High beam current storage at low energy for compact synchrotron
Transcription
High beam current storage at low energy for compact synchrotron
High beam current storage at low energy for compact synchrotron radiation rings (invited) H. Takada and Y. Tsutsui Sumitomo Electric Industries, Ltd., Osaka Research Laboratories. Shimaya, Konohana. Osaka 554, Japan T. Tomimasu and S. Sugiyama Electrotechnical Laboratory, Umezono, Tsukuba, Ibaraki 305, Japan (Presented on 1September 1988) A study ofhigh beam current storage at a low energy is being conducted on the compact electron storage ring NUl-I. In general, it is said that the stored beam lifetime is rapidly shortened as the beam energy decreases, and the high beam current storage is difficult to obtain. However, a stored beam current above a 350 mA was obtained at an injection energy of 100 MeV, and the lifetime of the stored beam is considerably long. For example, e-folding lifetime is about 2 h at 100 MeV. In this paper, we estimate the beam current decay rate due to the residual gas scattering, the ion trapping effect, and the Touschek effect, and make clear these contributions to the beam lifetime. It was clear that the Touschek lifetime is lengthened according to the bunch size growth, which is roughly explaine<;l by the longitudinal coupled bunch instability. INTRODUCTION Low-energy injection is of great benefit to the compact SR (synchrotron radiation) source for industrial use, because it enables a low-cost system. However, it is not easy to store a high beam current at low energy because of the short beam lifetime and the beam instabilities. The compact SR ring NIH-l was constructed by Electrotechnical Laboratory and Sumitomo Electric Industries, Ltd. I Its main purpose is to investigate the above-mentioned problems. The shortening of the beam lifetime at a low energy is mainly caused by the Touschek effect, but the ion-trapping effect and beam instabilities may also affect the beam lifetime significantly. Actually, it is reported that the SR ring, Aladdin, of the University of Wisconsin was not able to obtain high beam current without an ion clearing system. 2 Also, in NIH-l ion trapping has harmful influences at a high beam current, and the ion clearing is necessary in order to get the beam current above 200 mAo The observed decay of the stored beam current is considerably affected by the beam size change. This means that the Touschek effect has significant influence on the beam lifetime. The Touschek effect, first observed on the small storage ring AdA, 3 has been mainly theoretically studied4 - 9 and several experimental interpretations were also tried. 3 ,8,9 But experimental studies for a low-energy and high-current beams have been lacking. We have been quantitatively investigating the Touschek effect and other mechanisms which have great influences on the beam lifetime at a low energy. In Sec. I, the effects of the gas pressure and the ion trapping are evaluated. And the Touschek lifetime is discussed in connection with the bunch size growth at a low energy in Sees. II and III. Figure 1 is a photograph of the NUI-l and the main parameters are tabulated in Table I. The first beam storage was obtained in February 1986 and after several improve1630 Rev. ScLlnstrum. 60 (7), July 1989 ments stored beam currents ofabove 400 mA at a beam energy of 150 MeV and of above 350 mA at 100 MeV were achieved. I. DECAY OF STORED BEAM CURRENT If the rf accelerating voltage is enough, the following mechanisms are considered to be the causes of the decay of the stored beam current, (A) the Tousch~k effect, (B) the residual gas scattering, and (C) the ion trapping effect. It is necessary to evaluate these effects quantitatively in order to store high beam current at low energy. A. Touschek effect Electrons in the same bunch scatter each other elastically due to their betatron oscillation. If the change of longitudinal momentum after collision exceeds rf acceptance, these " FIG. I. Photograph ofNIJI-1. 