Low Temp Plasma 2

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 29

Low Temperature Plasma and Its Applications

I. Kinetic description of low-temperature plasma

Martin Mašek
Institute of Physics, Czech Academy of Sciences
Division of High Power Systems
Department of Radiation and Chemical Physics (PALS Centre)
e-mail:masekm@fzu.cz

winter term 2023


Summary of the last lecture (1)
N-body model for plasma description - Newton’s law of classical
dynamics:
dpk dvk
= mk = Fk = F (rk ). (1)
dt dt
drk
= vk . (2)
dt
Hamiltonian formalism:
∂H
ṙk = , (3)
∂pk
∂H
ṗk = − . (4)
∂rk
Hamiltonian is function expressing the generalized energy of the
system (e.g. H = vk2 /2 + (qk /mk )φ)
6N -dimensional phase space with vector X = (rk , pk )
R
Phase space volume Γ(t) = Ω(t) dX
Summary of the last lecture (2)
Conservation law of phase space volume is then:
d
Γ(t) = 0 (5)
dt
Poincaré reccurence theorem: All the mechanical processes are
reversible
Continuity equation in a “differential form”:
∂ρ
+∇·j=0
∂t
We defined scalar function f N (X, t) in 6N -dimensional phase
space as probability density of occurrence at a phase space point.
We call it the distribution function.
Using the continuity equation in the phase space we derived the
equation for the distribution function - Liouville’s theorem:
X
∂t f N + (ṙk ∂rk f N + ṗk ∂pk f N ) = 0 (6)
k
Summary of the last lecture (3)
We derived the Boltzmann kinetic equation in the case of elastic
scattering process:
∂fa ∂fa ∂fa
Z
dpb vab dσab {fa0 fb0 − fa fb }.
X
+ va + Fa = (7)
∂t ∂ra ∂pa b

The right-hand side term is called Boltzmann collision integral


Z
Iab [fa , fb ] = dpb vab dσab {fa0 fb0 − fa fb } (8)

Solution for the steady state (Maxwell distribution):


na (pa − ma v0 )2
 
fa (pa ) = exp − (9)
(2πma kB Ta )3/2 2ma kB Ta
Solution for the gas in the field of potential forces
(Maxwell-Boltzmann distribution):
na pa2 Ua (r)
 
fa (pa , r) = 3/2
exp − − (10)
(2πma kB Ta ) 2ma kB Ta kB Ta
Liouville’s theorem (3)
We substitute for the Poisson bracket:
∂f N X ∂f N ∂H ∂f N ∂H
 
+ − = 0. (11)
∂t k
∂rk ∂pk ∂pk ∂rk
Forces are assumed as binary interaction with centrally symmetric
potential. Hamiltonian of the set of N particles is then:
X p2 X
k
H= + Ukj (|rk − rj |) (12)
k
2mk k6=j

Liouville’s theorem:
∂f N X ∂f N X ∂f N ∂ X
+ vk − Ukj (|rk − rj |) = 0. (13)
∂t k
∂rk k
∂pk ∂rk k6=j

Liouville’s theorem with an external force Fk = F(rk ):


∂f N X ∂f N X ∂f N h ∂ X i
+ vk + Fk − Ukj (|rk −rj |) = 0. (14)
∂t k
∂rk k
∂pk ∂rk k6=j
One-particle distribution function

One-particle distribution function f (1) (x1 , t):

f (1) (x1 , t)
Z
= dx2 ... dxN f N (x1 , ..., xN , t) (15)
V
V is the volume occupied by the gas.
V −1 f (1) (x, t)dx determines the probability that particle is in the
infinitesimaly small volume dx = dpdr in the vicinity of phase
space point x = (p, r).
The one-particle distribution function normalized on the total
number of particles is often used:
Na (1)
fa (p, r, t) = f (p, r, t), (16)
V a
where a means the particles of kind a.
Two-particle distribution function

Two-particle distribution function f (2) (x1 , x2 , t):

f (2) (x1 , x2 , t)
Z
= dx3 ...dxN f N (x1 , .., xN , t) (17)
V2
Non-interacting particles move independetly on each other, thus
the two-particle distribution function is expressed as a product of
one-particle distribution functions:
Na Nb (2)
f (xa , xb , t) = fa (xa , t)fb (xb , t) + gab (xa , xb , t) (18)
V V ab
gab (xa , xb , t) is correlation term expressing the interaction between
particles.
Three-particle distribution function

