Compact FEL Based on Dielectric
Wakefield Acceleration
J.B. Rosenzweig
UCLA Dept. of Physics and Astronomy
Towards a 5th Generation Light Source
Celebration of Claudio Pellegrini
Catalina Island — October 2, 2010
FELs are Big Science
Size=$
Creating a compact FEL
 High brightness beam
 Very low charge (pC)
 Attosecond pulse
 Few 10-8 norm. emittance
J.B. Rosenzweig, et al., Nucl. Instruments Methods A, 593, 39 (2008)
 High field, short l
F.H. O’Shea et al, PRSTAB
undulator 13, 070702 (2010)
Hybrid cryo-undulator: Pr-based,
SmCo sheath 9 mm l, up to 2.2 T
 With HBB, large r, short Lg
 Lowers e- energy needed
 2 GeV hard X-ray FEL
FEL w/1 pC driver at 2.1 GeV
Scaling the accelerator in size
 Lasers produce copious power (~J, >TW)
 Scale in l by 4 orders of magnitude
 challenges in beam dynamics
 Reinvent resonant structure using dielectric
 GV/m fields possible, breakdown limited…
Resonant dielectric
laser-excited structure
(with HFSS simulated fields)
 GV/m allows major reduction in size, cost of FEL, LC
 To jump to GV/m, mm-THz may be better:
 Beam dynamics, breakdown scaling
 Need new power source…
New paradigm for high
field acceleration:
wakefields
Wakefields in dielectric tube
 Coherent radiation from bunched, v~c, e- beam
 Any slow-wave environment
 Powers exotic schemes: plasma, dielectrics
 Resonant or non-resonant (short pulse) operation
 THz regime easily w/in reach
 High average power beams can be produced
 Tens of MW, beats lasers… good for FEL, LC
 Intense beams needed, synergy with many fields
 X-ray FEL, ICS X-ray source, intense THz sources
Schematic of wakefield-based collider
J. Rosenzweig, et al., Nucl. Instrum.
Methods A 410 532 (1998).
(concept borrowed from W. Gai…)
• Similar to original CLIC scheme
• Study for plasma wakefield accelerator
• gg due to charge asymmetry in PWFA
• Not a problem for DWA…
• FEL can avoid all this complexity, use one module
The dielectric wakefield accelerator
Cerenkov scaling
dU
2
 n  1kdk k max
dz
Coherent Cerenkov scaling
dU
2
2

N
k

N
/

b max
b
z

dz
 High accelerating gradients: GV/m level
 Dielectric based, low loss, short pulse
 Higher gradients than
optical possible
 Unlike plasma, no charged particles in beam path…
 Use wakefield collider schemes
 CLIC style modular system
 Afterburner (energy multiplier) possible for existing linacs
 Spin-offs
 High power CCR THz source
Dielectric Wakefield Accelerator
Heuristic View
 Electron bunch ( ≈ 1) drives wake in
*

cylindrical dielectric structure
Dependent on structure properties
 Generally multi-mode excitation
 Wakefields accelerate trailing bunch
Design Parameters
a,b
z
 Mode wavelengths (quasi-optical)
4 b  a
l 
 1

n
n
 Peak decelerating field
eE z,dec




Extremely good
beam needed
Transformer ratio (unshaped beam)

Ez on-axis, OOPIC
4N b re mec 2

 8

a
 z  a
  1

R
E z,acc
2
E z,dec
T-481: Test-beam exploration
of breakdown threshold
 1st ultra-short, high charge beams
 Beyond pioneering work at ANL…
 Much shorter pulses, small radial size
 Higher gradients…
 Leverage off E167 PWFA
 48 hr FFTB run
 Excellent beam 3 nC, z ≥ 20 m, 28.5 GeV
 Goal: breakdown studies
 Al-clad fused SiO2 fibers
 ID 100/200 m, OD 325 m, L=1 cm
 Avalanche v. tunneling ionization studies
 Prediction: beam can excite Ez ≤12GV/m
T-481 “octopus” chamber
Beam Observations, Analysis
longer
bunch
ultrashort
bunch
Post mortem images
View end of dielectric tube;
frames sorted by increasing peak current
Breakdown determined
by benchmarked
OOPIC simulations
Breakdown limit:
5.5 GV/m decel. Field
(10 GV/m accel.?)
Multi-mode excitation – 100 fs, pulses separated by ps
— gives better breakdown dynamics?
E169 Collaboration
UCLA
H. Badakov, M. Berry, I. Blumenfeld, A. Cook, F.-J. Decker,
M. Hogan, R. Ischebeck, R. Iverson, A. Kanareykin, N. Kirby,
P. Mugglig, J.B. Rosenzweig, R. Siemann, M.C. Thompson,
R. Tikhoplav, G. Travish, R. Yoderz, D. Walz
Department
of Physics and Astronomy, University of California, Los Angeles
Stanford Linear Accelerator Center
gUniversity of Southern California
Lawrence Livermore National Laboratory
zManhattanville College
Euclid TechLabs, LLC
Collaboration spokespersons
E169 at FACET: overview
 Research GV/m acceleration scheme in DWA
 Goals


