Page 1
;\ •
...-v? • -"
TRITA-EPP-7U-06
BREAKDOWN AND PLASMA FORMATION
IN A ROTATING PLASMA DEVICE
by
B. Bonnevier and
A. H. Sillesen
Stockholm and Risd in ^arch 1974.
Department of Plasma Physics and Fusion Research
Royal Institute of Technology
S- 100 1*4, Stockholm 7j, Sweden
Danish AEC, Risd,
^ Roskilde, Denmark
*$?!• >
'---*•
Page 2
Abstract
Breakdown and formation of a plasma with high ion
temperature have been studied in the Puffatron at Risd.
The conditions for plasma formation is understandable along
the same lines as breakdown in an ordinary crossed field
discharge, Penning, PIG or magnetron discharge where ions
are not magnetically confined, as in the present experiment.
The growth of ionization occurs much faster than expected
from ionization by thermal electrons. This has earlier been
seen in some rotating plasma experiments. It is suggested
that work along the present lines can result in a time inde-
pendant arc discharge, where ions have thermonuclear energies.
Page 3
1.
S 1. Introduction
Different forms of crossed field discharges, that is dis-
charges with strong electric fields transverse to magnetic fields,
have been considered in the past. Here we will discuss such dis-
charges with time-independent magnetic fields or whose
time-dependance is unimportant for the main characteristics of the
discharge. Two types have been considered to a great extent by
different authors.
In the first type the electron motion, but not the ion motion,
is modified strongly by the magnetic field. Here we find work on
Penning discharges, P.I.G. (Philips ionization gauge) discharges
and coaxial diode or magnetron discharges. For the present dis-
cussion it is of special intrest that some of these discharges
can be obtained at neutral gas pressures from about 10 to 1 Pas-
cal. It is this great range of possible operating pressure which
has made these discharges so applicable in ionization pressure
gauges.
In the second type of discharge mentioned above, both electron
and ion motion are modified strongly. These discharges aie often
called rotating plasmas. A thorough review was given by Lehnert
(1971). These discharges give us one of the most efficient methods
known to produce a fully ionized plasma. In some of these dis-
21 -3charges a dense, n = 10 m , but fairly low temperature,
5 oT - 10 K, plasma is obtained. In other discharges, like
the present one, ions can directly gain energy from the
electric field and swiftly should be able to obtain the
10 keV energy around which nuclear reactions begin to take
place. Thus we have a direct conversion of high voltage direct
current energy into plasma ion energy. High voltage crossed field
discharges is an experimental field almost totally neglected,
Page 4
however» but with present day technique it should be fairly easily
accessible. In crossed field discharges considerable mass motion
velocities are obtained. Therefore, the centrifugal force might
give an extra, important, possibility for plasma confinement, added to
ordinary mirror confinement. This was for instance mentioned by
Bonnevier and Lehnert (1959), who pointed out, that 70 keV deuterons
could conceivably be contained by the centrifugal force created from
crossed electric and magnetic fields. Thus the low energy ions in a
mirror machine, with rotating plasma,are trapped. This is important,
because low energy ions swiftly get scattered out in an ordinary
mirror machine, due to the large Coulomb crossection. This gives a
considerable energy loss. This is further discussed by Bonnevier(197H).
Another advantage of rotating plasmas is that impurities
are centrifuged out, as discussed by Bonnevier (1966, 1970).
There is much scattered evidence of crossed field discharges.
In order to get a more unified approach it seems logical to examine
the ionization processes, the breakdown and the plasma build up in
crossed field geometries. Also as our interest lies in high vol-
tage effects, we started with the present configuration of the
Puffatron at Risd, where anode voltages up to 50 kV can be applied.
§ 2. Apparatus
The puffatron at Risd is shown in Fig. 1. The vacuum vessel
of the device has a diameter of 0.15 m and a length of 2.5 m. A
part of this vessel also serves as the outer electrode. The inner
electrode is a stainless steel rod, placed along the center line
of the apparatus. An axial magnetic field with a mirror ratio of
1.5 and a magnetic field strength of up to 1.6 Tesla in the cen-
tral parts of the machine is used, and a condenser bank of luF with
voltage up to 50 kV has been applied to the electrodes. In
normal operation the apparatus is pumped to a low base pressure, «
by oil free pumps. When magnetic field is already applied, gas,
from a small chamber of 50 mm , is injected through a fast valve
in the center of the machine, and finally the voltage between the
is switej?**} on,
Page 5
3.
