R.W. Aßmann, CERN SL-AP Snowmass 2001 July 7 th, PDF Free Download

R.W. Aßmann, CERN SL-AP Snowmass 2001 July 7 th, 2001

(almost a Higgs) November 2 nd, 2000, 7am

We can do this precision e+ephysics at the high-energy frontier! Push beyond 209 GeV at LEP (for the Higgs, ) Problem: This talk: Compare and recommend Sorry, folks: This is up to you

1. Introduction The LC design studies Simulation codes Beam dynamics beam energy 2. Single particle dynamics Optics design Example lattices Dispersion and energy spread 3. Multi-particle dynamics Transverse wakefield BNS damping SLC as the reference Emittance/wakefield optimization Predicted LC performance Required component performance 4. Transverse beam stability Predictions for LC designs LEP experience Stability studies (vibration damping) 5. Multi-bunch effects Knowledge of long-range wakefield Multi-bunch emittance growth Effects production errors/damage 6. Conclusion

This talk: It is not: Discuss some design choices Explain some trade-offs Illustrate some beam dynamics topics Use of simple example lattices, estimates Include some detailed, published LC studies A detailed review of all issues (would be a book) Common simulations for the three designs Any judgment on the different approaches Greg Loews International LC Study Group Talk based on my direct experience with SLC, CLIC, and NLC. But, more importantly: The work of many, many colleagues! Acknowledgements to all of them, especially: TESLA: R. Brinkmann and colleagues CLIC: D. Schulte, G. Guignard and colleagues NLC: T. Raubenheimer, P. Tenenbaum, A. Seryi and colleagues SLC: The old gang

The main linacs: The heart of linear colliders. Their crucial mission for the success of linear colliders: Provide beam energy of interest for physics Issues: High efficiency High accelerating gradients Reliable RF system Minimize cost per MV Transport high current and small emittance beams for high luminosity Issues: Transverse emittance dilution Energy sharpness, stability Transverse position stability Tails and beam losses Beam dynamics

TESLA SLC NLC/JLC CLIC (stage II) C.M. energy GeV 500 92 1000 3000 Luminosity 10 34 cm -2 s -1 3.00 0.02 3.40 10.0 Frequency GHz 1.3 2.856 11.424 30.0 Iris radius mm 35.00 10.00 5.00 2.05 Gradient MV/m 23.4 18.0 65.0 150.0 Bunch popul 10 10 2.0 3.5 0.75 0.4 # bunches 2820 1 190 154 f rep Hz 5 120 120 100 ge x (inj) mm-mrad 9.000 30.000 3.000 0.500 ge y (inj) mm-mrad 0.020 5.000 0.020 0.005 Basic scaling: Luminosity ~ Energy 2

Linac performance (static) characterized with: Beam current Transverse emittances Energy spread γε γε + γε + final inj disp γε wf Tolerance: γε disp + γε wf γε Wakefields ~ f 3 Beam parameters Tolerances on: Straightness of trajectory Centering in RF structures inj Required luminosity 5.000 mm-mrad (SLC) 0.020 mm-mrad (NLC/JLC) 0.020 mm-mrad (TESLA) 0.005 mm-mrad (CLIC) High gradient (high frequency) allows high beam energies, for the price of more stringent tolerances on alignment Tolerances scale with 1/f J.P. Delahaye et al For an optimized design

Test accelerators cannot test linac performance. Predict linac performance based on simulation codes Programs used in the context of linear collider studies: LIAR MAFIA TRANSPORT SAD MAD GUINEAPIG MERLIN TraFIC 4 MUSTAFA PLACET Q URMEL DIMAD GDFIDL LEGO WAKE TRACK FFADA PARMELA FLUKA GEANT CAIN OMEGA3P TAU3P HFSS PHI3P + many others (some nameless heroes) Computational activity for linear colliders is: manifold and redundant Especially: LIAR written for and tested against SLC linac Incorporates lot of experience from SLC Used for cross-checks of other programs (e.g. PLACET)

Beam energy is a crucial parameter for linear collider design. Target energy or energy range set by particle physics! Choice of technology determines the energy range: TESLA: JLC/NLC: CLIC: up to 800 GeV up to 1000 GeV up to 3000 GeV Linear collider design is not energy-independent Comparison of designs for different energies can be misleading! Compare designs for the same beam energy (G. Loew et al).

