© 1989 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.

EXPERIMENTAL MEASUREMENT OF NONLINEAR PLASMA

WAKE-FIELDS

J. 13. Rosenzweig,” P. Schocssow, B. Cole, 1%‘. Gai, R. Konecny, J. Norem! and J. Simpson
High Energy Physics Division, Argonne Nat,iottal Laboratory, Argonne, Illinois 60439
150 r-7 i ---r--r- r--
r- .-T---N-'-
Abstract
We report direct high resolution observation of nonlinear steep- ifl 41P1
ened plasma waves excited in the wake of an intense, self-pinched
electron beam. Oscillations in both accelerating and deflecting
/‘i i!-.
/‘-,‘i;
fields are measured, and analyzed in the context of linear and
:’’ 1I1 1.j;
’ I’ 11
nonlinear plasma wave theory. The degree of nonlinearity in the
wakr-fields is shown to be consistent with analytical predictions

I/ tI,I I”/ i
of the beam self-pinching. The impact of these results on plasma
acceleration and focusing schrmrs is discussed.

Introduction 15” L-1.

300
a-.1. i--l.
4ou
.. . I _i. .i ..L>_
SO0
>

1:ELAY ‘l‘i\iE (psec)

The Plasma Wakr-field Accelerator (PWFA), a promising ultra- Figure 1: Figure 1. Longitudinal n-akc-iield stall witness beam
high gradlent acceleration scheme which uses longitudinal electric energy cectroid LIE motion versus time delay, with plasma density
fields in plasma waves driven in the wake of a bunched relativistic of no = 7.3 x 10la cmm3.
electron beam, has been the subject of much recent experimental’
and thcorrtira12~~R investigation. Much of the theoretical effort to the expected reduction in beam size for the experimental condi-
date has concentrated on the linear regime of the PWFA, in which tions, and use this profile to estimated the wake-field amplitude
the amplitude of the sinusoidal electron density perturbations in at the fundamental frequency. Using the harmonic decomposition
the electron plasma waves 7~1 is much smaller than the ambient of the plasma wave as a perturbation theory, we then derive the
unperturbed plasma density ~0. The linear treatment is inade- expected amplitude of the higher harmonics in the wake-fields.
quate. to deal with large amplitude plasma wake fields, and also These calculations are shown to be consistent with the experi-
ignores the self-pinching of the driver bram,8 an effect that has mental data.
been proposed as a powerful final focusing lens (plasma lens3) for
high energy linear colliders. On the other hand, previous theoret-
ical analyses of the nonlinear regime of the PWFA have treated Experimental Method
only the one-dimensional case, due to the mathematical complex-
ity of the three-dimensional problem. The high intensity 21 MrV driving electron beam pulse used in
The only previous experimental test of the I’WFA was per- the present experiments has the following characteristics: num-
formed at the Argonne Advanced Accelerator Test Facility” ber of electrons per pulse N y 2.5 x 10’” (Q =: 4 nC), rms pulse
(AATF) with experimental conditions that satisfied the assump- length cr, = 2.1 mm, initial rms radius (T, 1.4 mm, and emit-
tions of both linear plasma response and negligible beam pinching. tance F = 7 x 10-s m-rad. The target plasma column is a provided
The experiments we wport here. also performed at the AATF, by a hollow cathode arc source of length I, 33 cm and variable
were executed with much higher driving beam current densities density no = (0.4 -- 7) x 10’” cm m3. The driving pulse is followed
present. These more intense beams undergo a self-pinch in the in time through the plasma by a low intensity, 15 MeV witness
plasma, increasing the beam current density further, which in beam which is variable in time delay. Roth beams are analyzed
turn increases the amplitude of the wake plasma waves into the for energy spectra and transverse deflections by a high resolution,
nonlinear regime. For moderately nonlinear waves (the plasma broad range spectrometer. At each point in witness beam delay
electrons hecome only slightly relativistic’) steepened wave pro- the centroids of the witness beam distribution in the energy ana-
files develop which can be decomposed into Fourier harmonics of lyzing and deflecting planes are calculated. The wake-field scans
the the plasma frequency wp :
measurements of longitudinal
Lf----
4Keznolme. We present herr
and transverse plasma wake-fields
in these experiments consist of many such points incremented
delay by fine time steps over a range of many oscillation
in
lengths.
which show rlearly this steepening and harmonic generation. The AATF, plasma source, beam diagnostics and data acquistion
Due to the nonlinearity of the plasma waves and the changing and analysis are desrribrd further in Rrfs. 1 and 10.
transverse profile of the driving beam, we must employ a hybrid of
linear and nonlinear plasma wave theories, as well as plasma forus-
ing theory to explain the experimental results we present, Using Experimental Results and Analysis
the theory of beam dynamics12,13 in the plasma lens, we estimate
For the purpose of illustrating most of the relevant physical
l Present address: Fermi National Acwlerator Lebomtory. phenomena within the scope of this paper, we show two represen-
CH2669-0/89/0000-0731%01.0001989 IEEE
PAC 1989
 1----’ 1I(~pr~.~“lall.e
‘T-’mar
‘~
I!I
0 \I$(~, \ :;I
1’
OF\ i 1 \ “;1
-! a \J\ ;:i, ii \
I ii dfi i?I 1
L-l_Ll_l.ae,~
30” 400 SO”
, , ~ ~, i
“E1AYTIME(pscc)
Figure 2: Figure 2. Transverse wake-held scan witness beam Figure 3: Figure 3. Longitudinal wake-field for the same scan
deflection plane ay centroid versus time delay for same scan as as in Figure 1, different delay range. Below 250 psec in delay,
in Fig. 1. the driving beam charge is 2.9 nC. Between 250 and 280 psec the
charge is raised ot 4 nC, where it remains for the rest of the scan.
tativr wake-field scans. The first scan, shown in Figs. 1 - 2, is
taken with a relatively low plasma density na = 7.3 x 10” cmm3.
The witness and driver beams are slightly misaligned in this scan
to allow observation of the longitudinal dependence of the trans-
verse wake-fields in the nonbend plane of the spectrometer.
/I :’ /;
Several qualitative remarks can be made upon inspection of ‘03, / ;I / \ \ i j /‘ij
Figs. 1 and 2. The first is that both the longitudinal and trans- ? 0,-I i / j 1’ !I 1 i-i
verse wake-fields are stable, oscillatory functions of the distance k2, i 1\,i i I,1, / ; ‘;\ //’ iI / / fl
behind the driving beam [ = ct -- z. Secondly, the longitudi-
- 1OG,ii
nal wake-fields W, have taken on a more saw-tooth appearance,
1 i,i
v ‘/1yi j
as would be naively expected from the one-dimensional
ear theory (cf. Ref. 7). Also, the transverse wake-fields
nonlin-
(W’r for -200 !
/ ILJ, ,.m,!.2m,.,21.1L. I ,--.! L. ~
a cylindrically symmetric driver) show a form consistent with the i;,J ,“/I
4,1 I1 ) I JO
differential form of the Panofsky-Wenzel throrem,r4 &W, = a,W,. ,I,, IAY ‘T,C,F (pwc)

