16.522, Space Propulsion
Prof. Manuel Martinez-Sanchez
Lecture 10: Electric Propulsion - Some Generalities on Plasma (and Arcjet
Engines)
Ionization and Conduction in a High-pressure Plasma
A normal gas at T <3000K is a good electrical insulator, because there are almost no
~
free electrons in it. For pressure >
~
0.1 atm, collision among molecule and other
particles are frequent enough that we can assume local Thermodynamic Equilibrium,
and in particular, ionization-recombination reactions are governed by the Law of
Mass Action. Consider neutral atoms (n) which ionize singly to ions (i) and electrons
(e):
n R e +i (1)
One form of the Law of Mass Action (in terms of number densities n=
P
j
kT
, where T
j
is the same for all species) is
nn
ei
() (2)=S T
n
n
Where the “Saha function” S is given (according to Statistical Mechanics) as
3
()ST
i
q
n
= 2
q ? π
e
?
h
2
?
2m kT ?
? ?
eV
i
e
kT
-
(3)
q
i
= Ground state degeneracy of ion (= 1 for H
+
)
q
n
= Ground state degeneracy of neutral (= 2 for H)
m
e
= mass of electron = 0.91 × 10
-30
Kg
k = Boltzmann constant = 1.38 ×10
-23
J/K
(Note: k = R/Avogadro’s number)
h = Plank’s constant = 6.62x10
-34
J.s.
V
i
= Ionization potential of the atom (volts)
(V
i
= 13.6 V for H)
Except for very narrow “sheaths” near walls, plasmas are quasi-neutral:
e
n= n
i
(4)
So that
e
n
n
n
()=S T (3’)
can be used.
2
16.522, Space Propulsion Lecture 10
Prof. Manuel Martinez-Sanchez Page 1 of 12
2
Given T, this relates n
e
to n
n
. A second relation is needed and very often it is a
specification of the overall pressure
P = (n + n
i
+ n )kT = (2n + n )kT (5)
e n e n
Combining (3’) and (5),
2
()
?
?
P
-2n
e ?
= ST n = ST
?
kT
?
()(n-2n
e
)
e
?
P
Where n= is the total member density of all particles.
kT
We then have
2
n+ 2Sn -Sn = 0
e e
S
2 (6)
n= -S + + Sn =
n
e
1+ 1 +
n
ST()
Since S increases very rapidly with T, the limits of (6) are
→ Sn
(
→ 0
)
(Weak ionization) n ?????
e
T0 T0→ →
?????
n
n
e
T→∞
→
2
(Full ionization)
G
Once an electron population exists, an electric field Ewill drive a current density
G
j through the plasma. To understand this quantitatively, consider the momentum
balance of a “typical” electron. It sees an electrostatic force
G G
F= -eE (7)E
It also sees a “frictional” force due to transfer of momentum each time it collides
with some other particle (neutral or ion). Collisions with other electrons are not
counted, because the momentum transfer is in that case internal to the electron
species. The ions and neutrals are almost at rest compared to the fast-moving
electrons, and we define an effective collision as one in which the electron’s directed
momentum is fully given up. Suppose there are ν
e
of these collisions per second
( ν
e
=collision frequency per electron). The electron loses momentum at a rate
JG
JJG
-m Ve ν , where V
e
=mean directed velocity of electrons, and so
e e
G JG
FFriction = -m Ve ν (8)
e e
On average,
16.522, Space Propulsion Lecture 10
Prof. Manuel Martinez-Sanchez Page 2 of 12
G G
F+ Friction =0, E
or
JG G
mVeν =-e E
e e
JG
? e ?
G
V=-
? ?
E (9)e
ν
?
m
ee ?
The group μ
e
=
e
is the electron “mobility” ((m/s)/ (volt/m)). The current density
m ν
ee
is the flux of charge due to motion of all charges. If only the electron motion is
counted (it dominates in this case)
G JG
j-en V ≡
e
e (10)
and from (9),
2
G
?
en
?
G
e
j=
?
?
m ν
?
? E (11)
?
ee
?
The group
2
e ν
e
σ = (12)
m ν
ee
is the conductivity of the plasma (Si/m).
Let us consider the collision frequency. Suppose a neutral is regarded as a sphere
with a cross-section area Q
en
.
Electrons moving at random with
(thermal) velocity c
e
intercept the
area Q
en
at a rate equal to their flux
nc Q
en
. Since a whole range of
ee
speeds ce exists, we use the average
value ce for all electrons. But this is
for all electrons colliding with one
neutral. We are interested in the
reverse (all neutrals, one electron),
so the part of ν
e
due to neutrals
should be nc Q
en
. Adding the part due to ions, e
n
ν = n cQ
en
+ n cQ
ei
(13)
e
e n
e
i
16.522, Space Propulsion Lecture 10
Prof. Manuel Martinez-Sanchez Page 3 of 12
NOTE:
c
e
is very different (usually much larger) than V
e
. Most of the thermal motion is fast,
but in random directions, so that on average it nearly cancels out. The non-
cancelling remainder is V
e
. Think of a swarm of bees moving furiously to and fro, but
moving (as a swarm) slowly.
