ARTICLE IN PRESS

0168-9002/$ - se

doi:10.1016/j.ni

�CorrespondE-mail addr

Nuclear Instruments and Methods in Physics Research A 566 (2006) 250–255

www.elsevier.com/locate/nima

Layout of bunch compressor for Beijing XFEL test facility

Xiongwei Zhua,�, Yingchao Dub, Xiaozhong Heb, Yufeng Yanga

aInstitute of High Energy Physics, Chinese Academy of Sciences, Accelerator Center, 19B Yuquan Road, Beijing 100049, ChinabAccelerator Lab., Tsinghua Univesity, Beijing 100084, China

Received 13 May 2006; received in revised form 4 July 2006; accepted 6 July 2006

Available online 31 July 2006

Abstract

In this paper, we describe the layout of the bunch compressor for the Beijing XFEL test facility (BTF). Our bunch compressor setup is

different from the usual one due to the space limit. The compensation X-BAND cavity and the first bunch compressor are separate in

distance. The electron bunch is decelerated first and then accelerated to enter the first bunch compressor. The simulation result shows

that our setup works well, and the nonlinear term is well compensated. Also, we present the result about the CSR emittance dilution

study. Finally, we develop a program to study microbunch instability in the second BTF bunch compressor.

r 2006 Elsevier B.V. All rights reserved.

PACS: 29.27.Bd; 41.75.�i; 41.60.Cr

Keywords: Bunch compressor; Beam dynamics; XFEL

1. Introduction

The Beijing X-ray free electron laser test facility (BTF)[1] is proposed to reside in the BEPCII linac which is thebest electron linear accelerator in China. BTF will not usethe existing DC gun preinjector, but will use photoinjectorto obliquely inject electron bunch to the main linearaccelerator of BEPCII through low energy transport lineDL1. At the same time, three accelerator sections will beremoved in order to install two magnetic bunch compres-sors (BC1, BC2) to compress the electron bunch. After theaccelerator section A46, the beam is extracted and injectedto the undulator through high-energy transport line DL2.BTF will provide the electron beam with the energy of1.2GeV, the energy spread of 0.15%, the normalizedemittance less than 2.5mmmrad, and the peak current of600A. Fig. 1 is the schematic layout of BTF, the two bunchcompressors divide the main linac into three sections: L1,L2 and L3. In particular, the compensation X-BANDcavity is placed immediately after the photoinjector andbefore DL1, since there is no available space in the mainlinac.

e front matter r 2006 Elsevier B.V. All rights reserved.

ma.2006.07.010

ing author. Tel.: +8610 88236505; fax: +86 10 88236190

ess: zhuxw@mail.ihep.ac.cn (X. Zhu).

2. Bunch compressor and beam dynamics

In order to obtain the BTF accelerator output para-meter, we need to choose the parameters of each section ofthe linac to get the correct accelerating and compressing.The parameters which we can change include the strengthof the bunch compressor, the energy of the bunchcompressor, and the accelerating phases of each linacsection. The optimized parameters of BTF accelerator areshown in Table 1.The two bunch compressors of BTF adopt the C type

four-dipole chicanes, and the main parameters are pre-sented in Tables 2 and 3.With the above parameters, we have done start-to-end

simulation. The photoinjector is composed of 1.6 cellsBNL-type RF gun and two S-BAND accelerating structuresections. The beam parameters from the photoinjector arethe energy: 153MeV, the relative energy spread: 0.14%,normalized horizontal emittance: 1.23mmmrad. The beamparameters after the high-energy transport line DL2 are theenergy: 1.2GeV, the relative energy spread: 0.12%, and thenormalized horizontal emittance 1.42mmmrad. BTF FELis proposed to adopt two-stage cascading HGHG method.With the beam parameters from our start-to-end simula-tion of BTF accelerator, the FEL output power is 1GW at

ARTICLE IN PRESS

GUN

Pamela Elegant

A0-A2 A3-A9 A11-A24 A27-A46

L1DL 1 BC 1 BC 2 L3L2

DL 2

TDA

UNDULATORX

RFGUN

NEW Bunch Compressores

New Injector Existing BEPCII Linac Undulators

A1'-A2'

Fig. 1. Schematic layout of BTF.

