Professional Documents
Culture Documents
Electronplasmaoscillations PDF
Electronplasmaoscillations PDF
Gottfried Wehner
Reuse of AIP Publishing content is subject to the terms at: https://publishing.aip.org/authors/rights-and-permissions. Download to IP: 132.248.182.1 On:
Thu, 19 May 2016 18:47:37
JOURNAL OF APPLIED PHYSICS VOLUME 22. NUMBER 6 JUNE, 1951
. Electron plasma oscillations are excited by a beam of fast electrons in a stabilized low-pressure mercury
discharge. Probe measurements reveal that the uhf fields are localized in thin layers the plasma density
and frequency of whlch follow Langmuir's law. The beam of fast electrons traversin~ such an oscillation
l~yer b.ec?mes velocit~ modulated, and excitation conditions result from drift time and bunching considera-
tions sHnilar to those ill a klystron. A sealed-off tube described covers a frequency range between 800 and
4000 Mc (five modes) without changing or matching any resonance circuit.
Reuse of AIP Publishing content is subject to the terms at: https://publishing.aip.org/authors/rights-and-permissions. Download to IP: 132.248.182.1 On:
Thu, 19 May 2016 18:47:37
ELECTRON PLASMA OSCILLATIONS 763
(1) The density is not constant along the axis of TABLE I. Plasma density measurements in the oscillation layer.
the anode space but decreases from the center towards
U. I. b a' Niemi }. meas. d }.cale
grid and repeller.
(2) The density increases with the anode voltage 270 0.05 0.6 1.1 X 1010 32 31.8
owing to the repeated reflections of the beam between 320 0.Q7 0.57 1.5XlOlo 29 27.2
400 0.1 0.5 2.1XlO'0 25.5 23
grid and repeller.
(3) The density depends very much on the repeller
U. = anode voltage in volts,
voltage which con trolls the number of fast electrons b I tJ =: anode current in amp .
reflected or absorbed, or the number of secondary a = distance between grid and oscillation layer in em.
d}. = wavelength in em.
electrons released at the repeller.
Some results of plasma density measurements right in where j+ is the space charge limited ion current density
the oscillation layer are shown in Table I. In the last two entering the ion sheath, ~0=1/36dOll, and M=mass
columns, the free space wavelength measured with a of the mercury atom. In one case, for instance, the grid
frequency meter is compared to the wavelength calcu- was U = 1000 volts negative to the plasma and the ion
lated with the measured plasma density N from sheath thickness was 0= 1 em, so thatj+= 1.24X ID-'A/
Langmuir's law f= (e2N /1rm)!. The agreement is so cm2 Figure 7 shows, for this case, the potential ('Po)
and field (Eo) distribution in and near the ion sheath.
o"0 Let us now assume that a layer of thickness .6x of
3 plasma electrons adjoining the ion sheath is shifted
4 by .6x in the direction of the plasma, leaving a layer of
",,- --........ ~200 ions behind, as shown schematically in the lower part
,3 I -
of Fig. 6, and investigate the field and potential change.
The oscillation frequency in the above case was 1000 Me,
and this corresponds to a plasma density of N = 1010/cm3
V
~ 1"100
The oscillation amplitude was about U ~= SO volt (from
Fig. 3), and a potential change of 2U ~= 100 volts
---
E yielded, with the poisson equation, a .6x value of about
~ 2
Z
/' 10 50 0.075 cm. Figure 7 shows the field El and potential 'PI
distribution for this case. The potential notch spreads
up to 0.2 em into the ion sheath, and the shifted elec-
I trons, of course, have the tendency to move back and
fill it up. In the potential minimum, the plasma elec-
trons have velocities higher than those electron veloci-
ties which belong to the maximum of the optical
o
o 5 10 15 mm 20 excitation curve; hence, the oscillation layer is darker
- DISTANCE FROM GRID
than the surrounding plasma. The characteristically
FIG. 5. Plasma density distribution along the axis in the colored seam, especially on the side of the ion sheath,
anode space. shows excitation by very slow electrons which are just
reversing their path.
