Proceedings of the 1999 Particle Accelerator Conference, New York, 1999

A MEASUREMENT OF HIGH GAIN SASE FEL INDUCED

ELECTRON BEAM MICRO-BUNCHING USING
COHERENT TRANSITION RADIATION*
A. Tremaine#, S. Anderson, P. Frigola, M. Hogan, A. Murokh, C. Pellegrini, J.Rosenzweig
University of California, Los Angeles,
405 Hilgard Ave., Los Angeles, CA., 90095

D.Nguyen, R.Sheffield
Los Alamos National Laboratory,
Los Alamos, NM, 87545, USA

Abstract studying forward emitted CTR, scattering effects in the

foil will be shown to cause a significant degradation in the
Coherent transition radiation (CTR) was used to study the
longitudinal modulations of an electron beam exiting the emitted signal.
UCLA/LANL high gain SASE FEL. The induced
longitudinal micro-bunching of the electron beam at the 2 BACKGROUND
exit of the undulator was measured with a frequency This section reviews the theory of transition radiation
domain technique using the CTR emitted when this beam (TR) needed to understand the experimental measurements
strikes a thin conducting foil. Formalisms for both CTR and also to point out the assumptions made in the standard
and SASE theories are related using the simulation code
model that may not be entirely accurate for this and future
GINGER in which the SASE FEL gain of the output
radiation and the micro-bunching of the electron beam are experiments.
given. Experimental results from the CTR measurement The emitted coherent radiation energy spectrum from a
will show the limit of standard transition radiation (TR) multi-particle electron beam striking a metallic foil is
theory is being approached and new analysis is needed. given by

1 INTRODUCTION
d 2U 2
2 d U
Diagnostics measuring very short periodic electron beam ≈ N 2
f ω
( )
modulations will be necessary for future experiments in dΩdω dΩdω 1 e _
which the modulating wavelength will be several microns (1)
and less. Up to the present, time domain measurements
such as the streak camera and interfermetric CTR [1] have where N is the number of electrons in the bunch, Ω is
reliably measured the longitudinal structure of electron the solid angle, ω is the frequency of radiation, and
beams to a resolution of several hundred femtoseconds.
→
However, as advanced accelerating techniques [2] and  iωr ⋅ nˆ  → 3
FELS [3,4] are becoming more common, a dependable f (ω ) = ∫ exp  S( r )d r
 c 
means of measurement for these very short longitudinal
(2)
electron beam modulations are needed. Using the CTR
frequency domain technique described here, a higher
is the Fourier transform of the beam particle distribution,
resolution than the time domain measurements can be
S(r ). Immediately from Eq. 1, one sees the emitted CTR
achieved.
spectrum has the same Fourier spectrum as the electron
A SASE FEL was used to induce the longitudinal
beam distribution and any modulations in the electron
electron beam modulation and as this beam strikes a thin
beam will be seen in the emitted CTR spectrum.
conducting foil, the emitted CTR will give information
The single electron energy spectrum for transition
about the electron beam spatial distribution. For the
radiation (TR) is given by the familiar relation,
SASE FEL process, this electron beam micro-bunching is
directly related to the gain of the SASE radiation and
using CTR, we are able to reconstruct the beam d 2U e2 sin 2 θ
= .
distribution at the undulator exit. The results presented dΩdω
1 e_
4π 2 c (1 − β cos θ )2
here agree well with the predicted performance of the (3)
SASE FEL given by simulation. Since we will be
_______________________
*
Work supported by DOE contract DE-FG-92ER40693
#
Email: tremaine@physics.ucla.edu

