Presentation on theme: "Collective effects Erik Adli, University of Oslo, August 2014, Erik.Adli@fys.uio.no, v2.01."— Presentation transcript:
1Collective effectsErik Adli, University of Oslo, August 2014, v2.01
2IntroductionParticle accelerators are continuously being pushed to new parameter regimes with higher currents, higher power and higher intensityParticle accelerator performance is usually limited by multi-particle effects, including collisions and collective instabilitiesA core element of particle accelerator physics is the study of collective effects. The understanding of collective instabilities has made it possible to overcome limitations and increase performance significantlyWe here describe the two most common effects, space charge and wake fields, in some detail.Example of increase in beam intensity in the CERN PS over the years
3Intense beams: example Compressed high energy electron beam (at the “FACET” facility at SLAC) :The beam can be defined by the following parameters :Charge per bunch: Q = 3 nC (N = 2 x 1010 electrons)Emittance: enx,y = 100 umBeam energy: E = 20 GeV (v ≈ c)Focused size at interaction point :sx = 20 um, sy = 20 um, sz = 20 um (Gaussian)Other key beam characteristics deduced from the above accelerator parameters:Peak current: Ipeak = Qpeak/t = Qc / √2p sz = 19 kA (cf. IAlfven = 17 kA)Beam density: re ≈ N / 4/3p(sx sy sz) = 6 x 1017 / cm3Plasma description:Electron transverse energy: Ee = ½ mvx2 = ½ m c2 x’2 ~ 1 meVElectron (transverse) temperature: Te ~ me / kB c2 x’2 ~ 100 KDebye length (shielding length): lD = √(e0kBT/e2n0) ~ 1 nm-> lD << sx, sy, sz (collective effects dominate over collisional effects)
4Space charge – field view point Two-particle interaction :x’y'zv2xlabxyzv1sFields transforms to the lab frame as :We observe beams, fields and forces in the lab frame. The force on particle 1 is F = e(E+v1 x B). Particle 2 generates no magnetic field in its rest frame, which gives the relation (BT – v2/c2 x E) = 0. The total transverse force on particle 1 in the lab frame is thus F = e(E - v1 x (v2/c2 x E)), or for parallel velocities :For v1 = v2 = v we get relativistic space charge surpression :Fx,y = eEx,y (1±v1v2/c2)Fx,y = eEx,y (1-v2/c2) = eEx,y / g2
5EM fields from a relativistic particle Fields in lab frameFields in particle’s rest frameLorentz Tranformationsv=0Ultra-relativistic limit (v=c) :Field compression-> “Pancake field”
6Space Charge – lab frame +FEvFB+FAlternate calculations: Gauss law's. Gauss law is valid also for relativistic moving (or accelerating) charges. In the rest frame, the beam sees an electrostatic field. In the lab frame, the moving charges produce a magnetic field.We assume a uniform density beam in shape of a cylinder, with beam charge density r = Ne / pa2L :Gauss law gives :Ampere’s law gives :Combine terms Fr = e(Er - vBf) :Rest frameLab framec+c+This is the direct space charge effect. “Direct” : does not take into account the effects of conducting walls surrounding the beam.Lab frame, v = c
7Space Charge for Guassian Beams KSC ~ Ipeak / (bg)3“beam perveance”Typical lattice focusing (FODO with ~10 m between magnets): <b> ~ 10 m -> Kb = 1/100 m-2For our example 20 GeV electron beam (a few slides ago) : KSC = 5 x 10-5 m-2 << Kb-> space charge completely suppressed at high Lortentz factorFor a 20 MeV electron beam a few slides ago : KSC = 5 x 104 m-2 >> Kb
8Non-linear defocusing. Gives tune shift and tune spread in rings. SummaryK. SchindlLinear defocusing. Gives tune shift in rings. Can be compensated by stronger lattice.Non-linear defocusing. Gives tune shift and tune spread in rings.
