1 1Office ofScience

MEMS Based Ion Beam Drivers forMagnetized Target Fusion

Thomas Schenkel, Berkeley Lab

San Francisco, August 29, 2017

Annual review of the ARPA-E ALPHA program

Compact, multi-beam accelerators

MEMS Based Ion Beam Drivers forMagnetized Target Fusion

LBNL:

Peter Seidl, Qing Ji, Arun Persaud, Will Waldron• Accelerators, beam physics, ion sources and beam transport, RF, …

http://atap.lbl.gov/

Cornell:

Amit Lal (Co-PI), Vinaya Kumar• MEMS fabrication, RF, Chip-scale particle accelerators, …

http://www.sonicmems.ece.cornell.edu/

2

MEMS Based Ion Beam Drivers forMagnetized Target Fusion - Summary

3

• What are you trying to do? • Our goal is to proof the concept of compact, multi beam ion accelerators that can be scaled to

very high beam energy (>1 MJ in ~1 µs) for compression of magnetized fusion targets at low cost (<$0.05/MJ)

• Why is this important? • Ion beams are very attractive drivers for MTF and for fusion plasma heating, but delivering the

required beam energy at low enough cost is very challenging• Why now?

• MEMS technology and solid state RF have matured significantly and we see an exciting opportunity to implement multi-beam accelerator concepts from the 1980s now with low cost fabrication and scalability

• Why is this hard? • ion beam transport in arrays of parallel (sub)-mm beams is challenging due to beam emittance

and space charge forces• alignment of beams is critical for efficient transport• ion acceleration in RF driven high voltage gaps has to be balanced with breakdown limits• efficient generation of acceleration voltages (>1 keV/gap) in a compact envelope

Our motivation is to develop low cost, scalable ion beam drivers for fusion plasma heating (and other applications)

• Ion beam drivers for fusion

o driver energy = Ekin * Ipeak * pulse length 1 to 10 MJ

o Heavy ion fusion (HIF): a few GeV, 0.1 MA, in nso Magnetized target fusion (MTF): ~0.1–1 MeV, MA, in µso (Tokamak) plasma heating: ~1 MeV, >10 A, >1 s

We need efficient and low cost drivers

4

Examples of high power ion accelerators at Berkeley Lab

5

• Pulsed induction linac (12 m)• 1 MeV, 2 ns, mm, ≥2 A peak• 200x drift compression• P. A. Seidl et al. NIM A (2015)

• Radio frequency quadrupole (RFQ)

• 2 MeV, 0.01 A, cw• 4 m long, 0.4 m cross section• Z. Zouhli, D. Li et al. IPAC2014

How can we scale ion beams to >1 MJ at

low cost?

• High Current Experiment• injection, matching and transport at

heavy ion fusion driver scale• 1 MeV, 0.2 A, 5 µs, ~12 m• 0.4 m cross section• M. Kireeff-Covo, et al., PRL (2006)

NDCX-II

Multiple-Electrostatic-Quadrupole-Array Linear Accelerator (MEQALAC) concept from 1980s (Maschke et al.)

6

• forming a high current beam from many small beamlets enables higher currents, higher beam power and higher current densities

• 1980s: ~1 cm beam aperture, electrostatic quadrupole (ESQ), lattice period a few cm

• G. Gammel, …, A. Maschke, … et al., IEEE Trans. Nucl. Sci. NS-28, (1981)• R. Thomae, et al., Mat. Science & Eng., B2, 231 (1989)

We use electrostatic quadrupoles (ESQ) to re-focus the ion beams for efficient transport

8

• beam acceleration with high voltage pulses from microwave driven resonant circuits (15 to 400 MHz) • LC, co-planar waveguides or strip lines

• high voltages per acceleration gap in the few hundred V range, experiments with >1 kV in progress

ions in from ion source

accelerated ions to detector

Q~87

We generate RF-acceleration voltages on the wafers with planar resonators

In the first two years we have developed the first MEMS based multi-beam accelerator

9

• A. Persaud, Q. Ji, E. Feinberg, P. A. Seidl, W. L. Waldron, T. Schenkel, A. Lal, K. B. Vinayakumar, S. Ardanuc, D. A. Hammer, Rev. Sci. Instr. 88, 063304 (2017), cover, June 2017

With multi-beam accelerators we can boost beam current and beam power >10X

10

• Ion current: current density from ion source (e. g. 100 mA/cm2) • area of extraction aperture(s)• Our 4” wafer based multi-beam accelerator, when extended to a 10x10 or 20x20 array of 1 mm

beams will have >10x higher current density (more current in a smaller footprint) than an accelerator with a single aperture (1 to a few cm2)

