You are on page 1of 9

Nuclear Instruments and Methods in Physics Research A237 (1985) 159-167 159

North-Holland, Amsterdam

FREE ELECTRON LASERS FOR THE XUV SPECTRAL REGION

J.B . MURPHY and C. PELLEGRINI


National Synchrotron Light Source, Brookhaven National Laboratory, Upton, NY 11973, USA

The generation of high intensity coherent radiation m the soft X-ray region from a free electron laser will require the FEL to
operate in the high gam or collective instability regime. In this mode of operation, which does not require a cavity resonator, the
radiation field grows exponentially along the undulator until nonlinear effects bring on saturation . We discuss the conditions that the
electron beam and the undulator must satisfy for the collective instability to develop. We present an example of an electron storage
ring with an undulator in a bypass section which satisfies these conditions . We present estimates of the output power that one can
expect from such systems.

1. Introduction 2. Principle of operation

The interest sparked by the operation of the first free In a FEL a relativistic electron beam and an elec-
electron laser (FEL) by Madey and his collaborators in tromagnetic wave traverse an undulator. The coupling
1975 has led to some exciting results [1-6]. In the last of the wave and the transverse electron current, induced
two years remarkable progress has been made toward by the undulator, can produce an energy transfer be-
the realization of the FEL as a source of high power, tween the beam kinetic energy and the radiation field
tunable, coherent radiation. FEL oscillators and ampli- energy if a synchronism condition is satisfied. This
fiers have been operated at wavelengths varying from condition relates the radiation field wavelength X, the
the centimeter to the near ultraviolet and at peak power
levels up to a hundred megawatts. This wealth of experi-
mental results is due to researchers at numerous labora-
o
undulator period Xo , the undulator field Bo , and
strength parameter K=eX BO/27rmc 2 , and the beam
energy y measured in rest energy units [10] :
tories : MSNW, TRW-Stanford, LASL, LBL-LLNL,
MIT, NRL, Columbia-NRL, UC-Santa Barbara, Orsay
and Frascati .
X= 2 Y°2 (1 + K2 (1)

The theory of FELs has at the same time reached a Notice that this wavelength is also the wavelength at
high level of completeness and is in good agreement which the spontaneous radiation from an electron
with the experimental results. As a result of the experi- traversing an undulator is emitted (111 .
mental and theoretical progress we now have a good An important property of the FEL is that the energy
understanding of the physics and technology of FELs, transfer between the beam and the radiation can be
which can be used to design systems operating in new enhanced by a collective instability producing an ex-
wavelength regions, like the XUV spectral region . ponential growth of the radiation [12] . When this insta-
The possibility of building a FEL operating at wave- bility becomes important the FEL is said to operate m
lengths shorter than 1000 A is a result of the progress the high gain regime. The existence of this regime is
made in producing high density relativistic electron very important for FEL operation in the XUV region
beams using electron storage rings. Storage rings spe- where we do not have optical components with suffi-
cially designed for FEL applications and capable of ciently high reflectivity and small absorption [13] .
accommodating undulators magnets 5 to 15 m long, Three modes of operation of an FEL can be consid-
should offer the possibility of producing coherent radia- ered. In the first mode, self amplified spontaneous
tion down to a few hundred Angström with average emission (SASE), the initial spontaneous radiation
powers of the order of watts and peak powers up to emitted by the electrons is amplified; this system does
hundreds of megawatts . One such ring is being built at not require any optical components but needs a high
Stanford University [7], while similar rings are also density electron beam and a rather long undulator [8,12] .
being studied at other laboratories [8,9] . In this paper The second mode is the FEL oscillator ; an optical
we want to bnefly review the different operation modes cavity is used to reflect back and forth the radiation for
of an FEL in the XUV region (sects . 2-4), we will then further amplification by another electron bunch; this
discuss in more detail, the self amplified spontaneous system can get by with a smaller electron beam density
emission mode (sects . 5-9) . and a shorter undulator but needs mirrors for the cav-

