Diapositive 1

1
Radio-Frequency Accelerators
Nuclear physics High Energy Particle
physics X-Ray and Neutron scattering for matter
studies the QUEST for higher energy higher beam
power ? maximal gradient (tunnel length) ?
maximal a.c. to beam power efficiency ?
unprecented beam quality // ? emittances,
energy spread
FOCUS on High Energy Electron RF Accelerators
only
2
What is RF acceleration ?
RF accelerating structure Converter of RF
power into E-field and hence into electron
energy Efficient acceleration if ? concentration
of E-field and ? synchronism of the e.m. wave
with the particles with a uniform waveguide vph
approches light velocity asymptotically for high
? the simplest and straigthforward methoduse
of disc-loaded structures with individual cells
coupled through the beam holes
3
TW structures
period of the structure (distance between the
irises) corresponds to 1/3 the distance the
electrons travel in one cycle at the operating
frequency 33 mm at 3 GHz or 3.3 mm at 30 GHz
E-field pattern
4
SW structures
In Standing-Wave structures superposition of
forward and backward travelling waves in a
resonant structure maximum efficiency when both
TW components contribute equally to the
acceleration this occurs when phase-shift / cell
? but group velocity (velocity of energy
propagation) approximately zero at 1st order and
field flatness very sensitive to fabrication
tolerance
5
A few figures of merit
and geometric optimisation of accelerating
structures Shunt Impedance as high as
possible(to minimize dissipated power for a
given Eacc) Surface fields as low as
possible(which limit the maximum achievable Eacc)
s.c. SW 1.3 GHz n.c. TW 11.4 GHz
r/Q 1 k?/m 10 k?/m
Q 1010 8000
r 107 M?/m 80 M?/m
Es/Eacc 2 idem
Hs/Eacc 4 mT/(MV/m) idem
6
beam quality
wakefields as small as possibleto preserve the
emittances during acceleration short-range
wakefields ? single bunch deformation
 banana effect  long-range wakefields ?
displacement of the bunch centroids tradeoff
between accelerating mode properties (small iris
for low group velocity) and dipole modes (large
iris w? a-3.5)
7
RF acceleration Limitation
Plot de Livingston inflection of the exponantial
growth in energy ? circular machines B 10
Tesla Synchrotron Radiation (e-) ? Linear
machines E 100 MV/m 50 MV/m for
SC practical limit Cost for 1 GeV
acceleration but also for MW beam power
production
ILC 0.5-1 TeV CLIC 0.5-5 TeV
8
Linear Colliders

• Parameters for the Linear Collider
• ( from the worldwide Study of the Physics
  Detectors for future Linear ee- Colliders)
• Baseline machine
• Ecm continuously adjustable from 200 500 GeV
• Luminosity and reliability to allow ?Ldt 500
  fb-1 (L 2 1034 cm-2s-1)in 4 years following
  the initial year of commissioning
• Ability to scan at any energy between 200 and 500
  GeV downtime to set up not to exceed 10 of
  actual data-taking time
• Energy stability and precision below 0.1
  machine interface must allow energy, differential
  luminosity spectrum with that precision
• Electron polarization of at least 80
• 2 intersection regions for experiments one with
  crossing angle to enable ?? collisions
• Allow calibration at the Z, but with lower
  luminosity and emittance
• ITRP chose the  cold  technology (August 2004)
• CLIC goes on the quest for multi-TeV Energy

9
Schematic layout

• High Energy
• ? High gradients
• High Luminosity
• ? small spot sizes at IP
• - Low emittances
• ? ?y a few tens of nm.rad
• - small energy spread
• ?E / E 10-3
• ? High Beam power
• P 10 MW / beam

10
Luminosity for Linear Colliders
wall-plug power constrained 100 MW
wall plug to beam power efficiency as high as
possible gt 10
energy loss by beamstrahlung constrained lt 3
for flat beam and ?y ?z
vertical emittance as low as possible 30 nm
spot size bunch population nb of bunches /
pulse rep rate enhancement factor due to
disruption
11
Linear Colliders the 3 main options

• The 3 main options which were considered
• superconducting standing-wave cavities
• operating at L-band (1.3 GHz), developed by the
  TESLA collaboration at DESY
• normal-conducting traveling-wave cavities
• operating at X-band (11.4 GHz), developed by the
  JLC/NLC collaboration at SLAC and KEK
• very high gradient normal-conducting
  traveling-wave cavities
• operating at 30 GHz, potentially capable of
  reaching beam energies up to 1.5 TeV, being
  developed by the CLIC collaboration at CERN.