0034-6748/89/071630-06$01.30 © 1989 American Institute of Physjcs 1630 TABLE I. Main parameters of.NIJI-1. Parentheses signify rQJltine operation." . ." ". ,. where Z is the atomic number of the gas. And tJc is given by . tJ = 80-160 MeV 230 Mev 13.256m 0.7m ( 1.37) ( 1.57) (0.46) 7 158.4 MHz 2kW Injection energy Maximum energy Circumference ' Bending radius Horizontal betatron tune Vertical betatron tune Momentum compaction factor Harmonic number Radio frequency Maximum rf power A ( 7 ) ..J(pz)Pz. c where A is the vertical half aperture of the vacuum chamber at the bending section, (P z) is the average vertical betafunction, and Pz is the maximum vertical beta-function at the bending secti~J.' . .. .. . . The decay rate due to this process is given by 1 dII· = (0'1+ 0'2 + O'3)dmcI dt v ' = 3.217X 1022 (0'1 + 0'2 + O'3)cPI, electrons are lost. This process is referred to as the Touschek effect. '.' . ' .. The decay rate ofthe stored beam current due to ~his effect is expressed b y ' . . . dII = aI 2, .. (1) dt T 1 where I is the stored beam current. Assuming that the vertical component of the transverse mo~entum is much less ,than the horizontal componentand transverse momentum is nonrelativistic, a is given by ,the . following formula4 : . _ a= C(E) r !ii~c,C(€) . ". O'~ (AE / E) ~ax Velr r ~ e- E +'.!-. f.oo = - 2 . + 2 E (3€-€ln€+2) 2 = -ln€-2.077 (2) In u e- "du U f.oo -e-"- d u E for (3) U €~I, where 70 is classical electron radius, c is the velocity oflight, r is Lorentz factor, O'~ is the horizontal beam divergence, (AE / E) max is the maximum relative energy deviation accepted by the rf system, Vis the bunch volume, e is the electron charge, Irr is the rf frequency, and € is defined by [(AE /E)max/( rO'~) ] 2. B. Residual gas scattering The total cross section which causes beam loss is composed of the Rutherford scattering due to nuclei, the bremsstrahlung due to nuclei, and the Moller scattering due to atomic electrons, these cross sections are given by 41Tdz 2 0'1= 2tJ2' r 0' 4 2 =--7 2 137 (4) c 0 Z Z ( (4-In I - -65)In-183 3 (AE/E)ma; Z1/3' + I) (5) 21T~Z 1 0'3 = - - - x - - - - r (AE/E)max 1631 Rev. Sci. Instrum., Vol. 60, No.7, July 1989 (6) (8) where d m is the density ofgas molecules, and P (Torr) is the pressure of the gas. In NUI-l the components of gas molecules are mainly CO·and H 2 and the ratio of their partial pressure is almost 1: 1 by measuring mass spectrum. The partial pressures of CO and H 2 are estimated with a monitored pressure, taking into account this ratio and the difference of gauge sensitivity for each kind of gas. The decay rate due to the residual gas scattering is calculated from these partial pressures.. Although the cross section increases as the beam energy de~ creases, SR induced gas desorption'is reduced at low" energy; Therefore, I- dI / dt Iv is not a: maii!. cause of lim~ting beam lifetime at low energy when the vacuum condition is good. At present, total pressure is almost 1 n Totr at a beam energy of 150 MeV and a beam current of 100 mA, and in this condi7 tion the contribution of I- dI/dt Iv to the. tota,l measured decat ratel:'- dI/dt 1m is estimated to be about 20%. c. Ion trapping effect In NIH-1 vertical beam widening and betatron tune shift and spread are observed as the stored beam current increases. These phenomena are caused by ion trapping... Assuming that the charge of ion is + 1, that transverse distribution of ions is a gaussian distribution which is a duplicate of that of the electron beam and that an 'electric field produced by ions is linear, the betatron tune shift Avx,z is given by J O'z,x d ; Px,zds, (9) O'x + O'z where d; is the density of the trapped ions, O'x,z is the transverse beam size, PX,z is beta-function, sis longitudinal length, and x and z means horizontal and vertical directions. Accordingly, d; can be estimated by measuring Avx,z and O'x,z' The betatron tune is measured by the rf-knockout method and the transverse beam size is measured by using the arrayed photodiodes. In calculating, we use the average beam size which is estimated according to the following relations: Avx,z = -70 r (O'x,z) = LL O'x,z (s)ds/L, O'x (s) =..J €xePx (s) Industrial application + rp<s)(~E /E)2, ( 10) (11) 1631 (12) Uz(s) =..jEz f3z(S) , 8E)2 (Ii E z = U 1 2· = [U xO (so) - Exo f3x (SO)] r/(SO) , ~, (so) f3z (so) , (13 ) (14) where L is the circumference of the ring, Exo is the natural horizontal emittance, and U x 0 and U z 0 are the observed horizontal and vertical beam size, respectively, at the longitudinal position s = So in the ring. In order to suppress the ion-trapping dc ion-clearing field is applied to button electrodes which are the same ones used for the beam position monitor installed at three points in the ring. Figure 2 shows the clearing voltage dependence of the