Table 2

BC1 parameters

Parameter Value Unit

Energy 0.344 GeV

Energy spread (rms) 1.53 %

Compressing ratio (rms) 2

R56 �31.5 mm

Total length 3.05 m

Project distance between the first and second

bend magnet

0.85 m

Project distance between the second and third

bend magnet

0.55 m

Bend length 0.2 m

Deflection angle 7.28 degree

Emittance dilution 7 %

Table 1

The optimized parameter of each section

Ein (GeV) Eout (GeV) sz-in (mm) sz-out (mm) sd-in (%) sd-out (%) frf (1) R56 (mm)

L1 0.136 0.344 0.87 0.87 0.2 1.53 �29 —

BC1 0.344 0.344 0.87 0.34 1.53 1.53 — �31.5

L2 0.344 0.700 0.34 0.34 1.53 1.11 �29.8 —

BC2 0.700 0.700 0.34 0.14 1.11 1.11 — �15

L3 0.700 1.18 0.14 0.14 1.11 0.12 40 —

Table 3

BC2 parameter

Parameter Value Unit

Energy 0.70 GeV

Energy spread (rms) 1.11 %

Compressing ratio (rms) 3.5

R56 �15 mm

Total length 7.77 m

Project distance between the first and second

bend magnet

3.2 m

Project distance between the second and third

bend magnet

0.17 m

Bend length 0.2 m

Deflection angle 3.6 degree

Emittance dilution 20 %

X. Zhu et al. / Nuclear Instruments and Methods in Physics Research A 566 (2006) 250–255 251

the wavelength of 45 nm and 0.41GW at the wavelengthof 9 nm.

3. X-BAND compensation cavity

In the electron longitudinal phase space, the relation ofenergy and time is not completely linear, the high ordernonlinear terms are incorporated due to the differentaccelerating phases, and the main contribution is from thesecond-order term. When the bunch is compressed, thesenonlinear terms can lead to a serious nonuniform oflongitudinal phase space, and the locally high peak currentwill occur to strengthen the CSR effect. Therefore, theemittance dilutes and the local energy spread deteriorates.Using a X-BAND harmonic cavity, the second-order termeffect can be compensated. In BTF, there is no availablespace to place the X-BAND cavity in the main linac, so weplace the cavity immediately after the photoinjector andbefore the low-energy transport line. The X-BAND cavityand the first bunch compressor are placed separately. Oursetup is different from the usual compensation method.The electron bunch is decelerated first and then accelerated.The simulation shows that this layout also works well, andthe second-order term is compensated effectively. In oursetup, the compensation cavity decelerating voltage is alsogiven as [2]:

eVx ¼E0 1� ð1=2p2Þðl2sT566=R3

56Þð1� sz=sz0Þ2

� �� Ei

ðls=lxÞ2� 1

.

In BTF design, ls=lx ¼ 4, R56 ¼ �31.5mm, T566E47.25mm, Ei ¼ 324MeV, E0 ¼ 7MeV, sz0 ¼ 870 mm and

ARTICLE IN PRESS

Fig. 2. The longitudinal phase space distribution without X-BAND cavity.

X. Zhu et al. / Nuclear Instruments and Methods in Physics Research A 566 (2006) 250–255252

sz ¼ 336 mm, the decelerating voltage is Vx ¼ 16.9MV, thecorresponding gradient is 28.2MV/m (the length ofcompensation cavity is 0.6m). Fig. 2 gives the ELEGANT[3] simulation result without the X-BAND cavity. Theabove figure shows the longitudinal phase space after thefirst bunch compressor. The following figure shows thelongitudinal phase space after the second bunch compres-sor. About 3 kA peak current occurs in longitudinal phasespace. When the beam goes through the second-bunchcompressor, the normalized emittance dilutes from 1.4 to1.82mmmrad. Fig. 3 gives the simulation result with thecompensation cavity. After the compensation, the bunchbecomes uniform. There is no abnormal peak current, andthe emittance dilution caused by CSR is very small.

4. Emittance dilution

When a short bunch goes through bent trajectories,coherent synchrotron radiation (CSR) occurs to deterio-rate the emittance. There are several simulation codes to

simulate this effect. We use TraFiC [4] to calculate theemittance dilution on BC1 and BC2. The initial parametersfor bunch and lattice are from the ELEGANT simulationresult. The TraFiC [4] simulation result is shown in Figs. 4and 5, which agrees well with the ELEGANT result.

5. Jitter effect

Although the two stage bunch compressors are opti-mized, they are still sensitive to the jitters of acceleratingphase, voltage, and bunch charge. Table 4 gives thesensitivities of RF phase and voltage, with each sensitivityleading to 15% increase of peak current or 0.1% increaseof energy spread. In addition, the sensitivity of bunchcharge is 2%.

6. Microbunch instability

A high brightness electron beam with a small amount ofdensity modulation can create longitudinal self-fields that

ARTICLE IN PRESS

Fig. 3. The longitudinal phase space distribution with X-BAND cavity.

