High Gain Intergran Equation Acrm PDF

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 9

IEEE JOURNAL OF QUANTUM ELECTRONICS, VOL. 51, NO.

7, JULY 2015 1100209

The High Gain Integral Equation


for CARM-FEL Devices
Silvio Ceccuzzi, Student Member, IEEE, Giuseppe Dattoli, Emanuele Di Palma,
Andrea Doria, Elio Sabia, and Ivan Spassovsky

Abstract— The theory of cyclotron autoresonance provide any new insight into the relevant Physics, the practical
maser (CARM) is formulated in a free electron laser (FEL) outcome consists in allowing the possibility of extending to the
like fashion. We get a one-to-one correspondence between the analysis of the CARM operation the wealth of practical formu-
equations describing the two devices and show that the wealth
of scaling relations developed for the FEL evolution can be lae used in the case of undulator-based FEL devices [3], [4].
exploited to describe the CARM dynamics too. The core of the strategy is that of analyzing the CARM
theory to find a one to one correspondence with the physical
Index Terms— Electron device, CARM, gyrotron, FEL.
quantities entering the description of a FEL. Although the
I. I NTRODUCTION problem of establishing a general theoretical framework
embedding a unified treatment of free electron coherent
I N A CARM (Cyclotron Auto-Resonance Maser) a moder-
ately relativistic e-beam is injected in a cavity where an
axial static magnetic field drives the electrons along a helical
generators has been discussed in a series of authoritative
researches [5]–[8], here we will address the problem from
a more modest (namely pragmatic) point of view, aimed at
path. Coherent emission towards a mode of the cavity occurs
peeking out the elements leading to practical scaling relations,
whenever the resonance condition
useful for the design of a specific system.
ω =  + kz v z (1) We will follow the already quoted seminal papers [1], [2]
where a careful analysis of the physics of CARM devices
is satisfied.1 In the previous equation  = (eB)/(mγ ) is the
(and of other free electron generators as well) has been
relativistic cyclotron frequency, v z is the electron longitudinal
worked out. The problem we will consider is that of an
velocity and k z is the field wave vector, linked to the phase
electron beam moving in the direction z in a constant magnetic
velocity by the identity
field pointing in the same direction of the electrons. The beam
ω experiences an energy exchange with the modes of the cavity,
vp = (2)
kz in which the system is embedded. The term cavity is used
The coherent emission process is driven by the bunching here improperly, we should refer to interaction region inside
mechanism: the interaction with the cavity mode electric field the waveguide for the single pass amplifier configuration and
determines an energy variation of the electron beam which to cavity for the oscillator.
transforms into a density modulation, coherent emission occurs Before getting into the specific, we will fix the elements of
when the electrons are bunched on a scale length comparable the game, i.e. the physical variables and the relevant scaling
to the mode wavelength. In these CARM type devices, when factors, which are defined according to the suggestions in [2]
the saturation occurs, the electrons are kept on resonance by and are summarized below
the fact that any variation of , induced by an energy decrease,
γ0 − γ
is compensated by a suitable modification of the longitudinal w=
velocity. This effect is in turn responsible for the intrinsic high γ0
⎛ ⎞
efficiency of these devices.  
βz ⎝ 
In the forthcoming sections paper we will reformulate the δ0 = 1− 1−s  ⎠
βp 1 − ββzp ω
CARM theory [1], [2] using the same language and formalism
 
adopted to study the dynamics of undulator-based Free 2 βz
Electron Lasers (FEL). Even though such a restyling will not u = 1 − w
β⊥2 βp
 
Manuscript received January 5, 2015; revised March 23, 2015 and 2 1 − ββzp
April 23, 2015; accepted May 4, 2015. Date of publication May 13, 2015;
date of current version June 2, 2015. = 2 (1 − β −2 ) 0
δ
The authors are with ENEA–Centro Ricerche Frascati, Rome 00044, Italy
β⊥ p
β⊥2 (1 − β −2 )
(e-mail: silvio.ceccuzzi@enea.it; giuseppe.dattoli@enea.it; emanuele.
p ωz
dipalma@enea.it; andrea.doria@enea.it; elio.sabia@enea.it; ζ =  
ivan.spassovsky@enea.it). 2βz 1 − βz c
Color versions of one or more of the figures in this paper are available βp
 
online at http://ieeexplore.ieee.org. βz 2
Digital Object Identifier 10.1109/JQE.2015.2432719 4|e| 1 − βp
1 The (1) is just a kinematical condition; its embedding within an appropriate Fs = [C J ]s s (3)
gain formula will be discussed later in the paper. γ0 m e c2 2β 3 (1 − β −2
p )2
1

0018-9197 © 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
1100209 IEEE JOURNAL OF QUANTUM ELECTRONICS, VOL. 51, NO. 7, JULY 2015

