towards more flexibility in operation of the xfel linac - desy

Towards More Flexibility in Operation of the XFEL Linac

J. Sekutowicz October 25th, 2013

DESY

Outline

1. Introduction and motivation a. XFELs and their technologies b. Continuous wave (cw) and long pulse (lp) operation modes c. Limitations of DF for the XFEL linac

2. Cw and lp operation; 3 experiments a. Dynamic heat load vs. DF b. Dynamic heat load vs. Eacc c. Dynamic heat load for Large Grain Cryomodule

3. Final remarks a. Possible upgrade scenario b. Subsystems for cw/lp operation c. Summary

2 J. Sekutowicz, Towards More Flexibility in Operation of the XFEL Linac, DESY, October 25th, 2013

XFELs are based on LC technologies and therefore they have low Duty Factors (DF) and limited flexibility in time structure of the electron and photon beam

Facility LC-project & technology

f [GHz]

Energy [GeV]

RF-pulses Max DF

[%] Length [µs] Rep. Rate [Hz]

LCLS SLC nc 2.8 14.7 3 120 <0.04

SACLA JLC nc 5.7 8 3 60 <0.02

Swiss-FEL JLC nc 5.7 5.8 3 100 <0.03

XFEL TESLA sc 1.3 17.5 1300 10 1.30

3

1. Introduction and Motivation: a. XFELs and their technologies

The XFEL facility, because it is driven by sc linac, has potential for substantially larger DFs and more flexibility.

J. Sekutowicz, Towards More Flexibility in Operation of the XFEL Linac, DESY, October 25th, 2013


Layout of the XFEL sc linac

E = 120 MeV, Slice energy spread: 50 keV Slice emittance: εx,y~1µrad σx,y < 0.8 mm

E= 17.5 GeV Slice energy spread: 1 MeV

Slice emittance: εx,y~1.4 µrad σx,y < 30 µm

101 cryomodules in total: 17 in the injector section (IS) and 84 in the main linac Each cryomodule houses eight 1.3 GHz 9-cell sc TESLA cavities.

4

σs= 2 mm

Ipeak= 50 A

σs= 0.02 mm

Ipeak= 5 kA


Inje

ctor

lina

c

4 cr

yom

odul

es

3rd

harm

onic

RF

sect

ion

Boos

ter L

inac

12 c

ryom

odul

es

Undulators

BC2

E=2 GeV

e- g

un

E=5MeV

BC1

Time structure of the XFEL nominal electron beam (sp, short pulse operation)

Nominal Energy GeV 17.5

Beam pulse length ms 0.65

Repetition rate Hz 10

Max. # of bunches per pulse 2700

Min. bunch spacing ns 240

Bunch charge nC ≤ 1

5



600 µs

100 ms 240 ns

~100 fs

XFEL will generate high average brilliance photon beam at very short λ:

1.6·1025 photons/0.1%bw/s/mm2/mrad2 @ 1 Å

Two additional features could make the XFEL photon beam even more attractive

An increased, to several µ-seconds, intra pulse distance between bunches, which will allow for less technically demanding detectors

A few kHz photon burst repetition rate, which will allow to use less expensive optical lasers (few kHz rep.-rate) for pump-and-probe experiments

For the nominal beam, these features will lead to substantially reduced number of bunches/s and significantly lower average brilliance.

6



1. Introduction and Motivation: b. Continuous wave (cw) and long pulse (lp) operation modes

We proposed (several years ago) additional operation modes;

Continuous wave (cw) operation for Eacc ≤ 7 MV/m Long pulse (lp) operation at Eacc > 7 MV/m with DF ≤ (7/Eacc)2 The bunch quality should be as high as for the nominal sp operation

7

Example of lp operation at Eacc = 14 MV/m with DF up to 25%


≥ 750 ms frep = 1Hz

~100 fs

1s

≤ 250 ms

4 µs, up to 0.5 nc

≥ 100 ms

~100 fs

≤ 33 ms

4 µs, up to 0.5 nc

The main XFEL linac will be split in 7 cryogenic-strings (CS = 12 cryomodules)

