Plasma Electron Trapping and Acceleration in a Plasma Wake

Field Using a Density Transition

H. Suk, N. Barov, and J. B. Rosenzweig

Department of Physics and Astronomy
University of California Los Angeles
Los Angeles, CA 90095

E. Esarey
Center for Beam Physics,
Ernest Orlando Lawrence Berkeley National Laboratory,
Berkeley, California 94720

Abstract

A new scheme for plasma electron injection into an acceleration phase of a plasma wake-
field is presented. In this scheme, a short single electron beam bunch is sent through an
underdense plasma with a sharp, localized downward density transition. Near this
transition, a number of background plasma electrons are trapped in the plasma wake field
due to the rapid wavelength increase of the wake wave in this region. The viability of this
scheme is verified using two-dimensional particle-in-cell (PIC) simulations . To investigate
the trapping and acceleration mechanisms further, a 1-D Hamiltonian analysis as well as 1-
D simulations have been performed, with the results are presented and compared.

PACS Codes: 52.40.Mj, 52.75.Di, 29.17.+w, 29.27.-a

Submitted to Physical Review Letters

 Compared to standard radio-frequency linear accelerators, advanced accelerators
using plasmas can produce a much higher acceleration gradient, in excess of 1 GeV/m.
Hence, extensive research on the plasma-based accelerators[1-4] has been performed in
recent years. For plasma-based accelerators, short, intense laser (laser wakefield
accelerator, or LWFA) or electron beam pulses (plasma wakefield accelerator, or PWFA)
are used to drive large amplitude plasma waves. In these schemes, the maximum
achievable accelerating gradient scales as the nonrelativistic plasma frequency, which is
given by ω p = (4πn0 e2 / me )1/2 , where n0 , e, and me denote the plasma density, electron
charge, and electron mass, respectively. In order to obtain a beam with small energy
spread, an very short (<< 1 psec) accelerating pulse must be injected into such a short
wavelength system This requirement is challenging to meet with an external injector,
however. This difficulty, which is more pressing in the LWFA, led to proposal of all
optical methods plasma electron injection schemes using two[5] or three laser pulses[6].
However, these optical methods require extremely accurate laser spatial and temporal
overlap, which again leads technical difficulties. To avoid these problems, Bulanov et al.
[7] proposed a self-injection scheme using a single laser pulse propagating in an
inhomogenous plasma. This scheme uses trapping of background plasma electrons from
wavebreaking induced by a gentle density decline, in which the density scale length
Ls = n0 / dn0 dz is much larger than the plasma skin depth k p−1 = v b / ω p , where vb ≅ c is
the driving pulse velocity. Bulanov’s scheme is much simpler than other plasma injection
methods, but leads to an injected beam with a relatively large phase spread. In addition, the
use of LWFA, in which the accelerating and focusing fields are fairly nonuniform, may not
in the end give as high of beam quality as the blowout regime of the PWFA[8,9], in which
a plasma electron-rarefied region is formed in the wake of the driving electron beam. Inside
of this rarefied region, an accelerating electron experiences acceleration dependent only on
longitudinal position, and focusing which is linear in offset from the axis, just as in more
conventional accelerators is still in direct contact with the plasma electron charge and
current.
In this Letter, we propose a new self-injection scenario, for the PWFA in the blowout
regime, where the beam density greater than the plasma density, nb > n0 (underdense
condition). It is known that in the PWFA in one-dimension, that self-trapping of
background electrons by the wave is very difficult[10,11], and it is conjectured that it is
even more difficult in two or three dimensions. Thus we propose to introduce a
discontinuity into the PWFA in order to self-trap plasma electrons. In this scheme, a single
short electron beam pulse is sent through an underdense plasma with a sharp downward
density transition k p Ls <1 marking the boundary between a dense upstream region (I) and
a less dense downstream region (II). In the blowout regime, the driving electron beam
generates a large wake field that is supported by plasma electron motion, which is initially
(in the wave’s decelerating phase) radially outward, and counter to the beam motion in z.
These initial directions are reversed in the accelerating phase of the wave, and the plasma
electrons eventually return to the axis in a large density spike, to near their original position
in z.
When the beam passes a sharp downward plasma density transition, the wavelength
of the plasma wake wave changes rapidly. In this situation, the plasma electrons just
inside region II spend much of their oscillation in region I before returning to near their
initial position in z advanced in wave phase compared to the expected cw region II
oscillation. At this position, for the expected electric field in a cw wave is zero, but the
faster oscillation of the electrons in region I in the transition case allows the field in region
II to be still accelerating. This proposed dephasing mechanism, which is justified in more
qualitative and quantitative detail below, allows plasma electrons to be trapped into the
acceleration phase of the wake field in region II. This phenomenon is shown in Fig. 1,
which displays a two-dimensional particle-in-cell simulation using the code MAGIC. This
injection mechanism is fundamentally different from the Bulanov's case for the LWFA. In
the Bulanov's case, plasma electrons are trapped in the second plasma oscillation period .
The trapped particles come from wavebreaking in the first plasma oscillation period in this
case, while in the new injection scheme proposed in this Letter, trapping of the plasma
electrons occurs in the first rarefied cavity and this is due to the sudden increase of the
plasma wavelength at the sharp density transition. The ion channel formed from blowout of
nearly all plasma electrons in the first rarefied cavity does not have plasma electrons so that
the trapped electrons do not have scattering with other plasma electrons during acceleration.
Furthermore, they are linearly focused in the electron-free ion channel[8,9]. Therefore, a
better beam quality would be possible in this new scheme.
We have explored density transition-induced particle trapping using 2-D PIC
simulations (both MAGIC, and a UCLA-specific code, NOVO-PIC) in order to illuminate
the physical mechanisms relevant to the trapping process. The simulation results shown in
Fig. 1 were obtained with an ambient plasma density of n0I = 5 ×1013 cm-3 for k pII z <11.2 ,