9Beam-beam effects Fx,y = eEx,y (1+v1v2/c2) Relativistic space charge supression is only for equal charge moving at the same velocity, in the same direction (and then only holds fully in free space). These requirement are often violated. An important example is the Beam-Beam interaction :Two colliding beams see the field of each other before collision. They may be strongly attracted, and may deform.Important limitation for collider luminosity. Considered the main challenge for LHC luminosity. We will revisit the topic of beam beam effects in the linear collider lectures.For v1 and v2 of opposite sign in our previous calculation (5 slides ago) we get :Fx,y = eEx,y (1+v1v2/c2)
10Wake fields Most of the material is G. Rumolo’s slides from CAS Course on wake fields
11Wake fields (general) 2b z z z L Source, q1Witness, q2zzz2bLWhile source and witness ( qi d(s-ct) ), distant by z<0, move centered in a perfectly conducting chamber, the witness does not feel any force (g >> 1)When the source encounters a discontinuity (e.g., transition, device), it produces an electromagnetic field, which trails behind (wake field)The source loses energyThe witness feels a net force all along an effective length of the structure, L
13Wake field framework approximations When calculating wake fields in accelerator, the calculations are greatly simplified, allowing for simple descriptions of complex fields, by the following two approximations :1) Rigid beam approximation: bunch static over length wake field is calculated2) Impulse approximation: we only care about integrated force over the length wake field is calculated, not the details of the fields in time and spaceWe will define wake function (wake potential) as such integrated quantities.
14Wake fields (general) 2b z L Source, q1Witness, q2z2bLNot only geometric discontinuities cause electromagnetic fields trailing behind sources traveling at light speed.For example, a pipe with finite conductivity causes a delay in the induced currents, which also produces delayed electromagnetic fieldsNo ringing, only slow decayThe witness feels a net force all along an effective length of the structure, LIn general, also electromagnetic boundary conditions can be the origin of wake fields.This is like a Green function, we need to define which is the pulse response (in time domain), in which we are interested!
151. The longitudinal plane zdp/p01. The longitudinal plane
16Longitudinal wake function: definition Source, q1Witness, q2z2bLThis is like a Green function, we need to define which is the pulse response (in time domain), in which we are interested!
17Longitudinal wake function: properties The value of the wake function in 0, W||(0), is related to the energy lost by the source particle in the creation of the wakeW||(0)>0 since DE1<0W||(z) is discontinuous in z=0 and it vanishes for all z>0 because of the ultra-relativistic approximationW||(z)z
18The energy balance What happens to the energy lost by the source? In the global energy balance, the energy lost by the source splits intoElectromagnetic energy of the modes that remain trapped in the objectPartly dissipated on lossy walls or into purposely designed inserts or higher-order mode (HOM) absorbersPartly transferred to following particles (or the same particle over successive turns), possibly feeding into an instabilityElectromagnetic energy of modes that propagate down the beam chamber (above cut-off), which will be eventually lost on surrounding lossy materials
19The energy balance The energy loss is very important because What happens to the energy lost by the source?In the global energy balance, the energy lost by the source splits intoElectromagnetic energy of the modes that remain trapped in the objectPartly dissipated on lossy walls or into purposely designed inserts or HOM absorbersPartly transferred to following particles (or the same particle over successive turns), possibly feeding into an instability!Electromagnetic energy of modes that propagate down the beam chamber (above cut-off), which will be eventually lost on surrounding lossy materialsThe energy loss is very important becauseIt causes beam induced heating of the beam environment (damage, outgassing)It feeds into both longitudinal and transverse instabilities through the associated EM fields
20Longitudinal impedance The wake function of an accelerator component is basically its Green function in time domain (i.e., its response to a pulse excitation)Very useful for macroparticle models and simulations, because it can be used to describe the driving terms in the single particle equations of motion!We can also describe it as a transfer function in frequency domainThis is the definition of longitudinal beam coupling impedance of the element under study[W][W/s]
21Longitudinal impedance: resonator T=2p/wrwrW||Re[Z||]Im[Z||]The frequency wr is related to the oscillation of Ez, and therefore to the frequency of the mode excited in the objectThe decay time depends on how quickly the stored energy is dissipated (quantified by a quality factor Q)