Regarding the electron kinematic quantities we will use equivalent to that exploited in the theory of high gain (SASE)
βz and β⊥ to indicate the (initial) longitudinal and transverse FEL devices [3], [4]. By SSA we mean that the dynamics of
reduced velocities respectively, while γ stands for the rela- the system is treated far from the saturation and that the (4)-(5)
tivistic factor (with γ0 being the relativistic factor associated can be linearized by keeping a lowest order expansion, namely
with electron initial energy) and β p (= v p /c) for the reduced by neglecting contribution in o(u 2 ), o(|Fs |u), o(|Fs |2 ).
phase velocity. The coordinate of propagation, namely the Within such a framework we reduce the previous equations
longitudinal coordinate, is z. The symbol denotes the to (we keep hereafter s = 1 and neglect all indices regarding
TE electric field mode amplitude and [C J ] is a contribution the harmonics)
depending on the geometry of the cavity where the interaction  
occurs, in the case of a cylindrical cavity it is given by a dF   1
≈ Ig eiθ θ + I g ueiθ θ , I g = b − Ig
suitable combination of Bessel functions, the index s stands dζ 0 0 2
for the s-th harmonics. Furthermore δ0 is the frequency du  
detuning and the condition δ0 = 0 corresponds to the resonant ≈ Re Fs e−iθ

interaction. d 2θ
According to [1] and [2] the CARM energy and phase ≈ (b − 1)|F| cos(θ + φ) +
dζ 2
equations, for TE modes interaction, can be cast in the form   
d 1 d
− b Q(ζ ) − Re i Fe−iθ
[1 − u] 2  −iθ 
s (7)
du dζ 2 dζ
= Re Fs e
dζ 1 − bu
dθ 1 The terms in brace brackets, in the last of (7), will be neglected
= [ − u − bQ(ζ ) for the moment, since they are vanishing for an initially
dζ 1 − bu
s s
 
unbunched e-beam and do not play a significant role in the
+ [1 − u] 2 −1 Re i Fs e−iθ small signal regime. Pursuing further the SSA we set
2
s  
dQ(ζ ) [1 − u] 2 d Fs −iθ
=− Re i e (4) θ ≈ θ0 + ζ + δθ
dζ 1 − bu dζ
δθ  1 (8)
The normalized u, θ variables are associated with the
electron energy and the electron-wave phase respectively,
Where δθ is a perturbation which, according to the third of (7),
while Fs accounts for the complex mode field amplitude,
can be written as
whose evolution is fixed by the equation
 ζ  
s 
d Fs [1 − u] 2 iθ δθ ≈ (μ − 1) (ζ − ζ  )Re F(ζ  )e−i(θ0 +ζ ) dζ 
= Ig e (5)
dζ 1 − bu 0
θ0 μ = b (9)
where ...θ0 denotes the average taken on the initial phase
distribution and Ig is the gain current, we will specify in the Furthermore the same level of approximation yields
next section. The quantity denoted by b is sometimes called the  ζ  

electron recoil parameter and is responsible for the mechanism u≈ Re F(ζ  )e−i(θ0 +ζ ) dζ  (10)
of auto-resonance itself. This quantity plays a role analogous 0
to the tapering factor in FEL operating with magnetic
undulators [3], [4]. The normalized field amplitude Fs has The first of (7) at the lowest order of perturbation reads
the dimensions of the inverse of a length and corresponds to  
dF
the force acted by the TE mode field on the electron, divided ≈ Ig ei(θ0 +ζ ) (1 + i δθ )
dζ θ0
by its mass.  
i(θ0 +ζ )
We will examine more carefully the physical meaning of + I g ue (1 + i δθ ) (11)
θ0
the previous variables and the relevant normalization factors
in the following. Here we just note that, for b = s = 0, the set inserting (9) and (10) in (11) and averaging on θ0 finally yields
of (4)-(5) reduces to the FEL equation and can be comprised
into the coupled pendulum and slow varying amplitude  ζ
dF i 
equation a là Colson [3], [4] as it follows = (b − 1)Ig (ζ − ζ  )F(ζ  )ei(ζ −ζ ) dζ 
dζ 2 0
d 2θ 
= |F| cos(θ + φ) Ig ζ  i(ζ −ζ  )
dζ 2 +i F(ζ )e dζ  (12)
2 0
dF 
= Ig eiθ θ
dζ 0 This equation is the CARM integral equation providing the
F = |F|e iφ
(6) SSA evolution of the field amplitude and differs from that of
FEL for the last integral on the rhs.
Let us now show that, in the small signal approximation (SSA), Eq. (12) holds in the high gain regime and the relevant low
the linearization of (4)-(5) leads to an integral equation mostly gain limit can be obtained using a perturbative expansion in
CECCUZZI et al.: HIGH GAIN INTEGRAL EQUATION FOR CARM-FEL DEVICES 1100209

where L denotes the length of the region through which the


interaction occurs. We introduce the number N of cyclotron
turns in an interaction time as

