4 th Annual X-band Structure Collaboration Meeting May 3, 2010 T. Higo (KEK)

Discussion on critical issues for testing:

breakdown identification, preparation for test, processing protocol, etc.

4th Annual X-band Structure Collaboration MeetingMay 3, 2010T. Higo (KEK)

The purpose and way of this timeslot

• I want the present discussion to focus on the high gradient performance evaluation of main linac accelerator structures for CLIC.

• I picked up some view graphs so that you are triggered to present comments and discussions.

• To do items 1—10 appeared here are those we think needed for KEK activities, but also common to all this community.

2010/5/3 4th annual X-band structuture collaboration meeting at CERN (T. Higo) 2

To do1. Summarize the present processing early stage2. Get BDR from medium field level3. Take BRD values carefully in statistical manner in mind4. Evaluate the evolution of BDR5. Try to correlate BDR with other observed results6. Measure dark current to monitor processing7. Evaluate for whole life of accelerator structure8. Do more with post-mortem by SEM9. Do tests with simple geometries10. Work hard to conclude present-day feasibility description


For completing feasibility study• Requirement for CLIC• Requirement for linear collider• Requirement for other applications

• Following discussions are for LC especially CLIC• Items needed for feasibility judgment

• Is it enough with evaluation items of on-going high gradient tests?

• Let us consider what we need and what we can.


What are needed to evaluate for judging the feasibility

• Fabrication and preparation until feeding power– Required processes, associated time and cost, …..

• Performance in processing without beam– Processing speed– Ultimate reach of Eacc field– Breakdown rate vs Eacc

– Evolution of breakdown rate– Breakdown rate vs pulse width/shape– Dark current– Damage during processing

• Performance with beam• Deterioration due to long-term operation

2010/5/3 54th annual X-band structuture collaboration meeting at CERN (T. Higo)

We have started a series of tests on 11.4GHz prototype structures

T18_#1 T18_#2 T18_#3 T18_#4 TD18_#1 TD18_#2

Cell make KEK KEK KEK KEK KEK KEK

Bonding SLAC SLAC SLAC SLAC SLAC SLAC

Test SLAC, done

KEK, done

SLAC, done

KEK, future

SLAC, in test

KEK, in test


We have also tested two quadrant type structures.

Let us take these examples in mind and optimize our tests in near future including 12GHz structures.

Basic studies with simple setup are ongoing


Single-cell C10/CD10 Waveguide DC spark Pulse heating

SLAC SLAC SLAC / KEK CERN SLAC

SW TW TW DC SW

On-going On-going Sleeping On-going On-going

These also are already on-going and let us extend these to complement the tests with prototype structures and help understand physics behind the results.

BackgroundAccelerator design in mind

• How high can / should the gradient be?– CLIC sets 100MV/m for 3TeV– We should ask ourselves whether it can be realized and when? – For a certain timing of the project, we should declare the

practical value. – Should it be 100? Can it be lower or higher?

• Tolerable breakdown rate?– During processing? – Through the accelerator life?

• What HOM performance is required?– Big interrelation between pulse heating and damping geometry


Key technical issues should be understoodin fabrication and preparation until feeding power

• Do we need diamond turning?• Do we need high temperature vacuum treatment of

material?• Do we need hydrogen high temperature process?• Do we need vacuum baking at 650C for a week?

• All these should be confirmed with dedicated tests to understood in their mechanisms to confirm.

• This can be accomplished by the systematic basic tests in addition to a series of prototype structure tests.


What can/should we learn from processing without beam?

• Higher field toward downstream– Especially in T18/TD18

• What is RF processing? Why do we need it?– Improvement or deterioration?– Inevitable process? – Can it be replaced by careful preparation?

• Breakdowns are inevitable? Tolerable amount?– Tolerable number of breakdowns– Tolerable breakdown size

• What determines the processing speed?– Ramping speed for recovery from fault

• Waiting time• Recovery path (width, power, others ….)


Difference in processing speed exists

• Processing speed– One of the important characteristics– Compare it w.r.t. time or #BD?– We should compare among experiments

• Breakdown identification is related– Related to how to identify breakdowns

• Missing energy (SLAC), current flush (KEK)• RF reflection, RF transmission• Need to critically compare, but not yet done fully

• Processing protocol is also related– Speed and path of recovery from BD


TD18_#2 test at KEKSome pulse: Run34 #38 and #39

This case was not analyzed reasonably.

