12/4/01 USPAS Lecture 10 1

LECTURE 10
Single particle acceleration:
Phase stability
Linear Accelerator Dynamics:
Longitudinal equations of motion:
Small amplitude motion
Longitudinal emittance and adiabatic damping
Large amplitude motion
12/4/01 USPAS Lecture 10 2
We now begin to consider the interaction between the longitudinal
electric field in accelerating cavities or waveguides, and charged
particles.
The most important feature of this interaction is the principle of
phase stability.
This applies both to linacs and synchrotrons, both for standing
wave and traveling wave structures.
It is this feature that allows us to accelerate simultaneously a group
of particles, with a spread in energies and a spread in time: a bunch
12/4/01 USPAS Lecture 10 3
L
RF Cavities
1 2 n
.......
s
Synchronous particle
a b
t
Cavity 1
Cavity 2
eV
b
t
a
a
b
acc
sin t E

s
s

12/4/01 USPAS Lecture 10 4

Consider 3 particles entering a string of rf cavities (the reasoning is
identical for a travelling wave structure). One is at the reference
energy (this particle is called the synchronous particle); one (b) is
slow, and one (a) is fast. The synchronous particle arrives at cavity
1 at time t
s
s

and gains energy E eV

s acc s
sin .
V
acc
is the effective accelerating voltage (includes the transit time
factor).

s
is called the synchronous phase.
For synchronism, the rf cavities must be spaced by L h
s
,
where h is the number of rf cycles between cavities (called the
harmonic number),
s
is the synchronous velocity after the cavity,
and

is the rf wavelength.
The fast particle, a, arrives at t t
a s
< and gains energy E E
a s
< .
12/4/01 USPAS Lecture 10 5
The slow particle, b, arrives at t t
b s
> and gains energy E E
b s
> .
The synchronous particle arrives at the next cavity at the same
phase
s
,(this is the definition of the synchronous particle: it is in
perfect synchronism with the rf). But, particle a, having gained less
energy and velocity, slips later, while particle b, with a higher
velocity, slips earlier.
In subsequent cavities, particles a and b will oscillate in phase
about the synchronous particle. This oscillation is called a
synchrotron oscillation.
Lets see how this works out quantitatively.
Linear Accelerator Dynamics:
Longitudinal equations of motion
12/4/01 USPAS Lecture 10 6
The synchronous particle has energy E
s
, and always arrives at an rf
cavity at a timet
s
s

relative to the rf zero-crossing.

The rf cavities are numbered by the index n. Well measure the
energy of non-synchronous particles relative to that of the
synchronous particle; then, at cavity n, the non-synchronous
particles time and relative energy are
t E E E
n n n s n
,
,

in which the time is measured from the zero-crossing of the rf in
cavity n.
The energy change between one cavity and the next is
E E
dE
dn
eV t
dE
dn
eV t
d
dn
E eV t t
n n
n
n n
s n
n s
n n n s
+
( ) ( )
( ) ( ) ( ) [ ]
1
sin , sin
sin sin
,


12/4/01 USPAS Lecture 10 7
in which V
n
is the effective accelerating voltage at cavity n.
Note that, strictly speaking, for rf cavities, this should be a
difference equation, not a differential equation. However, well be
focusing on cases in which the energy change per cavity is a small
fraction of the energy, so the use of a differential is appropriate.
How does the time t
n
change from cavity to cavity?
The change in the transit time from one cavity to the next is due to
the change in energy that has occurred as a result of the
acceleration in the cavity.
Let T
n
be the transit time from cavity n to cavity n+1.
Then we have
12/4/01 USPAS Lecture 10 8
t t T T
t t
dt
dn
T T
n n n s n
n n
n
n s n
+
+
+

1
1
,
,
t
Cavity n
t
T
T
s,n
n
Synchronous particle
t
t
n
s
t
t
n+1
s
Cavity n+1
Ts,n
We want to write this in terms of the small energy difference
n
,
so we Taylor expand
12/4/01 USPAS Lecture 10 9
T E T E
dT
dE
E E
T E T E
dT
dE
E
n n s n s n
E
n s n
n n s n s n
E
n
s n
s n
( ) ( )
( ) ( )
, , ,
, ,
,
,
+
( )