12
Summary of options (ITRP)

• TESLA operation of 9-cell cavities at gradients
  in excess of 24 MV/mUsing the newly developed
  electropolishing procedure for cavity
  fabrication, a 9-cell cavity, with input coupler,
  has exceeded the 35 MV/m goal
• JLC/NLC the original 1.8 m long structures
  operated reliably at 40-45 MV/m, but excessive
  breakdown rate when pushed to higher
  gradientserosion at the irises, and evidence of
  pulsed heating at the input couplers
• Redesigns focused on shorter (60 cm) structures,
  with lower group velocity and improved coupler
  design ?reached the design unloaded gradient of
  65 MV/m, with an acceptable breakdown rate
• CLIC Gradients required for the CLIC structures
  have been achieved using irises made from
  refractory metals, but only with short (15-30
  ns) RF pulsesThe elaborate two-beam pulse
  compression and frequency multiplication scheme
  requires extensive prototyping, planned for
  execution at CERNs test facilities (CTF3)

13
LC Parameters
TESLA SRF ? long RF pulse - peak RF power
modest (240 kW) - but large cryogenic planttwo
principal advantages higher efficiency and
reduced wakefields JLC/NLC short RF pulse (400
ns) to limit losses in Cu structure - peak power
quite high (56 MW), high peak power X-band
klystrons, together with a pulse compression
scheme CLIC Very short RF pulse (65 ns) - very
high peak power (gt200 MW) - no available RF power
source at this frequency and power ? two-beam
scheme to transfer power from a low-energy, high
current counter-propagating e-beam
14
Total Project Cost
TESLA Cost estimate 500GeV LC, one ee- IP 3,136
M (no contingency, year 2000) 7000
person.years
½ TPC
15
TPC as a function of the gradient
Length Nb of modules dominated
Cryogenic load dominated
flat around 35 MV/m sensitive to Qo
16
for NC X-Band Collider

• X-band collider cost of the linac balance
  between
• the cost of the power sources (which increases
  with gradient) and
• the cost of accelerator length (which decreases
  with gradient)
• Minimum when these are roughly equal (rather
  shallow)

17

• The rf pulse length is 1370 µs and the repetition
  rate is 5 Hz. At a later stage, the machine
  energy may be upgraded to 800 GeV c.m. by raising
  the gradient to 35 MV/m.
• RF accelerator structures consist of close to
  21,000 9-cell niobium cavities operating at
  gradients of 23.8 MV/m (unloaded as well as beam
  loaded) for 500 GeV c.m. operation.

• Low RF losses in resonators (Q0 1010 ,pure Nb
  at T2K)
• High AC-to-beam efficiency
• Long pulses/many bunches with low RF peak power
• Fast intra-train orbit energy feedback
  luminosity stabilisation
• Low frequency (f1.3 GHz), small wakefields ? f 3
• ? Relaxed alignment tolerances, good beam
  stability

18
The TESLA cavities are supplied with rf power in
groups of 36 by 572 10 MW klystrons and
modulators.
19
TESLA Test Facility (DESY)
TTF 2
20
Superconducting RF Test Facility (KEK)
SMTF _at_ FERMILAB
21
CLIC (Compact Linear Collider) concept
JP Delahaye, CERN
22
JP Delahaye, CERN
23
JP Delahaye, CERN
24
30 GHz RF power generation
JP Delahaye, CERN
25
CLIC Test Facility 3
26
gradient in NC structures
in 2000 ...
Chris Adolphsen, SLAC
27
State-of-the-art of high gradient NC structures
During the past 5 years, intensive efforts to
develop accelerator RF structures that meet the
high gradient performance requirements NLC/JLC 65
MV/m pulse length 400 ns CLIC 150 MV/m pulse
length 130 ns ? 60 ns (new parameter list)
NLC structure with power couplers and HOM
couplers on each side
28
NC structures Why high frequency ?
Peak Power per unit length