F Rs Ra Tr

FC-UP FC-Mid I_Rs Q_Rs


0 50 100 150 200 250 3000

20

40

60

80

100

120

BKD Rate (hr-1)Average gradient (MV/m)

50 ns 100 ns 150 ns 200 ns

This time, processed structure by progressively lengthening the pulse at constant gradient (110 MV/m)

Second T18 Structure Tested at SLACT18_#3 Chris Adolphsen, CLIC09


Whole history of processing of T18_VG2.4_Disk #2

0

20

40

60

80

100

120

0

500

1000

1500

2000

2500

3000

0 500 1000 1500 2000 2500 3000 3500 4000

MasterTable_Eacc_Trend till_090610

<Eacc> 51nsec Ushi<Eacc> 113ns Ushi<Eacc> 173n ushi<Eacc> 213n Ushi<Eacc> 253ns UshimotoEacc keep 253ns UshiEacc 51ns UshiEacc 113ns UshiEacc 173ns UshiEacc 213ns UshiEacc 253ns UshiEacc Usahimoto 2/21-3/23Eacc 252ns Usahi 4/1-4/7Eacc 252ns 4/7-14Eacc 312ns 4/14-16Eacc 412ns 4/23-27Eacc 331ns 4/28-5/1Eacc 252ns 5/1-11Eacc 412ns 5/11-6/10

Total BD<Eacc> MV/m Total# BD

RF-ON integrated (hr)

090610T18_#2, T. Higo


High Power Test begin at 12/03/2009 15:00

0 100 200 300 400 500 600 700 800 900 10000

20

40

60

80

100

120

Accumulated rf process time (hr)

BDR (1/hr)<G> for regular cell (MV/m)Pulse Width (divided by 10 ns)

TD18_#3 Faya Wang, 10 March 2010


Tp50ns250ns

TD18_#2, S. Matsumoto, in this meeting


To do -1

• These are some of recent examples.

• Let us summarize such processing characteristics

• And try to identify the driving source of difference– The way of processing or the structure itself?


We have done a series of tests on prototype structures


　 T18_#1 T18_#2 T18_#3 T18_#4 TD18_#1 TD18_#2 TD18_#3 Quad_#? Quad_#5Type Disk Disk Disk Disk Disk Disk Disk Quad QuadMake KEK/SLAC KEK/SLAC KEK/SLAC KEK/SLAC CERN KEK/SLAC KEK/SLAC CERN KEK

Material OFC OFC OFC OFC OFC OFC OFC OFC CuZrTest SLAC KEK SLAC KEK SLAC KEK SLAC SLAC KEK

Status Done Doen to be tested Done under test Done Done Done

#BD till goal 　 2000 280 * 　　 1500 80 　　Hrs till goal 　 4000 3300 # 　　 6000 800? 　　Trip source ME FC ME 　 ME FC ME ME FC

　　　　　　　　　　　　　　　　　　　　

Total number of BD 　 4000 　　　 >2000 $ 　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　　

*Till reaching 200ns#Before next width, stay before BDR<1/hr$On-going

MEMissing energyFCFaraday cup current burst

Goal100MV/m with >230ns

Should compare these processing-related values in a systematical manner.

Let us think What drives the processing?

• Number of breakdowns?– Surface processing through breakdowns

• RF processing time?– Field emission makes surface modification?

• Higher gradient or longer pulse operation?– Through higher dark current?

• Introduction of other perturbation?– Hot spot?


Is the present processing protocol the best?

• Any difference in its final performance– depending on the different protocol?

• Difference in such as – Peak power ramping speed during recovery– Width reduction and re-widening (SLAC)?

• Can we propose a processing in actual LC?– Any difference from what we do now?


How to define the ultimate field reach and operation level

• A significant deterioration may starts above some level in pulse energy or in peak power– Very stable operation may be possible before suffering from

many breakdowns– For example, see an experiment in DC spark test in next page

• Where should we stop further processing?– In case we can live within the level.