2.2 K forward 5 K forward 40 K forward

2-phase tube: 240 W (~20W/cryomodule) is the estimated limit for one CS

2 K, Gas Return Tube

80 K, Return 8 K, Return

1st limit: Heat load at 2K for each cryomodule should not exceed ca. 20 W

1. Introduction and Motivation: c. Limitations of DF for the XFEL linac

What are the technical and “practical” limits for DF ?


TESLA cavity was designed for ca. 1% DF (1992)

End groups (FPC and HOM couplers) are cooled by means of heat conduction

HOM coupler HOM coupler

FPC

9


2nd limit: Heating of the HOM couplers must not cause quenching of the end-groups


HOM antenna is exposed to the residual magnetic field of the FM

To reduce antennae heating all XFEL cavities will be equipped with high thermal conduction feedthroughs connected with copper braids directly to the 2-phase tube.

10

1. Introduction and Motivation: c. Limitations of DF for the XFEL linac 2nd limit: cont.,


Ti

Cu

Nb (RRR 400)

Sapphire

Mo

11

1. Introduction and Motivation: c. Limitations of DF for the XFEL linac 2nd limit: cont.,


High thermal conduction HOM feedthroughs with sapphire windows

HOM feedthrough with cable, thermal connection and load

Ti

Cu

Nb (RRR 400)

Sapphire

Mo

Other, “practical” limits

12



Existing Cryo-Plant High Duty Factor Operation

Capacity 2450 [W] (T=2K) 4980 [W] (T=1.8K)

3th limit: An upgrade of the cryogenic plant should be “doable”

Upgraded cryo-plant will have capacity of one unit at JLab

4th limit: New RF-sources will be added to klystrons used for the sp operation They should fit in the XFEL tunnel. We plan to have one RF-source/CM

IOT seems to be the best choice (at present): It is very compact, what makes it superior to solid-state amplifiers

It takes energy from the mains only when it generates RF-power. This makes IOT more efficient for the lp operation than a cw klystron

Other, “practical” limits, cont.

13



The second prototype will be delivered in fall 2013.

Parameter Unit Spec Measured (1stprototype)

Output P [kW] 120 85

Gain [dB] >21 22.3

Efficiency [%] >60 54

The first prototype of the IOT was built by CPI and delivered to DESY in 2009.

Goal: Measurement of dynamic heat load (DHL) vs. DF Technical conditions:

1. < Qload > = 1.5E7, 3 dB resonance width 87 Hz 2. Eacc= 5.6 MV/m 3. Only 3 cavities had new feedthroughs and new thermal connections 4. The measurements were done for DF = 38, 60, 75 and 100% (cw) 5. 7 cavities in operation (cavity #7 had detuned HOM coupler)

2. cw and lp operation; 3 experiments a. Dynamic heat load vs. DF

We equipped in 2011 Cryo-Module Test Bed at DESY with the IOT prototype Since May 2011 we tested 5 pre-series cryomodules in cw and lp modes

14

I. Example of the lp/cw run at 2K (cryomodule PXFEL3_1, June 2012)


Experiments with pre-series CMs at 1.8 and 2K (3 examples)

Measured dynamic heat load


0

5

10

15

20

25

12 AM 4 AM 9 AM 2 PM 7 PM 12 AM 4 AM

DHL [W]

Time [h]

DF=

75

%

DH

L=12

.4 W

Eacc

=0 M

V/m

DH

L=4W

DF=

60

%

DH

L=9.

7 W

DF=

38

%

DH

L=5.