that it may now be justified to consider these oscillations
These very simplified considerations are in no way
as "electron plasma oscillations."
complete or exact; however, they show that the visible
The well-known assumption concerning these plasma
oscillations is to suppose a perturbation of the quasi- discharge picture with oscillation layers can be ex-
neutrality of the plasma and to consider the plasma plained along these lines, and the calculated thickness
electrons oscillating about the stationary ions. 6 The of oscillation layer and seam are not in contradiction to
density distribution along the anode space in connection the observed data.
with the frequency-density law suggests, by itself, that
for one frequency, such oscillations can occur only in
one or two layers. In all our experiments, such layers UNOfSTURBED
were always detected near the plasma boundary, and it CASE
j+= (4/9)~0(2e/M)!(Ut/02).
6 L. Tonks and J. Langmuir, Phys. Rev. 33, 195 (1929). FIG. 6. Schematic picture of the oscillation layer.
Reuse of AIP Publishing content is subject to the terms at: https://publishing.aip.org/authors/rights-and-permissions. Download to IP: 132.248.182.1 On:
Thu, 19 May 2016 18:47:37
764 G.DTTFRIED WEHNER
~
C)
Anode current
Wavelength
0.16 amp
20cm
~ ~ Distance from grid to repeller 2cm
Distance from oscillation layer to grid 0.4 cm
1/) Modulation degree
Calculated focus distance F
t
1.8 cm.
J [L
V. TRANSIT TIME CONSIDERATIONS
~V / V Transit time considerations provide a much sharper
P ASM "'"
7: r"
~ ~ ELE TROI~
excitation condition than these bunching considerations.
,. 11-", ~ 'I, When the beam velocity (U a)! is plotted against j, re-
sults like those in Fig. 8, with several oscillation modes,
~ ~ . are obtained. In each mode the beam velocity is
proportional to the frequency and the proportionality
\ factor decreases (as experiments show) with the grid
50 1000 repeller distance. The proportionality factor between
U a l / 2 and j may be calculated from the following con-
/IX
10-4 X
siderations. Only those electrons which pass the first
oscillation layer when the field increases compose the
bunch. Hence, the electrons of such a bunch accumulate
around the electrons which pass the oscillation layer
FIG. 7. Field and potential distribution near t~e edge and
in the ion sheath when the discharge oscillates. when the field rises through zero. These electrons may
be called "reference electrons." The bunched beam
IV. BUNCHING OF THE BEAM ELECTRONS delivers energy to an rf field only when this field has
such a phase that the bunch is retarded. This considera-
No considerations have yet been made concerning the tion gives the wanted equation between transit time
mechanism which is responsible for the excitation of (which is a function of uat) and j, and there are two
these oscillations. It is obvious that the electron beam is excitation possibilities:
necessary for this purpose, and the observed velocity (1) Reflection.-The transit time tl of the reference
modulation of the beam points to considerations well electrons, from oscillation layer to repeller and back,
known from klystrons. Provided the mean free path
equals T/4X(4n+1), T=l/j and n=l, 2, 3 . . . , the
of the beam electrons is large, the beam leaving the same excitation conditions as in a reflex klystron.
oscillation layer and passing a certain drift space will (2) No reflection.-The bunched electron beam de-
also be density modulated. The distance F between
livers its energy in the second oscillation layer near the
oscillation layer and the place of maximal bunching
repeller. Considering the higher plasma density and
is given by F=3.68.6.107Uot/21rf~, w.here Uo=:,beam resonance frequency of the plasma in between these
velocity and ~=U~/Uo(U~=osClllatlOn amphtude).
two layers, it may be assumed that this plasma block
A resonance system (resonance frequency j) near this
is free of high frequency fields and oscillates like the
place should be able to transform kinetic beam energy
drift space electrode in the Heil generator. 6 The transit
into uhf energy. The repeller is of so much importance
time t2 of the reference electrons between these two
in getting these strong oscillations because it either
layers equals, in this case, T/4X(4n+1).
provides this resonance system in. form of a sec~nd
oscillation layer near the edge of the IOn sheath covermg
the repeller or it causes the beam to be reversed so that
the grid oscillation layer is "buncher" and "catcher,"
as in a reflex klystron. Reflection may take place also
when the repeller is positive to cathode but still nega-
tive to the plasma. In this case, the reflected beam
electrons are actually secondary electrons released at
the repeller.
