Best Available-Copy

REPORT SD-TR.80-37 ( /

A Review of the Science and Technology

of Cathodes from the Viewpoint of
Spacecraft TWT Applications

G. W. STUPIAN
Chemistry and Physics Laboratory
Laboratory Operations
The Aerospace Corporation
El Segundo, Calif. 90245

1 June 1980

DTic
Interim Report ~ LECTE~

APPROVED FOR PUBLIC RELEASE;

DISTRIBUTION UNLIMITED

0-. Prepared for

C) SPACE DMISION
AIR FORCE SYSTEMS COMMAND
Los Angeles Air Force Station
P.O. Box 92960, Worldway Postal Center
Los Angeles, Calif. 90009

80 6 19 n51
 This interim report was submitted by The Aerospace Corporation, -
El Segundo, CA 90245, under Contract No. F04701-79-C-0080 with the
Space Division, Contracts Management Office, P.O. Box 92960, Worldway '"':
Postal Center, Los Angeles, CA 90009. It was reviewed and approved for
The Aerospace Corporation by S. Siegel, Director, Chemistry and Physics
Laboratory. Gerhard E. Aichinger was the project officer for Mission-
Oriented Investigation and Experimentation (MOIE) Programs.
This report has been reviewed by the Public Affairs Office (PAS) and
is releasable to the National Technical Information Service (NTIS). At NTIS,
it will be available to the general public, including foreign nations.

This technical report has been reviewed and is approved for publica-
tion. Publication of this report does not constitute Air Force approval of
the report's findings or conclusions. It is published only for the exchange
and stimulation of ideas.

Gerhard E. AichBnger
Project Officer

FOR THE COMMANDER

va Br
Contracts Management Office

-i -'

• o,.': ..... fl-e .A,•-

2> -u'.:...
.• .
 U NC LASS IFLIFD
ECURITY CLASSIFICATION OF THIS PAGE (ITA" 0*1k 4. ......

SD TR-B -37

WORT PAGEHTNUII~R &0UETr

Th4 c -
rai n ,C: T~op
Aop ,-/ LO /'-
AT
1~' S......... MM
E/~~~~~PR S egundoD CaifO9V4I
... • .ED

! COTOF!;N
Y IOFD|ESNM FROMADDESS • •,WPPORTO

Gary W.r Sytemsa Comman Fý. i-79-C-#tp•S p

Los Angeles, Calif. 90009 74

4-. MONITORING AGENCY NAME & AOORES.Vi dftlfereent ro Controltlng Office) 15. SECURITY CLASS. (of thie repear)

Unclassified

Approved for public release; distribution unlimited

17. DISTRIRUTION STATEMENT (of the ebstract wntered in Block 20, it different bow Report)

IS. SUPPLEMENTARY NOTES

19. KEY WORDS (Continue on reveree itde If necesseary ad identify by block number)

* Thermionic Cathodes
Therm-.onic Em~is sion
Traveling Wave Tubes

_NJABSTRACT (Continue c on a& cIte

aiede neceesary and Identify by block number)
"
• he current state of the art of cathode science and technology is sum marized,
with emphasis on the types of cathode now used, or with potential use, in
U.S. Air Force communication satellites. The intent is that this report be
a stand-alone discussion of cathode construction and of the basic physics of
electron emission. Emphasis is placed on thermionic cathodes; however, one
nontherrnionic emitter, the field emitter, is discussed briefly. For the
thermionic emitters, the barium (calcium and strontium) oxide cathode and

ORM
D-IF.IACSITIIS
R•N1
OD LP 114133
L E 1 ICS
S I
UNCLASS OIFIED (I
/=/. E
_
• SECURITY CL..AX,. ATRON OF THIS PAGt (dWhen b~let n creat)'

44 -'- - ",- r."v*" " + , .

-i ' II I
 UNCLASSIFIED
SZURITY CLASS1PICAYIC#¶ OF THItS PAOR(WbtW DeAAtok"nQ

12. KYe. WOR'Ot C•ontA•,uer1)

2 ABSTRACT (Ccnfliwad)

dispenser cathodes (based on barium) are discussed, along with variant types
of the oxide and dispenser cathodes. Thorium-based cathodes and lanthanum
hexaboride cathodes are also desciribed. Some of the available data on
cathode failure modes and cathode lifetime are included. Electron emission
from solids has been studied with varying intensity since the last century,
and the present discussion, because of the scope of the subject, of course
is incomplete. Extensive literature references are given bowever.
It has been concluded that dispenser- impregnated cathodes'&-ill have to be
employed to meet future U.S. Air Force (SD) mission requirements. Of
these cathodes, the impregnated Type B and osmnium-coated Type M cathodes
offer the highest probability of success. There is no fundamental reason why
more exotic emitters, e.g., field emitters, could not ultimately be employed
in space applications. However, the difficulty in acquiring sufficient test
data in any reasonable time to flight-qualify components based on a new tech-
nology is so great that impregnated cathodes will almost certainly be utilized.
It is also concluded that studies of impregnated cathodes in which modern
surface analysis techniques are used for submicron surface elemental analy-
sis, especially if used in conjunction with techniques for mapping cathode
work functions with high spatial resolution, would be very valuable and should
ultimately permit identification of specific cathode failure mechanisms. Ap-
plication of this information could result in a marked improvement in cathode
reliability.

nabnnounoed
Justiticat$on

Distrtl-.ut r

A.a and /or

Dist special

UNCLASSIFIED
SECURITV CL.AM•IFCATION OF T1iS PAG•(PnI DOt Ent*Oes)
 "CONTENTS

I. INTRODUCTION ................................ 5
II. FUNDAMENTAL CONCEPTS OF TI-HERMONIC
EMISSION .................................... 7

A. Ernission Equations ......................... 7

B. Cathode Design Strategy ...................... 12

Ill. ALKALINE EARTH OXIDE CATHODES .................... 15

A. Construction .............................. 15
B. Principles of Oxide Cathode Operation ............. 16

IV. BARIUM MONOLAYER CATHODES ................... 23

A. Construction .............................. 24
B. Principles of Operation of Barium
Monolayer Cathodes ......................... 27

V. OTHER MONOLAYER CATHODES ................... 39

A. Thoriated Tungsten Cathodes ................... 39

B. Lanthanum Hexaboride Cathodes ................. 40

VI. TEST PROCEDURES AND LIFE-TEST DATA

FOR THERMIONIC CATHODES. ..................... 41

VII. FIELD-EMISSION CATHODES ..................... 45

VIII. CONCLUSIONS ................................ 47

REFERENCES ...................................... 49

APPENDIXES: A. EMISSION EQUATIONS ................. 53

B. WORK FUNCTION .................... 67

-- }_-
 FIGURES

i. Energy-Level Diagrams for a Metal and for a Semiconductor . 8

I 2. Parallel Plate Diode...............................Ii

3. Energy-Level Diagrams for Band Bending in n-Type

Semiconductor ................................. 20
. . . . . . . . . . . . Z5
4. L-Dispenser Cathode .................
5. rDpregnated Cathode ............................. Z6

j6. Potetial Energy Diagrams Illustrating Additional Work-

Function Lowering Obtained with Monolayer Cathode
as Result of Increasing Substrate Work Function ............... 35

7. Stanford Research Institute Field-Emission Cathode .......... 46

A-I. a. Energy-Level Diagram for Metal and Corresponding
Density of Electron States. b. At T = 0 K.
c. At T > 0 K.. ............................... 54

A-Z. Electron Energy as Function of Distance from Surface of

Metal in Absence of Applied Field ........................ 59

A-3. Electrostatic Potential as Function of Distance from Surface

in Presence of Externally Applied Electric Field. ................ 61
B-I. Electron Double Layer Formation at Metal Surface as Result
of Spreading of the Electron Distribution ............... 68

B-2. General Behavior of Work Function as Function of Surface

Coverage for Alkali or Alkaline Earth Element Absorbed on
M etal Substrate ................................. 72

B-3. Potential Energy Diagrams ........................ 74

n

" '" "tai' PRIIQ M T f"idJ

4I,,,T• ....... V ii,=• -•h

 L INTRODUCTION

The success of U.S. Air Force comrnnunications satellite missions

depends upon the availability of traveling wave tubes (TWTs) capable of
I
delivering adequate microwave power for extended periods. Traveling
wave tubes will, not in the near future, be replaced by solid-state devices
as high-power microwave amplifiers in frequency bands of interest to the
Air Force. The satisfactory performance of a TWT is dependent upon the
proper functioning of a thermionic cathode.

It is not certain that cathodes capable of delivering adequate current

densities with lifetimes necessary to meet SD mission requirements can
be produced with the present manufacturing technology. The production of
adequate cathodes fong esended space missions is probably one of the greatest
technological challenges in Air Force satellite programs. In view of the con-
siderable amount of work devoted to the study of cathodes during the era of
ascendancy of the vacuum tube, it has been widely assumed that the fabrica-
tion of cathodes suitable for use in space-qualified TWTs presents no extra-
ordinary problems. This assumption is invalid. Tubes in ground or airborne
applications generally can be replaced on a schedule or as necessary. A
lifetime of some thousands of hours was and is often adequate. The success
or failure of a modern communications satellite still hinges upon the per-
formance of cathodes developed for use under much less exacting conditions.
Various aspects of the cathode problem have been treated in several
published reviews. 3-7 Basic and applied results particularly relevant to the
utilization of cathodes in space applications have been compiled in this report.
This document is intended to be a stand-alone discussion of cathodes, including
fundamental considerations of electron emission and practical considerations
of cathode construction and failure. The literature in the field is so extensive
that no attempt can be made to cover every aspect of cathode science and
technology in exhaustive detail. It is hoped that the.. selection of topics and

-5-
the emphasis chosen will make the document useful to those wishing to
acquire some understanding of the spacecraft cathode problem. In addition,
extensive literature references are included.

The emphasis in this report is on thermionic emission simply because

thermionic cathodes have been predominant in tube technology to date, but one
other type of cathode with potential space applications, the field emitter, is

treated briefly. Basic equations governing therrnionic emission that are

fundamnental to any description of cathode operation are given. The various
types of thermionic cathodes then are described. There are many different
kinds of therrnionic cathodes, and some logical means of classification is
essential to an orderly presentation. Cathodes now used in TWTs fall into
two general categories:alkaline earth oxide cathodes and barium monolayer
cathodes. These cathodes and their variants are dealt with in Sections III
and IV. Monolayer cathodes other than those based on barium are discussed
in Section V. Thermionic-emitter life-test data are presented in Section V1.
Field-emission cathodes are discussed in Section VII. Conclusions and recom-
mendations regarding steps necessary to ensure the future availability of
cathodes capable of meeting Air Force satellite requirements are presented
in Section VIII. More detailed discussions of the emission equations and of
the concept of the work function are included in the appendixes.
In recent years, modern techniques for surface analysis, e.g., x-ray
photoelectron spectroscopy, Auger electron spectroscopy, and ion nicro-
probe analysis, have been developed; they make possible detailed studies of
cathodes not previously feasible. Most of the earlier work on cathodes pre-
dated these experimental advances. Although the modern techniques are now
beginning to be applied to the proble.n, much work on the characterization of
cathode surfaces remains to be done. As much of the newer work as possible
is included in this review.

-6-
 I]

jI. FUNDAMENTAL CONCEPTS OF THERMIONIC EMISSION

Vacuumn-tube technology is based on the manipulation of streams of

electrons by electric and magnetic fields. The electron source is the cathode-
Cathodes used in vacuum tubes are usually thermionic, i.e., electrons are
"boiled off" by raising the temperature; however, cathodes based on other
emission processes, e.g., photoemission and secondary emission, have
been employed in specialized applications. The study of thermnionic emission
has a long history. The escape of electrons from hot filaments was first ob-
served by Edison, and the still used alkaline earth oxide cathode was invented
by Weehnelt in 1903. Considerable effort was devoted to studies of thermionic
emission until the late 1950s when the rise of the transistor curtailed the
interest in vacuum tubes and in research directly or indirectly related to
vacuum tubes. During the 1960s, there was some renewed interest in therm.-

I
ionic emission for direct conversion of heat to electrical energy, but, for
the most part, thermionic emission has been neglected in recent years.
8

Thermionic emission is analogous to the evaporation of molecules

from a liquid. In both cases, particles must be given sufficient thermal
energy to escape from the emitting medium. For an evaporating liquid, the
necessary energy is called the heat of vaporization. For thermionic emission,
the energy barrier is called the work function and is defined as the energy
required to remove an electron at the Fermi level to an infinite distance. The
relation of the work function to the energy band structure of solids is illus-
trated for a metal and for a semiconductor in Figs. Ia and 1b, respectively.
The work function is a measurable quantity and plays a key role in understanding
the thermionic emission process.

