Electron Sources: Space Charge - Cornell Universityhoff/LECTURES/08S... · March 8, 2008 I.V....
Transcript of Electron Sources: Space Charge - Cornell Universityhoff/LECTURES/08S... · March 8, 2008 I.V....
Ivan Bazarov
Electron Sources: Space Charge
March 8, 2008 I.V. Bazarov, Electron Sources: Space Charge 2
Contents
• Space charge treatment– Debye length– Plasma frequency– Beam temperature
• Stationary distributions• Emittance growth due to excess free energy• Beam envelope equation• Concept of equivalent beams• Emittance compensation• Computational aspects
March 8, 2008 I.V. Bazarov, Electron Sources: Space Charge 3
Space charge
In a typical bunched beam from a gun, both charge and current density are functions of transverse & longitudinal coordinates. This makes space charge dominated behavior highly nonlinear.
For beam envelope equation we will assume that ρ and Jz are independent of transverse coordinate and that the beam is not bunched (aspect ration << 1).
March 8, 2008 I.V. Bazarov, Electron Sources: Space Charge 4
“Collisional” vs. “smooth” self-fields
It would seem that the easiest approach is to calculate Lorentz force of all electrons directly (this would encompass practically all the behavior of the beam). This is not feasible because the number of evaluations for each time step is ~ N2, with N ~ 1010. Taking the fastest supercomputer with 100 TFLOPS, one estimates ~ year(s) per time step (and one may need something like ~104 steps).
Instead, people use macroparticles in computer simulations with the same e/m ratio
When N → ∞ forces are smooth; when N → 1 grainy collisionalforces dominate. Envelope equation assumes the first scenario.
How to determine quantitatively “collisional” vs. “smooth”behavior of the space charge in the beam?
March 8, 2008 I.V. Bazarov, Electron Sources: Space Charge 5
Debye length and plasma frequency
Three characteristic lengths in the bunch:
bunch dimension; interparticle distance; Debye length
Interactions due to Coulomb forces are long-range; Debye length is a measure of how ‘long’ (screen-off distance of a local perturbation in charge).
apl Dλ
p
vD
x
ωσ
λ ≡
for nonrelativistic case: m
Tk
m
ne Bvp x
== σε
ω ,0
2
ne
TkBD 2
0ελ =
for relativistic case: γσ
γεω
m
Tk
m
ne Bvp x
== 3
0
2
ne
TkBD 2
02εγλ =
March 8, 2008 I.V. Bazarov, Electron Sources: Space Charge 6
Debye length: beam dynamics scenarios
aD >>λ YES
NO
collective forces are important
pD l>>λ YES “smooth” force; Liouville’s theorem canbe defined in 6-D phase space; if forces are
linear rms emittance is also conserved
NOfields of individual particles become important;
one ends up having 6N-D phase space to deal within the worst case; beam tends to develop ‘structure’
constcmc
xx vxpxn ==
σγσσσε ~
constTkBx =2γσ
single-particle behavior dominates; true when either energy or beam temperature is
large (emittance-dominated)
March 8, 2008 I.V. Bazarov, Electron Sources: Space Charge 7
Equilibrium distribution
0)(3
1
=
∂∂×++
∂∂
∑=i i
iii p
fBvEeq
q
f rrr&
0=×∇ Err
∫=×∇ pdfveB 30
rrrµ 0=⋅∇ B
rr
∫=⋅∇ pdfe
E 3
0εrr
Similar to thermodynamics and plasma physics, there may exist equilibrium particle distributions (i.e. those that remain stationary). Vlasov theory allows one to find such distributions (assumes collisions are negligible, but they are the ones responsible to drive the distribution to the equilibrium!). Without the derivation, Vlasov-Maxwell equations for equilibrium distributions f(qi,pi ,t) (i.e. no explicit time dependence, ∂/∂t = 0):
March 8, 2008 I.V. Bazarov, Electron Sources: Space Charge 8
Equilibrium distribution (contd.)
