PSI Scientific Report 2008

PSI-XFEL  9

or semiconductor surface by means of a laser beam. The

higher energies will be arranged at the tail of the bunch and

cathode is placed on the axis of a 2½ cell, 3 GHz accelerating

particles with lower energies at the head of the bunch. Due

cavity, which immediately accelerates the electron bunch after

to the nonlinearity of the 3 GHz accelerating field, the energy

extraction from the cathode. Solenoid and quadrupole mag-

chirp is slightly too large in the head of the bunch and too

nets in the subsequent structure focus the beam, to minimize

small in the tail. Therefore an X-band (12 GHz) cavity is intro-

the emittance at the exit of the gun complex.

duced before the bunch compressor to compensate for these

For the second gun, several options are possible. The decision

deviations.

on which will be based on the success of ongoing R&D work.

The bunch compressor (BC1) consists of a sequence of four

It will either be a newly-developed photo-electron gun or an

bending magnets, which create an orbit bump around the

alternative gun based on field emission arrays, where electrons

straight motion path in the linac. Since particles with higher

are extracted from a surface by means of high electric field

energies are subject to a smaller deflection in the magnets,

gradients (~ 5 GV/m). Such high gradients can be easily

their orbit lengths are shortened. They are consequently

achieved if the field is applied to micro- or nano-structured

moved from the tail towards the centre of the bunch. Simi-

surfaces where the field is strongly enhanced around tips with

larly, the lower-energy particles at the head of the bunch ex-

small apex radii. In order to mitigate space charge effects, the

perience larger deflections that result in a lengthening of the

energy of the beam is rapidly increased by passing the beam

orbit and a transition towards the bunch centre. The net effect

through a high-voltage and high-gradient diode configuration,

after BC1 is that the length of the bunch is reduced from 10 ps

before entering the first RF accelerating structure. A newly-

(for the 200 pC mode) to 450 fs.

developed high voltage pulser is currently being tested and

The subsequent Linac 2 (with the same cell structure as

further developed. Different surface materials are being

Linac 1) raises the energy to 2.1 GeV. At this point, the second

explored, to discover those which can sustain high surface

magnetic bunch compressor (BC2) is introduced, which re-

gradients without breakdown. Since this concept relies on a

duces the bunch duration to 30 fs, with a corresponding in-

longer initial pulse (40 ps fwhm with 5.5 A peak current), a

crease of the peak current to 2.7 kA. For the succeeding Linac 3,

higher compression is required to reach a sufficiently high

the transverse beam dimensions are already considerably

peak current at the entrance to the undulator. This compres-

smaller, due to the increased beam energy, permitting the

sion starts in the first accelerating cavity, which is fed by two

distance between the focusing quadrupoles to be increased.

frequencies (1.5 GHz and 4.5 GHz). In this way, the longitudi-

One cell here is constructed from four two-metre-long accel-

nal energy distribution in the beam can be suitably shaped to

erating sections between two adjacent quadrupoles, and has

reach a very effective velocity compression. In the low relativ-

a total length of 19 m.

istic regime, particles with different energies still have a

After Linac 3, the electron beam is extracted for the longer-

notable difference in velocities. If they are arranged properly

wavelength FEL lines Athos and Porthos. The nominal energy

in energy along the bunch, they move towards the bunch

at this point is 3.4 GeV, but will be reduced to 2.1 GeV for

centre, and the length is reduced.

Athos by not powering Linac 3. It remains to be verified by

After the gun complex, the bunch can be directed into a diag-

simulations whether the focusing lattice can remain un-

nostic line for complete characterization. A more conven-

changed, since the quadrupole strengths are matched to a

tional accelerating structure follows the gun and comprises

higher energy, otherwise a second extraction point after Linac 2

four S-band structures of 4 m length, surrounded by focusing

will need to be inserted.

solenoids. The maximum accelerating gradient is 20 MV/m.

Only for the 1 Ă…ngstrom wavelength of Aramis is an addi-

In the test setup for this injector presently under construction,

tional boost to 5.8 GeV required, provided by Linac 4, which

a bunch compressor will be placed at the end (250 MeV) for

uses the same cell structure as Linac 3.

test purposes. In the final layout, an additional accelerating

The electron beam quality is now sufficient for the lasing

section will be added (Linac 1) in front of the bunch compres-

process as the beam enters the undulators. The emittance is

sor, boosting the energy to 450 MeV. The higher energy will

reduced by adiabatic damping, and the bunch is longitudi-

alleviate the risk of emittance dilution due to space charge

nally compressed.

effects in the bunch compressor. Linac 1 comprises two FODO

In principle, an electron transversally accelerated in a mag-

cells, each of 10 m length, with two accelerating structures of

netic field emits a broad spectrum of radiation. However, in

2 m length between adjacent quadrupoles. One cell will pro-

an undulator the only wavelengths not to be eliminated by

vide an energy increase of 120 MeV on crest, corresponding

interference effects are those for which the electron beam lags

to an accelerating gradient of 30 MV/m.

behind the photon beam by one wavelength (or an odd integer

During the acceleration process prior to the bunch compressor,

multiple). Due to the long undulator structure, the intensity

an energy chirp will be introduced in the beam. Particles with

of the radiation steadily increases and becomes sufficiently