© 1993 IEEE. Personal use of this material is permitted.

However, permission to reprint/republish this material

for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers
or lists, or to reuse any copyrighted component of this work in other works must be obtained from the IEEE.

Propagation of Short Electron Pulses in Underdense Plasmas

N. Barov and J. Rosenzweig,
Department of Physics,
University of California, Los Angeles, CA 90024

Abstract can be divided into three qualitatively different regions.

The extreme leading edge, or head of the beam receives
Our program for an experimental plasma wake no focusing from the plasma, causing it to expand. The
field accelerator (PWFA) to take place at the Argonne body of the beam, which travels in the completely
Wakefield Accelerator (AWA) facility, in the recently rarefied ion channel, receives the maximum focusing
proposed blow-out regime]11 relies on the propagation force. III the trmsition region between these two, the
of an intense electron beam through an underdense beam evolution cannot be described with linear optics,
plasma with a minimum of degradation. This paper due to the remaining population of plasma electrons.
presents a near-equilibrium model of beam propagation For short pulses (az=kp‘l, kp = op/c), the evolution of
using the Maxwell-Vlasov equations governing the the beam head and the transition region are very
beam’s transverse behavior. Numerical results are important in determining the effective propagation of
presented which use this model simultaneously with the the beam over long distances.
plasma electron cold fluid equations. A solenoidal For the ultra-relativistic case, the beam head
magnetic field, which is necessary for high density expands freely, due to its finite emittance, retarding the
plasma containment, also provides an initial beam response of the plasma electrons at later time steps. This
equilibrium to begin the calculation. We compare the diminishes the focusing force for the next beam slice,
equilibrium mode1 with a discrete beam particle leading to what is called emittance driven erosion. A
simulation, which verifies the basic conclusions of the simple l-d model predicts that after some initial
equilibrium model, and shows the collisionless damping expansion, the erosion happens at a very slow rate. In
approach to equilibrium in the beam head. The initial addition, previous particle-in-cell simulation work by
matching requirements for the beam’s entry into the Krall, et al. [2] has shown that for emittancc-driven
plasma are examined. We also discuss the possibility of erosion, the point on the beam where the plasma
performing an adiabatic lens experiment. becomes completely rarefied, termed the pinch point,
Introduction moves very slowly, and a near equilibrium develops.
Inspired by this result, we develop a model in which
beam physics is described by a Maxwell-Vlasov
The recently proposed nonlinear blow-out regime
equilibrium and the plasma electrons are described by
for the PWFA[l], in which all of the plasma electrons
fully relativistic cold fluid equations. This model also
become expelled from the region of the beam, has the
includes the effects of a solenoidal magnetic field. Such a
advantage of linear focusing for the trailing
field is required for the containment of the plasma in the
(accelerated) bunch and a very high acceleration
PWFA, which can in some cases help to further stabilize
gradient. This regime requires a very intense, high
this erosion.
quality drive beam. Since the acceleration can be
prolonged by using a more energetic drive beam, the Numerical Treatment
long term propagation of this beam through the plasma of Beam Near Equilibrium
needs to be addressed.
When an intense electron beam propagates The response of the plasma due to the beam’s
through a plasma in the under-dense regime (no < nbem, electromagnetic field is modeled using a technique
no is the plasma density), a sufficiently long bunch developed by Breizman [3]. This model relies on a wake
length will cause complete rarefaction of the plasma field type assumption: any plasma perturbation
electrons, forming an ion channel. This is called the ion translates at the beam velocity. Further, the beam
focusing regime (IFR) due to the intense magnetic self- velocity is taken to be the speed of light. The plasma
focusing forces of the beam as it propagates through the electron currents can then be modeled by the cold fluid
ion channel. For a fully rarefied ion channel, the equation,
equilibrium beam radius cr is given by,
l/2
Or=
LJz& I
and the continuity equation. Implicit in using a cold
fluid model is the fact that the plasma electron velocity
where .s,,is the normalized emittance, re is the classical quickly becomes much greater than the initial
electron radius, and y is the Lorentz factor. The beam temperature of the plasma. The plasma can be treated as
0-7803-1203-l/93$03.00 0 1993 IEEE 2623
PAC 1993
 Propagation of Short Electron Pulses in Underdense Plasmas
N. Barov and J. Rosenzweig,
Department of Physics,
University of California, Los Angeles, CA 90024

Abstract can be divided into three qualitatively different regions.