1) Provide focusing for beam transport (no real challenge, FODO) 2) Minimize effects from kicks (wakefield) on the beam Offset y at s 2 due to kick at s 1 y( s2) = R12( s1, s2) θ ( s1) R = 12 β 1 β 2 E E 1 2 sin ( ψ ) Keep design beam size at s 2 constant (b 2 =const) For a given kick: y( s β 2 ) 1 Stronger focusing helps if kick amplitude does not depend on linac optics (wakefield kicks, injection errors) Weaker focusing helps for dispersive quadrupole kicks. (fewer quadrupole kicks)

TDR Dispersion-dominated optics design Wakefield-dominated optics design TESLA (.5 TeV) CLIC (3 TeV) NLC (1 TeV) Yellow report ZDR

y cell l b l b (250 GeV example) 72º 22-200 m ~ 40 m CLIC 80º 36-180 m ~ 80 m JLC/NLC 60º 390-576 m ~ 360 m TESLA Note: CLIC below 45 m for 250 GeV TESLA more sensitive to ground motion waves with Long wavelength Low frequency Larger amplitude

Not fully realistic 90 degree FODO lattice Cell length β y (max/min) 10.0 m 17 / 2.9 m 20.0 m 34 / 5.9 m 40.0 m 68 / 11.8 m 80.0 m 136 / 23.5 m CLIC NLC TESLA Injection energy: 10 GeV Acceleration length: 96 % of cell length Assume: 250 GeV Linac length: 1890 m (135.0 MV/m) ~189 cells a 10 m 4290 m ( 59.0 MV/m) ~215 cells a 20 m 10800 m ( 23.4 MV/m) ~135 cells a 80 m

Trajectory in the CLIC, TESLA, JLC/NLC control rooms Assume: 100 mm QD offset at start of linac, low current (no WF s) CLIC type JLC/NLC type TESLA type (all 250 GeV)

Example case CLIC type no acceleration no wakefields QD misaligned by 50 mm Filamentation Emittance saturates due to filamentation for large energy spread (chromatic phase mixing)

Correlated, fractional energy spread (CLIC/NLC: BNS) Linac Linac exit RF curvature contribution TESLA: < 0.06 % 0.06% 0.004 % NLC: < 0.80 % 0.25 % 0.042 % CLIC: < 0.55 % 0.35 % 0.023 % QD offset by 0.3 mm -> Y oscillation with amplitude ~ 1mm Example: σ E /E ε disp ε inj CLIC type 0.4% 0.0010 mm-mrad 0.004 mm-mrad NLC type 0.6% 0.0006 mm-mrad 0.030 mm-mrad TESLA type 0.05% 0.0000 mm-mrad 0.020 mm-mrad Dispersion no problem, if trajectory controlled on 1 mm level! (pessimistic assumptions: No wakefields would mean no BNS)

Interaction: Accelerated charge RF structures (small irises) (except TESLA) θ wf = W σ t en L 2E e struc ( z ) y1 0 Wakefield effect depends on: Intra-bunch and inter-bunch wakefields Offsets in rf structures (imperfections) Longitudinal distribution Charge Energy Optics RF phases Calculate effect with programs: Multi-particle beam dynamics Multiple interacting imperfections Chromatic, dispersive + wakefield errors Single-bunch and multi-bunch R. Assmann et al

Choice of technology determines radius of structure iris a: High frequency small a Low frequency large a Stronger wakefields (beam induced electro-magnetic fields) with smaller iris radius! Beam is closer to metallic walls

Bunch length: Transverse wakefield (at 1 σ z ): TESLA SLC NLC CLIC 300 mm 1100 mm 110 mm 30 mm TESLA 22 V/pC/m 2 SLC 1990 V/pC/m 2 NLC 11460 V/pC/m 2 CLIC 81000 V/pC/m 2 Injection energy: TESLA 5.0 GeV SLC 1.2 GeV NLC 8.0 GeV CLIC 9.0 GeV + Bunch intensity: TESLA 2.00 10 10 SLC 4.00 10 10 NLC 0.75 10 10 CLIC 0.40 10 10 BEAM

Represent bunch by two slices: separated by σ z each half charge θ wf = W σ t en L ( ) e struc z y1 2E 0 For 100 µm structure offset and 1 m structure length: TESLA: 0.7 nrad NLC: 86.1 nrad CLIC: 288.4 nrad SLC: 535.8 nrad Not done: Normalize to emittance Input structure length Wakefield kick for CLIC at 1.5 TeV almost down to TESLA at 5 GeV (1/3 larger)

Introduce correlated energy spread (RF phase) so that head and tail move together (same phase advance). No beam-breakup. NLC CLIC M. Woodley et al, PAC01 CLIC yellow report No BNS damping required for TESLA RF phase CLIC: ~ 6 degree

Single bunch emittance growth (SLC 1996/1997): R. Assmann, PAC97 γε 28 = κ γε initial + γε wf Problems due to poor emittance stability (drift towards larger emittances) multiplicative additive Simulation: κ = 1.06 γε wf = 2 Reasonable agreement with data from the SLC!

