dipole) emission

.
1 1 ⃗ 2
S≈ E ×B = |B| ⃗n
µ0 µ0
 2
⃗ 2= m 1
|B| |ä (t0 )|2 sin2 θ
4πε0 c 3 r2
| {z }
Some observations: ´
1
− Lecture 6: In general p em = c2
dV S

−For Larmor’s (dipole) emission

.
1 1 ⃗ 2
S≈ E ×B = |B| ⃗n
µ0 µ0
 2
⃗ 2= m 1
|B| |ä (t0 )|2 sin2 θ
4πε0 c 3 r2
| {z }
const
´
After integration over angles can see that dV S = 0, i.e. total momentum of the
emitted EM field is zero (!)

−More accurate treatment: the force appears when we take into account higher
order terms in 1/c expansion,

d 3r µ0 q 2
F rad = κȧ = κ , κ= .
dt 3 6πc
−Abraham-Lorentz force, radiation reaction force
Abraham-Lorentz formula:

d 3r µ0 q 2
F rad = κȧ = κ , κ= .
dt 3 6πc
We’ll use derivation from (Landau, Section §75):
−We’ll make 1/c expansion in expressions for retarded potentials and thus will find
EM fields created by moving charges
−We’ll assume that we have only one charge, and will find its interaction force with
its own emitted electromagnetic field
Radiation reaction (Abraham-Lorentz formula)
R
R = r − r ′ 
 
tr ≡ t − ,
Retarded potentials* c
ϕ = ϕ(0) + ϕ(1) + ϕ(2) + ϕ(3) +...
ˆ ′
1 ρ (r , tr )
|{z} |{z} |{z} |{z}
ϕ (r , t) = d 3r ′ c0 c −1 c −2 c −3
4πε0 |⃗r − ⃗r ′ |
ˆ
µ0 j (r ′ , tr )
A (r , t) = d 3r ′
4π |⃗r − ⃗r ′ |
Clearly expansion over 1/c gets dominant contribution from expansion of tr in argu-
ment of ρ (r , tr ),
X 1  R k  ∂ k
ρ (r , tr ) = − ρ(r , t),
k! c ∂t
k

The term ϕ(0) is monopole (Coulomb), not relevant for emission at all
The term ϕ(1) vanishes due to charge conservation for closed system
The term ϕ(2) is dipole emission, does not contribute to reaction force
The term ϕ(3) is additional term responsible for reaction force
Radiation reaction (Abraham-Lorentz formula)
ˆ ˆ
1 1 ... µ0
ϕ(3) = − dV ′ ρ r ′ , t0 R 2 A′ = − dV ′ ∂t j r ′ , t0 −



3
4πε0 c 6 4πc
ˆ ˆ
µ0 1 1
A(2) dV ′ ∂t j r ′ , t0 dV ′ ρ̈ r ′ , t0 R 2 =



=− − 2
∇
4πc 4πε0 c 6
ˆ
1
recall that µ0 = 1/(ε0 c 2 ) for this rea- dV ′ ∂t j r ′ , t0 +


=− 2
4πε0 c
son in A we make expansion only up ˆ
1 1
to the 1st order dV ′ ρ̈ r ′ , t0 R


+ 2
 Gauge freedom: fields (E , B) don’t 4πε0 c 3
change under Let’s assume any charges ρ(r , t) and cur-
1 ∂f rents j(r , t) might be represented as en-
ϕ→ϕ− , A → A + ∇f semble of n particles of pointlike charges
c ∂t
qi moving along the trajectories r i (t) with
where f is an arbitrary function. velocities v i (t) ≡ ṙ i (t)
Choose
ˆ
X 3
1 1 ⇒ ρ(r , t) = qi δ (r − r i (t))
dV ′ ρ̈ r ′ , t0 R 2


f =− i
4πε0 c 2 r 6 X
⇒ j (r , t) = qi ṙ i (t) δ 3 (r − r i (t))
to eliminate the scalar potential (ef- i
fectively switch to Weyl gauge)
Radiation reaction (Abraham-Lorentz formula)
ˆ X
∂t dV ′ j → qi v̇ i

Since R = r − r ′ , after replacement get

ˆ X
∂t2 dV ′ ρ r ′ , t0 R →


qi v̇ i

1 2X ∂A µ0 X
A′ = − qi v̇ i , E =− = qi v̈ i
4πε0 c 2 3 ∂t 6πc
Let’s assume we have just one charge and find its interaction with higher order EM
field:
µ0 q 2 2 q2
F rad ≈ q E = κv̈ , κ = =
6πc 3 4πε0 c 3
Average work of the force per unit of time for a charge moving along a closed contour
ˆ
κ T
F rad · v ≡ dt F rad · v =
T 0
ˆ T ˆ
κ κ T
dt v · v̈ = − dt v̇ 2 < 0
T 0 T 0

where we used that v · v̈ = d (v · v̇ ) /dt − (v̇ )2 and dropped the surface term
Radiation reaction (Abraham-Lorentz force)

