Download Electron emission from lead–zirconate–titanate ceramics

Electron emission from lead–zirconate–titanate ceramics
D. Flechtner, C. Golkowski, J. D. Ivers, G. S. Kerslick, and J. A. Nationa)
Laboratory of Plasma Studies and School of Electrical Engineering, Cornell University, Ithaca,
New York 14853
L. Schächter
Department of Electrical Engineering, Technion—Israel Institute of Technology, Haifa 32000, Israel
~Received 16 August 1996; accepted for publication 1 October 1997!
We report extensive experimental data on electron emission from lead–zirconate–titanate
ferroelectric ceramics. A 1–2 MV/m pulse is applied to a gridded ferroelectric cathode and diode
currents of up to 120 A/cm2 are measured across an A – K gap of 531022 m, with the anode at 35
kV. Both the current and the anode voltage pulse duration may extend to several microseconds. The
measurements extend previously reported data by nearly two orders of magnitude in the diode
voltage and by a factor of more than 3 in the diode spacing. Two major regimes of operation were
identified. In the first ;1 m s the ferroelectric cathode controls the electron flow through the diode.
Beyond this time plasma effects dominate the current flow. The results are of importance to the
development of novel cathodes for high current electron beam generation. © 1998 American
Institute of Physics. @S0021-8979~98!06401-9#
I. INTRODUCTION
Copious electron emission from a ferroelectric was first
reported by Gundel and Riege1 in 1989. The electron emission is based on the change in the polarization of a prepoled
ferroelectric sample. In contrast to ferromagnetic materials
which exhibit an external magnetic field, charge layers develop on any exposed surface of ferroelectrics, screening the
remnant polarization ( P r ;0.1– 0.5 C/m2! and substantially
reducing any external field outside the ceramic. In what follows we shall refer to this as the equilibrium state. The emission of electrons is triggered by pulsing the ferroelectric with
a rapidly rising electrical pulse providing an applied field
comparable to the coercive field (;13106 V/m). The rapid
change in material polarization leads to a surplus of free
electrons on the gridded front surface of the ferroelectric, a
fraction of which is repelled into the diode gap, enabling the
diode current flow. When the polarization state of the material is altered by external means ~e.g., by an applied electric
field! the ceramic is considered to be off-equilibrium. A variant on this basic concept include the use of rapidly rising
electrical pulses to switch lead–zirconate–titanate ~PLZT!,
from the antiferroelectric to the ferroelectric phase.1 More
recently, various groups have reported experimental measurements of emission at current densities on the order of
106 A/m2 ~Refs. 2–4! and theoretical explanations for the
high emission current densities have been proposed.5–8
The use of ferroelectric materials as advanced cathode
sources is appealing as they do not easily suffer from atmospheric contamination, operate at room temperature,1 are
reprateable,9 and have a high enough brightness
(109 – 1011 A/m2 rad2) 10–12 to be useful as electron beam
sources for a variety of microwave devices.1,2,13 Many other
applications have been proposed and are currently being
studied, including: triggers for low pressure switches;14 thin
film applications for flat panel display technology;15,16 and
a!
Electronic mail: [email protected]
J. Appl. Phys. 83 (2), 15 January 1998
trigger sources for plasma lighting devices. In the remainder
of this introduction we shall briefly describe the system under consideration, and outline the new results obtained.
In this article we present data describing the performance of diodes with ferroelectric cathodes operating at
1–50 kV and with diode spacings of up to 531022 m.
These data extend the parameter range previously reported
by factors of 50 in the voltage and more than 3 in the diode
spacing. We show, consistent with our earlier results,2 that
even at high voltages ~up to 50 kV! the current–voltage
(I – V) characteristic is, within a good approximation, linear.
We also report on the effects of delaying the application of
the diode voltage up to ;2.5 m s after triggering the ferroelectric source. We found that the anode current is strongly
dependent on the time delay between the ferroelectric and
anode voltage pulses. With short delays the beam emission is
found to be controlled by the ferroelectric, whereas with long
delays plasma fills the diode and dominates the current flow.
