An Introduction to the

Physics and Technology

of e+e- Linear Colliders

Lecture 1: Introduction and Overview

Nick Walker (DESY)

Nick Walker
DESY
DESY Summer Student Lecture
USPAS Santa Barbara 16 June,st2003
th
31 July 2002
 Course Content
Lecture:

1. Introduction and overview

2. Linac part I
3. Linac part II
4. Damping Ring & Bunch Compressor I
5. Damping Ring & Bunch Compressor II
6. Final Focus Systems
7. Beam-Beam Effects
8. Stability Issues in Linear Colliders
9. the SLC experience and the Current LC Designs
 This Lecture

• Why LC and not super-LEP?

• The Luminosity Problem
– general scaling laws for linear colliders
• A introduction to the linear collider sub-systems:
– main accelerator (linac)
– sources
– damping rings
– bunch compression
– final focus

during the lecture, we will introduce (revise) some important basic

accelerator physics concepts that we will need in the remainder of
the course.
Energy Frontier e+e- Colliders

LEP at CERN, CH
Ecm = 180 GeV
PRF = 30 MW
 Why a Linear Collider?

Synchrotron Radiation from e 2c 2

an electron in a magnetic field: P  C E 2 B 2
2
B
Energy loss per turn of a
machine with an average
bending radius :

4
C E
E / rev 


Energy loss must be replaced by RF system

Cost Scaling $$

• Linear Costs: (tunnel, magnets etc)

$lin  
• RF costs:
$RF  E E4/
• Optimum at
$lin = $RF

Thus optimised cost ($lin+$RF) scales as E2

The Bottom Line $$$

LEP-II Super-LEP Hyper-

LEP
Ecm GeV 180 500 2000

L km 27
E GeV 1.5

$tot 109 SF 2
The Bottom Line $$$

LEP-II Super-LEP Hyper-

LEP
Ecm GeV 180 500 2000

L km 27 200
E GeV 1.5 12

$tot 109 SF 2 15
The Bottom Line $$$

LEP-II Super-LEP Hyper-

LEP
Ecm GeV 180 500 2000

L km 27 200 3200
E GeV 1.5 12 240

$tot 109 SF 2 15 240

 solution: Linear Collider
No Bends, but lots of RF!

e+ e-

5-10 km

• long linac constructed of many RF

accelerating structures
• typical gradients range from 25100 MV/m

Note: for LC, $tot  E

 A Little History

A Possible Apparatus for Electron-Clashing Experiments (*).

M. Tigner
Laboratory of Nuclear Studies. Cornell University - Ithaca, N.Y.

M. Tigner,
Nuovo Cimento 37 (1965) 1228

“While the storage ring concept for providing clashing-

beam experiments (1) is very elegant in concept it seems
worth-while at the present juncture to investigate other
methods which, while less elegant and superficially more
complex may prove more tractable.”
 A Little History (1988-2003)

• SLC (SLAC, 1988-98)

• NLCTA (SLAC, 1997-)
• Over 14 Years of
TTF (DESY, 1994-)
Linear Collider
• ATF (KEK, 1997-) R&D
• FFTB (SLAC, 1992-1997)
• SBTF (DESY, 1994-1998)
• CLIC CTF1,2,3 (CERN, 1994-)
Past and Future

SLC LC
Ecm 100 500 1000 GeV
Pbeam 0.04 5 20 MW
*y 500 (50) 1 5 nm
E/Ebs 0.03 3 10 %
L 0.0003 ~3 1034 cm?2s-1

generally quoted as but we have a very

‘proof of principle’ long way to go!
 LC Status in 1994
1994 Ecm=500 GeV

TESLA SBLC JLC-S JLC-C JLC-X NLC VLEPP CLIC

f
[GHz]
1.3 3.0 2.8 5.7 11.4 11.4 14.0 30.0

L1033
6 4 4 9 5 7 9 1-5
[cm-2s-1]

Pbeam
16.5 7.3 1.3 4.3 3.2 4.2 2.4 ~1-4
[MW]

