1. Dielectric accelerator based 5th generation light source
Alexander Zholents, ANL
Physics and Applications of High Brightness Beams
March 28-April 1, 2016, Havana, Cuba

2. Single stage acceleration in a hollow dielectric channel*
or corrugated wall waveguide
7 ± ps 2b 2a Cu
Drive bunch
Main bunch
Dielectric 2a p t d
Drive bunch
Main bunch
Drive and Main from the same source bunch minimal timing jitter
Low cost device (likely)
Potential for:
high field gradients
high wall plug power efficiency
high bunch repetition rate
Quartz tube with 0.2 mm wall
Main DWA accelerator components
Copper tube
Electric field map ( GHz) Cu
Dielectric
group velocity = ( )c
*) G. A. Voss and T. Weiland, DESY M-82-10, 1982;
K. L. F. Bane, P. Chen, P. B. Wilson, SLAC-PUB-3662,1985;
W. Gai et al. Phys. Rev. Lett. 61, 2756,1988.

3. A concept of a high repetition rate multi-user FEL facility based on Collinear Wakefield Accelerator
Injector
Drive accel.
Main accel.
Undulator
x-ray optics
~25 m 20
Compact
Inexpensive
Flexible

4. A concept of a high repetition rate multi-user FEL facility based on Collinear Wakefield Accelerator
Drive accel.
Undulator
x-ray optics
DWA
~25 m 20
Low energy spreader
Up to 100 MV/m
High energy efficient
Tunable energy ~ a few GeV
Tunable Ipk> 1 kA
Rep. rate up to 50 kHz/FEL

5. Beam by design: beam shaper and why we need it
Road map to a high energy efficient accelerator: transformer ratio1,2
(Maximum field behind the drive bunch)
(Maximum field inside the drive bunch)
R =
E +
E - =
Other distributions possible, see also [2]
r(z)
E - E+
Double triangle distribution
R=5
z (mm)
R=3
Ez [MV/m/nC] -5
-10
-15 15 10 5
r(z)
z (cm)
E - E+
Double horn distribution
R< 2
E - E+
Wakefield (MV/m)
r(z)
Gaussian bunch
z (mm) 10
-10 5 -5
Goal is to extract maximum energy from drive bunch, up to 80%, and obtain highest energy for the witness bunch
DEwitness=0.8R Edrive
Bane et. al., IEEE Trans. Nucl. Sci. NS-32, 3524 (1985).
F. Lemery, P. Piot, Phys. Rev. Spec. Topics – Acc. and Beams, 18, (2015).

6. Mask + Emittance Exchange Beamline
Drive and Main Beam Generation
Mask + Emittance Exchange Beamline
Recent Demonstration at the AWA
e- beam profile after mask
e- beam profile before mask
Talk by Gwanghui Ha
Emittance Exchange Beamline
Mask
Witness
Drive
Shaping the drive bunch is critical to achieve high efficiency of energy transfer from the drive bunch to the wakefields and eventually to the witness bunch. Only bunches with a very specific shape (double-triangular) exhibit a quasi uniform energy loss for all electrons inside the bunch. This helps obtaining a high transformer ratio of the accelerating field in the wake to the decelerating field within the drive bunch.
~25 kW is deposited on mask
Drive Bunch: Double Triangle for High Transformer Ratio
Main Beam: produced from same mask to eliminate jitter

7. Drive bunch shaping using photocathode laser (2) *)
Linear ramp with parabolic head
r(z) 2 ~
r(z)
E+ ~ 1/√a
R ~ √a 2a
DEwitness=0.8R Edrive
High accelerator efficiency favors larger radius
*) F. Lemery, P. Piot, Phys. Rev. Spec. Topics – Acc. and Beams, 18, (2015)

8. Drive Bunch Beam Breakup Instability (BBU)
Examples of longitudinal and transverse wakefields
Wz (MV/m), Wx (kV/m/mm)
Cumulative collective instability arises from continuous exposure of tail electrons to transverse wake field*
tail
head
Snapshots of a single bunch traversing a SLAC structure
*) A.Chao, “Physics of collective beam instabilities in high energy accelerators”, New York: Wiley.