The actual time to onset of plasma dominated flow will depend, of course, on a number of parameters including the
diode spacing, and the amplitude and rise time of the trigger
pulse on the ferroelectric. These observations, which shed
light on the physics of the emission process, are combined
with a detailed examination of the diode current rise time,
amplitude, and even reversal, to ascertain the respective roles
of plasma formation in the diode compared to ferroelectric
processes in determining the current generated.
It should be noted that the results presented here apply
only to commercial ~Transducer Products Inc. or American
Piezo Ceramics! PZT 55/45 @ Pb~Zr0.55Ti0.45!O3# and to a
situation where the emission is enabled by the rapid pulsing
of the material around a minor hysteresis loop.17 No attempt
has been made to investigate the emission associated with
antiferroelectric to ferroelectric phase transitions, such as
that used in the work by Riege et al. In addition, no attempt
has been made to interpret the phenomena presented here in
terms of the microscopic processes which occur in ferroelec-
0021-8979/98/83(2)/955/7/$15.00
© 1998 American Institute of Physics
955
Downloaded 11 Oct 2004 to 132.68.1.29. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp
tric ceramics, but rather to present a macroscopic and device
oriented point of view.
This study has been carried out as part of an ongoing
program to develop new high current electron beam sources
for a variety of applications, especially as the driver for high
power microwave sources. The measured beam current of
;50 A at 40 kV, with a pulse duration of several hundred
nanoseconds, in a diode with a 531022 m gap convincingly
demonstrates the potential of ferroelectric sources for use in
high current electron guns. The relative role of ferroelectric
and plasma processes is important in determining the usefulness of the ferroelectric electron source to long life microwave devices and also for the development of ultrahigh
power sources where a limited cathode life may be acceptable. The duration of the emission, its lack of sensitivity to
impurities, and its uniformity is also of importance for these
applications.
The remainder of this article is organized as follows: in
Sec. II we summarize the results we have reported in the past
followed by detailed description of Sec. IV. A discussion of
the results and conclusions is presented in Sec. V.
II. SUMMARY OF PREVIOUS RESULTS
The results and the theoretical model summarized in this
section were reported in detail in Refs. 2 and 6. The system
consists of a planar diode with a ferroelectric ceramic cathode. The ferroelectric disk has two electrodes, the front electrode, facing the A – K gap, is gridded and is held at ground
potential. The back electrode is uniform and has a 1–2 kV
submicrosecond voltage pulse, V fe applied to it; the ferroelectric sample is 131023 m thick. The gridded electrode
serves as the cathode of the diode, which had A – K gaps of
<1531023 m. Diode current densities up to 1.2
3106 A/m2 were measured for anode voltages of ,1 kV and
gaps of order 131022 m. For these A – K gaps the anode
current duration is determined by the anode voltage
(<400 ns) and not by the voltage pulse, applied to the back
of the ferroelectric. The I – V characteristic of the diode is,
within a good approximation, linear. The diode impedance is
determined by the voltage applied to the nonlinear ferroelectric capacitor and by the gap spacing. If the ferroelectric is
not pulsed, the current measured on the anode is about two
orders of magnitude smaller. In the absence of a voltage
applied across the anode cathode gap electrons were detected
with energies up to a maximum of 60 eV. The Q – V curve of
the ferroelectric capacitor, determined experimentally from
the trigger current and voltage pulses, is highly nonlinear and
shows a hysteresis type character.