and n0II = 3.5 × 1013 cm-3 for k pII z ≥ 11.2 , plasma electron temperature kTe = 3 eV, and
stationary ions. Here we are parameterizing lengths in terms of the plasma skin-depth in
region II, k pII = 4πren0II . The ultra-relativistic drive beam used in the simulation had a bi-
 2
/ 2σ r 2 −ξ 2 / 2σ z 2
Gaussian density distribution nb ( r, ξ ) ∝ e
−r
e ( ξ = z − vb t ), of peak density
I II
nb = 2.4n0 = 3.4n0 , and dimensions k p II σ z = 1 , and k p II σ r = 0.56 . In Fig.1(a) we
observe the trapped population of electrons just as they return to near their initial position,
and the associated structure of the wake wave, which at this somewhat complicated at this
point. In Fig. 1(b), it is observed that, after traversing an additional few plasma
wavelengths past the boundary, the trapped plasma electron bunch is loaded into a well-
behaved blow-out regime wave. Further, this injected electron population is transversely
controlled by the uniform ion-focusing in the blow-out region of the wave.
The longitudinal phase space of the plasma electrons at the two times corresponding
to Figs. 1(a) and 1(b) are shown in Fig. 2. Figure 2(a) illustrates the significant amount of
plasma electrons as they are injected into the acceleration. Figure 2(b) shows that the main
trapped plasma electron have attained excellent momentum separation between trapped
particles (in the range of 5 ~ 15 MeV/c) and the background plasma. The population of
trapped particles occupies a longitudinal phase extent of δφ / 2π ≅ 9 %, with a large charge
of about 0.5 nC for these parameters. Careful comparison of Figs. 2(a) and 2(b) indicates a
very little phase slippage of the trapped electrons during acceleration.
In systematic simulation studies of this scheme, we varied the following parameters
describing the density transition: the sign of density change, the amplitude of the density
difference across the transition, and the density transition scale length. In all simulations,
the driving beam and plasma characteristics were kept similar, in which blowout is
complete and the plasma electron motion is nonlinear and moderately relativistic. In these
studies we observed trapping only in the case of transitions in which the density decreased
along the direction of the beam travel. In addition, it was found that the number of trapped
particles increased as the amplitude of density decrease was made larger. Finally, if the
length over which the density was linearly decreased from its initial to final value becomes
−1
larger than a plasma skin-depth k p = c / ω p , then the trapping disappears completely.
A further, definitive clue as to the trapping mechanism is that the trapped electrons
all initially dwell, as stated above, in region II. These observations have led to the
development of our trapping model, in which electrons are excited by the driving beam
passage initially travel in the negative z direction (as well as the positive radial direction)
under the repulsive space-charge force of the beam. Upon entry into region I, the electrons
experience wave fields which have a shorter oscillation wavelength, and can be dephased
and trapped. This model also qualitatively explains why the density scale length in the
transition must be short. If it is longer than the oscillation distance in z, then the plasma
electrons cannot access the region of significantly higher plasma electron density.
At this point, we would like to place our model on firmer analytical ground. As a
useful two-dimensional analytical model of plasma electron motion in the blowout regime
does not exist for the simplest wave cases, much less waves in the presence of a density
transition, we must begin with a more tractable model. In order to quantify our trapping
model, therefore, we proceed to develop a one-dimensional, fully relativistic analysis of
plasma motion in both regions, and propose a scenario in which the motion of the plasma
electrons over is not strongly affected by the existence of the transition. A Hamiltonian
analysis of the phase space trajectories of trapped electrons is thus made possible. This
analytical model is then compared to the results one-dimensional particle-in-cell
computations, in order to verify some aspects of the model and also point to limitations of
the model. Without discussing the model at all, some limitations are immediately apparent.
First, in the blowout regime, the plasma electron density vanishes in the wake of the drive
beam, while in one dimension nonlinear wave theory, the plasma density does not drop
below one-half of the initial ambient density n0 . In addition, an approximately one-
dimensional wave driven by an electron beam is somewhat of a practical improbability, as it
implies beam charge densities in excess of what are presently found. One-dimensional,
nonlinear plasma waves may be driven by ultra-short laser pulses, however, and so the
present analysis may have some direct application to density transitions in the LWFA.
Because of this, as well as its heuristic value, we proceed with the analysis outlined above.
Note that a large amplitude, nonlinear wave is necessary to allow the trapping process we
wish to describe, as for waves with amplitude well below the wavebreaking limit, the
plasma electrons do not move appreciably from their initial position, and do not have
relativistic velocities.
We begin the analysis by reviewing the nonlinear 1-D plasma wake-field theory
based on the fundamental work of Akhiezer and Polovin[12], and recent contributions in
the context of plasma accelerators[13-15]. In this theory, with the wave ansatz assumed
(all spatial and temporal dependences of the system can be expressed using only the
quantity τ = ω p (t − z /v b ) ), the differential equation governing relativistic cold fluid takes
the following form,