PAC
164 139 118 209 114 103 57 100
[MW]

y
100 50 4.8 4.8 4.8 5 7.5 15
[10-8m]

y*
64 28 3 3 3 3.2 4 7.4
[nm]
 LC Status 2003
2003 Ecm=500 GeV

TESLA SBLC JLC-S JLC-C JLC-X/NLC VLEPP CLIC

f
[GHz]
1.3 5.7 11.4 30.0

L1033
34 14 20 21
[cm-2s-1]

Pbeam
11.3 5.8 6.9 4.9
[MW]

PAC
140 233 195 175
[MW]

y
3 4 4 1
[10-8m]

y*
5 4 3 1.2
[nm]
The Luminosity Issue
2
Collider luminosity (cm s ) is
2 1 nb N f rep
approximately given by L HD
A
where:
Nb = bunches / train
N = particles per bunch
frep = repetition frequency
A = beam cross-section at IP
HD = beam-beam enhancement factor

nb N 2 f rep
L HD
For Gaussian beam distribution: 4  x y
The Luminosity Issue: RF Power

Introduce the centre of mass L

 E n Nf rep  N
cm b
HD
energy, Ecm: 4  x y Ecm

nb Nf repEcm Pbeams
 RF  beam PRF

RF is RF to beam power

efficiency.
 RF PRF N
Luminosity is proportional L HD
to the RF power for a given 4  x y Ecm
Ecm
The Luminosity Issue: RF Power
 RF PRF N
Some numbers: L HD
Ecm = 500 GeV
4  x y Ecm
N = 1010 Pbeams = 8 MW
nb = 100
frep = 100 Hz
Need to include efficiencies:
RFbeam: range 20-60% linac technology choice
Wall plug RF: range 28-40%

AC power > 100 MW just to accelerate beams and achieve

luminosity
The Luminosity Issues: storage ring vs LC

LEP frep = 44 kHz  RF PRF N

L HD
LC frep = few-100 Hz 4  x y Ecm
(power limited)
 factor ~400 in L already lost!

Must push very hard on beam cross-section at collision:

LEP: xy  1306 m2
LC: xy  (200-500)(3-5) nm2

factor of 106 gain!

Needed to obtain high luminosity of a few 1034 cm-2s-1
The Luminosity Issue: intense beams at IP

1  N 
L RF PRF  HD 

4 Ecm  
 x y 

choice of linac technology: Beam-Beam effects:

• efficiency • beamstrahlung
• available power • disruption
Strong focusing
• optical aberrations
• stability issues and
tolerances
 see lecture 2 on
The Luminosity Issue: Beam-Beam beam-beam

• strong mutual focusing of

beams (pinch) gives rise
to luminosity 3000

enhancement HD 2000
 x  y
• As e± pass through intense 1000

Ey (MV/cm)
field of opposing beam,
0
they radiate hard photons
[beamstrahlung] and loose 1000

energy 2000

• Interaction of 3000
beamstrahlung photons 6 4 2 0 2 4 6

y/y
with intense field causes
copious e+e pair
production [background]
 see lecture 2 on
The Luminosity Issue: Beam-Beam beam-beam

beam-beam characterised by Disruption

Parameter: 2re N z z
Dx , y  
 x , y  x   y  fbeam
z = bunch length,
fbeam = focal length of beam-lens

for storage rings, f beam  z and Dx, y 1

In a LC, Dy 10  20 hence f beam  z

Enhancement factor (typically HD ~ 2):

 D 3
  0.8  x , y 
H Dx , y 1/ 4
1  Dx , y 
x, y
 1  D 3  
x, y  
ln  Dx , y 
 1  2 ln 
 z




‘hour glass’ effect

 see lecture 2 on
The Luminosity Issue: Hour-Glass beam-beam

3 3

2 2

1 1

0 0
Y

Y
1 1

2 2

3 3
2 1 0 1 2 2 1 0 1 2
Z Z

 = “depth of focus”
reasonable lower limit for
 is bunch length z
 see lecture 2 on
The Luminosity Issue: Beamstrahlung beam-beam

ere3  Ecm  N2
RMS relative energy loss  BS 0.86  
2m0 c 2   z  ( x   y ) 2

we would like to make xy small to maximise luminosity

BUT keep (x+y) large to reduce SB.
 Ecm  N 2
Trick: use “flat beams” with  x  y  BS   2
 z  x