• Advantages
• lower RF energy per pulse (hence fewer rf
  sources)
• higher operating gradient with reasonablestructur
  e efficiency (hence a shorter linac)
• note higher gradient at higher frequency not
  valid for F? 3 GHz
• Drawbacks
• higher transverse wakefields generated in the
  structures by the beamwhich spoils the emittance
• short-range wakefields ? Constraint on the
  structure aperture to limit ?? growth minimum
  iris radius 17 of rf wavelength
• long-range transverse wakefields also need to
  be suppressed
• ? combination of dipole mode detuning and
  damping
• NLC/JLC structures, moderate damping achieved by
  coupling the cells to four circular waveguides
  (manifolds) that run parallel to the structure
• CLIC design achieves stronger damping with 4
  terminated waveguides attached to each cell

RF to beam power efficiency
29
Long-range Wakefield suppression
NLC/GLC, SLAC/KEK 11 GHz 65 MV/m, 400 ns
CLIC, CERN 30 GHz 150 MV/m, 60 ns
30
(No Transcript)
31
Main limitation RF breakdown

• RF breakdown fast and local dissipation of
  stored energy
• Several Joules of rf energy can be absorbed in a
  single cell, and in the process,surface melting
  and evaporation occurs in an area of a few 100
  µm2
• Strong electron emission, acoustic waves, gas
  desorption, X-rays and visible light is observed
  during a breakdown event
• The majority of breakdowns are concentrated in
  areas of high surface electric fields
• In areas with high surface currents but not
  necessarily high electric fields, surface defects
  like particles, voids and contaminants have been
  found to be sources of breakdown
• The stress imposed on the copper surface by
  pulsed heating resulting from high surface
  currents alone is also believed to lead to
  breakdown. Structure designs with a pulsed
  temperature rises lt 50 K appear to be safe in
  this respect

Very fast dissipation of stored RF energy High
electric surface fields leads to explosive
electron emission High magnetic surface fields
leads to RF pulse heating
32
Simplistic RF breakdown model

• RF processing can melt field emission points ?
  higher fields ?
• However, if too much energy in the breakdown
  event
• Electrons generate plasma and melt surface with
  enough energy ?
• molten surface splatters and generates new
  field emission points

High electric surface fields
Theory of explosive electron emission developed
for DC breakdown also relevant for the rf case
too, but cannot describe all the aspects of the
rf breakdown During breakdown, local plasma is
created copper ions from melting or
vaporization processes gas ions desorpted by
particle bombardment or local heating plasma
maintained to account for the amount of charge
observed and the erosion. A plasma spot model has
been developed which can predict pulse length,
gradient and material dependencies. Alternative
to field emission as the trigger of breakdown
surface fracturing directly from the high surface
field forces.
33
crater formed after explosive electron emission
? clean and smooth surface
Steffen Döbert, SLAC/NLC
34
Surface treatment for fast processing
35
Breakdown rates Material Cu
trip rate target based on operational argument
18000 structures 2 operational overhead 10 s
trip recovery ? 1 trip in 10 hours max trip rate
allowed to guaranty 100 availability
Steffen Döbert, SLAC/NLC, 2004
breakdown rates (obtained in 2004 at SLAC) in an
ensemble of 8 structures exponential function
of the average gradient Slope of the fitted curve
1 decade in breakdown rate for 7 MV/m of
ltGgt breakdown probability also grows
exponentially with pulse length at a fixed
gradient
36
Pulse length dependence Material Cu
G ? -1/6
Dependence of the maximum gradient on the rf
pulse duration blue curve breakdown rate of
0.1/h red curve onset of saturation in the
processing curve and correspond to a
breakdown rate of a few tens per hour G scales as
? -1/6 with pulse length for a fixed trip
rate can be explained by plasma spot
multiplication model ? Potential for higher
gradient at shorter pulse
Steffen Döbert, SLAC/NLC
37
Frequency dependence Material Cu
Kilpatrick s law Es max ? f not any more valid
for F gt 3 GHz (S-band) based on rest gas ions
being accelerated and bombarding the cavity
wall same limiting surface fields found for Cu
cavities at 30 GHz by the CLIC study and at
11 GHz by the NLC group
assuming pulse length scales with the fill time
...
Steffen Döbert, SLAC/NLC, 2004
38
A total of 8 structures were operated for more
than 1500 hours at 65 MV/m with a pulse length of
400 nsand a breakdown rate below 0.1/h
39
Use of refractory materials

