1. Recent Activities in NCRF Accelerators at SLAC-TID
Brandon R. Weatherford
KEK Accelerator Seminar – Sept. 19, 2018

2. Background – TID at SLAC
SLAC Technology Innovation Directorate (TID) was established to develop new technologies and applications – and new funding areas – using our core expertise in NCRF accelerators.
Many of these concepts require a compact footprint, high energy efficiency, and lower cost on a system-wide level.
This has driven interest in several aspects of NCRF systems:
Power Conversion - GREEN-RF Inverse Marx Circuit
RF Sources – High Efficiency Klystron Design
Accelerator Design – mm-wave and THz structures

3. GREEN-RF Inverse Marx Energy Recovery System
Mark Kemp, et. al

4. High Efficiency S-Band RF Source Programs at SLAC
SLAC has two “plug-compatible” programs to upgrade efficiency and power of 65MW S- Band 5045 klystrons used in LCLS
Redesign of 5045 klystron with a high-efficiency Bunch-Align-Collect (BAC) circuit
Increase system efficiency by recovery of energy in spent beam of klystron (depressed collector) – known as GREEN-RF

5. Conventional Depressed Collector Power Supply
In the conventional approach to depressed collectors, the power supply must be replaced with one appropriate for multiple collector stages
Power Supply
Upgrading existing systems is very expensive

6. Feed-forward Energy Recovery Scheme – GREEN-RF
Power Supply
Energy Recovery
Ideal Characteristics for “drop-in” retrofit for use of depressed collectors for increased efficiency :
Existing infrastructure is re-used so power supply remains the same
Energy is recovered by a separate component, and returned to the DC supply
Energy recovery is completely passive

7. An “Inverse” Marx Energy Recovery Modulator
Capacitors charge in series, and discharge in parallel
Each cell pre-charged to a voltage
The initial voltage at a tap is the number of cells below that tap times the pre-charge voltage
During pulse
Voltage rises in response to current to stage, and capacitance
In-between pulses
CLC resonant structure transfers energy in capacitors to modulator
Voltages re-settle to pre-charge value

8. Ability to Recover Rise and Fall of Pulse
RF is applied when the cathode current is flat
A key advantages of a depressed collector for pulsed klystrons is the ability to recover energy during the rise and fall time
Even for well designed modulators with short rise and fall, large fractions of the energy supplied to the beam are wasted
With proper energy recovery, could relax modulator rise and fall constraints, potential cost savings
Cathode current
Amplitude RF
Time
Energy wasted in collector during rise and fall times

9. Test Results at SLAC
First test of multi-stage depressed collector with energy recovery is complete
Collaboration between SLAC and CPI
Four stage collector on modified VKS-8262 tube (2.856 GHz, 5.5 MW peak, 6 us, 180 Hz, 45% efficiency)
The GREEN-RF concept was successfully demonstrated in test – Effective tube efficiency was increased to > 60%
Depressed Collector on VKS-8262
“Inverse” Marx power recovery module

10. High Efficiency XL Klystron Design Study for CLIC
Brandon Weatherford, et. al

11. High Efficiency X-band Design Study - Overview
SLAC/CERN collaboration funded by CERN to explore achievable efficiency for high peak power X-band klystron
Use recently developed techniques to improve existing SLAC XL series 50 MW, ~40% efficient klystron
Improve efficiency with novel bunching techniques such as Core Oscillation Method (COM)
Improve output cavity lifetime based on lessons learned from design and test of high gradient structures
Performance targets:
Frequency: 12 GHz
Efficiency: >70 percent
Peak RF Power: 40 MW minimum, 50MW goal

12. High Efficiency Klystron Design – COM Method
Conventional
Preliminary cavities are tuned conventionally for gain
One or more “penultimate” cavities are tuned high to squeeze the bunch
COM
I. Syratchev
I. Guzilov
A. Baikov
Most of the circuit sweeps electrons at the wrong phase into the right phase
Only once most of the electrons are at the right phase is the bunch compressed