Three-particle distribution function f (3) (x1 , x2 , x3 , t):

Na Nb Nc (3)
f (xa , xb , xc , t) = fa (xa , t)fb (xb , t)fc (xc , t) +
V V V abc
+ fa (xa , t)gbc (xb , xc , t) + fb (xb , t)gac (xa , xc , t) +
+ fc (xc , t)gab (xa , xb , t) + dabc (xa , xb , xc , t) (19)

The first term corresponds to the independently moving particles.


The next three terms shows the pair correlations in the group of
three particles.
The last term represents the triple correlation function.
In the same way we can construct higher and higher
number-particle distribution functions.
Vlasov-Boltzmann kinetic equation (1)
If the interaction between particles is relatively weak, so the mean
energy of the interaction is much smaller than the mean kinetic energy.
In this case we can neglect the correlation functions and thus the
distribution function can be written as product of one-particle
distribution functions:
f (x1 , t) f (xN , t)
f N (x1 , ..., xN , t) = ··· (20)
N1 NN
Liouville’s theorem (13):
Na Na X N
∂f N X X ∂f N X X b
∂f N ∂
+ vka − Uab (|rka − rlb |) = 0.
∂t a ka=1 ∂rka a,b ka=1 lb=1
∂pka ∂rka
(21)
Integrating over the coordinate and momentum space of all particles
except coordinates and momentum of one particle, we obtain:
∂fa ∂fa ∂fa ∂ X
Z
+ va − dxb Uab (|ra − rb |)fb (xb , t) = 0. (22)
∂t ∂ra ∂pa ∂ra b
Vlasov-Boltzmann kinetic equation (2)
It is possible to prove that1
∂ X
Z
dxb Uab (|ra − rb |)fb (xb , t) ≡ E(ra , t)
∂ra b

And kinetic equation becomes:


∂fa ∂fa ∂fa
+ va + qa E(ra , t) = 0. (23)
∂t ∂ra ∂pa
If we use the full Lorentz force, we will get the kinetic eq. for plasmas:
∂fa ∂fa h 1 i ∂f
a
+ va + qa E(ra , t) + va × B(ra , t) = 0. (24)
∂t ∂ra c ∂pa
The Lorentz force in this equation is self-consistent, so it means that it
is produced by all the particles in the plasma. The equation (24) is
called Vlasov equation.
1
See V.P. Silin, (Czech translation, Academia, 1976), pp. 170–171
Boltzmann kinetic equation (1)
We integrate the Eq. (14) over all variables except x1 , which belongs to
the particle of kind a. If we keep in mind that the distribution function
f N becomes zero for infinite momenta and at the boundary of the
considered system, we get

When Na  1, we can neglect the quantities ∼ 1/N and the last


equation can be rewritten as:
h∂ ∂ ∂ i (1)
+ va + Fa f (xa , t) =
∂t ∂ra ∂pa a
∂ X Nb ∂Uab (|ra − rb |)
Z
(2)
= dxb fab (xa , xb , t) (25)
∂pa b V ∂ra
Boltzmann kinetic equation (2)
It is clear that the interaction term in the one-particle kinetic
equation is determined by the two-particle distribution function.
So we need to construct similar equation also for two-particle DF.
But there is the interaction described by a three-particle DF and
so on.
Such chain of equations is called Bogoljubov equations.
We will not solve these equations, but instead of them we use the
correlation function.
Na (1)
fa (xa , t) = f (xa , t), (26)
V a
Na Nb (2)
f (xa , xb , t) = fa (xa , t)fb (xb , t) + gab (xa , xb , t) (27)
V V ab
The first term on the right-hand side of Eq. (25) leads to the existence
of self-consistent field. Thus, if we mark
∂ X
Z
F̃a = Fa − dxb Uab (|ra − rb |)fb (xb , t) (28)
∂ra b
from Eq. (25) we get
Boltzmann kinetic equation (3)