Already explored

At UCLA, BNL

Explore breakdown issues in detail
Determine usable field envelope
Coherent Cerenkov radiation measurements
Varying tube dimensions
CVD deposited
diamond
 Impedance, group velocity dependences
 Explore alternate materials
 Explore alternate designs and cladding
 Slab structure (permits higher Q, low wakes)
 Radial and longitudinal periodicity…
 Observe acceleration
Bragg fiber
 Awaits FACET construction
 Reapproval recently submitted
 Add AWA group to collaboration
Slab dielectric structure (like optical)
Observation of THz Coherent
Cerenkov Wakefields @ Neptune
 Chicane-compressed (200 m)
0.3 nC beam
 Focused with PMQ array to r~100
m (a=250 m)
 Single mode operation
 Two tubes, different b, THz
frequencies
 Extremely narrow line width in THz
 Higher power, lower width than THz FEL
Transverse wakes and slabsymmetric structures
 Transverse wakes at FACET
 Observable BBU with >10 cm
Simulated BBU @ FACET,
Initial, 10.7 cm distribution
(courtesy AWA group)
 Slab symmetric structures:
why?
 Can accelerate more charge
 Mitigate transverse wakes
4 GV/m simulated wakes for
FACET experiment
E-169 at FACET: Acceleration
 Observe acceleration
 10-33 cm tube length
 longer bunch, acceleration of tail
 “moderate” gradient, 1-3 GV/m
 single mode operation
 Phase 3: Accelerated beam quality
 Witness beam
Alignment, transverse wakes, BBU
z
r
Eb
Q
50-150 m
< 10 m
25 GeV
3 - 5 nC
FACET beam parameters for
E169: acceleration case
 Group velocity & EM exposure t  L /(c  vg )
 Positrons. Breakdown is different?
Longitudinal E-field

Momentum
distribution
after 33and
cm BBU
(OOPIC)
Witness beam,
acceleration
A High Transformer Scenario
using Dielectric Wakes
 How to reach high energy
with DWAs?
 Enhanced transformer ratio
with ramped beam
 Does this work with multimode DWA?
 Scenario: 500-1000 MeV
ramped driver; 5-10 GeV Xray FEL injector in <10 m
Symmetric beam R<2
Ramped beam R>>2
A FACET test for light
source scenario
 Beam parameters: Q=3 nC,
ramp L=2.5 mm,U=1 GeV
 Structure: a=100 m,
b=100 m, =3.8
 Fundamental f=0.74 THz
Ramped beam using sextupole-corrected
 Performance: Ez>GV/m, R=9Longitudinal
phase space
dogleg compression
after 1.3 m DWA (OOPIC)
10 (10 GeV beam)
R. J. England,
J. B. Rosenzweig,
and G. Travish,
Longitudinal
wakefields
PRL 100, 214802 (2008)
 Ramp achieved at UCLA.
Possible at ATF, FACET?
Multipulse operation: control
of group velocity
 Multiple pulse beam-loaded
operation in linear collider
 Needs low vg
N
Accelerating beam
Driving beam

 Low Q,  beams shorter,
smaller
 Can even replace large Q driver
 Use periodic DWA structure in
~-mode, resonant excitation
Example: SiO2/diamond structure
Standing wave wakes in
periodic dielectric structures
 4 pulse train excitation, 2-l separation
 Rms pulse length l/4, suppresses HOM
Initial multi-pulse experiment:
uniform SiO2 DWA at BNL ATF
 Exploit Muggli pulse train slicing technique
 400 m spacing, micro-Q=25 pC, z=80 m
 DWA dimensions: a=100 m, b=150 m
BNL multi-pulse experiments
 Array of 1 cm tubes
 Si02, also diamond
 325-660 m l
4-drive + witness in spectrometer
 Large aperture 490 mm
case first
 Use PMQs later…
 Operation of pulse train
with both chirp signs
 Sextupole correction used
 CTR autocorrelation
CTR autocorrelation and FFT
Recent results from BNL multipulse experiments
 Single, multi- bunch wakes
observed
 Wakes without mask give
fundamental resonant l
 ~490 m, per prediction
 Resonant wake excitation,
CCR spectrum measured
CCR autocorrelation
2nd harmonic
1st deflecting mode
 Excited with 190 m spacing
Fundamental
nd
(@noise level)
(2 harmonic)
 Misalignments yield l~300 m,
1st deflecting mode
Frequency spectrum
Towards GV/m: multiple pulse
DWA experiment at SPARC/X
 Uses laser comb technique
 Bunch periodicity:
190m (0.63 ps)
 0.5 of BNL case
 Scaled structure
 125 pC/pulse @ 750 MeV
 4 pulses + witness
 1 GV/m, energy doubling
in <70 cm
>1.1 GV/m wakes in
scaled DWA@SPARX
Honey, I shrunk the FEL
th
(not quite yet…CP’s 80 )
 FEL itself gets small with small Q, high
brightness beams; innovative undulators
 Lower energy needed
 Ultimate limit in optical undulators?
 Wakefields give very high field
 DWA gives a credible path
 Booster for hard X-ray FEL in few m
 Scaling to low Q synergistic, hard
TV/m simulated PWFA
using LCLS 20 pC beam
 Expect rapid experimental progress
 1st ATF; then FACET, SPARC/X, etc.