3

2 FEL THEORY



2.1 Introduction to the Theory of the FEL

Figure 2.1-1 Simplified diagram showing the electron beam motion in the wiggler and the
radiation process

Electron beam undergoes axial bunching in the wiggler due to energy spread
combined with the radiation field. Slightly higher and lower energetic electrons covers
slightly shorter and longer distance in the wiggler so head and tail catch up. Because of
this bunching the coherent radiation is produced.
By the time the first photon covers ct distance electron can cover vt distance. As shown
in Figure 2.1-1 vt is the length of the wiggler.
w w
N vt = Eqn 2.1-1
where N
w
is the number of wiggle in the wiggler and
w
is wiggler period. The path
difference between electrons and first photon emitted must be equal to N
w

s
where
s
is
the radiation wavelength.
Thus
s w
N t v c = ) ( Eqn 2.1-2
4
If we substitute N
w
with vt/
w
we get

) 1 (
=
w s
Eqn 2.1-3
which is approximately
2
2

w
s
= Eqn 2.1-4
Since the actual velocity in z direction is less than the electron velocity this expression
changes as follows
] )
2
(
2
1
1 [
2
2
2
mc
eB
w w w
s

+ = Eqn 2.1-5
We will come to this a bit later.


There are basically two kinds of wigglers
1) Helical
2) Planar
2.2 The FEL Equations with a Helical Wiggler in a Small Signal
Approximation

In this section we will introduce the FEL equations of motion with a helical
wiggler, which is a simpler case. We will solve the equations using a small signal
approximation.
A helical wiggler is characterized by a vector potential of a helical form:
2
1
) . ( c c e e A A
z ik
w w
w
+ =

r
Eqn 2.2-1
where
0 , 1 ,
2

= =

=
+
e e e e
y i x
e Eqn 2.2-2
As will be shown below, the electrons will follow helical path as depicted in Figure 2.2-1.
5

Figure 2.2-1 The electron trajectory in a helical wiggler
Eqn 2.2-1 can be written as:
)] sin( ) cos( [ z k y z k x A A
w w w w
+ =
v
Eqn 2.2-3
The magnetic field is derived from the vector potential:
)] sin( ) cos( [ z k y z k x A k A B
w w w w w w
+ = =
v v
Eqn 2.2-4
The Mechanical momentum can be written in terms of the canonical momentum P.
A e P
v v v
= Eqn 2.2-5
The canonical momentum is conserved before and after the field. So lets consider
only the perpendicular component of the equation. Since initially there is no momentum
in the perpendicular direction

P
v
will be zero.

= A e P
v v v
Eqn 2.2-6
Therefore

= A e
v v
Eqn 2.2-7
We can write the momentum in terms of the velocity.

= A e v m
v
v
Eqn 2.2-8
)] sin( ) cos( [ z k y z k x
m
eA
m
A
e v
w w
w w
+ = =

v
v
Eqn 2.2-9
)] sin( ) cos( [ z k y z k x
K
w w
+ =

v
Eqn 2.2-10
where K, which is called the wiggler magnetic strength parameter, is defined by
6
2
2 mc
c B e
mck
eB
mc
eA
K
w w
w
w w

= = = Eqn 2.2-11
In practical units
w w
B K

6
8
10 511 . 0 . 2
10 3

) ( ) ( 94 . 0 Tesla B cm K
w w
Which is usually of the order of 1

Energy Exchange
We want to use the wiggler and electrons to impart energy to an electromagnetic
field. So we have to introduce a radiation field and study the energy exchange between
the electrons and this field. As we understand from Eqn 2.2-10, the electrons move in a
helical orbit. This is not surprising because the vector potential is helical. We expect that
the electromagnetic radiation emitted by the electrons will also have a helical form. Thus
let us define:
. . ) , (
2
1
) (
c c e e t z E E
t w z k i
s s
+ =
+

v
Eqn 2.2-12
where k
s
is the wave number and w
s
is the frequency of the wave according to the plane
wave solution. The relation between the k
s
and w
s
is as follows:
c
w
k
s
s
= Eqn 2.2-13
In anticipation of the evolution of the electric field intensity along the wiggler, we
introduced a time-dependence in the amplitude. Normally there is no energy exchange
between the EM field and electron beam propagating in the same direction because their
scalar product would be zero (they are perpendicular to each other). Since there is a
transverse component of the electrons velocity in the wiggler there will be interaction
between the EM field and the electron beam.
Energy exchange is