• How to push
• the gradient further ?
• CLIC study group strategy
• Modify the RF design to obtain lower surface
  field to accelerating field ratio Es/Ea 2
• Investigate new materials that are resistant to
  arcing
• ? high melting point and low vapor pressure

Iris damage of CLIC structure at high field
W. Wuensch, CERN
W (3400) and Mo (2600) instead Cu (1800) on
iris
40
Changed to 60 ns with new parameters
41
NC high gradient structures Summary
NLC/GLC-collaboration achieved important
milestone for future high energy physics -
demonstrated 65 MV/m at 400 ns and less than 1
trip in 10 hours - Frequency dependence of
breakdown voltage is fairly weak above X-band,
pulse length dependence seems to
dominate CLIC-collaboration studies showed that -
New materials could provide a path for future
very high gradient (gt100 MV/m) applications -
Still missing consistent breakdown theory
42
Superconducting RF

• SC cavities offer
• a surface resistance which is six orders of
  magnitude lower than normal conductors (NC)
• high efficiency, even when cooling is included
• low frequency, large aperture
• high accelerating gradients
• Theoretical limit for the TESLA shape 45-50
  MV/m

43
Progress in accelerating field
H. Padamsee
44
Lutz Lilje (DESY)
45
SC Cavity preparation

• long and complex procedure
• High purity niobium sheets of Residual
  Resistivity Ratio RRR300 are scanned by
  eddy-currents to exclude foreign material
  inclusions like tantalum and iron
• Industrial production of full nine-cell cavities
• Deep-drawing of subunits (half-cells, etc. )
  from niobium sheets
• Chemical preparation for welding, cleanroom
  preparation
• Electron-beam welding according to detailed
  specification
• 800 C stress annealing of the full cavity
  removes hydrogen from the Nb
• Option 1400 C high temperature heat
  treatment with titanium getter layer to
  increase the thermal conductivity (RRR500)
  further
• Cleanroom handling
• Chemical etching (or electropolishing) to
  remove damage layer and titanium getter layer
• High pressure water rinsing as final treatment
  to avoid particle contamination
• Option Final bake out (120C) to change
  oxygen distribution near Nb2O5/Nb interface

46
Infrastructure
Infrastructure (DESY) Scanning niobium material
for inclusion Clean closed loop chemistry (Buffer
Chemical Polishing BCP) High Pressure Rinsing,
HPR, and clean room drying Clean Room handling
and assembling (Class 10 and 100)
47
Electron Beam Welding (CERCA)
48
(No Transcript)
49

• Gradient increased to 25 MV/min the 3rd
  production seriesof cavities by 2001 (TESLA-500
  specification)
• At the same time the spreadin the performance
  became smaller
• And ... An improved surface treatmentbecame
  available Electropolishing (EP)

