Recent High-gradient test result at KEK

Linear Collider Workshop 2012Arlington, Texas (WebEx)

25 October, 2012Toshi Higo and X-band group of KEK

Contents

• Performance comparison of prototype CLIC structures– Processing speed– Design parameters– Breakdown rate

• Proof with CLIC pulse operation• Some effort for understanding breakdown• Near-future studies in mind

22012/10/25 LCWS2012 (Higo)

Nextef facilitiesat KEK

KT-1X-band

KT-2C-band

NextefX-band

A

B

32012/10/25 LCWS2012 (Higo)

Prototype tests at A

Basic tests at B

4

A

B

Control RF source

for A

Power transfer to B

Nextef

2012/10/25 LCWS2012 (Higo)

Comparison of CLIC prototype structures

in initial processing

52012/10/25 LCWS2012 (Higo)

Comparison of CLIC prototype structures

• Collaboration from 2007 ---• T18 2008 --- 2009• TD18 2009 --- 2010• T24 2010 --- 2011• TD24 2011 --- 2012• TD24R05 2012 ----

62012/10/25 LCWS2012 (Higo)

CLIC test structures; T18 TD18T24TD24a series of nominal fabrication by KEK+SLAC

T18_Disk_#2

TD18_Disk_#27

T24_Disk_#3

TD24_Disk_#4

undamped

damped

d

d

2009

2010

2011

2011~12

2012/10/25 LCWS2012 (Higo)

SLAC/KEK typical fab/test flowDesign for

CLIC (CERN)

Fabrication of parts (KEK)

Bonding (SLAC)

CP (SLAC)

VAC bake (SLAC)

High power test (NLCTA-

SLAC)

High power test (Nextef-

KEK)

82012/10/25 LCWS2012 (Higo)

9

Comparison of initial processingstarting at 51nsec

2012/10/25 LCWS2012 (Higo)

10

Difference in processing speed among four structures

More BD’s are required for damped!

BD’s are needed for processing?

Can it be reduced?

T24 T18

TD18TD24

2012/10/25 LCWS2012 (Higo)

TD24#4 initial

processing

11

132ns, 100MV/m

Through

ACC-BD = 2400and

670 hours

2012/10/25 LCWS2012 (Higo)

ACC-BD

TD24R05 initial processing now

12

It has been processed to ~100MV/m at

132ns

in 500 hours at 50Hz

Accumulated ACC-BD events amounts only

800, much smaller than the early-tested

structures.51ns 91ns 132ns

51ns 91ns 132ns

Note: Most of ACC-BD were not taken. Those are included in FC-UP events shown in green dots.

132ns

51ns

91ns

2012/10/25 LCWS2012 (Higo)

13

Difference in processing speed among four+1 structures

T24 T18

TD18TD2451ns

91ns 132ns

TD24R05

Number if ACC-BD’s until reaching the same level in (Tp, Eacc)

RankingT24 TD24R05? T18 TD24 TD18

Magnetic field!?!?2012/10/25 LCWS2012 (Higo)

Structure parameter choice

2012/10/25 LCWS2012 (Higo) 14

Reduced electric field 18 24

15

Alexej Grudiev

T18

Undamped

0 2 4 6 8 10 12 14 16 180

50

100

150

200

250

iris number

P [

MW

] (b

lack

), E

s (

gre

en),

Ea (

red

) [M

V/m

],

T

[K

] (b

lue)

, S

c*50

[MW

/mm

2 ] (m

agen

ta)

8.1 12.5

148

232

2.7

4.4

76

126

53.0

37.4

Pinload = 53.0 MW, P

outload = 37.4 MW

Eff = 0.0 % tr = 0.0 ns, t

f = 0.0 ns, t

p = 100.0 ns

High Eacc

P (M

W),

Es

(MV

/m),

Ea

(MV

/m),

D

T(C

), S

c*50

(M

W/m

m2)

2012/10/25 LCWS2012 (Higo)

T24Undamped

0 4 8 12 16 20 24240

50

100

150

200

250

iris number

P [

MW

] (b

lack

), E

s (

gre

en),

Ea (

red

) [M

V/m

],

T

[K

] (b

lue)

, S

c*50

[MW

/mm

2 ] (m

agen

ta)

7.5 8.4

176

205

3.03.2

90

108

41.1

23.4

Pinload = 41.1 MW, P

outload = 23.4 MW

Eff = 0.0 % tr = 0.0 ns, t

f = 0.0 ns, t

p = 100.0 ns

Reduce EaccP

(MW

), E

s (M

V/m

), E

a (M

V/m

),

DT

(C),

Sc*

50 (

MW

/mm

2)