From Lecture 6, p. 32:
dt
t
dp
p
C
, in which, for a linac,

C

1
2
From relativistic kinematics,
dp
p
dE
E

1
2

.
Putting these together, we have
12/4/01 USPAS Lecture 10 10
dt
t
dE
E
C

2
so
dT
dE
T
E
E
s n
C
s n s n s n ,
,
, ,

2
The transit time for the synchronous particle is
T
L
c
h
c
h
c
s
s
s
s

in which the n subscript is understood. Then

dt
dn
T T
dT
dE
E
h
E c
E
n
n s
E
n
C
s s
n
s

2
The two differential equations that govern the longitudinal
dynamics are then
12/4/01 USPAS Lecture 10 11
dt
dn
h
E c
E
d
dn
E eV t t
n C
s s
n
n n s

( ) ( ) ( ) [ ]

2

sin sin
One second-order equation can be obtained by differentiating the
first equation and using the second:
d t
dn
h
E c
d
dn
E E
h
c
d
dn E
eVh
E c
t t E
h
c
d
dn E
n C
s s
n n
C
s s
C
s s
n s n
C
s s
2
2 2 2
2 2
( ) +
,

,
]
]
]
( ) ( ) [ ]
+
,

,
]
]
]

sin sin
We now assume that the energy of the synchronous particle, and
the rf wavelength, vary very slowly with n (compared to E and t),
12/4/01 USPAS Lecture 10 12
so we can ignore their derivatives. Then, we have the second order
nonlinear differential equation
d t
dn
eVh
E c
t t
n C
s s
n s
2
2 2
( ) ( ) [ ]

sin sin
Small amplitude synchrotron oscillations
Were going to start by restricting ourselves to small variations in
phase from the synchronous phase, to explore some features of this
equation.
Let t t t
n n s

If t
n
<<1, then we can expand and approximate
12/4/01 USPAS Lecture 10 13
sin sin sin sin
sin cos cos sin sin
cos



t t t t t
t t t t t
t
n s n s s
n s n s s
n s
( ) ( ) + ( ) ( ) ( )
+ ( )

in which
s s
t .
This gives us a simple linear differential equation
d
dn
t Q t
Q
eVh
E
n s n
s
C s
s s
2
2
2
2
2
2 0
2
( ) + ( )

cos
12/4/01 USPAS Lecture 10 14
This equation describes the small amplitude oscillations of a
particle about the synchronous particle, in both energy and time, as
it is accelerated in the series of rf cavities.
Q
s
is called the small amplitude synchrotron oscillation tune. It is
the number of synchrotron oscillations between rf cavities. It must
be positive for stable motion. For a linac,

C
s

1
2
and
Q
eVh
E
s
s
s s s
2
2 2
2

cos

and so

2 2
s
. We only have stable motion
for this range of synchronous phase.
Provided that Q
s
2
>0, the motion is simple harmonic:
Using
dt
dn
h
E c
E
n C
s s
n

2
, we get
12/4/01 USPAS Lecture 10 15


t t Q n E
h
E cQ
Q n
E E Q n t
E cQ
h
Q n
n s
C
s s s
s
n s
s s s
C
s
+

0 0
2
0 0
2
2
2
2
2
2
2
cos sin
cos sin

This can be written in the form of a matrix:

t
E
Q n
h
E cQ
Q n
E cQ
h
Q n Q n
t
E
n
s
C
s s s
s
s s s
C
s s
j
(
,
\
,
(

j
(
,
,
,
\
,
(
(
(
j
(
,
\
,
(
cos sin
sin cos
2
2
2
2
2 2
2
2
0

which, by analogy with the transverse case, suggests the
introduction of the longitudinal Twiss parameter
L
:

L
C
s s s s
C
s s
h
E cQ c
h
eVE

2 2
2
cos
12/4/01 USPAS Lecture 10 16
(To keep
L
positive, we need to define it in terms of
C
. For
C
<0, this requires a redefinition of E
n
to E E E
n s n n