22Longitudinal impedance: cavity A more complex example: a simple pill-box cavity with walls having finite conductivitySeveral modes can be excitedBelow the pipe cut-off frequency the width of the peaks is only determined by the finite conductivity of the wallsAbove, losses also come from propagation in the chamberRe[Z||]Im[Z||]
28Example: the Robinson instability To illustrate the Robinson instability we will use some simplifications:The bunch is point-like and feels an external linear force (i.e. it would execute linear synchrotron oscillations in absence of the wake forces)The bunch additionally feels the effect of a multi-turn wakedp/p0zUnperturbed: the bunch executes synchrotron oscillations at ws
29The Robinson instability To illustrate the Robinson instability we will use some simplifications:The bunch is point-like and feels an external linear force (i.e. it would execute linear synchrotron oscillations in absence of the wake forces)The bunch additionally feels the effect of a multi-turn wakedp/p0zThe perturbation also changes the oscillation amplitudeUnstable motionThe perturbation changes ws
30The Robinson instability To illustrate the Robinson instability we will use some simplifications:The bunch is point-like and feels an external linear force (i.e. it would execute linear synchrotron oscillations in absence of the wake forces)The bunch additionally feels the effect of a multi-turn wakedp/p0zThe perturbation also changes the oscillation amplitudeDamped motion
32Transverse wake function: definition Source, q1Witness, q2z2bLThis is like a Green function, we need to define which is the pulse response (in time domain), in which we are interested!In an axisymmetric structure (or simply with a top-bottom and left-right symmetry) a source particle traveling on axis cannot induce net transverse forces on a witness particle also following on axisAt the zero-th order, there is no transverse effectWe need to introduce a breaking of the symmetry to drive transverse effect, but at the first order there are two possibilities, i.e. offset the source or the witness
33Transverse dipolar wake function: definition Source, q1Witness, q2Dx1 (or Dy1)z2bLThis is like a Green function, we need to define which is the pulse response (in time domain), in which we are interested!
34Transverse dipolar wake function The value of the transverse dipolar wake functions in 0, Wx,y(0), vanishes because source and witness particles are traveling parallel and they can only – mutually – interact through space charge, which is not included in this frameworkWx,y(0--)<0 since trailing particles are deflected toward the source particle (Dx1 and Dx’2 have the same sign)Wx,y(z) has a discontinuous derivative in z=0 and it vanishes for all z>0 because of the ultra-relativistic approximationWx,y(z)z
35Discrete approximation of cavity transverse impedance Impedances in frequency domain calculated using electromagnetic solvers. Impedance can be approximated using a number of discrete modes (above 9 modes used).From CLIC decelerator design
36Dipole wake instabilities in linacs Single bunch: head drives tail resonantly -> banana shape, beam-break upLinac Beam break-up growth factor:ϒ~NWs/kbN: chargeW: dipole wakes: distancekb: betatron kTwo-bunches: one bunch drives the second resonantely
37Rings: A glance into the head-tail modes Different transverse head-tail modes correspond to different parts of the bunch oscillating with relative phase differences. E.g.Mode 0 is a rigid bunch modeMode 1 has head and tail oscillating in counter-phaseMode 2 has head and tail oscillating in phase and the bunch center in opposition
38Calculation of coherent modes seen at a wide-band pick-up (BPM) The patterns of the head-tail modes (m,l) depend on chromaticityhQ’=0Q’≠0 m=1 and l=0 m=1 and l=1 m=1 and l=2
39Coherent modes measured at a wide-band pick-up (BPM) m=1 and l=01Instabilities are a good diagnostics tool to identify and quantify the main impedance sources in a machine.
40Wake field (impedances) in accelerator ring design The full ring is usually modeled with a so called total impedance made of three main components:Resistive wall impedanceSeveral narrow-band resonators at lower frequencies than the pipe cutoff frequency c/b (b beam pipe radius)One broad band resonator at wr~c/b modeling the rest of the ring (pipe discontinuities, tapers, other non-resonant structures like pick-ups, kickers bellows, etc.)The total impedance is allocated to the single ring elements by means of off-line calculation prior to construction/installationTotal impedance designed such that the nominal intensity is stable
41We will talk more about wake fields in the lecture about linear colliders (tomorrow). For more details on wake fields, derived from first physical principles see the excellent book “Physics of Collective Beam Instabilities in High Energy Accelerators”, A. W. Chao (freely available; see course web pages).