∂fa ∂fa ∂fa


+ va + F̃a =
∂t ∂ra ∂pa
∂ X ∂Uab (|ra − rb |)
Z
= dxb gab (xa , xb , t) (29)
∂pa b ∂ra
The added term to the expression of force is self-consistent force of the
system. In the gases with smaller degree of ionization this term can be
neglected and just the collisional term expressed by the correlation
function remains. Right-hand side of the last equation can be expressed
as the generalized collision integral:

∂fa ∂ X ∂Uab (|ra − rb |)


  Z
= dxb gab (xa , xb , t) (30)
∂t coll ∂pa b ∂ra

This term does not describe only the effect of ordinary collisions.
Two-body problem (1)

r = r1 − r2 (31)
mr̈1 = F(r) (32)
Mr̈2 = −F(r) (33)
First, we add the equations (32) and (33)
mr̈1 + Mr̈2 = 0 (34)
Then, we multiply them by the second mass and subtract them
mM (r̈1 − r̈2 ) = (m + M )F(r) (35)
Two-body problem (2)
Eq. (34) can be directly integrated and gives
mṙ1 + M ṙ2 = const. (36)
Let us choose that inertial frame where the constant is zero and
integrate again:
mr1 + M r2 = const. (37)
In the chosen frame we have constant position vector defined as:
mr1 + M r2
R= (38)
m+M
This is the position vector of point C, which is called centre of mass.
Two-body problem (3)

Eq. (35) can be using r = r1 − r2 rewritten as:

µr̈ = F(r), (39)

where
mM
µ= (40)
m+M
is called reduced mass of the system. The two-body problem was thus
reduced to motion of one particle with mass µ. The solution to the eq.
(39) is:
M
r1 = R + r, (41)
m+M
m
r2 = R − r, (42)
m+M
Elastic collisions between particles
We write the energy of a particle in free space as:
p2
E = Ei + (43)
2m
Here, Ei is the internal energy of the particle and p2 /2m = mv 2 /2 is its
kinetic energy. In elastic collisions, there are no changes in the internal
energies, i.e., as
E1 = E10 , E2 = E20 , (44)
it follows from (43) that the total kinetic energy is
p21 p2 p0 2 p0 2
T = + 2 = 1 + 2 = T 0. (45)
2m1 2m2 2m1 2m2
By differentiating the expressions for particle positions (41) and (42)
with respect to time (g = v1 − v2 ):
m2
v1 = V + g, (46)
m1 + m2
m1
v2 = V − g. (47)
m1 + m2
Collisions between particles (1)

Differential scattering cross-section:


dσ b|db| b db
= = (48)
dΩ sin θdθ sin θ dθ
Collisions between particles (2)
Solid angle element:
dΩ = sin θdθdϕ (49)
Differential scattering cross-section:
dσ number of scattered particle per unit solid angle
=
dΩ number of incident particles per unit area
Let a given particle is at distance b from the axis through the
scattering centre S.
It is scattered through an angle θ to a detector located at polar
angles θ and ϕ and at distance L.
Detector area dS:
dS
dΩ = 2 = sin θdθdϕ (50)
L
b is called impact parameter
The particle scattered to dS must be incident the area:

dB = b|db|dϕ (51)
Boltzmann collision integral
Relation between the relative velocities before and after collision:
s
0 g0   2∆E
g = g − 2k(k · g) , g 0 = g2 + (52)
g m
Collision term:
∂ ∂ F ∂
 
+v· + · f (v, r, t)dv dr =
∂t ∂r m ∂v
= −b db dϕ g f1 (v1 , r1 , t) dv1 f (v, r, t) dv dr +
+ b db dϕ g 0 f1 (v10 , r1 , t) dv10 f (v0 , r, t) dv0 dr (53)
g 02 ∂g 0
dvdv1 = dvdv1 (54)
g 2 ∂g
b db dϕ = σ(g, θ) sin θdb dϕ (55)
b db
σ(g, θ) = (56)
sin θ dθ

∂f
 Z Z  g 03 ∂g 0 
= dΩ dv1 g σ 0 3 f (v0 )f1 (v10 ) − σf (v)f1 (v1 ) (57)
∂t coll g ∂g
Taylor expansion of the collision term (1)