The extreme leading edge, or hean of the beam receives
Our program for an experimental plasma wake no focusing from the plasma, causing it to expand. The
field accelerator (PWFA) to take place at the Argonne bony of the beam, which travels in the completely
Wakefield Accelerator (AWA) facility, in the recently rarefied ion channel, receives the maximum focusing
proposed blow-out regime111 relies on the propagation force. In the trurrsition region between these two, the
of an intense electron beam through an underdense beam evolution cannot be described with linear optics,
plasma with a minimum of degradation. This paper due to the remaining population of plasma electrons.
presents a near-equilibrium model of beam propagation For short pulses (oz=kp-l, kp = op/c), the evolution of
using the Maxwell-Vlasov equations governing the the beam head and the transition region are very
beam’s transverse behavior. Numerical results are important in determining the effective propagation of
presented which use this model simultaneously with the the beam over long distances.
plasma electron cold fluid equations. A solenoidal For the ultra-relativistic case, the beam head
magnetic field, which is necessary for high density expands freely, due to its finite emittance, retarding the
plasma containment, also provides an initial beam response of the plasma electrons at later time steps. This
equilibrium to begin the calculation. We compare the diminishes the focusing force for the next beam slice,
equilibrium model with a discrete beam particle leading to what is called emittance driven erosion. A
simulation, which verifies the basic conclusions of the simple l-d model predicts that after some initial
equilibrium model, and shows the collisionless damping expansion, the erosion happens at a very slow rate. In
approach to equilibrium in the beam head. The initial addition, previous particle-in-cell simulation work by
matching requirements for the beam’s entry into the Krall, et al. [2] has shown that for emittancc-driven
plasma are examined. We also discuss the possibility of erosion, the point on the beam where the plasma
performing an adiabatic lens experiment. becomes completely rarefied, termed the pinch point,
Introduction moves very slowly, and a near equilibrium develops.
Inspired by this result, we develop a model in which
The recently proposed nonlinear blow-out regime beam physics is described by a Maxwell-Vlasov
for the PWFA[l], in which all of the plasma electrons equilibrium and the plasma electrons are described by
fully relativistic cold fluid equations. This model also
become expelled from the region of the beam, has the
includes the effects of a solenoidal magnetic field. Such a
ad vantage of linear focusing for the trailing
field is required for the containment of the plasma in the
(accelerated) bunch and a very high acceleration
gradient. This regime requires a very intense, high PWFA, which can in some cases help to further stabilize
this erosion.
quality drive beam. Since the acceleration can be
prolonged by using a more energetic drive beam, the Numerical Treatment
long term propagation of this beam through the plasma of Beam Near Equilibrium
needs to be addressed.
When an intense electron beam propagates The response of the plasma due to the beam’s
through a plasma in the underdense regime (no < nbeZm, electromagnetic field is modeled using a technique
no is the plasma density), a sufficiently long bunch developed by Breizman [3]. This model relies on a wake
length will cause complete rarefaction of the plasma field type assumption: any plasma perturbation
electrons, forming an ion channel. This is called the ion translates at the beam velocity. Further, the beam
focusing regime (IFR) due to the intense magnetic self- velocity is taken to be the speed of light. The plasma
focusing forces of the beam as it propagates through the electron currents can then be modeled by the cold fluid
ion channel. For a fully rarefied ion channel, the equation,
equilibrium beam radius (Jr is given by,
112
CT-= r
-’ [-J2vw
El, 11 and the continuity equation. Implicit in using a cold
fluid model is the fact that the plasma electron velocity
where e,,is the normalized emittance, re is the classical quickly becomes much greater than the initial
electron radius, and y is the Lorentz factor. The beam temperature of the plasma. The plasma can be treated as
O-7803.1203.1/93$03.00@1993IEEE
2623
PAC 1993
a fluid only when the motion is laminar. This is satisfied
to a very good approximation in the first half cycle of In Figure 1, this method is applied to a lOi cm3
the plasma motion, making the cold fluid model well plasma, using the future AWA beam parameters 141:100
suited to studying the effects on the beam. These nC charge per bunch, ~~ = .7 mm, y = 300, and E, =400 x
