Laser-powered dielectric structures could provide a particle accelerator and light source on a chip within the next decade, enabling tabletop production of relativistic electron beams and x-rays. These structures allow for gigavolt-per-meter acceleration gradients using laser pulses to drive short electron bunches, with potential applications including medical and industrial imaging.

1. Laser-powered dielectric-structures
for the production of
high-brightness electron and x-ray beams
Gil Travish
Particle Beam Physics Laboratory
UCLA Department of Physics & Astronomy
on behalf of the MAP team
Material stolen from... lots of people including Chris Seers, Chris McGuinness,
Eric Colby, Joel England Charlie Brau, Jonathan Jarvis, Tomas Plettner

3. prediction
A particle accelerator “on a chip”, capable of
producing intense pulses of relativistic
electrons and x-rays will be widely available
in 10 years

4. no plasmas were harmed
in the making of this presentation

5. A laser-powered dielectric accelerator can provide
relativistic electron beams and x-rays in a chip-scale device
+ laser(s)

6. Our long term goal is to develop a mm-scale, laser-
powered, disposable, relativistic particle source
Large Application Space:
Industrial
• Petroleum Exploration
<10mm
• Non-Destructive Testing (NDT)
X-ray Photolithography
• Medical
x1000
• Cardiology
• Veterinary
• Medical Imaging
Defense
• Homeland Security & Military
>10m

7. Breakdown limits scale favorably with wavelength
and dielectric materials support high fields
GHz THz IR-VIS
10-15 fs
10-13
T-481 Eacc ~ Prf/λ
L
A ps
DWA S
10-11 E
R
Pulse Length [s]
Breakdown Limits
10 -9
Conventional Structure Du (1996) ns
~GV/m
10-7
us
10-5
Conventional RF
10-3
10-1
100 102 104 106 108 1010 1012 1014
Frequency [Hz]
in metals...

8. Of available power sources at wavelengths shorter
than microwaves, lasers are the most capable
lack of sources, materials and fabrication technology force us to
make a leap from Microwave to Optical

9. Optical-scale dielectric-structures promise GeV/m
gradients and naturally short bunches
+ very short pulses
+ very high repetition rate
+/- low charge
- no track record
- limited R&D work
! The red-headed stepchild of AA
Tolerances: Gradients x10-x100 metal
PWFA: ~300nm Structural control of fields
LWFA: ~30nm Many possible geometries
MAP: ~10nm Scalable fabrication

10. The choice of accelerator technology impacts the
possible light source configurations...
McGuinness
RF Optical
Gradient 10-100 MeV/m 1-10 GeV/m
Energy gain per period 1 MeV 1 keV
Repetition Rate 100 Hz 10-100 MHz
Charge per Bunch 0.1 - 1+ nC 0.01-1 pC
Bunch Length 1-100 ps 1-100 fs
key: charge and time scale; not gradient

11. Optical structures naturally have sub-fs time
structures and favor high rep. rate operation
Micropulse
Optical Cycles
femtosec
Laser Cycle
3.3 fs charge capture
< 1 fs
Macropulse
Laser Pulse
//
picosec
Fill Time ~ 1 ps Fill Time ~ 1-5 ps
100-1000 ns
(1-10 MHz)
Emitter Pulse
//
nanosec
Emission Time ~1 ns

12. An example of a soft x-ray FEL-based source
reveals the need for new undulator approaches
106 electrons; 108 photons Parameter Value
Wavelength 6 nm
Beam energy 25.5 MeV
Energy spread 10-4
Emittance (norm.) 0.06 µm (doh!)
Charge 1 pC (whew!)
Peak current 750 A
Undulator parameter 1
Lcoop/σL<1: 1-2 spikes Undulator period 20 µm
Focusing betafunction ~ 3 mm
Gain length 500 µm
FEL parameter ~3 x 10-3
Saturation length 6 mm (LOL)
x-ray flux per bunch ~5 x 108
Pellegrini and Travish