= = r d E e d mc dW
v
v
.
2
Eqn 2.2-14
If we take the time derivative of the equation Eqn 2.2-14 we get
7
E v e
dt
d
mc
v
v

2
Eqn 2.2-15
We can change the independent variable from time to position:
E e
dz
d
mc
v v

2
Eqn 2.2-16
Let us substitute the form of E from Eqn 2.2-12
( ) c c e t z E
K
dz
d
mc
z k t w z k i
w s s
. ) , (
2
) ( 2
+ =
+

Eqn 2.2-17
Let us introduce a phase variable ,
.) . ) , ( (
2
2
c c e t z E
mc
eK
dz
d
i
+ =

Eqn 2.2-18
where
t w z k k
s w s
= ) ( Eqn 2.2-19
Now lets assume that the amplitude of the electromagnetic field has the form:
( )
)
2
(
0
,

s i
s s
e A w t z E Eqn 2.2-20
where w
s
is the frequency and A
s0
is a coefficient and
s
is another phase factor. The
meaning of this definition will become clearer when we do the calculations for a bunch of
electrons. The phase
s
will provide an individual initial phase for each electron. To work
with a dimensionless variable, we define the coefficient a
s
as:
s
s
a
mc
eA

0
Eqn 2.2-21
If we substitute these in the Eqn 2.2-18, after a simple algebra we get the important
energy equation:
) sin(
s
s s
K a k
dz
d

+ = Eqn 2.2-22

Phase Equation
Using the definition of the phase variable (Eqn 2.2-19) we can take the derivative
with respect to z and we get
8
dt
dz
w k k
dz
d
s w s
1
+ =

Eqn 2.2-23
which can be written as

1
s w s
k k k
dz
d
+ = Eqn 2.2-24
The parallel velocity can be derived using Eqn 2.2-10.
2
1
2
2
2
2 2
)
1
1 (


K
= =

Eqn 2.2-25
Since
2
2
1

K +
term is small compared to 1, we can make a Taylor expansion and get
2
2
2
1
1
1

K +
+ Eqn 2.2-26
When we plug this in Eqn 2.2-24 we get the other important equation, the phase
equation:
2
2
2
1

K
k k
dz
d
s w
+
Eqn 2.2-27

Resonant Condition
For a certain value of the electrons energy, the phase does not change along the
wiggler. This condition is called the resonant condition, and we denote the resonant
energy as
0
= . For this value of energy, the phase evolution of Eqn 2.2-27 becomes
zero.
0 =
dz
d

9
Therefore, the condition for resonance becomes
2
0
2
2
1

K
k k
s w
+
= Eqn 2.2-28
So we get the central wavelength of the spontaneous radiation.
) 1 (
2
2
2
0
K
w
s
+ =

Eqn 2.2-29
This is equivalent to the statement that electrons slip one
s
behind the EM wave in one
wiggler period
w
.
In the next stage we consider electrons whose energy is near the resonant energy,
but not quite.
When
0
, but
0
<<
0

0
0
2
2
0
2
2
0
2
0
2
2 ) 1 (
2
1

=
+
=
w w s w
k k
K
k k
dz
d
Eqn 2.2-30
Therefore we can write the energy and phase equations of motion for
0

) sin(
0
s
s s
K a k
dz
d

+ = Eqn 2.2-31
0
0
2

=
w
k
dz
d
Eqn 2.2-32

Small Gain Approximation
Our next task is to solve this set of two equations. Lets consider the case where a
s

is essentially constant and
s
is independent of z.
Lets define
s
+ Eqn 2.2-33
Also, let us use a dimensionless position variable,
z k
w
Eqn 2.2-34
and a variable , which we call the detuning variable:
10
0
0
2