T = 2π N (17)

Accordingly, the product of the variables  and ζ , entering


the previously given equations, can be rewritten as
z
ζ = ντ, τ =
βz cT
δ0
Fig. 1. Cyclotron trajectories inside the interaction region. ν = 2π Nδ 0 with δ 0 = βz
(18)
1− βp
terms of the gain parameter Ig . At the lowest order, we obtain
  where ν plays the same role of the detuning parameter
Ig used in the FEL gain equations. It is worth stressing that
F = F (0) 1 + (ζ )
2 in (16)-(18) the number of cyclotron turns replaces the number
 ζ of undulator periods. We can accordingly establish the further
(ζ ) = (ξ )dξ correspondence with the period length as it follows
0
 ξ  
(ξ ) = i (b − 1)

(ξ − ζ  )ei(ξ −ζ ) dζ  βz
λ = 1− 
0
 βp
 s  ξ i(ξ −ζ  )  2πc
+i b − e dζ = (19)
2 0 
    iξ 
∂ s e −1 where  is understood as the relativistic Larmor radius.
= −i (b − 1) +i b− (13)
∂ 2  On the other side, if we define βz in terms of the longitudinal
The previous identities are easily recognized to reduce to the relativistic factor as2
low gain FEL approximation for b = s = 0, we expect 
therefore that the gain exhibits the usual antisymmetric shape, 1
βz = 1 − ∗2
with distortion appearing for increasing values of b and s. γ
To better appreciate both differences and analogies we need γ
γ∗ =  (20)
a step back, by assessing a more complete understanding of 1 + (γβ⊥ )2
the variables we used in this section.
we can recognize the correspondence with the undulator
II. T HE CARM L OW G AIN R EGIME strength parameter recognized as (β p ≈ 1 )
The choice of the normalization of the physical variables
entering the description of generators of coherent radiation βp ≈ 1
via free electrons is quite an important issue, because an K ∗ = γβ⊥ (21)
appropriate choice may provide the correct hint for the scaling
behavior of the relevant physical quantities. We will see The FEL equations are usually given using τ as an
how from the analysis of the variables and of the relevant independent variable, we recast therefore ζ in the form
normalization factors will emerge important indications.
In the previous section we have left unspecified the gain 1 − β −2
p
ζ = π Nβ⊥
2
 2 τ (22)
current, which is related to the e-beam current I0 and is βz
1 − βp
expressed as
 3
βz
2μ0 |e| β p 1 − βp
and rescale the CARM variables, by replacing in our equation
Ig = −2 0
I [C J ]2 (14)
mcγ β⊥ 4
(1 − β p ) d d
= χ −1
From the discussion of the previous section it also follows that dζ ⎛ dτ ⎞
the CARM resonance condition is fixed by 1 − β −2
⎜ p ⎟
 χ = ⎝π Nβ⊥
2
  ⎠ (23)
ωR = (15) βz 2
1 − ββzp 1 − βp

According to Fig. 1 the electron propagate inside the cavity A step further in our procedure is the introduction
by executing a number of turns during an interaction time of the following Colson like dimensionless CARM
L
T = (16)
βz c 2 It is to be noted that β 2 + β 2 = 1 − 1 .
z ⊥ γ2
1100209 IEEE JOURNAL OF QUANTUM ELECTRONICS, VOL. 51, NO. 7, JULY 2015

field amplitude
3
(1 − β −2
p ) N
2 2
|a| = χ |F| = 2π β⊥ 
2 2
2 [C J ]  E
γ0
1 − ββzp
 2
N
= 4π K ∗  [C J ] E
γ0
3
(1 − β −2
p )
2 e
=π  2 , E = 2
(24)
mc
2 1 − ββzp
The small signal gain coefficient is finally introduced as
   3  
I0 N βz −3
2πg0 = χ 3 Ig = 8π 4 1−
IA γ βp
Fig. 2. FEL (continuous line) and CARM (dashed line) gain vs. detuning
· γ 2 β p β⊥2
(1 − β −2
p ) [C J ]
2 2
 parameter for b = 1/2.
   3
J0 N  ∗ 2
= (2π)4π K [C J ] 
IA γ
where the PE expression in practical unit is given by
(1 − β −2 p )
2
I0    
 = π 2β p  3 , J0 = 2 MW A
 = · −4
γ |J |
1 − ββzp PE
cm 2
0.5109989 10
m2
4π mc3 Let us now comment on the parameter which is a distinctive
IA = ≡ Alfvèn current
Z 0 c |e| feature with respect to the FEL devices; it is defined as
Z 0 = μ0 c ≡ Free Space Impedance (25)
β⊥2
With such a restyling CARM and FEL equations can be easily b=   (30)
βz
translated one into the other. 2βz β p 1 − βp
It is worth stressing that the CARM field intensity is now
and controls the already quoted mechanism of auto-resonance.
normalized a là Colson [3], [4]
Even though it starts playing a role at the onset of the
I saturation, when the field intensity is large enough to
|a|2 = 8π 2 (26)
Is induce a significant electron energy variation, it cannot be a
being priori neglected in the gain formula too. In the low gain
 2 approximation we get indeed
c m e c2  
I = 2E G() = 2πg0 Re (ζ )|ζ =1
4π e ⎡
 2   2
c m e c2 γ 4 −2 ⎣ ∂ sin 
Is = ([cm]K ∗ [C J ])−2 (27) = πg0 (b − 1) 
2
8π e N ∂ 2
and in practical unit the expression of the two quantity I, Is   
1 sin 2 cos 