0168-9002/85/$03 .30 C Elsevier Science Publishers B.V . I. THEORY


(North-Holland Physics Publishing Division)
160 J. B Murphy, C Pellegrim / FEL for the XU V region

ity; at wavelengths smaller than 1000 A these mirrors 4. The SASE mode
have yet to be developed and their reflectivity is ex-
pected to be on the order of 50% . In the wavelength region below 1000 A, the best
In the third mode, or transverse optical klystron [14], accelerators available to produce the high density elec-
an external laser beam at the spontaneous radiation trons beams required to operate in the SASE mode are
wavelength is used to modulate the beam energy and electron storage rings. Existing storage rings, such as the
longitudinal density distribution leading to the emission VUV ring of the National Synchrotron Light Source at
of coherent radiation at the higher harmonics of the Brookhaven, can provide an average emittance on the
input laser. Of the three modes this is the one requiring order of 10 -s mrad, an energy spread of about 10 -3
the least stringent electron beam parameters . In ad- and a peak current of 60 A at an energy of 750 MeV
dition it does not need optical elements but on the other [151 . A ring like this, with straight sections capable of
hand it requires an undulator with rather strict magnetic accommodating undulators of 5 to 6 m, would allow us
field tolerances and produces the smallest coherent radi- to produce coherent radiation m the 1000 A region .
ation power. We believe that it is now possible to design a storage
ring with an energy of 700 to 1000 MeV, the same
energy spread and an emittance smaller by an order of
3. FEL growth rate
magnitude than that of the VUV ring, and peak cur-
In all three modes the FEL can be approximately rents to the range of 100 to 200 A. Such a ring would
characterized by one parameter, the FEL e-folding enable us to produce radiation m the wavelength range
length, 47rp, measured in number of undulator periods of 100 to 500 A using undulators about 10 m long [7-9].
N [12] Using this ring, the peak radiation power that one
KXos2p z~3 can obtain in the SASE mode is on the order of 10 -3
4 7rp = 41r times the beam peak power, or 100 MW . This pulse
87rc
( would have a duration of about 100 ps and a repetition
where S2 p is the electron beam plasma frequency, de- rate of 10 Hz, for an average radiation power of 0.1 W.
fined in terms of the electron density n o , and energy y, With the same system operating in the oscillator
by mode, one can obtain an average output power of the
1,12 order of 1 W, a pulse duration of about 100 ps and a
UP = (47rYe ('zn 0/y3) repetition rate determined by the ring revolution time to
re being the classical electron radius. be on the order of a few MHz, and a peak power of
For an oscillator to operate at short wavelength, about 10 kW . For this oscillator it is also possible, by
where the optical cavity losses can be on the order of modulating the system gain, to reduce the repetition
100% per round trip, one needs a value of 4apN on the rate and increase the peak power.
order of 1, i.e . a number of undulator periods For the TOK mode one can expect conversion ef-
ficiencies on the order of 10 -6 around the tenth
N - 1 /( 4 ?Tp) .
harmonic, so that starting with a 100 MW peak power
In the case of SASE [12] the value of 47pNu must be on laser at 2000 Â one should be able to produce about 100
the order of 10. W at around 200 A.
In the case of SASE and oscillator modes the energy In all of these cases the angular distribution of the
transfer from the beam to the radiation field is on the radiation is determined by the electron beam radius, a,
order of p, while in the TOK case the transfer from the and the radiation wavelength ; the characteristic angle is
input laser to the harmonics is rather small. on the order of X/a, i.e . of a few tenth of milliradians .
The expression (2) for the FEL growth rate applies The line width is on the order of the wavelength divided
only if two other conditions on the electron beam are by the electron bunch length, i.e. 10 -6-10 -5 , for the
satisfied. One is a condition on the beam energy spread, oscillator and the TOK mode . For the SASE mode it
which must be less than p; the second is a condition on depends on the details of the system and is intermediate
the beam emittance, which must be smaller than the between the oscillator limit and the inverse of the
radiation wavelength . If these conditions are not satis- number of periods in the undulator, i.e . between 10 - '
fied the radiation growth rate decreases and the output and 10 -5 .
laser power is reduced [8].
For wavelengths in the millimeter region and elec-
tron energy of a few MeV the value of p can be on the 5. FEL equations
order of 1 . In the VUV region with electron energies of
several hundred MeV, p is on the order of 10 -3 and In the remainder of this paper we will discuss the
one can expect an energy transfer from the beam to the high gain regime and the SASE mode of operation of an
radiation on the order of a few parts in a thousand . FEL. Following the work of other authors [16,17] we
J.B. Murphy, C Pellegrtm / FEL for the XUV region

