Beam-beam Experience at PEP-II W. Kozanecki, CEA-Saclay Introduction: PEP-II collision parameters & recent performance Interplay between e- - cloud & beam-beam issues Beam-beam limit studies Summary

PEP-II Collision Parameters J. Seeman, Jun 03 PEP-II Collision Parameters IP Parameter Design Peak performance (Jun 03) C-M energy (GeV) (e+: 3.1 ; e-: 9.0) 10.58 10.58 Crossing angle (mrad) 0.0 < 1.0 Luminosity (x 1033/cm2/s) 3.00 6.57 Number of bunches 1658 1034 LER current (mA, e+) 2146 1550 HER current (mA, e-) 750 1175 LER/HER current ratio 2.9/1 1.3/1 by*/bx* (cm/cm) 1.5 / 50 1.2 / 40+, 1.2 / 28- Emittance (nm-rad) (y/x) 1.5 / 49 1.8 / 30+, 1.8 / 49- IP rms beam size sy/sx (mm) 4.7 / 157 4.6 / 113 LER tunes (x/y) 38.64 / 36.57 38.52 / 36.57 HER tunes (x/y) 24.62 / 23.64 24.52 / 23.62 Beam-beam parameter (vertical +/-) 0.03 0.082 / 0.040 Beam-beam parameter (horizontal +/-) 0.03 0.109 / 0.040

Typical recent performance (May-Jun 03) Luminosity (mb-1) I+ x I- (mA-2) Specific luminosity (mb-1mA-2b-2) I+ x I- (mA-2) Near ½ integer (n+x/y ~ 0.52/0.57) e- y-size  with  e+ current e+ x-size  with  e- current Total luminosity: some tune-shift saturation, but potential for more L Specific luminosity scales (primarily) with e+ current

Interplay between e- - cloud & beam-beam issues F.-J. Decker, et. al., PAC’01, ‘03 Interplay between e- - cloud & beam-beam issues Bunch-by-bunch luminosity versus position along the whole train. Pattern: ‘by-4’ (8.4 ns spacing) with 7 additional big gaps, July 2000. The first bunches of each mini-train have a high luminosity, which drops to 40 % of its initial value at the end of the longest train. The long gaps clear the electron cloud, which slowly builds up again over along the mini-train. Solenoids had been installed in part of the straights only. Standard luminosity pattern in spring 2003 Pattern: ‘by-3’ (6.3 ns spacing) Mini-trains of 10 and 11 bunches are alternating. There is an ion gap of about 3%. In this pattern each mini-train has constant luminosity (except for bunches 1+3). Solenoids now cover most of the beam pipe in all straights and arcs.

Sensitivity to pattern details... “by-3’ pattern (6.3 ns spacing), trains with minigaps (6 Oct 03) Solenoids cover most of the beam pipe in all straights and arcs. Straight “by-3’ pattern (6.3 ns spacing), no minigaps (Oct 03) Solenoids cover most of the beam pipe in all straights and arcs.

Interplay between e- - cloud & beam-beam issues: impact so far... Bunch pattern optimization: maximize Ibunch , taking into account total-current budget (RF power, beam-heating problems) minitrain spacing (larger minigaps => better e- cloud suppression) minitrain length (shorter minitrains => less e- cloud buildup) # of minitrains (fewer minitrains => fewer ‘fragile’ bunches) need for current ramps at start of train (and/or minitrains) The severity of electron-cloud effects (for a fixed bunch pattern) has been steadily decreasing over the years Low-field (25-35 G) solenoids now cover most of the accessible beam-pipe sections. This system was upgraded this summer ( up to 3x higher field in the straights + cover remaining short sections) Vacuum-pipe scrubbing appears to have played a significant role Some e– cloud effects are no longer apparent single-beam e+ blowup at high I+ no longer observed (but what once I+  ?) In typical recent running, only 1st (few) bucket(s) in each minitrain affected by electron cloud (if any)

F.-J. Decker, et. al.,PAC’ 03 Interplay between e- - cloud & beam-beam issues: towards higher luminosities Impact of e- cloud may be more severe once higher currents force the use of a denser pattern (‘by-2’) e- cloud Pattern: ‘by-2’ (4.2 ns spacing), long trains (Feb 03) 20-25% luminosity loss after the first 5-10 bunches in each train Pattern: ‘by-2’ (4.2 ns spacing), short trains [2-4-2-4.-..] (Apr 03) The first buckets of the duplets have the same high L, while the L of the second ones drops about 50% over 320 ns. It remains constant for several hundred ns, then slowly grows to nearly 75% before it starts dropping again.