  Z Z
∂f
= dv1 dΩσg(f 0 f10 − f f1 ) =
∂t coll
Z Z 
m ∂f1 ∂f

= dv1 dΩσg − f+ f1 ∆vi +
m1 ∂v1i ∂vi

∂2f ∂2f
2
m ∂f ∂f1 m
  
+ f1 − 2 + f ∆vi ∆vj = (58)
∂vi ∂vj m1 ∂vi ∂v1j m1 ∂vi ∂vj

Due to the eq. (85) and (86) we have:

2π(ee1 )2 2π(ee1 )2 ∂2g


h∆vi i = − ψgi , h∆vi ∆vj i = − ψgij , gij = (59)
µm2 g 3 m2 ∂vi ∂vj

  Z 
∂f 1 gi 1 ∂f1 1 ∂f gij
 
= 2π(ee1 )2 ψ dv1 f− f1 + ×
∂t coll
µ g 3 m1 ∂v1i m ∂vi 2m2

∂2f ∂2f
2
m ∂f ∂f1 m
 
× f1 − 2 + f (60)
∂vi ∂vj m1 ∂vi ∂v1j m1 ∂vi ∂vj
Taylor expansion of the collision term (2)

The bracket on the last row we can rewrite as:


∂2f m ∂f ∂f1 m 2 ∂2f
   
f1 − 2 + f =
∂vi ∂vj m1 ∂vi ∂v1j m1 ∂vi ∂vj
∂ ∂f m ∂f1 m ∂ ∂f m ∂f1
   
= f1 − f − f1 − f (61)
∂vi ∂vj m1 ∂v1j m1 ∂v1j ∂vi m1 ∂v1i
and the whole last term of eq. (60) is expressed as:
 
∂ ∂f m ∂f1 m ∂ ∂f m ∂f1
   
gij = gij f1 − f − f1 − f −
∂vi ∂vj m1 ∂v1j m1 ∂v1j ∂vi m1 ∂v1i

∂f m ∂f1 ∂ m ∂
  
−f1 − f gij − gij (62)
∂vj m1 ∂v1j ∂vi m1 ∂v1i
The last term combines with the linear term:
1 gj 1 ∂ m ∂
 
− − gij − gij =0 (63)
µm g 3 2m2 ∂vi m1 ∂v1i
Taylor expansion of the collision term (3)
Finally, we get for the collision integral:
  Z  
∂f ∂ 1 ∂f 1 ∂f1

2
= π(ee1 ) ψ dv1 gij f1 − f (64)
∂t coll
∂vi m ∂vj m1 ∂v1j
In the last equation, we can swap the partial derivative with respect to vi and the
integral:
   
(ee1 )2 ∂
Z
∂f 1 ∂f 1 ∂f1 ∂Ii

=π ψ dv1 gij f1 − f = (65)
∂t coll
m ∂vi m ∂vj m1 ∂v1j ∂vi
We got the final form the Landau collision operator. It is sometimes convenient to
write it using function Ii :
1 ∂f
 
Ii = Dij − Ai f (66)
m ∂vj
Coefficient of momentum diffusivity:
(ee1 )2
Z
Dij = π ψ dv1 gij f1 (67)
m
Coefficient of dynamic friction:
(ee1 )2
Z
∂f1
Ai = π ψ dv1 gij (68)
m1 ∂vi
Taylor expansion of the collision term (4)
Let us use the notation:
∂2g ∂ 1 vi
gij = , g = |v − v1 |, =− 3 (69)
∂vi ∂vj ∂vi g v
We rewrite the coefficient of momentum diffusivity and the coefficient of dynamic
friction as:
(ee1 )2 ∂2 (ee1 )2 ∂2
Z
Dij = π ψ dv1 gf1 = 2π ψ G (70)
m ∂vi ∂vj m ∂vi ∂vj

(ee1 )2 (ee1 )2
Z Z
∂f1 ∂gij
Ai = π ψ dv1 gij =π ψ dv1 f1 (71)
m1 ∂v1j m1 ∂vj
Because
∂gij ∂ 1
 
=2 , (72)
∂vj ∂vi g
we have from (71)
(ee1 )2
Z
∂ 1
 
Ai = 2π ψ dv1 f1 (73)
m ∂vi g
We swap the partial derivative and the integral which results in
(ee1 )2 ∂ (ee1 )2 ∂H
Z
f1
Ai = 2π ψ dv1 = 2π ψ (74)
m ∂vi g m ∂vi
Rosenbluth potentials
In the previous we defined the Rosenbluth potentials:
Z
f1
H= dv1 (75)
g
Z
1
G= dv1 f1 g (76)
2
Applying the velocity space Laplacian operator on the Rosenbluth potentials reads