Also, the Fourier spectrum of the longitudinal wake-fields allows

Figure 4: Figure ‘1. Longitudinal wake-field scan, with plasma
us quantify the physical basis for the nonlinearity of these waves.
density of 7~” = 2.R x 10’” cm-‘.
The Fast Fourier Transform (FFT) of the longitudinal wake-field
in Fig. 1 gives the ratio of first harmonic to the fundamental am-
plitude in the wave of about 0.3. gest that significant self-pinching of the driver must have occured.
The 1-D nonrelativistic theory of nonlinear plasma waves gives This hypothesis is supported by several observations: (1) Mea-
the Fourier decomposition of the plasma electron density wave surement of large witness beam deflections at zero delay,’ indi-
in terms of the ratio n,/na. Note WC have used the symbol nr cating the presence of large focusing fields. (2) Improved trans-
here as the fundamental Fourier amplitude in the density per- mission of the aperture-limited driving beam throught the plasma
turbation, and we purposefully draw explicit equivalence to our source anode when plasma is present. (3) The driver beam im-
previous usage, where n, is the calculated linear perturbation at age in the spectrometer nonbend plane is greatly expanded with
this frequency. The amplitude of the m-th wave harmonic is given plasma present, implying strong overfocusing. Also, if we examine
by7 nml no - (nl/no)mmm/2mm’m!, and the longitudinal electric a different range of this scan, shown in Fig. 3, one in which the
field on axis associated with this charge density wave is! ignoring beam charge is diminished a factor of 0.72 from 4.0 to 2.9 nC, we
phase factors, observe a disproportionate reduction in the wake-field amplitude,
to a factor of approximately 0.35. This indicates an enhancement
, of the wake-field amplitude due to the the pinch effect.
To study self-pinching effects in detail, we move on to the anal-
where k, = +/c and ql,,,,(mk,o,) is a factor less than unity which ysis of an even more nonlinear scan, shown in Fig. 4, in which the
measures the degree to which the wake-fields are longitudinal.‘~4 plasma is nearly four times as dense as in the first scan. The
The ratio of first harmonic Ez to fundamental Er amplitude is plasma wavelength X, = 2n,‘k, in this case is twice as short
simply t/2(nI/n0)(~~2/~r1), which implies that the densit,y per- as before (0.65 mm), and is now shorter than the beam length
turbation at the fundamental frequency in the scan shown in 4~7, : 0.84 mm. This is the regime of the plasma lens, where
Figs. 1 and 2 is n,/n, x 0.33. This wave amplitude is not con- the self-forces on the beam can be well approximated by assum-
sistent with the predictions of linear theory if one ignores pos- ing total space charge neutralization of the beam and evaluat-
sible pinching of the driver beam. The linear theory the wake ing the resultant magnetic self-forces.‘2 Theoretical treatment of
wave amplitude for a bi-Gaussian driver is given by’?4 nt/no -- the beam dynamics in a plasma lens r3 allows calculation of the
21Vr, exp [-(kp~z)2/2]/kpuf. For the present case, with 6, taken distance inside the plasma required to achieve a pinch J and
as its initial value, this yields nIlno = 0.08, which is much smaller an equilibrium pinched beam spot size uf,. In terms of the
than the estimate from harmonic content. Thus we are led to sug- initial value of the p-function /?s = 0.3 m and the invariant
732
PAC 1989
( = Nr-,iJ47rr,6, 35 m *, where rn is the normalized emit- 2. P. Chen, J. M. Dawson, R. W. Huff and T. Katsouleas, Phys.
tancc. wr have (r& - c/C =- (0.44 mm)’ and s ‘v x/2m = 14 Rev. Lett. 54, 693 (1985).
cm. Thus we expect the beam to pinch to approximately one-third
3. P. Chen, Particle Accelerators, 20, 171 (198.5).
its original radius well within the plasma column. This pinrhing
is not uniform along the length of the beam, but is proportional to 4. R. D. Ruth, A. Chao, P. I,. Morton, and P. B. Wilson, Particle
