You are on page 1of 7

Nuclear Instruments and Methods in Physics Research A297 (1990) 24-30

24 North-Holland

Envelope instability in love-energy proton synchrotrons

P. Zhou and J.B. Rosenzweig

Fermi National Accelerator Laboratory *, P.O. Box 500, Batavia, IL 60510, USA

Received 1 August 1990

due to the instability of the beam

In this paper we investigate the limits on beam intensity in low-energy proton synchrotrons
examined in the context of space charge dominated ion beams, in particular for inertial
envelope . This instability has been previously
. We generalize the formalism of beam envelope evolution to include effects found in circular
confinement fusion applications
analyzed to determine intensity limits
accelerators, i .e . curvature focusing and momentum dispersion . Several example lattices are
can occur without depressing the phase advance below 90'
imposed by onset of envelope instability ; it is found that the instability
per cell .

l . Introduction Laslett, Smith and Haber [4] . As many synchrotrons are

designed with bare phase advance per cell near 90 ° to
The intensity limit for operation of low-energy pro- minimize the necessary aperture in the machine, it is
ton synchrotrons is often determined by the space charge reasonable to ask whether this effect is important for
tune shift [1], which can drive individual particle tunes weaker intensity beams found in circular machines . In
from the operating point onto a significant machine particular, this question was raised with regards to the
resonance . In this way a relatively small tune shift Fermilab booster, both in its present state and after the
(usually a few percent) can cause the emittance of a upgrade in injected energy from the linac [5] .
synchrotron beam to grow significantly . Experimentally, We review below the standard derivation of rms
this effect manifests itself by a linear asymptotic growth transverse beam envelope equation, and extend it to
of the emittance with peak beam current . This type of include the effects of design orbit curvature, momentum
behavior has been observed in experimental studies of spread and dispersion . The envelope equation obtained
the Fermilab booster [2]. is then analyzed for its stability properties under per-
In contrast to the scenario described above, linear turbation . This technique is used to examine existing
transport allows much higher intensity beams to propa- and proposed machines - the SSC low-energy booster,
gate without degradation of the beam emittance . the Fermilab booster, and the TRIUMF Kaon Factory
Space-charge tune shifts which are nearly as large as the booster - using their nominal lattices, intensities, emit-
bare (zero-intensity) tune can be tolerated given proper tances and momentum spreads, to determine possible
conditions in linear transport channels, as the beam problems with envelope instability in these devices .
travels much shorter distances in general, and does not
encounter the periodic resonance-driving defects found
in a circular machine . The limits of stable beam propa-
gation in a linear focusing channel have been demon- 2 . The envelope equation
strated experimentally by Tiefenback et al. [3] at LBL.
It was found that as long as the bare phase advance per
The first envelope equations for describing beam
cell a6 did not exceed approximately 90 *, the beam
propagation including the effects of the repulsive space
emittance was stable up to very high intensities . At
charge self-forces were derived by Kapchinskij and
phase advances above 90 ° , however, it was verified that
Vladimirskij (K -V) [6] using a special distribution func-
instability of the beam envelope caused emittance
tion (uniformly distributed population on a three-di-
growth in the transmitted beam. This result is in agree-
mensional hyper-ellipsoidal surface in four-dimensional
ment with the theoretical predictions of Hoffman,
transverse phase space) which yields a uniform spatial
density distribution within the elliptical beam boundary .
* Operated by the Universities Research Association under This uniform density distribution yields linear space
contract with the U .S . Department of Energy . charge forces, thus allowing a straight-forward extrac-

0168-9002/90/$03 .50 J 1990 - Elsevier Science Publishers B .V . (North-Holland)

P . Zhou, J .B. Rosenzweig / Envelope instability in low-energy , proton
synchrotrons '_5
tion of equations for the evolution of the beam envelope of eqs. (7) and (9) (along with the equivalent expres-
in a focusing channel, sions for 9), can be written in form identical to the
K-V equation,
E2 +- Q
a"+K(z)a__ _
a3 a+b
for the horizontal beam size a, and
E and
b"- K(z)6=b3 +
Qb (2)
for the vertical beam size b, where the focusing strength (12)
K = (aB/âx)(Bp) -1 is positive for a horizontally focus-
ing magnetic quadrupole, Ex, y is the total horizontal/ as previously derived by Sacherer [7] . All envelopes and
vertical emittance, taken by convention to be four times emittances in these equations are now rms quantities:
the rms emittance, and Q = 41/(10ß 3.y 3) with to = 31 the K-V equations in rms form are contained in eqs.
MA for protons . (11) and (12) as a special case.
In order to avoid use of unphysical K-V distribution
function, Sacherer generalized this treatment by con-
centrating on the evolution of the rms envelopes [7] . 3. Envelope instability
This approach is equivalent to following the second
moments of the Vlasov equation for the four-dimen- Solutions to eqs . (11) and (12) must generally be
sional transverse phase space of the beam. Assuming no found by numerical methods . In the case of periodic
x-y coupling, the second moments in the (x, pY) plane, focusing structures, application of periodic boundary
where px = x', satisfy the following equations : conditions to these equations gives the matched solution
for the rms beam envelopes . Beams will always have
x2' = 2xpx, (3) slight mismatches to the lattice, however, and the ques-

