High Gradient Induction Linacs for Hadron Therapy* G. J. Caporaso Livermore, CA 94551 USA 1 st International Workshop on Hadron Therapy for Cancer Erice, Sicily April 24 May 1, 2009 *Patents Pending. This work was performed under the auspices of the U.S. Department of Energy by under Contract DE-AC52-07NA27344 1
People of the Research Program 2
High gradient induction linacs for hadron therapy Dielectric wall accelerator (DWA) Early concepts High gradient possibilities Some current approaches Issues with the current approaches New architecture concept Summary 3
Induction machines are high current and low gradient Leakage current - + Coaxial feed line r r E dl = d dt S r r B ds V( t) = S( B) core B-field volt-second product current - + V accelerating gap B B s ²B H -B r Ferromagnetic core (area S) Core hysteresis loop 20 MeV, 2 ka, 70 ns AIRIX Moronvilliers, France Typical gradient <.5 MV/m 4
An early dielectric wall accelerator concept* Coreless induction accelerator Dielectric + - - Switch Insulator C L *A. I. Pavlovski, et. al. Sov. At. En. 28, 549 (1970) 40 MeV, 100 ka, 25 nsec** LIU 30 Sarov, Russia Gradient 1 MV/m **Courtesy of Anatoly Krasnykh, SLAC 5
High gradient insulators (HGIs) perform 2-5 x better than conventional insulators Closely spaced conductors inhibit the breakdown process Conventional Insulator High Gradient Insulator - - field emitted electrons floating conductors HGI structure forms a periodic electrostatic focusing system for low energy electrons Leopold, et. al., IEEE Trans. Diel. and Elec. Ins. 12, (3) pg. 530 (2005) Kapton + Emitted electrons repeatedly bombard surface + Emitted electrons repelled from surface Surface Breakdown Field Stress (MV/m) 100 High Gradient Insulators y = 104.27x 0.2003-10 Conventional Insulators y = 23.366x -0.2592 1 1 10 100 1000 10000 Pulsewidth (ns) 100 MV/M, 3 ns 3 mm HGI sample * U. S. Patent No. 6,331,194 6
A basic pulse generator is formed from two transmission lines All DWA configurations employ parallel plate transmission lines dielectric L conductors d w E One way transit time τ = L c ε r Impedance of each transmission line Z = 120π d ε r w Typical current flow in the line with a gradient E I = V Z = Ed Z = ε rwe 120π Example : E =100 MV /m, w =1cm,ε r = 3 4.6 ka 7
SiC photoconductive switch has demonstrated fast operation at > 30 MV/m average gradient* SiC offers the potential of high voltage, high current operation at elevated temperature with long lifetime and low jitter Optical Energy Injection Electrical Contacts SiC Peak Photocurrent (Amps) 25 20 15 10 5 150 V 350 V 550 V 700 V 900 V 1000 V Modulator/ Switching Region 0 0 2 4 6 8 10 12 14 Optical Energy (mj) High Gain Amplifier Region * Patent pending 8
DWA can be used in the single pulse traveling wave" mode to accelerate any charged particle* Along the wall E z τ t z/u ( r,t,z)= E z r,τ ( ) HGI characteristics imply that the highest gradients will be attained for the shortest pulses HGI E z (axis)/e z (wall) A high on-axis gradient is maintained as long as θ 0.3 This implies pulses in the range of a fraction to several ns Sech Super Gaussian Gaussian E z (wall) z w θ γw b w = full width at half maximum u = speed of wall excitation γ = Lorentz factor =1/ 1 u 2 /c 2 *patent pending 9
Stacks of Blumleins with independent switch triggers implement the virtual traveling wave* accelerator HGI Blumlein Proton source Laser Focusing Optical fiber distribution system HGI SiC photoconductive switches Monitor Stack of Blumleins * Patents pending 10
We are working with Tomotherapy, Inc. and CPAC to develop a compact proton DWA 200 MeV protons in 2 meters Energy, intensity and spot width variable pulse to pulse Nanosec pulse lengths At least 200 degrees of rotation Up to 50 Hz pulse repetition rate Less neutron dose (neutrons still produced in the patient) System will provide CT-guided rotational IMPT TomoTherapy has licensed the DWA technology from the Lawrence Livermore National Laboratory and CPAC has a Cooperative Research and Development Agreement (CRADA) with LLNL artist's rendition of a possible proton therapy system 11
F.A.S.T. Blumlein structure, proton injector and spark proton source Vacuum pump 5-Induction cells 7 Blumleins HGI Spark sources camera Thomson spectrometer 12
Proton injector and F.A.S.T. accelerator section Vacuum pump 5-Induction cells Laser line of sight Thomson spectrometer Spectrometer & camera 13
F.A.S.T. acceleration of protons is measured with spectrometer Accel. phase H + C ++ H 2 + or D + OFF Injector voltage Decel. phase accel decel Energy Signal proportional to F.A.S.T. voltage 14
Cast dielectric sustains interesting field levels in relevant configurations 4 cm x 56 cm x 0.8mm gap between electrodes Typical Trace Failure point ns At Failure
New SiC material fails at enhanced stress > 200 MV/m Failure of 3 Blumlein switches at > 30 kv Failures occur at electrode edges where computed field is > 200 MV/m Intrinsic breakdown field for pure material 10 mm x 10 mm x 1mm Average stress > 30 MV/m electrode Present work is focused on developing an integrated switch package that eliminates these enhancements solder SiC oil 16