equations are used along with the Maxwell equations for mm-mrad. The final peak density mimics the initial
V x E and V x B. The speed of light condition, gaussian density quite well. This run produced a
3/& = -a//act is used to eliminate the time variable in longitudinal wake field capable of sustaining greater
these equations, which yield a self consistent than a 1 GeV/m acceleration.
instantaneous representation of the plasma disturbance.
We use conducting wall boundary conditions in the Initial Matching: Time Dependent Aspects
radial direction and further assume that the plasma is
quiescent ahead of the beam. The equilibrium model of the last section relies on
The body of the beam is assumed to be initially a trumpet shaped beam, an condition which may be
matched to the linear focusing force of the ion channel. difficult to produce from a longitudinally uncorrelated
For the transition region, the nonlinear focusing brings profile. In this section we examine the requirements for
about a mixing of the transverse phase space in just a initial matching into the focusing channel and present
few betatron periods. If we assume that the emittance computational results which reveal the efficacy of such
increase, which is very difficult to compute, can be matching. In order to minimize the betatron oscillations
neglected, the phase space can be described by the of the beam head, the largest possible initial beam radius
equilibrium Maxwell-Vlasov equation, setting a//at = 0: must be used. This is important when the period of
Pr Jf(w,) +F (r) af(r,p,) = o
these oscillations is longer than the stopping distance of
$f(r.p,) = GT r the beam through the plasma. It is therefore
3pr
advantageous to use the focusing properties of the
where f(r,p,) is the distribution in the transverse phase
plasma in such a way as to reduce the radius of the
space, pr is the transverse momentum, y is the Lorentz
beam body, while leaving the beam head unaffected.
factor, m is the electron mass, and F, is the radial force
The initial ramp in the plasma density at the start of any
arising from the plasma fields. In addition, f(O,O)must
actual device can be useful for this task. We have
be constant due to conservation of phase space area.
studied the effect of a plasma profile which builds up as
The matching of the beam body implies that the
one side of a gaussian and remains flat afterward.
head will undergo betatron oscillations. For the purpose Although this does not represent the ideal focusing
of the model, we assume that the head is itself matched scheme, it is a good approximation of the experimental
to the solenoid. This is in part justified by the small
situation.
number of particles which feel no plasma focusing.
40 LII1 ~““‘l”“““‘l”“““‘~ 0.035 zj
----“beam (Ol/no
35
30
25
20
15
10
\
E L
or tail (cm)
0.0050 r- __ - ur for entire beam (cm)
I
5.0 \
/J ‘-‘I
1111111111,~1111111111 0.0 ’ ” ” ” ’ ” ” ’ ” ” ”
15 10 5 0 0 5 10 15 20
Distance behind Beam Head (k ml Propagation Distance (cm)
P 1
Figure 1. Final (equilibirum) and initial beam density Figure 2. Matching of the beam into plasma with a
normalized to plasma density using AWA beam gaussian ramp, IS = 2 cm. For this run, the beam has a
parameters, no =1014 cmm3,and a solenoidal field of virtual waist of 0.026 cm at a position 4 cm into the
B=2000 Gauss. The matched equilibrium is achieved device. The magnetic field was 2000 Gauss.
very early in the the beam pulse.
2624
PAC 1993
 To simulate these time dependent effects, we used is more realistic experimentally. Figure 3 presents
the same numerical techniques as in the last section to simulation results for such a lens. The body of the beam
treat the plasma dynamics. The beam in this case was is focused, in good agreement with the theoretical limit
composed of a set of discrete particles, treated self- assuming a completely rarefied ion channel. Note that
consistently with the plasma dynamics. It is assumed the or averaged over the entire beam never becomes
that the variation in no over the extent of the beam can small due to phase differences and the insufficient
be neglected. Figure 2 shows results, using the AWA focusing of the beam head.
beam parameters, for ramp which builds up as a
gaussian from no = lOi cmF3to no = lOI in 8 cm, and
0.060 tl
remains flat afterward. A slight mismatch of the beam
body is tolerable since it does not diminish the radial 0.050
force seen by the plasma electrons, which have already
left the region of the beam. 0.040
Through observing the evolution of the phase
space we have witnessed a new mechanism for the
suppression of erosion of the beam head. The presence 0.030
of any erosion diminishes the focusing strength of a
given z-slice very slowly. Thus, the slice’s or is 0.020
adiabatically increased until it becomes matched to the
solenoid, approximating the equilibrium discussed in 0.010

