TUPMN038 Proceedings of PAC07, Albuquerque, New Mexico, USA

COHERENT CHERENKOV RADIATION AS A TEMPORAL

DIAGNOSTIC FOR MICROBUNCHED BEAMS

Giancarlo Gatti, INFN/LNF, Frascati (Roma)

Alan Cook, James Rosenzweig, Rodion Tikhoplav, UCLA, Los Angeles, California

Abstract in terms of number of photons per unit frequency,

Cherenkov radiation of a relativistic e-beam traversing where α is the fine structure constant, n the medium index
a thin section of aerogel is analized, putting the stress on of refraction, β the particle velocity in speed of light units
the coherent contribution due to the intra-beam, θc the Cherenkov angle, and Ld the lenght of the radiator.
transverse and longitudinal structure. The use of this tool Defining the refracting index of the Cherenkov radiator as
as a temporal diagnostic for micro-bunched beams makes n=1+Δ, where delta ranges inside the interval 0.006-1.13
possible to improve the amount of collected power at the [4] the Cherenkov angle can be easily expressed as
microbunching frequency several orders of magnitude ϑ c = 2Δ
more respect to the uncoherent Cherenkov contribution. Hence it is natural to use this property to explore high
The non-idealities of a real beam are taken in account, frequency components (very fine bunch details), such as
and some techniques aimed on enhancing the coherent the microbunching deriving from the FEL and IFEL
part of radiation are proposed and analized analitically. processes (Fig. 1).
COHERENT LONGITUDINAL
DIAGNOSTIC
The use of e-beam emitted radiation (e.g. SR, TR, CR)
as a tool for complete reconstruction of the bunch
structure, must take in consideration the coherent
contribution of such a radiation, no matter which is the
physical source generating the phenomenon. More in
detail, assuming a process in which every particle of the
beam radiates in the same way [1], just an intra-beam
particle displacement (i.e. delay) will affect the total
coherent contribution.
In such a way it is possible to write a general form of
such a radiation, indipendently of the specific physical
process that generates it. Let's assume a single particle
angular spectrum S(k), where |k| is the vacuum wave Figure 1: Uncoherent Cherenkov over uncoherent
vector, then the whole bunch far field spectral response transition radiation photons vs. wavelenght
[Desy] will be:
T (k ) = S (k )( N + N ( N − 1) F (k ))
Form Factor Influence
with N is the number of particles in the bunch, and F(k) Restricting the present framework to the case of a
the so-called form factor , i.e. 3-D Fourier transform of microbunched beam at the fundamental wavelenght λr,
the bunch particle distribution f. The second term in the and assuming that the electrons distribution can be splitted
spectrum expression, the one to deal with, is the radiation in the product of longitudinal and transverse part, the
coherent contribution with its characteristic N 2 scaling. complete form factor expression can be written as:
The coherent contribution has extensively been used F(k)=Ft(kt)Fl(kl), with kt, kl respectively ksinθ kcosθ . As a
for longitudinal diagnostic (i.e. bunch lenght first example, assuming both a gaussian radial
measurement), and, in some case, for the whole beam (transverse) and longitudinal distribution (whose standard
reconstruction purpose, for example employing transition deviation are, respectively σx=σy=σt and σl), the form
radiation [2][3]. Hence, the total beam spectral response factors become:
will experience both the contribution of the single particle Ft (k t ) = exp[−(σ t k t ) 2 ] ,
emission and the collective bunch effect. +∞

∑A
2
Ft (k l ) ≈ n exp[−σ l2 (k l − n h k r )] 2
Cherenkov Radiation Coherent Diagnostic −∞
The advantage in the use of Cherenkov radiation would Here, the longitudinal part is expressed as a superposition
be given by the wide and flat spectral response (cut-off of the different microbunching harmonics with weight An
wavelenghts on the order of the electron classical radius) on the n-th component. The trivial integration over
∂N ph emission angles gives for the coherent contribution:
1
= L d α (1 − 2 2 )δ (ϑ − ϑ c )
∂k∂ϑ β n
02 Synchrotron Light Sources and FELs A06 Free Electron Lasers
998 1-4244-0917-9/07/$25.00 2007
c IEEE
 Proceedings of PAC07, Albuquerque, New Mexico, USA TUPMN038