Now we set x to fix SB, and make y as small as possible to

achieve high luminosity.
For most LC designs, SB ~ 3-10%
 The Luminosity Issue: Beamstrahlung
Returning to our L scaling law, and ignoring HD

 RF PRF  N  1
L  
Ecm   x   y

N  z BS
From flat-beam beamstrahlung 
x Ecm

 RF PRF  BS z
hence L
Ecm3/ 2 y
The Luminosity Issue: story so far

 RF PRF  BS z
L
Ecm3/ 2 y
For high Luminosity we need:
• high RF-beam conversion efficiency RF
• high RF power PRF
• small vertical beam size y
• large bunch length z (will come back to this one)
• could also allow higher beamstrahlung BS if willing to live with
the consequences

Next question: how to make a small y

 The Luminosity Issue: A final scaling law?

 RF PRF  BS z  y n , y
L y 
E 3/ 2
cm y 
where n,y is the normalised vertical emittance, and y is the vertical
-function at the IP. Substituting:

 RF PRF  BS   z RF PRF  BS  z

L 
Ecm3/ 2  n, y  y Ecm  n, y  y
hour glass constraint
y is the same ‘depth of focus’  for hour-glass effect. Hence  y  z
 The Luminosity Issue: A final scaling law?

 RF PRF  BS  y  z
L HD
Ecm  n, y

• high RF-beam conversion efficiency RF

• high RF power PRF
• small normalised vertical emittance n,y
• strong focusing at IP (small y and hence small z)
• could also allow higher beamstrahlung BS if willing to
live with the consequences
Above result is for the low beamstrahlung regime where BS ~ few %
Slightly different result for high beamstrahlung regime
Luminosity as a function of y

L (cm 2s 1 )
51034

 z 100 m  BS  1
z
41034

31034  z 300 m

21034 500 m nb N 2 f
L
4 x y
700 m
11034 900 m

200 400 600 800 1000

 y ( m)
 The ‘Generic’ Linear Collider

pre-accelerator
few GeV
source
KeV
damping extraction
ring few GeV 250-500 GeV & dump
few GeV final focus

bunch IP
main linac
compressor collimation

Each sub-system pushes the state-of-the-art in accelerator design

 see lectures 3-4
The Linear Accelerator (LINAC) on linac

Ez
c
travelling wave structure:
need phase velocity = c
z (disk-loaded structure)
bunch sees constant
field:
Ez=E0 cos()
ct 
Ez 2
c c
standing wave cavity:
bunch sees field:
z Ez =E0 sin(t+)sin(kz)
=E0 sin(kz+)sin(kz)
 see lectures 3-4
The Linear Accelerator (LINAC) on linac

Travelling wave
structure

Circular waveguide
mode TM01 has vp>c

No good for
acceleration!

Need to slow wave

down using irises.
 see lectures 3-4
The Linear Accelerator (LINAC) on linac

• Gradient given by shunt impedance:

Ez  PRF Rs
– PRF RF power /unit length
– RS shunt impedance /unit length
 2Q /  SW
• The cavity Q defines the fill time: t fill 
2 Q/ ls / vg TW
– vg = group velocity, ls = structure length
• For TW,  is the structure PRF ,out PRF ,in e  2
attenuation constant:
• RF power lost along structure (TW):
2 RF
dPRF E
  ib Ez
z

dz Rs would like RS to be
as high as possible
power lost to structure beam loading Rs  
 see lectures 3-4
The Linear Accelerator (LINAC) on linac