xpY = P +P x, (4)
tion remains as to whether the deviations from the
matched solutions are stable . To answer this concern, a
perturbation analysis is performed . which entails writ-
P2' = 2Px PP . (5)
ing the perturbed envelope equations obtained by first
Analogous equations apply in the (.y, p,) plane. Fol- order Taylor expansion for the rms beam sizes about
lowing Sacherer, we formally divide the force term the matched solutions . The eigenvalues of this system of
P --- F,/ß 2ymc 2 into a linear component - K(z )x due linear equations are then examined to see if they indi-
to the focusing elements and a part due to the space cate instability by moving off of the unit circle on the
charge self-force .F. eqs . (3)-(5) thus become complex plane. These instabilities occur when two ei-
genvalues collide cn the unit circle . If they collide at
x = 2xpx.
180' (on the real axis) then this is a half-integer reso-
xpx= -K(z)x +x~, +p , (7) nance between the perturbed envelope mode and the
focusing structure . This case is a termed "parametric"
P2 '= -2K(z)xpx + 2p., 3Y . (8) resonance . If both pairs of eigenvalues collide off the
real axis, then this is a "confluent" resonance between
The term p, 3~, is related to emittance growth, and can
the two envelope frequencies .
be ignored if one assumes that the rms emittance
In the previously considered example of a symmetric
2 FODO lattice, the resonanes are mainly confluent [8] .
Erms = x 2pY - (XPY) Necessary conditions for envelope instability in this
is either a constant or that its z dependence is a known case are
If, in addition, one assumed elliptical symme- (i) phase advance per cell ao > 90 °, and
try for the beam spatial density. the term x. depends (ii) depressed phase advance per cell a < 90' .
only on the rms beam dimensions .X= (x 2)1 2 and 03(=- In terms of design, therefore, it is advisable to choose a
)1/2 FODO lattice with rtJ <._ 90°, or failing that to make the
phase advance much larger than 90' . In machines with
(10) small space charge tune depression, the only way to
invite trouble with the envelope instability is to design
where we redefine Q -i~ Q/4 = IMo ß3 y3 ). In this way for ao just above 90' . This conclusion, which was
the hierarchy of moment equations is closed . With the reached for symmetric FODO lattices in linear trans-
additional resealing of E - E/4 = Erms, the rms envelope purt lines, is found to be valid for most cases in circular
equations, derived by differentiation of eq. (6) and use accelerators we examined. In addition, however, we find
26 P. Zhou, J.B. Rosenzweig / Envelope instability in low-energy proton synchrotrons

that in some asymmetric focusing systems that the and as a direct result of eq. (15), the rms beam size is
envelope instability can occur with a > 90' . the root sum of squares of the betatron and dispersion
4. The envelope equation : circular machines
where ap = p/p
The major changes to beam behavior in circular This approach is approximate in the presence of
machines which have relevance to deriving envelope space charge forces, which introduce coupling that pre-
equations are due to curvature focusing, which generally vents the simple decomposition of the description of the
serves to break the horizontal-vertical symmetry of motion shown in eqs . (15), (16). In this approximation
lattice, and to the presence of momentum dispersion, the horizontal emittance takes the functional form
which causes the horizontal beam profile to expand . 2]
The curvature effects on the horizontal motion merely 2=
1 + + ( .T,n~ _ d9Î )2 (19)
~ ~~
Ex EX aP
add to the horizontal focusing strength due to the
quadrupole fields while leaving the vertical focusing ,n2a 2
_ 2
strength unchanged, _ Ex X2_