T24 TD24

0 4 8 12 16 20 24240

50

100

150

200

250

iris number

P [

MW

] (b

lack

), E

s (

gre

en),

Ea (

red

) [M

V/m

],

T

[K

] (b

lue)

, S

c*50

[MW

/mm

2 ] (m

agen

ta)

7.5 8.4

176

205

3.03.2

90

108

41.1

23.4

Pinload = 41.1 MW, P

outload = 23.4 MW

Eff = 0.0 % tr = 0.0 ns, t

f = 0.0 ns, t

p = 100.0 ns

0 5 10 15 20 250

50

100

150

200

250

iris number

P [

MW

] (b

lack

), E

s (

gre

en),

Ea (

red

) [M

V/m

],

T

[K

] (b

lue)

, S

c*50

[MW

/mm

2 ] (m

agen

ta)

24.6 21.7

183193

3.3

2.9

94102

44.4

20.6

Pinload = 44.4 MW, P

outload = 20.6 MW

Eff = 0.0 % tr = 0.0 ns, t

f = 0.0 ns, t

p = 100.0 ns

Average unloaded of 100 MV/m

Increase of pulse heating DT ~ (Hp/Ea)^2 due to damping feature

16

Alexej Grudiev

2012/10/25 LCWS2012 (Higo)

Reduced magnetic field 18 24

0 2 4 6 8 10 12 14 16 180

50

100

150

200

250

iris number

P [

MW

] (b

lack

), E

s (

gre

en),

Ea (

red

) [M

V/m

],

T

[K

] (b

lue)

, S

c*50

[MW

/mm

2 ] (m

agen

ta)

29.1

47.0

155

226

3.2

4.4

79

120

57.5

34.3

Pinload = 57.5 MW, Pout

load = 34.3 MW

Eff = 0.0 % tr = 0.0 ns, tf = 0.0 ns, tp = 100.0 ns

P (M

W),

Es

(MV

/m),

Ea

(MV

/m),

D

T(C

), S

c*50

(M

W/m

m2)

Iris number

P

Ea

Sc

Es

DT

TD18 unloaded 100MV/m

17

TD18

Damped

TD24Damped

Reduce DT (or Hp)0 5 10 15 20 250

50

100

150

200

250

iris number

P [

MW

] (b

lack

), E

s (

gre

en),

Ea (

red

) [M

V/m

],

T

[K

] (b

lue)

, S

c*50

[MW

/mm

2 ] (m

agen

ta)

24.6 21.7

183193

3.3

2.9

94102

44.4

20.6

Pinload = 44.4 MW, P

outload = 20.6 MW

Eff = 0.0 % tr = 0.0 ns, t

f = 0.0 ns, t

p = 100.0 ns

TD24 unloaded 100MV/m

P (M

W),

Es

(MV

/m),

Ea

(MV

/m),

D

T(C

), S

c*50

(M

W/m

m2)

Alexej Grudiev

2012/10/25 LCWS2012 (Higo)

Reduce Ep

Reduce Hp/Ea and DT by reducing corner radius in the cell

18

TD24

TD24R05

Alexej Grudiev

2012/10/25 LCWS2012 (Higo)

19

Further reduce Hp/Ea

Average unloaded of 100 MV/m

Alexej Grudiev

TD24

TD24R05

2012/10/25 LCWS2012 (Higo)

0 5 10 15 20 250

50

100

150

200

250

iris number

P [M

W] (

blac

k), E

s (g

reen

), E

a (red

) [M

V/m

],

T [K

] (bl

ue),

Sc*5

0 [M

W/m

m2 ] (

mag

enta

)

24.6 21.7

183193

3.3

2.9

94102

44.4

20.6

Pinload = 44.4 MW, Pout

load = 20.6 MW

Eff = 0.0 % tr = 0.0 ns, tf = 0.0 ns, tp = 100.0 ns

0 5 10 15 200

50

100

150

200

250

iris number

P [

MW

] (b

lack

), E

s (

gre

en),

Ea (

red

) [M

V/m

],

T

[K

] (b

lue)

, S

c*50

[MW

/mm

2 ] (m

agen

ta)

21.1 19.0

184194

3.7

3.1

94102

40.5

18.7

Pinload = 40.5 MW, P

outload = 18.7 MW

Eff = 0.0 % tr = 0.0 ns, t

f = 0.0 ns, t

p = 100.0 ns

Max field and temperature rise

20

Ep/Ea Hp/Ea Sc/Ea2

TD18 1.97 5.9 0.52

TD24 1.95 4.5 0.37

TD24R05 1.95 4.1 0.41

T18 1.95 3.2 0.47

T24 1.95 2.6 0.37

2012/10/25 LCWS2012 (Higo)

Breakdown rate and flat or CLIC pulse

2012/10/25 LCWS2012 (Higo) 21

T24#4 BDR evolution at 252nsnormalized 100MV/m

22

Assuming the same exponential slope as that at 400hr BDR has kept decreasing.