d2  1 − β bβ  2 β nb 
 2 
= β + , (1)
dτ 2  1− β   βb − β n0 
b

where nb is the beam density, βb = vb / c is the normalized beam (as well as wave phase)
velocity, andβ = v /c is the normalized plasma electron velocity.
Since the plasma wave is impulsively excited by the driving beam, we concentrate
on the plasma dynamics in the region behind the beam. There, for a very relativistic beam
(βb →1), this equation can be rewritten as

1 1  1− β
x ′′ = 2 − 1 , with x = (2)
2 x  1+ β

This fluid equation is equivalent both to the Poisson equation and to the equation of plasma
electron equation of motion, so we can write the field and source quantities
n 1
= 1 + 2  , −eφ = m ec 2 (1 − x ) ,
1
and −eE = k pm ec2 x ′ , as well as the dynamical
n0 2 x
1 1 1 − x2 1 1
quantities γ = x + , β = , and βγ = − x , in terms of x.
2 x 1+ x 2 x
2

With these results, we can perform a Hamiltonian analysis of the plasma electron
motion. In the Galillean frame moving with the wave, the Hamiltonian is a constant of the
motion

H = −eφ + E − p = m ec2 (1 − x f ) + m ec2 xe . (3)

Here the quantity x f corresponds to the field (and the fluid electron dynamical) properties,

as outlined above and xe ≡ (1 − β e ) /(1 + βe ) describes the dynamical state of any electron,
fluid or otherwise, injected into the system. As the Hamiltonian is invariant, the difference
∆x = x f − x e is a constant of the motion. Knowledge of the value of H and the function x f
thus allows one to map the electron trajectories in phase space, and easily determine which
trajectories are trapped. With the Hamiltonian as written above, electrons with ∆x > x f ,min
are trapped, as when x f → x f ,min , xe approaches zero, and βe → 1. With this model in
mind, we can now discuss the approximate effects of the density discontinuity.
In this analysis we assume that the density discontinuity can be modeled by using
the nonlinear wave analysis in both regions on either side of a sharp density transition. The
fields on both the upstream and downstream sides of the moving (in the τ frame) boundary
are assumed to be given by the wave analysis. Note that at the boundary, the value of x f is
discontinuous, which implies a new value of H when the boundary is crossed by an
electron. It is this discontinuity in H which mathematically allows the trapping of initially
cold plasma electrons ( xe ≅ 0 ) in an ultrarelativistic phase velocity wave.
We also note that, because the system is one dimensional, at the density transition
boundary, which moves backwards in the wave frame at −v b ≅ −c , there is a self-
consistent discontinuity in the electric field of ∆E = 4πσ . Here σ is the net surface charge
density “absorbed” by the boundary due to displaced electrons from the upstream side, and
“emitted” by those which would be displaced to the downstream side. The induced field
discontinuity at the boundary is illustrated in Fig. 3. Note that while the peak field two
regions is the same, the field in the downstream, high density region has a larger spatial
gradient (higher plasma frequency).
While the electric field discontinuity at the density transition is self-consistent in
terms of displaced charge, it is of course unphysical in the sense that the fluid electrons are
not emitted or absorbed by the boundary between the two regions. The fictional absorption