write the FEL equations using the phase and energy as and write the FEL equations as [12]
electron variables and use the slowly varying amplitude
and phase approximation for the radiation field . These = 2p 11 - PZr2 ), i =1, 2, . . ., N, (lla)
equations can be written in a very general form includ-
ing the effects of space-charge fields and higher -- f A )
harmonics of the radiation field [18] . To simplify our I + c .c .
_ 7

1 I,
discussion we neglect these terms and use the results of
ref. [18] to evaluate their effects. Our notations are the
following: z is the electron beam and electromagnetic A=iSA+ I( ),
I'(
wave direction of propagation ; x and y are the trans-
verse coordinates ; Bo is the undulator magnetic field with
(we use a helical undulator for simplicity) and ao , Nu, S =~/p, (lld)
are its period and length in number of periods respec-
tively ; X is the wavelength of the radiation field; y is and D the detuning parameter
the electron energy m units of moc 2 ; ß~ = 1 is the ~ ° (Yô - YÂ)/2 Yii *
longitudinal electron velocity and ß 1 = K/y the ampli-
tude of the transverse velocity ; the electron phase rela- The dot indicates differentiation with respect to T. The
tive to the electromagnetic wave, ~5, is related to z and t angular brackets indicate an average over the particle
by p = 27rz/Xo + 2a(z - et)/X; the resonant energy YR initial phases, i.e . ( ) _ (1/N)E, where N is the num-
is related to Ao , X and K by y 22 = X0(1 + K 2 )/2a ; the ber of particles.
undulator frequency wo is co, = 27rc/8z /a o ; the ampli- From these equations we can show that the quantity :
tude, Eo, and phase, Bo, of the radiation field are H = (r)+ J A 1z (12)
combined to yield a complex amplitude a = i Eoe'a° .
is an invariant. In terms of laboratory variables this can
To write the FEL equations it is convenient to use a
be wetten as
set of normalized variables and introduce some quanti-
ties to characterize the beam properties [12,18] . We will E2
H = mc 2 no(y) + = constant, (13)
use the relativistic beam plasma frequency already in- 4~
troduced in sect . 3, which for a beam with energy which is seen to be the conservation of energy relation
dispersion is given by for the electron beam-radiation field system . It is also
/z
1,12 convenient, using eq . (9), to rewrite eq . (12) as
- ~ 4 7rren oc2 j
(4)
P
( Yo )3 Y - (Yo) 1 1 )
=P( I A Z- I Ao z > (14)
(Yo)
where (yo) is the average value of the initial electron
energy ; we introduce also the quantities which relates directly the change in the field amplitude
P=~K( (Yo) 1
2
2P l
2/3
(5) A to the average change in electron energy . One can see
4 J WO
from eq . (14) that, assuming I A o 1 « I A 1, the quantity
YR
p I A 1 2 measures the efficiency of energy transfer from
and the electron beam to the radiation field.
In integrating the FEL equations the maximum time
( Yâ is defined by the undulator length tm . = Nw lo /c . In
l'o=woll _ Yo >2 (6)
terms of the scaled time z this becomes
(
z
and a normalized time YR
7rn .ix=4wp( N_ (15)
)2 ~Yo) )
YR t. (7)
T=2wop
( (-Yo)

Using these definitions we can construct a set of 6. The FEL collective instability and coherent radiation
dimensionless variables generation

The system of eqs. (lla)-(llc) has been discussed in


ref. [12] where it has been shown that for S < 5 t,, = 1 .9
the system is unstable and the field amplitude A grows
a exp( i~o t) exponentially . Both the radiation field and the beam
-
A (10) bunching grow exponentially . We can characterize the
yz'
bunching by the parameter h = I (e - 'v ) 1 . The nonlinear
7
[ 4 rmc z(Yo)noP~