• Should we evaluate the performance step by step from lower field level?– See a simple schematic in peak power ramping and BDR

evaluation


Etched vs. Mirror surface

and BD field seems to be stable at the beginning.The first BD field is very high (~ 450 MV/m)

It seems to take longer time until saturation.The first BD field is low.

K. Yokoyama, private communication


Should confirm lower level operation


Evaluate lower level operation without (before) severe damage in peak power or in pulse energy ?? to confirm the robust and well-established levelThen go higher to evaluate at the nominal value with trying to detect when the deterioration starts

609080

100 110 Target120

To do -2

• Taking these consideration,

• Let us evaluate the BDR evaluation before reaching to the goal of 100MV/m

• Try to find the point over which deterioration starts (and possible the reason/mechanism)


Breakdown rate evaluation

• BD identification is needed• It changes in time• BD behaves in a statistical manner• …..


BD identification and threshold setting

MEReflectionFC current


In missing energy (SLAC) or current flush (KEK), or ??

OR

We should confirm the distribution and the threshold position. Do we have clear experimental evidence? Do we have information below threshold? Do we have any deterioration below threshold?


TD18_Disk_#2

Almost all confirmed to be spurious

To do -3

• Let us evaluate the present situation of how we identify breakdowns in a quantitative manner


Deduction of breakdown rate• = #BD/period?• We should be careful when to read the BDR values.

– Breakdowns appear in an irregular manner in time.– Some breakdown triggers following breakdowns in very

short time intervals.– Are the BDR values well saturated in a statistical view point?– In a smaller value case, the error is large due to limited time

to count many BD’s.• Evaluation period long enough?

– Even not, we should extract data. Just be careful to read and use these data points.


Breakdown rate measurements at 252ns

0.00 5.00 10.00 15.00 20.00 25.00 30.00 0.000

10.000 20.000 30.000 40.000 50.000 60.000 70.000 80.000 90.000

100.000

Run 34 (60MW)80 BD’s BDR [10^-6 /pulse/m]

0.00 10.00 20.00 30.00 40.00 50.00 60.00 70.00 0.000

5.000

10.000

15.000

20.000

25.000

30.000

Run 35+36 (55MW)41 BD’s BDR [10^-6 /pulse/m]



Run 45 (60MW) 59 BD's BDR [10^-6 /pulse/m]

0.00 5.00 10.00 15.00 20.00 25.00 30.00 35.00 40.00 0.0

50.0

100.0

150.0

200.0

250.0

300.0

Run 46 (55MW) 10 BD's BDR [10^-6 /pulse/m]

0.00 10.00 20.00 30.00 40.00 50.00 0.0

5.0

10.0

15.0

20.0

25.0

1st meas. 2nd meas.

Cascaded BDR Test at 100 MV/m @ 200ns

1 2 3 4 5 6 70

0.1

0.2

0.3

0.4

0.5

0.6

0.7

Frac

tion

of b

reak

dow

n ev

ents

Cascaded events

All eventsMulti-event taken as one event

Type 1 2 3 4 5 6 7

Events 77 23 5 5 1 2 1

33 hrs for the total test.

180 185 190 195 200 205 210 215

2468

pulses

Inpu

t ref

(arb

.u.)

24 26 28 30 32 34 362468

1012

pulses

Inpu

t ref

(arb

.u.)

150 160 170 180 190 200 2100

5

10

15

pulses

Inpu

t ref

(arb

.u.)

155 160 165 170 1752468

1012

pulses

Inpu

t ref

(arb

.u.)



Actual evolution of breakdown frequency

• Should evaluate the evolution in time– Keep decreasing?– Gets saturated? When? Why?– Any unexpected increase happens?

• Total number of breakdowns in life – should be expected from above information

• Speedup the evaluation– Evaluation at higher level may suffer from onset of severe

deterioration?– But we do not have any practical way to get precise data.

What to do?