In operation with V = 30 kV, B = 1.5 Tesla and a pressure p = 60 kP
of H2> in the small chamber behind the fast valve, a plasma with
20 —3m
20 —3a density of ^U^IO m , ion energy around 500 eV and estimated
electron energy of 20eV is produced. The density has been measured
by a He-Ne laser interferometer and the ion energy is obtained
from charge-exchange neutrals. The plasma flows along the magnetic
field lines and interacts with the walls after about 5ys. The
breakdown and plasma formation thus are very efficient in a device
of this type. This will be discussed further in the present paper.
Improvement of the confinement will be discussed elsewhere. A
description of the puffatron and some earlier results obtained in
this apparatus have been published by 0ster (1969).
i
5 3. Experiments
Here we wish to report on the discharge characteristics
obtained in the breakdown and plasma formation stages. Voltage
and current characteristics of the discharge were obtained. Further,
a logarithmic amplifier made it possible to analyse the current
in greater detail, in the way earlier reported by Rasmussen et al.
(1969). In Fig. 2 is shown that distinct types of discharge
characteristics were obtained for different combinations of ap-
plied voltage and magnetic field strength. The measure-
ments reported in this paper are made with positive inner elec-
trode. At low magnetic field or low applied voltage,
different types of low current discharges were obtained,
an example of which is shown in Fig.2a. In these discharges the
current rise was slow, up to a current plateau of some amperes.
In the other region of high voltage and high magnetic field
Page 6
strength the current rises to several kiloamperes, producing a
dense plasma with high ion temperatures, Fig. 2,b,c. the current
rise in this region was in many cases rather slow at currents less
than ten amperes but increased at higher currents by a factor
of ten.
The border V^ = f(B*,p) between the high current and the low
current regimes is conveniently devided in two parts, a low vol-
tage branch and a low magnetic field branch, Fig.3.
In the low voltage branch, the voltage V, increases steeply
when the pressure decreases: P = 150kP gives V, = 5kV,
p = 50kP gives Vfc = 25-3OkV and p = 40kP gives V^ higher than
U5 kV. These values are almost independent of the magnetic
field strength.
The low magnetic field branch depends slightly on the voltage.2
Experiments seem consistent with V. = g(p)«B. , where g(p) depends
on pressure only. For pressure p = 50kP, g(p) is very near to the
value obtained for proton cutoff. For protons leaving the anode,
calculations give the Hull cut-off condition
V = qB2(r2-r2)/8mr2.
With proton charge q and mass m, an anode radius r =0.025 and
3 2r=0.075 we get V=53»10 *B , where V is in volt and B in tesla. At
30 kV and varying pressures we get for the magnetic field B at the
border the following values: p=60 kP gives B=0.30 Tesla, p=50
gives B=0.52 and p=H5 gives B=0.80. The Hull cutoff at 30 kV is at
B=0.75 Tesla. Thus the exact position of the branches varies with
pressure.
The two branches of the Vb=f(Bb,p) curve meet at a certain
point. For psSOkP, Vj»25 kV, Bb~0.5 Tesla is the common point.
6
This corresponds to a velocity E/B of 10 m/s. In the measurements
with neutral detector referred to earlier, it was shown that the
most probable velocity of the neutrals corresponds to 2x(E/B),
what is also equal to the maximum velocity of a particle in cycloi-
Page 7
5.
within the whole high current regime. Near to the low magnetic
field branch, the ions will therefore get an energy near to the
applied anode voltage.