4H = −4πf1 (77)
Z Z
1 f1
4G = − dv1 f1 gii = dv1 =H (78)
2 g
∂Ii 1 ∂ ∂f
 
= Dij − Ai f (79)
∂vi m ∂vi ∂vj
 
∂Ii (ee1 )2 1 ∂ 3 G ∂f ∂2g ∂2f 1 ∂ 2 H ∂H ∂f
  
= 2π ψ 2
+ − + (80)
∂vi m m ∂ vi ∂vj ∂vj ∂vi ∂vj ∂vi ∂vj m1 ∂ 2 vi ∂vi ∂vi
 
∂Ii (ee1 )2 m m1 − m
= 2π ψ 4π f f1 + ∇H∇f + ∇∇G · ∇∇f (81)
∂vi m m1 m1
is electron-electron collision term.
Coulomb logarithm (1)

Coulomb potential:
e1 e2
U2 = . (82)
r
Impact parameter:
e1 e2 ϑ
b= cotg , (83)
µg 2 2
where
m1 m2
µ=
m1 + m2
is reduced mass. Rutherford scattering:
2
db 1 e1 e2 1

σ(ϑ, g) = b = ϑ
. (84)
dϑ sin ϑ 2µg sin4 2
Z Z
h∆vi i = dvf1 (v) dΩσ(g, ϑ)g∆vi (85)
Z Z
h∆vi ∆vj i = dvf1 (v) dΩσ(g, ϑ)g∆vi ∆vj (86)
Coulomb logarithm (2)
Let us define v coordinates as the g k ẑ. It means that g = (0, 0, g).
 
2m1 2m1 ϑ ϑ ϑ ϑ
∆v = − k(k · g) = + g sin cos cos ε, cos sin ε, sin (87)
m + m1 m + m1 2 2 2 2

k:  
ϑ ϑ ϑ
k= cos cos ε, cos sin ε, sin , (88)
2 2 2
and
ϑ
k · g = g sin
2
.  Z Z π 
2m1 2
h∆vi = 0, 0, − dv1 f1 (v1 )g 2π dϑ sin ϑσ(ϑ, g)kz (89)
m + m1 0
 Z Z π 2 
2m1 e1 e2 1 ϑ

h∆vi = 0, 0, − dv1 f1 (v1 )g 2 2π dϑ sin ϑ ϑ
sin2 (90)
m + m1 0
2µg sin4 2
2
 Z Z π ϑ

m1 e1 e2
 
1 cos 2
h∆vi = 0, 0, − 2π dv1 f1 (v1 ) dϑ ϑ
(91)
m + m1 µ g2 0 sin 2
Coulomb logarithm (3)
Coulomb logarithm: Z π ϑ
cos 2
ψ= dϑ ϑ
(92)
0 sin 2
Let us solve the integral: Z ϑ
cos 2
ψ= dϑ ϑ
(93)
sin 2
substition x = ϑ/2, dx = dϑ/2
Z
cos x
ψ=2 dx (94)
sin x
u = sin x, du = cos x du,
Z
1 ϑ
ψ=2 du = 2 ln |u| = 2 ln sin , (95)
u 2
but this integral diverges in the interval (0, π). When we “neglect” the infinity in the
integral limits and use ϑmin and instead of “zero” in the lower limit we will use the
Debye length:
Z π ϑ π
cos 2 ϑ ϑmin
ψ= dϑ ϑ
= 2 ln sin = −2 ln sin (96)
λD sin 2
2 ϑmin
2
Coulomb logarithm (4)

When we exactly solve the integral above, the contribution to the integral from 0 to
ϑmin is negligible.
We calculate ϑmin from the expression for the Debye length λD :
e1 e2 ϑmin
b = λD = cotg (97)
µg 2 2
It implies that
ϑmin 1
sin2 = 2 (98)
2

e1 e2
1+ µλD g 2

Coulomb logarithm:
e1 e2 2
ψ = ln(1 + ( ) ) (99)
µλD g 2

You might also like