I. THEORY
16 2 J.B Murphy, C Pellegrini / FEL for the XUV region

regime and saturation that follow the initial exponential 001


growth have also been studied in these papers .
In this paper we want to discuss a collectively unsta-
ble system using the parameters that apply to an elec-
a
tron beam obtained from a storage ring. a
w
We assume that the initial field amplitude is zero tn
and we introduce an initial noise in the electron phase
r
distribution so that the initial value of b = I (e "I) I is
1/t/Nx, Nx being the number of electrons in one w
z (c)
radiation wavelength . In fig. 1 we show the evolution of w
m (b)
the field amplitude I A I versus r, for different values of
the initial electron beam rms energy spread, cF,,,, and for
S = 0. One can see that for ae0 << p the field amplitude
I A I reaches a value of the order of unity, so that, from
eq . (14), we have an energy transfer efficiency of the
order of p, i.e ., at the peak of I A I we have transferred a Fig 2. Plot of o versus v for b = 9.1 X 10 -3 , p = 3 X 10 -3 and
fraction p of the beam energy to the radiation field. several values of the initial rms energy spread, a, = 0.1 p, 0 75
Fig. 2 shows the evolution of the rms beam energy p and p, labeled a, b and c, respectively.
spread for the same values of a, 0 as in fig. 1 . One can
see that when the field peaks the energy spread becomes
on the order of p provided a, 0 < p. 3) after traversing the undulator we have I A I = l,
The time needed to reach the peak can be seen from b=I and (Y,-p.
figs . 1 and 2 to be T = 10. Assuming 70 = i'R we can see
from eq . (15) that to reach the peak we need an undula-
tor with a number of periods Nw - 1/p. 7. The electron beam-undulator system
Let us summarize the results of this section:
1) the electron beam, undulator magnet, radiation As we wish to discuss the operation of an FEL over
field system is unstable, if S < 5,,,, and both the field a large wavelength range (30-2000 f1) we will consider
amplitude, I A 1, and the beam bunching, b, will grow operating the storage ring at energies ranging from
exponentially up to a saturation level where I A - 1 and 300-500 MeV. In addition we will consider 3 undulator
b-1 ; designs, a 5 mm period undulator for A in the range of
2) if the system initial conditions are I A0 = 0, bo 30-100 Â, a 1 cm period undulator for X m the range of
determined by noise, a, o < p, the electron beam will 100-250 f1 and one with a0 = 2.5 cm for A in the range
transfer a fraction p of its energy in a number of of 500-2000 f1 .
undulator periods of the order of 1/p . To calculate the undulator properties we assume it to
be of the hybrid (permanent magnet and iron) type and
calculate the magnetic field from [19]
z
B0 = 3.33 exp(-5 .47 g + 1 .8 g T, (16)
X0 aô
where X0 and g are the period and the gap, respec-
tively . A complete listing of the undulator data can be
found in table 1 . The output radiation wavelengths for
the 3 undulators are signified by 0's in figs. 5a-5c.
The electron beam described in table 2 can be ob-
tained in a storage ring, as we discussed in sect . 5 .
However, if we tried to install the undulators described

Table l
Undulator magnets
Period, A0 (cm) 0.5 1 .0 2.5
r
Gap, g (cm) 0.1 0.2 05
Fig. 1 Plot of JAI versus r for b = 9.1 X l0 -3 , p = 3 X10-3 Pump strength, B0 (T) 1.2 1.2 1.2
and several values of the initial rms energy spread, a,,= 0.1 p, Undulator parameter, K 0 .56 1.12 280
0.75 p and p, labeled a, b and c, respectively .
J. B. Murphy, C. Pellegrini / FEL for the XUV region 163

