SLAC-PUB-2498 AATF-80/21 April 1980 (T/E) BEAM EMITTANCE GROWTHCAUSED BY TRANSVERSEDEFLECTING FIELDS IN A LINEAR ACCELERATOR* Alexander W. Chao, Burton Richter, and Chi-Yuan Yao Stanford Linear Accelerator Center Stanford University, Stanford, California 94305 ** ABSTRACT The effect the emittance accelerator solved uniformly design of the beam-generated of an intense is analyzed by a perturbation accelerated tolerance bunch of particles in this paper. method for beam. transverse for The equation are applied the recently fields in a high-energy of motion cases of a coasting The results specifications deflecting on linear is beam and a to obtain proposed some SLAC Single Pass Collider. Submitted * ** Work supported to Nuclear by the Department Instruments of Energy, Visiting scientist from the University Hofei, People's Republic of China. and Methods Contract of Science DE-AC03-76SF00515 and Technology, -2- 1. Introduction Irthis paper we examine fields, generated through a conducting ticles which devices sors by the passage pipe, generated determining at their collision point intense bunch required ance of these the walls which will intense verse charge deflect particles dimensions are large wake field will of the bunch are deflected ture. Thus, travels of the pipe bunch of particles transverse linear reaction during colliders off-axis and leaves traveling travels energy rates of microns and the mini- the acceleration can degrade in colliding to the emittance down a pipe, behind of of the the perform- behind through be such that further as the bunch travels phase space occupied the point a structure will 2) all particles down along interacts wake field charge. the pipe, the particles If an whose trans- of the bunch, away from the axis by all it a transverse compared to the length in transverse increase. high devices. When a point with for succes- of the order related growth very in beam These linear dimensions useful important colliding achieving is directly emittance is 1) as possible system. to achieve dimension Significant problem of linear for beam transverse mum beam transverse the beam. rings center-of-mass require bunch of particles much attention beam storage deflecting of the same bunch of par- This to the luminosity the electron-positron beam devices of an intense fields. are now receiving to colliding of transverse on the emittance these the limits which the effect behind the the head of the structhe total area in the bunch -3- We examine appliGtion the effects to linear tions of motion tions by a perturbation uniformly for analytic are applied required for dimensions relative and then solve Finally, for bunches We calculate were zero. x(z,s), as a function and s, the distance of the bunch is very . ~(2,s) is The transverse with wake, the results an of this of an intensity and as if the displacement much less is uniform the displacement of of z, the longitudinal from the beginning of zero transverse wake field the bunch at the position charges equa- beam and a of the bunch (z is positive The approximation Thus the transverse effect, these the bunch as relativistically bunch is a good one in most cases of interest dimension strong the equa- beam applications. to the center head of the bunch), accelerator. find the cases of a coasting found. 3) is particular We first For the case of a very we treat on the bunch, position beam. with of Motion transverse a point method for colliding In what follows its accelerators. to the SJAC linac linear Equations wake fields in the bunch, solution analysis 2. electron particles accelerated asymptotic of these toward the of the dimensions for'the where the transverse than the size of the pipe. across the bunch. of the center of a slice In through z. force at z depends on the displacement z' > z and is given of all by 00 Fx(z,s) = e2 dz'p(z')W(z'-z)x(.z',s') / z (1) -4- where p is the line normalfzed to the total the transverse axis S density field produced to the retarded charge time the average W of the structure the case in electron The equation d ds charge, for of a particle N), displaced the point the displacement s so that (/pdz in the bunch, by a point z' - z behind ' = s - z' + z refers with in the bunch number of particles by x at a distance assumed that of the particles is e.W.x is from the and the field. We have changes sufficiently slowly can be used (certainly linac). of motion for x(z,s) y(s)x(z,s) ] + ($-$ can be written as = r o idn'p(z')W(z' -z)x(z',sl) (2) . where y(s) being is the energy the rest length mass of the particle; of betatron classical radius focusing focusing from a series bunch length retardation eq. W, y, of widely wave- s; and ro= e2/mc2 is the solve beam in which by a smooth function spaced quadrupoles. is much shorter first is the instantaneous X, and p are known functions, is provided (2) can be replaced coasting at position can be ignored, We will X(s) 2 of mc , m s in units of the particle. We assume that betatron of the beam at position than i.e., x(z' ,s') the than coming rather We also the betatron and that assume that wavelength on the right-hand the so that the side of by x(z',s). eq. (2) using a perturbation the bunch has constant energy method for as a function a of s. -5- 3. Coasting %r pendent Beam Solution a coasting of s. placement beam we have y(s) Consider of x=x0 the zeroth order a bunch injected and a slope solution for x1= 0. We expand x(z,s) into ko= 2~/h . , 0 in a series = X0 both the linac In the limit the bunch motion xb) (2,s) where we have defined = y, and X(s) with indea dis- of no wake field, is simply = xocoskos (3) of powers of the wake field co x(z,s) = c ,(n)(2,s) n=o (4) . and obtain the n th order term fromthe d2 ,W (z,s) + kix+z,s) = > ds2 which is a direct xcn)(z,s) = consequence fds'G(s,s')> 0 0 with G(s,s') the Green's (n-l)th order term by solving jdz'p(z')W(z'-z)x("-l)(z',s) 0 z of eq. (2). The solution rdz'p(z')W(z'-z)x("-l)(z',s') z (5) of eq. (5) is (6) function: G(s,s') = $ sinko(s0 s') (7) -6- Inserting x (0) from eq. (31, we obtain x(l)(z,L) xo(&)($ by iteration h = sin,,) . [dzlo(nl)W(zl- L coskoL + -8k 0 0 co co . sink z> - z> I dz2P (z,)W(z, “1 I dzlP (zl)W(zl z (8) - zl> . . . where we have taken field, one can ignore the deviation 1<<Ix 0 all higher from its is sufficient 1 which at the end of linac the terms of a particle It by x(l). lx(l) values than zeroth to keep only s=L. For a weak wake the first order order trajectory the first-order and is given term if requires Lro m I dzlo(zl)W(zl2yoko 2 The fact that of the resonant xb) a cos(kos), the sketch along should. tion If 1. and orders. pictures consequence and of the bunch looks of x (1) depends on the position at the head of the bunch x the wake field to higher of snapshot The value to s is a direct Since x(-l' a s sin(kos) situation. a series in fig. the bunch, Ix (1) 1 is proportional driving (9) z> << 1 is not weak, (1) vanishes one has to carry like z as it the calcula- -7- In many practical bunch?;as i.e., completed koL >> 1. reduces situations, many betatron In this case, is long oscillations we find enough so that before after reaching some albegra that the s=L, eq. (8) to ,b) (z,L) where Rn is defined Rn(z) the pipe z x0(-&y eikoL . R,(z) (10) by dz2P (z2)Nz2-zl>. dzlP (z,)W(z,-d = $- ($)n z .. I z s1 dznp (~~>“(z,y-z~-~) n-l (11) It is understood Note that that x (n+~,s) the variables A closed rectangular only the real h as the useful s and z, although of eq. property x(z,s) form can be obtained distribution part for = N/R 1 0 it is meaningful. is factorizable Rn if we approximate function, for jz[ < R/2 for Iz] >= R/2 p by a i.e., (124 w = w,z/I1 Both approximations to allow order tion a good assessment terms. for are close In particular the SLAG linac. (12b) enough to reality in many applications to be made of the importance the linear We find in is not. and W by a linear P that (10) wake is a quite of higher good approxima- -8- (13) and x(z,L) = (14) xoe where we have defined (15) The validity condition becomes In/41 CC 1. expression for eq. For intermediate (9) f or the first-order eq. In the limit approximation InI >> 1, one can find an asymptotic (14): values of lnl , the power series expression (14) is more accurate. 