−=
⊥Tk
renrn
B
)(exp)0()(
φ
)(1
)()(2
rrr selfext φγ
φφ +=
2/)( 220rmre ext ωγφ =
∫ ∫
−=
r r
self rnrr
erdrdr
0
ˆ
0 0
)ˆ(ˆˆ)(ε
ϕ
In particular, in aconstant focusing channel, equilibrium transverse density obeys a well-known Boltzmann relation
March 8, 2008 I.V. Bazarov, Electron Sources: Space Charge 9
Analytically, two extreme cases:
)0/( 0 →→ aTk DB λ
)1/( 0 ≥→ aDself λϕ
>≤=
=ar
arconstnrn
for ,0
for ,)( 0
−=
⊥Tk
rmnrn
B2exp)(
220
0
ωγ
uniform
Gaussian
Equilibrium distribution (contd.)
March 8, 2008 I.V. Bazarov, Electron Sources: Space Charge 10
Beam heating due to free energy
– Stationary (equilibrium) beam distributions (states) in a linear uniform focusing channel best described by a transverse Maxwell-Boltzmann distribution
– Nonstationary beams (nonequilibrium) such as mismatched initial conditions for the beam have a higher total energy per particle than the corresponding stationary beam (free energy)
– Collisional forces eventually will lead to a new stationary state at the higher energy per particle, result in increased beam temperature or emittance
March 8, 2008 I.V. Bazarov, Electron Sources: Space Charge 11
Beam temperature
focusing potential
beam self-potential
I headtail
zxbeam
eQ
R
FstdkT
σεσ
0
2
250
1~ ≈
∂∂
Equilibrium kTbeam ~ 100 eV!!
Diffraction limited beam at 1Å → εx
= 8pm at 5GeV → εnx = 0.08 µm
10 MV/m gradient → σlaser= 0.3 mm
Transverse temp. needed kT = 25 meV
March 8, 2008 I.V. Bazarov, Electron Sources: Space Charge 12
Space charge basics
near cathode after the gun relativistic energies
rEc
Bβ
θ =
⊥ force scales as 1- β2 = γ2
March 8, 2008 I.V. Bazarov, Electron Sources: Space Charge 13
Perveance and characteristic current
33320
2222
0 2 using ,
1
2 γβπεγβ
γβπεγ
mca
eIrrrr
ca
eIrrm =′′→′′== &&&&
rr p
2
2ω=&& r
a
Kr
2=′′ arKrr mmm ==′′ for
230
2
amc
eIp βγπε
ω =33
0
2
γβI
IK = kA17
30
14 230
0 =≈=e
mc
e
mcI
πε
Let’s derive beam envelope equation (i.e. we assume that self-forces are smooth). We have almost derived the equation already (previous lecture’s paraxial ray equation). Two terms are missing – due to space charge and emittance ‘pressure’.
Uniform laminar beam in the absence of external forces:
March 8, 2008 I.V. Bazarov, Electron Sources: Space Charge 14
Perveance and generalized perveance
=
2/1
0
2/3 )/2(4
1
meV
IK
πε
Nonrelativistic limit: , with3
0
3
0 2
2
mv
eI
I
IK
πεβ== 2/1)/2( meVv =
generalized perveance
perveance
March 8, 2008 I.V. Bazarov, Electron Sources: Space Charge 15
Emittance ‘pressure’ term
πε= area
x
x′
εΒ
εΓΒΑ−=slope
ΑΓ−=slope 0 z
ΓΑ−Α−Β
=
′′′′
==Σ εxxxx
xxxxxxT rr
1det[...]=
In a drift 0 → z: and
For σx: or
constxx =′→′zxxx ′+→
22 zz Γ+Α−Β→Β
const=Γ→ΓzΓ−Α→Α
,ΒΑ−=′ εσ x ΒΒ
=′′ εσ x0
3
2
=−′′x
x σεσ
March 8, 2008 I.V. Bazarov, Electron Sources: Space Charge 16
Beam envelope equation
011
2
1
22
1 22
223330
2
222=
+
−−
+′′
+′′+′′ nmc
P
I
I
mc
eB εγβσγβσ
γγγβ
σγβ
γσσ θ
adiabaticRF focusing
of cavity edge
solenoidspace charge
angular momentum ‘increases’ emittance
Β>>>> nn
I
I
I
I εγβσ
εβγ 22
02
2
0 2or ,
2
space charge dominated emittance dominated
Β<<<< nn
I
I
I
I εγβσ
εβγ 22
02
2
0 2or ,
2
From paraxial ray equation with the additional terms, one obtains
March 8, 2008 I.V. Bazarov, Electron Sources: Space Charge 17
Example of beam dynamics: 80 pCcharge
March 8, 2008 I.V. Bazarov, Electron Sources: Space Charge 18
Equivalent Beams Concept
– The actual beams have complicated transverse profiles and phase space distribution
– How useful is beam envelope equation derived for uniform cylindrical beams?