Table 2
Electron beam parameters m the bypass section

Energy E (MeV) 500.00 400.00 300.00


Beta horizontal, ßx (m) 3.00 3.00 3 .00
Beta vertical, ßy (m) 1 .00 1.00 100
Coupling, X 0.01 0.01 0.01
rms horizontal beam radius, ax (m) 101E-04 8 .10E-05 6.07E-05
rms horizontal angular spread, a, (rad) 3.37E - 05 2.70E-05 2.02E-05
rms vertical angular spread, a, (rad) 5.81E-06 4.65E-06 3 .49E-06
rms vertical beam radius, a, (m) 5.81 E- 06 4.65E-06 3 .49E-06

to table 1 directly in the ring such that the electron UNDULATOR


BYPASS
beam would pass through the small aperture of the SECTION
undulator on each revolution, it would become impossi- ~
ble to operate the ring . The small aperature (gap) of the
undulator would result in vanishing small beam life-
times due to collisions with undulator walls. The mini-
mum allowable gap depends on both the ring and
undulator parameters and must be determined experi-
mentally.
For this reason we propose to install the undulator
in a ring bypass, as shown in fig. 3. The electron beam
RF SYSTEM
would normally circulate in the ring, where the effect of
synchrotron radiation damping would produce the beam Fig. 3. Sketch of storage ring and bypass section .
properties of table 2. About once per damping time, of
the order of 50 ms for the storage ring illustrated in
table 3, the beam is taken into the bypass and focused left there for a time long enough for synchrotron radia-
in the undulator by a special quadrupole triplet . In tion damping to bring its characteristics back to their
going through the undulator the electron beam produces starting value . A more detailed discussion of the storage
the radiation, its energy is decreased by pE T and its ring and bypass system is given in the next section.
energy spread increases from its initial value to about p. As the electron beam circulates in the ring it per-
The beam is then taken back into the storage ring and forms both vertical and horizontal oscillations, the so-

Table 3
Storage ring parameters

Energy, E (MeV) 500 400 300


Gamma, y 978 782 587
Bending radius, R h (m) 4 4 4
Average radius, R_ (m) 15 15 15
Number of achromatic bends, M 6 6 6
RF voltage, V (MV) 1 1 1
Harmonic number, h 100 100 100
Number of bunches 1 7 1
Average current, Io (A) 0.10 0.10 0.10
Electron number, N 1 .97+11 1 .97E+I1 1 97E+I1
Synchronous phase, a, 1 .38E-03 5.67E-04 1.79E-04
Momentum compaction, a 1 .22E-02 1.22E - 02 1.22E - 02
Horizontal emittance, ex (mrad) 3.41 E -09 2.18E-09 1.23E-09
Vertical emittance, e, (mrad) 3.38E-11 2.16E-11 1.22E-11
Zero current bunch length, apo (m) 1 .99E-03 1 .42E - 03 9.24E - 04
Zero current energy spread, a, . 2.14E-04 1 .71E-04 1.29E-04
Synchrotron tune, v, 1 .97E-02 2.20E-02 2.54E-02
Synchrotron radiation loss, Uo (MeV) 1 .38E-03 5.67E-04 1 79E-04
RF acceptance, ERF 3.23E-02 3.61E-02 4.17E-02

I. THEORY
16 4 J B Murphy, C Pellegrini / FEL for the XUV region

called betatron oscillations . The betatron functions, N H


and ßv, which characterize the amplitude and period of
1(_7) ZR
proximately given by [21]
B

-13
the betatron oscillations, are a measure of the focusing a
6 M R d" (18)
properties of the magnetic lattice. Together with the
emittance, E, which is the area in the position-angular z
= 7 .7 X 10 Y.~ (mrad) . (19)
deviation phase space in which the beam is contained, EH
M-
the beta functions determine the beam size, i.e ., the rms where R B , R,, are the bending and average ring radii
beam height, a H = E H P H . The choice of the beta and M is the number of achromatic bends. An