Towards higher luminosities (2) parasitic crossings do matter! e- cloud 1 parasitic crossing @ minitrain edges (instead of 2)

Beam-beam limit studies Experimental procedure Fix one beam current (typically similar to physics conditions) Vary the current of the other beam from 0 to maximum possible; at each setting, optimize luminosity on tunes Measure L/bunch, specific luminosity Lsp, individual beam sizes s-x,y in- & out-of-collision Diagnostics Fast luminosity monitor (e+e-  e+e- g) Horizontal beam sizes: synchrotron-light monitor (SLM) Vertical beam sizes: SR-light interferometer

W.K., PEP-II Performance Workshop, Dec 00 HER b-b limit @ high e- current: I- =625 mA, I+ = 100-1400 mA, 597 bunches Collisions: LEB blowup in both x & y (N.B.: y blowup instrumental?) Single-beam: LEB blowup minimal LSP drops by ~ 25-30% LEB blowup in collision >> single-beam blowup, by an amount that depends on its own current => LSP drops! HER:SLM x & y size in collision flat with I+ No indication of b-b induced HER blowup (in this data set)

LER b-b limit @ high e+ current: I+ =1070 mA, I- =200-630 mA, 597 bunches LSP drops by < 5% LER:SLM x & y size in collision >> single-beam, flat with electron current (not “naive” b-b!) HER: I- dep. of x, y SLM size in collision, consistent with single-beam behavior (flat) LER b-b limit @ low e+ current: I+ = 90 mA, I- = 20-750 mA, 829 bunches

Typical current-dependence of specific luminosity & beam sizes, Jan-Apr 03 6 Feb 03 data

Wandering in tune space... LER  Apr 03: nx = .64 ny = .56 since May 1, 2003: nx = .52 ny = .57 HER nx = .57 ny = .64 ny = .62 The values quoted here are nominal, unshifted tunes

“Before” ½ integer vs. “after”: LER beam sizes LER x-size (28 Apr) +11 % LER current (e+) [mA] + 50-60 % wrt sx ( single-beam) + 8 % LER x-size (2 May) HER current (e+) [mA] LER y-size (2 May) + 12 % HER current (e+) [mA] LER y-size (28 Apr) +7 % LER current (e+) [mA]

“Before” ½ integer vs. “after”: HER beam sizes & Luminosity HER y-size (28 Apr) HER y-size (2 May) + 39 % LER current (e+) [mA] LER current (e+) [mA] L/bunch (28 Apr) L/bunch (2 May) Product of bunch currents [(mA2/b)2] Product of bunch currents [(mA2/b)2]

Original tunes (n+x/y ~ 0.64/0.56) LSP (2 May 03) Msrd: - 25 % Predctd: -25% Spec. luminosity [1030 cm-2 s-1 mA-2/b] LER current (e+) [mA] Original tunes (n+x/y ~ 0.64/0.56) e- size ~ independent of e+ current e+ size ~  with  e+ current (mostly x: proximity to 2/3?) Near ½ integer (n+x/y ~ 0.52/0.57) e- size  with  e+ current e+ size  with  e- current Specific luminosity scales (primarily) with e+ current => HEB the ‘weaker’ beam (at that time...) Total luminosity some tune-shift saturation (prob. HEB), but potential for more luminosity! Ibunch(e-) x Ibunch(e+) [(mA/b)2] L / bunch (1030 cm- 2s-1 b-1) Luminosity/bunch (2 May 03)

Impact of horizontal-tune change on IP spot size M. Baak, Sep 03 Measurement of luminous-region size by BaBar (dimuons & Bhabha’s) 1/sL = sqrt (1/s+2 + 1/s-2) nx  0.5: Reduction of horizontal spot size from ~ 100 m to ~ 70 m. No changes observed in longitudinal spot size. before April 28th after April 28th nx ~ 0.5 simulation (high I, well-chosen v+x ) (skipped these runs)

