High gradient linacs: new ideas, and some experimental data.

LBNL, November 18, 2011Beijing December 2003

FRANTEL/FELTECH 1

High gradient linacs: High gradient linacs: new ideas, and some new ideas, and some experimental data.experimental data.

An unorthodox view

F. Villa (FRANTEL)

Introduction• Maximun practical RF linac (warm) gradient 137 MV/m.

• The gradient is limited by dark current, breakdown. Field dependence as high as E8

• No substantial increase in the last twenty years

• Seven national labs and 137 (that number again!) have worked on the subject.

• High gradient: arbitrarily set at more than 430 MV/m ()

• Pros1) Higher gradient (shorter linac)

2) higher peak current, some Schottky help, for injector

3) higher brilliance.

• Cons– Simply said,( as of today) there is no switch with the right specs. Maybe there is another way.

• Will describe what has been done, my level of understanding, and some ideas.

• The presentation is the same as a talk in Beijing in 2003. Not much has changed since.

Beijing December 2003 FRANTEL/FELTECH 2


Consider “standard” acceleration techniques only. Parallel work in various flavors of “laser acceleration” not considered here.

• Apply a potential difference between two areas/electrodes in space, and a charge going through that potential will gain energy.

The limits for (maximum electric field ) are:• Work function, no field enhancement: , is the work

function in eV. For the max field is ~15 GV/m. This may not be a hard limit (magnetic insulation).

• Surface melting: depending on configuration, electrode material properties, limits vary from 12 to 26 GV/m for single pulse. Lower of course for multiple pulses at high repetition rate.

Achieved so far : 2.7 GV/m, 100 ps, single pulse. “Only” a factor of 5-6

to go.

mGVE /695.0 2 eV5.4


Present statusPresent status• Present status for RF linacs: 100 MV/m,

at 10-17 GHz (test set ups)

• Future (?): maybe 150 MV/m.

• But even if the structure will hold the electric field, the power sources+compression schemes will be a serious limit at 150 MV/m, even at “quasi optical” frequencies.

• Note: elastance scales with the square of the frequency, the power with the square of the electric field.


Work with RFWork with RF

Two parallel efforts to increase the E field: increase the available peak power from RF sources, increase the electric field at which the accelerator can be

operated. Requirements:

1. extreme (probably active) power compression2. dramatic reduction of dark current (from the present values)

The dark current problemThe dark current problemWhen high fields are present in a structure (generally a copper pipe with irises) an unwanted current appears in the structure. This current can be very high: it can generate beam loading at first, and catastrophic breakdown (arcing, sparks, surface damage). After a breakdown, the structure is

damaged, and the max field is lower .


Conventional wisdomConventional wisdom Field emission mechanism generates the dark current,

originating where the field is intensified by “imperfections” of the surface (protruding points and embedded dielectrics). The ratio between the applied field and the “intensified” field is called .

Most of quantitative data on dark currents are shown in the usual Fowler-Nordheim plot. A value of is quoted.

The fact that the values of range between 25-80 is ignored. After all, the measured current is only six orders of magnitude above the calculated value without the enhancement factor.

Breakdown in RF structures is frequency dependent. There is an “analysis of data” concluding that the breakdown goes as 7/8. Of course, only the available data that fit this dependence were used in the analysis. Those outside were ignored, as “not reliable”.


ConsequencesConsequences

• These “conventional wisdom” points have slowed down the progress towards high fields to the point where no progress has been done in the last fifteen years.

• Some light at the end of the tunnel: for instance, the work at CERN……


Experimental evidenceExperimental evidence1. Field emission, with very low, reasonable is the

“seed” for dark current and eventual breakdown.

2. This “small” current is amplified by secondary electrons and photons (and possibly ions) emitted from the surfaces bombarded by the initial field emitted electrons.

3. The breakdown voltage is strongly time dependent: high fields are possible with very short pulses, both in normal and superconducting materials.



Rough EstimateRough Estimate An extracted electron can produce, in average, secondary

electrons. It is a function of the voltage, but not a fast functional dependence. The value of includes all possible effects capable of producing/extracting new (not the original electrons emitted by tunneling) electrons from the surfaces. This quantity can be measured experimentally. It assumes a positive value, the number of electrons will grow as

for 1 initial electron. g/c , g is the gap between two electrodes, and c the speed of light.

t

tN )1(


ExampleExample• 1.5 mm gap has a transit time of 5 ps. There are 200

generations within 1 ns pulse. A value of alpha of 0.071 will give a “gain” of 106. Under the same conditions, a pulse 100 ps long will have a gain of 4.We have applied 2.7 GV/m on a 1mm gap, extracted 1 KA of current, and not seen any (measurable) dark current. No special precaution was used to clean/buff/polish the electrodes.

• Dark currents and subsequent breakdown are due to the amplification of a small field emitted current via secondary emission on the surfaces.


Dark current for 1 ns pulse (BNL)Dark current for 1 ns pulse (BNL)


Understanding the breakdown Understanding the breakdown mechanism is essential in order to mechanism is essential in order to

raise the electric fieldraise the electric fieldMeasure the dark current as a function of time: it should have the characteristic growth of an avalanche (difficult measurement)

Measure the integrated current as a function of pulse length, for constant voltage: initial current very low, growth with high fluctuations towards the end of the pulse (~1.5ns to 6 ns). This is the reason to go to the shortest pulse length possible.