,
functions in the bypass is determined by the require- achromatic bend, typically consists of two dipole mag-
ment of keeping p as large as possible, which requires nets with a horizontally focusing quadrupole in be-
small NH ßv, without violating the energy spread con-
tween, and is designed to focus all the entering elec-
dition a, < p. In effect for a beam with nonzero trans-
trons, regardless of energy, to the same point on excit-
verse emittances EH, E v , it is necessary to add to the
ing the bend . The vertical emittance is determined by
real energy spread an effective spread given by [20]
the coupling between horizontal and vertical oscillation
z due to magnet misalignment, E v =XEH .
1 27r Kz At zero or small current the rms energy spread and
ac .ee(=
21 Xo l I+K 2EVßv the bunch length are determined by synchrotron radia-
1 tion and are given by [22]
'Y E E l
(17)
+ 1+K 2 (V + Rvl aEo = 4.38 X 10-7YZZ . (20)
B
In what follows we will make sure that the condition aR a 
ac .eff < or, is always satisfied. apo = a~o (21)
v
In the cases that we will consider in the next section
the undulator length varies between two and three me- where P, = w,/w, is the ring synchrotron oscillation
ters and is determined by the condition N, = 1/p. This tune . At large currents the microwave instability [23],
length is also consistent with our assumption on the caused by the beam interaction with the broad-band
beta-functions. high frequency storage ring impedance can increase the
energy spread, ae , and the bunch length, ap . An increase
of ap reduces the value of p while at the same time aE
8. The electron storage ring and bypass section increases and the condition aE < p can be violated .
To evaluate this effect we use the approximate condi-

z aa° ,
The storage ring that we consider is similar to those tion [23]
used as synchrotron radiation sources, for instance the Z(n) <
National Synchrotron Light Source VUV ring [15] . Its eIP 27Eo aa for n >_ (22)
main characteristics are given in table 3.
P

Since we want to maximize the electron density to where I P is the peak current, related to the average
obtain a large value of p in the undulator, we have bunch current, Io. by
chosen a ring design which minimizes the beam emit- /
tance and the bunch length . When the beam enters the I p = (21r)' zR=°Io (23)
bypass section it undergoes additional focusing to in-
Op

crease p, as shown in table 2. and IZ(n)/n l is the effective longitudinal coupling


The ring has two 10 m long straight sections, one impedance of the ring.
used for the radiofrequency system and one for the From eqs. (21)-(23) an expression for the microwave
bypass switching magnets. The arcs joining the two long instability limited bunch length and energy spread can
straights each have three equal periods. Each period has be obtained
2 dipole magnets with a focusing quadrupole between
l aIo e Z
them and two quadrupole doublets on the external , (24)
ap=Rav1 n
sides. The ring energy dispersion is controlled by the 27rEY~
central quadrupole and is nonzero only in the dipoles vs
and in the region between them . aE= (25)
aRavaP,
The momentum compaction, a = (d E/E)/(d l/1),
relates the change in orbit length to the relative energy The storage ring coupling impedance is determined
deviation from the design energy Eo of the ring . For a by the vacuum chamber geometry and by the bending
ring with this magnetic structure the momentum com- radius in the curved section [23,24] and is a quantity
paction a, and the horizontal emittance, EH, are ap- difficult to calculate "a priori . . . ". However in modern
J B Murphi, C Pellegrtm / FEL for the XUV region 165

Table 4 the coherent betatron tune shift be smaller than the


Bunch length and peak current synchrotron tune [23]
Effective coupling impedance (S2) 0.1 10 10 el P
Bunch length, aP [ lL-wave limit] (cm) 0.95 20 44 Svß= ZT .~ff <~ ",,
(26)
Peak current, IP (A) 397 184 86
with the transverse coupling impedance ZTerf evaluated
from the longitudinal impedance as [23]
Z( n) ~ .
storage rings values of the order of 1 S2 have been ZT,etf - 2 h (27)
(R)2 fi
obtained . Since this quantity is very important in de-
termining the performance of our system we have cho- The ring described in table 3 will be free from
sen to use in our calculations three values of I Z(n )In I , transverse instability problems provided the effective
i.e. 0.1, 1 and 10 S2 . Let us notice that a 10 S2 coupling impedance can be kept on the order 1 SZ or less .
impedance is large, and is a pessimistic assumption, As a final measure of the ring's feasibility we com-
while a 1 S2 value is realistic and has been already pute the Touschek lifetime [25] T, . The Touschek life-
obtained . On the other hand, a 0.1 SZ value would time is the time in which losses due to Coulomb colli-
require a breakthrough in storage ring design . sions between electrons in the same bunch have reduced
The microwave instability limited bunch lengths and the beam current to half of its initial value. For the
peak currents, which depend on the value of the cou- range of ring parameters given in table 3 T, > 1 h.
pling impedance but not the energy, are given in table 4
The bunch lengths are typically a few centimeters and 9. Results
the peak currents are in the 100-400 A range.
To test the beam for stability against transverse In figs . 4a-4c we plot the FEL parameter, p, and the
coherent oscillations we have used the conditions that microwave instability limited energy spread, aE, versus