2 W

cw

DH

L=16

.5 W

Eacc

=0 M

V/m

DH

L=5W

2. cw and lp operation; 3 experiments a. Dynamic heat load vs. DF

DF [%] 38 60 75 100

Estimated DHL [W] 4.2 6.5 8.3 10.7

Additional Heat / End group [W] 0.07 0.23 0.29 0.41

Conclusion: Following operations of the XFEL linac should be possible for 20W limit: cw operation at Eacc ≤ 6.1 MV/m ………. lp with DFs of ca. 38% at Eacc = 11.5 MV/m


11.5

8.1 7.1

6.1

0

4

8

12

0

8

16

24

20 40 60 80 100

Eacc [MV/m]

THL [W]

DF [%]

THL= DHL at 1.8K + static 4.5WMax Eacc for 20W THL budget

Eacc vs. DF at 1.8 K (scaled from test I) 2. cw and lp operation; 3 experiments a. Dynamic heat load vs. DF

Goal: Measurement of DHL vs. Eacc at 2K Technical conditions:

1. < Qload > = 1.5E7, 3 dB resonance width 87 Hz

2. Only 2 cavities had new feedthroughs and new thermal connections

3. The measurements were done for DF = 20%

4. Eacc = 7.4, 9.0, 9.8, 10.5 and 10.7 MV/m

17

II. Example of the lp run (cryomodule PXFEL2_3, May 2013)



2. cw and lp operation; 3 experiments b. Dynamic heat load vs. Eacc



0

4

8

12

16

20

10 AM 12 PM 2 PM 5 PM 7 PM 10 PM 12 AM 2 AM

DHL [W]

Time [h]

E=7.4 MV/m DHL = 4.6 W

Eacc=0 MV/m DHL= 0.5W

E=9.0 MV/m DHL = 8.5 W

E=10.5 MV/m DHL=14.5 W

E=9.8 MV/m DHL= 11 W

E=10.7 MV/m DHL=16 W

Eacc [MV/m] 7.4 9 9.8 10.5 10.7

Estimated . DHL [W] 4.5 6.8 7.8 9.3 9.7

Additional Heat / End group [W] 0.01 0.1 0.2 0.32 0.39



20

40

60

80

100

16

18

20

22

24

6 7 8 9 10 11 12

DF [%]

THL [W]

Eacc [MV/m]

THL at 1.8KDF at 1.8K

DF vs. Eacc at 1.8 K (scaled from test II)

Conclusion: Following operations of the XFEL linac should be possible for 20W limit: cw operation for Eacc ≤ 7.8 MV/m ………. lp with DFs of ca. 24 % at Eacc = 11 MV/m.

Goal: Measurement of the DHL at 2K and 1.8K vs. Eacc and DF

Technical conditions:

1. < Qload > = 1.5E7, 3 dB resonance width 87 Hz

2. All cavities had new HOM feedthroughs and new thermal connections

3. 7 cavities had cells (not end-groups) made of LG-niobium, 1 of FG-niobium

4. DF range {13% - 100%}

5. Eacc range {5.7 MV/m - 14.7 MV/m}

20

III. Example of the cw/lp run with LG cryomodule (XM-3, Sept. 2013)


2. cw and lp operation; 3 experiments c. Dynamic heat load for Large Grain Cryomodule

1.E+10

2.E+10

3.E+10

4.E+10

0 4 8 12 16

Qo

Eacc [MV/m]

AC114 AC156 AC146 AC154AC157 AC158 AC151 AC152


XM-3 cavities: vertical tests at 2K (7 Large Grain +1 Fine Grain cavity)





0

4

8

12

16

20

11.09.13 12.09.13 13.09.13 14.09.13 15.09.13 16.09.13

DHL [W]

Date

Operation at 1.8 K Operation at 2 K

We performed 23 tests with that cryomodule varying DF, T and Eacc

0

4

8

12

16

20

0:00 2:30 5:00 7:30 10:00 12:30 15:00 17:30

DHL [W]

Time [h]