Lutz Lilje (DESY)
50
EP vs BCP
Improve surface quality --pioneering work done at
KEK Several single cell cavities at G gt 40
MV/m 4 nine-cell cavities at 35 MV/m, one at
40 MV/m Theoretical Limit 50 MV/m
Electropolished 1,3 GHz Elliptical Nb Cavities K.
Saito et al. KEK 1998/1999
51
Electro-Polishing (EP) provides less local
field enhancement High Pressure Rinsing more
effective Field Emission onset at higher field
Electro-Polishing Baking for 35 MV/m Low
Field Emission 800C annealing 120C, 24 h,
Baking high field Q drop cured High Pressure
Water Rinsing
52
Electropolishing delivers higher gradients
Potentially can avoid1400C treatment
EP vs standard Etching Single cell results from
mode analysis of multi-cells
Lutz Lilje (DESY)
53
How to push the gradient further ?
New Shapes for Higher Gradients Philosophy
Critical magnetic field is a brick wall near 180
mT Losses from field emission can be reduced by
HPR -gt Lower Hpk But we must raise Epk !!!
H. Padamsee (Cornell)
54
H. Padamsee (Cornell)
55
Single crystal Nb
P. Kneisels (JLAB) method to fabricate cavities
directly from Nb Ingot (cut sheets from the
"single Crystal No need for forging and rolling
of Nb
Q-slope still present in single crystal gt
Q-slope is not purely due to grain boundaries
Baking improves Q-slope
56
What about High-Tc superconductivity ?
Good candidate MgB2 (Magnesium
diboride)(Nagamatsu et al, Nature, 2001) High Tc
High critical magnetic field
400nm film on sapphire substrate Alp Findikoglu
of STC at LANL unpublished
_at_ 4,2 K 1,3 GHz Theor. limit Emax 80 MV/m but
Surface Resistance Rs still too large ... needs
more investigation using different elaboration
techniques (Hot Isostatic Press, Chemical Vapor
Deposition, etc)
57
NLC Efficiency
Simplified Layout of NLC/GLC RF System
? 9
C. Adolphsen (SLAC)
58
CLIC Efficiency
Wall Plug
30 GHz RF power
Main Beam
JP Delahaye (CERN)
59
TESLA Efficiency
R. Brinkmann (DESY)
Site power 140 MW
Sub-systems 43MW
Linac 97MW
RF 76MW
Cryogenics 21MW
Injectors
78
Damping rings
Water, ventilation,
Beam 22.6MW
65
? 20 from mains to beam
60
60
Developpement of new RF power sources Klystron
efficiency
61
Efficiency - Klystrons
multi-cathode gun
62
Efficiency - Klystrons
Exploring a Sheet Beam Klystron as an alternate
to the MBK High efficiency design using flat
beams instead of 6 beamlets Smaller with
simpler focusing, cavities, and cathodes
Intrinsically 3-D design however the tools exist
these days No experience with sheet beam tubes
The objective of an SBK (or an MBK) is to produce
very high power at a relatively low voltage, but
at high efficiency. In the MBK this is made
possible by operating several beams in parallel
in the same vacuum envelope. In an SBK, a single
beam is extended in one dimension and the current
is increased by having a wider beam, rather than
multiple beams. (Perveance per square is
low). This makes for a simpler mechanical
design. SBK design requires 3-D simulation
codes. There are no known operating or
manufactured SBKs, perhaps for that reason.
63
RF pulse compression
High peak power klystron are not available
C. Adolphsen (SLAC)
64
Efficiency RF to beam power in CLIC
JP Delahaye (CERN)
65
Future coherent light sources
The different schemes of Free Electron Lasers
"SASE" (Self Amplified Spontaneous Emission)
modulator
radiator
disp
FEL oscillator
"HGHG" (High Gain Harmonic Generation)

• low emittance e- beams (1 p mm.mrad) and small
  energy spread
• long undulators (100 m for ? 1 Å)

require

• high power at small wavelength (no mirrors)
• with 10-20 GeV linac l lt 1 nm down to 0,1 nm
• ultra short light pulse (bunch compression)

allow
66
The 4th generation light sources (in operation or
project)
Projet Labo Pays Type Mode LEL E GeV l nm Situation
TTF II DESY D SC SASE 1 6 En construction
X FEL DESY D SC SASE 25 0.1 APD, 60 financé
BESSY FEL BESSY D SC SASE 2.25 1.2 APD
SPARC-X Frascati I RT SASE 2.5 1.5 APS Prototype financé
FERMI ELETTRA I RT HGHG 3 1.2 APS
4GLS Daresbury GB SC HGHG 0.6 10 APS Prototype financé
ARC-EN-CIEL France F SC HGHG 0.7 0.8 APS
LCLS SLAC USA RT SASE 14 0.15 APD, financé
CHESS Cornell USA SC SASE 5 100 APS APD prototype
IRFEL JLAB USA SC Osc/ERL 0.2 10 En opération
LUX Berkeley USA SC HGHG 2.5 1.2 APD
MIT Bates USA SC HGHG 4 0.3 APS
SCSS KEK J RT SASE 1 3.6 En construction
JAERI FEL Tokai J SC Osc/ERL 0.017 5 Fonctionne
KAERI Corée K SC Osc/ERL 0.04 10 En construction
67
106 by FEL gain
103 by e- quality long undulators
Sept . 2001
Sept . 2000
30 nm (Jan. 2005)
68
LCLS X-FEL based on 1 km of existing SLAC Linac
1.5-15 Å
69
SCSS SPring-8 Compact SASE Source
soft X-ray SASE-FEL machine aiming at
demonstrating FEL operation below 10 nm
wavelength with 1 GeV electron beam in 20062007
combination of the short period in-vacuum type
undulator and the high gradient C-band main
accelerator ? compact machine (100 m long
tunnel) FEL Saturation within 22.5 m long
undulator line ? requires high quality electron
beam r.m.s normalized emittance lt 2 p mm-mrad
with 2 kA peak current HV pulse gun using
thermionic cathode (CeB6 single crystal)
subharmonic buncher system
70
TTF2 Tesla Test Facility (DESY)