Learnt how to go from Left case (start of tuning) to Right case (end of tuning) Methods: - 1-to-1 steering (steer flat) - RF phasing (energy profile) - Dispersion-free steering - Emittance bump tuning

Low repetition rates (5-120 Hz) Small beam sizes (shown was smaller area than LEP) No equilibrium state, no damping after damping rings Every pulse is different The beam is living Asymmetric beam distributions, tails due to wakefields Intense tuning needed to control beam sizes and stability (much better for super-conducting linacs) Wakefield effects can be corrected very efficiently (took a while for SLC to learn how)

Vertical emittance growth in the linac (normalized): ε y SLC 2 mm-mrad TESLA 0.01 mm-mrad JLC/NLC 0.015 mm-mrad CLIC 0.005 mm-mrad Measured Lower wakefields (SC rf) Parameter optimization (charge, bunch length, ) Minimize wakefields and dispersion! Stringent requirements (JLC/NLC, CLIC) on: - alignment (beam-based) - steering (magnet, rf structure movements) - feedbacks - beam instrumentation Innovative methods have been developed

TESLA NLC CLIC 1-to-1 1-to-1 1-to-1 Shunt method Shunt method Ballistic correction Dispersion-free steering Moving procedure (local) Multi-step lining-up Dispersion-free steering Emittance bumps Emittance bumps Shunt method ( k-modulation ) (FFTB, HERA, LEP, ) Dispersion-free steering (SLC, LEP, ) Ballistic alignment Multi-step lining-up Emittance bumps (SLC) Vary quad K, measure change in trajectory, fit BPM to quad misalignment Minimize orbit, dispersion information together Minimum orbit/dispersion -> Straight trajectory Hidden bump in orbit shows up as dispersion oscillation. Switch off quadrupoles. Beam defines straight axis. Measure BPM offsets to axis. Similar to dispersion-free steering. (Modify quad strengths) Introduce wakefield kicks that compensate wakefield kicks from imperfections.

ORBIT DISPERSION CORR. KICKS DFS: Simultaneously optimize orbit, disp., corr. Suggested for NLC (Raubenheimer et al). Developed for SLC (Assmann et al)! It even works for storage rings (it should work for future LC!)

CLIC (D Amico,Guignard, Schulte) TESLA (CDR) Ballistic correction Multi-step + bumps Dispersion-free steering

Requirements and predicted LC performance TESLA JLC/NLC CLIC Quadrupole offset *# 300 mm 50.0 mm 50.0 mm Quadrupole roll - 200.0 mrad ~100 mrad BPM resolution 10 mm 0.3 mm 0.1 mm BPM-quad offset 100 mm 2.0 mm n/a BPM offset (axis) n/a n/a 10.0 mm Structure offset # 500 mm 30.0 mm 10.0 mm RF BPM offset/resol n/a 5.0 mm 5.0 mm Mover resolution n/a 50.0 nm 0.5 mm Emittance growth (a) One-to-one 1000 % ~ 1000 % 2700 % (b) All methods 3 % 40 % 15 % * Initial offset, before beam-based alignment # Achievable performance depends on instrumentation, environment, accessibility, (e.g. worse inside of cryostats)

BPM to QUAD offset (NLC) Mover setp size (NLC) Tenenbaum/Raubenheimer, LINAC2000 P. Tenenbaum, PAC99 If specifications are not met, then performance deterioration!

Tolerances for the RF system: Energy Energy spread 50 cases simulated Phase jitter Emittance growth Emittance growth Tolerances JLC/NLC: T. Higo, K. Kubo and K. Yokoya, PAC99 Energy: 0.1 % RF amplitude: 2% RF phase: 3 degree

1) Beam position stability: Quadrupole vibration/drift drives coherent betatron oscillations. Feedback inefficient below ~ 0.04 * f rep Tolerance set by σ exit = (βε exit ) 1/2 2) Emittance stability: Betatron oscillation drives transverse emittance growth. TESLA JLC/NLC CLIC Quadrupole jitter 50-100 nm 10 nm 1.3 nm Offset jitter [σ y ] 0.5-1.0 0.3 0.25 A [10-7 µm 2 /s/m] 40 5 5 Orbit drift 1 σ y in 30s - - Emittance growth - 29% in 30 min 11% in 1min Correction w/o fdbk 15 min Correction w fdbk 10 h 3 days Note: 1) Take drifts as rough estimates! Different hardware, feedback constellation, tuning methods! No consistent operational procedure simulated 2) Different A reflects different geological conditions (sandy rock; site-specific)