13. ... the undulator technology has at least as much
impact on the FEL design.
PM Micro/Pulsed RF Optical
Period >1 cm 0.1 - 1 mm 0.1-1 cm 1-20µm
Parameter 1-10 <1 ~1 ~1
Gap 5 mm 1 mm 1+ cm 20-100µm?
Status Mature some SC work stalled paper
Focusing is an addition issue:
# n 4%
!opt "3 Lg
$ &

14. Optical-scale accelerator structures

15. At SLAC, the E-163 AARD team is producing a set
of laser-driven dielectric micro-accelerators
PBG
HC-1060 Fiber
10 µm
Woodpile
4 Layer Structure
(10/08)
2 Layer Structure
(6/08)

16. PBG-fiber-based structures afford large apertures
and scalability to HEP-length structures
input port! X. E. Lin “Photonic bandgap fiber accelerator,” PRSTAB 4, 051301 (2001)
Efficient coupling to the accelerating mode of a PBG fiber is
absorbing boundary! complicated by various issues: ~2.5 GV/m
➡overmoded: coupling to other modes drains away input power
➡extra modes are lossy and difficult to simulate
➡initial simulation results from overlap with accelerating mode: ~
12%
HFSS: custom dielectric waveguide coupler

17. Planar structures offer beam dynamics advantages
as well as ease of coupling power
MAP Logpile Grating
Flat beam LS: modes? coherence? undulator?

18. The MAP structure consists of a diffractive optic
coupling structure and a partial reflector
For gap a and
dielectric b-a
idealized
resonance:
( )
cot # k z ! " 1 b " a % = k z a ! " 1 ! !
$ &

19. The design of the relativistic structure is mature
and includes realistic material properties.
laser
Ez = E0 cos(! z c) !

20. Material measurements in the Nanolab are fed
back to simulations
Uniform ZrO2 showing grain size !"#$%&'()*% +),-./01)% 2$)-3.4% 2$)-3.4% F.'(%#.?% /533)'7@%
of tens of nanometers with "'()*% /5'(6/01" )*?.'@"5' 9)G>%
optimized condition 789:;3<=>% 9ABCD;E>%
!"#$% &'((% $)% *'(% *'+% #,%
123.456%789:;7<=>?@%59:%
-.#$% $'&+/$'$&% $% &0')% *'0% 3AB59C>89%98?%D3C?%
F;4D3.C%8,%D;?%98?%
E>)FG% $'0&% )0% )')% *'0% 7844896@%;C3:%D3"8.3%
1>#$% $'+% &&'+% (% )'G% H59:%I5B%B8CC>D6@%
B.893%?8%65C3.%:545I3%
H5I;J)("63% +),-./01)% 2$)-3.4% 2$)-3.4% F.'(%#.?% /533)'7@%
"'()*;@6K@7-.7)% "'()*% /5'(6/01" )*?.'@"5 9)G>%
789:;3<=>% '%9ABCD;E>%
H59:%I5B%?88%C4566%
J6$#)% &'K)/&'++% &L% L'&% $'K%
123.456%789:;7<=>?@%
E>#$% &'GK% &'G% 0'G% L'(% 68M%
Discovery Denton Sputterer NIO$% &')+/&'G% &G/&*% L'(/&G% &0'L% P3B8C><89%"57>6>?@%
:>Q7;6?%?8%R9:%
!#'"
Measured refractive index
of ZrO2/Y2O3 deposited by
!#&" sputtering
!#%"
!"
!#!"
!#$"
!"
(#%" (#&" (#'" (#)" (#*" (#+" Gaertner Ellipsometer
#$%&"'&!()*"+,-."