• 1993 CDR TTF1, "TESLA" technology development
• 2000 240 MeV beam, SASE at 120-80 nm with
  saturation
• 2004 TTF2 commissioning in progress with 5
  cryomodules
• 2005 SASE 10 nm, first  user  experiments

first lasing 30 nm Jan. 2005 450 MeV high FEL
gain 107
1 GeV SC pulsed Linac
71
TESLA X-FEL (DESY)

• Linac 1.6 km, 20 GeV, 23 MV/m
• 936 SC cavities, in 78 modules of 12 cavities
• TESLA technology (chosen by ITRP -August 2004-
  for the future ILC)
• l 0,1 - 10 nm (1st harmonic 0.05 nm)

Industrial mass production of cavities ( 1000)
and modules (gt 120)
72
French ARC-EN-CIEL proposal
Energy 700 MeV Injection 10 MeV Charge/bunch
1 nC Emittance 2 p.mm.mrad
Soft X-ray tunable coherent source(up to 1 keV )
120-10 nm
20 nm
TiSa
10 Hz - 10 kHz

• FEL oscillator (120-10 nm), SASE (200-7 nm),
  HGHG (100-0,8 nm)
• Seeding with high harmonics generated in gases
• additional Loops for energy recovery (ERL)
• Other uses plasma acceleration studies,
  Thomson scattering

73
cal for the best quality e- beams ex. emittance
sensitivity on LCLS saturation
eN 1.2 mm
P P0
eN 2.0 mm
P P0/100
74
RD on low-emittance Sources Fast Acceleration to
avoid space charge dilution
Ipk ? 50-100 A Q ? 1 nC eN ? 2 mm

• Ex. TTF2
• CsTe photocathode
• RF coaxial coupler
• Cathode Gradient 40 MV/m ? 60
• maximum dissipation lt 30 kW
• Emittance _at_1 nC lt 2 p mm.mrad

75
The Motivation for an ERL Light Source

• X-ray Experimenters Needs
• Higher brillianceallows one to work with
  smaller samples
• Higher coherent fluxallows one to capitalize on
  interference effects
• Shorter duration pulsesallows one to conduct
  pump-probe experiments
• These needs translate into a requirement for high
  average current electron beams with much smaller
  emittances and much shorter bunch lengths

Present Statususe a pinuse a pin-hole to select
a coherent x-ray beamray Future ERL sources
would change this dramatically3,000 fold
increase in hard x-ray coherent flux
C.K. Sinclair
76

• Typical ERL Light Source Parameters
• Beam Energy 5 GeV
• Fundamental frequency 1300 MHz
• Average beam current normal mode 100 mA (77
  pc/bunch)
• Average beam current short pulse mode -gt 1 mA
  ( 1 nC/bunch)
• Normalized transverse emittance at full energy lt
  below 2 mm-mrad rms
• Bunch length before compression 2 ps rms
• Bunch length after compression lt 100 fs rms
• Uncompressed ?E/E 2 x 10-4 rms

C.K. Sinclair
77
(No Transcript)
78
(No Transcript)
79
Needs very high brilliance photo-injectors
SCRF Photoinjector development (low emittance and
high beam power) BNL-AES
80
For high charge and short bunches
generation of strong HOM with large spectrum
Example for ERL at Cornell
Low number of cells per cavity
Special HOM absorbers
H Padamsee (Cornell)
81

• RF accelerators have a severe limitation in
  gradient( RF breakdowns)
• but the efforts continue ...
• CLIC has to demonstrate Ggt150 MV/m for longer
  pulse
• Plasma accelerators has potentially the
  capability to go much higher (one order of
  magnitude at least)
• For high beam power and dense beams
• Efficiency and Emittance preservation are
  key-issues