13. High Efficiency Klystron Design – COM Method
Conventional
Peripheral electrons brought in
Preliminary cavities are tuned conventionally for gain
One or more “penultimate” cavities are tuned high to squeeze the bunch
“Core” debunches via space charge; position cavities to counteract this
COM
I. Syratchev
I. Guzilov
A. Baikov
Most of the circuit sweeps electrons at the wrong phase into the right phase
Only once most of the electrons are at the right phase is the bunch compressed

14. Initial Design with “Ideal” Output Cavity (M = 0.9)
0.9 μK, 66% Fill Factor
11 Cavities
5% degradation in efficiency due to 2D effects. Simulated RF efficiency is 78.5%.
10-Layer model:
Radial Stratification = 1.67
Spent Beam Velocity = 0.42c (outer) to 0.67c (inner)

15. Effects of Variable Perveance, Fill Factor
Design
0.9 μK, 75% FF
0.9 μK, 66% FF
1.5 μK, 66% FF
Efficiency (%)
79.4*
78.5
72.6
r Stratification
1.32
1.67
1.73
Maximum I1/I0
1.83
1.84
1.82
Spent e- velocity
0.44 – 0.59
0.42 – 0.67
0.41 – 0.65
*For 75% FF, further improvement is likely with fine tuning of output cavity

16. Multi-gap Output Circuit – Existing XL-4 Design
MAGIC-2D used to model the standard XL-4 multi-gap output cavity
Imported beam from XL-4 klystron simulation
TW circuit, with variable phase advance
Beam parameters: 420 kV, 335 A
Efficiency = 38%
Gradient = 72 MV/m
119º
0.7π
88º
0.5π
117º
0.6π
π/2 mode = 0.5π
115 kV
120 kV
104 kV
115 kV
Maximum gradient in MAGIC simulation is 72 MV/m
Location depends on phase

17. Genetic Optimization of Multi-Gap Output
Genetic optimizer used to find ideal multi-cell cavity design
Results show that output efficiency can increased
Distributed power extraction may be particularly effective
4 Cell
2 Cell
4 Cell, extraction from each cell
Output circuit total length held roughly fixed at original 4 cell design

18. Current Status
Initial design had problem with out-of-band gain at high frequencies – an issue to watch out for in long COM tubes
Later iterations performed at CERN using integrated 2-D EM solver in KlyC-2D suggest 70% efficiency is possible
Optimization of new multi-gap output in MAGIC-2D is still in progress, using optimized circuit design from KlyC-2D

19. High‐Gradient Accelerators at THz Frequencies
Emilio A. Nanni, et. al

20. Rapid Development of THz Accelerator Technology
Curry, et al. arXiv: (2017).
M. Dehler, HG2017
Growing International Community:
Kealhofer, et al. Science (2016)
Healy, et al., UCMMT, 2017
Acceleration
 Nature Comm. 6 (2015): 8486.
Photoinjectors
Huang, W. R., et al., Nature Scientific Rep. 5 (2015).
Zhao, et al. PRX 8.2 (2018):
Beam-Driven GV/m Fields
M. Dal Forno, et al., PRAB 19.5 (2016):
Deflectors and
<1 fs Timing
R.K. Li, et al., Ultrafast Optics XI (2017) (2017)
Toward Externally Driven GeV/m Accel.
Impacting Diverse Areas of Accelerator Technology:
Precision Diagnostics and Beam Manipulation - <fs resolution
Ultrafast Electron Diffraction fC, <10 fs
X-ray Generation – few to 10s pC, low emittance
High Current, High Luminosity >>10s pC, bunch trains
Why do you care?
If you want fs, high brightness, compactness

21. Higher Frequencies Can Achieve Higher Gradients
Accelerating gradient is limited by breakdown (i.e. arcing or plasma formation)
Breakdown threshold for surface electric field
Demonstrated operation with ~1 GV/m surface fields
W. R. Huang, et al., Optica 3, (2016).
THz MIT/DESY (Esurf ~ 300 MV/m)
M. Dal Forno, et al., PRAB 19.5 (2016):
Beam FACET (Esurf ~ GV/m)
SLAC UED (Esurf ~ 150 MV/m)
Other Examples:
Wimmer L. et al., Nature Phys. 10, 432–436 (2014).
Nanni, E. A., et al.  Nature Comm. 6 (2015): 8486.
Huang, W. R., et al., Nature Scientific Reports 5 (2015).