A. EMISSION EQUATIONS

I * TEMPE RATURE-LIMITED EMISSION

The electron current emitted from a surface at temperature T(K)

depends on the value of the work function 0 and, ideally, is given by the
Richardson-Dushman equation

-7-

- . .- ... .........
 I. ........... ...

-F

CONDUCTION BAND

-EF
Ed

VALENCE BAND

b.

Fig. 1. Energy-'.,o Diagrams. a. Metal.

b. n-type semiconductor. Ep, and
Ed are :e Fermi elergy and donor
level energy, respectively, measured
from the bottom of the conduction band.

-8-

...........................
 J-- ATZOekT (ia)

where 3 is the electron current density in amperes per square centimeter.5

The universal (therrionic) Richardson constant A is
2
4wmek
A = -e-- O Alcm K (Ib)

A derivation of the Richardson-DDushman equation is given in Appendix A.

The Richardson-Dushman equation can be derived with the use of very
general thermodynamic arguments that are essentially independent of
the energy-band structure of the emitter. Equation (Ia) is thereiore valid
for all types of emitters with suitable slight modifications. An example of
such a minor modification arises because electrons may, instead of being
emitted, undergo Bragg reflection back into the bulk. A reflection coefficient
"Yis introduced to account for this phenomenon, and Equation (Ia) can be
rewritten:

J = A(i - -i)T~e-t/kT (2)

The work function depends on both the bulk band structure and on the
surface dipole layers brought about by surface crystallographic structure
and by the presence of ionizable surface species. It is customary to attempt
to fit emission data froin many different types of therrnionic cathodes with
an expression of the form of Eq. (2), although a plot of In J/9T versus I iT
does not always yield a straight line over a wide temperature range (for rea-
sons to be discussed later). However, with the use of the proper expression
for 0#, Eq. (2) is a useful starting point in describing thermionic emission
from a variety of emitters.

Equation (2) yields the naximum emission current density that can be
obtained from a cathode with a given work function at temperature T and

-9-
describes the situation known as temperature-limited emission. Another
electrode (the anode) is used to collect the emitted electrons in order to
actually measure the electron emission current from a cathode. For suf-
ficiently high positive anode voltages, the emission currenat is indeed de-
pendent only on temperature and not upon anode voltage.
2. SPACE-CHARGE-LLMITED EMISSION

The cathode in a TWT cannot be operated in the temperature-limited

region because under such conditions small changes in temperature result in
substantial changes in emission current. For example, for a cathode with a
work function of 1.8 eV onerating at 1100 C, (i/5)(8J/8T)-- 1. 3%/K. Stable
tube operation is difficult to achieve when slight changes in cathode termpera-
ture cause such changes in emission current. Traveling wave tubes are
operated in a space-charge-limited mode to minimize this possible instability,
i.e., anode voltage for a particular cathode temperature is kept sufficiently
low to ensure that a charge cloud is maintained in front of the cathode. The
charge cloud rather than the cathode itself may be viewed as the emitter.
With a simple diode geometry with planar anode and cathode assumed (Fig. 2),
the current-voltage relationship for the space-charge region is easily derived
by solving the Poisson equation to yield

j=499 1J2e" 1, v3/Z(3

where L is the cathode to anode spacing, and the other symbols have their
usual meanings. This expression is known as the Langmuir-Child law. It
is necessary that O/T be small enough that the emission current, as given by
the Richardson-Dushman equation, can sustain the space-charge cloud in order
for a cathode to evidence the behavior described by the Langmuir-Child law.
The Langmuir-Child law is discussed in more detail in Appendix A.

- o
 CATHODE ANODE

I HM
SPACE CHARGE

I. Fig. 2. Parallel Plate Diode

-p.
 3. PATCH EFFECTS

In the derivation of the Richardson-Dushrnan equation, it is assumed

that the work function of a surface is uniform, which is not the situation pre-
vailing for a practical cathode. The surface of a real cathode consists of
""atches" of differing work functions. The effect of cathode inhomogeneities
depends upon the emission region in which the cathode is being operated. A

cathode in a diode test configuration that is operated in the space-charge re-

gion is considered. Cathode current is practically independent of the cathode
work function since emitted electrons are essentially drawn from the space-
charge cloud. As the anode voltage becomes more positive, the higher work
function patches reac., saturation, i. e. , emission is described by the
Richardson-Dushman equation, and the slope of the J versus V curve de-
creases. Eventually, all patches will contribute a saturated emission current,
governed by individual work functions, to the total current. A Richardson
plot, in J/T versus i/T, will yield an "effective work function"
and an
"effective Richardson constant. " These quantities can be related to param-
eters characteristic of the patch distribution. 9 Patches are directly observ-
able with the use of several different experimental techniques.

B. CATHODE DESIGN STRATEGY

Even though traveling wave tubes are operated in the space-charge-

limited region, Eq. (2) is still of primary importance since it governs elec-
tron density in the space-charge cloud. The product of the work function and
the emission current represents power that must be supplied to the cathode
to sustain emission (in addition to radiative and conductive heat losses). A
lower work function, and thus for a given emission current, a lower oper-
ating temperature implies more stable cathode operation in general and a
longer operating lifetime as well as improved thermal efficiency. Thus, all
practical thermionic cathode designs have as a basis the preparation of sur -

faces with the lowest possible work function and the highest possible emit-
ting area.

• - 12-

:P.
 There are, however, additional constraints on cathode design and
fabrication. For example, the melting point of the cathode material must be
sufficiently high to ensure structural integrity at the operating temperature.
For some terrestrial applications, resistance to air exposure (when cold) or
to occasional ion bombardment by residual tube gases as a result of accidental
glow discharge is essential. Current density requirements also vary.

Mechanical stability is particularly important for TWT applications because

of focusing requirements.
Pure metals and alloys, semiconductor materials, and systems con-
sisting of work-function-lowering surface dipole layers on metallic substrates
have all been used as thermionic emitters for various applications. Of the
pure metals, only tungsten, rhenium, and tantalum have sufficiently high
melting temperatures to yield significant emission current densities. For
2
tungsten, J = 300 mA/cm at 2500 K, and the thermal efficiency is roughly
4 to 20 mA/W. i Required operating temperatures and attainable efficiencies
are not satisfactory for spacecraft TWT applications. Thus, in the following
discussion of "practical" thermionic emitters, the &-kalineearth oxide and
dispenser cathodes (semiconductor and monolayer, respectively) will
be emphasized since these cathodes and their variant types play a dominant
role in meeting the high-current density requirements of spacecraft TWT
operation.

"_'3_
-- •° 9
 M. ALKALINE EARTH OXIDE CATHODES

The alkaline earth oxide cathode has been in general use for many
years and is widely employed in traveling wave tubes. Current densities of
300 mA/cm are attained at about 7510C. However, operation at higher tem-
peratures seriously degrades cathode life and stability. Thus, although alka-
line earth oxide cathodes should meet some present Air Force mission require-
ments, projected demands for current densities of I amp/cm 2 for 7 to WO
years on orbit will ultimately mean that other cathode types will have to be
considered for use. The oxide cathodes are described in this section, and
some possible types of cathodes for future applications are discussed.
A. CONSTRUCTION

The alkaline earth oxide cathode consists of a layer of BaO, SrO, and
CaO on a nickel support. The alkaline earths are initially deposited as a
slurry of the carbonates with an organic binder. A typical mixture of ingre-
dients results in a final product with a 5/3/2 molar ratio of Ba/Sr/Ca. The
carbonate mixture is heated in vacuum to 900 to 1150'0 to decompose the
carbonates to the oxides. The final oxide thickness is typically approximately
50 ptrm.

Immediately after carbonate decomposition, the cathode exhibits low

emission and low conductivity. For efficient operation, the cathode must be
activated. The nature of the actual processes occurring upon activation is a
matter of contention. Certain reducing elements or "activators" in the nickel
substrate apparently play a vital role in the activation process. These ele-
ments, carbon, zirconium, magnesium, manganese, aluminum, silicon, and,
perhaps, tungsten, were originally added to the nickel to promote workability,
* not to enhance cathode emission.

SI
.:•• . " , q 'ok•. J°-1-- ._" S|,
 Several variants of the oxide cathode are in use, including the molded
nickel, ' Medicus, and coated powder cathodes. One could argue whether
some of these cathodes might perhaps be better classified as variants of the
dispenser cathode (Section IV); some might also logically be termed inter-
mediate between oxide and dispenser types.

In all of these variants of the oxide cathode, nickel powder in different

concentrations, sometimes with an activator, e.g., ZrHz, is mixed with the
alkaline earth oxides. Various advantages are claimed for these cathodes,
and various rationales have been presented to explain the improved perfor-
mance. These claimed improvements usually include not only higher current
density for a given operating temperature but often easier control of the shape
of the cathode surface and increased resistance to poisoning. In the
molded nickel cathode, the cathode coating is 69 wt% Ni powder, 30% alkali
carbonates, and i% activator. 12 In the Medicus cathode, the coating is ap-
proximately 75 wt% Ni. 13
The coated powder cathode (CPC), in contrast, containc only 0. 5 to
3. 0 wt% Ni. A rationale for the performance of this cathode has been devel-
oped14 and is presented below. The nickel network can evaporate during
operation and must be replaced with nickel from the cathode base; otherwise,
the CPC becomes a conventional oxide cathode. Further details of fabrication
techniques for these different cathodes can be found in the references cited.
B. PRINCIPLES OF OXIDE CATHODE OPERATION
1. GENERAL COMMENTS

A high thermionic-emission current is generally associated with a high

electron-energy density in the conduction band of the emitter and with as small
a work function as possible. In addition, the bulk electrical resistivity of the
emitter should be low in order to eliminate excessive voltage drops that could
cause undesired heating or spark discharges. These conditions may be

-t6-

•." ..... r ... •.

................................................
 satisfied by metals or by n-type semiconductors (Appendix A). The large
eiectronegativity difference between barium and oxygen implies that
BaO (and similarly, other alkaline earth oxides) would be highly ionic and
good electrical insulators. However, because of the doping that inevitably
occurs during cathode fabrication and activation, the oxide cathode material
is, in fact, an n-type semiconductor.

Any consistent theory explaining the mechanism of emission from the

oxide cathode requires an understanding of the origin of the donor levels
responsible for semiconducting behavior, and much experimental and theo-
retical work has been directed toward this end. A theory must also account
for certain experimental observations about oxide cathodes. These observa-
dons include the necessity for activation mentioned earlier, the fact that tem-
peratures derived from current-voltage plots (Boltzmann temperatures) may
disagree with pyrometric temperatures, 15 and the susceptibility of oxide
cathodes to poisoning by adsorbed electronegative species. A theory must
also explain the pulse behavior of oxide cathodes. Although an oxide cathode
can deliver only a few hundred milliamperes per square centimeter under
steady-state conditions, current up to 100 A/cm2 can be obtained for times
on the order of 10 msec by pulsing the anode voltage.

2. NATURE OF THE DONOR LEVELS

The identification of the donor levels in the oxide cathode is a difficult

problem, and the issue has still not been finally resolved. Whatever the nature
of the donor levels, the donor concentration in a cathode is a complex function
not only of processes internal to the cathode but of the interaction of the cathode
with the antbient tube environment. The cathode may be considered a reducing
agent and the remainder of the tube an oxidizing agent. An activated cathode
is intrinsically reactive, and chemical species originating anywhere in the
-- t
tube that are capable of reaching the cathode may affect it. A fairly complete
treatment of the various dynamic equilibria in an oxide cathode has been pre-
sented by Tischer, who considered simultaneous equilibria between oxygen

-17-
vacancies and F centers, between oxygen in the oxide and ambient gaseous
oxygen, and between gaseous oxygen and BaO.
Many investigators have given importance to the role of "excess
barium, " i. e.,. barium in excess of the stoichiornetric requirement for BaO,
in accounting for donor levels. Excess barium presumably in some way
activates the cathode hrough the formation of surface states. The excess
barium concept does not immediately explain the accompanying increase in
electrical conductivity in the activated cathode, since the barium excess is
small (this difficulty may be circumvented for barium segregated at grain
surfaces, especially for small grains where surface effects influence much
of the total BaO volume). Nergaard identified the actual donors in BaO as F
i7
centers, i.e., an oxygen vacancy occupied by an electron. The electrons
come from the barium atoms. When current is drawn from the cathode, barium
ions should drift away from the surface region into the bulk, and the pulse and
anomalous Schottky behavior of the cathode would be explained. The Nergaard
donor picture also provides a convenient explanation for the beneficial effect
of added nickel in the coated powder cathode. The nickel presumably coats
the (Ba, Sr)O grains, prevents the growth of large, inactive crystals, and
provides conducting paths through the oxide. As a consequence of these con-
ducting paths, the electric field resulting from the applied anode voltage has
a large component transverse rather than perpendicular to the surface. Donors
14
are retained in the surface region and emission is facilitated.
On the other hand, Plumlee concluded that the amount of excess
18
barium in BaO is insufficient to account for the necessary donor density.
This conclusion was based on measurements of the barium vapor pressure
over BaO, chemical determination of excess barium, and lack of correlation
between electron emission and measured barium excess in cathodes. Plumlee
proposed that the donor centers were OH" ions on the basis of chemical intui-
tion combined with mass spectrometer measurements.