9. Balakin-Novokhatsky-Smirnov (BNS) damping of BBU
Use FODO channel
Produce “chirp” in the betatron tune along the electron bunch using the energy “chirp”, and
Force tail to oscillate faster than head, thus averaging the impact of transverse wake fields.
Illustration for Dielectric Wakefield Accelerator (DWA)
Transverse oscillation of particles of a chirped beam
p = gb≈ E / mc2
main bunch
head
After 4 m of propagation
tail
drive bunch
After 8 m
Initial energy chirp ~15 % (peak-to-peak)
Particles of different energies have different oscillation periods in the FODO lattice

10. FODO channel (quadrupole wiggler) to control BBU
Wakefield accelerator
NdFeB
81.3
Tapered quadrupole gradient
Dielectric channel imbedded into quadrupole wiggler
Wz ~ Q/a2 W ~ Q/a3
Quadrupole wiggler E0
Main E0
Drive
dump
45 cm
High gradient hybrid quad
Bore radius = 1.5 mm.
Peak gradient = 0.96 T/mm.
Sub-micron precision in the magnetic center position.
Length = 40 mm.
Weight = 2.5 kg.
Magnetic force between top and bottom parts = 30.5 kg.

11. Slippage effect main, 250 pC
drive, 8 nC
main, 250 pC
Drive bunch and main bunch separation decreases because of their different energies
(ps)

12. Problem mitigation s
main
Move main bunch to second maximum (can be difficult if done using the mask)
Make adaptive frequency channel and always keep main bunch at or near to the maximum (easy)
Use drive bunch with higher energy (affects facility cost and energy efficiency) w1 w2 w3
< w2
< w3

13. Tracking results using 8nC drive and 250pC main bunch
After 2 m
Ein=400 MeV
After 20 m
Eout=2.0 GeV
1100
main
TM01 freq = GHz
main
3500
TM01 freq = 336 GHz
wake
current
wake
current
p=gb
p=gb
z (mm)
z (mm)
700
500 t t
500 50
x (mm)
x (mm)
-500
-50 t t

14. Main Bunch Shaping Reduce correlated energy spread in the main bunch
Ez (MV/m)
Gaussian main bunch
z (um)
110
100 90 80 70
Energy (MeV)
sE=5.3%
Reverse triangular main bunch*
Ez (MV/m)
z (um)
110
100 90 80 70
Energy (MeV)
sE=0.3%
Self-wake from main bunch matches curvature of drive wake
Can be further improved
*) T. Katsouleas et al., Particle Accelerators, 1987, Vol. 22, pp

15. Maximum energy gain is defined by the quad strength
Ez scales ~ 1/a2
Obtained using a two-particle model, this envelope is defined by the maximum attainable quadrupole gradient at each bore radius *.
Ez scales as ~ a1/2 on a boarder line of a stability region
100 nC
Ez, (MV/m)
10 nC
1 nC
stable region
Wz ~ Q/a2 W ~ Q/a3
a, (mm)
*) Gaussian peak current distribution of the drive bunch is assumed
BBU control favors larger tube radius a
*) C. Li, W. Gai, C. Jing, J. G. Power, C. X. Tang, A. Zholents, PRST-AB, 17, (2014)

16. Tolerances Misalignment of FODO quadrupoles < 1 mm
Random misalignment rms errors
5 mm initial orbit offset
No particle losses in the case of correlated variation with amplitude of 2 mm and period > 0.3 m.
Random quad roll angles: < 10 mrad
Straightness of the dielectric tube: better than 10 mm
at 20 m
Maximum amplitude is 10 mm and the period is varied