Luminosity decay due to vertical orbit drifts: L 30 2 1 0.3 10 cm s per minute ε 0.002 nm per minute Orbit correction ε ε 1.5% / min De/e ~ 1.5 % / min for best performance Luminosity stabilized with the vertical orbit feedback ( autopilot ) every 7-8 minutes (3% effect). Orbit stabilization: ~ 20 mm level. Both visible from experiments and beam lifetime BCT (faster)! No reason to be afraid of fast orbit stabilization

CLIC tolerance on vertical quadrupole vibration: 1.3 nm (vibration above 4 Hz) Measurements in the LEP tunnel Man passing by magnet 20 nm 2 µm 0.1 nm V. Shiltsev 1994 A. Seryi et al, CERN 1993 Ground stability in LEP tunnel much below required 1.3 nm (quiet) However, easily surpassed from human induced noise (running equipment)

CLIC Magnet Stability Study M. Aleksa, R. Assmann, W. Coosemans, G. Guignard, N. Leros, M. Mayoud, S. Redaelli, F. Ruggiero, S. Russenschuck, D. Schulte, A. Verdier, I. Wilson, F. Zimmermann - CERN: SL, PS, EST, LHC divisions involved - CERN test stand on main site (surface, close to road, accelerators, equipment). - Collaboration with SLAC/NLC. Contacts with DESY and FNAL. Since January 2001 fully approved www.cern.ch/clic-stability Goal: The goal of the proposed study is to show that the present design parameters of CLIC are feasible in a real accelerator environment, using and further developing latest cutting-edge stabilization technology and time-dependent simulation programs. Active and passive stabilization technology subject of intense industrial research and development. Applications: Chip lithography, electron-transmission microscopy, NMR devices, solid-sate physics, satellites, airplanes, gravitational wave detectors, lasers, E.g. If TEM can achieve 0.05 nm resolution why can t we use this? SLAC:Strong effort for final doublet stabilization (Seryi, Frisch, ) Number of recent papers by A. Seryi (see also T working group)

26th Advanced ICFA Beam Dynamics Workshop on Nanometre-Size Colliding Particle Beams CERN, September 2002 Amongst other topics, address many stability issues! Where is the limit? Hope for input from colleagues in our field and from other fields. Can we give a limit? Please contact R. Assmann or F. Zimmermann if you have ideas, input, special requests!

Input: Calculations + ASSET tests. Optimize design... CLIC structure: (scaled to 15 GHz) Measured (black) and calculated (red) transverse wakefield versus time [ns] Wakefield (V/pC/mm) I. Wilson et al EPAC2000 Time [ns] Very good agreement, except unexpected 7.6 GHz component HFSS: vacuum chamber to beam pipe transition, not the structure itself Time [ns]

TAU3P calculation for 10 cells compared with measurements: (RDDS accelerator structure, NLC) Dipole Mode Spectrum Dipole Mode Frequency (GHz) Amplitude Measurement Tau3P 16.868 (350) 16.89 (400) 16.440 16.46 16.280 16.30 16.176 16.18 16.098 16.10 16.034 16.04 Frequency [Hz] 1.58 1.68 10 10 Very good accuracy! Cho Ng, Brian Mc Candless, ICAP2000

HOM damping requirements for the TESLA superstructures Worst result (10 cases) Cavity misalignment 500 mm 21.7 MV/m gradient 27 modes: Q = 2e5 4 modes: Q = 1e5 N. Baboi et al EPAC2000 Measured HOM OK for TESLA Multi-bunch offsets at the end of the TESLA linac Calculated multi-bunch emittance growth along the TESLA linac e y = 20 E-09 m rad

Envelope of wakefield: (a) ideal R. Jones et al, EPAC2000 Wake Function [V/pC/mm/m] s [m] s [m] (b) 2 MHz rms error (c) 5 MHz rms error Emittance growth: BPM position [km] 4% 600% s [m] BPM position [km]

Linac beam dynamics is a very rich field. It depends strongly on choice of RF technology and beam energy. Predictions are based on detailed simulations. Simulation codes are connected closely to SLC experience. Good reproduction of SLC data. Solutions are published for LC proposals up to 3 TeV. Relevant beam dynamics is understood. No reasonable doubts on simulation results. But: What is reasonable input? Work ongoing to establish most realistic input data (test accelerators, magnet stability study, ) (fully base design on measured performance of components)