4. Accelerated Beam Solution We assume that as a result the energy of acceleration the beam energy yomc2 We assume that with constant, h(s) so that = X0 . in such a way that at injection, the strength beam energy of the beam increases y(s) linearly = y,(l+Gs), and G the acceleration of the focusing the instantaneous force betatron with s with gradient. in the linac wavelength scales remains -9- We first make a change of variable from s to a new variable u= l+Gs 2 ax -+Ldx+ u du du2 where we have defined by an iteration order solution at s=O by setting the initial -z)x(z',u) equation as in the coasting is obtained zero and demanding This x(z,u)=xo (17) again be solved The zeroth beam case. N(k)J = x0 . -q side of (17) to and dx(z,u)/du=O (&)-Jl(+)No(;u) where J If 0' J 1' No' and N are the usual 1 we expand x(z,u) obtained the x's from the (n-l)th on the left-hand hand side replaced a Green's x(~)(z,u) order side (4), - Jl(>&m Bessel functions. the n term x (n-1) replaced by x (n-1) . order by solving by x (4 The solution th term x Cd can be eq. (17) with and the x on the right- can again be obtained function: = J"du'6(u,u') 1 the Green's as in eq. (18) O kG Nl($)Jo($) with (2) or u=l: x(O) (z,u) using will the right-hand conditions of motion (2) then becomes dz'p(z')W(z' ko=2a/Xo. procedure Equation . x=------ in the equation function " yoG2u' rdz'p(z')W(z'z z)x(n-l)(z',u') (19) -lO- - G(u,u’) = In practice, + .[No(+ over.most betatron oscillation required to double eqs. and (20) argument x(O) i.e., for (z,s) = in the Bessel cos pi G(u,u’) The factor ,/l+Gs The terms recursion in eq. (z,s) 1) f-- (21) in the series ,b> where R,(z) = [ I the distance that is then appear in We can use the large to obtain J& the usual solution than the (211 = sin[>(u- is accelerator, functions functions kO $U- (20) u’)] The acceleration than unity. the Bessel - of a linear ko>>G. are much larger expressions u)N~> is much shorter the energy, and the arguments - Ji+ ut) of the length wavelength adiabatic (18) u)Jo($ =Oskos u')] 02) adiabatic damping factor. are = has been defined (23) in eq. (11) and I,(s) is given by the relation S In(s) = IO(s) = sin coskos ko(s1 s') 1In-l(s') (24) -ll- In general, Howevm, the solution if the beam energy than the beam energy solved to eq. (24) for In is rather complicated. at the end of acceleration at injection, i.e., (l+GL) is much higher >> 1, eq. (24) can be to yield I,(L) where taking the real Comparing the coasting results =. $- part beam results, (25) eq. beam results, eqs. (23) and (25), (lo), that the accelerated we note from the coasting substitution beam result to all with beam orders by rule Coasting Beam constant energy injection !?n(l+GLgn is understood. the accelerated can be obtained the simple e ikoL[& Accelerated --+ y, displacement x injection Beam energy y, + 0 length of linac L (26) where the last lation rule L+$ phase exp(ikoL). 2 is y mc . f result replace wake field apply energy beam result to the betatron oscil- at the end of acceleration reduces to the coasting beam G-to. For a bunch with linear does not The particle The accelerated in the limit Rn(yf/yo) a rectangular approximation, x0 by (yo/yf) l/2 charge eqs. distribution (14) and (16) still x0 and n of eq. (15) by and in the hold if we -12- Lr NW n In fig. koL=O, that 5. = ko(y;- 2 we have plotted ~r/2, n, and 3~/2 the value clear 2 that Misalignment analysis, the effect cision and it travels of low beam intensity. position that a displacement down the linac We will is such to 150. It large. into error. the linac in a straight is adiabatic. dz'p(z')W(z'-z) The accelerator position will error be treated In this with line the acceleration can be written as a consection, of the accelerator the case with of motion the accelerator is produced study is the transverse s. of strength of the bunch can be very injected d2x -+Ldx+ u du du2 where d(s) values of the bunch is equal caused by misalignment We assume the beam is The equation The wake field and the wake field with pipe. the approximation (27) the bunch for we have assumed that aligned of beam injection study tail ( ) along 2~r). the distortion is perfectly we will (modulus ' 2 lz --2 R Effects In the previous sequence x(z,L) of n at the very from fig. structure Yf y,) ai1< 0 perfect pre- in the limit acceleration under as [ x(z',u) of the pipe t as N - d(s)] (28) structure structures, at with the C i th eq. structure (28) misaligned contains by a distance an additional comes from the pipe misalignment. force d.. 1 Compared with term on the right-hand eq. (172, side that is -13- The zero-th order solution is asTurned to be injected of the bunch strictly line. without (28) error. is x ('I = 0 since The trajectory x (0) and is therefore follows The first-order alignments to eq. perturbation the beam of the head a perfect term comes solely straight from the mis- d.: 1 r d.R. x(l)(z,s) = - ; c i 5; = l sin[ko(s-si'l (l+Gs)l/2:l+GsZ)l/2 00 Rl'.Z' l I wq (29) where the factor sin an angular and the quantity kick, Note