• Steady state gradient drops over length of

structure due to beam loading

unloaded
av. loaded

  2 0 
2  e 1
Ez,l 1 ibrs  0  2
 
Ez,u  
2  1 e 0 
 

assumes constant (stead state) current

 see lectures 3-4
The Linear Accelerator (LINAC) on linac

• Transient beam loading

– current not constant but pulses! (tpulse = nb tb)
– for all LC designs, long bunch trains achieve steady
state quickly, and previous results very good
approximation.
– However, transient over first bunches needs to be
compensated.
V
unloaded

av. loaded

t
The Linear Accelerator (LINAC)
Single bunch beam loading: the Longitudinal wakefield

NLC X-band structure: Ez bunch

700 kV/m
The Linear Accelerator (LINAC)
Single bunch beam loading Compensation using RF phase

wakefield

RF
 = 15.5º

Total
The Linear Accelerator (LINAC)
Single bunch beam loading: compensation

RMS E/E

Ez

min = 15.5º
Transverse Wakes: The Emittance Killer!
Dtb

V ( , t ) I ( , t ) Z ( , t )
Bunch current also generates transverse deflecting modes
when bunches are not on cavity axis
Fields build up resonantly: latter bunches are kicked
transversely
 multi- and single-bunch beam breakup (MBBU, SBBU)
Damped & Detuned Structures

NLC RDDS1
bunch spacing

2QHOM
t 


Slight random detuning between cells causes HOMs to decohere.

Will recohere later: needs to be damped (HOM dampers)
Single bunch wakefields

Effect of coherent betatron oscillation

- head resonantly drives the tail
head eom:
d 2 yh
 k 2 y y 0
ds
tail tail eom:
head
d 2 yt
 k 2 yt k wf yh
ds
 Wakefields (alignment tolerances)
cavities
tail performs
bunch oscillation

accelerator axis tail

tail head Dy head
head
tail

0 km 5 km 10 km

higher frequency = stronger wakefields

 YRMS  1 E Z

NW  -higher gradients

f 3 E
 z
-stronger focusing (smaller )
N 
-smaller bunch charge
The LINAC is only one part
pre-accelerator
few GeV
source
KeV
damping extraction
ring few GeV 250-500 GeV & dump
few GeV final focus

bunch IP
main linac
compressor collimation

• Produce the electron charge?

Need to • Produce the positron charge?
understand how • Make small emittance beams?
to:
• Focus the beams down to ~nm at the
IP?
e+e Sources

Requirements:

• produce long bunch trains of 100-1000s @ 5-100 Hz

high charge bunches few nC
• with small emittances nx,y ~ 106,108 m
• and spin polarisation mandatory for e,
(needed for physics) nice for e+

nb N 2
Remember L scaling: L 
n
e Source
• laser-driven photo injector  = 840 nm
s
to n
• circ. polarised photons on l a se
r ph
o

GaAs cathode electrons

G aAs
 long. polarised e 20 m m cathode

• laser pulse modulated to

give required time
structure 120 kV

• very high vacuum

requirements for GaAs
n 10 5 m
(<1011 mbar)
• beam quality is dominated factor 10 in x plane
by space charge
factor ~500 in y plane
(note v ~ 0.2c)
e Source: pre-acceleration

E = 12 M eV E = 76 M eV

K K K
SHB
to D R in e cto r lin ac

solenoids
laser

SHB = sub-harmonic buncher. Typical bunch length from

gun is ~ns (too long for electron linac with f ~ 1-3 GHz,
need tens of ps)
 e Source
e
Photon conversion to e± 
pairs in target material

e
e
Standard method is e e
beam on ‘thick’ target e

(em-shower) ei
e
 e Source :undulator-based
250GeV e- to IP

e+e- pairs
NS N S NS NS N S
from
e- linac
S N S N S N S N S N ~30MeV photons
undulator (~100m)

0.4X target

• SR radiation from undulator generates photons ~ 30 MeV

• no need for ‘thick’ target to generate shower 0.4X0
• thin target reduces multiple-Coulomb scattering: hence
102 m
better emittance (but still much bigger than needed)
• less power deposited in target (no need for mult. systems) 5 kW
• Achilles heel: needs initial electron energy > 150 GeV!
 see lecture 5
Damping Rings