Kx =K+ (13) (20)

and the other term on the right hand side of eq. (14) is
In addition, in circular machines the x and y phase
advances are are chosen to be different to avoid cou- p _ n a2
x (21)
pling resonances. Thus the focusing strengths of the F -PT P P X .
and D lenses are usually different . These changes can be We now arrive at a new set of envelope equations for
trivially incorporated into the envelope equations . The circular accelerators
inclusion of the effects of momentum dispersion are not
2 E2
nearly as straightforward, and require careful treatment . _ 1( _a Q
X~~+ ~ K(`)+ (22)
Following the previous analysis on the evolution of 192 P` X) X X3 + X+0.1
second moments and ignoring the chromatic nature of
the ocusing we arrive at the equation for the rms
horizontal beam size X in a circular machine 2 Q
K(z)41 (23)
E2 P - E""~ + -ff +
X + [K(z) + P2
1 1 X- X Q
+ q= -
X 3 + PX' P where the vertical emittance is of course unchanged . In
(14) the absence of space charge (Q = 0) eq. (20) is exact.
This treatment obviously has a limited range of
The self-force term takes the same form as before, validity ; the separability of motion into betatron and
implying the assumption that the overall beam density dispersion components is an approximation due to the
still has elliptical symmetry. While the left hand side is coupling introduced by space charge. We thus expect
straightforward the right hand side terms still need that our envelope equations are valid only in the limit
some approximation in order to be expressed in known of small tune shifts, which fortunately serves our pre-
or solvable quantities . We follow the standard way of sent purposes well.
describing off-momentum orbits in accelerators, split- In the interest of completeness we note that we have
ting the motion into a betatron component x ß (the also tried to include an additional correction term in eq.
quantity we have simply termed x to this point) and a (17) to partially take into account the effect of space
dispersion component 71(0p/p)
. Thus we now write, charge force on the dispersion function
for the purpose of examining rms quantities
"+ [Kx(z) - 91= .
X=xP +9Î p X( X+ Y) P
P , (15)
The effect of this change was negligibly small, at least in
0 the range of beam currents close to envelope instability
P,=xR+~, (16)
The equation of motion for the dispersion function is Eqs. (22) and (23) are now used to provide the basis
taken to be of a perturbation analysis for several example lattices of
low-energy booster proton synchrotrons at injection,
 +K _ _1 including the effect of the beam bunching (bunching
71 x( `) ~1 - P (17)
factor) on the peak current . This is often where the
P. Zhou, J.B. Rosenzweig / Envelope instability in loK, -energy proton synchrotrons 27

space charge tune shifts are most important in a circular The method of analysis proceeds by converting eqs .
accelerator chain . (22) and (23) to the equivalent system of first order
differential equations and solving them numerically sub
ject to the appropriate periodic boundary conditions. In
5. Envelope instability in circular machines our case we have employed a shooting method with a
Runge-Kutta integration scheme. With Lhe matched
The perturbed envelope equations that are used to solutions for the given lattice, emittances, beam energy
test for envelope instability are, expanding eqs . (22) and and peak current, we then solve for the eigenvalues of
(23) about the matched solutions for (X, OJ ), the first order system equivalent to eqs . (25) and (26).
Q2 2
The rms phase advance per cell is calculated for each
SX~~+ K(z) + +~ +3E4 case by evaluating the phase integral over the length of
P X2
X the cell, L,
-QS0,l /'Ldz dz
(25) /'L
Jo P = E-KJo (27)
= X2 - (,,(F)2
with an analogous expression holding for a, In the
absence of space charge, this integral gives . the bare
betatron phase advance of the cell. On the other hard,
in the absence of momentum spread, eq. (27) gives the
rms betatron phase advance per cell. If both momentum
Note that. in the spirit of Sarherer's treatment, we do spread and space charge are present, there is no obvious
not perturb the terms inside the emittance expression . interpretation of eq. (27), as it contains some informa-

94 ._


90 . _

88 ~
É \ - erl~c' l


84 ; _
0 500 000 1500 2000
0 500 :!00 1500 2000
Peak Beam Current (mA)
Peak Beam Current (mA)

0 500 1000 1 :00 ;'000

0 500 1000 1500 2000 Peak Beam Current (mA)
Peak Beam Current (rnA)

