© 1991 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.

Ion Clearing by Cyclotron Resonance Shaking

P. Zhou and J.B. Rosenzweig*
Fermi National Accelerator Laboratory!
P.O. Box 500, Batavia, IL 60510

Abstract clearing electrodes, ions inside dipole magnets cannot be

easily cleared, since there are are often no clearing elec-
We discuss a new concept in ion clearing for storage trodes inside magnets due to tight space constraints, and
rings, that of resonant removal of ions in dipoles by shak- thus the only clearing mechanisms are Coulomb heating
ing the beam horizontally near the ion cyclotron frequency. and E x B longitudinal drift. The former is insignificant
This method of beam shaking is similar to the variations while the latter can be greatly reduced by the neutral-
on ion bounce shaking developed by Orlov, Alves-Pires ization itself. As a result the neutralization level inside
and coworkers at CERN, but has advantages in requiring the magnets is much higher than the rest of the machine
a narrower bandwidth of shaking frequencies and in much and deleterious ion effects are most likely caused mainly
higher achievable ion kinetic energies. The results of ana- by the ions in dipoles. The strong magnetic field inside
lytical theory, and computer simulations are discussed. magnets leads to cyclotron motion of ions in the horizon-
tal plane, which has, in general, a much higher oscilla-
tion frequency than that of the bounce motion, and there-
Introduction fore bounce shaking is not effective in the horizontal plane.
This explanation leads us consider the possibility of hori-
Accumulation of ions has been a limiting factor in the zontal shaking close to ion cyclotron frequency, which we
performance of antiproton accumulators. Besides the di- call cyclotron shaking.
rect method using clearing electrodes, beam shaking has
been proven very effective in further clearing ions[l] and
reducing the neutralization level in storage rings. In the Theory of Cyclotron Shaking
so called “resonant shaking”[2] a driving voltage with fre-
quency close to that of the ion oscillatory motion in the The theoretical analysis for cy$otron shaking is parallel
beam’s electric field (bounce frequency) is applied to the to that of bounce shaking[4], which uses the averaging tech-
beam, which responds by shaking, generating an oscillat- nique originally developed by Krylov and Bogoliubov[5][6].
ing transverse electric field which in turn drives transverse For completeness we reproduce analysis here.
ion motion. Ions traversing different beam sizes, and thus Let x be the transverse coordinate of ion in the horizon-
bounce frequencies, through slow longitudinal motion will tal plane and r the longitudinal coordinate, then, ignoring
be locked-on to larger amplitudes; in this way the neu- the space charge of the ions themselves, the equations of
lralization effects can be reduced[3]. The same or bet- motion for the ion are,
ter results should be achievable through modulation of the
driving frequency[4], w h ic h is called frequency modulation d=x
z= -w$ + -$!&(x - bcoswl)
shaking. This way the process is controllable and does
not rely on the ions’ longitudinal motion. To distinguish d=t dx
from what we are going to describe here we call this kind z= -WC-&-

of shaking bounce shaking because it is the ion bounce

motion that is being driven. where q and m are the charge and mass of ion respectively,
An interesting and important experimental fact is that w, is the cyclotron frequency of the ion, and b is the ampli-
the bounce shaking described above doesn’t work well in tude of the beam center oscillation driven by the applied
horizontal plane. This can be explained by noting that voltage. For a round Gaussian beam with rms beam size
while ions in straight sections are easily removed by the u the form of radial electric field is

*Present
Angeles,
address:
CA 90024
UCLA Dept. of Physics, 405 Hilgard Ave., Los E,(r)= V(l - e-5)
t Operated by the Universities Research Association under con-
tract with the U. S. Department of Energy where X is the beam line charge density.
0-7803-0135-8/91$01.00 @IEEE
1776