− 
is given by [3], [4] + b− 2
  2
2
(31)
  2
 4 −2
2 γ
MW 2
Is = 6.931215 · 10 ([cm]K ∗ [C J ])−2 In Fig. 2 we have reported the CARM small signal low gain
cm2 N 2
  function; for the undulator-based FEL and b = 0.5, it is
MW
I = 6.931215 · 10−2 ( E [m−1 ])2 (28) evident that the familiar antisymmetric shape is lost in the case
cm2
of CARM. It is worth noting that the gain function shape is
In the previous equations Is , called saturation intensity, is a not dissimilar from the case of devices operating with tapered
quantity of paramount importance in laser devices: it corre- undulators [9]. The relevant consequences will be discussed
sponds to the value of the field intensity which halves the in the forthcoming sections.
small signal gain. This is strictly true for low gain devices; Before closing this section, we note that the effect of
in the case of high gain, I = Is is a reference value marking inhomogeneous broadening induced by the e-beam r.m.s.
the importance of non-linear effects in the dynamics of the relative energy spread
system. The importance of FEL-like processes of Is (be they  
operating in low or high gain regime) is further stressed by γ
= σe
the following identity γ rms
βz βz
8β p 1 − β p m e c2 8β p 1 − β p can be accounted for through the coefficient
g 0 Is = γ |J0 | = PE (29)
π N 1 − β −2
p e π N 1 − β −2
p μ = 4Nσ (32)
CECCUZZI et al.: HIGH GAIN INTEGRAL EQUATION FOR CARM-FEL DEVICES 1100209

The latter can be exploited to quantify the gain reduction and with the initial amplitude fixed by
induced by the inhomogeneous broadening effects associated
"
3
with the non-ideal energy distribution, in a way analogous to fj = 1
that occurring in the case of undulator FELs [3], [4]. j =1
The condition to neglect the inhomogeneous broadening
effects is indeed μ < 1, which means that the relative energy "
3
δ j f j = −ν
spread should satisfy the following condition
j =1
 
γ m e γβz c "
3
<π (33) δ 2j f j = ν 2
γ rms eB L (39)
j =1
The physical content of previous equation is self evident and
amounts to saying that the momentum gained by the electron To get a first hint into the type of solutions we consider the
through the Lorentz force induced by the longitudinal velocity case with b = 1/2, ν = 0 (which means  = 0) thus ending
spread associated with (γ /γ )rms must be less than the up with growth rates specified by
electron relativistic momentum. δ 3 = −πg0 (40)

III. T HE CARM S MALL S IGNAL I NTEGRAL E QUATION so that the fastest growing root can be expressed as
z
After the above remarks, we go back to the (12) which, F(τ ) ∝ e lg
in terms of the new variables, will be written as 
 τ lg = √
dF  4π 3ρ
e−iντ = −i (1 − αν)πg0 (τ − τ  )F(τ  )e−iντ dτ  1  πg0  13
dτ ρ =
  0
 τ (41)
1  4π N 3
+i b − χπg0 F(τ  )e−iντ dτ  (34)
2 0 We have denoted by l g the CARM gain length in close analogy
βz to what has been done in the case of high gain FEL.
1 − βp
with α = The previous analysis has been devoted to the small signal
2π Nβz β p (1 − β −2 p ) regime, when the non-linear contributions are not playing
by following the same procedure of [3], [4] any significant role. We have pointed out the analogies
 τ (and difference as well) with respect to the case of ordinary
  −iντ  dτ  τ (F(τ )e
−2 −iντ undulator FEL devices and we have indicated how to take
(τ − τ )F(τ )e =D ) (35)
0 advantages from these formal analogies. The strong signal
where the operator D τ−2 is the inverse of the derivative regime, namely when non-linear effects due to the growth of
the field intensity becomes important providing a decrease of
operator [3], [4] and denotes a doubly repeated integral. After the gain and eventually its saturation cannot be accounted in
taking repeated derivatives of both sides of (34), we find that terms of simple analytical formulae. However the numerical
it can be transformed into the following third order differential integration of (4)-(5), performing a one dimensional simulation
equation procedure analogous to that proposed in [10], has shown
   