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5

0.00000

0.00001

0.00002

0.00003

0.00004

0.00005

0.00006

γεx,

y (m

rad

)

s(m)

xy

Fig. 4. Normalized emittance evolution on BC1.

0 4 8

0.00000

0.00005

0.00010

0.00015

0.00020

0.00025

0.00030

0.00035

γεx,

y (m

rad

)

s(m)

2 6

xy

Fig. 5. Normalized emittance evolution on BC2.

X. Zhu et al. / Nuclear Instruments and Methods in Physics Research A 566 (2006) 250–255 253

lead to beam energy modulation. Since bunch compressorintroduces path length dependence on energy, the inducedenergy modulation is then converted to additional densitymodulation that can be much larger than the initial density

modulation. The amplification process is accompanied by agrowth of energy modulation and growth of emittance.This is the mechanism for microbunch instability. We havedeveloped a program based on Heifet’s theory [5] to study

ARTICLE IN PRESS

Table 4

The sensitivities of phase and voltage

Parameter Sign DI=I0 ¼ 15% ðDE=E ¼ 0:1%Þ Unit

Mean L1 phase j1 0.15 degree

Mean L2 phase j2 0.15 degree

Mean L3 phase j1 0.15 degree

Mean L1 voltage DV1/V1 0.2 %

Mean L2 voltage DV2/V2 0.2 %

Mean L3 voltage DV3/V3 0.2 %

-20 0 20 40 60 80 100 120 140 160 180 200 220-0.2

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

Gf

λ(µm)

Gf

Fig. 6. Amplification factor Gf dependence on wavelength l.

Fig. 7. Amplification factor Gf dependence on b function.

Fig. 8. Amplification factor Gf dependence on a function.

X. Zhu et al. / Nuclear Instruments and Methods in Physics Research A 566 (2006) 250–255254

the microbunch instability in bunch compressor. Theprogram algorithm is optimized, and the running time isless than several minutes. Using our program, we study themicrobunch instability in BC2, and find that the mostharmful wavelength for microbunch instability is about90 mm, see Fig. 6.

We also study the effect of the twiss parameter at theentrance of BC2. At wavelength 50 mm, we change theentrance parameters, and then simulate the instability. The

result shows that the microbunch instability is sensitive tothe entrance twiss parameter. Figs. 7 and 8 give thedependence of Amplification factor Gf on initial twissparameter (a0, b0) at wavelength 50 mm.

7. Current chirp from the photoinjector

The X-BAND cavity is designed to compensate thesecond nonlinear term accumulated in L0, but it cannotcompensate the current chirp from the photoinjector. Thischirp can lead to anomoulous spikes in the bunchcompressor output that can lead to emittance dilutiondue to the CSR effect or the wakefield effect. In oursimulation, we also consider the current chirp from theinjector.The simulation result is showed in Fig. 9. We add a 70A

or so current chirp in the injector output, and get about2 kA current spike at the output of the BTF accelerator.The horizontal normalized emittance dilute from 1.23 to1.51 and about 0.1mmmrad contribution comes from thecurrent spike. This contribution is smaller than that fromthe nonlinear term accumulated on the L0 linac section.

8. Discussion

In this paper, we describe the design study of bunchcompressor for the BTF. Our bunch compressor setup isdifferent from the usual one due to the space limit. Thecompensation X-BAND cavity and the first bunchcompressor are separate in distance. The electron bunchis decelerated first and then accelerated to enter the firstbunch compressor. The simulation result shows that oursetup works well, and the nonlinear term is well compen-sated. Also, we present the result about the CSR emittancedilution by using TraFiC [4], which agrees well with theELEGANT simulation result. Finally, we develop aprogram to study the microbunch instability in the secondBTF bunch compressor (BC2), where the microbunchinstability is sensitive to the entrance twiss parameter.

ARTICLE IN PRESS

Fig. 9. The longitudinal phase space distribution with current chirp from the injector.

X. Zhu et al. / Nuclear Instruments and Methods in Physics Research A 566 (2006) 250–255 255

Acknowledgement

The work is supported by NSFC (10575114). Thanks goto the BTF study group and Prof. J. Gao.

References

[1] X. Zhu, et al., High Energy Phys. Nucl. Phys. 30 (Suppl. I) (2006) 126.

[2] Linac Coherent Light Source (LCLS) Design Report, SLAC-R-593,

2002.

[3] M. Borland, Adv. Photon Source LS 287 (2000).

[4] A. Kabel, M. Dohlus, T. Limberg, SLAC-PUB-8559, 2000.

[5] S. Heifet, Phys. Rev. ST-AB 5 (2002) 064401.