95 100 105 110 11510

-7

10-6

10-5

10-4

Unloaded Gradient: MV/m

BK

D R

ate:

1/p

ulse

/mBKD Rate for 230ns

250hrs

500hrs

1200hrs

900hrs

500 1000 150095

100

105

110

115

Time with RF on: hrs

Gra

dien

t: M

V/m

Gradient at 2e-6/pulse/m for 230ns pulse

Experiment DataPower Fit

Gradient Increase Over Time at a Constant Breakdown Rate

Hot Cell Flare-up

First T18 StructureTested at SLAC

T18_#1 Chris Adolphsen, CLIC09


BDR evolution with rf process

0 500 100010

-7

10-6

10-5

10-4

10-3

Accumulated rf process time (hr)

BD

R (1

/pul

se/m

)

100MV/m@100ns100MV/m@150ns100MV/m@200ns100MV/m@230ns

0 1000 2000 300010

-7

10-6

10-5

10-4

10-3

Accumulated rf breakdown



Breakdown rate

10-7

10-6

10-5

10-4

80 85 90 95 100 105 110 115 120

T18_VG2.4_Disk #2Breakdown rate for 252ns and 412ns

BDR(ACC) 412ns

BDR(ACC) 252ns

BDR(ACC) [1/pulse/m]

Eacc [MV/m]

0530-0610 (3700hr)

0525-0530 (3500hr)

0520-0525 (3500hr)

0515-0518(3400hr)

0511-0515 (3300hr)

0501-0507(3100hr)

0423-0427(2900hr)

0411-0414 (2800hr)

0403-0407 (2700hr)

0402-0403 (2700hr)

0401-0402 (2700hr)

0313-0323 (2500hr)

0227-0305(2300hr)

(700-1200hrs)

T18_#2, T. Higo,


To do -4

• Let us evaluate the evolution of breakdown rate

• Try to see whether it keeps decreasing or gets saturated or it deteriorate from some point?

• It needs long period operation, rather than tasting too many structures quickly. (I admit though that the latter is effective in some purpose.)


Breakdown rate vs pulse width/shape

• Evaluation as function of– Pulse heating temperature rise– Any other thought?

• These trials are important– for understanding the mechanism of trigger of BD

• Results are consistent or understandable?– With other observations?

• Breakdown location?• Breakdown timing in pulse

– Period of irradiation with RF before BD


BDR Pulse Heating Dependence

20 30 40 50 60 70 80 90 10010

-7

10-6

10-5

10-4

10-3

Peak Pulse Heating at Last Cell (K)

BD

R (1

/pul

se/m

)

T18, 900 hrsTD18, 100 MV/m @ 100ns,870hrsTD18, 100MV/m@ 150ns, 960hrs TD18, 100MV/m@200ns, 750hrsTD18, 115MV/m@150ns, 550hrsTD18, 120MV/m@150ns, 570hrsTD18, 100MV/m@230ns, 700hrsTD18, 105MV/m@230ns, 680hrsTD18, 100MV/m@50ns



Breakdown distribution at different stage

0 5 10 15 200

50

100

150

200

250

Regular Cell No.

BD

R E

vent

s

0 ~ 200 hrs200 ~ 400 hrs400 ~ 600 hrs600 ~ 700 hrs700 ~ 960 hrs 0 ~960 hrs



Timing statistics for Rs rise and Tr fall

Rs: 100—350ns for 252nsec pulse Most of Rs change sit around 360+-50nsec

Tr: 150—400ns for 252nsec pulse Most frequent when approaching to the end, from 300 to 370ns. There is a small bump at 170 nsec, at the beginning of pulse.


TD18_#2, T. Higo,

This Tr reflects the information of irradiation period before BD


TD18_#2, T. Higo,

Should revise the resolution of the analysis of BD location

BD location.

To do -5

• Try to correlate the BDR to other observables– Such as pulse heating – Pulse width and pulse shape– Breakdown location– Breakdown timing

• To understand the trigger mechanism to drive breakdowns


Pulse-to-pulse stable dark current

• Important to measure the evolution?– Amount of FE current– Beta values in FN – Decrease and saturation, or jump up?– Can be one of the measures of processing?

• Energy spectrum– To understand the FE position information– Possible relation to trigger of BD


TD18_#2Dark current evolution

especially at early stage of processing



To do -6

• Dark current should be observed as processing proceeds

• Try to find relation between stable FE current and reached high gradient performance

• To get relation between dark current to processing.

• This information may give us a way to monitor processing.