G
The number of electrons per unit volume that have a velocity vector c
e
ending in a
“box”
3
dc dc dc ≡ d c in velocity space is defined as
e e
z
ee
x y
G
3
fc
)
d c
(
e e e
(
G JG
Where fc , x
)
is the Distribution function of the
e e
JG
electrons which depends (for a given location x and time
G
t) on the three components of c
e
. In an equilibrium
JJG
situation all directions are equally likely, so f= f
e (
c c
)
= f ()only, and one can
e e e e
show that the form is Maxwellian.
3 2
2
mc
2 2 2 2
-
ee
f= n
? m ?
?
e
2kT
e
;
(
c = c + c
ey
+ c
ez
)
(14)
e e
?
e
e ex
2kT
e ??
π
With the normalization
3
fd c = n .
e e∫∫∫ e
The mean velocity is then
3
ce ≡
∫∫∫
c f d c
ee e
and direct calculation gives
8 kT
e
(15)c=e
π m
e
For Hydrogen atoms,
ce = 6210 T (m/s, with T
e
in K) (16)
e
NOTE:
If there is current, the distribution cannot be strictly Maxwellian (or even isotropic).
But since V
e
<< c
e
, the mean thermal velocity is very close to Equation (15) anyway.
16.522, Space Propulsion Lecture 10
Prof. Manuel Martinez-Sanchez Page 4 of 12
Regarding the cross sections Q
en
, Q
ei
, they depend on the collision velocityc,
e
especially Q
ei
. This is because the e-i coulombic interaction is “soft”, so a very fast
electron can pass nearly undeflected near an ion, whereas a slow one will be strongly
deflected. The complete theory yields an expression
Q
ei
2.95 ×10
-10
ln Λ
(m
2
) (17)
T
(T in Kelvin)
where
1
ln Λ -11.35 + 2ln T(K)- ln P(atm) (18)
2
so that ln Λ is usually around 6-12, and can even be taken as a constant (~8) in
rough calculations. For the neutral hydrogen atoms, the collisions are fairly “hard”,
and one can use the approximation
2
Q 2 ×10
-19
m (19)
e
H
Numerical Example
Consider Hydrogen at P=1 atm. For T ≥ 4000K , diatomic H
2
is not present anymore
(H
2
→ 2H). So ionization is from atomic hydrogen, H, for which
V = 13.6 volts,
i
q= 1,
i
q= 2,
n
so that
3
2
157,800
-
e
T
S= 2.42 ×10
21
T (m
-3
).
We also find
7.34 ×10
27
(m
-3
),n=
T
c =
e
6210 T (m/s),
σ = 2.821×10
-8
n ν
e e
(ν = ν
ei
+ ν ).
e en
16.522, Space Propulsion Lecture 10
Prof. Manuel Martinez-Sanchez Page 5 of 12
The results are shown below.
T(K) 5,000 6,000 7,000 8,000 10,000 12,000 14,000 16,000
-3
)
1.47x10
24
1.22x10
24
1.048x10
24
9.18x10
23
7.34x10
23
6.12x10
23
5.24x10
23
4.59x10
23
nm
S(m
-3
)
(
1.68x10
13
4.26x10
15
2.30x10
17
4.70x10
18
3.393x10
20
6.189x10
21
5.104x10
22
2.55x10
23
-3
)
4.97x10
18
7.216x10
19
4.906x10
20
2.072x10
21
1.545x10
22
5.565x10
22
1.203x10
23
1.716x10
23
nm
e
(
α 3.4x10
-6
5.9x10
-5
4.7x10
-4
0.0023 0.0215 0.1000 0.298 0.597
ce (m/s)
4.39x10
5
4.81x10
5
5.20x10
5
5.56x10
5
6.21x10
5
6.80x10
5
7.35x10
5
7.86x10
5
ln Λ 5.68 6.05 6.36 6.62 7.07 7.44 7.74
ν
8.01
ei
(s
-1
) 1.46x10
8
1.72x10
9
9.76x10
9
3.51x10
10
2.00x10
11
5.77x10
11
1.030x10
12
1.244x10
12
ν (s
-1
) 1.29x10
11
1.15x10
11
1.09x10
11
1.02x10
11
8.73x10
10
6.81x10
10
4.16x10
10
1.82x10
10
en
ν (s
-1
) 1.29x10
11
1.17x10
11
1.19x10
11
1.37x10
11
2.87x10
11
6.45x10
11
1.072x10
12
1.262x10
12
e
σ (Si/m) 1.09 17.4 116.3 426.7 1519 2434 3166 3836
Notice
(a) Coulomb-dominated ( ν >> ν
en
) for T > 8000 K
ei
~
(b) Rapid rise of ionization fraction α for 7000 < T < 10, 000 K
~ ~
100 Si/m for T > 7000 K
~
(c) Conductivity above
1, 000 Si/m T > 9000 K
~
Ohmic Dissipation - Stability, constriction
The conductivity σ increases rapidly with T in the fully Coulomb-dominated range,
0.0153 T
3
2
e
σ = (T
e
in K, σ in Si/m).
ln Λ
Notice also how, in this limit (which occurs at high temperature, as α approaches 1),
the conductivity becomes independent of the kind of gas in question, except for
small influences hidden in ln Λ .