Stacked Blumleins are subject to parasitic coupling Stripline impedances can be ten s of Ohms Switch Magnetic field lines close through adjacent layers, inducing currents in neighboring lines V Ideal stack output voltage t Output including parasitic coupling Z 0 /2 Z 0 Z 0 Z 0 /2 Radial lines completely isolate adjacent layers but have a very low impedance (sub- Ohm), requiring massive currents to support high gradients C L 17
Another DWA approach uses opening switches to interrupt current in an inductor* Opening switch is a drift step recovery diode (DSRD) *Courtesy of Anatoly Krasnykh, SLAC 18
There are several possible cell topologies employing DSRD s* SLIM concept *Courtesy of Anatoly Krasnykh, SLAC 19
Issues with the stripline/radial line architecture Architecture Stripline Radial Line New Parasitic coupling Yes No No Impedance Low Very low Low to High Switch count High Very high Moderate Input power High Very high Moderate to High Ringing output Yes* Yes* No Gradient capability High Highest Core may limit** Waveform control Limited Less limited Moderate External focusing No No Yes * Unless terminated with a matched resistive or beam load ** In addition to switch, material and HGI limitations of other architectures
Classical bipotential lens can be used to accelerate particles V conductor conductor 21
If we could move the lens at the right speed a particle could be continuously accelerated V conductor u conductor Move lens with speed u 22
Move the lens without moving the assembly - a moving virtual gap* V Move this region at speed u High conductivity material u High conductivity material φ V 0 Low conductivity material Need a material whose conductivity can be rapidly changed from a high conductivity state to a low conductivity state and back again * Patent pending Voltage is concentrated into a small gap 23
One way to do this is with photoconductivity* V Move this region at speed u Illuminated SiC u Illuminated SiC φ V 0 SiC not illuminated Laser light Voltage drop occurs across high resistivity region Need a material whose conductivity can be rapidly changed from a high conductivity state to a low conductivity state and back again * Patent pending 24
Induction machines have been used for decades to concentrate voltage Typical electron injector Induction cells gap Inductive voltage adders are induction concentrators 25
To what extent can this be achieved dynamically? Induction cells solenoid w i gap region l a Photoconductive switch C L Virtual gap* is created by shutting off photoconductive switches * Patent pending 26
Continuous model for moving virtual gap* Transmission line equations Acceleration field V x = g i L t Ri i x = C V t E a = Ri R = R o f ut x ( ) g = g o g ˆ ( t) Define dimensionless variables u (w) is speed (width) of the virtual gap ξ x w ψ ir o g o η R o t L Ω V g o w * Patent pending 27
Seek a traveling wave (similarity) solution for a long system ( 1 LCu 2 ) ψ fψ =1 E wr o Cu σ a = g o f ( σ)ψ ( σ) u is the speed of the virtual gap 1 LC σ ut x w = speed of an electromagnetic wave on the line = Lu η ξ µη ξ R o w There are two distinct regimes*: LCu 2 < 1 subluminal LCu 2 > 1 superluminal * Patent pending 28
Idealized solutions for constant accelerating field (subluminal) f = R ( σ ) R o 50 40 30 4 2 2 4 0.2 0.4 ψ = R o i g o σ 20 0.6 10 4 2 0 2 4 σ 0.8 1.0 E a g o 12 10 8 6 4 E Gain = g 1 1 + wr Cu a 1 o o u 2 4 2 0 2 4 σ 29
Idealized solutions for constant accelerating field (superluminal) f = R ( σ ) R o 50 40 30 4 2 2 4 0.2 0.4 ψ = R o i g o σ 20 0.6 10 4 2 0 2 4 σ 0.8 1.0 E a g o 12 10 8 6 4 Ea Gain = 1 + g o Lu R w o u 2 4 2 0 2 4 σ 30
3D EM simulations (XFDTD)verify the effect* grid coaxial feeds core switch tube Electric field (V/m) 120 100 80 60 40 grid conductivity (S/m) 15 16 14 12 10 8 6 4 2 0.10 0 5 10 15 20 25 30 35 40 45 50 time (ns) On-axis accelerating field in the middle of each switch tube segment (1 Volt drive per cell) 20 * Patent pending 0 0 5 10 15 20 25 30 35 40 45 50 time (ns) 31
Magnetic cores can enable the superluminal regime* Induction cells w Magnetic cores gap region l a Photoconductive switch C L i * Patent pending 32
A circuit dual exists using a helical inner conductor* Virtual gap* is created by shutting off photoconductive switches Induction cells w helix gap region l a C L * Patent pending 33
Traveling wave and finite length system solutions are similar ψ = R o i g o E a g o Charging phase η ξ Finite length solution η ξ 15 E a g o 10 E a g o 5 Traveling wave solution 5 10 15 20 σ ξ 34
Summary Key material strengths look to be consistent with 100 MV/m gradients for short pulses Near term goals Improve switch material Develop integrated switch package Add focusing to injector and characterize Source lifetime and repeatability quality New architecture idea to overcome parasitic effects 35