ion density di estimated by using Eq. (9). The ion density rapidly decreases and is almost saturated as the clearing voltage increases, and this saturating voltage is reduced as the beam energy decreases as shown in Fig. 2. This tendency can be interpreted with bunch size growth by decreasing energy which results in the reduction of beam potential. Bunch size change is described in detail in the following section. The influences on beam lifetime due to ion trapping are thought to be from the trapped ion scattering and the change ofTouschek effect due to the beam size change, as long as the operating point is not shifted to the dangerous resonance line. In order to understand them quantitatively the decay rate due to the trapped ion scattering I - dI / dt I; is calculated by use ofestimated d;. Figure 3 is an example of the decay rate plot. As shown in Fig. 3, the measured decay rate I - dI/dt Im without ion clearing is less than that with ion clearing in the beam current range 20-150 mAo This can be explained by the decrease ofTouschek effect due to the vertical beam size growth. Therefore in this region ofstored beam current, the Touschek effect has a much greater effect than that of trapped ion scattering. This agrees with the evaluated value of! - dI/dt I;. Above 150 mA, the measured decay rate without ion clearing considerably increases as the stored beam current increases. But an ion clearing suppresses the fast decay in high beam current and allows us to obtain more current. The 5r------~-----__._____, .aOMeV o 150MeV "" I FIG. 3. Decay rate of the stored beam current as a function of the stored beam current with clearing dc voltage of 0, - 0.2, and - 0.4 kV. Solid lines indicate measured decay rate and broken lines show estimated decay rate due to the trapped ion scattering. evaluated I - dI/dt I; considerably decreases as clearing voltage increases, and the ratio of I - dI / dt VI - dI/dt 1m can be reduced to about 10%. As a result, the ion trapping effect may limit stored beam current and shorten lifetime at high beam current without ion clearing, but an adequate ion clearing system can supress this. II. BUNCH SIZE A. Bunch size growth Since the Touschek lifetime depends on the bunch volume as in Eq. (2), it is important to know the bunch size. The horizontal, vertical beam size U x ' U z and the bunch length u/ is given by Ux =..jEx f3x + rlc uE/E) 2 , (15) Uz = ..jKEx f3z, (16) u/ = (cap/OJ s )/(uE/E), (17) E '0 FIG. 2. Trapped ion densities vs ion clearing dc voltage. Ion densities with 100 rnA at 80 MeV and 150 MeV are indicated by closed circles and open circles, respectively. where K is the coupling constant of betatron oscillation, Ex is the horizontal emittance, 1] is the dispersion function,· (uE/E) is the energy spread, a p is the momentum compaction factor, and OJ s /21T is the synchrotron frequency. In NUl-I, the observed bunch size is much larger than the natural bunch size. At first, it was thought to be caused by both the longitudinal coupled bunch instability and the multiple intrabeam scattering in a bunch. 10 The frequency spectrum of the beam signal shows several side bands separated by the synchrotron frequency around the harmonics of the revolution frequency, and these side 1632 Industrial application x o ------0 c o -0- o 0.5 Vdc (kV) Rev. Sci.lnstrum., Vol. 60, No.7, July 1989 1632 bands do not disappear even if a beam current decreases to 0.1 mAo Therefore, the growth of the longitudinal coupled bunch instability is significant in NIH-I. The bunch size, due to the longitudinal coupled bunch instability, is determined by the balance ofthe growth and the damping rate of the instability. The bunch length and the horizontal beam size due to this effect is given byll O'JO'/O = (lIIth O'x = ~Exf3x (18) )0.2, + rl(O'EIE)2(lIIth )0.4 for I>Ith , (19) where I th is threshold current and 0'/0 and (O'EIE)o are the natural bunch length and energy spread, respectively. The bunch size growth due to the multiple intrabeam scattering is expected to be not significant above 200 MeV. Therefore, we investigated at a beam energy of21O MeV and at lower energies in order to distinguish this effect. B. Stored beam current dependence The observed horizontal beam size O'x and the bunch length 0'/ depend on the stored beam current as shown in Fig. 4(a) and (b). They are almost proportional to 1°·2 at both energies of 80 and 210 MeV. This dependence agrees with the theory of the longitudinal coupled bunch instability, even though multiple intrabeam scattering is expected to be dominant at a beam energy of 80 MeV. The beam current dependence of the vertical beam size O'z is larger than that of O'x, however, it cannot be fully explained yet. As a result, the bunch volume V = (41T) 3/2 0'xO'zO'/ increases almost linearly as the beam current increases, and the Touschek lifetime is almost constant against the beam current. (a) .80MeV o 210MeV 200 E~ .;; 100 ..