Fig. 1. (a) Phase advance and growth rate per cell plotted against peak current at injection energy SSC LEB . no momentum spread.
(b) Phase advance and growth rate per cell plotted against peak current at injection energy, SSC LEB, Qr = 9 .3 x 10-4.
28 P. Zhou, J. B. Rosenzweig / Envelope instability in low-energy proton synchrotrons

tion about the off-momentum orbits coupled to the for the worse. This case is shown in fig . l b. The onset of
betatron orbits through space charge forces . the instability occurs at about 400 mA, which is quite
We now move on to the analysis of several low-en- close to the design value. This result can be explained
ergy synchrotron lattices, which will illustrate some by examining the space charge force term in the en-
interesting aspects of the envelope instability in these velope equations . If ao < 90 ° and there is no momen-
machines. The first example we take is that of the SSC tum spread, the quantity .X+ 9 is nearly constant, and
low-energy booster (LEB) FODO lattice [9], which has the space charge force has little modulation . The in-
an injection energy of 600 MeV, 360 mA peak current, crease in severity of the instability with momentum
and normalized rms emittance 0.75 mm mrad. The max- spread is due to the larger modulation of the space
imum ß function is 11.9 m and the maximum disper- charge force brought about by the periodicity of the
sion is ~1= 84 cm. The bare phase advances per cell are dispersion. The dispersion contribution to the beam
aXo = 93.6 ° and ayo = 90.0 °, and we can anticipate that width adds approximately in phase with the horizontal
this lattice may be susceptible to beam envelope insta- betatron contribution, introducing a strong periodic
bility. The instability growth rate is plotted along with driving term to the envelope equations .
the phase advances as a function of peak beam current In the Fermilab booster [10], one can expect that the
in fig. l a for the case of no momentum spread. We see large bare tunes (ayo = 102'* and axo = 100.5 °) would
that the instability can occur, somewhat surprisingly, keep the machine far from conditions which would
without both ax and ay being depressed to less than result in envelope instability . This is indeed true for the
90' . The onset of (confluent) envelope instability (ax = case with no momentum spread, shown in fig . 2a. The
90.6") occurs at 1120 mA, or approximately three times normalized rms emittance is taken to be 1.661T mm mrad
the design current, which is a comfortable margin . (10îr "Fermilab" emittance) in both dimensions, and
Inclusion of the design energy spread of up = 9.3 x the peak operating current is 560 mA, assuming N = 2.5
10-4 , however, causes the picture to change somewhat x 10 12 protons/batch and the bunching factor B = 0.25

a b

0 500 1000 1500 2000

Peak Beam Current (mA)

0 500 1000 1500 2000

P-ak Beam Current (mA)
Fig. 2. (a) Phase advance and growth rate per cell plotted against peak current at injection energy, Fermilab booster, no momentum
spread . (b) Phase advance and growth rate per cell plotted against peak current at injection energy, Fermilab booster, up =1 .5 x 10-3.
P. Zhou, J.B. Rosenzweig / Envelope instability in loxv-energV, proton sYnchrotrons 29

at 200 MeV injection energy. Instability onset occurs at advance per cell is 93° . On the other hand, the horizon-
approximately twice this current, assuming constant tal phase advance is far smaller, at 9to = 78.4 °. This
emittance . The calculation for momentum spread up = large tune splitting makes the machine relatively insen-
1.5 X 10 -3, displayed in fig . 2b, is again more pessimis- sitive to the envelope instability . Also, the horizontal
tic . The space charge force modulation lowers the insta- betatron width is larger than the vertical width due to
bility threshold current to less than 600 mA, which is tune and emittance differences (normalized rms c,0 = 23
right on the assumed maximum operating current. This mm mrad and c,ro =10.3 mm mrad), and is much larger
parametric resonance has a large growth rate ( - than the width due to dispersion. The cases without and
15%/cell), which peaks at 1.2 A and then is joined by with momentum spread in this machine are shown in
the other unstable mode above 1.45 A, generating a new figs . 3a and 3b, respectively. The momentum spread
large confluent resonant band at higher currents. actually has a small stabilizing effect in the TRIUMF
It should be noted that the envelope equations can machine. The envelope instability in both cases has a
all be recast in a dimensionless form which explicitly threshold current at 14 A, which is well above the
shows that the calculations presented here are depen- design peak current of about 2 A.
dent only on the ratio Q/E . Thus if the injection energy
is raised while keeping the number of protons per batch
and normalized emittance constant the threshold cur- 6. Conclusions
rent rises as ßy2. For the linac upgrade this factor is
1 .75, which raises the threshold to over 1 A. This treatment of the envelope instability in circular
The case of the TRIUMF Kaon Factory low-energy machines has pointed to the possibility of problems
booster [111 looks interesting on the surface, as peak near the nominal operating point of the Fermilab and
currents are very high and the initial vertical phase SSC low-energy synchrotrons at injection. There is good

go .-_

85 I ._ N5 . _


Hont-',al Hora :~ :~n',jl

75 '~ 75

0 5 10 15 20
Peak Beam Current (A) Peak Beam Current (A)