thr enclosed current at a given point. The profile which develops Accelerators 1’7 (1985) 171.
is effectively shortened on axis by a factor of &.
The FFT of thp scan in Fig. 4 gives a relative first harmonic 5. T. Katsouleas, Phys. Rev. A, 33, 2036 (1986).
amplitude value of I&/E, 7 0.48. The predicted value, using
6. R. Keinigs, M. Jones and J.J. Su IEEE Trans. Plasma Sci.,
0, = creQ = 0.44 mm and gZ .: 2.1 /fi mm in the linear response
PS-15, 199 (1987).
formula, is &/E, = 0.38, in fairly good agreement. In the first
scan, the equilibrium pinched beam radius necessary for the the- 7. J. B. Rosenzweig, Phys. Rev. Letf., 58, 555 (1987).
oretical estimate to agree with the measured harmonic content
d ea = 0.77 mm is not as small as our calculated value for the sec- 8. .J. B. Rosenzweig, ,Phys. Rev. A,‘38, 3634 (1988).
ond scan. This is dur to the fact that the beam length is shorter
9. R. Keinigs and M. Jones, Phys. Fluids, 30, 252 (1987).
than the plasma wavelength, and the beam does not encounter
maximum st rrngt h t ransversr wake-fields, as the plasma does not 10. H. Figueroa, W. Gai. R. Konrcny, .J. Norem, P. Schoessow,
react quickly enough to complctcly nrutralizr the beam charge. and J. Simpson, Phys. Rev. Lett. 60, 21 (1988).
The driving beam thus takes longer to pinch and does not acheive
such small spot, sizes. 11. D. IImstadter, R. Williams, C. Clayton, and C. .Joshi,
‘I’hr measured longitudinal wakc-firld amplitudes including the Ph”vs. Rev. Cett., 59 292 (1987).
effects of the gromrt,riral resolution of the witness beam’ arc, for
12. J. R. Rosenaweig, U. Cole, I). J. I,arson, and D. B. Cline,
the scan in Fig. 1, I%;, l.,lR McV/m, and for the scan in Fig. 4,
Particle Accelerators, 24, 11 (1988).
It;” -- 5.30 Me\‘,‘m. which is the largest accelerating gradient ob-
tained thus far in a plasma wave. It should be noted that these ac- 13. J. B. Rosenzweig and P. Chen? Phys. Rev. D, 39, 2039 (1989).
celerating gradients are low compared to particle-in-cell computer
simulations’” of the experimental situation, which indicate that 14. W. K. H. Panofsky and W. A. Wenzel, Rev. Sci. Instr., 27,
acceleration gradients of greater than 10 MeV/m were present, 967 (1956).
while confirming t.hr the nonlinear form of the wake-fields. The
15. J.J. Su, private communication.
discrrpancy is due to a variety of experimental effects, the most
important being the loss of maximally accelerated and decelerated
electrons into the background noise surrounding the core of the
witness beam image at the focal plane.

Conclusion
Althought we have dealt successfully with the data here in a
phenomenological manner, a more rigorous understanding of the
physics issues invloved would be desirable. The main questions
raised by this experiment have to do with the nonlinrar plasma
response. In these experiments, the beam is much narrower than a
plasma wavelength, and thus the plasma electron response and as-
sociated wake-fields cannot be well explained by a one-dimensional
theory. Conversely, one-dimensional theory predicts improved
transformer ratios in very nonlinear waves driven by long hrams
of density n,,/2; it is not possible to confirm this prediction by
extrapolating our results. On the ot,her hand, the status of the
theory and simulation of beam focusing in plasmas is more well
understood than the three-dimensional nonlinear plasma motion.
These experimental results are consistent with the theoretical pic-
ture of the self-pinch dynamics, and can be viewed as a succesful
test of the concept of the thick plasma lens, as the peak beam
density is increased by a factor of ten inside the plasma.13
This work supported by the U.S. Dept. of Energy, Division of
High Energy Physics, Contract W-331-109.ENG-38.

References
1. J. B. Rosenzwcig, D. B. Cline, B. Cole, H. Figueroa, W. Gai,
R. Konecny, J. Norem, P. Schoessow, and J. Simpson, Phys.
Rev. Lett. 61, 98 (1988).
PAC 1989