-18-
 3. THEORIES OF THE OXIDE CATHODE

Other aspects of cathode structure, in addition to the various

concepts concerning the nature of the donor levels presented in Sec-
tion Ill. B. 2, have to be invoked to more fully explain cathode function.

Chin et al. report that the grain structure of the cathode plays a
central role. 19 Barium ions adsorbed on the grain surfaces (both surfaces
internal to the cathode structure and external surfaces) cause band bending
(Fig. 3) that both lowers the work function and increases conductivity in the
surface region. Since the cathode grains are small, the perturbed region
constitutes an appreciable fraction of the cathode volume, and the bulk con-
ductivity is substantially affected. The correlation between electron emission
and conductivity thus is understandable.

The observed electrical conductivity of oxide cathodes over a wide

temperature range (600 to 1250 K) is best explained by the combination of
20
two mechanisms. At low temperatures, current flow is along paths from
grain to grain. At high temperature (?:800 K), the electron gas in the inter-
granular spaces is sufficiently dense that "pore conduction" by the electron
gas is dominant. This pore conduction model is supported experimentally by
(1) the observation that the high-temperature conductivity is suppressed by
xenon gas at a pressure of 25 atm, and (2) the observation that electrical
conductivity is not affected by a magnetic field below 800 K but decreases
with applied magnetic field above 800 K. 22 The pore conduction model for
electrical conductivity was combined with a single-crystal donor model for
thermionic emission in Surplice's "unified" theory of the oxide cathode. 22
The cathode is considered to be composed of small grains in which surface
effects dominate bulk effects. Surplice attributes the low cathode work func-
tion to band bending produced by surface states formed by adsorption of

-19.-

- -- - § .-, -.. ,..-.- -

•& _ -.--
j CONDUCTION BAND CONDUCTION BAND

BAND
VALENCEVALENCE
b.
a.

Fig. 3. Energy-Level Diagrams for Band Bending in n-Type

Semiconductor. a. Effect of ionized surface donors.
b. Effect of ionized surface acceptors.

. .................................... .. -....-.... , ,.L;•....~ -.

various residual gas species. One such adsorption process (as proposed
by Plurnlee) 18is

BaO + H0 - BaOHe + 0Z(g) + 2 g) (4)

where the BaOH. e species acts as a surface electron trap with a long lietime.
Conducting 'filaments" consisting of chains of metal atoms separated
by oxygen vacancies within the bulk oxide have been proposed by Dearnally
as an alternative explanation of various properties of oxide cathodes..3Oxy-
gen is removed, at the activation temperature, by reducing agents present
in the oxide coating. The oxide may becomeu locally conductive at certain
points at the metal-metal oxide interface through aggregation of vacancies.
It is postulated that metal-rich filaments then form under the influence of the
electric field usually applied during activatioi, and grow through the coating.
In Dearnally's view, emission from an oxide cathode is localized at the filament
ends and really represents emission from a metal. The high-emission cur-
rent density of oxide cathodes results from the very high local temperatures
produced by ohmic heating of filaments and through Schottky barrier lowering.
Cathode pulse behavior is explained by the breaking of filaments by thermal
agitation and the subsequent regrowth of filaments. Dearnally's model con-
tains many unknown parameters and would be difficult to quantify. The exis-
tence of sharply localized emitting regions is not consistent with spatially
resolved work-function distributions obtained experimentally for oxide cathodes.
The conducting filament concept does not appear to have gained much support.
4. FAILURE MECH-LANISMS

There are a number of known failure mechanisms for oxide cathodes.

Some mechanisms are intrinsic to conditions prevailing in a specific tube; they
include the breakdown of the potting compound used on cathode heaters, damage
by ion bombardment (to which ordinary oxide cathodes are quite susceptible),

-ii

4, ___
 poisoning by adsorbed electronegative species, and detachment of the oxide
coating from the substrate. Any one of these mechanisms may be of
practical importance, but they are outside the scope of the present discus-
sion because they are tube specific. Some general failure mechanisms
are discussed.

An attempt to draw excessive current from an oxide cathode will

eventually result in a disruption of the oxide because of the effects of re-
sistive heating. A resistance is associated with the oxide-ambient interface,
and this resistance can increase, especially during pulse operation, as a
result of donor depletion. In addition to the resistance of the surface region,
and of the bulk, a resistance is encountered at the oxide-substrate interface.
This resistance can increase with time as an insulating layer is formed
through the reaction of the BaO with impurities, e.g., with silicon to form a
barium orthosilicate. Excessive interfacial resistance can cause an internal
spark discharge during pulse operation if high currents are drawn. The
development of interfacial resistance can be controlled by proper initial puri-
fication of substrate material but can be a problem if quality control is not
exercised. Poisoning of a cathode can of course result when bulk impurity
species reach the surface. One recent study distinguishes between impurity
24
effects on the Fermi level and on the electron affinity. The formation of
high-work-function patches of CaO slag of about iO0 A thickness has also
been stated to result in failures of oxide cathodes. 25

Barium is continuously evolved from an oxide cathode during opera-

tion. Evaporation rates of barium, strontium, and calcium from oxide cathodes
have been measured by different methods including the use of radioactive
tracers 26 and measurement of the effect of evaporated material on the therm-
ionic emission of a tungsten probe wire. 2 7 The actual situation with regard
* to evaporation of material in a vacuum tube is complicated by the fact that
material can re-evaporate and redeposit on the cathode from other parts of
the tube. Barium loss ultimately should limit cathode life.

.22-
I
IV. BARIUM MONqOLAYER CATHODES

A '"rnonolayer" cathode consists of a metal base covered by an adsorbed

electropositive species. Mvore precisely, foz practical cathodes, the adsorbed
layer consists of an electropositive species coadsorbed with oxygen. A dipole
layer is forrwed, and the work function of the adaton- substrate system can be
lower than that of the initial substrate, and, in fact, lower than that of the bulk
adsorbate. An examination of the periodic table indicates that cesium and
barium monolayers should be particularly effective :a io[ov-ring the surface work
function since these elements are electropositive species of relatively large
ionic radius. This expectation is confirmed experiez.tally.

Monolayer cathodes used in TWTs to date have all been based on barium.
These cathodes are the dispenser and impregnated cathodes discussed later.
¶ Barium monxolayer cathodes have higher work functions than alkaline earth
oxide cathodes and therefore must be operated at a higher , <npmperatire for the
same emission current density. However, higher emission current densities
are possible than for oxide cathodes. Emission current &ensities
on the order
of several asnperes per square centimeter can be obtained at operating tem-
peratures of near 1100 C. These higher available current densities, com-
bined with enhanced resistance to spark discharge and less sensitivity to air
exposure during handling, make barium monolayer cathodes very attractive.

A surface monolayer of an alkali or alkaline earth on a substrate is

volatile at the temperatures required for the rnionic emission. Cesium is
more volatile than barium, and, barium, there•lore, is widely employed in prac-
tical cathodes. AUl barium monolayer cat1-,dee designs must provide for (1) the
formation of a low-work-function surface monolayer and (Z) the replenishment
of evaporative loss from this monolayer during the operating life of the
cathode.

_23-
 A. CONSTRUCTION

1. DISPENSER CATHODES

In a dispenser cathode, the two necessary conditions just described

are satisfied by placing the barium compound that forms the work-function
lowering layer in a cavity behind the base material, which is porous tungsten.
The material in the cavity forms a reservoir to replenish the material lost
from the cathode surface by evaporation.

original dispenser cathode, the Phillips "L-cathode, " is illus-

trated in Fig. 4. Z8 A mixture of BaCO 3 and SrCO 3 is initially placed in the
cavity. The porous tungsten plug has a density between 73 to 83% of the maxi-
mum theoretical density of tungsten. The carbonates are decomposed to
0
oxides by heating in vacuo at about i iOO C. Barium reaches the surface
during "activation, i.e., during further heating at a higher temperature.
More details concerning these processes, and the different roles of barium
and strontium, are given in Section IV. B.

An interesting variant of the dispenser cathode has been developed at

the Naval Research Laboratory (NRL).7 9 A regular mesh of iridium foil is
used in place of the porous tungsten plug. Barium oxide is placed in a cavity
behind the mesh. The chief advantage of this design is the possibility of very
precise control of the barium arrival rate at the surface by the control of the
mrnesh hole size and spacing. Further development work is in progress.

z. IMPREGNATED CATHODES
The "impregnated" cathode can be considered a type of dispenser
cathode in which the alkaline earth mixture is not placed in a separate cavity
but is distributed throughout the pores of the tungsten matrix (Fig. 5). In the
original impregnated cathode (A-cathode), the impregnant starting mixture
30
Swas SBaCO M0ZAIO
3 . The A-cathode was superseded by the B-cathode, in

which the starting irnpregnant is 5BaCO 3. ZA1 2 0 3 - 3GaCO 3 . 3 1, 32 The B-

cathode can now be considered the "standard" impregnated cathode. A slight
variation is found in the 4BaCO3- A 2 0 3 - CaCO 3 mixture of the Semicon S
cathode.

_24-
 POROUS
TUNGSTEN
~PLUG

I ' ••
!•.CAVITY
"••• •fBa,SriO CONTAINING

/-MOLYBDENUM
SUPPORT

HEATER

Fig. 4. L-Dispenser Cathode

-25-
 POROUS TUNGSTEN PLUG
CONTAINING IMPREGNANT

ILi
MOLYBDENUM
SUPPORT

HEATER

Fig. 5. Impregnated Cathode

-26-
 Still other variants of the dispenser cathode are based on tungstate
and scandate impregnants. In the tungstate cathode, a mixture of BaCO3 V
SrCO3 , and W0 3 is heated to form a compound that can be represented by
an expression such as Ba 5 Sr(WO 6 )2. This mixture, in turn, is mixed with
ZrHZ (an activator) and tungsten metal (90% by weight) to form a sintered
34
cathode compact. A similar procedure is employed to form scandate
cathodes.

3. OSMIUM DISPENSER CATHODES

The Phillips M cathode has considerable promise for future Air Force
35
applications. The M cathode consists of a standard impregnated cathode
on to which a thin (10, 000 A) surface layer of osmium has been deposited.
The resultant surface has a work function lower than that of a conventional
cathode. Phillips Corporation is no longer actively engaged in the manufac-
ture of the M cathode. Some details of their manufacturing process have been
given in a publicly available report commissioned by NASA;36 however, cer-
tain steps in the procedures are described only in very qualitative terms.

B. PRINCIPLES OF OPERATION OF BARIUM

MONOLAYER CATHODES
Some important questions that arise in connection with dispenser
(including impregnated) cathodes are: (1) What is the nature of the low-work-
function surface? (Z) How are the activating species produced? (3) What is
the mechanism by which these species reach the surface? and (4) What is the
rate of evaporation of active species from the surface? The question as to
what the mechanism is by which these species reach the surface involves both
gaseous transport through the pores of the tungsten and diffusion along the
tungsten surface. The (integrated) evaporation rate of the activating species,
i.e., the impregnant, ultimately determines cathode life.

The structure of the low-work-function surface is considered in this

section. The other questions posed will be addressed later for dispenser

-27-

...........................................
and impregnated cathodes. An atom of an electropositive element, e.g.,
cesium or barium, adsorbed on a metal surface will ionize if the ionization
potential of the adatom is less than the work function of the metal. In other

words, ionization occurs if the energy gained by transferring an electron to

the metal is greater than th.! energy required to remove the electron from
the atom. The electron transferred to the metal enters the conduction band
and is not localized. However, the adsorbed ionized species form a dipole
layer in conjunction with their classical image charges. The dipole moment
per unit area, i. e., the work-function change, is given by

&1= nd(MKS units) (5)

0

where n is the number of adatoms per square centimeter, q is the ionic

charge, and "d" the dipole layer separation (d t Zr, where r is the ionic radius).
The quantity calculated in Eq. (5) has the dimensions of an electrostatic po-
tential (in volts), i.e., change in energy per unit charge. The work function
has been taken to be an energy in the previous discussion. However, since
energy for electron emission is customarily measured in electron volts (the
energy change when an electron moves through a potential difference of one
volt), energy and electrostatic potential are related by a conversion factor of
unit magnitude. The difference between potential and energy should be kept
firmly in mind, however. Therefore, a capital t is used in Eq. (5) for poten-
tial, whereas the work function as an energy is denoted by a lower case 0.
Complications in the application of Eq. (5) arise in practice. Adatoms
interact, and the charge q is a function of surface coverage n. In addition,
in practical cathodes, both an electropositive species and oxygen may be
involved in the formation of tha dipole layer. A more detailed discussion of
these questions is given in Appendix B. The simple expression of Eq. (5)
does, however, indicate that good thermionic emitters may be produced
through the formation of appropriate surface monolayers on metallic
substrates.
 Tii

1. DISPENSER CATHODE

Schaeffer and White concluded that in the L-cathode the emitting surface
is covered with barium, without a substantial amount of oxygen on the emitting
surface or near the ends of the pores in the tungsten. 37 Barium coverage
was said to be less than a complete monolayer. These authors suggested
that a monolayer of residual oxygen initially present was removed from the
tungsten by evaporating barium very early in the life of the cathode. The
mechanism by which the oxygen removal is effected was not discussed. The
evaporating material was identified as barium metal, not BaO (removal of a
single surface monolayer of oxygen does not affect this conclusion).