Section IV was explained both qualitatively and quantitatively in terms of what we call the charge redistribution
model. In this model the dynamics of the ferroelectric are
determined by a Weiss theory adapted to ferroelectric materials with the material parameters determined from static and
dynamic measurements of the ferroelectric properties.6 A direct result of the dynamics is the hysteresis characteristic of
the Q – V curve. As the ferroelectric is pulsed, its state of
equilibrium is altered and a fraction of the excess surface
charge redistributes between the grid electrodes and the
A – K gap. Electrons remain in the gap as long as the ferro956
J. Appl. Phys., Vol. 83, No. 2, 15 January 1998
FIG. 1. Experimental configuration.
electric is an off-equilibrium state. Although it was not explicitly pointed out in our original reports ~Refs. 2 and 6! the
duration of the off-equilibrium condition can be determined
from the dynamic hysteresis curve, and was found to last for
about 2–2.5 ms. This result also is suggested by the data
presented later in Sec. V.
The electrons which redistribute in the gap oscillate in a
potential well induced by the ferroelectric electron emission
in the A – K gap. With no applied diode voltage any electrons
injected into the diode gap must give rise to a potential well,
the depth of which is set by the amount of charge expelled
from the grounded surface of the ferroelectric. In zero order
no net current is associated with the electron motion. This is
consistent with our experimental observation that the diode
current with zero applied gap voltage is much less than that
found with even modest gap voltages. When an anode voltage is applied the electron motion is altered and a net current
flows with, as expected following a perturbation of the equilibrium, a current proportional to the anode voltage. The linear I – V characteristic is determined by the amount of charge
~originating from the surface of the ferroelectric! in the gap
and it lasts as long as the ferroelectric is off-equilibrium.
Although this model does not seek to account for the microscopic processes at the surface of the ferroelectric, it explains
virtually all the experimental results listed above.
III. EXPERIMENTAL SETUP
In this article we report the results of a detailed experimental study of ferroelectric emission from a PZT cathode in
a ‘‘planar’’ diode geometry and, in the later parts of the
article, with the anode replaced by a cylindrical drift tube
immersed in an axial magnetic field. In this case the beam is
intercepted by a Faraday cup located well downstream of the
anode, and the anode does not intercept the beam current.
The basic setup used in our studies is illustrated in Fig. 1.
The front face of the 2.5431022 m ~1 in.! diameter ferroelectric sample has a 2.731023 m (1/8 in.) annular silver
grounding ring along the outer edge and a silver grid pattern
consisting of 200 mm lines on the remaining 1.931022 m
(3/4 in.) diameter. The other face has a centered 1.9
31022 m (3/4 in.) diameter silver electrode; the sample’s
thickness is 131023 m. The diode has an adjustable gap
(1 – 5031023 m), and the back electrode of the ferroelectric
is pulsed with a positive or negative 1–3 kV, 200 ns pulse,
V fe . The magnitude of this pulse is typically of the order of
Flechtner et al.
Downloaded 11 Oct 2004 to 132.68.1.29. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp
FIG. 2. Wave forms for 3 cm gap operation, showing ferroelectric current
(I fe), ferroelectric voltage (V fe), diode current (I an), and diode voltage
(V an).
or greater than that required to overcome the coercive force
of the ferroelectric sample; in our samples the coercive field
is ;1 MV/m.
An anode potential is applied via a length of charged
transmission line (Z 0 550 V), or a Blumlein ~14 kV, 300
ns, and Z 0 ;20 V!. Typically, we measure up to four parameters in any experiment: the voltage applied to the rear
electrode (V fe), the current through the ferroelectric (I fe),
the current flowing through the A – K gap (I an), and the voltage applied to the anode (V an). The entire system was maintained at a vacuum of 231024 Torr.