that term in this of this hand, with bation we do not driven = C i (S'Si) 2 "odiRi 2y;kzG error, with value is driven is necessary eq. (l+Gs)/(l+Gsi) . (l+Gs)l/2(l+Gsi)l/2 frequency apply. to carry (29) into As does oscil- may acquire in wake field. by substituting (11). On the other which of the system and it (1) . xw to the first-order does not of x by in eq. the natural condition up to the second order term can be obtained (z,s) a large term x (2) reason, an injection driving expect of the response has been defined by a force frequency For this calculation order R,(z) and the resonant the natural amplitudes. is characteristic - y)] the case for case is not the second-order late xW unlike system a consequence [ko(s large out the perturThe secondeq. (19). 1cosPo(s -'ii]* R2(Z) (30) If pipe we assume the misalignment structure to the next, errors di are uncorrelated from one -14- <x(1)2> = <x(2)2> = & R;(z) <d2> (-$-rEn3(zJ *R;(z) (31) C where <d2>lj2 that all is the rms value structures have the same length have approximated linac the sum over and we have also The factor the fact that l/NC If acquire i by an integral in expression the an additional (12b), ratio of <x (212 > to <x(lj2 we can use eq. by empirically beginning tributions growth controlling of the linac. factor . C In eq. over that (31), the length yf>> is not of l/N > under t o obtain these the injection in eq. R1(Z) of resonantly. <x (212) that C’ (12a) and a.'linear the quantities wake R,(z). x 0 first-and The is n2/1728. can be substantially offset The corresponding of the y,. driven assumptions due to misalignment we <x (112 > is a consequence distribution (13) have been obtained and we have assumed by x (1) is shown by the fact resonantly field for perturbation we assume a rectangular The emittance ai=L/N made the approximation the first-order That xC2) is driven does not of the misalignment and angle reduced x' 0 second-order at the con- (23): Jn(l+GL) 2G x sink L xA o o -<'OSkoL (32) x(2)(z,L) = -15- By choosing proper the first-order ‘X eq. if For example, is possible contribution, (30), eq. to cancel (29), either or the second-order by a corresponding the second order contribution misalignment term dominates, choose 0 0 X’ contribution, (32). one might of x0 and x: it misalignment misalignment from values an[(l+G~)/(l+Gs~)l 4G = diRi c i !?,n2(1+GL) (l+Gs i )1'2 ' 0 (33) so that angle the second-order cancels required contribution the second-order contribution x0 and x: have rms values <xi> = from the injection YflY 3Nc !?n(y,/;,) -ii- given offset and The from the misalignments. . by 2 i34) <d > 0 With x0 and x: given the sum of misalignment by eq. (33), the first-order and injection term obtained contributions by is Rn(l+Gsi) Rn(l+GL) (35) The rms value of this x (1) <x(l)2> = is given by Yf Rn y 0 l R;(z) (36) -16- which is l/3 Thus this of that the injection misalignment first-order 6. no injection the emittance conditions contribution growth not only but also cancellation effort. due to misalignment cancels significantly the secondreduces the contribution. Application to SLAC Linac The wake field and P. Wilson. 4) is linear collider the case with scheme of minimizing by controlling order for in z. for the SLAC linac In the regions The relevant operation has been calculated of interest parameters yf for the wake field the proposed single W(-z) pass arel) N = 5 x lOlo YO to us, by K. Bane u = 2.4 x lo3 = lo5 (50 GeV) 3 L=3XlO (1.2 Z = lmm 5 -5 W. = 5.9 x 10 m GeV) CT = 70 urn X m NC = 240 X0 = 100 m Under these (37) the asymptotic conditions, order to find cant emittance the proper tolerance criterion The value growth. - 1 for case has been shown in fig. of x(z,L) z = -0 z. If a wake field obtain a tolerance not having to eq. a signifi- (27), is 37 The bunch shape 2\/5 the maximum magnitude we require must be used in L, and 150 at z= -R/2. 