Sharply differing conclusions and results were reached by Rittner et al.

in a classic series of papers. 38, 39 Rittner et al. dealt with all four of the
questions raised earlier. Measurements of surface diffusion, Knudsen flow,
electron emission, and barium evaporation rate were carried out on tungsten
of varying porosity.It was deduced that the L-cathode surface is covered by
a complete monolayer of oxygen and that the oxygen, in turn , is covered by
a complete monolayer of barium. The oxygen on the tungsten surface is
responsible for both a lower work function than in the case of barium alone
on tungsten and for a greatly increased sticking time for adsorbed barium.

Rittner et al. state that BaO is formed by thermal decomposition

of the BaCO3 with which the cathode is initially charged. The BaO then reacts
with the tungsten in accordance with the following barium-producing reaction:

i i
2BaO + -3W -e- 2 Ba 3 WO 6 + Ba(g) (6)

The gaseous barium is transported through the tungsten pores by Knudsen

flow, and the barium picks up residual oxygen from the tungsten along the way.
Rittner et al. found that half of the barium content in the evaporant is in the

S-29-
 form of BaO. The Knudsen flow of the barium virougn tbhe pores is the rate-
limiting step in barium evaporation (although it may not be the rate-determining
step for barium coverage on the surface). Additional barium is formed through
the reaction

z SBa3WO6
BaW +!W
iWW -" BaWO4 + Ba(g) (7)

3 3 6 3 4 +ag()

However, this reaction is postulated to be accompanied by the release of an

oxygen-containing poisoning agent" (not further specified) that results in the
termination of cathode life.
The emission of the BaCO 3SrCO3-filed cathode is quite similar to

that of the BaCO 3 cathode: however, the lifetime can be an order of magnitude
greater. Rittner et al. attribute this lifetime increase to the fact that the
equilibrium pressure of Ba + Sr vapor is lower than for the case of barium
alone. The rate of barium release, which is far more than adequate to main-
tain monolayer barium coverage, is thus reduced, and cathode Life is increased.

Z. IMPREGNATED CATHODE
The impregnated cathode was also studied by Rittaer and his coworkers,
and, again, the questions of emitting surface composition, activation, and
transport and evaporation of the active species were addressed.40 The cath-
ode investigated had the impregnant composition 5BaO- 2A 2 0 3 ; it is known as
the A-cathode (later formulations included CaO). The oxides in this ratio
correspond to a eutectic mixture:

5BaO +A+
Z- 2 O 3 Ba3Al0O +! BaAl 0 (8)
232 3 26 2 2 4

Barium is generated (on thermodynamic grounds) in accordance with the

reaction

2 I 1 2
Ba 3 Al O0(s) +-IW(s)-
3 3- 3
BaWO 4(s) +- BaAl 0O(s) + Ba(g) (9)
4 3 2 0 4 (s+ag

-30-

-,,..•;• - •= .- • .•• ,'••-% - A,,

..
 The BaAl 2 O 4 [Eq. (8) or (9)3 can be considered inert with respect to the
production of Ba(g) since the reaction

4n-Al0+ -4 W - BaWO4 +-4- A1 3 + Ba(g) (10)

is,again on thermodynamic grounds, suppressed by the reaction of Eq. (9).

Barium reaches the surface by Knudsen flow just as it does in the L-cathode.
However, in contrast to the essentially constant evaporation rate of the
L-cathode, the barium evaporation rate from the impregnated cathode was
found to decrease as t- The experimental data indicate that the evapo-

ration rate is much higher than for the L-cathode during the first few
hundred hours of life and is always sufficient to maintain a complete barium
monolayer ; n the surface. Bariumn loss, of course, can occur both by
evaporation from the surface monolayer or by direct streaming of material
from the pores in the cathode matrix.
0. A. Haas and others at NRL carried out Auger analyses of
impregnated cathodes and have made scanning low-energy electron probe
(SLEEP) work function maps. 41 Auger electron spectroscopy provides an
elemental analysis of the outer three to four atomic layers of a specimen
surface; as a rough rule, elements present in concentrations as low as
101% of a surface rnonolayer are detectable. The SLEEP method proAdes
work-function distribution information with 3- to 5-p-m spatial resolution
and 30-mV energy resolution. In some cases, a correlation was observed
between the presence of sulfur and early failure. On other cathodes with
initially normal emission, more rapid than normal decay was found to
correlate with a decrease in the barium, calcium, and oxygen pe;"ks. This
observed doiletion is tentatively attributed to machi.ning effects during fabri-
cation that result in the blockage of pores in the tungsten. The cathode s'zr-
face was found to be more similar to a BaO surface than to a barium surface.
The BaO is one to several monolayers thick.

Springer and T. W. Haas of AFML measured chemical changes on

dispenser cathodes during activation and poisoning by reactive gases by means

-31-
 4
of Auger spectroscopy. 2 The concentration of surface species as a function
of temperature was determined for a Type B Phillips cathode. It wat, con-
cluded that in the normal cathode the ratio of barium to oxygen was i, which is
consistent with a tungsten surface covered by a monolayer of BaO.

Forman of NASA/Lewis has examined, by means of Auger spectroscopy,

both actual cathodes and tungsten surfaces on which various species have been
adsorbed. 4 3 44
' His results differ in some respects from those of others and
have generated some discussion. Forman compared Auger spectra from
barium and BaO deposited on cleaned tungsten and oxidized tungsten with
spectra of Phillips B and Semicon S cathodes. On the basis of the detailed
structure of the low-energy Auger peaks of bariucL, he concluded that a mono-
layer of barium on a monolayer of oxygen is present on the (tungsten) surface
of an impregnated cathode, as distinct from a layer of BaO on tungsten.
These barium Auger peaks in the range 50 to 80 eV of kinetic energy differ in
position, shape, and intensity for barium in the different chemical environ-
rrents. Although Forman stressed this point, a careful examination of
t papers by Rittner et al. and others does not indicate that distinguishing
between Ba-O-W and BaO-W surfaces is really germaine to the interpretation
of the essential results of these other investigators. It is not clear whether
describing a single layer of barium atoms on.top of a single layer of oxygen
atoms on a tungsten surface as an "oxide" or "not an oxide" is meaningful.
The difference between "BaO on W" and "Ba on 0 on W" would basically
have to do with the actual location of the surface dipole within a layer struc-
ture. As a scientific question, the issue merits further consideration, but
its importance for practical cathodes is not clear.

The disagreement between Forman and Rittner on evaporation rate is

more pertinent. On the basis of Auger measurements of barium desorption,
Forman concluded that surface barium coverage must be less than one mono-
layer for much of the cathode life. This observation is the basis of a model

-32-

to-t
for cathode barium coverage and for the eventual end of cathode life. The
extent of barium coverage represents an equilibrium between evaporative loss
and the rate at which barium reaches the surface. Early in the life of the
cathode, an adequate amount of barium is available, but as barium generated
at the pore ends near the surface is depleted, surface coverage and electron
emission decrease. Studies of "good" and "poor" cathodes have been described
by Forman in which inadequate emission has been attributed either to poison-

716
ing or, in fact, to barium depletion.
45

Forman's conclusions have been criticized. 4 6 Data on the B-cathode

were recently made available that indicate that emission is nearly constant
until the end of cathode life. Barium evaporation follows a t law, just
as for the A-cathode, and the total amount of barium given off, as calculated

by integrating the evaporation data to the end of cathode life, corresponds to

the amount of bariun initially present. The calculated maximum lifetime for
the particular cathode studied is 105 hr at 10600 C at the saturation emission
current density of 5 A/cm2 . The observed constant emission implies that
cathode barium coverage does not change with time, and, in fact, the avail-
able data indicate tiiat a full barium monolayer is always present until the end
of cathode life. As an explanation of the discrepancies between Forman's
results and the observations for B-cathodes, it was suggested that the deposited
barium films used by Forman to deterniine evaporation rate are not repre-
sentative of impregnated cathodes. Although the work function of a barium on
oxygen on tungsten surface is not critically dependent on oxygen content (in
agreement with Forman), b.ýrinun evaporation does depend strongly on oxygen
content. Forman's surfaces are described as oxygen deficient. On an actual
cathode, surface oxygen is presumed to be replenished from the BaO stream-
ing from the interior.
it is apparent that in spite of the considerable effort put forth by several
research groups over the past decades our understanding of impregnated cath-
odes is incomplete.

-33-
3. OSMflJM DISPENSER CATHODES

In a search for a lower work-function surface monolayer-substrate

system, it is possible to vary both the a4sorbate and the substrate material.
It perhaps seems paradoxical, but an increase in substrate work function re-
sults in a decrease in the minimum work function attainable for a given ad-
47
sorbate. This interesting result is explained by a qualitative argument.
The electrostatic potential energy near a metal surface is shown without and
with a surface monolayer of a readily ionizable cation species in Figs. 6a
and 6b, respectively. The work function lowering is given by Eq. (5), and is
proportional to the product of the number density of cations and their charge,
as in "ordinary" monolayer cathodes (Section IV. B). The available electron
of the adatom lies at the bottom of a potential well of depth Iads. 1ads is
measured with respect to the local electrostatic potential, which may differ
from the vacuum or zero level (Fig. 6).

As the number of adatoms increases, the local potential energy @(x)

(x is the spatial coordinate perpendicular to the surface) is depressed, but
the ionizable adatom level remains at lads below the local potential
O(x = r) (r is the cation radius). p(r) can be depressed only until lads
coincides with the Fermi level, at which point the greatest work function
lowering is achieved. If the adatom density increases further, the average
adatorn charge must decrease. The electrostatic potential energy is con-
strained to be at lads above the Fermi level at x = r and to be 0o above the
Fermi level just at the metal surface, In Fig. 6d, it is evident that these
constraints mean that an increase in substrate work function p. actually
lowers the work function of the monolayer system (provided lads < 0o)"

This qualitative explanation of the osmium dispenser cathode may be

4
placed on a more quantitative basis by a classical electrostatic argument.
A uniform layer of adsorbed cations on a metal with N cations/cmnZ is con-
sidered. The equivalent area per cation is rR 2= i/N, and r again denotes the
cation radius. The local potential at an ion can be calculated if it is assumed

-34-

i%.
• .6•,••
. .. . V . . .. |.... •
 E.F!
8F

a. 'Ib. d.

Fig. 6. Potential Energy Diagrams Illustrating Additional Work-

Function Lowering Obtained with Monolayer Cathode as

b.I
no adsorbed donor
Substrate
Result of Increasingspecies, Work Function,
potential
a. With
energy distribution
is just that for (assumed ideal) metal substrate. Work
function q0 is (by definition) energy interval between
Fermi level and vacuum level. b. When some adsorbed
present,
species isarereduced
donor function o tolayer
dipole %a. is
Adatoms and
formedappear
work from
as localized levels at D at an energy interval Iads below
local potential and distance r from the surface.
D coin-
c. At optimum adatorn coverage, donor levels has
level of substrate. Work function
cide with Fermi d. Potential energy curve is
minimum value isb at sub-
above Fermi level just
Fconstrained to lie at level at
lads above Fermi
strate surface and to value function
roniium
work of
locadatom distance r for A higher substrate work
iadsorbent-adsorbate system.

function (4o2, dashed curve) results in lower minimum

work function for the system than does a lower substrate
work function (4o1, sblid curve). Adatom concentration
for optimum coverage may differ for different substrate
work functions.