IV. EXPERIMENTAL RESULTS
A. Emitting surface
The silver grid is sputtered onto the ferroelectric in a thin
(<2 m m) layer, which is subsequently etched to a thickness
of ,0.5 m m. Profilometer measurements show that after this
process the surface roughness of the ferroelectric is comparable to the thickness of the metallic strips; the electrical
resistance of the strips increases as a result of the etching,
indicating a significant loss of electrical continuity in the
grid. Consistent emission from the cathode is observed after
the etching. A similar enhancement of the emission characteristics with thin silver layers has been noted elsewhere.19 A
typical set of measured characteristics as described above is
shown in Fig. 2. The ferroelectric current, I fe , is typically
about 40 A, considerably less than the ;100 A measured
prior to the grid etching. This is consistent with the reduction
in the capacitance of the ferroelectric as a result of the reduced continuity in the conducting grid. The fact that the
current does not return to the higher pre-etching level suggests that any plasma which develops does not cover the
complete surface of the ferroelectric, at least not during the
time the voltage pulse is applied to the sample. This result
differs from that reported by Shur et al.,20 who found that the
current through the ferroelectric increased to a value determined by the complete area of the ferroelectric surface and
from this they computed a plasma transverse velocity across
the ferroelectric surface.
J. Appl. Phys., Vol. 83, No. 2, 15 January 1998
FIG. 3. I – V characteristic for 3 and 5 cm gaps at 150 ns delay between
application of V fe and V an .
B. High anode voltage operation
We have extended our previous scaling results2 for the
diode current as a function of the anode voltage from <1 up
to 50 kV and from gap spacings of 331023 – 5031023 m.
The Blumlein pulser, which has a pulse duration of 300 ns
mentioned above was modified for use in these experiments
by inserting a 4:1 step up transformer between the pulser and
the diode. The source impedance is now ;320 V. Figure 3
shows the results of these measurements. The I – V characteristics are still approximately linear although there is a suggestion of current saturation in at least the 331022 m gap
data. The ‘‘resistances’’ of the gap for the 331022 and 5
31022 m spacings are about 300 V and 1000 V, respectively. For comparison, the theoretical model referenced
earlier6 predicts a typical impedance of ;250 V in the 3
31022 m case and ;900 V in the other case. The data
reported were obtained with ;150 ns delay between the
ferroelectric pulse and the anode voltage. The current wave
forms are quite flat and repeatable especially, as expected,
when the gap resistance is comparable to the source impedance, i.e., in the 331022 m gap case. The magnitude of the
current density exceeds that expected on the basis of the
Child–Langmuir law with zero emission velocity of the
electrons—exactly as in the low voltage case.2 Note that the
model’s predictions gives good agreement with experiment
over three orders of magnitude in the diode voltage and for
gap spacings from ;331023 to 531022 m. It is important
to note that the low observed emission energy of electrons
(<60 eV) cannot account for the current ‘‘excess’’ nor can
the plasma closure concept suggested by Mesyatz8 since, in
our geometry, the velocity of the plasma closure would be on
the order of 106 m/s in order to describe the measurements.
This value is much higher than that found in other pulse
power experiments and is thought to be unlikely to occur.
C. Delayed gap voltage
By using a Blumlein, instead of a direct current charged
cable, to apply the anode–cathode voltage, we can control
the delay between the start of the pulse applied to the ferroelectric and the start of the anode voltage pulse. The delay
Flechtner et al.
957
Downloaded 11 Oct 2004 to 132.68.1.29. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp
FIG. 5. Gap current as a function of delay between V fe and V an at 14 kV, 5
cm.
tance increases and, as indicated, the current drops. Results
displaying a similar time dependence have been obtained at
anode–cathode voltages of ;35 and 50 kV and agree with
the charge redistribution model predictions.
FIG. 4. Anode voltage ~top trace! and gap current ~bottom trace! for two
delays ~upper: 165 ns, lower: 1.2 ms!.