d 3r µ0 q 2
F rad = κȧ = κ , κ= .
dt 3 6πc
Full force in EM field:
F = q (E + v × B) +κ ȧ
Some challenges with F rad :
−in contradiction with Laplace’s determinism: for 3rd order equation

d 2r
m = F ext + F rad (1)
dt 2
it is not sufficient to know just coordinates and velocities at some time t0 to predict
evolution (should go beyond classical mechanics paradigms)
−Technically, F rad changed the structure of diff. equation in partial derivatives, and
thus we get extra unphysical (“runaway”) solutions, divergent in the limit κ → 0.
∗
Typically, for such solutions particle accelerates very rapidly to relativistic veloc-
ities. For such trajectory cannot use multipolar expansion at all, so it is NOT a
possible trajectory
⇓
−Should explicitly impose condition that for κ → 0 the solution should reproduce the
trajectory found in the limit κ → 0.
Radiation reaction (Abraham-Lorentz force)
In general we expect that dependence of trajectory on κ is analytic, so may develop a
perturbation theory:
X κn
r n (t) ≈ r 0 (t) + κ r 1 (t) + O κ2


r (t) =
n
n!
and solve (1) order-by-order:
−Leading order:
d 2r 0
m 2 = F ext (r 0 (t), ṙ 0 (t), t) (1a)
dt
−O(κ) correction (use components i, j to avoid confusion):

d 2 r1i ... X3 i
∂Fext (r 0 (t), ṙ 0 (t), t) j X3 i
∂Fext (r 0 (t), ṙ 0 (t), t) j
i
m = r 0 + j
r1 (t) + r˙1 (t) (1b)
dt 2 j=1
∂r0 j=1
∂ r˙0j
..............................................................................
...
In view of (1a), may rewrite r 0 as

... 1 dF ext (r 0 (t), ṙ 0 (t), t)

r0=
m dt
A particle of mass m moves in the central potential U(r ) over a circumference of
radius R. Find the direction of the radiation reaction force F rad . Demonstrate that
your result is compatible with expectation that particle looses its mechanical energy
due to emission of EM waves.
Emission in case of collision of particles

Evaluate the dipole emission for a

closed system of two charged particles.
Prove that for a closed system of two
particles with the same ratio of
charge-to-mass q/m the electric dipole
emission does not exist. Assume that
the motion is slow (typical velocities
v ≪ c), for this reason may use
classical (Newtonian) mechanics for
description of particle’s motion with
interaction given by the Coulomb’s law.
Emission in case of collision of particles
Evaluate the dipole emission for a
and relative motion with r 12 ≡ r 1 − r 2 and
closed system of two charged particles
mass µ = mm11+m
m2
2
intercating electromagnetically. Assume
In the CM frame, R = 0
that the motion is slow (typical
velocities v ≪ c), for this reason may ⇒ m1 r 1 = −m2 r 2
use classical (Newtonian) mechanics for
description of particle’s motion with m2
interaction given by the Coulomb’s law. ⇒ r1 = r 12 ,
m1 + m2
Prove that for a closed system of two m1
r2 = − r 12 ,
particles with the same ratio of m1 + m2
charge-to-mass q/m the electric dipole
Dipole moment:
emission does not exist.

For nonrelativistic motion can as- P = q1 r 1 + q2 r 2 =

sume that the energy/momentum of q1 m2 − q2 m2
= r 12 =
the EM field is tiny compared to m1 + m2
 
energy-momentum of particles, so Ṙ ≈ q1 q2
= µr 12 −
Pcm / (m1 + m2 ) ≈ const m1 m2
Classical Mechanics: can separate mo-
tion of center-of-mass ⇒If q1 /m1 = q2 /m2 , P = 0, no dipole
emission!
m1 r 1 + m2 r 2
R=
m1 + m2
Collision of particles p
= 1 + e cos ϕ
Evaluate the dipole emission for a r
closed system of two charged particles
intercating electromagnetically. Assume L2 4πε0 L2
p= = ,
that the motion is slow (typical mα mq1 q2
s
velocities v ≪ c), for this reason may
r
2EL2 32π 2 ε20 E L2
use classical (Newtonian) mechanics for e = 1+ = 1 +
mα2 mq12 q22
description of particle’s motion with
interaction given by the Coulomb’s law. In case of scattering (E > 0):
Prove that for a closed system of two
2 2
particles with the same ratio of mv∞ ρ
p= ,
charge-to-mass q/m the electric dipole α
emission does not exist.
s
2 ρ 2
 
mv∞
e = 1+
Dipole moment: α
 
q1 q2 where v∞ is the velocity and ρ is the impact
P = µr 12 − := µr 12 Q12 :
m1 m2 parameter

For v ≪ c relative motion is described

by potential U (r ) ∼ q1 q2 /r12
Always in plane orthogonal to ⃗L ≡ ⃗r ×
⃗p
Trajectory described by conical section
Motion in the Coulomb field - t-dependence
Dependence of r (t) from Classical Cartesian coordinates:
Mechanics course:
p = r − e x ⇒ x(t) = a (cos ξ − e)
t − t0 =
ˆ r
p p
dr y (t) = r 2 − x 2 = a 1 − e 2 sin ξ
= q (1)
2
(E − Ueff (r )) Dipole moment:
m
ˆ r dr P = µr 12 Q12 =
= r
L2
 