-35-

X . -. ~-
that one ion is removed from the surface, leaving a circular hole of radius R.
The corresponding distribution of charges is represented by a uniform posi-
tive charge layer at distance r above the surface and a negative image charge
layer a distance r behind the surface, and by a negative diwk (of the same
charge density as the original layer) at r above the surface and a positive image
charge disk at r behind the surface. The potential along the line of symmetry
between two parallel coaxial oppositely charged disks is easily calculated. With
the potential of the disks and of the uniform charge layers superimposed, the
potential energy difference between the center of an ion and the metal surface is

where e is the electronic charge. When N = Nopt, the optimum coverage for
1
work-function lowering, A& = Oo - ads' Furthermore, there is the relation
- . N
rez (12)
0 0 - ram
in op t e 0
0

An explicit expression thus can be obtained for rain in terms of 0o and 'ads'
and it is easily verified that dmrnin/d~o < 0, as asserted previously.
The investigation of different metal coatings of higher work function
than tungsten that do not react with the impregnant resulted in the develop-
ment of the Phillips M-cathode. Rhenium, ruthenium, iridium, and osmium
all meet tne essential requirements. An osmium plus 20% ruthenium coating
5000 to 15000-k thick is sputter-deposited on the surface of an ordinary dis-
penser cathode to produce the iVM-cathode. Osmium loss apparently does not
48
occur during the life of an M-cathode. Emission current density decreases

-36-

4 St°'
 as the result of the formation of an OsW2 alloy, wbich ha.s a work function
essentially equal to that of a pure tungsten surface and, after thousands
hours of operation, the M-cathode becomes similar to the standard B-cath-
ode without an osmium coating.

Analyses of M-cathode surfaces have recently been reported. 4 9 A

"Itcolumnar" structure was observed in scanning electron micrographs of the
sputter overcoat. The columns are i/2 to I pim across and may play a role
in the cathode function by increasing the effective surface area. X-ray
photoelectron spectroscopy data on an activated IA-cathode indicated that the
tungsten was metallic, not oxidized. 9

4. FAILURE MECHANISMS

Although a monolayer cathode ultimately must fail by depletion of the

active material, other failure modes, in practice, occur first. These failure
modes are under active investigation in several laboratories. Cracking and
dimensional changes have been observed. Porous tungsten is also not stable
with respect to a loss of pore volume by sintering.

-37-

- -1 - . -- .- .- iiT .- .- .- .- .- - . i .--- . -... . - -

 S32

V. OTHER MONOLAYER CATHODES

Monolayer cathodes, other than the barium dispenser and impregnated

types, have no history of use in spacecraft TWTs. A discussion of "other"
monolayer cathodes is included here both for completeness and because some
consideration has been given to the use of LaB 6 in TWTs.

A. THORIATED TUNGSTEN CATHODES

Thoriated tungsten cathodes have long been used in radio frequency

transmitting tubes. They provide rather high current densities and have long
operating life; however, their operating temperatures and work function are
significantly higher than those of cathodes based on barimn.

Thoriated tungsten cathodes are produced by adding i to 2% thorium

oxide to tungsten powder, then sintering, swaging, and drawing the material
to form wire filaments. The filament is heated to = 25008 C for several minutes,
then held at 1825'C for 30 -min. The normal operating temperature is !tci625 C.
Improved performance is achieved if the filament is heated in a hydrocarbon
atmosphere to form a tungsten carbide surface layer. The thermal efficiency
is higher, and the operating temperature of the thoriated tungsten filament
is lower than for pure tungsten; however, thoriated tungsten filaments are
more susceptible to damage by bombardment by residual gas ions.

Thorium monolayer cathodes analogous to barium dispenser cathodes

have been constructed also. One such cathode is made with a tungsten,
thorium, and tungsten carbide mixture in a tungsLen or tungsten rheniunm
cup. 50,51 In another design, a thorium carbide-tungsten eutectic is drawn
52
by capillary action into a tungsten dendrite matrix. This cathode is said
to be more resistent to hydrolysis and to be mechanically stronger than the

packed powder design. These cathodes were developed by General Electric.

It has long been thought that a surface monolayer of thorium is

53
responsible for the emissive properties of thoriated tungsten. Thorium,

-39-

tD
an electropositive species, could form a surface monoiayer that would result
in a surface of reasonably low work function (- 2. 6 eV for the thoriated
tungsten cathode). Recent measurements by means of Auger electron
spectroscopy confirm that the surface of thoriated tungsten is enriched in
thorium (Th/O z 40 for an activated cathode). 54 Carbonized thoriated
tungsten is apparently covered also by a thoriumn
monolayer.

B. LANTHANUM HEXABORIDE CATHODES

Lanthanum hexaboride cathodes are now widely used in scanning

electron microscopes and other laboratory instruments where a small,
bright electron source is desirable. The LaB 6 cathode is in the form of a
pointed, indirectly heated rod for such applications, although for other uses
other configurations could presumably be employed. The LaB6 cathodes
have good tolerance to exposure to air (when cool), which recommends
their use in laboratory instrumentation. The work function and operating
temperature of LaB 6 cathodes are higher than barium-based cathodes, but
the robust nature of LaB6 cathodes does suggest their possible use in TWTs.

One might question the classification of LaB 6 cathodes as monolayer

cathodes. It was originally proposed that the LaB6 cathode was enriched by
lanthanum that diffused to the surface to form a dipole layer. The latest
Auger evidence and mass spectrometric analysis of evaporant materials
are in contradiction with this hypothesis. The low work function of LaB 6
is apparently intrinsic to the material. However, it is likely that the outer-
most atomic layer of LaB 6 is indeed a lanthanum plane, and a surface
dipole does exist as it must at the surface of any material. LaB 6 happens
to have a dipole of proper sign (on the appropriate crystal faces) and a
sufficiently high melting temperature to make it a useful therrnionic emitter.
Other rare earth hexaborides have also been investigated, mostly by
Soviet researchers.

-40-
 VI. TEST PROCEDURES AND LIFE-TEST DATA FOR

THERMIONIC CATHODES

Requirements of different agencies concerned with TWTs for satellite

use may vary. Space Division has a requirement for a cathode capable of
delivering I to 2 A/cm 2 for 7 to 10 years (-•60, 000 to 90, 000 hr).

It is not possible to assess cathode performance without some criteria

for cathode failure and the end of cathode life. The end of cathode life in
TWT applications does not necessarily imply total cessation of emission.

The area of the cathode emitting surface is as much as 50 times Larger than
the final area of the electron beam. The cathode surface is concave, and
electrons are focused before entering the traveling-wave structure, Even a
slight degradation of cathode emission can change the space-charge distribu-
tion sufficiently to defocus the electron beam. The power in the beam is
more than sufficient to destroy the traveling-wave structure if the beam
wanders. A general discussion of TWTs is presented in Ref. 7.

Various empirical methods for assessing cathode condition are used.

One standard method involves shutting off cathode heater power, then recording
the time required for emission to drop to sonie fraction of its initial value.
This "'time-to-knee" method is valid only for comparing cathodes of identical
type and construction. In other life tests, cathode failure is defined as the
point at which cathode temperature must be increased above some arbitrary
-value to maintain a given current density. Still other criteria are, of course,
possible.

The selection of the "test vehicle" is another unresolved problem in

* •assessing cathode performance. Many tests have been carried out with
cathodes in a simple diode configuration. Unfortunately, the cathode interacts
-with the remainder of the tube. For example, evaporated material may be
transferred from the cathode to an intermediate surface and back to the

-41-

Z C••
 .---n'--.-.-...
.... . .. . ... . .. . .. . . .. . -. . ....-. .. .- * ..... . .. ... . .. . . . . .. . . . ... . . . .. . . -. . .. .

cathode. In some test configurations, TWT geometry is simulated, and

more widespread use of such configurations has been advocated. It must
then be decided whether appropriate bias and radio frequency voltages
should also be present. Inclusion of these additional factors greatly in-
creases the complexity and cost of cathode life tests. Considerable dis-
cussion at the recent Tri-Service Cathode Workshop was devoted to this
question of test standardization.

The lack of standardized test procedures complicates comparisons

of cathode life-test data. A further complication is the simple fact that the
very-long-term life-test data essential for space tube cathodes are just not
available. Life tests, especially those in which actual tubes or simulated
tube geometry are used, are expensive, and accelerated tests are suspect
because of the many variables involved. Extrapolation of observed
degradation to forecast ultimate tube life is not necessarily straightforward.

.4, Space experience with oxide cathodes in commercial satellites is

summarized in Ref. 7. For 100,000 hr, it is suggested that cathode current
"density should not exceed 0. 2 A/cm2.
Some long-term test data for dispenser cathodes are available. Infor-
mation given in Ref. 7 indicates that a lifetime in excess of 105 hr at 1035 0 C
for an S-type impregnated cathode is possible. This estimate is based on
an extrapolation of higher temperature data, and the current density is not
specified. There is some evidence from European manufacturers, including
both test data and flight experience, that dispenser cathodes may meet Air
Force requirements. More information, however, is needed.

Long-term life tests are now in progress at Hughes Electron Dynamics

Division and at Watkins-Johnson, Inc. The Watkins-Johnson data are of
particular interest since the B, M, S, and tungstate cathodes were all
2 57
evaluated in a simulated tube geometry at 2 A/cm, . 57The tungstate cathode
was found to be unsatisfactory in the Watkins-Johnson test, although other
58
published data are somewhat more encouraging. In contrast to the results

-42-

... _ _ S.. _. . - A"'#,

, -
 of Ref. 7, the S-cathode also did not perform very well. The impregnated
I cathodes, the B-cathode, and especially the M-cathode, seem to be more
promising for Air Force missions.

-3

• -43-
 VII. FIELD-EMISSION CATHODES

Processes other than thermionic emission, e. g., photoemission and

secondary emission, can be employed to provide an electron source. Of all
possible nonthermnionic cathodes, however, field emitters seem to be the
most appropriate for space applications.

Application of an electric field lowers the energy barrier for electron

emission (Schottky effect, Appendix A). Stanford Research Institute has
recently described a technique in which microcircuit fabrication methods
were used for producing an array of cones on a surface (Fig. 7). 59 About
6.4 x 105 cones/cm can conveniently be produced. At an applied potential
of 50 to 200 V on the molybdenum overcoat, each cone emits 150 p.A, for a
'-•

2
total emission of-95A/cm . Another field emitter cathode technology has
been described in which an array of tungsten fibers is produced in a UO0
rod. 60 The fiber density is I x 107 to 5 X 10 7/cm2 , and the sharp fiber
ends serve as field emission points on the surface. NRL has incorporated
an SRI-cathode in an actual tube (purely for demonstration purposes).

-45-
--- 1

EMISSION CURRENT
ANCONE; 95 Alcm TOTAL
"r156it0

MOLYBDENUM
OVERLAYER MOLYBLENUM CONES ,1 m

...........
SSUBSTRATE
Fig. 7. Stanford Research Institute Field-Ernmission Cathode

-46-
 Viii. CONCLUSIONS

Many different types of thermnionic cathodes, as well as some types

of field emitter cathodes, have been described in this report. Studies of
thermion.c emission from still other types of cathodes, including various
carbides, borides, rare earth oxides, and other rather exotic materials
can be found in the literature. However, cathode research in the United
States has not been funded extensively in recent decades. A number of
agencies have developed or adopted a "favorite" cathode and have

undertaken limited development work, but systematic studies of thernaonic

e•mssion by means of modern surface-analysis techniques have generally
been lacking. With the notable exception of work undertaken at NRL,
cathode studies in recent years usually have been initiated only in response
to specific problems.

The rest-its of a literature survey undertaken during the preparation

of this report may be of general interest. A computerized search (Lockheed
Dialog system) of the Inspec (IEE of the U. K.) data base was made with
the search terms thermionic cathode(s) or thermionic emission or oxide
cathodes. A total of 566 references was found for the period January i969
to May 1978. Several hundred references, from the mid-1970s to the pre-
sent, were sorted by country of origin. Nearly half of the papers originated
in the Soviet Union and Eastern Europe. About 20%0 of the publications were
from the United States, 17% from Western Europe, and 10% from Japan.
The studies of "exotic" thermionic emitters have almost all been carried
out in the Soviet Union.

Informed sources within the cathode community indicate that dispenser-

impregnated cathodes will have to be employed to meet futare spacecraft mis-
sion requirements. Of these cathodes, the impregnated B-cathode and osmi-
um coated M-cathode offer the highest probability of success. There are no
fundamental reasons why more exotic emitters, e, g., field emitters, could

-47-

--.. -
 not ultimately be employed in space applications. However, the difficulty in
acquiring sufficient test data in any reasonable time to flight-qualify com-
ponents based on a new technology, even with no gross problems in fabrication
and operation, is so great that impregnated cathodes will almost certainly be
utilized. Test data on impregnated cathodes are still not complete, but
encouraging data do exist.

Although impregnated cathodes are promising, there are many unre-

solved questions. Cathode perfozrnance varies. Reproducibility in
impregnated-cathode fabrication is essential if they are to be used with
confidence in space tubes. Reproducibility must ultimately depend on a
thorough understanding of impregnated-cathode operation, e.g., on the
importance of surface morphology and impurities. During the next few
years, additional studies of impregnated-cathode surfaces by means of
modern surface analytical techniques must be undertaken. Such studies
should correlate information obtained through submicron Auger analysis and
work-function mapping. Detailed analyses of this type should permit the
identification of specific cathode failure mnechanisms, the correlation of
cathode performance with these failure processes, and the development of
means of recognizing signatures indicative of premature cathode failure.
Having thus elucidated the basic failure mechanisms of TWT cathodes, the
objective of improved cathode performance would be met by means of
(1) recommendations on material and process improvements, (2) improved
methods for cathode quality control, and (3) guidelines for accelerated
testing and lifetime prediction.