– Concept of Equivalent Beams due to Lapostolle and Sachere (1971): “two beams having the same current and kinetic energy are equivalent in an approximate sense if the second moments of the distribution are the same.”
– E.g. rms sizes and emittances are the same for the two beams when compared at identical positions
March 8, 2008 I.V. Bazarov, Electron Sources: Space Charge 19
Emittance evolution due to space charge
We have seen that beam will evolvefrom space-charge dominated toemittance dominated regimes as itis being accelerated. At low energies,various longitudinal ‘slices’ of the bunch experience different forces due to varying current → ‘bow-tie’ phase space is common.
Important realization is that much of these space charge emittance growth may be reversible through appropriate focusing (and drifts), a so-called emittance compensation. Obviously, this emittance compensation should take place before (or rather as) the beam becomes ultrarelativistic (and emittance-dominated).
March 8, 2008 I.V. Bazarov, Electron Sources: Space Charge 20
Emittance growth in RF guns (without comp.)
RFkmc
eE20
2=α
)(sin
11
4
2
00
223
AI
I
k
k
xpeak
RF
scx
zxRFrfx
µφα
πε
σσαε
=
=
53
1, +
≈Agaussxµ
uniform)for mrad-mm 3.1( mrad-mm 0.4
mrad-mm 1.1
=
=scx
rfx
εε
↓α
↑α
λ/2λ/4
Kim analyzed emittance growth due to space charge and RF (NIM A 275 (1989) 201-218). His analysis applies to beam in RF guns. He has found:
e.g. 100 MV/m RF gun (λ = 10.5 cm): α = 1.64, phase (to minimize rf emittance) φ0 = 71° (φ → 90°), laser width and length σx = 3.5 mm, σz = 0.6 mm
March 8, 2008 I.V. Bazarov, Electron Sources: Space Charge 21
Emittance Compensation
8 nC
Carlsten noted in simulations that emittance can be brought down and gave a simple explanation for the effect (NIM A 285 (1989) 313-319).
March 8, 2008 I.V. Bazarov, Electron Sources: Space Charge 22
Emittance compensation from envelope eqn.
02
)(223
2
330
2=−−+
′′+′′γβσ
εγβσ
ζσγβ
γσσ nr I
IK
02
)(223
2
330
2=−−+
′′+′′γβσ
εγβσ
ζσγβ
γσσ nr I
IK
3302
)(
γβζσ
IK
I
req =
ζ
Serafini and Rosenzweig used envelope equation to explain emittance compensation (Phys. Rev. E 55 (1997) 7565).