21. Simulations including acceleration and beam
dynamics are underway.
Resonant Fields (@ t = 7 ps)
Input laser source
Incident laser • can correspond to actual Ti:Al2O3 laser
Ex (V/m)
Ex (V/m)
y(m)
t(s)
Ex (V/m)
x(m)
Energy Distributions
t(s)
Energy Gain

22. Prototype structures are starting to be produced.
Full scale structure DBR
Structure
Dimension:
300nmX250μmX1000
96.2nm
130.8nm
287.6nm
96.2nm
134.6nm
92.4nm

23. Integration of a full structure has been developed.
Process control improvements of fabrication is ongoing.

24. We are planning a ß=1 MAP beam de/acceleration
experiment at SLAC’s E163

25. How can we produce a low-beta structure?
at 1 GeV/m, each period only produces 1KeV
1000 periods only yields 1 MeV
1 TeV requires 1 billion periods

26. Creating a sub-relativistic MAP is hard: the
coupling and periodicity are one and the same
tapered structure two-color operation rapid change in velocity
laser light
1
Thick Glass Substrate
0.65
β
0.3
0 0.5 1
z
(cm)
DTL-like Solutions
The accelerating field
periodicity variation may die off before the
!/" particle fullly dephases
!
periodicity skipping
! 2! !

27. The low beta structure is now the critical technical
risk. Multiple approaches are being tried.
800 nm incident laser 800 nm incident laser
DBR
Periodic metal
matching
layer lets FP leak
acceleration out, but
reinforces
Reflective DBR standing wave
is short enough
to let F-P modes
leak out
800 nm 800 nm
400 nm 400 nm

28. Beam dynamics are challenging in optical scale
structures due to large transverse forces
Acceleration: coupling slot separation of βλ. Causes strong divergent force.
cannot achieve simultaneous transverse focusing and longitudinal stability
e-
FODO scheme proposed for focusing, stability (being studied)

29. Ultra-short period undulators

30. RF & Laser based undulators offer advantages but
demand excellent uniformity and are undeveloped
Good: Bad: Ugly:
large aperture betatron motion δ aU
high fields power loss along waveguide << ρ
smooth bore (wakefields) modes and cutoffs aU
tunable
Beating can create larger periods
RF waveguide undulators can work
Issues: 800nm + 1µm = 20µm
Readily available laser technology
Efficient path to longer periods
Better than OPO/OPA?
Ripples ok?

31. A grating based undulator can produce an
intermediate-period device
Plettner and Byer, Phys. Rev. ST Accel. Beams 11, 030704 (2008)
Barriers:
Smith Purcell parasitic radiation
Attosecond pulses and synchronization
Low fields?
Period limit? (300µm)

32. Beam powered devices have also been
considered: Image charge undulator (Wakefield)
Issues:
Another beam?
Advantage over RF?
Energy loss?
Acronym challenged (ICU)
Y. Zhang et al., NIM A 507 (2003) 459–463

33. A MAP-based undulator structure has been
designed
Undulator Period = Laser Phase Flip
waveplate
E-field
…… ……
λu >> λlaser
For E=3 GV/m,
Beqv=10 Tesla

34. Good mode quality has been
found but phase flips are hard laser

35. an all optical light source

36. It is possible to have an all-laser-powered x-ray
source using optical accelerator structures...
low energy high energy
+ +
optical undulator conventional undulator
= =
QFEL FEL but long
... but compromises must be made

37. A hard x-ray light source powered entirely by lasers
and on a laptop scale will be a Quantum FEL
Parameter Optical Und. Conventional
FEL Wavelength ~0.1 Å (10 keV)
Beam energy 10s MeV 100s MeV
Emittance (norm.) 0.06 µm
Current 2000 A
Charge 1 fC (whew! ~104 e-)
FEL Parameter (ρ) 10-5 10-3
Undulator parameter 10-3 ~1
Undulator period 1-20 µm 1 cm
Saturation length ~10 cm ~1 m
because !! / E " 6 # 10 $4
one photon emitted recoils > FEL bandwidth, ρ

38. We have the opportunity to develop a suite of on-
chip particle beam tools
guns sub-relativistic structures undulators
monolithic structures muons, protons, ions coherent THz/x-ray sources
IFEL accelerator
deflecting cavities focusing
ultra-fast sources ICS Gamma-Ray Source
all using laser-driven dielectric structure

39. Acknowledgments
Funding: Team:
NNSA Rodney Yoder
DTRA Jianyun Zhou (Postdoc - Fabrication)
UCLA Josh McNeur (Grad - Simulations)
DOE Hristo Badakov (Engineer)
Several past and present students...