1 that the intensity growth along the longitudinal coordinate
F  − i 2ν F  − ν 2 + b − χπg0 F  can be reproduced using a logistic type curve of the same
2
    type exploited to study FEL operating in the Self Amplified
1
= −i (1 − αν) + i b − χν πg0 F (36) Spontaneous Emission Regime [3], [4]. The parameters used
2
in the simulation come from the device specified in [11];
where the apex denotes derivative with respect to τ (we have they are the result of an optimization criterion, involving the
used the parameter ν instead of , for convenience reasons). attainable beam qualities, the cavity mode structure and a
If we choose b = 1/2, the (36) reduces exactly to the high CARM efficiency above 20%.
gain FEL equation. In Fig. 3 we have reported the CARM gain vs. ν for specific
The solution of (36) can however be written as values of the relevant parameters. As already noted for the gain
! curve with b = 1/2 also the intensity growth evolution of
F = 3j =1 f j e−i(ν+δ j )τ
(37) the full dynamics, with small signal approximation, is closely
F(0) = F0 F  (0) = 0 F  (0) = 0 similar to the high gain FEL and, in particular, as to the
intensity growth in the linear regime, we can use the following
with δ j being the growth terms fixed by roots of the cubic
relation
equation #      √ 
    z z π 3z
1 F(z) = 2 cosh − exp − cos + +
(ν + δ) + 2ν(ν + δ) + ν + b −
3 2 2
χπg0 (ν + δ) lg 2l g 3 2l g
2  √ 
     
1 z π 3z
= − (1 − αν) + i b − χν πg0 (38) − exp cos − (42)
2 2l g 3 2l g
1100209 IEEE JOURNAL OF QUANTUM ELECTRONICS, VOL. 51, NO. 7, JULY 2015

Fig. 3. CARM gain vs. ν when b = 0.1, χ = 1, α = 3 · 10−3 , g0 = 30


(continuous line) and b = 1/2, χ = 1, α = 3 · 10−3 , g0 = 30 (dashed line).
Fig. 5. Efficiency for the first harmonic (s = 1) as a function of the
non-dimensional  parameter for b = 0.39.

By assuming indeed that the efficiency is close to that of the


single particle we have
β⊥
2
b
η=  = (43)
2 1− βz −1 ) βz β p (1 − γ −1 )
β p (1 − γ

We end up with the following expression


A(z)
Pc (z) = P0 P0
(44)
1+ PF [ A(z) − 1]

P f = η PE (45)
The choice of the operating detuning parameter is not a
Fig. 4. CARM power growth vs. longitudinal position (continuous line) secondary quantity in the optimization of the device. We have
compared to a redefined analytical expression used for the FEL signal growth therefore reported in Fig. 5, the efficiency vs. , which
(dashed line); the parameters used for the simulation are reported in Table I.
exhibits a maximum for  = 0.04. The values of the various
TABLE I constants are chosen in such a way that
T HE PARAMETERS VALUES U SED FOR THE S IMULATION OF THE β⊥
= a ≈ 0.53, β p = 1.0008
CARM [11] H IGH G AIN S INGLE PASSAGE DYNAMICS F ROM βz
S MALL S IGNAL TO S ATURATION 
1 γ2 − 1
βz = , N ≈ 250
γ 1 + a2
Using furthermore the analogy with the high gain FEL we
introduce the effect of the energy spread inhomogeneous
broadening through the parameter [3], [4]
2σ
μ =
$ (46)
ρ
such a value can be in turn exploited to fix the energy spread,
which should be such that $μ < 1, to prevent any effect of gain
or efficiency reduction. Accordingly the beam energy spread
should be kept on the order of 0.1%.
The optimization of the efficiency is not a straightforward
task; it depends indeed on the choice of the value which
which reproduces fairly well the small signal growth of the
in turns fixes the quality of the e-beam. The choice of
field intensity, as reported in Fig. 4, where we also reported
an appropriate value for the phase velocity depends on the
the comparison with the numerical simulation based on the
geometry of the cavity through the dispersion relation
integration of (4)-(5).
 ω 2
We can preliminarily include the effect of saturation = kc2 + k z2
by using the same procedure adopted in FEL physics. c
CECCUZZI et al.: HIGH GAIN INTEGRAL EQUATION FOR CARM-FEL DEVICES 1100209

The formulae we have presented so far allow the possibility


of developing a preliminary analysis of the design of a
CARM device along the lines developed for the ordinary
FELs and summarized in [4]. The example we have discussed
so far is relevant to a scheme foreseeing the saturation in
only one pass inside the cavity. Namely the analogous of
FEL self amplified spontaneous emission. A different scheme
of operation will be discussed in the forthcoming section.