Tune spectra in collision Old tunes nx ~ 0.5 (June 03) ? n+x n+y n-x n-y

y-tunes in collision, vs. e- bunch current (LEB trickle, 12 Jun 03) LEB x-size + 64 % wrt sx ( single-beam) HER bunch current (e-) [mA/b] LER y-tune: b-b tune shift! Dn+ ~ .004 for 0.81 < I- < 1.07 A HER bunch current (e-) [mA/b] HER y-tune: not fully compensated HER bunch current (e-) [mA/b]

Determination of beam-beam parameters from luminosity & spot-size data

“Measured” beam-beam parameters

Beam-beam performance summary J. Seeman, Aug 03 Beam-beam performance summary

Summary (I) Electron-cloud effects have been minimized by a combination of scrubbing, solenoid-suppression, and bunch-pattern optimization Luminosity (& background!) optimization relies on a delicate balance between the currents, tunes, beam-beam parameters and e-cloud effects as these parameters vary along each bunch train. Current-dependence of the spot sizes in collision (@ nx ~ 0.5) LEB (HEB) sizes depend only on e- (e+) current sy(e-) grows by 30-70% wrt the e- single-beam size 30-40% is more typical of recent (stable) running : mostly ‘spontaneous’ ? sx(e+) grows by a factors of 1.5 to > 2 wrt the e+ single-beam size 50-60% is more typical of recent (stable) running: ‘spontaneous’ or effective optimum? offline analysis of luminous region size (using dimuons & Bhabh’as) corroborates the expected reduction in IP x-spot size (  simulation?) ...but the observed variations of IP spot size are not always consistent with the measured current-dependence of individual e+ and e- beam sizes

Summary (II) Beam-beam parameter measurements some tune measurements in collision are difficult (esp. LER nx) extracting x (or limits thereon) from L and s(SL)+,- x,y promising, but... it needs better control of SLM/interferometer systematics it would greatly benefit from the ability to ‘translate’ SL sizes into IP sizes the interpretation/analysis is complicated by dynamic-b effects. 1 measurement of vertical LEB beam-beam tune-shift vs. HER current! but interpretation not straightforward.. can we gain more from such parasitic monitoring? New diagnostics being commissioned gated cameras in HER & LER on-line measurements of IP positions & spot sizes by BaBar gated tune monitor PEP-II has recently achieved, near the ½ integer, beam-beam parameters xx/ xy of about .109 / .082 (.040 / .040) in the LER (HER)

Spare slides

In contrast to ‘classical’ single-ring collider, e+x,y  e-x,y (and b+x,y  b-x,y only) => s+x,y  s-x,y interpreting luminosity in terms of x requires additional knowledge and/or assumptions on individual IP spot sizes energy-transparency conditions x+x,y = x-x,y <===> I+b E+ = I-b E- (provided b+x,y = b-x,y , e+x,y = e-x,y , n+x,y = n-x,y...) largely violated in PEP-II best performance repeatedly achieved with I + / I –~ 1.3 (more typically 1.7 – 2, not 2.9!)

Typical performance since Run 4 startup I+ x I- (mA-2) Specific luminosity (mb-1mA-2b-2) I+ x I- (mA-2) Luminosity (mb-1)

Interplay between e- - cloud & beam-beam issues (2) At high I +, e- cloud strength varies along minitrain => e+ beam size varies (long range + within train) => Luminosity varies e- beam-beam tune-shift varies (=> e- beam size may vary ??) tunes optimized on the average only => slight L loss ‘raining’ buckets (rapid loss of charge, background spikes, flip-flop) electron-cloud enhanced beam-beam blowup of the e+ beam Gated-camera measurements (2001 data, 4-by-22 pattern) R. Holtzapple, PEP-II Performance Workshop, Jan 2002

R. Holtzapple, et. al., SLAC-PUB-9238 (2001 data) Beam-beam flip-flop Near the top of a fill: Several LER bunches have reduced beam size (~30% reduction in beam size) Several bunches - typically near the head of the train - have ~ ½ L 2 states of low-luminosity bunches: they have either reduced LER current or reduced LER beam size Ib+ = low; Ib- = average Ib+, Ib- = average; sx, y+ = small. Small LER bunches + low L implies transverse blow-up of the e- beam, since bunch currents are not particularly low for these.