Delayed breakdown with RF (I. Wilson, CERN CLIC)


Work to doWork to do

To increase the operating voltage on a structure with good reliability:

Reduce the avalanche gain (surface physics). Reduce the duration of the pulse. Determine the “memory” limit for a train of pulses.

In any case, a field 20-30 times higher (2.7 GV/m) than any obtained with RF has been proven, in presence of 1KA extracted current.


How to make subnanosecond How to make subnanosecond pulsespulses

Switching :1. Solid state: low efficiency, low voltage

2. Gas avalanche: efficient, high voltage, low rep rate

3. Shock waves in magnetic materials: at its limit around few ns, with present ferrite materials.

Fourier synthesis:Tested only at low voltages, with resistive load.


Experimental dataExperimental dataWe have obtained 2.7 MV, 110 ps, in the anode/cathode

gap of 1 mm

• By using the gas avalanche switch because it is relatively simple, inexpensive, and capable of fast risetimes (in the right conditions)

• The risetime limit of a pulse generated with this switch is limited by geometry, not by the switching mechanism.The pulse length is given by geometry as well.

• The switch is not capable of rep rates much above 20-30 pps: which makes diagnostics almost impossible. The fastest single shot scope has a bandwidth of 6 GHz (~70 ps risetime). Attenuating a 2.7 MV pulse down to a few tens of volts, with a short risetime is quite difficult. We “measure” the pulse height by looking at the extracted electrons energy, and we guess/deduce the pulse length in time.


Experimental setupExperimental setup


Measuring the emittanceMeasuring the emittance


Emittance measurement detailEmittance measurement detail


Beam profile after the 1D pepper potBeam profile after the 1D pepper pot


Measured emittance Measured emittance (0.4 GV/m, 0.8GV/m)(0.4 GV/m, 0.8GV/m)


Measured emittance Measured emittance (1.2 GV/m to 2.7 GV/m)(1.2 GV/m to 2.7 GV/m)


Jean Baptiste Joseph FourierJean Baptiste Joseph Fourier1768-18301768-1830


Fourier coefficientsFourier coefficients

• Triangular pulse

• Square pulse

22

22

1

12

kT

tT

tak

2

sin2

T

tk

T

tk

T

tak

for 05.0,463 kak nsTnstt

T200;3.0;667

first 60 harmonics have an amplitude of ~3 x 10-3


Advantages, disadvantages and Advantages, disadvantages and features of Fourier synthesisfeatures of Fourier synthesis

ProPro’’ss1. Short pulses can be generated at high rep rate.

2. The electron beam bunch length and rep rate are variable independently

3. Beam load compensation

4. Beam energy extraction efficiency increases by a large factor

5. Pre-compression modulator can be solid state.

ConCon’’ss1. Adding amplitudes without interacting with each other.

2. Load(s) may not be resonant.

3. Mode mixing.


Program for RF Fourier Program for RF Fourier compressioncompression

1. design and build a Fourier synthetic pulse system.

2. the first unit should be most general, not necessarily fit for an accelerator.

3. identify possible applications of the technology, and gather support from the appropriate agency.


ExampleExample

Detection and trigger of mines. (ground penetrating radar)

A big problem in many countries. A portative imaging radar can detect and possibly detonate the devices. A plus for foreign assistance. Builds up the image of the country “donating” the technology.

1. There are an estimated 60 to 70 million landmines in the ground in at least 70 countries in Africa, Asia, Europe, the Middle East and the Americas

2. Landmines remain active for an average of 50 years 3. An average of 2000 people are killed or injured by landmines each

month 4. The cost of neutralizing a landmine is (at present) between $300 to

$1000 5. The average cost for surgical care and prosthetics is $4000 per amputee 6. 1 out of every 236 people in Cambodia is an amputee as a result of

landmine injuries 7. 20,000 people in Angola are amputees due to landmines


Improvements with high gradientImprovements with high gradient

1. Increase photoelectron yield from Schottky effect, and lower photon laser energy

2. At the anode cathode gap: Increased brilliance and current extracted, reduced emittance. The emittance scales as Q/E. Brilliance scaling is a bit more complicated.

3. During acceleration, reduced beam loading (longitudinal emittance)

4. The total length of an FEL(accelerator+undulator), for a fix wavelength, scales as the inverse square of the gradient.


Schottky effectSchottky effect

• j is the current density

• I is photon beam intensity is the work function

AIEhRj i

21

MKSunitse,

4 0


StructuresStructures

Three structures ( there are many more):

1. Radial line

2. Plane wave transformer

3. Resonant line.


Radial line transformerRadial line transformer


Radial line transformerRadial line transformer


Radial line transformer gainRadial line transformer gain


Resonant lineResonant line


Resonant line detailResonant line detail


ConclusionsConclusions• Fields of the order of 10 GV/m are possible with 10-20 ps pulses,

and ~3 GV/m has been demonstrated for 100 ps: this field shrinks a 1TeV collider to 300 meters (from 10 Km).

• The possibility of keeping successive bunches well separated (by using Fourier synthesis) makes the detector a lot easier.

• High fields produce extremely low emittance, high brilliance beams, as seen experimentally and as predicted by various codes.

• First application of the high gradient linac will be a FEL.

• Is it too late for a collider? Maybe at the next energy scale….a 100Tev linac at 10 GV/m is only 10 km long.

High gradient linacs: new ideas, and some experimental data.

Documents

max field

intensified field

e field

field dependence

field enhancement

applied field

maximum electric field

high gradient linacs