IV. CARM O SCILLATOR DYNAMICS AND


C AVITY Q UALITY FACTOR
We have so far discussed the case of a CARM operating
in the high gain regime, namely we assumed an operation
in which the saturation is reached in one passage of the
Fig. 6. Log-log plot of the output efficiency vs. the quality factor for
e-beam inside the interaction region. We will consider now G M = 1.2, a = 0.05 (continuous line) and G M = 0.8, a = 0.1 (dashed line)
the alternative configuration, namely a CARM operating in the for a cavity length of 30 cm at a resonance frequency of 250 GHz.
oscillator regime, which implies the growth of the field after
a certain number of turns inside the optical cavity. Within
such a framework a number of quantities are of paramount
importance and should be carefully fixed. We will consider in
particular the starting current and the cavity quality factors.
To get a first and reliable evaluation, we will take advantage
from the already developed analogy with the conventional
FEL sources. We use therefore the saturation model underlying
the parameterization of [3] and [4] and express the equilibrium
intracavity power density, in terms of the main parameters of
the device, as it follows
 
√ 1 − ηa (1 + α)
Ie = ( 2 + 1) G M − 1 Is
ηa (1 + a)
ηp
a= (47) Diffractive factor Q a vs. active losses ηa .
ηa Fig. 7.

Where ηa are the active losses G M the maximum gain


(including the inhomogeneous contributions) and a the ratio
between passive (η p ) and active (ηa ) losses, the total losses are
given by ηa + η p . The saturation intensity Is will be specified
later. The out-coupled power will be specified by
I0 = ηa Ie = E(ηa , a, G M )Is

E(ηa , a, G M ) = ( 2 + 1)ηa
 
1 − ηa (1 + a)
· GM − 1 (48)
ηa (1 + a)
where E(ηa , a, G M ) represents the extraction efficiency. The
losses are linked to the cavity Quality factor by the identity
2πν L c
Qc = (49)
cηa
Fig. 8. Extraction efficiency vs. the active losses for G M = 2.1, a = 0.05
denoting with L c the cavity length, and ν the operating (continuous line) and G M = 0.8, a = 0.1 (dashed line).
frequency.
We can report the output efficiency vs. the quality factor
as shown in Fig. 6. Furthermore in Fig. 7 we have reported parameters of the continuous curve in Fig. 8 we have a
the Q factor vs. the losses. From the inspection of these last corresponding intracavity power of about 2.5M W ;
figures we infer that c) The corresponding Q-factor is around 104 ;
a) the operating quality factor (or what is the same the d) A Q-factor below 5 · 103 demands for cavity losses
losses) depends on the gain factor G M ; above 0.5.
b) if we impose an output power of 0.5M W , we find that An operative set of parameters is that reported in Fig. 8,
at the maximum output efficiency (ηa ≈ 0.2) for the where the following reference quantities have been
1100209 IEEE JOURNAL OF QUANTUM ELECTRONICS, VOL. 51, NO. 7, JULY 2015

Fig. 9. Intracavity enhancement factor Ie /Is vs. ηa with g0 = 1.8 Fig. 10. Q-factor at resonance when the cavity with Bragg reflectors is
and a = 0.01. excited with the TE82 mode.