Neither the calculation nor the measurement of F
with a movable probe can be done with much accuracy.
GRAPHITE GRID
The calculation cannot be made with much accuracy (6 HOLES
because the measurement of ~ is very inexact. On the PER CENTIMETER,
HOLE DIAMETER-
other hand, the density modulation as a function of the 1.4 MM)
distance has only a very flat maximum. The following
example shows that the calculated F is at least of the FIG. 8. Sealed off plasma oscillator.
order of magnitude of the electrode distances. 6 A. Arsenjewa Heil and O. Heil, Z. Physik 95, 753 (1935).
Reuse of AIP Publishing content is subject to the terms at: https://publishing.aip.org/authors/rights-and-permissions. Download to IP: 132.248.182.1 On:
Thu, 19 May 2016 18:47:37
ELECTRON PLASMA OSCILLATIONS 765
=i/' (4n+l),
t2 FIG. 9. Beam-velocity vs frequency diagram of the
j= 9000 Nl, plasma oscillator.
L=a-2d (Fig. 9),
d=550 Uo"/Nl/.8 anode current and may be found from the voltage
L=610 7 U oL t 2, current characteristic of the repeller. It was here in
d= thickness of ion sheath, the order of 10 to 100 rna. The vapor pressure limits
a= distance between grid and repeller, and are mostly given by the conditions for discharge stabili-
L=distance between the oscillation layers. zation. Within this range, the pressure is not very
From these, the following equation is derived. critical.
The repeller should be exactly parallel to the grid,
(n+i)610 7 U ot+10 7 Uof=a-j. otherwise output and efficiency become very poor.
For n= 1,2,3, and 4 with a=O, 6, the curves of Fig. 9, Sputtering of electrodes by ion bombardment is very
which are in good agreement with the measured points, rapid because both the voltage drop between plasma
result. and grid or repeller and the plasma density are com-
VI. PRACTICAL CONSIDERATIONS
paratively high. To cut down sputtering as much as
possible, the grid and repeller are made of graphite.
The output was measured by photometric means The sealed off mercury tube, without changing or
using small tungsten wire lamps. Maximal output was matching any resonance circuit, merely by changing
about 4 watts at 15 em wavelength, with an efficiency of anode voltage and current, covers a frequency range
4 percent. This small efficiency is obviously caused by between 800 to 4000 me. Oscillations started, in this
the small high frequency amplitudes ({3 was never case, even with 25 volts anode voltage.
found to be more than i). With a distance between
grid and repeller of 0.4 cm, the shortest detected wave- ACKNOWLEDGMENTS
length was 6.5 cm. In this case it becomes difficult to
The author wishes to express his thanks to his associ-
maintain the high required plasma density because the
ates in the Components and Systems Laboratory,
grid and repeller, being so close together, cause too great
Electron Tube Unit, who gave help in many respects,
a loss of ions.
The beam current is not to be identified with the to his supervisors, especially to Mr. Robert Cherpeski,
who made this research possible; to the Late C. Ohl who
7 G. Wehner, Jahrb. 1942 der deutschen Luftfahrtforschung
III, 24 (1942). skillfully built the sealed off tubes; and to Emil Benz for
8 H. Fetz, Ann. Physik 40, 579 (1941). his excellent job in glass work.
Reuse of AIP Publishing content is subject to the terms at: https://publishing.aip.org/authors/rights-and-permissions. Download to IP: 132.248.182.1 On:
Thu, 19 May 2016 18:47:37