• (storage) ring in which the bunch train is stored for

Tstore ~20-200 ms
• emittances are reduced via the interplay of synchrotron
radiation and RF acceleration

initial emittance
(~0.01m for e+)

 2T /  D
 f  eq  ( i   eq )e
final emittance damping time
equilibrium
emittance
 see lecture 5
Damping Rings: transverse damping
p replaced by RF such that pz = p.
y’ not changed by
photon (or is it?) since (adiabatic damping again)
g y’ = dy/ds = py/pz,
dipole RF cavity
we have a reduction in amplitude:
y’ = p y’
dp
dp

Must take average over all -phases:

2E c C E 4 2
D  where P  and hence D 
P 2  2 E3

LEP: E ~ 90 GeV, P ~ 15000 GeV/s, D ~ 12 ms

 see lecture 5
Damping Rings: Anti-Damping

E u
1 
ecB
E
0 
ecB
u
a  r 
ecB

particle now performs -oscillation about d x

new closed orbit 1  increase in emittance Q
dt
d x 2
0 Q  x
Equilibrium achieved when dt d
 see lecture 5
Damping Rings: transverse damping
2
D  3
suggests high-energy and small ring. But
E
E4 Remember: 8D
required RF power: PRF  nb N needed to reduce e+
 2
vertical emittance.
E2 Store time set by frep:
equilibrium emittance:  n, x 
 ts ntrain / f rep
an example:
• Take E  2 GeV radius:
ntrainnbtbc
• Bbend = 0.13 T    50 m 
2
• <P> = 27 GeV/s [28 kV/turn]
• hence D  148 ms - Few ms required!!!
Increase <P> by 30 using wiggler magnets
 see lecture 5
Damping Rings: limits on vertical emittance

• Horizontal emittance defined by lattice

• theoretical vertical emittance limited by
– space charge
– intra-beam scattering (IBS)
– photon opening angle

• In practice, y limited by magnet alignment errors

[cross plane coupling]
• typical vertical alignment tolerance: y  30 m
 requires beam-based alignment techniques!
 see lecture 6
Bunch Compression
• bunch length from ring ~ few mm
• required at IP 100-300 m

E /E E /E E /E E /E E /E

long.
phase z z z z z

space

RF dispersive section
 see lecture 6
The linear bunch compressor
initial (uncorrelated) momentum spread: u
initial bunch length z,0
compression ration Fc=z,0/z
beam energy E
RF induced (correlated) momentum spread: c
RF voltage VRF
RF wavelength RF = 2/ kRF
longitudinal dispersion: R56

conservation of longitudinal  c2   u2
Fc    c  u Fc2  1
emittance u

k RFVRF  z ,0 E c E  u 
RF cavity  c 
2
 VRF     c  1
F
E k RF  z ,0 k RF   z ,0 
 see lecture 6
The linear bunch compressor
chicane (dispersive section)
2
z  c z ,0 k RFVRF   z ,0  1
z R56 R56  2  2 2    2
 F u E  u  F

 z ,0 2 mm
 u 0.1%  2%
 z 100 m  Fc 20 VRF 318 MV
f RF 3 GHz k RF 62.8 m  1 R56 0.1m
E 2 GeV
 Final Focusing
final
doublet
(FD) IP

f1 f2 f2
f1 f2 (=L*)

Use telescope optics to demagnify beam by factor m = f1/f2= f1/L*

Need typically m = 300

putting L* = 2m  f1 = 600m
 see lecture 7
Final Focusing
L* 2  4 m
 y   n, y  y / 
 y 2  5nm   y 100  300 μm
remember y ~ z

f L *
at final lens y ~ 100 km

short f requires very strong fields (gradient): dB/dr ~ 250 T/m

pole tip field B(r = 1cm) ~ 2.5 T

normalised quadrupole strength: K1  1 B

r
o
B 0

where B = magnetic rigidity = P/e ~ 3.3356 P [GeV/c]