3 3
U y



0 5 10 15 20 0 10 15
Peak Beam Current (A) Peak Beam Current (A)
Factory booster .
Fig. 3. (a) Phase advance and growth rate per cell plotted against peak current at injection energy, TRIUMF Kaon
plotted against peak current at ir. zction energy, TRIUMF Kaon
no momentum spread . (b) Phase advance and growth rate per cell 10-3 .
Factory booster, up =1 .7 x
30 P. Zhot4 J.B . Rosenzweig / Envelope instability in low-energy proton synchrotrons

news for each case : at Fermilab, the linac upgrade takes References
the threshold current up to an acceptably high level,
and at the SSC the LEB is currently under redesign, [11 L.J. Laslett, Proc . 1963 Summer Study on Storage Rings,
with phase advance per cell near 113° [121 . This design Accelerators, and Experimentation at Super-High Ener-
should provide for a larger threshold current, although gies (Brookhaven National Laboratory, 1963) BBL 7534,
calculations should be performed to check this conclu- p. 324.
sion. [21 Y. Chao, J. Crisp, S. Holmes, J. Lackey and W. Merz,
In order to understand the potential harm that en- Proc. 1988 Europ. Particle Accelerator Conf. (World Sci-
velope instability can pose to circular machines better, entific, London, 1989) p. 663.
it would be instructive to run some multiparticle track- [31 M.G. Tiefenback and D. Keefe, IEEE Trans. Nucl . Sci.
ing codes which include space charge self-consistently . NS-32 (1985) 2483 ;
Also M.G. Tiefenback, LBL-22465, PhD thesis, UC-
These codes would be able to better include the effects
Berkeley (1986) .
of chromaticity (which may provide damping) and the [4] I. Hoffman, L.J . Laslett, L. Smith and I . Haber, Particle
effect of space charge on the dispersion function . Multi- Accelerators 13 (1983) 145 .
particle tracking has been done by Machida [131, for a [51 F. Mills and J. MacLachlan, private communication; and
variety of machines. These simulations display some C. Celata and L.J . Lasslet, Envelope Instability Status
unexplained emittance growth in the case of the SSC Report, unpublished .
LEB. This effect could not have been due to an accurate [61 I.M . Kapchinskij and V.V . Vladimirskij, Proc. 2nd Int.
calculation of envelope instability, however, as that Conf. on High Energy Accelerators and Instrumentation,
would entail evaluating the space charge kick many CERN (1959) p. 274.
times per focusing cell . This scheme could be easily [7] F.J. Sacherer, IEEE Trans . Nucl . Sci. NS-18 (1971) 1105 .
[81 M. Reiser, Theory and Design of Particle Beams, 1989 US
implemented to observe the instability computationally,
Particle Accelerator Summer School notes, unpublished
as the growth rate is high enough that one would only
book draft.
need track through a few tens of cells to see an effect. [9] M.A . Furman and L.K . Chen, Proc . 1989 IEEE Particle
If one becomes convinced that the envelope instabil- Accelerator Conf . (IEEE, New York, 1989) p. 1331 .
ity may be a problem in a machine, then observation of [10] E.L . Hubbard, ed ., Booster Synchrotron, Fermilab Pub-
the instability becomes an important issue . This could lication FN-405 (Fermilab, 1973).
be accomplished using the turn-by-turn beam profile [11] KAON Factory Proposal, TRIUMF, Vancouver (1985) .
f-nonitor currently under development for use in the [12] M. Syphers, private communication.
Fermilab booster [141. A beam which is envelope unsta- [131 S. Machida, Proc. 1989 IEEE Particle Accelerator Conf.
ble, where the growth rate is smaller than a turn period, (IEEE, New York, 1989) p. 1061 .
should display a growing oscillation in beam size when [141 V. Bharadwaj and J. Rosenzweig, private communication,
observed turn-by-turn . These oscillations should damp and
J. Krider, Nucl . Instr. and Meth . A278 (1989) 660.
as the emittance grows due to phase space filamentation
induced by the beam's nonlinear self-fields. As the
emittance growth the beam will become larger and the
space charge tune shift will diminish, taking the system
below the instability threshold.