: Eqn 2.2-35
Next, we define an angular frequency
s
, which we will identify later as the synchrotron
rotation frequency:
2
0
2
2

K a
k
k
s
w
s
s
Eqn 2.2-36
If we plug these definitions into the equations (Eqn 2.2-31, Eqn 2.2-32) we get the
celebrated FEL pendulum equations:

=
d
d
Eqn 2.2-37

sin
2
s
d
d
= Eqn 2.2-38
is analogous to the angle variable of a mechanical pendulum. If the input signal a
s
is
very small then
s
is also very small. In this case
s
becomes oscillation frequency in
the ponderomotive potential. This oscillation is called the synchrotron oscillation for
historical reasons (it describes the longitudinal motion of particles in a synchrotron).
Thus
s
is the synchrotron angular frequency.
The initial energy detuning is
0

Lets define

0
x ,
0

,
2
0
2

Eqn 2.2-39
is a dimensionless relative energy variable and x is the dimensionless position along the
wiggler scaled by the initial detuning.
By plugging these definitions into pendulum equations (Eqn 2.2-37 and Eqn 2.2-38) we
can obtain an even simpler form of the equations:

=
dx
d
Eqn 2.2-40

sin =
dx
d
Eqn 2.2-41
We define the small signal regime by <<1
11
In this regime we may apply a perturbation method and introduce iterations to
solve the equations. Having the solution for one electron (defined by the set of initial
conditions) we average over the phase of the individual electrons, and obtain the
change in energy of the ensemble.
We will skip the details of the derivation here and present the result.
<>=<
0
>+
2
x
3
f(x ) Eqn 2.2-42
where
3
sin
2
1
1 cos
) (
x
x x x
x f
+
= Eqn 2.2-43
By using the definition of the energy variable from Eqn 2.2-39 we can write
<

>=
2
1
< >=
2
0

<
0

> Eqn 2.2-44
and also by using Eqn 2.2-42 we can obtain the relation for energy loss.
<

>=
2
0

<
0
>= ) (
2
3 2 0
x f x

Eqn 2.2-45
We can use the definition of and x (Eqn 2.2-39) and get the following simplified
equation.
<

> ) (
2
1
3 4
x f
s
= Eqn 2.2-46

where
2
0
2
2

K a
k
k
s
w
s
s
= , z z k
w
w

2
= =
which are independent of the initial detuning
0
0
0
2

=
Naturally, when we use x it is at the end of the wiggler, that is at =k
W
L
W
=2N
W
.
Thus x is proportional only to the initial detuning. It happens to be the only variable
related to the initial detuning,
12

= =
0
0
0
4

w
N x detuning Eqn 2.2-47
In Figure 2.2-2 below we plot the gain function f(x) as a function of x. This plot has a
significant importance in FEL process, because depending on the value of the detuning
we can have positive or negative, minimum or maximum gain values. If we choose a
detuning which yields a positive gain value then we will have enhancement in the
radiation emitted by the FEL process. That means that electrons transfer their energy to
the radiation field. However if we choose a detuning which gives a negative gain value
then the electrons will gain energy from the radiation in the wiggler, which causes
acceleration of electrons. This process is laser acceleration, in this case an Inverse Free
Electron Laser (IFEL) process.