was varied between 0< t <2.5 m s. Figure 4 illustrates the
anode voltage and current for two different delays: 165 ns
and 1.2 ms. The average diode currents range from somewhat
,12 A in the first case to 50 A for the longer delay. Simulations performed with the charge redistribution model indicated that the current will drop for delays longer than 1 ms
and will vanish at 2.2 ms. Note that the parameters of the
ceramic used in this simulation are identical with these used
in Ref. 6 to simulate the performance when both voltages
~V fe , V an! were applied simultaneously. Experiments confirmed the ‘‘cutoff’’ in the diode current beyond delays of
2.5 ms. Figure 5 shows the anode current for various delays
at an anode voltage of ;14 kV. From the simulation ~see
solid line in Fig. 5! we found that the peak value of the
delayed anode current occurs when the voltage across the
ferroelectric has dropped to a value close to zero. A closer
look, using the simulation model, at the charge transfer
through the ferroelectric, and hence the charge within the
diode gap, indicates that it is still significant at this time as a
result of the hysteresis in the material Q – V characteristics,
i.e., the ferroelectric sample is still off-equilibrium. Measurements of the dynamic hysteresis loop using the current flow
through the ferroelectric and the voltage pulse used to trigger
the emission show that after (;1 m s) the ferroelectric has
not returned to its original state, and imply that the surface
charge is still about 10%–20%, i.e., ;(1 – 2)31022 C/m2,
of that required to screen the polarization field when the
ferroelectric is in equilibrium. This feature is unique to ferroelectrics and does not occur in linear dielectrics. Beyond
this time the charge in the gap diminishes, the diode resis958
J. Appl. Phys., Vol. 83, No. 2, 15 January 1998
D. Long pulse operation
A charged transmission line (Z 0 550 V) has been used
to provide an anode cathode voltage pulse of up to 5 ms in
duration. Typical results, using a ;3- m s-long cable, are illustrated in Fig. 6. Under these conditions it is clear that the
diode current starts during the ferroelectric pulse, drops to
approximately zero as the ferroelectric pulse ends, but then
rises to ;8 A and stays relatively constant for the round trip
time of the cable. This result illustrates clearly what we indicated earlier, namely that there are two phenomena which
control the diode current, namely for times when the ferroelectric capacitor is off-equilibrium the flow is controlled by
the change in the surface charge initially on the ferroelectric,
some of which is redistributed in the diode gap. At later
FIG. 6. Diode current for V an5500 V and A – K gap of 2 cm. Top trace: V fe
~1.3 kV/div!; middle trace: I fe ~37 A/div!; bottom trace: I an ~10 A/div!.
Flechtner et al.
Downloaded 11 Oct 2004 to 132.68.1.29. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp
FIG. 7. Anode current showing onset of reverse current flow in the diode for
2 cm gap, V an52 kV, 5 ms cable.
FIG. 8. Electron energy as a function of switching voltage magnitude and
polarity.
times as the ferroelectric returns to equilibrium, the plasma
generated when the ferroelectric was pulsed, takes over and
controls the flow.
An additional indication of the presence of a plasma in
the gap is revealed by the observation of a reverse current in
the diode as illustrated in Fig. 7. After the voltage wave of
the initial ~diode! current pulse has completed a round trip on
the cable the wave reflects from the diode end, and under
certain gap conditions, the diode voltage reverses producing
a reverse current. In this case a bright spot is detected on
what was previously the anode, but which now acts as a
cathode for the reverse current flow. The emission in this
case gives rise to current levels of the same order of magnitude as measured during the first diode pulse. Such current
levels can only be supported by a plasma in the gap region,
since the ferroelectric has returned to equilibrium by this
time (t>6 m s). A ‘‘cathode spot’’ has formed, where the
light emission occurred and electrically connects the plasma
in the gap with the graphite cathode. The time delay between
the initial diode current and the start of reverse current flow,
under similar conditions ~5 ms cable, 2 cm A – K gap, V an52
kV!, has a Gaussian distribution centered at 8 ms, with a
standard deviation of 61 m s. The scatter in the onset of
reverse current flow is associated with the time for surface
breakdown of the original anode ~now the cathode! and the
establishment of a connection between the plasma fill and the
cold electrode. For a higher diode voltage of 3 kV the average time drops to ;5 m s. A reduction in the cable length to
3 ms shows reverse current flow starting at about 4 ms. Based
on these data we conclude that plasma closure is completed
before the 5 ms round trip transit time of the cable. Measurements of this type may be used to establish an upper limit on
the plasma closure velocity which we assess as ;2 – 3
3104 m/s.