7.0 f 70
(a)

6.0 60

50 5.0 25

40 n 20
0 0
x K

30

20 20 10

10 10 5

o- -- -o-__-o
0
i i i 1 i i 1 i
300 400 500 OL----//
300 400 500 300 400 500
ENERGY ( MeV) ENERGY (MeV) ENERGY ( MeV)

Fig. 4. Plots of the FEL parameter, p, and the microwave instability limited rms energy spread, ae, versus energy for (a) Xo = 5 mm,
(b) X () =1 cm and (c) a o = 2 .5 cm The p values are signified by a 0 and the a, values are given by a O . Each figure displays the p
and a, values for 3 values of Zeff = 0.1, 1.0 and 10 S2 . A solid line corresponds to Z = 0.1 S2, a dashed line to Z = 1 92 and (- ) to
Z=10 12 . The lines are not fitted to the points, they are drawn simply to indicate trends .

L THEORY
166 J B Murphy, C Pellegrini / FEL for the XUV region

250rs--T--r- r --~f
L (b 800

700

2500
2001 600
3 3 3
200 w ,a
180 ô
500

a -~
o
2000 2000 ir
w

160 a
4 3
0
Y
400 Y 1500 \\ 1500,
140 Q 150 W \ a
â
120 300 0- \\
50
100
80
40
60
500 500
40 100 100
30
20
250 300 350 400 450 500 300 40050 ~ 300 00 00~
ENERGY(MeV) ENERGY(MeV) ENERGY(MeV)

Fig. 5 Plot of the output wavelength A and the estimated peak power output versus energy for (a) A () = 5 mm, (b) ao =1 cm and (c)
Xo = 2 .5 cm . The A values are signified by a O and the power values are given by a O_ For the power curves . a solid line corresponds
to Z = 001 S2, a dashed line to Z =1 2 and (- - -) to Z =10 S2

the beam energy for the 5 mm, 1 cm and 2.5 cm impedance. For a value of the order of 1 S2 one can
undulators, respectively . Recalling that the limit on the obtain peak powers on the order of 500 MW down to
allowable energy spread is a, < p it can be seen from wavelengths of about 500 f\ and 50 MW to 80 f1 ; the
fig. 4a that if the ring impedance can be kept at or power decreases sharply at lower wavelength . If it should
below 0.1 62 one can expect to obtain high intensity become possible to reduce I Z(n)/n I to 0.1 12 one could
coherent soft-X-rays in the range of 30-85 Â. From get peak powers on the order of 20 MW down to 30 A.
figs . 4b and 4c it can be seen that the energy spread in One should also remember that to this paper we have
the ring will not pose any problems for the generation concentrated our attention on the first harmonic pro-
of intense radiation in the range of 85-2000 Â. duction only ; however, from the results of ref. [18], we
Figs . 5a-5c are plots of the peak power versus en- know that the system will also produce higher harmon-
ergy for the three undulator designs. The peak power is ics and this can shift down the lower limit for soft X-ray
calculated assuming that the radiation pulse length is production .
equal to the electron bunch length. We want to emphasize that the results presented here
are preliminary, and that one might improve the system
performance by optimizing other ring parameters such
10 . Conclusions as the momentum compaction or the radiofrequency
voltage and frequency. To obtain a more complete
Using the system described, an electron storage ring understanding of the system one should investigate dif-
with an undulator m a special bypass section, we can fraction effects on the radiation due to the finite beam
obtain high intensity coherent radiation by sending the radius and consider a three dimensional calculation
beam through the undulator and using the FEL collec- taking into account the electron density variation in
tive instability to produce radiation. Compared to other both the transverse and longitudinal direction.
systems, such as an FEL oscillator or a transverse
optical klystron, this system has the advantage that it This work is supported by the US Department of En-
does not require mirrors to form an optical cavity or an ergy .
input high power laser to bunch the electron beam . On
the other hand, by its very nature, this system can only
produce high intensity, short radiation pulses with a References
repetition rate of the order of 10 Hz . [1] A presentation of many of these results appear m these
The storage ring needed to operate the system is Proceedings (1984 FEL Conf., Italy) Nucl . Instr and
characterized by a small transverse eirtittanee. The other Meth . A237 (1,2) (1985) .
important ring parameter is the longitudinal coupling [21 L.R. Elias et a1 ., Phys . Rev. Lett. 36 (1976) 717 .
J B Murphy, C Pellegrini / FEL for the XUV region 167