assumed: G M = 2.1, P0 ≈ 1 MW, ηa ≈ 0.6, equivalent to The preliminary calculation of specific mode operation with
Q a ≈ 3.5 · 103 . The intracavity power can accordingly be Bragg-type reflectors [12], [13] shows that the operating mode
specified around Pe ≈ 1.7 MW. The passive losses, associated TE82 has a Q-factor with suitable values for our purposes
with the ohmic dissipation, are not a parameter of secondary (see Fig. 10 where the chosen cavity parameters are reported).
importance, they specify the amount of cavity heating to be In this paper we have collected a number of thoughts on
cooled. With the above parameters the request of a = 0.05 CARM theory organized in an FEL type fashion. We have
implies an amount of power around 100 kW to be supported been able to provide a “translation” of previous researches
by the cavity walls. in terms of a language more familiar to people involved in
The numbers we have quoted are reasonable, but they should magnetic undulator devices (and hence to some of the present
be further specified in terms of gain coefficient, saturation Authors). This effort is however preparatory to a more accurate
intensity and so on. The saturation intensity and the e-beam analysis of the mode excited in a cavity discussed in this
power are linked by (note that Is , Ie and PE are understood paper, at least for the CARM oscillator configuration. Within
as power and not as power density, this is correct inasmuch this context a major role will be played by the necessity of
we assume that the beam cross section matches that of the suppressing unwanted modes, which may hamper the required
radiation mode) efficiency conditions. For this reason appropriate hollow beam
configurations have been studied. The beam, produced by a
g0 Is ≈ ec PE (50)
cathode with a corona emitter, is transported in the resonator
where ec is the intrinsic CARM efficiency. According to the in conditions suitable to match the TE82 mode and have
previous relations, the saturation intensity can be specified as almost negligible coupling with the neighbours modes. In a
ec PE forthcoming publication we will report on the possibility of
Is ≈ (51) exploiting an advanced Bragg reflector [14] to suppress the
g0
unwanted modes near the cutoff region. We will also comment
The small signal gain coefficient g0 can be determined from on the higher order harmonics CARM emission accompanying
the the identity discussed in the previous section. However the fundamental operating mode and make comparison with
assuming a total gain of 2.1, we obtain g0 = 1.8 after inverting gyrotron devices operating at second harmonics [15].
the relation [3], [4]
ACKNOWLEDGMENTS
G M ≈ 0.85g0 + 0.19g02 (52)
During the compilation of the present paper we have
We can fix g0 ≈ 1.8, a value compatible with an e-beam enjoyed many helpful discussion with many colleagues.
current of 5 A, which for a kinetic energy of 0.7 MeV amount In particular it is a pleasure to recognize discussions
to an e-beam power of 3.5 MW, therefore with an efficiency with Drs. G. P. Gallerano, F. Mirizzi, A. Tuccillo.
of 30% we find Is ≈ 0.6 MW. We can now fix the value of It is also a pleasure to recognize the help of
the cavity quality factor compatible with the previous requests. Professors N. Ginzburg and M. Glyavin for correcting
In Fig. 9 we have reported the ratio Ie /Is vs. the active losses; many of our misconceptions on CARM sources.
if we furthermore assume an equilibrium intracavity power of
about 2 MW, we get Ie /Is = 3.3, corresponding to active R EFERENCES
losses around 0.3 (Fig. 9) and, accordingly, to a Q-factor [1] V. L. Bratman, N. S. Ginzburg, G. S. Nusinovich, M. I. Petelin, and
below 104 (Fig. 7). P. S. Strklkov, “Relativistic gyrotrons and cyclotron autoresonance
Before closing this paper we should clarify some points masers,” Int. J. Electron., vol. 51, no. 4, pp. 541–567, 1981.
[2] A. W. Fliflet, “Linear and non-linear theory of the Doppler-shifted
relevant to the possibility of realizing an optical cavity with cyclotron resonance maser based on TE and TM waveguide modes,”
Q values corresponding to those we have previously derived. Int. J. Electron., vol. 61, no. 6, pp. 1049–1080, 1986.
CECCUZZI et al.: HIGH GAIN INTEGRAL EQUATION FOR CARM-FEL DEVICES 1100209

[3] G. Dattoli, A. Renieri, and A. Torre, Lectures on the Free Electron Laser Emanuele Di Palma was born in Ceprano, Italy, in 1969. He received the
Theory and Related Topics. Singapore: World Scientific, 1993. Laurea degree in mathematics from the La Sapienza University of Rome,
[4] G. Dattoli, P. L. Ottaviani, and S. Pagnutti, Booklet for FEL Design: Italy, in 1996.
A Collection of Practical Formulae. Frascati, Italy: Edizioni Scientifiche He was involved in the implementation of numerical methods for elec-
Frascati, 2007. tromagnetic wave propagation on different applications from 1997 to 2001.
[5] V. L. Bratman, N. S. Ginzburg, and M. I. Petelin, “Common properties He was appointed as a permanent position with the Italian Agency for
of free electron lasers,” Opt. Commun., vol. 30, no. 3, pp. 409–412, the Environment and Energy (ENEA), Rome, Italy, in 2003, where he has
1979. been collaborating with the Laboratory of Theoretical Physics and Applied
[6] T. Marshall, Free Electron Lasers. New York, NY, USA: Macmillan, Mathematics since 2011, and is involved in research on free electron laser
1985. application. His current research interests include the theory of special
[7] A. Gover, A. Yariv, and P. Yeh, “Prospectives of free electron lasers,” functions and application of analytical and numerical techniques for solving
Opt. Commun., vol. 18, no. 2, pp. 221–222, 1976. electromagnetic scattering and propagation problems for high-frequency high
[8] A. Gover, “Wave interactions in periodic structures and periodic dielec- power devices. He was a recipient of grants from ENEA, in collaboration
tric waveguides,” Ph.D. dissertation, California Inst. Technol., Pasadena, with the Italian Institute of high Mathematics from 1997 to 2001.
California, 1976
[9] G. Dattoli, S. Pagnutti, P. L. Ottaviani, and V. Asgekar, “Free electron
laser oscillators with tapered undulators: Inclusion of harmonic gen-
eration and pulse propagation,” Phys. Rev. ST Accel. Beams, vol. 15, Andrea Doria received the Laurea degree in
p. 030708, Mar. 2012. physics from the La Sapienza University of Rome,
[10] K. D. Pendergast, B. G. Danley, R. J. Temkin, and J. S. Wurtele, “Self- Italy, in 1987.
consistent simulation of cyclotron autoresonance maser amplifiers,” He collaborated in designing and operating user
IEEE Trans. Plasma Sci., vol. 16, no. 2, pp. 122–128, Apr. 1988. application experiments on terahertz interaction
[11] F. Mirizzi et al., “A high frequency, high power CARM pro- with biological organisms, solid-state systems, and
posal for the DEMO ECRH system,” Fusion Eng. Design, 2015, cultural heritage samples using the compact free
doi: 10.1016/j.fusengdes.2015.02.026. electron laser (FEL) sources of ENEA. He has been
[12] N. F. Kovalev, I. M. Orlova, and M. I. Petelin, “Wave transformation a Permanent Staff Member with the Experimental
in a multimode waveguide with corrugated walls,” Radiophys. Quantum FEL Group, ENEA, Frascati, Italy, since 1988.
Electron., vol. 11, no. 5, pp. 449–450, 1968. He has been the Principal Investigator of the joined
[13] C. K. Chong et al., “Bragg reflectors,” IEEE Trans. Plasma Sci., vol. 20, ENEA, INFN, and CNR research project named SPARC/SPARX, for the real-
no. 3, pp. 393–402, Jun. 1992. ization of a vacuum ultra violet and soft X-ray FEL operating in self-amplified
[14] N. S. Ginzburg et al., “Improving selectivity of free electron maser with spontaneous emission and external seeding regime and funded by the Italian
1D Bragg resonator using coupling of propagating and trapped waves,” Minister of Research. His research interests are Cerenkov-based FELs, com-
Phys. Rev. ST Accel. Beams, vol. 8, p. 040705, Apr. 2005. pact FEL sources in the millimeter and far IR regions, and coherence and cor-
[15] T. Idehara et al., “Development of a second cyclotron harmonic gyrotron relation effects on FEL power output. He is a member of the team for the study
operating at submillimeter wavelengths,” Phys. Fluids B, vol. 4, no. 1, of a CARM-type source for electron cyclotron resonance plasma heating.
pp. 267–273, 1992.