∂N ph 1 employment of different transverse shapes could

≈ L d N b2α (1 − ) exp[−(σ t k sin ϑ c ) 2 ] overcome this drawback, extending the cut-off frequency.
∂k β 2n 2

+∞ 2

∑A
−∞
n exp[−σ t2 (k cos ϑ c − n h k r ) 2 ]

It is worth noting that the main limitation in generating

high microbunching frequencies is given by the transverse
term strong suppression. Moreover, it can be defined a
“coherence angle”: for a given wavelength this is the
emission angle in which the transverse part “cuts” half of
the photons emitted. Beyond this limit there's very strong
suppression of coherence. Looking at the limitation for
the longitudinal (microbunching) wavelenght, in the
gaussian case above discussed, it is, for small values of Figure 2: Gaussian beam coherent Cherenkov and TR
the Cherenkov angle contribution
2
tan(ϑ cohr ) ≈ ϑ cohr ≤
σ t kr Looking at a hard edge uniform distribution beam on
This result shows how the transverse part influence the transverse dimension, with the same gaussian
grows for bigger angles. On the other side, when the distribution on the longitudinal one from the previous
emission is strongly peaked on a small angle, the example, the form factor changes. Being the electron
transverse particles don't influence each other. One way to radial distribution
“restore” coherence at a given wavelenght and a given 1
Cherenkov angle is to transversely squeeze the beam, and f (ρ ) = 2
rectσ t ( ρ )
πσ t
make it small, compared to such a wavelenght so that the
transverse particle displacementes contributions can add the form factor becomes
+∞ 2
inside a coherent lenght. J 1 (σ t k t ) 2
F (k ) = 4 N b2 [
σ t kt
] ∑
−∞
An exp[−σ l2 (k l − nk r ) 2 ]

PRACTICAL APPLICATIONS
with J1 first order, first type Bessel function.
It can be useful to compare the difference in the Once again, taking in consideration the number of
employment of TR and CR for some experimental photons inside a (narrow) longitudinal microbunching
situations. peak:
Considering a typical setting for the UCLA Neptune α 2 1 J (σ n k ϑ )
N ph ≈ 4 π L d N b2 An (1 − 2 2 )[ 1 t h r c ] 2
accelerator: γ=28, Nb=6E9 (i.e. Q=1nC), Δ=0.008, Ld=2.5 σl β n σ t nh k r ϑc
mm, σt=50 μm (that is a well focused beam), σl=500 μm
and looking at the contribution of the single n-th The improvement given by the form factor is showed in
microbunching harmonic, Cherenkov coherent photons Fig. 3, where the two distributions are compared.
are
2
α 2 1 ]
N CH
ph ≈ π Ld N b2 An (1 − 2 2 ) exp[−(n h k r σ t ϑ c )
σl β n

transition radiation ones, in the same case are:

α 2 γ
N TR
ph ≈ N b2 An [ ]4
2σ l π n h k r nh k rσ t
It must be noted, anyway, that for such a situation,
θc=7.2 deg, while the peak of TR is for θTR =2.04 deg, that
explains a stronger suppression. In Fig. 2 the two curves
are represented in a wavelenght range of 0-16 μm for a
microbunching factor |An|2 =1. It can be seen that TR
dominates for the wavelenght range (0.1-15 μm), even
though the decay of the Cherenkov due to the transverse Figure 3: Hard edge/Gaussian beam Cherenkov coherent
exponential factor is quite abrupt. This suggests that the contributions. Gaussian TR is also showed
02 Synchrotron Light Sources and FELs A06 Free Electron Lasers
1-4244-0917-9/07/$25.00 2007
c IEEE 999
TUPMN038 Proceedings of PAC07, Albuquerque, New Mexico, USA