,
).
Then the longitudinal motion is

t
E
Q n Q n
Q n Q n
t
E
n
s L s
L
s s
j
(
,
\
,
(

j
(
,
,
\
,
(
(
j
(
,
\
,
(
cos sin
sin cos
2 2
1
2 2
0

An invariant of the motion is
1
2 2

L
n L n
t E ( ) + ( ) constant =
L
in which
L
is called the longitudinal emittance.
12/4/01 USPAS Lecture 10 17
Longitudinal phase space is formed by the variables E
n
and t
n
.
In this phase space, these variables, evaluated at subsequent rf
cavities, trace out an ellipse, whose area is
L
.
E
t
E
L
L
( )
max

t
L L
( )
max

12/4/01 USPAS Lecture 10 18
The longitudinal emittance for a beam of particles is defined in the
same way as for the transverse emittance:
The rms longitudinal emittance is the area (divided by ) of the
ellipse containing 39% of the particles.
If the distribution is Gaussian, then we have

t
E
L L rms
L rms
L
( )
( )
2
2

,
,

t
E
L

The rms bunch length

of this collection of particles is given by
s c t c
s s L L rms
( ) ( )
2
2

,
12/4/01 USPAS Lecture 10 19
As in transverse phase space, the local phase space density in
longitudinal phase space is constant (Liouvilles theorem).
This theorem does not hold in the presence of particle losses,
dissipative processes (like scattering), or damping processes (like
radiation damping or cooling).
For (E, t) phase space, it does hold in the presence of
acceleration:
The longitudinal emittance
L
is an adiabatic invariant: it remains
constant even if the synchronous energy, velocity and phase
change, or if the rf voltage or frequency changes, as long as the
12/4/01 USPAS Lecture 10 20
changes are slow compared to a synchrotron oscillation period.
Thus, we have
E
mc eV
h
L
L
L s s s
( )
j
(
,
\
,
( max
cos

2 2 2 3
2
1
4
2
t
h
mc eV
L L
L
s s s
( )
j
(
,
\
,
( max
cos



2 2
2 2 3
1
4
2
So, as we accelerate the beam, or if we increase the rf voltage V,
the energy spread E V
s s
( )
( )
max

2 3
1
4
increases, but the time
spread t V
s s
( )
( )

max

2 3
1
4
decreases. This is called adiabatic
damping (in longitudinal phase space).
12/4/01 USPAS Lecture 10 21
Example: Fermilab Side-coupled linac
This machine accelerates a proton beam from 116 MeV (the output
of an Alvarez linac) to 400 MeV. There are about 450 cells
(cavities) in about 50 m, so each cell is about 0.11 m long (the cells
actually vary from about 0.08 m at the low energy end, to 0.13 m
at the high energy end).
The accelerating gradient is about 8.4 MV/m; the transit time
factor is about 0.85; so the acceleration per cell is about
V=8.4x0.85x0.11= 0.78 MV.
The rf frequency is 805 MHz, so =37.2 cm. The synchronous
phase is
s
= 58
o
.
The side-coupled cavity structure has phase
advance per cell, so h=1/2. The longitudinal emittance is
L,rms
=
6.4 eV-sec.
12/4/01 USPAS Lecture 10 22
Using these numbers, we find
Parameter 116 MeV 400 MeV Units

s
0.456 0.713

s
1.12 1.42
L 0.085 0.132 m
Q
s
0.032 0.0089
1/Q
s
30.4 112.1

L
9.9x10
-17
2.79x10
-17
s/eV

0.256 0.492 MeV

/
0.0022 0.0012

t
25 13 ps

s
3.46 2.82 mm
12/4/01 USPAS Lecture 10 23
Large amplitude synchrotron oscillations
We go back to two first-order nonlinear differential equations we
obtained on p. 11:
dt
dn
h
E c
E
d
dn
E eV t t
n C
s s
n
n n s

( ) ( ) ( ) [ ]

2

sin sin
Using the chain rule, and dropping the n subscript in what follows,
we can write
d
d
E
d
dn
E
dn
dt
dt
d
( ) ( )
12/4/01 USPAS Lecture 10 24
in which t is the phase of the particle under consideration.
Then we have
d
d
E eV
E c
h E
E
s
s s
C
s
s L

( )
[ ]

[ ]
sin sin
sin sin
cos
2
2 2
1
in which


L
C
s s s
h
E eVc
2
2
2 2
2

cos
has been used. So
Ed E d
s
s L
( )