While showing the importance of plasma effects the long
pulse experiments also provides solid evidence for the ferroelectric role in the diode current emission process. For a 5
31022 m gap and assuming typical plasma closure velocities of 23104 m/s any plasma effects on diode current flow
should be significant only at times longer than 2.5 ms. However in all the relevant data the diode current starts ,50 ns
J. Appl. Phys., Vol. 83, No. 2, 15 January 1998
after the peak of the ferroelectric voltage pulse and, in addition the rise time of the diode current pulse is essentially
independent of the gap size. In all cases the current level
exceeds that expected for space charge limited current flow
with zero initial electron velocity. Plasma velocities of
;106 m/s would be required to account for these observations. We assert that this velocity is too high for a plasma
closure process. This statement is based on a wealth of observational data well familiar to the pulse power community.
Typical plasma formation and closure processes were reported in the 1960s and 1070s by Bugaev and also by Parker
et al.21,22 It appears that while plasma in the gap controls
diode current flow at late times (t>20 m s) the initiation and
buildup of this current is controlled by the ferroelectric.
E. Dependence of electron energy on the ferroelectric
voltage
Several experimenters2,15 have reported pulses of energetic electrons crossing the diode gap when a voltage is applied to a ferroelectric. In our previous experiments this was
accomplished using a 1–2 kV, 200 ns full width at half
maximum positive pulse applied to the rear surface of a
1023 m thickness sample. In this case our results showed
emitted electrons with energies of up to 60 eV. This energy
was typically an order of magnitude less than that reported
by other groups. We have measured the emitted electron energy as a function of both applied voltage magnitude and
polarity and the results are shown in Fig. 8. For a positive 1.8
kV applied pulse, an anode voltage of ;2100 V was required to completely stop the flow of electrons across the
231022 m A – K gap. For a negative 1.8 kV applied pulse,
we needed ;2800 V to stop the electrons. These values
vary by <20% from sample to sample. The results show
also a linear increase of the maximum emitted electron energy with increasing voltage applied to the ferroelectric. The
electron energy dependence on the polarity of the applied
Flechtner et al.
959
Downloaded 11 Oct 2004 to 132.68.1.29. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp
switching pulse is a new observation, and may explain the
discrepancies between the results reported by different
groups.
The long pulse results show that plasma effects play a
crucial role in diode current flow on time scales >1 – 2 m s.
The presence of energetic electrons following the application
of the ferroelectric pulse, provides strong evidence that the
properties of the ferroelectric dominate the current flow on
short time scales. Note that the emission of energetic electrons cannot be described in terms of either the formation of
a plasma or its expansion into the diode gap.
V. CONCLUSIONS
Our cathode experiments2,6,12 use a commercial PZT ceramic ~LTZ-2! with a (55/45) zirconate–titanate distribution
which is near the tetragonal/rhombohedral ferroelectric phase
transition. Many other ferroelectric materials are also known
to emit. Lathanum doped compositions with high zirconium
concentrations ~e.g., PLZTx/65/35 or 2/95/5! are often used
as electron emitters.9,13,20 Some of these compositions lie
near regions of mixed phases, and such materials can cross
an antiferroelectric/ferroelectric phase boundary when pulsed
by fields of ;2.5 MV/m. PZT (55/45) remains ferroelectric
throughout the emission process and the electron emission
arises from the rapid changing of the ferroelectric state by
causing it to ‘‘swing’’ around a minor hysteresis loop. Our
results, comments, and conclusions are therefore restricted to
this regime and material.