PAC 1991
 The displacement of the ion from the beam center nor- values if the driving frequency is modulated from above
malized by the beam rms size u, i = $[z - bcos(wt)], is that frequency downward, which is the concept frequency
then described by modulation shaking based upon.
In general function G(a, zs) cannot be expressed in
2 = A(w2 -WC”) cos(w2) - w$ - x0) - w:f(Z) closed form. Results by numerical methods are shown in
Fig. 2. As x0 increases G not only drops in magnitud le but
where wb = dv qX mu is ’ t h e maximum ion bounce fre-
quency, A = b/c and xeu is the horizontal position of
the ion’s guiding center. Unlike in the case of bounce
shaking where the corresponding quantity is always zero,
ze here is in general not. For simplification we take
f(Z) = $ [l - exp (- i$,] which implies that we are only
considering ions close to the vertical center of the beam.
We look for the equilibrium solution of the form

Liz = a(t) cos(wt + e(t)) + x0

di?
-wa(t) sin(wt + e(t))
z= .
0
in which a(t) and 0(t) are slow varying functions relative
to the oscillation period. Averaging over a period yields: Figure 2: Function G(a, x0).
da (4 - W2jAsin0
also reverses its trend as a function of a. This means not
z= 2w
only the resonant frequency decreases, the direction of fre-
dO
(wz2;w2) [l - ; cos e] + 2G(a, x0) quency modulation that increases the ion oscillation ampli-
dl=
tude reverses also. Therefore the effect of cyclotron shak-
where G(a, x0) = & s,‘” f(zo +a cos 4) cos q5dq5. The max- ing with frequency modulation eventually vanishes. With
imum amplitude satisfies da/dt = dfl/dt = 0, therefore this discussion in mind, it is clear that cyclotron shaking
we have an implicit relation between the driving frequency will only be effective to ions close to beam center. For-
and equilibrium ion amplitude: tunately, these ions are precisely the ones which pose the
greatest threat to beam stability, because they slow the
w2 - w,2 _ G(a, x0) ions’ E x B drift inside dipoles by neutralizing the elec-
--- (1)
w: l&f tric field. The drift speed is N 103m/s for ions that are
la from the center of an unneutralized 100 mA beam in
Eq. 1 represents hysteresis which is well known for the Fermilab accumulator. This corresponds to a less than
non-linear oscillators. For x0 = 0, G(a,O) = $[l - 0.2% equilibrium neutralization level inside dipoles, which
e(-c)1e($)] and Eq. 1 is plotted in Fig. 1. The motion is much better than the current operation conditions. If
corresponding to the upper half of the left curve is unsta- the cyclotron shaking can excite the ions created close to
ble. The beam oscillation amplitude undergoes a jump at a beam center to large amplitudes, the drifting motion of the
particular driving frequency and it can reach much larger other ions will provide sufficient clearing.

1
Discussion

- Using the Fermilab antiproton accumulator parameters

.6 A=a.a04
----- -. A-0.02 with 200mA beam current, we compared the theoretical
l-3 .4 predictions on the equilibrium amplitudes with the sim-
.2
ulation results of both bounce and cyclotron shaking for
protons. As shown in Fig. 1 the data points, which corre-
0 sponds to u = 0.22cm,Aa = O.Olmm and no initial hori-
.9 .95 1 1.05 1.1 1.15 1.2 zontal displacement agree remarkably well with the analyt-
ical theory. The modulation of frequency, in both shaking
schemes, is also shown by simulation to be very effective,
as predicted by theory.
As seen above cyclotron shaking and bounce shaking are
Figure 1: Equilibrium Ion Oscillation Amplitude vs. Shak- very similar in many aspects. The equilibrium amplitudes
ing Frequency can be described by the same plot and hence both have