0-7803-5573-3/99/$10.00@1999 IEEE. 217

 Proceedings of the 1999 Particle Accelerator Conference, New York, 1999

Eq. 3 is derived by modeling single electron TR as a It can be seen from Eq. 7 that the CTR energy depends
collision between the electron with its image charge at the heavily on having a highly focused electron beam at the
metal/vacuum boundary and using the Lienard-Weichert foil since UCTR ∝ 1 / σ r4 .
fields for moving charges. In addition, the frequencies of It should be noted that Eq. 6 was found integrating over
emitted radiation is assumed much smaller than the the solid angle, but in the next section we will see the
characteristic time for the collision to take place, angular acceptance of the optical beam was only
θ acc = 15mrad . Also, the beam must propagate through
tcoll << trad per (4) the foil to emit forward CTR (at the back surface of the
foil) and degradation of signal due to scattering effects in
and the assumption that ω → 0 in the Lienard-Weichert the foil needs to be included. Both of these effects will be
fields is used. However, it will be shown below this accounted for in the theoretical analysis of this
assumption’s limit is being approached in this and future experiment.
experiments and modifications to existing standard TR
modeling need to be made.
The electron beam distribution exiting a SASE FEL is 3 EXPERIMENTAL SETUP
given by [5,6]
The CTR/SASE experiments were performed at the
Advanced FEL (AFEL) at Los Alamos National
 r2   z2  Laboratory in which the experimental setup has been
exp − 2
exp − 2 described elsewhere [7], but is reviewed briefly here.
 2σ r   2σ z 
∑ [1 + bn cos(nkr z )]
→
S( r ) = Important experimental parameters are given in Table 1.
2πσ r2 2πσ z n
(5)
Table 1: Electron beam and SASE FEL
parameters.
where Gaussian distributions are assumed in the radial and
Beam Energy E 17.5 MeV
longitudinal dimensions ( r, z ) and the longitudinal micro-
bunching profile superimposed on the longitudinal Charge/bunch Q 1.2 nC
distribution is given by a co-sinusoidal term with the Bunch Length τ 9.2 ps
harmonic wavenumber, kr = 2π / λ r , where λ r is the (FWHM)
longitudinal electron beam micro-bunching wavelength λ
Wiggler period u 2 cm
equal to the fundamental SASE radiation wavelength.
Higher harmonic, n , wavelengths are driven by the SASE On axis field B0 7.4 kG
FEL process and are included in Eq. 4, but only the
λr 13 µm
fundamental harmonic ( n = 1) induced micro-bunching FEL Wavelength
could be measured in this experiment. RMS beam size σ 180 µm
Integrating Eq. 2 about the solid angle, Ω , gives a line r
spectrum

The AFEL photo-injector uses a 10.5 cell L-band

4
dU N 2 bn2 e 2  γ 
[ ]
standing wave accelerator running at 1300MHz. A
=  exp −( k − nkr ) σ z .
2 2
 modelocked (108MHz) diode pumped Nd:YLF is
dk 4π  krσ r 
compressed using a fiber/diffraction grating pair and then
(6) amplified with a pair of flashlamp pumped Yd:YLF rods.
The emitted pulse train has 350 individual pulses separated
Notice, there is a peak in the emitted CTR line spectrum by 9.23ns each with a FWHM of 9.2 ps. When the laser
at the micro-bunching frequency as expected from Eq. 2. pulse train illuminates the Cs 2 Te cathode, an electron
Each peak is very narrow compared to the separation with train with nearly the same parameters as the laser train is
the neighboring harmonic Gaussian if σ r >> λ r . The created and accelerated down the beamline with each
total energy of the emitted CTR is found integrating Eq. 5 electron bunch having a charge of 1.2nC. Solenoids are
to be placed near the cathode for emittance compensation and
before the undulator to match the electron beam to the
UCTR =
( Nb1e)2  γ 
4
proper SASE FEL conditions.
  .
4 πσ z  krσ r  The 2m undulator was built from a collaboration
between the Kurchatov Institute and UCLA [8] and has a
(7)
magnetic period λu = 2.06cm , on axis field B0 = 7.4 kG ,

218
 Proceedings of the 1999 Particle Accelerator Conference, New York, 1999

and a normalized undulator field K ≈ 1. An insertable to pass a reasonable CTR signal to the detector and an
6 µm radiating foil was placed 1 cm behind the last intrinsic resolution of .177µm is estimated for the
undulator period. When inserted, the foil reflects all the modified monochromator setting. First, the SASE
SASE radiation (skin depth <50 nm) and the only light to radiation was scanned with the bandwidth broadened
continue down the optical beamline to the calibrated monochromator and results of this measurement are
HgCdTe detector is the forward emitted CTR. When the shown in Fig.1. A centroid at 12.8µm is seen in the
radiating foil is retracted, only the SASE radiation will SASE spectrum and as mentioned before, this is the
propagate to the detector. Since the SASE and CTR are at modulation wavelength of the induced electron beam
the same wavelength (see Eq. 6) and have the same source longitudinal micro-bunching, λ r . Next, the screen was
points, the end of the undulator, the collecting optics need inserted and the emitted CTR spectrum was scanned and
not be changed from the two measurements. the results are also shown in Fig. 1. As expected, the
The HgCdTe detector was placed about 3.5 m from the CTR spectrum is centered around nearly the same
source point which limited the angular collection of the wavelength as the SASE spectrum and is Gaussian in
optical beamline to just θ acc = 15mrad and Eq. 7 is not shape agreeing with Eq. 6. It should be mentioned the
entirely correct. To correct for this, numerical integrations CTR has been normalized to make it the same scale as the
of Eq. 1 are done out to θ acc . Also, θ acc forces collection SASE spectrum.
of the coherent transition radiation which is emitted at
θ coh = ( 2σ r kr ) −1 ≈ 8mrad and very little collection of
the incoherent light emitted at θ incoh ≈ 1 / γ = 28mrad .
Included in the numerical integration is the effect of
electron beam scattering within the foil. The forward
emitted CTR is derived from the electron beam
CTR (normalized), SASE (mV) 200 SASE
propagating through the foil and is emitted when the beam
travels from metal to vacuum at the foil back surface. CTR
Since the scattering angle is found to be θ scatt ≈ 8mrad ,
we find the transverse size of the electron beam ( σ r ) will
increase and the forward emitted CTR signal is degraded
(Eq. 1) by almost 40% compared to a signal assuming a 100
foil thickness of 0 µm .