BEAM CUT EFFECT The cut of the beam results in a big improvement
Improvement of the bunch response at short wavelenghts respect to the plain gaussian contribution; moreover it
could be achieved through transverse cutting of the beam. must be noted that, for modelling a realistic cut, the beam
Modelling the trasverse cut with a delta functions comb size has been assumed one order of magnitude bigger than
gives an idea of the spectral response extension given by the plain gaussian case. Even though, the spectral
acting a cut. Anyway, since the delta cut is not realistic, response of the cut beam is extended beyond the one of
the spectral extension would be infinite, that is obviously the tight focused beam.
a not physical result. Let's assume a periodic trasverse
modulation of the beam (e.g. Grid of wires). The only
assumption is for the cutting period (λ0, λ1) to be shorter
than the beam dimension. In such a way the distribution
along the cartesian coordinates will be
+∞
f t ( x) = f ( x) ∑A
−∞
L exp( jLk 0 x)

+∞
g t ( y) = g ( y) ∑C
−∞
M exp( jMk1 y )

using the Fourier expansion for the periodic modulation.

Still the longitudinal contribution will have the same form
h(z) very similar to the transverse one, but given by the
microbunching components. Taking in consideration the Figure 4: Coherent Cherenkov photons on a Periodic cut
gaussian case (but it would be valid anyway) and using gaussian beam
the previous assumption on the cut period:
λ0 ≤ 2σ x
it leads to CONCLUSIONS
2π π 2 The use of coherent Cherenkov radiation could be
k0 = ≥ ≥ extremely useful as a longitudinal diagnostic tool,
λ0 σ x σ x
expecially moving to short wavelenghts thanks to the
Cherenkov flat spectral response. Moreover, the limiting
This gives the possibility to write the expressions of the
factor is given by the trasverse form factor of the beam.
trasverse spatial spectrum dropping out the cross product
Different beam shapes cases have been taken in
terms:
2
consideration, showing the possibility of drastic
+∞ +∞ 2 2
~ ~ improvements, for example acting a transverse cut on the
∑ ∑A
2
F (k x ) = AL f (k x − Lk 0 ) = L f (k x − Lk 0 ) beam. Moreover still some effect such as divergence of
−∞ L =−∞
the beam, radiator dispersion, electrons and light
scattering inside radiator will be taken in consideration,
Assuming once again a full 3D gaussian distribution, as since they are likely to make the form factor suppression
beam dimensions will be larger, in order to allow a less steep.
physical easy way to cut. It will be σx , σy = 250μm for a
squared train of cut of period λ0, λ1 equal to 500μm REFERENCES
period and cut width Δ equal to 50μm so to get
[1] O. Grimm et al., DESY, April 24, 2006, internal
Δ Δ
AL = sin c( L) report
λ0 λ0 [2] Y. Shibata et al., Phys. Rev. E 50, 1479-1484, 1994
with longitudinal dimension of the beam still 500μm. [3] M. Castellano et al., Phys. Rev. E 63, 056501, 2001
Assuming the microbunching spectral line very sharp [4] J. Bar et al., Nucl Inst. And Meth. A, 538 (1), p.597-
respect to trasverse spectral distribution variation, the 607, Feb 2005
number of photons over this line is

∂N ph π ⎡ 1 ⎤ y
= L d N b 2α ⎢1 − ⎥ exp[−(k sin ϑ c sin φσ ) ]
2
∂φ 2πσ l ⎣ β n ⎦
2 2
+∞ 2

∑ −∞
AL exp[−(k r sin ϑ c cos φ − Lk 0 ) 2 σ x2 ]
+∞ 2

∑ −∞
BM exp[−(k r sin ϑ c sin φ − Lk1 ) 2 σ y2 ]

02 Synchrotron Light Sources and FELs A06 Free Electron Lasers

1000 1-4244-0917-9/07/$25.00 2007
c IEEE