Additional studies of other types of thermionic cathodes, and of non-

thermionic emitters as well, should also be funded. It should be possible,
by means of modern analysis techniques, to effect a considerable advance
in the understanding of the electron emission process. Research on cathode
technology is likely to be very cost effective in view of the enormous cost
of the communications satellites that are absolutely dependent upon it.

-48-

I-
 REFERENCES

1. Defense Satellite Communications Office, TOR-0074(4624-01)-2

The Aerospace Corporation (f February 1974).

2. C. Herring, and M. H. Nichols, Rev. Mod. Phys. Z, 185 (1949).

3. W. B. Nottingham, Thermionic Emission, Report 321, MIT

Research Laboratory of flectronlcs Technology, Cambridge,
Mass. (10 December 1956).

4. A. Venema, "Thermionic Emission, " in Handbook of Vacuum Physics,

Vol. II, Chapt. 2, A. H. Beck, ed., Pergammon Press, O~xford
(0966).

5. R.O0. Jenkins, Vacuum 19, 353 (1969).

6. R%. Strauss, J. Bretting, and R. Metivier, Proc. IEEE 69, 387

(1977).

7. G.A. Haas and R. E. Thomas, "Thermionic Emission and Work

Function, " in Techniques of Metals Research, Vol. 6, Part 1,
E. Passaglia, ed., Interscience, New York (1972).

8. J. M. Housten, and H. F. Webster, "Thermionic Energy Conversion,"

in Advances in Electronics and Electron Physics, Vol. 17, L. Martin,
ed., Academic Press, New York (i6U).

9. G. A. Haas and R. E. Thomas, J. Appi. Phys. 34, 3457 (1963).

10. G. A. Haas and R. E. Thomas, Surf. Scd. 4, 64 (1966).

11. %. D. Ryder, Electronic Fundamentals and Applications, Prentice-Hall,

'Englewood Cliffs, New Jersey (1959).

12. C. P. Hadley, W. G. Rudy, and A. G. Stoeckert, J. Electrochem.

Soc. 105, 395 (1958).

13. G. Medicus, 1978 Tri-Service Cathode Workshop, Naval Research

Laboratory, Washington, D. C., 1978.

14. D. W. Mauer and C. M. Pleass, Bell System Tech. J. 2375 (1967).

15. L. S. Nergaard, RCA Rev. 20, 191 (1959).

-49-

I.
t6. K. M. Tischer, Vakuum Tech. 15, 114 (1966).

17. L. S. Nergaard, RCA Rev. 13, 464 (1952).

18. R. H. Plumlee, RCA Rev. 17, 231 (1956).

19. T. N. Chin, R. W. Cohen, and M. D. Coutts, RCA Rev. 35, 521

(1974).
20. R. Loosjes and H. J. Vink, Phillips Res. Reports 4, 449 (1949).

21. R. C. Hughes and P. P. Coppola, Phys. Rev. 88, 364 (1952).

22. N. A. Surplice, Brit. J. Appl. shys.
(J. )hys.
D)I, 1245
(1968).

Z3. G. Dearnalley, Thin Solid Films 3, 161 (1969).

Z4. G. A. Haas, A. Sbih, and R. E. Thomas, 1978 Tri-Service Cathode
Workshop, Naval Research Laboratory, Washington D.C., 1978.
25. A. Shih, G. A. Haas, and C. Hor, 1978 Tri-Service Cathode Workshop,
Naval Research Laboratory, Washington, D.h.h 1978.
Z6. W. F. Leverton and W. G. Shepherd, J. App2. Phys. Z3, 787
(1952).

27. W. C. Rutledge, A. Milch, and E. S. Rittner, J. Appi. Phys. Z9,

834 (1958).

Z8. H.J. Lernmens, M. J. Jansen, and R. Loosjes, Phillips Tech. Rev.

11, 341 (1950).
29. R. E. Thomas, T. Pankey, aad G. A. Haas, 1978 Tri-Service Cathode
Workship, Naval Research Laboratory, Washington, D.C., 1978.
30. R. Levi, App2. Phys. 24, 233 (1953).

31. R. Levi, ARpp. Phs. 26, 639 (1955).

32. R. Levi, Phillips Tech. Rev. 19, 186 (1957/58).

33 Smicon Dispenser Cathodes, Semnicon Associates, Inc., Lexington,

Kentucky, undated technical bulletin.

34. R. J. Boadley, W. T. Boyd, T. J. Mall, and M. J. Silvka,

ECOM-01719-15-.3, Research and Development Technology,
Schenectady, New York (1969).

-50-

---------------------------
 35. P. Zalm and A. J. A. van Stratum, Phillip•s Txich. Rev. 27, 69
(1966).
36. A. Gupta and C. 0. Young, The Maar;factura of M-Tvpe Inpregnated
Cathodes, for Order No. C-Zi2-D, NASA/Lewie Research Center (1977).
37. D. L. Schaeff er and J. E. Whit- Y. pp.hys. 23.,5 669 (1952).•
38. E. S. Rittner, R. H. Ahlert, and W. C. Rutledge, J, Appi. Phys. 28,
156 (D957). 2 6 9

39. W.C. Rutledge and . S. Aplert, 3.

.App. 167 Py.Z8,
(1957).
40. E. S. Rittner, W, C. Rutledge, and R. M. Ahlert, J. A221. Phys. L_8,
t468 (1957).
41. G. A. Haas, H. F. Gray, and R. E. Thomas, J. Appl. Phys. 46,

3293 (1975).

42. R. W. Springer and T. W. Haas, J. Appl. Phys. 45, 5260 (1974).

43. R. Forrnan, TN 0-8295, NASA-Lewis (1976).

44. R. Forrnan, J. ApPi-. Phys. 47, 5272 (1976).

45. R. Forrnan, IEEE Trans. Electron. Devices ED?4. 56 (1977).

46. E. S. Rittner, J. Appl. Phys. 48, 4344 (1977).

47. P. Zalm and A. 3. A. van Stratum, Phillips Tech. Rev. 27, 69

(1966).

48. A. J. A. Van Stratum and P. N. Kuin, S. AppI. Phys. 42, 4436

(1971).

49. D. Jones, 1978 Tri-Service Cathode Workshop, Washington, D. C.

1978.

50. H. H. Glascock, Jr., S. Appl. Phys. 37, 4995 (1956).

51. H. H. Glascock, Jr., Rev. Sci. Inst. 43, 698 (1972).

52. H. H. Glascock, Jr., Rev. Sci. Inst. 47, 90 (1976).

V-51 -
 53. W. E. Danforth, "Thorium Oxide and Electronics," in Advances in
Electronics and Electron Physics, Vol. V, L. Marton, ed
a-Academic
fl~eas, New Yo-rk (1953j.
54. E. Ishikawa and H. Tobase, Japan J. Appl. Phys. 15j 1571
(1976).

55. J. M. Lafferty, J. Appl. Phys. 22, 299 (1951).

56. B. Goldstein and B. X. Szostak, 3. Appl. Phys. (to be published),

57. D. H. Smith, R. R. Bartsch, and G. Woda, CR-134602, NASA

(1974).

58. B. Smith and A. Newman, IEEE Trans. Electron. Devices

EDZ4, 279 (1977).

59. 3. Brodie and C. A. Spindt, 1978 Tri-Service Cathode Workshop,

Naval Research Laboratory, Washington, D.C., 1978.

60. W. L. Ohlinger, D. N. Hill, R. K. Feeny. and A. T. Chapman,

1978 Tri-Service Cathode Workshop, Naval Research Laboratory,
Washington, D. C., 19758.

.. S:

i - -~~... Z..... . b

It. i . . " ... | •

-• CI--.
 APPENDIX A

EMISSION EQUATIONS

A. THERMIONIC EMISSION FROM METALS

Thermionic emission from metals provides a convenient starting point

for a general discussion of thermionic emission. Certain concepts funda-
mental to an understanding of thermionic emission, e. g., work function,
are easier to visualize for a metallic emitter.

The thermal emission of electrons from a metal is analogous to the

evaporation of molecules from a liquid. In both cases, particles with
sufficient kinetic energy can overcome an energy barrier and escape. The
electrons in a metal constitute a degenerate gas and must be described by
means of Fermi statistics. At absolute zero temperature, each possible
energy level is occupied by two electrons to a maximum value known as the
Fermi energy (Fig. A- 1). This energy boundary between occupied and
unoccupied energy levels is sharply defined at 0 K. At a temperature T>0,
the boundary region is roughly of width kT (where k is the Boltzmann
constant). Only electrons in this interval, kT, are capable of absorbing small
(thermal) amounts of energy. The Pauli exclusion principle forbids thermal
excitation of electrons farther down in the energy distribution since no
unoccupied state is available to receive them. The Fermi energy is defined
at temperature T as the energy of that level with a probability of occupation
of 0. 5. The minimum energy that an electron must have to escape from the
metal (at absolute zero) is the energy difference between the Fermi level and
the vacuum level; this quantity is the work function. The work function is a
characteristic of the bulk metal and depends upon the surface state of the
metal, including crystal orientation. For the moment, it is simply assumed
that the work function is constant everywhere on the ideal metal surfaces
considered in the next several sections.

-53-

J~
 ++

j IE"_
n (E nIE
a.
T=0 T> 0r
b. c.

Fig. A- I. a. Energy-Level Diagram for Metal and

Corresponding Density of Electron States.
b. At T = 0 K. c. At T>0 K. EF is the
Fermi level:, is the work function.

-54-
1. RXCHARDSON-DUSHMAN EQUATION
Certain equations involving electron emnission current arise
repeatedly in any discussion of cathodes. The emission current from a
metal at temperature T is readily calculated. A standard derivation is
presented first.I

The momentum distribution for the electrons in a Fermi gas, i.e.,

the number of electrons per unit volume with momentum between$p and-$ +

nQp) I
h I + exp-(E EF)/k-],

where EF is the Fermi energy with respect to the vacuum level, and E is the
electron energy measured with respect to the Fermi level. The x axis is
assumed to be normal to the surface; the emitted electron current will
simply be (-e) times the rate at which electron4 cross the surface in the
x direction. Since the rate at which electrons in the moomentum range d$'
strike the surface is

vnnp eF--x af d$ %n(pT dE~dpYdpz

the emitted current is

h2
h
ffdp 4 dpI
dp[ dE
+ e xp( Er)/kT]
(A-2)

-55-
The zero of energy in Eq. (A-2) is taken as the bottom of the conduction
band (Fig. A-i). Since (E - EF)/kT >> I the above integral can be
approximated by

34qyA~e
= (kT) e-/kT (A-3a)

Equation (A-3) is prominent in any discussion of cathiodes and is known as

the Richardson-Dushman equation. The equation is often written in the

form

J =AT 2e- O/kT (A -3b)

where

2
4iirnek
hr3 120 A/(cm K2)

is the universal thermionic (Richardson) constant. As originally derived by

1/2
Richardson, the temperature dependence in the pre-exponential is T
not T2. Soon after the development of quantum mechanics, Dushman
2
correctly derived the T dependence (although Dushi an missed
by a factor of two I ýcause the electron spin still h riot been discovered
at the time of his derivation). Experimental valueu f the Richardson
constant can depart widely from the theoretical value 120 A/crnm . An
interesting correlaLion between measured values of the Richardson coefficient

and the heats of atomization of the emitting surfaces has been noted. Z How-
ever, this correlation evidently arises, at least in part, because of the
effect of the material evaporated from the cathode on the anode work function
and therefore is not intrinsic to the emission process.

-56-

-I
 It has been assumed that electrons reaching the surface with
F sufficient energy and with appropriately directed velocity inevitably are
C -emitted. There is a possibility that electrons may be reflected back into
the metal. Therefore, the Richardson-Dushman equation often is written
in the form

Iwhere
J = A(I - 7) T2e- /kT (A-3c)

Y is a reflection coefficient, which can, at least in principle, be

calculated.

Equation (A-3) was derived with the assumption of a particular energy

4 distribution in the metal. The equation can also be derived in a very general
way by means of statistical mechanics. 3 The emitter is taken to be in
equilibrium with the gas of emitted electrons outside the metal. This
derivation makes clear the interesting fact that measurements of emitted
current never yield information about the microscopic structure of the
emitter except to the extent that this structure influences the work function
or the reflection coefficient. Band structure effects on the reflection
coefficient, which are generally negligible, are treated in Ref. 4.