11 11Office ofScience

Compact, multi-beam accelerators

T. Schenkel, P. A. Seidl, A. Persaud, Q. Ji, W. L. Waldron A. Lal, K.B.Vinayakumar Sonic MEMSCornell ECE

• We now operate a multi-beam accelerator formed from a stack of printed circuit boards • Next step is scaling to higher beam current (from 50 µA to >1 mA) and higher beam energy (10 keV to

>100 keV)• This will be highly competitive/disruptive for mass spectrometry, neutron-generators, surface

treatment, ion implantation, plasma diagnostics, …• Scaling to ~1 MeV/m and >1 A for plasma heating will follow (e.g., neutral beam injectors)

12

extra slides

13

• beam acceleration with high voltage pulses from microwave (RF) driven resonant circuits (e. g. L-C, co-planar waveguides or strip lines)

• on-board acceleration per gap ~150 V to date, push to >1 kV/gap in progress

ions in from ion source

accelerated ions to detector

Electrostatic quadrupoles and RF driven high voltage gaps are the key components of our accelerator

14 14Office ofScience

We have demonstrated transport and focusing by ESQs in a 3x3 beamlet structure fabricated on PC boards

Platforms for focusing and transport

15

• ESQ options• PC board• Glass/metal• Silicon• 3D printed

• Evaluation metrics• Package density• Robustness (handling)• Beam transmission• Scalability• Cost

Platforms for acceleration

16

• Acceleration options• Off board RF• HV pulsing based on MOSFETS• RF driven

• co-planar waveguides• strip lines• LC circuits

• Evaluation metrics• Gradient per gap• Duty cycle• Scalability• Compactness• Cost

30 cm box

≈10 cm

10 cm

Co-planar waveguides for on-board RF high voltage generation

• Use CPW resonator to drive acceleration gaps.

• RF gen. power = 20W • Vdrive =10V (for 50Ω) • Q can be high, >1000 • Vres ~10kV • Next step: beam tests.

17

250 MHz

We envision that the compact RF accelerators fit in the footprint of the DEMO NBI system

18

Nucl. Fusion 57 (2017) 056026 P. Sonato et al.

modular: 2x10 RF

sources and injectors

Assume 60 Amps total current, 1 MeV D-, 3 Neutral Beam Injector ports on tokamak, and each NBI has:

• 46 accelerators, two rows per injector system • Accelerator components:

o 6-inch square wafers, 2 cm border on wafer.o 1080 beamlets/wafer, total current 0.43 A.

• Room for pumping/conductanceo 4 cm gap between accelerators

• Accelerators can be angled to achieve overall convergence to match port area at tokamak wall

• Assume injector J = 20 mA/cm2

o Beam-beam pitch = 3 mmo 2x margin for losses: Plasma source aperture dia = 1 mm,

inject 0.62 mA/beamlet (@ 20 mA/cm2), but need only 0.32 mA/beamlet

19

Accelerator array for NBI system is a converging group of beams injected into neutralization system. 1 MeV, 20 A

20

Accelerator array for NBI system is a converging group of beams injected into neutralization system. 1 MeV, 20 A

21

Matching quadrupoles + RF + focusing in RF drift section

22

Scintillator

Faraday cup

Retarding grid analyzer

MS RF+Q

With off-board LC-circuit: Eo = 11 keV + 4•(∆E/gap = 0.6 keV) = 13.4 keV

Transmission line model for the waveguide

23

After introducing loss into the ABCD matrix: 𝑘𝑘=𝑘𝑘 + 𝑖𝑖𝑖𝑖𝑄𝑄

𝑉𝑉𝑡𝑡𝑡 = 113𝑉𝑉𝑣𝑣𝑣𝑣 = 1𝑉𝑉RL=1MΩ

𝑉𝑉1𝐼𝐼1

= 1 𝑍𝑍𝑠𝑠0 1 × cos𝑘𝑘𝑘𝑘 𝑗𝑗𝑍𝑍0 sin 𝑘𝑘𝑘𝑘

𝑗𝑗𝑌𝑌0 sin 𝑘𝑘𝑘𝑘 cos𝑘𝑘𝑘𝑘 × 1 0𝑌𝑌𝐿𝐿 1 =

𝑉𝑉2𝐼𝐼2

*Because of high input capacitance (~14pF) of oscilloscope, we could not able measure high voltage directly.

Experimental setup used to measure high voltage from resonator

24

Gain calculated by dividing Resonator+SAW voltage by SAW voltage

- We used Surface Acoustic Wave (SAW) device to measure the high voltage(SAW has a very low input capacitance (~100 fF))

Resonance frequency and Quality factor of the resonator:

25

Impedance of Resonator + SAW and wires connectedInput Impedance of only resonator