§. C. Beam energy dependence The beam energy dependence ofthe observed bunch size is shown in Fig. 5. It is noteworthy that the bunch length increases just a little as the beam energy decreases. According to the theory of longitudinal coupled bunch instability, the bunch length is independent of the beam energy because the energy dependence of I?i,2 and 0'/0 are compensated in Eq. (18). If the multiple intrabeam scattering was dominant, the bunch length could be expected to considerably increase as the beam energy decreases from 210 MeV. Consequently, the bunch size growth due to the multiple intrabeam scattering is much less than that due to the longitudinal coupled bunch instability. The energy dependence of the horizontal beam size O'x also mostly agrees with the theory of the longitudinal coupled bunch instability. In addition, the vertical beam size O'z increases rapidly as the energy decreases below 150 MeV as shown in Fig. 5. Consequently, the bunch volume enlargement against the beam energy is considerable. D. rf accelerating voltage dependence In routine operation in NUl-I, the rf accelerating voltage Vrf is empirically chosen to be about 24 kV to obtain high injection efficiency and long lifetime, then the overvoltage factor is very large. Accordingly, the synchrotron frequency OJ s 121T is almost proportional to V~(2, and 0'I ex V rf- 0.2 and 0'x ex VOr f·3 are expected by the theory of the longitudinal coupled bunch instability. These dependencies almost agree with the observed data as shown in Fig. 6. The vertical beam size does not change much and is almost constant against the rfvoltage. As a result the bunch volume is nearly constant against the rf voltage. because the dependence of the O'x and the 0'/ almost compensate. Normally raising the rf voltage is not very effective in extending the Touschek lifetime, because the bunch volume is reduced with the rf voltage. But when the longitudinal coupled bunch instability occurs, the bunch volume is not reduced and it helps the Touschek lifetime to extend as the rf voltage increases. 50 10 20 50 100 200300 Vrf=24kV I=100mA I (mA) 2.0 (b) • (Jx .... (Jz 1 80M V r e o (Jx "(Jz _ r 21 OMeV 1.0 .s 10 'C: ..§. 0 05 ~ 0.5 ~ 100 0 E E 'oS c O'.e I 20 200 E O'x 0 E'O.5 0 o .s ~ 00 O'z 0.2 02 0.1 l.--=':5o:-------:-10~0:----:2:-:070 -:::3700;;---- 10 0.1 5 10 20 50 100 200 300 E (MeV) I (mA) FIG. 4. (a) Horizontal, vertical beam size (ITx,ITz ) and (b) bunch length (IT,) measured at 80 and 210 MeV as a function of the stored beam current. FIG. 5. Horizontal, vertical beam size (ITx,ITz ) and bunch length (IT,) vs electron energy E. Solid lines indicate the dependencies of E - 0.5 for ITx and constant against E for IT/. 1633 Industrial application Rev. ScLlnstrum., Vol. 60, No.7, July 1989 1633 10' ...---,--....,..--,---. E=110MeV I=100mA (JQ. 2 E E '0 1.0 ~ V~ o Vrf=24kV 200 O•2 E ~03 100 E ~ ~ ~ 50 0.5 10 20 30 Vrf (kV) FIG. 6. Horizontal, vertical beam size (O'x,uz) and bunch length (0',) vs rf accelerating voltage Vn" Solid lines indicate the dependencies of V~i' for O'x and V,r 0.2 for 0'" and the broken line indicate constant against Vn" 1 o35':;o:----:1-=o";:"o--1:;-;5~0:---::2:':OO::---:2:-!50 E (MeV) FIG. 8. Energy dependence of the Touschek lifetime with V,r = 24 kY. The error bar shows the variation of I - dI I dt 1/ taken at stored beam currents. Solid line indicates the calculated value by Eqs. (I )-( 3) substituting observed bunch volume. III. EVALUATION OF THE TOUSCHEK LIFETIME As described in Sec. I, the decay rate due to the Touschek effect is estimated by 1- I = I- I -I - I-I - I' df dt df dt T df dt m df dt i (20) v where I - df I dt Ii is sufficiently reduced by ion clearing. An example of this separation is shown in Fig. 7. It is verified that the beam lifetime is mostly determined by the Touschek effect and the Touschek lifetime f 11- df Idt IT is