2. LANGMUIR-CHILD EQUATION

A diode vacuum tube consisting of an ideal cathode and an anode is

11 considered. If a positive anode voltage VA is increased frorn zero, a limiting

current described by the Richardson-Dushman equation (Eq. A-3) is reached.
This situation is known as temperature-limited emission. At lower anode
voltages, emission current may be sufficient to produce a space-charge
region between the anode and cathode. The voltage between the anode
and the cathode must obey Poisson's equation:

d V PW (A-4)
dx

-57-

', . - r.
 where p(x) is the charge density. The equation of continuity requires that
J(x) = constant = J. If electrons are emitted from the cathode with zero
velocity,

(4 my = eV(x) (A-5)

Furthermore, since 3 pv, Eq. (A-4) can be written

-2dV 1/2
d V- t 'M (A-6)

One mnay solve for V(x). Utilization of the boundary condition that
at x L, V(x) = V, the applied voltage, results in the expression

o4c Z 1 3/2

relating the current and the applied voltage for a simple diode configuration.
__ This expression is known as the Langmuir- Child law. Electrons are in fact
not emitted with "zerokinetic energy as assumed in the above derivation.
A more complete solution to the diode current problem, including the effect
of the finite initial velocity of the electrons, is given in Ref. 5.

3. SCHOTTKY BARRIER LOWERING

The applied field itself can influence the emission current by effectively
lowering the cathode work function. An electron outside a metal (taken to
be a semi-infinite conducting sheet) "sees" an image charge and experiences
a potential (Fig. A-2):

• e
W(x) Tre x + constant (A-8)

-5$-

* p
 VACUUM

EF-

Fig. A-2. Electron Energy as Function of Distance from Surface

of Metal in Absence of Applied Field

-59-
As the electron approaches the surface, it will eventually "see" short-range
forces and will be affected by the details of the surface region. Beyond this
distance, which need not be further specified here, Eq. (A-8) will be valid.
If an external field, E. is now applied t" the metal, the potential outside
the metal is given by (Fig. A-3):

,()=16• e + Ex + constant (A-9)

This potential function has a minimum at

X (A-WO)

The difference between the potential (Eq. A-9) at xs and the potential at
infinity in the absence of the applied field represents a lowering of the
potential barrier for electron emission. The work function in the presence
of an applied field is (remembering to multiply the potential by the electronic
charge to yield an energy)

S=(A- IHa)

-o
- 3.79 x 10-5 VE- (A -iM b)

where 0o is the initial work function (in electron volts) in the absence of a
field, and E is the applied electric field (in volts per meter).

-60-

',¢
 xS
- VACUUM S
-•- -SI-,-
N

EFIDI qx)

N\

Fig. A- 3. Electý--ostatic Potential

as Function of Distance
from Surface in Presence
of Exter-nally Applied
Electric Field.

NL

-61-
 4. RETARDING-FIELD REGION

If the anode is biased negatively by a potential VA, an additional poten-

tial barrier must be overcome by the emitted electrons in the diode configura-
tion of Fig. 2. Strictly speaking, the additional barrier is -eVA + OA - OC,
where VA< 0, and OA and OC are the anode and cathode work functions, respec-
tively. The current density is

/eVA- OA+ c
S = JoexpQVA kT __ (A- 12)

where J is given by the Richardson-Dushman equation, Eq. (A-3).

In summary, a plot of current versus applied voltage in a diode con-

figuration, a standard test configuration, can be divided into distinct regions
of operation: (1) the retarding-field region; (2) the space-charge -limited re-
tion (Langmuir-Child law); (3) the saturation current region (Richardson-
Dushinan equation); and (4) at sufficiently high applied fields, the Shottky re-
action. These operating regions are encountered, in the order given, as the
anode voltage is made more positive and, in general, are observed for emit-
ters other than metals as well. Note that in the intermediate-field region the
Richardson-Dushman equation predicts the number of electrons that could be
emitted from a cathode at a given temperature. The Langmuir-Child law
yields the number of electrons that are emitted and actually reach the anode.
Most vacuum tubes, including all space TWTs, operate in the space-charge-
limited region. The space-charge cloud serves as a "fly wheel" to stabilize
tube operation. Small variations in cathode temperature should not, at least
in principle, have a significant effect on tube current.

B. THERMIONIC EMISSION FROM SEMIICONDUCTORS

• In the discussion of the Richardson -Dushman equation, care was taken

j to point out that the result is general. If the thermodynamic derivation is

considered, it is clear that nowhere in the derivation is it assumed that the

-62-

4*
emitter is a metal. Note also that the density of conduction electrons does
not enter directly into the Richardson-Dushmnan equation. Most of the con-
duction electrons cannot contribute to thermnionic emission because of the
exclusion principle. Only the occupation of energy states above the nearly
completely occupied states of the conduction band is of concern.

It does seem, intuitively, that the density of conduction electrons

should be important in thermionic emission. Observation of thernionic
emission from an insulator is not expected if only because the low
electrical conductivity would prevent electrons from the power source from
replacing emitted electrons without excessive voltage drop. The answer
to the question raised here is that the conduction electron density does enter
the Richardson-Dushman equation indirectly through its effect on the work
function.

It is possible to make a number of general observations pertinent to

therrnionic emission from nonmetals. These generalizations do not seem
to have been explicitly stated in the literature. In a semiconductor, it is
customary to refer energies to the bottom of the conduction band and to
introduce the electron affinity X, defined as the energy interval between the
bottom of the conduction band and the vacuum level.

The work function of a semiconductor (Fig. A-.Ib) is then

- F + X, (A- 13)

i. e., the interval between the Fermi level and the vacuum level, just as in
the case of a metal. The electron affinity for semiconductors with primarily
7
covalent bonding is on the order of 3. 5 to 4. 5 eV. Electron affinities in
ionic materials have not been as extensively measured; however, in BaO,
S~8
for example, the electron affinity is 0.7 to 1. 0 eV. It is expected that
electron affin,.ty should be generally lower in ionic than in covalent
materials. In an ionic compound, atomic shells are, to first approximation,

-63-
completely filled. An electron added to an ionic material has no conveniently
available orbitals in which it can be accommodated. Its binding energy,
i. e., the electron affinity, is low. In a covalent material, orbitals are more
readily available since the electron distribution about constituent atoms is
not completely spherically symmetric. Electrons *night, for example, be
placed in d-orbitals. Electron affinity thus would be higher than in an ionic
compound.

These considerations imply that any search for better semiconducting

therrnionic emitters should concentrate on ionic materials, or at least on
materials not strongly covalent. Even with heavy doping, the Fermi level in
a semiconductor cannot be moved far up into the conduction band, and the
work function is at least as large as the electron affinity. Studies of fairly
covalent materials, e.g., various uranium compounds, 9
confirm that these
materials do have higher work functions than are found for conventional
oxide cathodes (they wer e investigated for use in thermionic energy con-
version devices and were to be fission heated in a neutron flux).

These general observations may be made more quantitative. An

intrinsic semiconductor (an insulator is an intrinsic semiconductor with a
large bandgap) is considered. The Fermi level is related to the temperature
10
by the expression

EF = -1g+ 3- kT In (A-14)

The width of the energy gap is E ; the effective masses of holes and electrons
are given by inmh and me, respectively. The zero of energy is taken to be
the bottom of the conduction band. If Equations (A-13) and (A-14) are
substituted in Eq. (A-3c),

nV4
m E /Z+X
J 4
-A(
-) exp- kT (A-15)

-64-

. ..*1 - . .
 The quantity E]Z + X is large for an insulator or wide-band gap semiconductor,
and the thermionic emission current attainable at temperatures below the
melting point is insignificant, which is in agreement with intuition. For
silicon at its melting point, J = 280 nA/cm

Practical semiconductor cathodes are impurity materials and, since

"holes" cannot be emitted, must be n-type. The energy level diagram of a
semiconductor emitter (BaO) is illustrated in Fig. L. The donor levels are
located in the energy gap just below the conduction band. An expression for
the position of the Fermi level is readily found in standard reference works
i
on solid-state physics

Ed kTIn n h3 (A-16)
[d(7wnkT)J2

where Ed is the energy of the donor level. All energies are measured
relative to the bottom of the conduction band. Substitution of the above
expression in Eq. (A-3) yields
1/4/Fd/?-+.X
( - r=
r 3/
hi/Z 4 k51 4
t 14
exp kT %-7)

It is satisfying that this emission expression, known as the Fowler equation,

does depend explicity on the density of electron donors. Certain complications
limit the application of the Fowler equation. The validity of Eq. (A-16) and
therefore Eq. (A-17) is limited to temperatures below the range of operation
of practical cathodes. x is also a function of temperature. Equation (A-I7)
however, is useful as a guide to more refined cathode models.

-65-

P
*9
 APPENDIX A

REFERENCES

1. C. Kittel, Introduction Solid State PlhZsics, 3rd ed., *John Wiley and
Sons, New York (1966).

2. A. K. Vijh and P. Lenfant, Canadian J. Phys. 51, 111 (1973).

3. C. Herring and M. H. Nichols, Rev. Mod. Phys. 21, 185 (1949).

4. G. A. Haas and R. E. Thomas, "Therrnionic Emission and Work
Function," in Techniques of Metals Research, E. Passaglia, ed.,
Interscience, New York (197Z).

5. W. B. Nottingham, Thermionic Emission, Report No. 321, MIT

Research Laboratory of Ele-ctronics"(f0ecember 1956).

6. A. Venema, "Thermionic Emission," in Handbook of Vacuum Physics,

Vol. II, Chapt. 2, A. H. Beck, ed., Pergamrnon Press (1966).
7. D. L. Feucht and A. G. Milnes, Heterojunction and Metal-Semiconductor
Junctions, Academic Press, New York f1972).

8. L. S. Nergaard, RCA Rev. 13, 464 (1952).

9. G. A. Haas and J. T. Jensen, J. Appi. Phys. 34, 3451 (1963).

-66-

I I I I I I I
 APPENDIX B

WORK FUNCTION

A. WORK FUNCTION OF METALS

The work function has so far been simply accepted as a given quantity
for a metal. An understanding of work function variation among different
metals and among different crystal faces of a particular metal is required
in order to understand thermionic emission (Fig. B-la). The electron
distribution of a metal will spread beyond the outermost layer of surface
atoms. The boundary between the interior and the exterior of the metal
cannot be completely abrupt since such a boundary would require rapidly
kinetic energy.
changing wave functions representing electrons of high
By the spreading of the electron distribution outside the metal, the kinetic
energy of the electrons is reduced. The potential energy of the electrons,
however, will be increased. At equilibrium, the situation shown in Fig, B-lb
will nrevail. This configuration represents a separation of positive and
negative charge (Fig. B-Ic) and causes a surface dipole moment.

The removal of an electron from the metal is considered next. The work
function is defined as the change in (Helmholtz) free energy of the metal
when an electron at the Fermi level is removed, at constant temperature
and volume, to a position at rest at infinity. It is convenient to divide the
energy required into a bulk term, which represents the energy change if the
electron distribution around the surface atoms were the same as for bulk
atoms, and a surface term representing the energy change as a result of
the surface dipole moment D.

aOF +ED (B-I1)

T,V

-67-

--------
 a. bu
Spread'n of th Eletro W-nDstri-

Ic.d.
Fig. B-i. Electron Doable Layer Formation

distribution of electrons about suf-

face atoms were same as in bulk.
b. Actual charge distribution near
metal surface is represented by
density of stippling. Electrons are
found outside of volume that alone
would be occupied if surface atom
electron distribution were identical
to that in bulk. c. "Leakage" of
electrons beyond idealized surface
of Fig. ia results in charge dis-
tribution equivalent to dipole
layer, with negative side outer-
most, superimposed on idealized
distribution. Difference between
charge distributions of Figs. B-la
and B-lb. d. Characteristic wave-
length of electrons in metal is
large compared to dimensions of
surface features. Electron distri-
bution thus tends to be smoother
than surface geometry.

-68-

'%
 where F is the free energy, and N is the number of electrons in the metal.
The term&9F/8SN is the chemical potential and, in fact, is, by definition,
"the Fern-i energy.

The bulk (OF/3N) contribution to the work function can be handled rea-
sonably well theoretically. At T = 0 K, WFf&N can be replaced by SE/!N,
where E is the cohesive energy. Calculations of cohesive energy have been

made by Wigner and Wigner and Bardeen, among others. 3,4 These theories
yield E(N). The calculation of the surface dipole moment is more difficult.
The dipole contribution is estimated by subtracting the calculated bulk ternm
from the measured value of the work function. The surface dipole moment
can be very substantial, ranging from a few tenths of an electron volt up to
several electron volts. More recently, the Fermi-Thomas model was
S
applied rather successfully to the calculation of work functions. Both the
bulk and surface terms are given by the Fermi-Thomas model in terms of the
the lattice constant and one parameter. Agreement with experiment is
satisfactory (about ± M0%) for those metals to which the model is applicable.
The theory is not applicable to metals in which the diameter of the ion cores
is comparable to the lattice spacing, e.g., copper, silver, and zinc. The
theory, furthernmore, does not permit calculation of differences in work
function between crystallographic surface planes of the same metal.