43Overview of multi-particle effects Adapted from G. Rumolo’s slides fromUSPAS Course on collective effects
44General definition of multi-particle processes in an accelerator or storage ring Class of phenomena in which the evolution of the particle beam cannot be studied as if the beam was a single particle (as is done in beam optics), but depends on the combination of external fields and interaction between particles. Particles can interact between them throughSelf generated fields:Direct space charge fieldsElectromagnetic interaction of the beam with the surrounding environment through the beam‘s own images and the wake fields (impedances)Interaction with the beam‘s own synchrotron radiationLong- and short-range Coulomb collisions, associated to intra-beam scattering and Touschek effect, respectivelyInteraction of electron beams with trapped ions, proton/positron/ion beams with electron clouds, beam-beam in a collider ring, electron cooling for ionsMulti-particle processes are detrimental for the beam (degradation and loss, see next slides)
45Direct space charge forces (more details later in this lecture) xTransverse space chargeForce decays like 1/g2It is always repulsiveyLongitudinal space chargeForce decays like 1/g2it can be attractive above transitiondp/p0z
46Wake fields, impedances (more details later in this lecture) zW0(z)LseqModel:A rigid beam with charge q going through a device of length L leaves behind an oscillating field and a probe charge e at distance z feels a force as a result. The integral of this force over the device defines the wake field and its Fourier transform is called the impedance of the device of length L.
47Electron cloud Principle of electron multipacting: Example of LHC Electron multiplication is made possible by:Electron generation due to photoemission, but also residual gas ionizationElectron acceleration in the field of the passing bunchesSecondary emission with efficiency larger than one, when the electrons hit the inner pipe walls with high enough energy
48Several names to describe these effects... ‚Multi-particle‘ is the most generic attribute. ‚High-current‘, ‚high-intensity‘, ‚high brightness‘ are also used because these effects are important when the beam has a high density in phase space (many particles in little volume)Other labels are also used to refer to different subclassesCollective effects (coherent):The beam resonantly responds to a self-induced electromagnetic excitationAre fast and visible in the beam centroid motion (tune shift, instability)Collective effects (incoherent):Excitation moves with the beam, spreads the frequencies of particle motion.Lead to particle diffusion in phase space and slow emittance growthCollisional effects (incoherent):Isolated two-particle encounters have a global effect on the beam dynamics (diffusion and emittance growth, lifetime)Two-stream phenomena (coherent or incoherent):Two component plasmas needed (beam-beam, pbeam-ecloud, ebeam-ions) and the beam reacts to an excitation caused by another „beam“The performance of an accelerator is usually limited by a multi-particle effect. When the beam current in a machine is pushed above a certain limit (intensity threshold), intolerable losses or beam quality degradation appear due to these phenomena
49Example of coherent and incoherent effects Coherent: coherent synchrotron radiation (CSR)Previous calculations of synchrotron radiation in this course assumed each particle radiates independently Prad NIf particles are close with respect to the radiation wavelengths, the particles will radiate coherently (as one macro particle), Prad N2Significant effect for short bunch lengths, low energy beams, with large number of particles per bunchMay lead to instabilitiesIncoherent: intra-beam scatteringParticles within a bunch can collide with each other as they perform betatron and synchrotron oscillations. The collisions lead to a redistribution of the momenta within the bunch, and hence to a change in the emittances.This is effect is called IBS, intra-beam scattering. If there is a large transfer of momentum into the longidunal plane. It is called Touchek scattering.May lead to emittance growth and reduced beam life time.
50More examples Bunch in the CERN SPS synchrotron BNL-RHIC, Au-Au operation, Run-4 (2004)16hCoherent effects:When the bunch current exceeds a certain limit (current threshold), the centroid of the beam, e.g. as seen by a BPM, exhibits an exponential growth (instability) and the beam is lost within few milliseconds
51Single bunch versus multi-bunch effets Single or multi-bunch behavior depends on the range of action of the wake fieldsSingle bunch effects are usually caused by short range wake fields (broad-band impedances)Multi bunch or multi-turn effects are usually associated to long range wake fields (narrow-band impedances)Wake decays over a bunch length Single bunch collective interactionWWake decays over many bunches Coupled bunch collective interactionWW(z-z‘)l(z‘)dz‘W(kd)N(kd)zsds