 see lecture 7
Final Focusing: chromaticity
1
for a thin-lens of length l: K1l
f

  K1l yquad
yquad  K1l yquad 
1

yIP  f yquad  yquad 
f L * yIP2  yquad
2
 2  quad  y rms
2

for rms ~ 0.3% yIP2 20  40 nm

more general: yIP2  y rms

2

 is chromaticity  y K1 ( s )  ( s)ds

chromaticity must be corrected using sextupole magnets

 see lecture 7
Final Focusing: chromatic correction
magnetic multipole expansion:
1 1 1 
By ( x) B    K1 x  K 2 x  K 3 x  
2 3

 2 3! 

dipole quadrupole sextupole octupole

l
 k1 y quadrupole kn K n ds
2 -order kick:
nd
y  0

  k2 xy sextupole

introduce horizontal x  x  Dx

dispersion Dx y  k2 xy  k2 Dx y
   
geometric chromaticity

D need also to cancel

chromatic correction when k2  x
k1 geometric (xy) term!
(second sextupole)
 see lecture 7
Final Focusing: chromatic correction

dipole Dx IP

sextupoles

m 0 0 0 
FD
 
0 1/ m 0 0 
R 
0 0 m 0  L*
 
0 0 0 1/ m 
 see lecture 7
Final Focusing: Fundamental limits

Already mentioned that  y  z

At high-energies, additional limits set by so-called Oide
Effect:
synchrotron radiation in the final focusing quadrupoles leads
to a beamsize growth at the IP
minimum beam size:  1.83 re  e F   n 7
1 5
7

independent
of E!
 2.39 re  e F   n
2 3
occurs when 7 7

F is a function of the focusing optics: typically F ~ 7

(minimum value ~0.1)
 see lecture 8
Stability
• Tiny (emittance) beams
• Tight component tolerances
– Field quality
– Alignment
• Vibration and Ground Motion issues
• Active stabilisation
• Feedback systems

Linear Collider will be “Fly By Wire”

 see lecture 8
Stability: some numbers
• Cavity alignment (RMS):~ m
• Linac magnets: 100 nm
• FFS magnets: 10-100 nm
• Final “lens”: ~ nm !!!

parallel-to-point focusing:
 see lecture 8
LINAC quadrupole stability
NQ NQ

y  kQ ,i Yi gi kQ  Yi g i

* 1
sing1e quad 100nm offset
0.5
i 1 i 1
0
i
gi  *  i  * sin(i ) 0.5

 1
0 500 1000 1500 2000
for uncorrelated offsets
1
 *
 Y 2 NQ
100nm RMS random offsets
y*2  *  i Q,i i (ij )

 k 2
 sin
i 1
2 0.5

Dividing by  y   y ,n / 
*2 * * 0.5

and taking average values: 1

0 500 1000 1500 2000

y 2j N Q kQ2  
  2Y 0.32
 *2
y 2 y ,n

take NQ = 400, y ~ 61014 m, ~ 100 m, k1 ~ 0.03 m1  ~25 nm

 see lecture 8
Beam-Beam orbit feedback

e
IP

bb
y
FDBK
kicker
BPM

e

use strong beam- Generally, orbit control

beam kick to keep (feedback) will be used
beams colliding extensively in LC
Beam based feedback: bandwidth
10
5

1
0.5 g = 0.01 g = 0.1 g = 0.5 g = 1.0

0.1
0.05

0.0001 0.001 0.01 0.1 1

f / frep
f/frep

Good rule of thumb: attenuate noise with ffrep/20

Ground motion spectra
 see lecture 8
Long Term Stability
understanding of ground motion and vibration spectrum important
1 minute 1 hour 1 day 10 days
1

0.9

0.8
beam-beam
0.7
No Feedback feedback +
relative luminosity

0.6 upstream orbit

0.5 control
0.4

0.3
beam-beam
feedback
0.2

0.1

example of slow
0
0.1 1 10 100 1000 10000 100000 1000000
diffusive ground
tim e /s
motion (ATL law)
Here Endeth the First Lecture