Elio Sabia was born in Naples, Italy, in 1952.


He received the Ph.D. degree in physics from the
University of Naples, Italy, in 1978. His dissertation
Silvio Ceccuzzi was born in Rome, Italy, in 1983. concerned the study of the magnetic structure of a
He received the Italian (cum laude) degrees in elec- storage ring free electron laser (FEL).
tronics engineering and the Ph.D. degree in applied He was involved in the development of a sim-
electronics from Roma Tre University, Italy, in 2005, ulation code for a XeCl excimer laser in 1984.
2008, and 2015, respectively. He has been a Staff Member with the ENEA
After an internship with Telespazio and another Research Center of Frascati, Rome, Italy, since 1981.
one with Thales Alenia Space, in 2009, he started Since 1981, he has been involved in theoretical
a 3-years scholarship for scientific training across problems connected with particle dynamics in high-
European laboratories in RF systems for nuclear energy electron accelerators driven FELs. His current major areas of interest
fusion. In 2013, he joined ENEA for EUROfusion, and research include experimental activities on SPARC FEL, and theoretical
Frascati, Italy, as a Researcher. His main interests activities regarding the study on the theory of classical and quantum relativistic
are microwave components, nuclear fusion, and electromagnetic bandgap. equations and on the solution of nonlinear partial derivative equations.
Dr. Ceccuzzi has authored or co-authored over fifty works on technical
reports, journals, and conference proceedings. At the 19th Italian Meeting on
Electromagnetics (RiNEm 2012), he won the Sannino Award for outstanding
research on microwave components.
Ivan Spassovsky received the M.S. degree in quantum electronics and the
Ph.D. degree in plasma physics from Sofia University, Bulgaria.
For a couple of years, he participated the construction and test of Korean
infrared free electron laser (FEL) with KAERI, Daejeon. In 2002, he joined
the Italian Research Center, ENEA. At ENEA, he worked initially on a
Giuseppe Dattoli was born in Lagonegro, Italy, in 1953. He received the subterahertz FEL experiment. From 2006 to 2009, he was part of the team
Ph.D. degree in physics from the La Sapienza University of Rome, Italy, working on the realization of SPARC FEL facility. Since 1983, he has
in 1976. been part of different teams working on high power microwaves as the
He has been an ENEA Researcher and has also been involved in different development of 28-GHz gyrotron for plasma fusion activity with INPE, Brazil,
research projects, including high energy accelerators, free electron lasers, and and the design study and experimental work on 17-GHz second harmonic
applied mathematics networks since 1979. gyroklystron with the University of Maryland, USA. He is currently with the
Dr. Dattoli has taught in Italian and Foreign universities, and has received Department of Applied Mathematics and Computation, where he is involved
the FEL Prize Award for his outstanding achievements in the field. in the development of 250-GHz RF source for tokamak plasma heating.

You might also like