17. Measurement of quad’s magnetic center with a sub-mm precision*
quad, 0.6 T/mm
moving wire
meas
fit
r.m.s. error = mm
0.75mm
a = ±0.792 mm
Scan number
Transverse field distribution in 0.75-mm aperture quad
Pulsed wire technique will be used for a quadrupole wiggler
*) I. Vasserman, J. Xu, APS MD-TN

18. Adjustment of quad’s magnetic center with a sub-mm precision*
±50 mm adjustments of the magnetic poles 1 and 3 shift magnetic center by ±24 mm in x and ±6 mm in y
Impact of shifting:
Calculate the magnitude of shifting:
*) N. Strelnikov, E. Trakhtenberg, I. Vasserman and A. Zholents, APS MD-TN

19. Vacuum chamber RF main mode and HOM coupler every 0.5 m
RF coupler removes ~ 2 kW
RF coupler
BPM
Beam power in: kW
Beam power out: (32 kW drive +30 kW witness)
Vacuum chamber (rotated by 90º )
RF coupler field transmission efficiency
DT= 25 °C above cooling water temperature at 20 W/cm power dissipation

20. Dielectric Charging by Loss Electrons
Provide surface conductivity
TiN ultrathin film (<10 nm) coating
Coating
Brazing
Drain charge from the bulk
Material by design: attain properties of perfect dielectric at THz frequencies and DC conductivity
Candidate material: zinc metaniobate (ZnNb2O6), works at (10-150) GHz
R&D approach
Continue characterization and improvement of ZnNb2O6 or its equivalent
Explore CVD diamond with doping
Explore conductive material-based structures

21. Vacuum chamber prototyping
10 cm
Copper coated quartz tube
Copper coated quartz tube
Will compare vacuum in two units, one with and one without the tube

22. PLANS

23. The initial goal is to build a 0
The initial goal is to build a 0.5 m long accelerator unit and test it in LEUTL tunnel using APS injector linac
0.5 m
Will test:
alignment tolerance <0.75 mm rms quad-to-quad and <2 mm rms module-to-module

24. Linac Extension Unified Test Laboratory (LEUTL) at APS
Deflecting cavity
400 MeV
LEUTL tunnel
Beam parameters in LEUTL tunnel
Electron beam energy …500 MeV
Electron bunch charge 50 – 500 … 1000 pC
Beam emittance 0.5 – 1.5 mm-mrad
Beam energy spread 250 – 500 keV
Bunch length 100 – 1000 fs
Bunch rep. rate 6 – 30 Hz
PCgun installed at the head of the Linac
Laser (Nd:Glass)
Wavelength nm
Pulse energy 2 – 5 mJ
Pulse length 1 – 4 ps

25. S-band LCLS-type photocathode gun a the head of the Linac
APS 450 MeV injector linac
S-band LCLS-type photocathode gun a the head of the Linac
S-band deflecting cavity at a test stand during conditioning
LEUTL tunnel is ~ 40 m long and is “ready” to accept the beam
Go to "View | Header and Footer" to add your organization, sponsor, meeting name here; then, click "Apply to All"

26. Study cases Case I Case II Fundamental mode Freq. (GHz) 400 300
ID (mm), OD(mm), Length(cm)
1.5, 1.59, 50
2, 2.12, 50
Drive bunch charge (nC)
3.5 8
Double triangular bunch length (mm) 1
Drive/main bunch energy (MeV)
Bunch rep. rate (kHz)
100 50
Peak Accelerating Field (MV/m) 42 90
Power dissipation without and with THz field coupler per unit length (W/cm)
19, 5.4
58.5, 18.5
Transformer ratio 5
Main bunch charge (pC), length (mm)
50, 5
300, 10
Total DWA length (m)
~40
~25
Drive beam use, dump energy (MeV)
80%, ~ 70
80%, 80
Drive beam to main beam efficiency (%)
8.6 15
Main beam energy gain (GeV)
1.5
1.6

27. FEL simulations (illustration)
Undulator
period, cm
parameter, K
Energy, GeV
Charge, pC
Current, kA
Emitt, mm
RMS energy
spread, %
Pierce
parameter,
X-ray wave-
length, nm
Peak
power, GW
Bandwidth, % 3.8