[3] M Billardon et al ., Phys. Rev. Lett . 51 (1983) 1652 . [171 C. Pellegrini, in : Free Electron Lasers, eds. S. Martellucci
[41 B.E . Newman et al, Int. Quan . Flee . Conf . Pape r WHH1 and A.N . Chester (Plenum, New York, 1983) p 91 .
San Diego, California (1984) . [18] J.B. Murphy, C. Pellegnm and R. Bonifacio, Brookhaven
[5] T.J . Orzechowski et al ., ibid., post-deadline paper. National Laboratory Report, BNL-34156 (1984) .
[61 J.A . Pasour, R.F . Lucey and C.A. Kapetanakos, Phys. [19] K. Halbach, Proc . Bendor Free Electron Laser Conf., J. de
Rev. Lett . 53 (1984) 1728 . Physique 44 (1982) 211.
[7] J M.J . Madey, m : Free Electron Generation of Extreme [201 A. Luccio and C. Pellegnm, in : Free Electron Lasers, eds.,
Ultraviolet Coherent Radiation, eds., J.M .J . Madey and C. S. Martellucci and A.N Chester (Plenum, New York,
Pellegrini . American Institute of Physics, Conf Proc . Vol 1983) p. 28 .
118 (1984), p. 12 . [21] S. Krinsky, ref. [7], p. 44 .
[81 J. Murphy and C. Pellegrini, Int. Quant. Elec Conf ., paper [221 M. Sands, in : Physics with Intersecting Storage Rings, ed
WQQ2, San Diego, California (1984) : in course of publica- B. Touschek (Academic Press, New York, 1971) p. 257.
tion m J. Optic. Soc. Amer . B. [23] D. Boussard, CERN Lab II/RF/INT/75-2 (1975), J.M .
[9] K.-J. Kim et al ., OSA Annual Meeting, San Diego, Cali- Wang and C Pellegnm, in - Proc . 11th Int. Conf . on High
fornia (1984) . Energy Accelerators . Geneva (1980) ed ., W .S Newman, p.
[101 J.M .J . Madey, J. Appl . Phys . 42 (1971)( 1906 . 554; A. Chao, in : Physics of High Energy Particle Acceler-
[111 B.M . Kincaid, J. Appl Phys . 48 (1977) 2684 . ators, ed ., M. Month, AIP Conf . Series, no. 105 (1983) p.
[121 R. Bonifacio, C. Pellegnm and L.M . Narducci, Optics 353 ; C. Pellegrim, in : Physics of High Energy Particle
Comm. (1984) 373. Accelerators, eds., R.A Carrigan, F.R . Huson and M
[131 D.T Attwood et al ., in ref. [71, p. 294. Month, AIP Conf. series no . 87 (1982) p. 77 .
[14] B.M . Kincaid et al., in ref. [7], p. 110. [241 A. Faltens and L.J . Laslett, Particle Accel 4 (1973) 151 ;
[15] L. Blumberg et al ., IEEE Trans Nucl . Sci NS-26 (1979) V. Brady, A. Faltens and L.J . Laslett, Lawrence Berkeley
3842 . Laboratory, Report LBID-536 (1981) .
[16] W.B . Colson and S.K. Ride, m: Physics of Quantum [25] H. Bruck, Circular Particle Accelerators (Presses Uni-
Electronics, vol 7, eds., S.F . Jacobs et al . (Addison-Wes- versitaires de France, Paris, 1966).
ley, Reading, MA ., 1980) p. 377

I. THEORY