2012/10/25 LCWS2012 (Higo)

2012/10/25 LCWS2012 (Higo) 23

Nominal CLIC pulse

F(CLIC)

Rs

Tr

242012/10/25 LCWS2012 (Higo)

Only 3 breakdowns in 484 hour

operation with CLIC pulse at

FLT=100MV/m

1.6x10-7 bpp/m

ACC-IN

Rs

Tr

RsF

FC-UP FC-Mid

252012/10/25 LCWS2012 (Higo)

2012/10/25 LCWS2012 (Higo) 26

T24#4(final)

BDR summary on TD24 comparing to T24

BDR results of TD24#4• BDR decreases as processing, as usual• Larger BDR than T24 but much less than

TD18• CLIC requirement is met through 2000

hours processing• BDR seems still keep decreasing• CLIC requirement (3X10-7 bpp/m) was

actually confirmed in CLIC pulse

2012/10/25 LCWS2012 (Higo) 27

Various studies toward understanding of

vacuum breakdowns

2012/10/25 LCWS2012 (Higo) 28

Comparison of dark currentT18_Disk TD18_Disk T24_Disk

29

Undamped 90~100 MV/m damped 70~80MV/m

TD24_Disk

0.01

0.1

1

10

100

50 60 70 80 90100 200

T18_#2 Dark Current evolution081128-081224-090224-090414-090515

FC-Mid [microA] (081128)

FC-Mid [microA] (081224)

FC-Mid microA (253ns, 090225)

FC-Mid microA (253ns, 090414)

FC-Mid microA 090515

FC-Mid microA

Eacc [MV/m]

Eacc for peak dark current of 10 m 90MV/m 70MV/m 100MV/m (80MV/m)

(51ns processing)

Last year LCWS2011

2012/10/25 LCWS2012 (Higo)

Identification of BD location from RF pulse shape and reflected phase

302012/10/25 LCWS2012 (Higo)

TD24R05#2

BD cell #

No special location nor steep variation along the structure. Probably surface related mechanism is important to study.

1st-pulse BD

Niminal BD

BD timing in pulseevaluated by decay timing of transmission

Uniformly distributed in the pulse for nominal ACC-BD.

High probability at the beginning of pulse for the 1st-pulse BD.

No BDR increase was observed in time in the pulse!! Is it usual? Is it naturally understood? We see in many cases in structure test, but contradictory to the result we observed in the waveguide experiment!?

2012/10/25 LCWS2012 (Higo) 31

Switching mode operationin power and width

Time

Power

Width 2sec

2012/10/25 32LCWS2012 (Higo)

As one of the trials to study memory on following pulses.

Width and Power Failed!

More BD’s in higher Eacc. Obvious! Power dependence was not well resolved. need more spacing in power level. need more time to study in this mode.Lower power BD’s in 1st-pulse BD Because startup with less power setting for recovery routine. BDR characteristics on width is not evident. Need more statistics.

Ned more sophisticated experimental programming!

2012/10/25 LCWS2012 (Higo) 33

Setups being prepared to study basic characteristics and mechanism

of vacuum breakdowns in RF

2012/10/25 LCWS2012 (Higo) 34

Single-cell setup just as that established by SLAC

We will study breakdown characteristics taking much focus on the initial processing stage appearing at

medium field, 60-100 MV/m352012/10/25 LCWS2012 (Higo)

Preparation of setup in shield-B

36

WaveguideNEG

HEPA

GV

Cavity

2012/10/25 LCWS2012 (Higo)

Studies in mind

• Explore basic research in a simple geometry

• Center cell is such as the following1. Standard: KEK made – SLAC test2. Nominal: Heavily-damped3. Made of large-grain material 4. Undamped but all-milled5. All milled quadrant type6. Choke-mode type (take Tsinghua design?)

2012/10/25 LCWS2012 (Higo) 37

These are under preparation

Some studies in mind

38

Clean surfaceCrystal

characteristics

Damped cell

2012/10/25 LCWS2012 (Higo)

Quad with large chamfer radius

Conclusion • Finished high gradient test of four CLIC prototype

structures. • TD24 closest to actual CLIC3000 has estimated to meet CLIC

BDR requirement in full-flat pulse. CLIC pulse operation was actually confirmed to meet CLIC BDR requirement.

• Processing of TD24R05 recently started. Initial processing speed as function of # of ACC-BD showed better than TD24, and even better than T18, up to 132nsec . (preliminary)

• Basic study test stand is ready to start. The klystron for it is being evaluated whether to be recovered from water leakage into waveguide.

392012/10/25 LCWS2012 (Higo)