16.522, Space Propulsion Lecture 10
Prof. Manuel Martinez-Sanchez Page 6 of 12
V
One important consequence of σ = σ
( Τ )
is the tendency for current to concentrate
into “filaments”, or “arcs”. To understand this, consider the amount of work done by
electric forces to overcome the “friction” on the electrons due to collisions. The force
G
on the n
e
electrons in a unit volume is-en E, and these electrons reach a terminal
JG
e
velocity V
e
as they slide against friction. Hence the power dissipated per unit volume
is
GJG JG G
D
OH
=-en E.V =-
(
en V
)
. E
e e e e
GG
= j . E (Ohmic dissipation) (20)D
OH
G G
Since Ohm’s law gives j= σ E , we can put
j
2
= σ Ε
2
or D
OH
= (21)D
OH
σ
The simplest situation is one with
an initially uniform plasma subject
G
to a constant applied field E, such
as would occur between the plates
of a plane capacitor:
Regardless of the path taken by
the current, if the plates are large
and the gap is small, the field
E= remains unchanged. If we
d
now look at
= σ Ε
2
,D
OH
we see that the dissipation becomes large wherever the conductivity (hence the
temperature) is large. Starting from uniform temperature, if a small non-uniformity
arises such that T is higher along a certain path, that path becomes more
conductive, heats up due to extra Ohmic dissipation, and this reinforces the initial
nonuniformity. The result is a constriction of the current into a filament or “arc”.
In principle, the constriction process would continue indefinitely and lead to arcs of
zero radius and infinite current density. But as the temperature profile steepens,
heat will increasingly diffuse away from the hot core to the cooler surroundings, and,
provided it can be removed efficiently from there, an equilibrium is eventually
reached at some finite arc radius and arc core temperature. Clearly, the detailed end
result will depend on the details of the thermal management of the gas: the more
efficient the cooling of the background, the more the constriction can progress, and
the hotter the eventual arc. This counter-intuitive result (more cooling leads to
hotter arcs) is one of several paradoxical properties of arcs, all of them related to
their being the result of a statically unstable situation. We analyze this behavior
next.
16.522, Space Propulsion Lecture 10
Prof. Manuel Martinez-Sanchez Page 7 of 12
Sample Physical Properties of High-temperature Gases (Near Equilibrium)
Figure 1: Equilibrium Composition of Nitrogen at P=1 atm.
Note: N
2
replaced by N at ~ 7,000K, then by N
+
at ~ 14,000K and then by N
++
at
~ 29,000K. Electron density satisfies n
e
= n
+
+ 2n
++
For Hydrogen, similar, but all transitions at lower temperature (and, of course,
there is no H
++
)
16.522, Space Propulsion Lecture 10
Prof. Manuel Martinez-Sanchez Page 8 of 12
Figure 2, 3: Electrical Conductivity of Nitrogen and Oxygen
gas type
Note: Weak dependence on
Pressure
Units are mho/cm ≡ Si/cm. 1 Si/cm = 100 Si/m
16.522, Space Propulsion Lecture 10
Prof. Manuel Martinez-Sanchez Page 9 of 12
16.522, Space Propulsion Lecture 10
Prof. Manuel Martinez-Sanchez Page 10 of 12
Figure 4, 5: Thermal Conductivity of Nitrogen
Note: The molecular contribution Km is due to the transport of energy by the
random thermal motion of N
2
molecules. It increases weakly with T when there
are molecules, but, of course it disappears when they dissociate (T ≥ 7000K).
Similar physics applies to the atoms contribution K
a
, the ion contribution K
i
and
the electron contribution, K
e
(these rise rapidly at first, when these species first
appear, then drop rapidly out when they in turn dissociate).
The most striking new feature are K
D
(the “dissociation contribution”), and K
I
(the
“ionization contribution”). These are akin to heat-pipe effects, and they appear in
temperature ranges where dissociation (or ionization) are very sensitive to T
(around 7000K for K
D
, around 15,000K for K
I
). The mechanism is as follows:
When there is a temperature gradient, molecules are dissociating at a higher rate
16.522, Space Propulsion Lecture 10
Prof. Manuel Martinez-Sanchez Page 11 of 12
than they are forming by recombination in high-T sections, while the reverse
happens in low-T parts. The dissociation products are continuously diffusing from
hot to cold regions. When they are created (high T) they absorb the heat of
dissociation (very large) and they deliver it when they recombine (at low T). This
creates a net (strong) heat transport from high to low T, and hence a thermal
conductivity K
D
. The same description applies to K
I
, except the heat transported
is now the ionization energy. Note the very strong K
D
impact in the natural scale
Figure (5).
Figure 6: Thermal Conductivity of Hydrogen
Qualitatively similar to Nitrogen, the K
I
component is here even more evident
(around 16,000 K).
16.522, Space Propulsion Lecture 10
Prof. Manuel Martinez-Sanchez Page 12 of 12