Eacc = 7.04 MV/m DHL = 8.9 W

Eacc = 9.5 MV/m DHL = 16.1 W



Test examples: cw operation at 1.8K and 2K for Eacc 7 MV/m and 9.5 MV/m

Eacc= 0 MV/m DHL=0W

Eacc = 7.04 MV/m DHL = 13.5 W

Operation at 2 K Operation at 1.8 K

0

4

8

12

16

9:00 12:00 15:00 18:00 21:00

DHL [W]

Time [h]



Test examples, cont.: lp operation at 1.8K and Eacc = 14.7 MV/m

Eacc= 0 MV/m DHL=0.2 W

Eacc = 14.7 MV/m DF = 25%

DHL = 10.7 W



Conclusion from run III : DF vs. Eacc for T = 1.8 K

37.4%

0

20

40

60

80

100

0 3 6 9 12 15

DF [%]

Eacc [MV/m]

DF for 20 W THL

DF demonstrated



Comparison of cryomodules: cw operation at Eacc = 5.7 MV/m

PXFEL3_1 PXFEL2_3 XM-3 XM-3

Type of Nb FG FG LG LG # of cavities in operation 7 8 8 8 # of cavities with new HOM output 3 2 8 8 T [K] 2 2 2 1.8 Measured DHL [W] 19.5* 18.7 11.5 5.1 Ratio: M-DHL / (M-DHL of XM-3 at 2 K) 1.69 1.63 1 0.44 *scaled to 8 cavities

One must note here that we have not tested yet standard XFEL-type CM housing 8 FG cavities, all equipped with new HOM output lines.

To ensure bunch quality for cw/lp operations gradients in the injector section (IS) have to be as for the nominal sp operation

27

Gradients in the IS are > 7 MV/m, so we will need 136 low current 9-cell TESLA-like cavities operating cw at gradients up to 16 MV/m.

3. Final remarks a. Possible upgrade scenario


Inje

ctor

lina

c:

4 CM

E ac

c = 1

1 M

V/m

Boos

ter L

inac

: 12

CM

E ac

c = 1

5 M

V/m

E= 2 GeV

Inje

ctor

CM

E ac

c = 1

6 M

V/m

IS

BC2

e- g

un

BC1

Averaging results from the presented tests with FG pre-series cryomodules and assuming that

the injector section will be equipped with 136 new cw-cavities 12 out of 17 standard CMs from IS will be relocated to the end of ML

one can estimate end beam energies and DFs.

28

Operation mode Energy [GeV]

Eacc ML [MV/m]

Max DF [%]

sp 17.5 1.3 cw 7.6 7 100 lp 10 10 -> 35 lp 14 15 -> 15

Beam Energies and DFs for the scenario with new IS and 768 cavities in ML


3. Final remarks a. Possible upgrade scenario

We may expect these numbers to be higher for the standard XFEL cryomodules, because they will have better cooling of end-groups.

R&D program at DESY for cw-operating TESLA like cavity.


3. Final remarks b. Subsystems for cw/lp operation

For the modified HOM coupler, H is lower by 50%, antenna is shorter, F-part has better shape for cooling.

Modeling of HOM-coupler with hidden antenna

Cour

tesy

D.

Kost

in, D

ESY

H= 0.022 A/m Peak H field on the antenna for the same stored energy

Antenna is “hidden” in the output tube

H= 0.045 A/m

R&D program at DESY for cw-operating TESLA like cavity, cont.


Copper model of 9-cell cw-cavity is almost ready to test HOM damping

Nb-prototype is scheduled for next year

Should we use LG niobium?