4 MEASUREMENTS VS. THEORY

0
In order to accurately predict the expected emitted CTR, it 12 12.5 13 13.5 14
can be seen from Eq. 7 the bunching factor, b , needs to Wavelength ( µm)
be estimated. The bunching is predicted for these
Figure 1: CTR and SASE signals as a function of
conditions by the 3D FEL simulation code GINGER. For
wavelength with CTR scaled to SASE amplitude.
a range of parameters corresponding to experimental
uncertainties, the bunching is found to be b1 = .017 and
an estimated gain of 10 5 was achieved as reported in
References 5 and 8 for this system.
An estimation for the absolute energy of the forward 5 DISCUSSION
emitted CTR can now be calculated. Taking into
consideration foil scattering effects on the electron beam, Because there is good agreement between the predictions
angular acceptance, θ acc , the micro-bunching amplitude and measurements presented here, the formalism developed
above and the parameters in Table 1, an energy of 3.1 pJ above is assumed accurate and the electron beam
is predicted at the detector by numerical integration of Eq. distribution given by Eq. 5 is correct and no higher
1. It should be mentioned that scattering degradation and transverse modes are present. We have also demonstrated
θ acc each reduces the total amount of expected CTR at the the narrow angular spectrum expected for coherent
detector by about 40% and not including either will cause radiation by choosing an appropriate acceptance angle of
a significant overestimation of the signal. The energy the optical beam line to allow collection of the CTR and
measured at the detector was 2.7 pJ, agreeing well with insignificant collection of incoherent TR. These
the predicted number given above. conditions imply the electron beam micro-bunching is
Next, a Jerrell Ash monochromator was placed before uniform transversely across the beam.
the detector and line spectrum measurements were taken. A slight frequency shift was observed in the CTR
Because of the high attenuation of this optic, it was found spectrum center shown in Fig. 1. Looking at the
the monochromator bandwidth had to be broadened in order transition radiation model traditionally used, limits of its

219
 Proceedings of the 1999 Particle Accelerator Conference, New York, 1999

validity here are suspect. The criterion for using Eq. 3 for 407, 423 (1998).
the TR spectrum is given by Eq. 4. We see in this [8] M. Hogan et al., Phys. Rev. Lett. 81, 4867 (1998).
experiment the period of emitted radiation is
t = 4.3 * 10 −14 s and the assumption that this is much
greater than a collision time (for the electron/image charge
collision model) is questionable. If the condition in Eq. 4
is not applicable for the TR collision model, the more
general spectrum for TR from the Lienard-Wiechert fields
is found to be

∞ →
d 2U e 2ω 2 →   nˆ ⋅ r (t )  
=
dΩdω 4π 2 c
∫ nˆ × (nˆ × β ) exp iω  t − c  
 dt
−∞ 
(8)

where n̂ is the unit vector from the interaction to the

observation point, β is the velocity of the electron or
→
image charge and r (t ) is the trajectory of the particles in
the collision. Not only do the initial velocities need to be
known, but the physics of the particle trajectories during
the collision must be calculated. It is immediately seen
the spectrum in Eq. 8 contains additional phase
information not present in the standard TR spectrum given
by Eq. 3 and could account for the observed frequency shift
of the CTR spectrum shown in Fig. 1. As the frequencies
of emitted TR increase for future experiments, the
traditional spectrum from Eq. 3 will have to be replaced
by the more general TR spectrum given by Eq. 8.

6 CONCLUSION

The experiment and technique described here was shown to

reliably measure longitudinal beam modulations to a few
microns and less. Since this experiment was performed on
a SASE FEL, this measurement verified the crucial role of
micro-bunching in the SASE FEL gain process.
Simulations were used and the results agreed well with the
measurements described above, thus serving as an
independent check on the code predictions.

7 REFERENCES

[1] U. Happek et al., Phys. Rev. Lett. 67, 2962 (1991).

[2] J. Rosenzweig et al., Phys. Rev. Lett. 74, 2467
(1995).
[3] R. Bonifacio et al., Opt. Commun. 50, 373 (1984).
[4] J. Rosenzweig, G. Travish, and A. Tremaine, Nucl.
Instrum. Methods Phys. Res., Sect. A 365, 255
(1995).
[5] A. Tremaine et al., Phys. Rev. Lett. 81, 5816
(1998).
[6] D.C. Nguyen et al., “A High-Power compact
Regenerative Amplifier FEL”, PAC Conference,
Vancouver, B.C., Canada, 1997.
[7] N. Osmanov et al., Nucl. Inst. Meth. Phys. Res. A

220