includes solenoid and RF ζ tags long. slice in the bunch
For space charge dominated case in absence of acceleration
Brillouin flow σ′′ → 0:
March 8, 2008 I.V. Bazarov, Electron Sources: Space Charge 23
Emittance compensation
0...2
2
=
−
+−+′′
eqeqrK
σδσ
σδσδσσδ
( )( )
−=′
+=
zKKz
zKz
rr
req
2sin)(2),(
2cos)()(),(
ζδσζσ
ζδσζσζσ
)2sin(22
10
222 zKKrrrr reqr σσε ≈′−′=
0
Small oscillations near equilibrium:
Important: frequency of small oscillations around equilibrium does not depend on ζ. E.g. for beam with σ′(0,ζ) = 0 and σ(0,ζ) = σeq(ζ) + δσ(ζ) = σ0:
March 8, 2008 I.V. Bazarov, Electron Sources: Space Charge 24
Emittance compensation
• slices oscillate in phase space around different equilibria but with the same frequency
• ‘projected’ emittance reversible oscillations when δσ/σeq << 1, unharmonicityshows up when δσ is not small
• ignores the fact that beam aspect ratio can be >> 1 (e.g. at the cathode)
x
x′
March 8, 2008 I.V. Bazarov, Electron Sources: Space Charge 25
Emittance compensation
τττ ye
dy
d −
=Ω+ 22
2
focusing RF & solenoid represents ,)2/)(/( ,ln 000
Ω′≡≡ IIy ζγγστγγ
20
2
2/
41
1
2
)(2 ,
41
2
Ω+′=
Ω+=
−
I
Ieeq
y
eq γζ
γστ
2
γσσγ ′
−=′
eq
eq phase space angle is independent of slice ζ
Including acceleration term and transforming from (σ,z) → (τ,y) in the limit γ >> 1
Particular solution that represents generalized Brillouin flow or ‘invariant envelope’:
Matching beam to ‘invariant envelope’ can lead to ‘damping’ of projected rms emittance.
March 8, 2008 I.V. Bazarov, Electron Sources: Space Charge 26
Numerical approach
Ferarrio (INFN)
Using the beam envelope equation for individual slices leads to a recipe for emittance compensation, which works for simple cases (e.g. matching beam into long focusing channel / linac in the injector). For other more complicated scenarios one should solve the equations numerically (example: code HOMDYN).
Finally, particle trackingis indispensable foranalysis and designof the injector wherethe assumptions madeare invalid or theory istoo complicated to beuseful.
March 8, 2008 I.V. Bazarov, Electron Sources: Space Charge 27
Common space charge code strategy
• Majority of injector design codes use the strategy:– Assume nearly monochromatic beam
– Solve Poisson equation in the rest frame• various meshing strategies (simple uniform mesh for fast FFT
methods, nonequidistant adaptive mesh for distributions with varying density)
• Essentially removes granularity of the force
– Lorentz back-transform and apply forces including 3D field maps of external elements (cavities, magnets, etc.)
March 8, 2008 I.V. Bazarov, Electron Sources: Space Charge 28
Space charge codes
1) transform to rest frame of the reference particle
2) create mesh (charge) and cell grid (electrostatic fields)
3) create table containing values of electrostatic field at any cell due to a unit charge at any mesh vertex (does not need to be recalculated each time step)
Different approaches are used (e.g. envelope equation integration, macroparticle tracking, various meshing scenarios, etc.).Mesh method works as following:
March 8, 2008 I.V. Bazarov, Electron Sources: Space Charge 29
Space charge codes
4) assign macroparticle charges to mesh nodes, e.g. 1,2,3, and 4 vertices get QA1/A, QA2/A, QA3/A, and QA4/A respectively, where A1+A2+A3+A4 = A = ∆Z∆R
5) calculate field at each cell by using mesh charges and table, e.g.
6) find fields at macroparticle position by weighting
7) Apply force to each macroparticle
8) Lorentz back-transform to the lab frame
)4( ),3( ),2( ),1( EEEErrrr
AEAEAEAEA /))3()2()1(( 4321
rrrr+++
March 8, 2008 I.V. Bazarov, Electron Sources: Space Charge 30
Limitations
• Limitations of this method include:– Ignores interaction of charged beam with conducting
walls of the chamber (wake fields; can be added ad hoc)
– Fails if the beam has large energy spread
– Most of the time removing excessive granularity from space charge is justified because the actual beam has many more particles than ‘virtual’ beam
– However, the mesh method may lead to artificial ‘over-smoothing’ of the forces, underestimate intrabeam scat.
• There are powerful self-consistent Particle-In-Cell (PIC) codes. These require use of supercomputers.