at the boundary is also conceptually at odds with the physical model of trapping, in which
the plasma particles move significantly past the boundary. Nevertheless, we find good
agreement with this model and one-dimensional PIC simulations, as shown in Fig. 1.
Here we also displays the simulation-derived electric field associated with the same
physical parameters (beam surface charge density of 0.36 µC/cm2 , n0 = 3.5 ×1013 cm -3 for

z>0.5 cm, n0 = 5 ×1013 cm -3 for z<0.5 cm) as used in the analytical model. The agreement
is very good for the fields calculated in the region z<0.5 cm, which is the important zone
for electron trapping dynamics, as we shall see.
In order to determine how to connect the two values of H (i.e. when the electrons
encounter the boundary) in the two regions of interest, the position of the electrons as a
function of τ must be known. This can be accomplished by integrating the velocity,

τ τ
c 2  β  1 2 1 − xe2 
t2

z = c ∫ βdt =
ω p τ∫1  β −1  k p τ∫1  2x e2 
dτ = dτ (4)
t1

For the fluid electrons, x f = x e and we have k p z = 21 ∫ [n(τ ) − n0 ]dτ , a result which
emphasizes that the fluid electron motion in the wave is initially in the negative direction,
and arrives back at the starting position after one period.
Using the formalism we have developed, we can compare the dynamics of a
trapped electron from the 1D PIC simulation, to those derived from the analytical model.
The electron we follow in the simulation is initially located at z = 5.1cm (1 mm
downstream of the density transition). To employ the theoretical model, we must follow
the electron initially in the low density (downstream) region, calculate its position and
momentum crossing the transition, repeat the procedure in the high density (upstream)
region, and ultimately follow it back into the downstream region. The results of this
exercise are shown in Fig. 4, which displays the phase space trajectory of the electron in
both the model and simulation. Good agreement on the predicted trajectory is obtained in
the region of negative momentum, as expected from the agreement of the fields in the
upstream region (Fig. 3), while notable disagreement occurs when the electron
reapproaches the density transition from the upstream side. The disagreement in the
approach to the downstream region arises from the inconsistency of the discontinuous field
model we have developed, because of non-negligible charges passing through (not
stopping at) the transition. Note that even though the exact phase space trajectory in the
transition region is not accurately predicted, the final phase of the electron after trapping is
well predicted by the analytical model.
The self-injection scheme of plasma electrons which we have proposed in this paper
demands further theoretical and numerical investigations for the detailed issues such as
beam quality of the trapped plasma electrons, dephasing, driving-beam quality degradation
in the plasma, etc., and this will be done in the near future. To verify this new injection
scheme, we are planning an experiment with the existing argon plasma source [16] which
was originally developed for the underdense plasma lens experiment at the Neptune
Advanced Accelerator Laboratory of UCLA.
References

[1] C.E. Clayton, et al., Phys. Rev. Lett. 70, 37 (1993).