28. Summary Multiple stages possible in the next phase
Single stage high repetition-rate, soft X-ray FEL user facility
10 CWAs linacs driven by a single 400 MeV SRF linac
10 FEL 50 kHz bunch repetition rate
Compact, inexpensive, and flexible
Progress
Drive bunch shaping (triangular + quadratic component)
Control of beam breakup instability
Quadrupole wiggler, adaptive frequency channel
Small “main bunch” energy spread
Prototype quadrupole design
Prototype vacuum chamber design
Future development
alternative beam shaping techniques– critical
space charge effects
beam test– critical
break sections: BPMs, rf couplers, correctors, etc.- critical
Multiple stages possible in the next phase

29. Acknowledgement S. Doran, W. Gai, J. Power
Argonne National Laboratory, HEP
R. Lindberg, N. Strelnikov, Y. Sun, E. Trakhtenberg, A. Zholents
Argonne National Laboratory, APS
S.S. Baturin, C. Jing, A. Kanareykin
Euclid Techlabs
D. Shchegolkov, E. Simakov
Los Alamos National Laboratory
P. Piot
Northern Illinois University/Fermilab
Q. Gao, C. Li
Tsinghua University
G. Ha
Pohang University of Science and Technology

30. Thank you for your attention

31. BACKUP

32. 34 m Illustration using 3.5 nC drive and 50 pC main bunches g(t) x(t)
p = gb≈ E / mc2
3.5 ps
Ip(t)
34 m
the drive beam tail decelerates more,
develops more lagging, and
sees the wake’s accelerating field

33. Customizing the drive bunch current to reduce the required initial energy chirp*
Classical double-triangle bunch current distribution
Uniform wake field Wz ωt ωt
New bunch current distribution
Double-triangle with a parabolic content
Linearly growing wake field stimulates energy chirp Wz
Reduce initial energy chirp: 15%  9.5% ωt ωt
*) D.Y. Shchegolkov, E.I. Simakov, A.A. Zholents, IEEE Trans. Nucl. Sci., vol. 62 (6), pp. 1–8, Dec

34. Planar geometry waveguide
vacuum
dielectric
metal W X Y b a
(a=1.5 mm, b = mm, W=8 mm, =5) selected to obtain the same Wz as in the circular geometry
Drive
dipole
quadrupole
Trailing
particles Z Y
Even on-axis drive bunch will excite the quadrupole wake
Can quadrupole wake produce BNS-like damping?

35. Planar geometry waveguide
vacuum
dielectric
metal W X Y b a
(a=1.5 mm, b = mm, W=8 mm, =5) selected to obtain the same Wz as in the circular geometry
Drive
dipole
quadrupole
Trailing
particles Z Y
Even on-axis drive bunch will excite the quadrupole wake
Can quadrupole wake produce BNS-like damping?

36. Is there confinement by quadrupole wake?
Quadrupole wakefield and quad focusing are in “phase”
sx (m)
s (m)
Quadrupole wakefield and quad focusing are in opposite “phase”
Not much.
Although slightly weaker quads can be used to control BBU.
sx (m)
s (m)

37. Drive bunch shaping using photocathode laser *)
Laser pulse intensity
measured
parabolic fit
space charge
I (A)
I (A)
0.8 nC
head
tail
head
tail
Next to the cathode
At the end of the injector at 100 MeV
Without ramped peak current
With ramped peak current
… was proposed to remove significant quadratic energy chirp at the end of the FERMI FEL linac
*) Cornacchia, Di Mitri, Penco, Zholents, Phys. Rev. ST-AB, 9, (2006);
Penco, Danailov, Demidovich, Allaria, et al.,Phys. Rev. Lett, 112, (2014).

38. LINAC improvements New Photo-Cathode electron gun
Linac structure straightening (ongoing)
New slice emittance and slice energy spread diagnostics (ongoing)