The results presented above show that the current flow in
the diode is controlled by two different mechanisms. For
times when the ferroelectric capacitor is in an offequilibrium state, as measured from the dynamic hysteresis
loop associated with the voltage applied to the sample, the
flow is controlled by the charge, initially on the surface of
the ferroelectric, some of which is redistributed in the diode
gap. As the ferroelectric returns to its equilibrium state, the
plasma generated by electron emission from triple points
when the ferroelectric is pulsed,5,8 takes over and controls
the flow.
Electrons and plasma are generated in the first instance
because the large dielectric coefficient of the ferroelectric
capacitor causes very high surface electric fields to develop
at triple points. Electrons ejected from the ferroelectric can
dominate the current flow for short (;300 ns) pulses with
delays of up to 2–2.5 ms, the actual value depending on the
anode cathode spacing. For long pulses additional plasma
may be generated at the cathode by processes such as whisker evaporation and subsequent ionization. This leads to a
more rapid increase in the onset of plasma effects in the
diode current. It is not clear from the available data where
the exact boundary between the two regimes lies; it seems
most probable that at least the first ;1 m s of current flow in
a 531022 m gap is controlled by the ferroelectric charge
redistribution and that for longer times the flow of current in
the diode gap is facilitated by the presence of plasma, i.e., we
identify the most probable plasma closure velocity as ;2
3104 m/s.
During the period of time the current flow is dominated
by the ferroelectric sample the generation of electrons is con960
J. Appl. Phys., Vol. 83, No. 2, 15 January 1998
trolled by the pulse applied to the back of the ferroelectric
sample. This is in contrast to the usual field emission process
where the emission is induced by the voltage applied to the
A – K gap. In what follows we summarize the new results
obtained, and reported in this study, for a diode with a PZT
ferroelectric cathode:
~i! The anode current duration is, for a wide range of
A – K gaps (<531022 m), determined by the anode voltage
and not by the ferroelectric voltage. For typical times smaller
than the gap plasma closure time ~gap length/23104 m/s! the
behavior is controlled by the ferroelectric. Beyond this time
the behavior is controlled by plasma processes.
~ii! Current densities of up to 1.23106 A/m2 were measured for voltages of up to 50 kV and with gaps of order 5
31022 m. These current densities have been reported previously both by us and by others, however the pulse durations
achieved in these experiments are substantially longer than
any reported previously with high current densities. The long
high current density pulses are of relevance to the design of
high current electron guns for microwave source development. It should be noted that the emittance of these beams
has been measured, albeit at lower current densities, and is
well within the limits required for sources in the L – X bands.
~iii! The I – V characteristic of the diode is within a good
approximation linear. This result was reported previously for
low voltage (<1 kV) and for short diode gaps (<1.5
31022 m). The results reported here show the result still
holds good at 50 kV and with gaps of up to 531022 m.
During the ferroelectric control regime the gap impedance is
determined by the voltage applied to the ferroelectric capacitor and the gap width. At late times where the plasma effects
are dominant the gap impedances measured are significantly
lower. The energy of electrons emitted by the cathode in the
absence of an anode voltage depends on the polarity of the
voltage applied on the ferroelectric. In the case of a positive
voltage the energy is on the order of or ,100 eV and in the
case of negative pulse the energy is 800 eV. The energy
dependence on the switching polarity is a new result which
may help explain differences reported in the literature between the work of different groups.
~iv! The Q – V characteristic of the ferroelectric capacitor
is highly nonlinear during the triggering pulse duration and
follows a hysteresis like curve. The off-equilibrium state of
the ferroelectric has been shown, from earlier published data,
to last for at least 1 ms.
~v! The anode current increases with the delay between
the ferroelectric and anode voltage pulses. This increase is
limited to a delay of up to 1 ms which is also the time at
which the ferroelectric voltage drops to zero. Beyond this
time the current drops and it vanishes with delays longer
than 2.5 ms.
The results reported in this article suggest that PZT cathodes of the type examined may be useful for the generation
of high current electron beams with pulse durations of order
of 1 ms, and for longer durations with shorter life times if the
plasma formation leads to the degradation of the surface of
the cathode or a reduction in the strength of the electric fields
which can be maintained at the walls.