0.01
the last section. In the case of a weak solenoid, this
predicts that a small portion of the beam head will
become lost, while the rest of the beam propagates in a 0 4 8 12 16 20
near equilibrium. Z (cm)

Adiabatic Focusing Figure 3: Adiabatic lens for the AWA beam, no starts at
2.5 xlO1l and increases to 5 x1013 at z=15 cm. The body
There has been great interest in using a plasma of the beam (the core population containing half the
lens for the final focusing in a linear colliderJ51. Chen et beam) is focused quite efficiently, down to 100 microns
al. [61 has proposed making such a lens adiabatic, with a at z=16 cm.
smoothly increasing focusing strength. The advantages
of adiabaticity are to reduce the effects of synchrotron This work was supported by US Department of
radiation, lessening impact of the chromatic aberrations, Energy under Grant DE-FG03-92ER-40493
and diminishing the sensitivity to initial optical
mismatch. Because the equations governing the plasma References
motion have a simple scaling with no, the results of a
sub-GeV, experiment can yield a good prediction for [ll J.B. Rosenzweig, B. Brcizman, T.Katsouleas and J.J.
larger y and an no of lOis cme3 In addition, as no Su. Phys. Rev. A - Rapid Comrn, 44, R6189 (1991).
increases, the condition oz >> kp-l becomes easier to [21 J. Krall, K. Nguyen, G. Joyce, Phys. Rev. A, 44, (1991)
satisfy. Thus, a proof of principle experiment at low [3] B.N. Breizman, T. Tajima, D.L. Fisher, and P.Z.
energy would circumvent some of the experimental Chebotaev, University of Texas Fusion Institute Preprint
difficulties of an actual final focus experiment. The (19911,unpublished
AWA beam, with its high peak current, and 141 M. Rosing, et al., An &date on ArPonne’s AWA,
accompanying access to the underdense regime, can be 1991 IEEE Particle Accelerator Conference, I’.555
used to experimentally study such a device. The discrete [51 See B. Barletta, et al. , and references therein Plasma
particle method developed above has been used to Lens Exneriment at the Final Focus Test Beam, these
numerically study the adiabatic lens. The AWA beam is proceedings.
intense enough to cause blowout at very early times 161 I’. Chen, K. Oide, A.M. Sessler, S.S. Yu, Particle
after the arrival of the beam. In addition, the charge and Accelerators, 31(1990)
energy can be scaled down to study a large parameter
space.
The work by Chen, et al., suggests the use of a lens
in which the 0 function is linearly decreased. This lens
has the feature that the plasma density is increasing very
rapidly near the end. We have studied a lens whose
density increases only exponentially, which we believe

2625
PAC 1993