almost constant against the beam current because I- df I dt IT is almost proportional to 1. The beam energy dependence ofthe Touschek lifetime is shown in Fig. 8. And the solid line indicates the calculated value by use of Eqs. (1)-(3) substituting the observed bunch volume. Both f II - df Idt IT' by the use ofEq. (20) and the solid line have a minimum of around 150-180 MeV, and these tendencies are caused by the remarkable growth of the bunch volume as the beam energy decreases. Although Eqs. ( 1)-( 3) have some assumptions and approximations, the Touschek lifetime calculated by use of Eqs. (1 )-(3) roughly agrees with the value estimated from the measured decay rate as shown in Fig. 8. As a result, it was found that the Touschek lifetime can be evaluated by use of Eqs. (1)-(3), even if the stored beam size is considerably enlarged by the longitudinal coupled bunch instability at low energy. IV. CONCLUSIONS FIG. 7. Decay rate separation into that due to the Touschek effect, that due to the trapped ion scattering and that due to the residual gas scattering. Open circles represent the measured decay rate. Broken line and dashed line show estimated decay rate components due to the residual gas scattering and due to the trapped ion scattering. Closed circles indicate the measured decay rate subtracted by these components. The beam loss mechanism in low energy was investigated quantitatively in the compact SR ring NUl-I. Although the ion trapping may harmfully affect the stored beam, the beam decay due to this effect can be reduced by ion clearing. Consequently, the Touschek effect is the main cause which determines the beam lifetime. Since the Touschek effect depends on the bunch volume, the phenomena which affect bunch size are very important. In NUI-l the bunch size growth is remarkaple and it is roughly interpreted with the longitudinal coupled bunch instability except for the vertical beam size. In general, the bunch size growth at low energy is expected to be mainly caused by multiple intrabeam scattering, but our results indicate that multiple intrabeam scattering makes a small contribution by taking into account the energy dependence of the bunch size. The vertical beam size growth may be caused by other instabilities or the residual ion trapping, but cannot be fully explained now. The energy dependence of the Touschek lifetime is also affected by the rapid bunch growth as the beam energy decreases, consequently the lifetime has a minimum of around 150-180 MeV and becomes long at lower energy~ Since the Touschek effect is reduced by the bunch size 1634 Industrial application E=150MeV Vrf=24kV Vdc=-lkV 5.0 2.0 o c 'E 0 0 .. 0 0 o • • 0 0 "- 1.0 « E 0 0 I /; II 0 ~ '6 0.5 0 "...... " I 0 0 0 I /f I 0.2 //1 / 0.1 1 1 1 005 1 10 20 1 I / / / 50 100 200 I (rnA) Rev. ScLlnstrum., Vol. 60, No.7, July 1989 growth due to the longitudinal coupled bunch instability or multiple intrabeam scattering, the shortening of the beam lifetime at a low energy is confirmed to be no longer serious. But the other phenomena which are occasionally observed in NUl-I, such as an abrupt beam loss and an unusual fast decay, are apt to occur at a low energy. Further investigation of these problems has to be continued. Electrotechnical Laboratory, and engineers of Sumitomo Electric Industries, Ltd. for their support. IH. Takada, K. Furukawa, and T. Tomimasu, Opt. Eng. 27, SSO (1988). 2B. Schwarzchild, Phys. Today, March 19 (1986). 3c. Berrardini, G. F. Corazza, G. Di Giugno, G. Ghigo, J. Haissinski, P. The authors would like to thank the staff of the High Energy Radiation Section, Quantum Technology Division, Marin, R. Querzoli, and B. Touschek, Phys. Rev. Lett. 10, 407 (1963). 4H. Bruck, Accelerateurs Circulaires de Particules (Presses Universitaires de France, Paris, 1966). sB. Gittelman and D. M. Riston, HEPL-291, Stanford University (19Q3). 6U. V6lkle, DESY 67/S (1967). 7H. Wiedemann, PEP Note-27, SLAC (1973). 8H. Bruck and J. Le Duff, in Proceedings ofthe 5th International Conference on High Energy Accelerators, Frascati, 1965, p. 282. 9y' Miyahara, Jpn. J. AppI. Phys. 24, L742 (198S). 10M. S. Zisman, Lawrence Berkeley Laboratory Report No. LBL-19l91 Preprint (198S). liS. Asaoka, G. lsoyama, H. Mikuni, Y. Miyahara, and H. Nishimura, NucI. lnstrum. Methods 215, 493 (1983). 1635 Industrial application ACKNOWLEDGMENTS Rev. Sci.lnstrum., Vol. 60, No.7, July 1989 1635