§ U. Discussion
a. Current characteristics
In earlier experiments on a rotating plasma device at
Amsterdam, Rasmussen, Barbian and Kistemaker (1969), it was repor-
ted that in 50 per cent of the shots there was an exponential in-
-1 2crease in discharge currents from 10 A to a few times 10 A. Thecurrent rise could be accounted for by putting the discharge cur-
rent derivative -rr - irz s nn <av>. . . From simple theory theat dt n loniz
entity <ov>. . for ionization by thermal electrons was derivedloniz J
as a function of E/B. Good agreement between theory and experiment
was reported.
On the other hand, it has been pointed out by Lehnert (1966)
that the growth of ionization in the rotating plasma experiments
in Stockholm is almost two orders of magnitude faster than expec-
ted from ionization by thermal electrons. A fast growth of ioniza-
tion in a straight crossed field discharge has also been reported
by Eninger (1966) and Axnäs (1972). The fast ionization in these
experiments is related to the critical velocity phenomenon first
described by Alfven, By some processes electrons gain energy from
the relative motion of ions and neutrals. This enhances the ioni-
zation probability drastically. For reviews seé Danielsson (1973)
and Sherman (1973). In the present experiments, breakdown near the
low voltage branch can follow thermal ionizatioi» theory to about
10A. At higher currents a much faster ionization is going on. It
^
Page 8
is thus interesting "»-hat in the same shot we get the two behaviours.
The fast current rise in our experiment could be due to processes
ot importance for the critical velocity phenomenon. An increase in
the current rise would also occur if the electric field deviates
strongly from the electric vacuum field due to sheath effects.
In the present experiments, ionization of neutrals by fast ions
can not be neglected.
b. Border between low and high current regime
The border curve V, = f(B. ,p) presented earlier is very similar
to the curves presented for breakdown in a cylindrical magnetron
as presented by Redhead (1958) and Schuurman (1966). In their case,
only electrons are contained in the magnetic field and ion gyro
radii are much greater than dimensions of the apparatus. Ions
created in the gas will fall towards the cathode. There electrons
are produced by y-processes. The electrons drift into the discharge
volume, with reduced mobility due to the transverse magnetic field,
towards the anode. The theory of the discharge is thus equivalent
to the ordinary Townsend discharge. The Townsend breakdown criterion
can be written
-1) = 1 or
In (1 + i) = /adr
Here o is the first Townsend coefficient defined in the
usual way as the number of ionizing collisions performed by an
electron per unit displacement in the direction of the electric
field. At low pressures, the electron will move in cycloid orbit2^ffigwith diameter D * ̂ffig and hence a
be taken between the inner and outer electrode.
1/V * §§». The integral shall2mE
Page 9
7.
As the coefficient is placed in the argument of the
logarithmic function, very large changes in the value of y
are required to change the breakdown condition.
With a in the form obtained above, we see, that one
discharge condition is that there is a sufficient number of
gyroradii between the electrodes. This gives the upper branch
of the breakdown curve. A second requirement is that the electron,
in some part of its cycloidal orbit, must have energy enough for
ionization, or else gain this energy through elastic collisions.
But then a number of steps with length of order D has to be taken.
This lowers a. Thus the requirement is reasonably given >>y
V. < E-D = 2 - • £-3 e B2
This corresponds to the lower branch of the breakdown curve for
an infinitely long cylinder. It has been obtained experimentally
by other authors but very often electron emission from end plates
give a lower branch with constant voltage. For the condition tog
be fulfilled E/B>10 is needed for electrons.
The essential characteristics of the breakdown when ions
are not contained can thus be described by two processes. One
iB the ionization of neutral gas by electrons, the second is
electronemission from the negative electrode due to ion impact.
When the magnetic field gets greater the ion orbits begin to
be curved. This alters in the first instance the y-processes.
Page 10
- • v*. -
8.
The Y-process will be more efficient when ions hit the electrode
with a high tangential velocity. A further increase of the
magnetic field will make the ions contained. Then y-processes by
charge-exchange neutrals should be dominant. Further at high ion
energies the ionization probability of the neutral gas by ion
impact is not negligible. We will not judge at present which of
these processes are the most important ones.
Acknowledgement. We wish to thank Mr. Knud^Vilhelm Weissberfc
foT the construction of the logarithmic amplifier, and Mr, John
Petersen for his helpful assistance with the experiments,
Page 11
9.