1.E+10

2.E+10

3.E+10

4.E+10

0 4 8 12 16

Qo

Eacc [MV/m]

AC112 AC113 AC115 AC117 AC122 Z93Z100 Z104 AC155 AC153 AC154 AC158

LG vs. FG comparison of vertical tests at 2.0 K for cavities chemically treated at DESY

<Qo> = 1.7·1010

<Qo> = 2.2·1010



LG Vertical test at 1.8 K

2.E+10

4.E+10

6.E+10

8.E+10

0 4 8 12 16

Qo

Eacc [MV/m]

AC154 AC155 AC153 AC158

<Qo> = 4.1·1010



What is the benefit to use high Q cavities? Assumptions:

Linac has 140 cavities operating at 15 MV/m

7000 h operation per year

0.20 € for 1 kWh

10 years of operation




What is the benefit to use high Q cavities? Assumptions:

Linac has 140 cavities operating at 15 MV/m

7000 h operation per year

0.20 € for 1 kWh

10 years of operation


1.5E10; 24.7M€

1.7E10; 21.8M€

2.2E10; 16.8M€

2.8E10; 13.2M€

4.1E10; 9.0M€

0

10

20

30

1.E+10 2.E+10 3.E+10 4.E+10 5.E+10

Elect. Bill for Cryo of IS

[M€]

Qo



9 XF

EL L

G c

aviti

es

at 2

K, P

rep.

DES

Y

9 XFEL LG cavities at 1.8 K, Prep. DESY

24 XFEL FG cavities at 1.8 K, Prep. Indust.

Scaled !

24 X

FEL

FG

cav

ities

at

2 K

, Pre

p. In

dust

.

6 XF

EL F

G c

aviti

es

at 2

K, P

rep.

DES

Y

Seems that LG is much more attractive and can save operation costs, but…. Very recent results at Cornell and HZB showed that with a proper cool-down procedure one can enhance Qo of FG cavities by factor of 2. How should we proceed:

At first, we need to test the Cornell/HZB cool-down recipe to XM-3 cryomodule, to check if for LG material there is still some Qo improvement

Then, we should test FG cavities (AMTF) and 2-3 cryomodules, housing FG

cavities to verify the new cool-down procedure for the XFEL structures.

Having results of these experiments we will be able to make the decision on type of niobium.






Other R&D programs towards cw/lp operation 1. All superconducting electron gun (with sc cathode) needs more attention.

Open questions: How to recover at 2K QE of the sc cathode demonstrated at 300K? How to integrate better cathode plug in the clean sc cavity?

2. 3-rd harmonic cavities operating cw will be developed in collaboration with FERMI.

1.E+09

1.E+10

1.E+11

0 20 40 60

Qo

Ecath [MV/m]

Baseline Test with Nb-plugTest with Pb-coated plug

Courtesy P. Kneisel, JLab

1. The experiments showed feasibility of more flexible operations for the XFEL

2. We have demonstrated:

Gradients for beam energy of 7.6 GeV in cw mode and 14 GeV in lp mode

Benefit of operation at 1.8 K Benefit to use LG material for the standard cool-down procedure

3. We need to test several SERIAL XFEL CRYOMODULES to determine more precisely relation: DF vs. Eacc for the main linac.

4. Following R&D programs are in “slow” progress: For new SRF electron source For cw-operating low current cavities For new RF-sources For LLRF stabilizing accelerating gradient for cw and lp operation


3. Final remarks c. Summary

Acknowledgements

I want to express my gratitude to all Colleagues contributing to the experiments and R&D programs:

DESY: V. Ayvazyan, J. Branlard, M. Ebert, J. Eschke, T. Feldmann, A. Gössel, D. Klinke, D. Kostin, M. Kudla, F. Mittag, W. Merz, W.-D. Möller, C. Müller, R. Onken, D. Reschke, I. Sandvoss, X. Singer, W. Singer, A. Sulimov

TUL: W. Cichalewski, A. Piotrowski, K. Przygoda

EICS GmbH: W. Jalmuzna

WUT: K. Czuba

NCBJ: M. Barlach, J. Lorkiewicz, R. Nietubyc, J. Szewinski,

JLab: G. Ciovati, P. Kneisel,

BNL: T. Rao, J. Smedley


Thank you for your attention

towards more flexibility in operation of the xfel linac - desy

Documents