[ ]
sin sin
cos



2 2
12/4/01 USPAS Lecture 10 25
Assuming that
s
, and
L
are approximately constant during a
synchrotron oscillation, we can integrate both sides to give
1
2
2 0 0
2 2
E
s s
s L
( )
+
[ ]
+
[ ]
cos sin cos sin
cos


in which
0
is the phase for which E=0.
This can be written as

L
L s
s
L s
s
E ( ) + ( )
+ ( )
2
0 0
2
2
cos
cos sin
cos
cos sin
12/4/01 USPAS Lecture 10 26
This equation gives the curve in (E, ) phase space corresponding
to a large amplitude synchrotron oscillation. This curve is
sometimes called a phase space trajectory.
What do these curves look like? Heres a typical one

1
0

12/4/01 USPAS Lecture 10 27

The curve crosses the line E=0 at the two points,
0
and
1
, where

1
is given in terms of
0
by the equation
cos sin cos sin
1 1 0 0
+ +
s s
These two values of are the bounds of the motion, for this
particular phase space trajectory.
Phase space trajectories corresponding to larger values of
0
are
possible, up to a maximum
0,max
To see this,
we plot cos sin +
s
(for
s
= 0.1).
12/4/01 USPAS Lecture 10 28
-6 -4 -2 2 4 6
-1
-0.5
0.5
1
1.5

cos sin +
s

1
0

1,max
0,max
cos sin
0 0
+
s
cos
sin
,max
,max

0
0
+
s

0,max
and
1,max
, correspond to the maximum extent of bounded
motion possible. For larger values of , the motion is not bounded.

0,max
occurs at a minimum in the function cos sin +
s
.
12/4/01 USPAS Lecture 10 29
By differentiating the function, we see that this occurs at

0,max

s
The other bound to the motion may be found from
cos sin cos sin
cos sin
,max ,max


1 1
+ ( ) + ( )
+ ( )
s s s s
s s s
The phase space trajectory corresponding to the maximum
bounded motion

L
L s
s
L s
s s s
E ( ) + ( )
( ) ( )
2
2
2
cos
cos sin
cos
cos sin
is called the separatrix: it separates bounded from unbounded
motion
12/4/01 USPAS Lecture 10 30
-3 -2 -1 1 2 3
-1.5
-1
-0.5
0.5
1
1.5

Separatrix
1,max

0,max

b
Bucket
The synchrotron tune decreases as the oscillation amplitude
increases; on the separatrix, the tune is zero, and the period is
infinite. The area in phase space within the separatrix is called the
bucket.
12/4/01 USPAS Lecture 10 31
The phase space area occupied by the beam (the longitudinal
emittance) must be inside the bucket (typically, well inside: it
would correspond to one of the small ellipses in the figure above.)
The height of the bucket, E
b
, determines the energy acceptance
of the accelerator. This is given by setting =
s
, and E=E
b
in the
separatrix equation: the result is
E
b
s s
L

j
(
\
,
2 1
2

tan
The bucket represents the maximum stable area in phase space.
For zero synchronous phase (no acceleration), the bucket spans
the whole range of from - to .
12/4/01 USPAS Lecture 10 32
As the synchronous phase increases, the size of the bucket shrinks,
both in phase and in energy.
RF Waveform

s
=0
=0.5
=1.25

s
s
The bucket area, the area within the separatrix (in E, t phase
space), can be found by integrating over the bucket; the result is
12/4/01 USPAS Lecture 10 33
A
f
b
L
s

( )
16
2
in which the function f
s
( )is:
0.2 0.4 0.6 0.8 1 1.2 1.4
fsHradL
0.2
0.4
0.6
0.8
1
fHfsL
12/4/01 USPAS Lecture 10 34
For good performance, the longitudinal emittance of the beam
should be much smaller than the bucket area/.
Example: Fermilab proton linac again:

s
=58
o
,
L,rms
=6.4 eV-s
Parameter 116 MeV 400 MeV Units

s
0.456 0.713

L
9.9x10
-17
2.79x10
-17
s/eV

1,max
25 25 degrees

2,max
122 122 degrees
E
b
1.3 4.8 MeV

0.256 0.492 MeV