Flechtner et al.
Downloaded 11 Oct 2004 to 132.68.1.29. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp
ACKNOWLEDGMENTS
This work was supported by the U.S. Department of Energy and by the AFOSR under the MURI High Energy Microwave Program.
1
H. Gündel, H. Riege, E. J. N. Wilson, J. Handerek, and K. Zioutas, Nucl.
Instrum. Methods Phys. Res. A 280, 1 ~1989!.
2
J. D. Ivers, L. Schächter, J. A. Nation, G. S. Kerslick, and R. Advani, J.
Appl. Phys. 73, 2667 ~1993!.
3
A. Sh. Airapetov, A. K. Krasnykh, I. V. Levshin, and A. Yu. Nikitskii,
Sov. Tech. Phys. Lett. 35, 182 ~1990!.
4
K. Biedrzycki, H. Aboura, J. L. Chartier, and R. Le Bihan, Phys. Status
Solidi A 132, 205 ~1992!.
5
G. A. Mesyatz, Tech. Phys. Lett. 20, 8 ~1994!.
6
L. Schächter, J. D. Ivers, J. A. Nation, and G. S. Kerslick, J. Appl. Phys.
73, 8097 ~1993!.
7
G. Rosenman and I. Rez, J. Appl. Phys. 73, 1904 ~1993!.
8
V. F. Puchkarev and G. A. Mesyatz, J. Appl. Phys. 78, 5633 ~1995!.
9
H. Gündel, Integr. Ferroelectr. 2, 207 ~1992!.
10
S. E. Sampayan, G. J. Caporaso, C. L. Holmes, D. Prosnitz, D. O.
Trimble, and G. A. Westenskow, Nucl. Instrum. Methods Phys. Res. A
340, 90 ~1994!.
J. Appl. Phys., Vol. 83, No. 2, 15 January 1998
11
B. Jiang, G. Kirkman, and N. Rhinehardt, Appl. Phys. Lett. 66, 1196
~1995!.
12
D. Flechtner, J. D. Ivers, G. S. Kerslick, J. A. Nation, and L. Schächter, in
High Power Particle Beams ~San Diego, CA, 1994!, Vol. 1, p. 487.
13
T. C. Cavazos, W. L. Wilbanks, C. B. Fleddermann, and D. A. Schiffler,
Appl. Phys. Lett. 65, 2612 ~1994!.
14
H. Gündel, H. Riege, E. J. N. Wilson, J. Handerek, and K. Zioutas, Appl.
Phys. Lett. 54, 2071 ~1989!.
15
O. Auciello, M. A. Ray, D. Palmer, J. Duarte, G. E. McGuire, and D.
Temple, Appl. Phys. Lett. 66, 2183 ~1995!.
16
J. Asano, S. Iwasaki, M. Okuyama, and Y. Hamakawa, Jpn. J. Appl.
Phys., Part 1 32, 4282 ~1993!.
17
A. J. Moulson and J. M. Herbert, Electroceramics ~Chapman and Hall,
London, 1990!.
18
H. Riege, Nucl. Instrum. Methods Phys. Res. A 340, 80 ~1994!.
19
M. Okuyama, J. Asano, and Y. Hamakawa, Jpn. J. Appl. Phys., Part 1 33,
5506 ~1994!.
20
D. Shur, G. Rosenman, Ya. E. Krasik, and V. D. Kugel, J. Appl. Phys. 79,
3669 ~1996!.
21
S. P. Bugaev and G. A. Mesyatz, Sov. Phys. Tech. Phys. 12, 1358 ~1968!.
22
R. Parker, R. E. Andersen, and C. V. Duncan, J. Appl. Phys. 45, 2463
~1976!.
Flechtner et al.
961
Downloaded 11 Oct 2004 to 132.68.1.29. Redistribution subject to AIP license or copyright, see http://jap.aip.org/jap/copyright.jsp