Damage during processing or in operation• Deterioration are observed

– Cell frequency change• Especially non-uniform change: not easy to compensate

– Breakdown pits and rough surface profile• Larger field enhancement?• Cleaner surface?• Hard surface due to rapid cooling?

– Triggers to the following breakdowns• Degree of deterioration should be evaluated

– Easily subject to the systematic error in measurement– Well-established setup should be used


T18_Disk_#2 Bead pull amplitude plot before and after high power test


1. Some Eacc smoothness was deteriorated.2. Average frequency increased by 1.1MHz after 4000 hours

processing with 2500 breakdowns in total. 3. 1MHz ~40degrees/structureShould carefully measure to evaluate the effect.

Performance with beam• Field profile along structure changes

– Especially on T18 or TD18– Long-term stability is to be determined with beam

• Evaluation is needed with/without beam– On BDR– On BD location


Long-term operation• We should confirm the life-time stability

– The stability against practically long-enough operation at nominal field

• How to do it?– Design a structure with the same Es and Hs distribution without

beam as those with beam and test it without beam– Let us operate for full 2 months (1500hrs)

• At 10% higher field to gain x10 BDR• To simulate 2 years operation to estimate 10 year run• We can also get evolution information if exists

– Finally should inspect RF characteristics and proceed SEM inspection


To do -7

• Evaluate possible damage on structure due to processing

• Design a structure with nominal loaded Ea distribution without beam

• Estimate the stability over lifetime range

• To estimate the life time history of accelerator structure.


SEM inspection after high gradient test

• Postmortem inspection– Optical seems not enough, though non-destructive– SEM seems most powerful, though destructive for other than quad

• What should we evaluate– Any foreign objects? Try to find possible source!– Surface deterioration?

• On high electric field area• On high magnetic field area

• We should cut and see with SEM on the structures– T18_Disk_#1, #2, TD18_Disk_#2, #3?– Unless we have clear idea for another experimental plan


To do -8

• Let us proceed post-mortem inspection by SEM on most of the tested structures unless we have clear idea what to be tested in future

• Try to correlate this inspection to the obtained high gradient performance

• To convert a dead body into birth of another life!


Simple-geometry studies are needed for understanding the results on prototype structures

• Single-cell SW breakdown study– Performance taken from moderate field level (~80?)

• Milled cell• LG cell• Quad• Heavy damping• Etc.

• Modified C10/CD10 in 10 cell setup– SW needed?– TW


Evaluation of effect of damping port


vs orMed. Heavy

damp Heavy dampDDS

U U D U D U D U D

U U D U D U D U D

C10-modified

SW or TW in C10-modified

To do -9• Let us plan a series of high gradient test with

simple geometries from moderate field level• Try to correlate the results to the performance

observed in full-size prototype structures

• This should serve us with not only the understanding of the intrinsic limit but also some hints to go another direction to further higher gradient.


In summary: My conclusion with deferring to discussion

• Essential evaluation measure is still– BDR and early processing characteristics

• More information is needed to judge feasibility– Dependence on pulse shape– Evolution and estimated change in lifetime scale– Effect of beam loading– Long-term stability

• Simple-geometry studies must to be done– In parallel to understand the mechanism and physics– Single-cell or several to 10 cell setup– SW and TW

• Study at the moderate field level should be done– before addressing higher and target operation level


X-band strategy to take now• We need to show the feasible level based on now.• Characterize the structures of present-day design

(TD24) in low-60?, medium-80, goal-100, and higher-110 gradient levels.

• Conclude by ourselves the present-day feasibility of accelerator structures based on copper material.

• Show to outside world the feasibility at each field level.– Established level + level to be confirmed in near future, …..

• Then after, explore further developments including other design and other material or surface treatments.


Last to do -10

• Work hard with enjoying finding physics– Not only in a systematically well-organized manner for

CLIC– But also with various different view points to understand

the physics behind

• This is to conclude the feasibility– Of present-day established level– And judgment on 100MV/m– And possible idea to higher gradient


4 th Annual X-band Structure Collaboration Meeting May 3, 2010 T. Higo (KEK)

Documents

Transcript of 4 th Annual X-band Structure Collaboration Meeting May 3, 2010 T. Higo (KEK)