2 α = µQ12 (x(t) x̂ + y (t)ŷ )
m
−|E | + r
− 2m r 2

∂P ∂P/∂ξ
Let’s change integration variable: Ṗ = =
∂t ∂t/∂ξ
r = a − a e cos ξ
ˆ Q12 α
r P̈ = Q12 µr̈ 12 = − ⃗r 12
r
ma a e ξ(1 − e cos ξ) dξ
sin 3
⇒ t − t0 = a  = r12
α a e
sinξ
|Q12 α|
  
P̈  =
r  
ma r −a
 
=a
α
(ξ − e sin ξ) , ξ = arccos
ae
a2 (1 − e cos ξ)2
Motion in the Coulomb field

For a bound state of two charged

particles, evaluate the energy emitted
per period of rotation and compare it
with energy of the orbital motion for
the case of classical hydrogen atom
Motion in the Coulomb field
The intensity=max in direction trans-
verse to P̈ ∼ r 12 and vanishes in direction
For a bound state of two charged collinear to P̈ ∼ r 12
particles, evaluate the energy emitted The intensity I ̸= const
per period of rotation and compare it ˆ T
with energy of the orbital motion for
W = dt I =
the case of classical hydrogen atom 0
 µ  Q2 α 2 ˆ T dt
Q12 α =
0 12
P̈ = Q12 µr̈ 12 = − ⃗r 12 4 4 =
3
r12 | 6πc {z a } 0 (1 − e cos ξ)
const
|Q12 α|
 µ  Q2 α2 r m ˆ 2π
 
P̈  =
 
a (1 − e cos ξ)2
2 0 12 dξ
= =
6πc a2 aα 0 (1 − e cos ξ)3
| {z }
r const
ma  µ  Q2 α 2 r m
t − t0 = a (ξ − e sin ξ) , 0 12 2 + e2
α = 2
π
  6πc a aα (1 − e 2 )5/2
r −a
ξ = arccos
| {z }
ae const
 µ  2
I =
0
P̈ (t0 )
  As a function of a W ∼ 1/a5/2
6πc
| {z } W = min for e = 0 , increases for e → 1
const Explain (in simple words) why this happens
Motion in the Coulomb field

Using the results of the previous

slides, evaluate the total energy
emitted in case of collision of two
charged particles
Motion in the Coulomb field
The collision depends on scattering angle
cos ϕ0 = 1/e
Using the results of the previous ⇒Can repeat the same algorithm as in pre-
slides, evaluate the energy emitted in vious problem, but with different limits of
case of collision of two charged integration
particles
ˆ T
Trajectory ⇒hyperbola, e > 1
W = dt I =
The collision depends on scattering 0
angle ϕ0  µ  Q 2 α 2 r m ˆ ξ0 dξ
0 12
= =
6πc a2 aα 0 (1 − e cos ξ)3
| {z }
const

µ3 v∞
5
Q212 θ
= const tan3 [(π ± θ)×
3c 3 |α| 2

θ θ
(1 + 3 tan2 ) ± 6 tan
2 2
Synchrotron radiation
The EM emission leads to loss of mechan-
ical energy of a particle, for this reason this
A particle with charge q and initial
velocity ⃗
v moves in a constant emission is called magnetic bremsstrahlung
magnetic field B. Evaluate the EM (“breaking radiation”). Assume that the
emission for this case. Assume that rate of energy loss is small compared to the
v ≪c
energy of particle
 If energy loss is small, then can as- ⇒As expected, particle rotates with cy-
sume as a first approximation can dis- clotron frequency ωc around B
regard it when evaluating trajectory. 2nd derivative of dipole moment:
If we assume ẑ points in direction of
B, then EOM have a form P̈ = −qv0 ωc (sin (ωc t − φξ ) x̂ + cos (ωc t − φξ ) ŷ )
m v̇x = q vy Bz ,  
P̈(t) = qv0 ωc = const
 
m v̇y = −q vx Bz ,
ξ(t) : = vx (t) + ivy (t) Intensity of emission:
vx (t) = Re(ξ), q 2 µ0 v02 ωc2 q 4 µ0 Bz2 v02
I = =
vy (t) = Im(ξ), 12πc 12πc m2
⇒ ξ˙ = −i q ξBz /m -does not depend on time.
ξ(t) = v0 e −iωc t Note however that spatial distribution de-.
pends on orientation of P and thus oscil-
ωc = q Bz /m
lates with frequency ωc
vx (t) = v0 cos (ωc t − φξ ) , External observer: detected EM wave
vy (t) = −v0 sin (ωc t − φξ ) , intensity and polarization oscillate with
v0 frequency ωc
x(t) = sin (ωc t − φξ ) ,
ωc
v0
y (t) = cos (ωc t − φξ ) ,
ωc