94 at z=-0 this =- for of 0, according at z=O, Z expression 2. are 1.5x0 on the injection The corresponding for displacement displacement values z= 0 and 6.1x0 at z=-oz of of for of <ux, lx01 s 11 pm. we -17- This tolerance injectj%n on the injection stability since by a set of static inc term rather (33). given offset , is the bunch center given <x:>~'~ by eq. beam size size ox at the end of the linac, <d">1/2 growth 2 l/2 <d > this 0 by offset, optimizing beam size is found to be less is given injection After <d2>l12. the beam with the second-order of x0 and x: at oz behind at z = -oz for by con- growth, to be 0.25 2 112 the bunch center. than the transverse we demand a misalignment at <d > For beam tolerance of = 0.11 mm. The effect of the accelerator ing the reduction Since Since x'. of the required which at z=-oz, by injecting the resultant (36), perturbation 2.2. chaise = 0.35 conditions, and 0.62 on the jitter- For a particle and angle 0 rms value the injection <x("> 2>w x The expected is for be canceled tolerance can be minimized the optimum (34), can always by the second-order term. dominates, by eq. trolling effect determined contribution eq. are cominated (2) to xrms (1) is about xrms The misalignment empirically error the requirement magnet is + 1 grad. the first-order the ratio example, sets The corresponding kicker effects than x0, the injection magnets. of the injection Misalignment error, in luminosity the luminosity reduction factor is misalignment arising inversely R is approximately R = from the emittance proportional given is determined by examingrowth. to the emittance, the by (38) -18- where <x(~)~ > is given nosity reduction <d2>l12 order the rms orbit = 0.1 mm, the reduction in luminosity term is factor appreciable affect taken in the beam which different account become significant there if to investigate The third- will frequencies in the bunch, energies Since and thus differ- be a Landau damping effect is the spread (Ak 0 in betatron which in the betatron 0 A numerical tracking program is being pre- effect. Summary We have studied tance growth We first then 20%. For of the bunch and thus spread different Ak L>v this is about of the spread have slightly frequencies, 2 l/2 . <d > distortion, tail the lumi- noticeably. wave number in the bunch). 7. at the very comes from the energy particles ent betatron only the luminosity We have not pared 3, we have plotted fig. versus does not will by eq. (36).In of an intense it (4), accelerated (lo), and (11). a substitution the result from the coasting For practical ing a rectangular assumptions, wake fields on the emit- in a linear accelerator. particles in the bunch and cases of a coasting The coasting beam result For the case of an accelerated rule, applications, charge for method for beam. have found Under these of motion by a perturbation and a uniformly of transverse bunch of particles set up the equation solve by eqs. the effect eq. (26), that allows beam is given beam, we one to obtain beam result. we simplify distribution an asymptotic the calculation and a linear expression by assum- wake function. of the bunch shape, -19- eqs. (15), and (27), (16), can be obtained for cases with strong wake fields? We have looked acceleration effect at the effects sections. conditions collider. are applied We find in luminosity injection less applied. The can be minimized After by con- the minimization to beam emittance is given by (36). These results tion growth of the bunch. rms perturbation of the linac method is again on beam emittance the injection scheme, the expected eq. The perturbation of misalignment trolling caused by misalignments angle than +O.l to the recently that the emittance will be tolerable is within growth provided proposed SLAC single and the associated the jittering f 1 urad and the accelerator mm. pass reduc- in the misalignment is . Acknowledgments We would concerning this like to thank work. Dr. Rae Stiening for many useful discussions -2o- REFERENCES 1) Design 2) R. Helm and G. Loew, The Stanford Report R. B. Neal, also 3) for the SLAC Single New York, references W. K. H. Panofsky quoted Pass Collider Two-Mile W. A. Benjamin, Project Accelerator, Inc. (1968), (1980). edited p. 173. and M. Bander, K. Bane and P. B. Wilson, unpublished See therein. Rev. Sci, Instr. 39, No. 2, 206 (1968). 4) by CN-16 (1980). "Wake Function for the SLAC linac," -21- FIGURE CAPTIONS Fig. 1. Sketches four Fig. 2. of bunch shape for instances Bunch distortion a case with weak wake field at of time. at the end of accelerator ent values of total betatron wake field strength parameter for phase koL (modulus Q is taken four differ- 27r). The to be 150 at the bunch tail. Fig. 3. Luminosity reduction ment tolerance Collider. 2 l/2 <d > factor for R versus accelerator misalign- the case of the SLAC Single Pass k,s=O 4-80 3817Al Fig. 1 100 50 0 -50 - 100 -0.5 4-80 0 z/P Fig. 2 0.5 381485 R I 0 I I I I 0 4- 80 100 < d* > “’ I I I 200 (pm) 3814A44 Fig. 3