[2] K. Nakajima, et al., Phys. Rev. Lett. 74, 37 (1995).

[3] C. W. Siders, Phys. Rev. Lett. 76, 3570 (1996).
[4] F. Amiranoff, et al., Phys. Rev. Lett. 81, 995 (1999).
[5] D. Umstadter, J. M. Kim, E. Dodd, Phys. Rev. Lett. 76, 2073 (1996).
[6] E. Esarey, R. F. Hubbard, W. P. Leemans, A. Ting, and P. Sprangle, Phys. Rev.
Lett. 79, 2682 (1997).
[7] S. Bulanov, N. Naumova, F. Pegoraro, and J. Sakai, Phys. Rev. E 58, R5257
(1998).
[8] J.B.Rosenzweig, B.Breizman, T.Katsouleas, and J.J.Su, Phys. Rev. A 44, R6189
(1991).

[9] N. Barov, M. E. Conde, W. Gai, and J. B. Rosenzweig, Phys. Rev. Lett. 80, 81
(1998), N. Barov, J. B. Rosenzweig, M. E. Conde, W. Gai, and J.G. Power, Phys. Rev.
Special Topics—Accelerators and Beams
[10] T. Katsouleas and W. B. Mori, Phys. Rev. Lett. 61, 90 (1988).
[11] J.B. Rosenzweig, Phys.Rev. A 38, 3634 (1988).
[12] A. I. Akhiezer and R. V. Polovin, Sov. Phys. JETP 3, 696 (1956).
[13] R. D. Ruth and A. W. Chao, in Laser Acceleration. AIP Conf. Proc. 91, Ed. P.
Channell (AIP, New York, 1982).
[14] J.B. Rosenzweig, Phys. Rev. Lett. 58, 555 (1987).
[15] C. B. Schroeder, P. B. Lee, J. S. Wurtele, E. Esarey and W. P. Leemans, , Phys.
Rev. E 59, 6037 (1999).
[16] H. Suk, C. E. Clayton, C. Joshi, T. C. Katsouleas, P. Muggli, R. Narang, C.
Pellegrini, and J. B. Rosenzweig, IEEE Trans. Plasma Sci. (to appear Feb. 2000).
Figure Captions

Figure 1. Configuration space (r,z ) distributions of the plasma electrons from the MAGIC
code 2-D PIC simulation. The ambient plasma electron density is n0 = 5 ×1013 cm -3 for
k p II z < 11.2 , n0 = 3.5 ×1013 cm -3 for k p II z ≥ 11.2 . (a) Distribution when the beam center is
II II
at k p z = 18.6 , and (b) distribution when the beam center is at k p z = 35.2 .

Figure 2. Longitudinal momentum of the plasma electrons for Fig. 1(a) and (b).

Figure 3. Plot of electric field in two regions in beam’s wake (n0 = 3.5 ×1013 cm -3 for
z>0.5 cm, n0 = 5 ×1013 cm -3 for z<0.5 cm). Solid line indicates the field in the region
calculated from the analytical model, while the dashed line indicates the model-derived field
in the conjugate region. The thick dotted line is the field calculated from a one-dimensional
PIC simulation..

Figure 4. Phase space of trapped electron originating 0.01 cm downstream of plasma

density transition, for simulation and parameters of Fig. 1. The dashed line indicates the
phase space trajectory of the particle within the downstream region, and the solid line
indicates the trajectory in the upstream region. The thick dotted displays the results of the
1D PIC simulation.
 5
Discontinuity
boundary
eE (MV/cm)

∆E = 4πσ
-5
0 0.2 0.4 0.6 0.8 1
z (cm)

Figure 3
 4

2
βγ

-1
0 0.1 0.2 0.3 0.4 0.5
z-ct (cm)

Figure 4