22. Advantages of Operating at THz Frequencies
Additional advantages of high frequency structures:
Shunt impedance increases as
RF pulse energy decreases as
~300 GHz Structure
Shunt Impedance for TM01 π-mode Structures
SLAC/MIT/INFN High-Gradient Research
SLAC Source/Struc. Development

23. Comparison Between RF and THz Accelerators
Scaling structure design from S-band to the THz range
Parameters for 100 MeV/m Gradient
Frequency
3 GHz
300 GHz
Stored Energy [mJ]
8450
0.013
Q-value [x1000]
17.96
2.05
Shunt Impedance [MOhm/m] 55
514
Max. Mag. Field [MA/m]
0.3
Max. Electric Field [MV/m]
210
Fill Time [ns]
2000 2
Loss in 1 meter [MW]
181 19
decreases by
4 orders of magnitude
Potential to operate at 10s kHz vs 100s Hz

24. Prototyping of mm-Wave Structures
Assembly of structure and impact on RF and high gradient performance is a key concern
Prototyping effort to test assembly using diffusion bonding and/or brazing
Completed tests on 12 assemblies consisting of ~40 RF structures
Focus is structural integrity, RF performance, frequency shifts
Coaxial Probes

25. Structure Fabrication for High Gradient Test at 110 GHz
First test with split-cell and diffusion bonding
Applying Advanced Metrology for Close Loop Manufacturing
Pre-Bonding SEM
Mode Converter and Cavities
3 mm

26. 110 GHz High Gradient Structure Assembly Complete
Diagnostic Window
Vacuum Pump Out
Diffusion Bonded 110 GHz Structure
Dark Current Port
High Power Window

27. High-Power Testing of 110 GHz Accelerating Structures
Achieved 110 MeV/m Gradient (<25k pulses) → Power Available for 100s MeV/m
Breakdowns observed after power increased and rapidly process away
Improving transport, coupling, diagnostics
Measured Forward Pulse
Transmitted Pulse and Gradient
Impact of Breakdown on Transmitted Pulse

28. Bridging the Gap Between Single-Cycle and Quasi-CW Excitation for Optimized High-Field Performance
Laser-driven THz sources can produce high power pulses on 100s ps timescale
100 kW
Single-cell 300 GHz structure
3D printed prototype
Pursuing pulse-compression of electron-beam sources for very efficient nanosecond pulses
Nanosecond time scale preserves high-shunt impedance of structures
3 cavities
~ 600 MV/m for 1 MW
500 μm
Modeling high gradient and short pulses → multi scale
Power, coupling, beam aperture challenges
Additive manufacturing (3D printing) versus conventional machining
Materials (Cu, SS, Cu-Ag)

29. 300 GHz Single Cell Accelerating Structure
E field
Narrowband THz laser-based sources capable of producing 100s ps pulses
Utilizing over-coupled scheme to demonstrate high gradient
100 kW power
Shunt impedance ~ 600 MΩ/m
Emax/Eacc ~2.3
Diagnostic Port
S21 ≈ -42 dB
WR-3 Rectangular Waveguide
Over Coupled Cavity Port
Developing structure for high
gradient test with 100 kW,
100 ps laser generated source (w/ RIKEN)

30. THz streaking of femtosecond electron beams
Goal:
Characterize timing jitter and bunch length
Develop a timing tool for UED
THz manipulation/ acceleration of fs e- beams
3.1 MeV electrons, 50 um gap, 100 um thick slit
PPWG Deflector
Slope or streaking strength
4 mrad (353 pixels) = 0.49 ps, or 1 pixel = 1.4 fs
SLAC UED Facility – Strong synergy between testing and utilizing THz structures in ultrafast experiments
0.5 mrad
0.5 mrad
0.5 mrad
Here is an example of a structure with immediate impact at the lab
R.K. Li et al., arXiv: (2018)

31. Thank you!
Questions?
Power Conversion & GREEN-RF: Mark Kemp,
High Efficiency Klystrons: Brandon Weatherford,
mm-Wave and THz Accelerators: Emilio Nanni,