Figure 2.2-2 The gain function as a function of the normalized x variable, which is
proportional to the detuning (see text)


Small Gain Formula
Now that we have the energy transfer solved, we will estimate the small gain
formula. Lets assume that the optical beam size is equal to electron beam size with the
area A. We define the electron number density as n
o
so that the current would be given by
I
0
=ecn
0
A and lets take the electron bunch pulse length as t. The radiation field is
amplified only where it overlaps with the electron bunch, thus t is also the length of the
radiation pulse.
The radiation field energy is given by
13
At
Z
E
Energy
0
2
= Eqn 2.2-48
The gain is defined as the ratio of energy lost in the electron bunch to the energy of the
electric field in the wiggler. So the gain is
At
Z
E
e
t I
mc
G
0
2
0
0
2
0

= Eqn 2.2-49
where
e
t I
0
is just the number of the electrons in the bunch.
By using Eqn 2.2-46 we get
) (
2
3 0
2
0 0
2 4
x f
A
I
E e
Z mc
G
s

= Eqn 2.2-50
and also using the definitions of
s
and (Eqn 2.2-36 and Eqn 2.2-20) we can obtain
the expression for gain as
) (
) ( 2
) 2 (
3
0
2
0
0 0
2 2
2
0
x f c en
A w e
Z mc
K a
k
k
G
s s
s
w
s

= Eqn 2.2-51
Remembering
mc
eA
a
s
s
0
= we can also write the gain as
) (
2
3
3
0
2
2
0
2
0
x f
mck
K n e Z
G
w

= Eqn 2.2-52
In order to simplify the expression we can express the gain in terms of the Pierce
parameter , which is defined as
3
0
2
2
0
2
0 3
2
) 2 (

w
mck
K n e Z
= Eqn 2.2-53
Thus the gain in terms of the Pierce parameter is
) ( ) 4 ( 4
3
x f N G
w
= Eqn 2.2-54


14
Hamiltonian and Phase space
We had obtained the pendulum equations before in Eqn 2.2-37 and Eqn 2.2-38.
Now we will express the same equation in terms of the Hamiltons Canonical Equations.
We will start with a trial Hamiltonian as:
cos
2
1
2 2
s
H = Eqn 2.2-55
By applying the Hamiltons Canonical Equations to this expression we get

= = =
H
d
d
Eqn 2.2-56

and

sin
2
s
H
d
d
= = = Eqn 2.2-57
Which reproduces Eqn 2.2-37 and Eqn 2.2-38 (the pendulum equations). Therefore and
are canonical conjugate variables in Hamilton equations. (Similar to position and
momentum in classical mechanics).
Since our trial Hamiltonian fits these pendulum equations perfectly, it must be a
conserved quantity. Using this conserved Hamiltonian we can plot the phase space of the
electron trajectories.

Figure 2.2-3 Electron trajectories in - phase space, showing bounded and unbounded
trajectories and the separatrix.
In Figure 2.2-3 above we plot the derivative of the phase versus the phase itself
for different values of Hamiltonian H. As shown in Figure 2.2-3 some of the trajectories
are unbounded but some are bounded as closed loops.
15
For a certain value of Hamiltonian the trajectory intersects the phase axis at
..... 5 , 3 , . This special trajectory is called the separatrix, and it separates the
bounded trajectories (loops) from the unbounded ones.
The separatrix height is
s
2
The width of the separatrix is 2,which can easily be seen on the plot.
For small gain and small signal regime
0
>>
s
. The energy modulation varies
with z k x
w 0 0
= = .
Assume we have electrons with a given energy spread as shown in Figure 2.2-4.

Figure 2.2-4 Electron distribution in phase-energy space, before (left) and after (right
energy modulation by the radiation

Figure 2.2-5 The process of microbunching in a dispersive medium. The arrows show
direction of electron motion due to the modulation.
The interaction with the radiation modulates the energy of the electron beam as
shown in Figure 2.2-4. In the wiggler, the more energetic electrons bend less; therefore
cover a shorter distance than the less energetic electrons. This phenomenon causes the
more energetic electrons to catch up the less energetic electrons and that is called bunch
compression (specifically, magnetic bunch compression), leading to micro bunching. The
process is shown in Figure 2.2-6. Since the length of these micro-bunches is shorter than
16
the wavelength, these bunched electrons radiate coherently (in phase with the modulating
radiation). The strength of the radiation grows and we have gain. As radiation grows
s
also grows, therefore the separatrix height increases and radiation goes into the high
gain regime which will be discussed later.

Figure 2.2-6 The microbunched beam