1777

PAC 1991
 l I 4
the same hysteresis “lock on” effect which is crucial in re- --1 LJ_l-1-1-1-1-1-L'-'-~-~~-~-~-~-~-~-~-~
ducing neutralization efFects[3]. However because of the .e - 3
very different frequencies of the cyclotron and bounce mo- f i
tion, the two ways of shaking do have different properties. .6 - 1 -2
The cyclotron frequency w,, at least in our case, is much ;
higher than the bounce frequency wb. The correspond- -1
ing frequency spread in cyclotron shaking is a factor of
W,/wb = 15 smaller than that in bounce shaking. This dif-
ference affects the frequency modulation process because -250
“““““““““‘l”L’
-200 -150 -100 -50 0
the beam response to the driving voltage depends strongly Distance (cm)
on the frequency and that has a big impact on the ion re-
sponse (see Fig. 1). Ob viously the beam response to the
driving voltage varies rapidly around betatron sidebands.
If in the process of frequency modulation the beam re- Figure 3: Dipole Field and Ion Motion
sponse changes too fast, that could cause ions to loose the
lock-on and limit the effectiveness of modulated frequency
at a non-negligible rate in the Fermilab antiproton accu-
bounce shaking. The small frequency range of cyclotron
mulator .
shaking helps in stabilizing the beam response in the whole
In short we conclude that the cyclotron shaking is less
process of frequency change. The factor of 15 in the case
susceptible to the change of beam response to external
of Fermilab accumulator reduces the range to only a small
driving voltage and therefore may be more effective in re-
fraction of revolution frequency. Note that if the cyclotron
ducing ion neutralization effects. In addition, the large
frequency falls very near a betatron side band, a beam-ion
energy associated with exciting cyclotron orbits of radii on
instability may be excited. This subject is analyzed in a
the order of the beam size may aid significantly in final
separate paper[7].
removal of the ions from the beam. Experimental study
The theory presented earlier only deals with equilib- is needed, and is presently being pursued on the Fermilab
rium responses while frequency modulation inevitably in- antiproton accumulator.
troduces time varying effects. The simulation shows that The authors would like to thank R. Alves Pires and A.
cyclotron shaking with frequency modulation has a longer Poncet for their valuable help.
lasting transient ion motion which requires a longer mod-
ulation period. The long transient in the ion motion
in the case of driving cyclotron resonant motion is eas- References
ily understood by noting that there is a lot more kinetic
energy in a cyclotron orbit than in a bounce orbit of PI J. Marriner and A. Poncet, “Neutralization Experi-
the same horizontal amplitude. For the cases where cy- ments with Proton and Antiproton Stacks - Ion Shak-
clotron shaking are effective, the cyclotron kinetic energy ing”, Pbar-note 481, Fermilab Internal Note, March
is Ek = (x,w,)2m/2 N 500 eV, and the bounce kinetic 1989.
energy is smaller by a factor of (wb/w,)‘, or 2.2 eV. When Alves Pires, “Beam Shaking for the Fermilab An-
a bounce-shaken ion has exited the end of the magnets
PI R.
tiproton Accumulator”, Proceeding of the Fermilab III
through its E x B drift motion, its removal is still predi- Instability Workshop, 1990
cated on adequate clearing voltages, which may not always
be provided. If the ion is cyclotron-shaken, however, it can PI Yuri Orlov, “The Suppression of Transverse Instabili-
by virtue of its large energy easily escape the beam poten- ties in the CERN AA by Shaking the j Beam”, CERN
tial well and clear completely. This is indeed verified by PS/89-Ol(AR), 1989.
simulation. Shown in Fig. 3 are the normalized magnetic
field and a sample of ion motion driven by frequency mod- PI R. Alves Pires and R. DilZio, “On the Theory of Shak-
ing”, to be published
ulated cyclotron shaking. Notice that the tracks shown are
actually the alias of ion’s oscillation motion. The longitu- [51 N. Krylov and N. Bogoliubov, Introduction to Nonlin-
dinal positions that locked-on ions escape the beam and ear Mechanics (Princeton University Press, Princeton,
hit the vacuum wall are concentrated in a small range. N.J., 1943)
This fact provides a potential diagnostic to directly mea-
sure the effect of the cyclotron shaking, by collecting these PI Ronald E. Mickens, Introduction to Nonlinear Oscil-
energetic ions, and measuring their energy spectra. In ad- lations (Cambridge University Press, 1981) (1989) p.
800
dition, inside the magnet, if the cyclotron-shaken ion has
an elastic collision with an ion, a neutral, or a beam parti- PI J.B. Rosenzweig, “Beam-Ion Cyclotron Resonance
cle which redirects even 5% of its energy into the vertical Instability”, Proceedings of Fermilab III Instability
plane, it will escape the beam entirely. Preliminary calcu- Workshop, July 1990
lations indicate that this beneficial phenomenon may occur
1778

PAC 1991