It has been pointed out that the spreading of electrons beyond the
outermost atomic layer results in a surface dipole nomaent. The
actual electron distribution represents a compromise betveen potential and
kinetic energy. If a surface is not flat, but contains ridges and valleys,
the electronic potential energy is minimized if the electrons preferentially
bunch up in the ridges. This bunching increases electron kinetic energy
(higher frequency Fourier components are required to describe the electron
distribution), and the electron distribution on a rough surface will be

-69-

- I.-
 smoother than the surface topography if the spacing between the peaks and
valleys is on the order of atomic dimensions. The smoothing is equivalent
to an additional contribution to the surface dipole moment with the positive
end outward. The smoothing is more pronounced with rougher surfaces,
which accounts for the fact that the work function is highest on the most
atomically smooth surfaces.

Many attempts to relate the work function to various bulk atomic

, 6.
properties have been reported, e.g., to electronegativity, ionization
energy, 7 and to atomic radii8 (these quantities are related). Such theories
are intrinsically incapable of yielding the dipole contribution to the work
function and, therefore, cannot account for work function differences between
different crystailographic planes.
A semieinpirical theory by Steiner and Gytopoulous does, .½owever,
predict work functions and their crystallographic variation reasonably well.
The method is based on an expression for atomic electronegativity that
explicitly includes the coordination number of surface atoms. Details are
given in Ref. 9.

In summary, whereas work function calculatio-as have been carried out

for many metals with reasonable agreement with experiment, ab initio
calculations are not sufficiently accurate to account in detail for work function
differences between crystallographic planes. The qualitative nature of these
differences is understood and semiempirical theories are available. Note
that detailed calculations of work functions for different metals and for
different crystal planes of a given metal probably would not have much
importance in the construction of practical cathodes, The qualitative prediction
of trends, which can be accomplished, is of greater importance.

B. WORK FUNCTION OF MONOLAYER SYSTEMS

Monolayer adsorption of electropositive species on metals has been

extensively studied. Since these systems constitute perhaps the simplest

-70-

"''''• b-
"possible examples of chemisorption, they have been a favorite research
subje ct Lor both experimental '&d theoretical scientists, quite apart from
any considerations of applicaticn to cathode development.
A single electropositive atom, i.e., an alkali metal or alkaline earth

element, is considered to be adsorbed on a metal surface with a high work

function, e, g., tungsten. If the work function of the substrate exceeds the
ionization potential of the atom, then the adatom should ionize. The ionized
surface atom, together with its classical image charge, form a dipole. For
a surface layer of (widely separated) adatorns, the potential change at the
surface as a result of the dipoles formed is

AO= qd (MKS units) (B-2)

where n is the number of adatoms per square centimeter, q the ionic charge
(q = e for a completely ionized, univalent atom), and d the dipole spacing
(d = Zr, where r is the ionic radius and is the distance between the charge
center of the ion and its image charge). A4is an electrostatic potential
change expressed in volts. With the convention adopted in Section IV. B, the
corresponding energy change for an electron is denoted by A0. These quantities
are numerically equal if energies are expressed in units of electron volts. For
a complete surface monolayer (n :x 105 /cm Z) with d t 4 -A and q = e,
AO = 73 eV. This value is too large to be physically reasonable. If measured
values of 5 versus surface coverage n for various alkali or alkaline earth
atoms adsorbed on metal substrates are plotted, the behavior illustrated in
Fig. B-2 is found. At first, doidn is comparable to the value expected from
Eq. (B-2), then decreases in magnitude. O(n) goes through a minimum, then
increases slightly to nearly equal the bulk work function of the adsorbed
species at about one monolayer coverage. Since the minimumn value of
9(n) is actually less than the bulk work function of the adsorbed species,
Fig. B-2 indicates that a practical thermionic emitter based on the surface
dipole moment of a surface monolayer could be constructed. Such cathodes
are in fact in wide use.

-71-
 C)
CID

LL-

+min
0 Omin
FRACTIONAL SURFACE COVERAGE 0
4, Fig. B-Z. General Behavior of Work Function as
Function of Surface Coverage for
Alkali or Alkaline Earth Element
Adsorbed on Metal Substrate

-72-
 I The observed work function minimum is well understood, at least
10
qualitatively. The point is that as the adatom surface density n increases,
2' interaction of adatoms results in a decrease in the charge per atom, q.
Equation (B-2) therefore exhibits a minimum. Early discussions of the

problem employed depolarization models in which the field of each surface

Si
dipole contributed to a reduction in the moment of other surface dipoles.
More complete and recent models are firmly based on quantum
mechanics. 12, 13 As an alkali metal atom approaches a surface to a
distance d, the adatom energy levels are raised by e2 Ireo d (Fig. B_ 3). The
alkali ns level, which lies above the Fermi level, is also broadened, with a
tail extending below the Fermi level. In other words, the adatom is partially
ionized (q < e). As adatom coverage increases, the alkali level is lowered,
and the adatom layer is gradually neutralized. A complete adatom mono-
layer is essentially neutral and metallic. Work function changes in the
limit of complete monolayer coverage also have been treated
theoretically. 14p 15
Experimental studies of the adsorption of alkali and alkaline earth
species on metal substrates are too numerous to list in detail in this report.
Cesium and barium have been favorite adsorbates, whereas tungsten has
been a favorite adsorbent. Work also has been carried out with nickel
substrates. 6-20 Experiments with these materials are particularly
relevant to cathode technology. Low-energy electron diffraction has been
widely used to determine the structure of adsorbed monolayers, whereas
work function measurements generally have been carried out by means of
either field emission or the contact potential difference technique. The 0 (n)
behavior illustrated in Fig. B-2 is found quite generally. Coadsorption of
21
two electropositive species has also been investigated.

Coadsorption of an electropositive species with oxygen on metal

i0
substrates, e. g., sodium on oxygen-covered tungsten (Na-O-W),
Cs-/-W0 0"ZZ Ba-O-W,23 BaO-W,24 and Cs-O-Ni, 10,25,26 has been

t _73-

, .v3-
 Iads 0-E)

E 1ads

•/// -- LARGE SEPARATION AATOM

METAL POTENTIAL WELL POTENTIAL n(E)

WELL
WL ADATOM - DENSITY OF STATES
a.

FEads ads
b
bk,

RESULTANT METAL-ADATOM POTENTIAL ENERGY

FOR SMALL SURFACE-ADATOM SEPARATION ADATOM DENSITY OF STATES

Fi g. B- 3. Potential Energy Diagrams. a. Metal and ionizable

donor species, wide separation. (The density of the
states of the atom, also shown, is a sharply peaked
function.) b. Metal and ionizable donor species at
small separation. (Note that the adatom density of
states is broadened.)

-74-
 investigated. In all of these experiments, low-energy electron diffraction,
generally in conjunction with work function measurements, was employed.
The result most pertinent to cathode technology is that a double layer is
formed with the alkali or alkaline earth above the oxygen regardless of the
10
order of adsorption. The same general behavior of O(n) illustrated in
Fig. B-Z is observed with the alkali (alkaline earth)-oxygen-substrate
system, and the minimum work function is generally lower than in the
absence of oxygen. For example, with Cs-O-W (lIZ) for O-Cs-W (112)]
10 1
Omirk 1t eV, while •min t-1.55 eV for Cs-W (112).6

-75-

El.
..........................................................
, .<ka
.t .....
 7e

APPENDIX B

REFERENCES

I. C. Herring, "The Atornistic Theory of Atornic Surfaces," in Metal

Interfaces, ASTM, Cleveland (1952).

2. C. Herring and M. H. Nichols, Rev. Mod. Phys. 21, 185 (1949).

3. E. Wigntr, Phys. Rev. 46, 1002 (1934).

4. E. Wigner and J. Bardeen, Phys. Rev. 48, 84 (1935).

5. B. F. Rzyanin, High Temp. 11, (1973) (English translation).

6. W. Gordy and W. J. 0. Thomas, Phys. Rev. 24, 439 (1956).

7. C. F. Gallo and W. L. Lama, IEEE Trans. Ind. Appl. IA 10,

496 (1974).
8. R. 0. Jenkins, Vacuum 19, 353 (1969).

9. G. A. Haas and R. E. Thomas, "Thermionic Emission and Work

Function," in Techniques of Metals Research, Vol. 6, Part 1,
E. Passaglia, ed., Interscience, New York (197Z).

10. J. M. Chen, J. Franklin Inst., 298, 255 (1974).

11. I. Langmuir, J. Am. Chem. Soc. 54, 2798 (1932).

12. J. W. Gadzuk, J. K. Hartman, and T. N. Rhodin, Phys. Rev. B 4,
241 (1971).
13. J. P. Muscat and D. M. Newns, J. Phys. C. 7, 2630 (1974).

14. N. D. Lang, Phys. Rev. B4, 4234 (1971).

15. K. F. Wojciechowski, Surface Sci. 55, 246 (1976).

16. V. M. Gavrilyuk, A. G. Naumovets, and A. G. Fedorus, Soy. Phys,

JETP 24, 899 (1967).
f17. A, G. Fedorus, Yu, M. Konoplev, and A. G. Naumovets, Soy. Phys.
Solid State i1, 160 (1969).

_77-
18. V. K. Medvedev and T. P. Smerka, Sov. Phys. Solid State 15, 507
(1973).

19. D. A. Gordetsky and Yu. P. Melneck, Surface Sci. 62, 647 (1977).

Z0. V. K. Medvedev and A. 1. Yakivchuk, Soy.. ys. Solid State 17, 7

(1975).

21. Yu. M. Konoplev, A. G. Nautnovets, and A. G. Fedorus, Soy. Phys.

Solid State 14, 273 (1972).

ZZ. C. A. Papageorgopoulos and J. M. Chen, J. Vacuum Sci. and

Technol. 9, 570 (1972).

23. Yu. S. Vedula, V. M. Gavrilyuk, and N. D. Morgulis, Sov. Phys.

Solid State I, 1569 (1960).

24. B. J. Hopkins and K. J. Ross, Brit. J. Appl. Phys. 16, 1319

(1965).
25. C. A. Papageorgopoulos and J. M. Chen, Solid State Communications
1_33, 1455 (1973).

26. C. A. Papageorgopoulos and J. M. Chen, Surface Sci. 52, 40 (1975).

-78-
 LABOUTOt¥ O'ERATIONS

'iaa Laboratory Operations cf The Aerospace Corporation is conducting

experimental and theoretical investigations necessary for the evaluation and
application of acientifiL advances to new military concepts and systems. Ver-
setility and flexibility have been developed to a high degree by the laborato-
ry parsonnal in dealing vith the many problems encountered in the Nation's
rapidly daveloping space systems. Expertise in the latest scientific develop-
ments in vitil to the accoaplishment of tasks related to these problems. The
laboratories that contribute to chis research are:

harophysice Laboratory: Aerodynamics; fluid dynamics; plasmadynamics;

chauical kinetics; engineering mechanics; flight dynanmcs; heat transfer;
high-power gas lasers, continuous and pulsed, IR, visible, tUV; laser physics;
lager resonator optics; laser effects and counterasasures.

Chesistry and PhysicLaborstor: Atmospheric reactions and optical back-

grounds; radiative transfer and atmospheric transmission; thermal and state-
specific reaction rates in rocket plumes; chemical thermodynamics and propul-
sion chemistry; laser isotope separation; chemistry and physics of particles;
space environmental and contamination effects on spacecraft materials; lubrica-
tion; surface chemistry of insulators and conductors; cathode materials; sen-
sor naterials and sensor optics; applied laser spectroscopy; atomic frequency
standardsl pollution and toxic materials monitoring.

Elactrnni.cs Reaearch Laboratory: Elect-nagnatic theory and propagation

phenomena; microuave and semiconductor devices and integrated circuits; quan-
tum electronics, lasers, and electro-optics; coinonication sciences, applied
electeonics, superconducting and electronic device physic.l millimeter-wave
and far-infrared technology.

Hat.rials Sien o Development of na matcrial.; coavosita

tersls;
mal, iraphtet ar ceramics; polymeric materials; waspon effects and
hardaned materials; materials for electronic devices; dimensionally stable
materials; chemical and structural analyses; stress corrosion; fatigu, of
metats.

Spacerm&-ences Laboratory: Atmospheric and ionospheric physics, radia-

tion "-:nm the atmosphere, density and composition of the atmosphere, auroras
and airglow; magnetospheric physics, comic rays, generation and propageLion
of plas-m waves in the magnetosphere; solar physics, x-ray astronomy; the effects
of ruclear explosions, magnetic storms, and solar activity on the earth's
atmosphere, ionosphere, and sagnatoaphere; the effects of optical, electroma--
natic, and particulate radiations in apace on space systems.