References
' -n+l
Axnäs, I., Royal Inst. Technology, Stockholm,
Rep. TRITA-ÉPP-72-31, 1972.
Bonnevier, B., Lehnert, B., Ark. Fys. 16, 231-236, 1960.
Bonnevier, B., Ark. Fys. £3, 255-270, 1966.
Bonnevier, B. , Plasma Physics 13., 763-774, 1971.
Bonnevier, B., Report TRITA-EPP-74-10, Royal Institute of
Technology, Stockholm 1974.
Danielsson, L., Astrophysics and Space Science
24 > , 1973.
Eninger, J., Proc. VII. Intern Conf. on Phenomena in Ionized
Gases, 1, 520, Beograd 1966.
Lehnert, B., Physics Fluids $, 774-779, 1966.
Lehnert, B., Nuclear Fusion 11, 485-533, 1971.
Rasmussen, C.E., Barbian, E.P., Kistemaker, J.,
Plasma Physics ljL, 183-195, 1969.
Redhead, P.A., Canad. J. Phys. 3J5, 255-270, 1958.
Schuurman, W., Investigation of a low pressure Penning Discharge
Rijnhuizen Report 66-28, 1966.
Sherman, J., Astrophysics and Space Science 2£, 487-510, 1973.
Sillesen, A., 1972, to be published.
0ster, F.f Risd Report No 191, 1969.
Page 12
10.
Captions to figures
Fig. 1. The puffatron device. The length of the vacuum vessel is
2.5 m. The magnetic field is in the axial direction and has a
mirror ratio of 1.5.
Fig. 2. Discharge current in linear and logarithmic display at
different magnetic field strengths and condenser bank charged
to HO kV and a pressure p=50 kP. Curve with upper partto the left is proportional
to the current. Increasing current gives a downward trace,
with sensitivity 10 A/div. Curve with lower trace to the left
is proportional to the logarithm of the current, above 1 ampere
Time scale is 5 usec/div.
a) Magnetic field strength B = 0.65 Tesla
b) B = 0.70 Tesla,
c) B = 1.20 Tesla.
The points a»b,c are indicated on fig. 3.
Fig. 3. The two distinct discharge regions experimentally
obtained, for a pressure in the small chamber, behind the
puffvalve of 50 kP(kilopascal). The cutoff condition for
2protons according to the formula V = 53 B is indicated. The
points a,b,c refer to the oscillograms of Fig. 2.
Page 13
Pulse line
Slots
Spring arrangementHammer coil
Makrolon rodInsulator
Gas inlet
Pyrex insulator
Outer electrode (diameter 15cm)Inner electrode (diameter 23cm)
Diamagnetic loop
Thickening (diameter 5cro)
Valve openings
Coil
Extension of the inner electrode(diameter 1 cm)
nriii£orp- Sublimationion pump
pumps
k.it
Fig.1
Page 15
cx
CURRENT
DISCHARGE
CURRENT
DISCHARGE
Cutoff conditionfor protons
0.5 1.0 B [Testa]Fig. 3
Page 16
TRITA-EPP-74-06
Royal Institute of Technology, Department of Plasma Physics
and Fusion Research, Stockholm, Sweden
BREAKDOWN AND PLASMA FORMATION IN A ROTATING PLASMA DEVICE
B. Bonnevier and A.H. Sillesen, March 1974, 10 p. in English
Breakdown and formation of a plasma with high ion temperature
have been studied in the Puffatron at Risd. The conditions for
plasma formation is understandable along the same lines as
breakdown in an ordinary crossed field discharge, Penning,
PIG or magnetron discharge where ions are not magnetically
confined, as in the present experiment. The growth of ioni-
zation occurs much faster than expected from ionization by
thermal electrons. This has earlier been seen in some rotating
plasma experiments. It is suggested that work along the
present lines can result in a time independant arc discharge,
where ions have thermonuclear energies.
Key words Crossed field discharge, Magnetron discharge,
Rotating plasma, High voltage effects, Breakdown, Critical
velocity phenomenon, Plasma formation, Plasma heating,
Thermonuclear fusion.