Beam-beam flip-flop (2) Single Bunch Transition to low-luminosity state Initially at average beam size Luminosity dropped when rings were filled The bunch experiences a sudden change in luminosity/beam size This bunch went from one unstable state (low luminosity / small x, y) to the other unstable state (short lifetime) Transition between states is fast (~0.5 sec.) Horizontal beam size oscillation accompanies the transition

Beam-beam flip-flop (c’td) Possible Explanation of Beam Size Flip-Flop Dynamics The LER bunches at the front of the train have a smaller transverse beam size (lower electron cloud density). These small (strong) LER bunches blow-up the HER bunches. The resulting tune shift for the these LER bunches is smaller than “normal”, and as a result, they have a horizontal tune located near a resonance which gives them a shorter lifetime. The LER bunches lose charge. Eventually the HER becomes strong enough to flop itself, and the LER bunch, back to “normal” size. To confirm this theory, a 2nd gated camera has been installed in the HER and is being commissioned. Start of the store (high I): L/bunch (almost) uniform throughout each train. End of the store (low I): single-bunch L dropouts are prevalent in the 1st few trains

By-5 - 1/50th, 597 bunches , b*y = 1.25, July 00 HER b-b limit (continued) Collisions: LEB blowup in both x & y (N.B.: y blowup instrumental??) Single-beam or separated: LEB blowup minimal LEB blowup in collision >> single-beam blowup, by an amount that depends on its own current => LSP drops! LEB blowup with both beams present but out of collision, is similar to single-beam blowup No indication of b-b induced HER blowup (in this data set)

LER b-b limit @ low e+ current: I+ = 90 mA, I- = 20-750 mA, 829 bunches LSP drops by ~ 10-15% LER: definite, but moderate, increase in both x,y SLM sizes with electron current (as expected) HER: I- dep. of x, y SLM size in collision, consistent with single-beam behavior

LER b-b limit @high e+ current: I+ =1070 mA, I- =200-630 mA, 597 bunches LSP drops by < 5% LER:SLM x & y size in collision >> separated, but both are flat with electron current (not “naive” b-b!) HER: I- dep. of x, y SLM size in collision, consistent with single-beam behavior (flat)

Comparison of individual beam sizes with luminous-region measurements 6 Feb 03 data

Lsp =2.9 predicted from HER/LER beam size relative variation Very little HEB current-dependence of either LER or HER spot size LEB current-dependence : for 1.1 < I+ < 1.85 A, LER x-size: + 20% y- : + 10-15 % HER x-size: + 6% (but: large spread) y- : + 10% (but: large spread; may be consistent with ~0) Lsp : - 30% (measured) : - 21% (predicted from spot size variations) Lsp Lsp =2.9 predicted from HER/LER beam size relative variation 5 Apr 03 data LER bunch current (mA/bunch)

Specific luminosity vs. beam currents LSP Msrd: - 25 % Predctd: -25% I+ I+ (mA) I- (mA) LSP I- (mA)

LEB size vs beam currents LER x-size LER y-size + 8 % LER x-size LER y-size + 12 %

“Before” ½ integer vs. “after”: LER beam sizes + 8 % LER x-size (2 May) LER x-size (28 Apr) +11 % LER y-size (2 May) + 12 % LER y-size (28 Apr) +7 %

y-tunes in collision, vs. e- bunch current (LEB trickle, 12 Jun 03) LER y-tune: b-b tune shift! Dn+ ~ .004 for 0.81 < I- < 1.07 A LER y-tune knob HER bunch current (e-) [mA/b] HER bunch current (e-) [mA/b] HER y-tune knob HER y-tune: not fully compensated HER bunch current (e-) [mA/b] HER bunch current (e-) [mA/b]

Luminous spot sizes vs. beam currents M. Baak, Sep 03 No apparent dependence of the horizontal luminous size on LER or HER currents No apparent dependence of the longitudinal luminous size on LER current. after April 28th nx ~ 0.5

Current-dependence of measured SL spot sizes

Current-dependence of estimated IP spot sizes

Comparison of fitted & measured L & LSP

Comparison of fitted & measured L & LSP

But it does not always work so well... 3 May 5-6 Jun